Вы находитесь на странице: 1из 15

Tables des maTires eT Thmes

Tome 20. 2012. Mondes ruraux en Orient et en Occident II

Mondes ruraux en Orient et en Occident. 6 le vocabulaire J.-M. Carri, Nommer les structures rurales entre fin de lAntiquit et Haut Moyen
ge : le rpertoire lexical grco-latin et ses avatars modernes (2e partie). 7 les campagnes en priphrie de lempire romain A.S. Esmonde
Cleary, Northern Britain in Late Antiquity ; A. Stuppner, Die lndliche Besiedlung im mittleren Donauraum von der Sptantike bis zum Frhmittelalter ; A. Poulter, Goths on the lower Danube: their impact upon and behind the frontier. 8 les formes de lhabitat Cl. Negrelli, Le strutture del
popolamento rurale tra IV e IX secolo in Emilia Romagna e nelle Venezie ; E. Ario, El hbitat rural en la Pennsula Ibrica entre finales del siglo IV y
principios del VIII: un ensayo interpretativo ; M. Veikou, Settlements in the Greek countryside from 4th to 7th century: forms and patterns ; C. Duvette,
avec G. Charpentier, C. Piaton, Maisons paysannes dun village dApamne, Serilla (ive-vie sicles Massif calcaire de la Syrie du Nord) ; I. Taxel,
Identifying social hierarchy through house planning in the villages of Late Antique Palestine: the case of orvat Zikhrin ; D. Mattingly, M. Sterry,
V. Leitch, Fortified farms and defended villages of Late Roman and Late Antique Africa. 9 le rle des implantations ecclsiales G. Cantino
Wataghin, Le fondazioni ecclesiastiche nelle vicende delle aree rurali: spunti di riflessione per lOccidente tardo antico (IV-V secolo) ; Y. Codou,
Lglise et lhabitat dans le Midi de la France aux ve-xe sicles ; M. A. Cau, C. Mas, Christians, peasants and shepherds: the transformation of the
countryside in Late Antique Mallorca (Balearic islands). 10 les rapports entre villes et campagnes D. Fernandez, City and countryside in Late
Antique Iberia ; F. Cantini, Aree rurali e centri urbani tra IV e VII secolo: il territorio toscano.
Varia : V. Goncalves, Aleae aut tesserae ? Les significations dune opposition ludique dans la Rome dAmmien Marcellin ; A. J. Kosto, The transformation of hostageship in Late Antiquity ; Chr. Freu, Les salaris de la terre dans lAntiquit tardive ; St. Del Lungo, Provincia Lucania: topografia e
agrimensura in un paesaggio che cambia, dalla Tarda Antichit allAlto Medioevo (prima parte) ; P. Grossmann, berlegungen zum ursprnglichen
Grundriss der Kirche von Orlansville (Chlef, Algeria) und ein Beitrag zur Entstehung der christichen Basilika ; S. Ordez Agulla, J. Snchez
Velasco, E. Garca Vargas, S. Garca-Dils de la Vega, M. A. Tabales Rodrguez, Novedades arqueologicas de las sedes episcopales de la Btica
Occidental ; A. Uscatescu, Visual arts and paideia: the triumph of the theatre revisiting the Late Antique mosaic of Noheda.
Chronique : S. Ratti, Paens et chrtiens au ive sicle : points de rsistance une doxa ; B. Laszo Toth, Regards nouveaux sur le trsor de Nagyszentmikls, la suite dune publication majeure.
bulletin critique de de Sylvain Destephen, Adam Kosto, Sylvain Janniard, Hendrik Dey, Maria Grazia Bajoni, Jitse Dijkstra, Liudmila Khrushkova,
Jean-Louis Charlet, Antonino Metro, Jean-Pierre Coriat, Michael Whitby, Laury-Nuria Andr.

Rappel des tomes 1 20 (tous disponibles) :


Tome 1, 1993. les sarcophages daquitaine
Tome 2, 1994. la ttrarchie (293-312) : histoire et archologie (1re partie)
Tome 3, 1995. la ttrarchie (293-312) : histoire et archologie (2e partie)
Tome 4, 1996. les glises doubles et les familles dglises
Tome5, 1997. largenterie de lantiquit tardive
Tome 6, 1998. les gouverneurs de provinces
Tome 7, 1999. figures du pouvoir : gouverneurs et vques
Tome 8, 2000. le de dificiis de procope: le texte et les
ralits documentaires
Tome 9, 2001. la dmocratisation de la culture dans
lantiquit tardive
Tome 10, 2002. lafrique vandale et byzantine (I)

21 - 2013

Revue internationale dhistoire et darchologie (ive- viiie s.)


publie par lAssociation pour lAntiquit Tardive

Mondes ruraux
en orient et en occident - ii

21 - 2013

Tome 11, 2003. lafrique vandale et byzantine (II)


Tome 12, 2004. tissus et vtements dans lantiquit tardive
Tome 13, 2005. la vaisselle de bronze palobyzantine
Tome 14, 2006. conomie et religion dans lantiquit tardive
Tome 15, 2007. Jeux et spectacles dans lantiquit tardive
Tome 16, 2008. Lpoque des Thodoses
Tome 18, 2010. Lecture, livres, bibliothques
Tome 19, 2011. Christianisation et images
Tome 20, 2012. Mondes ruraux en Orient et en Occident (I)
En prparation :
Tome 22, 2014. LOrient chrtien de Constantin et dEusbe
de Csare

F H G
Begijnhof 67 B-2300 Turnhout (Belgium)
Tl. : +32 14 44 80 35 Fax : +32 14 42 89 19
periodicals@brepols.net www.brepols.net

Couv Antard21.indd 1
at_21_cover.indd 1

antigedad tarda Late antiquity


sptantike tarda antichit

antiQuit tardiVe

La revue annuelle Antiquit tardive (Antigedad Tarda Late Antiquity Sptantike Tarda Antichit), dite
par Brepols, est une revue multilingue rdige sous la responsabilit de l Association pour lAntiquit tardive
prside par Fr. Baratte et reconnue par le Centre national de la recherche scientifique franais.
Chaque numro est centr sur un thme principal mais comporte trois autres sections : des Varia dhistoire,
dhistoire du droit, darchologie et de philologie ; une Chronique sur des sujets ponctuels dactualit ; un Bulletin
critique rserv des comptes rendus gnralement dtaills douvrages importants, suivi de Notes de lecture sous
la forme de brves fiches.
Un rsum en anglais (ou en franais pour les articles en anglais) prcde chaque article. Antiquit Tardive est
dsormais accessible en ligne sur le site Priodiques de Brepols (pour plus de prcisions, voir lditorial 2005).

antiQuit tardiVe

ISBN 978-2-503-55066-4

9 782503 550664

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER
19/11/2013 12:29:16
21/11/13 09:03

AnTard, 21, 2013, p. 63- 76

GOTHS ON THE LOWER DANUBE:


THEIR IMPACT UPON AND BEHIND THE FRONTIER
$3

Les Goths du Bas Danube :


leur impact sur et au-del de la frontire
Cet article traite dun certain nombre daspects relatifs linstallation des Goths dans le Bas Danube. Il ne semble pas
que laccord de 382 ait eu limpact suppos jusqu maintenant. Bien quil soit admis, grce au bas-relief de la colonne
dArcadius, que les populations indignes reconnaissaient les Goths par leurs vtements et, bien entendu, par leur langue,
rien dans la documentation archologique ne permet de distinguer ces derniers des natifs . Par ailleurs, la culture de
Sntana-de-Mure / Cherineakov ne parat pas avoir exist : le seul lment trouv dans les installations attribues
cette culture (et aux Goths) est en ralit dorigine romaine et il ny a aucune preuve que les autres lments soient
gothiques . Enfin, la lumire des tmoins archologiques concernant la priode suivant la bataille dAndrinople et la
premire moiti du e sicle, il apparat que, quelle quait t lorigine ethnique des occupants, une nouvelle politique a
t engage vers 400, qui a entran la construction de nouveaux forts et le ramnagement de plus anciens de manire
accueillir des soldats-fermiers. Malgr la disparition du systme conomique des villas, lexploitation agricole a continu
jusqu larrive des Huns. Durant cette priode (400-445), lautorit militaire romaine et limpt de lannone ont jou un
rle capital dans la prservation de la paix et les garnisons de soldats, qui taient galement engages dans lagriculture,
ont assur la scurit sur la frontire comme lintrieur des terres, en mme temps quelles ont continu exploiter la
richesse des ressources agricoles. [Auteur, trad. de la Rdaction]

The catastrophic defeat of Valens in the battle of


Adrianople in 378 marked a signal change in the fortunes of
the Eastern Empire. Valens army was almost annihilated,
losing perhaps as many as 20,000 to 26,000 men, two thirds
of its total strength.1 For the lower Danube, the devastation

Abbreviations: Heather, Goths and Romans = P. J. Heather, Goths


and Romans 332-489, Oxford, 1991. IGL = E. Popescu (ed.),
Inscripiile Greceti i Latine din Secolele 6-13 descoperite n
Romnia, Bucharest, 1976. Liebeschuetz, Barbarians and Bishops
= J.H.W.G. Liebeschuetz, Barbarians and Bishops; Army, Church, and
State in the Age of Arcadius and Chrysostom, Oxford, 1990. Poulter,
The Transition to Late Antiquity = A.G. Poulter, The Transition to Late
Antiquity on the Danube and Beyond, Oxford, 2007.
1. N. Lenski, Failure of Empire: Valens and the Roman State in the Fourth
Century AD, 2002, Berkeley, 2002, p. 339. A more conservative view
estimates losses of c. 10,000-12,000; Heather, Goths and Romans,
p. 147. Whatever the exact number, it is certain that the remnants of

of Moesia and Thrace disrupted no doubt, at least in the


short term, the agricultural economy which underpinned
the regions evident prosperity under the Antonine, and
especially the Severan dynasties.2 The long term impact was
LQQLWHO\ZRUVH(VWLPDWHVDVWRWKHQXPEHUVRI*RWKVZKR
remained in the region vary but the total, including men,

WKHDUP\LQKHULWHGE\7KHRGRVLXVZHUHLQVXIFLHQWWRGHIHDWWKH*RWKV
despite relentless measures taken to levy new recruits. I would like here
to record my thanks to Wolf Liebeschuetz whose advice and comments
have made an invaluable contribution to my understanding of this
complex period at the end of the 4th century.
2. For the region bordering the Danube, plundering raids were already
devastating the lands around Marcianopolis when the Gothic revolt broke
out in 377 (Amm. Marcell., 31, 5, 7-9). Then, after the battle of Adrianople,
the whole of Thrace was at the mercy of the Goths (Amm. Marcell.,
31.6.4). For the destruction of the villa economy, see below, pp. 72-73.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

