Bukowski - Bludni Sin

Efthymios Rizos
Centres of the late Roman military supply network

in the Balkans: a survey of horrea
Beatitudine d(omini) n(ostri) Constantis victoris | ac triumfatoris semper
Aug(usti) | provisa copia quae horreis deerat | postea quam condendis
horrea deesse coeperunt | haec Vulc(atius) Rufinus v(ir) c(larissimus)
praef(ectus) praet(orio) per se coepta || in securitatem perpetem rei annonariae dedicavit1
In the happy times of our lord Constans, the victorious and triumphant
eternal Augustus. While supplies were anticipated that were insufficient
to fill the warehouses, eventually the warehouses started to be insufficient for the goods to be stored. The vir clarissimus Vulcatius Rufinus,
Praetorian Prefect, dedicated these (warehouses), which were commissioned by him, for the perpetual security of the annona.
This inscription celebrates the dedication of new depots for storing the supplies of the army at the Pannonian city of Savaria (Szombathely; Kom Vas/H) in A.D. 347-3502. It seems that, although Savaria already had
warehouses, increased annonary imports caused a storage crisis and the old facilities came to be insufficient.
Thus the text provides a unique reflection of a moment in the development of the infrastructure of the annona militaris in a province of the European frontiers of the empire. The securitas perpetua rei annonariae
was an important concern for the praetorian prefectures which managed this immense logistics system. In
the frontier provinces, where large quantities of imported and locally produced military supplies were accumulated, there was an urgent need for horrea, and such buildings must have been a priority in the building agenda of the state and the army. The warehouses of Savaria have not been located by archaeology,
but the inscription provides perhaps the most explicit testimony to the fact that the numerous late Roman
public storehouses/granaries known from the Danube and the Rhine provinces are associated with the annona. Indeed, most of them were built and functioned in the period of the late Roman annonary systems
establishment and most intensive function the tetrarchy and the 4thcentury. Their uniform architecture,
which followed the early Roman building tradition of hangar-like military granaries, is also indicative of
these buildings connection to a central building policy designed to provide infrastructure for the military
supply network3. Among the best-known late Roman horrea are those of Milan, Aquileia and Trier, and the
more recently excavated Building XV at the fort of Housesteads all ascribed to the period of the tetrarchy4.
CIL III 4180; Tth 2011, Cat. Nr 8. Translation by the author.

PLRE 1 (1971), 782 (Rufinus 25).
3 On military granaries of the imperial period in the West, see:
Salido Domnguez 2011; Rickman 1971, 213-290.
2
4 Aquileia
(prov. Udine/I): Mirabella Roberti 1965. Milan: A.

Ceresa Mori in: Sena Chiesa 1990, 102f. Trier: Eiden/Mylius
1949; Luik 2001; Rickman 1971, 264-265. Housesteads
(Northumberland/GB): Crow 2004, 92-94. 98f.; Rushworth
2009, 307-311.
Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013
659
They were oblong gable-roofed buildings divided into aisles by rows of pillars, with raised floors and strong
walls, often buttressed5. Several centuries after the end of Antiquity, their architecture was revived in the
late medieval tithe barns of northern Europe6.
As a permanent logistics network, the annona militaris entailed immense demands in both institutional and
material infrastructure. Its bureaucratic organisation is amply documented in late Roman legislation and has
been studied in considerable detail7. By contrast, its material traces and infrastructure have been studied
only partially8. This article attempts to redress some of this imbalance through a survey of public granaries
and warehouses (horrea) known from the Balkan provinces. Excavations since the 19thcentury have revealed
several late antique public warehouses in these regions, which allow us to follow the infrastructure of the
annona in remarkable detail. This rich architectural material can substantially promote our understanding of
the supply networks functioning, development and transformations by the end of Antiquity, and its impact
upon local economies and societies.
Here the examples are categorized chronologically and according to their settlement context. As we shall
see, the vast majority dates from the late 3rd and 4th centuries, while those from the 5th and 6th centuries are
much fewer. The large early group (late 3rd and 4th centuries) is discussed first, divided into three subcategories of settlement context (forts and small fortified settlements, villas and rural sites, and cities), while the
few later examples are discussed in separate sections.
Horrea of the late 3rd and early 4th centuries in military forts and small
fortified settlements
Several examples of military horrea come from forts of the Danube littoral where the annona was delivered
from both the Mediterranean and the neighbouring provinces of Thrace and Illyricum. In the early Roman
period, warehouses are known from forts and villas most of which were left in ruins after the 3rd-century
crisis. The making of a new frontier on the Danube in the period between Aurelian and Constantine led to
the restoration of many of the old defence centres and to the creation of several new ones. Tetrarchic and
4th-century horrea were found in a number of forts and settlements built or rebuilt during this period.
Capidava in Scythia Minor (jud. Constana/RO; fig.29, 4) was an important riparian fort on a bend of the
Danube, which in Late Antiquity is attested as both a military and civil centre, listed among the legionary
bases and cities of Scythia Minor. In the early 4thcentury, it received a circuit of very strong defences enclosing an area of about 1.5ha. A three-aisled horreum was built against the south wall of this fortification (dimensions 16.253218.75m) (fig.1). It was probably a granary, since it was equipped with a raised floor.
It was rebuilt perhaps in the late 4thcentury, retaining its original use. There is no evidence for it falling out
of use until the mid-6thcentury, when it was probably turned into a dwelling9.
Rickman 1971, 213-250.

Griswold 1999.
7 General on the annona militaris in Late Antiquity: Mitthof 2001
(the most exhaustive monograph on the subject); van Berchem
1937; Carl 2007; Carri 1994; Jones 1964, 448-469; Karayannopoulos 1958, 94-112; Schmidt-Hofner 2008, 117-163; Kissel
1995; Rickman 1971, 264-290. Local studies for Syria and Africa: Pollard 2000, 101-104; Pons Pujol 2008.
6
660
the built infrastructure of the annona: Rushworth 2009,

307f.; Fernndez Ochoa/Morillo/Salido Domnguez 2011;
Rickman 1971, 264-270; Whateley 2013. On the ceramic evidence: Karagiorgou 2001; Konrad 1999; Swan 2007.
9 Opri 2003, 26-33. 229-231. Excavation reports by various
teams working on the site (sector III of the Capidava excavations) from 1982-2012 can be consulted online at the website
Archaeological Excavations in Romania, 1983 2012: Preliminary Archaeological Reports (http://cronica.cimec.ro/).
8 On
E. Rizos Centres of the Late Roman Military Supply Network in the Balkans: a Survey of horrea
Fig.1 Capidava (jud. Constana/RO). Late Roman horreum. (After Matei 2000).
Still in Scythia Minor, a probably 4th-century horreum was partially uncovered at the fort of Flaviana (Cetatea
Ptulului; jud. Constana/RO)10.
In the neighbouring province of Moesia Inferior, a warehouse was found in the northern part of the riparian
fortress-city of Novae (near Svishtov; obl. Veliko Tarnovo/BG; fig.29, 23). It was a simple, two-aisled building (13m33m), probably with a raised floor, but with relatively thin walls. On its south side, there was a
room with hypocaust, interpreted by the excavators as a drying plant for grain. Built in the second quarter
of the 4thcentury, it remained in use until the late years of the samecentury, when it was destroyed by fire
and abandoned11. At least until the mid-5thcentury, Novae remained the base of the Legio I Italica, though,
at the same time, it also acquired the status of a city. It was probably a recipient of annona supplies imported
from the Mediterranean, which is also attested by inscriptions set up by primipilarii arriving from provinces
of the Aegean and the Levant until A.D. 43212. At the same time, however, Novae must have been the
gathering centre of supplies produced in the surrounding territory.
The neighbouring riparian fort of Iatrus (Krivina; obl. Ruse/BG; fig.29, 11), some 10km east of Novae, was
built probably under Constantine, in order to host an element of the Legio I Italica. Warehouses have been
located at it, but they belong to a secondary phase (the so-called period B), rather than to its original layout.
Period B starts around A.D. 350, when the buildings of the original Constantinian fort started to decline
and were gradually replaced by structures of a poorer nature. The only exceptions are two oblong two-aisled
10 Zahariade
11
1996, 225.
Dyczek 1997; 2002, 130-133.
12
Sarnowski 2005; 2013; ajtar 2013; Bresson/Drew-Bear/Zuckerman 1995; entries by U. Gehn in: LSA 1102-1103. 2445.
2590. 2596-2598 (Last Statues of Antiquity, online database at
http://laststatues.classics.ox.ac.uk).
661
Fig.2Iatrus (Krivina; obl. Ruse/BG). General plan of the fort in phase B/C. Buildings I and VII are seen on the left. The position of the
later Building XVIII is indicated in the centre (for its plan see fig.25). (After von Blow 2007).
buildings (Buildings I and VII), a bath, a peristyle complex (interpreted as the principia or praetorium) and a
Christian basilica. All these buildings stood at the west end of the fortified area, behind the principia of the
fort13. Buildings I and VII were most probably warehouses. Building I (14.90m38.90m, with 1.30-1.35m
thick walls) was two-aisled with a portico along its east faade, and stood immediately next to the bank
of the Yantra, thus being readily accessible from the river. It replaced an oblong structure of similar dimensions from the original Constantinian phase (Building XLVI), the use of which is unclear, but it is very likely
to have been also a horreum. Building VII was identical in plan, but larger and stood against the fortification
walls, following an angle in the walls (20.80m48.20m + 13.30m with 1.25m thick walls). Its plan and
size are strikingly similar to those of a horreum excavated within the Diocletianic fort of Palmyra. It has been
suggested that Building VII included an upper floor with rooms for barracks (fig.2)14. The buildings were
probably built around A.D. 350, and may be related to a fragmentary building inscription with the names of
the emperors Constantius II and Constans (co-emperors in 340-350). A coin of the same period (345-354)
found in the foundation of Building VII served as terminus post quem for their dating. The two warehouses
were destroyed and abandoned perhaps in the first decade of the 5th century15. Iatrus commanded the
confluence of the river Yantra (ancient Iatrus) into the Danube, by which it was connected to the territory
of Nicopolis ad Istrum. It was thus a point where the network delivering supplies from the Mediterranean
through the Danube met the regional supply resources of the Danubian plain.
The importance of supply bases behind the riparian frontier line proper is also demonstrated by warehouses
found at small fortifications in the interior areas like Storgosia in Moesia Inferior (Pleven/BG; fig.29, 33).
The site was a road-station under the Principate, but, in the late 3rd or early 4thcentury, it became a strong
13 On
the site of Iatrus (Krivina; obl. Ruse/BG) and its phases, see:
von Blow 2007; Herrmann/Wachtel 1979, 9-25.
14 Excavation reports by B. Dhle in: Klio 47, 1966, 57-151; B.
Dhle in: Iatrus-Krivina 1 (Berlin 1979) 53-62. For Building XLVI,
see Dhle 1999, 143. Cf. Dintchev 1999, who expresses doubts
662
about the use of Building VII as a horreum proposing its interpretation as a barrack. On the horreum of Palmyra, see note
88.
15 Excavation report by B. Dhle, Klio 47, 1966, 88. On the inscription: Wachtel 1999.
Fig.3 Castra Martis (Kula; obl. Vidin/BG). Quadriburgium with

possible storage rooms (1). (After Dintchev 2006).
Fig.4 Timacum Minus (Ravna; Zajearski okrug/SRB). 4th-century phase of the fort with the horreum. (After Petrovi 1986).
fortification associated with the Legio I Italica of Novae. A granary with strong buttressed walls stood in the
centre of the settlement. A large Christian basilica was built next to it in the 5thcentury, without causing any
damage. This may suggest that the horreum remained in use16.
In neighbouring Dacia Ripensis, the tetrarchic quadriburgium of Castra Martis (Kula; obl. Vidin/BG; fig.29,
33) was most probably also a supply base. This compact fortlet had two halls with rows of pillars on its
ground floor, which could be used for storage (fig.3). Perhaps under Constantine, the quadriburgium was
extended through a larger fortress attached to it. Rescue excavation within this later fortress located a building with a layer of burnt grain, probably a granary. Castra Martis appears as a city in the administrative list of
Hierokles (compiled under Theodosius II), suggesting that it was an autonomous fiscal district and production centre17. It lay at a small distance from the Danube and was readily approachable from the riparian civil
and military centres of Bononia (Vidin/BG) and Ratiaria (Archar; obl. Vidin/BG).
Another important artery of the late Roman military supply network in Dacia Ripensis was the Timok Valley
(SRB), which, as we shall see later on, has produced several important examples of 4th-century horrea. One
of them was the fort of Timacum Minus (Ravna; Zajearski okrug/SRB; fig.29, 35). Under the Principate,
Timacum Minus consisted of a small auxiliary fort (c. 1.5ha) and a civilian municipium. In the late 3rdcentury
A.D., the fort was rebuilt, and a horreum was erected in its centre. It had walls with buttresses and exterior
porticoes (fig.4)18.
Turning to Moesia Superior, an advantageous position comparable to that of Iatrus was occupied by the
fortified settlement located at the confluence of the river Poreka Reka (SRB) into the Danube (fig.29, 27).
The site is right at the middle of the Iron Gates district, just south of the riparian fort of Taliata (Veliki Gradac/Donji Milanovac; Borski okrug/SRB Serbia). The settlement, probably tetrarchic or Constantinian, was
protected by a long wall crossing the valley and isolating it from the riparian area. Behind the wall stood a
16 Kovacheva
1980; Kovacheva/Gerasimova-Tomova 1983, 105.

