Roman Frontier Studies

1979

I =--=
~=
---
' -
....1 1 . -
----

Papers presented to the 1 2 th

International Congress of
Roman Frontier Studies

Edited by
W. S. Hanson and L. J. F. Keppie

Part iii

BA R International Series 7 I (iii)

1980
 B.A.R.
B.A.R., 122 Banbury Road, Oxford OX2 7BP, England

GE i\""ERAL EDITORS

A. R. Hands, B.Sc., :.I.A., D.Phil.

D. R. Walker, :.I.A.

B. A. R. 871, 1980, "Roman Frontier Studies XII".

© The Individual Authors.
Price £28. 00 post free throughout the world. Payments made in currency
other than sterling must be calculated at the current rate of exchange and an
extra 1 O% added to cover the cost of bank charges. Cheques should be made
payable to 'British Archaeological Reports' and sent to the above address.
ISBN 0 86054 080 4
ISSN 0143 3067

BAR publishes monographs and collections of papers on all aspects of archae-

ology and on related subjects such as anthropology, classical and medieval
history and numismatics.
BAR publishes BAR International Series which is concerned with world archae-
ology and BAR British Series which covers the British Isles.
For details of all BAR publications in print please write to the above address.
Information on new titles is sent regularly on request, with no obligation to
purchase.
Volumes are distributed direct from the publishers. All BAR prices are inclu-
sive of r)ostage by surface mail anywhere in the world.

Printed in Great Britain

 CONTENTS

Page

vol. iii. Moesiae, Dacia

48. Rural communities (vici and komai) and their role in the organ-
isation of the limes of Moesia Inferior. A. G. Poulter 729
49. Vom obermCJsischen Limes nach dem Stlden: Via Nova von
Viminiacum nach Dardanien. Miroslava Mirkovic 745
50. Les forteresses du Bas-Empire sur le limes Danubien en
Serb le. Petar Petrovic 7 57
51. Untersuchungen des Forumkomplexes in der Colonia Ulpia
Oescensium (1975 .... 1978). Teofil Ivanov 775
52. Sacidava- an unusual design and construction method on the
later Roman limes. C. Scorpan 787
53. Recent research on the limes of Roman Dacia. N. Gudea 799
54. Les deux camps de Praetorium sur le Limes Alutnaus.
C. N. Vladescu and Gh. Poenaru Bordea 815
55.· Pedites et equites singulares in Dacia. C. L. Baluta 831

Oriens, Africa
56. Research on the Limes Palestinae:· a stocktaking.
Mordeschai Gichon 843
57. Towards a history of the Limes Arabicus. S. Thomas Parker 865
58. The frontier policy of Septimius Severus: new evidence from
Arabia. D. L. Kennedy 879
59. Trade routes to Arabia and the Roman army. B. Isaac 889
60. New research on the Euphrates frontier in Turkey.
J. G. Crow and D. H. French 903
61. The limes in the Kurdish Taurus. Timothy B. Mitford 913
62. Augustus' Egyptian frontier: Q'asr !brim?
W. H. C. Frend 927
63. Signification d'une frontiE:lre: nomades et sedentaires dans
la, zone du limes d'Afrique Pol Trousset 931
64. Un nouveau type d 'unite connu par 1' epigraphie Africaine.
Yann Le Bohec 945
65. Remarques onomastiques sur la liste militaire de Vezereos.
Jean-Marie LassE:lre 955
66. La ferme Romanette, .A"in Benia, .A"in Bent Soltane: fortins
ou fermes fortifiees? Nacera Benseddik 977
67. Carthage: the late Roman defences. C. M. Wells 999

General
68. The political significance of Augustus' military reforms.
Kurt A. Raaflaub 1005

69. Lixae. H. v. Petrikovits 1027

70. Urbanization and the north-west frontier of the Roman Empire.
W. Groenman-van Waateringe 1037
71. Barbarian invasions and frontier politics in the reign of
Gallienus. John W. Eadie 1045
72. Typology and late Roman fortification: the case of the
'Diocletianic type 11 • James Lander 1051
73. Military diplomata and war expeditions. Slobodan Dusanic 1061
74. British auxiliary regiments -origins and early nomenclature.
D. B. Saddington 1071
75. Funde von rClmischen Medaillons zu beiden Seiten des Rhein-
Donau-Limes. Jerzy Wielowiejski 1077
76. Soldaten in der rd'mischen Kunst. GlStz Waurick 1091
77. Zwiebelknopffibeln als kennzeichen von Soldaten auf spat-
rClmischen Denkmfilern. · Heinrick Zabehlicky 1099

Papers given at the Congress but published elsewhere:

J. E. Bogaers, 'King Cogidubnus in Chichester: another reading
of RIB 91', Britanniax (1979), 243-54.
Donald Moore, 'Venta Silurum - a frontier town in Wales'.
D. Baatz, 'Ein Katapult der Legio IV Macedonica aus Cremona',
Rdmische Mitteilungen forthcoming.
M. P. Spiedel, 'The guards of the praepositus', Fundberichte aus
Baden-Wdrttemberg 1980 forthcoming.
List of illustrations Page
Part (iii)
A. G. Poulter;
Fig. 48.1 Distribution of rural communities in Moesia Inferior. 731
Fig. 48.2 Regions in Moesia Inferior. 732

Miroslava Mirkovic
Abb. 49.1 RiSmischen Strassen im Zentralteil der Balkanhalbinsel. 746

Petar Petrovic
Fig. 50.1 Limes des Partes de Fer. 758
Fig. 50.2 Saldum. 761
Fig. 50.3 Donje Butorke, Zanes. 761

Fig. 50.4 Hajducka Vodenica. 763

Fig. 50.5 Mihaj lovac. 763

Teofil Ivanov
Abb. 51.1 Grundriss des Forumkomplexes von Oescus (VR
Bulgarien). 776
Abb. 51.2 Grundriss der Forum basilika in Oescus. 778
Taf. 51.1 Die Basilika von Oescus. 779
Taf. 51.2 Die Schwelle des Mittelschiffs der Basilika. 780
Taf. 51.3 Fries-Architrav des Mittelschiffes der Basilika. 781
Taf. 51.4 Zwei Fragmente von Karyatiden im Vorraum des
Mittelschiffes der Basilika. 782
Taf. 51.5 Karyatidenkopf vom Vorraum. 783

C. Scorpan
Fig. 52.1 Sacidava. 793
Pl. 52.1 Sacidava, Tower C. 788
Pl. 52.2 Sacidava, Tower B with re-used inscription. 788
Pl. 52.3 Sacidava, Tower B with re-used inscription. 789
Pl. 52.4 Sacidava, curtain wall at c. 792
Pl. 52. 5 Brick stamp - LEG XI CL(AUDIA) 795
Pl. 52.6 Brick stamp - LEG I ITA(LICA) 795
Pl. 52.7 Brick stamp - (LEG XI CL) TRAM(ARISCA) 796
Pl. 52.8 Brick with fluent Greek writing. 796

C. N. VU1de scu and Gh. Poenaru Bordea

Fig. 54.1 Releve du camp de Copa:ceni (Praetorium I). 817
Fig. 54.2 Releve du camp de Racovita (Praetorium II). 821
Fig. 54.3 Le limes Alutanus entre Castra Traiana et Caput
Stenarum. 827
Pl. 54.1 L'aile est du camp de Copaceni, entre la porta
praetoria et la tour du nord-est. 820
Pl. 54.2 L 'aile septentrionale du camp de Racovisa. 820
Pl. 54. 3 L' aile occidentale du camp. 822
Pl. 54.4 L'angle de sud-est, avec la saillie du front
renforcee de briques et 1'enceinte oriental e. 822
Pl. 54.5 La tour du nord-ouest. 822
Pl. 54.6 L'enciente meridionale avec la plate forme si~e
entre la tour du sud-est et la porta principalis dextra. 823
Pl. 54.7 L'angle du nord-est aux extremites en brique. 823
Pl. 54. 8 Conduit pour les eaux d 'ecoulement sous 1'enceinte
septentrionale. 823
Pl. 54.9 La porta praetoria avec ses tours interieures. 825
Pl. 54.10 La porta principalis sinistra avec la tour occidentale. 825
Pl. 54.11 Le groupe des pi~ces composant l'oecus, situees
a 1'ouest du praetorium. 825

Closca L. Baluta
Fig. 55.1 Materiaux de construction estampilles. 833
Fig. 55.2 MateriatlX de construction estampilles. 834

Mordechai Gichon
Fig. 56.1 Castellum "Be'er Sheba". 846
Fig. 56.2 The Judean limes in the Flavian period. 849
Fig. 56.3 En Yahav, the burgus. 849
Fig. 56.4 The limes in the Negev in the 2nd and 3rd centuries. 851
Fig. 56.5 The Limes Palestinae from Diocletian to the end of
the Byzantine period. 853

S. Thomas Parker
Fig. 57.1 Antonine limes (northern region): 106-192. 867
Fig. 57.2 Antonine limes (southern region): 106-192. 868
Fig. 57.3 Severan limes (northern region): 193-235. 869
Fig. 57.4 Severan limes (southern region): 193-235. 870
Fig. 57.5 Limes of Diocletian (northern region): 284-450. 872
Fig. 57.6 Limes of Diocletian (southern region): 284-450. 873

D. L. Kennedy
Fig. 58.1 North-east Arabia. 884
Pl. 58.1 Detail of building inscription from Qasr el- 'Uweinid
showing name of governor. 882
Pl. 58.2 Aerial view of Qasr el-Azraq £.· 1928 showing
earlier fort. 882
Benj amin Isaac
Fig. 59.1' The Roman road-system in the Negev and south
Transjordan. 891

J. G. Crow and D. H. French

Fig. 60.1 Euphrates Frontier in Turkey. 904

Fig. 60.2 The Eastern frontier - hypothetical defence

systems. 908

Timothy B. Mitford
Fig. 61.1 The limes in the Kurdish Taurus. 914
Pl. · 61.1 Ruins at Derikhale (Lacotena?). 917
Pl. 61.2 The gorge below Ht1§1Ukani. 919
Pl. 61.3 Fortlet? north of Ha~kento, guarding the Kerefto 9ay. 921
Pl. 61.4 Killik (Barzalo) from the south. 923
Pl. 61.5 Killik (Barzalo) -aqueduct core. 925
Pl. 61.6 Ruins near Taraksu-Iuliopolis? 926

Pol T rousset
Fig. 63.1 Organisation du limes dans le sud tunisien. 932
Pl. 63.1 La muraille de Bir oum Ali 938
Pl. 63.2 La muraille de Bir oum Ali. 939

J.-M. Lassere
Pl. 65.1 ILAfr. 27. Face principale. 973
Pl• 65.2 ILAfr. 27. Cote gauche. 974
Pl. 65.3 ILAfr. 27. Cote droit. 975

Nacera Benseddik
Fig. 66.1 Ruines de l'oued Ouerk. 982
Fig. 66.2 Ruines de Ai'n Benia. 984.
Fig. 66.3 Carte d'une partie de la Mauretanie Cesarienne. 989

C. M. Wells
Fig. 67.1 Carthage, northern sector. Topographic maps
1976-197 8. Excavation in relation to theoretical
street grid. 1000
Fig. 67.2 Carthage, main site. Excavations 1976-1979. 1001
Fig. 67.3 Carthage, northern sector. Section trenches
4B - 4C. 1003

H. v. Petrikovits
Taf. 69.1 Eine Grabinschrift aus Asciburgium (CIL 13, 12075). 1030
James Lander
Fig. 72.1 The core of the type. 1052
Fig. 72.2 Additions to the type. 1053
Fig. 72.3 Possible anomalies. 1055
 48. RURAL COMMUNITIES (VICI AND KOMAI) AND THEIR ROLE
IN THE OF THE LIMES OF MOESLA INFERIOR

A. G. Poulter

and ko~ai) are rarely discussed in classical

role in the Roman Empire is also limited by the scarcity of useful epigraphic
western provinces, and by the diversity of types
different provinces and in different regions within

centuries 'rural
communiti..e;s' identified with c:ertaintv, Although not all of
the settlements can be exactly located, the general location of almost all of
them is sufficiently secure to broadly define regional variations in both charac-
ter and ±8. and 48.
1
these
'rural communities' suggests the division of the province into four distinct re-
gions. Regions 1-3 will be discussed briefly before dealing with region 4 which
contrasts markedly with the others. It will be argued that region 4 is anomal-
ous in that its distinctive pattern of rural settlement is most likely to result
from direct imperial intervention, a policy determined by the peculiar military
problems encountered on the Lower Danube during the early 2nd century A. D.
Region 1: the North Bulgarian Plain from Lom to Ruse
The s the Planina (Haemus) and the flood-,plain
of the Danube are both fertile and well-watered (Gerasimov and Gulbov 1966,
57, 297£. ). Veterans and immigrants from other parts of the Roman Empire
settled around the legionary base at Oescus (Gigen) and in its immediate hinter-
land during the Julio-Claudian period (Gerov 1948-49, 30£. ). The auxiliary
·''"'·"'"~"' at Oescus and Novae (Steklen) vet·-
2.vJ;n::. settlement well to the south the Danube rov 19i11-53,
,, the Haem·:Is the provincial boundary between )"'·"·";;;:,'·:,J.a.
and Thrace, The native population was under military supervision
Julio-Claudian period: a Chwdian inscription records the career C. Baebius
Atticus who had been primipilus of the V Macedonica and l2Eaefectus civitatium
¥oesiae et~ TJ.:eballi~~. (CII-: V 1838=II;~§ 1349), A reorganization created new
'tribes' listed by ptolemy (Geogr. IlL 10, 4) which, in at least three cases,
possessed names identicaJ. with those of military bases on the Danube; Oetenses,
Dimenses, Appiarenses (M6csy 1974, 66L ). Quite probably, some military
control was exercised over the 'tribal' territories by the units whose names
they bore. Certainly, the Montanenses, west of the Iskur, remained subject
to military supervision into the 2nd century A.D. 4 Also, the 2nd century auxili-
ary base at Capidava retained control over an extensive territorium and it may
well be that the auxiliary garrisons In region 1 also maintained control over
simi.lar territoria until M. Aurelius created new municipia in Moesia Inferior. 5

729
Fig. 48. 1 Distribution of rural communities in Moesia Inferior

Key

Trajanic frontier between Thrace and Moesia Inferior

Hadrianic frontier between Thrace and Moe sia Inferior
komai and vici with Thracian names 0
vici vvith Roman names
legionary camp •
[j]
auxiliary· fort
city •
D
1. V. Buteridavensis 23. V. (nr. Tomis)
2. Chora Dagei 24. V. Am laid in a
3. V" . Quintionis 25. (K) Asbolodina
4. V. Par sal (- ?) 26. (K) Ke-
5. V. Secundini 27. (K) Sardes
6. V. Celeris 28. (K. )Val -
,..,
• • Laico s Purgo s 29 • (K) .Keribo stenon
8. V. [-]C 30. (K) Koruolenon
9. V (Neatfrnarea) 31. V. [-]diano
10. V. Clementianensis 32. K. Maskiobrias
11. V. Hi[-] 33. K. Ba[-J
12. V. illmetum 34. V. Saprisara
13. V. (Gilbiori) 35. V. Agatapara
14. V. Scenopesis 36. V. Pereprus
15. V. Rami[dava ?] 37. (V) Giridava
16. V. Vera [br ?]ittiani 38. (V) Perburdava
17. V. Petrus 39. V. Trullensium
18. V. Novus 40. V. Siamus
19. V. Urb(-] 41. V. Bapeni
20. V. Scaptia 42. V. Longinopara
21. V. 'I'urris Muca[poris ?] 43. Varovum Minor
22. V. Narcis[s]iani 44. V. Tautiomo sis
(K. Ermonaktos, beyond the Danube mouth, is not included in Fig. 48. 1 (et
note 2) ).

730
 )~
I 5\
) I
f/

-1

""'""""

PONTUS
( J

Jr- EUXINUS

0 50 I 00
• 0 6 ...
• I :=te-et-===:4 KM
-.:)
w
t\:1

• .I
~

/
_,..-----
......
______ REGION
..,.. 1-
••••
• ••
••
.
•• •• •
. . ... . .. • •

•• ••• • • •• • • • • • • • • • • • • • • • •
........ • ••

F:i.g. 48. 2 Regions in Moesia Inferior (for Key see Fig. 48. 1)
 Within region 1 'rural communities' were few in number and their names
are of Geto-Thracian or part-Greek, part-Thracian origin (Fig. 48.1). The
degree of romanization of even those communities which have left epigraphic
record of their existence is limited. Aurelius Victorinus was princeps vici
Tautiomosis. 2 The vicus Trullensium provides the only evidence for Roman
2
organization with magistri and lapidarii, albeit with Greek names. The
Trajanic foundations of Nicopolis ad Istrum, Marcianopolis and colonia Ulpia
Oescus provided no encouragement to the growth of 'rural communities'. A
Thracian, Aurelius Dyza, refers to his home as the vicus Saprisara within
the regio Nicopolitana (i.e. ficopolis ad I strum). The vicus Pereprus was
assigned to Melta (Lovech). .
Despite extensive Roman settlement and the Trajanic foundation of two
cities and a colonia, 'rural communities' remained few in number and the Geto-
Thracian pattern of settlement seems to have remained largely unaffected by
its inclusion within the Roman province.
Region 2: the North East Angle of Bulgaria
Few Roman roads traversed this dry tract of country (Fig. 48.1). As
late as the mid 19th century, this region was occupied largely by nomadic
pastoralists who migrated from the C arpathians to seek winter pasturages
south of the Danube (Kanitz 1882, vol. 3, 234). The absence of 'rural com-
munities' in this region during the Roman period is therefore hardly surprising.
It reflects a natural reluctance to establish settlements where the water-supply
is severely restricted. The military occupation of the Danubian limes in re-
a
gion 2 may itself be Flavian development. 6 There is no evidence, at least
as yet, that Durostorum (Silistra) was itself a legionary base before the
Trajanic period. 7 The apparent reluctance to extend the limes far beyond the
base of the I Italica at Novae before Trajan may be not unconnected with the
local geography south of the Danube and particularly the absence of a settled
native population, itself a consequence of the local geography and climate
which must have equally affected the pre-Roman population as it affe.cted the
settlement pattern in the 2nd century A .D.
Region 3: the Black Sea Coast from Nesebur to Istria
The territories of both Mesembria (Nesebur) and Odessos (Varna) were
fertile and had good water-supplies whereas, north of Callatis (Mangalia),
the climate is more unfavourable with low rainfall although temperatures are
le~s extreme than in the interior and humidity is higher (Pounds 1969, 559;
Radulescu 1974, 71£. ). During the Late Hellenistic and Roman periods, the
coast was occupied largely by hellenized Thracian and, on the Dobrogean coast,
Getic communities, some of which were organized as komai (Fig. 48.1; Vulpe
and Barnea 1968, 38).
Region 4: the Interior of the Dobrogea
Away from the coast, the natural fertility of the region is greatly offset
by its very low rainfall, lack of surface drainage and extremes of temperature,
features which are reflected in its natural cover of steppe vegetation (Radulescu
1974, 72 and 74; Murgoci 1911, 6£. ). Ovid describes the Dobrogea as an open,
treeless plain (Trist. III.lO, 71; III.12, 14; Ex Pont. III.l, 19--20). The
freezing of the Danube in winter allowed the Getae and Sarmatians to raid south

733
of the river (I'rist. III.lO, 56; Ex Pont. I. 2, 77 -80). There is no reason to
dispute Ovid's general observations and the insecurity of the Lower Danube,
expo sed as it was to repeated barbarian attacks, is unlikely to have allowed
'rural communities' to exist in the Dobrogea before the end of the 1st century
A.D. PUny's list of Moesian tribes concludes with a vague reference to
Scythians living near the Pontic coast (N. H. Ill. 149). Ptolemy knew only of
the Peucini, inhabitants of the island of Peuce, near the Danube mouth, and
troglodites in the Dobrogea (Geogr. III.lO, 4). As late as the Flavian period,
there is little evidence for the existence of permanent military bases along
the Dobrogean limes; supervision, rather than direct control, was probably
only maintained by naval forces under the praefectus classis Moesiae et Ripae
Danuvii. 8 It is only Noviodunum (Isacea), certainly a naval base during the
2nd century A .D., which has produced evidence for Flavian, possibly even
Neronian occupation. 9 The Dobrogean limes was however under construction
by A.D. 103 when Carsium (Hir~ova) was built (Parvan 1913b, 481f.) and the
establishment of other auxiliary forts along the Danube bank, if not completed
before the 2nd Dacian War, was under way soon after the conquest of Dacia
when the V Macedonica moved to Troesmis (Iglifa) and the XI Claudia arrived
at Durostorum. 10 It was during this construction of the final section of the
Danubian limes and probably after the 2nd Dacian War (105/106) that work was
begun on constructing the network of roads which criss-crossed the Dobrogea
in the 2nd century A.D. (Fig. 48.1; Vulpe et al. 1968, 95f.; Aricescu 1977,
78f. and 193f.; Parvan 1912, 576f.).
Despite the unattractive nature of its climate, the central Dobrogea was
rapidly settled by 'rural communities'. With the exception of Tropaeum
Traiani, which already had a Roman civilian community, possibly a vicus,
only six years after the dedication of the Tropaeum in A. D. 109 (CIL Ill 1247 0;
Popescu 1964, 185f. ), nineteen 'rural communities' are attested under Antonius
Pius of which four, at least, were already in existence under Hadrian (Chora
Dagei, Laicos Purgos, vicus Ulmetum, vicus Quintionis). 2 Those 'rural com-
munities' which are known to have existed in the central Dobrogea are all vici
with Roman names and, where exactly located, they lay close to Roman roads
(Fig. 48.1). Three were situated at the junction of several important routes
(vicus Ulmetum, vicus Celeris, vicus Urb .. ). The fact that these vici are
first attested under Hadrian and are clearly associated with the road-network
suggests that they were not founded until after the Trajanic consolidation of
the Dobrogean limes.
The Roman character of these 'rural communities' contrasts with the low
level of romanization in region 1. Four of the Dobrogean vici took Roman
personal names (vicus Secundini, vicus Celeris, vicus Clementianensis, vicus
Quntionis), one was named after a Thracian (vicus Turris Mucaporis) and
another had a Greek personal name (vicus Narcif{ s] iani). They possessed
all the characteristics of quasi-municipal communities. They passed their
own decrees (vicus Urb- ; d(ecreto) v(icanorum) CIL Ill 14441), annually
elected magistrates (Chora Dagei, v. Novus, v. Secundini, v. Urb[-], v.
Turris Muca [poris?], v. Hi[-], v. Celeris, v. Scenopesis, v. Scaptia,
Ulmetum, v. Vero[br]ittiani, v. Clementianensis, v. Narcis[s]iani, in the
three un-named vici and in the un-named komai near Urluia and_ Istria)2 and
quastors (v. Quintionis, v. Novus). 2 They possessed their own council

734
buildings (auditoria; v. Quintionis and in the unidentified kome near !stria), 2
and worshiped the gods of the Roman pantheon (Jupiter Optimus Maximus and
Iun.o Regina, Silvanus Sator, Hercules, the Nymphs and Silvanus and Ceres
Frugifera). They had their own territories (CIL Ill 12488; 1447) and were
engaged in agricultural production. 11 At least one vicus was within the terri-
tory of Tomis (Constanja) and Histria seems to have had several (CIL Ill 7466;
Gerov 1952-53i 387 no. 568; Fig. 48.1). 2 Ulmetum lay within the territorium
Capidavensis. 2
The four best documented vici are known to have contained native peregrini
who, when first attested under Antoninus Pius, already played an important
role in the administration of these 'rural communities'. At Ulmetum, the
villagers describe themselves as cives Romani et Bessi consistentes, at the
vicus Quintionis as veterani et cives Romani et Bessi consistentes. 2 The
Thracian tribe of the Bessi occupied the Rhodope mountains and the Stara
Planina during the first century A.D. (Strabo VII.5, 12; Pliny N.H. IV.40;
Sarafov 1969, 141£. ). Ovid refers to them only twice during his exile in Tomis
(Trist. IIL 10, 5; IV.l, 67) and it is the Getae, Scythians and Sarmatians who
are regularly mentioned throughout the Tristia and Ex Ponto (cf. Trist. Ill. 3,
5-6; V .10, 48; Ex Pont. !.2, 76). Ovid stresses the lack of rural settlement
(Trist. III.lO, 57-66; Ex Pont. !.3, 55-56; !.8, 50-60; II.7, 70; IV.2, 43-44).
If there were Bessi in the Dobrogea during the early 1st century A.D., they
were not living in settled 'rural communities'. The Lai, another Thracian
tribe which occupied territory near the headwaters of the Strymon, appear,
in the 2nd century A. D., settled alongside Roman citizens in the vicus Secundini
(cives Romani et Lai consistentes). They also gave their name to the 'rural
community' of Laicos Purgos, near Histria. 2 ·
The case of the vicus Quntionis is particularly interesting. Of seven
dedications erected by the villagers, the three earliest are dedicated to
Antoninus Pius and Aurelius Verus. 2 Two magistrates were elected each year
and one was regularly a Bessian, the other had a Roman name and two of the
seven quaestors recorded on the seven known inscriptions have Thracian names
(Lambrino 1948, 329£. ). It is in itself remarkable that the Bessi, who had
acquired a considerable reputation for brigandage and warfare in their oppo-
sition to Rome during the Julio-Claudian period, should so readily have as-
sumed an important role in the regulation of the vicus alongside veterans and
Roman citizens (Sarafov 1969, 141£. ).
The appearance of these Thracian tribes of the Bessi and Lai in the
Dobrogea has been explained as the result of an enforced colonisation carried
out by the Romans (Casson 1927, 97£.; Vulpe 1938, 186; Alfaldi 1939, 28£.;
Sarafov 1969, 149£.; Vulpe 1976, 287£.; Zah and Suceveanu 1971, 567£. ).
There were precedents for the transportation of native peoples into Moesia.
Aelius Catus settled 50, 000 Getae in the newly created province £· A. D. 2
(Strabo Geogr. VII. 3, 10; Syme 1971, 53£.) Tib. Plautius Silvanus brought
more than 100, 000 Getae into the province under Nero (ad praestanda tributa),
after the annexation of Thrace which included territory on the Lower Danube
(CIL XIV 3608; Pippidi 1967, 287). The introduction of new peoples to populate
the immediate hinterland of the limes provided a taxable source of income,
particularly useful for the levying of the annona and providing recruits for the
auxiliary forces. In the case of the Bessi and Lai, it is hardly possible to

735
ascertain from the evidence of four 'rural communities' whether the scale of
such a settlement matched those of Aelius Catus and Plautius Silvanus but their
removal to the Dobrogea by the imperial ;mthorities seems probable enough,
an operation which can hardly have been contemplated before the Trajanic
period. Since the arrival of the Lai and Bessi in the Dobrog~a and the estab-
lishment of the vici are both likely to have occurred only after the Second
Dacian War (105/106) and were carried out before the death of Hadrian (138),
it would seem not improbable that the Thracians were settled in the vici at the
same time as they were themselves established. If correct, this would suggest
that the Roman authorities were also involved in the organization and creation
of the vici: an action which would explain the entirely Roman character of their
organization and the very rapid settlement of a region which offered few natural
inducements to the creation of 'rural communities'. Official intervention may
equally explain the allocation of different tribes to particular vici and the role
which the peregrine settlers were permitted to play in their internal adminis-
tration. No evidence suggests that 'rural communities' were given any encour-
agement in the rest of the province but two factors can be suggested as reasons
why such a 'colonisation' was carried out in region 4.
The Trajanic preparations for the conquest of Dacia must have presented
a considerable problem for the supply of the substantial forces then stationed
on the Lower Danube. The camps along the southern approaches to the
Carpathians secured the fertile and well-watered plains of Walachia and
Muntenia (Fig. 48.1; Tudor 1968, 37; Vulpe 1961, 384f.). Whether the
Papyrus Hunt dates to shortly before or immediately after the Second Dacian
War (105/106), the document illustrates the measures taken after the occupation
of transdanubian territory to secure the corn supply (Fink 1958, 102; Syme
1971, 122f.; Vulpe 1960a, 337f.; 1960b, 325f. and 1961, 370f. ). Hadrian
abandoned the North Danubian Plain east of the Olt and withdrew its garrisons,
with the exception of that stationed at Barbo§i, close to the Danube on the Lower
Siret (Tudor 1968, 37; Vulpe 1961, 384f. ). The loss of the agriculturally rich
transdanubian territory must have created additional problems for the supply
of the Danubian limes, particularly for the recently organized Dobrogean
sector.
The establishment of the auxiliary and legionary bases along the Lower
Danube was complemented by the creation of important lines of communication
between the limes and the Black Sea coast and the construction of roads to
facilitate the rapid transfer of troops from one section of the limes to another,
avoiding the circuitous route along the Danube bank (Fig. 48.1; Aricescu 1977,
105f. ). The creation of the road system must have been accompanied by the
establishment of numerous mansiones and stationes to provide watering places
for the military and the cursus publicus in a region noted for its aridity. The
policing of the roads was, at least in part, undertaken by beneficarii, detached
from the legions for special duty (Aricescu 1977, 39; Farvan 1912, 576f.;
Vulpe et al 1968, 170f. ). There was a statio beneficarii in the vicus at rimicul
de Jos and similar stationes probably existed in other vici, particularly those
known to have been situated at important cross-roads (Parvan 1913b, 523;
Aricescu 1977, 104 and 148). The need to maintain these roads presented ad-
ditional difficulties in the Dobrogea since the region seems to have had no
settled population (at least until after the Second Dacian War) which could
readily be organized for the purpose.

736
 The 'rural communities' of the Dobrogea certainly played an important
part in solving both these problems. A fragmentary inscription from Ulmetum
refers to obligations in connection with the via publica between fixed points
within a defined territory (Parvan 1913a, p. 394 no. 31). Since the inscription
was discovered close to the ancient site and vici are known to have had their
own fines fixed by the governor, the duties mentioned in the inscription pro-
bably relate to the vicus itself (CIL Ill 12488; 1447). The responsibility of
'rural communities' for the provision of transport for the cursus publicus as
well as the upkeep of the imperial roads l.s a well-known feature of imperial
administration (Frend 1946, 46f. ), and the libellus addressed to Antoninus
Pius by the inhabitants of the Chora Dagei confirms that the Dobrogean 'rural
communities' were no exception (Stoian 1959, 369f.; Vulpe et al. 1968, 143f. ).
'Settled in the kome by the public road' ( xa,;ouwuv,;Ec; [xo:\. ~Jxov't£<; ,;'T)v
xw [llTlV Jrra:pa -,;1-)v OYJilOOL [ a:v 0Joov 11. 10-13). The villagers complained that
they were oppressed by the excessive burdens of the Litouriga and Angareia
(!!.1 13-15). They refer to the judgement of Antoninus Hiberus, gover~or of
Moesia Inferior£· A .D. 134 (Syme 1971, 222), which was issued in favour of
the village of Laicos Purgos, situated 'where the public road beings (?)I ( a.;}£ \I w6cu-
E't0 au-,;n ~ OT)IlOOLO: 06oc; 1;0 rrpw-,;ov ll. 2. 12-14). There were also duties
performed 'many times each year' (rroAAaxLc; ,;ou ~,;ouc; 11.2.6-7). The an-
gareia is certainly a reference to the duty of supplying the cursis publicus .
although precisely what the villagers meant· litourgia and what duties were
perfonned 'many times each year' remains uncertain although they probably
included the supply of the annona (Stoian 1959, 383f. ).
Two inscriptions from Thrace provide more detailed evidence for the
role of 'rural communities' in the Balkans for the supply of the military and
the upkeep of the cursus publicus. The libellus addressed to Septimius Severus
in A.D. 202 by the inhabitants of the emporium at Pizus refers specific~lly
to the creation of similar emporia by the imperial administration (lGBulg.
1690, 11.24 and 70). Komai were attached to the emporium and were obliged
to supply its garrison, soldiers responsible for the maintenance of the cursus
publicus {!._. 70 (e) £rr~,o,;a.;J!.lo'- a,;pa-,;d':i,;o:l.). In return the emporium was re-
lieved of the obligation to supply the city in whose territory it lay and other
military garrisons (!!. (e) 60-62). The villagers of Scaptopara (Gromadon-
Blagoevgrad) in their petition to Gordian ill in A.D. 238, refer to their loca-
tion between two military camps which the village had long been accustomed
to supply with the annona (IGBulg. 2236, 11. (e) 31-33). Both these 'rural
settlements' played a direct role in the supply of military units and supplied
facilities for public transport, duties which were already an important fea-
ture of rural organization in the Dobrogea, a century earlier. Given the
evident important service provided by the se settlements, it is not surprising
that all three libelli submitted by the inhabitants of the Pizus emporium,
Scaptopara and the Chora Dagei, include the threat to abandon the settlements
if their requests were not met (IGBulg. 1690a, !: 10; e, !: 54; IGBulg. 2236,
1]. 55-73; Stoian 1959,379 3a, ll.11-14). The villagers believed that the duties
they performed were sufficiently important to induce the emperor to intervene
on their behalf.
The existence of a comparatively large number of 'rural communities'
(vici) in the Dobrogea contrasts markedly with their rarity in regions 1-3.

737
In region 1, the native population was largely unaffected at the village level of
organization and, despite Roman immigration, those villages which can be
identified are of Thracian origin and appear little disturbed by the introduction
of Roman administration. Similarly, region 3 retained its local native and
Greek character throughout the Roman Empire. Region 2 remained without
'rural communities' on account of the limited agricultural potential of the re-
gion, whereas region 4, itself not obviously attractive to settlement and a
region which· seems to have remained largely unoccupied until after the
Trajanic reorganisation of the limes, received organized Roman and Thracian
settlements, some of which were certainly already in existence under Hadrian.
Their presence seems difficult to explain without acknowledging direct inter-
vention by the Roman authorities and an explanation can be sought in the need
to supply the garrisons stationed along the Danube and in the need to secure
internal communications behind the limes. If emporia were established in
Thrace for precisely these reasons, it seems not incredible that 'rural com-
munities' were also settled in the Dobrogea to fulfil similar functions.
A more precise context for the 'colonisation' of the Dobrogea can be sug-
gested. There was a Hadrianic reorganization of the canabae along the
Danubian limes which was probably also encouraged by their importance as
centres of recruitment and supply for the military garrisons under whose
control they lay (Mocsy 1972, 72f. and 1974, 142). Hadrian, as emperor,
visited the Lower Danube on three occasions and it may well have been during
one of these visits of inspection that the canabae aeliae at Durostorum were
given official recognition (CIL Ill 7474; Syme 1971, 204f. ). If cives Romani
consistentes ad canabas were considered worthy of Hadrian's particular atten-
tion, it could well be that it was he who initiated the settlement of cives Romani
et Bessi (et Lai) consistentes in vico for the Dobrogea where their potential
role must have been appreciated by an emperor who had had personal experi-
ence of the geographical and climatic as well as military problems presented
by the limes of Moesia Inferior.

NOTES

1. The principal ancient sources for 'rural communities' are: Festus (ed.
Lindsay) 502, 508, 371; !sod. Etymol. XV. 2, 11f. and Orig. 15. 2, 12.
The remaining references are summarized by Swoboda, kome, P. W.
Supp. bd. IV. s.950 and by VanBuren, Vicus, P. W. VIIIA2 s.2090f.
For further discussion; Mommsen 1887 Ill. i, 119f.; Marquardt 1884
1, 7; Rostovtzeff 1926, 230f.; Vulpe et al., 196 8 and Gerov 1952-53,
333-334.
2. Agatapara vicus CIL VI 32582; Gerov 1951-53 no. 169, 367, 334.
Amalidina vicus (23rd August (?)) CIL Ill 13743; Vulpe 1938, 194; 1964, 677;
Vulpe et al. , 1968, 206; Parvan 1913a, 347; cf. also on the scutum Durae -·
Europi M.M. Zaharide SCIV IV (1975), 26f.
'Aa~oA.ooE L vwv ( xwu.YJ) CIL Ill 1421433; Parvan 1913a, 347.
Bapeni vicus (on the Danube road, east of Augustae) Geogr. · Rav. IV. 7.

738
 Ba ••• xw!ln (Palamarca, nr. Popovo) Gerov 1951-53 no. 593, 388; V.
Velkov Byz. Bulg. 1 1962, 33; IGBulg. II no. 761.
Buteridavensis vicus (Sariurt) (1) CIL Ill 1447; Tocilescu 1900, 206-207;
E. Dorutiu-Boila, SCIV XV (1964), 132. (2) Parvan 1916 no. 30, 634£.;
.. --
I. I. Russu, SCIV VI (1955), 80; Vulpe et al., 1968, 193.
Celeris vicus (Vadul) (1) CIL Ill 7526. (2) ~rvan 1923 no. 53, 78;
Lambrino 1948, 324.
Clementianensis vicus (Mihai Kogalniceanu) .(1) CIL Ill 7565. - (2) Parvan
1913a no. 17, 368. (3) A. Radulescu, SCIV XIV (1963), 79£.
t~&ye:1.. xf)pa (nr. !stria) Stoian 1959, 379; Vulpe et al., 1968, 143.
•Ep!J.wvax-to~ xw11n (beyond the Tyras (Nlstrul)). Ftolemy Geogr. III.lO,
7; strabo Geogr. VII. 3, 16.
Giridava (vicus) (Pelishat) CIL Ill 12399; Gerov 1951-53 no. 386, 132.
Hi{ -] vicus (Dorobantul) CIL Ill 12494.
xe: ... xw11n (nr. Mangalia) CIL III 1421433.

KnpJ..~wcr-c[nJvwv(xw!J.n)Gerov 1951-53, 132; IGBulg. I no. 270.

KopuoAnv[wv] xw!ln (nr. Varna) BSA Varna VIII 1951 no. 48, 22; IGBulg.
I no. 301.

Aa1..xo~ ITupyo~ (xw 11 n) (nr. !stria) stoian 1959, 379; Vulpe 1968, 143.
Longinopara vicus (nr. Glava Panega) Gerov 1951-53, 334, no. 369, 377.
Macrxt.oSpl..(ad XWIJ.T) (nr. Razgrad) Gerov 1951-53, 334; no. 369, 377;
IGBulg. II no. 749.
Narcis[ s]iani vicus (nr. Constanta) D. Tudor, SCIV XIII (1962), 119£.
Novus vicus (Babadag) CIL Ill 14448.
Par sal (-?) vicus (nr. !stria) CIL III 12488; Vulpe 1938, 193, 195.
Perburdava (vicus) (nr. Svishtov ?) IBA! XIII (1939), 320.
Pereprus vicus (nr. Melta )Lovech)) CIL VI 2736. Gerov 1951-53 no.
168, 367.

Petrus vicus (Camena) CIL Ill 7612; Gr. Florescu, Dacia V-VI (1936-38),
426.

Quintionisvicus (nr. !stria) (1) F'arvan1916 no. 24, 617. (2) Parvan
Dacia II 1925, 214. (3) F'arvan 1923 no. 46, 55£. (4) Parvan 1923, no.
49, 62£. (5) Parvan 1923 no. 50, 67£. (6) Parvan 1923 no. 51, 71.
(7) Parvan 1923 no. 52, 74. (8) P"arvan 1916 no. 40, 127£.

Rami{ davaJ vicus (nr. Hirsova) SCIV II (1951), 125£.; ·Durotiu-Boila

1977a, 180£.

Saprisara vicus (nr. Nicopolis ad Istrwn) CIL VI 2933; Gerov 1951-53

no. 158, 366.

L:&p6e:wv ( xw!J.n) (nr. Mangalia) CIL Ill 1421433 .

.:'>c[apt ?]ia vicus (Palazu Mare) CIL Ill 7536.

739
 Scenopesis vicus (nr. Capidava) (1) Florescu 1958 no. Il, 88. (2) ibid.
1958 no. 19, 96f.
Secundini vicus (nr. Istria) (1) Parvan, Dacia II 1925 no. 41, 241£.
(2) Lambrino 1948 no. 9, 322£. (3) Lambrino 1948 no. 10, 323£.
(4) Parvan 1923 no. 61, 96£. (5) Parvan Dacia ll 1925 no. 43, 246£.
(6) Lambrino 1948 no. 13, 324.
Siamus vicus (nr. Ostrov) Gerov 1951-53 no. 246, 371.
Tautiomosis vicus (Krivodol) IBAI XIX 1955, 201.
Trullensium vicus (Kounino) (1) CIL Ill 14409 (12390); Gerov 1951-53
no. 251, 371. (2) CIL Ill 14412:r;-Gerov 1951-53 no. 288, 373. I.
Velkov 1958, 557f.
Turris Muca(poris ?J vicus (Anadolkioi) CIL Ill 7 533; SCIV X (1959),
139f.
Ulmetwn vicus (Panteliomon de Sus) (1) CIL Ill 12491. (2) CIL Ill
1421426. (3) Parvan 1913a, no. 8, 344£. (4) Parvan 1913a no. 16,
365£. (5) Parvan1914 no. II, 276.
Urb[ .. ] vicus (Rfmnicul de Jos) (1) CIL Ill 14441. (2) CIL Ill 14442;
Dorutiu-BoiUi, SCIV XV (1964), 131.
o{JaA.. • • ( XW~ll) (nr. Mangalia) CIL Ill 14214 33 •
Varovum Minor ~rovoder) Arheologiya IV (1962), 31.
Vero[br ?]ittiani vicus (nr. Cius) CIL Ill 12479.
[-]c vicus (nr. Istria) CIL Ill 12488.
[-]diano vicus (nr. Devniya (Marcianopolis)) CIL V 892.
Also a few 'rural settlements' can be identified although their names are
unknown:
1) vicus (Galbiori) Noi Monum, epigr. 1964 no. 4, 105.
2) vicus (nr. Tomis) CIL Ill 7466; Gerov 1951-53 no. 568, 387.
3) vicus (Neatirnarea) CIL Ill 12487.
Fiveother inscriptions mention 'rural communities' but these may refer
to some of those included in the above list. 1) kome (nr. !stria) Parvan
1916, 619£.; AEM XI 1887, 69. 2) kome (Urluia) AEM XVII (1894), 113;
Parvan: 1916, 619. 3) kome (Tra.stika) Gerov 1951-53 no. Ill, 363;
IGBulg. 750.
3. Hunt's Pridianum includes the Haemus (Stara Planina) within the Trajanic
province (Fink 1958, 113). The boundary between Thrace and Moesia
probably followed this natural frontier until the foundation of Nicopolis
ad !strum~. and Marcianopolis by Trajan. The Hadrianic frontier, fixed
in A. D. 136, lay north of Nicopolis ad I strum and closer to the Danube
(B. Gerov IBAI XVII 1950, 30; Fig. 2).
4. Cohors I Claudia Sugambrorum veterana eguitata was stationed at Mihail-
ovgrad before A.D. 134 (JRS ~I 1926, 74-75, no. 201; Krrut 1951 no.
1820, 185). A regio Montanensium is attested in A.D. 161-63 (CIL lll

740
 12385) which may well have been administered by its own decuriones
(Kazarov 1938 no. 589, Ill) but retained its auxiliary garrison (Vulpe 1976,
294f. ). In A.D. 256 a burgus was constructed both for the protection of
the castrenses and cives Montanenses (CIL Ill 12376).
5. A second century inscription from Ulmetum describes C. Iulius Quadratus
as Loci princeps and quinguennalis territorii Capidavensis (CIL Ill 12491).
There was no municipium at Capidava but only an extra-mural settlement
(Florescu 1958, 18f.; MCA VII (1961), 571f.). The relationship between
the canabae, territorium and the auxiliary base is difficult to define but
it seems likely that the territorium 1 if not directly controlled by the mili ...
tary,was under their jurisdiction (M6csy 1972, 142f. ). In A.D. 100 the
canabae at Dimum seem to have had their own fines which would imply
some territorial control (D. M. Pippidi, St. Cl. VI 1964, 338; Vulpe
et al., 1968, 52. For doubts about the reading cf. M6csy 1972, 142).
An inscription from Pleven may refer to a territorium Dimensium (P.
Aelius Victorinus exmag(ister) territorio Dianensium (sic Dimensium)
IAI I, 1910, 115-118). If cives Romani consistentes ad canabas did ad-
minister extensive territoria then they must certainly have themselves been
subject to military control (M6csy 1972, 132f. and 1974, 139f. ). Under
M. Aurelius, Durostorum and Troesmis were created municipia and this
reorganization must have involved the absorbtion of at least some of the
territoria (Vulpe 1976, 290f. ).
6. Sexaginta Prista (Ruse) had a garrison and canabae by A.D. lOO (V.
Velkov, Epigraphic a XXVII (1965 ), 90f.). Appiaria was an auxiliary base
by A.D. 79 (V ..Beshevliev 1952, 71 note 122) but there is no evidence
(at least as yet) to prove that there were auxiliary bases east of Ruse in
the Julio-Claudian period (see the doubtful cases of Noviodunum and
Troesmis, below notes 7, 9, 10.).
7. There is no evidence for the existence of an auxiliary or legionary garrison
at Durostorum before the arrival of the XI Claudia, first attested under
the governor Q. Pomponius Falco (CIL Ill 12470; 7 537) i.e. A. D. 116-
117 (Syme 1971, 89f.; 216).
8. E. Doru}iu-Boila 1977b, 89-100. The argument for the establishment of
a fully organized Dobrogean limes under the Flavians rests on very little
evidence cf. Barnea and Stefan 1972, 15-25; Al. Suceveanu, Pontic a IV
(1971), 105f.
9. At Isacea (Noviodunum) the Roman cemetery appears to have been in use
by the Flavian period (MCA VII 1961, 391f.). There might just have been
Neronian occupation (~efan 1972, 101 note 18).
10. For the XI Claudia at Durostorum, above note 7. The first record of the
V Macedonica at Troesmis is a dedication to Hadrian by veterans in the
canabae and an inscription recording an honesta missio of A .D. 134 (CIL
Ill 6166; 6178); Vulpe 1968, lOaf.; Syme , JRS XVIII 1928, 41f.;
Aricescu 1977, 32. One tile-stamp of the ala I Pannoniorum (CIL Ill
6242) is slender evidence for an auxiliary base at Troesmis before the
arrival of V Macedonica. The unit was in Moesia under Vespasian (CIL
Ill 14453; Kraft 1951, 155; Aricescu 1977, 54f.; Vulpe 1968, 64). Cf.
doubtful find of a second tile stamp (Peuce IV (1977), 97q.

741
11. See the agricultural scenes on the monument to C. Iulius Quadratus from
Ulmetum (CIL 111 12491) and the popularity of deities associated with agri-
cultural production; Ceres Frugifera, Silvanus Sator.
12. It has been argued that there was a regio Histriae independently organized
and distinct from the territory of the Greek city (Gr. Florescu, SCIV IX
(1958), 337£. ). The argument fails to convince (cf. Vulpe et al., 1968,
241 note 100). Ulmetum; CIL IU 12491; Parvan 1912, 79 and 1913a,
397; above, note 5.
ABBREVIATIONS AND BIBLIOGRAPHY

AEM Archaeologisch-epigraphische Mitteilungen aus Oesterreich-Ungaren.

Vienna.
Arheologiya Arheologiya, Sofia.
Aricescu, A., 1977. Armata mDobrogea Romana, Bucharest.
a
Barnea, I. and Stefan, Gh., 1974. 'Le Limes Scythicus des origines la fin
de 1'antiquite' in Pippidi, D. M. , (ed. ), Actes du IX Congres International
d'etudes sur les Frontieres Romaines, 15-25, Bucharest.
Beshevliev, V. , 1952. Epigrafski Prinosi, Sofia.
BSA Varna. Izvestiya na Varnensko Arheologichesko Drudjestvo.
Byz. Bulg. Byzantino-Bulgarica, Ebfia~

Casson, S., 1927. 'Thracian Tribes in Scythia Minor', J . .Roman Stud. 17, 97f.
Doru~iu-Boila, E. , 1977a. 'Epigraphisches aus Scythia Minor' in Pippidi,
D. M. and Popescu_, Em. (eels.) Epigraphica, Bucharest.
Dorutiu-Boila, E., 1977b. 'M. Arruntius Claudianus Praefectus 'Classis et
Ripae Danuvii', st. Cl. 17,. 89-100.
Fink, R. 0. , 1958. 'Hunt's Pridianum: British Museum Papyrus 2851', J.
Roman Stud. 48, 102-116.
F1orescu, Gr., 1958. Capidava: Monografie Arheologica, Bucharest.
Frend, W. C., 1956. 'A Third-century Inscription relating to Angareia in
Phrygia', J. Roman Stud. 46 , 46-56.
Gerasimov, I. P. and Gulb_Ev Dj. S., 1966. Geogi:afiya na Bulgaria, &>fia.
Gerov, B., 1948-49. 'Romanizmut medjou Dunava i Balkana', part 1, Godish-
nik na &>fiya Universitet Fac. Hist. -Philol. 45, Sofia.
Gerov, B., 1951-53. 'Romanizmut medjou Dunava i Balkana' part 2, Godish-
nik na Sofiya Universitet Fac. Hist. -Philol. 47-48, Sofia.
IBAI. Izvestiya na Bulgarskiya Arheologicheski Institut, Sofia.
IG Bulg. Inscriptiones Graecae in Bulgaria Repertae, &>fia.
Kanitz, F. , 1882. Donau-Bulgarien
t
und der Balkan, Leipzig.
Kazarov, G., 1938. Die Denkmttler des Thrakischen Reitergottes in Bulgarien,
Budapest.
Kr.aft, K. , 1951. Zur Rekrutierung der Alen und Kohorten am Rhein und Donau,
Berne.

742
Lambrino, S. , 1948. 'Le Vicus Quintionis et le Vicus Secundini de la Scythie
Mineure', Melanges Marouzeau, 319f. , Paris.
Marquardt, J., 1884. ROmische Staatsverwaltung, vol. 1.
MCA. Materiale §i Cercetari Arheologice, Bucharest.
Mocsy, A. , 1972. 'Das Problem der MiliHtrischen Territorien im Donauraum',
Acta Anti9ua Scientiarum Hungaricae 20, 132f.
Mocsy, A., 1974. Pannonia and Uwer Moesia, London.
Mommsen, Th. , 1887. Rnmisches Staatsrecht, vol. 3. i., Leipzig.
Murgoci, G. , 1911. 'Les Zones Naturelles des sols', Revue du Petrole 6-7,
6f.
Parvan, V., 1912. 'Cetatea illmetum I', Analele Acad. Rom. Memoriile
Sect. Ist. 2 34,' 496-607.
Parvan, V., 1913a. 'Cetatea Ulmetum 2', Analele Acad, Rom. Memoriile
2 .
Sect. Ist. 36, 245-420.
Parvan, V. , 1913b. 'Descoperiri Noua din Scythia Minor', Analele Acad. Rom.
Memoriile Sect. Ist. 2 25, 467-550.
Parvan, V. , 1914. 'Cetatea illmetum 3', Analele Acad. Rom. Memoriile
Sect. Ist. 2 37, 265-304.
Parvan, V., 1916. Histria 4,-13ucharest.
Parvan, V. , 1923. Histria 7, Bucharest.
Peuce. studii §i communicari de istorie, etnografie §i muzealogie, Tulcea.
Pippidi, D. M. , 1967. Contributii la Istoria Veche a Romaniei, (2nd edn. ),
Bucharest.
Pontica. Studii §i materiale de istorie, arheoloie §i muzeografie, Constanta.
Popescu, E. , 1964. 'Epigraphische Beitrllger zur Geschichte der stadt
Tropaeum Traiani', St. Cl. 6, 185-203.
Pounds, N. J. G., 1969, Eastern Europe, London.
R.idulescu, I. , 197 4. Judetul Constanta, Bucharest.
Rostovtzeff, M., 1957. The Social and Economic History of the Roman Empire,
(2nd edn. ) London.

Sarafov, T., 1969. 'Les Besses et Rome', Actes du Premier Congres Inter-
national des Etudes Balkaniques et Sud-est Europeens, 14-153, Sofia.
SCIV. Studii §i Cercetari de Istorie Veche, Bucharest.
stoian, I. , 1959. 'Sur la plainte des Pay sans du Territoire d 'Histria', Dacia
N. S. , 369f.
St. Cl. Studii Clasice, Bucharest.
~efan, Al. -s. , 197 4.
'Recherches de Photo-interpretation archeologique' in
Pippidi, D. M. (ed.) Actes du I.Xe Congres International d'etudes sur les
£rontieres Romaines, 95-108, Bucharest. .

743
Syme, R. , 1971. Danubian Papers, Bucharest.
Tocilescu, Gr., 1900. Fouilles et Recherches Archeologiques en Roumanie,
Bucharest.
Tudor, D., 1968. Oltenia Romana, Bucharest.
Velkov. , I. , 1958. 'Vicus Trullensium' in Izsledvaniya v chest na Akad. D.
Dechev, Sofia, 557f.
Vulpe, R., 1938. L'Histoire Ancienne de la Dobrogea, Bucharest.
Vulpe, R., 1960a. 'Muntenia §i Moldova de Jos fu Timpullui Traian fu
lumina unei noi lecturi a papirului Hunt', st. Cl. , 337-357.
Vulpe, R., 1960b. 'Les Getes de la Rive gauche du Bas Danube et les Romains',
Dacia N. S. 4, 309-332.
Vulpe, R. , 1961. 'La Valachie et la Basse-Moldovie sous les Romains',
Dacia 5, 365-393.
Vulpe, R. , 1976. Studia Thracologica, Bucharest.
Vulpe, R. & Barnea, I., 1968. Din Istoria Dobrogei, vol. 2, Bucharest.
Zah, Em. & &l.ceveanu, Al., 1971. 'Bessi Consistentes', SCIV 22, 567f.

744
49. VOM OBERMOSISCHEN Lif,IE.S NACH 'DEM SUDEN: VIA
NOVA VON VIf·1INACIUI:I NACH DARDAJ'·TI3N

Mireslava Mirkovi6

Die römischen Strassen im Zentralteil der Balkan-

halbinsel decken sich im allgemeinen mit den Richtun-
gen der Eroberungszügen, die die Römer aus dem 3üden
und aus dem \vesten geführt haben. Die Entstehung des
Strassensystems ist deswegen als Teil der Eroberungs-
geschichte zu betrachten.
Wie bekannt , kamen die Römer an den unteren
Donaulauf im Laufe der Feldzüge , die sie gegen die
Stämme an der Nordgrenze der Provinz Hakedonien ge-
führt haben. Bei der Verteidigung ihrer Provinz dran-
gen sie allmählich nach derri Norden vor und bereits im
ersten Jahrzehnt des !.Jahrhunderts v.u.z. besiegten
sie die Dardaner in ihrem Lande. .Spärliche Nachrichten
über diese Ereignisse erlauben uns kaum die Richtun-
gen nachzuziehen, die die römischen Truppen unter
Scribonius Curio, dem Sieger über die Dardaner im
gnadenlosen dreijährigen Kriege (75-72 v.u.Z.) und
dem ersten römischen Feldherrn, der an die Donau ge-
langte,einschlugen. Der Umstand, dass in den Quellen
auch eine Niederlage der Möser, des zwischen dem Ti-
macus (Tirnok) und dem Oi~cus (Isker) lebenden Stamme~,
er\vähnt wird, spricht dafür, dass die römische Trup-
pen die Donau östlich vom Eisernen Tor erreichten. 1
C.Patsch vertritt die Meinung, dass Scribonius Curio
in diesem Feldzug die ':!ardar-IVIorawafurche bis Naissus,
weiter das Timoktal benutzte, an dessen nündung er
die Donau erreichte. 2 :oem r.veg längst des Timaktals
dürfte auch M.Licinius Grassus, als er als Statthal-
ter f'1akedoniens in den Jahren 30-28 gegen die Höser

745
Abb. 49. 1 Römischen strassen im Zentralteil der Balkanhalbinsel
 und andere Stämme an der Donau kämpfte,gefolgt haben.3
Die Marschwege durch das Gebiet der Dardaner,Eöser und
Triballer,,die damals durchgeschlagen wurden, bilden
die Anfänge der später gebauten r•Iili tärstrassen.
Bei der Erörterung der Fra~e der Entstehung von
Strassen im Zentralbalkanraum kann Naissus in Dardani-
en nicht umgangen werden. In späteren Itineraren ist
hier der I:notenpunkt des Strassennetzes der Balkan-
halbinsel erwähnt. In Naissus trafen sich die Strasse
von Lissus, die in Richtung Donau (Ratiaria) verlief,
dann die Strasse aus Singidunum nach Constantinopolis,
auch diejenige die des Timok entlang an der Jon~u mün-
dete und auch die Strasse vom Süden. 4 JJas kann vor al-
lem durch die strategische Bedeutung von Naissus in
der frühen Phase der römischen Eroberungspolitik er-
klärt werden. ·:lie schon angenommen, zu der Zeit als
die Legionen aus fl[akedonien nach dem Horden verlegt
wurden, hatte das T"Iili tärkommando seinen Sitz in Nais-
sus.5~ahrscheinlich erhielt Naissus schon unter Augus-
tus die Strassenverbindung nach Hakedonien,vielleicht
auch nach Lissus. 6 Im weiteren Laufe des römischen Vor-
dringen nach dem Norden wurden dann die Strassen
strahlenförmig in Richtung zur Donau ausgebaut. Die
Frage der Richtungen dieser Strassen muss im Zusammen-
hang mit den Problemen der ersten römischen Feldzüge
gegen die Daker erörtet werden.
Schon unter Augustus wurden die Daker zu einem ·
wesentlichen Fa1{tor der römischen Politik an der un-
teren Donau. ~.vährend des ~vinters stiegen sie aus ihren·
Gebirgen hinab, drangen über die zugefrorene Donau ein
und plünderten das Gebiet auf dem rechten Donauufer.
Im Komplex der Fragen über die Kommunikationen auf
dem Balkan ist es wichtig zu erfahren , welcher Teil
des Donaugebietes gefährdet \:!ar und, im Zusammenhang
damit, in welcher Richtung die römischen Truppen
operierten. ~.'Jie es Florus, II 28, bestätigt, ihren ii.ll-
griffen war das rechte Donauufer in der Höhe der da-

747
 kischen Gebiergen ausgesetzt. Demnach kommt der östlich
vom Eisernen Tor liegende Abschnitt des Donaugebietes,
mit Durchgängen nach dem Süden an der fv1ündung des Ti-
macus und durch die links und rechts davon liegenden
Flusstälern in Betracht. Spätestens um die Mitte des
1. Jahrhunderts wurde dieser Korridor, der von der Do-
nau nach Naissus führte und der als ein Limes schon
in der Zeit der römischen Eroberung des Landes der
Möser durchgeschlagen wurde, als Verteidigungslinie
mit Strassen und Festungen ausgebaut. ~on dieser Be-
festigungslinie stammen die ältesten militärischen
Denkmäler der späteren Provinz Obermösien. Die mili-
tärische Kontrolle der Strasse des Timok entlang lag
wahrscheinlich im Befehlsbereich der Legion V Macedo-
nica. Die erste militärische Station nördlich von Nais-
sus befand sich in Timacum Minus. Ihre älteste bekann-
te Besatzung war die Kohorte I 1\'Iontanorum , von der
angenommen werden kann, dass sie schon zur Zeit Clau-
dius und Neros dort ihren Standort hatte.7 Das Ende
dieser Verteidigungslinie nach Norden bezeichnete die
Timacusmündung, in deren Nähe sich das Kastell Aquae
befand. In frühflavischer Zeit schlitzte diese Position
. 8
di~ Kohorte I Cantabrorum.

Unter dem Kaiser Tiberius befanden sich die Legi-

onen bereits an der Donaulinie. Die römische Regierung
versuchte seit dieser Zeit die grossen Schwierigkeiten,
die das Eiserne Tor mit seinen Sehtromschnellen und
steilen Ufern für die Verbindung zwischen den Legions-
lagern Pannoniens und jenen an der mösischen Donau-
strecke darstellte, zu bewältigen. Die unter Tiberius
bego~nenen Strassenarbeiten in der Oberen Schlucht des

Eisernen Tores wurden unter Claudius und den Flaviern

fortgesetzt; eine Kommunikation längst der Donau vlur-
de erst unter Trajan vollkommen ermöglicht.9
Auf Grund des bisher ausgelegten kann man schlie-
ssen, dass bis zur Eroberung Dakiens für Rom als Korn-
munikationsproblem im Zentralbalkan an erster Stelle

-748
 Verbindung zwischen Makedonien und der unteren Donau,
darrnach zwischen Naissus und der Donaustrecke östlich
vorn Eisernen Tor bestand.Nach der erfolgter Versetzung
der Legionen an die Donau war von grosser Bedeutung
die Verbindung an der Donau entlang. Die Legionslager
westlich vom Eisernen Tor ( Viminaciurn, vielleicht auch
Singidunum) hatten zuerst Verbindung mit den nächsten
Lagern in Pannonien un dann mit denjenigen an der un-
teren Donau.
Eine Ausnahme beim Ausbau des Strassensystems
stellte die Strasse im Tal der Flüssen r~nava und I·,1o-
rawa.dar, die vom Legionslager bei Viminacium nach
Süden abzweigte. Sie gehörte zu den frequentiertesten
im Römischen Reich, stammt aber allem Anschein nach
aus wesentlich späteren Zeit und·wurde vom Limes Rich-
tung Süden gebaut. Das Tal von Velika Morawa, durch
das in den vergangeneu Jahrhunderten viele Stämme nach
Süden bis Ma~edonien gelangten, war für die Römer von
Z\..rei trangiger ~edeutung zur Zeit ihrer Eroberung des
Zentralbalkans. Die Skordisker vom unteren Savelauf
und aus dem fviorawatal benutzten diesen ·\leg bis zum
Jahr 16 v.u.z. als freien Pass nach Makedonien. Der
F~ldzug, in dem sie von den Römern unterworfen-wurden,

kam aber nicht vorn Süden, sondern nach der Niederwer-

fung der Pannonier, vom \iesten her. 10 Auch unmi.ttelbar
nach dem Siege hatten die Römer keinen Grund südlich
der Donau , im Mlava- oder r1orawatal eine via mili taris
auszubauen, da das Gebiet im Nordwesten der späteren
Provinz Obermösien nach dem Rückzug der Skordisker
kaum besiedelt war. Die Dardaner wurden durch die
Strassen und die Festungen südlich von Naissus kon-
troliert, die Möser und Triballer - durch die fvlilitär-
strasse entlang des Timoktals. Die Verteidigung der
Donaustrecke von Budapest bis zum Eisernen Tor war
gemeins 8 me Aufgabe der pannonischen und mösischen Le-
gionen. Das abwechselnde Auftauchen der gleichen Trup-
penteile am pannonischen und mösischen Sektor des

749
Donaulimes bildet ein auffälliges Merkmal des Militär-
systems Ostpannoniens und_ 11'iestmösiens bis zur Erobe-
. 11
rung· Da k lens.
Die Strasse von Viminacium nach dem SUden konnte
zu jener Zeit entstanden sein, als das Donaugebiet
schon befriedet und Dakien erobert war. Sie wurde als
Fortsetzung der langen Kommunikationslinie , die Ita-
lien und den westlichen Teil des Reiches über den Bal-
kan mit den östlichen Provinzen verband, gebaut. Die
Entstehungszeit könnte auf Grund einer fragmentarisch
erhaltenen Inschrift aus Viminacium , die, wie es
scheint, Hadrianus gewidr::et ist, bestimmt werden. ;J.uf
Grund des erhaltenen TextbruchstUckes
AR
DIVI TR DI
EPOS TR
OS III P
•
NOVArJI QVA
NO COHP3N
IN DARDANIA
COMJVIEARE
FE
dürfte die Inschrift wie folgt rekonstruiert werden:
Imp.Caes]ar
Divi Tr[ aiani Parthici f.] di[ vi Nervae
n]epos Tr[aianus Hadrianus Aug.pont.max.
trib.pot •••• c]o(n)s(ul) III p(ater) [p(atriae)
•
per leg.Aug.pr.pr.viam] novam qua[e
coepta a divo patre suo !raia]no compenldio
facto per m.p. ad ....• Jin Dardania
direxit et munivit ut vehiculis] commeare[
] feLcit
12
Obwohl ein grosser Teil der Inschrift fehlt,
scheint es, dass gewisse Elemente als ziemlich ge-
sichert gelten können, und z1.1ar der Kaisertitel des
Hadrianus·und die Ergänzung viam novam , unter Be-
rUcksichtigung der Termini compendium und commeare ••-

750
 Die Ergänzungen in den Zeilen 5-9 sind nur exempli
gratia, nach dem üblichen Texten dieser Art von In-
schriften.13
Die via nova von Viminacium nach Dardanien konnte
die direkte Verbindung zwischen dem Donaulimes und dem
Binnenland der Provinz Obermösien darstellen.Uber die
Länge der Strasse undcden Endpunkt sirid keine Daten er-
halten geblieben. Dardanien begann südlich von der Sta-
tion Horreum Margi, nach der Angabe des Itin.Hieros.
564,8-565,7 - 61 römische Meilen von Viminacium ent-
fernt. Drei wichtige Funkte in Dardanien kommen in Be-
tracht als Endpunkt dieser Strasse: Scupi, als die
älteste Kolonie im Süden Obermösiens, metalla im Zen-
tralteil Dardaniens und Haissus als eine der wichtig-
sten :Strassenkreuzungen auf dem Balkan. Ich gebe den
Vorzug der dritten Möglichkeit. Die flavische Kolonie
Scupi muss schon früher im Norden mit der Station Nais-
sus verbunden gewesen sein und darüber hinaus mit Ra-
tiaria an der_Donau; die da:rdanische Bergwerke befanden
sich abseits der Hauptstrassen und waren mit ihnen
durch Nebenstrassen verbunden. Die Strasse Viminacium
- Naissus konnte die kürzeste Verbindung (compendium)
diastellen zwischen dem die Donau entlang zum Sch\varzen
Meer führenden Weg und der Strasse die vom Süden (Scu-
pi) oder Südwesten (Lissus) kommend Naissus mit Thra-
kien verband. Somit wurde das ~Jort compendium als Be-
zeichnung für die direkte Verbindung zwischen diesen
zwei Strassen gebraucht. nUS dem Wege aus Italien nach
Kleinasien ist das wohl die kürzeste Ver~~ndung, die
über den Zentralteil des Balkans verlief. Das auf dem
ersten Blick überraschende Ergebnis der Ergänzung der
cInschrift aus Viminacium , dass nämlich eine so \'fich-
tige Strasse erst zur Zeit des Hadrian gebaut vJurde,
steht, wie oben es zu zeigen versucht, im Einklang
mit der historischen und militärischen Situation auf
dem Balkan zur Zeit der römischen 3roberung.
Ursprünglich war diese Strasse vom Limes nach

751
dem Süden keine via militaris~5Als solche galten vor-
nehmlich Verbindungen zwischen Naissus und der Donau
östlich vom Eisernen Tor, die limites im Lande der
Triballer und Möser. Spätestens aber zur Zeit des
Marcus Aurelius muss sich der Charakter der Strasse
von Viminacium nach Dardanien geändert haben. Nachdem
sie immer mehr bei zahlreichen Feldzügen für den Trup-
pentransport aus dem \1esten und aus den an de-r Donau
liegenden Lagern im Osten benutzt wurde, musste sie
militärisch geschützt werden. N~ben den möglichen ~in­

fällen der sarmatischen Stämme aus dem Norden,stellten

auch die latrones eine Gefahr dar. In manchen Statio-
nen an dieser Strasse wurde in der zweiten Hälfte des
2. Jahrhunderts und später eine militärische Besatzung
nachgewiesen: in Horreum I"'argi ist die Anwesenheit
einer Vexillation der Legion VII Claudia, in Praesidium
Pompei - die zur Zeit des Harcus Aurelius gegründete
Kohorte Dardanorum und in Naissus sind die stratores
der. beiden obermösischen Legionen, VII Claudia und
IV Flavia bestätigt~6

752
 AWH~::;RKUNGEN

1. Ruf. Fest. VII,B; Liv. Per. 92 und 95; Flor.I,

39 u.a.
2. Patsch, 1932: 35.
3. Dio Cass. LI 23,2-27,1. Die Donau erreichte
er über Serdica.
4.Tabu1a segm.VII; Itin. Ant. 133,1-134,5; Itin.
Hieros. 564,7-566,3.
5. Domaszewski,l891:199 f.Vgl. auch ~.aitterling,

RE XII,l924: 1227 und ~.7luss, R~ XV,l935: 1592.

6.Domaszewski, A~I-~ 13,1890:151: Die 2trasse
11

Lissum-Naissus ist also gebaut ••• wahrscheinlich in

der Zeit als Augustus das moesische Eilitärkommando
errichtete 11 •
7. ~agner,l938:170.Siehe CIL III 8161 = Dessau,
ILS 2733.
8. Zbornik Fi1ozofskog fakulteta,Beograd, VII-1,
1964,87 ff. Vgl. Mirkovi6,1977: 171 ff.
9. A.et J.~a~el, ILiug. n.55-63. Vgl .• auch P.
Petrovi6, Starinar 21,1971: 31 ff. und J.Sasel, JR3
63,1973: 83 ff.
lo. Fapazoglu, 1978:343 (mit Literatur).
11. E.Mirkovi6, Epigr. Studien 5,1968: 180 ff.
12. Um das Jahr 1912 entdeckt, blieb die Inschrift
unbeachtet, da die damals bekannten sieben Frabmente
bei der ersten Veröffentlichung nicht gerade glücklich
zum Ganzen zusammengefügt wurden ( n.Vuli6,JOAI 15,
1912, Beibl.:215-216, der in Z.l den Titel des Antoni-
nus Pius erkannte). Doch dank der ingeniösen Ergän-
zung der rechten Seite der Inschrift wurde die ~rkennung

eines neuen und wichtigen Fragmentes unter verschie-

denen im Museum in Eo~arevac befindlichen Steinfrag-
menten ermöglicht. Vuli6 führt noch ein Fragment an,
das auf dem Photo nicht zu sehen ist und auf dem er
ama , di liest. Nach dem Kaisertitel liest er: novam
• • • • • • • in Da4t-ania] com (meare) fe (ci t.

753
 13. Vgl.z.B.CIL III 600 viam publicam guae a
col.Byllid. per Astacos ducit angustam fragosam ita
munit ut vehiculis commeatur. In der letzten Zeile
vielleicht per milites legionis VII Claudiae fe cit.
14. Line Analogie wäre I~in.Ant. 171,1: a Cabile
per compendium Adrianopoli usque m.p.LXXVIII. Vgl. auch
Itin.Ant. 119,1; 180,6 und 8. Es ist wenig wahrschein-
lich, dass parallel zu dieser Strasse noch eine andere
bestand, die bei der Iviorawa-Nündung nach dem Süden ab-
bog ( Kanitz, 1896: 60. Siehe auch Ch.riülsen,AEIVI 12,
1888;181 ff.)
15. Uber die Viae militares siehe J.hasel, 1977
235 f. Es ist allgemein angenommen, dass die Strasse
südlich von Viminacium eine via militaris war.
16. CIL III 12667-12668;· Spomenik 7l,l93l,n.
594 ( Horreum I'Iargi) ; CIL III 145 56 = JOAI 4, 1901,
Beibl. 163 f.,n.3 ( Fraesidium Pompei); CIL III
1674-1676 u.a. (Naissus).

LITERATUR
A.v.Domaszewski,l89l. Die Entwicklung der Provinz
~oesie. Neue Heidelb.Jahrb.I,l: 191-200.

F.Kanitz,l896. Römische Studien in .Serbien.Denk-

schriften d. Akad.l·/ien.
M.fviirkovi6,1977• Zum Problem der Militärgeschichte
Obermösiens im l.Jahrhundert. 6iva Antika 27,1977
F.Papazoglu, 1978. The Central Balkan Tribes in
pre-Roman Times, Amsterdam: A.Hakkert.
G~Patsch,l932. Beiträge zur Völkerkunde von Süd-

osteuropa,V: Aus 500 Jahren vorrömischer und römischer·

Geschichte Südosteuropas .ß. kad. :./ien,Philos. -hist .Klasse,
Sitz.-Ber.214.
J.Sasel, 1977. Viae militares. Studien zu den
Militärgrenzen Roms II. Vorträge des 1o.Internat.Limes-
kongresses in der Germania Inferior. Köln-Bonn: Rhein-
land-Ver:!._ag Gf·'lBH und R.Habelt Verlag GI·~3H.
V.Wagner,l938. Die Dislokation der römischen
Auxiliarformationen.Eerlin: Iunker und Dün:qhaupt \Terl.

754
 Summary

The oldest Roman roads in the area of the centrat Balkans co-
incide with the directions of the Roman marches of conquest, from
the south from Macedonia and from the west along the Save valley.
If the history of the origin of the roads is linked to the question of
the Roman conquest, several phases can be distinguished.
The link between Naissus and Macedonia is amongst the earliest
laid roads; the next phase of the Roman advance produced the roads
which run from Naissus in the direction of the Danube along the river
valleys of Timachus and Kiabrus. After the subjugation of the
Pannonians and the Scordisci the road in the Save valley was adapted
to military requirements. Its natural continuation is the communi-
cation along the Danube.
The only road in Upper Moesia which was built southwards from
the limes is that from Viminacium to Dardania in the Mlava and
Morava valley. As the shortest link between the communication on
the Danube and that which led via Naissus towards the east and the
south, this road originates at the latest from the time when the
Danube area had been pacified and Dacia had been conquered. One
may conclude that this road was only completed under Hadrian if the
restoration of the fragm.entary inscription from Viminacium (JÖAI
15, 1912, Beibl. 215f. ), of which a new fragm.ent was recently dis-
covered, is correct.