64

AnTard, 21, 2013

ANDREW POULTER

women and children may have been as many as 200,000.3


$EVRUELQJ VXFK VLJQLFDQW QXPEHUV LQWR WKH SUHH[LVWLQJ
Roman system of land tenure must have caused incalFXODEOH GLIFXOWLHV 6R PXFK FDQ EH DFFHSWHG RU DW OHDVW
presumed. However, the reality of the situation is less easy
to reconstruct. The literary sources for the period from the
death of Valens until the departure of Alaric are singularly
unhelpful, as will be explained below, and the almost total
ODFNRIKLVWRULFDODFFRXQWVIRUWKHUVWKDOIRIWKHth century
leaves a blank hole in the sequence of events leading up
to the arrival of Attila and the Huns. Inevitably, the only
possible way forward is for archaeological research to plug
the gap, at least as far as it can, and not always in adequate
ways resolve key issues of interest to ancient historians. A
joint Anglo-Bulgarian research initiative has been active
in researching the transition between the Roman and early
Byzantine periods on the lower Danube. It started with
the excavation of the Roman and early Byzantine city of
Nicopolis ad Istrum (1985-1991), and was then followed
by the excavation of the late Roman fort of Dichin (1996 DQGWKHGHYHORSPHQWRIDQHZIRUPRIVLWHVSHFLF
survey to explore the countryside. These two programmes
were followed by the excavation of a late Roman fortress,
Dobri Dyal (2011-). All three sites and the survey help
WRSURYLGHDUHPDUNDEOHLQVLJKWHVSHFLDOO\IRUWKHUVWKDOI
of the 5th century, which goes some way towards explaining
what actually happened, especially when compared with
reliable results from other excavations within the region.4
The settlement of the Goths
After the Roman defeat, the Goths were at liberty
to plunder the North Bulgarian plain (Dacia Ripensis,
Moesia II and Scythia) as well as the southern half of the
Thracian diocese, the hinterland of Constantinople.5 The
situation was critical; all attempts by Theodosius to restore
imperial control over Thrace met with little success; military

3. Eunapius (Hist. Fr. 42) claims that the number of Goths who crossed
the Danube was as many as 200,000; whether only warriors or the total
number, including men, women and children is unclear. This might just
be a rough estimate of the total number of new arrivals; it could hardly
EHWKHQXPEHURIJKWLQJPHQ1/HQVNLFailure of Empire, cit. (n. 1),
p. 354-345. As to the number of Gothic warriors which faced Valens in
the battle of Adrianople, a number of 20,000 seems to be a reasonable
assessment; Heather, Goths and Romans, p. 139, note 44.
4. The excavations at Nicopolis have been published: A.G. Poulter,
Nicopolis ad Istrum: A Roman, Late Roman and Early Byzantine City,
London, 1995; Id. Nicopolis ad Istrum, a Roman to Early Byzantine
City: The Pottery and the Glass, London, 1999; Id., Nicopolis ad Istrum,
a Late Roman and Early Byzantine City: The Finds and the Biological
Remains /RQGRQ 7KH QDO UHSRUWV IRU H[FDYDWLRQV DW 'LFKLQ
are in preparation, although the principle results and implications of
this large-scale programe have been published: Poulter, The Transition
to Late Antiquity. A report on the excavations at Dobri Dyal is to be
published in the Journal of Roman Archaeology, 26, 2013, pp. 362-383.
5. See above, n. 2.

operations against the Goths ended in defeat for the hastily


reconstructed army. For four years, the lower Danube was
at the mercy of the invaders and newcomers from across
the Danube, no doubt attracted by the prospects for plunder.
Eventually, in October 382, a settlement was reached
between the Goths and the Empire, though it was one
ZKLFKGLGQRWIROORZD5RPDQVXFFHVVRQWKHEDWWOHHOG6
What is certain is that the peace involved the provision of
land for the Goths to settle and farm.7 Themistius claimed
WKDW WKH *RWKV WXUQHG WKHLU ZHDSRQV LQWR SORXJKVKDUHV
However, such a rhetorical view does not help to reconstruct the terms of the settlement and it provides no useful
LQIRUPDWLRQDERXWWKHOHJDODUUDQJHPHQWV$PRUHVSHFLF
reference by Pacatus takes us a little further: some of the
Goths became farmers, others became soldiers. A year
before peace was agreed, after the death of the Gothic
chieftain Athanaric, some of his followers certainly not
those involved in the revolt and the Battle at Adrianople
ZHUHVWDWLRQHGLQQRUWKHUQ7KUDFHZKHUHWKH\NHSWZDWFK
RQWKH'DQXEHDQGIRUDORQJWLPHSUHYHQWHGDQ\DWWDFN
on Roman territory (Zosimus 4.34.5). Where the Goths
were given land is easy enough to determine. Themistius
refers broadly to Thrace, but Jordanes (Getica 25.133)
names Dacia Ripensis, Moesia and Thrace, provinces
close to or bordering the river Danube (J ).8 There is
also independent evidence for Gothic soldiers in Scythia at
some date in the 380s; Gerontius, the Roman commander
of troops within the city of Tomis, set upon a Gothic
garrison stationed close-by (Zosimus 4.40).9 The account
is plausible; during the 4th century, Tomis was the capital
of Scythia, and the residence of the dux Scythiae; members
of his ofcium were based there, as well as a military
detachment (IGL 2, 5, 30).
Unfortunately, the manner of the settlement of 382
remains uncertain. The allusions to the agreement provided
by Themistius and Pacatus must be considered seriously;
they were contemporaries of the events they describe, but
they did not write history, but panegyrics of Theodosius;
they were not impartial: eulogies of the emperor would
inevitably downplay any unacceptable aspects of the
agreement and exaggerate whatever interpretation might
meet with imperial approval. The current view holds that
the terms were uniquely favourable to the Goths, that they
retained a large degree of social and military freedom and
that they were only obliged to supply troops en masse,
under their own chieftains, when required to do so, and only

6. Cons. Const. 382 (CM I, 243).


7. Synesius, De Regno 21; Pacatus 22.3; Themistius, Or.16 and 34.
8. Jordanes knew the region well and is here likely to be correct, even if
his understanding of the nature of the settlement is much less reliable:
see below, pp. 64-65.
9. From the account it is clear that these Goths were serving the Empire
and were paid as regular soldiers, if somewhat better, it seems, that
the Roman garrison within the city.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

65

Fig. 1 Map of the Lower Danube region ( A. Poulter).

IRUVSHFLFFDPSDLJQVLQHIIHFWSUHVHUYLQJDVHPLDXWRQomous identity.10 However, whether or not Theodosius


had been forced to make radical concessions because his
military position was so weak remains uncertain.11

10. For example: Heather, Goths and Romans, pp. 161, 164, 165, 166,
177, 183; R. M. Errington, Roman Imperial Policy from Julian to
Theodosius,Chapel Hill, 2006, pp. 64-68; H. Leppin. Theodosius
der Grosse, Darmstadt, 2003, pp. 46-54; H. Wolfram, History of the
Goths, Los Angeles / London, 1987, p. 133.
11. Shortly before, the Goths were in Illyricum but had been forced back
into Thrace by the emperor Gratian. There is every reason to suspect
WKDWWKHGHYDVWDWLRQWKH\KDGDOUHDG\LQLFWHGRQWKHULYHULQHSURYLQFHV
ZRXOGKDYHPDGHLWGLIFXOWWROLYHRIIWKHODQGVLPLODUSUREOHPVRI
supply were to force the Goths in Gaul to agree terms, more favourable
to the Roman state than would otherwise have been the case;
Liebeschuetz, Barbarians and Bishops, p. 73. Blocking the passes over

Jordanes (Getica 27.141-29.146) provides what


appears to be the clearest description of the Goths military
role; he calls the Goths foederati, serving alongside the
Roman army and receiving payment for their participation. Unfortunately, this statement is suspect; writing
150 years after the event, his belief that the treaty of 382
granted similar terms to those held by Gothic warlords in
the second half of the 5th century is most probably anachronistic. Certainly, no other source uses the term foederati

the Haemus to prevent the Goths from reaching fresh food supplies in
Thrace was a tactic already applied in the early stages of the Gothic
uprising (Amm Marcell., XXXI.8.1). Despite the military weakness of
the Roman army, the Goths themselves may well have been anxious to
reach an accommodation with Theodosius.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

66

AnTard, 21, 2013

ANDREW POULTER

LQFRQQHFWLRQZLWKWKHDJUHHPHQW7KHUVWSRVLWLYHDWWHVtation of the term, in a technical sense, dates to AD 422


when the Gothic commander, Ariobindus, was comes
foederatorum in Thrace, although the nature and origin of
WKLVRIFHDUHREVFXUH12
The military arrangement which formed part of the
agreement is the key to understanding the terms of the
treaty and whether it favoured the Empire more or less
than it did the Goths. As has been pointed out by Heather,
the details of the treaty can only be surmised in the
light of subsequent events, but the view that the Goths
were obliged to support the Empire only when required
to do so, and that they fought as an independent force,
commanded by their own leaders, 13LVGLIFXOWWRDFFHSW
What we can glean from the sources does not fully
justify this claim. In 388, Theodosius marched West to
confront the usurper Maximus. Pacatus (32.4) observes
that Goths, the former enemies of Rome, marched under
5RPDQEDQQHUVFRPPDQGHGE\5RPDQRIFHUV$WIDFH
value, this seems to mean that the Goths were enrolled
in the army and that they were not an independent force,
but included recruits from Thrace and perhaps other
volunteers picked up on the march through Illyricum.14
That this did happen is supported by another reference
3DFDWXV   ZKLFK GHVFULEHV KRZ *RWKV RFNHG WR
MRLQ WKH HPSHURU DWWUDFWHG E\ 7KHRGRVLXV NLQGQHVV
(for kindness read pay and booty). There is no reason,
based upon the meagre sources available, to accept that
the Goths fought as a separate entity, or that they fought
under the command of their own leaders.
The next attempt at usurpation is a little different.
For the battle of the Frigidus against Eugenius in 394,
Theodosius employed a substantial number of Goths; the
JXUHRI*RWKLFGHDG 2URVLXV WKRXJKQR
doubt exaggerated for effect, at least suggests that Gothic
IRUFHV VXIIHUHG D VLJQLFDQW QXPEHU RI FDVXDOWLHV They
PXVW KDYH SOD\HG D FHQWUDO UROH DV FDQQRQ IRGGHU D
traditional Roman tactic, using auxiliary forces to bear the
EUXQWRIWKHJKWLQJLQVWHDGRIVDFULFLQJ5RPDQYHWHUDQV
LQWKHHOGDUP\7UXH*RWKVVXFKDV*DLQDVDQG$ODULF
were in charge of Gothic contingents (Zosimus IV.58) but
there is no reason to believe that they were chieftains,
in charge of their own clans; Gainas surely, and quite
SUREDEO\$ODULFZHUHRIFHUVLQWKH5RPDQDUP\XQGHU
the overall command of Timasius, the Roman commanderin-chief, and were not free to act independently or on their
own initiative. Until after the elimination of Eugenius,
there is no reason to believe that the Goths had acquired