111-116. The site and building are visible on Google Earth,
coordinates: 43 22 58 N, 24 37 46 E.
17 Hierokl., Synekdemos 655, 5; Atanasova 2005, 31f. 42. 46;
Dintchev 2007, 505. The site and building are visible on
Google Earth, coordinates: 43 53 26 N, 22 31 26 E.
18
Petrovi 1986; Mladenovi 2012, 183f. cat. nr. 1288. The site
and building are visible on Google Earth, coordinates: 43 38
29N, 22 15 17 E.
663
Fig.5 Poreka Reka (SRB). General plan of the site with the two warehouses (A and B). (After Petrovi 1982-1983).
quadriburgium (probably left unfinished) and a set of unidentified buildings including a bath-house and two
identical rectangular buildings (15m9m) with very strong internal pillars and walls (1.5-1.8m thick), which
are thought to be horrea (fig.5). Their walls are too thick for their modest dimensions, and most probably
point to a tall, tower-like construction. If that is true, they may have served as watch-towers too. The site
is believed to have been a fortified supply base serving the entire region of the Iron Gates. Very probably it
was the gathering centre of grain and other goods produced in the valley of the Poreka Reka. The life of
the settlement seems to have been short, not exceeding the late 4thcentury19.
Still in the Iron Gates district, a crucial point of the Danube was Trajans bridge with its twin forts of Drobeta
(Turnu Severin; jud. Mehedini/RO) and Transdrobeta/Pontes (Kostol; Borski okrug/SRB). A horreum was
built during the restoration of Transdrobeta/Pontes (fig.29, 26) as a military base, which took place surprisingly late, after the mid-4thcentury. A part of the horreum was located directly inside the east gate of the
fort: it was strongly built with buttressed walls and had its floor covered with hydraulic mortar. The building
was in use for a very short period of time and is thought to have been destroyed and abandoned in the late
4thcentury20.
Finally, in Dardania, an important supply-base was located on the site of Gradina near Pe (RKS; fig.29,
25). It consisted of four large warehouses (c. 12m75m each, with 1.20m thick buttressed walls) within a
fortified settlement of about 1.3ha (fig.6)21.
19
20
Petrovi 1977; 1980; 1982-1983.

Vasi 1999, 31-33. Mladenovi 2012, 129 cat. nr. 745.
664
21
Srejovi 1982/1983, 35-43. Mladenovi 2012, 170 cat. nr.

1155.
Fig.6Gradina
(Pe/RKS). Partly hypothetical plan of the
horrea (mostly unexcavated). (After Srejovi
1982/1983).
Warehouses at military forts can hardly be described as a novelty in Late Antiquity. Forts were the final destination of the annona militaris which, after all, was created for their provisioning in the first place. Horrea
can therefore be expected at most military sites of the Danube. Yet some of the sites described here are very
likely to have been storage bases not only receiving imported supplies, but also locally produced ones from
their own productive territories. One should note their location in valleys or near tributaries of the Danube,
which most probably played an important role in the transport and distribution of the supplies.
Horrea of the late 3rd and early 4th centuries at villas and rural sites
A substantial number of early Roman horrea and some of the largest late Roman ones are known from villas. The rebuilding of the Danubian provinces after the 3rd-century crisis included the partial restoration of
the villa network which had been a major provider of supplies for the early Roman army. However, the restoration focused on the largest villas only, while most of the smaller and middle-sized rural establishments
665
Fig.7Mediana (Ni; Niavski okrug/SRB). The excavated horreum. (After Srejovi 1993).
of the earlier period were abandoned for good. The new villas appearing around A.D. 300 are very few and
large, and some of them may have been bases of imperial or senatorial estates. The richest evidence for this
development comes from the provinces of Dacia Mediterranea and Dacia Ripensis where extensive parts of
the countryside were agriculturally exploited as imperial estates in Late Antiquity.
The main base of landownership in the region of Naissus (Niavski okrug/SRB; fig.29, 21) seems to have
been Mediana, a suburb of Naissus which was used by campaigning emperors as a temporary residence
and gathering point of troops in the 4thcentury. Archaeology has located an extensive early 4th-century villa
complex which has been associated with Constantine. The site centred on a sumptuous peristyle residence
flanked on the east and west by two compounds consisting of horrea and rows of barracks, the whole
complex being enclosed by a light precinct wall (figs7-8). The west horreum-barracks compound has been
excavated, while its eastern counterpart has been located by aerial photography. The excavated west horreum is one of the largest known late Roman warehouses (27m91m)22. It was three-aisled and contained
massive dolia, about 2m tall23. The building had a porch along its south faade and a complex of smaller
rooms on its west end, including a large wine-press with masonry-built vats. The press and the dolia suggest that the building was used both for storage and for industrial production and fermentation of wine,
revealing the main productive activity of this estate. Several early Roman parallels can be found in wineproducing villas of Gallia Narbonensis, like the villas of Le Molard at Donzre (dp. Drme/F), Les Toulons at
Rians (dp. Var/F), and Vareilles at Paulhan (dp. Hrault/F)24. The villa of Mediana is thought to have been
functional until the late 4thcentury AD, after which it was gradually transformed into a rural settlement,
perhaps suggesting that the imperial estate ceased to be exploited. It is significant that the site is located
on the left bank of river Niava, through which the products could probably be shipped down to the Velika
Morava (Margus) and the Danube.
22
Miloevi 2011 (with earlier bibliography). The site and building are visible on Google Earth, coordinates: 43 18 39 N, 21 56
47 E.
666
23
24
Brandl/Vasi 2007, 101. Mladevoni 2012, 44. 76 Nr. 209.

Brun 2005, 45-55.
Fig.8Mediana. Aerial photograph showing the excavated parts (left, including the horreum on the extreme left) and unexcavated buildings east of the residential complex (a possible horreum). (After Miloevi 2011).
Two mostly unexcavated buildings within the fortified enclosure of the imperial villa of Gamzigrad
(Zajearski okrug/SRB; fig.29, 9) are probably horrea. They occupied the area south and west of the temple with the crypt. The south one was an oblong rectangular structure with a porch along its north faade
(51.2m19.4m) and had a raised floor suggesting that it was a granary. The west horreum was larger
(30m45.1m) and had 5 aisles divided by four rows of pillars, and buttressed outer walls (fig.9). Both
buildings seem to belong to the original Galerian phase of Gamzigrad and can be dated to the first decade
of the 4thcentury. A third horreum (44m16m) was excavated about 200m west of the fortification, which
recent geophysical prospection has shown to belong to a small extramural settlement. The same geophysical survey located another extramural quarter, stretching over 7ha north of the fortification, which was
enclosed by a light precinct and comprised rows of barracks, an oblong three-aisled horreum (106m23m)
and other buildings, probably constituting the extramural partes rustica and fructuaria of the villa (fig.10).
This quarter looks very similar to the horrea-barracks compounds of Mediana, suggesting that they were
perhaps contemporary and served similar functions25.
Based on the investigation of the residential complexes, the villa of Gamzigrad is thought to have been left
unfinished after Galerius death in A.D. 31126. Given the isolated position of the complex, it is plausible to
think that no emperor or high official ever spent time there, so the palace lost its raison dtre. Yet does
this mean that the local imperial estate ceased to function as well? With its four storehouses/granaries,
Gamzigrad is one of the largest storage bases in the northern Balkans, suggesting that this great estate was
25 Von
Blow et al. 2009, 113-116. The authors believe that the

extramural structures predate the construction of the Galerian
fortified complex, dating from the 2nd or early 3rdcentury.
26
Petkovi 2011, 114-128 (with earlier bibliography).
667
Fig.9 Gamzigrad (Zajearski okrug/SRB). The fortified imperial villa with the two intramural horrea denoted byh. (After Srejovi 1993).
a major producer which probably supplied the forts of the area. This means that Gamzigrad, far from being
merely the megalomaniac project of Galerius for his own retirement palace, was also an integral part of the
provisioning system of the army a function which is unlikely to have been given up after the emperors
death. It is therefore possible that the site, both inside and outside the fortification, continued to function
as a storage base throughout the 4thcentury, even if the palace proper had been abandoned. This will only
be possible to confirm through the excavation of the horrea and agricultural installations, which has not
taken place yet. Like Mediana, the fortification of Gamzigrad was taken over by a rural settlement towards
the end of the 4thcentury.
In this context, it is perhaps worth pointing out another, less famous, villa complex which should be examined alongside the imperial complexes of Gamzigrad and Mediana, since it is comparable to them in size
and luxury. That is the partially excavated villa of Kostinbrod (BG), fifteen kilometres northwest of Sofia,
which probably served as a great estate-base, imperial or senatorial. No warehouses were located by the
limited excavations which focused on the residential complex, but, by analogy to the villas of Mediana,
Gamzigrad and Montana (see below), they are likely to have existed. The villa of Kostinbrod was functional
into the early 5thcentury when it was destroyed by fire. In the following centuries, it was taken over by a
poor rural settlement which in the 6thcentury was turned into a small fortified village27.
27
Dintchev 2003.
668
Fig.10 Gamzigrad. Plan including the extramural sectors revealed by excavation and geophysical survey. The two extramural horrea (12) are seen at the top and bottom left. The intramural ones (3-4) are not shown on this plan (see fig.9). (After von Blow et al. 2009).
A major villa cluster was located at Montana (Montana/BG; formerly Mihailovgrad; fig.29, 19), a settlement which in the imperial period included an auxiliary fortress and a civilian community epigraphically
attested as municipium Montanensium. Under the tetrarchy, the local fort was rebuilt as a mighty hilltop
citadel28. Montana had three large villas which probably functioned from the 2nd to the 4th centuries and
included several horrea. The largest of these country houses, villa 2, had no less than three granaries/warehouses of significant size, two of which had raised floors (fig.11). A villa with horreum was also excavated
in the vicinity of Ratiaria, near the village of Makresh (obl. Vidin/BG; fig.29, 16). All of these establishments
can be classified as large villas and present a remarkable continuity of occupation from the 2ndcentury A.D.
Despite suffering damage during the 3rdcentury, they were repaired and continued their life into the late
4thcentury29.
28 CIL
III Suppl. 12376. Binev 2003, 160-182. Wendel 2005,

45f. cat. nr. 1225 (with earlier bibliography). The results of excavations are published in Montana I (Sofia 1987).
29
Dintchev 1997, 25f. 32-40. Mulvin 2002, 92. 95f.
669
Fig.11The villa of Madara (obl. Shumen/BG) and

villa 2 of Montana (BG): H horrea. (After Poulter
1983/1984).
Granaries of the early 4thcentury have been located on unfortified sites of unclear character, perhaps villas,
at Knjaeva (Zajearski okrug/SRB; region of Timacum Maius in the Timok valley; fig.29, 13) and Maskar
(umadijski okrug/SRB)30.
Turning to the province of Moesia Inferior, great estates probably existed also on the plain of Pliska, which
lies half-way between Marcianopolis and Abrittus. It has been suggested, that the villa of Madara (obl.
Shumen/BG; fig.29, 15), existing since the 1stcentury A.D., may have been the base of an imperial estate
under the Principate. The villa was damaged during the 3rdcentury and in the 4th it was partially rebuilt.
That phase included at least one horreum (18.8m8.6m19 m8.5m) built against the north precinct
wall (fig.11). The warehouse included a wine press and dolia, showing that it was used for wine production, like the much larger warehouse of Mediana (fig.12). The buildings were destroyed and abandoned
in the late 4th or early 5thcentury, and the site subsequently became a village which was inhabited into the
end of Antiquity31.
The restoration and building of villas after the 3rd-century crisis demonstrates that the frontier areas retained
their role as chief suppliers of the army, and that the economic relationship between legions and estates
in the north Balkans was revived. The revival, however, was only partial and the villa landscape was not restored to its original complexity32. Instead it underwent a process of nucleation and centralisation, the new
villa network consisting mainly of very large estates. The close coexistence of horrea and residential buildings at the great imperial villas of the 4thcentury is a testimony to the complex character of these residential
complexes as centres of both power and production.
Great villas like these are known from many parts of Roman Europe in the 4thcentury, but are almost unknown in East. It is perhaps no coincidence that the most numerous, largest and most sumptuous examples
of the villas listed here come from a part of the Balkans which belonged to the West Empire until A.D. 395,
namely north Illyricum. They therefore may represent a social and agrarian reality chiefly pertaining to the
30
Srejovi 1982/1983, 35-43. Mladenovi 2012, 124. 151 cat.

nrs. 697. 958.
670
31
Dremsizova-Nelchinova 1984. Mulvin 2002, 90-91. Dintchev

1997, 79-74.
32 Henning 1987, 22-34.
Fig.12Madara villa. Horreum

with wine press and dolia (also
see fig.11). (After DremsizovaNelchinova 1984).
Roman West. Scholarship and excavations on such sites not only in the Balkans have focused on their
sumptuous residential buildings and mosaics, an approach leading to substantial misconceptions of their
character, overemphasising their residential and political aspects and underplaying their economic role33.
For example, earlier scholarship postulated that the Villa del Casale at Piazza Armerina (prov. Enna/I) had
no pars rustica and fructuaria, but was merely a pleasant country house of a senator or emperor. Recent research, however, located two large horrea southwest of the residential complex, showing that the villa was
indeed the centre of an estate34. The great villas of the northwest Balkans probably played a similar role.
Horrea of the late 3rd and early 4th centuries in cities

One of the salient features of the post-crisis restoration of the northern Balkans was the building of horrea
in cities. Early Roman horrea in the region are known almost exclusively from forts and villas. By contrast, the
majority of late Roman ones has been found at cities. This demonstrates that something changed in the provisioning system, with the addition of cities as a new major player indeed as the main annonary centres. In
areas near the frontier, cities took on an active role in the military organisation of the Late Empire35: written
accounts about the defensive role of new or newly-fortified cities like Amida (Diyarbakr/TR) in Mesopotamia, Karin/Theodosiopolis (Erzurum/TR) in Armenia and Dara/Anastasiopolis (Mardin ili/TR) in Mesopotamia demonstrate that the army relied on them for its provisioning and used them as strongholds in times
of war. They had to be prepared to withstand sieges and, at the same time, they were major producers and
gathering bases of grain and other supplies. Storehouses and fortifications are two categories of infrastructure provided in support of these functions warehouses are explicitly mentioned for Dara and Theodosi-
33 The
role of the great villas of Spain as annonary producers has

been suggested by Fernndez Ochoa/Morillo/Salido Do
mn
guez 2011, 284.
34
Pensabene 2010, 1-12.