755
 50. LES FORTERESSES DU BAS-EkPIRE bUR
LE LINiES DANU'BIEN EN SERBIE
Petar Petrovié

Lorsque les troupes romaines commandées par

Aurélien évacuèrent la Dacie, les régions au sud du Danu-
be subirent de sensibles modifications qui .se reflètent,
entre autre, dans les vastes travaux de construction. La
défense de la frontière danubienne, après une assez lon-
gue période de paix, revient au premier plan de la stra-
tégie militaire romaine et la grande intensité de ces tra-
vaux se poursuivra presque sans relache pendant un siècle.
Après les fouilles de protection poursuivies
plusieurs années lors de la construction de la centrale
hydroélectrique et du lac d'accumulation dans le secteur
des Portes de Fer 1 , nous sommes actuellement en mesure de
parler avec plus de précision du volume et de la qualité
de construction sur le limes. Sur plus d'une trentaine de
.
fortifications et tours explorées, disposées sur une lon-
gueur de 9o km environ de la rive du Danube (Fig. 50.1),les
~estiges de l'époque du Bas-Empire sont les mieux conser-

vés, soit qu'il s'agisse de constructions ajoutées, d'a-

daptations ou de nouvelles constructions. Elles montrent
en m~e temps que le limes danubien dans le secteur des
Portes de Fer a acquis disposition definitive au cours
du IVe siècle et qu'avec de petites interventions il est
resté le même jusqu 'à la fin de l'antiquité, qui dans ces
régions est déterminée par les conquêtes avaro-slave sur-
venues à la fin du VIes. et dans les premières décennies
· · e 2
du VII s.ap.J.C.
Dans le premier cas, lorsqu'il s'agit de re-
maniements ou d'adaptations., les travaux avaient pour
but de consolider les remparts par la construction d'un
mur parallèle longeant de près l'ancien (Novae, Boljetin,
Taliata, Karatas) et de toars d'angles à plan circulaire
ou en fer à cheval (BolJetin, Taliata); il semble que les
portes aient été murés beaucoup plus tard, dans la derni-

757
 LIMES DES PORTES DE FER N

Bolje9

~
0
• Forteresses
Ravna
""'Murs de defense

NOVAE Noms des sites anciens

Saldum Noms des sites modernes

0 5 10km

Fig.50.1
ère phase de l'existence du limes, vers la fin du VIe s. 3
Il semble qu'en général on n'ait reconstruit que les plus
grandes fortifications, celles qui n'ont jamais été tout
à fait détruites. 4
~e plus grand nombre de petits castella dans

le secteur des Portes de Fer a été construits au cours du

Bas-Emplre, le plus souvent sur un terrain non occupé (Bos-
man, Ravna, Porecka reka, rfalo Golubinj e, Hajducka vodeni-
ca, Sip, Donje Butorke), ezceptionnelle1ùent à l 'emplace-
ment de fortifications plus anciennes ayant survécu les
destructions du IIIe s. (Saldum). C'est au Bas-Empire qu'
appartiennent aussi les murailles en pierre, construites
à l'embouchure de petites rivières qui se jettent dans le
Danube par une large delta: Brnjica, Kozica et Porecka re-
ka5. Les burgi (Turski potok, Zidinac) et de nombreuses
tours d'observation, construites à des endroits soigneu-
sement choisis dans des zones de visibilité mutuelle, il-
lustrent de même la bonne protection de la rive du Danu-
be à l'époque.
Les castella susmentionnés, de Qb~ensions ~o­
destes, érigés de règle dans la gorge supérieure et infé-
r;ieure des Portes de Fer, attirent une attention toute
particulière. Il a été déjà remarqué que les guadriburgia
ou forteresses carrées de dimensions presque identiques,
avec des tours carrées aux angles, constituaient un grou-
pe à part: Porecka reka, Ravna, Tekija et Sip; il faudrait
y ajouter les forteresses de la rive gauche du Danube (Gor-
nea, Orsova, .Putinei, peut-être Insula Banului), qui; avec
celles qui ont été déjà citées, forment la couche la plus
ancienne, se rattachant chronologiquement à l'époque ~e
la Tétrarchie. 6
Les forteresses à tours circulaires ou tours
en U aux angles, qui marquent la phase de construction
suivante dans cette région, ont aspect typologique plus
complexe: 3aldu.m, Ravna (deuxiÈme phase), Malo Golubinje,
Hajducka vodenica et Donje.Butorke-Zanes. La ~ate de con-
struction de ces forteresses se place en général au Bas-
Errrpire, soit du IVe au VIe s. a:p. J. C. qpelques détails du

759
plan de la forteresse de Saldum, que nous avons exploré,
et sa stratigraphie, jettent plus de lumière sur ce pro-
blème et déterminent avec plus de· précision la date de
construction non seulement de Saldum, mais vraisemblable-
ment aussi d'autres forteresses similaires déjà mention-
nées.
On remarque tout d'abord que Saldum et Za-
nes sont construits selon un plan similaire (Fig. 50. 2 et 3).
Les deux sont rectangulaires, avec des escaliers sur le
rempart et des tours circulaires aux angles avec une en-
trée en entonnoir identique. Les deux ont une tour de for-
me unique, à base rectangulaire et mur semi-circulaire du
coté est: à Zanes sur le· rempart est, à Saldum à 1 'angle
nord-est. Il faut souligner que les deux tours ont été
construites en même temps que le rempart et que déjà au
cours des travaux la possibilité qu'elles aient été ajou-
tées ultérieure a été éliminée.Quelle était la fonction
d'une telle to~ dans un plan extrêment rationnel et sché-
matisé de la forteresse? S'agit-il uniquement d'une tour
de défense ou peut-être d'une chapelle? Cette deuxième
possibilité nous semble plus probable, surtout si on ti-
ent compte du fait qu'à Zanes elle est placée sur le rem-
part est, entre deux tours angulaires, où son importance
défensive n'a pu être grande. On pourrait croire que la
chapelle et la tour ont existé parallèlement à deux éta-
ges, mais malheureusement cette solution n'est pas appu-
yée par les trouvailles archéologiques qui font presque
totalement défaut dans les deux cas. 7 L'existence de pe-
tites églises à une nef a été, il est vrai, confirmée
dans d'autres castella aussi (Boljetin, Hajducka vodeni-
ca), mais celles-ci ont été construites bien plus tard
dans l'espace libre à l'intérieur de la forteresse, très
vraisemblablement au VIe siècle. ~uelques petites égli-
ses ont été construites à même le rempart; d'autres fois,
comme dans notre cas, elles en faisaient partie. Nais
nous les rencontrons dans un. autre contexte à l'intérieur
de la péninsule balcanique, dès le début du -Ve s.ap.J.C.
(Kamenica, Pirdopsko, Berkovica et autres). 8 Un autre ex-
760
Saldum

Fig. 5O.
2
0 10 20m

Donje Butorke, Zanes oanube

.
---~:::: ~-

Fig. 50.
3
0 10 20m

761
emple semblable, celui de la petite église dans le villa-
ge fortifié de Golemanovo Kale (région de Pleven en Bulga-
rie),remonte à l'époque paléobyzantine.9 Tout compte fait,
il semble qu'il n'y ait pas de parallèles directes pour
nos chapelles. En général, un très petit nombre de bâti-
ments dans nos régions, particulièrement sur le limes, se
situent à l'époque paléochretienne, bien que nous sachions
que certaines villes voisines du Danube étaient des évê-
chés au IV 8 siècle, avec de puissantes comunautées chré-
tiennes1~ Ce problèllie est iain d'être coLplèterr.ent étu-
.
dié. A noter qu'une crypte paléochretleJ.ll1e \,le forL:le rec-
tan~Jlaire et dotée de niches) a été mise au jour à Kara-
tas (Caput Bovis), une dizaine de kilomètres à l'est de
Zan es . 11
Il y a encore une particularité construc-
tive qui est présente dans le plan de nos fortifications.
Il s'agit d'une tour à murs épais, située au centre de la
forteresse et à_l'intérieur de laquelle il y a des piliers.
Un tel bâtiment a été découvert dans le castellum de Za-
nes (18,5 x 19 x 19,5 x 19 m), construit en pierre et bri-
ques, revêtus de gros blocs de pierre taillée (épaisseur
des murs 3,3o m), avec quatre piliers à l'intérieur (2,5o
x 2, 5o rn). Nous rencontrons à I~:ihailovac, fortification
voisine, un bâtiment central semblable (Fig.50.4} 12 . Il s'
agit du type de tours d'observation carrées aux murs puis-
sants, entourées de tous les cotés d'un mur. Ce type est
bien connu dans la partie pannoniènne du limes ( Budakalasz,
Leanyfalu, Pilismar6t et ailleurs), ainsi que dans les·ré-
gions occidentales de l'empire. 1 3 Pour la forteresse de
Zanes on pourrait citer de nombreuses parallèles dans les
provinces occidentales. Parmis celles-ci le burgus près
d'Asperden, d'après ses dilliensions, ses tours circulaires
aux angles et sa tour intérieure, est la plus proche. 14
Ces forteresses sont chronologiquement liées au program-
me de Valentinien Ier concernant la construction sur le
limes. Ajoutons aussi que la forteresse de Bajducka vode-
nica (Fig.50.5) avait aus-si la tour intérieure (l2,2o x
12 E.), au dessus duquel on avait construit une petite é-
762
Hajducka Vodenica
Fig. 50.4

Danube

o 10 20m

Mihajlovac

Fig.50.5

• 0
0 0

0 10 20m

763
glise au VIe siècle. 1 5
Il va sans dire que la similitude du plan
des forteresses de Saldum et de Zanes, auxquels les castel-
la de rJ::ihailovac, de Hajducka vodenica et de Malo Golubi-
nje16 sont très proches, ne peut par elle-même déterminer
la m~me date de construction. Les b~timents qui les carac-
térisent, appartenant soit au même schéma de fortificati-
on (tours d'observation au milieu de la forteresse), soit
à la même sphère de la vie spirituelle (chapelles), indi-
quent les liens entre ces forteresses et reflètent proba-
blement un même plan de construction~1 7
Voyons maintenant plus en détails ce que nous
offre la stratigraphie de Saldum, l'un des castella qui,
en raison de son bon état de présérvation et la richesse
des trouvailles, peut être un puissant appui à nos conclu-
sions. Au cours des recherches qui se sont poursuivies
pendant plusieurs années dans cette localité et qui ont
englobé plus des 2/3 de sa surface, cinq niveaux d'habita-
tion ont été constatés, comprenant l'époque de la moitié
du Ier s.de n.e. jusqu'à la fin du VIe s. Les couches I
- III appartiennent au Bas-Empire. Les deux premières sont
liés à la vie du camp au VIe s., tandis que le niveau III
correspond à la construction de la fortification et au
temps de son existence. Le niveau IV, endommagé par les
travaux de terrassement pour les remparts et les tours,
se rattache à une fortification plus ancienne, aussi en
pierre, dont le plan n'est pas connu dans sa totalité. Le
mobilier archéologique et les découvertes numismatiques
indiquent que cette fortification la plus ancienne existait
jusqu'aux dernières années du IIIe s. (monnaies d'Aurélien
et de Probus); l'absence de traces plus visibles de des-
truction et de déblai à ce niveau, y compris l'existence
d'une couche de nivellement, nous portent à croire que
lors de la reconstruction totale de la fortification on
avait utilisé tous les matériaux disponible sur place.
La couche III marque la vie de cette nou-
velle fortification. A l~intérieur on a trouvé plusieurs
baraques en poutres et rondelles de bois. Le contenu des
764
baraques, très riche et varié(céramique et verre, outils,
arm~s et autres) est cowplètement conservé grâce au fait
que les baraques avaient souffert d'un terrible incendie,
qui évidemment n'avait pas permis une évacuation paisible.
Ceci est confirmé par un grand nombre de monnaies décou-
vertes à leur intérieur (643 pièces). Les monnaies gisai-
ent sur le plancher en terre des baraques, pour la plu-
part endommagé par l'incendie. Elles ont été découvertes
en 2o groupes, qui s'étaient répandus sans doute des bour-
ses en cuir et qui comptaient le plus souvent 2o pièces.
Le plus grand pourcentage de ces monnaies (plus de 8o %)
a été frappé à Siscia pendant la règne de Valentinien Ier,
Valens et Gratien, entre les années 364 et 378. Les aut-
res pièces, à l'exception de celle de Constance II (3%
environ), ne représentent même pas 1;1~. Le rapport est
presque identique en ce qui concerne les découvertes de
monnaies isolées à ce niveau, mais hors des baraques. La
plus ancien exemplaire appartient à Constance II (351-361)~ 8
Il en resulte que la fortification de Sal-
dum a été anéantie par un incendie provoqué sans doute par
une des pénétrations connues des Goths entre les années
376 et 378 et par les troubles dans le bassin danubien
qui culminent avec la défaite des Romains à Adrianople et
avec la mort de Valens 1 9. Il est sûr qu'après ces événe-
ments la forteresse était restée en ruine. Sa reconstruc-
tion se reflète bien dans les objets archéologiques à par-
tir de Justinien. Les autres fortifications plus petites
des Portes de Fer ont connu un sort semblable 20 Faute d'
0

informatiors épigraphiques, il faut dater la construction

de castellum vers le milieu du IVe s., à l'époque de Con-
stance II ou plus probablement un peu plus tard à l'épo-
que de Valentinien Ier. Une date plus reculée serait dif-
dicilement acceptable, étant donné que l'église faisait
vraisemblablement partie du rempart. Les sources pour cet-
te période, mentiov_nant d'amples travaux de construction
de forteresses et des guerJ?es permanentes avec les Goths,
offrent un appui solide. 21 A nos régions se rapporte un
d8cret impérial de Valentinien Ier adressé au dux. de la
765
 Dacie Ripuaire, Tautomède (19 juin 365) , qui ordonne:" •••
à part les tours qui devraient être réparées, si besoin
en est, construisez (il est question du dux) chaque année
pour votre emploi de nouvelles tours dans des endroits
commodes " (Cod.Theod. X:V 1,13).
Comme preuve de l'activité constructive de
Valentinien rer dans cette région, on pourrait citer le
castellum de Gratia.na, inconnu dans les sour.ces plus an-
ciennes, qui tient son nom de celui du co-empereur (Not.
Dign.Or. XLI,26). L'endroit est cité comme siège de l'au-
xilium Gratianense, entre les~·· castella de Cup:pae et de
Taliata. Etant donné que dans ce secteur la disposition
topographique correspond aux données de la Notitia~2 ,
Gratiana devrait se trouver entre les sites modernes de
Golubac et de Donji Milanovac. Si on exclut Cezava, la
seule forteresse plus importante dans cet espace, l'anci-
enne Novae que la Notitia connait en tant que siège d'un
praefectus militum exploratorum'; théoriquement il ne res-
te que quelques castella plus petits qui viendraient en
considération: Brnjica, Saldum, Bosman (? ), Velike Liva-
.dice, Boljetin et Ravna. Parmi ceux-ci, Saldum est la
se~le forteresse reconstruite dans sa totalité au IVe s.
et de ce fait 1 'identification avec Gratiana semble pres-
que s'imposer. 23
Si notre consclusion est juste, la forte-
resse de Saldum, de m~me que les forteresses similaires
dont on a discuté, illustrent une fois de plus le program-
me ambitieux et l'activité constructive de Valentinien Ier
sur le limes. Tenant compte d'amples travaux qui ont été
exécutés en Pannonie et de ceux qui n'étaient pas moins
poussés sur le Bas-Danube, le secteur des Portes de Fer
qui se trouve au milieu, montagneux et difficilement fran-
chissable, mais très vulnérable à cause de la petite lar-
geur du Danube et des cataractes à cet endroit, avait des
raisons particulières, stratégiques et tactiques, d'être
bien fortifié et protégé.

766
NOTES:
1 Les informations préliminaires sur les fouilles de 1966
-l97o, publiées dans "Arheoloski Pregled" (Chronique des
fouilles), offrent une idée générale sur ces,,
travaux;
il en est de même du catalogue de l'exposition "Acrtcien-
nes cultures du Djerdap" (Eeograd, 1969), encore irrem-
placable.
1
Kondié,l974:39-54, pl.2-ll et . Boskovié,l978:
425-439,fig.l-23, donnent un aper9u sommaire. Il n'y a
pas longtemps qu'une nouvelle exposition sur "Les tré-
sors archéologiques dans la régions de Fort es de Fer 0
a été organisée à Bucarest et à Belgrade (1978), accom-
pagnée de deux catalogues contenant un choix de trou-
vailles archéologique. La modeste liste bibliographique
des travaux sur le limes danubien dans le secteur des
Portes de Fer se termine par deux études complÉmentaires
sur le castellum de Zanes, actueTiement Donje Butorke,
qui a été exploré ultérieurement, en 1972: Cermanovié
-Kuzmanovié, 1979:127-134 et pl.I-IV (sur la forteresse)
et Jankovié, l975:2ol-226 (sur le mobilier archélogique
de la nécropole et de la forteresse).
2. Sur les témoins archéolo~iaues des invasions avaro-sla-
u ....

ve dans l 'Illyricum byzantin, spécielercent dans le sec-

teur des Portes de Fer, voir l'étude très ar6umentée de
Popovié, 1975:478-485, avec bibliographie.

3 Les travaux de bouchage ou de retrécissement des por-

tes, effectuées auVe et VIe s., dûs aux invasions bar-
bares, sont attestées dans un assez grand nosbre de for-
teresses du limes, mais aussi à l'intérieur de lapé-
ninsule balcanique, cf. Ivanov, 1973:24-34,, avec preuves.
C'est ainsi que, vraiser;:.blabl ement au cours du règne de
Justinien, la porte oc>:identale de la forteresse de Ta-
liata a été bouchée et tr~~sformée en baptistère,tandis
qu'à coté de celui-ci a été érigée une église à trois
nefs (T:-?.echerches de R.l_jjubinkovié et D. kinié); c'est
de Eanière semblable qu'a été construite a11ssi l' égli-
se du castellum de :novae, actuel Cezava (Prioal<ovié,
l97o: pl.X~VIII).
767
4 La forteresse de Novae qui a été totalement reconstrui-
te au IVe s. (Pribakovié, l97o:89),est une exception;
par son plan, elle ressemble fort aux forteresses d'Oe-
scus, de Gamzigrad et autres, qu'on croit être caracté-
ristiques pour l'époque de Constantin le Grand ~cf.Ca­
nak-kedié , 1978: 174) .
5 La muraille ne représentait qu'un complément de défen-
se, étant donné qu'il y avait toujours à proximité nn
castellum avec garnison militaire. Cf. Petrovié,l977:
26o (Porecka reka); Pindié-Kovacevié,l97o:9l sq.(3r-
njic a); Petrovié, l97o: 86 (Koz ica ,près de SaldiDI'c).
6 Cf.Petrovié,l977:259-275. Considérant la même question,
Gudea,l974:l73-l8o rattache à l'époque de la tétrarchie,
outre les fortifications citées, d'autres aussi, situ-
ées dans les Portes de FerOrialo Golubinje, Hajducka vo-
denica, Kladovo), qui sont plus récentes et de par leur
plan diffèrent sensiblement des guadriburgia (voir ci
-dessous).
7 L'absence de matériaux de ces tours peut être le résul-
tat de leur mauvaise conservation: la tour à Salduw se
trouvait pour sa plus grande partie sur la rive même
et son contenu a été emporté par les eaux du Danube,
tandis que celle de Zanes avait été considérablement
endorr~agée par les buldozers. Il faut néanmoins menti-

onner que la tour de Zanes était vraisemblablement re-

couverte d'un toit en tuiles, ce qui est prouvé par un
assez grand nombre d'imbrices trouvés à l'interieur
( Cerrr1anovié-Kuzmanovié, 1979:129).
8 Qf. Jllitova-Dzonova, 1974:53 sqq., avec les plans de
quelques églises paléochretiennes fortifiées par des
remparts. Selon cet auteur, les remparts (periteihis-
me) font tout d'abord partie de l'architecture ecclé-
siastique et enseEble avec la cour intérieure (peribo-
los) servaient au déroulement de la liturgie.
9 Cf.Vetters, l95o:53 sqq.et .Abh.7-8. Dans les régions
du nord-ouest de l'Erripire, les petites églises chréti-

768
 ennes ou "garrison chapels", ne sont non plus connues
dans les camps militaires avant la fin du IVe siècle
(Petrikovitz, 1971: 2o3).
lo Cf.Zeiller,l918: 193-196.
ll Nikolajevi6,1978:684sq.et fig.5.
12 Prodanovi6-Zotovi6,1964:55-. Les fouilles de sauvetage
dans cette localité ont été effectuées en peu de ta1ps,
en 1964, mais el~es ont tout de même per~is de relever
le plan de la forteresse. Le directeur des fouilles,
Ljubica Prodanovié, collaborateur technique de l'Insti-
tut archéologique, a aimablement ois à notre dispositi-
on toute la documentation.
13 Cf.Soproni,l978:45 sqq. et note 178, avec un a.w.ple co~­

mentaire.
14 Le burgus de Stelli près de Wallbach en Suisse, et aus-
si les forteresses d~~s le Yorkshire en Angleterre
(stations d'observation sur la côte), appartiennent au
même type, cf. Hinz et autres, 1968:176 sq. et notes
9-lo.
15 Petrovié,N.l97o:55. La tour centrale est placée correc-
tement par rapport aux remparts (épaisseur de murs 7o
cm) et son espace intérieur est dallé de briques dont
les dimensions sont identiques à celles du rempart de
la forteresse.
16 Le plan du castellum de Malo Go lubin je n'est pas enti-
ère11ent connu. Cependant, d'après les tours circulaires
du rempart sud et les escaliers qui donnaient accès à
·celui-ci, ainsi que d'après la technique de construc-
tion, on a déjà noté que ce castellum ressemblait fort
aux forteresses de Saldum et de Donje Butorke,Zanes
(Popovi6,l97o:58sq.).
17 La forteresse de Zanes, actuel Donje Butorke, la seule
qui ait fait l'objet d'une plus ample inforLation, n'a
malheureusement pas_ une stratigraphie sûre, étant don-
né qu'elle a été très endommagée avant les fouilles.
769
 Le mobilier archéologique reflète· deux phase de vie
dans la fortif.~catio:r:t (IV 8 et VIe s. de n. e.). Les trou-
vailles étaient mélangées et n'ont pu être stratigra-
phiquement séparées pendànt la fouille (Jankovié,l975:
2o2). Les monnaies découvertes dans la forteresse, à
l'exception d'un~ ~ièce de Galère, appartiennent tou-
tes à un petit' tr~S:or (17 1,Jièces) qui contient des ex-
emplaires en bronze de Val~tinien Ier, V~lens et Gra-
tien; deux petites pièc-es de Théodose Ier (.378-.383)
sont les plus récentes. Notons qu'à l'ouest du fort on
a pu constater'une agglomération d'où proviennent des
monnaies de Galère et. de Probus. Dans le déblai devant
la tour centrale on a découvert l'inscription de con-
struction du temps de la Tétrarchie (Cerœanovié~Kuzma­
novié, 1978:13·o sq. ,cf.Petrovié,l977:263 sq. ), avec
une autre inscription plus ancienne réutilisée pour
l'érection de la tour. A.Cermanovié-Kuzmanovié, qui
a exploré la fortification de Donje Butorke, estime
que la tour d'observation au centre appartient à 1 'é-
poque de Diocl~tien (praesidium), tandis que le castel-
lum serait construit ulté.rieurement,
....,.,
au VIe siècle •
18' Nous sommes reconnaissants à Vladislav Popovié, profes-
seur à la Faculté des Lettres, qui a mis à notre dis-
position ses notes, nous permettant ainsi d'identifier
les monnaies.
19 Sur ces événements, voir Patsch, 1929:3 sqq.; Vetters,
195o: 28 sqq.; Velkov,l958:124 sqq.
2o Cf.Zotovié-Kondié,l971:43
21 èf. note 19.
·22 Dusanié,l974:422 et riote lo6.
23 Etant donné l '.importance certaine de Gratiana (une des
cinq cités en rv:oesia Prima d'après Hieroclès, (Synecd.,
p.657,2-6), on pourrait citer le fait que Saldum était
probablement le lieu où s·'effectuait la traversée du
Danube ( saldum = "passage tt en turque), et qu'en aval
de cette localité·, jusqu'à Taliata, s'étendait la Gor-
770
 ge supérieure pleine de remous et d' écuails. Ce qui
présente une difficulté complémentaire, ce sont les
toponymes Cantabaza, Smorna et Campsa de la liste de
Procope (De Aedificiis,IV 6,5) qui se trouvent dans
cette région (ils s'identifient selon Kondié,l97l-74:
53-58 à Saldu.m (? ), Boljetin et Ravna); il faut y ajou-
ter aussi Ad Scrofulas (Tab.Peut.VII) qui se trouve
de m~me sur le Denube de la Mésie. Cependant, ces topo-
nym1es ne devaient forcément pas coexister avec Gratia-
na et les plus importantes reconstruction des forteres-
ses d'autre part, pouvaient ~tre à l'origine du change-
ment de leur nom (Dusanié,l974-:4-22 et note lo6).

" ,
LISTE DES REFEREiiiCES
Boskovié,Dj. 1978. Apercu sommaire sur les recherches ar-
' .
chéologiques _du limes romain et paléobyzantin des Portes
de Fer. rv~EFRA, 9o : 4-25-46 3.
Cermanovié-Kuzmanovié,A. 1979. Rimsko utvrdjenje kod Kla-
dova. Starinar,28-29:127-l34 et Tab.I-IV.
1Dusanié,S. 1974. Fragment of a 1~ilitary Diploma from I•.'oe-
sia Superior. Germania ,52:4-oB-425.
Canak-IVedié ,N.. 1978. Gamzigrad: Communication de l 'Insti-
tut pour la protection des monuments historiques de la
République socialiste de la Serbie,XI. Beograd.
Gudea,N. 1974. Befestigungen am. banat er Donau-limes aus
der Zeit der Tetrarchie. Actes du IXe Congrès internatio-
nal d'études sur les frontières romaines:l73-l8o.
Hinz,H. ,Homberg,I., .!3ecker,G., 1\Cattner,H., Geissen,A.l968.
Ein Burgus bei Asperden,Kreis L'oers. Rheinische Ausgra-
bungen,3:167-212.
Ivanov,T. 1973. Sur certaines transforkations du système
des fortifications pendant la Haute époque byzantin (V-
VI s.).Arheologija,4 (Sofia):24-34.
Jankovié,Jj. 1975. Le ffiObilier archéologique de la né-
cropole et de la fort.eresse près de Kladovo. cit arinar,
?·+-25: 2ol-226.

771
Kondié,V. 1974. Ergebnisse der neuen Forschungen auf dem
Obermoesischen Donaulimes. Actes du IXe Congrès int~rna­
tional d'études sur les frontières romaines:39-54 et pl.
1-11.
r.Pitova-Dzonova, D. 1974. Peribolost pri ranohristijanska-
ta bazilika u nas. Proucvani,ja i konservaci,-ja na pametni-
cite na kulturata v Bl~arija, I, Sofia: 53-64.
Nikolajevié,r.· 1978. Les ensevelissements dans les égli-
ses paléochrétiennes sur le territoire de la Serbie.Arhe-
oloski Vestnik, XXIX (Ljubljana): 678-693.
Patsch,C. 1929. Beitrage zur Volkerkunde von Südosteuro-
pa, IV. Sitz.Ber.Acad.-~'Jien • .Fhil.Eist.Kl. :3 sqq.
Petrikovitz,v.H. 1971. Fortifications in the North-wes-
tern Roman Empire from the Thrid to the Fifth Centuries
A.D., JRS, 61: 178-2o4.
Petrovié,P. 197o. Sa1dum. Arh.Pregled, 12: 84-86.
Petrovié,P. 1977. Forteresse romaine à l'embouchure de la
rivière Porecka dans les Portes de Fer. Akten des XI in-
ternationalen Limeskongresses: 259-275.
Petrovié,N. l97o. Hajducka Vodenica. Arh.Pregled,l2:54-56.
Pindié,r/.-Kovacevié,r. l97o. Brnjica. Arh.Pregled,l2:9l-92.
POJ.?OVié,Lj. l97o. I\o"alo i Veliko Golubinje,Arh.Pregled,l2:
58-59.
Popovié,V. 1975. Les témoins archéologiques des invasions
avaro-slaves dans l 'Illyricum byzantin. Iw~BRA,87:445-5o4.
Pribakovié,D. l97o. Castrum Novae, Cezava. Arh.Pregled,
12: 86-91.
Prodanovié,Lj.-Zotovié,}..• 1964-. r~=ihailovac. Arh.PregleQ..,
6: 55-56.
Soproni,S. 1978. Der spatromische limes zwischen Eszter-
gom und Szentendre, Budapest: Akad~miai kiado.
Ve1kov,V. 1958. La construction en Thrace à l'époque du
Bas-Empire (d'après les ~crits). Archeo1ogia, X (Warsza-
wa-/Jroclaw): 124-138.
Vetters,H. l9)o. Dacia Ripensis . .Schriften der Balkankom-
mission, Ant. Abt. , 11. Vlien.
Zeiller,J.l918. Les ori5ines chrétiennes dans les provin-
ees danubiennes de l'empire romain. Paris.

772
t3otovié,Lj.-Kondié,V. 19?1. :rtor..can and early byzantiœfor-
teresses in Jjerla:p, r..-aterijali, VI. VIII Congrès des ar-
chéologues yougoslaves, ~eobrad: 37-54.

Summary

Sorne thirty fortified structures spread out along 90 kilometres

of the bank of the Danube in the Portes de fer (Iron Gates) sector
(excavations carried out 1966 ·1971) bear witness to the different
types of building of the late Empire. Apart from partitioning and
structural completion work carried out mainly on the most important
fortresses (Novae, Taliate, Boljetin), 3 types of construction took
place: (a) walls several hundred metres long preventing invasion
from the direction of the valley (Brnjica, Poreéka reka, Kozica);
(b) smaller forts, castella; (c) burgi and towers.
Particularly interesting are the forts in the (b) category, which
were built in two stages. First there are castella built on territory
that had not' y et been reached at the time of Diocletian and Constantine
(Ravna, Porecka reka, Tekija, Sip and others on the left bank of the
Danube), as can be gathered from historical sources, as well as from
the inscription of Kladovo. Probably belonging to the second stage
are the castella. most of which have survived the destruction of the
third century, who se size and layout are similar: Saldum, Golubinje,
Donje Butorke, Hajduék:a Vodenica and Mihailovac. Statistical data
on monetary circulation and archaeological finds enable us to date
their construction towards the middle of the fourth century, at the
time of Constantius II and Valentinian I.

773
 51. UNTERSUCHUNGEN DES FORUMKOMPLEXES
IN DER COLONIA ULPIA OESCENSIUM (1975-1978)

Teofil Ivanov

Die archäologischen Ausgrabungen und Untersuchungen in der römischen

Colonia Ulpia Oescensium beim Dorf Gigen (Bezirk Pleven), 5 km sildlieh der
Donau in der Provinz Untermösien, i!'l den letzten Jahren (197 5-197 8) konzen-
trierten sich ausschliesslich auf die ~läche des Forumkomplexes, genauer
auf dessen nördlichen Teil.
Noch im Jahre 1973 wurde die Flucht des decumanus maximus (Ost-West)
festgestellt und teilweise freigelegt. Er läuft an der südlichen Seite des
Forumkomplexes vorbei (Ivanov 1977a)(Abb 51.1). Das erste wichtige Ergeb-
nis war die Entdeckung und die Ausgrabung eines Teiles des cardo, der die
westliche Seite des Forumkomplexes berührt. Er ist 5, 90 m breit und mit
Kies belegt, während die anderen bis jetzt bekannten Strassen Steinplatteu-
pflaster haben. Die zweite Hauptstrasse-cardo maximus-ist noch nicht
sicher lokalisiert. In den römischen Städten Unt.ermllsiens und Thrakiens
(heute VR Bulgarien) läuft gewöhnlich eine der beiden Hauptstrassen durch
die ganze Stadt und·berUhrt eine Seite des Forumkomplexes, während die
andere Hauptstrasse im rechten Winkel auf die erste Hauptstrasse stösst und
den Buchstaben T, bildet. Sie hört am Forum auf und setzt sich hinter ihm
fort. Die Wagen fuhren daher durch die Strassen, die um das Forum herum-
liefen. Dies ist der Fall z.B. bei den Fora von Novae (Svischtov)(Cicikova
1977; Parnicki- Pudelko 1976; Sarnowski 197 8; Press 197 8), Nicopolis ad
Istrum (beim Dorf Nikjup, Bezirk Veliko Tarnovo) (Ivanov 1977b, abb. 2),
Serdica (heute Sofia), Pautalia (Heute Kjustendil), Augusta Traiana (heute
Stara Zagora) u. a. Dasselbe städtebauliche Prinzip ist auch in anderen
Städten ausserhalb der VR Bulgarien festzustellen.
Der Forumkomplex von Oescus liegt in der Stadtmitte und dat eine recht-
eckige Form mit folgenden Aussenabmessungen: Breite (Ost-West) ca. 100
m und Länge (Nord-Süd) ca. 200 m. Er umfasst also eine Fläche von ca.
2 ha. Die Gesamtfläche der Stadt in der ersten Periode betrug ca. 18 ha.
Der freie Platz im Forum hat folgende Masse: Breite (Ost-West) 57,60 m
und Länge (Nord- Stld) 98,20 m. Zwei Fünftel dieser Länge nimmt der
Forumtempel mit seinen zwei Höfen ein; er war wahrscheinlich dem Iupiter
gewidmet. Ein Forumtempel fehlt aber in Novae, Nicopolis ad Istrum,
Serdica und anderen Städten in Untermösien und Thrakien.
Das wichtigste Ergebnis der Ausgrabungenin Oescus in den letzten
Jahren ist aber die Entdeckung und teil~1eise Freilegung der Forumbasilika,
die sich ganz dicht an die Nord\•::::nd c2s Forumtempels anschliesst. Es
handelt sich um eine monumentale dreischilfige Basilika hellenistischen Typs
·mit folgenden Aussenabmessungen: Länge (Ost-West) ca. 1oo·m und Breite

775
 1;:....=
§

§
B A s I L I c A

!
Ia

\
l.t

.T. F=
\1

r-
1-

F 0 R u M

..
H

Cl
I
10
H
"'
< H
--
~~
u

=
:z
%: lo ~
IM ~
~~ ~.
:;- ~
~ ~
~ IM
JC

:z
iO
:::: IP
lac: I p
c TF

z I<
z Iu p

::::::: ~ ::::::-::::::::::-

0 10 20
•, · ··· • Metres

Abb. 51. 1 Grundriss des Forumkomplexes von Oescus (VR Bulgarien).

(Aufnahme: S. Daskalova und P. Vruev) -

776
 (Nord-Süd) 23,60 m (Abb 51. 2). Die drei Eingänge befanden sich an der
Westseite der Basilika, an der der neuentdeckte cardo verlief (Taf. 51.1
und 2). Die Länge jedes der drei Schiffe beträgt 72,45 m. Die Breite der
Seitenschiffe ist 3, 65 m und des Mittelschiffes 11, 50 m. Am östlichen Ende
der Basilika befand sich ein noch nicht ausgegrabener Raum, in dem ver-
mutlich das Tribunal stand. Unmittelbar hinter dem Haupteingang des Mittel-
schiffes ist ein viereckiger Vorraum vorhanden (6;. 20 m x 11, 50 m). Das
Mittelschiff hat eine zweistöckige Säulenstellung im korinthischen Stil. Die
Säulen des unteren Stocks sind rund. An Ort und Stelle sind runde Basen und
monolithische Säulenfragmente erhalten. Der Achsenabstand zwischen zwei
benachbarten Basen beträgt ca. 2, 90 m. Das ist die Länge der einzelnen
Blöcke des Freisarchitravs, von dem einige Fragmente gefunden worden sind.
Der freie Abstand zwischen den Säulen ist ca. 2, 15 m. Der Fries ist mit
einem reichen durchbrochenen pflazlichen Ornament verziert (Taf. 51. 3).
Der Architrav hat die ilblichen drei Streifen (fasces) und Eierstabornament.
Auf einem Friesfragment aus Kalkstein sind einige grosse Buchstaben einer
lateinischen Inschrift erhalten, die die Titulatur eines römischen Kaisers
enthalt, in dessen Regierungszeit die Basilika errichtet worden ist. Man liest
folgendes: [I]MP ..,II.. C[OS].... Im zweiten Stock befanden sich monolithische
rechteckige Pfeiler aus Kalkstein, die an der Vorderseite ein Hochrelief von
jungen Frauen (Karyatiden) hatten (Taf 51.4). Die Figuren standen auf einem
hohen Postament, vielleicht ca. 1, 80 - 1, 90 m hoch. Die Höhe der Pfeiler
betrug vermutlich 4, 30 m. Bis jetzt wurde kein vollständig erhaltener Pfeiler
gefunden. Die Höhe kann man aber auf Grund der gefundenen Fragmente
abschätzen. Fast alle Karyatiden ähneln einander, obgleich sie von verschiede-
nen Bildhauern gearbeitet worden sind. Sie stehen aufrecht mit noch vorn
gerichtetem Gesicht. Manchmal haben die Augen keine Pupillen, weil sie
von den Betrachtern infolge der grossen Höhe (ca. 10 m) nicht gesehen werden
konnten. Die Karyatiden stützen mit den beiden Händen oder nur mit der
rechten Hand einen Korb (Kalathos) auf dem Kopf, während die freie linke
Hand am Körper herabhängt (Taf. 51. 5). Sie sind in einem langen Chiton
und Kurzen, gegürteten Peplos gekleidet. Der Stil der Karyatiden, der
gesammten Architektur und die Stratigraphie des Gebäudes deuten als Datum
der Errichtung der Basilika frühestens auf die hadrianisch-antoninische Zeit.
Es ist interessant zu erwähnen, dass die Pfeiler mit Reliefdarstellungen
von Karyatiden und anderen Gestalten in der monumentalen römischen Archi-
tektur seit der Mitte des2 •. Jhs. die vollplastischen Statuen (Karyatiden u. a.)
gänzlich ersetzt haben. ·
Unter den Palmetten des Gesimses der Forumbasilika von Oescus gibt
es zwei Typen gleichzeitig nebeneinander. Bei dem einen Typ ist die Palmette
nicht unten mit einem Band versehen, bei dem zweiten Typ trägt sie ein Band.
Es gibt viele gut datierte Beispiele in der Architektur aus Kleinasien, Unter-
mösien und Thrakien zur Begründung unserer Beobachtung. Die Blätter der
Palmetten des Forumtempels in Oescus, der etwas früher als die Forum-
basilika ist, sind nicht gebunden. Aber die Palmetten des Gesimses der
Agora von Nicoplis ad Istrum sind immer gebunden. Die früheste bis jetzt
gefundene Inschrift von der Agora stammt aus dem Jahre 136 (Mihailov 1958,
n. 601). Also hat diese V@randerung in späthadrianischer Zei~ stattgefunden,
naturlieh nicht platzlieh und nic_ht tlberall im römischen Reich.

777
 E
0
N
-
0
0
u
<{
a:l
778
Taf. 51. 1 Die Basilika von Oescus. Blick von Westen (A~fuahme: T. Ivanov).
Taf. 51. 2 Die Schwelle des Mittelschiffs der Basilika (Blick von Norden. Aufnahme: T. Ivanov).
Taf. 51. 3 Fries-Architrav des Mittelschiffes der Basilika (Aufnahme:
T. Ivanov).
Taf. 51. 4 Zwei Fragmente von Karyatiden im Vorraum des Mittelschiffes
der Basilika ! (Aufnahme: T. Ivanov).
Taf. 51.5 Karyatidenkopf vom Vorraum (Aufnahme: T. Ivanov).
 Die Karyatiden von Oescus finden sich zwn ersten Mal in Untermasien
und Thrakien. Zweifellos verraten sie den Einfluss der griechischen Archi-
tektur und Kunst Kleinasiens aus hellenistischer Zeit. Diese Tatsache zeigt,
wie lebendig diese Tradition in der Römerzeit im heutigen Bulgarien war.
Viele Inschriften bestätigen die Anwesenheit von Leuten kleinasiatischer
Herkunft.
Ein gute Analogie zu de:f"architektonischen und künstlerischen Verwendung
unserer Karyatiden aus Oescus als Bestandteil des oberen Stocks der Forum-
basilika bilden die Pfeiler mit hohen Reliefdarstellungen an der Aussen- und
Innenseite eines affentUchen Portikus aus der zweiten Hälfte des 2. Jhs. u. Z.
von der Stadt Thessalonike, der unmittelbar südlich der Agora noch bis ins
19.Jahrhundert sichbar war. Die Darstellungen zeigen ver~chiedene Gestalten:
Mänaden, Dionysos, Ariadne und Leda und an der entgegengesetzten Seite
Nike, Aura, einen Dioskuren und Ganymed. Heute befinden sie sich in Paris.
Der Portikus ist in der Literatur unter der spanischen Benennung "Incantadas"
(in neugreichischer Sprache unter Uow'Aa. ) bekannt (Guerruni 1961; Mna.xl.a'ti:;Yjc;
1977).
In Oescus haben wir folgendes Schema des Forumkomplexes: die Forum-
basilika im Norden, den Tempel des Hauptgottes in der Mitte und das Forum
in Süden. Es unterscheidet sich vom gewahnlichen Schema, bei dem der
Tempel und die Basilika an der Schmalseite des Forums einander gegenfiber-
liegen.
Die neuen Ausgrabungen und Untersuchungen in Oescus haben gezeigt,
dass die Stadt und die Forumbasilika von den Goten in den Jahren 376-37 8
durch Brand zerstört worden sind. Die spätesten Bronzemünzen aus der
Brandschicht des Mittelschiffs der Basilika stammen aus der Regierungszeit
des Kaisers Valentiman I. (364-37 8). Wie gut bekannt ist, ereignete sich die
entscheidende Schlacht zwischen den Goten und Römern, in der Valens den
Tod fand, Anfang August 37 8 bei Hadrianopolis (jetzt in der Türkei).
In diesem Jahr (1979) wurden das ganze Mittelschiit und die nordwestliche
Ecke des Nordschiffs ausgegraben. Nächstes Jahr (1980)·möchten wir
möglicherweise das Nordschiff vollständig freilegen. Später werden wir
unsere Aufmerksamkeit auf den östlichen Teil der Basilika konzentrieren.
So h9ffen wir, dass wir hinreichend viele Belege für den gesamten Grundriss,
die vermutete graphische Wiederherstellung, die Datierung und die Geschichte
der Basilika gewinnen werden.

LITERATUR

Cicikova, M.' 1977. 'La fortification et l'urbanisation de Novae a la turniere

des dernieres fouilles archeologiques (1970-1974)', in Studien zu den
Militärgrenzen Roms II, KlHn-Bonn, 277-282.
Guerrini, L., 1961. 'Las Incantadas di Salonicco', Archaeol. Class. xiii,
40-70.
Ivanov, T., 1977a. 'Untersu~ungen zur Topographie und zur stadtlanlage von
Oescus in Untermoesien (Bulgarien)', in Studien zu den Militärgrenzen
Roms Il, Köln-Bonn, 339-35C.
IvanoY, T., 1977b. 'L'agora de Nicopolis ad Istrum d'apres les nouvelles
recherches'. Arheologija (Ebfia) xix, 16ff.

784
Mihailov, G., 1958/• Inscriptiones Graecae in Bulgaria repertae II, Sofia.
M~axLp~C~~, X. 1977. "ITEpl ~oD ouyxpo~~~a~o~ ~~; 'Ayop&~
8EC"l'C1a:f...ov[xr,~", 'Apxa:ta 1'·1a:xE6ovCu: II, 257 f.
Parnicki-Pudelko, S., 1976. 'Befestigungsanlagen von Novae', Ars Historica
1976, 179ff.
Press, L., 197 8. 'Travaux de l'expedition archeologique polonaise ~ Novae ',
Secteur ouest 1977, Latomus xxxvii, 960-65.
Sarnowski, T. 'Fortress of the legio I Italica at Novae,. in Fitz, J. (ed. ),
Limes: Acten des XI. Internationalen Limeskongresses, Budapest,
409-426.

SUmmary

Recent excavations and investigations (1975-1978) in the colonia

Ulpia Oescensium near the village of Gigen, District Pleven in the
province of Lower Moesia, were concentrated in the northern part of
the forum complex. New infgrmation was gained concerning town
construction and public architecture. The second unknown main
street (cardo maximus) was discovered west of the forum and was
partially exposed. The two main streets-cardo and decumanus-
define the west and south sides of the forum complex. This is the
firsttime tp.at such a special feature has been established in a Roman
town in the provinces of Lower Moesia and Thrace (now Bulgaria).
By way of contrast, usually one of the main streets runs right through
the town, while the second main street stops in front of the forum and
continues behind it, as for example in Novae near Svischtov, Nicopolis
ad Istrum near the village of Nikjup, District Veliko Tarnovo, and
Serdica (now Sofia), Patitalia (now Kjustendil).
The excavations have yielded the following dimensions for the
forum complex: Length (North-South)-183 m, and width (East-West)
-100 m. The temple of the main deity, Jupiter or the Capitoline
Triad, is located in the middle of the north side of the forum.
However the most significant discovery is the secular basilica,
of hellenistic type, which is tripartite in structure and dates from the
second half of the second century A. D. It is situated immediately to
the north of the forum complex. It is partially excavated. The width
is 22.30 m and it has a probable length of 100m (East-West). The
aisles are 3. 5 m wide and the nave 11.80 m. Each of these has its
own entrance. The nave has an ante-room (propylaea) and two-storeyed
colonnades on the long sides. The first storey consists of monolithic
round columns made of white chalkstone in the Corinthian style, and
the second storey consiRtR of monolithic supporting pillars with pro-
nounced reliefs of Cary<:~tids. The distance between the pillars is
2. 90 m. This is the firsttime that buildings from ancient times with
Caryatids have been found in Bulgaria.
During the Gothic incursion in 376-37 8 Oescus and the basilica
were destroyed, as is confirmed by the latest coins from the reign
of Valentinian I.

785
 52. SACIDAVA- AN UNUSUAL DESIGN AND CONSTRUCTION
METHOD ON THE LATER ROMAN LIMES

C. Scorpan

The Roman fortress of Sacidava is situated in the SW of Scythia Minor,

(the modern Dobruja) on the right bank of the Danube. The location of this
fortress is based on literary, epigraphical and archaeological information
(Not. Dig., Pars Or. XXXIX-XL; Gostar 1963, 299; Radulescu 1969, 349;
Aricescu 1970, 298; Scorpan 1972a, 301; 1972b, 349; 1974, 109; 1977a,
299; 1977b, 203).
Sacidav.a was built at the beginning of the second century A.D. and lasted
continuously, but with reconstructions, until the first half of the seventh cen-
tury. The oldest level belongs to the early second century .1 The defences
functioned until the end of Heraclius' reign; two further levels prove con-
tinuity in Roman occupation until the mid-seventh century. 2
Special attention may be drawn to the fact that the defences (seven curtain-
walls and seven towers) uncovered up to now, belong to the Late Roman epoch
(ranging in date from the end of the third to the early seventh centuries A. D.).
All the towers are rectangular in shape. The length of the frontal side of the
towers varies from 8. 60 to 14 m. 3 The lateral walls of the towers vary in
length from 4. 80 to 6. 85 m. The thickness of their walls varies from 2 to
2. 50 m. Some towers have been preserved underground to a depth of 8 m.
The precinct-walls and the towers of Sacidava were built using ingenious
methods previously unknown on the Lhnes or in the fortifications of the interior.
The outer face is built of big and very big blocks laid horizontally in regular
layers. Outside, the int~rstices were filled with fine plaster. At the corners,
the blocks were disposed crossways. The courses of masonry in the upper
part of the outer face were built of big slabs set on edge but lying transversely
and penetrating deep into the concrete core (Pl. 51. 1). The overwhelming
majority of the blocks and slabs in the tower face are epigraphic and architec-
tural monuments from the early Empire re-used as building stone at the be-
ginning of the fourth century (cf. Pl. 51. 2 and 3).
We present here only two towers, but they offer enough data and elements
to reflect the general situation:
. Tower B: A monumental, perfectly preserved tower, which still stands
to a height of 4 m. It is rectangular in shape. The front face is 9.13 m long,
the east side is 4.65 m long and the west side is 5.35 m long. A lot of spolia
has been used in this tower too. Six layers of facing blocks survive with the
seventh and the eighth layers leaving clear trace of caementa (Pl. 51. 2).
Very large blocks lying one over the other were used for the socle and the
foundation. Very big blocks, alternately crossed, were also used at the bases
of the corners.

787
 Pl. 52.1 Sacidava, Tower C.

Pl. 52.2 Sacidava, Tower B with re-used inscription.

Pl. 52.3 Sacidava, Tower B with re-used inscription.

\
 The majority of blocks used in the outer faces and base of this tower were
architectural and epigraphic fragments from the early Empire. For example,
there is in the uppermost course of masonry a squared block from an old
architrave that still retain~ on the external side the terminal letters of a
Latin inscription (...... HENS). It is the first and the only external inscrip-
tion which was not chiselled out. Also placed on the frontal wall was a grave
stele, with its inscription facing outwards. The inscription itself had after-
wards been completely chiselled out (Pl. 5L 3). An altar has also been incor-
porated with the inscribed face facing outwards at the eastern corner. The
whole inscription had been erased sometime after the block was built in.
Ten big altars were placed at the corners of the tower foundations; one in
marble is also very nicely finished and deserves special mention. On the
upper part of the tower, which is today missing rows Nos. 7 and 8, there
remain in the emplecton's mortar clear traces in negative of some inscriptions;±
Other slabs in the face are inscribed. The socle of tower B has been built
entirely of large architectural fragments carefully positioned according to
the alignments required by the architectural design.
Tower D was situated on the W side, that most vulnerable to attack. It
has the same rectangular shape. The front wall is 14 m long, the other two
5 m long. The face is preserved to a height of 6. 50 m (if we add the emplecton,
to a height of 8 m). My first remark is that the face of this tower has been
built using only large and very largp hlncl<s, placed lengthwise one over the
other in regular rows, giving an impression, .and reminding us of, the Greek
tradition of massive walls. Beside the socle, which is itself very ingeniously
built, we can see eight rows of blocks. The corners of the tower, even their
foundations, have been built of very big blocks. Old monuments have also
been used .here, especially altars (in the face) and architectural fragments
(in the socle). Also of interest is the fact that some of the slabs and blocks
were specially cut in steps for a perfect join.
Some of the construction methods used in the building of the new Late
Roman precincts are extremely interesting, representing as they do the appli-
cation of a general architectural design responsive to local conditions. The
first characteristic of the Later Roman fortress of Sacidava is that all its
towers were rectangular. None of these towers, even those in the corners,
has a rounded front-rounded towers are the distinguishing feature of the later
epoch. Another characteristic feature of the faces of the towers is the massive-
ness of the lower parts where very large slabs are very carefully placed, a
feature which gave the towers an extraordinary power of resistance. In all
the towers of Sacidava (a remarkable proportion) old monumental, epigraphic
and architectural fragments of earlier times have been used. The inscribed
faces of the monuments scarcely ever face outwards. Nevertheless a few do
have the inscription to the outside. Neither the shape of the stone, its size,
nor the place in the wall-face (all the inscriptions with their face to the inside
are untouched) justify this fact. We feel inclined to think that at the moment
of building of each tower, such exceptions or such negligence were overloo 1<ed
and tolerated. But later on~ when Christianity became intolerant, the pagan
formula from the early Roman altars and funeral inscriptions was chiselled
out. Only one inscription ( ..•.•. HENS), a word-ending without pagan signifi-
cance, remains untouched. All the towers at Sacidava were strengthened in

790
the second half of the sixth century with a double wall on the inside which
provided solid platform-bastions for the artillery of that time (the most mas-
sive bastion is sited at the back of tower C, in the SW corner of Sacidava, a
vulnerable point (Scorpan 1977b, 236, figs. 7, 8)).
Another interesting fact to be noticed is the evident difference between
the massiveness, the power, and the monumentality of the towers on the one
hand and the curtain walls on the other, which have only a shallow foundation
(30 cm - 1 m) and are faced in a degenerate opus guadratum (which we meet
with in fact at all the other fortresses along the limes). Nevertheless, sur-
prising elements have appeared on the curtain-wall too. Curtain c has at its
base 3-5 steps which represent a local method of reinforcing the resistance
against landslips, given the marked slope which starts right from the wall
and the shallowness of the foundation (Pl. 52.4). Curtains f and B. have
been entirely plastered on the external surface with distinct skill and elegance.
Probably this method is a local innovation to protect the wall against rain-water.
Still, the wish for beauty must not be excluded; curtain B has plaster on its
external surface decorated with incised spiral motifs disposed in regular lines.
The same plaster technique and same decoration were found on the internal
walls of a house from the fourth-sixth century (Domus I).
We will examine now the principal entrance gates of the fortress. On the
S side only one narrow gate (for pedestrians) has been discovered; it was later
blocked. The principal gate, flanked by two square towers and a massive
additional fortification, has been located on the E side (Scorpan 1976, 235,
243, figs. 2.6). A second principal gate was discovered exactly at theW
limit of the main street that starts from the E gate. This W gate has a par-
ticular structure. I interpret its function as follows. Because of the need
to place a regular sequence of towers along the easily accessible west wide,
and because Tower E was of necessity placed at the W extremity of the main
street (the latter being one of the earliest elements of the site-plan, and not
to be altered) the second main gate was opened without the normal flanking
towers of the Late Empire, in the middle of the front face of Tower E (Fig.
52.1). After the abolition of the pedestrian gate, a little square tower was
erected between towers A and B~ on the S side, in order to fortify the defence.
On the W side, a break in the defence wall (curtain e), was repaired but at
the same time the thickness of the wall was doubled with the construction of
a supplementary external wall (1. 50 m thick).
That is how Sacidava revealed to us new and original methods in con- ·
struction and innovation. At the time of the rebuilding and extension of the
Later Roman precinct (fourth century) account was taken, in a consistent way,
of the topography so that the defence wall exactly follows the crest line over
the natural slopes.
I have it as a general rule that the elements of the fortification were
erected by stages and in a certain succession. First, the perimeter line was
established (with via sagularis), which accentuated the crest and the slope by
deepening the ditch outside and sometimes bringing clay on to the crest. Then
the foundations were dug and the towers erected. And lastly, when two
neighbouring towers were finished, the curtain-walls were attach~d to them.

791
Pl. 52.4 Sacidava, curtain wall at c.
<!
~
0
-
()
<!
Cl)
.
C)
LL.
793
 Stratigraphic and numismatic evidence (Valens coins on the construction
level) and the fact that only towers G, F and B have inscriptions built in with
their faces outside, makes us .think that the erection of the new precinct of
Sacidava began with the S side (the SE corner) and progressed gradually with
towers A, C, D, E and so on, until the end. The period may cover the years
from Diocletian until Constanti.us II {providing that the old precinct was quickly
repaired after the catastrophe of 295 A. D. ), which permitted a systematic
construction of the new walls in the new technique.
As regards the size of stones on the faces of the towers, the towers of
Sacidava are clearly superior in massiveness and monumentally to the other
Roman fortifications along the Lower Danube and Scythia (cf. Stefan 1940,
401; Barnea 1961; 1971, 44; Florescu 1958; Parvan 1911). These obser-
vations, the data and the analogies direct us to a new conclusion. The analo-
gies that we can draw between Histria and Sacidava are important and signifi-
cant as regards the architectural conception and construction.
a) The sizes of the frontal sides of the towers are relatively similar (at
Histria the tower D=9.40 m, E=12. 58 m, G=l5. 33 m); but the towers of
Sacidava lie closer together.
b) The facing blocks from towers at Sacidava are bigger, more massive
and,given the height to which the towers B and D of Sacidava survive, overtop
Histria (Histria tower E=2.30 m, G=2.66 m, F=3. 80 m), (Condurachi 1954).
c) At Histria behind the large tower G, the curtain wall is thickened by
1 metre achieving a platform 5 m broad.
d) Some corner towers at Hi stria (T, F, G) were built in the same way
as Sacidava.
,,
e) At Histria, but rarely and at larger and irregular distances, some slabs
are placed on edge, transversely, with their long sides going in to the core.
At Sacidava they are placed with greater regularity.
f) The massiveness of the bases of all the Sacidava towers can be seen
at Hi stria too; at the towers of the principal gate.
g) At Histria too, some blocks have been especially cut in steps, in order
to link in with the blocks of the lower courses.
h) Epigraphic monuments have also been reutilised in the walls at Histria
but proportionately less than at SaCi-.:.a.v a.
It is at this moment not possible to make a definitive statement about this
problem, but I am personally of the opinion that we must recognise here many
elements of Oriental-Greek influence and tradition, viz. exclusively rectangular
towers, the massive technique of the large blocks in the walls, blocks super-
imposed crosswise at the corners, alternate rows of headers and stretchers,
the massiveness of the bases of the towers, regular series of slabs set into
the core, stones cut in steps for perfect superimpositio.p., the decoration on the
external face of the curtain and supplementary walls at the back of the towers
and on the outside of the curtain. All these data predispose us to think of an
architect or builder from the Greek-Oriental world coordinating the work.
Also we must remember the four highly significant epigraphic documents of

794
Pl. 52. 5 Brick stamp - LEG XI CL(A UDIA)

Pl. 52.6 Brick stamp- LEG! ITA(LICA)

Pl. 52. 7 Brick stamp - (LEG XI CL) TRAM(ARISCA)

Pl. 52. 8 Brick with fluent Greek writing.

cohors I Cilicum, the oriental names and Greek letters ( z, 6) in the inscrip-
tions from Sacidava, and finally, fluent Greek writing on the bricks (Pl.
52. 8).
I have presented here an interim explanation of the design and method of
construction of the fortress of Sacidava, so original in the Later Roman epoch
and quite unique on the Lower Danube limes from the period of Diocletian down
to Heraclius.

NOTES

1. Some parts of the base of the W precinct wall, and a building with mar-
tared stone walls discovered near this wall, have been dated to the 2nd
century. Inscriptions testify to the following military units at Sacidava
in the 2nd and 3rd centuries: cohors II Gallorum, cohors I Cilicum, and
a detachment of legio V Macedonica; cf. Aricescu 1974, 261; Scorpan
197 4, 113, pl. 27; 1977b, 203. Moreover, in 1979 three altars with
Latin inscriptions were discovered referring to cohors I Cilicum (two of
the 2nd century and one of the time of Philip the Arab). This new material
suggests that the cohors I Cilicum was stationed for a lengthy period at
Sacidava (Castellum Cilicum). · Also in 1979 we discovered some bricks
stamped by legio I Italica, legio XIII Gemina, and legio XI Claudia
Tra(s)marisca (Pl. 52.5-7).
2. Scorpan 1972b, 349; 1973, .267; 1978, 155. Finds from the first half of
the 7th century are coming to light in every new campaign of excavation.
3. Tower G, in the SE corner: 8. 65 m; Tower F: 10 m; Tower B: 9.13
m; Tower A: 10.21 m; Tower C: 8.6 m; Tower D: 14 m; Tower E:
10.26 m.
4. The tombstones were discovered in the summer of 1976 in the demolition
layer on the outside of tower B; cf. Scorpan 1977b, nos. 3-5.
5. Note that from Histria and Sacidava have come bricks bearing identical
stamps of the Emperor Anastasius. Different stamps have been discovered
in other fortresses of the limes. We must emphasise once again the
weaknesses of some complicated but false typologies for Late Roman for-
tifications, e.g. Florescu 1972, 23-4. Sacidava most certainly does not
belong to the so-called "Carpathian-Danubian type" as proposed by Florescu
197 8. For a condemnation of the tendency to classify the Late Roman
fortresses under types cf. Petrikovits 1971, 203.

BIBLIOGRAPHY

Aricescu, A., 1970. 'Quelques precisions sur la carte de la Scythia Minor',

Dacia xiv, 298-309.
Aricescu, A., 1974. 'Deux Inscriptions des alentours de Sacidava', Pontica
vii, 261-4.
Barnea, I., 1961. Dinogetia, :Bucharest.

797
Barnea, I., 1971. 'Dinogetia et Noviodunum, deuz villes byzantines du Bas-
Danube', Revue des Etudes Sud-Est Europeenes iii, 343-362.
Condurachi, E., 1954. Histria, I, Bucharest.
Florescu, G., 1958. Capidava, I, Bucharest.
Florescu, R., 1972. 'Limesul dun~rean in perioada tirzie a imperiului
roman', Buletinul Monumentelor Istorice iii, 23-4.
Florescu, R., 1978. 'Sacidava' in Condurachi, E. (ed. ), Dobrogea, 305,
Bucharest.
Gostar, N., 1963. 'Monumente epigrafice inedite din lapidariul Muzeului
regional de arheologie Dobrogca', Studii Clasice v, 299-313.
Parvan, V., 1911. 'Cetatea Tropaeum', Buletinul Comisiunii Monumentelor
Istorice iv, 96.
Petrikovits, H. von, 1971. 'Fortifications in the North-Western Roman Empire
from the Third to the Fifth Centuries A.D. ', J. Roman Stud. lxi, 178-218.
Radulescu, A., 1964. 'Un miliar de la Decius la Rasova', Revista Muzeelor
iv, 349o
Scorpan, C., 1972a. 'Sacidava si unele probleme stratigrafico-cronologice
ale limesului (Sec V in arheologi.a dobrogeana)', Pontic a v, 301-28.
0

Scorpan, C., 1972b. 'Date arheologice referitoare la sec VI-VII pe teritoriul

Dobrogei', Pontic a v, 349-72.
Scorpan, C., 1973. 'Sap~turile arheologice de la Sacidava (1969-71)',
Pontic a vi, 267-331.
Scorpan, C., 1974. 'Sacidava - a new Roman fortress on the map of the
Danube Limes' in Pippidi, D. M (ed. ), Actes du IXe Congr~s International
0

d'~tudes sur les Fronti~res Romaines, 109-116, Bucharest-Cologne.

Scorpan, c.' 1977a. 'Rezultate ale sapaturilor arheologice de la Sacidava I

Pontica xt 229-51.
Scorpan, C., 1977b. 'StNes funeraires in~dites de Sacidava' in Pippidi,
D. M and Popescu, E. (eds. ), Epigraphic a (Travaux dedi~s au VIle
0

Congres d'~pigraphie greque et latine) 203-221, Bucharest.

Scorpan, C., 1978. 'Descoperiri arheologice diverse de la Sacidava'
Pontica xi, 155-80.
Stefan, G., 1940. 'Dinogetia I- Risultati della prima campagna di Scavi
(1939)', Dacia vii-viii (1937-40), 401-25.

798
 53. RECENT RESEARCH ON THE LIMES OF ROMAN DACIA
N. Gudea

This report comprises the results of archaeological investigations carried

out during the last ten years with the aim of recovering the defensive system
of Dacia in general and its limes in particular.1 The report considers the
general military organisation of the Dacian Roman provinces, the problems
of the limes, technical details of the fortifications, together 'With new informa-
tion about troops and the line of advanced watch and signal towers, with the
aim of acquainting a larger circle of scholars with the results of Romanian
archaeological research into the military history of Roman Dacia (lmowing
that not all the Romanian publications are familiar to the majority of scholars)
and adding new data to the military history of the Roman Empire.
Because of the good state of preservation of the Dacian forts, systematic
excavation has revealed interesting, and important data concerning the prob-
lems of military history of the limes, especially plans of the fortifications,
their construction system, architecture, dating and the tactical movements
of the troops. Although archaeological excavation and research on the limes
of Dacia first began in the middle of the last century the number of forts
which were well known remained limited and the manner and methods of
excavations were not advanced until 1948. We can state that proper research
into the defensive system began after that date when excavations took on a
systematic and organised scale. 2 But the majority of the forts have been
examined during the last 10-15 years. A great number of forts were identi-
fied and important details obtained concerning the building system, the stages
of construction, inte:r'ior organisation and the nature of the archaeological
material. A more precise opinion about the organisation of the defensive
system could then be presented. It was possible to establish the place, the
role and line of the limes and the stages of the defence, troops etc. 3 Of
course, we cannot pretend that the problems of Dacian military history are
definitely solved, but if unsolved questions still exist, the data we possess
today leaves less room for purely theoretical speculation. We use the term
theoretical speculations because, in spite of the progress made in the last
decades and although good monographs and studies have appeared dealing
with the military history of Dacia, in the specialised literature there are
still in use a great number of older books dating from the so-called romantic
period of research, which present a picture of the defensive system of Dacia
which is, in varying degrees, incorrect. 4
In order to present the defensive system of Dacia one must take into
consideration the general aspect of the Carpathian relief on the one hand,
and the organisation of the defensive system on the Middle and Lower Danube
before the conquest of Dacia, on the other. Recent research has confirmed
that the conquest of Dacia was first of all the result of military necessity.

799
Dacia's relief offered an excellent strategic position, which was necessarily
used in the defensive system of the Roman empire. The high Transylvanian
plateau, surrounded by mountains, formed a natural bastion. The domination
of this, strengthened later by artificial defensive works, secured peace for
the Empire not only in the province of Dacia, but also on the neighbouring
parts of the Middle and Lower Danube. The changes that took place in the
organisation of the defence of Pannonia Inferior (Mocsy 1962, 614-6 and 632-
8), Moseia Superior (Gudea 1977 and 1978) and Moesia Inferior5 clearly
show the importance of the Dacian bastion and the part played by it in the
political and military history of this area.
The study of the defensive system of Dacia is difficult because of the
lack of written sources. All the conclusions connected with the defensive
system, the organisation and the line of the limes and position of troops are
merely the result of archaeological excavations and discoveries. This fact,
based on the existence of more than one hundred fortifications at different
levels of knowledge and research, can produce difficulties in presenting
hypotheses of a definite character. However, we can affirm that, as a re-
sult of the excavations done in the last years, knowledge of the defensive
system of Dacia goes beyond the first stage in which all the fortifications
were identified. We have passed on to a second stage in which the fortifica-
tion began to be excavated in order to establish the stages and the system of
their construction, the organisation of their interior, and to gain familiarity
with the archaeological material.
In the last ten years excavations have been carried out in the following
fortifications:6
Bulci in 1976-78 by S. Ferenczi and M. Barbu; Berzovia in 1968-70,
1972 and 1976 by M. Moga, F. Medelet , R. Petrovszki and M. Zahariade;
Orl]OVa in 1967 by F. Medele}; Teregova in 1969 (Gudea 1973b); Jupa in 1965-
78 by M. Moga, F. Medele~, D. Benea and R. Petrovszki; Ve~ei in 1968-78
(Floca and Marghitan 1973); Abrud in 1977-78 by V. Moga; Bologa in 1967-
76 (Chirila and Gudea 1973 and Gudea 1972, 1973a and 1977); Buciumi in
1963-76 by M. Macrea, E~ Chirila, V. Ludtcel, C. Pop, N. Gudea, A. Matei
and S. Simon; Moigrad-Pomet in 1977-78 by E. ChiriHt and N. Gudea;
Tii~a in 1978 by D. Protase and G. Marinescu; Brmcove§ti in 1971-73 and
1975-78 (Protase and Zriny 1975); Olteni in 1968-69 by Z. Szekely; Hoghiz
in 1975-78 by D. Protase; Feldioara in 1973-78 (Gudea and Pop 1974b);
Boro§neul Mare in 1973-74 by z. Szekely; Rf~nov in 1969-74 (Gudea and
Pop 1971; 1973; 1974a; 1975); Rucar in 1971-75 (C:ltaniciu 1976); Boi~a in
1968-76 by N. Lupu; Pietroasele in 1973-76 by G. Diaconu; Tirg§or in 1970-
78 by G. Diaconu; Urluieni in 1973 by I. B. Catiiniciu; Valea Toti~ei in 1976
by I. B. Cataniciu; Cfmpulung-Muscel in 1968-78 (Popescu and Popescu
1970); Sl~veni in 1968-78 by D. Tudor and G. Popilian; Eno§e§ti in 1977
(VHidescu and Poenaru-Bordea 1978); Re§ca in 1968-76 by D. Tudor and
C. Vladescu; Simbotin in 1969 by D. Tudor, N. Hampar~mian and P.
Purcarescu; Bivolari in 1967-69 (Tudor et al. 1970; Vladescu and Poenaru-
Bordea 1969); Radacine§ti in 1971-72 (Vllidescu and Poenaru-Bordea 1974);
Tite§ti in 1972 by C. Vladescu and G. Poenaru-Bordea; Racovi~a-Copaceni
in 1973-74 by D. Tudor, C. Vlad"':'(''~ 2-nd G. Poenaru-Bordea; Bumbe§ti in
1968; GiHiu in 1975-79 by D. Isac; Gherla in 1968-69 by D. Protase and I.

800
Chifor; Turda in 1973-78 by M. B~bulescu, Z. Milea, A. Hopfrteanu and
A. Catina§.
Research was conducted on the advanced line of watch towers of the
limes: in the NW section excavations were carried out in front of the forts
at Bologa in 1968-1971, at Buciumi between 1971-1976 and at Porolissum
between 1976-1979 all by N. Gudea. On the N. sector surface excavations
were made in order to identify the towers in front of the forts at Tihau and
Ca§ei in 1970-1974 and Orheiul Bistri~ei on the N. sector in 1977-1979 both
by St. Ferenczi.
Excavations were carried out at the wall called transalutanus, at Ro§iori
in 1972 and at Urluieni in 1973 both by I. B. Catliniciu. The data obtained,
however, provided no new in sights.
As is mentioned above, the extent of the research, its duration and the
methods employed are different. As a consequence of this the results shed
light for the time being, more on the specification of the general aspect of
the organisation of the defensive system than on the establishment of a speci-
fic or characteristic element for each fortification. Dating evidence is much
less readily available. The dimensions of many forts, their elements of
fortification, the building system and general dating have not been established
yet. We have got details of plan, architecture, stages of construction and
inner organisation only for a small number of forts: Bologa, Buciumi, Rf~nov,
Feldioara, Slaveni, Turda and GiHiu. Except for three (GiHi.u, Gherla, Turda),
all the excavated forts belong to the limes, hence the title of my paper.
Research on the advanced line of watch and signal towers of the limes is
in a similar situation. On the NW sector we are checking the number, char-
acter, form and dimensions of the construction, as well as the dating of the
towers (Gudea 1971). On theN and NE secti~ns many suspected points have
been identified as Roman towers (Ferenczi 1969; 1972; 1973a, 197 4 a and b).
But at only two such points has excavation taken place where it was possible
to establish the exact dimension of the towers (Ferenczi 1974c; 1975;
Ferenczi and Ferenczi 1978).
The following sets out briefly the main conclusions of this research with
reference to the fortifications belonging to the province of Dacia only. The
fortifications on the bank of the Danube which probably belonged to the South-
Danubian provinces Moesia Superior and Moesia Inferior are omitted. The
defensive system of the Dacian provinces has been considered as a whole,
not individually for each of the provinces. The conclusions are broken down
into 3 parts: the composition of the defensive system; the structure and line
of the limes; and technical details concerning the building system and methods
of construction at various stages.
Turning first to the defensive system. In the last 2-3 years we have
come to the conclusion that this vva.s made up of three well defined parts each
determined by its position and tasks (Gudea 1977a, 853-4, 860-2).
1. In the interior, along the circle of mountains, was the line of forts
which had to control the territory in front of the limes and had to face the
first attack. In its turn the limes proper consisted of successive elements in
depth: the roads advancing at right-angles or through the forts (headquarters
of troops), and the line of advanced watch and signal towers.
801
 2. Behind the limes at a certain distance from crossroads were some
forts forming an intermediary line of defence, their intention being to support
the sector of the limes in front of them.
3. The centre of the defensive system consists of what we have called
the central defensive disposition, made up of the forts of the two legions.
For the most part this breakdown is probably correct, but for the time
being it must be considered as a hypothesis until relevant data can be obtained
by research. The function of this structure is confirmed only by the position
of fortifications and troops in the following way: in the centre of the defensive
system are the legions, pre-eminently infantry units, on the intermediate
line are only units of cavalry and bowmen, on the limes, in accordance with
the sector, there are mixed troops with light concentrations of infantry (in N. W.)
or cavalry (in N).
The limes itself, which constituted the largest and most important part
of the defensive system, offers the largest possibility for research. However,
there are still unsolved problems because research work is not on the same
advanced level on all sectors of the limes. In the sectors where problems
are more difficult, research is slower. In the last two decades Romanian
historiography accepted a certain line for the limes (Macrea 1960, 376-79 and
1969, 229-33). In the early form of the province the limes included the
Transylvanian plateau, the entire Banat and Western Oltenia. In the second
stage the limes also included a part of Muntenia, as far as the so-called
transalutanus wall. '!'his line represents the limes at its largest, between
A. D. 118-250.
The unsolved problems and objections concerning the line of the limes
concern the SE and SW sectors of the province. Although the excavations
carried out at Bulci and Cenad seem to confirm the existence of some fortifica-
tions, they do not constitute clear-cut evidence. 7 In the same sector along
the river Theiss no excavations have taken place at the sites_ of suspeQted Roman
forts. 8 As far as sector III (Lederata-Ber zobis-Tibiscum) is concerned, the
results of excavations carried out at Berzovia and Jupa are not sufficiently
clear, particularly regarding the dating and duration of occupation there.
Therefore any of the following three hypotheses are potentially valid. 9 The
first maintains that the SW border of Dacia was there; the second states that
this sector (III) was the boundary only up to A. D. 118, after which date the.
boundary was moved to the East (along sector IV: Orsova-Jupa); the third
hypothesis questions that sector III was ever a boundary and maintains that
the sector on the east (IV) constituted the boundary. Not even the problem
of sector IV is sufficiently clarified: the sounding made at Teregova has
shown that the fort had two phases and it had a long existence, in accordance
with older information about Mehadia and Jupa. It is maintained that this
line was the western boundary after 118 or 159 A. D.
Concerning the organisation of the western sector important information
has been provided by the excavations carried out in the western Carpathians.
Scholars believe that the earthen wall at the exit of the rivers Cri§ul Alb and
Cri§Ul Negru from the mountains constitute a part of the western limes
(Dumistrascu 1969). -

80_2
 The new data about the forts of SE Transylvania bring into discussion
another interesting problem. Generally all the forts situated on the bank of
the Olt in Transylvania and on the bank of its confluent Rful Negru, were
considered as part of the limes (Macrea 1969, 232). 10 This is only partly
true. The forts at Bre~cu, Comalau,- Boro§neul Mare could have belonged
to the limes, but the forts situated between Rf§nov and Boitp. do not. This is
supported by the fact that between Bran and the Turnu Ro§U pass it is not
possible to cross the mountains. South from the Bran pass as far as the
Danube ran the line of the wall called transalutanus. Excavations carried out
at the forts in this sector have revealed that all the forts exhibit two stages
of construction; they are of usual proportions and all of them were garrisoned
by complete auxiliary units. There are also signs that all these forts were
permanently occupied. It appears a little strange if we think that on both
alignments south of the Carpathians there are only one or two big forts (on
the Olt at SHiveni and Racovi~a, and on the Wall at Urluieni .and Sapata de jos),
the others being very small, unable to quarter a complete auxiliary unit.
The sections in the SE of Dacia, in the Romanian plain, constitute a real
problem. Recent excavations carried out at Tirglj)or and Pietroasele failed
to make clear the dating of these forts. As no other forts have been investi-
gated on this line, we are left for the time being, with an ambiguous solution
that these forts were initially outposts of the defence of Moesia Inferior, built
under Trajan, if not earlier, and the duration of their existence was limited
to A. D. 118 when the defensive system of Dacia was set up (V ulpe 196 0, 523-
4; Macrea 1969, 49).
The dating of the forts on the line of the wall and on the wall proper was
not clarified by the excavations carried out at Putineiu, Cfmpulung Muscel
and Rucar. It is maintained that the I?urgus at Rucar was also abandoned in
A. D. 118. At the same time the star, wall fort at cimpulung Muscel is sup-
posed to date from the middle of the second century A. D. (Popescu and
Popescu 1970). Confronted by the persistently held hypotheses concerning
the dating of the wall (Tudor 1978, 526), another hypotheses stands out, namely
that the wall of the fortifications dates from the same period as the defence
organised along the Olt, the two functioning simultaneously (Cataniciu 1978,
344; Tudor 1978, 256).
The excavations carried out at the fortifications along the Olt revealed
new data, but these do not solve the problem of the limes alutanus. From
the long line of fortifications, supposed to be situated at the distances men-
tioned by the Tabula Peutingeriana, not all are confirmed archaeologically
(Tia Mare, Re§lca, Simbotin, Jiblea) at least up to the present moment
(Cataniciu 1978, 233 and 235; VHtdescu and Poenaru-Bordea 1974b, 255).
The rarity of the forts on the Olt favours the acceptance of the hypothesis
which proposes the simultaneous functioning of the two lines (the Alutanus
and the Transalutanus). The plan and architecture of the forts from the Olt
passes (Bivolari, Tite~ti, Copaceni Radacine§lti) have definitely been estab-
lished. But the question arises: why was such a powerful defence of the
passes towards Transylvania necessary under Hadrian if the earthern wall
had already been built in front? Such a situation would favour the dating of
the wall later than the building of these forts and so after A. D. 140. Connec-

803
 ted with the stages of organisation of the defence on the Olt we must stress
that the existence of an earlier phase, with an inner earthern wall, was not
identified except at Slaveni. But this situation was not clearly demonstrated
stratigraphically (Tudor 1978, 301-7).
Recent research has offered important data about technical details of
the manner and building system of the forts in two stages. The map of the
forts with earthern walls indicates rather clearly the line of the limes in
Transylvania. Less well known is the situation in Western Oltenia. At a
fairly large number of excavated forts it could be established that the phase
with earthern wall dates from Trajan's reign (Gudea 1975). This dating is
based on stratigraphy as well as on the corroboration of the construction of
the fort agreeing with the arrival of the auxiliary garrison unit. In such way
the forts at Vejel, Bologa, Buciumi, Moigrad-Pomet, Moigrad-Citera and
Ri~nov could be dated, and by analogy all the forts that have earthern inner
walls, built in the same m:1.nner, and at which the relation between the earth ern
wall and stone wall are the same. Unfortunately only at Bologa and Buciumi
is anything known about the interior organisation of the fort at this stage. At
the others we have information only on the construction of the defences (Vetel,
Moigrad-Pomet, Brincovenesti, R1~nov. Hoghiz, Feldioara), although of
Feldioara there is also some inform~ t:ion about the gate.
We can state today that the turf wall was built in the simplest possible
manner, consisting of successive strata of beaten earth mixed sometimes with
other materials. In the places where the groun.d (sandy or loess) did not offer
a natural binder, artificial building methods were used. Such was the situation
at Hoghiz where the mass of earth of the wall was supported by three lines
of posts (Horedt 1953; Protase 1977, 196-200 and 1978, 307-9.). AtRf~nov
traces of posts were found in the structure of the wall. In both cases the soil
, was sandy or loess (Gudea and Pop 1971, 13-4). The excavations at Ri~nov,
Bologa and Hoghiz have revealed that in the turf-wall stage no berm exists.
Interesting problems concerning the relation between the wall and the ditch
were revealed. It is well known that the earth taken from the ditch constitutes
the major part of the wall. This statement was confirmed by research in
Dacia. But there are some exceptions: at Teregova and Moigrad Pomet
the earth neffssary for building the wall was brought from elsewhere (Gudea
1973b, 99).
The second stage of the limes began in A. D. 118, when its line was ex~
tended along the Olt river, and the wall was built up to the Danube. In this
phase the part played by the sector in the SE of Transylvania, as well as by
the forts on the line of the Jiu changed. The fort walls began to be rebuilt in
stone, although earth ern walls continued to be used both in new constructions,
like those on the wall called transalutanus, and in older forts which were only
rebuilt in stone later.
The earliest epigraphic attestations date from A. D. 138-140 and they
refer to the forts from the passes of the Olt at Radacinesti (CIL 12604 and
12605), Copaceni (CIL 13795 and 13796) and Bivolari (CIL 12601 and 13793).
The style and manner of their building, and common architectural elements,
allow us to date other forts in the same period by analogy (T_ite~ti, Racovij;a)
(Tudor 1978, 289-90 and 307-8). Technical characteristics of these fortifi-

804
cations are: very small dimensions, gates with quadrilateral prominent
bastions and prominent corner towers.
The second stage of construction of the stone forts seems to be under
Antoninus Pius. From this period probably date the stone forts at Vej:el,
Moigrad-Citera, Inlaceni, Ri;mov, Hoghiz 7 Feldioara and C1mpulung-Muscel
(Daicoviciu 1931; Macrea 1961; Gudea and Pop 1971; Gudea 1977). There
is, however, no precise information concerning their dating. Among the forts
built in this period two have unusual technical characteristics. The excava-
tions at Hoghiz show the existence of an unusual building method also in the
stone-wall phase. The wall is double, the two walls that formed the faces
were of different thickness, (1.15. and 1. 05 m) and were connected by trans-
verse walls at a distance varying from 3 to 7 m. There were no corner towers,
the gates have quadrilateral bastions projecting beyond the line of the inner
wall; the space in their interior is very small.
The information acquired at Ri~nov is also interesting: generally in
Dacia at the forts which had a primary earthern wall phase, the ultimate
stone wall was placed in such a manner that its base was contained by the
wall, and the old wall constituted the revetment behind the later wall. At
R1~nov we are faced with such a situation on two of the sites (NW and SW).
On the other two sides (NE and SE) the base of the stone wall was placed in
the ditch of the earthern fort, which was then levelled in front, at the rear
the old rampart and a part of the ditch were built up and enlarged forming a
new rampart to match the height of the wall. In front of the wall three suc-
cessive parallel ditches were dug.
The third stage of building of the forts with stone walls dates from the
beginning of the third century A. D. (Macrea 1957 and 1969, 223-4). This
was epigraphically identified at the fort at Moigrad-Pomet. By common
architecture and a date given by epigraphy, this dating can be applied to the
forts at Boh)ga, Buciumi and Ca§ei. All these forts have a common architec-
tural element: their gates have bastions with semicircular projection. At
Bologa and Moigrad-Pomet one can find similarities in the manner of construc-
tion of the inner wall. Here the earthern wall is very well preserved, and
survives to a height of 2 m. At Buciumi and Caf?ei the bastions of the gates,
the corner towers and the yard were inhabited. At Bologa and Moigrad we
face another situation. At Bologa only the bastions of the porta decumana
and one of the bastions of the porta principalis dextra were inhabited, at ·
Moigrad Pomet only one gate bastion was inhabited while the internal towers
and the corner towers were not inhabited. More interesting are the possibili-
ties of construction at the gates and corner towers. At Bologa at the bastions
of the porta praetoria and porta principalis sinistra, the stone wall surrounded
the end of the earthern wall, the side towards the wall was not closed. The
northern bastion of the porta praetoria was filled with stones already at its
building. At Moigrad-Pomet the bastions were built complete, their interior
was divided into two, the semicircular part was separated from the quadri-
lateral one by a wall. In none of the forts was this semicircular part inhabi-
ted; the eastern bastion of the porta sinistra and the eastern bastion of the
porta dextra were filled with yellow earth coming from the walL Neither the
corner towers nor the interval towers were inhabited. Even more in the

805
eastern corner, higher than the rest of the site (a volcanic rock in fact), the
inner wall is not closed; it stopped on both sides of the rock, and no tower
was built.
A last stage in the works of fortification takes place towards or at the
middle of the third century A. D. (Macrea 1969, 443). It consists of works
aimed at strengthening the defences by either refurbishing the gates or blocking
them off totally or partially. Archaeologically this situation was identified
at Bologa, where porta praetoria and porta decumana were blocked: at
Buciumi, where there are a lot of buildings behind the inner wall; at Moigrad,
where repairs were made on the inner wall; and at Inlaceni and R1smov,
where the gates were partially blocked. At Bologa and Buciumi later buildings
were identified which do not respectthe older organisation of the fort.
With regard to the advanced line of watch and signal towers of the limes,
excavation revealed the following: Generally the towers are situated in front
of the forts on hills at a distance varying between 3 and 10 km. The towers
are not placed in a straight line but in depth according to the requirements
imposed either by the relief or by the necessities of the defence. No speci-
fication can be made with regard to the construction phases of the towers.
We presume that some wooden towers existed, but we have evidence for stone
towers only on the NW section. At some of these, two phases of construction
could be identified. Such is the case at tower nr. 2 at Bologa and nr. 15 at
Buciumi. At Bologa a tower of the first phase was dismantled, and with its
stones another tower was built at some distance from it. On the NW sector
40 stone towers were identified in front of the forts at Bologa and Moigrad,
and two burgi with earthern walls. The towers have circular or rectangular
plans. The reason for the variation in plan cannot be ascertained since they
alternate. In the Bologa sector from the thirteen recorded towers we know
nine plans, seven of which are square and two circular. The majority of
towers in Bologa sector have square plans, and others are circular. The
dimensions of the towers are between4,50-12 m. Those with circular plans
are usually bigger, perhaps because of the method of construction. In a
lower area, two km.away south of tower no. 30, an earthern wall was built
with its ditch orientated westwards. In front of the complex at Moigrad the
situation is not as clear. The advanced line of towers is completed in the
gorge by a stone burgus descending to the level of the road. South from this
burgus in the direction of the tower on Magurita, 200 m away, a stone wall
was identified ascending the slope. A similar wall or rampart would appear
also to have existed heading towards Pognior. On the N sector, from the
fort at Tihau as far as Orheiul Bistritei some 50 points have been identified
which are suspected of being Roman towers. Two of them were excavated
at Cap1lna and Zagra. In both cases the plans are quadrilateral. Roman
towers were identified on the N sector of the eastern limes, but they have
·not yet been excavated.
As a consequence of the excavations and research at the forts new infor-
mation has been acquired with regard to the troops)2 This mainly refers
to three aspects of the problem: the identification of some military units un-
known or unattested in Dacia; the specification of the stations of some units
already known; and the completion of general information with regard to a
series of units.

806
Units previously unattested:
Legio Ill Gallica. Stamps of L m G type were found in the fort on the Pomet
hill at Moigrad (Gudea 1978b, 66).
Legio VII Gemina Felix. Stamps of LVII G F type were discovered in the
Roman fort on the Pomet hill at Moigrad (Gudea 1978b, 66 and 79; 1976a,
109-114).
Cohors Ill. Stamps of three types and several variants were found at Moigrad
(C lll; Ill C; C III D). The unit cannot be specied with certainty (Gudea
1978b, 68 ar,td 71-4).
Cohors IIII Betasiorum or Betavorum. Numerous stamps of a sir,tgle type
(C Ill B+ E) were found at Olteni (Go star 1966, 1 78-9).
Numerus Illyricorum. Stamps of this unit were discovered in the last century
at Hoghiz (Bak6 1975, 141-6).

