archaeologies of slavery and servitude (1)

8/17/2019 Archaeologies of Slavery and Servitude (1)

http://slidepdf.com/reader/full/archaeologies-of-slavery-and-servitude-1 1/19

rchaeologies of slavery and servitude

bringing

ew World perspectives to Roman ritain

Jane Webster

Vegetus, assistant slave of Montanus the slave of the august emperor, has

bought the girl Fortunata, by nationality a Diablintian, for 600

denari i.

She

is warranted healthy and not likely to run away. (Wr iting tablet, London);

A likely yellow girl about 17 ar 18 years old, has been accustomed to ali

kinds of house and garden wark. She is sold for no fault. Sound as a dollar.

(Broadside 1833,Charlestonl.é

Modern historians and sociologists, knowing full well the need to make

every allowance for particular variations in one society after another, have

disclosed nevertheless the universalist features of slavery across time and

place: to pretend otherwise is futile. (K.Bradleyl.é

This contribution addresses the archaeology of slavery, not as Roman archaeologists now

know it, but as 1hope it might one day appear. 1explore some of the approaches to material

culture developed by archaeologists who study slavery in the New World 4 a nd argue that

similar strategies might help us to see more clearly some of the least visible persons in the

Roman world: slaves and other servile groups.êI take Roman Britain as my starting point and,

to illustrate some of the things that New World approaches to the complex material world of

the unfree might offer, 1develop a case-study on slave-quarters (accommodations) that moves

from the plantations of Virginia and the Carolinas to prisons

erg astula

and roundhouses in

Roman Britain.

The proposals made here are offered in the hope of encouraging a revival of interest in an

inexplicably moribund field. Slavery, colonialism and empire have always gone hand in hand,

and the archaeology of slavery is in the ascendant from South and West Africa to Brazil, the

Caribbean, and the former slaving-ports of Britain. These are places, of course, where slavery

has a continuing legacy, and where descendant communities have fought hard for the experi-

ences of their ancestors to be acknowledged and articulated. No similar imperative exists to

promote the study of slavery in the Classical world, and perhaps for that reason an Archaeo-

logy of Roman slavery has never emerged as a sub-discipline in its own right. It is time it did,

and here 1put forward some suggestions, all informed by work on modem historical slavery in

N America, that 1hope will stimulate debate.

In the Roman world, and throughout the American colonies, slave-worked estates, gang

labour, rebellions, slave markets and auctions, and legal frameworks for bondage and manumis-

sion were facts of everyday life. Put this way, it does not seem at all surprising that an intel-

lectual c ross fertilisation go es on between ancient and modem

his torians

of slavery, and has

done since the 19705.Today, many ancient historians regard the comparative study of slavery

Writing tablet from No. 1 Poultry, London, dating to A.D. 80-120:see R.Tornlin, T he girl in question :

a new text from Roman London,

Britannia

34 (2003)41-5l.

From a broadside announcing a Public sale of Negroes , offered by Richard Clagett of Charleston, S.e.



6

Webster

as a stimulating pursuit, not a misguided one. Archaeologists, unfortunately, take quite the

opposite view. It is remarkable that the bibliography of the most recent Roman archaeological

synthesis? does not contain a single reference to New World slavery studies. But a lack of cross-

fertilisation is notable even among N American classical archaeologists, who might have been

expected to foster comparative

strategies f

Where is the archaeology of Roman slavery?

The archaeology of Roman slavery is clearly in the doldrums.? This is so even with refer-

ence to Roman Britain, a province for which the textual evidence is so limited that it is only

via the study of material culture that we can hope to gain some understanding of slave owner-

ship and slave experience. Despite this, the recently published 'roadmap' for the future of

Romano-British archaeologyl'' contains only one reference to slavery.' and makes no specific

reference to archaeological strategies for studying the enslaved. Part of the problem, no doubt,

is that slavery is an unpleasant subject - one at odds with the kind of Roman world that (even

today) many archaeologists wish to envisage. This is very noticeable with reference to Roman

Britain. For example, in

Roman villas and the countryside

(a book in which slavery is mention-

ed just three times), G. de Ia Bédoyere remarks that

'slave' is an emotive word and one to which we exclusively, and understandably, attach negative

connotations. But it was part of Roman life and not necessarily abused in the way we interpret it.1

2

It is hard to envisage any historical

context

in which slavery and abuse were not entwined, yet

many archaeologists instinctively suppose, as de Ia Bédoyere seems to do, that enslavement

was less awful in the Classical world than, say, in 19th-c. South Carolina.l '

6 Amongst ancient historians, interest in modern slavery has stimulated some explicitly comparative

work: see, e.g., H. D'Arms, SIaves at Roman convivia, in W. Slater (ed.),

D ining in a clas sica l context

(Ann Arbor 1991) 171-83;W. Phillips, Continuity and change in western slavery: ancient to modern

times, in M. Bush (ed.),

Serfdom and slavery: studies in leg al b ondage

(London 1996)71-88; and Bodel

in this issue.

7

F. Thompson,

The archaeology of G reek and Rom an slavery

(London 2003).

8 An interest in New World plantation archaeology flickered briefly in Italian archaeology in the 1980s,

notably in the work of A. Carandini at Settefinestre

Settefinestre. U na villa sch iavist ica neil Etruria

romana. Vol. 1. La villa nel suo insiem e [Modena 1984]),which he concluded with a chapter entitled

Schiavitü antica e moderna a confronto . In it he examined plantation sites in Virginia, South Carolina

and Louisiana. Although he suggested (188-89)that the study of N American plantation buildings might

potentially help us to identify ancient slave-quarters, comparative archaeology was not the central

focus of his discussion; as a Marxist scholar, Carandini primarily looked to the New World for evi-

dence to support his contention that slavery was the mainstay of the Roman economy, and that, when

slavery declined, Rome's economic fortunes also declined. Nevertheless, his explicit comparison ofNew

World and Roman archaeological data remains unique. The nearest anyone has come to this approach is

T.Smith, who notes briefly, in his Roman villas: a stu dy in social eiructu re (London 1997)229, that the

rows of identical buildings sited along one side of the villa courtyard at Levroux-Trégonce (Indre) are

spaced apart just like a Virginia plantation . Since the 1970s, very little compara tive work has