the role of cities in defense, see: Pollard 2000, passim; Crow
2007.
35 On
671
opolis36. The combination of sophisticated fortifications and horrea in

the period of the tetrarchy and Constantine can be archaeologically
observed on several sites in the Danube and Rhine provinces, especially
at Keszthely-Fenkpuszta (Kom. Zala/H), Kapospula-Alshetnypuszta
(Kom. Tolna/H), Sgvar (Kom. Somogy/H), Trier, Strasbourg and Maastricht37. A connection between urban fortifications and the annona
system has also been suggested for strongly fortified cities in north
and west Spain and south-west Gaul, even though no horrea have
been located there so far38. The Middle and Lower Danube provinces
and their direct neighbours clearly fit this model, since warehouses and
walls were indeed the most prominent kinds of infrastructure provided
to cities of that region in the late 3rd and 4th centuries.
In the early 1880s, Grigore Tocilescu, one of the fathers of Romanian
archaeology, discovered a robust rectangular building (56m20m)
with three aisles and buttressed walls (c. 1.6m thick), standing on the
central crossroad of the city of Tropaeum in Scythia Minor (Adamclisi;
jud. Constana/RO; figs13; 29, 36)39. Due to the early date and unsophisticated character of the excavation at Tropaeum, no certain conclusions can be drawn about the duration of the buildings life. Later
additions included a narrow oblong room with an eastward apse, resembling a chapel, which was built along the north side of the buildFig.13Tropaeum (Adamclisi; jud. Con
ing, taking over some of the adjacent street. A portico along the east
stana/RO). Central horreum (basilica
forensis) of the late Roman city. (After
side also a later addition contained large dolia, over 1.5m high
Tocilescu 1900).
and 1.2-1.5m in diameter. Buildings of uncertain date and function
took over the interior of the structure when it fell out of its original
40. Tocilescu, interpreted it as a forum basilica of the age of Trajan. Later research abandoned the idea
use
that a forum existed on the site and established that the building belonged to the tetrarchic period, when
Tropaeum was rebuilt and fortified after its destruction during the 3rd-century crisis. Its interpretation as a
basilica, however, was not questioned until Andrew Poulter and more recently Ventsislav Dintchev correctly
argued that the structure had all the features of a late Roman granary41. The discovery of this important
building by Tocilescu took place in a time when the architecture of Roman granaries and warehouses was
little known. Besides, its location at the centre of the city also created expectations for civil buildings, thus
leading its excavator to his erroneous interpretation. As we shall see later on, however, this is not the only
example of a horreum built within an urban settlement in the early 4thcentury. The city of Tropaeum was
so radically redesigned and rebuilt after its mid-3rd-century destruction, that it came to be regarded as a
newly-built city, representative of a new understanding of urbanism. The dedicatory inscription from one
36
Amida (Diyarbakr/TR): Amm. 18.9.1-3. Theodosiopolis: Moses of Chorene 3.59, quoted by Adontz 1970, 119. Dara (Mardin ili/TR): Procopius, Buildings 2.1; Ps.-Zachariah of Mitylene,
Syriac Chronicle 7, 6 (see Greatrex et al. 2010). See also the
discussion in Crow 2007.
37 Pannonia: Borhy 1996; Heinrich-Tamska 2011, 655-658. 677.
679-681 (with earlier bibliography on the sites). Strasbourg: Gis
singer 2002. Trier: see note 4. Maastricht: Panhuysen 1996.
General on the Rhine and Upper Danube fortifications with occasional references to warehouses: Johnson 1983, 137-195.
672
38 Fernndez
Ochoa/Morillo/Salido Domnguez 2011; cf. criticism

by Arce 2011.
39 Tocilescu 1900, 91. On Tocilescu and his role in the birth of
archaeology in Romania, see: Niculescu 2011, 400-403.
40 Prvan 1912, 138-143. Barnea 1968, 467f.; 1979, 20. 251.
The site and buildings is visible on Google Earth, coordinates:
44 05 30 N, 27 56 39 E.
41 Poulter 1983/1984, 119. Dintchev 2005, 283. Zahariade
2006, 107-108.
Fig.14Zaldapa (near Abrit; obl. Dobrich/

BG). Partly hypothetical plan of the horrea in
the centre of the fortified city. (After Dintchev
2005).
of the gates declares that the civitas Tropaeensium was built ad confirmandam limitis tutelam a civilian
settlement with a strong supportive role in the defence of the frontier42. The strong defences and central
position of the horreum explain the meaning of this statement and are evocative of the centrality of defence
and the annona in the life of the community.
A few years after Tocilescus excavations at Tropaeum, the father of Bulgarian archaeology, Karel korpil,
found parts of a similar building on the site which he thought was Abrittus, but which was later identified
as the late-antique town of Zaldapa in Scythia Minor (near the village of Abrit [formerly Aptaat]; obl. Dobrich/BG; fig.29, 38). It consisted of two longitudinally linked three-aisled halls which formed a complex
of overall dimensions 101m18m (48m18m each warehouse, with an intermediate common antechamber) (fig.14)43. Its very thick walls (1.2-1.5m) were strengthened with buttresses. The building stood at the
centre of the settlement. Following Tocilescus ideas about Tropaeum, korpil believed that his find was also
a basilica, but he never published anything about his excavation, and, in the meantime, the buildings he uncovered have completely disappeared. Luckily, his notes and plans survived, and, almost acentury after the
dig, Sergei Torbatov used them to produce a short monograph about the site, accompanied by plans based
on korpils sketches and on recent fieldwork. Torbatov, following the original excavators view, also interpreted the central building as a basilica. His views, however, received convincing criticism by Dintchev who
proposed that the building was a granary44. The same issues discussed about Tropaeum apply at Zaldapa.
The settlement seems to be a newly-founded city of the tetrarchic or Constantinian periods, very similar and
contemporary with Tropaeum, but significantly larger (25ha) (fig.15). The granary at its centre shows that
it had a similar role as a settlement supporting the provisioning of the frontier forces.
In the 4thcentury, the walled area of the city of Istrus (or Histria) in Scythia Minor (Istria; jud. Constana/RO;
fig.29, 12) was dominated by four irregularly rectangular buildings, three of which were three-aisled and
one two-aisled (dimensions 25.75/24.8m12.62/12.7m; 17.5m17.3m; 16.65m11.8m; 24.1/23.75m
7.25/6.05m) (fig.16)45. They stood near the main gate and were accessed through a street leading to a
small porticoed courtyard. On the east side of the courtyard there was an apsed hall with a spacious rectangular antechamber, whose form and size recall the principia building at the fort of Iatrus. The buildings
were initially interpreted as civil basilicas of the 6thcentury. Stratigraphic observations led later excavators
of Istrus to propose a dating to the 4th century without questioning the interpretation of their use. The
42 CIL
3 13734; also see discussion by Poulter 1983/1984, 119;

2007a, 35.
43 The same plan can be seen at the horrea of the villa of Piazza
Armerina (prov. Enna/I; see note 34).
44 Torbatov
45
2000, 37-47; 2003, 96-99. Dintchev 2005, 282f.

G. Florescu in: Histria I, 1954, 114. 116. 122. Suceveanu/Scorpan 1970. The site and buildings are visible on Google Earth,
coordinates: 44 32 50 N, 28 46 24 E.
673
evidence about the period of their function is virtually lost, but they are thought to have functioned at
least during the 4thcentury46. Recently, V. Dintchev
suggested that the interpretation of these buildings
should be reconsidered, suggesting that they were
actually granaries/warehouses. Their architectural
type and positioning next to the walls and near the
gate support this interpretation. If this hypothesis is
correct, they constitute one of the most remarkable
clusters of late Roman horrea in the entire region.
Istrus, as a port on the Black Sea, may have been one
of the points where the annona from the Mediterranean was delivered, but at the same time it must
also have been the gathering centre for supplies produced in the surrounding plain.
Another example of a probably misinterpreted
Fig.15Zaldapa. General plan of the city: 1 Christian basilicas.
horreum can be found at the port-town of Calla2 horrea. (After Dintchev 2009b).
tis in Scythia Minor (Mangalia; jud. Constana/RO;
fig.29, 3). It was a three-aisled trapezoid structure
(17m13.4m15.5m14m) built with its narrowest side against the north walls of the city (fig.17). It
was accessed through an enclosed courtyard, on the east side of which stood a suite of three rooms with
a small veranda. The finds are mostly unpublished and the phases of the complex are only very roughly
known. In the late 5th or 6thcentury, the complex was refurbished, and perhaps converted to some official
use; a peristyle was built on the courtyard and marble decoration was added. Small storage rooms with dolia were created in the south part of the warehouse. The building was interpreted as a public building, bath,
Christian basilica or private house47. Once again, however, the form, size and position have more parallels
in warehouses than civil or ecclesiastical buildings. Its positioning perpendicularly against the fortifications,
in particular, is also observed at the horrea of el-Lejjun (gouv. Kerak/JOR), Tokod (Kom. Komrom-Esztergom/H) and Cluijk (prov. Noord-Brabant/NL)48.
The strongly fortified town of Abrittus in Moesia Inferior (Razgrad/BG; fig.29, 1) was also equipped with
a horreum (56.25m20.2m), similar to those of Tropaeum and Zaldapa (fig.18). Located directly south of
the west gate of the city, it was three-aisled with rows of pillars and strong buttressed walls. Its position and
design are strikingly similar to those known from the inner fortifications of Pannonia49. At some point, probably later in the 4thcentury, a second warehouse was built north of the west gate, just opposite the already
existing one. Unlike the first horreum, which stood about 5m from the wall, this was built directly against
the defences. The substructure of its walls show that it was also three-aisled, but smaller (34m17m),
without buttresses and probably divided by continuous walls rather than rows of piers. The two buildings,
which were correctly recognized by their excavators as horrea, are believed to have been destroyed and
abandoned in the late 4th or mid-5thcentury50.
46 A.
Suceveanu in: Histria VI, 1982, 86f. Condurachi 1957;

1971. Dintchev 2005, 280-282.
47 Barnea 1977, 128-134. Condurachi 1971. Zahariade 2006,
109.
674
48 El-Lejjun
(gouv. Kerak/JOR): Crawford 2006. Tokod (Kom.

Komrom-Esztergom
/H): Visy 2003, 84-86. Cluijk (prov.
Noord-Brabant/NL): Johnson 1983, 146.
49 See note 37.
50 Radoslavova/Dzanev 2003, 133f.
Fig.16 Istrus (Histria/Istria; jud. Constana/RO). The late Roman warehouses sector: 1-4 horrea. 5 apsidal building. (After Histria I,
1954).
Fig.17 Callatis (Mangalia; jud.

Constana/RO). Late Roman horreum
complex: Left 4th-century phase.
Right late 5th to 6th-century phase.
(After Barnea 1977).
Still in Moesia Inferior, a public warehouse can be found at the centre of Nicopolis ad Istrum (Nikjup; obl.
Veliko Tarnovo/BG; fig.29, 22). The north wing of the Trajanic/Hadrianic agora of the city consisted of
a monumental Ionic stoa and a sumptuous hall of unknown functions, which were damaged during the
3rdcentury. In the subsequent period of repair they were replaced by two structures built with opus mixtum
masonry, with outer dimensions 19m43m (west) and 19.45m67.4m (east) (fig.19). These late antique
buildings were divided into three aisles by rows of large pillars, and at least parts of them probably had
wooden floors raised on a suspensura built of reused limestone blocks. A coin of Constantine found in the
mortar of the masonry provides a terminus post quem for their construction. It is unclear when the life of the
two buildings ended; the excavation reportedly produced 197 coins, dating from Gratian (367-383) to Leo
I (457-474) and clay floors of later structures over the destruction layer. The two buildings were destroyed
by an earthquake, perhaps in the late 4th or early 5thcentury. They are unlikely to have been functional later
than the mid-5thcentury, when the early Roman city of Nicopolis declined and was gradually replaced by
675
Fig.18 Abrittus (Razgrad/BG). General

plan of the city: 2-3 horrea. (After Dintchev
2009b).
the late antique fortified site51. Based on their position on the agora, both buildings were interpreted by
their excavators as civic, the east one being regarded as a basilica. The thought is plausible, given the topographical context, but the architecture and austere utilitarian form of the buildings are incongruent with
the setting. There was evidently no effort to integrate them in the monumental aesthetics of the agora. The
so-called civil basilica does not even seem to have had access to the piazza; it had an entrance on the neighbouring street, while its south wall seems to have formed an uninterrupted new north limit of the agora.
The construction of the two warehouses did not necessarily mean the end of administrative business on
the agora of Nicopolis ad Istrum, although the replacement of the monumental Ionic stoa by the utilitarian
horreum caused a substantial reduction of the space available and considerable aesthetic degradation. Nevertheless, an effort was made to keep up the monumentality of the civic ensemble, and five statues were
relocated and set up in front of the south wall of the east horreum (civil basilica).52 Given the fact that
cities and settlements near Nicopolis received horrea rather than basilicas in the same period, it seems convincing that this city also went through the same process. It is, of course, remarkable that Nicopolis gave up
such a fine structure to build a warehouse, but the choice can be understood, if we consider the advantages
of accessibility through the cardo and decumanus maximus. The same reason seems to have dictated the
building of horrea at the centre of Tropaeum and Zaldapa and, as we shall see later on, at Cabyle (Kabile;
obl. Jambol/BG), Sirmium (Sremska Mitrovica/SRB) and Scupi (Skopje/MK).
Further south, in the Late Roman province of Thracia, the settlement of Cabyle (Kabile; fig.29, 2) was organised as a city in the early 4thcentury, after several centuries of occupation as an auxiliary fort and village
51 Ivanov
/Ivanov
1994, 73. 97. Vladkova 2007, 208-214.

