Download file in PDF format: TRAC 1993: Slavish Nonsense or the Talking Tool (pp. 122–140)
Slavish nonsense or the talking tool
I should begin with my excuses for being a contributor amongst this
august body of learned, but no less trendy for that, Romanists, for I am at
heart a medieval archaeologist The main excuse is that it seemed
pointless leaving the subject of slavery to Roman archaeologists as there
is no sign of any bothering themselves with it Let me demonstrate just
how silent Roman archaeologists are on the subject
Occasionally, as in the case of J. T. Smith (1978; 1982) or Ramsay MacMullen (1987) (although he is no archaeologist, but his work is cited by
them), the scholar takes the hard line that there were scarcely any slaves
in the north-west of the Roman empire. But more commonly the issue of
slavery is simply skirted.
In Richmond’s revision of Collingwood’s The Archaeology of Roman
Britain, for instance, no form of dependent labour features in the index
(while “spokeshave” does!). And while I managed to find one abstract
reference to slavery (p. 133), all references to actual dependent labourers
on villas (chapter seven) are to “farm-workers”. John Percival is similarly
quiet on the question of slave residence in The Roman Villa, mentioning
slavery only twice according to the index: once to mention its demise and
once while quoting the social historian Rostovtzeff. In The Archaeology of SLA VrSH NONSENSE OR TIlE TALKING TOOL 123
the Roman Economy, Kevin Greene (p. 69) explicitly admits that the question of slaves on villas will not be addressed, being too intractable. In The
Roman Villa in Britain, edited by Rivet, slaves appear in the index four
times only, three times in Rivet’s own contribution on social and economic aspects. Look them up and three of the four entries are to Strabo
and the export of slaves from Britain in the Iron Age. And the fourth reference is in the index by mistake! Rivet mentions servile labour but in the
medieval sense of serfdom not slavery. While slaves make the index of
Anthony King’s Roman Gaul and Germany some thirteen times, most of
these instances recount the same anecdote about the Gauls willing to exchange a slave for an amphora of wine or stress that there were no slaves.
Slaves, for example, were not involved in mining (p. 121) and had no part
in the production of Samian pottery (p. 130). We also hear that “villas
were often large enough to accommodate a substantial number of fannworkers” (p. 101) but that there is “little support to the idea that there
were many slaves in Gaul and Germany, particularly in the countryside”
(p. 95). So, who were King’s “fannworkers” if not slaves? “Clients”
(clientes) in the early empire and “tenants” (colon i) in the late empire.
Martin Millett’s The Romanization of Britain has slaves down in the index
five times. Numbers one and two are references to Strabo again and the
exportation of slaves in the Iron Age. Tum to Millett’s references three
and four and we find that slaves appear in figures tabulating social personae recorded in epigraphic evidence. Millett mentions slaves one more
time, when they are being replaced by the coloni in the late empire.
Almost comically, when mentioning slavery many archaeologists make
reference to Strabo and an Iron-Age slave trade (before Roman Britain)
and a further reference to the “end of slavery”1 (after Roman Britain).
There is, seemingly, no middle! If this were not odd enough, the most
common aspect of slavery mentioned by archaeologists is in the context
of trade. Note that mentions of slaves by Roman archaeologists regularly
accompany illustrations of amphorae and other Roman goodies, in the
context of trade goods, or the mentions accompany plans of hill forts;
slavery is equated with tribal warfare, it is almost never suggested that
these Iron-Age slaves might have been part of an indigenous institution
of slavery. Thus, slaves spring from somewhere and are traded, but then
disappear from the narrative. One must assume that British scholars see
them all being swallowed up by that cesspit of human excesses known as
Italy. Moreover, it is extraordinary that a Romanist should characterise
the late empire as a time when slavery disappeared as the institution
gave way to a more medieval fonn of serfdom when the same scholar has 124 SAMSON
already said that slavery was uncommon, and that rural slavery was almost unknown in the western provinces. Would this not mean that the
fourth and fifth century saw the transfonnation of a few uIDan domestic
slaves into a multitude of servile peasants? In any case, I (1992b; 1994)
have tried to show that rural, back-breaking agricultural slavery did not
end at all in the late Roman period, but thrived in the early Middle Ages,
at least it did in Gaul and Spain. 2
Now, I have an idea that rural slavery was much more common in the
western empire than is usually accepted, certainly much more common
than is accepted by British archaeologists, and even more common than is
accepted by most Marxist historians. I suggest that there were millions of
slaves in the western empire at anyone time. Perhaps lover argue the
point for polemical reasons, but it is sadly in need of far more strenuous
arguing than even I can manage.