Archaealagies af slavery and servitude

163

This observation goes part of the way to informing the absence of a fully-developed

arc haeology of Roman slavery . A more fundamental factor, perhaps, is that many scholars

believe slaves to be beyond the reach of archaeology. Most Romanists assume slaves to be

arc haeologically invisible , leaving no clear fingerprint (other than shackles, collars and

buIlae 4 for excavators and finds specialists to identify. Indeed, the oft-repeated assertion of

slave invisibility pops up in R. P. [ackson s Epilogue to E. A. Thompsori s recent synthesis, even

after 260 pages of archaeological data.l Classical archaeologists still appear to believe that

slaves must remain invisible in the archaeological record because, as a body, they do not

leave a

specific

material fingerprint. A similar assumption once pervaded slavery scholarship

in the New World, but archaeologists there have long since accepted that they must study

slaves through the imposed material culture of their owners - a material fingerprint that

makes them (at first sight) barely distinguishable from the poorest European-Americans. They

have developed innovative strategies to help them interpret the artefacts they do find -

strategies which are grounded in a willingness to regard the form an artefact takes as being of

less importance than the logic informing its use. This understanding has helped New World

archaeologists not only to identify the

presence

of slaves in the archaeological record, but to

say something meaningful about slave

experience .

A more detailed look at work going on in N American slavery forms a prelude to examining

the approach taken in the Roman world. What do our New World counterparts do that we do

not, and what might we learn from them?

Seeing slavery the New World

The archaeology of African-American experience first emerged in the Americas in the

1960s.1

6

Some important plantation Great Houses had been excavated before this point, but

the Anglo-élite owners of these colonial homes, not their slaves, had been the target of that

work. Attitudes began to change in the 1960s and 1970s, as both the black consciousness and

Ne w History m ovements began to impact upon historical research in the U.5.A. In the 1980s,

African-American slavery emerged as a major research theme in American historical archaeo-

logy, and the topic remains a dominant one. The archaeology of slavery is particularly strong

along the eastern seaboard, home of the original 13 colonies established by the British after

1607. Much important work has focused on the tobacco plantations of Virginia and on the cotton

and rice plantations of N and S Carolina ?

N American work focuses on the complex ways in which slaves brought beliefs and practices

of African origin to bear upon the use and interpretation of slave-made or European-Amerícan

material things. One term commonly used to describe this adaptive mixing of Anglo-American

and .African cultural elements is creolization - to denote the varied processes of multicultu-

seeking archaeologists taking a comparative approach to Roman slavery. Several respondents informed

me that slavery in the classicalworld was better than inN America.

4 Bullae are small bronze plates bearing an inscription. These were attached to collars worn around the



164 e ster

ral adjustment through which African-American society was created. Several classes of creole

artefact and behaviour have been identified in New World slavery contexts, including colono-

ware pottery, architecture, tobacco pipes, and dietary practices.l''

At this point, it must be admitted that New World archaeologists do have one significant

advantage over their Classical counterparts: New World slaves were sourced mainly from W

Africa and from the are a of the Congo and Angola; though these places are characterised by

great ethnic diversity, they are all, to put it simply, African As a result, archaeologists and

anthropologists have been able to identify some broadly similar pattems of culture and belief

that, in the southem states of the U.s.A. in particular, inform the recognition of African

'survivals' (known as Africanisms ) in the creole material-culture of slaves of African des-

cent. In the Roman world, contrast, it was unusual for slaves of a given household or estate to

share a common ethnic origin; thus, it might be assumed, we cannot look for the Roman

equivalent of Africanisms .

One possible response to this problem is discussed below. In the meantime, I need only

emphasise that the majority of the artefacts used

y

N American slaves were of European-

American derivation, and that Africanisms (where present) reflect both a dialogue with,

and a manipulation of, European-American objects. At times, the explicitly African part of

that dialogue was quieter than at others. In Virginia, for example, where centuries of inter-

action preceded segregation, even slave-made artefacts, such as colonoware pottery vessels,

frequently took Anglo-American forms.ê? Crucially, New World archaeologists do not see this

as a problem inhibiting the study of the material world of the unfree: on the contrary, the

acceptance that African-Americans documented themselves through manipulation of a largely

'given' material culture is a

starting point

for analysis of the material world of slaves. Take,

for example, L. Ferguson's commentary on the work of [ohn Otto:

[ohn Otto ... shows that in the nineteenth century, slaves on Georgia's Cannori's Point plantation

were eating from glazed and decorated English bowls; on the other hand the planter's family were

using predominantly plates. Otto argues that slaves may have been using these bowls for eating

African-styled meals; Africans had been utilising ceramic bowls as their primary serving dishes

since long before colonial times. Thus, while the artefacts or 'lexicon' of slave meals at Cannori's

Point were European, the shape of those artefacts implies a foodways structure, or 'grammar', that

was strongly African. An ignorant visitor might observe that sIaves had adopted European

tablewares but didn't know quite how to handle them, preferring bowls to plates; a more informed

observer might see West African rules of etiquette employed with a new kind of bowl.

21

Central to creolization theory is the understanding that slave experience is best grasped

focusing on the ways slaves

used

the material things at their disposal. This is bome out by a

detailed look at slave-quarter architecture, the most relevant category of New World creole

material culture in terms of the concems of this article.

The architecture of slave-quarters has long been a major point of interest for plantation

archaeologists. Studies begin from the recognition that plantation slaves were usually required

to build their own accommodations. It has become clear that, although slaves were generalIy

expected to follow (to a lesser or greater degree) a European architectural blueprint, the



Archaealagies of slavery and servitude

165

results - a fascinating mixture of Anglo-American and African forms and techniques -

represents a truly creole hybrid. Important studies of slave-built accommodations have been

made throughout the plantation belt, but particularly in Virginia and the Carolinas. As early

as the 1970s, archaeologists working at the Yaughan and Curriboo plantations (Berkeley

County, SC) showed that the builders of the slave houses here employed cob-wallíng , a tech-

nique for building in day common in W Africa.P Since then, others have demonstrated further

similarities between W African homes and S Carolina slave-housing, including a preference for

gabled roofing and twin-roomed, rectangular floor plans; use of porches and verandas;

maintenance of African

y rd

systems for the organisation of shared social spaces; and the

preference (in the earliest slave-quarters) for outdoor hearths. American archaeologists have

also explored the origins and development of the 19th-c. hotgun house - perhaps the most

explicitly African-derived architectural form in the U.5.A.23 Others again have shown that

at least two aspects of African-derived architecture - southern timber framing and the use of

the veranda or porch - were gradually adopted by Anglo-American builders, becoming part of

the Anglo-American vernacular.