Dintchev 2009a, 76-78.
676
52 Ivanov/Ivanov
1994, 60. 63.
Fig.19 Nicopolis ad Istrum (Nikjup; obl. Veliko Tarnovo/BG). The Roman agora: 1 east horreum (basilica). 2 west horreum (aedes
7). (After Dintchev 2009a).
of veterans. Its fortifications were rebuilt and a large horreum (37.35m12.45m) was erected at the centre
of the settlement. The excavation has not been published and little is known about the finds. Two buildings
that were probably also horrea were built against the north and east wall of the fort of Cabyle (a separately
fortified part of the late Roman town). These buildings also remain unpublished and their chronology cannot be defined with certainty53. Cabyle lay near the river Tonzus (mod. Tundzha), a tributary of the Hebrus
(mod. Maritsa), which may have been used for transporting annonary supplies.
Another newly-founded late Roman city in Thracia was Diocletianopolis (Hissar; obl. Plovdiv/BG; fig.29,
8). A 4th-century horreum (23.4m15.3m) was excavated within the citys military quarter. It was a simple
rectangular building used for storing goods in jars54.
Two storehouses were recently excavated at Maronea (Agios Charalambos; perifria Anatolikes Makedonias kai Thrakes/GR; fig.29, 17) which was probably the only notable port in the western part of the Late
Roman province of Rhodopa in the 4thcentury (fig.20). The buildings were two parallel two-aisled halls,
8.9m45m, located near the port, in what seems to have been an area of monumental buildings. The
excavator ascribed them to the early 3rdcentury and their abandonment the late 4th. Maronea must have
been a gathering centre of the annona collected from Rhodopa, whence the goods were shipped to the
frontier55.
One of the largest horrea-bases has been found at Serdica (Sofia/BG; fig. 29, 29), provincial capital of
Dacia Mediterranea and one of the most important hubs of communications, defence and administra53
Excavation reports in:

1978, 67f.; 1979,
70; Nikolov 2003, 318-323. The site and buildings are visible
on Google Earth, coordinates: 42 32 58 N, 26 28 59 E.
54 Madzharov
1993, 150f.
2003, 14f. The site and buildings are visible on
Google Earth, coordinates: 40 52 27 N, 25 30 40 E.
55 Kokotaki
677
Fig.20Maronea (Agios Charalambos;

perifria Anatolikes Makedonias kai
Thrakes/GR). View of the excavated warehouses. (Photo E. Rizos).
tion in the central Balkans. Rescue excavations in

central Sofia revealed a row of at least eight horrea filed along the western city-wall (fig. 21). The
warehouses were three-aisled with an approximate
average size of 20m35m each and were dated to
the early 4th century. Once again, these buildings
were initially misinterpreted as civil. As Dintchev has
already argued, they have the architectural features
and typical position of late Roman warehouses56.
Serdica was probably a supply-centre of regional importance, connected to the Danube by the valley of
the river Iskur (Oescus).
A central location, on the crossing of the main streets,
was chosen for the building of a large warehouse at
Sirmium (Sremska Mitrovica/SRB; fig.29, 31), probably under the tetrarchy. The building (44m22m)
was 5-aisled with 4 rows of 6 pillars and probably
had a raised floor. It seems that it stood near monumental buildings (fig. 22)57. Sirmiums role in defence and administration was extremely important,
since until the 440s it was the main residence-city
of the Praetorian Prefects of Illyricum. Built on the
river Sava (Savus) it was easily approachable from
the Danube.
56
Fig.21 Serdica (Sofia/BG). Warehouses along the southwest

walls. (After Dintchev 2005).
678
57
Dintchev 2005, 279f.

Boskovi et al. 1974, 608-610. M. Jeremi in: Srejovi 1993,
113f.
Fig.22Sirmium (Sremska
Mitrovica; Sremski okrug/SRB).
Central horreum. (After
Boskovi et al. 1974).
A large 4-aisled horreum (15m65m) occupied a central position on the decumanus maximus of Scupi
(fig.29, 28), provincial capital of Dardania. The building is very similar and probably contemporary with
the horrea of Gamzigrad, Mediana and Sirmium. A small public bath was built within the horreum after
its abandonment. The baths plan recalls the 6th-century bathhouses found at Cariin Grad (Jablaniki
okrug/SRB)58. Scupi may have been a supply-centre of regional importance for forts in south Dardania and
north Macedonia, which seem to have been densely garrisoned regions in Late Antiquity. The river Vardar
(Axius), on whose left bank Scupi lay, must have been used for the transport of the goods.
In north Dardania, a notable-supply base was Soanica (RKS; fig. 29, 20). Under the Principate, it was
epigraphically attested as Municipium DD (Dardanorum?), a civil centre closely related to the mines of
the area. In Late Antiquity, the site is not mentioned by the sources or inscriptions, but it is likely to have
preserved its civic/municipal status until the late 4thcentury, when it was destroyed and abandoned. With
a sizeable unfortified settlement of about 30 ha, Soanica played an important role in the annona network
and was equipped with a complex of two three-aisled horrea with strong buttressed walls (43.3m16m
and 42.7m15.5m). The buildings took over what seems to have been the forum and religious centre of
the early Roman town, respecting a pre-existing temple which stood on the rear side of the courtyard between them. The complex was accessible from the street through a portico, in an arrangement reproducing
the plan of the twin horrea of Trier, Milan and Aquileia (fig.23). A very thin layer of abandonment was discerned under the debris of the collapsed horrea, but it produced no dating finds. The buildings are unlikely
to have continued their life beyond the destruction and abandonment of the town in the late 4thcentury59.
Finally, a 4th-century horreum was found within the fortified city of Horreum Margi in Moesia Superior
(uprija; Pomoravski okrug/SRB; fig.29, 10)60. Horreum Margi and Municipium DD were probably two
of numerous supply-bases that used the river Velika Morava (Margus) and its tributaries to send their products to the Danube.
58 Korakevik
2002, 65-74. Interestingly, a bath was built within

an abandoned part of the horreum at Housesteads during the
4thcentury: Rushworth 2009, 311.
59
erkov 1970, 14-46; Mladenovi 2012, 24-26. 44. 199 cat.

nr. 1450. The site and buildings are visible on Google Earth,
coordinates: 43 03 06 N, 20 48 34 E.
60 Mladenovi 2012, 84 Nr. 287.
679
Fig.23 Municipium DD (Soanica/RKS). Double

horreum. (After Srejovi 1982/1983).
GENERAL REMARKS ON THE TETRARCHIC PERIOD AND THE 4TH CENTURY

It is clear that the most numerous and largest late Roman granaries/warehouses in the Balkans were urban.
They thus allow us to identify the involvement of cities in the military supply network as a salient aspect of
the tetrarchic military and administrative reforms in the region. The cities with their large storage facilities
now probably performed productive functions previously confined to the countryside, probably making up
for some of the numerous villas that were abandoned in the mid-3rdcentury. This fits with the impression
mentioned earlier that the productive landscape of the north Balkans became more nucleated and centralised, now focusing on urban centres and a few very large estates. The involvement of cities in the annona
had major implications for their socio-political character. Under the Principate, the cities were bases of the
landed class and hubs of interregional commerce, political and cultural activities. They were in contact with,
but clearly separated from sites of a primarily military and rural character. Now, however, their involvement
in military supply meant a convergence with these realms and a degree of compromise of their urban character. The central positioning of warehouses and the fact that they outnumber virtually every other category
of public building in the 4thcentury, except fortifications, reflects the institutional centrality of the annona
in the administrative life of cities and its impact upon urban economy, society and culture.
The geographical positioning of late 3rd- and early 4th-century supply-bases, including cities, villas and forts,
is remarkable for the way it takes advantage of the transport possibilities of rivers and their valleys, especially
the network of tributaries of the Danube (fig.29). The Yantra (ancient Iatrus) is a very illustrative example, since it linked the territory of Nicopolis ad Istrum to the forts of Iatrus and Novae. Similarly, the Iskur
(ancient Oescus) connected Serdica with the city of Oescus (Gigen; obl. Pleven/BG) and its surrounding
680
villas; the Ogosta (ancient Augusta) connected Montana with the fortress-town of Augustae (Hurlets; obl.
Vratsa/BG), while the Timok (ancient Timacus) connected Timacum Minus, Gamzigrad and Aquae. The vast
river system of the Velika Morava (ancient Margus) was evidently a major transport artery through which
products could travel to and from the Danube. It is also important to stress the presence of supply-bases at
a distance of about 50km from the Danube: Tropaeum, Zaldapa, Abrittus, Nicopolis ad Istrum, Storgosia,
Montana, Castra Martis, Gamzigrad, Timacum Minus and Horreum Margi. This suggests that a supportive
network of settlements, very similar to the better known inner fortifications in Pannonia, was also provided
for the Lower Danube frontier. Similar support zones dotted with fortifications and horrea probably existed
also in Raetia, so that we may now regard them as a common pattern of defensive policy and organisation
all along the Danube frontier61.
The Theodosian Period

The building of warehouses, which flourished under the tetrarchs and Constantine, seems to have lost much
of its intensity after A.D. 350. As we saw above, the only examples dated to the third quarter of the 4thcentury are those of Iatrus and Transdrobeta/Pontes. Does that mean that the needs of the supply-system had
been covered, or that the investment of the empire on the annona network yielded less than expected? It
is possible that changes in the system of military supply and payments affected the built infrastructure of
the annona. In the second half of the 4thcentury we have the first recorded steps towards the adoption of
adaeratio, namely the commutation of rations into ration allowances, which gradually became an established
practice by the 5thcentury62. Yet the introduction of cash payments by no means led to the abolition of payments and supply in kind and cannot provide a satisfactory explanation of the decline of warehouse building.
Zosimus reports that, during Valens confrontations with the Goths (367-369), special care was taken to
secure the food supply of the troops and that annonary imports were delivered to the Danube cities thanks
to the efficient administration of the newly-appointed praetorian prefect of the East Auxonius63. Themistius
also insists on the same theme in his panegyric given at Constantinople after the peace with the Goths
in 370. Yet he goes on to mention civil unrest and food shortages prevailing prior to Valens restoration
of order: Supplies for the military units have been dispatched, and a surplus of necessary things has
been imported in advance [ ,
]. Thanks to this, the guards are not forced to fight against the empires subjects instead
of the barbarians, and to abuse the former because of their own destitution, while refraining from the latter because of the peace. For, I know not how, our soldiers have reversed the normal order of courage and
fear: they despise the barbarians, while exceedingly fearing the farmers, and it is more terrible for them to
be accused by the farmers than myriads of Scythians attacking!64.
Disruptions in annonary imports from the Mediterranean during the revolt of Procopius (365-366) probably increased disproportionately the pressure upon local farmers, causing tensions with the soldiers. The
supply of the Danube troops relied on a balanced mixture of local production and external imports, which,
if disturbed, could destabilise the frontier provinces, and that seems to have happened in the 360s. These
events combined with the wars of the late 370s probably contributed to the sharp decline of villas and rural
61 Mackensen
63 Zosimus
62 On
1999, 234-239.
adaeratio, see: Karayannopoulos 1958, 103f.; SchmidtHofner 2008, 212f. 315-320.
64 Themist.
4.10.3-4; Lenski 2002, 127-137.

or. 10, 138 b-c (translation by the author).
681
Fig.24 Dichin (obl. Veliko Tarnovo/BG). Early 5thcentury horreum. (After Poulter 2007b).
supply-bases in the Balkans. All the villas with warehouses mentioned above, and many urban horrea, were
abandoned by or around A.D. 400. By the mid-5thcentury, it seems none of the 4th-century horrea remained
functional.
Newly-built examples known from the Theodosian period are very rare and small in size. A group of small
warehouses (c. 14m x 8m each) occupied the western half of the fort of Dichin (obl. Veliko Tarnovo/BG), a
few kilometres west of Nicopolis ad Istrum, which is thought to have been built as a fort for a small garrison
of federates around A.D. 400 (fig.29, 7). The warehouses were poorly built with mud-bricks and at least
five of them had raised floors (fig.24). They were destroyed with the rest of the fort in the 470s, during an
invasion which left the site in ruins. Quantities of seeds including cereals, millet and pulses were found carbonised in the buildings, alongside pottery and amphorae from the Danube region and the Mediterranean65.
Roughly contemporary with the horrea of Dichin, a large granary was built at the fort of Iatrus (Krivina; obl.
Ruse/BG), following the destruction of the two 4th-century horrea of the site (fig.29, 11). The structure
(building XVIII) stood northeast of the Christian basilica, and consisted of two rooms, the north of which
measured 18.7m6.3m19.3m6m and featured a raised floor (fig.25). The building was erected in
the early 5thcentury and was destroyed around A.D. 450, perhaps during the Hunnic wars66. Similar warehouses, dated to the early 5thcentury, were also recognised at the fortified settlements of Montana in Dacia
Ripensis and Shumen in Moesia Inferior67.
The few newly-built horrea of the Theodosian period demonstrate that the annona network continued to
take advantage of rivers as transport arteries; supply centres continued to exist in areas and sites known to
have played a role in the network during the 4thcentury, namely the region of the Yantra (Dichin and Iatrus),
and the Ogosta (Montana). The plain of Pliska probably also retained its importance for the supply-network,
with Shumen now rising as a new central place. On the other hand, the construction of new warehouses
now follows the broader transformation of the settlement landscape in the northern Balkans after the
war with the Goths. This was the age of small fortified settlements and hilltop-sites like Dichin and Dobri
Dyal (Veliko Tarnovo/BG) near Nicopolis ad Istrum (both of a size of 0.5-1ha), Shumen, Pernik, Sadovets,
Cherven, Zikideva (Tsarevets Hill; obl. Veliko Tarnovo/BG), and many others68. Excavations on sites of this
65 Poulter
2007b, 87-91; 2004, 244-246; 1999. Swan 2007.

Grinter 2007. Dintchev 2010.
66 Kuzmanov 1991, 112-116. Stanev 1995, 124-127.
682
67
68
Dintchev 2005, 284f.