How to prove my point? Well, there seem to be three major lines that
might be followed
PROOF 1: HISTORIOGRAPHY
The first is to do some historiography, deconstruct the work of past and
contemporary historians and archaeologists, the truly great and Martin
Millet too. Archaeologists certainly have little to say about slavery. A
historical critique would have us ask whether the cause is indeed the
scarcity of ancient slaves, simply the absence of evidence, or something
else. And if there are other reasons why archaeologists have ignored
slavery, what are they and why are they.
One could probably show that anti-Marxist and anti-Communist sentiment, conscious or unconscious, not the absence of evidence has largely
been to blame for the minute amount of archaeological research on slavery. You may wish to note the next time you read why an archaeologist is
going to skip over the question of slavery whether there is any mention
of the wrong-headedness of Marxist theory. In the typical western historian’s misunderstanding of what is meant by a “slave economy”, it is
generally taken to mean that a huge amount of general production was
undertaken by slaves. The Marxist notion of “slave economy” is rejected
on the grounds that the larger portion of all ancient production was performed by the free, by tenants, and by servile peasants. Ramsay
MacMullen (1987) went one step further and suggested that most slaves
did not perform real productive labour at all, being primarily cleaners,
baroers, and sex-workers. In short there were not that many slaves and SLA VISH NONSENSE OR TIlE TALKING TOOL 125
those few there were did not count anyway. By a slave economy orthodox Marxists mean (or meant, since they are probably now all dead or
converted) only that the role played by slave production was somehow
crucial in detennining the structure of the society. In other words, slaves
need not have been numerous to have been significant to historical developments. As it happens, most orthodox Marxists have assumed that
slaves in the western empire were not at all numerous. And I believe
them too to be wrong.
It is worth noting that in Branigan and Miles’s edited Villa Economies
there are only two mentions of slaves and they coincide with reference to
Carandini’s excavations at Settefinestre. Carandini is overtly Marxist
scholastically and emphasises slavery in all his works. Indeed, the word
is in the very title of his book on the villa excavations. Note too that there
is the unconscious non-reflexive parrot-fashion recital of what others
have already said. Thus, although there is nothing remarkable about
Koln-Miingersdorf, the villa appears time and time again when an
archaeologist wishes to briefly mention slaves and pass on to other
things. There is nothing so special about it that the same could not be said
of hundreds of other villas in western Europe. When a Roman archaeologist wishes to pay lip service to slavery, he or she digs through a few
handy books and, consulting the index and without being any the wiser,
reproduces either Settefinestre in Italy or Koln-Miingersdorf in Germany
or Montmaurin in France just as dozens of others have done before, and
only because their original excavators had the courage to mention slaves.
This original mention of slavery is then faithfully duplicated in a synthetic work.
In the typical British villa excavation report or synthetic work, there are
few if any mentions of slaves. Villa workers are called field hands or farm
workers. If the British archaeologist mentions slavery, the chances are
almost odds on that the passage is quoting or paraphrasing a foreign
archaeologist or just maybe a social historian. Most British archaeologists
make the mistake of believing that the terms farm hands and field hands
and labourers are neutral and cover up our ignorance of the exact form of
their dependency on the villa grandees. In fact, the terms are appropriate
to the modern world and carry with them the modern implications of
Of course, this is the result of more than overt or unconsious shyness
towards Marxism, for these very bourgeois forms of dependent labour so
regularly ascribed to villa workers fit neatly in the long disciplinary tradition of western historical scholarship, which has concentrated on things 126 SAMSON
of bourgeois values, such as the doings of great men and the world of the
mind rather than social history (which only today is beginning to rival
political and intellectual history) and the oppression of women and the
poor. Yet, even today the rich and their mansions are far more often the
object of study than the poor, who lived such uninteresting drab lives in
uninteresting and drab houses.
Oddly, despite the slow rise of social history and feminist studies in
the last three decades, this period has seen further influences on Roman
archaeologists which have neutralised these gains. And the result is an
even greater tendency to overlook ancient slavery. In particular, there has
been a marked reaction against old-style classicism as archaeology has
come more into its own. The organiser of the very first TRAC went so far
as to say that Roman archaeologists and Roman historians should work
independently, only coming together from time to time to check results.
This was a natural reaction against the /I archaeology-as-handmaiden”
syndrome (one already experienced in medieval studies). The attitude is
perfectly understandable, utterly indefensible, but perfectly understandable. Imbued with the classics, it was once easy for Romanists in Britain
to paint a picture of the villa using the pallet of Columella, Cato, Pliny,
and Varro. Twenty years ago, John Burke in Life in the Villa in Roman
Britain was still able to write in the old style (p. 41):
But the slaves remained constant; or, rather, there were new
generations and perhaps different races of slaves. The civilising
influences which led the Romano-British to ape their conquerors
and practise some of the refinements of living, including the art of
philosophical conversation over the dinner table, did not extend
to the treatment of those unfortunates whose labours made possible the accumulation of so much wealth. There were slaves in
Rome and in the provinces, and always had been.