12. Heather, Goths and Romans, p. 262. A possible context for this command
is noted below, p. 75.
13. Heather, Goths and Romans, p. 160; H. Wolfram, History of the Goths,
Berkeley, 1987, pp. 133-134.
14. Liebeschuetz, Barbarians and Bishops, p. 29.

any unparalleled level of independence which marks them


out from all previous settlements of barbarians within the
Empire. The claim that few Goths entered into military
service15 is not easy to accept, or at least to presume. In
particular, the regular distribution of Roman units along
the Danube, as seen in the Notitia Dignitatum, is unlikely
WR UHHFW WKH UHDO VLWXDWLRQ DIWHU WKH FROODSVH RI PLOLWDU\
control in 378. There is a cogent argument against the
proposition that this section of the Notitia dates back to
the late 4th century.16 Consequently, the Notitia Dignitatum
cannot be used to maintain that Roman units were still
operating on the lower Danubian frontier. But the possibility exists that, apart from voluntary enlistments in the
HOGDUP\*RWKVDOVRWRRNRYHUWKHUROHRISURWHFWLQJWKH
frontier. Could not the reference by Zosimus (4.34.5) to
Goths keeping watch on the Danube and protecting the
(PSLUH EH RI ZLGHU VLJQLFDQFH LQGLFDWLQJ WKDW *RWKLF
forces served in the Roman army, replacing regular units
lost c. 378 or subsequently withdrawn to join Theodosius
GHSOHWHG HOG DUP\" $V H[SODLQHG EHORZ WKH DUFKDHRlogical evidence tends to support such a proposition.
If the activity and status of the Goths in the 380s remains
uncertain, what is clear is that all changed in 395 when
Alaric rebelled against the Eastern Empire. His demands for
JROGDQGRIFLDOVWDWXVDVDKLJKUDQNLQJRIFHUSUREDEO\
the post of magister militum SUHVDJHV WKH ELWWHU FRQLFWV
between Constantinople and Gothic leaders in the second
half of the 5th century. The army Alaric commanded was
most certainly an independent force, serving their leader and
QRWWKHLQWHUHVWVRIWKH(PSLUHXQOHVVLWSURYHGSURWDEOHWR
do so. Its commander must have learnt and adopted Roman
military tactics and the core of his army must have been
the troops which survived the battle of the Frigidus. The
success Alaric achieved marks a step change in the military
capability of Gothic forces. Even after Adrianople, the
Goths had been incapable of capturing walled cities.17 Under
$ODULFVLHJHZRUNVZHUHXVHGWRFDSWXUH,WDOLDQFLWLHVUVW
amongst which was Rome itself. The commander and his
soldiers represented a new generation which had not been
involved in the Danube crossing and subsequent events. No

15. Heather, Goths and Romans, pp. 63-64.


16. There is an unexpected regularity in the system of forts and units listed
in the Notitias description of the lower Danubian frontier, but this is
not what would be expected post-Adrianople. No Theodosian units are
recorded: this is most surprising if this section of the Notitia postdates
a reorganization of the limes which, most certainly, would have been
a high priority following the agreement of 382. The distribution of
late Roman legionary tile-stamps indicates that legionary detachments
were outposted, upstream and downstream from the primary bases of
the legions. This arrangement matches the description in the Notitia
and, since the tile-stamps must surely belong to the Diocletianic
reconstruction of the frontier, this section of the document is most
SUREDEO\ FRPSLOHG ZLWKLQ WKH UVW KDOI RI WKH th century and not
later; see A. G. Poulter, The transition to Late Antiquity, in Idem, The
Transition to Late Antiquity, pp. 32-34.
17. See below, p. 69.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

doubt many, even most, of the Gothic force had served in


the Roman army, and Alaric applied what he had learnt to
gain the upper hand in his relations with Constantinople.
This involved bleeding the provinces on the lower Danube,
especially Macedonia and Dacia. 18 The lower Danubian
provinces seem to have been largely immune from further
PROHVWDWLRQ %XW LW ZDV WKH QDO GHSDUWXUH RI $ODULF DQG
his Gothic army for Italy in 408 which paved the way for
a period of relative calm on the lower Danube or so we
must presume until the Hunnic invasions again wreaked
havoc throughout the region.

67

The column, the central monument for the new


forum of Arcadius, was under construction by AD 402.
,WV FDUYHG VFHQHV PXVW UHSUHVHQW WKH RIFLDO LPSHULDO
version of events surrounding the defeat of Gainas and
his Gothic army.19 Although only the base survives, the
drawings, made before the its demolition, and preserved
LQWKH)UHVKHOGIROGHULQWKHOLEUDU\RI7ULQLW\&ROOHJH
Cambridge, provide a remarkably detailed and apparently
accurate copy of three sides of the column.20All
columns erected in Rome, and then in Constantinople,
invariably offered a convenient opportunity for imperial
propaganda; the column of Arcadius is no exception.
Some episodes in the removal and subsequent death
of Gainas were portrayed in detail and are as close to
the truth as artistic licence would permit, especially the
failed attempt by the Goths to build boats and cross to
$VLD0LQRU(YHQVRWKHPRVWVLJQLFDQWHSLVRGHLQWKH
fall of Gainas, the massacre of Goths by the citizens of
Constantinople, is omitted from the narrative; it was an
aspect of the story which did no credit to Arcadius, the
Christian emperor.21 But it is the detail, in particular the
representation of the Goths, which is of interest here.
Especially when events depicted occurred a considerable distance from the location of such a monument,
details could be used, not necessarily to provide a true
image, but to offer a narrative which could be clearly
understood by its intended audience.22 However, in

the representation of the imperial city and the participants in the revolt of Gainas, it is improbable that the
reliefs on the column of Arcadius distort the reality,
at least in the details of appearance of Romans and
Goths. The people addressed, in this case the citizens
of Constantinople, were all too familiar with the dress
and appearance of the Goths who had been only recently
expelled from the capital. Any misrepresentation would
confuse rather than assist the onlooker, trying to follow
the story told in the reliefs. One scene is of particular
interest (J $). It shows the enemy being escorted
by clean shaven, Roman soldiers with spears or pikes
and wearing plain, no doubt linen cloaks. In contrast,
three horsemen ride by; all have beards and wear a short
chamlys, made of coarser material, surely intended to
represent sheepskins, an item of clothing commonly
worn by Goths.23 Beards were also known to have been
a Gothic fashion, as was long hair, all shown in this and
other scenes.24 Conspicuous, behind the riders, although
apparently inserted without any connection with the
primary subject of the scene, there is a cart, containing
people. Carts were used by the Goths for transporting
goods and their families and, when threatened, were used
to form a defensible lager (carago).25 It is reasonable to
assume that the inclusion of the wagon and its occupants
was to remind the citizens how the enemy relied upon a
barbarian mode of transport, emblematic of the Gothic
lifestyle; its inclusion must have been intended to ridicule
the barbarism of this non-Roman custom. The same
scene shows a woman, not riding a horse as the male
Goths behind her, but directly facing the onlooker in an
anatomically impossible position for riding the horse
VKHDSSHDUVWREHVLWWLQJRQ(YLGHQWO\WKHVLJQLFDQFH
RIWKLVIHPDOHJXUHPXVWKDYHEHHQXQGHUVWRRGE\WKH
onlooker. It seems that presenting her face on was done
to emphasize the fact that this person was female. One
possible explanation could be that this was not a human
JXUHDWDOO'XULQJWKHSHUVHFXWLRQRIWKH&KULVWLDQ*RWKV
in 369-372, Athaneric had ordered that a pagan image
should be sent to each village in a cart and that all were
UHTXLUHG WR PDNH VDFULFHV WR LW WKRVH ZKR GLVREH\HG
(Christians) would be burnt alive (Sozomen VI.37).

18. Heather, Goths and Romans, pp.199-224.


19. Liebeschuetz, Barbarians and Bishops, pp. 121, 273.
20. For a full discussion RI WKH )UHVKHOG GUDZLQJV DQG WKH RWKHU OHVV
accurate copies, see Liebeschuetz, Barbarians and Bishops, appendix II,
pp. 273-278.
21. The burning of a church in which the Goths had sought refuge was not
an act which did much for the emperors public image; Liebeschuetz,
Barbarians and Bishops, pp. 121-122.
22. Trajans column shows legionaries in lorica segmentata and
auxiliaries in scale armour or chainmail. For the citizens of Rome, this
presented the image of Roman power, assisted by faithful auxiliary
units. However, the monument of Adamklissi on the lower Danube,
closer to the scene of battle, provides a different image where

legionaries and auxiliaries both wear chainmail, which is likely to


mean that, at least on the lower Danube, lorica segmentata was not, at
the time, worn by legionary troops. For the column in Rome, a handy
convention was used to distinguish legionaries from auxiliaries.Very
few of the onlookers in Rome would have known better.
23. Liebeschuetz, Barbarians and Bishops, p. 274.
24. J.H.W.G. Liebeschuetz, Habitus barbarus: did barbarians look
different from Romans?, in P. Porena, Y. Rivivre (dir.), Expropriations
et conscations dans les royaumes barbares, Rome, 2012, pp. 13-28.
(forthcoming).
25. U. Wange, Die Gotenkriege des Valens, Studien zu Topographie und
Chronologie im unteren Donauraum von 366 bis 378 n. Chr., Frankfurt
am Main, 1990, pp. 152-157.

Finding Goths on the Column of Arcadius

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

68

AnTard, 21, 2013

ANDREW POULTER

Fig. 2 Scenes on the


Column of Arcadius (Trinity
College, Cambridge).