Seeing slavery in Roman Britain

In 1994, K R. Bradley challenged ancient historians to try to penetrate the psychological

world of the Roman slave ,24 and, in so doing, help others to understand what it meant to be a

slave. Approaches to Roman slavery by Old World archaeologists, however, tend simply to

seek indirect evidence for the pres ence of slaves. We assume their presence (in the apparent

absence of direct evidence) in quarries, mines and mills, on agricultural estates, and in other

work-places demanding large labour forces; and, serenely unaffected by the advances made by

New World archaeologists, we continue to believe that the

ex perience s

of these invisible

work-forces lie beyond our grasp, in places that archaeology cannot reach. For that reason, few

have sought to ask how Roman slaves might have used the objects at their disposal.

In Roman Britain, we have looked for sl aves principally in agricultural contexts, the

rationale being that large villas would probably have relied on field and domestic slave-

labour. Imperial estates would certainly have used slave labour, but few specific candidates

have been suggested for Britaín.P As in Gaul and Germany. some very large villas have been

síngled out as possibly employing large bodies of slaves or estate-workers, accommodated

either in the main villa complex or in associated non-villa settlementsP At Gatcombe, it has

22 Good intraductions to plantation architecture can be found in: Ferguson ibid. 63-82; S. [ones, The

African-American tradition in vernacular architecture, in Singleton 1985 (supra n.16) 195-213;Deetz

(supra n.17) 212-3l.

23

The seminal studies here are

Vlatch, Shotgun h ouses, N atural H istory 87.2 (1977)50-57 and id., The

A fro American tradition in d ec or ative art s (Cleveland, Ofl 1978).For a more general overview, see Deetz

(supra n.l7). Entered through the gable end, the shotgun house derives from WAfrican exemplars in both

plan and dimensions: even the name probably comes frorn the Yoruba (Nigerian) word io gun meaning

place of assembly. Shotgun houses first appeared in New Orleans in the early 1800s,brought there by

emancipated blacks from Haiti. They amalgama te French building techniques (Haiti was formerly a



166

Webster

been proposed that slaves (or other workers) may have occupied some of the many subsidiary

buildings located inside the walled villa enclosure, along with the non-villa settlements loca-

ted within the boundaries of this 6000-ha estate. A related argument is that aisled buildings

sited alongside winged corridor and courtyard villas could have housed labourers, tenants or

slaves.ê? 1 return to this below. Finally, it has also been suggested that overseers or bailiffs

could have managed large, slave-run, villa estates for absentee landlords. This was proposed,

for example, at Llantwit Major, where a 4th-c. villa was converted to bailiff occupation.ê'' In

all of these Romano-British cases, the presence of slaves has been inferred, rather than demon-

strated. Worse still, the interrogation of the material culture found within putative slave-

accommodations has had little or no part to play in making, or supporting, these inferences.

Putting New World lessons into practice: ergastula (slave-prisons)

Let us try to put some of the American ideas into practice by looking briefly at a category of

'slave specific' building known 'to have existed in the Roman provinces: the

ergastulum.

In the

Roman world, as in N America, many men and women actively resisted slavery. Troublesome

Roman slaves were often confined in purpose-built slave-prisons. In addition, private

ergastula

(or

carce resv

were a feature of many agricultural estates and rural villas. In such contexts,

er gast ula

served as communal work-houses for chained slaves

se rvi vin cti

who worked in the

fields.F In the 1st c. A.D. Columella described ergastula as semi-subterranean structures lit by

narrow windows, placed high enough above ground-level to prevent egress by

inmates.P

Few

such slave prisons have been positively identified (I return to this below), and only two

buildings in Britain have tentatively been placed in this category.

The first of these lay within Colchester, at the corner of Insula 15. Measuring 24.6 x 14.2 m

and set within its own walled enclosure, it was a semi-subterranean rectilinear building. The

principal building appears to have been constructed in the mid-2nd C.

34

It was partitioned into

four internally, with two larger rooms to the north and south being separated by parallel

partition walls enclosing a narrow inner chamber and a wooden staircase. With exterior walls

lm thick, and concrete floors 0.5 m thick in the larger rooms, this was a very solid structure. In

the largest and most subterranean of the inner spaces (room

D

timber beam slots were cut into

the concrete floor. No small finds were recovered from the unsealed floors, which were covered

with later rubbish deposits of varying dates. In his discussion of the purpose of this enigmatic

structure, M. R. Hull observed that the presence of a sunken chamber (room

D

containing beam

slots to which shackles might feasibly have been attached, might suggest the building was an

ergastulum .

35

He went on to reject this suggestion, however, because the room was faced with

good-quality tiles laid in pink mortar. This finish, Hull felt, was too elaborate for a slave

prison. He concluded, somewhat tentatively, that the structure was a mithraeum, built along

similar lines to that at Heddernheim (Frankfurt-am-Mainj.ê

A second possible

rg stulum

was excavated by

D

[ohnson at Chalk (Gravesend) in 1961

(fig. 1).37Measuring 13.9 x 3.9 m, this rectangular structure was smaller than the example at

28

Ibid.71.



Archaeologies of slavery and servitude

167

r---·-··------···-----··---···--·-··-······--··-·····-- -- ---1

I CHALK,

NR

GRAVESEND, 1961 ,

I

: - - - - - _ _ _ _ _ / - _ _ _ _ _ _ _ _ _ _

-

1 I I

I r of..m l z : : : n m

< - j , 1

r

c h i m i iC H - r w l l ti l a r c r

\

, p U l 1 d

h«

I

, k u l l A 1 Iwl,

' I : p

------

n ic h e s J c r la m p s

1 5

• •

Fig. 1. Plan of erg stulum at Chalk, Kent after Johnson [supra n.37] 116,

by

kind permission of the author .