Dichin (obl. Veliko Tarnovo/BG): see note 65. Dobri Dyal (obl.
Veliko Tarnovo/BG): Poulter 2013.
Fig.25Iatrus. Early 5th-century horreum (Building XVIII) (see fig.2

for its location). (After Kuzmanov 1991).
category have shown that they were inhabited by a population of men, women and children (found in burials), including farmers and soldiers (finds include both weapons and agricultural equipment). They clearly
played a role as centres of production and storage of food supplies, but their potential was very restricted
in comparison to annonary centres of the 4thcentury. Very often, their location at high altitudes precluded
agriculture on a serious scale69. Things were probably better at settlements on plains, as suggested by the
horrea of Dichin, though archaeobotanical evidence from both Dichin and Nicopolis ad Istrum shows a
significant recession in the cultivation of wheat and cereals, and a greater emphasis on the production of
legumes and millet, suggestive of horticulture within or near the walled areas in other words, a serious
change in the scale of production70.
69 Ch.
Kirilov, Entstehung, Charakter und Niedergang der befestigten Hhensiedlungen im stlichen Balkan (5.-7. Jh. n.Chr.)
[Lecture given at the workshop New Cities in Late Antiquity,
Istanbul, 9-10 November 2013].
70
Poulter 2007b, 77.
683
At the same time, settlements embodying the previous socioeconomic reality, namely villas and cities,
enter a period of neglect, if not crisis. Despite the
wars and dangers of the late 4thcentury, there appears to be no effort to improve the walls and infrastructure of cities north of the Haemus. Purposefully
or by necessity, emphasis now is on small fortified
settlements in sharp contrast to several notable
urban fortification projects south of the Haemus in
southern Thrace and Macedonia (notably at Nicoth
th
Fig.26Zikideva (obl. Veliko Tarnovo/BG). Late 5 or 6 -century
polis ad Nestum, Apollonia-Sozopolis, Selymbria,
granary in the centre of the settlement. (After Ditchev 2006).
Constantinople, Thessalonica, Stobi, Dion and Actia
Nicopolis). At the same time, villas and forts were
transformed into villages, often fortified, while the quality of building throughout the Danube provinces fell
precipitously71. The agricultural, civil and military landscape of the Roman Lower Danube was approaching
its end.
The transformation of the Roman military system from the 380s on was probably instrumental to the
change. As the regular army forces of the Danube were extensively replaced by communities of warriorfarmers (limitanei and foederati), thecentury-long interdependence between villa-based landownership and
military units was broken. With the departure of the regular army, one of the main components of the Roman institutional and economic networks was removed, followed by its main civilian partner, the villa-based
landownership. The result was the disintegration of settlements embodying these two societal domains,
namely forts, villas and cities, and their replacement by a society of warrior farmers whose mark on the
archaeological record was small fortified settlements.
The late 5th and 6th centuries

The departure of the rebellious Ostrogothic leaders and the effective dissolution of the autonomy of the
foederati allowed a tighter control over the Danube provinces in the 490s and better conditions for the development and restoration of the Roman economy and society. The empire invested heavily on building; for
the first time since the 4thcentury, urban fortifications and forts were rebuilt on a scale and at an intensity
suggestive of a generously funded and consistently planned building drive. Unlike the Theodosian period, the
Anastasian and Justinianic building schemes put great emphasis on cities and forts of the earlier periods, and,
quite remarkably, they even produced a number of new cities like Zikideva, Iustiniana Prima (Cariin Grad;
Jablaniki okrug/SRB), and the new fortified section of Nicopolis ad Istrum. The aim was the restoration of
Roman power, with its institutional and material infrastructure, to levels preceding the disruptions of the
5thcentury. In the administrative sphere, this period was marked by the reorganisation of provincial administration through the foundation of Iustiniana Prima as the prospective new capital of Dacia, and the creation
of the quaestura exercitus72. In this framework, warehouse building seems to have resumed, for the last time
in Antiquity. We currently know seven sites that received warehouses in the late 5th and 6th centuries.
71 On
the decline of villas, see: Henning 1987, 35-41; Poulter

2004.
684
72 For
these changes see: Sarantis 2013, 787-790; Torbatov 1997;

Curta 2002; Gkoutzioukostas/Moniaros 2009.
Fig.27 Tzoides (obl. Sliven/BG). Southwest corner of the fortification with 6th-century horreum (H). (After Shtereva 2006).
A horreum stood in the central-north part of the late-antique city of Zikideva in Moesia Inferior (Veliko
Turnovo/BG; fig.29, 39), which can be dated to the late 5th or 6th centuries when the city was founded
and fortified. Simply designed and modest in size (12m6m), the building had rectangular pillars probably
supporting a raised floor (fig.26)73. In the same region as Zikideva, geophysical survey and excavation at
the southern fortification of Nicopolis ad Istrum have located a series of oblong buildings (c. 6m80m), at
least parts of which could be horrea (fig.29, 22)74. To the south of the Haemus, two two-aisled warehouses
of the late 5th or 6th century (8m55m) have been excavated at the south-west corner of the fortified
settlement of Tzoides (Sliven/BG; fig.29, 32) in the province of Haemimontus (fig.27). Tzoides/Sliven is
very close to Cabyle, which, as we saw earlier, was an important gathering centre with large horrea in the
4thcentury75.
Turning to north Illyricum, similar storehouses to those of Tzoides, though smaller (c. 9m29m), have been
excavated northwest of the circular piazza of Iustiniana Prima (fig.28; 29, 5). The discovery of a mill and
bread-oven next to these buildings support their identification as granaries76. More or less contemporary
seems to be an also two-aisled warehouse with an outer pier-supported portico found within the fort of
Taliata (Veliki Gradac, Serbia; fig.29, 34). It was built during the restoration of the fort under Justinian77.
73
Dintchev 2005, 285.

Poulter 1995, 40f. 201.
75 Shtereva 2006, 51. The site is visible on Google Earth, coordinates: 42 31 35 N, 22 20 00 E.
76 Reports by D. Mano-Zisi in: Starinar 3-4, 1952-1953, 165; Starinar 5-6, 1954-1955, 158-159. 164. Kondi/Popovi 1977,
71-74; Bavant 1984, 276. Recent excavation on the same site
74
revealed a new warehouse of the same type: Ivanievi (forthcoming). Flour and bread production next to granaries was
probably a usual practice, since relevant installations have been
located next to horrea at Keszthely-Fenkpuszta (Kom. Zala/H;
belonging to the 4thcentury) and at Palmyra (tetrarchic): Heinrich-Tamska 2011, 681; Kowalski 1998, 203f.
77 Mladenovi 2012, 219 cat. nr. 1655.
685
Finally, 6th-century warehouses housing jars were

found in Macedonia, in the fortified settlements of
Vinica (MK; fig.29, 37) and Louloudies (near Kitros;
periphria Kentrikes Makedonias; fig.29, 14). Both
sites have been interpreted as episcopal centres by
their excavators, although only Louloudies can be
conclusively shown to have been a site of ecclesiastical character; Louloudies is a compact quadriburgium enclosing a basilica, a residential complex and
buildings for storage and workshops. The excavations of the site also produced quantities of LR2 amphorae which are thought to have been used for the
transportation of annonary olive oil78.
These are the largest and strongest warehouses/
granaries built in the Balkans after an interval of
almost two centuries. Horrea like those of Nicopolis, Sliven, Veliki Gradac, and Cariin Grad, demonFig.28 Iustiniana Prima (Cariin Grad; Jablaniki okrug/SRB).
strate that the tradition of designing and building
Circular piazza with warehouse to the northwest (H). (After Bathese structures continued with only insignificant
vant 1984).
changes79. Of course, the number of the examples
is very small in comparison to the 4thcentury. This is
significant given the fact that none of the earlier horrea (except perhaps Capidava and Callatis in Scythia
Minor) seems to have remained functional or to have been rebuilt in the 6thcentury. However, this impression may be the product of the current state of research and thus could change in the future. It is
no coincidence that the Yantra region features so prominently throughout this survey: 4th-century horrea are known from Nicopolis and Novae, 5th-century ones from Dichin and Iatrus, and 6th-century ones
from Zikideva and perhaps Nicopolis. This shows that the area consistently kept its importance in the
supply-network despite changing conditions, and that new storage facilities were built in every period.
This continuity was most probably not exceptional. What is exceptional is the intensity of archaeological
attention the Yantra region has received, which has rendered its diachronic role in the local supply network
so visible.
It must also be pointed out that most of these warehouses are found on sites that had the official status
of a city (Nicopolis ad Istrum, Zikideva, Tzoides, Iustiniana Prima), suggesting that the cities once again
became centres of the annona network. Warehouse building in cities seems to have resumed together
with the rebuilding of urban fortifications in clear contrast with the Theodosian period when, as we saw
earlier, newly-built warehouses and fortifications are known only from small fortified settlements. Since the
dominant civic institution of the 6thcentury was the church, and the sites discussed here are attested as
episcopal sees, it can tentatively be suggested that the ecclesiastical authorities were now involved in the
process of collecting military supplies. The power of the church as a landowning institution is reflected in
78
On Vinica: Mikuli 2002, 256f. cat. nr. 169. The site is visible
on Google Earth, coordinates: 425238 N, 223009 E. On Louloudies: Karagiorgou 2001, 143; Poulter et al. 1998, 463f. 483485. 505f.; Marke 2004. The site is visible on Google Earth,
coordinates: 402036 N, 223604 E
686
79 Very
similar buildings of a slightly later date were discovered at

Recopolis in Spain. Although initially interpreted as a palace,
it was recently proposed to interpret them as granaries. Arce
2011, 296f.
Fig.29Sites of late Roman horrea in the Balkans. (Map M. Athanson).
a novel of 544 which gave special permission to the bishops of Tomis and Odessus to alienate ecclesiastical property for the redemption of captives80. The church probably managed substantial estates of land
and may have been a partner of the state in supplying the troops. The presence of granaries just next to
the fortified acropolis of Cariin Grad, which perhaps protected the headquarters of the most important
churchman of north Illyricum, could support this view. Similarly, the warehouse at the fortified ecclesiastical
centre of Louloudies is a witness to the power of bishops as managers of estates in Macedonia. Bishops are
otherwise epigraphically known to have built granaries in the same period in the cities of Arethusa (Rastan;
gouv. Homs/SYR) and Tella-Constantina (Viranehir; anlurfa ili/TR) in Mesopotamia81.
80
81
Jones 1964, 896f.; Nov. Iust. 120 (544), 9.

Arethusa: IGLS 2081. Tella/Constantina: Humann/Puchstein
1890, 405; Greatrex/Lieu 2002, 240-243. For a plan of Tella/
Constantina and earlier bibliography on the city, see: KeserKayaalp 2013; Mundell Mango 2000, 192f. fig.9.
687
Despite the efforts of Anastasius and Justinians governments, the recovery of productivity in the Balkans
was slow and difficult. Anastasius ordered that the emergency measure of coemptio should be permanently
applied in the diocese of Thrace, because frequent barbarian incursions made it difficult for supplies to
be produced and collected in normal fashion. The situation probably remained unchanged in the times of
Justinian, since the relevant decree was included in the Codex Justinianus. Justinians decision to extend the
validity of this law to the whole of Thrace and Illyricum demonstrates that the woes of the Moesian farmers
were wide-spread. It seems that the hardships echoed in these laws were not isolated, but rather episodes
of a prolonged period of impoverishment, aggravated by the behaviour of the Roman landed class: Justinian also intervened for the return of fields to poor peasants in Moesia Inferior, which had been usurped by
landowners in exchange for usurious loans82.
Conclusions
These are the most significant examples of public warehouses known from the Balkan lands of the East Roman Empire between the late 3rd and the 6th centuries. Their architectural and structural features and the
finds they produced could be only cursorily discussed here, but I hope to have shown that these buildings
are reliable witnesses to the built infrastructure of the late Roman supply-network of the army, its functions,
geographical background and diachronic transformations.
The current state of the evidence allows the reconstruction of a picture that can be summarised as following. The ambitious effort of the empire under the Tetrarchs and Constantine to secure a well-equipped supply network for the Danube army produced a large number of horrea in the late 3rd and early 4th centuries.
The revival and building of large villas and the involvement of cities as principal gathering bases suggest that
the production of supplies was now organised in a highly centralised way with the involvement of the civil
authorities and great landowners. However, it seems that the building of public warehouses lost much of its
importance after the mid-4thcentury, and declined even further after the wars and crises of the 380s. Great
estates and villa-based agriculture collapsed, leaving a gap that may have been partly filled by small fortified
settlements83. The role of cities as annonary centres also declined, and all the urban horrea gradually fell
out of use. A new building drive, reflected in the few late 5th- and 6th-century warehouses, probably aimed
to restore the supply-network to its initial effectiveness, but with limited effect.
One may anticipate that this picture will be altered by future research, but it is hard to predict whether the
broader pattern emerging from this survey will change substantially. The rarity of horrea in the 5th and 6th
centuries is problematic and may be due to the fact that the 4thcentury is better represented in our archaeological record, since the most extensively excavated sites are cities, forts and villas with a continuous
life from the Principate. The settlement landscape of the 5th and 6th centuries, and its principal elements,
have been less extensively investigated, and it is likely that several examples of warehouses from that period still await excavation the discoveries at Dichin confirm that very clearly. Yet it seems unlikely to me
that the numbers and size of early 4th-century horrea will be surpassed, since they probably represent a
coemptio: Jones 1964: 840; Cod. Iust. 10.27.1, 10.27.2 (

5-10) (decree of Anastasius, 491-505). On the landowners of
Moesia Inferior: Nov. Iust. 32-34 (535).
82 On
688
83
Poulter 2007b, 92-94.
true high point in this domain of building, in the context of a great investment by the state on the system
of the military annona.
Appendix: a note on the Eastern Frontier