How many of the Goths in Constantinople were still


pagan is an impossible question but it seems quite likely
that many still worshiped their traditional gods, like
Athaneric himself who had died in Constantinople in
381, less than twenty years before the revolt of Gainas.
3HUKDSV WKH IHPDOH JXUH ZDV WKH HIJ\ RI D SDJDQ
JRGGHVVWLHGWREXWQRWULGLQJDKRUVH7KHVLJQLFDQFH
of this part of the scene, if correctly understood, would
be that the depiction of Goths as being pagans, like the
appearance of the cart, was used to remind the citizens
of Constantinople how barbaric practices distinguished
them from Romans. That this was a conscious attempt
to ridicule the Goths would not be out of keeping with
their portrayal as barbaric people, justly defeated by the
emperor, supported by divine providence. However,
none of these attributes of the Goths can be found in
the archaeological record, nor is the knowledge that they
used a clearly Germanic language any help; dead bones
do not speak.
However, further up the column, but still in clear view for
anyone looking up from ground level, there are scenes which
depict, in a rural landscape, triangular structures occupied
E\PDOHDQGIHPDOHJXUHV J%). It has been suggested

that these structures were tents.26 In a late 5th century illustration (Ilias Ambrosiana Miniatura XXXVII) a military
tent is shown with vertical not slanting sides, and so
must have been supported by a timber framework, allowing
soldiers to stand up inside while also providing room for
more than a couple of occupants: a similar design to the
tents used by the Roman imperial army.27 At least two of
those reclining within these structures have beards which,
as with the procession out from Constantinople, was surely
LQWHQGHGWRLGHQWLI\WKHPDV*RWKV7KHWULDQJXODUSUROH
indicates that the roof came down to the ground, a wigwam
VKDSHG VWUXFWXUH ZKLFK FORVHO\ UHVHPEOHV WKH SUROH QRW
of a Roman tent, but of a grubenhaus. Since the inclusion
and prominence of these structures was considered so

26. Liebeschuetz, Barbarians and Bishops, p. 176.


27. For what it is worth, all the tents depicted on Trajans Column have
vertical side-panels and would therefore not have had a triangular
SUROH$WHQWIURP9LQGRODQGDDOVRKDGVLGHSDQQHOVDOORZLQJVROGLHUV
to stand within it, and large enough, just about, to accommodate 8 men
(a contuburnium); C. Van Driel-Murray, A Roman tent: Vindolanda
Tent 1LQ9$0D[HOG0-'REVRQ GLU Roman Frontier Studies
1989, Exeter, 1991, pp. 367-375.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

important, it seems reasonable to conjecture that, like the


FDUWVDQGSRVVLEO\WKHLPDJHRIDJRGGHVVWKHLQFOXVLRQ
of these structures was intended to ridicule the use of this
primitive form of housing which has not been found in the
region before at least the very end of the 4th century. But,
if Grubenhuser were associated with the Goths, then at
least some of them must have been observed using such
structures. At this point, it appears to introduce a possible
way of identifying the presence of Goths and, as will be
seen below, in the case of the Roman fort of Dobri Dyal, the
DSSHDUDQFHRIWKHVHVXQNHQEXLOGLQJVPLJKWDWUVWJODQFH
be one way of proving the existence of a non-Roman,
perhaps Gothic settlement.
Can a material culture
identify the presence of Goths?

69

from it. House types include timber rectangular buildings


and Grubenhuser. Cemeteries include both inhumations
DQGFUHPDWLRQVDQGDPRQJVWVPDOOQGVWKHRQO\FRPPRQ
denominator is the presence of purely Roman objects which
were evidently exported in large quantities to the inhabitants North of the Danube. Presumably, Goths in the 4th and
5th centuries did use what was commonly available but so
did native Getae.31 At Nicopolis, from the middle of the 4th
century, there existed a substantial number of mudbrick
houses, very different from the mortar built houses within the
FLW\6LQFHLQWKH*RWKVRI8OODKDGEHHQVHWWOHG
between Nicopolis and the Haemus; it would be tempting
to believe that this relatively poor extra-settlement was
RFFXSLHGE\*RWKV+RZHYHUQRWDQ\RIWKHQGVDUHRWKHU
than typically Roman and provide no grounds for assuming
WKDW WKH RFFXSDQWV FRXOG EH LGHQWLHG E\ WKHLU PDWHULDO
culture. Although too large a question to be dealt with in
this paper, there is equally no proven association between
the rich hoards found to the North of the Danube and the
Goths. All the techniques of production were well-known
in the Mediterranean world by the 3rd century and, for
example, the so-called Germanic symbolism of the eagle
motif is well attested within the Late Roman Empire. These
treasures must have been owned by barbarian chieftains.
The acquisition of these items and regularly associated
purely Roman objects are most likely to represent gifts,
manufactured, not by German settlers, but by craftsmen
in the imperial workshops of Constantinople. It cannot be
maintained that the recipients were Germanic in origin. The
owner may have been a Goth but, equally, could have been
any powerful prince who had accepted alliance with Rome
and was consequently favored with precious gifts.

It has been widely assumed that the Goths can be


LGHQWLHGZLWKPDWHULDOQGVLQWKHDUFKDHRORJLFDOUHFRUG
Unfortunately, pottery is not one of them. During the
2nd-4th centuries, the regional market was dominated by
oxidized red wares but, commencing around the middle of
the 4th century (at least at Nicopolis), new black ware forms
ZHUHLQWURGXFHGQRWDEO\WKHQHZDUHNQRZQDVIRHGHUDWL
ZDUHZKLFKHPSOR\HG5RPDQIRUPVEXWVRPHQHZYHVVHO
types, not previously attested in the region.28 By the 5th
century, black wares dominated and red ware production
FHDVHG H[FHSW IRU WKH UDUH DSSHDUDQFH RI LPSRUWHG QH
wares. But, this change in fashion need not have any implications for the ethnicity of the population this type of
pottery is so widespread throughout Thrace that it must
have been used by all the inhabitants, Thracians, Getae
DV ZHOO DV QHZFRPHUV 1RU GR VPDOOQGV KHOS LQ GLVWLQguishing between locals and new immigrants. Even the
wide-spread use of composite bone combes, so typical of
the period, have no proven connection with Goths. Romans
had hair, though perhaps generally not so long, but which
also required grooming.29 Another misguided presumption
LV WKDW WKH 6vWQWDQDGH0XUH  &KHUQLDNKRY FXOWXUH
was developed by the Goths and that, wherever it occurs,
both North and South of the Danube, it must attest their
presence.30 The material evidence is not heterogeneous, far

28. R. K. Falkner, The pottery, in A. G. Poulter (dir.), Nicopolis ad Istrum:


the Pottery, cit. (n. 4), p. 734; Id., Vagalinski Burnished Pottery from
the First Century to the Beginning of the 7th Century AD from the
Region South of the Lower Danube6RD>%XOJDULDQ@
29. Settled close to the city of Nicopolis c. 347/348. Goths may well have
OLYHGLQDQH[WUDPXUDOTXDUWHULGHQWLHGGXULQJH[FDYDWLRQVEXWDOOWKH
PHWDOQGVDUH5RPDQDQGGRQRWVKRZDQ\EDUWEDULDQWUDLWV$3RXOWHU
Nicopolis ad Istrum: The Finds, cit. (n. 4), pp. 15-64; Id., Invisible Goths
within and beyond the Roman Empire, in J. Drinkwater, B. Salway (dir.),
Wolf Liebeschuetz Reected, London, 2007, pp. 169-183.
30. Cf. P. Heather, The Goths, Oxford, 1996, pp. 14-25. H. Wolfram,
History of the Goths, Berkeley, 1987, pp. 61, 73, 419.

Forts, towns and the countryside


LQWKHODVWTXDUWHURIWKHth century

The most intensively researched of the frontier forts is


Iatrus.32$OWKRXJKLWVIRUWLFDWLRQVDUHPDVVLYHO\GHIHQVLYH
the internal layout of the fort contained a range of structures
not dissimilar to those of the Early Empire. It had a standard
principia, probably a praetorium, and barracks, arranged
either side of the main colonnaded road from the eastern
gate, as well as other barracks at the western end of the
defences, behind the principia.33 The fort shows no sign of

31. A. Poulter, Invisible Goths, cit. (n. 29), pp. 171-174.


32. L. Bartosiewicz et al., Iatrus-Krivina V. Sptantike Befestigung und
frhmittelalterliche Siedlung an der unteren Donau (Studien zur
Geschichte des Kastells Iatrus, Forschungsstand, 5), Berlin, 1989, and
G. von Blow, The fort of Iatrus in Moesia Secunda: observations on the
Late Roman defensive system on the lower Danube (fourth-sixth centuries
AD), in Poulter, The Transition to Late Antiquity, pp. 459-478.
33. More barracks may well have existed on the northern side of the
fort, but here erosion of the Danube bank has destroyed most of the

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

70

AnTard, 21, 2013

ANDREW POULTER

GHVWUXFWLRQ E\ UH EHIRUH WKH PLGGOH RI WKH th century.34
1HLWKHU KDYH DQ\ WUDFHV RI GHVWUXFWLRQ EHHQ LGHQWLHG
within the legionary fortress of Novae (Svishtov) in the
late 4th or early 5th century.35 But, outside the western gate,
built over a peristyle building (which had burnt down in
the 3rdFHQWXU\ DQHZODUJHVWUXFWXUHZDVGHVWUR\HGE\UH
at the end of the 4th century.36 Some additional evidence
comes from northern Thrace, south of the Danubian
frontier. A quadriburgium at Koula, with one of its corner
towers still standing 16.30 m high, must have been built in
early 4th century. Here, a reinterpretation of the stratigraphy
points, not to three destruction levels as the excavator
DVVXPHGEXWWRRQO\RQHWKDWFDQEHUPO\GDWHGWRWKH
middle of the 5th century and not earlier.37
Somewhat more substantive is the information available
IRUXUEDQVXUYLYDO7KHFLW\RI3KLOLSSRSROLVZDVIRUWLHG
by AD 172 and the walls remained in use during the 4th
and into the 5th FHQWXU\ZLWKRQO\DVOLJKWPRGLFDWLRQWR
the curtain around the main gate.38 There was no reduction
LQWKHIRUWLHGDUHDDQGQRHYLGHQFHIRUDGHVWUXFWLRQOHYHO
datable to the end of the 4th century.39 Large, luxurious
houses were built in the late 3rd and continue in use until the
6th century.40 True, during the revolt of Procopius against
Valens, the city was besieged by the forces of Valentinian,

northern wall and with it a substantial portion of the interior.