Colchester.ê It was two storeys high, but no further internal spatial divisions were noted. The

foundations were cut

c.2

m deep into the natural chalk, with walls

C . O . S

m thick raised on this

subterranean base. A row of niches (probably for lamps) was found high up on the E wall. The

first phase of use at Chalk could not be dated, but the building was modified some time before

the end of the 3rd c. and destroyed by fire in c.300.

39

Prior to its destruction (that is, during

Phase III), the upper storey functioned as a store-room, whilst a small group of people appear

to have been living and working in the subterranean basement area. The northwest (and best

lit) comer of the Phase-Ill basement floor produced 37 antler pins and a quantity of antler-

working débris, indicating that at least one person had been engaged in pin-manufacturing

here. He or she was not alone: pits cut into the basement floor contained the remains of 3

infants, allless than a year old. Johnson went on to suggest that the people working here might

have been slaves, and that during Phase III the building may have functioned as an erg stu-

lum

His case rested in part on the architectural details outlined above, and in part on the

unusual nature of the finds from the Phase-Ill basement floor. A total of 11 whole or fragrnen-

tary ceramic vessels were recovered, three of which bore cross marks, incised after firing.

Johnson reasonably concluded that these were marks of (illiterate) ownership. Yet literacy was

in evidence too, because 6 graffiti were aiso found, inscribed on ceramic sherds: four of these

were very fragmentary inscriptions in Latin characters, but one made use of Greek script

< l >1ÍÀtKt : to Fel ix ), and the sixth took the form of a trident, incised on a rouletted

beaker.t

On

the basis of these finds, Johnson concluded that at least one of the occupants of the Chalk



8

Webster

basement dwellers here were preoccupied with naming, and even those who were not literate

felt a need to inscribe ownership upon their possessions. Unfree persons, in many historical

contexts, often share that concem, and bring it to bear upon the artefacts at their disposal. The

artefacts from Chalk thus seem to provide a tantalising glimpse of what it meant to be a slave

in Roman Britain. The identification of further

ergastula

(and the detailed interrogation of

their assemblages) should rightly be a priority for archaeologists interested in Roman slav-

ery.42

rgastula

were dedicated spaces for slaves; we know broadly what they looked like; and

we can be confident that they were widespread. Indeed, agricultural estates were provided

with ergastula as a matter of course .43 Why then, have so few examples actually been

found ?44It is surely a possibility that some sites have avoided identification simply because

slave prisons are something we are neither looking for nor expecting to encounter. Perhaps it is

time to look again at published semi-subterranean buildings similar in scale to Chalk, and to

re-assess their artefact assemblages through better-informed eyes. The group of excavated

subterranean features originally interpreted as villa cellars might offer a good starting

point.

If we could build up even a small corpus of likely slave prisons, we would have a group

of artefact assemblages that might - if we ask the kinds of questions New World archaeo-

logists favour - unlock the door to the material world of Britain s s er vi v in cti

Looking for enslaved Britons

As was noted above, households or estates in the Roman world potentially drew slaves from

diverse sources. It might be argued, then, that in the Roman provinces the diverse origins of

slave groups would make it very difficult to isolate the bedrocks of shared, pre-slavery tradi-

tions upon which creolization processes would necessarily be founded. Was not the Roman slave

body, in other words, too disparate to permit the approach I am advocating here?

One answer to this problem (for Britain at least) might be to begin any search for slaves and

other servile groups amongst those whose pre-conquest traditions

are

well understood:

indigenous Britons. Epigraphic and other textual evidence points clearly to the presence of

enslaved (and manumitted) Britons in Roman Britain. This is potentially good news for slavery

studies, because we know a great deal both about Iron Age lifeways and, increasingly, about the

maintenance of pre-existing traditions during the Roman period. It is important to grasp that

the desire to maintain pre-existing lifeways can often be especially deep amongst slaves and

servile or oppressed minorities, and that it is amongst these groups that creole adaptations are

most likely to occur. It follows from this that, if we can isolate instances in which pre-Roman

lifeways appear to be maintained and/or adapted in otherwise highly Romanized settings

(for example, within the context of villa complexes or forts), then we might, in some cases

perhaps, be glimpsing servile groups servicing civilian and military élites.

pigraphic evid ence

Epigraphy is an important source of information on slaves and, in particular, former slaves.

The latter are far more visible epigraphically than the former, and the epigraphic commemo-



Archaealagies af slavery and servitude

9

be highly

complex.s

Little work of this type has been undertaken for Roman Britain, for the

good reason that epigraphic references here to slaves

servi

and freedmen

liberti

are ex-

tremely rare. Only a handful of references to servi survive from Britain.s'' and only 30 to

freedmen (most of those come from the military zone and refer to the former slaves of military

personnel, busy dedicating rnonuments to their patrons' ).

The British epigraphic record may be slight, but it points to the existence of enslaved

Britons

as well as foreigners.ê'' In theory, subjects of the empire could not legally be enslaved.

The enslavement of Britons within

Bri ta nnia

might therefore seem a surprising phenomenon.

But there were many ways in which Britons might have become slaves in their own land. First,

some may have been enslaved at the time of the conquest, during the steady expansion of the

province, or as a result of insurrection. Second, from the dose of the Republican era it was the

practice in many provinces to top up the slave supply from internal (rather than external)

sources; these sources induded vernae,5 l orphans, exposed infants and other íoundlíngs.V

children sold as a result of poverty, self-sale for debt, and penal condemnation to slavery. In all

such cases, Britons could legally have been enslaved.

Many pre-Roman lron Age archaeologists now accept that a slave trade existed in Gaul and

Britain before the conquest, and that this trade supplied the Roman market. Taking this point

a little further, it seems reasonable to assume (as was also the case in the Hellenistic world

and, much later, in W Africa) that those pre-conquest communities developing a slave trade

would have done so in part because forms of slavery existed at home. Textual evidence

47

For Roman Italy, see, e.g.,H. Mouritsen, Roman freedmen and the urban economy: Pompeii in the first

century AD, in F. Senatore (ed.),

Pompei tra Sorre nto e Sarno: A tti dei terzo e quarto ciclo

di

co nfere nze di

geologia, storia e archeologia , Pompei, 1999-2 00 0

(Rome 2001) 1-27, and Mouritsen (supra n.46).

