Edited J~,
Jennifer R. Da\ris and Michael cCormicl<
Ashgate: Aldershot 2008
Chapter 2
Strong Rulers - \Veak Economy? Rome,
the Carolingians and the Archaeology of
Slavery in the First Millennium AD
Joachim Henning
Rome did not die in a battle. Rome's death lasted centuries. lt survived for a
thousand years in Byzantium, and Charlemagne attempted a rebirth in the eighth
centurv. Curious though it may seem, important elements for understanding
Rome's decline emerge from an analysis of \Vhat happened. or more precisely.
of what went \vrong with Europe's economy \Vhen powerful and. mibtarily,
successful Frankish rulers tried to reinvent the Roman villa estate.' Conversely,
how the late antique agrarian economv fostered Rome's downfall, at least in the
\X'est, illuminates the Carolingian situation. This chapter will use archaeological
data to explore comparatively the problem of productiYity in the late Roman
and Carolingian countrysides, focusing on the interplay of coercion, the logic
of farming organization, and technology.
Archaeology addresses this c1uestion by analyzing material remains of
late Roman and early medieval times. Founded on natural sciences such as
botany. soil micromorphology, zoology, chemistry and climatology, the results
of setrlement archaeology are particularly important. Nor can archaeologists
ignore recent research based on written sources. \\'e should not misconstrue a
dense written record as the objectively illuminated realjt)· of that time, and we
underestimate the tight connection of medieval writing and power structures
at our peril.2
h1r the htc antit)UC roots of the Carolingian estate, see Peter Sarris, "The Origins of
the ll.hnorial Economy: :-.Jew lnsights from Late Amic(uity", English llislorical Rnitw, 61
) (2004):
279-311.
I am most grateful to l\.like :'llcCormick and the undergraduate and graduate students
and colleagues oi Harnrd's Histor)· Department for a semester of remark:lble discussions of
the carlr mcdienl written record, its relations to archaeological findings. and the advances of
scientific archaeology. This unforgettable cxp,:rience peaked in tbe New Directions conference
and two interdisciplinary workshops I h:~d the pleasure of organizing with illike at Harvard
in aununn 2005. I am especially grateful to .Jcnnifer D:wis and i\like for their paticnc.: with an
arclucologist·s foibles.
34 ]be Lo11c~ Aloming of :\ledieralEt1rope
\\'riting was in the hands of the mighty.3
It is not therefore impossible that
the eighth century's sharp increase in documents such as polyptychs, donations
or capitularies reflects the growing economic power of the Carolingian high
nobility rather than a more producti\·e organization of the whole society.
Overlooking this simple insight, earlier work has sometimes tended to e<.Juate
abundant charters and economic efficiency. In light of the scarce written
evidence between 500 and 700 AD, we should be cautious about assuming an
economic "awakening of the eighth century", so long as we do not reallv knmv
how economic (especially rural) organization worked in the first post-Roman
centuries, that is before the Carolingian manor spread oYer Europe.4
In the light
of archaeology, it could turn out that the establishment of the Carolingian villa,
the bipartite estate or manor, \vas merely an awakening of pO\ver structures
whose impact on the rural economy was more quantitative than qualitatiYe.
\\'hen eYaluating the success or failure of modern organizations, it seems
normal to begin with their economic efficiency and their ability to react Aexibly
to changing challenges. How can we consider the end of the "Roman empire as
a business concern" or its "hostile takeover" by new social forces without such
an economic perspective? Territorial co:1qucst as it occurred in the late Roman
West does not necessarily transform the economy. Yet the very principles of
agriculture. the basic form of economic production of the pre-modern world,
changed fundamentally in this period. Judging from the growing archaeological
record, rural 1-'illae and smi, villas and slaves, surviYed in the Frankish heartland
mostly as terms in the written records. In contrast, rural settlement patterns
and social structures as they appear on the ground. for instance in cemeteries
as well as in agricultural practices, indicate a broad new logic of organization.5
for more details, see \'\:alter Pohl and Paul Herold (eds), Vom Nutzen du Schreibm.r,
Denkschriften der Osterreichischen Akademie der Wissenschaften, Philosophisch-historische
Klasse, 306 (Vienna. 2002) and Rosamond J\lcl(ittcrick (cd.), The l'srs o/ Uterm]' in Earl)· Medieml
Europe (Cambridge. 1995).
• Jean-Pierre Devroey, ·The Econonw". in Rosamond McKitterick (ed.). Tbr Earb• Middle
Ages. Europe 400-1000 (Oxford, 2001), pp. 97-129, here p. 104.
Chris Wickham's discussion of the "logic of the economic system" is most stimulating:
r~md tmd Po11·rr (London, 1994), pp. 77-913. For him, in the second century the sla,·c mode of
production declined, which, at its core, had Cflllsisted of the agricultural use of chattel slaves.
The new S\'Stem that unfolded in the second and third centuries would ha\·e consisted of: "the
combination hetween a greater autonomy for what can now be called peasants, tl1e dominance
of an economic system based essentially on subsistence agriculmre, and the end of effecti\·e
imen·ention by landlords in the procedure of production ..." (p. 85). Wickham is skeptical that
hte or prm•incial ,·illas can he associated with sbverv (for example. p.137: ..\Vho is to sa\· that every
villa was a slave ,·ilia?"), but the archaeological t1nds presented in this paper suggest coerced labor
in the late Roman countn•side. From an archaeological perspecti,·e. I would hesitate to locate the
crucialmrning point in the second and third centuries inside the Roman Empire. I ha,·e argued
that the new economic logic did indeed begin rhen, bur outside rhe empire, and that it graduallr
expanded into former Roman territories in the fourth and t1fth cenmries: Joachim Henning,
..Germanisch-romanische Agr:trkontinuitat und -diskontinuitiit im nordalpinen i(ontinentaleuropa
- Teile eines Systemw:mdels' Beohachtungcn aus archaologischer Sicht". in Dieter I Eigermann.
\Volfgang Haubrichs and .Ji.irg Jarnut. with Cla11di:; Giefers (eds). AkkulturatiorJ- J>ro/Jieme ri11rr
gmllaniscb-roma11iscbm Kldtursp11hese i11 Spittalllike mu/ jr1ibem Jllittelalter, Erg:lnzungsbiinde zum
Romr, tbr Carolii(V,ians and JI,IIICI)' in the First Mi/lmnium AD 35
\X'hat arc the reasons for this reorganization that had begun some hundred
years before Clovis conquered Gaul?
The nature of the reorganization itself may supply the answer. The shift
from the classical Roman villa to the use of tenant labor in the late empire
was an important step toward a new organization of the rural world. But \vas
this change a complete turnaround, which soh·ed tbe basic problem of earlier
villa organization, namely how to increase the productivity of rural tabor in
an essentially exploitatiYe system? The use of slaYe or dependent labor could
be highly oppressive and required a combination of incentives and coercion.
Roman agricultural writers had long discussed this problem with surprising
frankness. Columella, for instance, advised '"'1nning m·er selected s)a,·es to the
owner's side by intensive conversation and com·incing them that their good
work was indispensable for the whole villa community. Unfortunately, this
approach probably worked mainly on slaves of rather limited intelligence. The
other slaves - the critical ones for improving producti\•it\·, for example in the
highly sophisticated vineyards- wound up doing their fieldwork in chains6
Archaeological finds of Roman-era slave chains (see figure 2.1) throw
doubt on the idea that a shift to the self-managing eo/onus on late antilJUe estates
actually soh-ed the problem of servile productivity.7
Although Rostovtzeff
presented only those well-known first-century iron shackles from the villas
around Pompeii, iron shackles and fetters are not confined to the first two
Rc:dlcxikon der Germanischen Altertumskunde, 41 (Berlin, 2004), pp. 396-435. On the other
hand, \'\lickham's description of the "new logic" seems an exccllt:nt starting point for defining the
post-Roman economic system.
• Columella. De re mslic<l !.3 and 7-8. Sec also Andrea Carandini. "Columella's Vine\·ard
and the lbtionalit:y of the Roman Economy", Op11s, 2 (1983): 177-203.
· Since 1990, I ha,·e identified m·er 400 iron sla\·e or prisoner shackles from all o\'er
Europe, from the pre-Roman Iron Age through around 1500 AD. I am especially grateful to r.
Hugh Thompson (d.l995), who had independently begun collecting iron shackles from Roman
and pre-Roman times. for a most generous exchange of archaeological data between 1991 and
1993. Due to technical problems and errors that arose in the posthumous publication of his
'11!t Archarolog)' of Greek and Roman Slal'ery (London, 2003) and in his late article "Iron Age and
Roman Sla\·c-Shackles". Archaro/~(al]o11mal. ISO (1993): 57-168, his typology and datings rcCjuire
re\'ision in small but sometimes decisi\·e details. I\I\' own study. which focused on post-Roman
shackles (''Gefangenenfesseln im slawischen Siedlungsraum und der europaische Skla,·enhandel
im 6. bis 12. Jahrhundert", GmJJ<mia, 70 (1992): 403-26). probably arri,•ed too late for him to
iully exploit. Earlier (\''i'illiam H. Manning, Catalogue of tht RoJJJano-Britis!J Iron Tools, 1~1/il(~·'· <llld
lJI'tapo11s i111he British Museum (London, 1985), pp. 61-4) as well as more recent efforts to classify
Roman sh:tckles (l~rnst Kiinzl, "Schlosser und fesseln", in Ernst Kiinzl, Dit Alammmmbmlt <1111
drm Rhtill bei Nmpo!~ l\!onographien des Riimisch-Germanischen Zentmlmuseums, 34 (4 \'ols,
l\lainl, 1993), \'ol. I, pp. 365-78), either rely on too small a selection (Manning studies onhRoman
material from the UJ<..') or seem too general (Kiinzl only distinguishes between two types,
and misidcntifies some objects. for ex:unple an animal chain from the iron hoard of Bnwic in
Serbia; the objects from l\lannersdnrf in Austria are plow, animal :md kitchen chains). l'\'ly new,
more detailed typology incorporates as many aspects as possible of Thompson's classification
and combines these with Ill\' organization of the post-Roman material colleC!ed in central and
eastern Europe (1-lenning. "Celangenentesseln''). The full publication \\~th classification and
dctaik·d dating of the abuncbnt material is in preparation.