Locations for units already attested:

Ala I Bosporanorum. Numerous stamps grouped under three types with
variants attest the existence of the fort at Criste§lti and its garrison (Gudea
and Z rinyi 1977).
Cohors I Augusta Ituraeorum. Some inscriptions and stamps were discovered
in layers relatir,tg to the earthen rampart phase of the fort at Buciumi. We
presume that this unit was stationed in this fort until A. D. 114, after which
it was moved elsewhere (Chirila and Gudea 1972, 11 7-8; Petolescu 1974,
5 99-6 01; Gudea 1976b, 519-2 0).
Cohors I Ulpia Brittonum. At Bologa the COH I BRITT tile stamps could be
dated to the first period of the fort~· A. D. 106-114. At Moigrad a different
type of stamp was found. We presume that the garrison was initially stationed
at Bologa and was then moved to Moigrad (Gudea 1977d, 129-34).
Cohors II Numidarum. Bricks and tiles with three types of stamps (C N;
COH NUMID; COH NUM ANT) are known from Feldioara. The latest seems
to be the stamp bearing the Antoniniana epithet (Pop 1975; Gudea and Pop
1977' 23 7).

Further information on units already attested:

Numerus Palmyrenorum Porolissensium. Graffiti refering to this unit were
identified in the temple of Bel, built in A. D. 210, at Moigrad. Where this
unit was stationed remains unknown.13
Numerus Palmyrenorum Tibiscensium. Inscriptions and tile stamps were
identified attesting the presence of the unit at Jupa (Russu 1977, 226; Moga
and Russu 1974, no. 157).
Cohors II Flavia Commagenorum. The reading of the COH F L Comm stamps
from Ve~el was confirmed (half of the text is written in capital letters, the
other half in cursive letters). The older and incorrect reading of the text
suggested the presence of Cob. II Hisp. in this fort (Floca et al.- 1970, 91;
Petolescu and Marghitan 1974, 256-7; Petolescu 1972; and Gudea 1976b, 519).

807
Cohors I Vindelicorum. There are stamps of COH I VN and C I V type
attesting the presence of this unit at Jupa {Russu 1977, 225-6; Moga 1970).
Cohors I Sagittariorum. There are stamps of C I S type and an honorific
inscription in the principia of the fort at JuPa.. It is thought that this is the
same unit that was stationed at Drubeta · (Moga and Russu 1974, no. 55 a-d;
Russu 1977, 225, nos. 251-2; Moga 1970, 146).

I have tried above to summarise the results of research on the limes in

Rumania. For further details of clarification the reader isreferredto the
extensive bibliography appended.

NOTES·

1• In recent years we have tried to provide an overall view of the defensive

system of Dacia and its limes, as well as a complete list of the fortifi-
cations (Gudea 1974a; 1974b; 1977a; 1977b). The present article
contains a number of ideas already Bxpounded, but further expanded with
regard to technical conclusions.
2. Systematic research activity began in 1949 with the foundation of the spe-
cialist Institute of the Academy of the Socialist Republic of Rumania
(history and archaeology) in Bucharest, whose activity was planned and
financed by the state.
3. Older works on the limes and on the military organisation of the province
in general consisted of a simple ennumeration of the fortifications. This
has been shown to be incomplete and contain many errors, especially
concerning the chronological sequence. For the province as a whole see
Christescu 1937, 115-127 andMacrea 1969, 218-35; for large parts
thereof see Daicoviciu 1945 8 9f and Tudor 1942 with the later editions of
1958, 1968 and 1978.
4. It will suffice in this context to mention the two fundamental studies on
the limes, though both deal only in part with the limes of Dacia (Fabricius
1926, 641-5 and Forni 195 8, 1 076-80). We should not forget the way in
which a series of authors of both older and more general recent studies
view the limes of the province (Homo 1925, 188-9, 210-13; 1933, 463-5,
503 ...5; Chapot 1927, 430-1; I..ongden 1936 233-4 and map 8; Istoria ·
Universa1.3. (Bucharest 1959. Translated from the Russian), 648;
PropyHien Weltgeschichte Ill (Frankfurt-Berlin-Vienna 1963), 360 with
map; Saeculum Weltgeschichte II (Freiburg-Basel-Vienna 1966), 500-
503.
5. Hitherto the organisation of the defensive system of Moesia Inferior in
the 2nd and 3rd centuries A. D. has not been the object of systematic
study, but see Zahariade 1978 ~or some preliminary results.
6• Concerning fortifications investigated over the last ten years see:
'Sapaturile arheologice din E. s. Romania', Studii ~i Cercerari de Istorie
veche 20 (1968), 482-8; 21 (1970), 504-9; • Les fouille~ archeologiques
en R. S. Roumanie', Dacia 15-(1971), 359-93; 16 (1972), 325-58;

'808
 17 (1973), 361-97; 19 (1975), 269-308; 20 (1976), 273-86; 21 (1977),
357-74; 22 (1978), 348-62.
7. A report on the results of archaeological research in 1977-78 by S.
Ferenczi and M. Barbu presented at the annual research meeting. For
the discovery of tile-stamps, see: Russu 1977 246-9. A recent paper
by Petolescu (1977) has brought together all data on this sector.
8. Recently the problem of the S.W. border of Dacia has been discussed by
Daicoviciu and Daicoviciu (196 7).
9. Ferenczi (1973b and 1974d) has recently discussed the first two hypotheses
taking into account almost all of the older literature. He repeats an
hypothesis formulated by Protase (1975) and appears to accept it. On the
third hypothesis see: Brandis 1901, 1969-1970 and after him Fabricius
1926' 641.
10. In the last two studies of the limes in Roman Dacia we have supported
this hypothesis also •
11. The results of the excavatic::.: 3.t Moigrad have not yet been published.
12. To the older studies of the army of Dacia (Christescu 1937, 166-236;
Wagner 1938 and Szilagyi 1946) maybe added the work by Benes (1970).
The latter two works have been checked and corrected by Russu (1957,
360-2; 1972 and 1974).
13. New and unpublished results of excavations at Moigrad.

809
 BIBLIOGRAPHY

Bako, G., 1975. 'Numens Illyricorum la Hoghiz', Studii §i Cercetari de

Istorie veche 26, 141-6.
Brandis, E., 1901. 'Dacia' in Paulys Real-encyclop!i.die der classischen
Altertumswissenschaft IV. 1 . Stuttgart.
Benes, J. , 1 970. 'Die rf.}mischen Auxilliarformationen im unteren Donauraum',
Sbornik Praci Filosoficke Fakulty Brnenske University E 15, 159-209.
Cataniciu, B., 1976. 'CerceU!ri in castellum la Rucar', Studii $)i Cercetari
de Istorie veche 25, 277-8.
C[taniciu, I. B., 1978. 'Nouvelles recherches sur le limes de la sudest de
la Dacia' in F itz 1978, 333-52.
Chapot, V., 1927. Le Monde romaine, Paris.
Chirila, E. and Gudea, N., 1972. Das Rl1merlager von Buciumi, Cluj.
ChirilA, E. and Gudea, N., 1973. '$antierul arheologic Bologa', Materiale
s;i cercetari arheologice 10, 115-123.
Christescu, V., 193 7. Istoria militara a Daciei romane, Bucharest.
Daicoviciu, C. , 1931 • 'M icia. Cercetari asupra castrului', Anuarul
Comisiunii monumentelor istorice, Secpa pentru Transilvania 2 (1930-31),
3-43.
Daicoviciu, c., 1945. La Transylvania dans l'antiquite, Bucharest.
Daicoviciu, C ~, and Daicoviciu, H., 1967. 'Noi consideratii asupra Daciei
Malvensis 1 , Acta Musei Napocensis 4, 73-81.
Dumitra51cu. S., 196 9. 'Contribu~ii la cunoa~terea granitei de vest a Daciei
romane', Acta Musei Napocensis 6, 483-91.
Fabricius, E., 1926. 'Limes' in Paulys Real-enclyclopttdie der classischen
Altertumswissenschaft XIII, Stuttgart.
Ferenczi, S., 1969. 'Cu privire la apararea hotarului de Nord al provinciei
Dacia', Studii $i Comunicari Muzeul Satu Mare 1, 91-110.
Ferenczi, S., 1972. 'Cercet~ri s;i rezultate noi pe limesul de nord al Daciei
romane', File de Istorie 1, 37-45.
Ferenczi, S., 1973a. 'Contribu!ii la cunoa§lterea limesului roman de la
nordul Somes;ului Mare. Studiu preliminar', Sargetia 10, 79-104.
Ferenczi, S., 1973b. 'Cerceta..ri ~i rezultate noi pe limesul Daciei intracar-
patice', Apulum 11, 191-222.
Ferenczi, S., 1974a. 'Die Nordstrecke des dakischen Limes vom Cri~ul
Repede bis zu den Ostkarnaten' in Pippidi 1974, 201-5.

810
Ferenczi, s., 1974b. 'Zur Verteidigung der Nordgrenze der :Provinz Dakien'
in Birley, E., Jarrett, M. G. and Dobson, B. (eds.) Roman Frontier
Studies 1969, 204-211, Cardiff.
Ferenczi, S., 1974c. 'Investiga}ii noi pe limesul de N §1i NE al Daciei
Porolissensis 1 , File de Istorie 3, 181-a.
Ferenczi, S., 1974d. 1 Neue Forschlln.gsergebnisse uber den limes des
innerkarpatischen Dakiens 1 , Dacia 18, 127-36.
Ferenczi, S., 1975. 1 Limesul Daciei Porolissensis tntre Valea Zagrei SJi
Valea Mure§lului', Sargetia 11-12 (1974-75), 285-89.
Ferenczi, G and Ferenczi, S., 1978. 'Neue Forschungen und Ergebnisse am
Nordostlimes der Dacia Porolissensis' in Fitz 1978, 365-76 •.
Fitz, J. (ed.), 1978. Limes:Akten des XI Internationalen Limeskongresses,
Budapest.
Floca, 0. and Marghitan, L., 1973. 'Noi contributii privitoare la castrul
roman de la Micia', Sargetia 10, 43-57.
Floca, 0., Marghitan, L. andFerenczi, s., 1970. Micia. Grupulde cup-
toare pentru ars ceramic~, Deva.
Forni, G., 1958. 'Limes' inDizionarioepigraficoiV, 1076-80, Rome.
Go star, N., 1966. 'Studii epigrafice (II) 1 , Arheogia Moldovei 4, 1 78-9.
Gudea, N., 1971. t Limesul roman tn zona castrului de la Bologa', Acta
Musei Napocensis 8, 507-30.
Gudea, N., 1972. 'Castrul roman de la Bologa', Apulum 10, 121-50.
Gudea, N., 1973a. 'Castrul roman de la Bologa. Sa~turile arheologice
din anul1969', Crisia 3, 109-37.
Gudea, N., 1973b. 'Sondajul arheologic de la Teregova', Banatica 2, 97-101.
Gudea, N. , 1974a. 'Sistemul defensiv al Daciei roman I. Stadiul actual al
cercetarilor', Apulum 12, 182-92.
Gudea, N., 1974b. "Das Verteidigungssystem des rl1mischen Dacien',
Saalburg Jahrb. 31, 41-49.
Gudea, N., 1975. 'Sistemul defensiv al Daciei romane II. citeva observa~ii
in legatura cu faza de pamint a castrelor', Anuarul Institutului de Istorie
;;i Arheologie din Cluj-Napoca 18, 71-87.
Gudea, N., 1976a. "l.egio VII Gemina in Dacia', Studii ~i Cercetari de Istorie
veche 27, 109-114.
Gudea, N., 1976b. 10bservatii in legatura cu unele note epig-rafice 1 , Studii
§i Cercetari de Istorie veche 27, 517-21.
Gudea, N., 1977a. 'Der Limes Dakiens und die Verteidigungssyteme des
obermoesischen Donaulimes von Traianus bis Aurelianus' in Temporini,
H. (ed.) Aufstieg un Niedergang der Romischen Welt II, ~. 851-876,
Berlin.

811
Gudea, N., 1977b. 'Limesul Daciei romane de la Traianus la Aurelianus',
Acta Musei Porolissensis 1, 97-113.
Gudea, N., 1977c. 'Materiale arheologice din castrul roman de la Bologa',
Apulum 15, 169-215.
Gudea, N., 1977d. 'Cohors I Ulpia Brittonum in Dacia', Studii ~i Cercetari
de Istorie veche 28, 129-34.
Gudea, N., 1978a. 'Die M iliUirorganisation an der Nordgrenze der Moesia
Superior w!ihrend der Romerzeit in Dakien' in Fitz 1978, 223-36.
Gudea, N., 1978b. 1Descoperiri arheologice §i epigrafice mai vechi sau mai
noi la Porolissum', Acta Musei Porolissensis 2, 66-75.
Gudea, N. and Pop, I., 1971. Das Rl!imerlager von Rl$nov (Cumidava),
Brasov.
Gudea, N. and Pop, I., 1973. 'Castrul roman de la Rl~nov. Sapaturile arheo-
logice din anul1971 ', Cumidava 7, 13-24.
Gudea, N. and Pop, I., 1974a. 'Castrul roman de la Rl~nov. Sapaturile
arheologice din anii 1973-1974 ', Cumidava 8 (1974-75 ), 55-64.
Gudea, N. and Pop, I. , 1974b. 'Castrul roman de la Feldioara-Fagaras',
Cumidava 8 (1974-75), 39-54.
Gudea, N. and Pop, I., 1975. 'Cercetari arheologice recente tn castrul
roman de la Rf~mov', Studii ~i materiale de muzeografie >?i istorie
militara 7-8 (1974-75), 55-78.
Gudea, N. and Pop, I., 1977. 'Un castru nou descoperit in sistemul defen-
siv al Daciei romane: castrul roman de la Feldioara', Pontica 10, 333-
338.
Gudea, N. ani Zrinyi, A., 1977. 'Contributions epigraphiques a l'histoire
de ala I Bosporanorum' in Epigraphica. Travaux dediees au Vile Congres
d'epigraphie greque et latine, 223-232, Bucharest.
Homo, L., 1925. L'Empire Romain, Paris.
Homo, L., 1933. Le Haut-Empire, Paris.
Horedt, K., 1953. 'Cercetari in regiunea Hoghiz-Ungra', Materiale $i
cercet:!ri arheologice 1, 785-98.
Longden, R. P., 1936. 'The Wars of Trajan' in Cambridge Ancient History
. XI, 223-52, Cambridge.
Macrea, M., 1957. 'Ap:lrarea granij:ei de V 5Ji NE a Daciei pe timpul
1mparatului Caracalla', $tudii $i Cercetari de Istorie veche 9, 215-45.
Macrea, M., 1960. 'Dacia in timr:-·~~! ~j:::tplhirii roman~' inistoriaRo~~i
I, Bucharest.
Macrea, M., 1961. 'Santierul arheologic Porolissum', Materiale !?i cercetari
arheologice 7, 3 74-6.
Macrea, M., 1969. Viaj;a fn Dacia romana, Bucharest.

812
Mocsy, A., 1962. 'Pannonia' in Paulys Real-encyclop~die der classischen
Altertumswissenschaft XI, Stuttgart.
Moga, M., 1970. 'Garnizoana romana de la Tibiscum', Acta Musei Napocen-
sis 7, 140-3.
Moga, M., and Russu, I. I., 1974. Lapidarul Muzeului Banatului, Timi~oara.

Petolescu, C., 1972. 'Cohors Hispanorum la Micia', Sargetia 5, 43-9.

Petolescu, C., 1974. 'Note epigrafice (II) 1 , Studii ~i Cercet&ri de Istorie
veche si Arheologie 25, 599-601.
Petolescu, C., 1977. 'L'organisation de la frontiere romaine dans la vallee
du Mures' inHaupt, D. and Horn, H. G. Studien zu den Milita!rgrenzen
Roms II, 367-72, Cologne-Bonn.
Petolescu, c. and Marghitan, L., 1974. 'Tigle si di.ramizi stampilate din
thermele de la Micia', Muzeul N~tional 1, 247-58.
"'
Pippidi, D. M. (ed.). 1972. Actes du IXe Congres International d'Etudes
sur les Frontieres romaines, Bucharest.
Pop, I., 1975. 'Cohors Numidarum Antoniniana in Feldioara-Fagaras',
Studii s>i Cercetari de Istorie veche 26, 289-93.
Popescu, M. and Popescu, E., 1970. 'Raport preliminar asupra sapaturilor
efectuate in anii 1962-67 fn castrul roman de linga Clmpulung Muscel',
Materiale ;?i cercetari arheologice 9, 251-63.
Protase, D., 1975. 'Castrul.roman de la Surducul Mare', Banatica 3, 345-
8.
Protase, D., 1977. 'Castrele romane cu dublul zid de incinta descoperite
in Dacia', Sargetia, 13, 196-2 00.
Protase, D., 1978. 'Rtfmisches Lager m it doppelten Umfassungsmauern' in
Fitz 1978, 307-9.
Protase, D. and Zrinyi, A., 1975. 'Castrul roman de la Brfncovene~ti',
Marisia 5, 5 7-68.
Russu, I. I. , 1957. Review of Szilagyi, X. A Daciai ert1drendszer helyt1rsegi
es a katonai teglabelyegek in Dacia 1, 360-363.
Russu, I. I., 1972. 1Auxilia Provinciae Daciae', Studii &'i cercetari de Istorie
veche 23, 63-77.
Russu, I. I., 1974. 'Die Hilfstruppen am Limes Daciae Superioris 1 in Pippidi
1974' 215-26.
Russu, I. I. , 1977. Inscriptiile :Uaeiei Romane Ill, 1, Bucharest.
Szilagyi, J. 1946. A Daciai erl:Jdrendszer helytlrsegi es a katonai teglabelyegek,
Budapest.
Tudor, D., 1942. Oltenia romana, Bucharest.
Tudor, D., 1978. Oltenia romana (4th edn.), Bucharest.·

813
Tudor D., Vllidescu, G. and Poenaru-Bordea, G., 1970. 1Arutela I-II.
Rezultatele primelor doua campanii de slip~turi arheologice 1967-68 in
castrul roman de la Poiana Bivolari 1 , Studii ~?i materiale de muzeografie
si istorie militara 2-3 (1969-70), 8-45.
Vladescu, C. and Poenaru-Bordea, G., 1969. 'Date noi in legatura cu cas-
trul Arutela 1 , Studii ~i Cercetari de Istorie veche 20, 101-111.
Vladescu, C. and Poenaru-Bordea, G., 1974a. 1 Primele sapaturi arheologice
in fortificatia romaru de iinga Radacine§lti I ' Studii g;i Cercetari de Istorie
veche 23, 4 77-85.
Vladescu, C. and Poenaru-Bordea, G., 1974b. 'Fortifications romaines sur
le limes alutanus dans la zone du massive de Cozia' in Pippidi 1974, 247-
258.
Vladescu, C. and Poenaru-Bordea, G., 1978. 1Cercetari in castrul de la
Arcidava, sat Eno~e§lti', Studii ~i materiale de muzeografie §i istorie
militara 11, forthcoming.
Vulpe, R., 1960. 'Muntenia §i Moldova in epoca roman!' in Istoria Rom[niei
I, Bucharest.
Wagner, W., 1938. Die Dislokation der rtSmischen Auxiliarformationen in
den Provinzen Noricum, Pannonien, Moesien und Dakien, Berlin.
Zahariade, M., 1978. 'The structure and functioning of the Lower Moesia
Limes in the 1st and 3rd centuries A. D.' in Fitz 1978, 385-98.

814
 • 54. LES DEUX CAMPS DE PRAETORIUM
SUR LE LIMES ALUTANUS
Cristian N".Vladescu & Gh. Poenaru Bordea

Les deux camps romains de Praetorium tiennent

une place de toute première main dans l'ensemble des
fouilles urchéologiques entreprises depuis plusieurs
années par le Musée militaire central·de Bucarest a-
fin d'étudier le système des fortifications du limes
Alutanus. Ce système englobait aussi les camps d'Aru-
tela, Radacine§ti, Tite§ti et Copaceni. présentés lors
du ·rxe Congrès international des frontières romaines.
tenu à Mamaia (Roumanie) en 1972~ cependant que l'en-
semble de Romula, qui s'en rattache également, a fait
1' objet d'une communication au Congrès du limes de
Székesfehérvar, en 1976.
Là où l'Olt touche à son défilé, sur sa gauche,
était sise l'antique localité de Praetorium- de nos
jours, village· de Copaceni, commune de Racovita, dé-
partement de Vîlcea. entre les deux cours d'eau dits
"pîrîul Clocoti9" et "rîul Baie9ului". Elle se trou-
vait située, donc, à vingt kilomètres plus au nord du
càmp d' Aru tel a, qui veil·lai t au débouché du défilé.
Deux camps romains ont été localisés dans cet-
te zone. Le premier, dans les limites du village de Oo-
paceni, surmontant un promontoire haut d'environ 16 m.,
le second à 5oo m. plus au nord, dit "le camp de Raco-
vita". Les premières recherches archéologiques en ces
lieux appartiennent à Pamfil Polonic, qui y travailla
en 1894 sous la direction de Gr.G /rocilescu. Exceptant
les inscriptions (1), les résultats de ces recherches
•
sont restés sous la forme de notes manuscrites (2)
jusqu'à ce que, bien plus tard, les eût valorisés D.
Tudor (3). Ces fouilles allaient être reprises par le
Musée militaire central en collaboration avec l'Insti-
tut d'archéologie de Bucarest, en 1973 pour le camp de
Copaceni (4) et en 1976, 1978 à Racovita (5).

815
 Le camp de Copaceni (Praetorium I) fut bâti en
138 par le numerus bur~a;iorum et veredariorum Daciae
Inferioris, sur l'ordre de Titus Flavius Constans, le
procurateur impérial de la province (6). Deux ans plus
tard, cette fortification devait être élargie par les
soins de la même unité militaire: on lui ajouta deux
tours. A cette époque, c'est-à-dire en l4o, la fonction
de procurateur provincial était tenue par Iulius Aqui-
la Fidus ( 7) (Fig. 54. 1).
En ce qui concerne son système de construction,
celui-ci s'avère analogue à celui utilisé dans le cas
des camps d'Arutela, Radacine9ti et Tite9ti, qui assu-
raient la protection du massif de Cozia sur ce tronçon
du limes.Ce qui lè caractérise, ce sont les éperons
intérieurs de la maçonnerie de pierre, semblable à
l'enceinte, qui remplace l'agger (8).
Les parements de cette-dernière sont en pier-
res de taille, dont les rangées alternent avec des dal-
les, solidement-- fixées au mortier; quant aux espaces
libres entre ces rangées, ils sont comblés de pierres
fichées, toujours au mortier. Le mur mesure 1,5o m. de
large et il s'est conservé sur une hauteur de o,6o à
1,30 m. Sur les 64 m. enregistrés en 1894, le côté est
du camp a pu être exploré encore sur une longueur de
41, 5o m. (Pl. 54.1}. A 28 m. de son angle du nord-est
était pratiquée la porta praetoria, avec une ouverture
de 3 m. Elle était flanquée à l'intérieur de deux
tours carrées, avec le_côté de 2,4o m. et des ouvertu-
res d'accès larges de o,4o m. aménagées dans leurs an-·
gles nord-ouest. Encore plus renforcée sur son front
à la hauteur des tours, l'enceinte mesure jusqu'à 2,lo
m. de large. A l'est, on a dégagé deux éperons inté-
rieurs, du même appareil et construits en même temps
que l'enceinte, situés entre la tour septentrionale de
la porte et celle qui se dresse à l'angle nord-est du
camp. Cette dernière tour comporte une pièce à peu
près ovale, avec les diamètres de 2 et 4 m. et une ou-
v~rture de o,8o c~- Là encore l'enceinte est renforcée

816
 1
1
1
L·-·-·- -·-·-·-·-·-,
i-·-·-·-· -·-·-·-i
1
i
1
1
i
i
-·-·-·-'--·-· ~

r·-·~

-i 'i !
1 1
i i
!
!

Fig. 54. 1 Releve du camp de Copaceni (Praetorium !). 1973.

817
 jusqu'à mesurer 2,4o m. de large. Le plan irrégulier de
cette tour, dont le côté sud est pratiquement un éperon,
alors que son côté ouest représente, par la courbe qui
le prolonge vers l'est, l'adaptation toujours à un épe-
ron, ce plan confirmerait donc sa construction ultérieu-
re, mentionnée par l'inscription de l'an 14o. Quant au
côté septentrional du camp, il a été exploré sur une
longueur de 16,5o m., jusqu'au point de rupture du ter-
rain éboulé à l'occasion d'une crue anormale de l'Olt.
Sur le tronçon conservé, on a dégagé seulement deux é-
perons.
Comme il ne s'est conservé qu'un quart de tout
le camp, les vestiges mis au jour consistent en deux
murs, .larges de o,8o cm., restes d'un bâtiment de gran-
des dimensions, situé dans la praetentura dextra et
vis-à-vis, dans la praetentura sinistra, d'un petit
bassin en brique, crépi au mortier. Ce petit bassin,
long de 4, 6o m. et large de 1, 2o m., doté d'un fond.
en cuvette, serv~it peut être d'abreuvoir aux chevaux
de la troupe.
L'unique horizon de l'habitat attesté au point
de vue stratigraphique par nos fouilles a livré quel-
quasi documents archéologiques, récupérés soit lors des
premières recherches effectuées en ces lieux, soit par
les toutes dernières explorations. Il s'agit de frag-
ments céramiques, dont une jarre à provisions haute de
o,65 m. et avec une inscription en grec accompagnée de
la représentation d'un poisson, ainsi que de pièces
d'équipement, harnachement et armes (pointes de flè-
ches et de lances, l'umbo d'un bouclier, fibules, pen-
dentifs de cuirasse et un fragment de spatha) . A ceci
s'ajoutent encore quelques monnaies d'Antonin le Pieux
et Iulia Mamaea (9). La borne milliaire de l'an 236 (~)
mise au jour devant le camp nous fournit le terminus
post guem pour le moment où celui-ci dût être anéanti
et qui se place vers le milieu du IIIe siècle.
Fort probablement, le camp de Racovita (Praeto-
rium II) a été bâti juste. a12rès la fin de celui de Co-

818
 paceni, héritant de la fonction stratégique de celui-
ci(Fig.54.2). Ce deuxième camp s'est conservé intégra-
lement. De forme rectangulaire, il mesure sur les cô-
tés nord (Pl. 54. 2) et sud 118, 3o m•• cependant que ses
côtés est et ouest (PI;54_.~_3) . sont longs de lo6, 8o m.
Les pierres de carrière,qui composent les parements de
1' enceinte sont de dimen~ons à peu près égales et
liées au mortier, revêta-nt l'emplecton fait de galets
et beaucoup de mortier. La muraille, large de 1,5o m.,
qui s'est conse~ée sur u~e hauteur oscillant de o,84
m. jusqu'à 1, lo ·m., étai tr bâtie sur des as si ses en
mortier avec un sou"tiassement constituant à la hauteur
du sol un socle dè pierre~ large de o,15 à o,2o m., é-
galement visible à l'intérieur et à l'extérieur. Les
angles du camp sont arrondis et munis de tours inté-
rieures de plan trapézoïdal, ainsi que de saillies du
front avec une moyenne de o,25 m. et aux extrémités
renforcées de briques; ctfs saillies sont organiquement
liées à l'enceinte sur une longueur qui peut aller jus-
qu'à 1,3o m. et alternant avec les pierres du parement
(Pl. 54. 4) • Les chambres des tours sont inégales et la
même inégalité se manifêste lorsqu'il s'agit de l'épais-
seur
t
des murs ou des dimensions des ouvertures d'accès
(Pl. 54. 5) •
A l'extérieur, l'enceinte est munie de plate-
formes faisant saillie sur le front de la muraille
(Pl. 54. 6)·: de même que dans· le cas des tours, leurs
extrémités sont en brique (Pl. 54. 7) ; il y a deux plate-
formes de chaque côté, entre les tours d'angle et le$
portes. Elles sont de dimensions variables, d'une lon-
gueur qui peut aller de 3,95 jusqu'à 6,lo m•• pour une
·épaisseur moyenne de o,25 m. Il convient de mentionner·
la présence d'un voûte en brique, avec l'ouverture
large de o,-34 m. et haute de o, 54 m., qui est le dé-
bouché d'un canal, fort probablement d'évacuation;cette
voûte se trouve dans l'aile septentrionale, à 36 m. de
la tour sise au nord-ouest (Pl.54.8) • L'aqueduc, diri-
.gé depuis un réservoir de captation, a été localisé

819
Pl. 54. 1 L'aile est du camp de Copaceni, entre la porta praetoria
et la tour du nord-est.

Pl. 54. 2 L'aile septentrionale du camp de Racovita.

Fig. 54. 2 Relevé du camp de Racovita (Praetorium Il). 1978

821
 Pl. 54.. 3 L'aile occidentale du camp.

Pl. 54. 4 L'angle de sud-est, avec la saillie du front renforcée de

briques et 1'enceinte orientale.

Pl. 54. 5 La tour du nord-ouest.

Pl. 54.6 L'enceinte meridionale avec la plate forme sise entre la tour
du sud-est et la porta principalis dextra.

Pl. 54. 7 L'angle de nord-ouest aux extremites en brique.

Pl. 54. 8 Conduit pour les eaux d'ecoulement sous 1'enceinte

septentrion ale.
 dans une colline· qui se dresse à l5o m. sud-est du camp.
Il s'agit d'un conduit composé de tuyaux en terre cuite
longs de 29-35 cm., avec l'un des bouts ~vas~s et re-
liés au mortier.
Les portes sont flanquées de deux tours intéri-
eures, leur front dessinant des saillies, mais il y a
une différence de dimensions entre ces tours, les plus
grandes sont celles de la porta praetoria (4,6o x 5,lo
m. - Pl.54.9 ), les plus petites celles de la porta de-
cumana (2 x 3 m.). cependant que les tours des portes
principales, de dimensions moyennes (Pl~ 54.10) ne sont
guère parfaitement égales entre elles. Tout comme dans
le cas des tours d'angle, leurs murs, ainsi que leurs
entrées sont d'épaisseur variée, de même, du reste, que
leur ouverture, dont les dimensions changent selon le
cas.
A l'intérieur, on n'a dégagé que le praetorium,
avec ses côtés inégaux: le côté oriental, dans lequel
est ménagée la porte. mesure 21,88 m·.. cependant que
le côté septentrional est long de 29,o8 m., le côté oc-
cidental de 25,4o m. et le côté méridional de 28,84 m.
L'édifice se pose de manière légèrement asymétrique par
rappprt à l'axe des portes praetoria et decumana. Bien
qu'on n'ait pas encore achevé de le fouiller, trois di-
visions s'ébauchent dès à présent; celle de l'ouest.
l'oecus,se composant ·elle-même de trois compartiments
(Pl.54.11). L'extérieur des murs du praetorium a ét~ bâ-
ti suivant le même appareil que celui de l'enceinte,son
côté occidental ~tant le seul à avoir les angles renfor-
cés de briques-. De même, les murs des chambres qui font
partie du groupe oecus sont revêtus à l'extérieur, du
côté du couchant, de briques, cependant que leurs pa-
. rois intérieures sont de pierre.
Vraisemblablement, c'est dans ce camp que fut
transféré le numerus burgariorum et veredariorum, qui
l'a même construit, peut être, au sein de la même gar-
nison de Praetorium. En effet, au stade actuel de la re-
cherche, aucun indice ne ~uggère la présènce de quel-

824
 Pl. 54. 9 La porta praetoria avec ses tours intérieures.

Pl. 54. 10 La porta principalis sinistra avec la tour occidentale.

Pl. 54. 11 Le groupe des pièces composant l'oecus, situées a 1'ouest

du pr aetor ium.
ques autres unités militaires en ces lieux - naturelle-
ment, la mise au jours de documents archéologiques
plaidant en ce sens n'est pas exclue.
L'horizon archéologique très mince, avec quelques
rares éléments de culture matérielle (fragments céra-
miques, une pointe de lance et une autre de piqua, un
boulet de ballista et des pierres à fronde), atteste
la brièveté relative de cet habitat, qui aura duré à
l'éventuel jusqu'à la retraite imposée par Aurélien.
L'investigation du voisinage immédiat du camp de Raco-
vita a permis la localisation du point où la principa-
le route stratégique de la vallée de l'Olt, qui pas-
sait devant la porte prétorienne du camp, était inter-
sectée par celle assurant la liaison avec le camp de
Tite9ti, sur la variante plus longue du versant orien-
tal du massif de Cezia; celle-ci constituait par la
même occasion la ligne supplémentaire du limes, sur le
tronçon correspondant au défilé (Fig. 54. 3) • La garnison
de Praetorium avait pour mission d'assurer tout à la
fois le trafic ët la défense entre Pons Vetus au nord,
le camp de Tite9ti à l'est et celui d'Arutela au sud.
Une large visibilité locale, ainsi que des possibili-
tés de liaison étaient assurées aux fortifications de
i

Praetorium grâce à quelques points de guet aménagés

dans les tours bâties sur les collines dressées au
nord-est du ruisseau du Clocoti9 et au sud de la ri-
vière Baie9ului, au lieu~it "Turnul lui Doanca" (11).
De cette manière, au sud, la visibilité était assurée
jusqu'à Arutela. On serait donc en droit d'avancer dès
maintenant l'hypothèse des liaisons visuelles, vers le
nord comme vers l'est, avec d'autres pointsde guet,na-
turels ou aménagés par la main de l'homme, ceci d'au-
tant plus que la distance jusqu'au point fortifié de
Pons Vetus dépassait de beaucoup celle normale entre
les camps du limes Alutanus.
Les recherches prat~quées dans les deux camps de
Praetorium confirment les conclusions en ce qui con-
c~rne la portée du système dêfensif de la Dacie romai-

826
 e Ocna Sibiuliu
trJ
Su ra Hom ba
Mica 0 Hosman

• Gu}terifo
Ro}ia

• 0 Feldioara

ungord +
++e+
Ca~olt
Socodate
~
0
Cisnëidioara
; Brodu

......... .

CAPVT STENARVM : ·::·:: : ::

-~ Turn~ R·~;~ () :: ( :::. :::.·w.:

:::::::::::::::::..::::::::::::::::::::::::::::::::::::::::-:.:::::....:::::::::::::::::::::::::::::::::~
.

·.·.·.·.·.·:.·.·.·.
Tit~~.~i ·-:-:-:-:-:-:-:-:·:·.

......
. ......... .
Peri jan i
Pripoa~ele

N\ S E
• Radacine~ti

!Il

•
Dobnceni
CASTRA
Simbotin
TRAIANA

01 ten i

Rimnicu Vilcea

Fig. 54. 3 Le limes Alutanus entre Rusidava et Caput Sternarum,

avec le tronçon des abords du massif de Cozia.
827
ne sur le limes Alutanus. Ces recherches offrent de
nouvelles données relatives aux objectifs militaires
de cette région. Elles contribuent de la sorte à une
meilleure connaissance de l'activité développée par
l'armée romaine, qui fut un facteur actif du processus
de romanisation sur la frontière orientale de la pro-
vince.

N o t e s
1 Gr.G.·Tocilescu, AEivi, 19, 1896, l, p.84, no 16.
2 Idem, Sapaturi arheologice în Oltenia, ms. Biblio-
thèque de l'Académie de la R.S.R., t.5133,pp.lo-ll,
l9,33.38,4o,42,44; Valuri, limes, drumuri 1 cet8ti,
ms. t.5139, pp.36,ll5,128. Archive Felonie, I, ms.B,
p.85, I Varia, l-71.
3 D.Tudor, Oltenia roman8 4 , Buc.,l978, pp.287-29o,32o.
4 Cr.M.Vladescu, Gh.Foenaru Bordea, Castrele hadria-
nee din valea carpatina a Oltului, Oltenia, Studii
~i Comunicari. Istorie, Craiova, 1974, p.54-55·
5 Lt.colonel Cristian M.Vladescu, Gh.Poenaru Bordea,
Castrul de la Racovita (jud.Vîlcea), communication
à la XIIe Session annuelle de rapports concernant
tes résultats des recherches archéologiques effec-
tuées en 1978, Oradea, 9-ll mars 1979. Cristian M.
Vladescu. Armata romana în Dacia Inferior (thèse de
doctorat), Buc., 1978, p. 75-81.
6 CIL.III, 13795 = ILS, 89o9. IDR, II, no 587,p.227-
228.
7 CIL.III, 13796 = ILS, 918o. IDR, II, no 558,p.228.
8 Cristian M.Vl8descu et Gh.Poenaru Bordea, Un seg-
ment din Limes Alutanus. Fortificatiile romane din
jurul masivului Cozia, Buletinul Monumentelor Isto-
rice, XLI, 3, 1972, p.27-32.Cristian M.Vl8descu et
Gh.Poenaru Bordea, Les fortifications romaines sur
le limes Alutanus dans la zone du massif de Cozia,
Actes du IXe Congrès International d'études sur les
frontières romaines, Mamaia, 6-13 septembre 1972,
Bucure9ti, Këln-Wien, 1974, p. 247-257.

828
 9 D.Tudor, op.cit., p.29o, mentionne, par erreur, la
monnaie à Racovita au lieu de Copaceni.
1o CIL.III, 14216, 19. IDR,· II, no 589, p.228-229.
11 Gr.G.Tocilescu, ms. t.5133, p. 2o.

Summary

The two Roman camps of Praetorium standing 500 metres apart in

the present-day villages of Copaceni and Racovita, district of Vficea,
were built one after the other, for the same purpose, on the left bank
of the Alutus (Olt). The camp of Copaceni, built in A. D. 138 by the
numerus burgariorum et veredariorum Daciae Inferioris and extended
in the year 140, was destroyed when the Olt burst its banks, prob-
ably towards the middle of the third century. Following this, the
second camp of Racovit,a, measuring 118.30 metres by 106. 80, was
built; it was occupied until about the time of Aurelian. The archae-
ological researches of 1973, at Copaceni, and 1976-1978, at Racovi"t;L,
brought to light: in the former, similarities with the other stone-
built camps_ to be found around the Cozia massif (Arutela, Rà:dacine§ti,
. Tite~ti) ; and in the latter, which is in a fine state of preservation,
quite marked differences in plan and method of construction (again in
brick). The two fortified camps of Praetorium guaranteed surveil-
lance of the strategically important limes Alutanus road over its
shortest distance between Arutela, to the south, and Pons Vetus, to
the north.

829
 55. PEDITES ET EQUITES SINGULARES IN DACIA

Closca L. Baluta

Le legatus Augusti praetorius ou consularis de la Dacie romaine, qui, par

mandat de l'empereur, était aussi le commandant suprême des armées, avait,
comme dans les autres provinces de 1' Empire romain ayant le même statut,
1
une unité de garde et d'escorte propre, formée de milites singulares.
Cette unité est attestée en Dacie sur des monuments épigraphiques à
caractère honorifique, votif ou funer2.ire, sur des plaques votives et sur des
statuettes. Elle est egalement attestée, avec son nom collectif, sur des
matériaux de construction estampillés.
Le nombre des epigraphes, connues jusqu 'à present, qui attestent 1'existence
de cette unité est de 10, dont 1'une a été decouverte dans la capitale de la Dacie,
illpia Traiana Sarmizegetusa, une autre sur le limes septentrional de la province,
à Negrile§ti, et huit dans le double centre urbain de la Dacie, Apulum. Le
nombre des tuiles ou autres matériaux de construction portrant 1'estampille
collective de 1'unité est de 131, dont 129 proviennent d'Apulum (Baluta et
Berciu, sous presse).
Le fait que ces témoigngages épigraphiques se trouvent dans une telle pro-
portion à Apulum s'explique par la présence quasi permanente du gouverneur
suprême de la province carpatique-legatus Augusti Daciae (Daciarum)-dans
cet important centre militaire, économique, administratif et routier de la
Dacie, siège permenent de la XII:f legion Gemina au long des 17 decennies de
domination romaine au nord du Bas-Danube.
Les militaires faisant partie de 1'etat-major du commandant suprême ou
d'autres services mineurs attestés epigraphiquement, avec leurs fonctions
respectives, dans la province carpatique sont les suivants: M. Calventius ·
Viator centurio legionis IIII Flaviae Felicis, exercitator equitum singularium ...
(CIL III 7904), sur un autel votif d'Dl pia Traiana Sarmizegetusa; P. Aelius
Crescens, duplicarius ( ?) S... C... eques singularis ... , sur en autel honor-
ifique d'Apulum (CIL Ill 1160); D. M. Aelius Genialis ex singularibus ... vexil-
larius, Valerius librarius ... , sur un monument funéraire d'Apulum (CIL ITI
7799); M. Aurelius Sila(nus=Silvanus ?) act arius equitum singularium ... , sur
une statuette en bronze du dieu Sol, à Apulum (Zefleanu 1949, 170f.); Iulius
Gracilis, beneficiarius ex singularibus, sur une plaque votive d'Apulum (Radu
1961, 105, no. 13, fig. 13; Rus su 1970, 164f. , fig. 2); Dasatus Scenobarbi
(filius) eques alae Batavorum ex Numero Sin.gularium stipendiarum ... Bers( ... )
Ingenuus decurio ex Numero eodem ... , sur un autel funeraire d'Apulum (CIL
III 7800; Christescu 1937, 196); ... eques alae lllyricorum ex Numero Sin.gu-
larium ... , sur une epitaphe funeraire d'Apulum (Radu 1961, 114, no. 29, fig.
29; Rus su 1967, 174, no. 8, fig. 8; 1977, 100, fig. 5); Mucasènus Ce(n)sorini

831
aeques ex singularibus consularis ... , sur un monument funéraire figuratif
d 1Apulum (CIL III 1195) et ... decurio equitum singularium, sur un ex-voto
funéraire d 'Apulum (CIL III 7787) .. A Negrile~ti est attesté, sur un autel votif,
... P. Aelius Attilianus, decurio ex singularibus consularis (Gabor 1911, 433).
A noter que certains des soldats, sous-officiers et officiers mentionnés
sur ces épigraphes ont un seul nom "barbare", comme Mucasenus, Dasatus,
d'oÙ, selon toute apparence, on peut déduire qu'ils n'avaient pas encore la
citoyenneté romaine. Cependant, la plupart d'entre eux ont cette citoyenneté,
ainsi qu'il ressort du moins de leurs noms propres de type romain, corrects.
Cette situation semble d'ailleurs être exceptionnelle, vU. que les milites
singulares consularis ne pouvaient, en fait, être citoyens à pleins droits,
ètant donné que les cavaliers des alae ne recevaient la qualité de cives qu'au
moment de leur libération-honesta missio. (Mateescu 1923, 184; Grosso 1965, 641J,
Les matériaux céramiques portant 1'estampille de 1'unité de garde et
d'escorte du gouverneur de la Dacie comprennent presque toute la gamme des
produits nécessaires pour la construction des édifices. 2 Les estampilles
appliquées sur ces materiaux sont r1 'une remarquable diversité et variété
typologique. Le nom collectif de 1'unité apparan sur ces estampilles sous les
formes: soit Pedites Singulares (Speidel1972, 299; 1974, 371) ou seulement
Singulares, soit Equites Singulares3 soit Numerus Singularium (Cagnat 1892,
117; Howell 1937, 1327, ~?37; Benes 1970, 192, no. 24).
Les estampilles portant le nom de la formation d'infanterie sont de plusieurs
types: P{edites) S(ingulares), avec la lettre Pliee au bord du cartouche 4 (Fig.
55. 1. 1); P(edites) S(ingulares), avec la lettre P indépendante5 (Fig. 55. 1. 2);
P(edites) SIN(gulares), avec la boucle de la lettre P pleine 6 (Fig. 55.1. 3);
PED(ites) S(ingulares), avec la lettre S gravée en "positif" (Fig. 55. 1. 4).
PED(ites) S(ingulares), semblable à la précédente, mais d'une forme plus
elégante1 (Fig. 55. 1. 7); PED(ites) SIN(gulares), avec des lettres renversees
et inversées (Fig. 55.1.10); PED(ites) SIN<gulares), avec certaines lettres
tangentes aux bords du cartouches (Fig. 55. 1. 8); PED(ites SING(ulares), avec
la lettre S gravée en "negatif" (Fig. 55. 1. 5). PED(ites) SING(ulares), avec
les hampes des lettres inscrites sur les bords du cartouche (Fig. 55. 1. 9);
PED(ites) SING(ulares), avec hedera distinguens entre les lettres abrégées 9
(Fig. 55. 1. 6); P(edites) S(ingulares) C(onsularis), avec les lettres à coupe
classique 10 (Fig. 55. 2. 4); P(edites) SIG(ulares) COS(ularis), avec omission
de la lettre N dans les épithètesll (Fig. 55. 2. 3).
Les estampilles au nom générique de L'unité, Singulares, sont d'un seul
type, avec deux variantes: SINGVL(ares), de dimensions réduitesl2 (Fig.
55. 2. 6) et SINGV... , avec des lettres de grandes dimensions (Fig. 55. 2. 5).
L'unité de cavalerie a un seul type d'estampille, qui ne se trouve que sur des
tuiles creuses. EQVIT(es) SING(ulares)13 (Fig. 55. 2. 7). &Ir certains matéri-
aux céramiques l'estampille de l'unité, Numerus Singularium, a été appliquée
dans deux variantes du même type: N(umerus) SING(ularium), avec des lettres
renverséesl4 (Fig. 55. 2. 2), et N(umerus) SING(ularium), avec les lettres
disposees normalement15 (Fig. 55. 2. 1).
Sur certains estampilles de l'unité d'infanterie apparru"t l'épithète Consularis.
Cela indique que, fort probablement, l'unité ne portait le nom de Pedites
Singulares Consularis (ou Numerus) que lorsque la province avait un gouveneur ·

832
 cO
lO
.
lO
•
.....bD
~
833
 :-···~?-·.
'.·.;··.t,;.··<.:
.
•• , . '"'!
: ~-·' .-· .·./
.
: '..:;·. __
' ...
; ....... .·.•
.
...
' . ~ .' '
• • 1 • •
(o'\~~~.·1··
~] . . ·(
••. 1-
.
N
LC')
.
LC')
~4
de rang consulaire. A souligner 1'absence de 1'epithète Consularis dans le nom·
de 1'unite de cavalerie, Equites Singulares, autant sur les monuments epi-
graphiques que sur les estampilles, pour la periode durant laquelle la province
était gouvernée par un praetorius. La modification du rang du gouverneur de
la province a été déterminée en Dacie par 1'evolution administrative et terri-
toriale de la province et par les étapes d'organisation de ses troupes d'oc-
cupation. 16
Presque tous les matériaux de construction estampilles ont été decouverts
sur le territoire de la colonie Nova Apulensis, 17 dans quelques zones archéo-
logiques groupées autour de 1'ensemble qui comprend les thermes, des temples
et des édifices publics, 18 y compris le siège présumé du gouverneur (Berciu
1948, 192; Berciu et Popa 1964, 302; 1965, 180 n. 79). Les constructions
de cette zone archeologique, qui sont toutes pourvues d'installations de chauf-
fage central, sont bâties avec des matériaux céramiques produits soit par les
ateliers spécialisés des unités de Numeri, soit par les ateliers locaux de la
xme legion Gemina, d'oÙ il résulte que ces édifices à caractère public sont
le fruit d'une activité commune de ces ateliers.
Sur le rapport numérique, on remarque la disproportion entre les matéri-
aux portant 1 'estampille des Pedites Singulares et ceux portant 1'estampille
des Equites Singulares, les premiers ~tant beaucoup plus nombreux que les
seconds. En ce qui concerne le rapport entre les matériaux portant l'estampille
de 1'unite de garde du gouverneur et ceux appartenant à la xur~ légion Gemina,
avec ou sans noms de personnes, ces derniers sont, tout naturellement,
infiniment plus nombreux.
La plus ancienne attestation documentaire des troupes de garde et d'escorte
du gouverneur de la Dacie est 1'auteul votif d'Ulpia Traiana Sarmizegetusa, qui
date des premières décennies après la conquête; la plus récente est 1'epitaphe
d' Apulum, ainsi que des matériaux de construction estampilles mis au jour dans
les; zones situées au nord et au nord-ouest de l'ensemble comprenant les
thermes qui datent de la fin du ne siècle ou du début du siècle suivant.
La date exacte de 1' apparition des deux unités de garde du gouverneur de
la Dacie ne peut être établie. Des données concluantes dans ce sens ne nous
sont offertes ni par la chronologie des types de moules des estampilles, ni
par la datation approximative des édifices respectifs.
Nous ne connaissons pas non plus le point précis où était le siège de 1 'unité
de garde du gouverneur, dans le cadre de la grande garnison d'Apulum. Nous
ignorons, de même, si cette unité speciale avait son propre castrum ou
castellum, ou si son siège se trouvait dans le camp de la~ legion Gemina.
Les matériaux épigraphiques dont on dispose ne nous apprennent rien au
sujet du grade du commandant de cette unité, ni sur sa qualité, ni sur son
rapport hiérarchique avec le commandant de la légion.
Les estampilles de 1'unité de garde du gouverneur de la Dacie, avec leurs
caractéristiques typologiques, éclairent un chapitre trop peu connu encore de
1'histoire militaire romaine. De même, les donnees assez riches des epi-
graphes complètent le répertoire anthroponomastique de la province carpatique,
ainsi que la liste des fonctions de 1'etat-major ou d'autres services mineurs
du commandant suprême de la Dacie romaine.

835
 NOTES

1. On les appelait Singulares d'après leur mode de recrutement; cf. CIL rn,
p. 2501; Cagnat 1909, 1346-7.
2. Briques parallelipipediques-rectangulaires ou carrées (lateres), briques
cylindriques pour hypocauste, tuiles de couverture (tegulae hamatae),
tuiles pour placage de parois (tegulae sine marginibus), tuiles pour hypo-
causte (tegulae mammatae), tuiles demicirculaires de couverture
(imbrices), etc.
3. Cf. Cagn.at 1892, 789-91; Liebman-1907, 312; von Domaszewski 1908,
50-3; 1911, 19ff. ; Mateescu 1923, 86 n. 3; Fiebiger 1927, 237;
Webster 1969, 269; Speidel 1965; Levi 1965, 467f. ; Durry 1966, ~42 ;
Grosso 1966, 900-9; Panciera 1974, 221; Benes 1970, 192, no. 22;
Speidel 1978.
4. Cserni 1892, 26, no. 74; 1903, 98, Plan 1; 1901, 529, no. 80; Oehler
1898, 118, no. 5; Szilâgyi 1946, 58, no. XJX. 287.
5. Cserni 1908, 42; Oehler 1898, 118, no. 5; Szilâgyi 1946, 58, no. XlX.
287. Une tuile ài'estampille P. S. d'un type complètement different,
dans un cartouche en forme de tabula ansata, datant probablement de la
periode de la conquête de la Dacie, a été decouverte à Buridava (Dacie
Inférieure). Elle pourrait plutôt être assignée au corp de Pedites Singulares
du legat de la Mesie Inférieure, cf. Tudor 1964, 346; 1965, 185-6; 1978,
341, fig. 60. 11; Benes 1970, 192, no. 23; IDR TI 564.
6. Cserni 1890, 37, no. 2 a-d; 1901, 529, no. 79; Oehler 1898, 118, no. 4;
CIL Til 12633 c (ad ill 8084. 33); Téghts 1911, 499f. ; Szilagyi 1946, 57,
no~ XVITI. 282; Berciu 1948, 195, no. c; Benes 1970, no. 23.

7. Berciu 1948, 195, no. b. Au J'viusée des Portes de Fer de Drobeta.:Turnu

Severin il y a une brique portant une estampille semblable, mais plus
petite, que 1'on considère comme provenant d' Apulum, cf. Tudor-
1965, 185 n. 23, fig. 6; Benes 197 0, 192, no. 23; D. Benea (1978, 206,
no. 196, fig. 24) cite des matériaux en céramique portant la même
estampille à Germisara ( ?). Mais vu· les différences de dimensions du
cartouche que des lettres entre 1'estampille d'Apulum et celle du Musée
des Portes de Fer, nous ne croyons pas que la brique en question provienne
d'Apulum.
8. Cserni 1890, 39, nos. 9-11, 40, no. 11 ou il reproduit l'observation de
Jung (1891, 99) "dans les estampilles nos. 8, 10 et 11, N. P. nous
rappelle N(umerus) P(almyrenorum)" ,lecture erronée, considérent SIN
inversé, les lettres N. P. Cserni (1891 ,- 23; 1902, 12f., no. 113) crQit
qu'elle est privée, la premiere de ce genre decouverte par lui; cf.
Cserni 1801, 529, no. 82-3, ·où il reproduit l'estampille no. 9 de Cserni

836
 1890 et do~e aussi une lecture correcte pour les nos. 10 et 11; Oehler
1898,118, no. 3; CILID12633d @!!III8074.33); Szilagyi1946, 57,
XVIII. 283-4.
9. Cserni 1890, 39, no. 8; 1891, 22 et 27,no. 33; 1897, 44,no. 8; 1901, 529,
nos. 76, 81; CIL III 12633 a @9_ III 807 4. 33); Szilagyi 1946, 57, nos.
XVIII. 278-9; XIX. 285; Benes 1970, 192, no. 23.
10. Cserni 1890, 38, no. 3; 1904, 103; 1908, 42, no. 4; 1901, 529, no. 78;
CIL III 12633 g @&III 8074.33); Szilâgyi1946, 57,no. XIX.286; Benes1970,
192, no. 23.
11. Cserni 1890, 38, no. 3; 1901, 529, no. 77; Oehler 1898, 118, no. 2;
CIL III 12633 f @9. III 807 4. 33). Szilagyi 1946, _58, no. XIX. 288, a lu
P(edites) SI(ngulare)S CO(nsulari)S; Benes 1970, 192 no. 23 (P. SIS. COS).
12. Cserni 1891, 26, no. 97, plan IV; 1901, 530, no. 88; Szihigyi 1946, 57,
no. XVIII. 281; Benes 1970, 192, no. 23.
13. Cserni 1891, 24, no. 23; 1901, 524, no. 31; CIL rn 12633 h ~rn 8084.
33); Szilagyi 1946, 57, no. XVTII. 280; Benes 197 0, 192, no. 22.
14. Cserni 1890, 38, no. 4 et 39,no. 6; 1891, 36,no. 1 b, plan I2; 1894, 12,
28; 1897, 44, no. 9; 1903, 94, plan I, 98 1 no. 3, plan I et 101 no. 3; 1908,
42,no. 5; 1901, 529, nos. 73-4; CIL rn 12633 a-b @9. III 8074. 33);
Szilâgyi 1946, 56, no. XVTII. 270; Berciu 1948,195, no. a (une tuile sur
laquelle sont incisées a main levée les terminasons de deux noms ... ATVS
et ... ANVS); B~nes 1970, 192, no. 23.
15. IDR III 1, 243, fig. 188, XLI, un exemplaire probablement d'Apulum.
n n'est point exclu que ce soit un abject importé d'Apulum dans Banat.
a
16. IDR III. 1, 16-26, les chapitres qui se referent l'evolution territoriale
et administrative de la Dacie romaine (avec une ample bibliographie).
17. Cserni 1901, 141, 173; Daicoviciu 1945, 128-9, no. 4 et 170-1; 1950,
225-8; Russu 1949, 145-161;Florescu 1949, 162-9; Tudor 1968, 144-82;
Vittinghoff 1968, 135ff. ; 1971, 410; Macrea 1969, 125ff. ; Popa et Aldea
1972, 209-220; 1973, 490ff.; Daicoviciu et Daicoviciu 1973, 97.
:
.. ' .- --- .- .- ., -
18. Les fouilles sont publiees dans Az alsofehermegyei tHrtenelmi, regeszeti
es termeszettudomanyi tarsulat evkHnyve I (1888)-XVIII (1917); Cserni
1901; Archaeologiai ÊrtesitH; et Muzeumi és Kt5nyvtâr ÊrtesitH I (1907)-
XII (1918 ).

837
 BIBLIOGRAPHY

ATE. Az alsofehérmegyei tortenelme, régészeti és természettudomanyi

târsulat évkOnyve.
Baluta, C. L. et Berciu, I. sous presse. 'Sigilla tegularia Peditum et Equitum
singularium in Dacia', Latomus.
Benea, D. , 1978. 'Caramizi §tampilate rn Muzuel Por~ilor de Fier',
Apulum xvi, 206.
Benes, J. , 1970. 'Die rOmischen Auxiliarformationen im Unteren Donau',
Sbornik Praci Filosofike Faculty, Brnenske University.
Berciu, I. , 1948. 'Descoperiri arheologice rn Apulum', Apulum iii (1946-48),
192-5.
Berciu, I. and Popa, A., 1964. 'Exceptores consularis in Dacia', Latomus
xxiii, 302-10.
Berciu, I. and Popa, A., 1965. 'Monumente epigrafice din Apulum', Apulum
v' 180-7.
Cagnat, R., 1892. 'Equites singulares', in Daremberg, C. and Saglio, p;:-·
(edd. ), Dictionnaire des antiquités greques et romaines, Paris, II, 789-91.
Cagnat, R., 1905. 'Numerus', in Daremberg, C. et Saglio, E. (eds. ),
Dictionnaire des .antiquités grecques et romaines, Paris, IV, 117-8.
Cagnat, R., 1909. 'Singularis', in Daremberg, C. and Saglio, E. (edd. ),
Dictionnaire des antiquités grecques et romaines, Paris, IV, 1346-7.
Christescu, V. , 1937. Istoria militara a Daciei romane, Bucure~ti.
Cserni, A. , 1890. .ATE iii, 37.
Cserni, A., 1891. ATE iv, 22, 27.
Cserni, A., 1892. ATÉ v, 26.
Cserni, A. , 1894. ATÉ vi, 12, 28.
Cserni, A., 1897. ATÉ ix, 44.
Cserni, A., 1901. 'Alsofehér vârmegye torténelme a romai korban', Alsofehér
varmegye monografiaja ii. 1.
Cserni, A. , 1903. ATÉ xii, 98.
Cserni, A., 1904. ATÉ xiii, 103.
Cserni, A. , 1908. ATÉ xiv, 42.
Daicoviciu, C., 1945. La Transylvanie dans 1'antiquité, Bucarest.
Daicoviciu, C., 1950. 'A§ezarea autohtona de la Apulum', Studii si cercetari
de istorie veche i, 225-8.
Daicoviciu, C. et Daicoviciu, H. , 1973. 'Urbanisation et romanisation dans
la Dacie trajane', Vestigia xvii; 97-8.
Domaszewski, A. von, 1908. Die Rangordnung des rOmischen Heeres, Bonn.

838
Domaszewski, A. von, 1911. 'Equitibus Singularibus Praesidium',
Archaeologisch- Epigraphische Mittheilungen ~, _19ff.
Durry, M., 1966. Review of Speidel, M. P. Die Equites Singulares Augusti,
in Révue des Etudes Latines xliii, 542.
Florescu, G. , 1949. Apulurn iii, 162-9.
Fiebiger, 1927. 'Singulares', in Wissowa, G. (ed.) Paulys Real-encyclopttdie
der Classischen Alterturnswissenschaft, IliA, 237. -
Gabor, F. , 1911. 'Két ramai felirat daciabOl' ,_ Archaeologiai ÉrtesitO xxxi,
433.
Grosso, F. , 1965. 'il diritto latina ai militari di eta flavia', Riv. Cult. Class.
e Med. vii, 541-560.
Grosso, F., 1966. 'Equites SiDooulares Augusti', Latomus xxv, 900-9.
!QB Inscrip\iile Daciei Romane, ed. Bussu, I., 1975-77.
Jung, J., 1891. 'Siebenburgische Inschriften', Archaeologisch-Epigraphische
Mittheilungen, 97-100.
Liebenam, W., l~_Q7. 'Equites Sinooulares', in Wissowa, G. (ed. ), Paul,ys
Real-encyclopttdie des Classischen Altertumswissenschaft VI, 312-21.
Levi, M. A. , 1965. Review of M. Spejdel, Die Equites Singulares Augusti,
Athenaeum xlii, 467-8.
Macrea, M., 1969.- Viata mDacia romana, Bucure~ti.

Mateescu, C. G. , 1923. 'Le tracee nelle epigrafi di Roma', Ephemeris

Dacoromana i, 184.
Oehler, J., 1898. 'Antike.Denkmruer in Siebenburgen', Jahreshefte des
Osterreichischen Archaeologischen Institutes i.
Panciera, S., 197 4. 'Equites singulares-nuove testimonianze epigrafiche',
Rivista di Archaeologia Cristiana iii, 221-47.
Papa, A. and Aldea, 1. A., 1972. 'Colonia Aurelia Chrysopolis', Apulum x,
209-20.
Papa, A. and Aldea, I. A., 1973. 'Colonia Aurelia Apulensis Chrysopolis,'
Vestigia xvii, 490-2.
Radu, D., 1961. 'Materiale epigrafice din Muzeul Regional Alba Julia',
Apulum iv, 105.
Rowell, H. T. , 1937. 'Numerus', in Wissowa, G. (ed. ), Paulys Real-
encyclopttdie XVI, 1327-41, 2537-54.
Russu, 1. , 1967. 'Note epigrafice. Seria a X-a. Inscrip\ii din Dacia
revizuite', studii §i cercetari de istorie veche xviii, 17 4-6.
Russu, 1., 1949. 'Apulum', Apulum iii, 145-161.
Russu, 1., 1970. 'Note epigrafice. Inscrip\ii din Apulum', Studii Clasice xii,
164-5.
Russu, I. , 1977. studii §i cercetari de istorie veche §i arheologie xxviii,
100-1.

839
Speidel, M. , 1965. Die Equites Singulares Augusti, Bonn.
Speidel, M. , 1972. 'The pedites singulares Pannoniciani in Mauretania',
American Journal of Philology xciii, 299-305.
Speidel, M., 1974. 'Pedites Singulares' in Wissowa, G. (ed.) Paulys Real-
encclop!':ldie der classischen Altertumswissenschaft, Suppl. XIV, 371-5.
Speidel, M., 1978. Guards of the Roman Armies. An Essay on the Singulares
of the Provinces, Bonn.
Szilagyi, J., 1946. A Daciai erodrendszer helyorségei és a katonai téglabely-
egek (Dis s. Pann. ser. 3, xxi), Budapest.
Teglas, G., 1911. 'Neue Beitr::tge zur Inschriftenkunde Dakiens' Kl.io x, 499-
505.
Tudor, D. , 1964. 'Pedites singulares à Buridava', Dacia viii, 345-51.
Tudor, D. , 1965. 'Comuml'ri epigrafice: II', studii §i cercetari de istorie
veche xvi, 185-6.
Tudor, D., 1968. Orase, tfrguri §i sate mDacia romana, Bucure§ti.
Tudor, D. , 1978. Oltenia romana, 4th ed. , Bucurestji.
Vittinghoff, F. , 1968. 'Die Bedeutung der Legionslager fttr die Entstehung
der rOmischen Statte an der Donau und die Dazien', in Claus, M. ,
Haarnagel, W. and Raddatz, K. (edd. ), Studien zur europ!':lischen Vor-
und FrUhgeschichte, 135ff.
Vittinghoff, E., 1971. 'Hauptprobleme der Urbanisierung rOmischer Provinzen',
in Acta of the Fifth Epigraphie Congress, Cambridge 1967, Oxford. 407-11.
Webster, G. , 1969. The Roman Imperial Army, London.
Zefleanu, E. , 1949. 'Note epigrafice la Apulum: II', Apulurn iii, 170-71.

840
 Summary

The guard and escort unit of the governor of Dacia-legatus

Augusti praetorius or consularis-is attested both on inscriptions
and tile stamps. Soldiers from the supreme commander's general
staff or from other minor offices are present together with their
functions on inscriptions of an honorific, votive or funeral character.
The collective name of the unit is present on tile stamps. The infan-
try formation-Pedites Singulares-is known on six types of stamp
and on some variants, while the cavalry-Eqvites Singulares-is
known on only a single type of stamp. The name Numerus Singularium
is represented by a single type of stamp with two variants, while the
generic name-Singulares-is attested on a single type of stamp again
with two variants.
Eight of the ten inscriptions and 129 of the 131 tegulae stamps
come from Apulum (colonia Aurelia Apulensis or colonia Nova
Apulensis). The concentration in a single centre is due to the almost
permanent presence of the Supreme Governor of the Carpathian prov-
ince-legatus Augusti Daciae (Daciarum)-in Dacia's most important
centre-Apulum-the permanent headquarters of the Xlllth legion
Gemina throughout almost 170 years of Roman domination over the
northern part of the lower Danube.
The oldest documentary attestation of the unit is the votive altar
from the capital of the Roman Dacia- Ulpia Traiana Sarmizegetusa-
dating from the first decades after the conquest, while the latest is
the epitaph from Apulum and the tile stamps from the northern and
north-western areas of the baths dating from the end of the 2nd cen-
tury and the beginning of the 3rd century A.D.
The Governor's residence in Apulum is already known, but the
unit's headquarters within the large garrison from Apulum has not
yet been defined.

841
 56. RESEARCH ON THE LIMES :PALAESTINAE:
A STOCKTAKING

Mordechai Gichon

Since its creation in 1948, Israel has carried the burden of constant war
with all its neighbours and of absorbing at the same time a population six
times as large as its initial number in 1948 and hailing largely from under-
developed countries.
By necessity, archaeological research had to be low on the list of public
expenditure. Against this, the speed and extent of modern development im-
mediately began to encroach upon ancient vestiges that for centuries had been
undisturbed by hiunan activities. The Roman and Byzantine remains suffered
most, being in most cases the most exposed to interference, since they had
been frequently preserved as visible ruin-s above the surface.
Only the dogged efforts of the seriously understaffed Israeli governmental
department of antiquities prevented the unrecorded destruction of many sites.
This meant continuous work on more often than not hasty rescue excavations
that often did not allow for the necessary leisure to follow up all clues so as
to come to comprehensive conclusions. Nor did they always provide sufficient
published evidence to permit a critic~l appraisal.
Planned research aimed explicitly at the study of the Limes Palaestinae
has so far been limited to Tel-Aviv University, and even there work on other
fields had to be carried out too.
Aside from our surveys and excavations at En Boqeq, Tamara, Tsafit
and Malatha, 1 excavation on major Limes sites was nearly always ancillary
to the main effort of probing into earlier and lower levels. One colleague
has however devoted much effort to excavating sites of purely Roman and
Byzantine character: Rudolf Cohen, supervisor of Antiquities, southern dis-
trict. Alas, much of his effort was directed to rescue operations which were
hamstrung by shortage of time and funds. Second to him, Dr. Z. Meshel
must be mentioned, though his work was conducted within the much wider
scope of general regional studies, without special emphasis on the Roman
period. For the most part, only very preliminary publications of their work
(and mostly without plans, sections and illustrations of the small finds) are
available, either in Hebrew only, in the Hebrew- language Archaeological
News (Hadashot Archeologiyot) of the Israel Department of Antiquities, or in
the "Notes and News" of the Israel Exploration Journal.
Excavations have been carried out by the late Y. Aharoni in biblical Arad
and Tel-Sheba, both of which had Limes fortresses built above the earlier
levels. Whereas Arad's Roman level has not yet been published even in a
preliminary way beyond a very brief mention, the final report of the Tel Sheba

843
castellum was published in 1975, of which more later. The biblical outpost,
south of the Beer Sheba valley, Aroer, has been under investigation since
1976/7 by A. Biran. Its uppermost fortress was destroyed during the Bellum
Iudaicum of Vespasian and Titus (Biran and Cohen 1976, 139; 1977, 249; 1978,
197). Nabatean towns that continued into Byzantine times have been under
intermittent excavation by A. Negev since 1958. 2 He was joined in 1975 by
Y. Tsafir, who excavated the castellum of Rehovot in the Negev and parts of
the nearby town (Tsafir 1975; 1976).
Beno Rothenberg has enriched our knowledge of the Araba valley and that
of Roman Sinai. 3 The extremely important complementary and comparative
material from the Limes Arabicus has been amply summarized by Graf (1978,
1-26) and the findings of a recent comprehensive survey have been prelimi-
narily published by Parker, who must be commended for his quick publication
(197 6 ' 19). 4
A convenient start to a summing up of our present state of research on
the Limes Palaestinae and the insight gained from it, is the enumeration of
some of the conclusions drawn by me in 1968 in my Ph. D. thesis:
a. The Roman Limes is of Flavian origin and is based upon former border ·.
defence.
b. The shape of the Roman castella and burgi continues pre-Roman patterns.
c. Trajan occupied the Negev immediately on his occupation of Nabataea.
d. Under Hadrian, the Roman defences were largely moved northwards and
eastwards.
e. The Severan period witnesses a renewed penetration into the Negev, in-
cluding the rehabilitation of Oboda and the foundation of agricultural
settlements of limitanean character.
f. The great Diocletianic-Constantinian reorganisation may have been, to a
certain extent, anticipated by Aurelianus and his immediate successors.
g. The Araba valley became the central connecting link between the western
and the eastern part of the Limes Palaestinae when Diocletian added to it
all of southern Arabia, both for lengthwise and lateral communications.
h. From its beginnings, the Negebite Limes was constructed to hold off the
Arab nomades, and in times of growing danger from Persia and its Arab-
ian satellites, it gave additional safety against a Persian or Persian-
inspired thrust.
The first site to be excavated on the main line of the Judean Limes was
Arad, dug by Y. Aharoni and associates from 1963 onwards. The present
writer supervised the Roman level, without having a part in the consequent
processing of the material. As already stated, no report beyond short state-
ments is yet available for the Roman level (EAEHL I, s. v. Arad; Aharoni
and Amiran 1964, 131-47).
The quadrangular castellum, measuring 25 x 30 m, followed the central-
courtyard type, with rooms arranged around it and leaning on the curtains.
Typical of many limitanean installations of the Roman period, the place was

844
kept very clean and nearly void of pottery. Aharoni dated the fort, on ceramic
evidence and by one Greek ostracon, to the 1st cent. A.D. With our present-
day knowledge, the date may be extended to include the first half of the 2nd
century. Later the fort seems to have been abandoned until the late Byzantine
period. It overlay a strong Hellenistic-Hasmonean m)pyoc; 5.
TEL SHEBA castellum (Fig. 56.1) was excavated by V. Fritz in 1974
under Aharoni's directorship (Aharoni 1973, 83-9, with plan, pl. 2). The
final report has only a ground plan; sections and pottery drawings are missing.
The shape of the castellum is slightly trapezoidal; the sides measured between
31-32 m. It too follows the courtyard pattern. The pottery, scarce in the in-
side, but more abundant on the outside,largely belongs to the 1st and 2nd cen-
turies. Such was also the impression of the present writer who has had the
privilege of examining it. The picture tallies with that of the writer's previous
survey of that site. However, the question still lingers whether the Herodian
material from around the fortress should be associated with it and not with an
earlier Hellenistic fortification discovered below. 6
Numismatic evidence from within the castellum proves its existence in
A.D. 112. The relevant coin, found in one of the inner partition walls, has
been taken as a terminus a guo for its erection (Fritz 1973, 86-7, n. 7;
Kindler 1973, 90-96). Proof is, however, lacking for this partition's absolute
contemporaneity with the outer walls. Another coin, dated to 251-3 (Volusianus,
minted in Neapolis), found in room 362, affords a terminus post guem for the
abandonment of the castellum. On the artificially levelled surface outside the
fort, a coin of Const_antius II (327 -337) was picked up, as well as 4th century
folies (Fritz 1973, 87 n. 7; Kindler 1973, 96). Three other coins from out-
side the castellum belong to Augustus, to Domitian and to an unidentified 1st
century emperor respectively. As long as no information on the Hellenistic
to Herodian (?) pre-castellum level (?) is available, no final verdict on the
relationship between these coins and the castellum can be given (Fritz, loc.
cit~; Kindler, loc. cit.). ·
The castellum was refurbished according to Fritz in the early Arab
period. This again tallies with the observations from our survey on the
scarcity of late-Byzantine wares. among the surface finds. These include,
however, pottery belonging to the late 6th and early 7th centuries by its simi-
larity to well stratified material from En Boqeq and Mount Nebo. 7 Fritz has
noted the relative smallness of the castellum. This is correct. However,
none of the many Judean Limes castella seem to have been able to accommo-
date more than part of the units stationed in them. In this respect Judaea
differed from Arabia. Fritz 's functional comparisons between Arabian
"Zwischenkastelle" and Tel Sheba based upon size are therefore misleading
(1973, 87-8). There is, however, no doubt that the great abundance of Byzan-
tine vestiges from present-day Beer Sheba as against the scant pottery of that
period from the Tel, points to the former as base for the eguites Dalmatae
Illyriciani that were, according to the Notitia Dignitatum (Or. LXXITI.18),
stationed at Berosaba. The presence of the late-Byzantine pottery in the
castellum may indicate a refurbishing of this place as late as Heraclius'
reign. 8 The absence of Roman vestiges from the numerous rescue excavations
carried out in Beer Sheba since 1948 by the municipal archaeolQgists J.
Yesreoli and D. Berman9 indicates on the other hand that in the pre-Byzantine
period no large-size military establishment can be placed here with any proof
whatsoever.
845
 E
"'
J
I !
I /
' I I
I I I ,'
I I I I
I I
I I
I /
I I I 1
I I I I
I I I I
I 1 I ,'
I ,
I
I
I
I'··~·--,
I-
I I
I I
' I
I I
1 I
I I
I I
I /
I
.
I
I
I
I _.
I
I
I
a.
I 1
I '
I /
' I
I I
I
I I
'
I ,'
:./
.
.....bD
~
846
 In the spring of this year, the p.resent writer, ably assisted by Z. Shaham,
conducted a rescue excavation in the civil settlement at the feet of Tel Malatha.
Unfortunately funds ran out before we penetrated below the 4th century levels.
Surface finds are, tl'owever, most copious and cover the Hellenistic-Herodian
and early-to-late Roman period abundantly: they include Megaran bowls,
Rhodian jar handles, Eastern sigillata and later disc-shape lamp sherds. 10
This is a repetition of the picture gained from the examination of a trial
trench sunk in 1971 in the N. W. corner of the castellum on top of the te1. 11
No drawings have been published but the excavator reports pottery of the 1st
century of the Roman and the Byzantine period, and the existence of a Hellen-
istic fortification beneath the Roman site. It is not yet clear whether the
rcupyo<; of the Herociians (Jos. Ant. XVIII. vi. 2 .147) is to be identified with
the first phase of the Roman castellum or with the last phase of the Hellenistic
building.
Taken all together, the evidence adduc€9 by the present writer for an early
pre-Diocletian limes along the wadi line Nahal Besor- and Nahal Beer Sheba
is now safely established by excavation results. Further excavation must
prove its Flavian origin by other than pottery evidence. So far numismatic
proof exists from Trajan onward (Fig. 56. 2). There has also been much
argument over the immediate occupation of the Nabataen Negev by Trajan
and its subsequent history. The present writer's view of a Roman take-over
and continued Roman presence has been completely vindicated by additional
archaeological evidence, though, so as not to appear too self-righteous and
omniscient, I hasten to add that I accepted the idea of a complete later Roman
withdrawal-which was not so.
A. Negev, who has been one of the exponents of the absence of permanent
. Roman military and civil sedentary presence in the inner Negev, has been able
by; his own continuous researches to prove to himself and to us that there was
eo ntinui ty •
His excavations at Mampsis proved, by abundant pottery, coins, bulae
seals and a host of other evidence, continuous existence from Trajan to the
Arab conquest.12

A similar pic:!ture emerges now at Oboda where Negev excavated north

of the Nabatean pottery kiln a 2nd-3rd century quarter (Negev 1977b, 68).
A courtyard-type building there of 19 x 12 m has yielded coins of Antoninus
Pius, of Philippus Arabs and of 4th century date. Thus the time-gap between
the inscriptions known from this town, pertaining to A. D. 126, A.D. 242 and
Byzantine time, is being bridged (Negev 1967a, 47, 55).
Excavations by the present writer at Migdal Tsafit have revealed beneath
the numismatically well-defined reconstruction layer of the Illyrian emperors
,, from Claudius to Carus, a first floor of typical 2nd ang 3rd century pottery
(Gichon,1974a, 18, 22-3). -·

The courtyard-type castellum or road station at H. Hazaza, north of

Avdat (Oboda), which measures 51 x 22 m, was dated ceramically and numis-
matically in R. Cohen's excavation to the 3rd and 4th century (Cohen 1971a, 27).

847
 Key to figs. 2, 4 and 5.

e town

0 town and castle

• fort
• fortlet
• watch-tower
*- road-station
.&.M road-station (military)
• defended waterpoint
• limitanean fortified farmstead
.. limitanean settlement
road
secondary roads and tracks
definition uncertain
·-:-:-: boundary between main and forward area of defence approximate
existence as late as period indicated by title of map not sure

..
EWSA
wadi
other type of settlement
~-
excavated site

site abandoned during the period covered by the map

Roman and ancient Hebrew names in upper case, modern names in lower case
letters.
H = Horva (Hebrew for 'ruin') M = Mezad (Hebrew for 'fortlet')

848
 N A
8 A 20......
I I I
Fig. 56.2 The Judean limes in the Flavian period.

a
N

ROOM2

ROOM1

ROOMJ

0 2m

Fig. 56. 3 En Yahav, the burgus after Joseph (197 8, 14). ·

B49
Also much of interest is the sm-all (8 x 8 m) b~t very elaborately built burgus
at Har Haluqim, which is near Hazaza (Cohen 1976~ 47-50). Its ceiling was
supported by columns and a N3batean-type staircase tower leading up to the
missing second storey. The pqttery is clearly early 2nd to 3rd century.
Severan coins from Petra were picked up by R. Cohen during the excavation.
R. Cohen and Z. Meshel cooperated in the excavation of Mazad Maagara,
4 km west of Sde Boqer on the Oboda-Gaza rbad (1970, 25; Meshel 1974,
66-8). The site has a small Roman guadriburgus superimposed upon a square
Israelite Iron-age fort. The dimensions are about 35 m 2 and a great central
cistern was discovered. No: plans or sections have been published. The
Iron-age fort was refurbished in Hellenistic times and also used by the
Nabateans. The Roman edifice was established by ceramic evidence (?) as
definitely pre-Byzantine.
Two seaons of excavation were dev'Qted by R. Cohen to the great castellum
of Hazeva, 13 where a tower and parts of the curtain rooms and the courtyard
were unearthed. Hazeva in the Araba valley is the Negebite terminal of the
Scorpion pass road, which led thence to Aila or Petra respectively. Abundant
coins and pottery date the erection of this castellum-most probably by the
Nabateans-to the beginning of the 1st century A.D., and show that it existed
without a break to the time of Constantine. Incidentally, our conclusions drawn
by surface exploration have been·completely vindicated. Byzantine pottery
is conspicuous by its complete absence. Although I would ~till for some
small-scale protection of the source in Byzantine times, this rules out
Aharoni's identification with the Tamara mentioned in such late-Byzantine
sources as the Madeba map and the Berosabae Edicts (Alt 1921, 8; Avi-Yonab
1954, sect. 2). Incidentally, the-square corner towers of Hazeva, their
corridors and the mode of wall-construction are reminiscent of En Boqeq of
4th century A.D. date (Gichon 1971, 392). As to the pattern, see below.
The walls of rough ashlar faces and rubble filling can be traced back in the
Negev through the Herodian into the First Temple period-a warning against
their use for dating purposes.
A similar picture to that from Hazeva, but which shows Byzantine reoccu-
pation, comes from En Rahel (Rothenberg 1967b, 126), a road station in the
Araba, 20 kms south of Hazeva, while coins and pottery date the guadriburgus
of En Yahav, lying between the latter and Hazeva, to the 3rd century A.D.14
This structure measures 20 m2 and has its rooms leaning on to the curtain
walls.
We conclude that the Romans did continue to hold the Negev, including
the Arab a south of the Beer Sheba valley, after their occupation of Arabia
(Fig. 56. 4). Their posts follow the main roads. Since these linked the
main water sources, they controlled at one and the same time the traditional
routes of travel, commerce and war. Road research was carried out during
the period under survey by the author, by A. Negev, B. Rothenberg, z.
Meshel, I. Roll and B. Isaac.l5
A. Negev proved the shift of importance from the Petra-Moyet Awad-
Oboda road to that of the Scorpion pass as the major· tra_ns-Negebite axis
(1966, 89). Meshel proved the burgus on the Ramon cliff to be completely
Nahatean (Meshel and Tsafrir 1974, 103; 1975, 3). In both instances

850
 uo~

II '
I
I
I
I
I
'\
\
\
\
'
'\I
\
',
' ...
"

Fig. 56.4 The limes in the Negev in the 2nd and 3rd centuries.