48 The best discussion of Rornano-British epigraphy remains A. Birley, T he peo ple o f Roman Brita in (Lon-

don 1979)145-50.Birleyhighlights the difficulties arising from the common practice of bestowing Greek

names upon slaves. The following

RIB

inscriptions on stone mention

servi: RIB

21, a base from London

set up by the provincial slave, Anencletus;

RIB

717, a building stone from Malton, Yorks;

RIB

902, an

altar from Old Carlisle; and RIB 1436, a tombstone from Haltonchesters. The latter is particularly

interesting in that it tells us that the slave Hardalio's tombstone was set up by a guild

collegium

of

fellow-slaves.

RIB

560, a tombstone from Chester, was erected by a master

d om inus

for 3 slave

children in his household. One of 3 stamps on the inside face of a wooden barrel-stave from Aldgate,

London, preserves the name of the slave Onesimus. A defixio from Bath B ri ta nnia 1982,298-99) con-

tains a petition by a group of 11people, three ofwhom (Cunitius, Lavendus and Mallonius) are stated

to be slaves. To this very smal group of references to

servi

can be added correspondence between slaves

on Vindolanda writing-tablets: A. Bowman and D. Thomas, The Vindo/anda writing tab/ets Tabulae

Vindolandenses l

(London 1994) 301, 302, 347; and the recent find of a writing tablet from No. 1

Poultry, London, documenting the purchase of the Gal icslave-gírl Fartunata: Tomlin 2003(supra n.l ).

The Fortunata tablet highlights the sometimes complex nature of slave purchase and ownership in

provincial Britain. Himself the property of an imperial slave named Montanus, Vegetus was wealthy

enough to expend 600

de nari i

on a

sl ve-girl



7

e ster

certainly points to a variety of Iron Age social relationships placing individuais in ambiguous

but certainly servile positions that might, either as a result of increased slave-trading for

Rome or following conquest by Rome, have been re-defined simply as slavery.P Fosterage was

common in Iron Age Britain, and some categories of fosterage (of foundlings, for example) might

well have produced subordinate sub-classes whose servitude would have continued under

Romanrule.

It has also been suggested that the emergence and elaboration of aisled buildings may

reflect a gradual social degradation of the extended families of increasingly powerful villa-

owners.ê This work (along with

T. Smith s much-debated extended family model for villa

occupancy=) reminds us both that the exact status of kin groups resident within villas could

have lain anywhere on a spectrum from equal, through tenant, to slave, and that there are

likely to have been different levels of servitude in Roman Britain. The strategies that might

help us to find one servile group (for example, servi vincti or chained field hands) may not

work elsewhere (for example, with semi-servile kin, or co loni or household slaves).

Villas villa estates and other agricu ltural settings

Villas must surely remain primary targets for archaeologists interested in Roman slavery

and servitude, but little real progress has been made in this area, with the presence of slaves on

villa sites often inferred but nowhere proven. A key problem= has been the difficulty of

isolating dedicated s lave-quarters on Roman sites. At first sight, this puts Romanists at a

distinct disadvantage when compared with their New World counterparts, who are often able

to study plantation slaves living in purpose-built, segregated quarters. On the other hand, as I

suggest below, we have surely excavated enough villas in Roman Britain by now to have at

least some idea where to begin looking for their servile residents.

But let us begin by addressing this old problem in a New World way. N American slave-

quarters are at once both archaeologically visible, and a key source of information on artefact

creolization. This is in large measure because slaves generally buili their own accommodatio n.

This leads to an obvious question: if Roman slaves had to build their own quarters, what would

they build? Given the potentially disparate origins of the slaves resident on any given site,

this is difficult to answer. But if we limit this question to enslaved or servile Britons, then an

answer does come to mind: they might well build roundhouses.

Roundhouses in villa contexts

There were large numbers of roundhouses, and a huge variety of housing traditions, in Roman

Britain, and I do not mean for a minute to suggest that the majori ty of roundhouses were built by

(or for) slaves: indeed, in some counties, such as Northamptonshire, substantial stone-built

roundhouses in non-villa contexts appear to represent the homes of wealthy families. At the

same time, however, Roman-period roundhouses were clearly key loci for the maintenance of

53

Referring to Gaul, Caesar Be 6.13) noted that Gallic peoples were oppressed by debt and servitude,




171

/

/

- -

-

/ /

/

/

, /

--_

.. . . •

_

-

,

\

\

\

.

,

\

I,

I I

, I

I

Siri, OF LATE R B

il

flOUSE

)(JOO()

(lOCJO()

I

, I

I

I

I

I

I

I I

I

üC)üUU

O 80J O

: = : = t = : : : : - = : - =

h

I ' I I I

Ir _. I.• __ , ,

,

: :

;ITE )F

j f;fNclrIA

I

I

l. . .....-_-_ -:. -. -:; ...._-.-:..--:::...-.-.-dl

-

-

- -

:

l-.::;_;

---_.

. . . - - . : - - . : = - - · , : 1

O•• • _-=IO==:::J

2D

__ . -=JO=======- :=::J H e : : '1

Fig. 2. Vindolanda: Severan annex and location of the roundhouses (Birley, Blake and Birley [infra n.61] p.

14; copyright Vindolanda Trust, reproduced by perrnission).

pre-existing ways of dwelIing,57and aspects of their presence in villa contexts are, to say the

least, intriguing (see below).