36 'Jl,e Lm(g Momi11g of Medin/rll Europr
centuries AD and their ''plantation slavery".~ \'('hen, in the later third century,
Germanic irwaders surprised and devastated the Roman farmhouses of the
western provinces, the farms' cellars \Vere filled not only with iron tools, but
also with slave chains (sec Figure 2.2).9
Afterwards some buildings of such
former villas experienced new and different uses. Rural populations occupied
former villa sites, although some impressive building compounds, sometimes
even palaces (which typically lacked production installations) still existed in
the countryside.10
The period of the second Germanic wave of devastation of
the mid-fourth to the early fifth century produced an archaeological picture of
shackle finds that differs surprisingly little from earlier timesn Coins assign the
best-dated examples of such chains to the second half of the fourth century in
a late Roman well on a Rhenish villa site, and in a Roman iron hoard uncovered
in the Palatinate near a villa that might still have been functioning in the fourth
century.12
The archaeological data from some 18 villa sites that mostly date from
the earlier third through the fourth centuries is revealing. Even more shackles
and related objects come from late antique rici (non-agrarian hamlets. small
l'l!ichael Rostcwtzeff, .fodal a11d Ecoi/Olii/C History· of the Romrm Empire (Oxford, 1926),
pi 9.2.
9
Pedro l'llanucl Berges Soriano. "Las ruinas de ·EJs Munts"', hiformadoll Arq11eo/~gica.
3 (1979): 81-7. here p. 84 (Aitafulla. Spain: shackles with human bones from a Roman ,·ilJa
basement, destroyed in the second hali oi the third century); Rudolf L:mr-Bclart ami Victorine
\'On Gonzcnbach, "R6mische Zeit",jabrbuch derJchll'eiifrischm Gesellscbriftfiir Urgeschicbte. 37 (1946):
66-86, here pp. 76f., pi. 10, fig. 2 Qona. s,.~tzerland: shackles from a cellar or secondary building
of a Roman ,·ilia, late third century); Willem Caes, "De landbouw in Noord-Gallie: Een onderzoek
naar de Gallo-Romeinse landbouwwerktuigen uit Belgie" (unpublished licentiate thesis, Lem·en
Uni,·crsitv, 1984), catalog Qette. Belgium: shackles from a Roman ,·ilia cellar. with agricultural
implements, destroyed in the third century). from a late Roman basement: l\'iark Reginald Hull.
Romr111 Colchester, Society of Antiquaries Report, 20 (London, 1958), pp. 106-11:!, pi. 21. f-rom an
oil mill's cellar belonging to a Roman villa. found with human bones in l'llilrcu, distrito de Castel
Branco. Portugal: I thank the exca,·ator, felix Teichner, for sharing this unpublished find from the
second half of the third cemury.
1
" The palace-like building of Bad Kreuznach (built around 250 :\D) w'ts transformed into
a lmrgus, a iortified tower, in the iourth century: Ronald Knochlein. "Die nachantike Nutzung
der Bad Kreuznacher PalastYilla", Mr~iiiZI'r ArcbJoloJ!.iscbe Zeitscbrift, 2 (1995): 197-209: cf. too the
large villa of Echtcrnach: Jc:annot l'lletzler, Johny Zimmer, and Lothar Bakker, "Die romische
Villa \'On Echternach (Luxemburg) unci die Ant1inge der mittelalterlichen Grundhc:rrschaft", in
\\~llter Janssen and Dietrich Lohrmann (eels). Villa-Curti.r-Grm(girl (Munich, 1983), pp. 30~5. here
p. 38, fig. 5. i\t Vieux-Rouen (Somme, Prance)- <http://crdp.ac-amiens.fr/crdp/picarcheo/
PDf/fichesagache_C.pdf>- archaeological im·estigation showed that the large second-century
Roman \'illa was transformed into a palace \\~thour signs of farming installations in the fourth
centun·. Further late Roman rural palaces: Bad Durkheim (Rheinland-Pfalz: with a fourth-century
b11rgus). Saint-Just-en-Chaussee (Oise. France), Bbnzr-les-fismes (Aisne, France), l'llogorjdo
(Bosnia and 1-lcrzego\'ina), Kostin brod (Bulgaria) and seYeral cases around Trier (Konz,
Leudersdorf. Nennig, \\'ittlich, and so on).
11
See especially the well-dated shackles from Epiais Rhus (France:. tourth-centur\' Roman
\"ilia context), Cologne (a wc.:ll \\;th late Rom'tn material), Mainz (fourth-cemury building with a
set of iron tools). Great Chcsterfort (UK, late Roman iron hoard).
12
Respecti\·elv, at Bengel: Woltgang Binsteid, "l'lletallgerat aus einem ri\mischen Brunncn",
in Riif11er·IIIIIsflierte. 2 (Cologne, I975), pp. IH3f., and Helmur Bernlurd, "Der spiitri)mischc
Depot fund von Lingenfek\", Mitteilm{f,rtl des Hist01iscbm Vereins der P(r~liJ 79 (19Hl): 5-103.
Rome, tbe Carolingian.r cmd Sla!!f1J' in the Fi1:rt Millmflitllll AD 37
towns), fortifications, urban centers. and e\·en burials (Figure 2.2).13
These late
Roman iron manacles, shackles, and fetters illustrate that the Roman agricultural
system continued to rely on physically coerced labor. \\'hether tenants or slaves
dominated on late Roman villas, it is hard to deny that the countryside was
organized by the elite and exploited for the aristocracy's O\Vn benefit.14
The
Code......- Theodosimms, which indicates that iron shackles could also be used for
co/oni who fled from an estate, shows that independently of their legal status,
agricultural workers were subjected to coercive force in order to make them
work."
To my knowledge, no late antique rural settlement structure in the territories
under continuous Roman control compares to the distinctive rural structures
documented in large numbers outside the imperial frontier and from post-Roman
Gaul or, very exceptionally, in areas temporarily occupied by Germanic groups
in the mid-fourth century.16
These settlements \vere surely true yilJages - that
u Virton "Chateau Renard" (Belgium, shackles from a fourth-century fortification):
documentation of the Sen·ice n:uional des fouilles, Brussels. I am grateful to A. Cahen-Delhare
and C. f-Iassart (Brussels) for these unpublished data. Tekic C'Trdtanoncka gradina", Pozeskosla,·onska
zupanija, Croatia: a male skeleton with a riveted iron neck-ring without gra,·e goods in
a late fourth-century cemetery of which 140 gra\-cs, some with gra,·e goods, ha,·e been excavated:
Dubravka Sokac-Stimac, ''Rimska nekropola na Trciitanovackoj gradini", Pozdki zbomik (Sicwonska
Pozega). 4 (1974): 115-40. here p. 129. with fig. I would like to thank the exc;n·ator, DubraYka
Sokac-Srimac, for her very detailed and partially unpublished information, as well as Dragan
Bozic. Ljubljana, for his linguistic help.
H On the question of late Roman coloni, sec Elio Lo Ctscio (ed.), Terre, proprietari e CO!Itadini
dell'i111pero rommro: dall'cifjillo apwrio al colrmalo lardMntico (Rome, 1997). I would like to thank J. KYle
Harper for his ad\'ice about late antique slavery.
" Code.v Theodo.rimuu 5.17.1 (332 AD): see also Codex justinimiiiJ 11.53.1 (371 AD). 11.51.1
(386 AD), 11.52.1 (392 1\D). For the progressive conflation of shwes and co/01ri after the third
century and the debate about any difference between them, see Klaus-Peter Johnc. ''Von der
Koionenwirtschaft zum Kolonat", in Klaus-Peter Johne (ed), Gmllschqft rmd 117irtscbaft dn
RomiJchm &ichrs im 3. fabrhrmdert (Berlin, 1993), pp. 64-99.
16
An early exception dated to the mid-fourth centunc comes from east of the Seine
estuary, near Rouen. The publication of the first half of the excanted settlement area showed
the reconstruction of an organized settlement consisting of multiple farmsteads separated b\·
fences. thus a "true village": Paul Van Ossel, "Die Gallo-Romanen ais Nachfahren dcr romischen
Prm·inzialben)lkenmg", in Alfried Wieczorck t:t al. (eds), Die Fr{//tkm.· IFegberriter Buropa.r (2
mls, Mainz, 1996), pp. 102-9, here p. 108. fig. 80. The complete st:ttlement layout published
fi,·e ye;trs later b,- Valeric Gonzalez. Pierre Ouzoulias and Paul Van Ossel. "Saint-Ouen-duBrcuil".