But the easy use of classical texts has ended. Brooches are the new texts.
Burke was happy to stress the monstrosity of Roman masters towards
their ubiquitous slaves. And this image of Romans is something of a
cliche drawn from the classical texts, the result of reading too much
Suetonius and Juvenal. It clearly needed some revision, but the “New
Archaeology” wanted rid of all that old-fashioned culture history.
Romanists were the least affected (or infected) by New Archaeology, but
they were nevertheless touched, shared some of the seventies’ enthrallment with “continuity”, experienced some of the infiltration of a smidgin
of social anthropology into British archaeology, and a few youngs ones SLAVISH NONSENSE OR lHE TALKING TOOL 127
have even adopted the post-processual cockiness that everything before
Foucault was wrong. This has conspired to produce images of RomanoBritish society regulated or dominated by kinship ties, ideological meanings, and “central-place” geography. Down with the toga! In Britain,
Romans were just down-the-line gift-exchanging Iron-Age tnbesfolk in
rectangular houses or, worse, contextual manipulators of ideological
So, you see, Roman archaeologists are happy not seeing slaves for several reasons. Slavery falls outright into the arena of social history, and
a Marxist circus at that, a place where few Roman archaeologists stray.
Traditional, conservative, bourgeois Romanists prefer military and
imperial political history, annies and administration; New Archaeologists
prefer Thiessen polygons, positive systems’ feedback, and any number of
mathematical equations; both utilise economic models that are often capitalistic in their assumptions and applications; Hodderite contexualists
are largely trapped in the mind game of “meanings” and seem to believe
that ideological thought is the moving force in history; the trendiest postprocessualists rarely get beyond a reiteration of all the essential theoreticallitany of agency, recursivity, differance, and phenomenological hypertextual routinisation; while most practical Roman archaeologists, like
practical archaeologists of all periods, spend most of their time counting,
measuring, and typologically ordering their pottery, brooches, and sites.
To those who do not believe that the disciplinary tradition plays a dominant role in Roman archaeologists not mentioning slavery, I ask this question. How is it that almost unreadable abstractions on “fields of discourse” may end up in the bibliographies of Roman archaeologists, along
with references to French semioticians, while the superb works of the
social historians Keith Bradley and Keith Hopkins are almost never to be
PROOF 2: POSITIVE EVIDENCE
The second line of proving the existence and extent of slavery is the
simplest We could just muster all the positive evidence there is. For
archaeologists this is not easy. Can we recover what might be interpreted
as the archaeological paraphernalia of slave markets, such as the market
itself or slave tallies for accounting? Are ownership tags known in the
archaeological record? Now, this question I invented as a rhetorical one,
not being a Roman archaeologist and shying away from anything as
common as evidence. So, you can imagine my surprise when I promptly 128 SAMSON
came across three examples of slave collar tags which read, “Stop me
from running away and bring me back to my master, Viventius in the
yard of Callistus” and “Restrain me so that I don’t escape, and take me
back to my master Pascasius in the colonnade in Trajan’s market” and “I
have run away, so keep hold of me. When you return me to my master
Zoninus you will receive a reward” (all quoted and referenced in Massey
and Moreland 1978, 62). Remember Juvenal’s Rutilus in the Satyricon
(XIV. 15), liTo his trembling household he is a complete monster, never
happy until he has called for a torturer and some poor wretch is being
branded with a hot iron – and all because of two missing towels.” Have
we yet found any branding irons bearing the initial HiE, for II this is a
fugitive”? How frequently do we find chains for humans? It took me only
a moment to find a list of twelve sites in France yielding slave chains
(Famechon, Somme; Moulets-et-Villemartin, Gironde; Annecy aux Ilettes;
Conde-sur-Iton, Eure; Liberchies, Belgium; Epiais-Rhus, Val d’Oise;
Vertault, Cote-d’Or; Compiegne, Grand, Vosges; Saintes; Alesia; Strasbourg). And what about that impressive chain from near Cambridge with
collars for six? And what was the purpose of the cryptoporticus often
found under the forum (as at ArIes, Naroonne, Reims, etc.)? Some suggest this was a slave barracks for the town’s public slaves. Can we spot
slaves in Roman sculpture? Many of the household servants depicted
around the representation of the deceased on tombstones must be slaves.