In the current state of our knowledge, a survey of late antique horrea from the provinces of the Eastern
frontier yields much poorer and less coherent results than those deduced from the Balkans. Unlike the many
examples known from Europe, the securely identified late-antique warehouse/granaries in the East are very
few. Nevertheless, the written sources confirm that urban horrea were a necessary kind of public building,
especially in times of war: warehouses were among the public buildings erected in the new cities of Theodosiopolis in Armenia (built under Theodosius II) and Dara-Anastasiopolis in Mesopotamia (under Anastasius)84; in 531, Justinian sent the former praetorian prefect Demosthenes with funds to build granaries in
the cities of the East, preparing them for the imminent war with Persia85.
A late-antique warehouse used to be visible until the early 20thcentury at Tella-Constantina in Mesopotamia. The same building is most probably commemorated by an inscription dating from 543, which names a
bishop as the commissioner of its construction. A bishop also built a warehouse at Arethusa in Syria in 49886.
The archaeologically investigated examples of horrea in the East are few. The best known and most extensively excavated and studied ones are those of Caesarea Maritima. These include an interesting variety of
buildings, some of which are most probably to be associated with the exports of annonary products from
Palestine, and some seem to be privately owned87.
Examples of warehouses from military sites include a warehouse at the south-east corner of the fort of elLejjun, one at the south end of Diocletians fort at Palmyra and two tri-partite horrea near the main gate of
the fort of Nag al-Hagar. All of them date from the tetrarchy or the early 4thcentury. The Palmyrene example
seems to have been rebuilt under Justinian88.
These few examples do not represent the actual situation in an area where military presence was heavy
throughout late antiquity, and where the annona functioned intensively and uninterruptedly. That so few
granaries are known should rather remind us of how much archaeological work remains to be done on the
Roman frontier in the East.
Acknowledgements
I wrote this paper as a Research Fellow at Ko University Istanbul, funded by the Scientific and Technological Research Council
of Turkey. Aspects of it were presented and discussed at seminars
of the Rmisch-Germanisches Zentralmuseum in Mainz, the Open
University of Cyprus and Oxford University. For their invitation and
hospitality, I am grateful to Professor Falko Daim and Dr Benjamin

Fourlas in Mainz, and to Professor Georgios Deligiannakis in Nicosia. I also wish to thank Dr Byan Ward-Perkins, Professor Jim Crow
and Dr Alexandra Teodor for reading the text and for making useful
suggestions. Dr Michael Athanson produced the map of figure 29.
84 The
85 Malalas
granaries complex of Theodosiopolis (Erzurum/TR) was

known as the Augusteum: Moses of Chorene 3.59, quoted by
Adontz 1970, 119). Dara: Ps.-Zacharias of Mytilene, 7.6 (166)
(see Greatrex et al. 2010). Cyril and Marlia Mango believed that
the complex of vaulted oblong halls preserved on the west walls
of Dara were horrea: Mango 1985, 40 note 20; Mundell Mango
2000, 193. Recent excavation, however, revealed a fountain on
the south side of the building, suggesting that the structure was
probably a cistern.
18.63 (467).
note 81.
87 Patrich 1996; Stabler et al. 2008.
88 El-Lejjun: Crawford 2006, 235-240; Gregory 1997, 356. Palmyra: Kowalski 1998, 203; Intagliata 2014, 189-190. Nag alHagar: Wareth/Zignani 1992, 202.
86 See
689
Sources
Cod. Iust.: Codex iustinianus. Recognovit et retractavit Paulus
Krueger = Corpus Iuris Civilis, v. 2 (Berolini 1877).
Hierokl., Synekdemos: Le synekdmos dHirokls et Lopuscule
gographique de Georges de Chypre: texte, introduction,
commentaire et cartes par Ernest Honigmann (Bruxelles 1939).
Malalas: The chronicle of John Malalas. Translation by E. Jeffreys / M.
Jeffreys / R. Scott (Melbourne 1986).
Nov. Iust.: Novellae. Recognovit Rudolf Schoell; opus Schoellii

morte interceptum absolvit Guilelmus Kroll = Corpus Iuris Civilis,
v. 3 (Berolini 1895).
Themist. or.: Themistius, Orationes. Recensuit H. Schenkl, opus
consummavit G. Downey (Lipsiae 1965).
Zosimus: Zosime. Histoire nouvelle. Traduit et comment par
F.Paschoud (Paris 1979).
Moses of Khorene: Mose de Khorne; nouvelle traduction de

larm
nien classique par Annie et Jean-Pierre Mah (daprs
Victor Langlois); avec une introduction et des notes (Paris 1993).
References
Adontz 1970: N. Adontz, Armenia in the period of Justinian. The
political conditions based on the Naxarar sytem [translated with
partial revisions, a bibliographical note, and appendices by N. G.
Garsoan] (Lisbon 1970).
Arce 2011: J. Arce, Horrea y aprovisionamiento en Hispania (ss. IVVI). In: J. Arce / B. Goffeaux (eds), Horrea dHispanie et de la
Mditerrane romaine (Madrid 2011) 287-297.
Atanasova 2005: J. Atanasova (ed.), . . Razkopki i Prouchvanija 33 (Sofia 2005).
Barnea 1979: A. Barnea (ed.), Tropaeum Traiani 1. Cetatea. Institutul de arheologie Bucureti. Biblioteca de arheologie 35
(Bucureti 1979).
Barnea 1968: I. Barnea, Perioada Dominatului (Sec. IV-VII). In:
I.Barnea / R. Vulpe, Din Istoria Dobrogei 2. Romanii la Dunrea
de Jos. Biblioteca Historica Romaniae 4 (Bucureti 1968) 369556.
1977: I. Barnea, Les monuments palochrtiens de Roumanie.
Sussidi allo studio delle antichit cristiane 6 (Citt del Vaticano
1977).
Bavant 1984: B. Bavant, La ville dans le nord de lIllyricum. Pannonie, Msie I, Dacie et Dardanie. In: Villes et peuplement dans
lIllyricum protobyzantin. Colloque, Rome 1982. Collection de
lcole Franaise de Rome 77 (Rome 1984) 245-288.
van Berchem 1937: D. van Berchem, Lannone militaire dans
lempire Romaine au IIIe sicle. Mmoires de la Socit nationale
des antiquaires de France 8, 1937, 117-202.
Binev 2003: M. Binev, Montana. In: R. Ivanov (ed.), Rimski rannovizantijski selite v Blgarija [Roman and Early Byzantine Settlements in Bulgaria] (Sofia 2003) 160-182.
Borhy 1996: L. Borhy, Non castra sed horrea Zur Bestimmung
einer der Funktionen sptrmischer Binnenfestungen. Bayerische
Vorgeschichtsbltter 61, 1996, 207-224.
Boskovi et al. 1974: D. Boskovi / N. Duval / P. Gros / V. Popovi
1974, Recherches archologiques Sirmium. Campagne FrancoYougoslave de 1973. Mlanges de lcole Franaise de Rome.
Antiquit 86, 1974, 597-656.
Brandl/Vasi 2007: U. Brandl / M. Vasi (eds), Roms Erbe auf dem
Balkan: sptantike Kaiservillen und Stadtanlagen in Serbien.
Sonderbnde der Antiken Welt (Mainz 2007).
690
Bresson/Drew-Bear/Zuckerman 1995: A. Bresson / T. Drew-Bear/

C.Zuckerman, Une ddicace de primipilaires Novae pour la victoire impriale. Antiquit Tardive 3, 1995, 139-146.
Brun 2005: J. P. Brun, Archologie du vin et de lhuile en Gaule
romaine (Paris 2005).
von Blow 2007: G. von Blow, The Fort of Iatrus in Moesia Secunda: Observations on the late Roman defensive system on the
Lower Danube (fourth-sixth centuries). In: A. G. Poulter (ed.), The
transition to Late Antiquity on the Danube and beyond. Proceedings of the British Academy 141 (Oxford 2007) 459-478.
von Blow et al. 2009: G. von Blow / U. Wulf-Rheidt / T. Schler /
M. Opelt / G. Breitner, Das deutsch-serbische Gemeinschaftsprojekt Romuliana-Gamzigrad. Bericht ber die Arbeitskampagnen 2004-2007. Germania 87, 2009, 113-116.
Carl 2007: F. Carl, Tu tantum praefecti mihi studium et annonam
in necessariis locis praebe: prefettura al pretorio e annona militaris nel III secolo d.C. Historia 56/1, 2007, 82-110.
Carri 1994: J. M. Carri, Diocletien et la fiscalit. Antiquit Tardive
2, 1994, 33-64.
erkov 1970: E. erkov, Municipium DD kod Soanice. Dissertationes et Monographiae 10 (Beograd 1970).
Condurachi 1957: E. Condurachi, Histria lpoque du Bas-Empire
daprs les dernires fouilles archologiques. Dacia 1, 1957, 245263.
1971: E. Condurachi, Problema unor basilici cretine de la Histria
i Callatis. Pontica 4, 1971, 173-189.
Crawford 2006: P. Crawford, The Horreum at el-Lejjun (Area M).
In: S. T. Parker (ed.), The Roman Frontier in Central Jordan: Final
Report on the Limes Arabicus Project, 1980-1989. Dumbarton
Oaks Studies 40 (Washington 2006) 235-240.
Crow 2004: J. Crow, Housesteads: a Fort and Garrison on Hadrians
Wall (Stroud 2004).
2007: J. Crow, Amida and Tropaeum Traiani: a comparison of
late antique fortress cities on the Lower Danube and Mesopotamia. In: A. G. Poulter (ed.), The transition to Late Antiquity on the
Danube and beyond. Proceedings of the British Academy 141
(Oxford 2007) 435-455.
Curta 2002: F. Curta, Quaestura Exercitus Iustiniani: the Evidence of
Lead Seals. Acta Byzantina Fennica 1, 2002, 9-26.
Dintchev 1997: V. Dintchev, [Roman Villas in the Present-Day Bulgarian

Lands] (Sofia 1997).
Gissinger 2002: B. Gissinger, Recherches sur le site fortifi de Strasbourg durant lantiquit tardive: le castrum dArgentoratum.
BAR International Series 1024 (Oxford 2002).
1999: V. Dintchev, Sur la charactristique de Iatrus (deuxime

moiti du IVe- dbut du Ve s.). In: G. von Blow / A. Mileva (eds),
Der Limes an der unteren Donau von Diokletian bis Heraklios
(Sofia 1999) 165-174.
Gkoutzioukostas/Moniaros 2009: A. E. Goutzioukostas / X. M.

Moniaros,
(527568): Quaestura Iustiniana Exercitus.
22 (Thessalonike 2009).
2003: V. Dintchev, Scretisca

[The late Roman residence Scretisca and the early Byzatine settlement
]. 30 (Sofia 2003).
2005: V. Dintchev,
Thracia Dacia [Late antique public granaries from Thracia and
Dacia]. In: M. Milcheva (ed.), Stephanos Archaeologicos in honorem professoris Ludmili Getov. Studia Archaeologica Universitatis Serdicensis Supplement 4 (Sofia 2005) 277-285.
2007: V. Dintchev, The fortresses of Thrace and Dacia in the Early
Byzantine period. In: A. G. Poulter (ed.), The transition to Late
Antiquity on the Danube and beyond. Proceedings of the British
Academy 141 (Oxford 2007) 478-546.
2009a: V. Dintchev,
: Philippopolis Nicopolis ad Istrum. In B. Petrunova / A. Aladzhov /
E. Vasileva (eds), Laurea in honorem Margaritae Vaklinova (Sofia
2009) 63-81.
2009b: V. Dintchev, Scythia
Moesia Secunda. In: Ch. Popov / A. Tenchova,
(Sofia 2009) 414-448.
2010: V. Dintchev, The Northwest Sector of the Fortress Gradishteto near the village of Dichin, Veliko Turnovo Region. In:
Stephanos Archaeologicos in honorem Professoris Stephcae
Angelova. Studia Archaeologica Universitatis Serdicensis 5 (Sofia
2010) 449-474.
Dhle 1999: B. Dhle, Beobachtungen zur Bebauung und Bauabfolge im Westabschnitt von Iatrus (Perioden A bis D). In: G. von
Blow / A. Mileva (eds), Der Limes an der unteren Donau von
Diokletian bis Heraklios (Sofia 1999) 143.
Dremsizova-Nelchinova 1984: Ts. Dremsizova-Nelchinova,
. , .
11 (Sofia 1984).
Greatrex/Lieu 2002: G. Greatrex / S. N. C. Lieu, The Roman Eastern

Frontier and the Persian Wars II: 363-630 A Narrative Sourcebook (London, New York 2002).
Greatrex et al. 2010: G. Greatrex / R. R. Phenix / C. B. Horn / S.
P. Brock / W. Witakowski, The Chronicle of Pseudo-Zachariah
Rhetor: Church and War in Late Antiquity. Translated Texts for
Historians 55 (Liverpool 2010).
Gregory 1997: S. Gregory, Roman Military Architecture on the Eastern Frontier (Amsterdam 1997).
Grinter 2007: P. Grinter, Seeds of Destruction: Conflagration in the
Grain Stores of Dichin. In: A. G. Poulter (ed.), The transition to
Late Antiquity on the Danube and beyond. Proceedings of the
British Academy 141 (Oxford 2007) 281-285.
Griswold 1999: J. W. Griswold, A guide to Medieval English Tithe
Barns (Portsmouth 1999).
Heinrich-Tamska 2011: O. Heinrich-Tamska, Die sptrmische Innenbefestigung von Keszthely-Fenkpuszta: innere Chronologie
und funktioneller Wandel. In: O. Heinrich-Tamska (ed.), Keszthely-Fenkpuszta im Kontext sptantiker Kontinuittsforschung
zwischen Noricum und Moesia. Castellum Pannonicum Pelsonense 2 (Rahden/Westf. 2011) 653-702.
Henning 1987: J. Henning, Sdosteuropa zwischen Antike und
Mittelalter. Archologische Beitrge zur Landwirtschaft des
1. Jahrtausends u.Z. Schriften zur Ur- und Frhgeschichte 42
(Berlin 1987).
Herrmann/Wachtel 1979: J. Herrmann / K. Wachtel, Stand und
Probleme der Ausgrabungen in Krivina nach den Grabungskampagnen 1966-1973. In: Iatrus-Krivina 1. Schriften zur Geschichte
und Kultur der Antike 17 (Berlin 1979) 9-25.
Humann/Puchstein 1890: K. Humann / O. Puchstein, Reisen in Kleinasien und Nordsyrien (Berlin 1890).
Dyczek 1997: P. Dyczek, New late Roman horreum from sector IV

at Novae. In: A. Biernacki / P. Pawlak (eds), Late Roman and early
Byzantine cities on the lower Danube from the 4th to 6thcentury
A.D. (Poznan 1997) 87-94.
Intagliata 2014: E. E. Intagliata, Late Antique and Early Islamic Palmyra/Tadmur. An Archaeological and historical reassessment
[unpubl. diss. Univ. Edinburgh 2014].
2002: P. Dyczek, A late Roman settlement on the site of the vale

tudinarium at Novae. In: L. Ruseva-Slokoska / R. T. Ivanov / V.
Dintchev (eds), The Roman and late Roman City (Sofia 2002)
130-133.
Ivanievi (forthcoming): V. Ivanievi, Main Patterns of Urbanism

in Cariin Grad (Justiniana Prima). In: E. Rizos / A. Ricci (eds.),
New Cities in Late Antiquity. Documents and Archaeology. Bi
bliothque de lAntiquit Tardive (Turnhout forthcoming).
Eiden/Mylius 1949: H. Eiden / H. Mylius, Untersuchungen an den

sptrmischen Horrea von St. Irminen in Trier. Trierer Zeitschrift
18, 1949, 73-106.
Ivanov/Ivanov 1994: T. Ivanov / R. Ivanov, Nicopolis ad Istrum (Sofia