Cermmria. 79 (2001): 43-61. here p. 46. fig. 3. no longer shows the fences. They are
now assigned to the Galla-Roman period and figure on a different map (ihid., p. 45, fig. 2). The
intense historical debate in France about the rioe of "true \'illages·· has culminated in the model
of a gradual "mutation" of Rom;tn ,-illas into ''true \'illages", a process which is thought to han:
been completed in the seventh and eighth century. In his forthcoming "De la ·,·ilia' au ,·iJhge:
l..es premices d'une mutation", in Jean-l\larie Yante and Anne-;..brie Bultot-Verie,·sen (cds).
Autourdu "1illage": Efllblissemmts h1111Mins. jifl<!'l,l'J d colllllllllldlltis rmwlu mlrt Jnfle et Rl>in (Lou\·ain-laNem·e,
forthcoming), Van Osscl makes the Gist: for the late de,·elopment oi "true \'illages" in the
se\·emh centurv, without discussing Saint-Oucn-du-Breuil. l\ly ctrlier skepticism about the site's
dating (llenning. "Agrarkontinuitar··. p. 421) was ill-founded, for we now know that a coin ho:trd
associated with tht: settlement must ha,·e been assembled between 345 and 350 AD md bidden
some time thcrt:after. Since hoards are tvpically markers of destructiYt: t:\Tnts. this find seems more
38 Till' I /!IIJ!, Momi1(~ of ,I Irdiel'ni curope
V ' · ""-··......
I
'
/• .. ,.-· .. •
'}. .
Figure 2.1 Find map o f Roman iron shackles.
J--;P._
"ry"' ;
A
'... \
"'-,
~· ' V
..
Figure 2.2 Roman iron shackle contexts.
Roman iron shackles
0 0 · 150AD
0 150 • 300 AD
Iron shackle contexts
Roman civil context (0·500 AD):
• villa rustica T cellar X well
Lock shackles:
c from burial ~with skeleton
Reveted fetters:
RMIIt, t/1t CaroliJ(~imu and Jlal<!i' ,;, tbe Firsi.llil!.m;iJIIII --11) 39
Figure 2.3 Post-Roman iron shackJes.
Cl ?
\
.':~ ;::,..
: e<:'8
'\'-· ~·
~-....~,~..~.l
·-.._
~,.. ..,
i
.•.
'
0
~· ;._
,;;~_
)' ' i.
-..._
0
Figure 2.4 Post-Ro man tron shackle contcxt·s.
,. Post-Roman iron shackles
~-"t , 0 500 • 750 AD
• 750 · 1000 AD
Iron shackle contexts
Post-Roman (500·1000 AD):
0 fortified site I ringfort I refuge
1!1 town /trading place (lortified)
Cl burial
40 Tbt 1111(!{ Jlloming if hlrdirmlE11rope
is, composed of separated farmsteads with different buildings serving different
domestic and dwelling functions. From the third to early fifth centuries, such
"row settlements., can be observed only outside the limes, the Roman frontier. 17
I,ate Roman rural structures, on the other hand, whether they occupied a former
Yilla or belonged to a hillfort settlement, look unimpressively poor and lack any
visible division into farmstead units separated by enclosures or fences. 1
K What
Germanic finds occur on late Roman villa or rural settlement sites. as well as
tombs with Germanic grave-goods inside Roman Gaul or the Rhineland, could
reflect a gradual integration of barbarian populations into the late anti'JUe rural
system. Although they came from areas with deYeloped Yillage structures, these
newcomers seemingly were pre\·ented from bringing their village economy
onto the late Roman estates. \'\'e may deduce that they became either quasi
coloni or landlords.'9
\\1ith the exception of Ouen-du-Breuil, no Germanic
or Roman settlement with that true village layout is known from inside the
limes, whereas they are abundant outside it. And such true villages beyond the
li111es were not inhabited exclusively by Germanic peoples.20
Nevertheless, the
only Roman territories which have provided convincing village layouts are
those from which Roman troops and administration had been \vithdrawn. for
instance the astonishingly early cases from Toxandria in the Lmv Countries, an
early Frankish settlement area,21
or the fl,l?,li deml!lales in southwestern Germany,
from around 300. The archaeological evidence thus suggests that the new logic
likely to he related to an unp<:aceful end of the village after 350 AD than to its foundation, a~ K:trl
I lcinz Lenz assumes in a way which is difficult for me to follow: ..Gcrmanische Siedlun~cn des
3. bis 5. Jahrhundcrts n. Chr. in Gallicn", Flrrichtr der Romiscb.Grm;auischm Kommis.rio11. 86 (2005):
349--445. I am most grateful to Paul Van Osscl (Paris) and Karl 1-!einz Lcnz (frankfurl am i\lain)
for generously making their unpublished studies a\·ailablc and for the stimulating discussions !hat
followed. Both authors' interpretations assume that Rome would ha\·e had to give permission for
settlements oi this type to be built. Another explanation, howcn:r, would be that Roman land was
illegally occupied. This seems to me more probable in the light of written sources and that coin
hoard: contemporaries clearly attest a series of '·illegal" settlement e\·ents by Germanic tribes
that started shortly beiore the middl<: of the fourth century in nearly all parts of inner Gaul,
reaching almost to the Atlantic coast; Zosimus. Historia, 3.3.1; Ammianus i\hrceUinus, Rt.rguftlt,
16.2. I-7; Julian, Epittula adAthmimsu, 277Df, 279Af. As Libanius says (Oratio, 18.34), Germ:ms
had seized farmlands and plowed and han·ested in many parts of Gaul. Starting in 355, Caesar
Julian·s military campaign put a bloody end to this early. illegal attempt to replace the late-Roman
agricultural system. Roman troops destroyed or confiscated the Germanic grain han·est in Gaul
(Ammianus, 16.11.11; Libanius. Oratio, 18.52). The situation at Ouen-du-Breuil seems to me to tit
this historical context perfectly.
,. I lc.:lena I lamerm\·, Ear!J Mediem/ Settluumls: The Arcbarolog)' of Rltral Commtmitiu i11
Nonhu•ul Europr, 400-900 (Oxford, 2002), p. 55.
'" Paul Van Ossel, F.tabli.rtemmfs mmr1."\· de i>·ltJtiquitr fardil'f dai/S lr uord de la Gaulr, ( ;aJiia,
Supplement, 51 (P:tris I992).
•• The case oi i\limne-.\larboue seems to indicate a landlord of Germanic origin: i\lichclc
Blanch:ml-l..<:mce. '"La villa :1 mosai.<Iu<:s de 1\·lienne-Marbouc (Eure-et-l.oire)'", Gallia. 39 (I 9Hl):
63-83.
21
' According to l.ibanius, Omtio I8.3-l. the Ccrm:ms brought inhabitants of Gaul to th<:
right bank of the Rhin<: to work their fields, so it is no \\·onder Ammianus i\larccllinus, 17.1.7,
saw 1hcr<: rural buildings I hat resembled thost: of Gaul.
21
St:c Hmning, "t\grarkontinuit:lt", p. ~25, fig. 7.
Romr, the Cttroli'!~irJIIJ a11d Sldl¥!1)' i11 tbe First lllillnmium A/) 41
of family- and farmstead-based f:mning could not be fully realized within late
antiquity's social, fiscal, and legal structures. It needed lands where the Roman
writ no longer ran.22
The archaeological evidence from Toxandria shows that such territories did
not tumble back into prehistory once they left Roman administration. On the
contrary, the village-system that now took root there displays unusual dynamism
in applying advanced agricultural technology. That dynamism contrasts sharply
with the rural landscapes Rome still controlled south of the new fortified fillieS
running from Cologne to Bavai that now bounded Toxandria.23 There. Belgian
archaeologists have recently disproven the myth that tl1c heavy wheeled plow
had spread across late Roman Gallia Belgica. The iron plow components found
on villas or rural sites devastated by Germanic invaders in the later third or
fourth century arc unimpressive: a//weU-dated plowshares belong to the simple
ard or stick-plow known in these area-; since the pre-Roman Iron Age24
As a
new find from Aquitaine confirms, the situation extended across Gaul.25
With a few exceptions from territories where the imperial administration was
already \Veak or nearly non-existcnr (for instance rural hillforts of the .Moselle
area or Middle Danube sites devastated by the Huns). adYanced agricultural
equipment. including the hea,·y wheeled plow and the true long scythe, occurs
predominantly along the approaches to the late antique /imes.26
The most
impressive assembly of agricultural equipment comes from Osterburken,
beyond the Rhine frontier, and has to be dated to the late fourth to fifth
century.27
This data contradicts the often-stressed technological back\vardness
of small agricultural units of peasant farmsteads, their autarchy and limited
ambition to produce a surplus. In fact, the significant quantity of late Roman
ceramic imports found beyond the liJJiel8
in such ,·illages attests to exchange.
- In Lhe late third and fourth centuries, the Roman army reh'lllarly attacked Germanic
groups which had tried to settle in Gaul \\ith their families, where we would expect them to haw
founded ,;uages. Such subjugated groups were "resettled" to distant regions, often chained and
sold to the local landowners: .\.11 patttgYrid latini 4 (8).9.1 and 6 (7).4.2. Such contacts with ,;ua
owners were perhaps not unconnected with reactions such :ls those of 366 AD when Germanic
barbarians attacked and destrm·ed ,·iJias in Gaul: Ammianus J\larcellinus. 27.2.2.