On a fragment of a French stone we see a carving of what is clearly a
slave, for he is being freed by a touch of the vindicta, a rod, used in manumission ceremonies (fig. 1).
Feminists will be most acutely aware that the infrequent reference to
women in antiquity and the even less frequent reference to women by
modem scholars does not imply that they did not exist The same is
probably true of slaves. So, finding little evidence for slaves does not necessarily mean there were no slaves. My impression is that there is
scattered and inferential evidence for slavery, but it is generally ignored,
and that there is not much of what you might call solid evidence for slavery, not much solid evidence, that is, if we overlook one source. Tomb
inscriptions reveal some eye-popping facts. Sure, there are few slaves
commemorated, and those are almost all domestic slaves of high standing. But then there are few enough of the common rural, but free
peasants, such as the coloni. Although historians (including Marxist historians) consistently accept that in the late empire coloni probably vastly
outnumbered slaves, they are even rarer in the epigraphic evidence.
Almost non-existent to be honest For a period of three hundred years SLA VISH NONSENSE OR TIlE TALKING TOOL 129
Figure 1. Sculpted fragment of a manumission scene. The slave is about to be touched by
the vindicta, the rod, which is the same rod that probably beat him on many occasions
before this ceremony (Musee Royal de Mariemont). 130 SAMSON
from Britain to Spain, from Sardinia to Dalmatia, and from Hungary back
to Britain (more or less all of Europe except Italy) there are only fourteen
coloni recorded epigraphically. (There are more epigraphic references to
semi than coloni in Britain.) But there are huge numbers of inscriptions remembering the deceased freedman and freedwoman, the liberti.
If we did some simple mathematical modelling, in the style of a New
Archaeologist, we might say that one in ten slaves were freed by the time
of their death (and here I am exaggerating in order to weaken my proposal
in the spirit of the null hypothesis, for it is likely that a much smaller
percentage were freed). Freed slaves, on the whole, must have been
poorer than ordinary citizens, but let us say that free people were neither
more or less frequently commemorated with tomb stones (again I am
purposefully weakening my own proposal, for I could legitimately suggest
that freed slaves were more likely to lack the wherewithal of the free to
put up stones). A quick calculation based on the actual figures of social
groups commemorated on stones would make the slave population, on
average, somewhere between forty and eighty per cent of the total
population in the western empire. And this is a conservative estimate.
Now, it is no accident that libmi are well and truly over-represented
and that the free did in fact make up much more than a third of the population. Freed slaves, and even more so their children who probably set up
most of the stones, had a vested interest in making public statements that
recalled their manumission, not wishing former masters or their successors to try to reassert their former ownership. (This, I trust, you will find
a more reasonable explanation than MacMullen’s suggestion that somehow the liberti were simply addicted to epigraphy.) But even if liberti
were over-represented, there must have been relatively large numbers of
slaves in some provinces in the west for freed slaves to be so common
among our tombstone inscriptions.
PROOF 3: THEORY
The third approach is the least straightforward, the most theoretical, and
therefore the most appropriate to TRAC. We can investigate or theorise
about the nature of slavery itself or of slave societies as entities and look
for corroborating evidence for slavery. For example, what is the role of
the state in the maintenance of slavery and does it manifest itself materially? The ability of one person to sell another is the most classic element
of slavery. The acquisition of new slaves and the vitality of the slave
trade is regularly assumed by bourgeois historians as all important to the SLA VISH NONSENSE OR TIlE TALKING TOOL 131
maintenance of slavery. Without a source of new slaves, many historians
believe that slavery is doomed to die out Be that as it may, can archaeology suggest widespread slave trading by plotting a one-way flow of
traded goods and hypothesise that the archaeologically unrecoverable
objects traded in the other direction were slaves? I believe it was Andy
Fitzpatrick who showed that three rim sherds of Dressel lA amphorae are
the trade equivalent of one strong adolescent male slave or an older
woman with two children.3
Here, however, my interest is in the nature of slave labour, particularly
in its organisation and control by masters. While the evidence is not unequivocal, we have masses of it – the villa excavations.
Slave labour: theoretical premises
Time itself was not ordered by slaves. Time to eat, time to rise, time to
sleep, even time to fornicate. Slaves’ labour was not their own, nor was
the produce of that labour. Certainly slaves did not own the means of
production, the livestock, equipment, or buildings. Here I have gone
from the least easily demonstrated archaeologically to the most easily.
We surely cannot demonstrate the chronological patterns at the daily
level. At the next level, and despite the pessimism and dismissiveness of
some British archaeologists, I believe that the organisation of labour can
be demonstrated. But more of that shortly.