851
the present writer's previous erroneous views were corrected (1963, 88).
On the other hand, my contention for a Byzantine use of the former road has
been proved by the confirmation of Byzantine finds near Moyet Awad (see
below) and trial trenches in Tahenat Moahila.16
Whereas Meshel ascribes the Petra-Moyet Awad-Oboda road to the
Nabateans, Isaac (see his article, pp.889f. below) has recently claimed Roman
origin for this road and its anepigraphic milestones.
We shall content ourselves with an even more concise review of the
Byzantine period. Generally speaking, it was typified by a strengthening of
the existing Limes.
Coin evidence from Tamara (Probus 276-282), Tsafit (Aelianus 270-275)
and En Yahav proves that in some instances the work of reconstruction had
already begun under the Illyrian emperors, soon after the reduction of the
Palmyra.17
Not by any means may all of the Byzantine work be attributed to Diocletian
and Constan.tine. En Boqeq' s earliest coin belongs to Constans (337-350)
(Gichon 1971, 396). Nessana's garrison bears the name of Theodosius II
(408-450).18
If we do confine ourselves to the Arab a valley, we can prove by excavation
for the Middle and late- Byzantine period, the 5th, 6th and 7th centuries:
a. the existence of a 40m2 fortress at Yotvata, possibly the Ad Dianam men-
tioned as road station on the Tabula Peutingeriana;~9 b. the existence of
a road station at the mouth of Nahat Par'an (Rothenberg 1973, 41).
Surface finds including coins on the site prove the existence of an exten-
sive and well guarded farm settlement near the by then ruinous Moyet Awad. 20
A plan of the quadriburgus (?) of that period from En Tamar was published
by B~ Rothenberg. Its surface finds included some very late-Byzantine wares.21
Taken altogether, the string of strong-points along the western Arab a both
in Roman and Byzantine times proves the assumption that a road must have
linked them and served as latent connection between each, a fact formerly
doubted by Rothenberg (1971, 215). 22
From the more northern and western regions, we would like to mention
two quadrangular eastella: (1) at H. Rahaba, 55 x 33 m, on the Mampsis-
Oboda road (Cohen 1975, 33) and (2) at Mezad Yirqeam (Cohen 1972, 35)
between Mampsis and Hazeva, 23 x 24 m. Both werel)f the courtyard type
(Fig. 56.5).
It will be remembered that the biblical Judean prototypes for the Negebite
Roman fortifications belong to two sub-types: those with protruding towers
and those without. Elsewhere, the chain of similar buildings has been followed
up by the writer from the First Temple period down to the eve of the Roman
conquest (Gichon 1968, Text I, 94-114; Text II, 332; 1967, 180ff., 190ff.;
1972, 44). Further links have meanwhile come to light. However, we shall
contain ourselves with Negev's excavation in the castra on the outskirts of
Oboda. According to him, contrary to my opinion, it never even passed into
Roman hands, but existed only during the 2nd half of the 1st century. Such

852
 0 2Qkm
! lllll!!l!l 1
- J

Fig. 56.5 The Limes Palestinae from Diocletian to the end of the Byzantine
period.

853
is his interpretation of the pottery. Nabatean ostraca that mention camel
and donkey trains and coins prove to everybody's satisfaction the basically
Nabatean character of this castrum. Negev was able to correct Musil's plan
of 1904, which had round-corner and horseshoe-shaped intermediate towers,
to include only square towers. The fact does, however, stand out that we
have now a reasonably well dated fortress that con:bines the local character-
istics of protruding towers and rooms leaning into the curtain walls with
strigae and camp roads (Negev 1975, 49; 1976b, 56; cf. Gichon 1972, 48ff. ).
Add to this the evidence of the early Roman guadriburgus of probably
Nabatean origin from Hazeva (see above) and the existence of the 2nd century
quadriburgus of En Yahav (see above) that did not even survive into the Byzan-
tine period, and the ~vidence for the early adoption of this pre-Roman type
by the imperial engineers is to my mind conclusive. A further example of
the Roman application of the outjutting square towers occurs at the quadri-
burgus Mezad Maagora (see above). There a Roman structure with protruding
towers was superimposed upon a square, tower less Israelite fort of the court-
yard type.
To this should be added our evidence from Mezad Tamar - Tamara.
The towers there are leaning onto the walls, but are differently dressed,
and have much deeper foundations penetrating the early Nabatean layer.
They are however prior to the 3rd century refurbishing. So far, the best
explanation we do have to offer is that these towers were added to the pre-
viously towerless structure, either in a very late Nabatean phase, shortly
before the Roman take-over, or preferably soon after that (Gichon 1976, 89f. _,
fig. 15; 1977, 448, fig. 2, taf. 42.1-2, 46.1; EAEHL IV, 1194).
Of great interest is the hook- or gamma-shaped building discovered by
Z. Meshel near the Sirpad (1975, 30) water-holes on the western fringes of
the Negev. This 11. 2 m long building has the shape of similar structures
proposed by the present writer to have served as guard stations of mounted
troops (1967, 182; 1972, 56££.). This pattern also goes back in the Negev
to Iron Age prototypes as proven by Aharoni and others (Aharoni et al. 1960,
103). However, there is no proof as yet for its adoption by the Romans be-
fore the Byzantine period, although the writer did discover a similar structure
containing abundant Herodian pottery near the Dead Sea shore (Gichon 1968,
Pottery II, 4, 143; Aharoni and Rothenberg 1960, 25, fig. 4). The Sirpad
guard-post contained only Byzantine pottery. Its interest is that it· proves .
that in the "outer Limes" too, Roman regular or semi-regular outposts were
permanently posted wherever water could be obtained. These regions have
been less surveyed than others. The Sirpad post could however not have
been an isolated phenomenon.
A peculiar feature of the castella of Rahaba (Roman- Byzan-tine), Ylrqeam
(late Raman-Byzantine) and Hazaza (2nd-4th century) is a partition-wall that
runs through the central courtyard and divides the buildings into two, with
either no direct communication at all or a single narrow passage only. 23
The same arrangement has been observed at Uza. There, the existence of
rooms leaning to either side of the partition and the towers at both ends have
been taken to mean that this former 1st Temple period and -later Herodian
stronghold had been subsequently enlarged. While I myself have not examined

854
this point particularly, I.. Roll and B. Isaac have revisited this site and given
this feature special attention. Their im~ression is that there are no structural
proofs for an addition, but the contrary. 4
It might be that these partitions marked off a military from a civilian
part in these edifices. Elsewhere we have argued that the Negev castella
and burgi doubled as stations for the cursus publicus and for civilian traffic,
which was a main source of revenue for the government, and carried commo-
dities deemed essential for t\he imperial economy. The rabbinical Mishnaic
simile of the caravan taking~overnight shelter inside a desert fort would thus
be best explained (Gichon 1974b, 535).
A peculiar ground plan is that of the castellum parvulum at Maale Zin,
north of Oboda, which is rectangular with only two corner towers that pro-
trude in one direction only and create a kind of propugnaculum for the gate. 25
There exists only one exact parallel known to me, coming also from the Limes
Palaestinae: Mezad Hatrurim, north of En Boqeq. 26 Surface finds there
were inconclusive. Maale Zin has proved upon excavation by R. Cohen to
be of 6th-7th century date.
To terminate this necessarily selective survey, mention must be made
of the work of Harding (1952, 36; 1953, 8) 9 A. Jamme (1959, 150; 1967),
Milik, Ryckmans (1956, 5) and Winnet (1959, 146; 1973, 54), as well as
that of A. Negev (1967, 253; 1977) on the Saffaitic, Thamudian and Nabatean
inscriptions of Arabia, Palestine and Sinai. To draw the conclusions, a
separate paper is ne.eded.
Let us chart the general picture emerging. The main conclusion is that
the Arab tribes appear as the main and constant threat on our limes. The
Thamudians seem to have been the most active on the Negebite and southern
Trans-Jordan borders. Scholarly consensus places their "graffiti"-mostly
short, laconic blessings, requests or commemorative inscriptions-into the
1st-4th centuries. The contemporary Saffaitic graffiti from the areas north
of Thamudian settlements are a little more explicit, and mention wars together
with or against Rome (CIS 1292, 4438, 1952, 3064, 3688, 3721, etc.).
Neither can, however, be dated accurately. Winnet, who in the past
differentiated between four different periods of Thamudic inscriptions (A - D)
according to palaeographic and textual divergencies (Winnet 1937) has lately
come to the conclusion that those differences are to be interpreted as regional
characteristics (Winnet and Read 1970, 69-70).
These difficulties in dating makes it also impossible to accept without
reservation A. Negev's view that the destruction level of "mid Nabatean"
settlement at Oboda and related phenomena in other Negebite towns are to be
attributed to a Thamudian and/or Saffaitic invasion before the great enterprise
or reconstruction of the Nab ate an realm by Rabel II (after 7 0 A. D.) (Negev
1969, 12). With no exact archaeological stratification published for any of
these sites to relate to and correlate with the evidence marshalled from
buildings, inscriptions and small finds, the undated graffiti are just not con-
clusive.
There is, however, no doubt ~h?.t destruction and desertion of sites in
the Negev were the outcome of Thamudian and Saffaitic incursions. Their

855
constant threat made the permanent manning of the Limes a necessity from
the first appearance of these tribes on the fringes of southern Judaea, i.e. ·
at latest from the Flavian period onwards.
Recent work on the famous Rawwafa inscription has albeit finally proved
that by the days of Marcus Aurelius the Thamudian "ethnos" was already under
some kind of Roman tutelage27 and most probably employed to ward off the
threats of other tribes, such as those anonymous "Arabs" mentioned as having
"devastated the land" in a Sinaitic graffito from 189 A.D. (Euting 1891, I,
no. 463).
From scattered mentions upon tombstones and from other inscriptions it
seems that before the second half of the 2nd century it was Rome's policy to
try and enlist single tribal sheikhs under such titles as " cpuA.apxot; ",
" cr'tpa'tTJYO<; ", " e:-&\Japxoc;; ", " cru\lcS uo~.oc; " and the like, rather than to try
and pact with paramoun~ ·chlefs that ruled large tribal confederations (Devresse
1942, 263-5; Peters 197 8, 323; Kammerer 1929, 330ff. ).
It may be no coincidence that this policy does finally change during the
tetrarchy, which in all fields aimed at streamlining the government apparatus
through centralisation. According to the Nemara inscription, Imru al Qais
of Hira emerges as "King of all Arabs" in about 300 A.n.28 This development
was hastened by the disappear.ance of Palmyra, which had formerly policed for
Rome directly or indirectly much of the Arabian desert borders.
Yet already Imru al Qais' case proves that the more influential the great
tribal heads became,_ the more independent became their policy, either by
leaning over towards the Sassanides or by veering indep~ndently between the
two great powers.
Thus the supreme tribal heads such as the 6th century Ghassanides either
would not or could not always stem the tides of tribal incursions. This the
more so, since Rome never relegated full authroity to them, but in the im-
mediate belt surrounding the outer fringe of the permanent fortifications,
phylarchs were directly responsible to the dux limitis. These introduced a
further factor that was able to influence developments. 29 The following years
have been recorded as years of border unrest, and war from the 4th century
onwards: 340-42; 353; 364-78; 472/3; 491-518 (not all these years, but
certainly 502; 525-28; 532 (? ); 533; 582/3; 629 /30-first Muhammadan
attempt at conquest (Gichon, forthcoming).
This list is of course very incomplete: it suffices, however, to illustrate
the necessity for the constant watch on the Palestinian marches.
A survey of the Limes is incomplete without recalling the many agricultural
sites discovered during the period under consideration mostly by the same
scholars mentioned above. Barring the towns, practically no excavations
were carried out, and some at least of the so-called Byzantine sites may have
to be dated to the Severan period.
Moreover, only some of the regions have been comprehensively surveyed,
so that no true picture for all of the Negev may be presented. Those parts
that have had systematic attention may; however, furnish a good example
for all those regions where the then more abundant winter :rains enabled major

856
agricultural efforts. Such is the region between Oboda and Yeruham, where
over 30 "Byzantine villages" have been discovered in an area of 100 km2
(Cohen 1965, 20; 1966a, 30; 1966b, 23; 1967, 37; 1969, 33-4). All are
conglomerations of buildings guarded by towers in the immediate surroundings
and within their midst. Partly they consist of structures of a stereotype
pattern that gives the impression of being erected by public rather than private
initiative. In some instances single-room dwellings, all of the same size
4 x 4 m, highten this impression. 30 One wonders if the settlers inhabiting
these basically lived in tents erected beside the stone houses that served
either as stores or winter quarters. A similar phenomenon is observed in
the present-day Negev. Some of these settlements will have belonged to the
limitanei, others to federate tribes that were becoming sedentary.
A commercial and administrative (?) centre for that region, which inclu-
ded a caravanserai and a monastery, was discovered at Yeruham. Excavations
prove its floruit for the 5-6th centuries (Cohen 1967a, 123-4).
Since the work of Alt and Avi-Yonah, the northern flank of the Limes
Palaestinae and its hinterland east of Gaza have been taken to be main food-
producing areas, the surplus of which might have been partly used to feed
the frontier garrisons. These tasks could feasibly have been alloted in the
Byzantine periods to the two saltus situated here, 31 the Saltus Constantiniaces
and the Saltus Geraritike. Excavation in the mounds of Tel Gamma (Yemme),
Shera (Sheria) and He sa (Hessi),. situated in these parts, prove that from the
Persian to possibly the eve of the Roman period these sites housed large grain
stores. 32 The public character of those at Tel el Farcah have been proved
by aramaic ostraca of the Persian period (Naveh 1972, 184-9). Here .as else-
where emerges the continuity from long before the Roman conquest.

NOTES

1. For En Boqeq, cf. EAEHL II, s. v. For Tamara, cf. EAEHL IV, ~·
For Tsafit, cf. Gichon 1974a, 16-40. Since it is supposed that students
of the limites imperii are aware of the author's research and that of his
associates, mention of their work will be kept to a minimum. For a .
concise Hebrew summing up of 110 sites investigated by the author, cf.
Gichon 1975, 149-166.
2. For Oboda, (= Eboda ~ Ptol. V.16.4) cf. EAEHL II, s.v. and biblio-
graphy there; for later work, see below, and Negev 1977b, 68. For
Elusa, cf. Negev 1976b, 89-95; 1974, 87-90. For earlier work there,
cf. EAEHL II, s.v. For Mampsis, cf. EAEHL Ill, ~·and bibliography
there. For Subeita (= Sobota, cf. EAEHL IV, ~··and bibliography).
Negev's Nabatean researches have furnished most valuable material
for the research into the transition of eastern Palestine (Nabatea) into
the Roman province of Arabia, as well as for its relationship with Judaea
before and after the take-over (cf. inter alia Negev 1969, 1..:.15, and, for
a discussion of some aspects, cf. Graf 1978).
3. Rothenber~ 1967b with bibli_ography; 1970, 4-29 and bibliography. The
scop~ of th~s pape: does not permit an investigation of Sinaitic problems
despite th~1r beanng on the Limes Palestinae, nor into Rothenberg's
metallurg1ea,l discoveries which prove that the Limes was not only of
8ti7
 military but also of economic importance to the Roman Empire; cf.
Rothenberg 1962, 5; 1972, 208ff. For Roman researches, cf. also
Rothenberg 1971.
4. Arabia too cannot be covered in these pages, although the above research
and also articles by scholars such as Bowersock and Speidel on that
province's garrisons are a necessary complement to investigation of the
Limes Palestinae.
5. Arad is one more example of the fact that, in the Negev, sedentary life
continued well into the Umayyad period, and that while the population
continued to inhabit some sites, such as Subeita and Nessana, Muhami:nadan
Arabs took over others, especially those of purely military character.
6. Aharoni 1973, 7. No details have been published as yet, apart from the
Hellenistic temple, cf. Aharoni 197 5, 163.
7. The final report on En Boqeq will be published soon. Cf. meanwhile,
Gichon 1971, 404-6; On Nebo, cf. for example, Schneider 1950.
8. As at En Boqeq, cf. Gichon 1971, 396; and at Tamara, Gichon 1976, 85.
9. For reports, see issues of Hadashot Archeologiyot from 1962 onwards.
10. Cf. Gichon 1979, 10-13. The final report, with comprehensive plans
and sections, is scheduled to be finished by February 1980.
11. Cf. Kochavi 1967, 272-3; EAEHL s. v. Malhata, Tel; Kochavi 1970, 24.
12. Cf. above, n. 2. " ... no gap between the late Nabatean and the late
Roman periods" (Negev 1967a, 51).
13. Cf. Cohen 1972, 36-8. A plan of Hazeva (= 'Ein Hash = Hrejbet Hosob)
was drawn by Musil before the Roman castellum was incorporated into
the Turco-British police station (Musil 1907, II, 208, fig. 145). The
connection between the towers and the curtain-walls is incorrectly ren-
dered. Rothenberg 1971 must be altered accordingly.
14. Excavated by Y. Porat and A. Joseph (1977 and 197 8). The nearby farm
or guard houses (fig. 56. 3) we!:'e pu.~lished by Joseph (1978).
15. The author's work is incorporated in his Ph. D. thesis (Gichon 1968).
Work carried out by I. Roll, B. Isaac and the above, in the framework
of the Israel Milestone Commission, including the survey of the Deragot
and Scorpion passes, is being processed for publication. Current in-
formation may be obtained from the Commission's secretary, E. Shenhav,
Division of Archaeology Dept. of Classical Archaeology, Tel Aviv
University.
16. Cohen, Meshel and Tasfrir 1965, 24£. This fact is not mentioned in
· Meshel and Tsafrir 1974; 1975.
17. For Tamara, cf. EAEHL IV, 90; for Tsafit, Gichon 1974a, 23.
18. For Nessana, cf. EAEHL Ill, s. v.; Kraemer 1958, 41, pap. no. 15
dated 30.5.512:&pl~~o~ ~wv xa~ocrLw~Evwv 9Eo6ocrLaxwv
(The Numerus of the most loyal Theodosians").
19. Excavations at Yotvata have been carried out intermittently since 1974
_by Z. Meshel and B. Sa ss (1974, 27 3). See also reports in Hadashot
Archaeologiyot 1975-78. The· site includes a Bronze Age settlement, an
Iron Age fortress, a Tetrarchic (?) quadriburgus (?) and a late-Byzantine
858
 or early-Arab fort or Han, in which Hellenistic and Nabatean pottery was
found by Rothenberg (1967, 144, site no. 65) prior to Meshel 's soundings
which prove the early origin of this structure, or its substructure.
Early Roman pottery was found in different places, including the Iron Age
fortress hill and "tower burials". The absence of sufficiently documented
excavation reports is tantalising, but the continuity of settlement on the
site, which since Alt (1935, 24-7, 30) is identified with the Ad DIANAM
of the Tabula Peutingeriana (cf. Avi-Yonah 1976, 26) seems well established.
20. The farming enterprise started either in the late Nabatean or early
Roman period (Trajan-Hadrian) since material from the Moyet Awad
fortress was used in its construction, cf. Gichon 196 8, Text I, 185£.;
Pottery II, 70-2c. In 1970 I. Roll. reported surface finds of Byzantine
coins, including one large medallion, possibly Justinian. The site was
revisited in 197 3 by Z. Meshel, who came to the same conclusions (197 3,
39).
21. Rothenberg 1971, 215, fig. 103. Surface survey from 1962 by the present
author has shown that clear pre-Byzantine material is also present on
the site.
22. 1971, 215. To the sites mentioned in this paper more must be added,
cf. Rothenberg 1967 a, 324ff.
23. No plans of the above castella have been published; the arrangement
may be visualised as being like that at Uza (cf. Meshel 1975, 30).
24. For the ground-plan of Uza, cf. Aharoni 1967, 4, fig. 1. 2.
25. Cf. Cohen 1971b, 37-8. Area 15.4 x 12.4 m. Size of towers 3 x 3 m.
26. Gichon 1975, 156, fig. 9. First surveyed by Frank (1934, 200). Nearest
in shape are Harbawa and Mqehil, cf. Poidebard 1934, 55, pl. LX-XI.
27. Summed up by Graf 197 8, 9-10, with reference to J. T. Milik, G. W.
Bower sock (197 5, 521), etc. For his own conclusions, cf. Graf 197 8, 15.
28. Cf. Raswan 1964, no. 1. For more recent discussions, cf. Altheim and
Stiehl 1965, 3204; von Wissmann 1960, 885 with relevant bibliography;
Bower sock 197 5, 521-3.
29. For the research into background history of the Arabs versus Rome and
Persia, the bibliography accompanying Graf's study (1978) provides·an
excellent base.
30. Cohen 1966a, 30. Other settlements with building of a very stereotyped
character include the mining or quarrying settlement not far from Elath
at Horvat Bodeda, excavated by Z. Meshel and B. Sass (1974, 284-5).
It could have been a penal colony like that at Phainon to which, according
to Eusebius, De Mart. 7. 2, the Christian martyrs were banished by
Arbanus the praeses Palaestinae. The little settlement partly excavated
by P. Bar Adon near En Gedi may also have been a forced-labour camp.
Apart from the similarity of each single room-unit, there is no proof for
his proposal that this place should be identified as a monastic establish-
ment (1968, 18).

859
31. Sites from the area belonging to the saltus or their vicinity, mentioned
in Hadashot Archeologioyot,include (inter alia) Horvat Gerarit (synagogue
floor.), Kissufim (church floor), and The Great Winepress north of
Zeelim; cf. Bakar and Araq 1973, 33.
32. For Tel Gamma, cf. Van Beek 1972, 245; for Tel Shera, cf. Hadashot
Archeologiyot xlviii-ix (Jan. 1974), 78£.; liii (Jan. 1975), for Tel el
Hesi, cf. Hadashot Archeologiyot xi (Oct. 1971), 16; Worrell and Toombs
1971, 232.

BIBLIOGRAPHY

Aharoni, Y., 1967. 'Forerunners of the Limes: Iron Age Fortresses in the
Negev',Israel Explor. J. xvii, 1-17.
Aharoni, Y. (ed. ), 1973. Beer Sheeba I, Tel Aviv.
Aharoni, Y., 1975. 'Excavation at Tel Beer Sheeba, 1973-4', Tel Aviv ii,
163-5.
Aharoni, Y. and Amiran, R., 1964. 'The Excavation at Tel Arad - Preliminary
Report on the First Season 1962', Israel Explor. J. xiv, 131-47.
Aharoni, Y. and Rothenberg, B., 1960. In the Footsteps of Kings and Rebels,
Tel Aviv (Hebrew).
Aharoni, Y. et al., 1960. 'The Ancient Desert Agriculture of the Negev, V',
Israel Explor. J. x, 23-26, 97-111.
Alt, A., 1921. Die Griechische Inschriften der PaUlstina Tertia, Berlin.
Alt, A., 1935. 'Aus der 'Arabah II', Zeitschrift der Deutschen PaUlstina
Vereins 1viii, 24££.
Altheim, F. and Stiehl, R., 1965. Die Araber in der Alten Welt, II, Berlin.
Avi-Yonah, M., 1954. The Madaba Mosaic Map, Jerusalem.
Avi-Yonah, M., 1976. Gazetteer of Roman Palestine, Jerusalem.
Bakar, H. and Araq, H., 1973. Hadashot Archeologiyot xlv (Jan. 1973), 33.
Bar Adon, P., 1968. Hadashot Archeologiyot xxviii (July 1968), 18.
Biran, A. and Cohen, R.' 1976. 'Aroer, 1976 '' Israel Explor. J. xxvi,
139-40.
Biran, A. and Cohen, R.' 1977. 'Aroer, 1977 '' Israel Explor. J. xxvii,
250-1.
Biran, A. and Cohen, R.' 197 8. 'Aroer, 197 8'' Israel Explor. J. xxviii,
197-9.

Bowersock, G. W ., 1975. 'The Greek-Nabatean Bilingual Inscription at

Ruwwafa, Saudi Arabia' in Cambier, G. (ed. ), Le mond grec: Hommages
a Claire Pr~aux, Bruxelles, 513-522.
CIS Corpus Inscriptionum Siniticarum.

860
Cohen, R.' 1965. Hadashot Archeologiyot xv (July 1965), 20.
Cohen, R., 1966a. Hadashot Archeologiyot xvii (Jan. 1966), 30.
Cohen, R.' 1966b. Hadashot Archeologiyot xviii-xix (July 1966), 23.
Cohen, R.' 1967a. 'Mesad Yeruhan',Israel Explor. J. xvii, 123-4.
Cohen, R., 1967b. Hadashot Archeologiyot xxiv (Oct. 1967), 37.
Cohen, R.' 1969. Hadashot Archeologiyot xviii-xix (Jan. 1969), 33-4.
Cohen, R.' 1971a. Hadashot Archeologiyot xxxix (Feb. 1971), 27-8.
Cohen, R.' 1971b. Hadashot Archeologiyot xl (Oct. 1971), 37:8.
Cohen, R.' 1972. Hadashot ArcheologiYot xliv (Oct. 1972), 36-8.
Cohen, R.' 1975. Hadashot Archeologiyot liii (Jan. 1975), 33.
Cohen, R.' 1976. 'Excavation at Horvat Haluqim', Atiqot xi, 47-50.
Cohen, R. andMeshel, Z., 1970. Hadashot Archeologiyot xxxvi (Oct. 197 0),
25.
Cohen, R., Meshel, Z. and Tsafrir, Y., 1965. Hadashot Archeologiyot xvi
(Oct. 1965), 24-5.
Devresse, R., 1942. 'Arabes-Perses et Perses-Romains, Lakhmides et
Ghassanides' Vivre et Penser 1942, 263-5.
EAEHL. Encyclopaedia of Archaeological Excavations in the Holy Land, 1975,
Jerusalem.
Euting, J., 1891. Sinaitische Inschriften, Berlin.
Frank, F., 1934. 'Aus der 'Araba I', Zeitschrift der Deutschen PaHlstina
Vereins lvii, 191-280.
Fritz, V., 1973. 'The Roman Fortress', in Aharoni 1973, 83-89.
Gichon, M., 1963. The Limes Palestinae in the Time of Diocletian, in Elath:
the 18th Archaeological Convention, Jerusalem, 88.
Gichon, M., 1967. 'The Origin of the Limes Palaestinae', in studien zu den
MiliUtrgrenzen Roms, Koln-Graz, 175-193.
Gichon, M., 1968. The Limes in the Negev from its Foundation to Diocletian
Times (Ph.D. thesis) Jerusalem.
Gichon, M., 1971. 'Das Kastell En Boqeq ', Bonner Jahrbtt.cher clxxi, 386-406.
Gichon, M., 1972. 'The Plan of a Roman Camp depicted upon a Lamp from
Samaria', Palestine Explor. Quart. civ, 38-58.
Gichon, M., 1974a. 'Migdal Tsafit, a burgus in the Negev (Israel)', Saalburg
Jahrbuch xxxi, 16-40.
Gichon, M., 1974b. 'Towers on the Limes Palaestinae', in Pippidi, D. M.
(ed. ), Actes du IXe congr~s international d'etudes sur les fronti~res
romaines, Bucuresti-KBln, 513-544.

861
Gichon, M., 1975. Eretz Israel xii, 149-66.
Gichon, M., 1976. 'Excavation at Mezad Tamar-"Tamara"', Saalburg Jahr-
buch xxxiii, 80-94.
Gichon, M., 1977. 'Mezad Tamar-"Tamara". Vorbericht der Grabungen
1973-1974 ',in Studien zu den Militfirgrenzen Roms II, Koln-Bonn, 445-452.
Gichon, M., 1979. Hadashot Archeologiyot lxix-lxxi (July 1979), 10-13.
Gichon, M. , forthcoming. 'The Military and Political Developments on the
Limes P alestinae: Table of Events'.
Graf, D., 197 8. 'The Saracens and the Defence of the Arabian Frontier',
Bull. Am. Sch. Or. Research ccxix, 1-26.
Harding, G. L., 1952. 'Epigraphy: the Thamudic and Lihyanite Texts, Bull.
Inst. Arch. x, 36-52.
Harding, G. L., 1953. 'The Cairn of Rani', Ann. Dept. Antiq. Jordan ii,
8-59.
Jamme, A., 1959. 'A Sa.ffaitic Inscription from the Negeb', Atiquot ii, 150-1.
Jamme, A., 1967. Thamudic Studies, Washington.
Joseph, A., 197 8. Nophim xi-xii, 10-14.
Kammerer, A., 1929. Petra et la Nabatene, Paris.
Kindler, A., 1973. 'The Coins', in Aharoni, Y. (ed.), Beer Sheba I, Tel
Aviv, 90-6.
Kochavi, M., 1967. 'Tel Malhata', Israel Explor. J. xvii, 272-3.
Kochavi, M., 1970. Qadmoniot iii, 24.
Krae:rher, C. J., 1958. Excavations at Nessana: Ill, The Non-Literary
Papyri, Princeton.
Meshel, Z., 1973. Hadashot Archeologiyot xlv (Jan. 1973), 39.
Meshel, Z., 1974. History of the Negev in the Time of the Kingdom of Judah
(Ph.D. thesis), Tel Aviv (Hebrew).
Meshel, Z., 1975. Hadashot Archeologiyot liv-v (July 1975), 30.
Meshel, Z. and Tsafrir, Y., 1974. 'The Nabatean Road from 'Avdat to
Sha'ar Ramon: I', Palestine. Explor. Quart. cvi, 103-18.
Meshel, Z. and Tsafrir, Y., 1975. 'The Nabatean Road from 'Avdat to
Sha'ar Ramon: 11', Palestine. Explor. Quart. cvii, 3-21.
Meshel, Z. and Sass, B., 1974. 'Hurvat Bodeda',Israel Explor. J. xxiv,
284-5.
Musil, A., 1907. Arabia Petraea, Wien.
Naveh, J., 1972. Two Aramaic Ostraca from the Persian Period', in
Offenheimer, B. (ed. ), Bible and Jewish History, Tel Aviv, 184-9
(Hebrew).

862
Negev, A., 1966. 'The Date of the Petra-Gaza Road', Palestine Explor.
Quart. xcviii, 89-98.
Negev, A., 1967 a. 'Oboda, Mampsis and Provincia Arabia', Israel Explor.
~· xvii, 46-55.

Negev, A., 1967b. 'New Dated Nabatean Graffiti from the Sinai', Israel
Explor. J. xvii, 253-5,
Negev, A., 1969. 'The Chronology of the Middle Nabatean Period', Palestine
Explor. Quart. ci, 1-15.
Negev, A.' 1974. Hadashot Archeologiyot xlviii-ix (Jan. 197 4), 87-90.
Negev, A.' 197 5. Hadashot Archeologiyot lvi (Oct. 1975), 49.
Negev, A.' 1976a. . Hadashot Archeologiyot lix-lx (Oct. 1976 ), 56.
Negev, A., 1976b. 'Survey and Trial Excavation at Haluza (Elusa), 1973',
Israel Explor. J. xxvi, 89-95.
~--;.__JO;...;;.;;.__;;_..;;..

Negev, A.' 1977a. The Inscription of Wadi Haggag, Sinai (Qedem 6), Jerusalem.
Negev, A.' 1977b. Hadashot ArcheologiYot lxiii-iv (Oct. 1977), 68.
Parker, S. T., 1976. 'Archaeological Survey of the Limes Arabicus-A Pre-
liminary Report', Ann. Dept. Antiq, Jordan xxi, 19-31.
Peters, F. E., 1978. 'Romans and Bedouin in Southern Syria', J. Near East-
ern Studies xxxvii, 315-26.
Poidebard, A., 1934. Le trace de Rome dans le desert de Syrie, Paris.
Porat, Y. and Joseph, A., 1977. Had as hot Archeologiyot lxi-ii, (April 1977 ),
50.
Po;rat, Y. and Joseph, A., 1978. Hadashot Archeologiyot lxvi-ii, (April1978),
54.
Raswan, C. R., 1964. Revue Egyptienned'Archeologie, i.
Rothenberg, B., 1962. 'Ancient Copper Industries in the Western Arabah,'
Palestine Explor. Quart. xciv, 5-71.
Rothenberg, B., 1967a. Survey of the Eloth Area, Elath (Hebrew).
Rothenberg, B., 1967b. Negev: Archaeology in the Negev and the Arabah,
Ramat Gan (Hebrew).
Rothenberg, B., 1970. 'An Archaeological Survey of the Sinai', Palestine
Explor. Quart. cii, 4-29.
Rothenberg, B., 1971. 'The Arabah in Roman and Byzantine Times in the
Light of New Research', in Applebaum, S. (ed.) Roman Frontier Studies
1967, Tel Aviv, 211-223.
Rothenberg, B., 1972. Timna, Valley of the Biblical Copper Mines, London.
Rothenberg, B., 1973. Hadashot Archeologiyot xlv (Jan. 1973), 41.
Ryckmans, J., 1956. 'Aspects nouveaux du probleme thamoudeen', Studia
Islamic a v, 5-17.

863
Schneider, H., 1950. The Memorial of Moses on Mount Nebo, Ill. The
Pottery~ Jerusalem.

Tsairir, Y., 1975. Hadashot Archeologiyot lvi (Oct. 1975), 42-4.

Tsafrir, Y., 1976. Hadashot Archeologiyot lix-lx (Oct. 1976 ), 59-60.
VanBeek, G. W., 1972. 'Tel Gamma', IsraelExplor. J. xxii, 245-6.
Winnet, F. V., 1937. 'A Study of the Lihy anite and Thamudic Inscriptions',
Univ. Toronto Oriental Stud. iii.
Winnet, F. V., 1959. 'Thamudic Inscriptions from the Negev'. Atiqot ii,
146-9.
Winnet, F. V., 197 3. 'The Revolt of Damasi, Safaitic and Nab ate an Evidence',
Bull. Am. Sch. Oriental Res. ccxi, 54-7.
Winnet, F. V. and Reed, W. C., 1970. 'Arabian Inscriptions of Ancient Re-
cords from North Arabia', Univ. Toronto Near and Middle East Series vi,
69-70.
Wissmann, H. von., 1960. ''Badw", in Encyclopaedia of Islam I.
Worrell, J. E. and Toombs, L. E., 1971. 'Tell El-Hesi', Israel Explor. J.
xxi, 232-3.

864
 57. TOWARDS A HISTORY OF THE LIMES ARABICUS

S. Thomas Parker

Previous meetings of this Congress have shown little interest in the Limes
Arabicus, the £· 360 kilometre frontier extending from southern Syria to the
Red Sea. This neglect is certainly understandable; until recently no survey
of the entire frontier had been attempted and not a single military site of the
frontier had been excavated. Our knowledge of the limes was based largely
on several surveys of the late 19th and early 20th centuries (Brtlnnow and von
Domaszewski 1904-09; Musil 1907; Butler 1909). These scholars provided
gqod descriptions of the extant forts and the road system and published some
excellent plans. There was little evidence, however, regarding the historical
evolution of the frontier from its creation by Trajan in 106 to the Arab con-
quest in the early 7th century. A handful of building inscriptions and a few
references in the literary sources provided most of our historical information
(Bowersock 1976, 219-229).
This situation, however, is now rapidly changing. In 1976 the first com-
prehensive survey of_ the frontier was conducted under my direction. Most of
the major military sites of the limes (some 41 in all) were surveyed. A topo-
graphical analysis was made of each site to determine its function within the
system; a surface ceramic sample was collected at each site. The samples
of pottery furnished evidence concerning the occupational history of each site'-
andthus for the history of the limes as a whole. The dating of the pottery was
based oc J. A. Sauer's refined ceramic chronology for Palestine and Trans-
jordan, developed from the Heshbon excavations (Sauer 1973, 1-5). Several
dated building inscriptions and a handful of surface coins served as a control
ohthe ceramic evidence (Parker 1976, 19-31). The purpose of this paper is
to present a brief outline of the history of the Limes Arabicus in the light of
this recent research, I
The primary purpose of the limes was to control the incursions of the
nomadic Saracen tribes of north-western Arabia. The limes protected a
thickly inhabited region along the fringes of the desert. Local Roman forces
were also responsible for the protection of the lucrative caravan traffic pas-
sing into the Empire from India and southern Arabia. The threat posed by the
Saracen tribes was ordinarily not conquest and occupation of Roman territory
but endemic raiding of caravans, farms, villages, and towns.
The Trajanic annexation of the client state of Nabatea in 106 resulted in
the creation of the new province of Arabia. A major military highway, the
via nova Traiana, was constructed between 111 and 114; it extended from the
provincial capital of Bostra in the north to Aqaba (ancient Aila) on an arm of
the Red Sea. A significant number of Nabatean forts and watchtowers were
taken over by the Romans. It appears that these former Nabatean posts formed

865
the initial framework of the Limes Arabicus (Parker 1976, 26). It is important
to point out, however, that the Nabatean defensive system was not taken over
in toto by the Romans. For example, a number of Nabatean watchtowers in
the Hisma, east of the southern end oi the via ~· were not occupied by the
Romans after the annexation (Graf 1978b). The Romans also built several new
fortifications about the time of the annexation, such as Qasr el Kithara and el
Quweira in the extreme south. In general, it seems that the Romans developed
a linear type of defence, with most forts located on or near the via ~· A
glance at the maps of the limes during the Antonine era will show this linear
system of defence (see figs. 57.1 and 57. 2). The only area in which a signi-

--
ficant number of forts was located east of the via nova was the north. The forts
in this sector, such as Qasr Usaikhin and Qasr el Hall ab at, controlled ace ss
through the Wadi Sirhan, a natural migration route between southern Syria
and the interior of the Arabian peninsula (Glueck 1944, 7-17). In the south,
there is evidence of treaty arrangements with nomadic tribes as early as the
reign of Marcus Aurelius. These tribes may have assisted in the defence of
the frontier by this period (Graf 197 Sa, 1-26).
A significant strengthening of the limes occurred under the Severans (193-
235). The primary focus of this activity centred on the north-western outlet
of the Wadi Sirhan (see fig. 57.3). Several forts were constructed or rebuilt
in this period, including Qasr el Hall ab at (A. D. 213; Littmann, Magie and
stuart 1910, no. 17) and Qasr el 'Uweinid (A.D. 198-211; Parker and Mc-
Dermott 1978, 61-66). Surface pottery also suggests occupation at Qasr
Usaikhin (Parker 1976,_23). All these posts guarded the Wadi Sirhan. An
inscription of 201 from Motha, an important town in the north, may indicate
the employment of Gothic gentiles in the defence of this sector (Speidel 1977,
712-716). Farther south, at least one fort (Jurf-ed-Darawish) was apparently
built in the Severan period. The earliest pottery recovered from this post
dates to the Severan era. The ceramic samples suggest that all the other
forts of the Antonine period in the south remained occupied under the Severans
(see fig. 57.4: Parker 1976, 24-25).
Numerous milestones of Severan date may be cited to supplement the buil-
ding inscriptions and ceramic evidence. In fact, more milestones are extant
in Arabia from the Severan era than from any other comparable period (Thomsen
1917, 90-91). These milestones indicate the repair of the entire length of the
via~ plus several branch roads. 2

What can explain this burst of activity along the limes in the Severan period?
First, it must be remembered that the Severans pursued an aggressive, ex-
pansionst policy along the eastern frontier. Septimius fought two major cam-
paigns in Mesopotamia between 194 and 199 and annexed the north-western
portion of Mesopotamia. Caracalla began further military operations against
Parthia but these were cut short by his assassination in 217. It seems evident
that the Severan repairs of the road system were made, at least in part, to
facilitate the movement of troops from Arabia and Palestine northward for the
Mesopotamian campaigns. The large number of fortifications constructed in
Arabia in this period, however, suggest that considerations of local security
were also important. The blockage of the Wadi Sirhan may therefore indicate
that nomadic tribes were attempting to push into southern Syria from the Ara-
bian peninsula (Parker 1981 ).

866
 X SHUHAR
26

MA'AN
0

XSHUOAYYID
• QII~ANA JORDAN
3S

ANTONINE LIMES

106-192

llM£5 ARAB! CUS SCU1HERN REGION

KEY: .
lEGIONARY CAMP(CASTRA) •
AUXlllARY FORT (CAS1£LLUt--1) •
WATCHTOWER X
C1TY 0
CARAVANSERAI A
SCALE 1 :7SO.OCO
0 5 10 20 JiJ 40 50 KllOME"TERS
--- -- -- --
_j
SAUDI ARABIA ------
--- S.T. PARKER OR.A.BOELRAZEQ

Fig. 57.2

868
 --
Ou!!ein

• Hallobat
3"

Usaikhin•
JORDAN 4
.Azraq
s
'Uweinid.
6
0 SEVERAN LIMES
JIZA
193-235
...13 Zebib
liMES ARABICUS- NCRTHERN REGION

KEY:
LEGIONARY CAMP (CASTRA l )(
A.UXILIARY FORT
WATCHTOWER
CITY
((J..SiELLUM)
•
X
0
CARAVANSERAI
SCALE 1:150,000 •
0 5 10 20 :m 40 ~0

KILOMETER5

~R 5 I FAROAN

Fig.57.3

869
 X Shuhar
26

MA'AN
0

X.Shudayyid
'Qirana
JORDAN

SEVERAN LIMES
193-235
l
~
liMES ARABlCUS SOUTHERN REGION

KEY:
lEGlONARY CA MP(C,A.SiRA) )(
39 AUXlllARY FORT (CAS1ELLU~)
WATCHTOWER
CllY
•
X
0
CARAVANSERAi A
SCAlE ' :7 50,0 00
0 S 10 20 JO L.O 50 KllCMETERS .
--- -- -- -- 12=~--~===---~==
SAUDI ARABIA ------
--- ST. PARKER OR.ABDELRAlEO

Fig. 57.4

870
 In the mid 3rd century the Limes Arabicus was disrupted by the shock of
the Palmyrene invasion and by the general political chaos of the period.
After the defeat of Palmy ra the Romans were faced with the enormous task
of reconstruction and reorganisation of the Arabian frontier. Aurelian pro-
bably began this work by replenishing the Arabian army with units drawn from
the West (Speidel1977, 717-718). The central figure in this revitalization of
the limes, however, is Diocletian. Building inscriptions indicate that several
new castella were constructed in this period, including Deir el Kahf (A. D.
306; Littmann, Magie and Stuart 1910, no. 228). Qasr Azraq (A.D. 292-
306; IGR 3.1339), and Qasr Bshir (A.D. 306; CIL 3.14149). Milestone
inscriptions indicate considerable work on the road system during the Tetrarchy
(Thomsen 1917, 92-93). The ceramic evidence from the 1976 survey clearly
supports this picture of a substantial military build-up; surface pottery sug-
gests a large increase in the number of occupied forts, from 14 to 30 (Parker
1976, 27, see figs. 57.5 and 57 .6). Most important among these new con-
structions were the two legionary camps of Lejjun, in the central sector,
and Udhruh, in the south. Ceramic evidence also suggests that the castella
of al-Qastal, Khirbet ez-Zona, Qsar eth-Thuraiya, and Khirbet el Fityan
were constructed in the Diocletianic era. It is significant that most of these
new foundations, plus Bshir mentioned above, were erected in the central
sector of the frontier. This sector, roughly the area east of the Dead Sea,
had previously been only lightly defended but was now heavily fortified. All
the fortifications in the central sector were built some distance east of the
via nova. By the 4th century it appears that the Romans had abandoned the
Antonine concept of a linear defence and had shifted to a defence in depth. 3
The Limes Arabicus was now a broad fortified zone, 9_. 20-30 kilometres
in depth.
The reasons for this dramatic build-up are not entirely clear. The Dio-
cletianic policy was to strengthen all the frontiers (Zosimus 2. 34), but it
seem:s probable that the major build-up in Arabia was at least partially a res-
ponse to increasing Saracen pressure. Diocletian himself conducted a cam-
paign against the Saracens in 290, but few details of this have survived (Latin
Panegyrics 11.5.4; cf. Ensslin 1942, 27).
The Limes Arabicus was at its height of effectiveness and strength in the
period from Diocletian to the death of Theodosius II in 450. Considerable
epigraphic· material, some literary references, and archaeological evidence
from the forts of the frontier all suggest substantial and sustained military
activity. By the late 4th century Ammianus describes Arabia as castrisque
oppleta validis et castellis (14. 8.13). The list of the army units in the Notitia
suggests that the Arabian frontier was strongly garrisoned (Oriens 34, 37).
Mile stone inscriptions indicate that the road system was maintained through
most of the 4th century (Thomsen 1917, 92-93). The majority _of the Roman
military building inscriptions date to the period from 305 to 412.4
This picture of a heavily fortified frontier in this period is confirmed by
the ceramic evidence of the survey, which suggests that more forts were occu-
pied during the 4th century than in any other period (Parker 1976, 27-28, 31;
see figs. 57.5 and 57. 6). It seems clear that this substantial effort was a
response to serious Saracen incursions. A major Saracen inva_sion led by
Queen Mavia is attested during the early 370s (~ufinus, H.E. 2.6; cf.

871
 ........ ,"',
' .... ....
.... ,
' ~-
Ba•jq- ... .______ .,*' , , "'
2 ~:-:-::E:-:-l_------ - - - - /
-JIMAL '0 • Kahf
Qu!!ein 1

•
3 Hallo
. bot

Usaikhin.
JORDAN 4
• Azraq
s

LIMES OF
DIOCLETIAN
284-450
LIME 5 AFlABICUS- NOR THE RN REGION

KEY:
lEGIONAAY CAMP (CAST RA) )(
AUXILIARY FORT
WATOiTOWER
CITY
(CASTELLUM)
X
•
CARAVANSERAI
SCALE 1:750,000
....0
0 s 10 10 30 LO
KILOMETERS

~S Abu lulcba

S T" F'ARKER OR 5 I F'A'lR..\~

Fig. 57. 5

872
 30 UDHRUJ?:

MA:AN
0 • • MUTRAB.
~AMMAM33 J~ .

X.
3635
QIRANA JORDAN

LIMES OF
DIOCLETIAN
284-450
liMES ARAB\CUS SOUTHERN REGION.

KEY:
lEGlONARY CAMP(CA51RA) )(
AUX'1L\ARY FORT(CASTELLUM) •
WATCHTOWER X
C1TY 0
CARAVANSERAI A
SCALE l :75C,DC:J
0 S 10 20 30 40 50 KILOMEiERS
IC=~--~==~--==~
ST. PARK!:R OR.ABDElf<AZE~

Fig. 57.6

873
Bowersock 1976, 225-226). Jerome mentions another Saracen invasion in
410 (Epistle 126; cf. Bowersock 1976, 226). The frontier was restored,
however, after both these attacks.
The period after 450 reflects a gradual decay of the Arabian frontier.
Both epigraphic and ceramic evidence suggest that no new forts were built
after the early 5th century. Indeed, most of the fortifications of the Diocletianic
system were abandoned during the 5th and 6th centuries. In the central sec-
tor, heavily fortified in the late 3rd through early 5th centuries, only three
castella show evidence of occupation after the early 6th century. Most of the
forts of the southern sector were also abandoned by£.· 500 (Parker 1976, 28).
Only in the north is there epigraphic evidence of continued military activity:
inscriptions of 491-518 and 529 from Hallabat (Littmann, Magie and Stuart
1910, nos. 20, 18).
This gradual abandonment of the limes was caused by several factors.
The collapse of the Danubian frontier following the disaster of Adrianople in
37 8 resulted in the transfer of troops from Arabia. For example, the eguites
Nona Dalmatae, who built the burgus at Jimal in 371 (cf. n. 4 above), were
subsequently withdrawn to the field army based near Constantinople (N. D.
Or. 5. 37). Mounting problems on the Danube and continuing Sassanid pressure
in Mesopotamia may have caused the emperors to neglect the south-eastern
frontier. The limitanei seem to have declined in both quality and quantity
during the 5th and 6th centuries (Jones 1964, 649-654). · Regional security
was increasingly in the hands of Arab phylarchs and their foederati. Justinian
accelerated this process by further weakening the limitanei ( Procopius,
Anecdota 24.12 ff.) and shifting primary responsibility for defence to the
Ghassanids (Procopius B. P. 1.19. 8-13). The accompanying maps of the
Arabian frontier under Justinian reflect this change (see figs. 57.7 and 57. 8).
No longer was there any continuous limes; only small clusters of forts re-
mained occupied in certain limited areas. Justinian's shift of reliance from
regular Roman forces to Arab foederati was not necessarily a mistake. This
policy probably freed military and financial resources for use in his Persian
campaigns, in the defence of the Danube, and in the reconquest of the West.
In fact, the Ghassanids capably defended the south-eastern frontier of the
Empire for some years. But the short-sighted policies of Justinian's suc-
cessors, who progressively weakened the Ghassanids without any corresponding
strengthening of the old limes, paved the way for the Muslim conquest in the
early 7th century.
The 1976 survey of the Limes Arabicus has established the outlines of the
historical development of this long neglected frontier. Excavation of key sites,
which be·gan with our soundings of the castellum of Jimal in 1977, holds great
promise. Several other projects are now planned, including excavation of the
legionary camp of el Lejjun, scheduled for 1980. Such projects will make
further significant contributions towards a history of the Limes Arabicus.

874
 Sal'lcflod

_
...
--

8 Hallobat
3 .

Usaikhin•
JORDAN 4
.Azraq
5

0 LIMES OF
JIZA
JUSTINIAN
.......... ~Zona 532-636
LIMES ARABiCUS- 1\0RlHERN REGION

KEY:
lfGIONARY CAMP (CASlRA l X
AUXILIARY FORT (CASTELLUt-1) •
WATCH TOWER X
CITY 0
CARAVANSERAI A
SCALE 1:750,000
0 s 10 10 lC 40 50

KILOMET ER S

5 T. F'.I.RKER OR SI 9.R~..t.N

Fig.57.7

875
 30 UDHRUl;I

MA'AN
0

JORDAN

LIMES
JUSTINIAN
OF l"=!:7
532-636
llt--!ES ARASICUS SJU1HERN REGION

••
KEY:
lEGIONARY CAMP(CASTRA)
AUXlllARV FORT (CASTELLUM)
WA1CHTOWER X
CllY 0
CARAVANSERAI A
SCALE 1 :7 SO.O CO
___ 0 5 10
IC=~--~==~--==~
20 )0 l.O SO KllOME1ERS

..... ______ S.l. PARKER

--- DR.ABOELRAIEQ

Fig.57.8

876
 NOTES

1. The Preliminary Report of the survey was published several years ago
(Parker 1976). The Final Report has been completed and will be published
by the American Schools of Oriental Research, which sponsored the pro-
ject. Funding was provided by the Shell Oil Foundation, the U. C. L. A.
Friends of Archaeology, the U.C.L.A. Patent Fund, and the Kyle Kelso
Research Fund. The author is grateful to these organisations for their
support.
2. Thomsen 1917, 90-91, who recorded 12 milestones of Pertinax (193), 19
of Septimius (primarily dated 194-200), 31 of Caracalla (212-214), 18 of
Elagabalus (218-222), and 4 of Severus Alexander (222-235). Several
more have appeared subsequently.
3. For the adoption of this strategy throughout the Empire, cf. Luttwak
1976, 130-190.
4. These include inscriptions from the forts at Deir el Kahf (A.D. 306,
Littmann, Magie and Stuart 1910, no. 228; A.D. 367-375, Littmann,
Magie and Stuart 1910: no. 229), Qasr Bshir (A.D. 306, CIL Ill 14149),
Qasr Azraq (A.D. 326-333, Bowersock 1971, 241), Qasr el Ba'iq (A.D.
412, Littmann, Magie and Stuart 1910, no. 21), and Umm el-Jimal (A.D.
371, Littmann, Magie and Stuart 1913, no. 233; A. D. 412, Littmann,
Magie and Stuart 1913, no. 327, cf. Parker 1980).

BIBLIOGRAPHY

Bowersock, G. W ., 1971. "A Report on Arabia P rovincia", J. Roman Stud .

. lxi, 219-242.

Bowersock, G. W., 1976. "Limes Arabicus", Harvard Stud. Class. Phil.

lxxx, 219-29.
Butler, H. C., 1909. Ancient Architecture in Syria (Publ. of the Princeton
University Archaeological Expedition to Syria in 1904-5. II. A.2: The
Southern Hauran), Leyden.
Brtinnow, R. and Domaszewski, A. von, 1904-9. Die Provincia Arabia, 3
vols., Strassborg.
Ensslin, W., 1942. Zur Ostpolitik des Kaisers Diokletian, Munich.
G[ueck, N., 1944. "Wadi Sirhan in North Arabia", Bull. Am. Sch. Oriental
Res. xcvi, 7-17.
Graf, D. F., 1978a. "Saracens and the Defense of the Arabian Frontier",
Bull. Am. Sch. Oriental Res. ccxxix, 1-26.
Graf, D. F., 1978b. "A Preliminary Report on a Survey of Nabatean-Roman
Military Sites in Southern Jordan", Ann. Dept. Antiq. Jordan xxiii,
forthcoming.
Jones, A. H. M., 1964. The Later Roman Empire, 284-602, 3 vols., Oxford.

877
Littmann, E., Magie, D. andStuart, D. R., 1910. GreekandLatinlnscrip-
tions (Publ. of the Princeton University Archaeological Expedition to
Syria in 1904-5 111. A.2: The Southern Hauran), Leyden.
Littmann, E., Magie, D. and Stuart, D. R., 913. Greek and Latin Inscrip-
tions (Publ. of the Princeton University Archaeological Expedition to
Syria in 1904-5, II.A.3: Umm idj-Djimal), Leyden.
Luttwak, E. N., 1976. The Grand Strategy of the Roman Empire from the
First Centucy A. D. to the Third, Baltimore.
Musil, A., 1907. Arabia Petraea, 2 vols., Vienna.
Parker, S. T., 1976. "Archaeological Survey of the Limes Arabicus: A
Preliminary Report", Ann. Dept. Antiq. Jordan xxi, 19-31.
Parker, S. T., 1980. "The Roman Barracks", in DeVries B. (ed. ), Umm
el-Jimal 1977, Ann. Am. Sch. Oriental. Res. forthcoming.
Parker, S. T., 1981. The Historical Development of the Limes Arabicus,
Cambridge (Massachusetts), forthcoming.
Parker, S. T. and McDermott, P. M. , 1978. "A Military Building Inscrip-
tion from Roman Arabia", Zeitschrift fUr Papyrologie und Epigraphik
xxviii, 61-66.
Sauer, J. A., 1973. Heshbon Pottery 1971 (Andrews University Monographs
7), Berrien Springs (Michigan).
Speidel, M. P., 1977: "Exercitus Arabicus", in Temporini, H. and Haase,
W., (eds. ), Aufstieg und Niedergang der r6mischen Welt, II. viii, 687-730.
Thomsen P., 1917. "Die romischen Meilensteine der Provinzen Syria,
Arabia, und Palaestina", Zeitschrift des Deutschen PaHlstina-Vereins,
xl, 1-103.

878
 58. THE FRONTIER POUCY OF SEPTIMIUS SEVERUS:
NEW EVIDENCE FROM ARABIA

D. L. Kennedy

The imperial frontiers "inherited by any emperor were not those designed
by his predecessor(s). Rather they were arrived at by accident and compromise;
the outcome very often of unfulfilled, incomplete or unsuccessful plans • 1
Given the basic premise that a primary function of any emperor was to ade-
quately and efficiently protect his subjects and their lands, the scope for
improvement-whether voluntarily undertaken or forced upon them by outside
circumstances-was essentially restricted to two choices: on the one hand,
the frontiers could be advanced or retracted to more suitable lines and/or,
on the other, the frontier army could be strengthened. They were not mutually
exclusive. Traditionally Romans resolved frontier problems by war and an-
nexation, and it took a man of the imagination and courage of Hadrian to give
up recent conquests. However, if Hadrian was accused of faint-heartedness,
Trajan, Marcus Aurelius and Septimius, for all that their approach was more
acceptable, did not escape charges of glory-seeking. Despite the apparent
contrast between them, it is im{XJrtant to bear in mind that all were seeking
the same ends-only the means differed. A number of emperors increased
the numerical strength of the army without any major revision of the frontiers
or even a sector of them, but army reforms and increases were an invariable
and complementary part of all major frontier revisions throughout the second
century. In this respect, Septimius Sev~rus is no different from Marcus or
Trajan.
The military policy of Septimius is to some extent overshadowed and
obscured by the circumstances of his accession. Great as were his external
wars they were eclipsed by four years of civil wars waged on a greater scale
and much more immediate to Romans. Likewise his treatment of the army
was seen by contemporaries as securing his position and rewarding the men
who had brought him to power and were needed to maintain the dynasty
[Herodian, III. 4-5]. Several recent studies have clarified and interpreted
Septimius' army reforms and his two great external campaigns in Mesopotamia
and Britain are considerably better understood [Birley, E., 1969; Smith
1972; Birley, A. R., 1971 253ff, 283ff.]. Septimius emerges now as a
traditional military emperor securing the Empire by aggressive expansion
to better lines or into positions of greater strength. His success is, of course,
partly due to his predecessors; after several generations the peoples of
Mesopotamia were more accustomed to the Romans on their fringes and less
likely to react violently to the apparent imposition of alien rule. In the first
century we can witness the peaceful absorption of client-kingdoms within or
along the frontiers; by the end of the second century such petty rulers were
all outside the frontiers but their absorption is merely the other side of the
same coin [Luttwak 1976, passim and esp. 114ff. ].

879
 The character of Septimius' frontier policy can be briefly summarized.
In the East his new province of Mesopotamia both absorbed neighbours who
were becoming romanized and gave Rome a strategically commanding position
permanently overbalancing her Parthian rival. In Britain, too, the probable
intention to annexe southern and central Scotland can be seen in the same
light [Birley, A. R., 1971, 253ff]. Archaeological evidence has revealed
advances along the Dacian frontier under Septimius to more commanding
positions [RE,XIII (1926) 645; Tudor 1977, 401f; Popescu and Popescu
1. 968] and, even better known as the result of intensive archaeological activity
by French and British scholars, in the southward advance in Africa [Salama
1955; Fentress 1979, 114-17; Rebuffat 1972; Birley, A.R., 1971, 216-9;
Goodchild 1954]. There we find new forts occupying more advanced positions
in the Mauretanias, Numidia and Tripolitania. Neighbouring nomadic tribes
were brought within the Empire and strategic points were seized on mountain
passes or oases. It is the latter which are of most relevance to the situation
in Arabia because of the many similarities of terrain and of the neighbouring
peoples. In Tripolitania in particular the Septimian foundations of Bu-Ngem,
Gheria el-Garbia and Ghadames were thrust far out into the desert onto major
oases [Goodchild 1954; Rebuffat 1972]. It is my contention that a similar
pattern is now beginning to emerge in Arabia.
I am only one of a number of people who have been impelled to undertake
research on Arabia as a result of the paper of Prof. Bower sock in the Journal
of Roman Studies for 1971. Bower sock himself noted that the Arabian limes
had attracted little attention at limes congresses: none in fact until this one
at which we have 2 papers. Furthermore we can now look forward, to the
excavations soon to be carried out on the great legionary fortresses of Lejjun
and Udruh. The eastern fringes of Arabia, like the south of the African pro-
vinces, are relatively inhospitable: desert with a scattering of oases and
water-holes delineating the possible routes. In north-eastern Jordan, Trajan's
great highway running south from Bostra towards Philadelphia and then on to
the Red Sea, forms a rough demarcation line between the cultivated river
valleys to the west (£!_. Gerasa = "Antioch on the Golden River") and the desert
and pre-desert to the east. Deserts are of course not dead and (in northern
Jordan at least) it is clear that there was considerable human activity to the
east of the via nova Traiana. Qasr el-Hallabat, a few miles east of the road
is a Nabataean and Romano-Byzantine site on the marginal land between desert
and sown [FAES, II.A. 2, 70-7; II. A.2.App.xvi-xviii]. The modern village
relies on water carted in from a well drilled a mile or more to the east, but
there is ample evidence for ancient military and civilian settlement clearly
sustained by the extensive and elaborate water collection and preservation
system of cisterns and a large reservoir. 2 Although there has been habitation
in the lands east of Hallabat [see e.g. Rees 1929], the flint, chert and basalt
wastes, by and large, are unsuitable for agriculture. Some 35 miles south-
east of Hallabat and 50 miles south-south-east of Philadelphia lies the large
and again-populous oasis of el-Azraq [Nelson 1974]. It is the key site in the
region and is now a quite substantial settlement through which a major highway
passes on its way down the Wadi Sirhan into Saudi Arabia. The route is an an-
cient one [Bowersock, forthcoming], although today the traffic in luxury goods
is moving in the opposite direction. Plentiful water is available throughout the
year and pumps even carry water by pipe to H5 on the IPC oil pipeline and

880
another carries it north-westwards to Irbid. It was Lawrence's headquarters
in the winter of 1917-18.
Although reconstructed in the 12th century, the fort was clearly Roman
in design and the style accorded well with the altar to the Tetrarchs still to
be seen there [Bowersock 1971, 241 and pi. XIV .2 = IGRR, Ill, 13391. For
many years, the fort was seen as a link in the Diocletianic re-organization
of the limes. Indeed, Deir el-Kahf to the north of Azraq is not only very
similar in style but it has produced a building inscription of 306 [PAES, II.A.
2., 145-8; III. A, 2., 126]. Further substantiation of this sort of date was
provided by the newly published Azraq inscription of 326-33 [Bowersock
1971; 241; pl. XIV.3 ]. ·Since Prof. Bowersock wrote new discoveries
have modified the dating of military sites in this reign, and some preliminary
statement has already been made on this subject [Graf 1978, i2f. ; Bowersock
forthcoming]. Turning to the evidence:
1. In 1976, S. T. Parker and P. McDermott recorded a new building inscrip-
tion at Qasr el- ruweinid 6 miles south-west of el-Azraq fort:
Pro salute et ingolumitate domi
norum n(ostrum) imp(eratorum) Aug(ustorum) L.Septimi Se
veri Pii Pertinaci[s] et M .Aureli An
[[tonini Pii Felicis et P. Sept(imi) Getae]]
FL[. ]E[. ]T
no[ •• ] Severia[n
They observed that it must fall between 198 and 211 and gave 209-11 as their
own preference. My own inspection of this difficult text (PI. 58. 1) in 1978 and an
examination of photographs and squeezes has enabled me to present an improved
reading. The most significant emendation for the purposes of this '(:nper is
in line 5 where :pERJ?ET can be read. We need have little hesitation in restoring
here, where the governor's name could have been expected, L MARIVS]/
PERPET[VVS • • • • Brother of the well-known Severan general and biographer
iBirley, A. R., 1971, 308-14], he governed Arabia~· 200-23 [cf. MacAdam
1979, 48ff.].
2. Prof. Bower sock presented a partial reading [Bower sock 1971, 241] of
a Latin inscription preserved on a photograph taken by Sir Aurel Stein in 1939
and, on the same page, he referred to another inscription mentioned in Stein's
brief re}X>rt which the latter had assigned to 201. Bowersock thought that. they
came respectively from el-Azraq and from Qasr el-Useikhin, 10 miles north-
east of el-Azraq. My own investigations have revealed that only one inscription
is involved in Stein's rather ambiguous report and that he had recorded it at
'Uweinid. Pere Re ne Mouterde provided the preliminary reading for Stein from
which his date came, and this, together with my own reading from the photo-
graph, has produced the following text:
CASTELLVMETS
PRAESIDIVM SEVE
RIANVM VEXILLATIO
LEG Ill CYRENIC BALN
MV CIA [NO ]ETF A13 [IAN]O [CO] S
EXTRVXIT

881
Plate 58. 1 Detail of building inscription from Qasr el- 'Uweinid showing
name of governor.

Plate 58. 2 Aerial view of Qasr el-Azraq £· 1928 showing earlier fort.
(Crawford Collection, Institute of Archaeology, Oxford.)
 Castellum et <s) praesidium Severianum. Vexillatio
leg(ionis) Ill Cyrenic(ae) (sic) baln(eum) Mucia[no] et
Fab[ian]o [co](n)s(ulibus) extruxit
L. Annius Fabianus. and M. Nonius Arrius Mucianus were consules ordinarii
in 201 [Degrassi 1952, 57]; i.e. the work was carried out during the gover-
norship of L. Mar ius Perpetuus.
3. At Azraq in 1978 I recorded fragments of three milestones, two of which
preserve enough to be restored to give the names and titles of Septimius ·
Severus.
(a) SEVERVS
AVGARABADIA
PARTHMAXPP
(b) IMP CAES L SEPTIM.:
VERVS PIVS PERT
They should probably both be restored in the same fashion and the war titles
in (a) give us a date of 198-211-possibly before 209 if we infer an earlier
date from the absence of Brit(annicus). We can however find support for a
probable narrower date of£· 200-4. for these stones. This comes from an
examination of the road-building in Arabia under Septimius Severus, which
falls into two phases. First, at least 27 milestones bearing the names of the
governor P. Aelius Severianus Maximus, initiated by Pertinax in 193, but
continuing in Septimius' name in 194 [Thomsen 1917, 90]. Then there is the
work initiated by Septimius himself and carried out by L. Marius Perpetuus.
For this we have 4 stones [Thomsen 1917, 90]. I have suggested elsewhere
that it is a plausible inference that all road-making in Arabia under Septimius
was carried out in these two phases of 194 and.£. 2 00-2 [Kennedy 1979]. It
would follow that the two Azraq stones would belong therefore to the latter date
and the two building inscriptions from 'Uweinid of this same governorship
provide strong support.
The cumulative effect is to show Septimius responsible for major building
work at ruweinid and, a few miles to the north, for a road. The road is cer-
tainly that still clearly preserved running due north to Deir el-Kahf. As we
have seen, this latter fort has a building inscription of 306 but the Princeton
Expedition, who recorded that text, noted that two adjoining walls of an
earlier, smaller castellum were encapsulated in those of the early 4th century
fort [PAES, II. A. 2., 147]. Support for an earlier date is provided now by
pottery from the site [Parker 1976, 23].
But what of Azraq itself? There is no epigraphic support for a Severan
occupation there but, with a new fort at 'Uweinid, a newly built road to the
north, probable Severan occupation at Deir el-Kahf, it is just not possible
that no garrison was installed at Azraq itself, the fulcrum of the regional
network. Such positive evidence may some day come to light, but in the mean-
time the prima facie case for a Sever an .or earlier fort is given some support
by the evidence of pottery [Parker 1976, 23] and old aerial photographs taken
before the modern settlement grew up around the fort walls. The photographs
clearly reveal the outline of a large part of an earlier, larger forfbeneath,

883
 Fig. 58. 1
NORTH EAST ARABIA
•CHANATA

GADARA
• ADRAA
• BURQU
•
CAPITOL IAS
BOSTRA •
,. DIU M •SALCHA
I

~
I
I
I

aJ
aJ
~
• THANTIA
',, GERASA
•
UMM EL QOTTEIN

DEIR EL KAHF

•
HALLABAT

• ASEIKHIN
PHILADELPHIA

AZRAQ
~Septimian milestones
•ESBUS
•
UWEINID

e MADABA
0 20 60 100 KMS

-~~-----~-~----~~~~------------
c::.-.. ....
and to the west of, the fourth century castellum. The enclosure has the diag-
nostic playing-card shape and rounded corners of forts up until the third
century.. I could find little trace of it today on the ground amongst the houses,
but I have little doubt that the photographs preserve the Severan fort, largely
re-used to build the castellum. It might, of coarse, be earlier, but we have
no evidence (Pl. 58. 2).
The evidence now available from epigraphic sources and supported by
surface pottery finds clearly reveals the military activity in the north-eastern
desert of Jordan under Septimius Severus. The emperor made good use of
his governor Le Marius Perpetuus and it is significant that the new activity
in Arabia is happening at the same time as much of that in Africa, especially
the new oasis sites in Tripolitania.
A final matter remains for discussion: the garrison of these desert forts.
This brings us back to the matter alluded to at the outset: the reform of the
army as a related feature of frontier modification. It goes without saying
that the external campaigns of Septimius required the services of an enormous
number of soldiers and~ for some years at least, additional garrisons were
required; the extensive conquests and the closer frontier network of Septimius
necessitated a great many additional soldiers. The new legions and the addi-
tional auxilia went a long way towards meeting this requirement. Septimius
Severus was not the man, however, to employ his soldiers inefficiently and
it can be argued that the increase in pay and privileges was intended not only
to make enlistment more attractive but to compensate for the greater demands
of service.
One aspect of this combination of greater demand and efficient use of man-
power can be seen in the treatment of the legions. The use of vexillations
from individual legions or groups of legions was well established by the end
of the first century for campaign or in construction work-one has only to
think of the Jewish War and the building of Hadrian's Wall [Saxer 1967].
Indeed, the use of vexillations as permanent out-posted garrisons is lmown
before Septimius, e .. g. the legionary vexillation and ala established at New-
stead under Domitian and again under Antoninus Pius [Frere 1974, 143 and
175]. As a general rule, however, legions remained concentrated in and
around specific fortresses;bases. By the end of the second century though
there were obvious cases where this procedure was absurd. The possibility
of internal rebellion had receded in most provinces, and in some there was
no longer (possibly never had been) the need for a huge concentrated strike
force of some 5 000 men to check external threats. It comes as no surprise
therefore to find extensive use of detached vexillations as garrisons under
senior centurions in and after the Severan Period. They turn up in the Mrican
forts already mentioned and the vexillations of more than one legion at
Carpow on the Tay and Dura-Europos on the Euphrates are well-lmown
[Saxer 1967, 104-8; 95.; Leach and Wilkes 1977, 59]. Given the character
of Arabia, one would not be surprised to find vexillations of Ill Cyrenaica
playing a prominent role as garrisons in frontier forts in the early third
century. The 1Uweinid inscription seen by Stein certainly supports this view
as does the unpublished altar from Jawf in Saudi Arabia set up by a centurion
of the legion [Speidel1977, 694; Bowersock, forthcoming]. It is open to de-
bate that one reason why the legionary fortress at Bostra has continued to

885
elude fieldworkers is that from the 3rd century at least, there was only a
small camp within the city-as at Dura-Europos-for the headquarters staff
and perhaps no more than a cohort or two. The fortress itself, if it ever
existed, would have been obliterated by subsequent building which continued
through into the Islamic period. One should not place too much weight on it,
but one might suspect that in Arabia the legion had always been scattered:
first around the centres of population, later pushed out into the frontier forts
[cf. Speidel1977, 694f.]. Certainly, no single scrap of evidence has been
found in north-eastern Jordan outside of Qasr el-Hallabat for the presence
of any auxiliary unit [PAES, II.A.2. 21; Speidel1977, 706].

REFERENCES

Birley, A. R., 1971. Septimius Severus, The African Emperor, london.

Birley, A. R., 1974. 'Septimius Severus, propagator imperii' in Pippidi,
D. M., (ed.) D'Etudes sur les Frontieres Romaines, 297-9, Bucarest.
Birley, E., 1969. 'Septimius Severus and the Roman Army', Epigraphische
Studien 8, 63-82.
Bowersock, G. W., 1971. 'A Report on Arabia Provincia', J. Roman Stud.
61, 219-42.
Bower sock, G. W., forthcoming. 'Nabataeans and Romans in the Wadi
Sirhan', Proceedings of the Pre-Islamic Arabia Conference, forthcoming.
Degrassi, A., 1952. I Fasti Consolari dell'impero romano, Rome.
Fentress, E. 1979. Numidia and the Roman Army, Oxford.
Frere, s. s., 1974. Britannia, london.
Good child, R. G. , 1954. 'Oasis forts of Legio m Augusta on the routes to
the Fezzan', Proc. Brit. Sell. Rome JX, 56-68.
Graf. D., 1978. 'The Saracens and the Defense of the Arabian Frontier',
Bull. Am. Sch. Oriental Res. 229, 1-26.
Graham, A. J., 1973. 'Septimius Severus and his Generals, A.D. 193-7'
in M. R. D. Foot (ed.) War and Society, I..ondon, 336-45.

Kennedy, D. L. , 1979. 'The Date of the Arabian Governorship of Q. ScriboniuS

Tenax', Z. Papyrol. Epig. forthcoming.
Leach, J. D. and Wilkes, J. J. , 1977. 'The Roman Military Base at Carpow,
Perthshire, Scotland: Summary of Recent Investigations (1964-70, 1975)'
in Fitz, J. (ed.) Limes: Akten des XI Internationalen Limes Kongresses,
47-62, Budapest.
Leaning, J. B., 1971. 'The Date of the Repair of the Bridge over the River
Chabina', Latomus 30, 386-9.
Luttwak, E. N. , 1976. The Grand Strategy of the Roman Empire, Baltimore-
london.

886
MacAdam, H. I., 1979. Studies in the History of the Roman Province of
Arabia. An unpublished Ph. D. thesis submitted at the university of
Manchester.
Mann, J. C .. , 1974. 'The Frontiers of the Principate' in TemEor~i, H. (ed.)
Aufstieg und Niedergang der Rtlmischen Welt ii , 1, 508-33, Berlin.
Nelson, B .. , 1974 .. Azraq, Desert Oasis.
PAES = Publications of the Princeton University Archaeological Expeditions
to Syria in 1904-5 and 1909, Leyden.
Parker, S. T. , 1976.. 'Archaeological Survey of the Limes Arabicus: A
Preliminary Report', Ann. Dept. Antiq. Jordan 21, 19-31.
Parker, S. T., and McDermott, P., 1978. 'A military building inscription
fromRomanArabia' z .. Papyrol. Epig. 28, 61-6.
Popescu, E. and Popescu, E., 196 8, 1Castrul roman Jidava-C1mpulung',
Studii §i communicari (Pitesti) I, 67-79.
Rebuffat, R., 1972. 'L'arrivee des romaines a Bu-Ngem', Libya Antigua lX-
X (1972-3), 121-34.
Rees, L. W. B., 1929. 'The Transjordan Desert', Antiquity III, 389-407.
Salama, P., 1955. 'Noveaux remoignages de 1'oeuvre des Severes dans la
Mauretanie cesarienne', Libyca 3, 329-63.
Saxer, R. 196 7. Unt-ersuchungen zu den Vexillationen des rHmischen
Kaiserheeres von Augustus his Diokletian, KHln-Graz, (= Epigraphische
Studien 1).
Smith, R. E., 1972. 'The Army Reforms of Septimius Severus', Historia
XII, 481-500.
Speidel, M. P ., 1977. 'The Roman Army in Arabia' in Temprini, H. (ed.)
Aufstieg und Niedergang der RHmischen Welt II, 8, 687-730, Berlin.
Thomsen, P .. 1917. 'Die rHmischen Meilensteine der Provinzen Syria, Arabia
und Palaestina', Zeitschrift des Deutschen PaUtstina-Vereins 40, 1-103.
Tudor' D.' 1977. I Le role defensif du camp roman de Slaveni I in Haupt, D.
and Horn, H. G. (edd.) Studien zu denMilit:l'rgrenzenRoms II, 399-403
KHln-Bonn.

887
 NOTES

1. An excellent discussion of frontiers can be found in M ann 1974.

2. A full discussion of the forts at Qasr el-Hallabat, el-Azraq, el- 'U~~inid
and Useikhin, together with a treatment of associated features, forms
the subject of a report now in progress and arising from a survey I con-
ducted in north-eastern Jordan in July and August 1978. Likewise, full
publication of the new texts and revised inscriptions mentioned below,
together with a number of others not relevant to this discussion, will
appear in the same publication~ A preliminary report is forthcoming in
the Annual of the Department of Antiguities of Jordan.
3. Leaning (1971) has shown that L. Marius Perpetuus was still in Syria
Coele responsible for building work near Samosata in 200.

888
 59. TRADE-ROUTES TO ARABIA AND THE ROMAN ARMY

Benjamin Isaac

Roman roads were usually constructed during the subjugation of a region

or after its incorporation in a province. 1 I shall now attempt to show that a
Roman road was constructed on the territory of a client-king and we shall see-
paradoxically-that this road ceased to be used when the client-state was made
a province. 2
One of the major highways of the Nabataean kingdom, and certainly the
most important of the Nabataean Negev, led from Petra to Gaza on the coast
of Judaea. Stretches of it have been studied by several archaeologists 3 and
one section is the subject of a very fine report (Meshel and Tsafrir 1974; 1975).
The road was provided with way-stations, water-reservoirs, guard-towers
and forts. Parts of the route were marked with kerbstones and uninscribed
milestones. As noted in the report, the latter are found only on parts of the
road running through open country where there was a danger of going astray.
The milestations have two milliaria each. Traces of cutting and levelling are
visible at various stretches of the road,and on steep gradients the road was
supported by a retaining wall.
The route will have been used by the Nabataeans and their predecessors
long before it was constructed in this manner. The date of the construction
in its final form and the period when the road was used have been established
by ~rchaeological finds. The milestones are uninscribed. The pottery found
on the sites of the road is dated by all the archaeologists who have studied it
as material of the first century A.D. or a little earlier. 4 There are no finds
of the late second century A. D. This is confirmed by coins found on three of
the sites. 5 The road, therefore, would seem to have been constructed when
Au~stus ruled in Rome. It was not used after the annexation of Arabia in
106. 6
The archaeologists who have studied the road and established its date feel
confident that this was a Nabataean road, i.e. a road built by the Nabataeans.
But was it? I am not at all certain that it can be proved that the road-stations
and other buildings are Nabataean architecture. 7 There is, in fact, nothing
which serves to identify the builders with certainty as Nabataeans. The pot-
tery and coins on the sites are Nabataean of the first century. But Roman
soldiers could have used Nabataean pottery and coins just as well. The road
has all the features typical of Roman roads in Judaea. 8 Uneven surfaces were
levelled and retaining walls built to support the road on steep gradients. This
is a familiar sight on roads undeniably Roman. Are there parallels undeniably
Nabataean? In short, it can be argued that no features of the road or its sites
are identified beyond doubt as either Nabataean or Roman. The only reason
why it might be considered Nabataean is that it ran through Nabataean territory

889
Note on the map. Fig. 59.1

The map of the Negev and south Transjordan is based on the 1:250.000 map
of the Roman road-system prepared by Dr. I. Roll of the Israel Milestones
Committee. The roads in Transjordan are given as they appear in existing
surveys of the area. Some of the routes in the Negev marked as uncertain
are as yet insufficiently explored. Others are mentioned in ancient sources
but are hardly or not visible on the surface. It may be noted that the exist-
ence of an important ancient route through the 1 Arab a to Aela is denied by
some scholars (cf. Rothenberg 1967, 211ff.; 1971, 160ff.; contra Aharoni
1954, 9; 1953, 113; 1958, 129. While we agree that there is no evidence of
ancient road-building there, Eusebius, Onomastikon 8, 7ff. (Klostermann)
clearly indicates that people travelled from Mampsis through the 1 Arab a to
Elath.