23 See Ilenning. ·Agrark;mtinuitiit", p. 424. fig. 6.
24
Berangcn: de L:weleyc and Agncs Vokaer, "De Plinc :1 ~lageror: Araire ou plaumoratum?".
in Ph~li~~e i\Iignot and Georges Raepsact (eds). Le sol ff /'amire dmrs I'Antiquitt: (Brussds, 1998),
pp. 2.>-.J.J.
5
i\lichel Fcugere, "Outillage agricole et quincaillcrie antique de Valentine". in Michcl
Feugerc and i\litja Gustin (eds). Iron, lllacksmitbs and Tools (l\lontagnac. 2000). pp. 169-78.
16
See Hcnning, "Agrarkontinuitiit", and Joachim Hcnning, "Zum Problem dcr Entwicklung
materieller Produkti\·kriifte bei den germanischen Staatsbildungen", Klio. 68 (1986): 128-38.
•· Joachim Iknning, "Zur Datit:rung von \\'erkzeug- und t\gr:trger:itefunden irn
gerrnanischen Landnahmegebiet zwischen Rhein und oberer Donau", Jahr/JJJcb des RDmischGrmranischm
Zmtralmusmms Mailt'{. 32 (I 985): 570-94. See also Henning, "Agrarkontinuitaf'.
p. 400, fig 1.
2
' Bernd Kaschau, Die Drebscbribmkeramik aJtsdm PllliJJ!,rabungm 1967-1972. Der Runde Berg
bci Urach, ,·oL 2 (Stuttgart, 1998). The same is :rue for "Cermanic" ,·ilbgcs west of the Rhine:
Gonz:uez. Ouzoulias and Van Ussel, "Saint-Oue:1-du-Bn:uil", pp. 49-51: Gm· de Boe, '·Un village
42 Tl~e /_o1(f5 Moming nf MedieralEurope
This in turn indicates that "more eating or less working"21
was not the only
option when peasant households produced a surplus.
\\'ithout wishing to revive theories of climatic determinism for the falJ
of Rome we ought, ne\·ertheless, to consider some recent results from
dendroclimatology. Burghart Schmidt has developed a ne\v method which
interprets changing cycles of Europe;;n-wide homogeneity of tree growth.
It seems to indicate that in the long run Rome failed to react to a dramatic
climatic change and its aftermath.3
" ln the decades around the middle of the
third century. a first significant unfavorable shift occurred towards colder and
dryer conditions. \'{'hen such a situation recurs se\·eral years in a row, it seriously
affects the food supply. I \vould not argue that this climate change caused the
overall transformation. Hut it may help explain why serious attempts to resolve
the problem of boosting productive labor on Roman estates only occurred after
conditions changed: in the last centuries of antiquity, some first steps toward
an agrarian organization different from the classical villa system attempted to
respond to changing enYironmcntal circumstances even under the old Roman
legal conditions. Decreasing crop yields and worsening conditions for animal
husbandry connected with such climate change must have aggranted the
problem of rural productivity inside the empire; at the same time, they must
also have caused problems in the unconquered lands beyond the li111es and
encouraged de,·elopment there.
Attaching slaYe-like coloni to the arable land while continuing the ancient
rhythms of agricultural work apparent!~· failed to resolve the problems arising
from more complicated ecological conditions and the increased human mobility
they may have entailed. A solution that in the long run was to prove stable
and superior to attempted adjustments inside the prm·inces came from outside;
it ne,·ertheless resulted from close contacts with the late Roman economy,
including, especialJy. selective adaptation of elements of the Roman technology
of production. Rural technological improvements, that is, changing the rural
production cycles, a more intensiYe style of production of a sort typical for
family-based farm units. and a spreading of agricultural risks over the whole \·ear,
combined with a broader social framework for human mobility to constitute
the responses to these social and ecological changes.
The eighth-century awakening is more closely related to the fall of Rome than
ma~· appear. First we haYe to observe unequi,·ocallv that the rural technological
improvements we have just described continued to be used bv post-Roman
peasants, even in those households that powerful aristocrats subjugated to the
gcrmanique de la seconde moitie du IV' sil:clc it Neerharen·Rekem", in J\larcd Ottc and Jacqucs
\'\'illcms (eds), La ciri/isatio11 111iroril~~itllllf dmu le bossinmosrtlt (l.iege, 1986), pp. 101-10.
~· \'\'ickham, I..md rmd Power, p. 22·1.
>• Burghart Schmidt, \\'olfgang Gruhle, Andrcas Zimmermann, and Thomas fischcr,
"l\logliche Schwankungcn n>n Gerreideertriigcn. Befundc zur Rheinischen Linearbandkcramik
und Riimischen Kaiserzeit", Arrhiiolo..P,isdm Kom.rpmtdmzblat( 35 (2005): 301-16, here p. 306. t1g. 6:
Burghart Schmidt and \'\'olfgang Gruhle, '·Niedcrschlagsschwankungen in \Vesteuropa wiihn:nd
der lctztcn 8000 Jahre'·. /lrchiiologisrhe.< Korrespondfllzblafl, .)3 (2003): 281-300, here p. 293, fig. 9.
/{ol!/f, thr Caro!Ji\~ians and SlatV:f)' in tbe First Millmnium .·! D 43
bipartite estate which arose in Carolingian times. Although written sources
of this period clearly testify to this, I prefer to emphasize the archaeological
evidence.
There is, for example, no doubt that, like many other technological
improvements, the heavy wheeled plow that turned over the sod was used in
early post-Roman times.31
This equipment was well suited to an important role
in the three-field-rotation system. ln fact, finds of seeds in the immediate postRoman
fifth century signal a new spectrum of crops. That ne\V crop complex,
with its balance between winter and summer grains, fitted well into the cycles of
the three-field system. It also made people more independent of the changing
post-Roman climatic conditions. for the three-field system's separate winter
and summer crops spread agricultural risk over the whole year. At the latest,
the turn to that new type of crop cultivation is clearly visible in i\lenwingian
times, if not already during the J\ligration period, centuries before the Frankish
version of the manorial system would be established.32
i\!oreover, it has often
been stressed that the three-field system is typical of agriculture based on cooperating
farmsteads of village communities,33
so the absence of older traces of
this system in connection with Roman villas is not surprising. And the evidence
11
h1r technical details of the swivel plnw and its appearance in the Roman \'\'est, see
llenning. "J\gmrkontinuitiit". pp. 405-!7. Por iron parts of a wheeled swivel plm\· from the
.Middle Danube area, see Dragoljub Bojovic, ·'Ostava rirnskog poljoprinenog alata iz scla Borovic
kod Ohrenovca", Got!ifnjak <~radn Be~~mdn, 25 (1978): 185-96.
12
Joachim 1-lenning, "Landwirtschaft der Franken··. in \\1ieczorck et :tl. (eds), Die Frankm:
ll''tgbtrriter Europas, pp. 774-85; Joachim Henning, "Did the Agricultural Revolution' c;o East
with Carolingian Conc[uest?" OxfordjoJmw/ of Archaeology (forthcoming).
33 Sec l\l:tx Weber, Die roi11ische Agmrgescbichte in ihrtr Bedmtungjiir dns Jtac1ts- und l'tii~Jtrrcht
(Stuttgart, 189 1), re-edited by Horst Baier et al. as Ma_" lf"e/Jer Guamfau~r,a/Jt, Abtcilung I: Jtl11ijim
und Redm, ml. 2 (fiibingen. 1986), p. 297, on how the three-field rotation system was unim:tginable
for Roman villas: ..... weil die Dreifelderwirtschaft ... keine \X'irtschaft eines lndividuums,
sondcrn einer Dorfgemcinschaft ist und mit dem Flurzwang untrennbar zusammenhiingt"; for
further references, see Wilhelm Schneider, Arbettm '{ftraln!llarmischm Friihgeschicbte, \'ol. 14, Ar/Jeitnr
'?firAgmr:P,mhichlt, part 2 (fiibingen, 1987), s.v. 'Dreifelderwirtschaft", pp. 42-92. The debate has
been whether the demesne's fields were intcgr:tted into the C\'cles of exploitation of peasant fields
("Flurzwang'). This has been shown to be the case at \X'issembourg and other estates. b·cn so,
some ha,·c thought that big blocks of reserTe :lelds existed separately from the village's arable
land: Gertrud Schri)dcr-Lembke, "~ebenformen der alten Dreiielderwirtschaft in Deutschland",
in Agrico/tura e !llondo mrak irr Oaidmte nel/'alto llltdioeJ'O, Scttimane, 13 (Spoleto, 1966). pp. 285-306,
here p. 281!. Whether the three-field rotation system was first introduced on the demesne in the
eighth century or on peasant land much earlier is ·'impossible to answer" from the written sources:
t\dria:tn Verhulst, The Carolilw~m EcorroJJD' (Cambridge, 2002). p. 62. Nc,·ertheless, the older \'icw
of August l\kitzen, Jiedrlul(g and Aganresm der !Festgmllallfff 1md Os~r,ermamn. der Krltm, /{oiJirr,
Fimrtlllmt!Jirm:m (3 vols, Berlin, 1895), vol. 2, p. 594, which assumed that the manor inlluenced
the peasant communities, has found a certain echo in the historical literature. Palcobouny is
deliniti\·cly resoh-ing the debate by documenting the rise, from the fourth or fifth century, of
a new crop system consisting of winter and SL;mmer cereals: Karl-Hcinz Kni.lrzer, "Oher den
\l;randcl dcr angebauten Ktirnerirtichte und ihrer Unkraut:Ycgct:ttion auf eincr nic:dcrrheinischcn
l.i.lflfEiche", in Udelg:ml Korber-Grohm: (cd.), Fe,;t.rrbri_fr Maria Hopj(Cologne, 1979), pp. 1·17-63.
here p. ISC..