MacMullen is only one in a long list of bourgeois historians who have
insisted on legal definitions to distinguish coloni from slaves, maintaining
that they cannot be lumped together. Archaeologists naturally cannot
recognise legal status, with the result that, like Harald von Petrikovits
nearly half a century ago, they abandon the effort to distinguish between
“dependent peasants”. Although Richard Hingley is not afraid to mention slavery in his Rural Settlement in Roman Britain, a typical pasage runs
like this (p. 67):
This model [of two unequal “houses” on one villa] specifies that
aisled houses were the homes of labourers, tenants and slaves on
a villa estate. The alternative model [is] that aisled houses were
the homes of extended kin groupings and that the evidence does
not appear to suggest a rigid division between landowner and
Almost every mention of slaves is in a bundle of other forms of dependency, usually, as here, including tenants (is this a translation of coloni?)
and labourers. Thus, Hingley allows for the existence and presence of – ———
slaves in rural Roman Britain, but he refuses to try to recognise them.
This is an improvement on Millett and Greene but ultimately is a capitulation to the bourgeois historian.
In fact, the ancient legal definition of freedom and un freedom or
colonus and serous is largely irrelevant Or it is to us, at any rate. Medieval
history demonstrates that thirteenth-century peasants hated the slur of
being a villein. Even if economically better off than some neighbouring
free tenants, villeins might well pay good money to escape their status.
But medieval history also demonstrates that legal definitions of an absence of “freedom” hardly made a villein a “slave” in any sense of the
word that a good social historian would use today. For us, the real importance is the degree and nature of control a Roman lord or master had
over his or her “labourer”. A colonus was no “tenant” if he lived in special
quarters on the villa of his dominus, was not free to leave, and did not organise his own labour but did what he was told. Such a man was very
nearly a slave, and extreme forms of servile dependency may well have
left obvious spatial and material traces.
When archaeologists investigate villas for possible slave quarters, they
start with Columella and his description of the ergastula. Well, this prison
block clearly was not common. And one reason why Settefinestre features so commonly in archaeologists’ references to Roman slavery is that
it has something that passes for the ergastuIa (fig. 2). But very few pay attention to the passage of Columella’s in which he assumes that slaves will
be quartered according to their jobs. Cattle herds might be expected to
sleep in the byres, others in the barns. In a sense, this is how KolnMungersdorf became so well known. A few buildings were postulated as
perhaps holding slaves in addition to their obvious agricultural functions.
When I first started looking at villa plans in search of possible accommodation of slaves I was particularly excited by the French doublecourtyard villas. At Warfusee-Abancourt you can see that a pale imitation of the great house is sometimes to be found. It is interpreted as the
residence of the actor, the overseer, who was more than likely a slave.
Roman writers, such as Varro and Columella, have much to say about the
actor, his qualities and age and personality and his duties. They even say
something of his mate or “wife” in inverted commas – for slaves could
not legally marry – and how good looking she should be if the master
picked her out for him, and how she was to behave and who she was to
supervise. Reading between the lines there appears to be a division of SLAVISH NONSENSE OR THE TALKING TOOL
del mOnitor .
o 5 10″,
Figure 2. Plan of the slaves’ quarters at Settefinestre (after Carandini 1988). Labelled
rooms include cells (celie), infirmary (infermeria), kitchen (cucina), store room (magazzinoJ, “porter’s cell” (cella ostiariaJ, and the overseer’s lodgings (alloggi del vilicus).
labour between genders, which also had a spatial element and something
of an agricultural-domestic division as well.
Well, my initial excitement stemmed from the belief that the aciores
were more likely to be organising and supervising slave than free labour,
but for the moment let us look at the actor’s home. It was probably built
by and owned by the master(s) in the great domus. Indeed, all the buildings in the outer courtyard were surely owned by them. It seems to me
that the outer buildings are surprisingly well built to house animals and
“farm hands”. The French villas found by Agache through aerial photography, for example, are only recognised because their foundations, at
least, were in stone. If we are to accept the presence of labourers on the
villa we probably should begin by accepting that they did not live in their 134 SAMSON
own homes, that the buildings they inhabited were more substantially
built than those of their social equals who might live in their own homes
far from any villa, that they lived within a compound in which an
overseer possibly dwelt, whose residence mirrored that of the landlord
and lady, and whose status might well be unfree. So, what sort of
dependency held these “hands” to the villa owners?
Can the spatial organisation of the villa throw light on the nature of the
control of that labour? I have argued (1992a) that the enclosure wall had
more than the role usually imputed to it of preventing access or demarcating where access was no longer legally or socially permissible. The
enclosure wall was more than some cosmological symbol, the beloved
explanation of the ideology-crazed post-processualists. More than keep
out animals, robbers, and taxmen, it kept people, particularly slaves, in.