34. G. von Blow, Die Entwicklung des Siedlungsbildes von Iatrus in der
Periode B/C, L. Bartosiewicz et al., Iatrus-Krivina V, cit. (n. 32), pp. 2930; Id., The fort of Iatrus, cit. (n. 32), pp. 466-467.
35. T. Sarnowski, Die principia von Novae im spten 4. und frhen 5. Jh.,
in S. Konrad, L. Vagalinski (dir.), Der limes an der unteren Donau von
Diokletian bis Heraklios (Vortrger der internationalen Konferenz
Svishtov, 1.-5. September 1998)6RDSS
36. K. Dimitrov, Kum vuprosa za rasvitieto na kompleks extra muros
(sector VIII A) na Nove v Dolna i vtora Miziya, in Phosphorion, Studia
in honorem Maria Chichikova6RDSS>%XOJDULDQ@
37. Three destruction levels are claimed, one under Valens, another in the
early 4thFHQWXU\WKHQDQDOGHVWUXFWLRQDQGDEDQGRQPHQWLQWKHPLG
5th century. Two sections have been published. In the western sector, there
ZDVDSULPDU\RRUOHYHOZKLFKZDVEXUQWWKHQFRYHUHGE\WZROHYHOV
of burnt debris in the report which cannot be interpreted as occupation
GHSRVLWVWKHWRSVRIERWKGXPSVDUHLUUHJXODULQSUROHDQGLWLVDGPLWWHG
LQ WKH UHSRUW WKDW WKHUH ZHUH QR RWKHU RRU VXUIDFHV$WDQDVRYD et al.,
Razkopki i Mpoouchvaniya 33, Kastra Martis, Kvadribourgii i Kastel
%XOJDULDQ SDQGJ7KHPDWHULDORYHUWKHSULPDU\RRU
FDQ EH FRQGHQWO\ LGHQWLHG DV SDUW RI WKH VDPH UHGHSRVLWHG GXPS
of destruction material. This interpretation is supported by the second
SXEOLVKHG SUROH IURP ZLWKLQ WKH QRUWKHDVWHUQ WRZHU ibidem, p. 39,
DQGJ$JDLQRQO\RQHFRPSDFWHGVXUIDFHZDVIRXQGDERYHZKLFK
there was a single deposit of soil and burnt debris. In neither case was the
rubble removed and the latest coin was an issue of Theodosius II. There
is no reason to believe that there was more than one destruction level and
that it dated to about the middle of the 5th century.
38. V. Gerasimova, Zwei frhchristliche Stifterinschriften aus der kleinen
Basilica in Plovdiv (Philippopolis), in Archaeologia Bulgaria, 2, 2002,
p. 78-79; E. Kesyakova, Philippopolis prez rimskata epoha6RD
pp. 23-24 [Bulgarian].
39. L. Botusharova, E. Kesyakova, Sur la topographie de la ville de
Philippopolis lepoque de la basse antiquit, in Pulpidava: Semaines
philippopitaines de lhistoire et de la culture Thrace, 4, 1980, pp. 265270; E. Kesyakova, Philippopolis, cit. (n. 38), pp. 81, 93.
40. E. Kesyakova, Philippopolis, cit. (n. 38), pp. 18-19.

but was never captured (Amm. Marcell. XXVI. 10.4). The


walls of Augusta Traiana, another large town, enclosed
KDE\WKHWLPHLWZDVIRUWLHGXQGHU0DUFXV$XUHOLXV
A late 4th century destruction is suggested.41 However, this
is by no means certain; large town houses with notably
elaborate mosaics, built in the early 4th century, seem to have
continued in use without interruption into the 5th century.42
A large extramural house survived the 4th century and was
later converted for use as a church.43 Serdica had originally
FRYHUHG  KD ZLWKLQ LWV IRUWLFDWLRQV DQ DUHD VXEVWDQtially increased to 84 ha in the late Roman period, probably
during the 4th century, and seems to have continued into the
5th century. Nor is there any sign that Pautalia, in western
Thrace, suffered destruction at the end of the 4th or early
5th century.44 Nicopolis ad Istrum continued to be defended
by its late 2nd-century defences and there is no reason to
believe that the Goths were able to capture and destroy
the city.45 The agora remained in use and presumably its
civic buildings.46 There was no sign of a decline in the
range of agricultural products reaching the city in the late
4th or early 5th century. The only indication that these were
XQVHWWOHGWLPHVLVWKHUHGXFWLRQDQGQDOGLVDSSHDUDQFHRI
the extramural settlement. Marcianopolis is a case apart.
Its size, c. 70 ha, makes it one of the largest cities in the
Danube plain. One impressive house of 4th century date,
decorated with mosaics, has been excavated in the centre
of the town although this appearance of prosperity may
ZHOOUHHFWQRWORFDOZHDOWKEXWWKHFLW\VIXQFWLRQDVWKH
capital of Moesia II and, more particularly, its role as a
military base, and imperial residence during Valens Gothic
wars (367-369). That it had an arms factory is not at all
surprising.47
Whether, as might be expected, many forts along the
frontier were evacuated during the Gothic revolt remains
unknown. But, certainly cities remained largely protected
behind their defences, even when their walls were, for
the time, simply the old-fashioned circuits built in the late
2nd century. Ammianus Marcellinus accredited Fritigern
ZLWKWKHYLHZWKDWWKH*RWKVNHSWSHDFHZLWKZDOOV $PP
Marcell. XXX1.6,4). This seems to have been generally true.
The fate of the countryside is more problematic. Some
villas survived at least until the very end of the 4th century.48

41. Kr. Kalchev, Arheologichesiyat rezervat Augusta Traiana Beroe,


prouchvaniya I problemi, in Sbornik, 1998, pp. 9, 92.
42. Ibidem, pp. 53-55.
43. Ibid., p. 53.
44. V. Katsarova, Pautalia i neinata territoriya prez I-VI vek, in Veliko
Turnovo, 2005, p. 146 [Bulgarian].
45. A. Poulter, Nicopolis ad Istrum, cit. (n. 4), pp. 33-34.
46. P. Vladkova, The late Roman agora, in Poulter, The transition to Late
Antiquity, p. 210.
47. See Not Dig. Or XI. 34, and G0LKDLORYHSLJUDFDQRLQIzestia
Burgas, 11, 1965, pp. 150-153.
48. See below, for the villas which do not have been directly affected by
the events of 378, pp. 72-73.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

The ornate villa of Ivailograd, not surprisingly, since it


lay within the territory of Adrianople, would seem to
have been destroyed and abandoned about the time of the
battle.49 That villas did not survive in the years immediately
following Adrianople might be expected, but surprisingly
many of those which have been excavated do seem to have
continued to exist, even a decade after the event.50
Continuity and change
RQWKHIURQWLHUF
Following the departure of Alaric and his army for Italy
in 407, it seems that military control was re-established
over the Danubian frontier. The provision of local
VXSSOLHVIRUWKHDUP\DQGWKHFRQVWUXFWLRQRIIRUWLFDWLRQV
was underway in 408 (CTh, 11,17,4). In 412, orders were
issued for the repair of military river craft in Scythia and
Moesia II (CTh, 7, 17, 1). In 443, measures were again
WDNHQWRPDLQWDLQWKHHIIHFWLYHQHVVRIWKH'DQXELDQHHW
in Thrace and Illyricum (NTh, 24.5)
On the ground, excavations support the evidence of
the law codes. An inscription from Berkovitsa, south of
Montana, proves that imperial building took place at some
point between AD 408 and 423.51 Novae, still the base of
legio I Italica, is one of the most informative sites. Nothing
is known about military accommodation in the fortress,
apart from the 3rd century abandonment of one barrack,
excavated immediately to the South of the porta principalis
dextra. The legionary baths also appear to have no longer
functioned during the 4th century. But excavations in the
principia have demonstrated that coin-loss continued down
to the late 440s.52 Also, from the area of the headquarters
building, small statue bases were recovered. They had been
set up by primipilarii in 430, 431 and 432.53 The primipilarii were responsible for the shipping of supplies to the
legion and the inscriptions helpfully inform us that they
were involved in the provision of the annona from the
Mediterranean provinces of Hellespontus and Insulanea.
Moreover, since the ceremonial role of the principia still
existed, presumably it also continued to be the headquarters
of the legion. Even so, there is good reason to believe that

much had changed in the way that imperial authority was


maintained over the region.
The singular case of Iatrus illustrates changes which
were no doubt taking place along the length of the
ORZHU 'DQXELDQ IURQWLHU 'XULQJ WKH IRUWV UVW SHULRG RI
occupation (period A), its layout, as described above,
exhibited the essential characteristics of a fort of the early
HPSLUHH[FHSWIRUWKHGLIFXOW\RISURYLGLQJDVWDQGDUGSODQ
within a curtain wall which followed the topography and
which did not adopt the traditional playing-card form.54
The end of period A is problematic.55 The argument that
period A ended as early as c. 350/360 is insecure.56 A date
c. 400 would seem more likely.57 In the ensuing period
(period B/C) the principia still stood but was partly
dismantled and used simply for metal-working, as was
the praetorium.58 Barracks were dismantled and only
fragments of the mortar-bonded walls were incorporated
within irregularly planned houses and workshops, mostly
built of stone with earth bonding.59 Such a radical change in
the layout of the fort suggests a fundamental change in the
character of its garrison and structure of command. There
was no destruction deposit separating periods A and B/C
which might have required rebuilding on this scale.
South of the Danube, auxiliary forts continued to be
occupied, although by much smaller garrisons than the
nominally 500 strong units of the Antonine and Severan
periods. The quadriburgium at Koula survived until late
in the reign of Theodosius II.60 But it is unlikely that the
fort retained its strictly military function after c. 400.

49. Caution is required, even here, since no coin list has been published.
In favour of a connection with the battle or immediately following, see
Ya. Mladenova, Armira, krai Ivailovgrad6RDS>%XOJDULDQ@
V. Dinchev, Rimskite vili6RDSS>%XOJDULDQ@
50. See below, pp. 72-73.
51. A. Poulter, The transition to Late Antiquity on the lower Danube:
the city, a fort and the countryside, in Poulter, The Transition to Late
Antiquity, pp. 93-94.
52. T. Sarnowski, Die principia von Novae, cit. (n. 35), pp. 57-58.
53. Idem, Drei sptkaiserzeitliche Statuenbasen aus Novae in Niedermsien,
LQ 0 0LUNRYL GLU  Rmische Stdte und Festungen an der Donau,
Akten der regionalen Konferenz, Belgrade, 2005, pp. 151-152.