44 Tbt I111Ig !Ifomi1~~ of MediemlE11rope
for this system among c;ermanic tribes begins only once they had intensively
adapted late Roman agricultural equipment.
It is for nmv unclear whether the gap bet\vcen the relatively good
archaeological evidence for advanced agricultural iron implements from
500 to 700 AD34
and that of the late ninth to tenth centuries35
reflects a real
decline ill everyday use of such tools in the time of Charlemagne, but welldated
finds of the eighth and early ninth centuries are still laclcing. This could
turn out to be only a problem of survival of evidence. The development of
settlement structures in the eighth century seems more alarming, however.
Large-scale excavations of ;\[erovingian settlements in Germany and France
have uncovered layouts of surprisingly large and strictly organized villages,
consisting of clearly separated farmsteads.36
They arc obviously the direct
and prosperous successors of the village structures of the third to early fifth
centuries known from northern Germany, the Netherlands and Denmark, and
\vhich shortly thereafter appeared in the former Roman territories, starting
\Vith Toxandria and southwestern Germany. This continuity of nearly half a
millennium peaked in the se\·enth century when some farm owners among
these village communities st:uted to create their own separate cemeteries within
their farmsteads. Their burials supply extraordinarily rich grave-goods such as
have been found in Lauchheim (Baden-\Xlurttemberg).37
We now understand
that such graves, which typify the upper stratum of l\lcrovingian cemeteries
in general, need nut always reflect a manorial aristocracy. The buried persons
were sometimes just wealthy peasants, living in a peasant society characterized
by internal socio-economic differentiation. Social status may have played a role,
but the bigger space of one of the farmsteads at Lauchheim also indicates a
substantial difference in landed property. The same observation can be made in
the MerO\·ingian ,·illage of Kirchheim, near Munich.
Examples like these leave no room for earlier romantic ideas of an
egalitarian peasant society. This was a society that must haYe produced a surplus
.M See 1-lenning, "Agrarkontinuit~t", p. 426, fig. S.
·'' Karl 1-licl~cher, "Fragc:n zu den Arheitsgeraten der Bauern im Mittcl:llter", ZeitschriftJiir
.·lj!,mrgescbichte, 17 (1969): (~3.
36
Vitry-en-Artois, Belgium: Etienne Louis, ''A Dc-Rom~nized L1ndscapc in Northern Gaul:
The Scarpc Valley", in \'i'illiam 1\owden, Luke La\·an. and Carlos i\hchado (cds), Recmt Research 011
tbe lAte Allfiq11e Comittyide (Lcidc:n, 2004), pp. 479-503. here p. 495, lig. 7: Genlis, France: lsabelle
Catteddu, "L'habitat merovingien de.: Gcnlis", in Claude Lorren ~nd l'atrick Pcrin (eds), L'babitat
mraldu hmlf,\lqJ·m/lge (Rouen, 1995). pp. 185-92, here p. 186, pi. I: Bussy-Saint-Georges. France:
~atalie Buchez, "Un habitat du haut Mm·en .\ge :1 Bussy-Saint-Gcorgcs", in L'habital mral. pp.
109-12, here p. Ill, with fig.: 1--:irchheim: R:uncr Christlein, ""Kirchheim bei .i\lunchen··. Das
ArrhizO/Oj'jscht Jahr i11 R~yem 1980 (1981): 162-3. farmsteads separated by fences but situated
side-by-side are clearly :lttestcd c\·en in limited ex~a,·ations: Cristina Gon~ah•es, ..Drancy (SeineSaint-Denis)'·,
in franc;:ois Gcntili, Annc Lcfcne and Nadinc Mahc. (cds), L'babital mral du baut
J/~rm Age. Supplement au Bulletin archeoiogique du Vexin fran~ais, I (Guiry-en-Vcxin, 20!J3),
pp. 56-63. here p. 57. fig. I.
l' Ingo Stork, ""Fricdhof und Dorf, Herrenhof und :\ddsgrab". in Karlheinz I·uchs et al.
(eds), Dit.A!tmJamml (Uirn, I')1)7), pp. 2'J0-31 U.
Ron1e. tbe Carolingim1s a11d J'lat't?l)' ill tbe I·irst Millmmlo11 AD 45
big enough to afford imports from around the Merovingian world, from the
eastern ~Iediterranean, even from Africa and lndia.38
It was a society whose
craft production occurred in central places associated with evidence of trade and
exchange. Thanks to more than twenty iron hoards from the early post-Roman
period (fifth-seventh centuries) and to a very large number of l\Ierovingian
burials that are \veil eguipped with iron items, we have ample information about
the widespread, everyday use of iron in post-Roman western Europe. In sharp
contrast, however, to the Roman and late Roman situation. where iron shackles
for humans came to light from almost every third iron hoard and from many,
many settlement contexts, shackles drop nearly to zero in the early post-Roman
period (see Figure 2.5). The only three finds come from outside the l\ferovingian
realm.39
No doubt the slave trade continued in the post-Roman centuries and
crossed the Continent, especially from the British isles to the south;4
" some
aspects of the slave-like treatment of rural serfs may have survived as well.
Nevertheless, judging from the shackle finds, the situation must have changed
fundamentally.
Together with the disappearance of the villa and the S\vift rise to dominance
of true villages in the Frankish heartlands, the surprising fall of western Europe·s
''iron shackle curve" from late-Roman to J'vlerovingian times (Figure 2.5) seems
to signal a complete turnaround in the social organization of the rural world.
The problem of productive labor on late Roman estates was apparently resolved
in a simple but conseguential way. An empire, its socio-economic fabric, and
especially its legal system had to be destroyed and replaced by another one
that was based upon a peasant societr that (at least in the rural sphere) did
not necessatily need iron shackles to function. I would go so far as to say that
the new logic aimed to avoid as much as possible such means of organizing
efficient agrarian production. \X'e will return to this shortlv.
\X'hat about Yillages from c. 700 to 900? Archaeologically speaking, this period
is again problematic, but not because of a lack of evidence. New excavations
in France show that villages display a signil1cant difference from earlier ones.
Small numbers of sunken floor huts had formerly been used as secondary, nonJ>
Cart Pause. ··Obcrn:gion:tlcr Giiteraustausch und \\'irtschaft bei den Thiiringern
der Merowingcrzcit", Zritscbriji jiir Arcbiiologit des Millelaltm. 29 (2001): 7-30: Cart Pause,
"Merowingerzeitliche i\lillcfioriglasperlen", Rbemiscbe.r La11dmmm11111 Bom1. 3 (1996): 63-5; Hdmut
Rorh, "Zum Handel der C'vlerowingcrzcit aufgrmd ausgewiihlter archaologischer Quellen'', in Der
Ha11del dufriibm Mittelalters, Klaus Duwel. 1-lerbert Jahnkuhn. llarald Sicms and Dieter Timpe
(eels). Untm11rb1111gm if' 1-lmideiJmd Verk.ehrder ror- 1111dfrii~epcbirbtlicbm Zeil i11 Mittrl- und t\'ordmropa,
,·ol. 3, Abhandlungen der Akademie der \'(.'issemchaften in Gottingen. Philosophisch-1-listorische
Klasse. 3. Folge, 150 (Gottingen, 1985), pp. 162-91.
l? Aldaieta, Spain (sixth cemury): llorst W.'olfgang Bi)hme. "Der Friedhoi von Aldaicta
in Kantabrien", Ada Pmebi.rtorica et .·lrrbaeologica, 34 (2002): I.J5-SO, here p. 148, fig. 9: Caricin
grad (lustiniana Prima): two shackle rings from the sixth-century le,·el. I :~m grateful to Vujaclin
I,·anisc,·ic (;'\rchaeological lnstinue. Belgrad) for making the unpublished pieces a\'ailable; Lagore.
Ireland (crannog): Thompson ··slave-Shackles", pp. 84-5. figs 29-.)0 (se,·emh century).
'" D:wid Pdteret, "Slave Raiding and Sl:we Trading in Earh· Enghnd'', .·lt(~lo-Jax011 F11gland,
9 (1981) 99-114.
46 Tbe 1-0I{f? Jllomir{fl, of AlediemlEurope
Roman Iron shackles
,--~......,................
I \
60 250 BC- 1500 AD
I \
: I \
: I \
:I I
: '. :: .,.....50
(Absolute frequency
of site occurence)
" ' :I I ;
I I :
. I I :
; I I ;
.... . '······· i ..........• ~
40 ··············t··· .................
I; I:
I' I:
: ! Carolingian
I: li : .-• :
............... ·: : .. -~ . .,.,.. ""-t\ .~ ..............................
I • ; I ; \
/ i : ' \ M:edieval
... L. .... j .......:......... ~....;...............~.............
, . ' .. .
/ ~ 1: ,......