We know so little about slavery in Roman Britain that it might seem speculative, at best, to

propose not only that servile groups of indigenous origin built their own accommodations but

also that, given the choice, they might have built roundhouses. But we have one intriguing

piece of evidence suggesting that this might have been the case. An enigmatic and as yet unique

group of late Znd-c. roundhouses has been identified at Vindolanda, a fort on the Stanegate just

south of Hadrian's WalI (figs. 2-3 , These roundhouses are the only examples ever found in the

context of a Roman military site. Their presence there is

all

the more extraordinary given that,

although the roundhouse remained the dominant civilian architectural form in post-conquest

N

Britain, the zone of the WalI itself witnessed a rapid and standardised shift to rectilinear

building forms. lndeed, as S. Clarke has emphasized in his work on extra-mural settlement at

the fort of Newstead.P roundhouses do not even occur within v o

TheVindolanda roundhouses have been subjects ofdebate ever since the first two were found

in 1935, Between then and 1978 a total of 7 examples were located. Their spacing (back-to-back

in rows of 5, facing onto streets) suggested (as proved to be the case) that there might be more



172

e ster

Fig. 3. Vindolanda: possible reconstruction of the roundhouses (Birley, Blake and Birley [supra n.61] p. 18;

copyright Vindolanda Trust, reproduced by pennission).

awaiting discovery. In P. Bidwell's comprehensive study,59these roundhouses were interpreted

as accommodation for conscripted labour. Contrasting the regimented layout of the houses with

their varied (but clearly vernacular) construction, Bidwell suggested that the occupants had

been allowed to construct housing in their own style on plots laid out by a military surveyor.s''

The discovery in

1997

and excavation of two more roundhouses at Vindolanda prompted a

fur-

ther re-examination'' during which environmental samples were taken from a hearth in one

building and a clay floor in another: both produced mammal bone, and oat and hulled barley

grains. Six more huts were found in 2000 62 and one of these was fully excavated, but the only

finds located on an otherwise very clean floor were two bronze studs. On an artefact-rich site

such as Vindolanda, the paucity of finds from the roundhouse floors is in itself instructive, and

in my view Bidwell's interpretation of these buildings as the self-built dwellings of forced

labourers who chose to build in the round remains the most convíncing.s''

59 P. Bidwell, The Rom an fort of V indolanda at C hesterholm Northumberland (London 1985).

60

Ibid.31.



Archaealagies slavery and servitude

173

Roundhouse 369

Líoe 01 Mfldiaval plO()QhinO

m

.::::::':':':_:"'::::-.:-:-="":::>.

_ .c.= .::.::..:: ::f

Fig. 4. Redlands Fann, Stanwick: plan of Structure 369 (Keevill and Booth (infra n.63] p. 23; copyright P.

Booth and G. Keevill, reproduced by permission).

Perhaps the same explanation can be put forward to account for the occurrence of roundhouses

in villa contexts. It is well known that many Romano-British villas overlie Iron Age round-

house settlements. Indeed, the progression from circular 'hut' to rectilinear stone villa is one of

the classic markers of the 'Romanization' of 5 Britain. 50 what are we to make of roundhouses

occurring within villa complexes of 2nd-4th c. date? It is surely a remarkable fact that one of

the largest circular dwellings excavated in Britain was found not on an lron Age site, but in the

grounds of a mid-3rd c. villa at Redlands Farm

(Northantsj.v'

Measuring 13.7 m in diameter

internally, this building (5tructure 369) had entrances to the east and west, and appears to

have served simultaneously as a sleeping place, kitchen and byre (fig. 4). A series of intercut-



174

Webster

ting hearths or ovens at the centre suggested that the house was in use for a lengthy period. It

sat in a courtyard area, close to the heart of the villa complex, but separated from the main

residence by a small rectangular barn. What are we to make of Structure 369 and other villa

roundhouses like it? As scholars have pointed out, we cannot see them wholly in terms of the

progressive Romanization of an Iron Age building tradítion.s No less importantly, villa

roundhouses in some regions (including 3rd- and 4th-c. examples from Oxfordshire) actually

post-date the decline of the domestic roundhouse form in the surrounding landscape.s To put it

another way, villa-owners were apparently maintaining a tradition that had died out on

lower-status sites.

As G. Keevill and P. Booth have pointed out in their discussion of villa roundhouses in

Oxfordshire and Northamptonshire, where circular buildings were in contemporary use with

more obviously Romanised structures, it is easy to infer a social distinction between the

inhabitants of the different buildings, but less easy to define what this distinction

was . ?

Smith regards the juxtaposition of rectilinear and circular domestic structures as a variant on

his proposed extended family model for villa occupancy.s but exca vated villa roundhouses

produce noticeably fewer finds than other buildings in villa complexes, making it very difficult

to argue for equity b etween their occupants s?

Keevill and Booth?? raise the interesting possibility that villa roundhouses might be read

as a very pointed statement about relative social status: that is, powerful villa-owners may

have exercised dominion over their inferiors by housing them in circular buildings. Similarly,

in his discussion of the Holme House roundhouse, Harding remarked that the native

building type had no doubt been retained because it was cons idered adequate as living-

quarters for farm-labourers ar slaves, to whom in any case, native ways may

have seemed

preferable to the trappings of Romanisation=.? In my view, it seems highly likely that villa

roundhouses may have been the dwellings of servile groups whose status lay somewhere along

the continuum between slavery and degraded kin discussed above, and who had built their own

accommodations.

With this in mind, let us approach these structures in the way that N American archaeolo-

gists would assess slave-built plantation houses. Focusing on both architecture and the manipu-

lation of artefacts within slave-built accommodations, plantation archaeologists explore the

The fact that roundhouses were built and occupied alongside, or even successive to, buildings of

rectilinear plan in itself implies (as D. Harding notes with reference to the Holme House roundhouse

[Holme House P iercebridge : excavat ions 1969-70 (Edinburgh 1984) 18]) that the roundhouse form was

de liberately retained or revived in parallel with dominant Roman practice . See also R. Hingley, The

imperial context in Romano-British studies and proposals for a new understanding of social change, in

P. Funari, M. Hall and S. [ones (edd.),

Historic al archaeology backfrom the edge

(London 1999) 145-46.

Keevill and Booth (supra n.64) 41-42.

67

Ibid.40

68 Smith 1985(supra n.55).




175

dialectic between African lifeways and an imposed material culture. In that dialectic, they

find slave experience and slave culture. Can we use this approach with villa roundhouses?