71

54. See above, pp. 68-69.


55. B. Dhle, Die Siedlungsperiode A in Iatrus, in L. Bartosiewicz et al.,
Iatrus-Krivina V, cit. (n. 32), p. 25.
56. This date is argued on the basis that a coin of Constantius II came from
the reconstruction level for period B; J. Hermann, Stand und Probleme
der Ausgrabungen in Krivina nach den Grabungskampagnen 19661973, in G. von Blow, D. Schieferdecker, H. Heinrich (dir.), IatrusKrivina 1. Sptantike Befestigung und frhmittelalterliche Siedlung an
der unteren Donau. Ergebnisse der Ausgrabungen 1966-1973, Berlin,
1979, pp. 11-12. G. von Bulw, Die Entwiklung, in L. Bartosiewicz et al.,
Iatrus-Krivina V, cit. (n. 32), pp. 29-30; Id., The fort of Iatrus, cit. (n. 32),
p. 466. The discovery of a coin of Constantius II in the construction
deposit (which also contained two coins, one of Honorius and another
of Arcadius or Honorius) does not support the contension that the fort
was reorganized at such an early date. For the regular use of 4th-century
coinage well on into the 5th century, see P. Guest, Coin circulation in the
Balkans in Late Antiquity, in Poulter, The Transition to Late Antiquity,
p. 299. Surely, coins of Arcadius and Honorius point to the beginning of
period B in the early 5th century.
57. See below, for the close similarity between Dichin and Iatrus, p. 74.
$W'LFKLQWKHFRLQQGVSRLQWWRDFRQVWUXFWLRQGDWHRIFGHVSLWH
the presence of some earlier 4th-century coins which continued in use,
even to the very end of the 5th century: P. Guest, Coin circulation, cit.
(n. 56), p. 299.
58. D. Stanchev, S. Conrad, Die Objekte XLI und XVIII, in G. von Blow,
B. Bttger, S. Konrad et al. (dir.), Iatrus 6:Iatrus-Krivina, Sptantike
Befestigung und frhmittelalterliche Siedlung an der unteren Donau.
Ergebnisse der Ausgrabungen, 1992-2000 (Limesforschungen, 28),
Mainz, 2007, pp. 119-125.
59. G. von Bulw, The fort of Iatrus, cit. (n. 32), p. 468.
60. See above, note 37.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

72

AnTard, 21, 2013

ANDREW POULTER

Buildings were of mudbrick and, notable amongst the


QGV ZHUH DJULFXOWXUDO WRROV LQFOXGLQJ VLFNOHV DQG
ploughshares.61
The small fort of Dichin sat on a low ridge, 10 m above
the river Rositsa and c. 11 km west of the city of Nicopolis.62
It was not on a Roman road but was located on the bank of
a navigable river which gave access to the fertile country to
the west and, downstream to the river Iantra, then north to
the Danube. Its date is remarkable: unlike the known late
Roman forts on the Danube which were mostly built in the
Tetrarchic period, Dichin was constructed c. 400. Its walls,
using brick courses alternating with stone facing blocks,
was superbly designed and had both circular corner towers
and rectangular medial towers set astride the curtain-walls;
certainly the defences must have been designed and built
by Roman military engineers. However, inside the fort,
the layout, from the start, did not conform to traditional
practice. No principia was found and, despite the lavish
use of mortar in building the walls, almost all the interior
structures used only earth-bonding with rough limestone
blocks. The superstructure was completed in mudbrick,
protected by tiled roofs, identical to the buildings excavated
at Iatrus. Large rectangular buildings were roughly aligned
along the central roadway whereas the western quarter
ZDV RFFXSLHG E\ D VHULHV RI JUDQDULHV ZLWK UDLVHG RRUV
unexpectedly supported by mudbrick bases, not ones of
mortared stone or tile. The fort survived, unusually, beyond
 WR EH QDOO\ GHVWUR\HG F $'  $ GLVPHPEHUHG
female skeleton was found in the destruction deposit.
+RZHYHUWKHPLOLWDU\IXQFWLRQRIWKHIRUWLVFRQUPHGE\
the discovery of three shield bosses, javelins, a plumbartum
DQGVFDOHDUPRXU6LJQLFDQWO\WKHUHZHUHDOVRDJULFXOWXUDO
implements including ploughs, scythes and sickles. About
22 km south-east of Nicopolis ad Istrum, the fort of Dobri
Dyal sits atop a steep-sided hill, overlooking what must
have been a main road heading east towards Marcianopolis
and the Black Sea coast. After two years of excavation, the
main characteristics of the site have been established. A
QHO\EXLOWLUUHJXODUFLUFXLWIROORZVWKHFUHVWRIWKHKLOOWKH
outer face supported by regularly spaced pilasters; even at
the corners, it did not have towers; the slope is so steep it was
no doubt neither necessary nor practical to provide them.
$JDLQWKHSOHQWLIXOFRLQQGVSURYLGHDUHDVRQDEO\SUHFLVH
dating. Like Dichin, it was built c. 400. In order to provide
DWZDONZD\VDURXQGWKHVXPPLWPDVVLYHTXDQWLWLHVRIFOD\
were dumped against the slope and held in place by walls
of stone and earth. It seems that the interior was stepped
from the top down to the defences, providing level areas
for buildings, some of which were certainly contemporary
with the construction of the fort since they bonded with

61. I. Atanasova, I. G., Kabakchieva, A. Iotsova, Castra Martis, quadriburgium


I kastel 5D]NRSNLL3URRXFKYDQL\D 6RDS>%XOJDULDQ@
62. A. Poulter, The transition, cit. (n. 16), pp. 82-94.

the retaining walls. What is remarkable is that the interior


was carefully divided into two equal sections. Upstanding
earth-bonded buildings lay on the eastern side of the fort,
but the other half had no stone structures at all.63 Instead, the
open area was occupied by Grubenhuser and it seems that
this part of the fort was used only for these structures. No
similar Grubenhuser have (so far) been recognized on the
other side of the hill where the buildings were of stone and
earth. None of the Grubenhuser contained Slav pottery,
only standard late Roman wares. It seems possible that
these very un-Roman structures were contemporary with
the usual earth-bonded structures.64 Unlike Dichin, the life
of this fort was more typical for the region; its latest coins
were issues of Theodosius II and provide a terminus post
quem of 435 for its abandonment, apparently following the
GHPROLWLRQRIEXLOGLQJVQRWDIWHUDQ\GHVWUXFWLRQE\UH
The creation of these new forts in the interior marks a
departure from the 4th century. Located at important road
junctions, they must have provided the internal security
which was now required, ranging from impressive defences,
like those illustrated above, down to much smaller police
posts such as that at Gostilitsa (0.8 ha),65 south of Nicopolis.
Survival and the militarization
RIth-century towns
Survival is easily demonstrated for the cities, especially
in the case of the wealthier ones south of the Haemus and
DWVRPHUHPRYHIURPWKHWXUPRLOZKLFKKDGDILFWHG'DFLD
Ripensis, Moesia II and Scythia. However, there were differences. Already during the 4th century, some had garrisons.
Tomis, capital of Scythia, was the residence for the dux
Scythiae and his ofcium.66 Marcianopolis under Valens,
was the capital of Moesia II and certainly had a garrison:
the comes per Thracias was based there on the eve of the
Gothic revolt (Amm. MarcellXXXI.4,9-5,9). No doubt it
remained a key military centre, within easy reach of the

63. This was established by geophysical survey, carried out by Dr Michael


Boyd, as well as by excavation.
64. Forts along the Danube regularly contain Grubenhuser but these are
usely ascribed to the 9th/10th century. The only other case on the Danube
where these structures might have been built within the Roman period
is at Intercissa on the middle Danube. Here, three Grubenhuser were
FXWLQWRWKHQDO5RPDQOHYHOZLWKLQWKHFRQQHVRIWKHODWH5RPDQ
fort. It seems possible that they were in use before the beginning
of the 5th century; they contained black burnished ware, as well as
handmade pottery, normal late Roman ceramics, and one of them
produced two coins of Valentinian: S. Soproni, Die letzten Jahrzehnte
des Pannonischen Limes (Verffentlichung der Kommission zur
archologischen Erforschung des sptrmischen Raetien, Der
bayerischen Akademie der Wissenschaften), Mnchen, 1985, pp. 4448, 63-64.
65. B. Soultov, Arheologicheski proouchvaniya v Diskondouratera ot
1958-1961 godina, in Izvestiya Veliko Turnovo, 3, 1966, pp. 28, 35,
42 [Bulgarian].
66. See above, p. 63.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

Danube, up the road to Durostorum (J). It is probable


that barracks existed along the inside of the curtain-wall at
Philippopolis.67 The massive defences at Diocletianopolis
contained what are certainly barracks along its southern
and eastern curtain-walls.68 Whether Nicopolis ad Istrum
KHOG D JDUULVRQ LV XQNQRZQ EXW SHUKDSV VLJQLFDQWO\ LW
did not acquire the new style defences in the late Roman
period.69 The agora was no longer in use by the middle
of the 5th century.70 Extra defensive precautions were
XQGHUWDNHQ GXULQJ WKH UVW TXDUWHU RI WKH th century; the
external ditch was widened and deepened and a mudbrick
revetment (proteichisma) was added to the outside edge
of the berm. Occupation continued beyond AD 430 but
ended in the violent destruction of the city, most probably
within the second quarter of the 5th century.71 Thereafter the
Roman city was abandoned, only to be replaced by a new
IRUWLFDWLRQLQWKHODWHth or early 6th century AD, but one
which was very different from the urban centre it replaced;
the 6th century defences served ecclesiastical and military
needs and contained very few if any civilians.72
The towns in southern Thrace have been extensively
excavated and provide useful information. Philippopolis
survives, not only the turmoil of the late 4th century but
exhibits a remarkable level of prosperity throughout the 5th
century. Large, luxurious houses, decorated with high quality
mosaics, were built in the late 3rd and continued in use into 6th
century.73 A synagogue was destroyed in the mid 3rd century
but was rebuilt and continued in use until the middle of the 6th
century.74 There were changes but civilian occupation would
seem to have continued; although the eastern baths were
abandoned, they were replaced by a new two-storied building
at the end of the 5th century.75 Notable is the construction of
DODUJHEDVLOLFDGHFRUDWHGZLWKQHPRVDLFVPORQJDQG
39m wide; erected in the 5th century; it also functioned as
late as the 6th century.76A large public bath suite went out of
use in the mid 5th century but was subsequently rebuilt.77 One
small basilica with mosaics, attributed to the late 5th century,