:/ J :: ......,
· ,,# , A f:·············································Pre-Rpman : ' 1
: t,,., :
30
20
10
Merovingian
BC 250 0 250 500 750 1000 1250 1500 AD
Figure 2.5 Frequency of iron shackle finds 250 BC-1500 AD.
residential domestic buildings. In the eighth century, their numbers increased
signi6cantly41
1\loreover, a dozen excavated rural settlements in the Frankish
heartlands show that such simple huts began to receive stone m·ens or heating
facilities and so were adapted as dwellings, apparently indicating a growing
number of ill-el1uipped inhabitants of rural sites.42
For French lands, according
to Fdith Peytremann, this is a ne\v development unprecedented in the Germanic
rural world of the third to the fifth centuries.43
After peaking in the eighth
century, in the following century the new arrangement declined rapidly west
of the Rhine. But to the east, it blossomed again. brieflv. in the forecourts of
41
See the imprcssin: examples from f'rance: Nadine Beague-Tahon and J\IurieUe GeorgesLero\·,
"Deux habitats ruraux du haut Moym Age en Champagne Crayeuse", in /,'ba/Jitat mm/.
pp. 175-83. here p. 181, fig. 8. p. IS2. fig !J
·~ Drancy: Conc;ah-cs, "Drann", p. 60. Goudelancoun-les-Pierrepom, :-vier. Saint Dizicr.
Poses, Thieux, Trcmblay, Saint-Gibricn, Ensishei:n, Eply: F.dith l'eytremann, .· lrcbf(,/ogir de /'habitat
mraldtms le Nord de la Fmnce, \'01. I (Saint-Germain-cn-L:wc, 2003). p. 276.
'' Ibid.
Ro111e, thr Cflfolit(~iaus aud Slai!I!IJ i11 the Fir.rt Millnmi11111 A[) 47
royal ( )ttonian rural estates or palaces44
Compared to the l\lerovingian period,
traditionallonghouses decreased in number while sunken floor huts increased,
as can be seen west of the Rhine, for instance in the difference between the
~lerm·ingian and Carolingian phases at Speyer-Vogelsang.45
In a word, eighthand
ninth-century peasants look rather poorer than their predecessors.< )rganized
rural settlements on the earlier. true village modeL seem to survive mainly
outside the Carolingian heartland, for example in Saxony, Viking Denmark,
and neighboring areas.46
Could this help explain \vhy it took the Carolingians
half a century to conquer Saxony, even as they subdued the fortified centcrs
of Italy in a fe\v months? \\'hy Carolingian rulers had so little success against
invaders first from Denmark, then from other Scandinavian areas? From the
ninth century the invaders e\·en settled northwest of Paris, and that area rapidly
developed into Normandy, the strongest principality of the old Frankish \'('est.
One wonders whether Viking success resulted from a dramatically weakened
western European rural world, a weakness which may have arisen from serious
stresses caused by the establishment and spread of the powerful Carolingian
bipartite estate. Could Carl Hammer be right when he secs Bavaria falling back
into a world of rural slavery when the Frankish manorial system arrived there
under the Carolingians?47
l\luch would speak for J\lichael l\fcCormick's opinion that Charlemagne
had little more than human beings to offer for trade with the i\rabs.4
H \XIhere
did these enchained human wares come from? The occurrence of post-Roman
iron shackles shows an interesting progression (Figure 2.5). After an all-rime
low under the l\lerovingians, the absolute number of shackle finds surges in
Carolingian times (including in the Viking North), forming an impressive postRoman
peak unparalleled down to 1500 t\D. This peak seems to support a
significant place for the sla,·e trade in propelling the Carolingian economy.
Surprisingly, howc\'er, with the exception of two pieces found in the Seine
estuary near Rouen. the geographical distribution indicates that shackles are
absent from the inner territories of the Carolingian empire (see figure 2.3).
Nevertheless it would not be surprising if shackles emerged from the river
finds still awaiting analvsis in French museums. The written sources seem to
•• Peter Donat, Gtlnsrr: Klostrrhof 1111d kiitr{{liche Rriststation du 10.-12. ]<rbrhmrdrrts, \'\'ciman:r
~lonographicn zur Ur· und friihgeschichte, 34 (Stuttgart, 1999): Paul Grimm, "fi'llrda: F.inr
KiinigJflfalz am Kxffhdusrr 2, Schriften zur Ur- und f-riihgeschichtc, 40 (Berlin, 1990).
<\ lldmut Bcrnhard, "Ausgrabungcn in der friihmittclaltcrlichen Siedlung Spcyer
'Vogclsang'", in Adri,un Van Doorselaer (ed.), De mtroringiscbr hescbmiT(I!, in de Jrhtldtmllri (Kortrijk,
1981), pp. 223-38.
"' Saxony: Roli Lhn:nt:inger, "Vier GehOftc des 9. Jahrhunderts aus llesel. l.kr. l.ccr",
Narhrirbtm a111 l\·irdrrstubsms Urgesrhirble, 63 (1994): 39-72: Denmark: 1\nne Nissen Jauhcn,
"[.'habitat rural au Danemark \'ers 200-120(f', in L l>abitat rural, pp. 213-22.
,. Carl I. Hammer, A I Arge-smle J'/are Socit(J' of tbe Earb' Midrllr ./(f!,U: .li<ll'f! and tbrir Fa111ilirs
in Ear~· Jllfflirml l3tlltllia (Aidcrshot, 2002).
•• Miclm:l illcCormick, ''New Light on the ·Dark Ages"', /'a.<t ami Presmt, 177 (2002):
17-54; i\lich:.cl ~lcC:ormick, 'Jb~ Ori,l!,iiii of the F.11ropM11 F.ronoiiiJ': Co1111111111irrtlions aud Co!lllm•m,
A.D. J00-900 (Cambridge. 2001).
48 The lol(l( Momi11,.g rf' Medieval E11rope
imply sla\'~ transport through the Frankish lands, for the Church criticized it
in so far as Christians could be affected.49
lron shackles do occur outside the
Frankish empire from Iceland to Ireland and England, through Scandinavia
to the east~rn Slavlands and Byzantium's doorstep in southeastern l·:urope.
The concentration of Roman shackle finds exactly in the territories that would
become the heartlands of the Frankish empire contrasts dramatically with their
absence there in post-Roman times, even as abundant Carolingian-era shackle
finds cluster in a broad corona around continental \Vcstern Europe. I think
this retlects a fundamental change in the pattern of slavery and slave trade
in post-Roman Europe. The slave trade had long been oriented towards the
western prm·inces - Italy, Gaul, Spain -where rural exploitation of sla,·es was
the backbone of intensive agriculture from the period of Roman conguest until
late antiguity.50
But the logic of the post-Roman rural system developed in the
opposite direction. There is no doubt, ho\Ve\'cr, that in the same centuries when
tl1c Carolingian ruling elite intervened powerfully in tl1e countryside to establish
tl1eir estates, the European slave trade expanded anew. ~fcCormick has shown
that the Carolingians \'er~' probably were involved in that trade and capitalized
on it, and that the Ottonians did the sameY The absence of iron shackles in
the rrankish hcartlands shows however that the enslavement and selling abroad
of the western European rural peasant population was exceptional. The slave
extracting grounds \vere in fact the non- or semi-Christianized ncighboring
lands \Vhcre Frankish armed forces, local chiefs or trade-loving Vikings \vere
active in the slave-hunting business.52
It seems to me highly significant that
Carolingian and Viking-age shackles in the Frankish periphery never occur in
open, unfortified or predominantly rural settlements (see rigure 2.4). As a rule,
they come from major fortified sites: ringforts (for example, Denmark's famous
Trellcborg, whose name means ''slave castlc"),53
Slavic ramparts (for example,
Kniazha gora, "Princely hill" in the Ukraine), Irish crannogs Oocal aristocrats'
•• Charles Verlinden. "Problemes d'histoire economique iranquc !"', IVI'IIe beige dtpbiloiOJ!,it r1
d'bistoirt, 12 (1933): I090-95; \X1altraut Blcibcr. i\ial11ralu·irtscbajt 1111d ll""an-Ctld-Brzitbm~P,tll '{flisrbm
Jommt 1111d Loin ll'iihrmd des 7. Jabrbmtderts, frmchungen zur mittelaltcrlichen Geschichte, 27
(Berlin, 1981), pp. 73-5.
~· The situation in the eastern Roman Empire was different because the slarc system, at
least according to documentary sources. was more dc,·eloped. Howe,·er. we need to in\"estig:ue
the situation from written sources since findings of. for instance. shackles, in the cast arc \"cry
rare. for example, in Israel. no archaeological e\•idencc of slawry has been found, although the
Talmud mentions traders who arri,·ed \\ith 20,000 iron shackles in order to bur lews: Samuel
Krauss, "/{;/mlldiJdlt .-Jrrbao/Oj!,it, ml. 2 (Leipzig, 1911), p. 96. . .
31
Joachim 1-lenning, "Neue Burgcn im Osten", in 1\chim I lubel and Bernd Schncidmiilkr
(eds). /l11jbmcb i11s ::;}l'tiltJabrtallsmd (Ostfildern, 2004). pp. 151-81. here p. 176.
51
Sec the reflections on the origin of Carolingian-era slares in Joachim llenning. "Sl:wery
or freedom' The Causes of Earl\' Mcdie,·al Europe's Economic Adnnccment'", F::\IE, 12
(2003): 269-77, hen: p. 272: i\lichaell\lcCormick, "'ComplexitY. Chronology and Context in the
Early i\lcdie\·al Economy", EAIE, 12 (2003): 307-23, here pp. 308-12.