As discussed above, there is much to suggest that various categories of unfree Briton existed

both before and after the conquest. It also seems reasonable to suggest that (as was common

practice in the New World) some of these servile Britons would have built their own accommo-

dations. This was certainly likely at Vindolanda. At the same time, we know a great de al

about Later pre-Roman lron Age indigenous architectural traditions, and about the logic in-

forming the use of dwelling spaces. Specifically, it is clear that the movement of the sun around

E-facing roundhouses was an important structuring principle for many communities in lron Age

Britain.F The Iron Age concept of the sunwise path can perhaps be seen as an equivalent of

the African cosmological traditions informing plantation architecture and

artefacts ê

How,

then, did the sunwise path fare in the Roman world? It seems reasonable to suggest that the

cosmologicallogic of the sunwise path would soon be irrelevant to those living in buildings

with corners.õ but what if one

continued

to live in the round? One possibility here might be to

look at artefact patterning on villa roundhouses to see if Iron Age cosmological principles are

maintained within them, and to ask at the same time whether potentially creole artefact

categories can be identified in such locations.

Roundhouses have been fully or partially uncovered in the course of large-scale villa

excavations at a number of sítes. A brief look at the work at Barnsley Park and Bancroft will

highlight the problems, but also the potentials, that emerge when attempting a re-analysis of

the kind advocated here. Six circular structures were identified at Barnsley Park, a villa

complex occupied from AD. 140 to 380.

76

Two of the circular structures (F and G) were discovered

in the S yard of the villa complex and dated to Phase 2 (275-315); two more (K and L) were

uncovered in the N yard and dated to Phase 3 (315-40); finally, two further examples (Q in the

N yard and R in the S yard) were dated to Phase 4 (340-60). Structure Q was still in use in 360-

72

Important studies include A. Fitzpatrick, Outside in: the strueture of an Early lron Age house at

Dunston Park, Thateham, Berkshire, i n ido and E. Morris (edd.),

The lr on Age in W essex: re cent work

(Trust for Wessex Arehaeology 1994)68-72;A Fitzpatriek, Everyday life in lron Age Wessex, in A

Gwilt and C. Haselgrove (edd.),

Reco nstructing Iron Age societie s

(Oxford 1997) 73-86; and

Oswald,

A doorway on th e past, ibid. 87-95.

73 See Ferguson (supra n.20) on the Bakongo eosmogram as a referent structuring the use of slave-made

eeramics in SCarolina.

74 Taylor (supra n.57) has suggested that patterns in the usage and layout of spaee observable in lron Age-

and Roman-period roundhouses were maintained in 2nd- and 3rd-c. aisled buildings in parts of the E

Midlands. Taylor did not look specifieally at eosmologieal referents, preferring to foeus on the

segregation of private and publie spaees within buildings; on which see also Hingley (supra n.27), and

id., Public and private spaee: domestic organisation and gender relations among lron Age and Romano-

British households, in R. Samson (ed.),

The socia l archaeology of houses

(Edinburgh 1990) 125-49. Nor

did Taylor assess how far, if at ali, eosmological referents intimately linked with building in the round



176 Webster

75 (Phase 5/6) and thus survived the major reconstruction of the villa in c.360.

77

The excavators

initially dismissed the idea that these buildings could have been houses, asserting that any

circular or irregular dry stone outline must surely be regarded as purely a dry-stone walled

structure, perhaps informally roofed, or perhaps unroofed, as an animal pen .78 As several

subsequent writers have

suggested. ?

it is more likely that some (perhaps all) of these struc-

tures were roundhouses.s?

On balance, it seems very likely that the Bamsley Park buildings represent servile quarters,

but, having made that suggestion, what more can be said about them? The key point is that

none of the previous interpretations was based on analysis of the artefacts from these

structures.

It

is not the case that there are

finds from the Bamsley Park roundhouses; the

problem is that no-one has thought to look beyond the circular form of the buildings to the use

to which artefacts were put within them.ê We can note, though, that several of the

roundhouses here had E or SE-facing.entrances, like the Iron Age roundhouses which were

regularly oriented towards sunrise at the equinoxes and midwinter.V A good number of finds

appear to have come from the general area of building Q, a substantial (9.5 m diameter)

circular building with an E-facing entrance.

The final report on excavations at the extensive villa and temple-mausoleum complex at

Bancroft'ê provides information on two circular structures (Buildings 4 and 11) associated with

a villa of the late 1st to late 2nd c., and another (Building 12) constructed in the late 2nd or

early 3rd c., at a time when the villa site otherwise appears to have been devoid of human

activity.ê These buildings were excavated under modem conditions and using a quadrant

excavation system that would, potentially, facilitate detailed analysis of artefact-patterning

within circular buildings.ê'' But even here problems emerge. Building

4

(8.6 x 6.0 m overall) lay

at the centre of the villa farmyard. There was no evidence for an entrance, and the floor was of

beaten earth. This building was first located, and partially excavated, in 1976; the original

excavators reported the presence of post-holes and a pit in the interior, but they were not

subsequently relocated. The more substantial (13 m diam. externally) Building 11 was sited on

the edge of the farmyard area; it probably had an E-facing entrance.f Part of the interior on

the E side was surfaced with cobbles, and the remains of

a

hearth were located at the centre.

Pottery of 2nd-c. date was recovered from the interior, and the building was interpreted as

77

Webster 1981(supra n.75) 9l.

78

Webster and Smith 1982 (supra n.75) 89.

79

Including Hingley (supra n.27),Smith 1985(supra n.55), and Keevill and Booth 1997(supra n.64) 37.

8 Smith (1985, supra n.55) has also attempted fit Barnsley Park inta his 'extended kin' model of villa

accupancy, suggesting that three families, at first living in circular buildings and holding land in joint

ownership, were converted into a single modified kin group living within a single hall villa. Webster

and Smith (supra n.68)87have countered with the suggestion that the roundhouses represent peripheral

occupation by inferior workers in outbuildings with animais .

81

Part of the difficulty here lies with the presentation of the finds data in the published reports (Webster

1981, supra n.75; Webster and Smith 1982,supra n.75). Simply put, it is not possible to determine what




177

providing accommodation for farm workers.V ln the case of Building 12 (11.7 m diam.), no trace

of an entrance was found, and the interior was devoid of contemporary features.