890
 q,'tt TDJJDPPA / ELEUTHEROPOLIS TO JERUS LEM

~r:t;
~c; /
.~J-------- I HEBRDNo
I ~c ~ . ra
. '!;.q,"-'~ /\.e,za \ ~ EN&~DI :
~/ / ,, I ~
~'?1 , / ' I .,...-- -,
/ / \ ' °CHERMELA
I t ' I 7,
,.(~o RHINO COLURA \ \..... (

'
I
& PELUSIUM \
\
\
'
\ '\.
.., /..,
....
\ ,""'
~
I ' BEROSABAC.:_
r ' . . .' - ..........
II '\ /" .I "'~" f/l"g, '
\ nus}o'
/t
\ / \
\ /.1 \ /~""
\ / I ",.
I ( \ I
I ) /
...... /
I _...,---~j~ESSAU
' ) /-
,./ I 1_..;
I \ \ '
\ I ,.\DBDDA
,,
l I
\ I

A-""
_, ..... ----} ,.
I

...
.._
""
I .,/ )

V (
. I
I I \
J I \
\ I \
' I '
\ I '
l \ '
~
l,
'
',
'
)
I
.~-__,,
'\
'
'\ I
'
\
'
\'
' .....
\ } TO S. ARAliA

' ~ I
\\ =
~ I
I
\ ~ I
) ~ I
( I
\ (
\
\ I
./
\ I
\ I
\ I
~-
•
,
lroAR;DE&CLYSMA ----........ -- '
\
'\
I
'\

. '
'--!_, } ----
RDMIJI ROAD
PRESUMED I:DURSE OF RIIMIN
ROAD OR ANCIENT ROUTE

L ----------------~\=-~~'----~------------------------~
891
 However, the question is decided by the presence of
road. Milestones, of course, are the first characteristic
over the Empire. It has been suggested that the Nabataeans,
9
the Romans, erected milestones (Meshel and Tsafrir 1975, 15).
This presupposes that the Nabataeans were familiar with Roman milestones
"''""'~"'""''"" them, However, at the end of the first century B. C. there was
,:>A.JU)';J.cc Roman stone in the entire region. In the eastern provinces
only one road to have been constructed under Augustus (6/5 B. C.,
in Plsidia) .1° Not until A .D, 56 do we find the first dated milestone in Syria.
The st milestone attested in Judaea was set up in 69 (Isaac and Roll 1976).
The the first century B. C. would have had to travel to the west-
ern to see milestones.
Another point to be considered is the distance between the milestones.
The stations are found approximately one mile apart. This shows that the
stones were set up by and for the use of people counting in miles. This points
to Roman, not to Nabataean builders. 11 Significant, finally, is the fact that
two set up at station. It is characteristic of Roman bureauc-
s were persistently added on spots previously marked,
. In north Judaea groups of six milliaria erected in
less than a hundred years are common. The Petra-Gaza road therefore must
be st Roman road known in the region and one of the
antium. The presence of two milestones at each station,
moreover, s that Romans remained responsible for the state of the road,
some time r Hs constru.ction.
l~. historical survey might explain the existence of this solitary
Roma.JJ. road on territory of a vassal state. Petra was one of the main
collecting and striliuti~ centres of goods imported from south Arabia, east
Africa, India and China. These goods reached Petra and thence Gaza and
Rhinocolu.ra (el Arish) through a great number of intermediaries, first of all
the Arab s at Gerrha and south Arabia. These controlled the traffic
in oriental spices, aromatics, precious stones from India and Ceylon, silk
from China and so on. The Nabataeans were, by the end of the first century
B. C., in every respect a client-state of Rome. Contemporary sources keep
stressing that the Nabataeans, like other Arabs, were accomplished traders
and but no warriors. Unlike that other client, Herod of Judaea, their
not care much about public affairs and, least of all, military affairs. 13
They were subject to Rome. At one time Augustus considered giving Arabia
to Herod, says Josephus. 14 Whether true or not, the remark shows the extent
to the Nabataeans had lost their independence. Augustus tried to extend
Roman influence further southwards. In 25 B. C. Aelius Gall us tried and failed
to south Arabia. 15 Strabo, in his description of this campaign, em-
phasizes that the expeditionary force suffered from bad roads impeding pro-
gress and leading the troops astray (xvi. 4. 24). Allegedly the Nabataeans
were interested in the failure of the expedition. The aim of the expedition had
been to conquer the kingdoms of south Arabia reputed to be fabulously wealthy
and a source of various luxury goods. 16 The attempt to establish a foothold
in south Arabia ended in failure. 17 Another expedition to Arabia was despatched
about 25 years later, in A.D. 1, under Gaius Caesar (Bowersock 1971, 227;
Raschke 1978, nn. 836-7. 902). It may o ... may not be a coincidence that

892
Augustus was not at all well disposed to Aretas IV, king of the Nabataeans. 18
In any event, we are not told what were the aims of the expedition of Gaius
Caesar, but we do know that it was carefully planned, possibly because of the
experiences of Aelius Gallus. King Juba of Mauretania dedicated a scholarly
treatise on Arabia to Gai~s.19 Aelius Gallus had sailed from Egypt to Leuce
Come. Gaius, on the other hand, operated in the Negev and marched as far
as the Red Sea without penetrating south Arabia. 20 I cannot resist the temp-
tation of connecting the construction of a Roman road from Gaza to Petra with
this expedition.
The results of this campaign are reflected in the Periplus Maris Erythraei,
a document written in the first century A.D. (Bowersock 1971, 223; Fraser
1972, n. 327; Raschke 1978, .663ff. ). In a well known paragraph it is stated
that a control-post and point of trade existed at Leuce Come (White Village)
on the Red Sea coast and on Nabataean territory (Periplus 19). Strabo describes
it as a great emporium (xvi. 4.23). The Periplus notes that there was a har-
bour there and a fort controlling the road to Petra. A duty of 25 per cent was
charged by a receiver and the place was guarded by a centurion with troops.
Mommsen already .saw the implications of the presence of a centurion at Leuce-
Come ( 1856, 479, with n. 30). He coi!Qluded that Roman troops patrolled
the caravan-road from there to G~za via Petra21 Thus what Moll'1.~sen deduced
from a literary source has now been confirmed by archaeological evidence.
In the first century A.D. the cities in the north of the Nabataean kingdom
became increasingly affluent; these were the cities of the Decapolis, first
mentioned as such in sources of the second half of the century. 22 Our sources
are not informative about the events leading to the annexation of the kingdom.
Marcus Ulpius Traianus, father of the emperor, seems to have been interested
in the development of cities of the Decapolis (Bowersock 1973, 133ff. ). He
certainly was responsible for-the construction of the first two Roman highways
in the region, one linking it with the port of Caesarea in Judaea in 69 (lsaac
andi Roll 1976 ), the other in 7 5, leading to Palmyra and thence to the Euphrates
(Seyrig 1932, 270-2). Trajan continued as emperor what his father had started
as governor. In 106 the Nabataean kingdom was annexed and two major steps
followed immediately. A legion was based at Bostra in the north of the prov-
ince (Speidel 1977, 689ff.) and a highway was constructed from Bostra to the
Red Sea, more or less corresponding to the ancient Transjordanian caravan
route, the so called King's Road.2 3The new road provided easy access from
south Syria and the Decapolis to Aela on the Red Sea. Apart from the strategic
importance of the road-which is obvious-the aim was to draw traffic from
the Red Sea to the Hellenistic cities in the north of the province. All the cities
east of the Jordan prospered in the second century and continued to do so till
late in the third. 24 Direct trade between Egypt and India, and across the
short desert route via Palmyra, may well have increased in importance (cf
Warmington 1974, 99; Rostovtzeff 1932, 102; 1957, 57 5 n. 15). But the pros-
perity of caravan-cities like Philadelphia, Gerasa and Petra shows that they
did not suffer, possibly because eastern commerce as a whole was much more
regular in the second century than in the first (Rostovtzeff 1957, 604ff. and
n. 19). In the late third century Aela became the fortress of the legion X.
Fret. 25 This shows a need for increased security measures in the south. It
shows too that at this time the south end of the Via Traiana was _a key position.

893
 I shall conclude with a few words about the Petra-Gaza road after 106.
The sites along the road show that the road was not used in the second century,
when the Negev had become part of Arabia provincia. There might be a con-
nection. Roman policy would seem to have been as follows: the organization
of Arabia as a province was followed immediately by an extensive project which
aimed at diverting traffic to the Hellenistic cities in the north of the province. 26
These prospered as a result. On the other hand, the main road through the
Nabataean Negev, linking Petra with the Mediterranean,was left unprotected.
It is hard not to see this as deliberate policy of the Roman government. If it
had wanted to keep this road safe it could have done so. Archaeology shows
the eventual result of this policy. There were three important stations on the
roads from Petra to Gaza and Rhinocolura, namely: Oboda, Elusa and Nessana.
All three seem to have declined in the second century, only to revive in the
Byzantine period. 27
Summing up, I think it can be said that, in Arabia, the Romail army appears
as an instrument of imperial trade-policy both inside the province and across
the frontier.

NOTES

1. See e.g. Stevenson 1939, 130, for an idealising view of the Roman
road-system.
2. I am grateful for the generous assistance afforded by Prof. S. Applebaum
and Dr. I. Roll.
3. Frank 1934, 270-6 with comments by Alt 1935; Kirk 1938, 233-4;
Glueck 1953, 12ff.; 1955, 11; Gichon 1963, 86ff.; Rothenberg 1967,
130-7; Negev 1966, 89-98. The road from Petra to Gaza is mentioned
by Pliny, NH vi 32, 145. Pliny describes the position of Petra on the
cross-roads of trade-routes, one leading to Gaza, another to Syria
and thence to Palmyra, and a third to Charax Spasinu.
4. Kirk 1938; Gichon 1963; Rothenberg 1967, 131-2; 1971, 217; Negev
1968, 96-7; Meshel and Tsafrir 1975, 14-5. Bowersock 1971, 239-40
suggests that the road might have been still fortified in the second cen-
tury. But the pottery on the sites shows them to have been abandoned
in the first half of the second century.
5. Negev 1968, 96 (Aretas IV?); Rothenberg -1967, 232: Aretas IV (9
B.C.-A.D. 40) and Malichus II (A.D. 40-71) or Rabel II (71-106).
My colleague Dr. I. Roll has kindly informed me that he has found a
coin of Rabel II in the Grafon fort. According to Prof. M. Gichon
Byzantine coins have been found on several sites.
6. The pottery and coin-finds furnish approximate dates only and a con-
nection with the annexation of Arabia is hypothetical. More precise
dates may be expected when the results of the excavations of Drs.
Meshel and Tsafrir are published.
7. For an army-camp of 100 m 2 at Avdat see Musil 1908, 94 (fig. 65),
107; 118 (fig. 88); 122-4. Partial excavations were carried out in
the fifties and a few years ago by A. Negev (1959, 27 5; 1977, 621).

894
 The fort was said to have been abandoned in the second century (1959)
or in the first (1977). However, coins of Constantine have been found
and I personally picked up pottery on the surface which Dr. I. Roll
identifies as late Roman (third or even fourth century). The plan of
the fort is undubitably Roman, but it has four external corner-towers,
which would suit a late date. The fort might have existed longer than
suggested so far, and later alterations should not be excluded. That
the fort contains only barrack blocks and no other buildings is remark-
able. Nearby a square stone enclosure could mark a temporary or
marching camp.
8. E. g. the road in Judaea from Bethlehem southwards,for which see
M. Kochavi 1972, 26 and map 3 (Hebrew); and the roads in the Judaean
desert discussed by Harel (1967, 18-25). Rothenberg (1967, 136-7)
argued that the milestones of the Petra-Gaza road are recognizable
as Nabataean because they are small in size. However, milestones
come in all sizes and a similar small milestone has been discovered
recently on the road SE of Kurnub ('Scorpion's Ascent') which is gen-
erally accepted as Roman, first-third century. see below n. 27.
9. Incidentally, I must emphasize once again that milestones did not
mark city-boundaries, as claimed by Meshel and Tsafrir 1975, 5.
Boundaries were marked, if at all, by boundary-stones (cippi); cf.:
Isaac 1978, 57-9; Isaac and Roll 1979, 151.
10. Dessau 1916, 5828; Magie 1950, 463-4; followed bytwo roads under
Claudius, one in Bithynia, A~D. 45 (Dessau 1916. 5883) and one in
Pamphylia, · A.D. 50 (Dessau 1916, 215). See also Magie 1950, 488.
11. The distances between the milestations are tabulated by Meshel and
Tsafrir 1975, 5. They vary considerably and the authors note that
measurements could have been taken by marking a thousand double
steps from station to station.
12. See now Raschke 1978. Among earlier studies, cf. Jones 1974,
chapter 7; Warmington 197 4; Miller 1969; DelbruecR: 1956, 8-58.
Charlesworth 1924 is semi-popular. Until the second century this
was primarily a trade in Luxury goods, although in antiquity spices
were used for many purposes, such as ointments, drugs, cosmetics,
incense and so on; cf. Miller 1969, 2 .. For the silk-route which had
Petra as one of its western terminals, see Boulnois 1963. For trade
with China, see also Ferguson 1978, 581ff. Sources relating to trade
to Petra in the second and first centuries B. C.: Diodorus iii 42;
Strabo xvi 3, 3 (766); 4, 2 (767) ff. The importance of Aela is noted
by Strabo xvi 2, 30 (759); 4, 4 (768).
13. For recent literature on the Nabataeans see: Bowersock 1971; Negev
1977; Schtlrer 1973, appendix ii; starcky 1966. On the Nabataean
kings and their army see: Strabo xvi 4, 23 (780); 4, 24 (781); Josephus,
Ant. xiv 2, 3 (31). Brigandage in the Red Sea punished by the Egyptian
fleet: Strabo xvi 4, 18 (777); Diodorus, iii 43, 5. Brigandage in
Trachonitis fostered by Syllaeus, the Nabataean epitropos and corn-
batted by Herod: Jos. Ant. xvi 9,1 (271) ff. The Nabataeans used

895
 to practise brigandage in Syria before they became subject to the
Romans": Strabo xvi 4, 20 (779). Merchants from Arabia Felix used
to be robbed in Damascene country before the Roman army established
security: Strabo xvi 4,20 (756); see also Pliny, NH vi 32, 162. At
least one procurator of Judaea intervened in Nabataea: Cuspius Fadus
(44-?46), who suppressed the 'arch-brigand' Tholomaeus (Ptolemaeus)
when he inflicted great harm upon Idumaea and upon the Arabs (Jos.
Ant. xx 1, 1 (5 )). See also below, n. 21.
14. Jos. Ant. xvi 9, 9 (333-5 ). For other indications that the Nabataeans
were a vassal-state see Charlesworth 1924, 254 with note on pp.
63-4. Strabo xvi 4, 21 (779) tells us that in his time there were many
Romans and other foreigners in Petra. The recorded presence of
Romans in Petra has found archaeological confirmation with the dis-
covery that the Qasr al-Bint,a building previously considered typically
Roman of the second century A.D. (Bowersock 1971, 225-6),dates
to the first century B. C.
15. Strabo xvi, 4, 22 (780) ff. Strabo stresses the commercial motives
for the expedition. See further: Dio liii 29; Res Gestae 26; P liny
NH vi 32, 160-1. For literature on the expedition see: Bowersock
1971, 223, n. 19; and now (Wissmann 1978, 308-544; Raschke 1978,
901-03). In Egypt, too, much was done under Augustus to facilitate
and protect traffic from the Red Sea to Alexandria, cf. Warmington
1974, 14 ff. However, this did not lead to a drastic change in pat-
terns of trade, see below nn. 16 and 17. For the territorial ambitions of
Augustus, see. Brunt 1963, 170-176.
16. South Arabia was a centre of production and one of the greatest of the
entrepot areas, cf.: Periplus, passim; Raschk:e 1978 pp. 836-7, 902.
Strabo xvi 9, 19 (778) notes that the tribes of south Arabia themselves
sailed to Ethiopia and used to sell the products of this region together
with the local spices. For literature on south Arabia see above n.
15 and, especially, Pirenne 1961.
17. Strabo xvi 4, 24 (781-2). Syllaeus the Nabataean epitropos was made
the scapegoat, but it is likely that resistance to Roman penetration
was as widespread among Nabataeans as among other Arabs. Later,
early in the reign of Aretas IV (9 B. C.-A.D. 40), Syllaeus was exe-
cuted in Rome (ibid.), following the repeated complaints of Aretas
(Jos. Ant. xvii 3, 2 (54 ff. ); BJ ii 5, 1 (6 8); for Syllaeus, see
Stein 1931, 1041, and Schdrer 1973, 581). This shows again to what
extent the Nabataenas had become subjects of Rome, Strabo xvi 4,
24 (781) records that, following the failure to conquer south Arabia,
more trade went to Egypt direct (from India and Arabia to Myos
Hormos and thence to Coptus). Direct trade to Egypt became easier
in the first century A. D. with the discovery of the use of the monsoon
by Alexandrian merchants. Instead of sailing from port to port
around the Arabian peninsula, merchants could sail from Berenice
to India and back with only one stop at Ocelis or Cane in Arabia
(Pliny, NH vi 26,101 ff.; Periplus 57; cf. Warmington 1974, 44 ff.).
However, it has been shown that direct traffic to Egypt, bypassing
Arabia, was very slow in developing. In the first century B. G. and

896
 afterwards the Arabian routes retained their previous importance.
See: Preaux, 1952, 257-81 and Fraser 1972, 173-84.
18. Jos. Ant. xvi 9, 4 (295); 9, 9 (353-5). Under Tiberius, Aretas
again got himself into trouble with the Roman government (bee below,
n. 22).
19. Pliny, NH vi 31, 141; another treatise was prepared for Gaius by
one Dionysius of Charax:. Some scholars take this to be an incorrect
reference to Isidore of Charax. See: Pliny, NH xii 56; xxxii 10;
cf. Miller 1969, 15 n. 3.
20. Pliny vi 32, 160: Romana arma solus in earn terram (§£_. S. Arabia)
adhuc intulit AeHus Gallus ex equestri ordine; nam C. Caesar
Augusti filius prospexit tantum Arabiam. For Leuce Come see
the reference in Speidel 1977, 688-9, n. 1. Leuc.e Come marked
the southern frontier, first of the Nabataean kingdom, later of the
Roman provice. As pointed out by G. W. Bowersock, Arabia pro-
vincia included all of the Nabataean littoral on the Red Sea, beyond
Aqaba (1975, 518-9); and, similarly, Paribeni 1927 14-5; also Raschke
1978, nn. 905-6.
21. De Laet 1949, 306ff. has shown, I think, convincingly, that the duty
charged at Leuce Come was Roman and could not possibly have been
Nabataean. See also: Raschke 1978, 664. We may note that Aelius
Gallus sailed from Egypt to Leuce Come and there started his march
southwards. Gaius, on the other hand, extended Roman influence
overland as far as Leuce Come and no further (see above, n. 20).
As pointed out by Mommsen, the fact that the officer in command at
Leuce Come was a centurion shows that Roman troops were involved,
not Nabataeans. Without this piece of evidence it could still have
been argued that the Petra-Gaza road was constructed by Roman
troops and patrolled by Nabataeans, not very probable in itself. In
fact, we do not know whether a regular "Nabataean army" existed
in the first century B. C. (for the references, see Schdrer 1973,
577-8). After Pompey's intervention in Syria, Nabataean troops
are mentioned solely as auxiliary troops levied to support the Romans
(Schdrer 1973, 580). There were two exceptions: first when the
Nabataean defended themselves against Herod invading their territory
at the command of Antony in 32-1 B. C. (Jos. Ant. xv 5, 1 (108-20);
5, 4-5 (147 -60); BJ i 19, 1-6 (364-85)); next during disputes between
Herod Antipas and the notoriously high-handed Aretas IV :in A. D. 36-7
(Ant. xviii 5, 2 (113-4)). It is significant that both kings dispatched
others as commanders (see above, p892 and n. 13) and that Tiberius
considered the actions of Aretas a casus belli.
22. The Decapolis is mentioned first by Pliny, NH v 16, 74 and Josephus,
Vita 65 (341-2); 74 (410) and BJ iii 9, 7 (446). Parker (1975, 437)
has pointed out that references to the Decapolis as a group of cities
are missing in earlier sources such as Strabo. Parker further has
shown that the sources refer to the Decapolis as a geographical re-
gion (east of the Jordan from Damascus in the North to Philadelphia
in the South). There is .no evidence for the existence of -a league or

897
 confederation as thought previously by various scholars. On the
Decapolis from Pompey to Trajan, see now Bietenhard 1978, and A.
Negev 1978, passim. The main source of information on the Deca-
polis remains Kraeling 1938. For Gerasa in these years, see
Kraeling 1938, pp. 39 ff.
23. For the via nova Traiana, see Thomsen 1917 and references in
Bowersock 1971, 237-8. For reflections on some aspects of its
military function, Bowersock 1976, 219-29. Rashke's remarks _(1978,
n. 966) about the road are unconvincing.
24. See Kraeling 1938, 46 ff.; Rostovtzeff 1932, 66 ff. The period of
prosperity started in the first century A.D., cf. Kraeling, 1938,
395 ff. and above n. 22.
25. Notitia Dignitatum, Orientis LXXIII, 30. Of the legionary fortress,
presumably in or near modern 'Aqaba, we know nothing. See
references in Avi-Yonah 1976, 27 s.v. Aila (Berenice).
26. Cf. Rostovtzeff 1932, 68. A section of the Roman fleet operated in
the Red Sea at least from the time of Trajan, cf. Rostovtzeff 1957,
604 n. 19; Warmington 1974, 96; Raschke 1978, 648-9. In the
second century a series of Roman police-stations guarded the road
to Medina (Seyrig 1941, 218-23). It is possible, but not certain, that
at this time Leuce Come was superseded by Aela as control-post and
point of trade, as argued by Warmington 1974, 92, 102. In Egypt, too,
Traj an was active in diverting traffic northwards. He cleared one
canal (leading to Arsinoe) and constructed another between the Nile
and Clysma (Suez) cf. Raschke 1978, 649. He further established
a permanent garrison at both ends of the new canal (Ftolemy iv 5;
Eutropius viii 3; cf. Warmington 1974, 95-6). After his conquest
of south Mesopotamia Trajan drew up fresh tariffs for the eastern
trade passing through Spasinu Charax (Fronto, Pr. Hist. 16 (Haines,
vol. ii, p. 214).
27. Nessana and Elusa may have been abandoned in this period, see Negev
1966, 95-6. As for O:>oda, Negev (ibid.) writes that "shortly after
A.D. 126 Oboda was stormed by a flood of new nomad tribes", who
burned down the city. "After a long gap a new small town was built
at Oboda at about A.D. 242". Similarly Negev 1963, 121; 1967, 47.
However, Negev later suggested that Oboda was abandoned at this
time without destruction (1969, 6, 12; cf. the comments by Bowersock
1971, 225). Now, in a recent publication Professor Negev does not
mention the abandonment of Oboda in the second century (1977, 633).
It is not made clear whether this reflects a change of mind or new
discoveries. In a brief note Mr. R. Cohen of the Department of
Antiquities (1979, 36) states that he and Prof. Negev have discovered
living-quarters of the second and third century, showing that Oboda
continued to be inhabited in this period. He further says that he does
not agree with Negev's dating of Nabataean pottery. Unlike these
three places,Kurnub (Mampsis) farther north flourished in the second
century and afterwards as well (Negev 1967, 48; 1969, 8-9). This
was an important station on the road from Petra through the 'Arabah,

898
climbing the 'Scorpion's ascent' (v. supra n. 8). From Kurnub, Malatha could
be reached, whence roads led to Hebron and Beer Sheba (for the ascent, see
Harel 1958, 145-52). At Kurnub there is evidence of a Roman military pres-
ence (Mann 1969, 211); further references in Speidel 1977, 710-1. Sites of
the road mentioned above contained pottery of the first-third centuries collected
by Meshel and Tsafrir 1975, 15 and n. 25.

BIBLIOGRAPHY

Aharoni, Y., 1953, Eretz Israel ii, 113-6.

Aharoni, Y ., 1954, 'The Roman Road to Aila', Israel Explor. J. iv, 9-16.
Aharoni, Y., 1958, Eretz Israel v; 129-34.
Alt, A., 1935, Zeitschrift des Deutschen PaUlstina-Vereins lviii, lff.
Avi-Yonah, M., 1976, Gazetteer of Roman Palestine.
Bietenhard, H., 1977. 'Die syrische Dekapolis von Pompeius bis Trajan' in
Temporini, H. and Haase, W. (edd. ), Aufstieg und Nidergang der
rl5mischen Welt, Berlin-New York, II. 8, 220-61.
Boulnois, L., 1963, La Route de la Soie.
Bowersock, G. W ., 1971, 'A Report on Arabia Provincia',J. Roman Stud. lxi,
219-42.
Bowersock, G. W ., · 1973, 'Syria under Vespasian',J. Roman Stud. lxiii, 133-40.
Bowersock, G. W ., 'The Greek-Nabaeaean Bilingual Inscription at Ruwwafa,
Saudi Arabia'_, in Cambier, G. (ed. ), Le monde grec: Hommages Clairea
, Preaux, Bruxelles, 513-22.
Brunt, P.A., 1963, Reviewof: HansD. Meyer, DieAussenpolitikdes
Augustus und die Augusteische Dichtung • J. Roman stud. liii, 170-6.
Charlesworth, M. P., 1924, Trade-routes and Commerce of the Roman Empire,
Cambridge.
Cohen, R., 1979, Qadmoniot xii, 36.
Delbrueck, R., 1956, 'Sudasiatische Seefahrt im ALtertum', Bonner Jahrbftcher
clv-vi (1955-56 ), 8-58.
Dessau, H., 1916, Inscriptiones Latinae Selectae, 1892-1916, Berlin.
Ferguson, J., 1978, 'China and Rome', in Temporini, H. and Haase, W.
(edd~ Aufsteig und Niedergang der r~mischen Welt, Berlin-New York, II
9, 581-603.
Frank, F., 1934, 'Aus der 'Arabah I: Reiseberlchte', Zeitschrift des Deutschen
PaHlstina-Vereins lvii, 191-280.
Fraser, P. M., 1972, Ftolemaic Alexandria, 3 vols., Oxford.
Gichon, M., 1963, Elath, The ~8th National Archaeological Convention,

899
Glueck, N., 1953, 'Explorations in Western Palestine', Bull. Am. Sch.
Oriental Res. cxxxi, 6-15.
Glueck, N., 1955, 'The Third Season of Explorations in the Negeb', ~·
Am. Sch. Oriental Res. cxxxviii, 7-29.
Harel, M., 1967, 'Israelite and Roman Roads in the Judean Desert', Israel
Explor. J. xvii, 18-25.
Harel, M., 1969, Bull. Isr. Explor. Soc. xxi, 145-52.
Isaac, B. H. , 1978, 'Milestones in Judaea, from Vespasian to Constantine',
Pal. Explor. Quart. ex, 47-60.
Issac, B. H. and Roll, I., 1976, 'A Milestone of A.D. 69 from Judaea: The
Elder Trajan and Vespasian', J. Roman Stud. cxvi, 14-19.
Jones, A. H. M., 1974, The Roman Economy, Oxford.
Kirk, G. E., 1938, 'Archaeological Exploration in the Southern Desert', Pal.
Explor. Quart. lxx, 211-35.
Kochavi, M., 1972, Judaea, Samaria and the Golan: archaeological Survey
1967-8.
Kraeling, C. H., 1938, Gerasa, City of the Decapolis, New Haven.
Laet, S. J. De, 1949, Portorium, Brugge.
Mann, J. C., 1969, 'A note on an Inscription from Kurnub', Israel Explor.
!!_. xix, 211-4.
Meshel, Z. and Tsafrir, Y., 1974, 'The Nabataean Road from 'Avdat to
Sha'ar-Ramon (Part I)', Pal. Explor. Quart. cvi, 103-118.
Meshel, Z. and Tsafrir, Y., 1975, 'The Nabataean Road from 'Advat to
Sha'ar-Ramon (Part Il)', Pal. Explor. Quart. cvii, 3-21.
Miller, J. I., The Spice Trade of the Roman Empire, Oxford.
Mommsen, T., 1856, RlSmische Geschichte, 1854-6, Berlin.
Musil, A., 1908, Arabia Petraea, 2 vols., Vienna.
Negev, A., 1959, 'Avdat', Israel Explor. J. ix, 274-5.
Negev, A., 1967, 'Nabataean Inscriptions from 'Avdat (Oboda)',Israel ExplO'r.
~· xiii, 113-24.

Negev, A., 1966, 'The Date of the Petra-Gaza Road', Pal. Explor. Quart.
xcviii, 89-98.
Negev, A., 1967, 'Oboda, Mampsis and Provincia Arabia',Israel Explor. J.
xvii, 46-55.
Negev, A., 1969, 'The Chronology of the Middle Nabataean Period', Pal. Explor.
Quart. ci, 5-14.
Negev, A., 1977, 'The Nabataeans and Provincia Arabia' in Temporini, H.
and Haase, W. (edd. ) Aufsteig und Niedergang der rlSmischen Welt, Be rUn-
New York II 8, 520-686.

900
Paribeni, R., 1927, Optimus Princeps, 2 vols. (1926-7), Messina.
Parker, S. T., 1975, 'The Decapolis Reviewed', J. Bibl. Lit. xciv, 437-41.
Pirenne, J., 1961, Le royaum.e sud-arabe de Qataban et sa datation, Louvain.
Preaux, C., 1952, 'Sur les communications de 1'Ethiope avec 1'Egypte
hellenistique ', Chronigue d' Egypt, xxvii, 257-81.
Raschke, M. G., 1978, 'New Studies in Roman Commerce with the East', in
Temporini, H. and Haase, W. (edd. ), Aufsteig und Niedergang der
rlSmischen Welt, Berlin-New York, II 9, 605-1361.
Rostovtzeff, M., 1932, Caravan Cities, Oxford.
Rostovtzeff, M., 1957, The Social and Economic History of the Roman Empire,
2nd ed., revised by Fraser, P. M., Oxford.
Rothenberg, B., 1967, Tsephunot Negev (Archaeology in the Negev and the
'Arabah; Hebrew).
Rothenberg, B., 1971, 'The 'Arabah in Roman and Byzantine Times in the
Light of New Research' in Applebaum, S. (ed. ), Roman Frontier Studies
1967, Tel Aviv, 211-223.
Schit.rer, E., 1973, The History of the Jewish People in the Age of Jesus
Christ, vol. 1, rev. by Vermes, G. and Millar, F. G. B., Edinburgh.
Seyrig, H., 1932, 'Antiquites syriennes: 9. L'incorporation de Palmyre ?l.
1'Empire romain', Syria xiii, 266-77.
Seyrig, H., 1941, 'Antiquites syriennes:: 37. Postes romaines sur la route
de Medine', Syria xxii, 219-23.
Speidel, M., 1977, 'The Roman Army in Arabia' in Temporini, H. and Haase,
; W. (edd.) Aufsteig und Niedergang der rlSmischen Welt, Berlin-New York,
n 8, 687-730.
Starcky, J., 1966, Supplement au Dictionnaire de la Bible, Vol. 7, Paris.
Stein, 1931, 'Syllaeus' in Wissowa, G. (ed.) Paulys Real-encyclopttdie der
classischen Al tertumswissenschaft, IV A, 1041-4.
Stevenson, G. H., 1939, Roman Provincial Administration till the Age of the
Antonines, Oxford.
Thomsen, P., 1917, 'Die rdmischen Meilensteine der Provinzen Syria, Arabia
und Palaestina', Zeitschrift des Deutschen PaUlstina-Vereins xl, 1-103.
Warmington, E. G., 1974, The Commerce between the Roman Empire and
India (2nd edn), London-New York.
Wissmann, H. Von, 1978, 'Die Qeschichte des Sab~errechs und der Feldzug
des Aelius Gall us' in Temporini, H. and Haase, W. (edd.) Aufsteig und
Niedergang der rlSmischen Welt, Berlin-New York, II 9, 308-544.

901
 60. NEW RESEARCH ON THE EUPHRATES FRONTIER IN TURKEY

J. G. Crow and D. H. French

Recent research on the Euphrates frontier in Turkey has done much to

illustrate the history and topography of the limes (Mitford 1974a; 1977; Harper
1977; Hellenkemper 1977; Wagner 1977a; 1977b) but work is still needed to
clarify both its chronology and the details of frontier defence. Much of the
surviving evidence is threatened by the construction of two new dams on the
Turkish Euphrates. 1 The first located south of Malatya at Karakaya in the
Taurus Gorge will flood large areas of the river bank in the Malatya plain,
although the legionary fortress of Melitene (Eski Malatya) lies outside the area
of the proposed lake.
The second dam is being built south of Samsat (Samosata) at Karababa and
will flood a number of important classical sites including the legionary base
and city of Samo sat a. As part of the rescue programme organised by the
Middle Eastern Technical University, Ankara, the British Institute of Archae-
ology at Ankara has begun excavations at Tille, which is in the area threatened
by the lower dam.
Excavations at Tille, May-July 1979
The village of Tille lies on the right bank of the Euphrates next to the mod-
ern road bridge which links Adiyaman to Urfa (Ozdogan 1977, 92); it is about
60 ;km east of Adiyaman and 45 km north of Samsat, in the province of Adiyaman
and the ancient kingdom of Commagene. (Fig. 60.1).
The main area excavated was north and south of the Adiyaman- Urfa road,
at points where road construction had revealed walls in the road embankments.
This recent work had destroyed a large part of the structures investigated.
On the north side of the road only Roman structures were excavated. They
represented successive additions or reconstructions of a major building now
largely lost. Below the Roman levels it was possible to observe Hellenistic
remains (a coin of Antiochus was found in the rubble wall). As yet the Roman
structures cannot be interpreted but a number of features (ash, water channels,
hypocaust bricks, tiles) suggest a bath house.
South of the road the remaining fragments of the 'bath building' were dis-
covered beneath two metres of soil. A fine stretch of masonery and an internal
coarse mosaic floor were found almost immediately. Subsequently a floor of
small rectangular tiles set in herringbone fashion was discovered on the west
side of the main wall. All rooms and areas were completely empty. The
wash soil began at floor level, indicating that the buildings were abandoned
and never reoccupied.
No Roman coins were found; pottery was poor and rare. J:iles were how-
ever abundant and 119 stamped ·examples were found. The impressions give
the names of four military units.
903
 B1a c k Se a

••
0
fortress
fort
watchtower

• town
Roman road. certain
Roman road . possible
..)( pass
Y,z, bridge
0

~
80km
above 2000 metres
1000-2000
0 50 miles 0-1000

Fig. 60.1 Euphrates Frontier in Turkey.

904
 The first group is from Samosata: LEG XVI F F (legio XVI Flavia Firma).
This legion which originally served at Satala was transferred to Samosata after
Trajan's Parthian war (:Mitford 1974, 164-167; Crow 1979, 6).
The second group is incomplete: .•. AC AUG perhaps ala or cohors
Thrac(um) Aug(usta). Both alae and cohorts with the name Augusta Thracum
are known from Syria in inscriptions and diplomas (Jarrett 1969, 217-219;
Rey- Coquais 1978, 68-69 n. 326, 327, 329). Ala Syrica Auro-rsta is also po~s!ble.
The third group is rare and obscure: VIIM. If it does refer to a military
unit it presents problems. Legio VII Macedonica is an unlikely candidate (ILS
2695), and no other seventh legions are known with a beginning in M. Similarly
no auxiliary unit except perhaps an anonymous VII milliaria; but the stamp is
not broken after M, it is complete. VII M may perhaps be connected with the
enigmatic entry in the Antonine Itinerary which records a leg VII at Samosata
(186, 6). Legio VII Claudia appears only briefly in Syria (Rey-Coquais 1978,
68, n. 324). Some of the legions which appear on the Itinerary aru>ear to be
Tetrarchic in origin. 2 Perhaps an otherwise unknown VII Martia is to be
sought.
The last group is the most frequent, 86% of all legible stamps: AL FL
AUG; this can be restored as al(a) Fl(avia) Ag(ripPiana). There are at least
four versions incorporating the name Agrippiana:
Ala Agrippiana
Ala Agrippiana Miniata
Ala F1avia Agrippiana
Ala II Flavia Agrippiana
Of these, the Ala II Flavia Agrippiana is attested once only: at Akhisar (in
Asia) (IGRR IV.1213, date: ?post-Caracallan). Flavia Agrippiana occurs
twice: at Palmyra (in Syria) (AE 1933. 211) and (restored) at Gradiste (in
Macedonia) (ILS 1. 2724). Ala A.grippiana, on the other hand, is found on five
inscriptions: at Gradiste (in Macedonia) (vide supra, line 5), (restored) at
Vaison (in Gallia Narbonensis) (AE 1967, 287), at Worms (in Germania Slperior)
(ILS. 2503) at Grenoble (in Gallia Narbonensis) (CIL XIT. 2231) and El-Hit
(ancient Eitha in Syria) (IGRR 111.1140); in citing this last inscription, Rey-
Coquais (1978, 68 and n. 331) expands A.grippiana to II Flavia Agrippiniana;
Speidel also (1977, 514 n. 10), quoting Cichorius, PW I-2, 1229, A.grippiana,
misprints Agrippiniana. An Ala Agrippiana Miniata is attested on a diploma
(CIL XVI. 69, a. 122 Iul. 17); the title is restored on an inscription from
Cherchel (in Mauretania) (AE 1960. 245).
An Ala Agrippiana Miniata, on this evidence, served in Britain (and in
Germany?) while an Eastern unit, the Ala Flavia Agrippiana, can now be
identified and located on the basis of the Tille evidence. On the other hand,
the origin and location of the unit, Ala II :F1avia Agrippiana, is not known;
Birley (1978, 271) suggests a Flavian formation.
No trace of an auxilliary fort has yet been found. The size of the presumed
bath building and the large number of military tiles from what represents less
than a quarter of the complex which survives to be excavated, strongly suggest
a military context. In future seasons we hope to test for a fort in the immediate
vicinity.
Results of Fieldwork
Fieldwork irl the Tille area in 1978 produced important results for our
understanding of the defence of the frontier. South of .Tille three sites were
found.
905
 The first at s1tma Tepe on the edge of a steep river cliff where the narrow
plain of Tille and Tilbe ends. The structure was apparently square; t?e s?uth,
west and east walls could be traced in part but the north wall on the cliff s1de
had been eroded away. The south wall measured 10.70 m and was approxi-
mately 75 cm wide. It was built of roughly worked ashlar blocks apparently
without mortar. No ditch or surface finds were noted. The site commands
fine views over the river and northwards towards Tille.
South-east along the river cliff, 1050 m from Sitma Tepe ·another site was
located. It survives as a slight mound with some worked stone blocks, no other
finds were noted. Whether this represents a similar structure to Sitma Tepe
is uncertain, but it does provide a necessary link with the next site at Kavu9er
Sergirik.
Kavu<;er Sergirik is situated on a low knoll at the end of a steepsided spur
overlooking the narrow plain of Grik (Ozdogan 1977, 98) and the modern ferry
crossing to the village of Pelis on the east bank. The surviving structure is
square, approximately 21 by 21 m with walls of similar width and construction
as S1tma Tepe. The east wall is best preserved, whilst on the north side its
course is indicated by a low earth and stone bank with perhaps a ditch beyond.
A ditch would only be necessary on the north side. No internal structures
could be made out.
The final site found in the Tille area is situated 7 50 m north east of Tille
Htlyflk on the north edge of Kep1r Tepe above the meeting of the Degirmen pay
and Euphrates. A slight mound was noted about 12 m across, with some cut
stone and stone cllippings. The site commands good views to the north and
east and a tower would be intervisible with Sltma Tepe to the south.
The plans of only two of the four sites are certain and there is no positive
dating evidence to suggest that they are Roman. The regularity of plan of both
Sltma Tepe and Kavu<;er Sergirik, together with their location at good vantage
points overlooking the river makes it very likely that they belong to a regular
defence of the west bank of the Euphrates, which is probably Roman in date.
Tacitus provides little detail as to how Corbulo defended the line of the Parthian
invasion route into Syria (Annals XV 2), although it is clear that this was not
limited to the Euphrates ripa but included forts to protect watering places pre-
sumably west of the river. The clearest description of how the frontier was
defended is by Ammianus Marcellinus. In 359 at the beginning of Sapor's in-
vasion of Mesopotamia he records that fribunes were sent to order the defence
of the west bank of the Euphrates with forts and sharp stakes and all kinds qf
defences. (Amm. Marc. XVIII, 7, 6. )3 But here again towers are not specifi-
cally mentioned although perhaps understood in the phrase 'omnigue praesidiorum
gene re'.
Most of our knowledge of frontier defences below the level of auxilliary
forts derives from limites rather than ripae. Systems of towers and fortlets
associated with river frontiers are only sporadically known, a late Republican
group of watch towers is found around the Walensee in Switzerland, although
not perhaps a true ripa (Wells 1972, 54-6). The best documented systems of
towers along a river are all late in date: from the upper Rhine above Basle,
where at least 39 towers are known of fourth century date (Degen 1972), in
Hungary (Soproni 1967) and from the Djerdap in Yugoslavia where towers are
noted on the map of the frontier defences but not yet described (Kondic 1974,
fig. 2). 4 From Syria, Poidebard saw anumber of towers between the forts
of the Strata Diocletiana and elsewhere on the desert frontier • but none were
noted along the Euphrates bank (19'34, 39 and passim).

906
 The importance of the Tille group of towers is to confirm that an integrated
system of towers and fortlets is to be found on the Euphrates frontier. All are
located on high ground above the river plain and their spacing appears to be
regulated by the need for intervisibility as on the German frontier, rather
than a regular spacing as found on Hadrian's Wall. It is not possible to be
dogmatic about this however, as in stretches of low ground where the river
plain is wide, the post-Roman wash deposits are frequently deep, as we found
at Tille, and all traces are likely to be lost beneath this deposit.
The conclusion that spacing was irregular is however confirmed by dis-
coveries along the main military road between Kemah and Erzincan. 5 A gro':!p
of three towers south of the village of Alpkt!Jy were found on isolated spurs
which projected into the narrow river plain. Two are poorly preserved, but
the northernmost which lies about 100 m from road can be seen to be about 8
m square. This size has been noted at other Cappadocian towers: at KlJbbe
above the main road from Melitene to Samosata, near yit Harabe and at
Zabulbar (see Appendix I). The Alpkt!Jy towers were spaced between 800 and
1000 m apart.
Hypothetical Defence Systems on the Euphrates (Fig. 60. 2)
On the basis of the known evidence from the Euphrates frontier we have
attempted to design a series of hypothetical defence schemes to demonstrate
the variability which the present evidence suggests.
Diagram I: the river bank is level and the main military road runs straight
with forts, fortlets and towers, and a legionary fortress is set back from the
river. It represents the simplest form of system uncomplicated by terrain.
The only sector whe-re this might be found on the ground is in the _Malatya basin,
although the major road runs directly to the legionary base at Melitene, with
perhaps a military way close to the river (Ozdogan 1977, 45, P50/18).
Diagram II: shows the main road following the river bank which is inter-
rUP,ted by spurs on which are set towers. Possibly a subsidiary military way
follows the river bank. Evidence for such a system is found close to Alpkt!Jy
between Kemah and Erzincan (Appendix I).
Diagram Ill; with N, this represents the system where the river runs
into steep gorges. In Ill the main frontier road avoids the river line and takes
an easier route inland. A fort is located between the road and the river, to
retain control over both. Towers are located on the main road and are pushed
forward to overlook the gorge. A military way possibly runs at the foot of
the gorge and along it. Elements of this system are found between Zimara
and Dascusa. A tower has been found overlooking the Euphrates gorge above
the fort at Sabus (yit Harabe) next to the modern road to Kemaliye. A possible
tower on the main military road which crosses the San!;i!;ek yayla was noted
by :Mitford on the pass between Pa~a Mezra and <;anakci (Appendix I). A nar-
row, stone road was seen close to the tower at Zabulbar (Mitford 1972, 299).
Diagram IV: the main road runs away from the river which is marked by
severe gorges. Where possible towers are located close to the river, a military
way is associated with them. This system represents the sector of the major
Euphrates gorge through the Taurus. Possible towers in this area are recorded
near Taraksu, overlooking the Gerger c;ay, and near Gerger Kale (Appendix
I). We observed a narrow, wel_l built stone road, 3.5 m wide, close to Gerger
Kale.
Diagram V: the river has wide bends and meanders, again the main road
is distant but the· military way follows the river bank except where it is too

907
 .. :

>:.: . ·....· :·:

....

. . . .
·. .: ... ·.. . . . . . ' .. ..
. . . . ··. ·:··:. ·; · ........... · .... .

.· ...... :·:.·.-·.···
~~---~....;__...:-

=::s·;::;~)y-)·-··-•- ·i:;x;;;•··•··;ilill
:: ..-:: ·.: ·: ;.. :.... ·...:. ~<: :\: ·.-:.:·~:·/ .· .~.·~<~;·:.: ·.~· . :·: : ~·=.;:·.:.:.;.: \": ":? ..::::.'it~<:·~

•.
.::. \/·.=:>=·::.i: ....<(·
. I .. ..~ ...... ~ .. . ..

••••• • . .. ·····: •..

·.· ..... · ·.· ... N
0
<D
())

_I
lL

908
steep or unstable. Towers are located along the river or beside the military
way. This represents the evidence as seen around Tille. No trace of a mili-
tary way has yet been located near to Tille, but further south alongside the
course of the Samosata aquaduct near Kosan, we observed traces of rockcutting
for a narrow road; probably the military way.
Conclusion
We have not attempted to discuss the evidence for the frontier system
north of the Erzincan plain to the Black Sea. No longer a river frontier, little
evidence is recorded. Watch towers and a possible outpost fort are described
by Billioti in the account of his excavations at Satala in the late nineteenth cen-
tury (Mitford 1974b, 230).
On the Euphrates there are two distinct frontier systems: one strategic,
the other tactical. The former is represented by the legionary bases and the
major military road which ran from Trapezus to Zeugma. For some of its
length this coincides with the tactical system on the river bank, but for ease
of communications it frequently lies back from the river, in places as far as
40 km. To ensure tactical control of the river a system based on forts, fort-
lets and watch towers, linked by a military way was used, often in locations
where we might now consider the qifficulties of the terrain made defence on
the ground unnecessary. This is especially true of the Taurus gorge where
such a system remains to be proven. The Romans themselves were aware of
the vulnerability of this sector. The traitor, Antoninus, himself a soldier in
Mesopotamia, advised Sapor in 359 to lead his invasion army to Earzalo and
Claudias, in precisely this area, since there the river could be forded even
when it was impossible to cross further south. (Amm. Marc. XVIII, 7, 10).
The final problem is t~know how far the systems are contemporary. We
have yet to date any of the structures discussed above: are we in effect con-·
fusing the Stanegate with Hadrian's Wall. Only with further excavation and
fieldwork can we hope to understand the Roman frontier on the Euphrates.

909
 APPENDIX 1

Site of known towers and fortlets on the Euphrates Frontier from South - North.
Vilayet
Name Type Dimensions Bibliography
(Province)
Kavu9er Sergirik Adiyaman Flt 21 x 21 m
Karatflbe Adiyaman Tower? ?
Sltma Tepe Adiyaman Tower 10. 70x?10 m
Kep1r Tepe Adiyaman Tower? ?
Gerger Adiyaman Tower ? Ainsworth 1842,
276
Gerger yay Adiyaman Tower ? Mitford 1977,
508
Taraksu Adiyaman Tower ? Mitford 1977,
508
Near Kttbbe pass Malatya Tower 8x 8m
DfilHik Tepe Malatya Flt ? Mitford 1972,
329
Korpanik Elazig Flt ? Mitford 1972,
282-291
Bahadin Bridge Elazig Flt? 40 x 40 m Mitford 1972,
300-302
Zabulbar Erzincan Tower 8x 8m Mitford 1972,
299
Handeresi Erzincan Flt or 18 x 24 m Mitford 1972,
Mansio 29
Pass between Pa~a Erzincan Tower? Mitford 1972,
Mezrae and 9anakci 242, 406
Near 9it Harabe Erzincan Tower 8x8
AlpklSy 1 Erzincan Tower ?
Alpktsy 2 Erzincan Tower ?
Alpktsy 3 Erzincan Tower
Satala Giimii~hane Tower Mitford 1974b,
230

910
 NOTES

1. Important surveys of the areas to be flooded are Serdaroglu 1977 and

(}zdogan 1977; more fieldwork rema ins to be done.
2. The location of leg. I Jovia at Troesmis and of ll Herculianae at Novio-
dunum by the Antonine Itinerary seems to have the order wrong. Leg. I
Jovia is known from the Notitia Dignitatum to have been at Noviodunum
(Or. XXXIX, 32) and II Herculianae at Troesmis (Or. XXXIX, 29),. the
latter's presence is confirmed by ILS 683, 2781. Despite this error the
Itinerary seems to record Diocletianic units on the Lower Danube section
at least.
3• 'citerioris ripae Euphratis castellis et praeacutis sudibus omnique prae-
sidiorum genere communibant'. Rolfe's translation of 'castellis' as
'towers' is surely inco~rect as Vegetius states 'castellum parvulum, guem
burgum vocant (Epit. Rei Milit. IV, ~10, 18-19)~

4. Two freestanding towers have been described by Petrovic which form part
of the linear defence across the Porecka reka (1977, 260, 262). They are
5 m square and have produced coins of Constantine I.
5• Evidence for the- frontier road north of Kemah demonstrates that the fron-
tier followed the Euphrates into the Erzincan plain and did not run north
to Melik. Serif as Mitford has claimed (1974a, 165-166; 1977, 509-510).
From the Erzincan plain it must have crossed the high Sipipkor pass to
Satala. A number of old roads may be seen on the Sipikor, but it has
~ been impossible to determine which represents the frontier road.

BIDLIOGRAPHY

Ainsworth, A., 1842. Travels and Researches in Asis Minor, Mesopotamia

Chaldea and Armenia, London.
Birley, E. , 1978. 'Alae named after Commanders', Ai1clent Society 9, 257-273.

Cheesman, G., 1914. The Auxilia of the Roman Imperial Army, Oxford.
Crow, J., 1979. 'The Roman Frontier on the Turkish Euphrates', Yayla 2,
5-7.
Degen, R., 1972. 'Sp!trlSmische Befestigungen am Rhein; Weisch, Koblenz
und Zurzach', Helvetica Archaeologica 2, 41-49.

911
Harper, R., 1977. 'Excavations at Pa~nik dreni, Turkey 1968-1971,
and at Dip si Faraj, N-Syria 1972-1973', in, Haupt, D. and Horn, H. G.
(eds. ), studien zu den .Milit!rgrenzen Roms II, Koln, 453-460.
Hellenkemper, H., 1977. 'Der Limes am nordsyrischen Euphrat. Bereich
uu einer arch!ologischen Landesaufnahme' in Haupt, D. and Horn, H. G.
(eds. ), studien zu den .Milit!rgrenzen Roms II, KlHn, 461-471.
Jarrett, M., 1969. 'Thracian Units in the Roman Army', IEJ 19, 215-224.
Kondic, V., 1974. 'Neue Forschungen auf dem obermoesischen Donaulimes'
in Pippidi, D. M., (ed. ), Actes du IxEl Congres International d'Etudes
sur les Fronti~res Romaines, Bucuresti, 39-54.
Mitford, T., 1972. 'The Roman Frontier on the Valley of the Upper Euphrates
from the Black Sea to Samosata', (Unpublished D. Phil. Thesis) Oxford.
Mitford, T., 1974a. 'Some inscriptions from the Cappadocian Limes',
J. RomanStud., 64,160-175.
Mitford, T., 1974b. 'Biliotti's Excavations at Satala' Anat. Stud. 24, 221-244.
Mitford, T., 1977. 'Roman Frontier in Cappadocia' in Haupt, D. and Horn,
H. G. (eds. ), Studien zu den .Milit!rgrenzen Roms II, Koln, 501-516.
6zdogan, M., 1977. Lower Euphrates Basin 1977 Survey METU Series 1,
Number 2, Ankara.
Petrovic, P., 1977. 'Forteresse Romaine a 1' embouchure de la rivi~re
Pore~ka dans les portes de Fer', in Fitz, J. (ed. ), Akten des XI Inter-
nationalen Limeskongresses, Budapest, 259-275.
Poidebard, A., 1934. Le Trace de Rome dans le desert de Syrie BAH XVIII,
Paris.
Rey-Coquais, J. P., 1978. 'Syrie Romaine de Pompee h Diocl~tien' J. Roman
Stud. 68, 44-73.
Serdaroglu, iL, 1977. Surveys in the Lower Euphrates Basin 1975 METU
series 1, Number 1, Ankara.
Soproni, S., 1967. 'Burgus-Bauinschrift vom Jahre 372 am pannonischen
Limes', in studien zu den Militargrenzen Roms Kllln, 138-143.
Spiedel, M. P. , 1977. 'The Eastern Desert Garrisons under Augustus and .
Tiberius' in Haupt, D. and Horn, H. G. (eds. ), Studien zu den Militttr-
grenzen Roms II, KOln, 511-515.

Wagner, J., 1977 a. 'Vorarbeiten zur Karte, "Ostgrenze des romischen

Reiches" im Ttibinger Atlas des Vorderen Orients' in Fitz, J. (ed. ),
Akten des XI internationalen Limeskongresses, Budapest, 669-703.
Wagner, J., 1977b. 'Legio IIII Scythica in Zeugma am Euphrat' in-Haupt, D.
Horn, H. G. (eds. ), studien zu den Militargrenzen Roms II, Kllln, 517-529.
517-529.

Wells, C., 1972. The German Policy of Augustus, an examination of the

archaeological evidence, Oxfo!d.

912
 61. THE IJMES IN THE KURDISH TAURUS

Timothy B. Mitford

At the 1Oth Limeskongress, I gave a general outline of my fieldwork

since 1963 along the Euphrates Frontier in Cappadocia. This year I would
like to discuss in more detail its passage of the Kurdish Taurus, and in
particular the gorge dramatically described by Pliny between the Plains of
Melitene and Samosata.
This is a part of the ripa never to my knowledge visited, certainly not
described, in modern times. Yet it holds a section of the limes comparable
in length with Hadrian's Wall. There similarity ends. For no other section
of the imperial frontier encountered such massive obstacles in its construction,
or hides itself behind them in its research.
But the abrupt sides and depths that have always made the gorge inhospi-
table are well suited to another PJ.rpose: three huge dams are already under
survey or construction at Karakaya, below Bibol, and below Samosata. In
the next decade they _will alter irreversibly a terrain and a way of life unchanged
since the days of Corbulo.
Few travellers have even approached the ripa. Hogarth turned north from
Eski KJihta to Melitene in 1891. So too did Earl Percy, who was dissuaded
from exploring the eastern bend of the Euphrates. But he re'(X>rted that he
had "talked to more than one educated native who had travelled through .the
villages on the east bank between Gerger and Malatya, and they all agreed
that there were no traces of anything of the kind (Roman roads), and that a
French scientist had made a careful but fruitless investigation a few years
before".
Freya Stark rode through the lower part of the gorge, from Gerger as
far as Tillo. But no account of the northern part exists, and a few observations
may be of value.
It took me a week to walk south through the main gorge from PUttlrge to
Gerger, perhaps a hundred miles, with a mule and a Kurdish drover. A
hungry and thirsty journey, for it was Ramazan, and my companion sat
miserably on my mule, descending only to pray. I had hoped to find along
the course of the river traces at least of a frontier road, in continuation of
the long sections that I have followed north of Melitene; and evidence per-
haps for some of the several forts known from the Geographers~ In a road
~ ripam I was disappointed. Indeed, as a formal road it probably did not
exist. For the Euphrates flows through a trench in places more than a mile
deep, and the precipitous mountains ides are cut by ravines and cliffs.
Villages-many in the north once Armenian, and south of Tillo speaking Zaza
and broken Turkish-are scattered above the right bank, linked at best by a

913
 I
I
I
I
I
I
9 MELITENE ARMENIA
' .,I
I
0
c L.
7789

'•Haltkhan
c .. \
... - ...
L.
7149 •
Bur~ Koy
L. HaKn~uru
.. :..k•-.....
... _
Ti+irim
-e 'of
8399 c;:unkU~
' - - -a e Kerar Kale
1
•

L.
8556

L.
7411
L.

~':
8228

L.
Nemrud Da~
7238

US~ SEVER
BRIDG;,'

,,
,,

o'PERRE
Siverek
I
\
I
•
\
LIMES
'\ 0
\
IN THE
0

'
KURDISH TAURUS
'\
\
0 5 10

M I lE S

Fig. 61. 1 The limes in the Kurdish Taurus.

914
rough mule track; a continuous switch-back that climbs across shoulders
many hundreds of feet above the river, and drops abruptly towards its bed
at the mouths of torrents.
Except to snow in ,January and February, this difficult path is not
the main outlet for most villages~ They are linked instead to their district
centres~ at Ptlttlrge and Gerger, by radial mule tracks that cut more or less
directly over mountains. Such primitive communications effectively
determine the shapes of these two adjoining kazas, which lie each at the eastern
extremity of rcsp•'2ctive vilayet. But the demands of geography have not
changed. Along the Cappadocian limes as a whole, a site found suitable for
a fort-the of good water is often the key-or a line followed by an
ancient invariably still in use for a village or a modern track.
The converse also applies~
So the common vilayet boundary must roughly preserve the ancient
boundary between Cappadocia and Commagene in the Kurdish Taurus; and,
more importantly, communications in Roman times must have followed simi-
lar lines between positions still occupied by villages. Thus the existence in
the s two distinct routes between Melitene and Samosata is
readily explained: a direct route through the Kurdish Taurus; and a series
stA~Lticms, Joi.a"1.ed to it by individual radial links, and to each other
nr,P<I"!:8r-~ll1-~~ 1-~·<:lf'·lr suitable at best for pack animals.

The Euphrates cannot be seen as a highway, but a few words on

its navigation may of interest. Outside the gorge, two quite different
types of Keleks-rafts typically of some 40 goatskins
tied together long ash poles-are ideally suited for transporting heavy
loads down the river. I took passage on one, for example, which carried
nearly a ton of corn, as combined ballast and cargo, from Zimara through
the AntiTaurus gorge. Once at their destinations, they can be deflated and
carried back piecf;meal to their starting point. Kayiks, on the other hand,
are used to cross the river. They are best imagined as overgrown punts,
and crab their way slowly between fixed points on either bank where the
current is sluggish.-
But in the gorge of the Kurdish Taurus, navigation can be discounted.
The river is a barriere Indeed Pliny's description shows that the Euphrates
was. navigable in Roman times only as far as the mouth of the gorge, and below
the lowest , 40 miles above Samosata. Von Moltke passed through
by raft in 1840, hoping to find it suitable for the passage of guns for the army
of Hafiz Pasha@ Huntington, who followed in 1900, found that several rapids
could only be negotiated by portage. Only above the junction of the ~iro yay
was there regu1ar traffic. No other navigators have left accounts. My own
intrepid kelekcis had no hesitations in the AntiTaurus gorge. But they spoke
with awe of the Kurdish, of rapids high as a house, and refused to attempt it.
Passage across the river is a lesser problem. In Ottoman times, a
minor caravan route led northwards from Gerger to the ytink:U~ ferry (this
was Freya Stark's route). The ferry was replaced as recently as 1968 by a
fragile suspension bridge e But I have seen regular kayiks only at Ptrot
(Tomisa) in the Malatya Plain, and at Killik (Barzalo) where the gorge first
widens to allow traffic along the ripa itself. There are said to be other

915
kayik crossings in between, accessible only with great difficulty, opposite
the valleys of the ~iro 9ay, and close to the site of the Karakaya dam.
In Roman times, the Kurdish Taurus contained a series of stations
between Melitene and Samosata. The Geographers reflect two distinct
sequences and routes.
The first, preserved in the Antonine Itinerary, had only three intermediate
stations, at Miasena, Lacotina and P(lrre. It evidently led directly across
the mountains by an easy route, for Constantius passed through Lacotena in
36 0 on his way to Edessa. The description per ripam in the Itinerary thus
applies strictly only to the long section between Analibla and Melitene. This
was presumably the route followed by Corbulo in 62, travelling by the shortest
route, with magna vis camelorum ..
The second is known from Ptolemy and the Peutinger Table, and from
its derivative, the Ravenna Cosmography. The lists show a marked similarity.
Between them they contained seven intermediate stations-at Corne, Metita,
Claudias, Juliopolis (only in Ptolemy), Barzalo, Heba (only in the Peutinger
Table), and Charmodara. That the sequence ran broadly per ripam is con-
firmed not only by the schematic map in Peutinger, and by the coordinates
in Ptolemy; but particularly by the reference to Claudiopolis in Pliny's
navigational account-apud Claudiopolim Cappadociae cursum ad occasum
so lis agit (Euphrates)-and by Ammianus' reference to castra duo praesidiaria,
Claudias et Barzalo. Moreover, Metita was garrisoned, according to the
Notitia Dignitatum, by a miliary cohort, m Ulpia Petraeorum; it was part
mounted, with archers, and was evidently a relic of the first century auxiliary
garrison of Cappadocia.
The lines of these two routes continue to defy identification. Research
has preferred to concentrate on the direct. On its course between Samosata
and Melitene, the only fixed point has until recently been the Severus bridge
over the Chabina in the southern foothills of the Kurdish Taurus. From there
Hogarth and Yorke struck north in 1891 without success. The Antonine mile-
stone found by Jtfrg Wagner in the village of B ibo in 1970 implies that a road
did continue northwards from the bridge, and gives a perspective to Severus'
claim that he restored-restituTI'-his magnificent bridge. The temple site
at Derikkale (Lacotena), which I visited in July 1965, was also in use under
Severus (Pl. 61.1). It lies on a saddle some three hours north of the bridge,
may indicate a continuation of the line of the road.
At its northern end, there are some indications that the Roman followed
the line of the modern road from Malatya to Ptlturge. For the course of the
Ottoman route is clearly marked by the ruins of two bans: Halikhan at the
southern a:lge of the Malatya Plain, below the abrupt northern slopes of
Yorke Dag; and Ham Ktlrtlk some two miles from the steep summit of the
pass leading over to the ~iro yay; by a fountain shaded by an ancient plane
tree at Titirim, -where I heard tell of an "Iranian" inscription and a sculp-
tured lion-on the more gentle southern slopes of Yorke Dag, some three
miles above the ~o pay; and by the great apparently Byzantine fortress of
Kerar Kale, in the bed of the ~o pay and commanding what must be a more
ancient crossing.

916
 The positions of Selcuk and Ottoman bans are always interesting: and
particularly in eastern Anatolia, for they tend to preserve the lines of Roman
roads. So, for example, at Topah YaZI Kl1y, some 48 miles north west of
Melitene, a Selcuk ban lies close to a cluster of Roman milestones beside
the modern road.
Between Kerar Kale and Derikkale the Roman road evidently followed a
fairly direct line yet to be resolved. It is likely to have passed close to Peras
Kale and Tepehan, where sections of an old road and the ruins of large bridge
abutments are said to survive. Miasena should probably be located not far
below and west of Ptlturge.
The positions of the seven stations along the ~ are more obscure.
Along the Cappadocian limes to the north, forts tend to guard crossings, or
points where the right bank allows easy access to the west-thus Z imara
beyond the AntiTaurus, Sabus high on its southern flank, and Dascusa close
to the Keban crossing. The same may be inferred in the Kurdish Taurus at
three points of particular weakness: the wide valleys of the eyiro, the Tillo
and the Gerger 9ay. So it is likely that forts were positioned to command
their junctions with the Euphrates, in the vicinity of Keferdis, Tillo and
Gerger respectivley. Perhaps for similar geographical reasons, the villages
of Keferdis and Tillo are still unusually large and important. But apart from
a vaulted Ottoman bridge at Keferdis. I could find no trace of antiquity at
either. Nevertheless, Corne should be located on the southern slopes of the
Siro valley, probably close to Keferdis. For in the valley itse1f, I found
)

ancient remains only- .at BUrc K~, once an Armenian village, three hours
)

west of Keferdis-a 12 metre square stone "tower" with a concrete core and
crude courses, on a hill 50 metres above a dry tributary of the Siro Cay.
Whether it was an Armenian church, or Byzantine, is impossible '
' to judge. It
does not appear to be Roman.
· The Euphrates valley, if only for its very wildness, is far more interesting.
From Keferdis to Tillo is a journey that could be compressed into two extremely
long and difficult days, across the most broken part of the entire ripa. The
mule track, in places no more than a meagre path, skirts precariously around
bare mountainsides at an average height of some 1500 feet above the river
(PI. 61.2).
Two and a quarter hours from Keferdis, below the two halves of the
small Kurdish village of Hu~i~ukam, the Euphrates changes dire.ction sharply
to the south. High above the bend, on the tip of a narrow promontory dividing
the village, is a natural mound some 15 metres high; and on the top are
traces of crude structures. From there the ridge of Yorke Dag and the right
bank below Keferdis are clearly visible. But to the south the view is soon
obstructed by the flanks of the gorge. It is tempting to associate so fine a
vantage point with a continuation of the signalling system inferred in the
AntiT aurus. On the promontory are several Ottoman graves similar to the
groups scattered along the frontier road 100 miles further to the north.
The track continues high above the river for a further three hours-past
Tilman, known for its almonds and grapes-before descending sharply through
the hamlet of Kerefto to cross the mouth of the Kerefto gorge in the. bed of

917
Pl. 61. 1 Ruins at Derikhale (Lacotena ?).
PI. 61. 2 The gorge below Htl§ttk:ani.
the Euphrates. From there, a direct path across the mountains leads in
three hours to Keferdis. The crossing is guarded by the remains of a fortlet
or watch tower standing on a natural promontory 400 feet above: a structure
7 metres square, the walls faced inside and out and constructed without con-
crete; from a small sherd of Samian type it is tempting to infer an early
date, and a position garrisoned by an auxiliary detachment (PI. 61. 3).
An hour and a half's steep climb through poplars and mulberries leads
to Haskento, a pretty Kurdish village spread out in a bowl over-looking the
'
Euphrates. Again without trace of antiquity. I stayed with the Mukhtar, in
his beautiful house far above the river. He advised me not to go on to M idye,
for I would be shot.
Above Haskento, the track climbs steadily for an hour to a bare ridge
2000 feet above' the Karakaya dam site, hidden in the gorge below. From
there the fortlet at the Kerefto qay is easily seen. A further hour leads to
a treeless promontory, surrounded by cliffs rising sheer for three thousand
feet above the river. This is the highest point on the ~ with a fine view
south towards the Tillo valley. The track keeps its height for an hour, dipping
through the hamlet of Kerkinos with its goats and almond trees; then threads
carefully down a fault in the 2000 foot high semicircle of crags and screes
that reflect the sun down on to Midye. It is a large village, and-as always-
friendly, spread out like Ha~kento among fruit trees and walnuts and fertile
fields, in a hanging valley a mile across. I had hoped to find there traces
of Metita, for Midye is perhaps the most isolated village on the ripa-four and
a half hours from Ha§k~;:mto, and five and a half to the nearest village to the
south. As at Kerefto, a direct path crosses the mountains to Keferdis in six,
and to PUturge in eight hours. But no coins, no masonry in reuse, no inscriptions~
Communications run north from Midye, and the villagers have only
infrequent reason to venture south towards Tillo. It is a linguistic frontier
as well; to the north Kurdish, and to the south unwritten Zaza. So only a
rough track leads down in 90 minutes to the He skin Cay,
t
carving a deep and
narrow gorge into the Euphrates. It is spanned by a bridge of tree trunks,
100 feet above dark water. The track ends there, leaving a four hour scramble
over a waterless mountain thick with stunted oak trees. At the top is a track-
less waste that is the haunt of bears, and on the south side 2000 feet of scree
,
above the Tillo Cay. This must have been effectively the border between
Cappadocia and Commagene.
The Tillo valley is not as wide as the Siro,
J
and from its river the track
leads in just over an hour to Tillo. It is a large village of 300 houses, with
abundant fields, a primary school and a new mosque: but without traces of
antiquity. From it a track leads directly to Gerger in eight hours, unaffected
by winter; and to PUturge also in eight. But the latter is closed by five months
of snow. Freya Stark followed it, and found it extremely cold.
On the opposite side of the Euphrates the mountains fall back to allow
access from CUnkUs and the Plain of Siverek. Three hours to the east of
' )

Tillo was the ferry, now replaced by the suspension bridge. The valley is
exposed to infiltration from across the Euphrates. It deman~ed a powerful
garrison, for the legions at Melitene and Samosata were at least two and

920
three days' forced march away respectively. A part mounted miliary cohort
would have been well suited. There is little choice but to locate Metita close
to Tillo. For the mountains to the north, except at inaccessible Midye, hold
no place for it.
An excellent track, in places paved, leads east from Tillo towards the
ferry. It takes nearly two hours to Zengeto, 30 Kurdish houses set among
trees and terraced fields above the Euphrates, tobacco and peppers drying on
their flat roofs. This must have been the route followed by Ammianus in his
flight from Amida.
Another hour to the corner above the suspension bridge, and then the moun-
tains draw back from the right bank, leaving a high terrace across which the
track winds through fields and villages. Haburman is two hours away, with
lOO houses and 67 children, speaking Zaza, ·and talking of two old coins, since
sold. An hour to Deyro, a poor and dirty village of 70 houses, well watered
but unwelcoming, and without school or mosque. And from there nearly two
hours to Bibol.
It is a desolate village, hemmed close above the river by an amphitheatre
of dark cliffs, and shown by scattered foundations to have been once larger
than it is today. It has :few fields or trees, but the Mukhtar's garden was
full of carefully tended vegetables. Like Midye, it has an intangible sense of
history, beyond the obvious association of the name with the Tto:>..t.sof Claudio-
polis or Iuliopolis. It is certainly a candidate, for it has a stream of good
water, and the caravan route passed below, with a small paved section leading
around the cliff on the southern bank of the Bibol ~ay. But it is perhaps too
close to Tillo-seven hours_.and to Killik-two hours-to have contained· a fort.
High above to the west hangs the bare conical peak of Ziyaret Tepe, a twin to
the peak of Nemrud Dag at the opposite end of the same range.
South of Bibol the track descends into the bed of the Euphrates for the first
time, excluding the mouth of the Kerefto ~ay, since the Malatya Plain; and
skirts along the edge of a narrow alluvial plain. Here the river runs slowly,
nearly 200 metres wide. Just above is the site (now abandoned) of the Bibol
dam, bedded in a vast volcanic plug. Two hours below it, round a bend where
the river turns sharply to continue the wide valley leading down from Siverek,
the village of Killik stands beyond rich fields on a rocky promontory jutti.D.g into
the Euphrates (Pl. 61. 4). With 80 Zaza houses packed tightly together, it is
the first village actually on the ripa since P1rot (Tomisa). A copious sj>ring wells
up in the fields 500 metres behind the village. The name for the area in gen-
eral is Berzelo Cem, the "field of Berzelo," and the spring is Berzelo ~e§mesi.
The coincidence of name, like modern Zimara in the hills an hour behind
Pingan (Roman Zimara), and the evidence for Roman occupation leave little
doubt that this is the site of Barzalo, one of Ammianus' castra praesidiaria,
well placed to guard the easy approach from Siverek.
There are cut stones in the houses, and one has a deep well. I was shown
a worn ~ from Antioch, probably of Trajan, and told of inscriptions found
and buried, of a rock tomb with three undisturbed graves, of multiple graves
with hundreds of skulls near the school, of foundations nearly everywhere
beneath the village and its fields, of water pipes of earthenware set in

922
.
~
concrete, 40 cm long and 15 in diameter, similar to pipes at Satala. Two
piers, apparently from an aqueduct, survive in a field 150 metres from the
village. still standing to a height of 4 metres, they show traces of the original
facing on all four sides, around a core of concrete and some bricks (Pl. 61. 5).
They are 2 metres long and 3 wide, and 16 metres apart. Two to three hours
above the village, on a huge rock outcrop easily visible from it, is said to be
a castle, with walls and rock cut inscriptions .
•
Below Killik the track follows the low bank of the Euphrates for half an
hour, and leads to a busy kayik crossing. Soon after, it leaves the river,
climbing slowly through Ninyat-a small village in despair over an almost dry
fountain-to reach a low saddle overlooking the broad valley of the Gerger
Qay. At the top are two mounds some 7 metres in diameter, from which
Nemrud Dag is visible. Similar mounds in the Anti Taurus are perhaps
associated with signalling.
From the saddle the path descends to Helim, a poor village just above the
right bank, its meagre fountain supplemented from the river. Euphrates
water, claim the villagers, is very healthy to drink. An hour and a half to
the west is Taraksu. In the houses are several large ashlar'blocks, said to
have come from a ruin field above the mouth of the Gerger Qay.
This proved to be a site 40 minutes from Taraksu, and half a mile from the
Euphrates, where a rock outcrop stands 15 metres above the surrounding
cotton fields. On it is a rectangular building, 8 metres long and 4 metres
wide, with fine ashlar facing on a core of rubble and concrete (Pl. 61. 6). All
around, over a radius of 150 metres, are the foundations of buildings, and the
fields are full of pottery. The construction is similar to the tower at Derikkale,
evidently dating from the second century. It is clearly not a fort. If the iden-
tification of Killik with Barzalo is correct, it is likely to have been Claudio-
polis or Juliopolis; for the valley of the Gerger Qay, above all other parts of
the Kurdish Taurus, is suitable for the foundation of a civilian rather than a
military settlement, close to the Antiochene fortress of Gerger Kalesi. The
turn to the west reported by ·Pliny occurs not far above Taraksu, and the site
is altogether more attractive than the cramped and infertile position of Bibol.
From it a direct path is said to lead over the mountains to PUtttrge, in about
eleven hours; and accordingly I paid off my drover and sent him thirstily on
his way.
~, in summary, this account has not located the stations spread along the
ripa. But it has perhaps narrowed .down the areas of search for Corne and
Metita, and has proposed identifications for Barzalo and Claudiopolis. It has
established that no formal road followed the right bank, and suggests rather
that stations were linked by individual tracks leading across the mountains
towards the direct road from Melitene to Samosata. And it identifies the
probable route followed by the latter.

924
.
~
.
~
 62. AUGUSTUS' EGYPTIAN FRONTIER: Q'ASR !BRIM?

W. H. C . F rend

The well known account by Strabo (Bk. xvii.l. 5, 3-54) of the relations
between the emperor Augustus' Prefects of Egypt and the Ethiopians, has
usually been interpreted to mean that the Roman advance down the Nile in re-
prisal against Ethiopian raids against Syene was short-lived. Strabo records
the victory of Publius Petronius over the army of the Ethiopian Candace (Queen)
at Ibrim (Premnis) in 23 B. C. Then after advancing as far as Napata, Petronius
fortified and garrisoned Ibrim with 400 men "with food for two years", before
returning to Alexandria. The assumption has been that after this period the
garrison was withdrawn and the Roman frontier pulled back northwards to
Hierasykamenos.1
There are reasons today to challenge this assumption and suggest instead,
that U>rim played a prominent part in the frontier policy of successive Prefects
of Egypt during Augustus' reign. Though the eventual date of evacuation re-
mains unknown, the total occupation may be measured perhaps in terms of
decades rather than years.
The evidence on which this belief may be based has come from a remark-
able series of chance finds made during the Egypt Exploration Society's ex-
peditions to the site in 1974 and 1973. In March 1974, the writer was working
for the Society under the direction of his colleague Rev. Professor J. M.
Plutnley with the object of finding evidence for the dating of the dry stone walls
that form the surviving outer wall of the fortress. After clearing rubbish from
the line of the south wall he moved to the angle formed by the base of the pro-
minent tower that dominates the south-west corner of the site. The tower had
been built on a series of well-cut steps approx. 0. 70 m deep. These had been
excavated from the rock, which before the raising of the water-level of the
Nile resulting from the High Dam, would have stood some 200ft (60 m) above
river level on a steep incline.
The steps were wide enough to allow a good deal of material to accumulate
on them. Below the top layer containing Nubian material, a hard compact
layer was encountered, which contained papyrus fragments of forty Greek and
Latin documents dating by their writing to 1st c. B. C.-1st c. A.D. (Frend
1975; Weinstein and Turner 1976). The Greek included a fragment of the Iliad
(viii. 273-276) and two considerable fragments of Bks ii and v of the Odyssey
(ii. 72-100, 107-108, 110-111, 120, 122-125) and (v. 122-133, 135-141,
165-171), but also such material which seemed to be lists of military stores.
Provisions are mentioned including wine, peas, grain, lentils and onions.
There is a letter mentioning the rank of beneficiarius (Weinstein and Turner
1976, 125), fragments of another letter referring to a sum of money of at least
two talents, and possibly to military provisions (Weinstein and Turner 1976,

927
122), and six private letters. The Latin documents include what seems to be
a set of soldiers accounts and the beginning of a letter written in a large for-
mal hand " ... Varius Valerio suo s al pl. .. " ("Varius to his friend Valerius
very many greetings") (Weinstein and Turner 1976, 127).
In the opinion of their definitive editors, M. E. Weinstein and E. G. Turner,
the documents most likely were connected with Petronius' expedition: "It is
hard to resist the view that the force is that led by the Roman, Petronius in
23 B.C." (1976, 115). Strabo says that Petronius fortified Premnis, and the
tower would seem to be his work. 2 Resting on a rock-cut step some ten feet
from the angle formed by the tower and wall, two amphorae were found made
of hard reddish ware with long necks which one associates in Britain with the
first half of the first century A. D. The layer in which all these finds were
made was extensive and compact 3 and hardly appeared to the excavators to be
the result of an occupation lasting only two years.
Why did Petronius choose !brim for his garrison? It was certainly a strong
point and a religious centre which the Meroites had fought hard to retain, but
there were other fortified positions in the area, such as Djebel Adda some 50
miles to the south. What may have increased the attraction of !brim however,
was that Petronius had been preceded only six years before by an expedition
led by Cornelius Gallus, Augustus' first Prefect of Egypt. Gallus was a thrust-
ing general, who advanced through Thebaid and beyond the First Cataract
(Dessau 1893, no. 8995). This was too far for Augustus who recalled him,
and he committed suicide in 26 B. C. after being accused of peculation.
Apart from his military prowess, Gallus was a poet, credited with four
books of Amores and Elegiacs. One line, a pentameter of an elegy has survived.
"Uno tellures dividat amne duas". 4 This meagre harvest has now been sen-
sationally increased. During the 1978 E.E.S. expedition directed by Mr.
Robert Anderson, a poem by Gallus written in honour of Augustus (?) was found. 5
Did this poem reach Ibrim through Ga1lus, or in the knapsack of an officer
who admired him? Had he returned thence to Phllae where he received envoys
from the King of MerlSe and extended Roman protection over his Kingdom, -a
policy which Augustus disavowed (Shinnie 197 8, 245-6). There followed the
Roman withdrawal, and perhaps the rejection outright of the protectorate status
imposed on the kingdom by Gallus symbolised by raids by Meroite troops
against Syene, and Petronius' punitive expedition. This time the Romans stayed.
Temples were built at Kalabscha and Madamud to the north in honour of Augustus
and Tiberius respectively, and near Abu Simbel, 55 miles south, in honour 'of
Augustus. 6 Seen thus, ilirim may be reckoned an important frontier post,-
and perhaps the pivot of the advanced Roman defences south of the First Cataract
in Augustus' reign.
One final problem is posed by the discoveries, namely the original setting
of the famous bronze head of Augustus, found by Garstang at MerlSe in 1912,
just in front of a small temple which he believed had been built in connection
with it (Shinnie 197 8, 250). It is usually thought to have been the result of the
Meroite attack on Syene which provoked Petronius' expedition to !brim. How-
ever, in 1974 there was discovered at !brim itself a circular pavement 4 m
in diameter near the south wall of the fortress, that is, facing southwards
down the Nile (Plumley 1975, 16, fig. 4 and pl. xi.i). One is tempted to

928
compare it with the great concrete base found at Richborough, and the landfall
of the Claudian invasion of Britain, which seems to have been part of an im-
perial monument (Bushe-Fox 1926, 35-6 and 1928, 12-13). Did the !brim cir-
cular foundation serve a similar purpose? It was soon covered by the line of
a temple wall. How better to symbolise the restoration of Meroite sovereignty
over !brim than to remove any statue of Augustus that may have stood there
in MerlSe?