The simple expedient of an enclosure wall meant not that slaves were
trapped in, as if in prison, but should they be found outside when they
were not supposed to be, after curfew perhaps, they could be instantly
recognised or charged with trying to escape. They could not plead ignorance. According to Columella (1.8.12), no slave should leave the fann
without the overseer’s permission. Cato, in his Agricultura (5), may just
give us evidence of an evening curfew, when he wrote of the overseer’s
He must be the first to get out of bed, and the last to go to bed.
Before that he must see that the farm is shut up, that everyone is
asleep in their proper place, and that the animals are provided
Varro, in his Agricultura (2.10.1 and 2.10.6), makes the distinction between herdsmen that go back to the farm every night and those who stay
away – who are of special concern. Conversely, Columella (12.3.7)
stresses that the female actor is to make a close inspection that none of the
field slaves is still inside the farmyard when the team has gone out to
work, for it is symptomatic of the malingerer.
The existence of an enclosure wall is almost a necessary pre-requisite to
the day-to-day maintenance of slavery if flight is at all a common fonn of
slave resistance, and there are whole books devoted to the subject of
slave flight in the Roman period (e.g. Bellen 1971). It is no accident that
medieval servile labour – serfdom – did not involve such enclosures.
Now, almost every villa in the north-west of the Roman empire had an
enclosure. It may have had nothing to do with social control of labour
and everything to do with penning in rare, colourful and exotic animals, SLAVISH NONSENSE OR 1HE TALKING TOOL 135
such as peacocks and peahens, or it may have been thought essential to
signify the axis mundi lest the sky fallon top of them. But the fact that the
accommodation for the workers surely belonged to the villa and was located so close to the master or overseer makes it hard to see the labourers
as anything other than closely controlled servants, presumably slaves. It
is a pity that British archaeologists have made no effort to descnbe how
they envisage the relationship between villa owners and so-called farm
workers. Could the workers have been indentured servants? Perhaps.
Itinerant waged farm labourers taken on at an annual hiring fair seem
unrealistically anachronistic. A tenant, in the sense of a cash- or tributepaying renter of land, is surely out of the question, for these workers
most assuredly worked the villa’s home fields rather than their “own”.
Indeed, this is implied by British archaeologists when they use terms like
field or farm hands. And yet, “tenant” is one of the most common terms
used by Roman arcaheologists (to translate colonus one must assume) to
describe the labourer housed in a villa outbuilding.
It is not true that archaeologists in Britain have totally ignored social
relationships among the inhabitants of villas. Twenty years ago, J. T.
Smith first drew our attention to villas with what appears to be two
houses and suggested joint ownership, presumably of close kin (siblings
one might suppose). This opened the way to seeing villas as perhaps
accommodating extended kin. The argument is interesting, not least
because it holds out the possibility of seeing Romano-British society as
essentially prehistoric or early medieval. Villas would have been more
egalitarian than I envisage them, but the vision is not one I can share, for
it seems to suffer from two deficiencies.
One problem with the argument is that the supposed duplication of
domestic “units” is only rarely encountered on villas. Many of the supposed examples are just not very believable. For instance, Hingley (1989,
figures 33 and 34) offers the plans of villas with “two relatively equal
houses” and further draws attention to villas with separate aisled halls,
which are suggested to hold either estate workers (following Richmond)
or a separate branch of the family (following Smith). The problem is that
the “second” house is seldom obvious and that which is presumably
being posited as a second house (in some cases I cannot tell) is usually
quite different in plan and even in size. This is most extreme in the case
of aisled halls, which regularly shelter one room into which alone could
be fitted four or five of the biggest rooms of the main luxurious villa
domus. At Winterton and Sparsholt, for instance, the domus and aisled hall
are so different in plan that one really has to imagine the buildings func-136 SAMSON
tioning quite differently. Even to suggest they might be two variations on
the same theme is to admit that we really have not a clue what was happening in the various rooms of Roman villas. But the idea that there was
little differentiation between the lesser kin who all bunked down in the
hall of the aisled building of a villa such as Winterton does seem like
nonsense. While we may not be able to label each room and space at
Winterton villa, the layout in plan simply reeks of specialisation. The
space was highly ordered and one might presume also hierarchically ordered. Moreover, rooms in Roman houses have quite a special nature
compared to prehistoric and to medieval rooms which seems to be little
appreciated by archaeologists, too happy with their plan drawings.