1994).
Fernndez Ochoa/Morillo/Salido Domnguez 2011: C. Fernndez Ochoa / A. Morillo / J. Salido Domnguez, Ciudades amuralladas y annona militaris durante el Bajo Imperio en Hispania:
una cuestin a debate. In: J. Arce / B. Goffeaux (eds), Horrea
dHispanie et de la Mditerrane Romaine (Madrid 2011) 265285.
Johnson 1983: S. Johnson, Late Roman Fortifications (London

1983).
Jones 1964: A. H. M. Jones, The later Roman Empire 284-602: a
social economic and administrative survey (Oxford 1964).
Karagiorgou 2001: O. Karagiorgou, LR2: a container for the military annona on the Danubian border? In: S. Kingsley / M. Decker
691
(eds), Economy and Exchange in the East Mediterranean during

Late Antiquity (Oxford 2001) 129-166.
Karayannopoulos 1958: J. Karayannopoulos, Das Finanzwesen des
byzantinischen Reiches (Mnchen 1958).
Keser-Kayaalp 2013: E. Keser-Kayaalp, The Church of the Virgin at
Amida and the martyrium at Constantia: two monumental centralised churches in late Antique Northern Mesopotamia. Olba
21, 2013, 411-418.
Kissel 1995: T. Kissel, Untersuchungen zur Logistik des rmischen
Heeres in den Provinzen des griechischen Ostens (27 v.Chr. - 235
n.Chr.). Pharos 6 (Sankt Katharinen 1995).
Kokotaki 2003: N. Kokotaki, .
.
17, 2003, 14-15.
Kondi/Popovi 1977: V. Kondi / V. Popovi, Cariin Grad:
utvreno naselje u vizantijckom Iliriku (Beograd 1977).
Konrad 1999: S. Konrad, Zur Typologie und Funktionsbestimmung der Amphoren aus dem Kastell Iatrus. In: G. von Blow/
A.Mileva (eds), Der Limes an der unteren Donau von Diokletian
bis Heraklios (Sofia 1999) 175-188.
Korakevik 2002: D. Korakevik, Scupi: gradska teritorija (Skopje
2002).
Kovacheva 1980: T. Kovacheva,
. In: (Pleven 1980) 22-23.
Kovacheva/Gerasimova-Tomova 1983: T. Kovacheva / V. Gerasimova-Tomova, . 2, 1983, 105117.
Kowalski 1998: S. P. Kowalski, The camp of Legio I Illyricorum in
Palmyra. Novaensia 10, 1998, 189-209.
Kuzmanov 1991: G. Kuzmanov, Ein sptantikes Wirtschaftsgebude (Objekt XVIII). In: Iatrus-Krivina 4. Schriften zur Geschichte
und Kultur der Antike 17 (Berlin 1991) 112-116.
ajtar 2013: A. ajtar, A newly discovered Greek inscription at Novae (Moesia Inferior) associated with pastus militum. Tyche 28,
2013, 97-111.
Lenski 2002: N. Lenski, Failure of Empire: Valens and the Roman
State in the Fourthcentury AD (Berkeley, Los Angeles, London
2002).
Luik 2001: M. Luik, Rmische Wirtschaftsmetropole Trier. Trierer
Zeitschrift 64, 2001, 245-282.
Mackensen 1999: M. Mackensen, Late Roman fortifications and
building programmes in the province of Raetia: the evidence of
recent excavations and some new reflections. In: J. D. Creighton/
R. J. A. Wilson (eds), Roman Germany: Studies in Cultural Interaction. Journal of Roman Archaeology Supplement 32 (Portsmouth 1999) 199-244.
Madzharov 1993: K. Madzharov, (Sofia 1993).
Mango 1985: C. Mango, Le dveloppement urbain de Constantinople (IVe VIIe sicles). Travaux et Mmoires du Centre de Recherche dHistoire et Civilisation de Byzance. Monographies 2
(Paris 1985).
Marke 2004: E. Marke,
692
. In: . 5-15
(Athens 2004) 27-45.
Matei 2000: C. Matei, Capidava Cetate 2000. Cronica cercetrilor
arheologice din Romnia. http://cronica.cimec.ro/detaliu.asp?k=
1199&d=Capidava-Topalu-Constanta-Cetate-2000 (27.9.2014).
Mikuli 2002: I. Mikuli, Sptantike und frhbyzantinische Be
festigungen in Nordmakedonien: Stdte, Vici, Refugien, Kastelle.
Mnchner Beitrge zur Vor- und Frhgeschichte 54 (Mnchen
2002).
Miloevi 2011: G. Miloevi, A residential complex at Mediana:
the architectural perspective. In: G. von Blow / H. Zabehlicky
(eds), Bruckneudorf und Gamzigrad. Sptantike Palste und
Grossvillen im Donau-Balkan-Raum. Akten des internationalen
Kolloquiums in Bruckneudorf vom 15. bis 18. Oktober 2008. sterreichisches Archologisches Institut, Sonderschrift 45 (Bonn
2011) 168-169.
Mirabella Roberti 1965: M. Mirabella Roberti, Ledificio romano nel
Patriarcato, supposto palazzo imperiale di Aqiuileia. Aquileia
Nostra 36, 1965, 46-71.
Mitthof 2001: F. Mitthof, Annona Militaris. Die Heeresversorgung
im sptantiken gypten. Ein Beitrag zur Heeresgeschichte des
rmischen Reiches im 3. bis 6. Jh. n.Chr. Papyrologica Florentina
32 (Firenze 2001).
Mladenovi 2012: D. Mladenovi, Urbanism and Settlement in the
Roman Province of Moesia Superior. BAR International Series
2367 (Oxford 2012).
Mulvin 2002: L. Mulvin, Late Roman Villas in the Danube-Balkan
Region. BAR International Series 1064 (Oxford 2002).
Mundell Mango 2000: M. Mundell Mango, The Commercial Map
of Constantinople. Dumbarton Oaks Papers 54, 2000, 189-207.
Niculescu 2011: G. A. Niculescu, Disciplinary Identity and Autonomy at the Beginnings of Archaeology in Romania. In: G. Klaniczay / M. Werner / O. Gecser (eds), Mutliple Antiquities Multiple
Modernities: Ancient History in Nineteenth century European
Cultures (Frankfurt am Main 2011) 381-405.
Nikolov 2003: D. Nikolov, (IVVII .). In: . Studia
Archaeologica Supplement 2 (Sofia 2003) 318-323.
Opri 2003: I. Opri, Ceramica roman trzie i paleobizantin de
la Capidava n contextul descoperirilor de la Dunrea de Jos (sec.
IV-VI p.Chr.) (Bucureti 2003).
Panhuysen 1996: T. A. S. M. Panhuysen, Romeins Maastricht en zijn
beelden (Maastricht 1996).
Prvan 1912: V. Prvan, Cetatea Tropaeum. Consideraii Istorice
(Bucureti 1996).
Patrich 1996: J. Patrich, Warehouses and Granaries in Caesarea
Maritima. In: A. Raban / K. G. Holum (eds), Caesarea Maritima: a
Retrospective after two Millennia (Leiden 1996) 146-176.
Pensabene 2010: P. Pensabene, Villa del Casale e il territorio di Piazza Armerina tra tardoantico e medioevo. In P. Pensabene (ed.),
Piazza Armerina. Villa del Casale e la Sicilia tra tardoantico e medioevo. Studia Archaeologica 175 (Roma 2010).
Petkovi 2011: S. Petkovi, Gamzigrad-Romuliana in der Zeit nach
dem kaiserlichen Palast. In: In: G. von Blow / H. Zabehlicky (eds),
Bruckneudorf und Gamzigrad. Sptantike Palste und Grossvillen
im Donau-Balkan-Raum. Akten des internationalen Kolloquiums

in Bruckneudorf vom 15. bis 18. Oktober 2008. sterreichisches
Archologisches Institut, Sonderschrift 45 (Bonn 2011) 114-128.
Radoslavova/Dzanev 2003: G. Radoslavova / G. Dzanev, Abritus.

In: R. Ivanov (ed.), Roman and early Byzantine settlements in Bulgaria (Sofia 2003) 110-148.
Petrovi 1977: P. Petrovi, La Forteresse Romaine lembouchure

de la rivire Poreka dans les Portres de Fer. In: J. Fitz (ed.), Akten
des 11. internationalen Limeskongresses, Szkesfehrvr 1976
(Budapest 1977) 259-275.
Rickman 1971: G. Rickman, Roman granaries and store buildings

(Cambridge 1971).
1980: P. Petrovi, O snabdevanju rimskih trupa na erdapskom

limesu. Starinar 31, 1980, 53-63.
1982-1983: P. Petrovi, Poreka Reka, sabirni centar za snabdevanje rimskih trupa u erdapu. Starinar 33-34, 1982-1983, 285291.
1986: P. Petrovi, Timacum Minus und die Kastelle im Timok-Tal.
In: Studien zu den Militrgrenzen Roms 3. 13. Internationaler
Limeskongre Aalen 1983. Forschungen und Berichte zur Vorund Frhgeschichte in Baden-Wrttemberg 20 (Stuttgart 1986)
514-518.
Pollard 2000: N. Pollard, Soldiers, cities and civilians in Roman Syria
(Ann Arbor 2000).
Pons Pujol 2008: L. Pons Pujol, The annona militaris in the Tingitana: observations on the organisation and provisioning of the
Roman troops. In: P. P. Funari / R. S. Garraffoni / B. Letalien (eds),
New perspectives on the ancient world: modern perceptions,
ancient representations. BAR International Series 1782 (Oxford
2008) 145-153.
Poulter 1983/1984: A. G. Poulter, Roman towns and the problem of late Roman urbanism: the case of the Lower Danube.
Hephaistos 5/6, 1983/1984, 109-130.
1995: A. G. Poulter, Nicopolis ad Istrum A Roman, late Roman
and early Byzantine city. Excavations 1985-1992. Journal of Roman Studies, Monograph 8 (London 1995).
1999: A. G. Poulter, Gradishte near Dichin: a new late Roman
fortress on the Lower Danube. In: G. von Blow / A. Mileva
(eds), Der Limes an der unteren Donau von Diokletian bis Heraklios (Sofia 1999) 207-227.
2004: A. G. Poulter, Cataclysm on the Lower Danube: the Destruction of a complex Roman landscape. In: N. Christie (ed.),
Landscapes of change: rural evolutions in Late Antiquity and the
Early Middle Ages (Aldershot 2004) 223-254.
2007a: A. G. Poulter, The transition to Late Antiquity. In: A. G.
Poulter (ed.), The transition to Late Antiquity on the Danube and
beyond. Proceedings of the British Academy 141 (Oxford 2007)
1-50.
2007b: A. G. Poulter, The transition to Late Antiquity on the
Lower Danube: the city, a fort and the countryside. In: A. G.
Poulter (ed.), The transition to Late Antiquity on the Danube and
beyond. Proceedings of the British Academy 141 (Oxford 2007)
51-97.
2013: A. G. Poulter, Dobri Dyal: a late Roman fortress on the
Lower Danube. Journal of Roman Archaeology 26, 2013, 362382.
Poulter/Beckman/Strange 1998: A. G. Poulter / M. Beckman /
P.Strange, Field survey at Louloudies: a new late Roman fortification in Pieria. The Annual of the British School at Athens 93,
1998, 463-511.
Rushworth 2009: A. Rushworth, Housesteads: Roman fort The

grandest station. Excavation and survey at Housesteads (Swindon 2009).
Salido Domnguez 2011: J. Salido Domnguez, Horrea Militaria:
aprovisionamiento de grano al ejrcito en el Occidente del Imperio Romano. Anejos de Gladius 14 (Madrid 2011).
Sarantis 2013: A. Sarantis, Military encounters and diplomatic affairs in the North Balkans during the reigns of Anastasius and
Justinian. In: A. Sarantis / N. Christie (eds), War and warfare in
Late Antiquity: current perspectives. Late Antique Archaeology 8
(Leiden 2013) 759-808.
Sarnowski 2005: T. Sarnowski, Drei sptkaiserzeitliche Statuenbasen aus Novae in Niedermsien. In: M. Mirkovi (ed.), Rmi
sche Stdte und Festungen an der Donau (Beograd 2005) 223230.
2013: T. Sarnowski, Accepta pariatoria und pastus militum. Eine
neue Statuenbasis mit zwei Inschriften aus Novae. Tyche 28,
2013, 135-146.
Schmidt-Hofner 2008: S. Schmidt-Hofner, Reagieren und Gestal
ten. Der Regierungsstil des sptrmischen Kaisers am Beispiel der
Gesetzgebung Valentinians I. Vestigia 58 (Mnchen 2008).
Sena Chiesa 1990: G. Sena Chiesa (ed.), Milano Capitale dellImpero
Romano 286-402 d.C. [exhibition catalogue] (Milano 1990).
Shtereva 2006: I. Shtereva (ed.), - 1.
28 (Sofia 2006).
Srejovi 1982/1983: D. Srejovi, Kasnoantika itita u Maskaru.
Balcanica 13/14, 1982/1983, 35-43.
1993: D. Srejovi (ed.), Roman Imperial towns and palaces in
Serbia (Belgrade 1993).
Stabler et al. 2008: J. Stabler / K. Holum / F. H. Stanley / M. Risser/
A.Iamim, The Warehouse Quarter (Area LL) and the Temple Platform (Area TP), 1996-2000 and 2002 seasons. In: K. G. Holum/
J.Stabler / E. G. Reinhardt (eds), Caesarea Reports and Studies:
Excavations 1995-2007 within the Old City and the Ancient Harbor. BAR International Series 1784 (Oxford 2008) 1-17.
Stanev 1995: D. Stanev, Grabungen im Kastell Iatrus von 1984
bis 1988. In: Iatrus-Krivina V. Schriften zur Geschichte und Kultur
der Antike 17 (Berlin 1995) 123-139.
Suceveanu/Scorpan 1970: A. Suceveanu / C. Scorpan, Stratigrafia
Histriei Romane tirzii n lumina spturilor din 1969 i 1970 n
sectorul central. Pontica 4, 1970, 155-172.
Swan 2007: V. G. Swan, Dichin (Bulgaria): Interpreting the ceramic
evidence in its wider context. In: A. G. Poulter (ed.), The transition to Late Antiquity on the Danube and beyond. Proceedings
of the British Academy 141 (Oxford 2007) 251-280.
Tocilescu 1900: G. Tocilescu, Fouilles et recherches archologiques
en Roumanie (Bucharest 1900).
Torbatov 1997: S. Torbatov, Quaestura exercitus: Moesia Secunda
and Scythia under Justinian. Archaeologia Bulgarica 13, 1997,
78-87.
693
2000: S. Torbatov, The Late Antique city of Zaldapa (Sofia 2000).