NOTES

1. For a discussion of the relations between MerlSe and the Roman empire
under Augustus, see Shinnie 1978. strabo indicates that in contrast to
Napata which Petronius razed and then evacuated, he "fortified Premnis
better" and established a garrison there. For the assumption that the
Roman occupation of !brim was brief, see Plumley 1964, 5.
2. This would seem to be evident enough and contradicts W. Y. Adams'
opinion that the S. W. tower was ''Bosnian and dated from the sixteenth or
seventeenth century" (Plumley et. al. 1977, 37). My investigations east
of the podium also convinced me that the main dry stone walling in that
area was Nubian, and that Bosnian additions had been confined to the top
seven courses of the fortification.
3. A view reinforced by Sir Laurence Kirwan in a letter to me after visiting
the site in 1976.
4. See, Frag. Poet. Latin., ed. Morrell, p. 99. (I owe this reference to
Mr. J. G. Griffith of Jesus College, Oxford).
5. To be published in the Journal of Roman Studies in 1979 by Robert Anderson.
I owe my information about the discovery to Professor Plumley.
6. Now reconstructed and set up in the Rijksmuseum at Leiden.

BIBLIOGRAPHY

Bushe-Fox, J. P., 1926, First Report on the Excavations of the Roman Fort
at Richborough, Kent, London.
Bushe-Fox, J. P., 1928, Second Report on the Excavations of the Roman Fort
at Richborough, Kent, London.
Dessau, H., 1893-1916, Inscriptions Latinae Selectae, Berlin.
Frend, W. H. C., 1976, 'Some Greek and Latin Papyri of the period 50 B. C.-
50 A.D. from Q'asr !brim in Nubia' in Proc. XIV International Congress
of Papyrologists (Oxford 1974), 103-111, London.
Plumley, J. M., 1964, 'Q'asr !brim 1963-64', J. Egypt. Archaeol., 50, 3-5.
Plumley, J. M., 1975, 'Q'asr !brim 1974', J. Egypt. Archaeol., 61, 5-27.
Plumley, J. M., Adams, W. T. & Crowfoot, E., 1977, 'Q'asr !brim 1976',
J. Egypt. Archaeol., 63, 29-47.

929
Shirutie, P. L., 1978, 'TheNiloticSudanandEthiopia'inFage, J. D. (ed.)
Cambridge History of Africa, vol. II, 245-52, Cambridge.
Weinstein, M. E. & Turner, E. G., 1976, 'Greek and Latin papyri from Q'asr
!brim', J. Egypt. Archaeol., 62, 115-30.

930
 63. SIGNIFICATION D'UNE FRONTIERE : NOMADES ET SEDENTAIRES
DANS LA ZONE DU LIMES D'AFRIQUE

Pol Trousset

L'opposition des nomades et des sédentaires a fourni, parallè-

lement à celle de la montagne et de la plaine, un schéma explicatif
de prédilection à l'historiographie de l'Afrique du Nord antique. Au
r8le négatif du montagnard rebelle faisait pendant celui de cet au-
tre protagoniste de "l'Afrique oubliée" : le nomade pillard refoulé
au désert par l'occupation romaine. A l'un ou à l'autre ou bien en-
core aux deux à la fois mais à des degrés variables suivant les au-
teurs, il convenait_ d'attribuer l'échec final de la romanisation de
l'Afrique (LEVEAU 1977 : 205).

Selon cette même représentation, c'est au long du limes conçu

beaucoup
, mieux depuis les travaux de Baradez comme une zone frontiè-
re (frontier) que comme une frontière linéaire (boundary), et ceci
malgré l'illusion des murs ou fossés en réalité discontinus qui en
soulign&ient le tracé, que s'opérait l'affrontement entre nomades et
sédentaires. Le limes ainsi perçu n'était pas seulement un moyen de
différenciation spatiale, mais un moyen de différenciation temporelle
dans la mesure où il séparait deux modes de vie incompatibles ayant
chacun son rythme propre. On avait là une des meilleures illustrati~

ons possibles de la frontière "isobare" d'Ancel exprimant un rapport

de force et d'organisation dans une zone critique de transition cli-
matique (GUICHONNET-RAFFESTIN 1974 : 16). La mise en place de ce
dispositif de défense anti-nomades, de mieux en mieux connu dans la
. complexité de ses articulations
.. stratégiques et logistiques (forti-
fications, voies de desserte et de pénétration, etc.), apparaissait

9'31.
 TH EVESTE Tebessa ·- - - -..... --- - ...-- ,-..-. ,..- ---,.-.-
- . - .-....-- -~
--- -
- .-- .--...- -..
---
J __ _ _ ,.-._,... _. . - _ .- _ _ _ _ , . . , , _.... . - -- -- - .-.

---- ---------.,.,.--
Gafsa ..,~.- - - - .---- ~ -~- --~ -J · . -.-- ...--.-.,­
--- ~ _,--. ..-_-- _._ .-

Besseriani o
El AHida -----_
--- - - -...,-.,.-_-
___.,..•__ ..._-- .-- ---.-_,------- ______ ___ ----,.-...
.- ---- --- .- ...--- - -
_.,... .. ......
_..,.... ___ ____ ,_
-. - _
_,...,.....__.,
. _....,-....-.,-
_,..... ~.,.. -
,..,.-
...
_.,-.,...,.-....,-
-
-- ....... .-..;-
_. - - .........
~ ___ ___...-...-
,--.,-.
___..

....-,.-
___ _- ._...-.....,.;-- _,.-._,..-...
..,.-. ............... -

/ .
/ ;. ·:·.
, ··=:
ù,-·.·.:.
'·' .. ... . ·....... .
Tel mine

''
'
''
,)- 'r .....' ...... ~-: :·. -..: ;~.: .
~,f
es-- S_o_·· ; ~ ~-

TISAVAR O
Ksar Rhilane
Grand \\
-:..
('t)

D
•
C

•
o camp. fort. fortin du l imes

fermes fortifiées Erg

- ......
tu
= fossatum clausura du Chareb
}r- '"""\
voie et milliaires
_ _ _1_1
Ortental .....l
TI L LI BARt QJ
R~?mada
piste caravanière
-yi
relai optique .1/
/ //'
centuriation du Fejaj s, ArJIJn o, ·-. ,.f'
.,
\\ "' r,,
1ones de cultures \

::::::=::;::> b
-> mouvements sa i sonniers des
semi nomades du Nefzaoua
des sahariens \
\
\ CIOAMUS
1. A . M o...,___ _ _ _ __ _______
lO.......
Okm J. LENNE PJfA Z A N f f ~ Rhadames

Fig .. 63.1 Organisation du limes dans le sud tunisien

comme une tentative grandiose mais finalement désespérée pour mettre
"l'Afrique utile" à l'abri des entreprises de destruction issues du
monde instable de .la steppe et du désert.
A la lumière des travaux les plus récents effectués sur les di-
vers secteurs du limes d'Afrique (EUZENNAT 1976 : 526-'538; EUZENNAT-
TROUSSET 1978 135-156; SALAMA 1973 : 337-345 et 1976 : 577-595;
REBUFFAT 1976 395-419; TROUSSET 1978 : 125-177), dans quelle me-
sure un tel schéma est-il aujourd'hui encore recevable? Vouloir
traiter d'un aussi vaste sujet dans les limites étroites d'une com-
munication serait une bien aventureuse gageure si mon propos n'était
pas d'abord de mettre en relief les présupposés qui en sous-tendant
cette représentation du limes lui avaient conféré l'autorité des vé-
Tités acquises une fois pour toutes. A partir de quelques exemples
concrets empruntés aux résultats de prospections dans le Sud tunisien
je tenterai néanmoins d'apporter plusieurs correctifs nécessaires à
cette vision traditionnelle des rapports de Rome et du nomadisme sur
la frontière saharienne de l'Empire.

*
* *
C'est en effet à travers un faisceau de perspectives en trompe.
l'oeil que s'est développée à propos de ces rapports, une image du
limes dont la première forme remonte à une idée de Gsell (GSELL, 1933:
-149-166) et dont, malgré les réserves pourtant nombreuses que celle-
ci avait suscitées (LESCHI 1942 : 47-60; PICARD 1960 : 66-67), le·s
derniers sont encore perceptibles dans la synthèse récente de Béna-
bou (BENABOU 1976: 69-73, 209-211, 443-444) : privés de leurs ter-
rains de parcours du Tell, puis poussés progressivement au désert
par l'extension de la colonisation romaine, les nomades de la steppe,
devenus sahariens (grâce au chameau selon Gsell) en sont réduit à se
heurter désespérément à la barrière du limes qu'ils finiront par
forcer.

933
 La première erreur de perspective imputable à la nature même
de la documentation archéologique et épigraphique qui privilégie la
période romaine, consiste à isoler celle-ci en portant à son seul
crédit un processus de sédentarisation qui s'inscrit plus vraisembla-
blement dans un cycle de très longue durée puisque ses effets conti-
nueront de se faire sentir localement sur ces franges sahariennes en
CyrénaYque, jusqu'à l'invasion hilalienne (DE PLANHOL 1968 : 131-
133). Commencé du reste bien avant l'intervention de Rome, ce proces-
sus peut s'expliquer par une lente maturation de la société indigène
au contact du monde méditerranéen (CAMPS 1960 : 209-213; THEBERT
1978 : 78). Il suffira de rappeler que dès l'époque des royaumes
numides, une forteresse comme celle de Maktar surveille le franchis-
sement de la Dorsale par les nomades de la steppe et qu'en préfigu-
rant celle de Tacfarinas contre Rome, la révolte d 1 Aphter contre
Massinissa montre bien que le problème des rapports entre agricul-
teurs et nomades n'est certainement pas né de la conquête romaine,
m~me s'il a pris alors une forme nouvelle ( KOTULA 1974 : 52). Au
demeupant, le cas même d'Aphter qui se réfugie chez les Garamantes
.ne doit-il pas faire illusion plut8t que la fuite au désert, c'est
la sédentarisation qui est la tendance la plus constante d'une socié-
té nomade.dans son rapport avec les agriculteurs sédentaires dès lors
que la balance des forces s'inverse en faveur de ceux-ci (BOURGEOT
1972 : 85, 88).

Par ailleurs, la mise en évidence à la suite de Despois, par

Lawless (LAWLESS 1972 : 450-464), d'une aire spécifique d'ancienne
civilisation rurale présaharienne fondée sur l'irrigation et la cul-
ture en terrasse, interdit de mettre sur le compte exclusif de Rome
les aménagements en fait non datés, décrits par Baradez sur le pied-
mont saharien de l'Aurès en inclus dans la zone du limes.Rappeler
l'existence de ces noyaux de vie sédentaire ayant leur dynamisme
propre et liés aux montagnes dominant le prédésert depuis les monts

934
du Hodna jusqu'au Jebel Nefousa, paraissait utile avant d'évoquer la
diffusion avec les mêmes techniques, d'établissements agricoles d'é-
poque romaine telle qu'elle peut ~tre prouvée dans le sud tunisien
par exemple ou dans la région des oueds en Tripolitaine (REBUFFAT
1977 : 357-399, 407-408). Il faudrait y ajouter bien sûr les oasis
des Ziban, du Jérid et du Nefzaoua où les chefs libyens avaient dé-
jà leurs yyrgoi (Diod. Sic. III, 49, 3) ainsi que le lointain Fezzan
intensément cultivé gr~ce à son réseau de foggaras et sur lequel s'
appuyait la puissance des Garamantes (DANIELS 1970 : 36-44).

Réciproquement, aucune preuve archéologique ni épigraphique ne

permettra jamais d'affirmer qu'en arrière du limes, la mise en cultu-
re consécutive à l'occupation romaine ait été si complète qu'elle n'
ait laissé subsister aucune possibilité de transhumance; c'est tout
le contraire qu'on est en droit de supposer quand on considère com-
bien l'agriculture surtout céréalière que pratiquent du reste les
nomades et l'élevage extensif représentaient depuis le Tell jusqu'au
prédésert deux activités complémentaires dans le système agro-pasto-
ralitraditionnel (BOUKHOBZA 1977 : 207-219). Cette complémentarité
explique les multiples déplacements saisonniers en hiver en direction
des p~turages sahariens, en été vers le Tell pour la 'achaba; à cette
occasion se nouent entre groupes parfois très éloignés entre eux,
des relations régulières justifiées par l'appel de main d'oeuvre pen~

dant la moisson, l'utilisation des chaumes et les échanges commer-

ciaux dont les nomades étaient les agents (CAPOT-REY 1941 : 171-
186).

Ainsi convient-il de ne pas durcir à l'excès l'opposition des

genres de vie de part et d'autre du limes mais d'admettre comme pro-
bable que celui-ci, loin d'être une barrière hermétique entre le Tell
et la steppe ou bien la montagne et le prédésert suivant les cas,
étaient plus vraisemblablement un moyen de contr~le et de régulation
des déplacements et des échanges par lesquels s'exprimait cette

935
nécessaire symbiose entre régions d'économies complémentaires. Attri-
buer à Rome les moyens de modifier en profondeur la nature même des
choses serait un anachronisme; sa domination s'est coulée dans des
structures préexistantes : on la retrouve partout marQuée aux lignes
de clivage de l'économie et de la société traditionnelles. Elle se
manifeste aussi bien dans l'organisation des marchés au saltus Be-
ggensis (C.I.L. VIII, 270), dans la perception du tarif de Zaraï QUe
dans la création des camps le plus souvent établis à la rencontre
entre groupes différents à Besseriani entre montagnards transhu-
mants et semi-nomades du prédésert, entre ces derniers et les noma-
des sahariens à Bu Njem et à Remada (EUZENNAT-TROUSSET 1978: 142).
Le même effet de sédentarisation ne manQuait pas de se produire au-
tour de ces implantations au prix sans doute d'un abaissement rela-
tif des "valeurs nomades" (CAPOT-REY 1962 331-335).

C'est cette conception plus réaliste QUe la vision purement

coercitive autrefois présentée QUi est en voie d'être admise par
Benabou lui-même malgré la persistance déjà signalée chez lui de la
thèse du refoulement et gr~ce aux travaux les plus récents, notam-
ment dans le chapitre fort bien venu consacré au thème du nomadisme
dans le bel ouvrage de J.M. Lassère (LASSERE 1977 : 348-362) ainsi
QUe dans une étude très éclairante donné.e par Ch. R. Whi ttaker où le
même problème est abordé de façon neuve à la lumière de l'ethnologie
(WHITTAKER 1978 : 331-362).

C'est également en tenant compte de ces données Qu'il faudra

interpréter sans dramatiser la coupure QU'elles représentent, les
clausurae du Sud tunisien. Si une incertitude demeure au sujet de la
datation sans doute tardive du fossatum du Tebaga, on peut tenir pour
probable celle des murailles du Chareb (TROUSSET 1978 : 167-175) :
construites sous le règne de Trajan en liaison avec le cantonnement
des Nybgenii (C.I.L. VIII, 22787, 22788) entre le territoire de Cap-
sa (Gafsa) et les rives du Chott el Fejaj, c'est à dire à la limite

936
du prédésert, elles barraient tous les passages à travers la monta-
gne aujourd'hui encore fréquentés par les semi-nomades du Nefzaoua
pour se rendre dans leurs terres de cultures collectives du Bled Se-
gui. En dépit de l'aspect impressionnant qu'a gardé le plus impor-
tant de ces ouvrages, celui de Bir oum Ali, il ne s'agit pas à pro-
prement parler d'une défense militaire car aucun camp ou castellum
de quelque importance n'a été reconnu à proximité et sa fonction
était moins d'interdire tous les déplacements que de les rabattre
pour les filtrer vers quelques passages obligés comme celui de Bir
oum Ali où la présence d'un guichet aujourd'hui effacé par la piste
confirme cette interprétation. Une explication séduisante inspirée
par les descriptions contemporaines de Despois et de Capot-Rey rela-
tives au contrôle de la 'achaba par l'administration française à l'
entrée du Sersou est suggérée par l'étude de Whittaker (CAPOT-REY
1941 : 172; DESPOIS 1958 : 224; WHITTAKER 1978 : 350). Elle consis-
te à voir dans cette muraille la limite d'une zone d'attente (waiting
zone) pour les troupeaux et leurs bergers, voire la main d'oeuvre
sai~onnière en fonction du calendrier des récoltes. Interprétée de
cette manière et compte tenu des indices de sédentarisation recueil-
lis dans cette m~me région, la clausura de Bir oum Ali serait à la
fois le signe et l'instrument de la conversion du semi-nomadisme des
Nybgenii en une simple transhumance (LASSERE 1977 : 360).

* * *
En conclusion on pourra se demander si l'erreur d'optique rele-
vée encore dans des études récentes ne provient pas en fait d'une re-
présentation très stéréotypée du nomadisme reprise d'Ibn Khaldoun
pour des raisons idéologiques (BOUKHOBZA 1977 : 208). Or le problè-
me des rapports en+,re nomades et sédentaires comporte de multiples
paramètres écologiques, socio-économiques, historiques dont aucun ne
doit ~tre négligé (WEEXTEEN 1975 : 195-198)

937
Pl. 63. 1 La muraille de Bir oum Ali
 Il faut tenir compte également du fait que la première inter-

prétation du limes a été élaborée de Donau à Baradez par des militai-

res qui, en fonction des conceptions stratégiques du temps et en
pensant à d'autres frontières, ont eu tendance à s'exagérer la pres-
sion extérieure à laquelle était soumise celle dont ils recherchaient
les traces. D'où peut-~tre une certaine impression de trompe-l'oeil
à la vue du schéma qu'ils nous ont laissé de ce dispositif stratégi-
que quand on sait que plutet qu'un moyen de combat, il n'était qu'un
moyen de contrele.

Sans négliger pour autant la menace pesant à basse époque sur-

tout, sur les villes cetières de Tripolitaine et sur l'isthme tuni-
sien, on conviendra que ce n'est pas le long de cette frontière "iso-
bare" que se creuseront les cyclones qui emporteront l'empire afri-
cain de Rome.

Bibliographie

BENAEOU, M. 1976. La résistance africaine à la romanisation. Paris.

BOUKHOBZA, M. 1977. Nomadisme et crise de la société pastorale en

Algérie. Quelques points de repères historiques et socio-économiques
: Actes du Colloque international de l'Institut de Recherches Médi-
terranéenne. Sénanque, mai 1976 207-219.

BOURGEOT, A. 1972~ Nomadisme et sédentarisation : le processus d'in-·

tégration chez les Kel Ahaggar. Revue de l'Occident Musulman, 11:
85-97 ..

CAMPS, G. 1960. Massinissa ou les débuts de l'histoire. Libyca, 8

1-320.

CAPOT-REY, R. 1941. La migration des Saïd Atba ou "la Zénétie res-

suscitée". Revue africaine, 85 : 170-186.

940
CAPOT-REY, R. 1962. Etat actuel du nomadisme au Sahara : Les problè-
mes de la zone aride (UNESCO, Recherches sur la zone aride, 18).

DANIELS, C.M. 1970. The Garamantes of Southern Libya. Stoughton.

DE PLANHOL, X. 1968. Les fondements géographiques de l'histoire de

l'Islam. Paris.

DESPOIS, J. 1958. L'Afrique blanche, 1, l'Afrique du Nord. Paris.·

EUZENNAT, M. 1976. Les recherches sur la frontière romaine d'Afri-

que (1974-1976) : Akten des XI. Internationalen Limeskongresses.
Budapest.

EUZENNAT, M. et TROUSSET, P. 1978. Le camp de Remada. Fouilles iné-

dites du Commandant Donau (mars-avril 1914. Africa, 5-6 : 111-190.

GSELL, S. 1933. La_Tripolitaine et le Sahara au IIIe siècle de notre

ère. M.A.I., 43 : 149-166.

GUICHONNET P. et RAFFESTIN, C. 1974. Géographie des frontières.

Paris.

KOTULA, T., 1974. L'Affaire des Emporia : problème d'histoire et de

chronologie. Africana Bulletin, 20 : 47-61.

LASSERE, J.M. 1977. Ubigue populus. Peuplement et mouvements de po-

pulation dans l'Afrique romaine de la chute de Carthage à la fin de
la dynastie des Sévères (146 a. C. - 235 p. C.). Paris.

LAWLESS, R.I. 1972. L'évolution du peuplement, de l'habitat et des

paysages agraires au Maghreb. Annales de Géographie, 81 : 450-464.

LESCHI, L. 1942. Rome et les Nomades du Sahara central. Travaux de

l'Institut de Recherches sahariennes, 1 : 47-83.

LEVEAU, P. 1977. L'opposition de la montagne et de la plaine dans

l'historiographie de l'Afrique du Nord antique. Annales de Géogra-
phie, 86 : 201-205.

941
PICARD, G. 1959. La civilisation de l'Afrique romaine. Paris.

REBUFFAT, R. 1976. Une zone militaire et sa vie économique. Le limes

de Tripolitaine. Armée et fiscalité dans le monde antique.Paris,
Colloque CNRS 936: 395-419.

SALAMA, P. 1973. Un point d'eau du limes maurétanien. Maghreb et Sa-

hara, études géographiques offertes à Jean Despois. Paris, 339-349.

SALAMA, P. 1976. Les déplacements successifs du limes en Maurétanie.

Akten des XI. Internationalen Kongresses. Budapest.

THEBERT, Y. 1978. Romanisation et déromanisation en Afrique : his-

toire décolonisée ou histoire inversée? Annales Economies Sociétés
Civilisations, 33, 64-82.

TROUSSET, P. 1978. Les bornes du Bled Segui, nouveaux aperçus sur

la centuriation du Sud tunisien. Antiquités africaines, 12 : 125 -
177.

WEEXSTEEN, R. 1975. Révolution agraire et pastoralisme. Les problè-

mes akraires au Maghreb. CNRS, Paris.

WHITTAKER, C.R. 1978. Land and Labour in North Africa. Klio, 1, 2

331-362.

942
 Summary

All the recent research into the limes of the Roman provinces
of Africa has shown the importance of nomadism in explaining both
the degree of Roman penetration and the organisation of a frontier-
zone in the regions bordering the Sahara. The ancient writings
which refer to this phenomenon are greatly lacking in precision, the
inscriptions are few and difficult to interpret, all of which poses the
problem of the Roman conceptions about nomadism and semi-nomad-
ism in Africa, given even that these conceptions were subject to
change over the years.
The view of the anarchie wanderings of the pillaging inhabitants
of the Sahara is also encountered in certain parts of contemporary
historiography which ·appear to ignore the reality of nomadism às a
coherent system of territorial occupation and as a social network
operating over long distances, at times generating friction, at times
cohesion. ·
Since no other source of information is available for the period
in question, we are obliged to refer to the data provided by ethnogra-
phy or human geography and to re-interpret the archaeological re-
mains discovered just under a century ago in the limes zone in the
light of this.
With the addition of certain, hitherto unknown, Roman construc-
tions from southern Tunisia to this controversial dossier of new ele-
ments, we shall show that the aim of the Roman administration was
less to prohibit seasonal migration than to control its course with
regard to its distance and duration.

943
 64. UN NOUVEAU TYPE D'UNITE
CONNU PAR L 1 EPIGRAPHIE AFRICAINE
Yann Le Bohec

L'examen d 1 une inscription publiée en

1925 (1) et trouvée à Kherbet ouled Arif (Lambiridi)
servira de point de départ à notre réflexion.En voici
le texte :

A.
1
vDD-NN·AVREL·(-,,
~~--------------~------~----
""\r,
i1
î 1
/"

! P RA E P· A E Q ·A L · p. P E I·A'
1
1
- - .

)dQ N C O. L L A T 1S A l

! ~1 1\ r 1 r 1 0

B• (ascia) A VS SC Il DEO
VOTVMSOLVITI

c.
~L PRO·SEETV N
~
COMMELIIONIBVS
1

SVB-CVRA·POMPETH o
l!Wf P P· N N EA 1 V L lW\
/\V

945
 L 1 éditeur ne comprenait que trois élé-
ments : il s'agissait d 1 une dédicace en l'honneur de
Carin et Numérien (A,1);il y était question de soldats:
une aile et.des commilitones sont mentionnés (A,3;C,2);
enfin,une somme d 1 argent avait été rassemblée (A,4).Si
ce troisième point semble plus contestable,il n 1 y a pas
à revenir sur les deux premiers : les seuls co-empereurs
appelés Aurelii et dont le cognomen de l'un commence
par 11 C11 sont bien Carin et Numérien.Mais on peut aller
plus loin,si l 1 on accepte l'hypothèse que deux de ces
trois fragments constituent un tout cohérent et qu'on
peut les assembler ~ conformément à une suggestion de
M.H.-G. PFLAUM,il faut admettre que le morceau B appar-
tient à un autre texte,car il n 1 a pas sa place ici.Re-
marquons que cette inscription ,à la l.1,porte un cep,
insigne du centurion : une ascia,symbole funéraire,ntau-
rait rien à faire sur une dédicace (2).
A la première ligne,il y a sans doute
un souhait abrégé : gNv]v(ictissimis);les deux empe-
reurs sont déjà inuictissimi sur un autre texte de Nu-
midie (3),et ont également fait 1 1 objet d 1 une acclama-
tion (4).En A,l.3,le nom de l'aile est celui,bien con-
nu en Afr i que , de 1 1 a 1 ( a) p ( ri rn a ) E. [an n (on i or u m)] ( 5} , et
en B,à la fin de la 1.2,nous avons peut-être celui du
p r a e po s i tu s , le co g no rn en P r o s ~ e {s.)~ ~s) ( 6 ) ; r erna r quo n s
d'ailleurs que le numéro de l'aile des Pannoniens se
trouve parfois écrit en lettres et en abrégé,et non en
chiffres.En outre,à la 1.4,il est fait mention d 1 un ~­
merus collatus : ce type de corps vient d 1 être ·identifié
récemment (?);ajoutons que la forme aeguites pour egui-
~ est attestée en Afrique dès 1 1 époque sévérienne(B).
Enfin,en C,l.4,on peut penser qu'il y a le nom du gou -
verneur M.Aurelius Oecimus,devenu célèbre pour l'ample
moisson de dédicaces au 1 il a laissées en Numidie (9);
qu 1 a-t-il fait faire ici ? Peut-être un temple,comme à
Markouna (Verecunda) (1D);nous ne le savons pas.Mais
nous pouvons proposer une lecture de ce texte :

946
 [Inu]u(ictissimis) dd(ominis) nn(astris duobus) ~-
f
r e1 ( i i s ) [ar i na !!! a
Nu rn er i an o 1 f:Te rn p 1 u rn? •• ·] Pro s-
re[c]~[s~ praep(ositus) aeg(uitum) al(ae) p(rimae)
.E.Gann(oniorum)]~ camm~li~onibus 1 [et a] eg(uitibus)
n ( urne ri ) co 11 a ti ~ [ u] 2, [e
e c ( u ni a ) f e c ( i t ) ,] ~ cu ra
J J,
Po rn p e~ J 1 [ ... M• Au r el i u s De c ~ ~s p ( r a es es ) p ( rou i n-
ci a e) N( u rn id i a e) , [ • ":.]
Cette interprétatian,si elle est accep-
tée,offre une meilleure compréhension du texte et pré-
sente trois centres d 1 intérêt : d 1 abord,elle fournit
une nouvelle pièce à ajouter au dossier de M.Aurelius
Decimus;ensuite,elle montre que l 1 aile des Pannoniens
se trouvait encore en Afrique sous Carin et Numérien,
alors que sa trace était considérée comme perdue à par-
tir du milieu du IIIème s. (11);enfin,elle mentionne un
numerus callatus.

La question qui se pose maintenant est

de savoir ce qu'est un numerus collatus.Avant d 1 en cher-
cher une définitian,il ne sera pas inutile de rappeler
comment a été découvert ce type d 1 unité.Une inscription
provenant de Si Aoun,dans le sud tunisien (12),mention-
nait un~( ) ~( ),dans lequel on avait voulu voir
un n(umerus) col(onorum) (13):
sa 1 u te I p: pp ( er at ar u rn) .!J.!l ~~ (ost r or u m tri u m)]j
.!:!:..!?. fh.
Septimi Seueri Pertinac(is) 1~ M.Aureli Antonin!,
Au 9 9 ( us ta ru m duo r u m) , [ !,.! 1.. Sep ti mi ((Geta e ~, Ca es ar i s •
Q.Anicius Faustus,ca(n)s(ul) de(signat~~~: jpraesi-
dium poni iussit ~ 1 cura Aemili Emeriti,dec(urio..;.
~) al(ae),jpraepositi coh(ortis) 11. Fl(auiae) Jl-
frprum) ~ ~( ) ~( )•
Mais un texte de Bu Njem,trouvé et publié par R.REBUF-
FAT,rendait cette lecture impossibl8,et commandait de

••• consecrauit M[ •• J,a

lui substituer celle de n(umerus) cal(latus)
Felixs,(centurio) [le-
g(ionis) lll Aug(ustae) lw(iae) u(indicis) Seueria-
(14) :

ti
n a] ~' p ( r a e ) p ( a s i tu 5 ) u 8 x 11 ( at i 0 ni 5 )] 1 m u 8 xi 11 a-

947
 tionem leg(ionis) lll(Aug(ustae~ ~~ numerum ~­
latum fac(iendum) cur(auit).
Ainsi,il n'y a jamais eu en Afrique de numerus colonorum.
Pour la question de la définition,nous
avons deux termes à examiner.Numerus ne présente pas
trop de difficulté,bien que le mot ait deux acceptions
différen~es : soit il s'applique à tout corps de troupe
qui n 1 est ni une légion,ni une aile,ni une cohorte (15),
soit,à partir de Trajan ou Hadrien,il peut également dé-
signer une unité ethnique de caractère barbare (16).Pour
le participe passé du verbe confero,collatus,on a la si-
11
gnification de rassemblé 11 (17) .Mais que peut-il y avoir
là derrière ? Pour essayer de le savoir,regroupons notre
documentation qui est constituée de quatre textes les
trois mentionnés plus haut,de Kherbet ouled Arif (Lambi-
ridi),Si Aoun et Bu Njem,et un dernier d 1 El-Mahder (Ca-
~),dont l'interprétation est très difficile (18).
(Pro sa 1 ut e I rn ppp ( er at or u rn) Ca es s s (ar u rn tri u rn) ~ j ~L •
Seetirni 5 eu dl ri et 1'1. Au rel ( i) 1 Anto ni ni et 1 [(;'~~~~­
:~ ~:J] J~:~~ o )] (~) Au 999 ( ustorum tr ium), 2.! r.Iuli ae
J
Au g { u s t a e ) , !!!.ê.l t r i ( s ) Ge a s t r a r u ( rn ) j e t Au g g ( u s t o r u m
duorum), tatiusgue 1 domus diuinae, j dedicante Subatia-
œ 1 Proculo,leg(ato) Auggg(ustorum trium),, ~ ..s!!-
L!0 1
C.Iuli Paulini,(centurionis) cah(ortis) lJ ~­
r(orum)J ~ _Πcollato ..ΠCXXVI, 1 kal(endis) Aug(ustis)
d ( 0 min 0 ) n(0 s t r 0 ) An t 0 nin 0 ll1. 1 aet Geta llTI c0 ( n)-
s(ulibus)~ ~ l ~·
C 1 est la 1.11 qui se trouve au centre du
débat.Dn en a en effet donné une version que nous appel~

lerons "économique" : uex(illatio) n(ummo) collato ~­

mero) CXXVI (19);il faudrait alors traduire : 11 sous la
responsabilité de C.Julius Paulinus,centurion de la Deu-
xième Cohorte des Maures,le détachement,ayant collecté
des fonds pour un total de 126 ••• ,a acquitté san voeu 11 •
Mais cette lecture ne va pas sans difficultés.En effet,
les inscriptions africaines mentionnant des dépenses(20)
ont en général un formulaire stéréotypé : quand la som-

948
me n'est pas indiquée,on trouve en général aere collato,
mais pas nummo callato (21);quand elle est précisée,cela
se traduit par une expression en quatre termes,à savoir
la préposition~· la précision de 1 1 unité monétaire,le
montant de la somme et le substantif nummum,normalement
abrégé en ~,et placé à la fin.Toutefois,la fréquence de
ces mentions fait que l 1 on sous-entend souvent l'un de
ces mots,à l'exception bien sûr du chiffre,encore que
celui-ci se voit souvent privé de l'indication des mil-
lisrs.Mais la formule que 1 1 on nous propose ici est pour
le moins exceptionnelle dans l'épigraphie africaine : la
tournure diffère des formules habituelles,il manque deux
des quatre éléments usuels (~et l'unité monétaire),et
nummum serait placé en tête.De plus,la somme,sans être
invraisemblable,paraît modique s 1 il faut lire 126 -ce
qui,d 1 ailleurs,n 1 est pas un chiffre rond comme à l'accou-
tumée - et très importante s 1 il faut sous-entendre mille.
Et,en outre, le spécialiste des prix africains,R.DUNCAN-
JONES,n1a pas jugé devoir intégrer les données de ce tex-
te à ses listes (22).
Alors,sans toutefois éliminer catégori-
quement l'interprétation économique,on peut en proposer
une autre qui serait militaire.Car,cela nous semble es-
sentiel,on connaît maintenant l'existence des numeri col-
lati .Ajoutons que les inscriptions de Si Aoun et El-
Mahder présentent dans leur composition un remarquable
parallélisme : l'une et 1 1 autre commencent par la for-
mule~ saluts Imperatoris (23),suivie de la mention du
légat et du sous-officier qui a pris la responsabilité·
de la réalisation (~ cura);de plus,ces deux textes sont
très proches dans le temps : le premier date de 198,le
second de 208 (24).Nous proposons de lire : uex(illarii),
n(umero) collato n(umero) CXXVI, ••• u(otum) s(oluerunt)
l(ibens) a(nimo),et de comprendre : 11 les membres du déta-
chement,constitués en numerus collatus,au nombre de 126,
••• ont acquitté leur voeu avec empressement 11 .De fait,le
verbe soluit,-erunt appelle un sujet qui,dans ce texte,
ne peut être que uexilla.tio,-arii;ensuite,le groupe~-

949
merus collatus,qui existe,peut fort bien être employé
comme ablatif absolu;enfin,numero avec le sens de "au
nombre de" est d'un emploi très classique, - il vient
d'être étudié chez Cicéron (25).
Ainsi,on voit un peu mieux ce que peut
être un numerus collatus : il s 1 agit d 1 une unité compo-
sée d 1 hommes regroupés,peut-être à partir de plusieurs
camps ou corps de troupe,pour une mission définie.On
pense évidemment à une sorte de uexillatio,mais avec
des caractères originaux.Le numerus collatus,en effet,
accompagne parfois un détachement légionnaire (26),une
aile (27) ou une cohorte (2B),mais à El-Mahder i l sem-
ble opérer isolément (29) et à Lambiridi il ne comprend
que des cavaliers.De faible importance numérique (cent-
vingt-six hommes une fois),il est confié à un centurion
légionnaire (3D),auxiliaire (31) ou à un décurion d'ai-
le portant le titre de EEaepositus (32).Enfin,l 1 exis-
tence des numeri collati se limite au IIIème s. on
trouve leur trace dès 198 à Si Aoun (33),en 208 à El-
Mahder (34), sous- Sévère Alexandre à Bu Njem (35) et sous
Carin et Numérien à Kherbet ouled Arif (36).Cette dis-
persion géographique et chronologique rend difficile-
ment acceptable 1 1 idée qu'il y ait eu une seule unité
appelée numerus callatus : de fait,il semble bien que
nous ayons là des troupes constituées pour une circons-
tance précise,l'installation d 1 un praesidium à Si Aoun
(37),la construction ou la consécration d'un lieu sacré
à Bu Njem (38) et Kherbet ouled Arif.
Il reste maintenant à espérer que,gr~ce
à de nouvelles lectures d'inscriptions déjà connues,et
à des découvertes faites dans d'autres provinces,il se-
ra possible d'approfondir notre connaissance des numeri
collati,qui nous sont apparus comme un cas particulier
de uexillatio.

950
Notes.

La présente communication n'aurait ja-

mais pu stre présentée sous la forme qui est la sienne
si elle n'avait bénéficié de nombreuses remarques de
Monsieur M.Le Glay,Professeur à l'Université de Paris,
et de Monsieur H.-G.Pflaum,Professeur à l'Ecole Prati-
que des Hautes Etudes.Qu'ils soient ici remerciés.

1. Colin 1924-1925 257-259~

2. Il est assuré que l'ascia a été en usage dans l'A-

frique romaine : J.Carcopino, Notes sur les épita-
phes africaines à ascia, Mél.M.Andrieu(1957) 79-
81.En dernier lieu,mais avec un exemple pris en
Maurétanie : R.Rebuffat, L 1 ascia de l'épitaphe de
Sex.Iulius Iulianus, Bull.Archéol.Algér. (V,1971-
1974) 195-206.5i la signification précise de ce
symbole a été discutée,son caractère funéraire n'a
jamais été mis en cause : M.Le Glay, Libyca(II,
1954) 492 et ibid.(III,1955) 395;chronique annuel-
le de P.-M.Duval dans la Rev.Et.Anc.

3. f.,4222Q

4. f . ,4221 ..
5. Manna 197D
11,19-20,28-29,37,60-62;Le Bohec, art.
à paraître;Trausset 1978 559-576.0n trouve une a-
bréviation analogue dans A.E.,1948,214,1.5 ala
pr(ima) Pan(noniorum). ---

6. I.Kajanto, The Latin Cognomina(1965), 354 et 407.

7. Rabuffat 1966-1967 97-98, d 1 où A.E.,1972,677, et

1969 206;Euzennat 1972 19 n.1;Trousset 1974
120;Le Bohec,art. à paraître.

B. c.,B793 = 18019 = A.E.,1929,70;il est vrai que,dans

ëe dernier texte,on peut lire : [ uexill)a(tio) ~­
g(uitum).

9. H.G.Kolbe,Die Statthalter Numidiens von Gallien bis

Konstantin, Vestigia (IV,1962), 21-2B;B.E.Thomasson,
R.E.,Suppl.{XIII,1973) 321;J.Marcillet-Jaubert, A
propos de M.Aurelius Decimus, Z.P.E.(XIV,1974) 249-
251.

10. f.,4221.

11. J.Baradez, Limes-Studien = Carnuntina (III,1956) 6.

12. I.L.Tun.,1;Trousset 1974 120.

13. Il y a bien eu dans 1 1 armée romaine une ala cola-

951
 norum : Arrien, Ect., I;Inscr.gr. ad res rom. Ber-
tin.,III,1144;A.E.,1907,57, 1926,150 et 1930,1 B;
C.,B934 et p.973 =Dessau, I.L.S.,1400;C.Cichorius,
R7E.(I,1894) 123B;F.Cumont, SyJia(V,1924) 352;J.
Carcopino,Rev.archéal.(1924,II 325;Manna 1970 63.
Mais il faut tenir compte de la différence existant
entre une aile et un numerus.

14. Rebuffat 1966-1967 97-98 et 1969 206 n°3 (A.E.,

1972,677);Euzennat 1972 19 n.1;Trausset 1974 120.

15. H.T.Rowell, R.E.(XVII,2,1937) 1328,et Yale Cl. St.

(VI,1939) 74-75.Sur les num~ri,voir en outre : F.
Vittin~hoff,Historia (1,1950) 389-407;J.C.Mann,
Hermes(LXXXII,1954) 501-506;H.Callies, Ber.rom.-
9 e r m• Ko mm • ( XL V, 1 9 6 LI ) 1 3 0 - 2 7 7 ; 1~1 • 5 p e i d e 1 , Au f s t i e g
und Niederg. rom. Welt(II,3,1975) 202-231 et Guards
of the Roman Armies(1978) 22-23.
16. H.T.Rowell, art.cité 1329.Voir n. précédente.

17. Th.l.l.(IV,1906) 173 et suiv.On a une unité dont

le nom est ainsi composé,la cahors III collecta :
A.E.,1957,340 et Euzennat 1972 19 n:Ï:

18. C.,4323 = 18528;R.Saxer, Untersuchungen zu den Ve-

xillationen des romischen Kaiserheeres von Augus-
tus bis Didkletian,Epigraphische Studien(I,1967)
104 n°310.

19. H.-G.Pflaum, Carrières(II,1'960) 649;le même,à la

ligne précédente,préfère la lecture cohortis (mil-
liariae) Maurorum.Les éditeurs du C. avaient chOI-
si la même version que H.-G.Pflaum:mais en mettant
un point d'interrogation après nummo.R.Saxer,pass.
cité,ne donne ce texte que jusqu 1 au mot uexilla-
..ilJ?., et ne prend pas parti pour la sui te·.
20. Duncan-Jones 1974 396 p. reprend plusieurs art.,en
particulier,pour l'Afrique,P.B.S.R.(1962) 47-115
et P.B.S.R.(1963) 159-177.0n trouvera la liste des
indications de prix dans l'épigraphie africaine
aux pp. 89-114.

21. c.,4599,5363,5365,5368,6710,6711,9663,14291,14372,
14612,19697,23226,25850,25935,27828,25934;Th.l.l.
(IV,1906) 176.Voir cependant ~,2438 : pecun1a
conlata (mais pas nummo conlato).

22. Duncan-Jones 1974 89-114.

23. Dans l'inscription d'El-Mahder (C.,4323 = 18528),

il s'agit d'une restitution,très vraisemblable
au demeurant.

24. Le légat Q.Anicius Faustus est consul désigné :

952
 B.E.Thomasson,Die Sttatha1ter der rëmischen Provin-
zen Nordafrikas(II,1960) 197-201 et R.E. Suppl.
(XIII,1973) 319;cdt Donau,B.C.T.H. (1909) 42-43,
par erreur : novembre 197.

25. T.Scarpat, Paideia(XXV,1970) 213-214.Ajoutons un

exemple épigraphique africain :"des statues au nom-
bre de trois 11 ,num.III (B.C.T.H.,1911 CCXL v. A.E.,
1911,97 et 1912,17).

26. A.E.,1972,677.

27. Colin 1924-1925 257-259.

28. I.L.Tun.,1.

29. f..,4323 = 18528.

30. A.E.,1972,677 : (centurie) [1eg(ionis) III

tae)~l ·iae) u indicis) SeuerianaJ~' rae
uex ill ationis •
-
31 • f.,4323 = 18528;Colin 1924-1925 257-259.

32. I .. L.Tun.,1.

33. Voir n. précédente.

34. f.,4323 = 18528.

35. A.E.,1972,677.

36. Colin 1924-1925 257-259.

37. I.L.Tun.,1 : Q.Anicius Faustus,co(n)s(ul) de(sig-

natus), praesidium poni iussit.

38. A.E.,1972,677 : ••• consecrauit M. [ •• ]us Felixs,

(centurie) [le ionis III Au ustae}JI iae) u(in-
dicis) Seuer~anaJ~' rae os~tus ~ '11 ationis~.
Une autre inscription de Bu Njem compor ait peut-
être la mention d'un numerus collatus,mais i l a pa-
ru préférable de ne pas l'utiliser ici car i l s'a~
git d'une restitution totale,l'expression n'ayant
pas été conservée (ce texte,encore inédit,sera pu-
blié par R.Rebuffat dans Libya Antigua).

SUmmary

A new examination of already published texts provides us with

four African inscriptions mentioning numeri collati. These appear
to be units similar to vexillationes, but made up of only a small oum-
ber of soldiers brought together for a precise mission, the men
having been recruited from different troops or camps. The inscr~p­
tions are dated from the time of Septimius Severus to that of Carinus
and Numerianus.

953
Abréviations et bibliographie.

L'existence des numeri collati venant

tout juste d'être établie,la bibliographie sera néces-
sairement sommaire.

A.E. L'Année Epigraphique.

a.c.T.H. Bulletin du Comité des Travaux Histori-

ques.

c. Corpus inscriptionum latinarum,vol.VIII.

Colin 1924-1925 J.Colin, Etude sur une inscription tri-

ple· de Lambiridi, Atti Pontif.Accad.
romana archeol.(III,1924-1925) 257-259.

Duncan-Jones 1974 R.Duncan-Jones, The Economy gf the

ROman Empire(1974) 396 p.

Euzennat 1974 M.Euzennat, Comptes-Rendus Acad.Inscr.,

(1972) 19 n.1.

I.L.Tun. A.Merlin,Inscriptions latines de la Tu-

nisie(1944). ·

La Bohec Y.Le Bohec, Les auxiliaires de la Troi-

sième Légion Auguste,A propos du livra
de M.G.Manna,B.C.T.H. (sous presse).

Manna 1970

R.E. A.Pauly et G.Wissowa,Real-Encyclopadie

der klass. Altertums.

Rebuffat 1966-1967 R.Rebuffat, Bu Njem 1967, Libya Ant.

(III-IV,1966-1967) 97-98.

Rebuffat 1969 R.Rebuffat, Deux ans de recherches dans

le sud de la Tripolitaine, Comptes-Ren-
dus Acad.Inscr.(1969) 206.

Th.l.l. Thesaurus linguae latinae.

Trousset 1974 P.Trousset, Recherches sur le limes tri-

politanus (1974) 176p. et 36p1.

Trousset 1978 P.Trousset, Le camp de Gemellae, Akten

des XI. internat. Limeskongresses {1978)
559-576.

954
Addendum :

Monsieur R.Rebuffat a pu relire l'inscription de Bu

Njem mentionnée plus haut (Année Epigraphigue,1972,

677);en outre,il en a retrouvé divers morceaux épars

dans la forteresse.Il faut désormais la dater de Ma-

ximin le Thrace,et non,comme cela avait été fait

jusqu'à présent,de Sévère Alexandre;on sait également

qu'il s'agit de la consécration d'une~ Cerei.

 65. REMARQUES ONOMASTIQUES
SUR LA LISTE MILITAIRE DE VEZEREOS
(ILAfr 27)

Jean-Marie Lassère

Gr~ce au dévouement de Si el-Hachemi· ben Çouf, la lis-

te mutil~e des soldats de Vezereo~ a trouvé sa place dans
les dépendances de la Maison de la Culture de Qbili(Ké~~
Prouvée le 2o mai 1919 par le colonel Donau au centre de
la cour du fort de Bir Ghzen ( ou Sidi Mohamed ben A~ssa)
qui domi~e l'Oued el-Hallouf, elle fut transportée dans ~
bordj du commandant français du cercle de Qbili puis, ce
b~timent devenu un hOtel, dans la Maison de la Culture.Ce
voyage n'a pas été sans dommages pour notre document, en
particulier pour les premières lettres de la ligne 6 de la
race principale, où se trouvait précisément le nom du ffite;
en fait cette ligne était déjà mutilée en 1919 : le V ~

tial n'était que partiellement lisible. A.Merlin, dans sa

communication à l'académie dea Inscriptions, déclare mroir
discerné "peut-être le haut d~un trait oblique dirigé de
droite à gauche, c'est-à-dire qui se présente comme le
sommet de la seconde branche d'un V". Cette lecture avait
été vérifiée par P.Trousset lors de sa première visite à
2
la pierre • La forme Vezereos est donc à préférer à Beze-
reos, qui est pourtant dans l'usage à cause des leçons de
l'Itinéraire d'Antonin et de la Notitia Dignitatum. Mais
il s'agit là d'une évolution du phonème Q vers une bllaooa-
1 e cons tri cti ve ( b) bi en connue des phoné ti ci ens et f ré-~
quente à partir du Ile s. 3 • La graphie Vezereos apparente
notre toponyme à quelques autres noms de lieux de-l'Afri-

955
que4. Mais l' intértt essentiel du texte réside dans la lis-
te nominative des soldats qui composent la garnison.
L'inscription comportait quatre parties, aujourd'hui
inégalement conservées :
sur la face prin8ipale, une dédicace de 6 lignes par une
uexillatio tirée de plusieurs centuries de la Legio III
Augusta pour le salut des trois empereurs septimiens etde
Julia Damna;
sur la face latérale gauche, trois doubles colonnes de
noms, dont seule la seconœ est à peu près complète;
sur la face latérale droite, il ne reste d'une disposi
tian identique que deux fragments qui correspondent k la
fin de quelques lignes de la colonne I et au début de œr-
taines autres de la col. II. Les précédents éditeurs les
0 nt négligées; un réexamen récent de la pierre, avec con-
trôle sur photographie, a permis la restitution hypothét~

que de neuf noms de soldats (voir dessin);

sur la face postérieure, traces de quatre lignes, où en
tout quatre lettres sont lisibles; l'intértt onomastique
en est encore moindre.
La ligne 2 de la face principale commençait par lœ let-
tres AVGGG; le troisième G n'avait pas été martelé, (mais
n'est plus lisible aujourd'hui) , alors qu'à la fin de la
1.3 les mots ET P SEPTIMI GETAE l'ont été; après quoi l'on
à regravé, dans une place un peu trop large BRIT PAR
GERM. Théoriquement, c'est seulement de 209 à 211 que les
Trois Augustes ont régné; mais, comme l'avait observé R.
Cagnat, certaines inscriptions africaines font de Géta un
Auguste dès 199. Ces précisions sont intéressantes pourdes
rapprochements entre notre monument et d'autres listes 5 •
M~me incertitude pour le martelage : Caracalla est procla
mé Germanicus en 213, mais quelquefois plus tOt.
6
Récemment réexaminée par Y.Le Bohec , la 1.6 de la fa-
ce principale peut être lue VEZERE! PRAETEN[dit.. et] T(itus)
GEN~ ou -iiV[s••• , ce qui nous permet d'ajouter un nom à
la liste, celui de l'officier qui commandait la uexillatm
et qui sans doute était un centurion 7 .

956
 Restitutions

1 fi
1 1
\ 1v
vs
OLVS IMI,/ Vol us ( ius) Im [Pet ra tus
/N\)IVSVIC.T Mu]ndius Vict(or
N ["'-] 1 V 5 1 M PE Ajn [n) ius Impe [tratus
AOt\lS

AC'

11 (/ 1
/AN NI 7 Anni (us •••
A NT/ Ant [istius ou onius
.. ANt>ll>l
. c] andidi[ us •••
t.o VOL V// Volu [sius
Vl\/5 G]auius
Pl/5 v Cu?]pi[u) s Vi fctor
vs (plutôt qu'~lpius, car on ne
voit pas de ~race d'une bar-
re horizontale)

957
 Dans l'ensemble de la liste nominale, on ne relève pas
de noms où les diphtongues aient été simplifiées : pas d'
Elius, ni d'Emilius; pas davantage de ces petits faits de
langue qui fourmillent dans les inscriptions africaines da
IIIe s. : pas de Cresces pour Creacens, de Bictor ou deBi.-
bian1;1s; dans une inscription officielle (où de surcro1t le
site ~tait eDcore appelé Vezereos), cela parait très nor-
mal, à la différence de ce qu'on observe un siècle et de-
8
mi plus tard dans l'Album de Timgad •

LES P RAENOMIHA
Il y avait environ 300 noms de soldats sur la pier~ ce
qui représente une assez grosse garnison 9 • Mais seuls 103
de ces noms (et 112 si l'on accepte mes tentatives de res-
titution) ont subsisté. On remarque vite l'aspect unifor-
me de l'onomastique : seuls onze individus (eu plut0toou-
ze10) font conna1tre leur prénom. La proportion est bien
sdr plus forte qu'elle ne l'est 150 ans plus tard dans l'
Album de Tim~ad~ où deux individus seulement sur 263 por-
tent un prénom, ce qui au Bas-Empire n'est plus qu'une
11
survivance aristocratique • Ce pointage des prénoms offM
un intér~t supplémentaire si l'on fait une comparaison
avec quelques autres listes militaires qui, en gros, sont
à peu près contemporaines :

1. CIL VIII, 18065, de 162 p~ un seul prénom pour en-

viron 90 individus;
2. CIL VIII, 18068, de 198 : tous les soldats sont dési -
gnés par leurs tria nomina complets;
3. CIL VIII, 2556, entre 199 et 211 13 tria nomina com-
pl e ts.
4. CIL VIII, 2557, qui commence par les mêmes titulatures
imp~riales que le texte de Vezereos et est datée, in fin%
de 203 : sur 36 noms, 13 sont sans prénoma.
5. CIL VIII, 2618 (de 211-212) : sur la face b, la seule
qui soit intacte, 14 tria nomina et 2 ~ nomina sans p~

noms.
12
6. AE, 1917-1918, 29 (du début du Ille s. ) :l'état de
la piêrre écarte toute certitude, mais le prénom semble

958
d'un usage général.
7. AE, 1917-1918, 57 (m~me époque, ou peut-~tre 225, cf.
Cagnat,BCTH, 1917, p.CLXXIII): 7 prénoms sur 55.
8. CIL,VIII, 2586 (vers 225, Cagnat, .L.l.-): 41 hommesavec
prénom et 8 sans prénom.
9. ~VIII, 2564 (sous Sévère Alexandre) 4 prénoms à
peine sur une centaine.

A comparer ces textes ( et l'on pourrait en citer d'au

tres, en particulier hors de l'Afrique), on constate que
si la disparition du prénom dans les listes militaires
africaines semble bien se placer entre l'avènement de~

Aurèle et l'époque sévérienne, il serait vain de chercher

à fixer une date plus précise, puisque la liste de198 est
plus complète que celle de 162. Mais on est à court d'ex-
plication devant l'anomalie qui fait que, dans un docuœm
officiel, le prénom appara1t de façon irrégulière, car on
constate vite que ce ne sont pas régulièrement les nansdes
centurions qui sont au complet 13 • On est réduit pour l'i.m-
tant à supposer (mais sans pouvoir le vérifier à cause de
l'état de la pierre) que l'indication irrégulière du pré-
nom servait à distinguer des soldats pourvus de gentilices
et de ~gnominaidentiques. En fait certaines compara~
14
sont peu probantes , soit à cause des dégradations des
pierres, soit parce que les listes ne font conna1tre qu'
une partie de l'effectif d'une unité. On remarque aussi
que dans plusieurs des listes précitées, la place du pré-
nom a été respectée, comme pour un complément ultérieur.

LES COGNOMINA
Pour les 112 individus dont les noms figurent encore
sur la pierre, on relève 47 cognomina différents 1 5, dont
voici la liste :
Adiutor; Ammon ( ou Ammon(ius)); Antoninus; Aue tus;
Bea tus; Catullinus; Cres cens; Do na tus; Faustus; Fe 1 ix;
Fortunatus; Fuscinus; Honora tus; Ianuarius; Impetratus;
Ingenuus; Iucundus; Iulianus; Iustus; Iuuenalis; Libe -
ralis; Marcellus; Martialis; Maximus; Niger; Nouatianus ;
Primus; Priscus; Proculus; Quartus; Quintus; Rogatus; Ru-

959
·fus; Sallustianus; Salutaris; Saturninus; Saturus; Seueri-
anus; Siluanus; Successus; Verecundus; VerissimUrs; Vibia-
nus; Victor; Victorianus; Vitalis; Urbanus. Sans attein%e
celle qu'on observe dans l'Album de Timgad, cette iisper-
sion des cognomina est déjà grande et correspond à l'ab-
sence de prénom : on a le souci, particulièrement explica-
ble dans l'armée, d'individualiser chaque soldat. On ne
relève en effet que deux groupes de deux homonymes : A~s
Saturninus (II, 14 et 15; encore le~ distingue-t-on par
l'abréviation de leur grade); Iulius Donatus (II,33 et la,
9); le texte, officiel, ne fait pas conna1tre d'~gnomina
qui (dès le Ier s. 16 ) eussent pu les distinguer.
Dans l'ordre d'importance, ces cognomina se classent
ainsi (entre parenthèses, leur fréquence dans l'indexcqr-
nominum du CIL VIII)
Felix :8 (1254); 5aturninus : 7 (735); Victor : 5 (774);·
Donatus et Fortunatus : 4 chacun (362 et 425); Proculus ,
Rogatus et Siluanus : 3; Catullinus, Ianuarius, Impetra
tus; Iuuenalis, Maximus et Primus : 2; aux 5 Victores s'
ajoute aussi un Victorianus. Les cognomina qui reviennent
avec la plus grande fréquence (4 ou plus) dans notre liste,
Felix, Saturninus, Donatus, Fortunatus, ainsi que le g~

pe Victor-Victorianus, sont parmi les plus communs en Mn-

que (le dernier groupe est le plus fréquent dans l'Album
de Timgad). Les autres cognomina africains sont : Âuctus,
Crescens, Faustus, Honoratus, Ianuarius~ Impetratus, l.olla
18
tianus, Rogatus, Sallustianus • Parmi les autres, an peut
s'intéresser à Ammon, dont les formes dérivées, Amonius ,
Ammonius, Ammonianus, Hammonius, n'apparaissent que neuf
fois dans l'index du CIL VIII, dont deux dans la ville de
garnison qu'est Lambèse et six dans le port de Césarée~

a dee rapports historiques avec l'Egypte 19 •

Antoninus : si le gentilice Antonius est encore assez r~

re à l'époque en Afrique (il n'est pas sdr qu'il figure

dans notre liste), Antoninus peut avoir été choisi en sou-
venir d'un empereur Aurelius Antoninus, Antonin le Pieux,
Marc Aurèle ou Commode. Kajanto pourtant y voit un bon
20
exemple d'un ~gnomen formé sur le nom paternel •

960
 Catullinus est peu fréquent en Afrique (17 fois à l'index
du CIL VIII); dans la liste qui nous préoccupe, ce surnom
·est porté par un Euped(ius) et par un Fannius.
'
Siluanus est un cognomen tres .
popula~re se 1 on K aJan
. t o 21 ,
mais M.Leglay insiste sur l'importance, à Lambèse, du culw
22
de Silvain introduit par ~~6 militaires venus du »anube •
En e f fe t , à Vez er e o s , t ro i s s o 1 da t s (qui vi e nn en t de .1 a ga.I'-
~ison de Numidie), Bennius, Heluius et Pomponius, portent
ce ~gnomen.

Puis on aborde le groupe des ~gnomina repris d'adjec

tifs : Felix, Iucundus, Iustus, Iuuenalis, Liberalis, Ma -
ximus, Niger, Primus, Priscus, Quartus, Quintus, Rufus, 14.-
2
.. fr é quen t s d'es 1 a p é r~o
talis, tous t res . d e repu
, b 1 ~ca~ne
. . ' •

Dans les uns on trouve des idées de chance (Felix, dont le

sens est analogue, finalement, à celui de Ianuarius) chères
aux Africains, mais représentés aussi dans d'autres part~

de l'Empire. Egalement des particularités physiques ou mo-

rales, voire des- circonstances particulières, comme pour
Proculus, celui, selon Festus (225) qui est né pendant 1 1

absence de son père.

Formé sur un participe présent, on ne trouve que Cres-
cens : cette formation est encore peu fréquente à l'époque
e 24
sévérienne; elle est beaucoup plus commune au IV s. • Plus
nombreux sont les ~gnomina formés sur des participes pas-
sés : Auctus, Donatus,Faustus, Fortunatus, Honoratus, Ro-
gatus, Successus : il s'agit en effet d'un type de forma-
tion plus ancien dans les milieux humbles de l'empire 25 •
Plus rare, et m@me unique, le cognomen Salutaris, formé~
26
un nom, qui est absent des indices du CIL VIII • Enfl.n des
_QQgnomina formés par la suffixation d'autres noms : des 5;9--
gnomina spécifiquement latins suffixés en -ianus 27 et qui
deviennent Nouatianus, Seuerianus, Victorianus; un adjectif
28 d
·
su ffi x é en -1nus : F usc~nus;
· es gen t.~ 1.~ces su ff.~xes
' en
-ianus : Iulianus, Sallustius; 9 vibianus.
Il faut surtout remarquer qu'aucun ~gnomen n'est cons-
titué par un nom grec ou oriental en dehors d'Ammon, ou
par un nom sémitique ou af~icain non traduit. De même, en

961
dehors de Saturninus, on ne relève aucun nom théophore.
,>

On doit conclure à l'aspect très af~icain de la majorité

de ces ~gnomina : hormis Salutaris, tous sont connus en
Afrique.

LES GENTIL! CES

Reste à faire maintenant l'étude des soixante gentili-
ces qui subsistent sur la pierre; pour certains, qu'on~

père dans la liste suivante à l'aide d'astérisques, il s'

agit de restitutions hypothétiques.
M.Acilius; Aelius; M.Aemilius; Agrius; Annius; Antisns;
Arruntius; C.Attius; C.Aufidius; Aurelius; Bennius;+Can-
didius; Clodius; P.Clementius; Cornelius; +Cupius; Denti-
lius; Domitius; Eppius; Errius; Eupedius; Fannius; Flaui-
us; Florius; +Gauius; Q.Gellius; Geminius; Granius; H~­
us; Hostilius; Iulius; Iunius; Lucilius; Lurius; Marius ;
Mundius; Mussius; Papirius; Papius; Petronius; L.Pinarius;
C.Pompeius; Pomponius; Pontius; Porcius; Publilius; Safi-
dius; Scribonius; Sedius; Silicius; Statilius; Stertinius;
Sulpicius; A.Surius; Valerius; Varius; Veturius; Vitelli-
us; +Volusius.
Nul besoin d'insister sur les gentilices impériaux :
Aelius, Aurelius, Flavius, Iulius; on peut noter l'absen-
ce, du moi•s dans les vestiges de l'inscription, des Ul-
pii3@. Il n'est pas s~r qu'Antonius y ait figuré 3 \ On
passera vite aussi sur les gentilices empruntés à des
proconsuls de l'Africa, en sigrcalant pourtant les homony-
mie~ avec les gouverneurs les plus anciens : on relève en
effet dans la liste un M.Acilius Fortunatus, un M.Aemïtius
Felix, ainsi que les gentilices Domitius, Iunius et Volu-
sius. Il y a bien sdr l'inévitable cortège des noms e~­
mement fréquents en Afrique, mais aussi dans d'autres pro-
vinces : Cornelius, Geminius, Hostilius, Marius3~, Pompe-
ius, Pomponius, Sulpicius, Valerius ••• En dehors d'eux
quelques noms méritent une particulière attention.
Tout d'abord ceux pour lesquels on dispose de repères
chronologiques qui prouvent leur ancienneté dans la pro-
vince : Agrius, qui est attestS ~ Utique au I r s c
e • ~· '

962
pr,cis~ment par un militaire33; Bennius, connu à la fois
dans la Carthage de César et dans la confédération cirté-
enne 34 ; Eppius, représenté dans le conuentus d'Utique au
milieu du Ier s. a.c. 35 ; Heluius et Pontius, noms qu'onre-
trouve dans le mur d'amphores de Carthage, à la fin de la
37
république 36 , Pinarius, que fait conna1tre Cicéron ; Sta-
tilius, signalé en Tripolitaine au !er s. p.C, par Suéto -
ne 38 ; Veturius, attesté d2s le !er s. p.C. à Simitthus par
un vétéran sans doute d'origine italienne
39 ; ce nom pour-
tant n'est pas tr~s courant en Afrique (32 dans l'index no
minum du~ VIII).
On est aussi frappé par quelques noms, trois exactemen~

qui sont étroitement liés à une région de l'Afrique la

Cirtéenne d'abord, représentée par un Florius, dont le nom
est fréquent à Cirta et à Celtianis, et par un Antistius (
et peut-~tre par deux) dont l'origine est bien s~r Thibi-
lis. On sait que la Cirtéenne, tràs peuplée, a vu beaucouy
de ses enfants s'éloigner vers d'autres régions africaines
comme le montre la dispersion des gentilices 3ittius et
Sallustius. L'aut~e région qu'évoque le troisième nom, Au-
f·idius, est celle d'Hippone où les Aufidii, riches en ter-_
res céréalières, embarquaient les grains vers Ostie 4 ~ Ces
noms sont ainsi des jalons pour l'étude du recrutement r~

gional en Afrique.
Bn revanche, on est frappé par des noms qui sont rares
en Afrique Dentilius, attesté seulement deux fois, lune
et 1 'autre à Lamasba, une bourgade. numide dont le peuple-
ment était d'origine vétérane; Mundius, qu'on ne rencontre
qu'à Lambèse, dans la famille du vétéran Mundius Honoratu~

et à Thignica (ainsi qu'au voisinage de cette ville); 3e-

dius, gentilice dont on ne conna1t que~ exemples afri-
cal.· ns41 •. ce nom semble pra t•1.quemen t propre a' l'Af r1.que
· ,
en particulier à Cirta et à sa région, puisqu'il est i n -
connu en Italie et pour le reste n'appara1t qu'une fois~
42
Gaule Narbonnaise • Enfin Surus ou plutOt Surius sous
la première formej le nom est répertorié une fois h Arsa-
cal, et trois fois sous la seconde, à Carthage, à Thabar-
busis et aux Mouias, dans. le nord de la Cirtéenne. En de-

963
bors de l'Afrique, on le rencontre en Italie et en Gau~e ,
en particulier ~ Arles, ainsi qu'en Germanie
43
; la seule
indication d'ordre ethnique nous v.ent dun texte d'Alisca
(Szegzard en Pannonie inférieure) relatif à un soldat de~
I~re cohorte des Vindéliciens, Surius, Essimni filius· ~~
4

p eup 1 e a 1 p e .3 t r e 4- 5 d e s Ca t t e nat e s • I1 a i s c e t t e i nd i c a t i o n est

trop isolée pour nous renseigner sur l'origine du soldat
d e V e z e r-e o s •
Il s'ajoute à cette courte liste quelques noms hap~

en Af:tique : Candidius et Clementius, forgés l'un et 1 'au-

tre sur un adjectif selon une formation bien connue; mais
ils sont attestés aill;urs 46 ; Safidius, inconnu eL Italie,
se rencontre en Afrique à Thuburnica à 47 , à Calama 48 et
Celtianis 49 ; on peut ajouter, pu~q~e le soldat de Vezereos
a pour ~gnomen Ammon, que Safidius n'est pas connu en E-
gypte. Autre nom pour lequel on a des doutes, Errius, car
les fautes R pour N et R pour P sont fréquentes; il faurat
àlors lire Ennius ou Eppius (attesté dans la liste); on
conna!t pourtant à Thignica le sénateur Erius Fanius Gemi-
.
n.1.anus SO ; 1 e m~:::me.
,. - nom se 1.~ t aussl.. une f o~s
. 'a Rorne 51; el...I.U.n
......P:

Euped(ius) Cat(u]l(l:iJnus_. Ce nom, peut-ttre forgé sur l'adjec-

tif EltnÉét.oç, est si totalement inconnu qu'on en vient à
~e demander, par hypothàse, s'il ne serait pas une traduc-
'
tion maladroite du nom punique Namphamo, dont la signifi-
cation (quid aliud significat, quam boni pedis hominem? )
nous a été transmise par S.Augustin
52 •
Les gentilices et les cognomina rares en Afnque condu~

sent à se demander si les militaires qui les portent sont

des Africains ou proviennent d'autres régions de recrute-.
ment. Il convient donc de compléter cette analyse en je-
tant un rapide coup d'oeil sur quelques groupes gentilic~
..Q.Q,gnomen :
Dentilius Adiutor : il est difficile de se prononcer à
cause de la grande rareté de ce nom en Afrique, où seuls
deux habitants de la bourgade de vétérans de Lamasba le
portent. Il est tout aussi rare dans le reste de l'Empire.
Le cognomen Adiutor est représenté 49 fois à l'index cog-
nominum du CI~ VIII (avec en plus deux fois Adiutrix); il

964
se rencontre aussi ailleurs et il est impossible de se pnr
noncer.
Hostilius Salutaris est remarquable par son ~gnomen ,
jusqu'ici inconnu en Afrique, mais fréquent ailleurs, en
particulier en Italie; or Hostilius est ~ù nom r~équent

dans l'ensemble de l'empire et non dans la seule Afriqu~:

impossible, là aussi, d'en dire plus.
Q.Gellius Iuuenalis : le ~gnomen est fréquent à tr~

le monde romain; le nom Gellius n'est pas très abondamment

représenté dans l'épigraphie africaine, bien qu'on puisse
citer cinq QeGellii dans nos ~ravinees. Peut-~tre le centu
rion Gellius (qui d'ailleurs n'était pas présent à Vezere-
'"
os) est-il originaire d'Italie ou d'une province d'Occi-
dent?
Veturius Proculus enfin porte un 'riom très italien, mais
un gentilice ancien en Afrique, où il est au total assez
abondamment représenté (32 dans l'index du CIL VIII).
Aucun de ces quatre cas n'est donc absolument révéla -
teur d'une origine non africaine. Pour Safidius Ammon, les
chances sont sans doute plus grandes, àmoins de voir dans
son ..Q.Q,gnomen les traces d'une adhésion à un culte oriental.
Il nty a en revanche guère de doutes pour les 107 autres;
il faut donc conclure à l'homogénéité africaine du recru-
tement du poste. L'intér~t de cette constatation vient d'
une comparaison avec les listes lambésitaines précitées ,
qu'on voudrait, pour finir, passer rapidement en revue.
1 .. CI 1 VI II , 1 8 0 6 5 ( 1 6 2 p..tlLt.) : 1 e s p at ri e s d es mi 1 i ta ir es
ne sont pas indiquées, maib l'onomastique apparaît peu a-
fricaine d'allure : Annius Corus (a,19); Sulpicius Olympi~
linus (b,10); Iulius Prouincialis (b,11); Aelius Lepidi-
nus (b,12); Atidius Auster (b,14); Aelius Isidorus (c,9);
Claudius Promptus (c,12); Vitellius Aedilis (c,16); Kami-
us Probus (c,18); Menonius Varro (c,22); Aetrillius Furen-
nianus (d,14); Aelius Menecratianus (d,17; il est pourtant
né à Zitha, sur les bordes de la petite Syrte, cf AE,1911,
97, mais il est sans doute d'ascendance orientale); Thor~~

nius Potitus (d,19); Antonius Nereus (d,21); Cordius Ascle-

965
piodorus (e,20); Libellius Primitiuo~ (e,22). Il Y a pro-
bablement une demi-douzaine d'Orientaux, et peut-~tre une
dizaine d'Italiens ou de provinciaux d'Occident dans cette
liste ..
2. ~IL VIII, 18068 (198 p~): les patries sont indi ...
q_uées; tous les hommes sont africains.
3. ~IL VIII, 2556 (entre 199 et 211): pas de mention de
la patrie, mais l'onomastique ne r~vèle pas d'allog~nes.
4. CIL VIII, 2557 ~203 p.C.) : pas de patries, mais on
relève un Valerius Apollin(~ris) (a,19) dont le ~nomen
53
est fréquent parmi les Orientaux et un Iulius Sigillia-
nus (b,19) qui pourrait ~tre Italien.
5. GIL VIII, 2618 (211-212) pas de patries, onomasti-
que d'allure romano-africaine.
· 6.. AE , 1 9 17- ~ 1 9 1 8 , 2 9 ( dé bu t du II I e s .. ) : 1 es p ~tri es
sont indiquées, tous les soldats sont africains.
70 JE, 1917-1918, 57 (m~me époque) : la patrie d'un hom-
me est indiquée : M.Aurel(ius) Hermias, Ale(xandria)(16);
pour les autres, on remarque des noms peu africains, wire
exotiques : Arellius Apollonius ( 13); Cornelius Aratoxen
(25); Sextus Lateranus (sic, 29); Postumius Pacuius (sic,
30); Aelius Glyconianus (32); Aurelius Nilammon ( 45); Pom-
peius Cottinus (60).
8. CI!J VIII, 2586 (m~me époque) : on connait l'origine
de trois soldats de cette liste : C.Aelius Iulianus, de
Sarmizegetusa (20); T.Aelius Victorinus, de Siscia (22);
M.Aure~ius) Nicostratus, de Tarse (26); on remarque aussi
le nom de Valerius Daphnus (54) dont l'origine est incon-
nue.
9. ·~IL,VIII,2564 (sous Sévère Alexandre) : pas d'origi-
nes indiquées; mais tantOt des noms italiques, tantOt une
onomastique orientalisante, ou dont on a peu d'e~emples

en Afrique : 1ulius Hebenus (1,21); Titinius Laurent(ius)

(1,22); Aurelius Leon (1,29); Iulius Gaianus (I,48); Iu-
nius Congius (1,81); Peticius Catullin~s (1,82); Aurelius
Princeps (!,84); Hortensius Neptunalis(1,88); Iulius La-
bracenus (II,2); Aurelius Apollonius (11,27); Aemilius ~

966
rannus (II,35); Caelius Iouinus (II,37); Aurelius Derisor
(11,65); 3taberius Cinnus (II,74); Aelius Lucidus (II,76);
Gauius Acutus (II,94); Sempronius Vinitor (II, 106).

Certes, à Lambèse, à la m~me époque, les allogènes ne

constituent plus qu'une faible partie du recrutement de la
légiono Ils n'en sont pourtant pas totalement absents. En
revanche, il n'y en avait sans doute pas, ou guère, dans ffi
poste de Vezereos, à l'effectif appréciable. Il n'en faut
pas chercher bien loia les raisons : on les trouverait dmB
les conditions climatiques assez dures (auxquelles pourœnt
des Syriens du désert se seraient ai.sément accoutumés) et
dans la nature des missions, qui exigeaient des patrouïUes
d'abord une bonne connaissance du pays, des mentalités et
des coutumes de ses habitants, des pasteurs semi-nomades
des step-pes; mais aussi, on a tort de n'y pas penser, une
. 54
pratique suffisante de leur langue : le commandement attEn-
dait nécessairement des Romano-africains en poste à Veze -
reos qu'ils recueillent du renseignement auprès des tribus
dont il devait surveiller et canaliser les mouvem~nts.

N 0 T E S

1. Voir la bibliographie récente de P.Trousset, Recherches sur

le limes tripolitanus, du Chott el-Djerid à la frontière tuniso-
libyenrie, Paris, CNRS, Etudes d'Antiquités africaines, 1974, p •
75-77, n° 72; sur notre pierre, p~ 76-77, n° 2.
2. La pierre a souffert d'autres dégradations : un éclat de la
face gauche a été,-semule-t-il, perdu, et il ne reste que quel-
ques lettres des lignes de la-colonne I de cette face (soldats
des centuries d'Errius Saturninus et d 1 A.Surius Iustus).
3. Voir sur ce point D.Nordberg, Manuel pratique du latin médi~
val, Paris, l96b, p. 21; R.A. Haadsma et J. Nuchelmans, Précisde
l;tin vulgaire, Groningen, 1963, p. 26, n° 19 et p. 3l,n° 30; V.
Vaananen, Introduction au latin vulgaire, Paris, 1963, p. 51-52;
ces deux derniers ouvrages font remonter cette évolution au Ier
Se p.c. en citant des exemples pompéiens.
On sait que si l'Itinerarium Antonini Augusti remonte à,une car

967
te du ~emps de Caracalla, la rédactidn qui nous est ind1recte -
ment parvenue ne date que de l'époquè-de la Tétrarchie.
Pour des exemples africains de la m~me mutation consonnar.tique,
mais plus tardifs, cf P.Miniconi, in Tablettes Albert1ni, Actes
privés de l'ép.2_que vandale (fin du Ve s.), Paris, 1952, p.6? •.
4. Vazi Sarra; Vasampus; Vasidice;· Vazaiui; les Vesatences;Ves-
cera. Si l'on ramene notre toponyme à une carcasse consnnantique,
on peut y distinguer l'élément in1tiail V, fréquent dans l'onomas
tique "liby::::ue" et (peut-être !") la r 1acine ZR, qui désigne en
général .les fruits sans noyaux, et dabs la région plus particu -
lierement le figuier, un des rares arbres q_ui précisément se
rencontrent au~ourd'hui. avec le datt~er ~t l'olivier, dB~s les
quelques vergers installés à l'amont ict'une murette, dans les val-
lons. Mais dans l'état actuel de nos ~albutiements sur la langue
libyque, il ne s'~git que d'une très fragile hypothèse.
5. On trouve ces listes énumérées infra. Noter en particulier
la liste CIL VIII, 2557, qui provient· du temple d'Esculape à
Lambèse, ~où une titulature identiq'p.e a été martelée et re -
gravée exactement de la m@me façon qu~à Vezereos. Or ce texte
est un règlement de collège daté -de 203 •.. j
6 .. Y. Le Bohec, Notes ·prosopographiqüês kJr la Legio III Augus-
ta, ZPE, 31, 1978, p. 188-192~ -- --
7• Cf. ILAfr 26, de 201 p.C., avec un martelage comparable.
8. A.Chastagnol, L'Album municipal de Timgad, Aptiquitas, 3,22,
Bonn, 1978, VI-ll2p.
9. Cf. CRAI, 1921,- p. 247.
10. A la ligne 52~ il est diffdïle d.e·· décider s'il faut lire
7 ASVRI IVSTI ou 7 A(uli) SVRI IVSTI; les déux noms, Surius et
Asurius, sont attestés, en particulier en Afrique. On penchera
plutet pour la seconde interprétation, car le A déborde la comn
ne à gauche, ce qui est le cas de to~s les autres prénoms de la
liste, sauf pour Q.Gel.lius Iuuenalis (II,44).
11. A.Chastagnol, L'Album de Timgad, p. 50.
12 .. Postérieure à 205 selon J •. Gascou, La poli tique m'..lnicipale
de l'empire romain en Afrique proconsulaire de Trajan à Septime
Sévère, Paris et Rome, 1972,-p. 179.
13. On trouve dans notre texte, col.I, 1.45 : 7 ERRI SATV[rn~
et, I,52, 7 A SVRI IVSTI ; II,37, 7 FLAVI YICTORIS et II,44,
7 Q GELLI IVVENALIS, etc •••
14. On peut observer dans CIL VIII 2586 des soldats aux genti-
lices peu courants en Afriq~(L.Considius Paulus, C.Galuentius
Ianuarius, L.Orbius Felix, Q.Duronius· Pri,inus) po urt an t désignés
par leurs tria nomina.
15 .. Quatre E.Q,gnomin.a sont de lecture malà:isée à cause de l'état
de la pierre : Aemilius Sa(••• (III,36); P.Clementius c(... (III
14); Domitius Arno[ ••• (III,33); Iulius PQ••• (I,47); le cognomen
de 14 individus a ~omplètement disparu. 1 . ·

16. Cf. I.Kajanto, Supernomina, a study in latin Epigrap~y,

Comm. Hum. Litt. vol. 40, Helsink;i, 1966, p. 7-b.
17. Le nom Ianuarius, lié au présage favorable du-premier mois

968
de l'année, aurait en quelque sorte la m!m~ signification que
Felix, ou Faustus, cf. I~Kajanto, Onomastic Studies in the ear-
ly christian Inscriptions of Rome and Carthag~, Acta Instituti
Romani Finlandiae, vo..L •. II,l, Helsinki, 1963, p .. 22, n~l.
18. J'ai proposé, pour ces noms, des rapprochements avec des
noms punic;ueè~ dans Vbique Populus, p .. 451-454 ..
19. Lambèse (CIL VIII, 3362 et 3575); Césarée (9362;93bb;2125~;
21333;21334;21442); le dernier texte, dont la restitution est
hypoth~tique, est un fra~ment des environs de Capsa (11247).

20 ... I.Kajanto, Onomastic Studies, p ... r_64.

21 .. 0.1 ... , p.6l~

22. M.Leglay, Saturne africEin, histoir~, Paris, 1966, p.243 et

notes.
23. I.Kajanto, Onomastic Studies, p. 61.
24~ A.. Chastagnol, L!Album de Timgad, p. 57 et notes.
25. I ...Kaj:::nto, Onomastic Studles, p. 61.
26. Ibid, p. 81; l'A. ne r~pertorie pas Salutaris, mals relève
en re~he Sal utius; Sal uL:,r'is est pourtar. t très fréquent :
CIL II, 1085, Ilipa (B~tique) : L.Cominius Vipsanius Saluta-
ris, domo Roma (ép. de Septime Sévère).
III,5254, Ce1eia (Norique) : (Marius) Salutaris

2195, Salone (Dalmatie) : Clodius Salutaris

13028, " L.S.Salutaris
6065, ,Ephèse (Asie) : C.V]ibius, C.f...Ouf(entina tribu)
Salutaris, cf. H.G.Pflaum, Les carrières procu-
ratoriennes équestres sous le Haut-Empire ro -
9
main, III, P.• lOS7 Tet 1Q9~ ~ entre. ,,8,~ e\
cr.C.1.L II.L,rl-L et .l'-t-'-_.,5, •
96.
IV (Pompei), 4588 : Salutaris, curator.
V (Cisalpine) : 8672 (Concordia): Sa~utaris.
6076 (Milan) : Pupius Salutaris (?)
623 (Tergeste) : C.Paconius Salutaris.
X, 6229 (Capoue) T. P(omponius?) Salutaris.
3947 Il T.Flauius, T.lib. Salut3ris, augustalis
3958 Il A.Canuleius, A.f. Pal. Salutaris,
domo Canusi ..
!1
4410 L•Tarquinius Sëlutaris
Il
3939 Caeciiia Salutaris (affranchie?)
690 (Sorrente) Q.Valerius Salutaris, Aug.
4873 (Venafrum) : Sex.Pulfennius, C.f. ~er(entina tri-
bu) Salutaris.
6154 (Formies) : Q.Planius Salutaris.
5611 (Fabrateria Noua) : Gauia Salutaris.
8048, 17, sur un mortier du temple de Mercure à Pompei:
Cn. Domiti Salutaris.
XI, 126 (Ravenne) : Flauia, Q.f. Salutaris (3 femmes de
ce nom).
1564 (Faesulae): Sa1utaris (femme)
3232 lNepet) : L.Paconius Romanus, P.f. Salutaris.
3b27 (Veli) : T.Flauius Sa~utaris.