Roman rooms are quite small. Many rooms are under twelve feet wide
and would be at home on a ” Brookside” -style estate. If you place the
plans of an Iron-Age round house, a Roman villa, a Dark-Age hall, and a
Jacobean manor all side by side, what is most remarkable is how the
space of a Roman house was so much more divided up into small rooms.
To see a Roman master as some Brian Blessed-like character throwing
bones over his shoulder while a packed crowd of dependants, the familia,
slobbered, belched, and snored in a large communal hall is to borrow
a thoroughly fanciful medieval image and make it even more inappropriate.
The second problem with the egalitarian extended family interpretation is that a large villa may well have held a hundred people or more.
More, at any rate, than an extended family. At Winterton some fifty
rooms have been excavated. In his introduction, Hingley (1989, figure 3)
introduces us to the extended family. The model of a three-generation
family, borrowed from Thomas Charles-Edwards, has only twenty-five
members, which include the two great grandparents who were probably
no longer living when the youngest grandchild was born and three
women who presumably would have married and gone to live with their
husbands, assuming a patrilocal society. In other words, the villa would
have been populated with just six kids, two teens, eight young adults,
and four middle-aged adults. That makes fifteen able bodies, but surely
they did not all plough and hoe and harvest and muck out and weave.
Villa owners were at the apex of society and presumably did not often
break into a sweat At least some of these fifteen must have swanned
about, and, when not bathing, been fed grapes while reclining. Perhaps a
villa estate could well have been worked by ten strong women and men;
I am no farmer and have no practical knowledge of these matters.
If a Winterton-sized villa needed far more workers (ancients and SLAVISH NONSENSE OR TIlE TALKll’IG TOOL 137
archaeologists alike asume a rough figure of one hundred, cf. Samson
1989), to stock it with yet more extended family would have required all
the descendants from an ancestral pair six or even seven generations back
in time. Now, this is not an extended family as anthropologists know it,
this is a true lineage. This postulation turns villas estates into some sort of
tribal kibbutz, containing cousins five times removed, which while quite
a sweet picture of Romano-British society is also quite daft
The most unsatisfactory aspect of villa archaeology must be our almost
total inability to ascribe functions to the dozens of rooms we find on any
plan (this observation has also been made by Millett 1990, 197). Where is
the master bedroom, the study, the childrens’ play room, the library, the
billiard room, or the guest’s room? Where did the nanny sleep? Where
was the maid’s room and where did she keep the linen and mops? Was
there a granny flat? We are ignorant of the layout of most of the big
house (where past excavations have concentrated), so it is not surprising
that little can be said about the arrangements for agricultural staff. But
one thing seems almost unanimous among scholars and that is the assumption that many labourers did shelter somewhere in the buildings in
and around the villa enclosure wall.
While I cannot mystically divine slavery from a villa enclosure wall
and a well-built stone agricultural building against it that mayor may not
have afforded accommodation to workers, at least I have a plausible
theory that says that both of these bits of architecture point to intensive
control of the labourers by the villa owners, which implies slavery rather
than tenantry. Almost all Roman archaeologists make no explicit theoretical statement about slavery or dependent labour. Most do not even
mention slaves but prefer terms that better fit modern capitalism, deluding themselves that they are somehow “neutral” terms. I may be wide of
the mark supposing that this is the result of an explicit loathing or an unconscious shared western political and cultural fear of Marxism, a backlash against culture history and classicism by New Archaeologists and
post-processualists alike, or just the stupid parroting of what is found in
the secondary sources. But one thing is certain, most archaeologists’ thin
descriptions of dependent labour on Romano-British villas, devoid of reference to slavery but full of references to “farm hands” and “tenants”,
are, frankly, pure nonsense. 138 SAMSON
1. Here is the seldom-seen bedrock of most Romanists’ assumptions on slavery
in the western empire: the slaves produced from early imperial expansion
died out naturally (somehow they did not reproduce in large enough
numbers), the opportunities to recruit new slaves dried up in time, slave
owners finally realised that it was in their economic interest to let their
slaves go, because as slaves they were lazy and good for nothing, and that,
in any case, there was no slavery anyway. Thus, we find Peter Salway, who
rarely mentions slavery in Roman Britain, arguing on the page headed
“slaves, labourers and tenants” (p. 605) (1) that “under the Empire, as
conquests became less frequent … the availability of new slaves must have
become less”, (2) that “there is no direct evidence in Roman Britain for the
great slave-run estates or latifundia of Republican Italy”, and (3) that “by
imperial times Roman writers were doubting their economic efficiency and
preferring tenant farmers” .