2003: S. Torbatov, Zaldapa. In: R. Ivanov (ed.), Roman and early
Byzantine settlements in Bulgaria (Sofia 2003) 96-99.
Tth 2011: E. Tth, Lapidarium Savariense: Savaria rmai feliratos
kemlkei (Szombathely 2011).
Vasi 1999: M. Vasi, Transdrobeta (Pontes) in Late Antiquity. In: G.
von Blow / A. Mileva (eds), Der Limes an der unteren Donau
von Diokletian bis Heraklios (Sofia 1999) 31-33.
Wareth/Zignani 1992: U. A. Wareth / P. Zignani, Nag al-Hagar: a

fortress with a palace of the Late Roman Empire. Second preliminary report. Bulletin de lInstitut Franais dArchologie Orientale
92 (1992) 185-210.
Wendel 2005: M. Wendel, Karasura III: Die Verkehrsanbindung in
Frhbyzantinischer Zeit (4.-8. Jh. n.Chr.). Schriften des Zentrums
fr Archologie des Schwarzmeerraumes 6 (Langenweissbach
2005).
Visy 2003: Z. Visy (ed.), The Roman army in Pannonia: an archaeological guide to the Ripa Pannonica (Pcs 2003).
Whateley 2013: C. Whateley, El-Lejjun: logistics and localisation on

Romes eastern frontier in the 6th c. AD. In: A. Sarantis / N. Christie (eds), War and warfare in Late Antiquity: current perspectives.
Late Antique Archaeology 8 (Leiden 2013) 893-924.
Vladkova 2007: P. Vladkova, The Late Roman agora and the state
of civic organisation. In: A. G. Poulter (ed.), The transition to Late
Antiquity on the Danube and beyond. Proceedings of the British
Academy 141 (Oxford 2007) 208-214.
Zahariade 1996: M. Zahariade, The Roman frontier in Scythia Minor

(1980-1994). In: P. Petrovi (ed.), The Roman Limes on the Middle and Lower Danube (Belgrade 1996) 225.
Wachtel 1999: K. Wachtel, Fragment einer Bauinschrift aus den Jahren 340/350, gefunden im sptantiken Kastell Iatrus. In: G. von
Blow / A. Mileva (eds), Der Limes an der unteren Donau von
Diokletian bis Heraklios (Sofia 1999) 195-199.
2006: M. Zahariade, Scythia Minor: a History of a later Roman

Province (284-681) (Amsterdam 2006).
Zusammenfassung / Summary / Rsum

Zentren des sptrmischen Militr-Versorgungsnetzwerkes auf dem Balkan: eine berblicks
studie zu den sptantiken horrea
Der Beitrag ist eine Studie zur Bebauungsinfrastruktur der annona militaris in den Balkanprovinzen des Rmischen Reiches vom spten 3. bis 6. Jahrhundert n.Chr. mithilfe einer Untersuchung der ffentlichen Warenlager (horrea). Durch Ausgrabungen seit dem 19. Jahrhundert sind mehrere sptantike horrea in dieser
Region bekannt, die uns erlauben, die Infrastruktur der annona bemerkenswert detailgenau zu betrachten.
Dieses reichhaltige Architekturmaterial kann unser Verstndnis von der Funktion, Entwicklung und Um
bildungen des Versorgungsnetzwerkes bis zum Ende der Antike grundlegend voranbringen. Die Beispiele
werden chronologisch und gem ihres Siedlungskontextes untergliedert. Die groe Mehrheit datiert ins
spte 3. und 4. Jahrhundert, whrend die Belege aus dem 5. und 6. Jahrhundert wesentlich seltener sind.
Die umfangreiche frhe Gruppe (sptes 3. und 4. Jahrhundert) wird zuerst diskutiert, unterteilt in drei Untergruppen nach ihrem Siedlungskontext (Militrlager und kleine befestigte Siedlungen, Villen und lndliche
Siedlungen, Stdte). Die wenigen jngeren Beispiele werden in eigenen Abschnitten behandelt.
Die ehrgeizigen Bestrebungen des Reiches unter den Tetrarchen und Konstantin, ein gut ausgestattetes Versorgungsnetzwerk fr die danubische Armee sicherzustellen, ergaben eine groe Anzahl an horrea im spten
3. und 4. Jahrhundert. Der Bau bzw. Wiederaufbau groer Villen sowie die Einbeziehung von Stdten als vornehmliche Sammelpltze lsst vermuten, dass die Produktion der Versorgungsgter nun in hochzentralisierter
Weise durch die Einbeziehung der zivilen Autoritten und Grogrundbesitzer organisiert wurde. Jedenfalls
scheint der Bau ffentlicher Lagerhuser nach der Mitte des 4. Jahrhunderts seine Bedeutung eingebt zu haben und nach den Kriegen und Krisen der 380er Jahren weiter niedergegangen zu sein. Grogrundbesitz und
Villenwirtschaft brachen zusammen und hinterlieen eine Lcke, die teilweise von kleinen befestigten Siedlungen gefllt worden sein knnte. Die Rolle der Stdte als Proviantlager ging ebenfalls zurck, und allmhlich
kamen alle stdtischen horrea auer Nutzung. Eine neue Bebauungskampagne, vertreten von den wenigen
Lagerhusern des spten 5. und 6. Jahrhunderts, zielte mglicherweise auf ein Wiedereinrichten des Versorgungsnetzwerkes zu seiner ursprnglichen Leistungsfhigkeit, aber sehr wahrscheinlich mit geringem Erfolg.
694
Centres of the Late Roman Military Supply Network in the Balkans: a Survey of horrea
This article presents a study of the architectual infrastructure of the annona militaris in the Balkan provinces
of the Roman Empire from the late 3rd to the 6th century AD through a survey of public warehouses (horrea).
Excavations since the 19thcentury have revealed several late antique public warehouses in this region, which
allow us to follow the infrastructure of the annona in remarkable detail. This rich architectural material can
substantially promote our understanding of the function, development and transformations of the supplynetwork into the end of Antiquity. The examples are categorised chronologically and according to their
settlement context. The vast majority dates from the late 3rd and 4thcenturies, while those from the 5th and
6th centuries are much fewer. The large early group (late 3rd and 4th centuries) is discussed first, divided into
three subcategories of settlement context (forts and small fortified settlements, villas and rural sites, cities),
while the few later examples are discussed in separate sections.
The ambitious effort of the Empire under the Tetrarchs and Constantine to secure a well-equipped supply
network for the Danubian army produced a large number of horrea in the late 3rd and early 4th centuries.
The revival and building of large villas and the involvement of cities as principal gathering bases suggest that
the production of supplies was now organised in a highly centralised way with the involvement of the civil
authorities and great landowners. However, it seems that the building of public warehouses lost much of its
importance after the mid-4thcentury, and declined even further after the wars and crises of the 380s. Great
estates and villa-based agriculture collapsed, leaving a gap that may have been partly filled by small fortified
settlements. The role of cities as annonary centres also declined, and all the urban horrea gradually fell out
of use. A new building drive, reflected in the few late 5th- and 6th-century warehouses, probably aimed at
restoring the supply-network to its initial effectiveness, but probably to limited effect.
Les centres du rseau de ravitaillement militaire tardo-romain dans les Balkans: une tude synthse sur les horrea tardo-antiques
Cet article tudie linfrastructure btie de lannona militaris dans les provinces balkaniques de lempire
romain de la fin du 3e au 6e sicle ap. J.-C. par une prospection des entrepts publiques (horrea). Depuis le
19e sicle, plusieurs fouilles ont mis au jour des entrepts publiques de lAntiquit tardive dans ces rgions,
permettant de suivre trs en dtail linfrastructure de lannona. Il sagit dun matriau architectural tres
rich, qui permet dapprofondir de manire substantielle notre comprhension du fonctionnement, du dveloppement et des transformations du rseau de ravitaillement jusqua la fin de lAntiquit. Les exemples
sont catalogus chronologiquement et selon le contexte dhabitat. La grande majorit date de la fin du 3e
et du 4e sicle, ceux des 5e et 6e sicles tant bien moins nombreux. On discute dabord le grand groupe
(fin 3e et 4e sicles) subdivis en trois catgories de contextes dhabitat (forts et petits habitats fortifis,
villas et sites ruraux, et les villes), pour aborder ensuite travers diffrentes sections les quelques exemples
plus tardifs.
Les efforts ambitieux dvelopps par lempire sous les Ttrarques et Constantin en vue de scuriser le rseau
de ravitaillement, bien quip, pour larme du Danube ont cr un grand nombre dhorrea la fin du 3e
et au dbut du 4e sicle. La construction ou restitution de grandes villas, ainsi que limplication des villes en
tant que bases principales de stockage, suggrent une production trs centralise des rserves avec la participation des autorits civiles et des grands propritaires. Il semble, cependant, que la construction dentrepts publiques ait perdu beaucoup dimportance aprs le milieu du 4e sicle et dclint encore davantage
aprs les guerres et crises des annes 380. Les grands domaines et lagriculture base sur les villas seffondrrent, laissant un vide combl peut-tre partiellement par de petits habitats fortifis. Le rle des villes
comme centres annonaires dclina galement et les horrea urbains tombrent peu peu en dsutude. Une
695
nouvelle campagne de constructions, reflte par quelques entrepts de la fin du 5e et du 6esicle, visait
probablement restituer au rseau de ravitaillement son efficacit dantan, mais probablement avec peu de
Traduction: Y. Gautier
succs.
696

Bukowski - Bludni Sin

Uploaded by

Copyright:

Available Formats

Bukowski - Bludni Sin

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bukowski - Bludni Sin

Uploaded by

Copyright:

Available Formats

Efthymios Rizos

Centres of the late Roman military supply network

CIL III 4180; Tth 2011, Cat. Nr 8. Translation by the author.

(prov. Udine/I): Mirabella Roberti 1965. Milan: A.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Rickman 1971, 213-250.

the built infrastructure of the annona: Rushworth 2009,

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Fig.3 Castra Martis (Kula; obl. Vidin/BG). Quadriburgium with

1980; Kovacheva/Gerasimova-Tomova 1983, 105.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Petrovi 1977; 1980; 1982-1983.

Srejovi 1982/1983, 35-43. Mladenovi 2012, 170 cat. nr.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Brandl/Vasi 2007, 101. Mladevoni 2012, 44. 76 Nr. 209.

Blow et al. 2009, 113-116. The authors believe that the

Petkovi 2011, 114-128 (with earlier bibliography).

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

III Suppl. 12376. Binev 2003, 160-182. Wendel 2005,

Dintchev 1997, 25f. 32-40. Mulvin 2002, 92. 95f.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Fig.11The villa of Madara (obl. Shumen/BG) and

Srejovi 1982/1983, 35-43. Mladenovi 2012, 124. 151 cat.

Dremsizova-Nelchinova 1984. Mulvin 2002, 90-91. Dintchev

Fig.12Madara villa. Horreum

Horrea of the late 3rd and early 4th centuries in cities

role of the great villas of Spain as annonary producers has

Pensabene 2010, 1-12.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

opolis36. The combination of sophisticated fortifications and horrea in

Ochoa/Morillo/Salido Domnguez 2011; cf. criticism

Fig.14Zaldapa (near Abrit; obl. Dobrich/

3 13734; also see discussion by Poulter 1983/1984, 119;

2000, 37-47; 2003, 96-99. Dintchev 2005, 282f.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Suceveanu in: Histria VI, 1982, 86f. Condurachi 1957;

(gouv. Kerak/JOR): Crawford 2006. Tokod (Kom.

Fig.17 Callatis (Mangalia; jud.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Fig.18 Abrittus (Razgrad/BG). General

1994, 73. 97. Vladkova 2007, 208-214.

1994, 60. 63.

Excavation reports in:

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Fig.20Maronea (Agios Charalambos;

tion in the central Balkans. Rescue excavations in

Fig.21 Serdica (Sofia/BG). Warehouses along the southwest

Dintchev 2005, 279f.

2002, 65-74. Interestingly, a bath was built within

erkov 1970, 14-46; Mladenovi 2012, 24-26. 44. 199 cat.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

Fig.23 Municipium DD (Soanica/RKS). Double

GENERAL REMARKS ON THE TETRARCHIC PERIOD AND THE 4TH CENTURY

The Theodosian Period

4.10.3-4; Lenski 2002, 127-137.

Jahrbuch des Rmisch-Germanischen Zentralmuseums 60 2013

2007b, 87-91; 2004, 244-246; 1999. Swan 2007.

Dintchev 2005, 284f.

Fig.25Iatrus. Early 5th-century horreum (Building XVIII) (see fig.2