969
 5850 (Iguuium) : P.Pisentius Salutaris.
6310, I,8 (:Pisaurum) : C;..Tedius Sal.utaris.
7432 b (environs de Viterbe) : Salut[ari~
8119·~ ~53 : +L..O ....Salut •. (vase de Sovana).
XIII, 2256 (Lyon) : Salutar1s.
3392 (Reims) : Sal.utaris •.
6664 (Mogontiacum) Vi]ctorius Salutaris.
6983 " : Vitalinius Salutaris, ~q.leg~
XXII P..!.P..!-
XIV:, 1575 (Ostie) : Salutaris.
1456 " : Nasennius Salutaris.
2326 (Ager albanus) : Aur(elius) Salutaris.
2724 (Tusculum) : L.Maius Salutaris.
27. I.Kajanto, Onomastic Stud1es, p. 62.
28 .. I .. Kajanto, .ibid, p .. 61-62, note la tendance à l'altération
de la forme orJ_gine11e des cognomina par 1' adjonction de suffi-
xes en -inus ou -ianus. Il donne l'exemple de Rufus (qui est
attesté dans notre liste) et de ses dérivés \qu'on n'observe pas
ici)- Cette tendance à la dérivation s'observe déjà à l'époque
népublicaine •
2~. Il faut noter la fréquence du nom Sallustius et de ses dé-
rivés dans l'Africa nova et dans la Cirtéenne, et leur rayonne-
ment vers d'autres régions, en particulier celle des garnisons.
30~ Un pointage dans les colonnes de 1' index nominum du CIL VIII
montre l'infériorité numérique des Ulpii (217) par rappor-t--aux
Aelii (623), aux Aurelii (520), aux Claudii (372), aux Flauii (
586); je renonce à compter les Iulii.
31. Le nom Antonius, surtout fréquent en Afrique au IIIe s. et
plus tard, serait lié en grande partie à la romanisation sous~
Gordiens, cf. A.Chastagnol, l'Album de Timgad, r. 51. Dans les
liste,s militaires précitées, on trouve cinq Antonii dans CIL VIli
18065' (162 p.Cl, 2 dans CIL VIII 2556 ( entre 199 ~t 211):--2"
dans AE 1917-i918,57 (début du IIIe s.) et un dans CIL VIII 2564
\ sou;-Sévère Alexandre).
32. Sur cenom, fréquent parmi les descendants des Gétules roma-
nisés par C.Marius, et fidèles aux traditions'militaires de leurs
anc.~tres, cf. J .Gascou, MEFR, 81,2, 1969, p.557-568.
33 .. ILTun 1241.
34~ CIL VIII, 24865; 24868, pour Carthage; pour la datation, cf
Ant.Afr, 7, 1973, p.30-31 et 142; pour la C-irtéenne, cf. H.G ..
Pflaum, Remarques sur l'onomastique de Castellum Celtianum, Car-
nuntina, 1956, p .. 126-151 (le nollï Bennius est étudié à la p.l31)
35. Bell.afr. 89, 5.
36. R.P. Al.Delattre, Le mur d'amphores de la colline Saint-
Louis à Carthag~, BCTH, 1894, p. ~9-119.
37. Ad fam. XII, 24, 3.
38. Dlu.Vespe 3 , 1.
39. CIL VIII, 25646; cf. H.G.Pflaum, RSAC, LXXI, 1969-l.971, p.
61-62-.-

970
 4u. M.Fasciato, Inscriptions proveaant d'Ostie, REL, XXVII,i9~9
p.34; J.Rougé, Recherches sur l'organisat1on du commerce mari-
time en Méditerranée sous l'empire romain, Paris, 1966, p.260;
3DB-Jü9.
41. L'index du CIL VIII confond Sedius et Saedius, ce qui porte
le total à 14. ---
42. CIL XII, 1767 (Valence).
43. CIL V : 4182 : Q.Surius, à Murianello, entre Crémone et
Brescia; 7219 : Mogetius Surius et Surius Clemens, Mogeti f. à
S.Michele della Chiesa (Alpes Cottiennes); 3804, Suria Dometia,
Vérone; 1392, Suria,O. 1. Pyrallis, à Aquilée.
CIL VI(Rome) : 27028, S:ex.Surius Sabinus; 27029, Suria Tertulla
et Sex.Surius Seuerus.
CIL IX: Sur~us Frontonianus (3814), Lecce.
CIL XI: c:.surius, C.l ... Epaphroditus ( 5544), Assise.
CIL XII: Arles, M.Surius, C.f. Albinus (707) et Syria, L.f.Lu-
cilla (885); Nimes, s·uria Nauica (3932); Grenoble, L.Surius Iu-
cundus (22i6); Tertius Surius Ganus (5679,71, tuile de Vienne).
CIL XIII: 6737 : Surius Felix, Mogontiacum; 767r3 : Surius Quin-
tus, Kretz, Germanie' supérieure; 8848 : Surius s:edulus, à Blan-
kenheim, Germanie inférieure.
44 .. AE 1935, 103.
45. Pline, N.H., III, 137.
46. Gandidius : C;IL III, 7648, à Sebesvaralja : G;andid.ius Pa -
truinus, Praef. cOb: I Aeliae Gallorum; CIL XII,722, Arles :
Q.Candidius Benignus et sa fille Candidia Quintina, qui apparatt
aussi en CIL XII; 775 •. Surtout, treize mentions dans CIL XIII :
3643, C.Candidius Piscator, Trèves; 5486, Candidia Paterna, Di-
jon; 6026, Candidius Euhodus, sur les bords de la Zinsel, en
Germanie supérieure; 6243, C •.Candidius Martinus et Candidia (ou
Martinia) Dignilla, Borbetomagus, Germ. sup.; 6396, Candidius
Qu&rtus, 6399, L •.G:andidius Mercator, et C.Candidius Calpurnianus,
tous trois à Heidelberg, Germ. sup •. ; 6999, Candidia Urbana, à
Moguntiacum, Germ. su~ 7925, Sex.Candidius Maternus, à Zulpich,
Germ.inf.; 88'r7, Sex. Candidlus Maternus encore, à Blankenheim,
peut-@tre le m@me que le précédent; 8336, t!!.Candidi[us .....Jer ,
Ci:.olonia .Agripp.inensium; 8858, M..Candidius Celerinus, loc .incert.
Clementius : CIL III, 3424 et 10424, relatifs à Clementius
Siluius, V.E. (.Aquincum); CIL. XI, 2621 = 6689,78a, tuile prove-
nant des ruines de Runella;;-: Sex.Clemen(tiust) Probus;CIL XIII
2861, anneau d'or trouvé dans les sources de la Seine : Clem(~-)
Montiola; 4241, Rascheid (Trévires), Cl~~en[~ia Poppa;8833,
Clement1a Iassa,~oc.incert.; 11267, Clementia Magna, Intaranum
(Entrains). ·
47. CIL VIII, 14721.
48 •. ILAlg I, 402.
49. ILAlg II, 3181.
50. CIL VIII, 15204.
51. CIL VI, 17256.
52. Augustin, ~.XVII; Eupedir.s a été o.ublié par les auteurs de
l'index nominum des ILAfr; la restitution Euped(ius) est celle

971
du nouvel index, paru dans Karthag~, XI, 196i-1962, p.-179-208.
53~ Cf • .L.Robert, RPh, XIII, 1939, p. 180.
54. I.e bilinguisme est prouvé pour un certain nombre de soldats
africains, comme le montrent :
1° : les inscriptions "latina-puniques" (inscriptions_militai-
res composées en néo-punique, mais gravées en caracteres la-
tins et étudiées par M.G. Bertinelli Angelli, Termini romani ,
}JUblici e sacri in epigrafi "latino-libiche", Studi di Storia
antica in memoria di Luca de Regibus, G@nes, 1969, P• 217-224,
cf. AE 1972, 676;
2° : l 1 épitaphe certainement bilingue du vétéran C-Iulius Gae -
tulus (CIL,VIII, ,5209 = Gsell, ILAl, 137 = J.B.Chabot, Recueil
des Inscriptions libyques, 146; cf. J.Gascou, MEFR, t.82, 1970,
p.725); d'autres ont pu etre bilingues sans que leur épitaphe
l'atteste;
3o : enfin la mention (rappelée par Y. Le Bohec dans la discus-
sion de cette communication) d'un soldat de Bou Njem détaché
cum Garamantibus, nation dont il parlait donc nécessairement la
langue, cf. AE 1975, 869,b.

Note additionnelle à propos des restitutions proposées pour le

côté gauche : après un nouvel examen du document, je ne vois
p~s de raisons d'éliminer, l. 22, la restitution ~~PI ~}s.

Summary

The text of the inscription ILAfr. 27, a list of soldiers from the
limes Tripolitanus, dating between A. D. 199 and 211, can be partly
completed by a recent re-examination. Studied in itself and compared
with sorne other African military inscriptions from the reigns of
Marcus Aurelius to Seve rus Alexander, it off ers two points of interest:
firstly an onomastic one, because of the small proportion of the tria
nomina, for very few soldiers appear to have a praenomen; secondly,
although the origin of the men is not pointed out, the study of the
nomina gentilicia and of the cognomina shows that the unit was almost
entirely recruited in Africa, while at the same time soldiers from
the Danube area, from the Orient and also from Italy appear in the
legionary lists from Lambaesis. Sorne assumptions can be proposed
about the reasons for the homogeneous recruitment of the garrison
of a border fort.

972
Pl. 65. 1 ILAfr. 27. Face principale.
Pl. 65. 2 ILAfr. 27. Côté gauche.
Pl. 65. 3 ILAfr. 27. Côté droit.
 66. LA FERNE RO:rt..AIŒTTE,AIN BENIA,AIN BENT SOLTfu~E:

FORTINS OU FERMES FORTIFIEES?

Nacéra BENSEDDIK

Dans une communication qu'il R présentée au dernier congrès sur

" Les déplacements successifs du limes en 1·'Iaurétanie Césarienne" ,
P.Salama a dressé une liste (pp.584-586) des camps et forteresses de
la frontière sévérienne.Qu'entend-on exactement par forteresse? Si
l'on donne à ce mot le sens d'établissement militaire (1),deux des
"forteresses" de cette liste,la ferl!le Romanette (n°21) et Ain Benia
(n°28) posent en réalité un problème d'identification.
Loin de prétendre clore cette étude du limes de Césarienne en-
treprise depuis longtemps par P.Salama,cette communication a pour but
d'apporter une information complémentaire sur la portion comprise en-
tre:Ain Teukria et Numerus Syrorum.Les vestiges de la ferme Romanette
de Ain Benia et Ain bent Soltane appartiennent-ils à des avant-postes
du limes dans cette région? Ont-ils eu une autre fonction?

La ferme Romanette,Ain Benia,Ain bent Soltane

~vant-postes du limes?

Etant donné la documentation existante,comment se présente

cette portion du limes dont nos trois sites sont supposés assurer
la protection?

Le limes

L'occupation romaine ayant été dès le départ mal acceptée,elle

fut mise en danger,~ l'intérieur de la province par les montagnards,

977
au Sud par les tribus nomades toujours refoulées au-delà du limes.
Tout l'effort romain a alors consisté à encercler par des routes
les massifs qu'on ne pouvait occuper et interdire aux nomades - ou
du moins contrôler sévèrement - l'accès de la province par une
route-frontière qui serait une sorte de barrière infranchissable:ce
fut la praetentura,établie sous le règne des Sévères (2) à la limi-
te de la zone steppiqûe pour relier un certain nombre de camps et
de villes fortifiées;désormais la liaison Zaraï - Numerus Syrorum
était possible.
Dans la région qui nous intéresse,cette voie allait de Ain
Teukria (ex-Bourbaki) à Tiaret en passant par Ain Tissemsilt (ex-
Vialar),Sidi Hosni (ex-Waldeck Rousseau).Entre Tiaret et Aïoun Sbi-
ba on n'a trouvé aucun vestige de voie,mais il en existait proba-
blement une.On peut de nouveau suivre cette route-frontière de
A!oun Sbiba à Maghnia (Numerus Syrorum),par Taoughzout,Takhmaret
(Cohors Breucorum),Benian (Ala Miliaria), Timziouine ( ~), Sidi
Ali ben Youb (Kaputtasaccora),Hadjer Roum ( Altaua) et Tlemcen (~­
maria).
Sur cette partie de la f~ontière romaine de Maurétanie Césarienne,
les témoignages archéologiques,épigraphiques,nous permettent de
'
compter Ain Teukria,Columnata,Aïoun Sbi•a,Cohors Breucorum,Ala Mi-
liaria,~,Kaputtaaaccora,Altaua,Pomaria et Numerus Syrorum parmi
les points fortifiés importants du limes (3).
Ain Teukria (ex-Bourbaki) était très probablement une place
importante de cette frontière.Des ruines romaines étendues -dont
les vestiges d'un rempart- ont été- signalée-s par Stéphane Gsell (4)•.
D'autre part,un détachement de la IIème cohorte des Sardes y est
passé,sous le règne de Gordien III,peut-être pour une quelconque
action dans les abords. montagneux (CIL VIII,21523).

Columnata (Sidi Hosni). Au XIXème siècle,on y distinguait en-

core,paratt-il,les vestiges d'une enceinte de 40m x20m et,à c&té,
d'une constructic·n carrée de 6m de côté (5).!1 est cependant diffi-
cile d'attribuer ces vestiges au Haut-Empire plut&t qu'au Bas-Empi-

978
re.L'épigraphie du Haut-Empire ne nous y signale ni stationnement,
ni passage de troupe;en revanche,sous le Bas-Empire,Columnata a
joué un rôle militaire de premier plan puisqu'elle a été le chef-
lieu d'un des districts militaires de la frontière: le limes colum-
natensis (6).

Tiaret pose un problème car on ne sait pas avec certitude si

cette ville où on a retrouvé,au XIXème siècle,les vestiges de trois
enceintes munies de bastions carrés se trouvait sur la praetentura
(7); dans cette zone,le limes est incertain et la voie ne pourra
être suivie de nouveau qu'à partir de Aioun Sbiba.

Aïoun Sbiba. Le Lt Fort y a signalé,au début du siècle,des rui-

nes entourées d'un mur à double revltement que coupent,à quelques
mètres d'intervalle,de gros blocs de pierre réguliers (8). P.Salama,
qui l'identifie à Cen (9),la considère comme un camp d'aile.Il n'y
a là pourtant aucune certitude car les trois inscriptions militai•
res qui en proviennent concernent,l'une le cursus d'un personnage
qui a été 2raepositus alae Parthorum (aile stationnée à Kaputtasac-
~) sous le règne de Gallien ou après (10),la deuxième un vétéran
d'une troupe dont le nom n'est pas précisé ( CIL VIII,21558) tandis
que la troisième (Lib.t.III,1955.2ème sem.,pp.342-352) mentionne
l'inauguration d'une fontaine,entre 212 et 217,par le commandant
d'une aile dont on n'est pas s'O.r qu'il.s'agisse de l'aile des Par-
thes.

Cahors Breucorum (Henchir Souik,Takbmaret) était le camp de la

IIème cohorte quingénaire des Breuques (11).

Ala Miliaria (Benian) (12) était le camp d'une aile milliaire,

l'ala Miliaria (BSGA0.,1936,p.109; CIL VIII,21568 a et b).

~ (Timziouine). On ne connatt pas avec certitude le nom de

la troupe qui occupait cette place forte,mais la Ière cohorte mon-

979
tée des Pannoniens,si elle n'y a tenu garnison,y a séjourné suffi-
samment longtemps pour réaliser la voie Lucu-KaputtasaccorM dont on
a deux bornes datées de 209-211 (CIL VIII,22602/4; 22611) (13).

Kaputtasaccora (Sidi Ali ben Youb) (14).Ce camp de l'aile des

Parthes a été établi au nord des monts de Tlemcen,barrant la vallée
de l'oued Mekerra à sa sortie des montagnes.

Altaua (Hadjer Roum) (15). Le camp ainsi que l'agglomération

civile ont été installés sur le plateau du m3me nom.Le site fut oc-
cupé dès 201 (CIL VIII,9831 et 9833) par la !!ème cohorte montée
des Sardes (16).

Pomaria (Tlemcen) (17) était-elle la ville de garnison de 1'~­

la Exploratorum Pomariensium (CIL VIII,9906,9907) ou son camp,comme
le laisserait supposer le "balneum castrorum" de l'inscription CIL
.llll,9908?

Numerus Syrorum (Maghnia) (18). Point extr3me de cette cha1ne

de routes et de forteresses,le camp fut établi dans la vallée de la
Mouila pour surveiller la plaine des Angad.Un numerus de 1000 ar-
chers peut-~tre,le numerus Syrorum,y tenait garnison (CIL VIII,9962
et 9964).

La ferme Romanette,Aïn Benia,

Ain bent Soltane

Bien que cette praetentura marqu!t la limite de l'occupation

romaine de la Césarienne,des vestiges romaine,rares il est vrai,ont
été signalés au-delà de cette ligne:c'est le cas des trois sites
qui nous intéressent ici.
Il est possible que ces ruines soient celles d'avant-postes établis

980
au sud du limee,soit pour protéger des colons qui s'étaient aventu-
rés jusque là (19),aoit pour renforcer la praetentura.

La ferme Romanette.(Fig.66. 1)(20).

C'est une ferme coloniale qui a été édifiée sur des ruines,dans
une plaine stérile et nue si on excepte le lit de l'Ouerk qui passe
juste au pied. A. Joly qui s'est rendu sur les lieux en 1897 en a
donné la description suivante:" J'ai pu reconna1tre encore les tra-
ces de quelques constructions faites en moëllons liés par un très
bon ciment,avec des cha1nes d'angle en pierres de taille.L'épaie-
seur des murs était d'environ Om50.Je remarquai un mur d'une épais-
seur considérable (plus de 2m) dont les parois étaient en pierres
de taille d'énormes dimensions,quelques-unes longues de plus de 2
mètres avec Om50 à Om80 de queue.
Le remplissage était fait en blocage de moëllons liés par un bon
mortier jaun!tre,se tenant bien dans la masse,mais qui s'effritait
seulement un peu à la surface sous le choc du marteau.Beaucoup de
pierres de taille étaient éparses çà et là aux environs.J'ai pu re-
connaître aussi la trace bien nette des fondations d'une tour d'an-
\
gle un peu plus que demi-circulaire et dominant la berge de la ri-
vière.Les murs en très belles pierres de taille liées par d'excel-
lent ciment avaient environ 1m20 ou 1m25 d'épaisseur et le diamètre
de la tour était de 8 à 10m.Je reconnus plus loin la trace de deux
autres tours analogues maie moins distinctes sur le bord de la ri-
vière.La plus éloignée de la première me fit l'effet d'être aussi
une tour d'angle.Quant à celle qui se trouvait dans l'intervalle ,
ce pouvait être une tour destinée à flanquer un rempart car elle é-
tait semi-circulaire.Les trois tours s'alignaient sur une direction
perpendiculaire à celle de ce qui m'a semblé le reste d'un rempart
et à une distance d'une trentaine de mètres de ce reste.Les deux
plus éloignées étaient séparées par un espace d'environ 60m;les
vestiges du rempart pouvaient se prOlonger sur 15m environ ••••• Le
mur extérieur de la ferme Romanette,celui qui bordait la rivière, a
dû être construit sur les fondations mAmes du rempart ou mur d'en-

981
 OUED OUERK

/ ..
1

6
2

0 50
Metres

., 3

Fig. 66.1 Ruines de l'oued Ouerk, d'apres A. Joly

(Archives du Service des Antiquites de l'Algérie).

1. Fondations de tours nettes

2. Fondations un peu confusés
3. Portion de mur antique conservé
4. Debris de mur
5. Mur epais de 3 rn
6. Fondations de chambres

982
ceinte que flanquaient les tours.
Quelques pierres de taille portaient des ornements sculptés;des
phallus,un niveau de maçon triangulaire et un ornement en forme de
croix" (21).
La ferme Romanette se trouve à environ 60km des hiberna de
l'ala Sebastena. P. Salama pense même qu'un détachement de cette
aile occupait cet avant-poste chargé de surveiller une piste venant
du Sud et se dirigeant vers les hiberna.D'autre part,le plan de ces
ruines avec des fondations de tours semi-circulaires et circulaires
et un reste de rempart,la solidité de la construction (murs d'une
épaisseur qui atteint 2m) permettent d'abonder dans ce sena.A. Joly,
qui,le premieraa décrit les ruines,écarte l'idée d'une aggloméra-
tion civile et opte pour une colonie militaire restreinte,un poste
fortifié qui,avec Benia,défendait au Sud le Sersou.Stéphane Gsell ,
lui,parle d'une avant-ligne postérieure à l'établissement du limes
sévérien (22).

Ain Benia.(Fig.66.2)

A une quarantaine de kilomètres de Tiaret,sur la rive gauche

de ~'oued el Faïdja, R. de la Blanchère signalait,en 1883, l'exis-
tence de vestiges d'une forteresse romaine (23).Elle aurait eu pour
mission de protéger le Sersou au Sud;le plan des ruinea,avec ces
deux bastions rectangulaires qui défendaient l'entrée,peut en effet
être celui d'un poste fortifié.

Ain bent Soltane.

Parlant d'Altaua,O. Mac Carthy dit "qu'elle avait comme d~­

pendances immédiates quelques postes fortifiés destinés à couvrir

la vallée supérieure de l'Iss~r dont elle fermait l'entrée.A sa
tête même,à Meurbah,on voit les restes d'une tour et,au-dessous ,
quelques vestiges antiques à Sidi Brahim et à Ain bent Soltane des
Beni Smail" (24).
Stéphane Gsell,dans son Atlas archéologigue,ne signale à Ain bent

983
 AIN BENIA

0 25
L-.---------~ Metres

-1 ...

Fig. 66.2 Ruines de Ath Benia (R. de la Blanchère, Voyage d'etude dans une
partie de la Maurétanie Césarienne, pl. VII).

984
Soltane qu'une hache de pierre polie (25).
Presque un siècle après le rapport d'O. Mac Carthy, P. Courtot rap-
porte,dans une étude qu'il a consacrée à Altaua,qu' "à 5kms en a-
mont de cette cité,dans la vallée de l'Isser,on voit les ruines
d'un fortin romain près d'une source,l'A!n bent Soltane" (26).

Fermes fortifiées?

Aïn bent Soltane

L'exemple de Ain bent Soltane montre bien quel danger il y a

à suivre aveuglément les récits des voyageurs du siècle dernier.On
eüt souhaité qu'O. Mac Carthy nous d!t pourquoi il considérait les
"quelques vestiges" de Ain bent Soltane comme un poste fortifié
(27)! 80 ans plus tard,P. Courtot reprenait cette identification en
précisant qu'il avait vu "les ruines d'un fortin romain",mais il ne
donnait pas plus de détails (28).
En fait,pour ce site,l'état de la documentation ne nous permet pas
plus de parler de fortin que de ferme fortifiée.Peut-Atre n'étai-
ent-ce m3me pas des vestiges romaine.

Ain Benia et la ferme Romanette

La qualité des sources est heureusement différente pour ces

deux autres sites;nous disposons de descriptions détaillées et de
plans laissés par ceux qui.les ont vus.
L'arrière-pays de m3me que le type de leur architecture donnent
toute vraisemblance à l'hypothèse de deux grandes fermes fortifi-
ées.

L' .!.rrièr_!! ~l:!!

Les ruines de Ain Benia sont situées dans le fond d'une riche
vallée du djebel Nador,sur la rive gauche de l'oued Faïdja et près

985
d'une excellente source.Cette large vallée aux nombreuses sources
est comme "une vraie oasis de montagne" (29) après le désertique
Sersou qui s'étend au pied du Nador.

Les vestiges antiques de la ferme Romanette dominent la berge de

l'oued Ouerk dont l'humidité du lit tranche avec la sécheresse de
la plaine9Bien sür il est impropre à la grande culture car maréca-
geux et très encaissé,mais il reste la culture de subsistance et
l'élevage du gros bétail et,surtout,l'eau,ici,est pure et abondan-
te ..

L'f!rchi t!_c1u~
L'existence de tours carrées dans le cas de Ain Benia (29bis),
circulaires ou semi-circulaires dans celui de la ferme Romanette,de
même que le vestige de rempart relevé dàns cette dernière,ne dimi-
nuent en rien la vraisemblance de l'hypothèse que ces deux ruines
appartiennent à des uillae rurales.
On a souvent tendance,en effet,à donner à la tour une signifi-
cation exclusivement militaire.Pourtant,toute une série de tours
connues,en Afrique avant la conquête romaine (30) et ailleurs dans
le Bassin Méditerranéen (31),ont été de simples annexes d'enclos
rustiques. S'appuyant sur des recherches archéologiques et la pros-
pection du terrain,J.H.Young a remarqué que les tours existaient
dans toutes les grandes fermes de l'Antiquité où elles ont servi de
magasins pour les céréales,l'huile,le vin et autres produits, ou
d'ateliers (32).
Il est vrai aussi que dans ces m3mes exploitations agricoles la
tour a servi de poste d'observation pour les travaux qui se dérou-
laient dans les champs,le bétail qui était dans les pâturages,et de
refuge en cas d'attaque des brigands (33). Elément de l'architectu-
re civile,la tour a joué aussi un rSle défensif,mais pas nécessai-
rement militaire;lequel des deux aspects a précédé l'autre,c'est là
un problème qu'il ne noua appartient pas ici de résoudre.Il suffira
de dire avec P. Grimal que la maison à tour est restée en usage

986
surtout parce qu'elle a pu offrir une certaine protection contre
les brigands et les pillards;c'est une des raisons -probables de
l'absence de tours dans les uillae rusticae italiennes (34) et de
leur présence dans les fermes situées sur les frontières du monde
antique,sur lee territoires expos~s aux attaques (35). Si,dans les
domaines italiens,la construction de tours -même si elles sont une
réminiscence des maisons fortifiées de la période républicaine-
répond à un souci esthétique et de prestige,en Maurétanie Césarien-
ne elle a un but défensif et lee uillae deviennent des habitations
fortifiées sans que l'on puisse pour autant parler de dispositif
militaire.
Inclure Ain Benia et la ferme Romanette dans ce dispositif sem-
ble donc désormais moins évident (36) et l'hypothèse de deux ex-
ploitations agricoles fortifiées appartenant soit à une personnali-
té indigène (37),soit même à un vétéran (38),n'est pas à écarter.

Quatre exemples de militarisme non fondé

Le_ch!te,!_u_de Kao~ i3.2)

Grâce à l'inscription qui surmontait l'entrée,le Capitaine
Marchand qui l'a fouillée a réalisé qu'il s'agissait là d'une gran-
de demeure fortifiée et non d'une forteresse (40). Cependant le vo-
cabulaire employé dans sa description et dans celles d'autres visi-
teurs est resté équivoque: " •• vaste maison fortifiée ••• ch!teau for-
tifié •• " (41),puie " •• entrée du fort •• " (42) chez le Capitaine
Marchand; " •• chAteau de Kaoua •••• habitation (43) ••• castellum"
(44),puis sur la même page " •• existence sur ce point des habitati-
ons qui devaient abriter les esclavee,les serviteurs et probable-
ment une partie de la garnison de la place",chez Lacave-Laplagne
qui a réédité le mémoire du Capitaine Marchand.

La_fer.E!,e_du Naso.r
Cette grande ferme qui se trouve entre Tipasa et Cherchel a été

987
prise,en raison de sa porte monumentale,pour un château fort, un
"castellum","une forteresse officielle" (45).
En réalité,dès la fouille de A.Ballu,les fonctions agricoles de cet
établissement étaient démontrées et la ferme du Nador s'avère Atre
une des plus grandes uillae de la région (46).

La_régio,a de_Ténès _
Ceux qui ont sillonné l'arrière-pays de Cartennae au XIXème
siècle y ont reconnu de nombreux vestiges de fortins.Pour la plu-
part,nous ne disposons d'aucune description ,d'aucun témoignage ir-
réfutable.Nous n'avons pas non plus cette confiance de A.Berbrugger
dans ce qu'on pourrait appeler "le flair archéologique" des mili-
taires (47).En réalité,ces "forts" se sont avérés être,pour la plu-
part,des fermes ou des huileries (48).

Le§. _" _E.a_!tra_aJ!ziel!sia.:_

De nombreuses ruines dispersées sur le territoire d'Auzia
(Sour el Ghozlane,ex-Aumale) auraient été autant de camps destinés
à constituer la ceinture défensive de la colonie: Tseniet ben You-
cef, Hedjia, Djebel Dira, Oued Djinenne et Oued el Akhal auraient été
les "castra auziensia" (49).
Le poste de l'oued Djinenne,par exemple, est "un rectangle de 60
pas sur )O";les autres sont décrits avec la m3me imprécision.
D'autre part,il semblerait que l'expression "in territorio auzien-
si" de l'inscription CIL VIII,9047,qui ne concerne que le territoi-
re urbain de la colonie et non un territoire militaire,ainsi que le·
mot "castrorum" -élément du titre de Iulia Domna,"Mater Castrorum"-
d'une autre inscription ,soient à l'origine de la militarisation de
toutes les ruines avoisinantes (49).

Ce que j'appellerais "le militarisme archéologique" est loin

de constituer une particularité de la Maurétanie Césarienne;sans

988
 1

voie romaine existante

voie romaine suppossée

piste probable

CAESAREA Localité antique

Taoughzout Localité moderne

• ouvrage militaire

*
sites concernés par l'article

1
1
1 Ferme
1
1

. - -*
1
~mord janet

Ben Beho Aïn Benia

~----.-- Aiounsbibo

Ai bent
Soltane

Fig. 66.3 Carte d'une partie de la Maurétanie Cesarienne, d'après P. Salama,

Reseau Routier de 1'Afrique Romaine, Alger 1947.
aller très loin,les fermes fortifiées du limes tripolitanus (région
de Tebaga) ont également été,pendant longtemps,considérées comme des
ouvrages militaires (50).
A la lumière de ces exemples d'identification militaire contestable
mais systématique au XIX et même au début du XXème siècle,il con-
viendra d'attendre une information archéologique complémentaire,dea
découvertes épigraphiques,pour attribuer définitivement à la ferme
Romanette,Aïn Benia,Aïn bent Soltane,une fonction militaire ou une
fonction agricole.Le but de ma communication était moins de résoudre
un problème d'identification des vestiges que de jeter le doute sur
une militarisation abusive,rarement fondée sur des arguments scien-
tifiques,des vestiges de la Maurétanie Césarienne.

1: SALAMA (P.), Les Déplacements successifs du limes en Maurétanie

Césarienne,Akten Des XI Internationalen Limeskongresses,Budapest ·
1978,pp.584-586 et carte n°4.

2: La bibliographie concernant la question est donnée par SALAMA (P)

art.cit.,pp.588-589, plus BIISEDDIK (N.), Les Troupes auxiliaires
de l'armée romaine en Maurétanie Césarienne sous le Haut-Empire,
Thèse de Doctorat de IIIème cycle (sous presse).

3: BENSEDDIK (N.),op.cit.,troisième partie. P. Salama cite,dans sa

communication,d'autres points fortifiés sur cette portion du li-
~;ceux-ci ont cependant encore besoin d'Atre soumis à un exa-

990
 men çri tique car la ferme · Romanet te, Ain Beni a, Ain bent So 1 tane .
·ne sont probablemént pas des cas isolés.

4: GSELL (St.),Atlas archéologique de l'Algérie,Alger-Paris 1911,

23,27.

5: Ibid.,33,15; 22,127 et~·

6: Id., Le Christianisme en Oranie avant la conquête arabe,BSGAO.,

1928,p.25,n.1 (Bull. du Cinquantenaire). ALBERTINI (E.), La
Route-frontière de Maurétanie Césarienne entre Boghar et Lalla
Maghnia,BSGA0.,1928,pp.34-35 (Bul~.· du Cinquantenaire). CADENAT
(P.), Vestiges paléo-chrétiens dans la région de Tiaret,Libyca
(arch./épigr.).t.V,1957,pp.90-97. SALAMA (P.), Les Déplace-
ments ••• ,p.585,considère Columnata comme un camp du limes sévé-
rien de Césarienne.Aucune présence de troupe n'y est pourtant
attestée et des vestiges de rempart et d'une construction car-
rée peuvent aussi bien appartenir à une agglomération civile,
fortifiée ou non.La liste des effectifs du limes de Césarienne
et de leurs stationnements que donne l'auteur plus loin,p.590 ,
ne comprend d'ailleurs pas Columnata.Cette remarque vaut aussi
bien pour Ain Teukria.
Pour l'importance de Columnata au Bas-Empire,voir la Notitia
Dignitatum,Oc.XXV,12 et 2Q; XXX,3 et~ (éd. O.Seek,Berlin
1876).

7: CAGNAT (R.),op.cit.,pp.660-662. GSELL (St.), Atlas arch.,33,

14.

8: FORT (Lt.),Les Ruines romaines d'Ain Sbiba,BSGA0.,1908,pp.21-

25.

9: Cette identification reste hypothétiq~e car Cen n'est attesté

par aucun document sinon un milliaire trouvé en remploi à

991
 Aioun Sbiba (P. SALAhA, Aioun Sbiba,identification de la
ville ror1a.ine, Li tyca arch ./épit,T., t .III, 1er sem., 19 52, pp.
173-177).

10: AE.,1954,135. Pour la datation de l'inscription,voir CIUISTOL

(Jo!.), La Carrière de Traianus E:ucianus et l'origine des ~­
tectores,Chiron (à paraître),pp.16 et 17.

11: Cil VIII,21560 J. 21562. GSELl (st.),op.cit.,33,23.

12: GSELL (st.),ou.cit.,32,93.

13: !bid.,32,46.

14: Ibid.,31,76. CIL v~II,9827 et 9828.

15: GSELL (St.), op.cit.,31,68.

16: Dans son article s~J.r }_es déplacements successifs du limes de

Césarienne (ci-dessus cité),P. Salarna dit ~u'Altaua était
aussi le camp de l'ala !la Thracum ( l' "ala Ia Thracum" de
l'article est certainement une faute d'impressio~). Zn fait,
les deux inscriptions d'Altaua ~ui mentionnent cette aile
( CIL VIII,21721 et AE.1932,31) concernent un décurion de
l'aile II des Thraces, T. Iulius Germanus,~ui se trouvait à
Altaua comme praepositus de la IIème cohorte des Sardes. Il
a pu venir à Altaua avec un détachement de l'aile des Thra-
ces ( apparemment stationnée à Caesarea),comme il a pu y
arriver seul avec pour mission de prendre temporairement le
cow~dement de la cohorte.Ce type de mouvements a été étu-
dié par DO~~ZEWSKI (A. VON), Die Rangordn~ des. rërnischen
Heeres (2ème éd.,Dobson,Cologne 1967),p.53; pour des exem-
ples d_e Césarienne, voir TIE!'JSEDDIK ( :N. ) , op. ci t. , p. 154 et
n.1 (du nanuscrit).

17: GS:S:..L (St.), Atlas arch.,31 ,56.

992
18: I~id. ,41, 1 •

19: GSELL (st.), op.cit.,23,21 à 26; 33,17 à 22,117,129 et 130, a

signalé dans le Sersou des établissements de colons au-delà du
limes.

20: Ibid., 34,57. Crœonique archéologique africaine ,MEFR., t .XX, 1900

p.140.

21: JOLY (A.), Ruines romaines de l'oued OUerq près de Chellala ,

EACTH.,1898, p.188. La figure n°1 du . présent article repro-
duit un relevé de ces ruines fait par A.Joly en 1899 mais res-
té,à ma connaissance,inédit.

22: GSELL (St.),art.cit.,p.140.

2): LA BLANCRK'1.E (R. de), Voy36e d'étude dans une partie de la

J'ilaurétanie Césarienne, extrait des Archives des !'~lissions Scien-
tifiques et Littéraires,t.X,1883,pp.7D-71 : "C'est un carré à
peu près régulier de 32 à 35 mètres de côté,orienté exactement
par ses faces.L'entrée,dans le côté oriental,est défendue par
deux bastions carrés,de 4 mètres sur 4m50,laissant entre eux
un espace de 4m80 de large.La porte devait avoir 2 mètres de
haut,ou un peu plus,sur m1e largeur proportionnée.Du dedans on
va dans les bastions par tme ouverture de un mètre de large.
L'enceinte est un mur de 3m15 de hauteur au-dessus du sol ac-
tuel fait de pierres de taille sans mortier et terminé par une
corniche fort sirnple.Les pierres,qui ont 45 centimètres de
haut,70 centimètres de large et de 60 centimètres à 1m10 de
long,sont bien taillées,bien ajuatées,quoique parfois,à l'in-
térieur,avec le secours d'éclats plus petits,parées seulement
sur la face extérieu.re,et réunies entre elles par des crampons
de métal.Tout indiquerait une assez bonne époque ••••• Dans
1' intériem,,il n'y a pour ainsi dire aucune trace de construc-
tion,sinon deux pierres près de l'angle s.e. qui doivent indi-
quer une porte;celle de gauche a un reste de sculpture mainte-

993
 nant tout à fait indistinct. Au-dehors de l'angle N.O. et de
l'angle S.E. partaient deux rnurs,aujourd'hui rasés,de même ap-
pareil que ceux du château,mais sans crampons,et certainement
de peu de hauteur.Ils enfermaient un espace tout plein de gros-
ses pierres qui paraissent taillées".
GSELL (St.), Atlas arch. ,34 ,26. GAGNAT (R.), L'Armée romaine
d'Afrique et l'occupation militaire de l'Afrique sous les empe-
reurs,Paris 1892,p.652.

24: BAC C.AHTHY ( 0.) ,Algeria Romana,R .Af., t .I, 1856 ,p.1 02.

25: CSELIJ (St.) ,o;p .ci t. ,42 ,82.

26: COURTOT (P.), Essai historique sur Altaua d'après l'épigraphie,

R.Af.,t.LXXIX,1936,p.407.

rille c~~TRY (o.),art.cit.,p.102.

28: COURTOT (P.),art.cit.,p.407.

29: LA BLAKCHERE (R. de),op.cit.,p.70.

29: LEVEAU (Ph.),Recherches historiques sur une région monta6~euse

bis
de Yaurétanie Césarienne:des Tigaua Castra à la mer,l·'l.EFR..,t.L
XXXIX,1977,pp.274-275,signale dans la région comprise entre
Ténès et Chercbel,sur le site d'Ararfi,une ferme antique flan-
quée de trois tours carrées saillant de 4,6 à 5,3m.

30: DIODORE DE SICILE,III,49, dit à propos des Libyens:"leurs chefs

ne possèdent pas de villes mais ils ont quelques tours assises
au bord de l'eau ,dana lesquelles ils conservent le restant de
leurs vi vree".

31 : CRU-:.AL (P.), Les Na.iscns à tour hellénistiques et romaines ,l"lEFR.

t.LVI,1g39,pp.28 à 59 et surtout pp.47-49,donne l'exemple de la
Palestine où il était courant d'élever des tours dans les

994
 champs. Ed.~eyer,qui rapporte le fait,ajoute que ces tours
"n'avaient rien de commun avec des fortificationa,mais
qu'elles étaient bâties par des particuliers pour leur usage
personnel" (Hermes 1920,p.101,cite le Nouv. Test.,Marc,12,1).

32: YOill{G (J.H.), Studies in South Attica Conntry Esta te at Sou-

nion,Hesperia XXV,1956, cité par NOWICKA (M.), Les }~isons à
tour dans le monde grec,Varsovie 1975,p.16.

33: Sur l'origine politico-juridique,sociale ou économique du

banditisme dans l'Empire Romain,voir FUU~-ZUCKNU·~n; (L.), A
propos d'une inscription de SUisse, Latomus t.XXIX 1 1970,fasc.
2,pp.451-473·

34: NCw'TCKA (:r.~.), op.ci t. ,p.122.

35: Ibid. Pour les menaces qui pesaient sur les plantations afri-
caines,voir l'inscription de Henchir Mettich, rrJémoires pré-
sentés par divers savants à l'Académie des Inscriptions et
Belles Lettres,1ère série,t.XI,1901,pp.31-81. CU~ (E.), Le
Colonat partiaire dans 1' Afrique romaine d'après l' inscrip-
tion de Renchir I"Lettich,Hérnoires présentés ••••• ,pp.83-146.

36: Ces sites ont été considérés co~me des forteresses par ceux
qui les ont vus et décrits pour la première fois: R.de la
Blanchère et A.Joly. Cette identification militaire a été
reprise par CAGNAT (R.),L'Armée •• ,p.652; GSELL (St.), Atlas
arch. ,34, 26 et 57; id. ,cr..ronique •••• ,hEFR., t .xx, 1900,p .140;
SAl..AEA (P.), :r;ouveau.x témoigna~;es de 1' oeuvre des Sévères en
:tviaurétanie Césarienne, Lib. ,a.rch./épigr., 1953 ,p .259 ,n.178 et
récemment dans les Déplacements successifs •• ,Akten des XI
In~ernationalen •.•• ,p.585,n°21 et 28.

37: Il faut rendre justice à Stéphane Gsell qui évoqua pour Ain
Ee:1ia la possibilité qu'elle fût la demeuxe d'un chef indi-
gèrJe (Atlas arch.,34,26). A.Joly avait également émis cette

995
 hypothèoe pour la ferme Romanette,maio l'a rejetée aussitôt
(art.cit.,p.190).

38: Les hasards des découvertes épi5Taphiques ont permis de cons-

tater que certains vétérans se fixaient dans les localités
voisines du camp où ils avaient servi;d'autres,au contraire,
regagnaient leur pays d'origine à leur congé.Probablement dès
le IIème siècle,certainement au IIIème,on prit l'habitude
d'accorder aux vétérans des parcelles de terre à cultiver (R.
CAGNAT,op.cit.,pp.484 et suiv. H.PAVIS D'ESCURAC-DOISY,Lam-
bèse et les vétérans de la legio III Augusta,Hommages à Al-
bert Grenier,coll.Latomus,8,Bruxelles 1962,2,p.576. J.KOLIDJ-
DO,Le Colonat en Afrique sous le Raut-Empire,Paris 1976,p.109
n.85) et un certain nombre d'autres avantages. De cette façon
étaient assurés la culture des terres situées aux frontières
de 1 1 Empire,la défense de ces frontières et le recrutement de
l'armée.On comprend,dans ces conditions,l'avantage d'un colo-
nat militaire,non sur le limes,mais au-delà,à la ferme Roma-
nette ou à Ain Benia (ce type de colonat n'a pas dÛ être très
différent de celui qu'ont constitué les limit~1ei dès le rè-
gne de Sévère Alexandre;voir à ce sujet D. VAN BERCH~1, L'Ar-
mée de Dioclétien et la réforme constantinienne,Paris 1952,pp
40-41.).

39: GSELL (St.), op.cit.,22,63.

40: ~illRCHAND (Capitaine), Occupation romaine dans la circonscrip-

tion d'Ammi Moussa,BSGA0.,1895,pp.209,218-220.

41: Ibid.,p.209.

42: Ibid.,p.218.

43: LACAVE-LAPLAGNE, Notes sur quelques ruines romaines relevées

dans la commune mixte d'Amrni Moussa,BSGA0.,1911,pp.38 et 41.

996
44: Ibid.,p.42. Le mot castellum suscite encore de vives discus-
sions: est-ce une forteresse (Tacite,Ann.XV,3), un château
fort (st. Gsell,Les Y~numents antiques de l'Algérie,t.I,Paris
1901,p.100. Lacave-Laplagne,art.cit.,pp.38 et 42.) ou une
agglomération rurale (P.A. Février,Inscriptions inédites rela-
tives aux domaines de la région de Sétif,relanges Piganiol,
~,pp.217-228. s. Lancel,Lee Actes de la Conférence de
Carthage en 411,l,Paris 1972,pp.141-143)?

45: GSELL (St.), op.cit.,pp.100,102.

46: BALLU (A.), Rapport •• ,en 1909,BACTH.,1910,pp.100-101, a mis en

évidence des installations de vinification et de pressage des
olives. LEVEAU (Ph.), Caesarea de ~~urétanie et son territoi-
E!• Thèse de Doctorat d'Etat 1979 (inédite),p.435,confirme que
seule la façade est fortifiée et que c'est là l'un des éta-
blissements agricoles les plus importants qu'il ait découverts
lors de sa prospection de la région.De nouvelles fouilles, me-
nées par une-équipe algéro-italienne,ont été effectuées sur ce
site entre 1972 et 1975,laiseant prévoir un plan plUB compli-
qué de ces ruines.

47: E.ER13RUGGFE (A.), Antiquités du cercle de Ténès, R.Af.,t.II,18-

57-58,p.275: "Quant aux vestiges les plus nombreux ••••• ,lors-
que leur caractère essentiellement militaire nous est affirmé
par un homme de métier [il s'agit du Lt Colonel Lapa.sse"f] ,
nous noua en rapportons à son appréciation,aauf dea chances
d'erreur exceptionnelles;car il peut très bien arriver qu'un
établissement purement agricole ae forme sur un emplacement
stratégique".

48: LEVEAU (Ph.), Recherches ••• ,~~.,t.LXXXIX,1977,pp.257-311 et

surtout p.258,n.5.

49: DE CAUSSADE,Traces d'occupation romaine dans la province d'Al-

ger,Y~moirea de la Société Archéologiques de l'orléanais,1§21,

997
 p.243 et suiv; p.290,l'auteur donne le texte de l'inscription
qui lui a permis d'affirmer l'existence dea castra auziensia :
Iuliae/ Aw~tae)/ Fatri/ Caatrorum/ Auzien~es •••• (CIL VIII,
9032).

50: TROUSSET (P.), Recherches sur le limes tripolitanus du Chott

El Djerid à la frontière tuniso-libyennetParis 1974,pp.136-139.

Slmmary

Under the r~le of the Severi, a praetentura, a kind of impassable

frontier road, was established on the edge of the plains in Mauretania
Caesarea, in order to link sorne of the camps and fortified town s. In
the area which interests us, the road runs from AJh Teukria to Maghnia
with a break between Tiaret and Albun 3>iba. This, theoretically, was
the boundary of Roman occupation in Mauretania Caesarea, although a
few Roman remains have been found beyond that limit; su ch as the
Romanette farm, Ant Benia and Ant bent &>ltane. Do the se remains
belong to sorne outposts on the limes responsible for protecting the
more adventurous settlers, or for guarding the praetentura? On the
other hand, are these the remains of fortified farming developments?
Existing records allow us to attribute one or other of these ftmc-
tions to the Ho manette farm and to A1h Benia, though it is not certain
which function. The information for AJh bent &>ltane is too patchy to
enable us to be as certain as those who found and reported the remains.
How often has a military significance been systematically but
wrongly attributed to sorne 'good strategie position' or 'perimeter 20
by 30 paces' in the 19th and even at the beginning of the 2oth century?
Only archaeological research and epigraphie discoveries could help
us to learn more about these sites.

998
 67. CARTHAGE: THE LATE ROMAN DEFENCES

C. M. Wells

Three teams taking part in the UNESCO "Save Carthage" campaign have
excavated sections of the late Roman city wall, the so-called Theodosian Wall
of circa A.D. 425.1 Of these, the Italians have not yet published a report;
their dig was on the northern edge of the city, west of the cardo maximus.
The British excavated part of the Wall on the south side of the city; their
interim reports have appeared in the Antiquaries Journal from 1975 onwards.
The most extensive invesitgation of the Wall has however been by the second
Canadian team, directed by Dr. Edith Wightman and myself, 2 on the north of
the city, east of the cardo maximus, between the odeon and the Damous el-
Karita (Fig. 67 .1, showing the site at the end of the 1978 season). In this sec-
tor we have shown that the Wall, now heavily robbed, lay at the foot of the
Teurf el-Sour escarpment, which was itself formed by erosion after the Wall
was robbed, the ground level having been higher inside than outside the Wall.
Indeed the whole plateau south of the Teurf el-Sour has proved to be artificial,
the result of Roman build-up.
Our main site lies along decumanus VI, the decumanus ultimus of the
original Augustan colony, 3 between cardines II and Ill east (Fig. 67. 2, end of
1979 season), The Wall at this point was built on the line of decumanus VI,
largely obliterating it, and abutting against the outer walls of the two houses
on the south side of the street. These were solidly built, with mosaic floors,
and they continued to be occupied after the Wall was built, probably until the
end of the 5th century, after which they were destroyed or abandoned, although
the site was reoccupied in the 7th century. One house projects further into
the street than the other; the Wall consequently varies in width, from 3 m to
2 m.
The Wall then turns to the north-east, leaving a stretch of decumanus VI
just west of cardo Ill as a dead end with no function save possibly to give ac-
cess to the Wall. This stretch was used as a garbage dump and has yielded
a most important deposit of pottery from the second quarter of the 5th century. 4
On its new line the Wall passes over the end of a cistern lying north of decu-
manus VI, which was reduced and filled; this too has yielded much pottery
and also glass.
Where it turns, the Wall reveals a section through decumanus VI, with
a well-built vaulted drain 1. 5 m high. Where this drain underlies the Wall,
it has been filled with stones and mortar and serves as a foundation for the
ashlar blocks which formed the facing of the Wall at this point. Two survive
in situ, but traces of others can be seen in the mortar on top of the drain.
The core of the Wall has also mostly been robbed, but it was set in a charac-
teristic grey mortar which still adheres to the surfaces against which the
Wall rested.

999
I /
Fig. 67.2

2CC2

/ ',~/
/ Room"3--

BEB
CARTHAGE
NORTHERN SECTOR
CANADIAN EXCAVATIONS
Main Site, excavations 1976-79 \
------------
0 1 2

RECORDING I DRAWING
3 4 Sm

ALAIN EPIDMfR 78
A
 The date of the r@bbing is not known. The fill of the robber trench
yielded a very few fragments of medieval Islamic pottery," while we found
resting on an undestroyed portion of the core, 1. 5 m below the present ground
level, in the robber trench, a rotten canvas bag containing a wad of newspaper
and some 200 dynamite fuses. The newspaper contained an item about a pro-
test from Crowley (Sussex) Parish Council to Lord Leathers, Minister of War
Transport, that Italian prisoners-of-war were getting a ride to work, while
British workmen had to cycle or walk. U we could be sure that this was not
a later intrusion, it would mean that the robber trench was still open, and the
Wall presumably still being robbed, after May 7th, 1943, the date when the
British captured Tunis.
Further east, we have undertaken a number of trial trenches along the
line of the Wall under Dr. Wightman' s direction, although nowhere as yet have
we extended a trench northwards to see if there was a ditch; the British found
one on the south of the city, the Italians established that no ditch existed on
the north-west.
Between cardo ill and cardo IV, the Wall follows the line of the rural
centuriation, which began north of decum.anus VI. It then turns east again,
and here in the angle (Fig. 67 .1, 2CC7) we have found and excavated a tower,
roughly square, 8 m wide and projecting 7. 5 m from the Wall. Here the facing
of the Wall was of substantial limestone blocks bonded with fine hard whitish
mortar. The interior of the tower at foundation level comprised a stone packing
set in red clay or with grey mortar poured over. At a higher level were lar-
ger, roughly squared sandstone blocks on mortar platforms, few of which sur-
vived.
We thlnk it likely that the Wall had towers at all the external angles. At
the next such angle to the east, the size of the platform visible in the bank of
the Teurf el-Sour makes us wonder whether there may not have been a large
polygonal tower. Parallels with the contemporary defences of Constantinople
may be instructive.
Between cardines V and VI, there is, exceptionally, a short stretch of
Wall still upstanding (Fig. 67 .1, 2CC4). A section across the Wall (Fig. 67.
3) showed that here, contrary to what Was observed on the main site, it was not
built against existing- structures, although the ground level inside the Wall
had been raised after construction, making it 3. 25 m above the ground level
outside, and suggesting the presence of structures at a higher level in the.
vicinity. The Wall foundations at this point were 3. 5 m wide, as on the British
site. The facing stones had been robbed. We reopened and widened a trench
dug here by Poinssot in 1923, and were able to establish that what he and
Lantier had taken for a medieval earth rampart was two phases of an intervallll!!?
road, the earlier of which may p.erhaps be ~ssociated with Count Belisarius's
repairs to the Wall in the 530s. 5 Other possible traces ~f Belisarius's acti-
vities include a heap of constru.ction material in one of the houses on the main
site, up against the back of the Wall, the house itself having presumably been
abandoned.
By the early 5th century, when the Wall was built, Carthage had expanded
beyond its theoretical northern limit of decumanus VI; indeed further east than
our sector, along the coast, there was a decumanus VII, if not VIll, between

1002
 Trench 48
+ + + + + +

+ +
Road
~I' 1 surfaces
~;;, ~
1
.. <h • oc,
f';;;;;
_~•·:-;:-~- £"-"~:\:; •:• ,---- __.
+ + +
·, • I .
I V
' , I
• I • '\ "
( . r a \]

-' .+ . . . +
Interior ground level
after Wall built
..... (red clay)
0
0
i:,:)
+ + -t + + + + + + +
Build· up going
with Wall foundation
~
4C
+ -t -t + + + + + +

-t
CARTHAGE
NORTHERN SECTOR
CANADIAN EXCAVATIONS +
Section, Trenches 48- 4C, 1978

------
------
o
EMW78
1

Fig. 67. 3
2 3 4 5m

Red clay packed against

Wall foundation
cardo IX and the sea. The Wall shows every sign of having been built in a
hurry, in a somewhat irregular and haphazard manner, to enclose the built-
up area. It failed however to keep the Vandals out, and the subsequent decline
of Carthage is also a matter of archaeological record, though not a subject
for this paper.

NOTES

1. For the UNESCO campaign and summaries of the various excavations,

cf. the CEDAC Carthage Bulletin (no. 1, Sept. 1978; no. 2, June 1979).
The date (A.D. 425) is given by Chronica Gallica (see Mommsen 1892,
658). On its reliability, cf. report by F. M. Clover in Humphrey 1978,
9; also Hurst 1975, 36-8.
2. Brief preliminary reports in the bulletin of the Classical Association of
Canada: Echos du monde classigue/Classical News and Views xxi (1977),
15-23; xxii (1978), 9-12; xxiii (1979), 15-18. A slightly fuller report
on the first two seasons' work (1976 and 1978) is due to appear in Journal
of Field Archaeology late in 1979. A report on the 1979 excavations will
appear in Echos du monde classigue/Classical News and Views xxiv (1980).
3. On the street layout, the basic artiCle is Saumagne 1924; also Davin 1930;
Saumagne 1931.
4. For a preliminary report, see Neuru, forthcoming.
5. Poinssot and Lanthier 1923; on Belisarius' activities, Procopius, de aed.
vi. 5; de bello Vand. xxiii. 19-20.

REFERENCES

Davin, P., 1930. "Etudes sur la cadastration de la Colonia Iulia Carthago",

Revue Tunisienne 1930, 74-85.
Humphrey, J. H., 197 8. Excavations at Carthage 1976 conducted by the
University of Michigan, Ann Arbor, vol. 4.
Hurst, H., 1975. "Excavations at Carthage 1974: First Interim Report",
Antiq. J. 1v, 11-4 0 .
Mommsen, T., 1892. Chronic a Minora I, Berlin.
Neuru, L., forthcoming. In Antiquities Africaines.
Poinssot, L. and Lanthier, R., 1923. "Notes de topographie carthaginoise:
une enceinte de Carthage", Comptes Rendus de 1' Acad. des Inscriptions
1923, 306-11.
Saumagne, C., 1924. "Colonia Iulia Karthago",Bull. Archaeol. 1924, 131-40.
Saumagne, C., 1931. "Les recherches recentes sur la topographie de
Carthage ", J. des Savants 1931, 145-57.

1004
 68. THE POI.JTICAL SIGNIFICANCE
OF AUGUSTUS' MILITARY REFORMS

Kurt A. Raaflaub

Suetonius writes in eh. 49 of his Divus Augustus: "Of his military forces
he assigned the legions and auxiliary units to the various provinces... The .
remainder he employed partly for the protection of Rome, partly as palace-
guards.. • • He also standardized the pay and allowances of all the soldiers,
wherever they were stationed, designating the duration of service and the
rewards due on its completion according to each man's rank, in order to keep
them from being tempted to revolt after their discharge on the excuse that they
were either too old or too poor to earn an honest living. In order to have
sufficient funds always available for maintaining the soldiers and paying the
rewards due to the veterans, he established a Military Treasury, supported
by new taxes • "
This is a brief summary of Augustus' military reforms which partly
emerged as the result of experiments and a series of changes during his entire
reign, partly were introduced by specific legislation in 13 B. C. and A.D. 6.
From other sources we know enough about the details: the amount of pay;
the duration of service (12 years for praetorians, 16 for legionaries after
13 B.C., extended to 16 and 20 years respectively plus five years of service
in special reserve units in A. D. 6; some more for the auxiliaries); the
conditions of discharge, especially the kind and amount of retirement rewards
(down to 13 B. C. usually consisting of a piece of land, afterwards of a lump
sum equalling about the pay of 13 years); the origin of the enormous capital
necessary for both these purposes (before the establishment of the aerarium
militare from Augustus' private fortune, afterwards from a 5% inheritance
and a 1% sales tax); the marked difference in status, rewards and treatment
between auxiliaries, legionaries and the praetorians who enjoyed many special
privileges.
Even more important than the details is the basic significance of those
reforms. They formally institutionalized the Roman army as a standing army
of professionals. They replaced the insufficient temporary improvisations
of the Republic by a permanent organization that was far better able to serve
the needs of the empire and therefore basically remained unchanged during
the next two centuries. And they represented in many respects a valid and
final answer to some of the most difficult social and political problems of the
Late Republic-problems which the Senate had never managed to solve and
which had decisively contributed to the succession of civil wars, to military
dictatorship and the breakdown of the Republic. To obtain a clear understanding

1005
of the political significance of Augustus' military reforms is, therefore, a
matter of prime importance for every attempt to understand the emergence,
character and significance of Augustus' Principate.
A proper understanding can only result from appropriate questions. That
one has to start with a careful examination of the problems Augustus had to
face and of the solutions he offered--individually and in their mutual relation-
is obvious enough. On that basis the following questions seem to be especially
important: 1) Were the reforms of Augustus, if compared with the problems
and needs, adequate, that means: were they radical, comprehensive and
efficient enough? In other words: Did he really solve the problems or did
he only cure some of the most obvious symptoms? 2) Since the military re-
forms are only one part of a comprehensive rearrangement of the social and
political order (which Sir Ronald Syme has called the "New State"), what is
their place in this complex system? How do they relate to regulations and
measures in other political areas? 3) The effectiveness of reforms has to
be measured by the disappearance or recurrence of the problems which they
were designed to solve. How, then, did the relationship between armies and
state or society develop in the decades after Augustus ? Are there any events
or phenomena that reveal problems comparable to the ones at the end of the
Republic?
My premise, accordingly, is simple, obvious and demanding at the same
time: The military reforms, as central as they may be, can only be judged
adequately, if they are analyzed in the framework of their political context,
their historical background and their consequences and effects. It is necessary,
therefore, to go back at least to Marius and the important changes connected
with his name, to follow the later development at least down to the period of
the next great civil war of A. D·. 68-69, and to comprise the whole complex
of political problems which Augustus had to face.
Modern scholars usually include only one or two of these three historical
dimensions in their study of periods that appear as the "threshold" or ''border
area" (rather than "line") between two distinctly different epochs. The reign
of Augustus which obviously m~rks such a fundamental transition, is no ex-
ception. It is generally approached either from the side of the Republic and
then seen as the very end of a specific epoch, or from the Empire and there-
fore taken as the very beginning of a new era, or else studied rather narrowly
by itself. But hardly ever is it comprehended as the central piece of a long
period (extending over at least about two hundred years) of gradual transition
and only occasional abrupt changes in almost every area of Roman constitu-
tional, political and social life. Here, as elsewhere, only a "three-dimen-
sional" approach which deliberately bridges the threshold between the epochs
can give us the full historical perspective and understanding.
In this paper I intend to present a first (and preliminary) summary of the
results that can be reached by such a broad approach. I shall therefore con-
centrate on the main lines of development and on showing the complexity of
the problems involved. After a brief study of the preconditions and the details
of the regulations of Augustus I shall especially try to find a more differen-
tiated answer to the question of their effectiveness and appropriateness. By
applying this method of analysis to. one important part of the settlement of

1006
Augustus, I hope to contribute to a better overall understanding of the possi-
bilities of and limitations on solving political and social problems under the
specific conditions that characterized the Principate of Augu.stus.

II
What, then, was the nature of the problems to be solved? They were,
roughly, the result of a basic contradiction between theory and practice of
the Late Republican military system. In theory (which determined the legal
situation, the institutions and the official policy of the Senate) the traditional
principles of the citizen-militia of a city-state were still valid: All citizens
who met the census-requirements were obliged to serve in a fixed number of
yearly campaigns; armies were formed by senatorial decree, temporarily
and in emergencies only. In practice, however, since Marius the recruitment
of volunteers, mainly from the mass of rural proletarians, had become the
rule. For this new type of soldier the service was a means to secure a living
rather than a civic duty; the hope for material rewards like good pay, rich
booty and especially an adequate veteran "pension" was his main incentive.
The Senate which had never regulated or even legalized, only tolerated the
improvisation of Marius, was, for several reasons, unable and unwilling to
fulfill these demands. Instead, the commanding generals became leaders
and representatives of the soldiers and their interests, in military life and
beyond. The traditionally close and personal relationship between soldiers
and generals thereby was considerably intensified and developed into a specific
kind of patronage and clientele that was based on effort and personal loyalty
on one side, military and social responsibility and success on the other.
To understand the political implications of this ''army-clientela" one has
to consider two characteristic features of Roman society. First, despite the
recent changes in the social composition of the armies, the traditional identity
of soldiers and citizens was still generally valid. Consequently, soldiers on
leave or veterans could vote and, if necessary, even form effective pressure
groups in the assembly. Second, the original unity of military and civil com-
petence in the Roman magistracy was still expressed by the simultaneousness
or at least succession of civil and military functions in the pattern of political
careers. A general could, therefore, not only use his soldiers or veterans
for direct }.X>litical blackmailing by military force, he could also, after the
end of his command, exploit the voting power of his veterans in order to
achieve certain political goals. Thus the army-clientela could be transferred
directly into political clientela. Since the veterans could not permanently be
present in Rome, this specific kind of support could probably be activated
only for votes on a few special topics (like agrarian bills, regulations or
extensions of provincial or extraordinary commands), in elections and in
emergencies for the protection of their patronus. But these were the most
important occasions for the formation of personal power.
These are the main preconditions of a process in which the Roman pro-
letarian soldiers, led by unscrupulous and power-hungry members of the
senatorial aristocracy that had long since lost its unity and homogeneity,
became a decisive factor in political life. In the course of this process the
use of armies for political purposes was not only steadily increased and

1007
refined, the soldiers became increasingly aware themselves of their important
role. The result and climax was the development of something like an inde-
pendent policy of the legions in the years after Caesar's assassination. Led
by their centurions and tribunes, the soldiers forced their generals to avoid
unnecessary battles and conclude treaties and alliances. Their political goal
was to maintain the unity among the heirs of Caesar an9. thereby to improve
the chances for the realization of the promised distribution of money and,
much more important, of land.
To sum all this up in two ugly but appropriate catchwords: The lethal
crisis of the Republic was characterized by the militarization of politics and
the politicization of the military. This was not the only but certainly the most
important and immediate factor that caused the breakdown of the Republic.
Whoever undertook it to end the age of the civil wars and establish a sound and
lasting JX)litical order, faced in the first place the necessity of eliminating
this factor. When Octavian emerged as victor and sole ruler after Actium,
it was, indeed, his most urgent and important task to make sure that the
soldiers and officers unlearned their political role and that the military and
political spheres were henceforth clearly separated from each other again.

III
It is pretty safe to assume that Augustus had to attack the problems on
two levels: He had ideally to exclude, realistically at least to reduce as far
as possible (a) the. need and readiness of the soldiers to interfere (or to be
used as an instrument) in politics; (b) the probability or even the JX)SSibility
of any p:>litical use of military power by anyone (except, of course, by the
Princeps himself).
Before we enter the discussion of these two aspects we have to consider
Augustus 1 freedom of action and range of possibilities. On first sight, one is
tempted to think that they were almost unlimited. Not only did Auguetus after
Actium in fact control the entire military and p:>litical p:>wer and was he ap-
parently able to use it at will. But there are indications that a majority of his
comtemporaries after three generations of civil wars desperately expected
him to restore peace once and for ever and that they were ready to accept
rather profound changes and the establishment of some kind of a unified and
superior political and military leadership, because they understood that only
a strong princeps civitatis could prevent the outbreak of new civil strife.
But there were decisive limitations: First, Augustus had to work with
the existing comJX)nents of Roman society and p:>litics. He especially had to
base his planning on the existing capacitiE-s. For example, there was, for
several (practical and other) reasons, simply no way to completely and quickly
replace the senatorial class in political, administrative and military leader-
ship. Second, he had bound himself by the promises he had given when he faced
the necessity of winning the supp:>rt of the leading classes for the struggle
against Sextus Pompeius and Antonius. The "restoration of the Republic" was
from then on promise and program, inevitable and compulsory for the estab-
lishment of peace and order as well as for his own credibility. ·whatever we
think about the realization of this program, in several imiXJrtant matters
Aug·ustus had no choice: He could not rep lace the citizen. army at once

1008
by mercenaries or any alternative military system. He could not completely
change the traditional unity of civil and military functions in the Roman
magistracy. And he could not exclude the Senate or the individual senators
from the administration of the provinces.
Third, there was the unfortunate experience of Caesar, the Dictator.
And Octavian was a rather cautious man. The mistakes of his adoptive father
certainly influenced his considerations quite distinctly. The lesson was clear:
A dictatorship based only on military power and the enthusiasm of the lower
classes without strong support in the upper classes was bound to fail. It was,
under these circumstances, at present not advisable, if not impossible, to
try to rule Rome without or against the Senate and especially the old nobility.
Keeping all this in mind, let us now turn to Augustus' actual measures:

IV
In order to depoliticize the armies he had first of all to reduce their need
and readiness to assume a politica.l rol2. He therefore had to eliminate the
reasons that had in the past induced or even forced them to politicize. This
could be done (a) by fuliilling old promises and complying with the justified
demands of the soldiers, and (b) by eliminating their dependence on their
commanding generals for pay and rewards during and after their service.
To take care of the veterans, accordingly, was Augustus' most urgent
concern. Between 41 and 13 B.Co, several hundred thousand men were set-
tled in a vast number of colonies in Italy and the provinces. The land assign-
ments, at least down to the years after Actium, however, were dictated by
immediate necessities; they satisfied the soldiers (who had expressed their
demands once more very clearly in the mutiny of Brundisium of 30 B.C. ),
they eliminated the legacy of the civil wars and at the same time helped to
reduce the armies to a reasonable size. Despite a few additional measures
the pressure of these emergency-actions for a long time prevented Augustus
from developing a general solution for the problem in the future 0

Such long-term solutions w_ere the aim of his military reforms of 13 B. C.

and A.D. 6, which I have described b~fore. In essence, he thereby officially
recognized and legally regulated the professional status of the soldiers and
their corresponding social and material needs. While retaining in principle
the character of the army as an army of Roman citizens, he nevertheless
was able to draw the necessary consequences from the century-long develop-
ment since the creation of the proletarian army by Mar ius 0

The introduction of clear and binding regulations at the same time was
designed to eliminate the traditional dependence of the soldiers upon their
generals. They now received their pay and retirement rewards from the
emperor or the financial administration in Rome, no longer from their tem-
porary commander. The previous form of army-clientela was thus consider-
ably weakened. It could be expected that the inclination of the soldiers toward
following a general into a political revolt would henceforth be reduced. Our
sources mention this quite rightly as the main purpose of the reforms.

1009
 V

It was, however, obvious that by merely satisfying the basic material

needs of a professional army-as big an improvement as that might have been-
further political interventions could not sufficiently be prevented. Strong
supplementary measures which made an inappropriate use of the armies, if
not impossible, at least very difficult, were indispensable. The legal and
practical preconditions for such measures had been created early. Already
in 38 B. C., Augustus had underlined the importance of his military leader-
ship by assuming an unprecedented praenomen imperatoris. In 2 9 B. C. , it
was accepted and given permanence by the Senate; from then on it was part
of the titulature of the Princeps. The long-term proconsulship which was
voted to him in 27 B. C., made him governor of the three most important
military provinces of Spain, Gaul-Germany and Syria. He thereby not only
became the supreme commander of at least four fifth of the armed forces
with the right of nominating the generals of armies and legions, but was also
able to exert a dominant influence on decisions concerning the "foreign policy"
in the most important frontier regions. His control in these two areas became
almost absolute when Illyricum was added to his share in 12 B. C. and the
Macedonian legions were moved out of the province into the recently conquered
territories of Moesia later in his reign, so that Africa with one, at times two,
legions remained the only senatorial military province.
This long-term concentration of the supreme command of almost the entire
armed forces in one hand was, despite certain precedences in the Late Republic,
unique and new. Its basic significance is obvious. Using those powers,
Augustus in the subsequent years introduced a vast complex of measures· which
were indispensable for the stabilization of his own position and the political
order.
1. The drastic reduction of the number of legions, resulting in a danger-
ously unfavorable relation between the size of the empire (especially the length
of the frontiers) and the size of the army, must have been motivated, not only
by financial considerations and calculations of the available manpower, but
also very much by reasons of internal security: The fewer the legions the
smaller the probability of revolt.
2. Similar considerations could have been res{X>nsible for the decision
to distribute the legions and auxiliary troops in scattered camps all along the
border of the empire and thereby to avoid any large concentrations in the
same area. We know little about the location of the legions in the time of
Augustus, but the existing indications make it probable that garrisons with
two legions were exceptions, actually confined to Germany. Relatively little
rx>wer was thus concentrated in one place; potential troublespots could be
isolated and played off against each other. That some of the most serious
troubles later originated L11 areas of such (permanent or temporary) concen-
trations, proves the accuracy of Augustus' assessment.
3. That the legions were stationed in the frontier regions, as far away
from Italy and Rome as possible, certaic.:y was a (psychologically and politi-
cally advantageous) demonstration of the break with the past (the civil wars),
of the geographic separation of civil and military spheres and of the (at least

1010
officially) sole and only future purpose of the army: the defense of the fron-
tiers, warfare against external enemies. One should, however, be aware
that Augustus thereby sacrificed the goal of a complete reintegration of the
soldiers into Roman society. The split between unarmed and armed citizens
which had emerged during the civil wars with disastrous consequences, was
perpetuated and deepened. The military professionalism soon led to the de-
velopment of a military class with the well-known effects visible already in
68-69, but much more clearly in the late 2nd and 3rd centuries.
4. Besides these more technical precautions, Augustus concentrated
his efforts in the development and consolidation of a uniquely close relation-
ship between the soldiers and himself and his family. Here as in many other
areas he drew the logical conclusions from the experiences of three generations
of civil wars. He, as all the great men of the Late Republic, had won his
powerful position through the loyalty of his victorious armies. Their loyalty
had been carefully built upon successful leadership and generous patronage.
Like his adoptive father he had, after his final victory, inherited with the
command also the patronage of all existing Roman troops. He was not going
to give it away any more; command and patronage of the whole army were
henceforth monopolized by the Princeps and his family-at least as far as
possible.. As long as all the soldiers and all the veterans felt (and were fre-
quently and strongly enough reminded) that they were the clients of the Princeps
and only the Princeps, as long as no conflicting loyalties were given a chance
to develop, there was little danger of p:>litical interference. Every action
connected with the army, from recruitment, to warfare and triumph, to reward
and discharge, had henceforth to happen in the name of and through agents of
the Princeps. A few examples may illustrate that:
The sacramentum was sworn to the Princeps by all units, even those
stationed in senatorial provinces. Donatives were distributed to the soldiers
from the private treasury of the Princeps, not only after victories and at
triumphs, but also on important political occasions. It is especially revealing
what Cassius Dio reports about the presentation of C. Caesar, the presumptive
successor, to the army in 8 B. C. : "Augustus granted money to the soldiers
not as to victors ••• but because then for the first time they had Gaius taking
part with them in their exercises" (55 .6 .4). The discharge benefits in land
or money were distributed by the Princeps and, at least until A. D. 6, out of
his money. Augustus himself tells us how many hundreds of millions he spent
for land purchases or cash rewards of no less than 300,000 veterans ~
Gestae 3. 3, 16 .if.). His personal representatives distributed the money to
the soldiers in the provinces at the occasion of their discharge. All the dozens
of veteran colonies he founded of course bore the name of either colonia Julia,
Julia Augusta or Augusta: a powerful and permanent reminder. And when
the financial demands of his discharge program finally forced him to transfer
these expenses to a new public treasury, he made sure that the initial funding
came again out of his pocket and that only cities and (foreign) vassal kings,
but not Roman individuals, were allowed to contribute to it: there must not
be any conflicting merits and claims.
5. The treatment of the centurions was a matter of special concern
because, in the last civil wars, they had been the leaders and speakers of the

1011
legionaries and therefore exerted strong political influence. Augustus offered
them high material advantages, generally improved their status and increased
their competences, and created for them a new and differentiated career-
scheme that included the possibility of social promotion into the ordo equester.
The centurions thus became a privileged corps of officers; they owed every-
thing to the Princeps, they thanked him with unconditioned loyalty, and hence-
forth represented an important element of stability.
6. However, the ·allegiance of soldiers, officers and veterans was not
based only on material rewards or on social care. It was formed and developed
through personal contact, leadership and success. Here, too, the Princeps,
his family and his trusted friends claimed a rather rigorous monopoly. The
armies and legions were (with few exceptions) commanded by legati Augusti,
officers who were nominated by and acted in the name and under the auspicia
of the supreme commander. All their actions and successes, therefore,
were in fact attributed to the Princeps, which is best illustrated by the 21
salutations as imper~tor he claims in the Res Gestae (4 .1 ), and the 55 thanks-
givings noffered to the immortal gods on account of the successes on land and
sea gained by me or my legates acting under my auspices" (4 .2). To be hailed
as imperator and to celebrate a triumph soon became the exclusive privilege
of the Princeps and his family-members. The last triumph celebrated by a
person not belonging to this circle dates to 21 B. C. , the last imperator ac-
clamation to A. D. 24. The importance attributed by Augustus to military
success and triumph as basis for an especially close relationship between
general and army is further under lined by the fact that only very few of the
early triumphatores are known to have obtained another important military
command afterwards.
It was the natural consequence of this principle that all major military
actions should actually be led by the Princeps himself or at least by his closest
relatives. It is therefore significant that even after 27 B. C. Augustus person-
ally commanded the last stages of the pacification of Spain and Gaul and was
at least present himself in the East for some time; that he delegated his
stepson Tiberius to regulate the Armenian succession and negotiate the resti-
tution of the standards from the Parthians in 19 B. C.; that he sent his yotlllg
grandson C. Caesar to a similar mission twenty years later; that he was
present in Gaul when the drive toward shorter and safer frontiers in the Rhine
and Danube areas was launched with the conquest of Rhaetia and Noricum;
that these conquests were conducted by Drusus and Tiberius, etc. Further,
it should at least be mentioned here that consequently also Augustus' external
policy (at least as far as it included the possibility of military engagements)
must largely have depended on considerations of internal security, and that
the Princeps obviously only abandoned his cautious policy of consolidating
already conquered territories and started a new program of systematical
expansion when his own family finally could provide able commanders for the
big armies necessary for such conquests.

VI
It is hardly necessary to continue along these lines. There can be little
doubt that Augustus' comprehensive program of recognizing and fulfilling the
basic needs of a professional army ·and monopolizing its leadership and

1012
patronage indeed succeeded in reducing to a great extent the danger of its
detrimental politicization. This certainly is an impressive political achieve-
ment that deserves our appreciation or admiration, even though its long-term
effectiveness has often been overestimated in modern scholarship. There is
no space here to deal with individual opinions. The vast majority of scholars,
however, seems to agree that the system of Augustus, based on the two pillars
that have been described above, was sufficient to pacify the armies for two
centuries. In this view, the civil wars of 6 8-6 9 and 193-197 appear as "inter-
ruptions11 or "accidents", as grave but still surprising eruptions of disobedi-
ence which should actually have been excluded by this system.
This conclusion seems to me to be questionable. It is the purpose of the
following discussion to find out how far it can be maintained. The primary
question is: Were Augustus' measures in the military sphere really compre-
hensive and fundamental enough to permanently eliminate the danger of a new
politicization of the armies? My doubts are based on the following consider-
ations:
1. Some of Augustus' measures, e.g. the change of address from
"Comrades" to "Soldiers" (Suet. Div. ~· 25.1) or the establishment of the
aerarium militare as a state treasury, show a deliberate effort to rationalize
his relationship to the armies, to put some distance between Princeps and
soldiers. Nevertheless, the army remained a personal army. For several
reasons this could not be avoided. It was even indispensable because, in view
of the basically unchanged social and political structures, the army could not
really be confined to the function of defending the frontiers. For the sake of
peace and stability (and contrary to the appearance and official representation
that have been mentioned above), its duty also was to secure the power and
rule of the Princeps. To be sure: In the place of a plurality of possibly
competing personal armies there remained only one now; and the Princeps
had indeed concentrated the whole army-clientela on his person and was de-
termined to prevent the formation of rivalling clientelae. Nevertheless, the
old mechanisms were still working. Since it had only been possible to achieve
stability by intensifying and expanding the very methods which previously had
prevented stability, the possibility remained that in changed times and circum-
stances these same methods might again threaten stability. Even if clientela
was generally taken as a strong obligation, even if it was hereditary, it still
could be replaced or superseded by other obligations; temporary allegiances
or advantages might prevail.
2. The loyalty of the client had always largely depended on the success
and care of the patronus. Certainly, Augustus had fulfilled the demands of
the soldiers and improved their situation tremendously. Accordingly, it can
be argued, the social needs of soldiers and veterans no longer were a political
issue. Nevertheless, there were upsetting factors: It took quite a long time
until the new regulations worked satisfactorily; in A. D. 14, the complaints
about their failure were the main reason for severe difficulties. Moreover,
the pay of the soldiers was, as far as we can see, hardly generous; nor was
it ever adapted to inflation between Caesar and Domitian. One further gets
the impression that even the retirement rewards, if compared with the costs
of land, did not allow the veterans to establish themselves very comfortably.