;; l.cszck Pawcl Slupecki, 'Joms,·ikingalog, Jc">msvikings. Jomsborg/\\"olin and Danish
Circular Strongholds", in l'rzemyslaw Urbat1cZ\·k (ed.), Tbr Nrigbbom:r rf Poland intbr lOth cm/Ill")'
(Warsaw, 2000). pp. 49-59, here pp. S4-S.
Rome, tbr Carolit{?,icms and Skll'fl)' in tbe First Millmnill!ll AD 49
man-made island settlements in lakes) and especially an impressive series of
fortified early trading centers, some of which have been called proto-towns
or just plain towns: Dublin, \\'inche~ter, Haithabu, Stare i\'lesto (the biggest
center of ninth-century :Moravia), Nitra (residence of the prince of eastern
l\[oravia) and Preslav (the capital of the Bulgarian empire). \'('ritten sources
attest the slaving background of some of these centers. which mushroomed
in the Carolingian era and which archaeologists often prize as signs of an
upturn of the rural economy in their hinterlands. To some extent that might
well be true, altl10ugh it remains to be proved. The slaYe trade, howeYer, and
its archaeologically well-proven markers, must be taken into account when
explaining the sudden rise, brief flourishing, and mysterious disappearance of
many of these trading places in post-Carolingian times. 1t could turn out that
their early medieval trajectory peaks in unison with our iron shackle curve.54
Silver coins found around such trading places do not contradict this vie\v.55
According to written sources silver regularly accompanied that business and
facilitated far-reaching economic connections, for example to the Arab world.
Silver coins flowed into the Carolingian lands. and \vere minted there as well. \'\le
cannot be sure that they testify exclmively and generally to a strong economy
relying on increasingly efficient rural production. lt is worth repeating that
technologically speaking, all the known important post-Roman agricultural
improvements had already been invented centuries before the Carolingians. \X'e
might rather suspect that the application of Carolingian pmver structures to the
western European countryside triggered developments that were, in the short
term, unfavorable to the inner texture of at least some peasant settlements
within bipartite estates. This \vill have compromised their efficiency. As far as
we can tell, a reinforcement of '·normal" villages seems to occur no later than
the tenth century, for example at La Grande Paroisse in the Paris basin% This
is exactly the period when the lord's reserve, the crucial element of bipartite
s.. I \Vould tend to agree \vith Frans \'erhaeghe. \\"jth Christophcr Lovcluck and Joanna
Story, "Urban De\'elopments in the.: Age of Charlc.:magne''. in Joanna Story (ed.), Charlemagne:
Empiretmd .loriel)· (1\!anchester, 2005), pp. 259-87, that "the importance of emporia in the range
of urban settlement in the Carolingian period ha~ been m·erstressed" (p. 269).
" Sec llcnning, ·'Neuc Burgcn", pp. 173-81: Sebastian Brather, "Friihmittelalterliche
Dirham-Schatzfunde in Europa··. in Zeitscbrift .fiir Arrbiiolo,gir des :\lillekrlters, 23/24 (1997):
73-153, here pp. 182-6; Sebastian Brather "Friihmittelalterliche Dirham-Schatz-und -Einzclfundc
im siidlichen Ostseeraum··. in Sebastian Bra6er, Christel Biicker, and f.lichael Hocpcr (eds),
Arrbiiologir als Soi.falge.rrbirbte, Studi:t Honoraria, 9 (Rahden 1999), pp. 179-97; and in England:
Mark Blackburn. "'Producth·e· Sites and the P:t:tern oi Coin Loss in England. 600-1180", in Tim
Pcstcll and Katharina Ulmschneider (eels). Market.< in Em:l)' JllrdiemlEurope. Fmdi11._~ a11d "Produrtiz¥!"
.l'iles, 650-850 (Macclesf1eld, 2003), pp. 20-3C>, here p. 22, fig. 3.1: ~lichacl Metealf. "Variations
in the Composition of the Currency at Diiferent Places in England". in Marlut.r in Early Medin'f11
Europe, pp. 37~7.
'• Michel Petit. "La Grandc.:-Paroisse (Scine-ct-Marne)", in i\lichel Petit and Monique
Depraetere (eds). L '1/e-de-Fmnrr de Clovis ,/ Ht(~Uts Cape! (Paris, I993). pp. !99-200: :-.1ichel Petit, "La
Grande-P:uoisse", in Jean Cuisenier and Remy Guadagnin (eds). Un nlk{l!,t 1111 temps de Cbt~rle"''ll'."t
(Paris, 1988), pp. 147-9.
50 V1e Lo!{l!, illomil(l!, rf i\Iediel'fllEurope
manorial organization, which had floutished in Carolingian times, declines
north of the Alps.
Finally, let us look at exchange in the eighth century. J.ately scholars have
stressed the upturn of long-distance trade to the North and the establishment of
trading places or proto-towns with increasing craft activities, as well as grmving
agricultural production around them. Even though many proto-towns may
ha,·e owed some of their success to the slave trade, the growth of new centers
cannot be denied. But they cluster on the Frankish borderlands, on the Atlantic
and North Sea coast, the Baltic coast with Reric and the Elbe-Saale-frontier with
its trading places such as 1\fagdeburg and Erfurt described in Charlemagne's
capitulary of 805.57
They are even more prominent outside the Frankish empire,
at places such as Haithabu in Denmark, or Frisia, Sweden, England, and Ireland.
The archaeological evidence for significant tmvn deYelopment in the Prankish
heartlands, the homeland of the much-admired Carolingian manor, however,
is still missing. The old Roman urban centers are now deliveting more and
more archaeological evidence that specialized craft production flourished in
1\ferm·ingian times5
8
Gregory of Tours' sixth-century Paris was a Ji,·ing city
with workshops and markets. Paris's l\fusee Carnavalet is full of finds from that
period, but offers nearly no items or structures from the eighth century. Frans
Theuws has shown that, after lively l\[erovingian craft production in several
Meuse valley towns, a hiatus ensued in the eighth century.5
9
The same holds for
Cologne. According to recent excavations the center was not abandoned and
ruralized in post-Roman times, as scholars had previously assumed60
Instead,
the 1\lerovingian period saw flourishing craft production, including highly
57
MGH LL, 2. Capitnla1ia regum Fmncomm, ed. Alfred Boretius. \'ol. I (1-lano,·er, 1883).
no. 44, c. 7, p. 123.
58
Cologne: Marcus Trier, "~uln im fruhen !.litrelalter··, in Joachim l-lenning (ed.) Europa
im /0. Jahrhnndert: Archtiologie riner Arifbmchszeit (Mainz, 2002), pp. 301-10; Namur: Jean Plumier,
"Namuco Fit: Namur mcrovingicn", in XX' ]o11mies intemationales d',,rchiologie miro1ingimne, Jean
and Sophie Plumier-Torfs and i\Iaudc Rcgnard (cds), Bulletlil de liaison, 23 (1999): 29-32: for
further Roman settlements in the Mcuse \'alley that were Merm·ingian production ccntcrs sec
Jean and Sophie Plumicr-Torfs, Maude Rcgnard and \X'im Dijkman (cds), 1Hosa l'Voslm: La Mmse
miroti11gitm1ede Vmlrm aMaastricht. Carnets du patrimoine, 28 (Namur. 1999); Mainz: Egon \V'amers.
Die friihmiffelallerlirhm Lu!fimde tillS der Uihrsh"tlSJe, Mainzcr archiiologische Schriftcn, 1 (Mainz.
1994). pp. 162-75. with well-attested i\ft:rovingian craft production; written sources: Stcphane
Lebecq. "Les echanges dans la Gaule du nord au \'1' sii:cle". in Richard !lodges and \X.'illiam
Bowden (eds), The Si:-:tb CmfiiiJ': Production, Distribution rmd Demm1d, The Transformation of the
Roman \\1orkl. 3 (Leiclen, 1989), pp. I85-2U2; for Merovingian craft production in Roman urban
centers of Gaul (Paris. Gene\·a, Bonn, and so on), see Helnnn Roth, Kmuf und Ha11dwerk imji·iiherl
Miffellllter (Stuttgart, 1986), pp. 51-4; for ongoing urban functions see: S.T. Loseby, "Gregorv's
Cities: Urban Functions in Sixth-century Gaul'·. in Ian \\'ood (ed.), Frrmks and Alammmi in the
Merwi1(gia11 Pfli&d (San i\larino, CA, 1998). pp. 239-84.
'• Frans TI1euws. "\\'here is the Eighth Cc.:ntury in the Towns of the i\leuse Valley'", in
Joachim Henning (ed.), Post-Roma11 ToJI'IIS: Trade a:ld.l'elllemmt i11 Europe tmd ~yzrmlitll/1, ml. 1 (Berlin
and new York, 2007), pp. 153-64.
'" 1-leiko Steuer, Die Fnmke11 i11 J.:.o/11 (Cologne, 1980); lleiko Steuer. "Stadtarchiiologic.: in
Koln", in 1-ldmut_lager (ed.) Stadtkemfomhun:;,, \'criiffentlichungen des 1nstiruts fur Yergleichende
Stiidtcgeschichtt: in \lunster, series:\: Darstellungen, 27 (Cologne, 1987). pp. 61-102.