This stereotypical explanation for the end of slavery is also employed for
the Middle Ages (see Samson 1994). The “explanation” sees slavery as
somehow uneconomical and therefore even almost unnatural, which beggars belief and can only be ascnbed to a mad free-market capitalist mentality
that says that people only work hard if working for themselves. It makes me
wonder how, if Roman masters and medieval lords were able to work out
that they would receive increased profits from less direct exploitation but
greater productivity after freeing their slaves, plantation owners of the
American South were unable to make the same calculation. Moreover, I
know of no abolitionist literature that ever put forward this argument that
slavery was “uneconomical”. It seems just possible that abolitionists refrained from making such claims for fear that plantation owners would,
rightly, have seen them as economic Simpletons.
2. I apologise for referencing no fewer than five different pieces of my work,
the medieval ones of debatable relevance. I hate this self-indulgence in
others, but a bout of paranoia has convinced me that no one has ever read
anything I have written. This is no more than a desperate bid to remind
others of these often-overlooked gems.
3. Dear reader, this text contains many instances of the first person pronoun
because I actually did write these words and think these thoughts. I believe
there is nothing wrong with admitting that the thoughts are my own and not
some objective analyses that sprang from a disembodied undefined authorscholar. (Indeed, the thoughts are more completely my own than might be
the case of other scholars because my friends do not seem to have the time or
inclination to read my work or pass comments any more.) As it stands, Dr
Reviewer, the paper does not read like an unrevised lecture. It was almost
completely rewritten for publication. I simply prefer clear and unpretentious
prose, especially if it occasionally amuses. The reference to Fitzpatrick, as far SLA VISH NONSENSE OR 1lIE T AU<ING TOOL 139
as my memory now fails me, is totally spurious. I apologise to Andrew for
any distress this may cause him.
Bellen, Heinz 1971. Studien zur Sklavenflucht im romischen Kaisemich.
Bradley. Keith R. 1984. Slaves and Masters in the Roman Empire: a study in
social control. Tournai: Latomus.
Branigan, Keith and David Miles (eds) 1988. Villa Economies: Aspects of
Romano-British Villas. Sheffield: John Collis.
Burke, John 1978. Life in the Villa in Roman Britain. London: Batsford.
Carandini, Andrea 1988. Schiavi in Italia. Gli strumenti pensanti dei Romani fra
tarda Repubblica e medio Impero. Rome: La Nuova ltalia Scientifica.
Collingwood, R. G. and Ian Richmond 1969. The Archaeology of Roman
Britain. London: Methuen.
Demarez, I.-D. 1987. Les batiments a fonction economique dans les fundi
de la provincia Beglica. Amphora 50, 1-36.
Greene, Kevin 1986. The Archaeology of the Roman Economy. London:
Hingley, Richard 1989. Rural Settlement in Roman Bn·tain. London: Seaby.
Hopkins, Keith 1978. Conquerors and Slaves: sociological studies in Roman
History, vol. 1. Cambridge: Cambridge University Press.
MacMullen, Ramsay 1987. Late Roman slavery. Historia 36,359-382.
Massey, Michael and Paul Moreland 1978. Slavery in Ancient Rome. London:
Millett, Martin 1990. The Romanization oj Britain. An Essay in Archaeological
Interpretation. Cambridge: Cambridge University Press.
Percival, John 1976. The Roman Villa. London: Batsford.
Rivet, A. 1. F. (ed.) 1969. The Roman Villa in Britain. London: Routledge.
Salway, Peter 1981. Roman Britain. Oxford: Clarendon Press.
Samson, Ross 1989. Rural slavery, inscriptions, archaeology and Marx: a response to Ramsay MacMullen’s “late Roman slavery”. Historia 38, 99-
Samson, Ross 1990. Comment on Eleanor Scott’s “Romano-British villas
and the social construction of space”. In Ross Samson (ed.), The Social
Archaeology of Houses, 173-180. Edinburgh: Edinburgh University
Samson, Ross 1992a. Knowledge, constraint and power in inaction: the
defenseless medieval wall. Historical Archaeology 26, 26–44. 140 SAMSON
Samson, Ross 1992b. Slavery, the Roman legacy. In Hugh Elton and John
Drinkwater (eds), Fifth-Century Gaul: a Crisis of Identity, 218-227.
Cambridge: Cambridge University Press.
Samson, Ross 1994. The end of early medieval slavery. In Allen Frantzen
and Douglas Moffat (eds), The Work of Work. Servitude, Slavery and Labor
in Medieval England, 95-124. Glasgow: Cruithne Press.
Smith, J. T. 1978. Villas as a key to social structure. In Malcolm Todd (ed.),
Studies in the Romano-British Villa, 149-156. Leicester: Leicester
Smith, J. T. 1982. Villa plans and social structure in Britain and Gaul.
Caesarodunum 17, 321-351.