73

existed just within the defences.78$SDUWIURPWKHPRGLFDtions carried out at the east gate and a cutting through an
existing tower on the southern defences, providing a small,
secondary entrance into the city, the late second century fortiFDWLRQVUHPDLQHGLQXVH79 There are no signs of destruction
or a slackening in the pace of urban regeneration in the 4th or
5th centuries AD.80 At Augusta Traiana, it has been suggested
that the town was destroyed at the end of the 4th century
but the evidence seems based on historical grounds rather
than upon archaeological evidence.81 The impressive circuit,
enclosing 48.5ha, is not reduced in size. However, not all
buildings remained in good repair. Some large town houses
with mosaics of the highest standard, possibly laid, not by
local craftsmen, but by Constantinopolitan specialists in the
early 4th century, would seem not to have continued in use
beyond the 5th century but were subdivided into smaller units
with additional walls of mudbrick.82 Further signs of trouble
around the middle of the 5th century include the destruction
level which ended occupation of an extramural building.
But, this was perhaps exceptional or accidental. Another
large town house survived and was, at a late date, turned
into a cemetery church and was surrounded by inhumation
burials.83 Pautalia may well have been burnt down towards
the middle of the 5th century but occupation had continued
LQWRWKHUVWKDOIRIWKHth, even though mud brick replaced
mortar in new buildings and the forum seems no longer to
be in use.84

67. L. Botusharova, E. Kesyakova, Sur la topgraphie, cit. (n. 39), pp. 265-270.
68. K. Madzharov, Diocletianopolis6RDSS>%XOJDULDQ@
69. There are strong indications that a military garrison was billeted
in the city during the 3rd century, but Nicopolis, as far as its archaic
defences suggest, remained of no particular consequence in the late
Roman period; A. Poulter, Nicopolis ad Istrum, cit. (n. 4), pp. 28-29.
70. P. Vladkova, The late Roman agora, cit. (n. 46), p. 210.
71. A. Poulter, Nicopolis ad Istrum, cit. (n. 4), pp. 34-35. The fact that
not only the interior of the city but also the proteichisma were burnt
and destroyed, could hardly have happened accidentally. Weapons and
armour in the bottom of the newly-cut ditch and on the cobbled roadway
outside the southern gate support the view that the destruction followed
a hostile attack.
72. Ibidem, pp. 34-47.
73. E. Kesyakova, Philippopolis, cit. (n. 38), pp. 18-19.
74. Ibidem, p. 81.
75. Ibid., p. 19.
76. Ibid., pp. 20-21.
77. Ibid., p. 43.

The destruction of the Roman rural economy

The pattern of coin-loss provides a generally reliable


LQGLFDWLRQDVWRZKHQYLOODVZHUHQDOO\DEDQGRQHG1RUWKRI
the Haemus range, coins from the villa of Gorna Oryakovitsa
end with issues of 392-94.85 There is less certainty about the
three large villas around Montana, although the published
evidence strongly suggests that their use terminated at the
very end of the 4th century.860DQ\ZHUHGHVWUR\HGE\UH
and the variety of metalwork left behind suggests that they
were abandoned in haste, as in the case of Ourovene near
Vratsa.877KHWHUPLQDOGDWHIRUWKHIWHHQYLOODVLWHVVXUYH\HG
within the territory of Nicopolis cannot be securely
established without excavation. However, none produced

78. V. Gerasimova,. Zwei frhchristliche Stifterinschriften aus der kleinen


Basilica in Plovdiv (Philippopolis), in Arch. Bulg., 2, 2002, pp. 78-79.
79. E. Kesyakova, Philippopolis, cit. (n. 38), p. 23.
80. L. Botusharova, E. Kesyakova, Sur la topographie, cit. (n. 39), pp. 265-270.
81. See above, p. 69.
82. Kr. Kalchev, Arheologichesiyat rezervat, cit. (n. 41), pp. 53-55.
83. Ibidem, p. 53.
84. V. Katsarova, Pautalia I, cit. (n. 44), pp. 146-147.
85. This simple, small villa near Pernik produced a terminal issue of
392/395: V. Dinchev, Rimskite vili, cit. (n. 49), pp. 28-30.
86. Montana 1, Montana 2 and Montana 3: ibidem, pp. 32-37.
87. Ibidem, pp. 41-42.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

74

AnTard, 21, 2013

ANDREW POULTER

FHUDPLFVGDWDEOHDIWHUF&RLQQGVDUHLQYDULDEO\th
century in date. In each case, their association with burnt
PXGEULFNSURYHVWKDWWKHYLOODVKDGEHHQGHVWUR\HGE\UH
and several produced complete agricultural implements; it
would be surprising if they had been simply abandoned.88
In southern Thrace, the villa of Chatalka existed down to
at least the reign of Honorus (395-423).89 To the west, the villa
of Kralev Dol also survived into the 390s.90 Even though the
occupation of these sites as functioning villas seems to end in
the last decade of the 4th century, many of the sites acquired
simple mudbrick structures which were used for some time
after the villas themselves had ceased to exist.91
What is clear is that, as in the systematic study of villas
within the territory of Nicopolis, the villa economy along the
Danube and in Thrace came to an abrupt, perhaps violent end.
0RUHVLJQLFDQWO\QRDWWHPSWZDVHYHU\PDGHWRUHEXLOGWKH
ruined estate centres. The Roman villa system and, with it,
the traditional basis of the rural economy, appears to have
ceased c. AD 400. However, as described below, this does
not mean that agricultural production did not continue during
the 5th century. But the evidence that this was so comes not
from the villas but from the forts and, exceptionally, from
one city in the north Bulgarian plain.
The military and their supplies
Agricultural production and stock rearing between 400
and 450 continued around the city of Nicopolis.92 Although
WKLVLVWKHRQO\XUEDQFHQWUHZKLFKKDVSURGXFHGDVWUDWLHG
TXDQWLHG DVVHPEODJH LW LV UHDVRQDEOH WR EHOLHYH WKDW
where cities existed (especially in Thrace), the farming of
ODQGLQWKHYLFLQLW\RIWKHIRUWLFDWLRQVPXVWKDYHEHHQVWLOO
taking place to sustain the urban population. It is unproven
but probable that it contained a garrison, as argued above
was the case in the less-exposed cities of southern Thrace.

88. A. Poulter, The transition, cit. (n. 16), p. 82; Id., Site-specic survey:
the methodology, in Id., The Transition to Late Antiquity, pp. 583-595.
89. V. Dinchev, Rimskite vili, cit. (n. 49); D. Nikolov, The Thraco-Roman
Villa Rustica Near Chatalka, Stara Zagora, Bulgaria (BAR, Sup.
Series 17), Oxford, 1976, p. 45.
90. This simple villa near Pernik produced a terminal issue of 392/395;
V. Dinchev, Rimskite vili, cit. (n. 49), pp. 28-30.
3RVWYLOODXVHRIVLWHVKDVEHHQUHJXODUO\LGHQWLHGLQFOXGLQJ2XURYHQH
Vrtatsa: V. Dinchev, Rimskite vili, cit. (n. 49), pp. 41-42; Veselchani,
Kurdzhali (southern Thrace): ibidem, pp. 54-58, the three villas around
Montana: ibidem, pp. 32-37, and more generally: ibidem, p. 25. Whether
the new residents living in and around the abandoned villas were local
inhabitants or newcomers remains unknown. Since the villas were all
generally insubstantial structures, the reoccupation of the site would
probably have taken place immediately or soon after the villas were
abandoned by their occupants. No coins have been associated with this
period of occupation; presumably they were not used for a long time:
notably, none have produced 6th century coins.
92. See in A. Poulter, Nicopolis ad Istrum: The Finds, cit. (n. 4): M. Beech,
pp. 154-197 (large mammal and reptile report), Z. Boev, M. Beech,
pp. 242-253 (bird bones), M. Buysse, pp. 15-292 (botanical remains).

In the forts, the evidence for continued agriculture is


compelling. At Dichin, fully processed crops included
wheat, barley, millet and rye as well as peas, beans, lentil
and bitter vetch.93 A wide range of meats was consumed;
wild species were represented by pheasant, partridge,
black grouse and pigeon, boar and deer. Domestic animals
were also available: pig, cattle, sheep/goat and, above
all, chickens.94 Skeletal representation analysis implies
that some animals, notably sheep/goat and pigs, were
butchered on site and probably kept close to the fort. The
cattle elements were dominated by upper limb bones,
indicating that prepared carcasses or joints of meat were
SUREDEO\EURXJKWWRWKHVLWHIURPIXUWKHUDHOG+RZHYHU
the discovery of cow-bells suggests that cattle or perhaps
oxen were also kept close by. Unlike the city of Nicopolis,
the occupants showed a marked preference for pike.95 Dobri
Dyal would seem to have been equally well-supplied with
local produce, including limited quantities of wild species
(rabbits and hares, wild birds and boar) but primarily
the diet consisted of domestic animals: pig dominated
the assemblage, followed by sheep/goat and then cattle.
Skeletal representation again suggested that many of the
larger domestic animals were supplied to the fort as joints
of meat, even though some primary butchery did take place
on site.96 Seeds included lentils, barley, wheat and beans
and cover much the same range of produce as at Dichin.97
%RWK VLWHV FRQUP WKDW WKH ODQG ZDV EHLQJ LQWHQVLYHO\
IDUPHGGXULQJWKHUVWKDOIRIWKHth century.
This is not to say that the occupants of Dichin and
'REUL '\DO ZHUH WRWDOO\ VHOIVXIFLHQW $W 'LFKLQ
3RQWLFW\SH DPSKRUDH 2SDL %  IURP WKH &ULPHD DQG
Sinope were present as well as LRAI amphorae, probably
from Cilicia or possibly North Syria, LRA2 from the
Argolid, East Aegean types and Tunisian as well as local
products, copies of LR1/2. Dobri Dyal has also produced
imitations of LRA1, LRA2 from the Argolid, other
amphorae from Sinope and types typical of the Samos
JURXS7KHUHLVDKLJKLQFLGHQFHRIORFDOEODFNQHZDUH
IRHGHUDWL ZDUH  DQG W\SLFDO th century coarse wares,
including locally produced transport amphorae. Notable
DOVR LV WKH SUHVHQFH RI VRPH KLJK TXDOLW\ LPSRUWHG QH
ware: Late Roman C (red slip) and Pontic Red slip.98
$OWKRXJK WKH SUHVHQFH RI ORFDO DPSKRUDH FRQUPV WKDW
UHJLRQDO SURGXFWLRQ UHPDLQHG LPSRUWDQW GXULQJ WKH UVW
half of the 5th century, the importation of amphorae from