Rome. the Carofi,(giml! ,md.lltn'l!l)' in tbe First ;1/i//mniulll .-JV 51
specialized installations such as gbss ovens61
\X'hen production actiYitics next
picked up is still in dispute but it seems to be about the tenth century at the
latest. According to Simon Loseby, 1\farseille was still a "great port" under rhe
1\ferovingians, but declined in the eighth century under the Carolingians to a
monastic site of local importance at most.r'2
It started to revive only in the tenth
century. In the eighth century Venice may have been far from replacing that
gatc\vay to the eastern Mediterranean. It probably started its career as a port of
slave trade to the south, first under the auspices of the Carolingians and then
on its u\vn account - probably the reason it survived after Carolingian power
declined. The tenth-century reviYal of so many of the old Roman centers is
now a familiar phenomenon. at least in continental Europe.63
\'\'bile specialized production declined in the old Roman centers in postMerovingian
times, it moYed in part to places that were better controlled by
the aristocracy, as is particularly well documented for monasteries, paramount
administrative centers of the estate system. Although specialized craft production
in monasteries was not absolutely new- the J\ferovingian monastery of Saint
Denis had had its workshopsM -archaeologically attested monastic workshops
increase significantly in the eighth and ninth centuries. New discoveries of
glass workshops from the abbeys of Lorsch, Fulda and Corvey in Germany,
from San Lorenw al Volturnu in ltaly, and from Barking abbey (England),65
61
Colognt:'s Heumarkt are:t seems to show two main periods of production and tnult:. ()ne
is the layer underneath the market floor (dendrodated to 957 AD), which has yielded excellent
Mcro\'ingian finds, whereas Carolingian and Ottonian finds came to light exclusi\·ely abm·c the
market pa\•ing. For details, see 1-lelnmt Roth a:1d i\larcus Trier, "Ausgewiihlte Fund.: des 4. his
11. Jahrhundcns :ms den Ausgrabungt:n auf dem Heumarkt", KolnerJahrbHcb, 34 (2001): 759-91.
Trier looks simil:tr: Lukas Clemens, "Archiinlogische Beobachtungen zu friihmittclaltcrlichen
Siedlungsstrukturen in Trier", in Sabin.: Felgenhauer-Schmicdt, Alexandrine Eibncr and llcrbcrt
Knittler (cds), 7.uiJ(ben Ron1rrsiedhmg 11ndmitteltrlterlicherStad, Beitriige zur 1\littelalterarchiiologie in
Osterrt:ich, 17 (Vienn:1, 200 1), pp. 43-66, here p. 45, fig 2 (early Mcrovingian workshop activities:
fibula mold), pp. 58-9 (written sources about Carolingian agricultural and ,-ineyard acti,·itit:s in
the town area), p. 60, fig. 18 (increasing activity in Ottonian times).
''
2
Simon T. Loseby, "Marseille and the Pirenne thesis, J", in The .'ii:xth CmlllfJ', pp. 203-29:
Simon T. Loseby, "Marseille and the Pirenne thesis, 11", in lnge Lyse llansen and Chris \'\'ickham
(eds), Tht J.~mg Eighth Cmtnry, The Transformation of the Roman World, 2 (L.eidcn, 2000),
pp. 167-93.
61
Frans Verhaeghe, "Continuity and Ct:ange: Links between .\ledieval Towns and the
Roman Substratum in Belgium", in Rudof De Smet, 1-lenri '.lelacrts and Cccilia Safrens (.:ds),
St11dia Varia Bm.wllmsia (L.em·en, 1990). pp. 229-53.
"' Sec the l\lermingian molds for casting fibulas: Patrick Perin. "Lcs moulcs de fond.:urs de:
Saint-Dcnis (Scine-Saint-Denis)'', in Petit and Dcpraetere (eds), L'lle-de-Francr, pp. 279-!ll.
''
1
Class production in the imperial abbeys: l\larkus Sankc. Karl 1-lans Wedcpohl, and
Andrcas Kronz, "Karolingerzeitliches Glas aus dem Kloster Lorsch". ZeiiJChrijiflir /lrchiiol~~ie drs
Mitlda/lm, 30 (2002): 37-75: Abbey of l'ulda: Thomas Kind, Karl 1-lans Wedepohl and J\ndrc:as
Kronz, ''Karolingerzeitliches Glas und n:rschiedene !Iandwcrksindizien aus dem Kloster
Fulda", Zfit.rrbrift flir /lrrhiiolo,~it du lllilleltJ/Im. 31 (2003); 61-93: Hans-Georg Stephan, Karl
Hans \'\hlcpohl and Gerald Harmann, "Mitte:alterliches Glas aus dcm Rcichsklostcr und dc::r
Stadtwiistung l.on·ey", Ger/1/allia. 75 (1997): 673-715:Judv Ste\·enson, ":-Jinth-centmy ( ;Iassw:tre
PrnducJion at San Vincenzo al Volmrno". in f<rans Verhaeghe (eel.). Maleric~l Ctdttm i11 Jllfdirml
Europe (Zellik, 1997), pp. 125-36; Barking abbey: Julian Henderson, "Le \-erre de Dor.:stad?"
52 The Long :Homing of llledieml E11ropr
of pottery workshops from the Heiligenberg monasteries (Germany) and from
the abbey of St George in Baralle (france)M' as well as many other specialized
craft works (comb production, bronze foundries, and so on) make clear that the
Carolingian plan of Saint Gall's \vorkshops was anything but wishful thinking.
1\lonastic substitutes for ··normal" tO\vns seem to mark a Carolingian detour in
European town development that \vas dearly paid. These curious "monasterytowns"
remained but an episode. Along with the bipartite estate system, they
soon declined. The true towns \vhich replaced them often revived sites with
older Roman and Merovingian town traditions.
If this all reflects the impact of the installation of the mighty Carolingian
power strucmres of the bipartite estate system accompanied by an upmrn of
European slave trade, it looks rather like an accidental rebirth of Rome that \vas
imposed on a formerly flourishing peasant society. So \Ve \Vould have to agree
with 1\fichael McCormick that Pirenne was both right and wrong. And this is
my conclusion: Pirenne was probably right when he stressed that the eighth
cenmry brought a visible stagnation and sidetracked, if it did not set back.
parts of Europe's economy. The roots of that setback must lie in a problematic
reorganization of the rural world \vhich distorted the productive logic of smallunit
agriculture and blocked its potential efficiency. Adriaan Verhulst once
pointed out that the Capit11/are de [,.;//is \Vas likely closely connected to the famines
of 792-93 and 805-(> AD.67
The frankish king's orders were anything but a \vise
handbook of agriculmral knowledge. On the contrary, as Alfons Dopsch long
ago maintained, the capitulary shows the bankruptcy of the bipartite manorial
system. If this is right, Pirenne is also wrong: it \vas not .Muhammad who was
responsible, but Charlemagne. 1\Iy impression is that, in the economy of earlv
medieval Europe, places or periods with relatively weak power strucmres were
more innovative and efficient so long as they had access to the most adYanced
technical improvements of late antit]Uit\·.68
This raises serious doubts about the
currently dominant view, which suggests that strong or eYen centralized power
structures were indispensable in boosting peasants' small production units to
greater producti\"ity and efficiency. Historv seems neYertheless to have offered
in Daniele f'oy (ed.), Le mn de llmtiquiti tardll't et dtt haul m~rm ~ge (Guiry-en-Vexin, 1996),
pp 51-5
66
f'or pottery production at the Heiligenberg monastery (GermanY), see Peter Marzolff.
..Die bcncdiktinischcn Bcrgkloster auf dcm 1-Iciligcnbcrg bei 1-leidelberg", Beitriige if',.
:\Jittrlallerarrhiiologie in Oskrreirb, 12 (1996): 129-45: for Barallc (r:rancc), see Alain Jaques, "Un
atelier de production de ceramiques au ham moyen age aBaralle'·, in Gauheria. 31 (1994): 89-100;
Alain J:tques, ..lJn four de potier du haur moyen age 3 Baralle'·, Rmte du Nord, 58 (1976): 73-86.
6
- :\driaan E. Verhulst, "Karo1ingische Agrarpolitik: Das Capitulare de Villis unci die
Hungersnote von 792/93 und 805/06", Zdtsc!JI?ftfiir / lgrarg1·.<rhirhte. 13 (1965): I75-89.
•· f'or earh· retlections in that direction see: Joachim Hcnning...,\.inschaitsarchao1ogie des
Friihmittelalters und aktuelle Fragen der Geschichtswissenschaft... J'tudijlli ifY:Sii /Jrcheolo.~ickibo
ri.rtm•11 !ilm·m.rkq akadi11tie ried. 25 (1988): 41-6, here p. 44. Michaell\lcCormick, "Um 808: Was der
friihmitteblterliche Ktinig mit der \\'irtschaft zu tun hatte", in Bernhard Jussen (ed.), Die Alacbt
des Kiin~f!,S: J-hrrsrhtifi i11 Europa 1'01!1 Friihmittelalter bis in die T''>l-mvif (Munich, 2005). pp. 55-71, makes
the s:tme point abour merchants and relati\·eh· weak ro\·a1 or srate power strucn1res.
Ro111e. the Cllrolil(l',ians tlfld J!aJ,'fry in the Firs/ Millenniu111 AD 53
its own answer: Rome disappeared, and so too did its Carolingian rebirth. Other
centralized power structures have disappeared recently, but peasant structures,
self-managing and self-determined economies have continued to exist. The
Carolingian villa \vas probably responsible for a really Yery long - I would say
too long - eighth cenntry, if not at all for a, literally, tongue durie of the early
:Middle Ages.