In order to creolize villa roundhouses, we need to know what was on their floors, and

where. As this brief look at Barnsley Park and Bancroft makes plain, that information is diffi-

cult or impossible to reconstruct, even on sites excavated in recent years. We know, however,

that a good number of these buildings faced east; and it is interesting to note that a number of

excavators have found traces of internal, and sometimes radial, partitioning segregating villa

roundhouse interiors.P Both features suggest the maintenance of pre-existing cosmological

referents. As for artefacts, we may als o have to come to terms with the possibility that one of

the clearest s ignatures of servile villa accommodation may actualIy be a paucity of finds. In

this context, a number of likely villa roundhouses have been identified from the air but not

excavated.ê? more may yet appear. It should be clear by now that the question we need to ask

in future is not What form did villa roundhouse artefacts take? , but How were they being

used? The answer might be that the occupants of villa roundhouses were using artefacts

according to an underlying, Iron Age logic. If so, we can begin to ask whether we might be listen-

ing to the dialogue of the unfree. At the very least, we can begin to re-assess old data in depth,

and put new strategies in place that will inform future work.

Further thoughts on villas

Dedicated slave-quarters might be hard to find on villa sites, but surely we have excavated

enough villas in Britain to have so m e idea as to where to begin to look for servile, as opposed to

élite, residents. Villa roundhouses, I have suggested, are one possibility. What else? One

obvious target might be aisled buildings and related ancillary structures situated within

double- or multiple-compound villas. Despite all that we now know about variations in the

date and function of aisled buildings in Britain, this idea still has considerable merit.f? The

central difficulty is our inability to differentiate with confidence between buildings housing

extended families, semi-servile kin , tenants, servants and slaves. But we have to start some-

where, and villas with compounds exhibiting clear economic differentiation (including Wood-

chester and Chedworth [Glos.], North WraxalI [Wilts.], Bignor [Sussex], and Gorhambury

[Herts.]) seem as good a place as any. At Gorhambury, for example, the compound to the east of

the villa contained aisled buildings and an isolated bathhouse, and is usually interpreted as a

wo rkers compound . Why do we not look again at the assemblages from this and similar

large sites to see what more can be said about the use of artefacts within these architectural

spaces? We might also explore whether some of the ancillary buildings on these villa

complexes might not have served as rural

erg stul

In so doing, we might find spaces in which

(and artefacts through which) servile groups shaped their identity in Roman Britain.

One other villa context in which we should be looking for slaves is amongst the dead. In the

New World, some of the best evidence for slave experience comes from plantation cemeteries.V



178

Webster

Because slaves buried each other, and because grave goods were commonly interred even with

Christian slaves, cemeteries provide important evidence about aspects of the social lives of

slaves that were largely hidden from slave-owners. These aspects include inter-slave hierar-

chies, economic networking, and the maintenance of African spiritual traditions and social

categories. Even in the New World, plantation cemeteries are not easy to find, and in the

Roman world the situation is far worse. We know very little about the disposal of dead slaves

in Roman Britain, for example, and have yet to identify a slave cemetery; we cannot even be

certain that such sites existed. Certainly, few sites in the Roman world as a whole have been

identified as slave cemeteries. The best-known candidates are cemeteries associated with the

mining settlements of Portugal, such as Dehesa and Corta Lago (Riotinto), which have pro-

duced cremated remains of possible slave-workers.P Examples of slave cemeteries in villa

contexts are especially rare: a cremation cemetery located outside the enclosure wall of the

villa at

Koln-Müngersdorf

contains

55

burials (Flavian to 2nd c.) that may represent the

remains of slaves or farm-workers dispatched with few grave goods, and without cremation

urns;94 in Britain, the villa at Hambleden has produced 100 infant burials, leading to specula-

tions about infant exposure on a

slave-run establishment i

The number of villa cemeteries identified in Britain remains small (perhaps because, until

recently, we have rarely looked beyond the walls enclosing villa complexes), but are there

ways in which we could look for slaves among the human remains from the excavated sites? At

Patuxent Point (Md.), to give just one example, physical anthropologists examining human

remains from a 17th-c. farm graveyard identified massive

upper-torso

development in the only

body on the site to be interred with grave goods: they suggested that the young man may have

been a slave who lived and died alongside his owners after a brief life

of

hard manual

lab-

our ?

What more might we be doing with the skeletal and cremated human remains already at

our disposal in Roman Britain, not to mention with new finds?

Conclusions

Some readers may feel that paucity of finds will hamper all the avenues of enquiry have

outlined. My response is that we don't know if that will actually be the case until we try. The

error we make in the study of Roman slavery is that we under-estimate the power of artefacts,

assuming that they can have no part to play in articulating stories of slavery and servitude.

lndeed, amongst the British sites discussed above, Chalk stands out as the sole example where

interpretation

a building as a slave dwelling was informed by observations about the use and

manipulation

artefacts. Most

the sites discussed produced material culture: they just do not

produce very much of it. But we will never come doser to seeing slavery and servitude if we are

not prepared to extract every nuance, and wring every piece of information, out of the little we

do have. Perhaps the most important lesson to be gleaned from the New World is the

under-

standing that we cannot see the worlds the unfree made for themselves without constant refer-




9

ence to the ways in which they manipulated the few (often unremarkable) artefacts at their

disposal. The archaeology of Roman slavery - insofar as such a thing exists - remains, by

contrast, heavily reliant on the bigger picture offered by Classical literature. As a result, we

have drastically under-estimated the potential of the small things through which Roman

slaves, like their counterparts in the New World, wrote themselves into history.

[email protected] School of Historical Studies, Univ. of Newcastle-upon-Tyne

Acknowledgements

Versions of this paper were first aired in seminar series at Kíng s College London and at Durham. I am

grateful to colleagues and students from both institutions for their comments and ideas. My appeal to

ROMARCH brought very helpful information from a number of colleagues, including E. Fentress, M. T. Boat-

wright and Bode . I owe special debt to N. Cooper. who commented on the Chalk assemblage, and to H. Mou-

ritsen, who provided a summary in English of Carandini (1984,supra n.8)and commented on an earlier draft

of this paper. D. Mattingly and E. Fentress also commented on an earlier draft and suggested many improve-

ments to the textoI should also like to thank the Caird Trustees (National Maritime Museum, Greenwich),

whose award of a Caird Senior Research Fellowship supported my research on slavery between 2001 and

2003.

archaeologies of slavery and servitude (1)

Documents