93. P. Grinter, Seeds of destruction: conagration in the grain stores of


Dichin, in Poulter, The Transition to Late Antiquity, pp. 283-284.
94. C. Johnstone, A short report on the preliminary results from the study
of the mammal and bird bones assemblages from Dichin, in Poulter,
The Transition to Late Antiquity, pp. 288-292.
95. R. Parks, The sh bones (report in preparation).
96. Preliminary reports by R. Billson and F. Bowen (unpublished).
97. Preliminary report by P. Grinter (unpublished).
98. Information from J. Timby and P. Reynolds.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

AnTard, 2 1 , 2 0 1 3

GOTHS ON THE LOWER DANUBE: THEIR IMPACT UPON AND BEHIND THE FRONTIER

WKH0HGLWHUUDQHDQDORQJZLWKQHZDUHVPXVWPHDQWKDW
luxury items (oil certainly, and probably wine) were being
supplied for the military in the form of the annona.
Goths in residence?
On the lower Danube, the abandonment of standard
military buildings (such as the principia and praetorium at
Iatrus) suggests a rapid and complete change in the system
of military organization. Since this was apparently not
preceded by any destruction, the change did not appear to
have been violent but part of a systematic reorganization.
As to the forts build c. 400 at Dobri Dyal and Dichin,
their massive defences were well-built and could only
have been erected by Roman military engineers. What is
remarkable is that, from the start, the interior buildings
conformed to no traditional military layout. No principia
was ever built at Dichin. Its rows of houses and storerooms were almost exclusively built with mudbrick and
earth-bonded stone walls. Mortar was used but sparingly:
a striking contrast with the well-mortared defences. There
seems to have been a disarticulation between the Roman
provision of building materials and the occupants of the
fort. Even so, the appearance of imported goods, including
DPSKRUDHDQGHYHQVPDOOTXDQWLWLHVRI1RUWK$IULFDQQH
ZDUH GHPRQVWUDWH WKDW ERWK IRUWV ZHUH EHQHWLQJ IURP
the annona, no doubt organized and supplied through
the fortress of Novae.99 7KH UHFRYHU\ RI PLOLWDU\ QGV
armour and weapons at Dobri Dyal and Dichin, as well
as Iatrus, proves that the occupants performed military
duties. But, equally, the presence of numerous agricultural
implements at Iatrus, Dobri Dyal, Koula and especially
Dichin must mean that the inhabitants of these forts were
also engaged in agriculture. The discovery of a female
skeleton in Dichin proves that the fort did not just contain
soldier/farmers, but no doubt their families as well.100
Who these people were, it is impossible to say. It would
be tempting, even likely, that these forts contained Goths,
settled on the frontier and in its hinterland. However,
there is no proof that this was the case. As in the contemSRUDU\VHWWOHPHQWVQRUWKRIWKH'DQXEHWKHQGVRILURQ
copper-alloy and bone are Roman in type and form, and
FDQQRWEHUHJDUGHGDV*RWKLF7KLVDEVHQFHRIDOLHQ
material is itself what might be expected, particularly
since before crossing the Danube, as argued above, there
GLGQRWH[LVWDSXUHO\*RWKLFFXOWXUDODVVHPEODJH,Q
terms of jewellery and mundane artefacts, the Goths are
likely to have been using the same objects as the native

75

provincials. True, as described above, in appearance and


dress as well as language, most Goths would have been
easily distinguished from a Thracian or inhabitant of
Constantinople. But their distinguishing characteristics
are ones which do not survive in the archaeological
record. An exception, it could be argued, is the use of
Grubenhuser, such as those depicted on the column
of Arcadius, where they were associated with Goths
and probably used as an object of derision since
civilized Romans did not build such simple structures.
Grubenhuser have not been found dating back to the
Roman or even late Roman period, except possibly at
Intercissa and more convincingly at Dobri Dyal. But if
this structure had been introduced by the Goths, the rarity
of its appearance would seem inexplicable. A possible
explanation might be that Grubenhuser were used but
not widely; none were found in Dichin. Moreover, the use
of Grubenhuser cannot be taken as an ethnic indicator,
but rather denoted low status and a lack of resources. Up
until the early 20th century, Grubenhuser were used in
Bulgaria.101 They have advantages: they can be erected
without the use of a professional carpenter and do not
involve the acquisition of good quality timber. They are
cheap to construct. Also, there was no need to purchase
roof tiles. At Dobri Dyal, the western half of the fort
was occupied by such humble buildings which were
certainly occupied and were not just for storage; they
KDG FHQWUDO UHSODFHV ,Q FRQWUDVW WKH HDUWKERQGHG
and mudbrick buildings on the eastern side of the fort
involved greater investment; structural timbers were
needed and invariably the supply of roof-tiles to create
an overhang, protecting the mudbrick walls from getting
wet and eventual collapse. Consequently, the presence of
Grubenhuser suggests that those living on the western
slope were not as well supplied with materials as those
occupying the houses on the eastern side. Unfortunately,
there presence does not mean that these structures must
only have been used by Goths.102

99. For the role of Novae as a supply centre for the annona, see above,
p. 70.
100. A. Poulter, The transition to Late Antiquity on the lower Danube, cit.
(n. 51), p. 86.

7KHPLOLWDU\LQIUDVWUXFWXUHF
Some elements of military organization survived:
notably the supply system, organized from the legionary
base of Novae, was still operating during the 430s. The
occupants of Dichin, Dobri Dyal, Iatrus and other forts on
WKH'DQXEHIURQWLHUPXVWKDYHEHHQUPO\XQGHUPLOLWDU\

101. N. Spasova, Ouzemnoto zhilishte v Nikipolski I Plevensko, in Monoumenti


I Pamenitsi na Kultura, 16/3, 1976 (Bulgarian), pp. 26-34. For early
19th century examples, see A. Poulter, Nicopolis ad Istrum, cit. (n. 4),
pp. 49, 126-128, 172, and a 9th/10th century grubenhaus: ibid., pp. 166-170.
)RUIXUWKHUGLVFXVVLRQRI*RWKLFFXOWXUHDQGWKHIUXLWOHVVVHDUFK
for buildings which could be described as Gothic, see A. Poulter,
Invisible Goths, cit. (n. 29), pp 169-183.

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

76

AnTard, 21, 2013

ANDREW POULTER

jurisdiction. The new policy was systematically implemented; the fortresses of Dobri Dyal and Dichin were
built to contain soldier/farmers and were never intended
to house a regular garrison. They could not have served
as foederati, at least not in the sense in which the term
was applied to the Gothic armies during the second half of
the 5th century.103 But, if the forts were occupied by Gothic
settlers which, despite the lack of archaeological proof, still
seem likely, then the term, if applied to the inhabitants of
these forts, had a totally different meaning. Speculative but
possible, given the apparent status of the soldiers in these
forts, is that the mysterious post of comes foederatorum
was a command exercised over these new military bases,
possibly also applying to Iatrus and other frontier forts.
They were certainly part of the military establishment and
KDG PLOLWDU\ GXWLHV EXW WKH\ ODFNHG WKH UHJXODU RIFLDO
military support, as in the provision of building materials,
and were living with their womenfolk and no doubt
children. That they were engaged in farming and care of
livestock is not surprising; the villa-economy which had
performed this role during the 1st-4th centuries AD had been
totally destroyed. Whether this is the whole story remains
unknown. Other settlers, who had no military duties, may
have taken over the countryside. But until more intensive
survey is carried out, there is no way of knowing what had
happened to the abandoned villa estates.
The collapse of the system
The end of this period of relative prosperity can be
dated with some precision. On the frontier, Iatrus was burnt
to the ground in the second quarter of the 5th century; it
remained unoccupied until c. 500.104 The same fate befell the
legionary fortress of Novae.105 With such key installations
abandoned, it can be safely assumed that the Danube was

103. See above, pp. 64-65.


104. G. von Bulw, The fort of Iatrus, cit. (n. 32), pp. 468-470.
7KH ODWHVW FRLQQGV IURP WKH principia are issues of 416-422:
T. Sarnowski, Die principia von Novae, cit. (n. 35), p. 58.

not defended after c. 450. In the interior, the quadriburgium


of Koula was burnt down at some point late in the reign
of Theodosius II; it was never reoccupied. The latest coins
from Dobri Dyal provide a terminus post quem of 435 for
its abandonment, after which, like Koula, it was never put
back into commission. It would be not unreasonable to link
these destruction deposits with the invasion of the Huns of
Attila in 447 and the subsequent treaty which handed over
Dacia Ripensis to the Hunnic Kingdom; Marcianopolis was
also certainly destroyed by the Huns at this time. Despite
this apparently plausible explanation, there were exceptions.
$OWKRXJKWUDFHVRIEXUQLQJZHUHLGHQWLHGSDUWLFXODUO\LQWKH
gate-tower at Dobri Dyal, none of the excavated buildings
produced roof tiles; they must have been removed for reuse
HOVHZKHUH DQG WKH PDVV RI VWRQH EORFNV ZKLFK OOHG WKH
excavated building with earth and bonded walls must be the
result of systematic demolition. This implies that the fort was
not taken by a hostile force but was evacuated. Surprisingly,
one fort was not given up: Dichin. It continued to exist down
to the very end of the 5th century; because it was not located
close to a Roman road, it might have escaped the attention of
Hunnic raiding parties.
7KH UVW KDOI RI WKH th century witnessed a dramatic
change in the character of Roman military control in the
region: a decision which was inevitable, not only because of
the need to absorb the Gothic settlers, but also because the
agrarian system of villa estates had collapsed. If the annona
brought in luxury imports and even basic foodstuffs, there
must still have been a need for locally available supplies
to support the military. Although the possibility remains
that the Goths were settled in the new forts, this is still
uncertain: archaeology is incapable of providing an answer.
Whatever their status and ethnic origin, the employment
of soldier/farmers proved an effective solution: it worked
quite well for c. 50 years before the arrival of the Huns
put an end to the system. Rather than 382, the crucial date
for the implementation of these reforms would seem to be
later, around the year 400. What happened in the interval
between 382 and 400 is another problem which, at present,
cannot be resolved.
University of Nottingham

BREPOLS PUBLISHERS
THIS DOCUMENT MAY BE PRINTED FOR PRIVATE USE ONLY.
THIS DOCUMENT MAY NOT BE DISTRIBUTED WITHOUT PERMISSION OF THE PUBLISHER

Вам также может понравиться