1013
All this helps to understand why the profession of soldier soon lost so much
of its attraction especially in the traditional Roman recruiting areas-a process
which can clearly be derived from the statistics listing the origin of the legion-
aries in different periods.
It is hardly surprising, therefore, that the donatives more and more were
considered to be a regular part of the pay; they were taken as a right, no
longer as an exceptional and voluntary gift. In general, such material consid-
erations again became very influential, even more than one should expect
anyway; and that means that they could play a decisive role in determining
loyalties. Furthermore, the privileges and power bestowed on the centurions
created tensions between soldiers and officers that occasionally erupted in
ugly acts of revenge and could, under certain conditions, be exploited by the
generals. To loosen the discipline thus became a possible method of gaining
popularity and forming bonds of loyalty against the Princeps.
3. Augustus had, as far as possible, concealed the fact that the army
also had to serve as the main guardian of internal stability. His successors
departed from this policy; for several reasons they emphasized this function
which was mainly exercized by the praetorian guard. From Tiberius on, it
was concentrated in Rome and was increasingly used to exert pressure on the
Senate and to influence political decisions. In this context, the do natives
became a means of securing the loyalty of the troops in critical situations,
especially at the discovery of (real or alleged) conspiracies or at imperial
successions. In such cases, unlike after victories or at triumphs, the troops
did not receive a reward for an extraordinary military achievement, but
rather because they had simply declared their loyalty or even because they
had failed to show any disloyalty. The consequences were grave: loyalty
again became venal, the do natives became the precondition of loyalty, the
troops 11 maden the emperor. Certainly, they preferred to transfer their
allegiance within the family of the Princeps because they were bound by
clientela and were interested in the continuity of their material privileges.
The ascendance of Claudius is a characteristic example for this attitude.
But if there was a choice of possibilities, money could influence the decision
for one candidate and against the other. Such was the case of Nero. From
this point, to the proclamation of an outsider who excelled by birth and social
prestige, it might have been a smaller step than one is inclined to think. And
this was probably even more true for the legionaries who did not enjoy the
special privileges of the praetorians.
4. Certainly, until A. D. 68, and then again until193, no major civil
war broke out. We should not forget, however, that between the ascendance
of Tiberius and the death of Nero there were several military revolts. Even
though, before 68, they failed without exception, their examination provides
us with some useful clues.
In A. D. 42,. for example, the governor of Dalmatia, L.Arruntius Camillus
Scribonianus, tried to instigate his troops to revolt against the emperor. He
had been one of the candidates of the Senate for the succession of Caligula
and feared now, like others, the suspicion and revenge of Claudius. His
attempt failed completely: after only five days the soldiers decided to stick
to their oath of allegiance. However, initially they had accepted the proposal

1014
of revolt, and what caused them to change their minds was not their bad con-
science because they had planned to break their oath, but the political program
of their leader. When he promised the restoration of freedom and Republic,
they could only think about trouble, disorder and civil war, nothing desirable.
Usually, however, such revolts did not fail because the soldiers refused
to participate. Quite the contrary. Two early cases may illustrate that
(there are more):
In A. D. 18, Tiberius sent Germanicus to the East. He had watched the
successes and the JJOpularity of his nephew among the troops in Germany with
increasing suspicion in the years before and decided to prevent similar develop-
ments in the East. Cn. Calpurnius Pi so was installed as governor of Syria
and obviously had clear orders to secure the loyalty of the Syrian legions and
keep them from any close attachment to Germanicus. 110n reaching the army
in Syria 11 , writes Tacitus, "Piso was lavish with gifts, bribes, and favours
even to the humblest soldiers. He replaced company-commanders of long
service and the stricter among the colonels by his own dependants and bad
characters. He allowed the camp to become slack, the towns disorderly, and
the men to wander in undisciplined fashion round the countryside. The demor-
alization was so bad that he was popularly called 'father of the armyw (fac.
Ann. 2.55.5). However exaggerated this picture may be, it vividly recalls
the methods Sulla had used to spoil his army and prepare it for the imminent
civil war. Piso had acted on orders of Tiberius, and the outcome of events
later prevented him from using the weapon he had sharpened. But the weapon
had been there.
Soon after the death of Augustus, in A. D. 14, a mutiny broke out first
among the Pannonian, then among the German troops. It was caused by the
discontent about the setbacks in the realization of Augustus' military reforms,
by the exhaustion of the soldiers, the lack of homogeneity among the troops
and their hatred against the centurions, in short: mostly by the effects of
the military catastrophes of A. D •. 6 and 9. The legions therefore mainly
asked for social and financial redresses; in Germany they also wanted
Tiberius to be replaced by the more popular Germanicus. The loyalty of
Germanicus and the other generals, as well as the clever use of favorable
circumstances, finally brought the revolt down. But it is most revealing to
read the comment of V elleius Paterculus who, after all, was a contemporary
and an experienced officer: 11 • • • the army ••• wanted a new leader, a new
order of things, and a new Republic. Nay, they even dared to threaten to
dictate terms to the Senate and to the emperor., They tried to fix for them-
selves the amount of their pay and their period of service. They even resorted
to arms; the sword was drawn; their conviction that they would not be
punished came near to breaking out into the worst excesses of arms. All they
needed was someone to lead them against the State; there was no lack of
followers" (2. 125. lf. ).
The situation of A. D. 14 certainly was extraordinary: The first and
therefore especially difficult and uncertain succession in the Principate
coincided with exhaustion and widespread, probably rather justified, discon-
tent among the troops. But a system devised to exclude political interference
by the military could only be efficient if it also worked under extraordinary

1015
circumstances. Exhaustion and discontent of the soldiers hardly were any-
thing surprising in a Roman army; one simply had to expect that from time
to time and build a sufficient safety margin into the system.
The analysis of all military revolts down to A. D. 68 provides us with a
clear result: Only once the initiative originated, as in 14, at least partly
with the troops. Characteristically, these were, in 68, the German legions
whose long-term commanders Nero had forced to commit suicide, whose pay
was delayed, and who, by crushing the revolt of Vindex, had been elated by
military success. Even here, however, it remains a strong possibility that
their general, L. Verginius Rufus, had been actively involved in the plot. In
all other cases the generals took the initiative when they succeeded in gaining
popularity and (at least temporarily) the following of the legions against their
distant patronus in Rome. Two conclusions seem obvious: (a) The loyalty
of the troops was indeed no firm and unbreakable factor, and (b) the most
serious problem were not the troops but the army commanders.
5. To sum these considerations up: There are enough reasons not to
attribute the lack of successful military revolts down to A. D. 68 and again
to 193 only or even mainly to the effectiveness of Augustus' military reforms
proper or of his attempt to monopolize the army-clientela. Rather, we have
to consider another decisive factor: Civil wars had only become possible
because soldiers and generals had been united in their determination to fight
for their individual interests against their common opponent, the Senate.
Civil wars could, therefore, only be prevented, in the future, if the possibility
and interest of soldiers and generals to use their military power for political
ends, could be reduced· to a minimum. That means: the depoliticization of
the armies presupposed the depoliticization of the generals. Accordingly,
we have to discuss again, and more in detail, the problems of military leader-
ship.

VII
Facing this situation, Augustus had to devise a system of safeguards that
made it impossible and unnecessary or uninteresting for generals to interfere
in politics. His task was made especially difficult because, as I have pointed
out before, he had to respect the framework of Republican traditions and
structures as well as the existing capacities. Radical solutions were there-
fore excluded. He could, for example, not generally recruit his legionary
and army commanders from the equestrian class (which, moreover, at
present hardly was able to provide enough skilled and experienced personali-
ties); more than a few exceptions were impossible. Accordingly, a basic
precondition of his success was his ability to win the cooperation of at least
large parts of the senatorial class and especially its leading' families. The
details of this life-long process have been carefully investigated by several
scholars and are very illuminating.
What he gained thereby was on one hand credibility for his program of
res publica restituta and support for his goal of peace, order and stability,
on the other hand a supply of personnel in reserve for the leading positions
in army and provinces. This, however, was at the same time useful and
problematical. Many senators could still rely on a centuries-old system of

1016
family relationships and clientelae in Rome, Italy and the provinces; they
therefore possessed enormous social prestige, they were proud, ambitious,
independent-potentially powerful and therefore dangerous. It was essential
that they be kept away, if not from office and honour, at least from the sources
of real power. The competences derived from his extensive pro consulship
and his unchallenged auctoritas enabled Augustus to exercise a tight and almost
complete control on all promotions into the top military positions. All the
measures he introduced to secure his monopoly of leadership and patronage
of the army were at the same time components of a personnel policy that was
based on an elaborate and comprehensive system of safety controls. This
personnel policy deserves a closer look. Its main content which careful
prosopographical research by many scholars during the last decades has
taught us to understand, can be summarized in two principles: (a) Only a
strictly limited group of most reliable persons was admitted to the highest
military positions at all. (b) A system of administrative provisions was
designed to make it most difficult even for those chosen men to develop more
than a minimal amount of personal power. This can be illustrated by the
following details:
1 • Members of the clan of the Princeps (including the husbands of
numerous female family members), his close friends, his fellow combatants
of the early days and again their relatives were given high priority. Such
close relationships were traditionally seen as warranty of loyalty and relia-
bility. The descendants of the leading populares of the Late Republic had
generally much better chances than those of the families most closely connected
with the narrow optimate oligarchy and with the enemies of Caesar. Members
of this second group were allowed to reach the consulate and to govern sena-
torial provinces, but only very few who had either joined Octavian early or
married into his clan got positions in the provinces of the Princeps. After
23 and especially after 12 B. C., members of the old Republican nobility
appear frequently in the consular fasti. This was the demonstration of their
reconciliation with the "New State" -but not more. Between 12 B. C. and
A.D. 2, only three noble consulars entered the ranks of legat! Augusti. Of
26 remaining nobles who did not serve in the provinces of the Princeps,
thirteen governed at least senatorial provinces, thirteen non at all.
The nobiles were, however, still well represented in the governorships.
Their share was especially high in the very honorable senatorial proconsulates
of Africa and Asia (75%). In the big military provinces of the Princeps they
were less successful: they obtained still almost 5O% in Spain and Illyricum,
only one third in Syria and Germany. Even more interesting is the comparison
with the figures of the patrician nobility. Augustus himself promoted many
noble families to patrician rank as sign of especially high distinction. 29%
of the senators were patricians, but they obtained 53% of the consulships,
52% of the important priesthoods. On the other hand, they got only between
25 and 33% of the governorships. Again: Augustus was obviously very
generous in allowing the senatorial nobility to occupy the most honorable
positions, but he was clearly less enthusiastic in admitting those men to the
positions controlling the real power. The beginning of the later sharp division
between senators with purely civil and those with military careers is clearly
visible.

1017
 Besides the members of his own and his friends' families, and besides
a few equites to whom Augustus entrusted the top positions in Egypt and in
some recently conquered territories, he mainly employed "new men" in his
provinces. Between 40 and 50% of the governors of Spain, Gaul-Germany and
Syria are sons of senators of non-consular rank. One has to compare those
figures with the Late Republican ones to understand the real proportions of
this enormous change. Those men who owed their promotion iJ?.to the social
elite to the Princeps alone, could be expected to be absolutely reliable. Like
the centurions, but on a higher level, they represented an element of stability,
especially since, lacking the usually indispensable amount of inherited social
prestige, they could hardly become a challenge for the Princeps himself-at
least not yet. Their families together with the old noble clans formed the
new nobility of the early Principate. And not unexpectedly, their sons were
treated with much more reluctance. Only few of them are lmown to have
reached the same high positions in military provinces as their fathers had
before. Obviously with good reason: High office, governorships and military
commands of ancestors bestowed high social prestige on their descendants.
The Vitellii, for example, belonged to the homines no vi of Tiberius. The
youngest of four brothers was three times consul and became the most trusted
lieutenant of Claudius. His son was made governor of Lower Germany by
Ga1ba-without previous military distinction. He was welcomed by the army
"with open arms, as a gift from the gods. After all, here was the son of a
man who had held three consulships ••• " (Suet. V it. 7). Tacitus simply
remarks: "That seemed to give him enough legitimacy" (I!ist. 1. 9 ) . Only
a few weeks after his arrival he was proclaimed emperor by the troops.
2. The career and performance of the generals was carefully planned
and supervised. As soon as they became, not only successful, but too proud
of themselves, too ambitious or too independent, their careers came to a
quick end. Their commands were usually short. The three years that
Maecenas recommends in Dio's famous speech (52.23) and that appear as a
rule in the second century, are attested under Augustus only in very few
cases; they may have been more frequent, but longer commands certainly
were extremely rare. Two or three years were long enough to get familiar
with the specific problems of a province, too short for the development of
dangerously close ties with the army, especially if the command of larger
operations was reserved to members of the immediate entourage of the
Princeps.
Even our scanty material allows us to see that a sequence of several
such short governorships in different provinces formed the normal pattern
of the surprisingly short careers. In addition, those commands were usually
separated by years of leave, civil functions in Rome or senatorial governor-
ships. The same person could absolve two commanding functions in the same
region, but hardly ever in the same province. In the few cases of exceptions,
the persons involved belonged to the closest friends of the Princeps. Also
very rare are subsequent governorships of members of the same family in
the same province. And the same is true for contemporary commands of
close relatives in several military provinces. Again with obvious good reasons
that are aptly illustrated by the difficulties caused to Tiberius and especially
Caligula by Cn. Cornelius Lentulus Gaetulicus who for many years governed

1018
Germania Superior with four legions, while his father-in-law commanded
another four legions in I..ower Germany and a brother-in-law very probably
was in charge of Pannonia with three legions.
All these measures seem to me to indicate a certain tendency: During
the Late Republic active generals had used their armies to exert political
pressure in order to secure an exceptionally powerful position in political
life faster and more effectively than was possible through the normal career-
patterns. As a result, the ultimate goal of political ambition no longer was
to become princeps senatus, i.e. the most distinguished of the former consuls,
but to belong to a very small group of potentates whose power was based on
the long-term command of several provinces and legions rather than on their
influence in the Senate. Now, under Augustus, this plurality of principes
who still (at least formally) had competed for primacy among equals, had
finally been replaced by one Princeps whose position, moreover, no longer
was open to competition. Since, therefore~ the previous main target of sena-
torial ambition had become unattainable-and had to be unattainable for the
sake of lasting peace and stability-, this ambition itself had to be redefined
and given a new orientation. The new pattern of senatorial careers accordingly
provided, in addition to the traditional magistracies and governorships in
senatorial provinces, a series of new civil offices and the prestigious and
rewarding positions in the provincial administration and command of the armies
of the Princeps. This, however, implied an almost complete change in the
character of office: The dependent legatus, closely controlled and limited
in his freedom of action, replaced the independent, king-like proconsul. At
the same time the offices of the traditional cursus. including the consulate,
increasingly lost their political significance. Previously political competition
in Rome had always aimed at obtaining influence, power and rule in a very
basic and concrete sense. It might, therefore, not be exagger3:ted to conclude
that the internal pacification of Rome could, in the last resort, only be realized
by depoliticizing the entire senatorial aristocracy. In spite of the program
of "res publica restituta", the time of Augustus saw the (very cautious, but
obvious) beginning of this process. It should be added that, by assigning for
the first time in Roman history some legionary and provincial commands
permanently to eguites, and by introducing the basic components of what was
later fixed as the equestrian tres militiae, Augustus also laid the foundations
of the slow but inevitable process of demilitarization of the senatorial class
and its replacement by the ordo eguester-which was itself created and organ-
ized as such by Augustus.

VIII
It is time to summarize the results of our discussion: The internal
pacification of the empire was not only, perhaps not even mainly, due to
Augustus' military reforms proper. Their effect certainly should not be
underestimated. They satisfied the most urgent needs of the soldiers, they
secured a basis for their professional existence and their pension and there-
by considerably reduced their readiness to interfere in politics. This readiness
was further decreased by the monopolization and careful cultivation of the
army-clien.tela by the Princeps and his family. But the very nature of such
a professional army (which, moreover, for political reasons could not be

1019
fully integrated into Roman society but was forced to live a secluded life on
the edges of the civilized world) a priori limited the effect of these measures.
The chances of a revival of its political activity could therefore be minimized
but not eliminated, the latent inclination of the soldiers to politicize continued
to exist. This danger could only be checked effectively on the political level,
by controlling the leadership of the armies.. The Senate and personnel policy
of Augustus, therefore, was most important for his program of internal paci-
fication. Only a restricted group of tested and trustworthy men, bound to the
Princeps by close personal ties, was henceforth allowed to handle the delicate
instrument of real power-and their possibilities to handle it were restricted
and tightly supervised.
The solution which Augustus developed for the separation of the political
and military spheres, therefore, was broad and comprehensive. It was suc-
cessful because it combined all the three possible and necessary factors:
the social care for and material satisfaction of soldiers, officers and veterans;
the intensive, steady cultivation of the personal ties of clientela; and the dili-
gent provision for and supervision of trustworthy leadership. The skilful
combination of these three elements was so effective that the system was able
to survive, even if one or the other component occasionally failed or was
neglected by the successors. The reign of Tiberius, for instance, was marked
by some extremely long governors hips. The reasons may have been complex,
but the names of the persons involved show that he selected his men very
carefully. In addition, his long career as commanding general provided him
with such a bonus of loyalty within the armies that, after the mutinies of A. D.
14, there were only minor difficulties. Caligula was overthrown without the
participation of the provincial armies. He did not live long enough to destroy
the reputation he enjoyed with the soldiers as a son of Germanicus; and,
whatever his shortcomings in other areas, he obviously was well aware of
the fundamental significance of a close relationship to the armies. Claudius
lacked any such personal relationship when he became emperor. But he made
up for that by personally visiting and leading his troops at the end of the first
stage of the conquest of Britain.
The system only broke down when Nero neglected all three of its compo-
nents at once. He did not care about the material needs of the army and let
the pay and retirement rewards fall in arrears; he not only did not establish
any personal relationship to the soldiers but, by his extravagant behaviour,
antagonized especially parts of the important corps of centurions; he not
only made grave mistakes in the selection of governors, but by arbitrarily
putting to death some successful commanders, he decisively alienated the
corps of his generals. It is, therefore, not surprising that the breakdown
finally occurred but that it occurred so late (and this again can at least partly
be explained by some provisions introduced by Augustus). The attempt of
the German legions to proclaim Verginius as emperor, the revolt of the
Spanish troops under GaTha and the subsequent civil wars, accordingly, were
no accidents; they were logical and inevitable. Under the specific circum-
stances of Nero's later reign the permanent but latent tendency of the armies
to reassume their political role was bound to break into the open.

1020
 Contrary to the appearance, then, the complex of problems that so
largely had been responsible for the end of the Republic, had not really been
solved. Under the conditions of the Augustan Principate with its deliberate
and necessary continuation of m.ost of the fundamental Republican structures,
these problems could not be solved, they could only be brought under control
and suppressed. To insist on this negative statement, does not mean that
Augustus failed or is to be blamed for the failure of later generations. On
the contrary: he did all that could be done, and he did it thoroughly and with
deep insight. It rather means that overoptimistic and sometimes superficial
judgments of modern observers ought to be corrected. Illegitimate interference
of armies in politics could and can never be stopped or prevented by measures
on the military level only. Adequate changes and provisions on the political
level are indispensable, probably even more important; there is no lack of
later, even contemporary evidence. And so the secret of Augustus' successful
suppression of the politicization of the armies is to be found rather more on
the political than on the military side. The events after the assassination of
Domitian and Commodus were going to prove that again.

NOTE: Earlier drafts of this paper have been read at Brown University in
Providence and the Free University in Berlin. I am grateful to my colleagues
at both universities and to others who have read the text (esp. to G. Bowersock,
Ch. Fornara, M. Krebs, R. MacMullen, Chr. Meier and J. van Sickle) for
valuable comments and suggestions. The responsibility for the published text
rests of course entirely with me. Due to the complexity of the problems
involved and the limited space available in these "Actan the format of a lecture
has basically remained unchanged. I hope soon to be able to present a thorough
documentation and more detailed discussion in a forthcoming publication.
For most readers my main sources of information (ancient and modern) will
be obvious. However, a few remarks on the "Forschungslage" and some
selected references to important or most recent publications seem to be
necessary:
I. Modern scholarship on the period and problems covered by this
paper generally shows two remarkable gaps. One of them concerns the
temporal dimension: The lack of a wide-spaced diachronic approach which
''bridges" the "threshold" between the Late Republic and the Principate, has
been described above (part I}. H. H. Scullard's From the Gracchi to Nero
(3rd ed. London 1970) or R. MacMullen's Roman Social Relations, 50 B. C.
to A.D. 284 (NewHaven-London 1974) belong to the few exceptions. The
other desideratum is an equally comprehensive synchronic inclusion of the
whole range of Roman society. For a long time research on the crisis of
the Late Republic and the emergence of the Principate has been preoccupied
with one part of society only. Both the students of prosopography, trying to
clarify the composition and changes of the "parties" struggling with each
other (best represented by R. Syme 's already classic The Roman Revolution)
and the students of the working principles of Roman political life and of the
structural characteristics of Roman "constitution" and institutions (from
M. Gelzer's Die Nobilitltt der rtlmischen Republik [1912=Kleine Schriften I
(1961) 18ff.], to Chr. Meier's Res publica amissa [Wiesbaden 1966]) have

1021
 tended to concentrate almost exclusively on the upper classes and to neglect
the remarkable influence of the will and needs of the lower classes, the pro-
letarian masses, the soldiers. In reaction to this one-sidedness, some
scholars (most prominently P. A. Brunt in numerous publications, e.g. in
"The Army and the Land in the Roman Revolution", J. Roman Stud. 52 [1962]
69ff., and in Italian Manpower [Oxford 1971]; or, e.g. H. Aigner, Die
Soldaten als Machtfaktor in der ausgehenden rtlmischen Republik [Innsbruck
1974], esp. 148 ff. H. Schneider, Wirtschaft und Politik. Untersuchungen
zur Geschichte der sp!!ten rt.Jmischen Republik [Erlangen 1974], esp. p. 250ff.
Id., Die Entstehung der rljmischen M iliUtrdiktatur. Krise und Niedergang
einer antiken Republik [KtHn 1977] ) more recently have focused their attention
on the social and economic situation and the important historical function of
the Roman and Italian lower classes in the period of the "Roman Revolution".
As necessary and important, however, as it is to stress those neglected
aspects, the result often enough is again unbalanced and unsatisfactory. This
is clearly visible even in some of Brunt's conclusions (cf. e.g. J. Roman
Stud. 52 [1962] p. 84, or Social Conflicts in the Roman Republic [London 1972],
p. 149), and it is amazingly obvious in the works of Schneider (passim) and
Aigner (e.g. 1. c. p. 176: "· .es ist ••• nt.Jtig zu betonen, dass Augustus •••
die 'RtJmische Revolution' auf der Ebene gestoppt hat, auf der sie sich haupt-
sttchlich bewegt hat: auf der Ebene der Resozialisierung der Waffen ftlhren-
den Proletarier. "). By concentrating on either the upper or the lower classes
one is bound to miss factors which are indispensable for a proper understanding.
A comprehensive explanation of the fundamental changes that were witnessed
and brought about by Augustus and his contemporaries therefore has to take
into account the role of all social classes. In the context of this paper, it
needs especially to be emphasized that the power struggles within the nobility,
the permanent collision of individual pretensions to a form of power which was
increasingly taken as absolute and separated from its original aristocratic
basis, were as necessary a precondition for the civil wars and the breakdown
of the :Republic as the social misery and discontent of the proletarian soldiers
(cf. above, end of part VI). I am, of course, not the first to insist on the
importance of Augustus' tight control of military leadership. And most of
my "facts" are well-known. Rather, it is my aim to reach, on the basis of
systematical considerations, a more balanced and comprehensive understanding
and to attribute the proper weight to each of the several social and political
factors that made it possible to overcome the threat of arma civilia and found
the pax Augusta.
II. The following bibliographical notes are, of course, by no means
supposed to be complete. Recent systematical bibliographies have been pub-
lished by K. Christ, Rt.Jmische Geschichte. Eine Bibliographie (Darmstadt
1976), and M. Clauss, "Ausgewtthlte Bibliographie zur lateinischen Epigraphik
der rtJmischen Kaiserzeit (1. -3. Jh.) ", in: Aufstieg und N iedergang der
rt.Jmischen Welt, ed. H. Temporini (ANRW), vol. II 1 (Berlin-New York
1974), pp. 796ff.: cf. esp. pp. 819ff.
1 • For the political role of the soldiers in the Late Republic cf. mainly
E. Gabba, Republican Rome, the Army and the Allies (Berkeley- Los Angeles
1976) pp. iff., and 20ff. P.A. Brunt, "The Army and the Land ••• ", I.e.
R. E. Smith, Service in the Post-Marian Army, Manchester 1958. F. B.

1022
Marsh,. The Founding of the Roman Empire (2nd ed. London 192 7) is still
very valuable. Cl. Nicolet, Le metier de citoyen dans la Rome republicaine
(Paris 1976) pp., 122ff .. 150ff .. 173ff .. , and Rome et la conquete du monde
mediterraneen, vol. I: Les structures de 1'Italie romaine (Paris 1977) pp.
300ff .. , covers both the teclmical and political aspects.. The shortcomings of
J. Har·mand, L'armee et le soldat a Rome de 107 a 50 av.notre ere (Paris
1967), have sharply been pointed out e.gs by M. Rambaud, "Legion et armee
romaine", REL45 (1967) pp .. 112ff .. The dissertations of H. Aigner (I.e.)
and E. Erdmann, Die Rolle des Heeres in der Zeit von Marius bis Caesar.
Milit::trischeund politische Probleme einer Berufsarmee (Neustadt/A is eh 1972),
are useful but leave many questions open (see my review in Gnomon 49 (1977)
486ff.). For the years after44 cf. W. Schmitthenner, "Politik und Armee
in der sp!tten RI'Jmischen Republikn, Histor. Zeitschr. 190 (1960) iff. H.
Botermann, Die Soldaten und die ri'Jmische Politik in der Zeit von Caesars.
Tod his zur Begrtindung des Zweiten Triumvirats (Mtlnchen 1968).
2. Augustus t military reforms: V. G ardthausen, Augustus und seine
Zeit (2 vols. Leipzig 1891-1904) pp. I 626ff. with II 340ff., still gives the
most complete references. A. Momigliano, "I problemi delle istituzioni
militari die Allt:,ousto 11 , in: Augusto, Studi in occasione del bimillenario
augusteo (Rome 1938) pp .. 195ff., gives a useful 11Forschungsbericht 11 • Short
chapters e.g. in: A. H. M Jones, Ay.gustus (London 1970), pp. 110ff. M.
Hammond, The Augustan Principate in Theory and Practice during the Julio-
Claudian Period (2nd ed. Cambridge, Mass. 1968) pp. 148fL F. Vittinghoff,
Kaiser Augustus (GI'Jttingen 1959) pp. 75ff. All of these, of course, have to
be consulted for other aspects too. For Augustus 1 supreme command cf.
esp. H. Nesselhauf, "Von der feldherrlichen Gewalt des ri'Jmischen Kaisers",
Klio 30 (1937) 306ff. For the praenomen imperatoris cf. R. Syme, "Imperator
Caesar. A Study in Nomenclature 1 ' , Historia 7 (1958) 172 ff. J. Deininger,
''Von der Republik zur Monarchie .. Die Urspriinge der Herrschertitulatur
des; Prinzipats" ANRW I 1 (1972) 982ff. For the monopolization of the army-
clientela cf. A. von Premerstein, Vom Werden und Wesen des Prinzipats
(Mtlnchen 1937) 13ff., esp. 99ff. (the sections on the "Kaisereid" now have
to be modified: cf. P. Herrmann, Der ri'Jmische Kaisereid (GI'Jttingen 196 8).
V. Fadinger, Die Begrtlndung des Prinzipats (Berlin 1969) 272ff.). R. Syme,
The Roman Revolution (Oxford 193 9) 2 76ff. and passim. J. Gage, Les classes
sociales dans !'empire romain (2nd ed. Paris 1971), 59 ff .. For Augustus'
dealing with the centurions cf.. B. Dobson, ''The Significance of the Centurion
and 'Primipilaris' in the Homan Army and Administration", ANRW II 1, pp.
392ff., with rich references. For the colonization program cf. F. Vittinghoff,
Rl:Jmische Kolonisation und BUrgerrechtspolitik unter Caesar und Augustus
(,Mainz 1951). F. T. Hinrichs, "Das leg ale Landversprechen im Bellum civile,
Historia 18 (1969) 521ff. P.A. Brunt, Italian Manpower (Oxford 1971) 332ff.
A more recent but hardly satisfactory discussion in H. Chr. Schneider, Das
Problem der Veteranenversorgung in der spttten ri'Jmischen Republik (Bonn
1977) 2 06ff.
3. For the armies of the early Principate cf. G. Webster, The Roman
Imperial Army of the First and Second Centuries A.D. (London 1969). G. R.
Watson, The Roman Soldier (London 1969), both mainly concerned with the
technical aspects. M. Grant, The Army of the Caesars (N. Y. 1974), what-

1023
ever its other shortcomings, at least underlines the political problems. Cf.
further H. G. Pflaum, "Forces et faiblesses de 1'armee romaine du haut-
empire", in: Problemes de la guerre a Rome, ed. by P. Brisson (Paris
1968) 85 ff. J. Harmand, "Les origines de l'armee imperiale. Un temoignage
sur la realite du pseudo-principat et sur l 1 evolution militaire de l'occident",
ANRW II 1, 263ff. For the social and economic problems of the professional
army and the veterans cf. Watson, 1. c. 89ff. Brunt, Italian Manpower 342ff.
Id., "Pay and Superrumeration in the Roman Army", Proc. Brit. Sch. Rome
18, 1950, 50 ff. Esp. important: G. Forni, I1 reclutamento delle legioni
da Augusto a Diocleziano (Milano-Roma 1953) esp. 28ff. 119ff., and, more
recently: "Estrazione etnica e sociale dei soldati delle legioni nei primi
tre secoli dell'impero", ANRW II 1, 339ff., esp. 390ff. Forni also provides
the basic statistics for the change of recruiting areas in the 1st and 2nd century
A. D. The development from military professionalism to military class is
discussed by J. Gage, I.e. 133ff. 249ff. However, the political function of
the army during the Principate has not been systematically discussed in recent
years. For a discussion of some important aspects cf. J. Bleicken, Verfas-
sungs- und Sozialgeschichte des rl:Jmischen Kaiserreiches, vol. I (Paderborn
1978) 213ff. For a discussion of the military revolts down to 68 one has to
refer to the biographies of the individual emperors. For 68/69 cf. G. E.F.
Chilver, "The Army and Politics A.D. 68-70 11 , J. Roman Stud. 47 (1957) 29ff.
P. A. Brunt, "The Revolt of V index and the Fall of Nero", Latomus 18 (1959)
531ff. H. Grassl, Untersuchungen zum Vierkaiserjahr 68/69 n. Chr. Ein
Beitrag zur Ideologie und Sozialstruktur des frUhen Prinzipats (Diss. Graz
1972).
4. For Augustus' relationship to the upper classes (and many related
problems) cf. mainly R. Syme's Roman Revolution (a work to which this
paper owes much more than it shows). The thorough analysis of P. Sattler,
Augustus und der Senat, Untersuchungen zur rl1mischen Innenpolitik zwischen
30 und; 17 v .Chr. (Gtlttingen 1960), and nJulia und Tiberius. Beitr!tge zur
rtlmischen Innenpolitik zwischen den Jahren 12 vor und 2 nach Chr., in: Id.,
Studien aus dem Gebiet der Alten Geschichte (Wiesbaden 1962), iff., now in:
Augustus, ed. by W. Schmitthenner (Darmstadt 1969) 486ff., covers at least
a large part of Augustus' reign. For the control of military leadership and the
the careers of legati Augusti cf. E. Birley, "Senators in the Emperor's
Service", Pro c. Brit. Acad. 39 (1953) 197ff. "Be:IDrderungen und Versetzun-
gen im rtlmischen Heere", Carnuntum Jahrb. 1957/58, 3ff. Birley's conclu-
sions have been slightly modified but basically confirmed in several publications
of G. Alftlldy (e.g. in "Die Generalitttt des rl:Jmischen Heeres", Bonner Jahrbb.
169 (1969) 233ff.), W. Eck (e.g. in"Befbrderungskriterien innerhalb der
senatorischen Laufbahn, dargestellt an der Zeit von 69bis 138 n.Chr.",
ANRW II 1, 158ff.) and others. Criticism and doubts are expressed e.g.
by F. Millar, "The Emperor, the Senate and the Provinces", J. Roman Stud.
56 (1966) 156ff. B. Cam];i>ell, ''Who were the 'viri militares'?", J. Roman
Stud. 65, (1975) 11ff. The conclusions of all these scholars, however, are
mostly based on evidence of the 2nd half of the first and the second centuries.
For the first half of the first century and the time of Augustus the sources
are much poorer. How much of the later system was already fully developed
by Augustus himself, remains therefore uncertain to some extent. The
available source material has carefully been collected and arranged by R.

1024
Szramkiewicz, Les gouverneurs de province a l'epoque Augusteenne, 2 vols.
Paris 1976. Despite its lack of historical perspective and penetration, this
work is extremely useful for its statistics. Here, at last, is a synchronic
study on the governors of all provinces of the empire in a given period. For
many questions this is much more helpful than the diachronic studies concen-
trating on the administration of one single province, that are so popular
among prosopographists (cf. also W. Eck, I.e. 158-161). A similar systema-
tical study for the subsequent period from Tiberius to Nero is (despite R.
Syme's Tacitus) most desired. H. H. Pistor, Prinzeps und Patriziat in der
Zeit von Augustus his Commodus (Diss. Freiburg 1965 ), and A. Bergener,
Die ftthrende Senatorenschicht im frt.!hen Prinzipat (14-68) (Diss. Bonn 1965),
at least Hll some parts of the gap. Again, some discussions of the problems
involved can be found in more recent biographies of individual emperors.
Cf. , e.g., for Tiberius: R. Seager, Tiberius (Berkeley-Los Angeles 1972).
G. AlftHdy, "La politique provinciale de Tibere", Latomus 24 (1965), 824ff.
W. Orth, Die Provinzialoolitik des Tiberius (Diss. MUnchen 1970).

1025
 69. L I XA E
H.v.Petrikovits

Die antiken Quellen sprechen von lixae und Händlern, die der Truppe
auf dem Marsch folgten oder i~ den canabae legionis wohnten. Seide
Gruppen, lixae und Händler, sind in zeitgenössischen Quellen vonein-
ander unterschieden 1 ). Offensichtlich lebte~ sie von den SectUrf-
nissen der Truppe. Man fragt sich zunächst von welchen, denn fUr die
SachgUter, die der Soldat fUr sein tägliches Leben und fUr den Kampf
brauchte, sorgten Verwaltungs- und Handwerkssoldaten sowie eine
Transportkolonne. In einer Legion unterstanden sie dem praefectus
castrorum. Auch zivile Handwerker versorgten die Truppe. Sie arbei-
teten entweder ausschließlich fUr eine bestimmte Einheit oder sie
lieferten ihre Erzeugnisse fallweise. Wozu also noch lixae? Gegen
sie und gegen die Händler gingen römische Kommandeure mehrmals vor.
Sie ärgerten sich Uber die Aufdringlichkeit und den oft demoralisie-
renden Einfluß der Händler und lixae. Offensichtlich boten sie der
Truppe Leistungen an, die den Soldaten angenehm, dem Kommandeur aber
UberflUssig erschienen.

Von den antiken Schriftstellern erfahren wir wenig Uber die Tätigkeit
der lixae. Appian schreibt, daß im Jahr 134"vor Chr. P.Cornelius
Scipio Africanus die Disziplin der römischen Truppe vor Numantia
wieder herstellte; indem er Zivilisten vertrieb, die sich unerlaubt
im Militärlager aufhielten. Es ist da von Händlern, leichten Mädchen,
Wahrsagern und Opferpriestern die Rede: "E f-.L TC 0 r 0 L. , Sc 'T -y -l 0 r: l ,
ur~\)'TEL.C, Nonius Marcellus, ein Grammatiker des frUhen
-8-U'TC'(L..

4.Jahrh.n.Chr.,wollte wissen, daß die lixae den Soldaten Wasser ins.

Lager und sogar in ihre Zelte brachten. Auf diese Nachricht darf man
aber nicht viel geben, weil im 4.Jahrh. die alte Bedeutung von lixa
schon verschwommen war. Nonius Marcellus behauptete, daß "die Alten ..
mit lixa das Wasser bezeichneten. Das ist unwahrscheinlich.

Unsere dritte schriftliche Quelle ist Iustinus, der den Feldzug des
Antiochos III. im Jahr 212 vor Chr. gegen die Parther schildert. Er
erwähnt lixae, die dem Heer des Königs folgten, unter ihnen coci,
pistores---,~enici 2 ). Man kann verstehen, daß die Soldaten einmal

1027
Ptwas anderes essen wollten als ihre Rationen und daß die scaenici
ihnen die Zeit vertreiben sollten. Ob man aber die Nachricht des
Justin ohne weitere Prüfung von einem hellenistischen Heer auf das
römische Ubertragen darf, ist fraglich, auch wenn eine innere Wahr-
scheinlichkeit dafür spricht.

Im folgenden wollen wir einige weitere Zeugnisse prüfen, ob sie die

i~achricht Justins auch für das römische Heer gelten lassen. Es han-

delt sich um delege für Schauspieler(-innen), Tänzer(-innen), Musi-

ker und Gladiatoren.

Jede große Garnison - und manche kleinere auch - hatte ein Amphithe-
ater. t:s dientE~ vor allem als 'ludus', als Übungsplatz. Deshalb ivar
seine Arena im Vergleich zum Zuschauerraum größer als bei einem zi-
vilen Amphitheater 3 ). Ein ludus ist zweimal auf der Trajanssäule
dargestellt. J.Wahl hat gezeigt, daß in militärischen Amphitheatern
sowohl Gladiatoren als auch Tierkämpfer auftraten. Sie standen in
einem engen Verhältnis zur Truppe. Formulierungen auf Inschriften
weisen darauf hin, daß einige Gladiatoren Dauerangestellte der Truppe
waren wie die Zivilarb~iter, die wir am Anfang erwähnten 4 ). Man wird
erwägen milssen, ob sich nicht lanistae, Gladiatorenunternehmer, mit
ihren Gruppen zeitweilig oder dauernd bei der Truppe aufhielten. Eine
hölzerne sica, ein Obungskrummschwert, mit dem ein Thraex geübt haben
dürfte, wurde im LagerOberaden gefunden. Sie bestätigt Wahls und un-
sere Vermutungen. S.v.Schnurbein wird sie demnächst in der 'Germania'
vorlegen. Auch Tierjagden werden in den militärischen Amphithe~tern
vorgeführt worden sein. Durch Inschriften sind militärische Tier-
fänger für Löwen und dären bezeugt, denen Fährtensucher (vestigia-
tores) behilflich waren 5 ).

In den Lager-Amphitheatern traten auch Schauspieler(innen), Tänzer(-

innen) und Musiker auf, wenn kein Bühnentheater in der Nähe war.
Durch eine Inschrift in Lugudunum (Lyon) lernen wir einen Schauspie-
ler kennen. Er setzte in der schola pollionum, dem Freizeitraum der
Küstungs- und Schwertfeger, einen Weihealtar für die kaiserlichen
Gottheiten. Sein Name T.Fl(avius) Super ist durch den Spottnamen
Cepula erweitert (das Zwiebelchen). Er war scaenicus, honesta
missione missus ex leg. XXX U.v.p.f. (207 n.Chr.). Dieser Schau-

1028
spieler gehörte also bis zu seiner Entlassung der 30. Legion an, die
ihren Standort in Vetera (Xanten) hatte und seitSeptimius Severus
ständig eine Untereinheit nach Lyon abkommandierte. Es ist möglich,
daß 11 das Zwiebelchen 11 seine schauspielerischen Fähigkeiten schon
während seiner Dienstzeit zum besten gab. Aus zwei Inschriften der
militärischen Vigiles, der Feuerwehrtruppe in Rom, erfahren wir näm-
lich, daß sie und die Flotte in Misenum für besondere Feste eine
ganze Theatertruppe aufstellten. In ihr gab es archimimi (Schauspie-
ler), stupidi (dumme Auguste) und scurrae (Clowns). S·ie alle gehört81
der Truppe als Soldaten an; sie können sogar immunes oder principa-
~s gewesen sein 6 ). Wenn Iustinus scaenici zu den lixae zählt,
denkt er an Berufsschauspieler, während die Truppenangehörigen
Laienschauspieler gewesen sein werden. Vielleicht gab es scaenici,
die um des sicheren Verdiensteswillen ständig vor der Truppe auf-
traten. Inschriftlich sind beim Dux ripae Euphratis in Dura-Europos
-rpcxy~oo C (tragische Solosänger) und unoxp vrn,:C , ihre Gehilfen,
bezeugt 7 ).

Eine Tänzerin oder Schauspielerin für die Truppe scheint uns auch
jene Polla Matidia gewesen zu sein, der ein Veteran der Legio II Aug.
in Asciburgium (Moers-Asberg) einen Grabstein setzte. Die Legio II
Aug. stand von 9-17 n.Chr. in Mogontiacum (Mainz), von 17 bis 43 in
Argenterate (Straßburg). Es ist nicht zu entscheiden, von welchem der
beiden genannten Standorte der oberen Rheinarmee der Freund der Polla
Matidia an den Niederrhein verschlagen wurde. Aus dem Ort des Grab-
steins innerhalb einer ty~ologischen Stelenreihe, die viel Wahr-
scheinlichkeit für sich hat, möchte man das Bildwerk in das 2.Jahr-
zehnt des 1. Jahrh.n.Chr. setzen 8 ). Der Name der Bestatteten paßt
gut zu der frühen Datierung. Polla ist die monophthong'Erte Form von
Paulla, 1 die Kleine'. Der Gebrauch eines weiblichen Vornamens war
während der Republik recht häufig, verschwand aber im 1 .Jahrh. der
Kaiserzeit. Paulla war als weiblicher Vorname nicht selten. Aus Köln
kennen wir zwei Paullae und die Mutter Vergils hieß Magia Polla. Das
Gentiliz Matidius(-a) scheint etruskischer Herkunft zu sein und wird
vor allem in Mittelitalien gefunden, seltener in Norditalien. Aus
einer Familie, die in Vicetia (Vicenca) in der venetischen Region
zu Hause war, stammte die Nichte Trajans Matidia, die im Jahr 112

1029
Taf. 69. 1 Eine Grabinschrift aus Asciburgium (CIL 13, 12075).
mit dem Titel Augusta ausgezeichnet und nach ihrem Tod unter die Göt-
ter versetzt wurde 9). Nach dem Vor- und dem Gentilnamen las man seit
A.Oxe in Z.1 der Grabinschrift Sp(uri) f(ilia). Das ist falsch. Auf
dem Stein steht SIBE , das ist sive, darauf folgt Olumphia (Olympias).
Dies ist also kein Cognomen, sondern ein Obername, wie ihn der oben
angeführte T.Flavius Super 'Cepula' führte. Derartige griechische
Künstlernamen waren für Schauspieler und verwandte darstellende Be-
rufe kennzeichnend. Es wäre reizvoll, den Zusammenhang dieser Ober-
namen mit spätrömischen Supernomina zu überdenken 10 ).

Frauen traten im frühkaiserzeitlichen Mimus noch nicht auf. Aber in

den Pausen einer Pantomime sah man eine emboliaria, eine Schauspie-
lerin im Zwischenspiel. Weibliche pantomimae gab es zwar schon vor
Seneca, aber ob sie schon im 2.Jahrzehnt des l.Jahrh., zur Zeit der
Polla Matidia, auftraten, ist fraglich. Im Mimus sah man eine salta-
trix, eine Tänzerin. Vielleicht war Polla Matidia Tänzerin. Eine
Parallele aus Karthago wäre dann eine Thyas saltatrix, deren bürger-
licher Name Metel(e)ia Rufina war. Ob sich Polla Matidia längere Zeit
in dem Auxiliarvicus Asciburgium aufhielt, sagt die Grabinschrift
nicht aus. Immerhin sind auch stadtrömische Schauspieler und Tänzer
in die Provinzen auf Tournee gegangen. Eines von mehreren Beispielen
war denanonyme Tänzer (Pantomimus?) der Severerzeit, der in Rom, in
vers~hiedenen Regionen Italiens und in Provinzen auftrat, sogar in

der Germania inferior. Ein solches Format werden wir Polla Matidia
freilich nicht zuschreiben dürfen 11 ).

Die Funde tönerner Theatermasken im Auxiliarvicus Regensburg-Kumpf-

mühl, im Theater des Kastells Heddernheim, im Kastell Wiesbaden, in
der Militärtöpferei Holdeurn bei Nijmegen und in Vechten (Fectio)
mögen es als wahrscheinlich erscheinen lassen, daß auf den Truppen-
bühnen oder in militärischen Amphitheatern nicht nur Mimi, sondern
auch Pantomimen, Atellanen und Komödien gespielt wurden, wie sie auf
dem Programm der bürgerlichen Theater geläufig waren 12 ).

Die spätrömischen Tragöden und die frührömische Tänzerin oder Schau-

spielerin Olympias waren nicht Angehörige der Truppe, sondern lixae
im prinzipatszeitlichen Wortsinn. In diese Gruppe scheint uns auch
ein Wasserorgelspieler zu gehören. Sein Grabgedicht für seine Frau,

1031
die auch hydraularia war und eine gute Stimme hatte, ist in Aquincum
erhalten. Ihr Mann bezeichnete sich als hydraularius salarius leg.II
ad(iutricis). Unter salarium verstand man nicht den Sold des Solda-
ten, sondern Vergütungen für hohe Offiziere und Beamte und ihre Be-
gleitung, für Ärzte und selbst für arme Senatoren. Mit diesem hydrau-
larius T.Aelius Iustus scheinen wir einen Zivilisten kennenzulernen,
der für die Truppe tätig war und dafUr ein Salär bekam. Die Wasser-
orgel wurde nicht nur im Zirkus und im Amphitheater, sondern auch
oei Schauspielaufführungen gespielt 13 ).

Zu den lixae gehörten also Zivilisten, die für die Truppe verschie-
dene Dienstleistungen verrichteten. Ausdrücklich sind Wahrsager,
Opferpriester, Schauspieler(innen), Tänzer(innen, Musiker(innen) und
Dirnen bezeugt. Nach Betrachtung der angeführten Zeugnisse scheint
es uns wahrscheinlich zu sein, daß es bei prinzipatszeitlichen und
vielleicht spätantiken römischen Heeren lixae im Sinne des Iustinus
gegeben hat. Das Wort lixa selbst hat man aber in der Spätantike
nicht mehr richtig ve~standen. Zeitgenössische Schriftsteller und
Philologen verwechse1n calones mit lixae oder haben andere abwegige
Vorstellungen von der Bedeutung dieses Wortes 14 )

ANMERKUNGEN

1. ßell.Afric. 75,3; Val.Max. 2,7,1; Tac.ann. 2,62,3; hist. 4,15,3.

2,. Appian.Iber. 85; Non. p.48,17 und 62,2; Iustin. 38,10,2. Die
sprachliche Herkunft des Wortes scheint unklar zu sein (etrus-
kisch?): Walde und Hofman~ 1938, Bd.l, 317 und 872.
3. v.Petrikovit~ 1980.

4. Wahl, 1977, 108-132, besonders 130 f.

5. Wahl, 1977, 128 f. Rostovtzeff, 1952, 44-50, Nr. 954 f.
6. Weihestein des T.Fl.Super: Wuilleumier, 1963, 234. Listen der
Vigiles: C.I.L. 6,1063 f.
7. Rostovtzeff, 1952, 30-40, Nr. 945-950.
8. C.I.L. 13,12075; Gabelmann, 1972, 100 f. und 133.
9. Zu Polla: Oxe, 1907, 21 f. Paulla Petronia: C.I.L. 13,8337.
Polla Poblicia L.f.: Galsterer, 1975, 55, Nr. 216. Magia Polla:

1032
 Büchner, 1955, 1038. Das Gentiliz Matidius: Schulze, 1904, 274 f.;
Pflaum, 1963, 525.
10. Römische Schauspielernamen: Bonaria, 1956; Bonaria, 1959.
11. Wüst, 1932, 1753; Wüst, 1949, 850-852; Bonaria, 1965, 184;
Blume, 1978, 128-130; zum anonymen Pantomimen: Sordi, 1953,
104-121.
12. Castra Regina, 1979, 90 und 141. Woelcke, 1929, 76 f.; Wahl,
1977, 124; van Hoorn, 1960, 75 und 79; O.R.L. B, Nr.31 (Kastell
Wiesbaden), 121; Bieber, 1930, 2092-2105.
13. C.I.L. 3,10507. Salarium: Rosenberg 1920. Römische Wasserorgel:
Ti t t e1 , 191 4 , 60- 77; Kaba , 1976 ; Wi 11 e , 1967 , 203- 21 0.
14. Veg.mil. 1,10 (lixas, guos galiarios vocant) verwechselt die
calones galearii mit den lixae. v.Petrikovfts, 1975, 58. Unver-
standen in den Glossarien, in den Schol. Hor. sat. 1 ,2,44, bei
Nonius Marcellus (Anm.2) und Fulgentius, serm.ant. 55.

BIBLIOGRAPHIE

Bieber, M., 1930; 'Maske', R.E. 14,2070 f.

Blume, H,-D., 1978: Einführung in das antike Theaterwesen.
Bonaria, M., 1956: Mimorum Romailür-um Fragmenta.
t3onaria, M., 1959: 'Dinastie di pantomimi latini·', Maia 11.
Bonaria, M., 1965: 'Emboliarius', R.E. Suppl. 10, 184.
Büchner, K., 1955: 'P.Vergilius Maro', R.E. 8A, 1021 f.
Castra Regina, 1979: Jubiläumsausstellung (Katalog)
C.I.L.: Corpus inscriptionum Latinarum.
Gabelmann, H., 1972: 'Die Typen der römischen Grabstelen am Rhein',
~ 172, 65 f.
Galsterer, B.und H., 1975: Die römischen Steininschriften aus Köln.
van Hoorn-Groneman, J.J.C., 1960: 'Römische Maskenfragmente',
Bull.Ant.Besch. 35, 75-79.
Kaba, M., 1976: Die römische Orgel von Aguincum (Musicologia
Hungarica N.F.6).
O.R.L.: Der obergermanisch-rätische Limes des Römerreiches.

1033
Oxe, A., 1907: 'Römischer Grabstein einer Frau, gefunden in Asberg
(Asci burgi um) •, ~ 116, 19f.
v.Petrikovits, H., 1980: 'Canabae legionis', Neue Ergebnisse archäo-
logischer Stadtforschung, hrg. Deutsches Archäologisches Institut
(im Druck).
v.Petrikovits, H., 1975: Die Innenbauten römischer Legionsiager
während der Prinzipatszeit (Abhandlungen der Rheinisch-Westfäli-
schen Akademie der Wissenschaften 56).
Pflaum, H.-G., 1963: 'Rez. Kajanto, I., Onomastic Studies etc.'.
Rev.Et.Lat. 523 f.
R.E.: Real-Encyclopädie der class.Altertumswissenschaft, hrg. Pau1y,
Wissowa G., Kroll W. et al.
Rosenberg, A., 1920: 'Salarium•, R.E. lA, 1846f.
Rostovtzeff, M.I., 1952: The Excavations at Dura-Europos, Prelim.
Report 9, hrg. Rostovtzeff, M.I. et al.
Schulze, W., 1904: Zur Geschichte lateinischer Eigennamen (Neudruck
1966).
Sordi, M., 1953: 'L'e~igrafe di un pantomimo ecc.•, Epigraphica 15,
1953 ' l 04-121.
Tittel, 1914: 'Hydraulis', R.E. 9, 60f.
Wahl, J., 1977: 'Gladiatorenhelm-Beschläge vom Limes', Germania
55': 108f.
Walde, A. und Hofmann, J.B., 1938: Lateinisches etymologisches
Wörterbuch, 3.Aufl.
Wille, G., 1967: 'Musica Romana'.
Woelcke, K., 1929: 'Fundchronik. Histor.Museum Frankfurt a.M.',
Germania 13, 75f.
Wüst, E., 1932: 'Mimos•, R.E. 15, 1727f.
Wüst, E., 1949: 'Pantomimus', R.E. 18, 833f.
Wuilleumier, P., 1963: Inscriptions latines des trois Gaules.
( Supp 1 . Ga 11 i a).

1034
 Summary

The epitome of Justinus, an extract from Pompeius Trogus'

historiae Philippicae, makes mention of lixae which followed the
army of King Antiochos III in 212 B. C., including in their ranks coci,
pistores and scaenici. This observationwas also applied to the
Roman army. Indeed epigraphic and archaeological evidence for
actors, dancers and musicians (both male and female) supports this.
They offered the troops entertainment which the soldiers welcomed
but which the commanders on occasion found harmful for discipline.
An epitaph from Asciburgium is interesting in this connection (CIL
13, 12075 from Moers-Asberg on the lower Rhine in Germany). A
veteran composed it for Polla Matidia sibe (sive) Olumphia (Olympias,
a stage-name). She was presumably a dancer or an actress who
followed the troop unit together with other lixae. Another example
of a lixa was a scaenicus from Lugudunum who had the abusive name
'the little onion'. He was honesta missione missus ex leg. XXX
U. v. p.f. (A.D. 207). Perhaps there were scaenici who for the sake
of a secure income continually appeared before the troop unit. The
epigraphic and archaeological evidence of gladiators, animal catchers
and trackers argues for the fact that gladiatorial battles and animal
fights took place in the troop amphitheatres, which were primarily
training areas (ludi).

1035
 70. URBANIZATION AND THE NORTH-WEST FRONTIER
OF THE ROMAN EMPIRE

W. Groenman-van Waateringe

My proposition is that for a successful Roman occupation the pre-existence

of a native infrastructure was an essential condition on two counts, firstly,
for the food supply of the Roman army and secondly for the imposition of the
Roman administrative structure. The necessary infrastructure in this case
is an urbanized level of society, such as described by Cunliffe (1976a and b)
for Southern England. Support for this proposition will be drawn from a
comparison of the native social structure in the regions lying on either side
of the Limes in NW Europe. Cooter (1976, 94) in his thesis entitled 'Pre-
industrial frontiers and interaction spheres: aspects of the human ecology
of Roman frontier regions in Northwest Europe' points out the importance of
the "correlation of the frontier zones with pre-existing tribal hierarchies",
but immediately warns that "this demands a more extensive knowledge of the
Iron Age and Roman period" than he himself "or perhaps any other, possesses".
I do not wish to claim this knowledge for myself, but, utilizing some of the
more recent studies-amongst others, Cunliffe (1976a and b) for Southern
England and Waterbolk (1975, 1977) for the Northern Netherlands, I will
make an attempt.
As a result of his numerous publications concerning China, and his own
observations of the differences between what he calls the dynamic southern
frontier of the Chinese empire and the static northern frontier, Owen
Lattimore (1962) felt able to conclude that the degree of economic integration
between conqueror /occupier and the native population is of paramount impor-
tance. If the conqueror is unable to make use of a pre-existent organization
in the occupied region, permanent acquisition is difficult to achieve, from
a military viewpoint, for logistic reasons, from a civil viewpoint for economic
reasons. Thus, China was gradually able to absorb fresh territories on her
southern frontier, since the area to the south of the Yangtze was already
being affecte9. by higher cultures from the Indo-Burmese direction and was,
as a result, economically and administratively uniform with the rest of the
Chinese empire. On the northern frontier, however, the Chinese came upon
communities organized at a much lower level-'barbarians'-with which
economic integration was far more difficult to achieve, if it is accepted that
economic integration is a function of the possibility of transporting bulk goods
at a profit. Or, as Lattimore succinctly expresses it (1962, 479) "In a pre-
industrial civilization, when food has to be transported either by grain-con-
suming human porters or by grain-consuming draught animals the margin of
profit is eaten up within approximately 1 00 miles. 11

1037
 This brings us directly to the measures which were necessary to main-
tain the Roman army in the far corners of the Empire, for as Cooter (1976,
91) comments "the logistics of provisioning standing armies concentrated in
any one part of the empire was nightmarish.''. Even though a marching army
could have provisioned itself by foraging from the local population, we should
not think of this as an easy solution since, to name but a single example, the
three legions put into the field by Varus consumed more than 10 tonnes of
grain per day, quite apart from the fodder for the pack animals and the riding
horses of the cavalry. For a permanent military presence, lasting over two
centuries, as in our regions, pillage was no solution, and was, moreover,
inconsistent with the idea of the Pax Romana. It must be stressed that the
establishment of the string of Roman garrisons along the northern frontier
not only resulted in a sudden aggregate of soldiers, but also exerted a con-
siderable draw on the native '(X>pulation; witness the growth of the vici and
canabae. Thus we see a sudden and drastic increase in the number of people
dependent on the labour of others to meet their primary needs. There was
no question of a gradual growth towards larger units or centres such as the
millennium of development in Southern England which I will proceed to sketch
in a moment. These great concentrations of population arrived suddenly and
demanded immediate, adequate and daily provision of food, as well as the
institution of a network of general supply. As much of the requirement as
possible had to be drawn from the immediate vicinity on account of the limited
freshness of foodstuffs and the high cost of land transport. How could such
demands be met?
There are three possibilities:
1 • Self-sufficiency. This is not the place to pursue this point further, but
reference may be made to Von Petrikovits study (1960) of this aspect.
2. Imposition of taxes (levies) on the native population in return for military
protection.
3. , Trade.
In both the latter alternatives the local population must have possessed
some form of centralization if the Romans were to exploit them effectively
(i.e. economically). Of even greater importance in both instances is that the
native economy must already have produced a surplus, since over-exploitation
cannot be maintained for centuries. Fentress (1979, 177) also touches on
this point in her study of the Roman army in Numidia: " In order to sell
agricultural products to the army-or to merchants selling to the army-a
farm must have produced a surplus. This is obvious enough, until it is re-
membered that the tribal economy of the area was probably producing exclu-
sively for subsistence before the Roman occupation." This implies that the
preconditions for a successful occupation were met only in those regions where,
prior to the coming of the Romans, developments had already taken place
which had resulted firstly in the production of agricultural surplus (and also
a surplus of labour) and secondly in the emergence of a social structure
capable of concentrating that surplus in one place and of redistributing it by
means of trade •

1038
 Archaeologically, such developments should be recognizable in defended
storage places for goods, central to the local settlements involved in produc-
tion and well situated with. regard to transport facilities. Can such locations
be identified in NW Europe! and, if so~ can they be correlated with the Roman
occupation? The answer may be found in the comparison of two regions, namely
Southern England and the Netherlands north of the Rhine.
For Southern England, we material in the studies
presented by Cunliffe (i 976a, b with references) Towards the end of the 2nd
millennium B. C, CunHffe sees several co-existent types of settlement. 1.
farmsteads, sometimes in small groups, as the smallest social unit, 2. en-
closed areas not inhabited but which have functioned as a
sort of cattle for the several farmsteads in vicinity at particular
periods of the year~ and which at same time served as meeting places
for a more dispersed and 3. so-called plateau-enclosures
which functioned as assembly places for. the farming population drawn from
a larger area than that the meeting places mentioned previously. Cunliffe
assumes that this settlement pattern reflects a stable egalitarian society.
In of B. C. we see emergence of
hill forts through.out much Southern Britain. These may located on the
site of a former farmstead, or on one of the old central places just mentioned,
or on a ne'vF U) construct of the fort,
the necessary manpower must have been diverted from agricultural pursuits-
where it can no longer have been essentiaL This brings us to the subject of
a surplus of manpower and also Excavations in certain of these
early has be interpreted as
huts for grain storage. this interpretation correct, then it might be
permissible to think terms of communal stora.ge of surplus products and
perhaps also of organized redistribution~
• From the middle of the first millennium B. C. an imrx>rtant change occurs
in the distribution pattern of the hill forts as it had hitherto existed. A number
seem to have gone into disuse while others seem to increase in importance.
Archaeologically this process is reflected in, for instance, the very large
numbers of storage pits datable to this period (on one site no less than 5500
represent a period of four centuries) and rows of granaries, which were re-
built a number of times. The remaining hill forts were now better fitted out
for defence and it is i<'1 this period that the epithet !fort' seems first to be
justified. By mapping the forts and utilizing natural boundaries such as rivers,
it can be shown that the territories served are approximately equal in size
(£. 100 sq km). The trend of the development continues~ eventually resulting
in hill forts dominating an area approximately four times as large.

The function of these hill forts may on the existing archaeological evidence
be seen as centres within clearly defined territories~ maintained by an
authority which could mobilize the surplus of manpower and of agricultural
products when necessary. The forts served the hinterland as religious centres,
markets, centres for redistribution and possibly also in the specialized
production of manufactured goods. Cunliffe sees the permanent settlements
in the hill forts as the precursors of an urbanized social structure.

1039
 Rapid social and econ~mic developments in the first century A. D. led to
the growth of trading contacts ·with the Continent (import of amphorae with
Mediterranean wine) which· in turn stimulated new, specialized, extra-terri-
torial trading centres (cf .. Polanyi 1957: ports of trade). In the last century
B. C. and the first A. D. even larger units-the territorial oppida-came into
. being, controlling areas of over 5000 sq km, within which one or more distinct
settlements might be located •. It is the territorial oppida which in general
tend to survive throughout the Roman occupation. The market area now be-
comes too large and satellite markets spring up besides the lower order set-
tlements. The complexity. of the organization is proof that the process of
urbanization is fully unde~way. The Roman administrative structure was
projected onto the existing ·organization, and we indeed see that civitas capitals
are generally situated in or near to such territorial oppida.
To turn now to the Netherlands north of the Rhine~ Here a few structures
have been recognized whicp Waterbolk (1975, 1977) suggests are to some
degree comparable in function to the hill forts of Southern England. Here I
will concentrate on assessing how far these sites fit into Cunliffe's model as
detailed above.
I refer to three structures situated 3-4 km from one another, each
roughly square in shape, bounded by an earthen bank which is reinforced on
both inner and outer faces by· a palisade. Outside the bank is, although not
in all periods:~ a shailow ditch. Similar enclosures are said to occur else-
where in the province Drente. The beginning of these structures must be
dated to the Pre-Roman Iron Age and they seem to continue in use into the
Roman period. Roman imports-second century Samian Ware-were found
on one of the three sites, although Waterbolk considers this material to form
a terminus ante quem for the period of use of the enclosures. The familiar
type of long house (living quarters and byre under one roof) for this region
is entirely absent. Howev:er, rectangular post arrangements which could be
interpreted as granaries, do occur. All in all, Waterbolk arrives at the
conclusion that these defended sites should be regarded as stock enclosures
and depots for the harvest surplus. Their situation on the upper reaches of
small rivers which form a direct link with the low-lying terp (house-mound)
region suggests a function related to the interaction between the areas of
higher sandy soils and the. terpen. That these roughly contemporary defended
sites, in themselves relatively small, are situated only a short distance from
one another is, according to Waterbolk, an indication of the extremely re-
stricted area which they s.erved.
If we compare this situation with the model proposed by Cunliffe, then
these continental defended sites to the north of the Rhine would seem to stand
right at the beginning of the whole process of development. There is hardly
evidence for extensive surpluses of foodstuffs or organized redistribution,
let alone for the first steP.s to an urbanized social form. Neither do ports of
trade or organized long distance trade yet enter the question. The virtual
absence of Roman imports also indicates that these storage places cannot
yet have assumed the functions which Cunliffe attributes to the territorial
oppida.

1040
 Certain other sites on the Continent are relevant to the discussion. At
Bentumersiel in NW Germany~ early Roman imports, some of them military
in origin, have been associated with the campaigns of Germanicus in 15-16
A., D. (Ulliert 1977). But the traces of constructions and the osteological
material (absence of young animals) (Zawatka & Reichstein 1977) are not
consistent with the normal type of native farmstead, nor do they bear a
characteristically military stamp (Brandt 1977). Could this also be a central
place in Cun.liffeYs sense, but then for military use? A centre with a similar
function might have been located at Bruggeburen, near Winsum in Friesland,
according to the finds of early Roman imported material. Something similar
could also be suggested for Het Hain at Krommenie, in the province N Holland
(Halliertsma 1 Trimpe Burger 1964; Helderman, 1971) in view of
the quantity early Roman material found there, in marked oo_ntrast to the
paucity of such finds in the native settlements roundabout (Groenman -van
Waateringe~ Glasbergen & Hamburger i 961)~ Even so, various interpre-
tations are possibleQ Het Hain might have been a place of assembly for the
native po:pulation and at the same time a sanctuary, where Roman imports.
whether o r fragmentary~ functioned as a medium of exchange. But
it been a native market place which ·,vas used by the
Romans (perhaps from the fort at Flevum ?) to obtain their necessities from
the natives. At present there is no evidence from the Nether lands north of
Rhine of social structures as advanced as those existing
in Britain on the eve of the Roman conquest.
"\Vhat about the other pa.rts of the ---
' ,,
Limes?
study by (1976) en1phasises the geographical spread the
urbanized settlements of the first two centuries B. C. They extend a broad
zone from SE England to Slovakia, but are absent from the north European
plain. and Scandinavia: the frontier between the urbanized and non-urbanized
area thus roughly corresponds to the N frontier of the Roman empire, which
stretched from Britain to Hungary. The great extent of these oppida stimu-
lated extensive trading networks for the provision of their inhabitants and led
eventually to the emergence of a market economy making use of coin.
The hill forts to the north of the Limes in Central Germany north of the
Danube and in Czechoslovakia apparently do not continue into the Roman
period, but were abandoned towards the end of the first century B. C. , an
event frequently associated with the Marcomannian invasion of i0-15 B. C.
In this region, urbanized settlements only reappear in the Slavonic period.
We may also note a passage from Cooter (i 976, 93) "In much of the eastern
Mediterranean, Rome had inherited Hellenistic kingdoms that had usually
been far more sophisticated than the Italian allies of the Roman Confederacy.
Provincialization was also a relatively easy matter in the more developed
portions of Carthaginian Africa and Spain and in regions of southern Gaul
long associated with Greek colonies such as Massilia'' (cf. also Nash i 976).
I believe that it is now possible to conclude that a successful permanent
Roman occupation was only IXJSSible in those regions where the Romans were
confronted with a well-organized proto-urban or urban structure, which they
could utilize for the supply of their armies and upon which they were able to
project gradually their social and administrative system. For its food supply,

1041
the Roman army was heavily dependent on pre-existent central places, ·where
local produce was concentrated and where a market economy with long distance
trading networks was fully functional.
The Romans were able to extend beyond the 'frontiers of urbanization'
in certain geographically favoured areas of Europe. The occupation of Britain
for example extended further north and more to the west than the area covered
by Cunliffe's territorial oppida, in Gaul and the Low Countries the Atlantic
coasts and the course of the Rhine formed the frontier.
Since water transport was very much less expensive than land transport
(see, for example, Peacock 1978) the westward courses of the great rivers
in Qaul enabled expansion outside the directly economically integrated area,
and in the Low Countries the Rhine had the same effect. This situation cor-
responds almost exactly to that described by Lattimore for the structure of
frontiers (1962, 480), when he distinguishes three zones of influence with
ranges of decreasing order:
a. an outer radius with unification by military action.
b. an inner radius with centralization under uniform civil administration.
c. a central area with economic integration.
The outer radius "reached into territories that could be invaded, with
profit in plunder or tribute, or for the purpose of breaking up barbarian con-
centrations dangerous to the state, but that could not be permanently annexed".
This I would equate with military actions across the Limes into German
territory. The inner radius "reached over territories that could, after con-
quest, be added to the state" i.e. for example the NW fringes of the Roman
Empire, outside the originally urbanized or proto-urbanized areas, whereas
real economic integration was restricted in practise to these latter areas.

Translated into English by Dr. Carol van Driel-Murray.

1042
 REFERENCES

Brandt, K., 1977. 'Die Ergebnisse der Grabung in der Marschensiedlung

Bentumersiel/Unterems in den Jahren 1971-1973.' Probleme der KUsten-
forschung im stldlichen Nordseegebiet 12, 1-31.
Collis, J., 1976. 'Town and market in Iron Age Europe' in B. Cunliffe &
T. Rowley (eds.), OppidainbarbarianEurope, 3-23, Oxford.
Cooter, W. s., 1976. Preindustrial frontiers and interaction spheres:
aspects of the human ecology of Roman frontier regions in Northwest
Europe e Unpublished dissertation. Ann Arbor, University of Michigan.
Cunliffe, B., 1976a. 'Hill-forts and oppida in Britain' in G. de G. Sieveking,
I. H. I..ongworth & K. E. Wilson (eds. ), Problems in economic and
social archaeology, 343-358, London.
Cunliffe, B. , 1976b.. 'The origins of urbanisation in Britain' in B. Cunliffe
& T. Rowley (eds. ), Oppida in barbarian Europe, 135-161, Oxford.

Fentress, E. W. B., 1979. Numidia and the Roman army, Oxford.

Groenman-van Waateringe, W., W. Glasbergen & M. F. Hamburger, 1966.
'Een boerderij uit de eerste eeuw na Chr. te Krommenie (N. H.)' in W.
Glasbergen & W. Groenman-van Waateringe (eds. ), In het voetspoor van
A. E. van Giffen. 110-123, 178-179, Groningen.
Halbertsma, H., 1964. 1Assendelft .. Met naschrift van J. A. Trimpe Burger',
Nieuwsbull. KNOB 17, 100-101.
Helderman, E. J., 1971. 'Enige resultaten van vijftien jaar archeologisch
onderzoek in de Zaanstreek', Westerheem 20, 52-83.
Lattimore, 0. , 1962. Studies in History, Oxford.
Nash, D., 1976. 'The growth of urban society in France' in B. Cunliffe &
T. Rowley (eds.), Oppida in barbarian Europe, 95-133, Oxford.
Peacock, D. P. S., 1978. 'The Rhine and the problem of Gaulish wine in
Roman Britain' in J. du Plat Taylor & H. Cleere (eds.), Roman shipping
and trade: Britain and the Rhine provinces, 49-51, London.
Petrikovits, H. von. 1960. Das rl:Jmische Rheinland: archnologische
Forschungen se it 1945, Kl:Jln.
Polanyi, K., i 957. 'The economy as instituted process' inK. Polanyi,
C. M. Arensberg & H. W. Pearson (eds.), Trade and market in the early
empires, 243-270, New York.
Ulbert, G., 1977. 1 Die rl:Jmische Funde von Bentumersiel', Probleme der
KUstenforschung im sUdlichen Nordseegebiet 12, 33-65.

1043
Waterbolk, H. T., 1975. 'De 'versterking naar Romeins patroon' bij Zeijen
opnieuw bekeken 1 in J. s. Boersma et al~ (eds.) Festoen. (Scripta
Archaeologica Groningana 6.) 63 7-65 2, Groningen-Bus sum.
Waterbolk, H. T. , 1977. 'Walled enclosures of the Iron Age in the North of
the Netherlands', Palaeohistoria 19, 97-172.
Zawatka, D. & H. Reichstein, 1977. 'Untersuchungen an Tierknochenfunde
von den rljmerzeitlichen Siedlungspltttzen Bentumersiel und Jemgum-
kloster an der unteren Ems /Ostfriesland 1 , Prob le me der Ktistenfor schung
im sttdlichen Nordseegebiet 12, 85-12 8.

1044
 71. BARBARIAN INVASIONS AND FRONTIER POLITICS
IN THE REIGN OF GALLIENUS

John W. Eadie

Confronted by defective and notoriously tendentious sources, and beset

by agnostics who insist that further research will not produce verifiable inter-
pretations, historians of the third century A.D. have not surprisingly sought
refuge in convenient orthodoxies. The employment of preconceived generali-
zations to bridge awkward gaps in the extant evidence and to explain particular
events is especially common in modern reconstructions of the period between
the death of Severus Alexander and the accession of Diocletian.
One of the more popular generalizations, one that appears in most of the
standard accounts of the period, is the assertion that each time vexillations of
the German or Raetian legions were transferred to other frontier zones the
barbarians attacked Roman settlements and forts on the Rhine and Danube
(e. g. AlflHdi 1939, 159). Thus, it is commonly believed that Gallienus' deci-
sion in 258 to commit legionary and auxiliary vexillations from the German
garrisons to indefinite service in Italy and Pannonia critically weakened the
Rhine defences and invited the barbarian attacks the following year. These
attacks in turn persuaded the dispirited remnant of the Rhine army, the argu-
ment runs, to support the rebellion of Postumus (2591260), to murder Galli-
enus' son Saloninus and his guardian Silvanus, and to establish the rival
imperium Galliarum. In other words, students of the period have concluded
that two of the most important events of Gallienus' reign resulted from his
decision to transfer troops from the Rhine zone.
Several inscriptions from Sirmium, the headquarters of the expeditionary
force, attest the transfer of the vexillations to Pannonia after 258. The first of
of these (CIL ITI 3228 = Mirkovi6 19 = Saxer 101) reveals that British as well
as German troops participated and that auxiliary units accompanied the legions
to which they were regularly attached.
[I(ovi) O(ptimo)] I M(aximo) I Monitori p<ro> I salute
adque incolumitate I d(omini) n(ostri) Gallieni Aug(usti) I
et militum I vexill(ationum) leg(ionum) I [g]ermaniciana(rum) I
[e]t brittanicin(arum) [cu]m auxili(i)s I [e]arum I
[.... V]italianus [.... pro]tect(or) Aug(usti) n(ostri)
[praepo]situs I [v(otllm) p(osuit)
Since Valerian's name has been omitted we may infer that the altar was erected
sometime after 260 (pace Fitz 1976, 13); nothing in the text, however, indi-
cates when the units arrived in Pannonia. Presumably they had served with
Gallienus on the Rhine until 258, remained with the emperor during his stay
in Italy, and then set off for Pannonia either in 26 0 to combat Ingenuus and
Regalianus (Saxer 1967, 55) or in 261 to take part in the campaign against the
east"!rn usurper Macrianus (Sa~ell 1961, 14 ff. ).

1045
 Although the number and identity of the vexilllations serving in Pannonia
cannot be determined, it is likely that one of the British units was the detach-
ment of the XX Valeria that had been transferred to Mainz in 255 (CIL XIII
6780). With respect to the composition of the German vexillations we are
on firmer ground since three inscriptions from Sirmium document the parti-
cipation of two legions, the I Minervia regularly stationed in Bonn and the
VIII Augusta from Strasbourg.
(Mirkovic 32 = Sasel and &tsel 273)
D(is) m(anibus) I Restitutius I Silvanus I (centurio)
leg(ionis) I Men(erviae) I Aur(elius) Martius I heres
et I Aurelia Bona I sec(Unda) heres I coniugi benelmeren(ti)
f (aciendum) c (uraverunt)
(Mirkovic 35 = §asel and Sa~el 27 4) .
[ • . . . . Jmus . . . . ] nius Samlminus (?) I
im(a)g(inifer) leg(ionis) I VIII Aug(ustae) I fecit
Verina coiux
(Mirkovi6 36 = §asel and Sasel 272)
D(is) m(anibus) I Gratius Arillius I et Cludius Clamolsus
(centuriones) leg(ionis) VITI Aug(ustae) I bello
Serdicensi I desideratis schol[la] (centurionum)
leg(ionis) s (upra) s(criptae) collelgis bene merenltibus
d(e) s(ua) p (ec::unia)
If the bellum Serdicense designates the campaign against Macrianus, as Sasel
(1961) has suggested, the last monument, and possibly the other two as well,
must have been erected in 261 or shortly thereafter.
Whether the other German legions- XXII Primigenia (Mainz) and XXX Ulpia
Victrix (Xanten)-were represented in the expeditionary force Gallienus had
assembled is uncertain. We know that all four German legions had participated
in the remarkably similar strikeforce conducted by Septimius Castinus in 206-
208 "adversus defectores et rebelles" in Pannonia (CIL XITI. 1047112 = Hoffiler-
Saria 362 = Saxer 86-88). On at least two other occasions during the century,
however, vexillations had been recruited from individual German legions.
A unit of XXII Primigenia was stationed at Romula (in Dacia) during the reign
of Philip the Arab (AE 1940, 13 = Tudor 1968, 495-496) and a vexillation of
the VIII Augusta, originally intended for Gordian Ill's Persian campaign, was
commemorated on coins of Philip the Arab from Heliopolis (Saxer 95 -96). In
short, there is no reason to believe that all four German legions were invariably
involved in extra-provincial expeditions. Even if Gallienus did recruit a
vexillation of the normal size (500...;1000) from each of the German legions, no
more than 4000 men would have been available for duty in Italy and eventually
in Pannonia. Is it likely, one may ask, that a transfer on this scale would
have invited the barbarian attacks in.259l260 or impairedthe effectiveness
of the Rhine army in resisting these attacks?
Most historians have accepted at face value Aurelius Victor's statement
(32.1; cf. Zos. 1. 28·29) that the frontierforts in southern Germany were

1046
evacuated when Valerian recalled the troops from Raetia in 253 to combat his
political rivals in Italy. Moreover, many believe that this withdrawal "in-
spired" the Alamannic attacks later in the year that purportedly devastated
much of Raetia and Germania Superior. The evidence of destruction in this
zone, however, is far from conclusive. On a given site it is usually impossible
to distinguish the destruction horizons of 253 and 259/60. Moreover, although
inscriptions that can be securely dated after 250 have not been found in the
frontier forts. a building inscription from the region of the fort at Heidenheim
(CIL III 5933) indicates clearly that the Romans had not relinquished entirely
their claim to transdanubian territory. Similarly, in the region of the civitas
Ulpia Sueborum Nicretum two milestones (CIL XIII. 9103 [Ladenburg] and 9111
I_Heidelberg]) reveal that Roman officials, and presumably Roman troops,
were still active in this sector of the transrhenane zone under Valerian and
Gallienus.
It was the Frankish threat to Germania Inferior, and not the Alamannic
attacks in the south, that evidently prompted Gallienus' return to the Rhine in
256. We know that Gallienus selected Cologne as the headquarters for his
campaign and set about strengt~_~n:ing the to-wn's def~nces_by repa~ing some of
the towers on the existing wall (CIL XIII. 8261). He also relocated in Cologne
the mint that he had transported from Viminacium. His overall strategy cannot
be determined, but there is no reasonto doubt Zosimus' statement (1. 30. 2)
that his principai objective was the defence of the Rhine crossings. That he
envisaged a major campaign across the Rhine is unlikely. Indeed, according
to Zosimus (1. 30. 3) 1 he entrusted the defence of the transrhenane zone, or
at least a portion of it, to a German chief with whom he had negotiated a client...
treaty. It may well have been this traditional diplomatic solution that led the
author of the Laterculus Veronensis (Riese 1892, 208) to conclude that Gallienus
had abandoned the transrhenane civitates .
. The two German 11 victories" that Gallienus celebrated in these years pro-
bably were won in Ir.Jerior rather than Superior. Nothing in the archaeological
record or the literary accounts suggests that he attempted to recover territory
in southern Germany that had been ntost" to the Alamanni. On the other hand,
when Gallienus departed for Italy in 258 the garrisons and towns in the central
section of the Rhine zone, the stretch between the legionary camps at Xanten
and Bonn, were securely in Roman hands (cf. Demougeot 196'9, 490; Drinkwater
1973, 252ff. ). A remarkable achievement when one considers that the Franks
had managed to penetrate the defence perimeter in the north and, after a brief
stay in Gaul, had besieged and severely damaged Tarraco in Spain (Orosius 7.
22. 8, 41. 2; Aur. Vict. 33. 3; Eutropius 9. 8). One can only conclude that
Gallienus' strategy in Inferior had achieved its purpose.
Although the raids continued in Inferior over the next two years, there are
no reports of catastrophes or major losses of territory. Even when the garri-
son was withdrawn from the transrhenane fort of Niederbieber around 260 the
defence of the left bank apparently was not impaired. The garrisons and towns
in this region were evidently intact when Postumus claimed them in 259/26 0
and they remained for more than a decade the bulwark of the imperium GalUarum.
Clearly, Gallienus' .decision to transfer troops from the I Minervia in Bonn,
and perhaps from the XXX Ulpia Victrix in Xanten, had not incapacitated the
Roman defence network in Inferior.

1047
 The suggestion (e.g. Demougeot 1962, 9; deBlois 1976, 6) that the trans-
fer of troops from southern Germany may have provoked the Alamannic attack
in 259/60 is neither necessary nor persuasive. As the Alamanni had been
raiding Roman territory for several decades, there is no need to suppose that
they were responding to a particular opportunity in 259/260. Moreover, it
is difficult to believe that they would have had knowledge of troop movements
in Mainz or Strasbourg, towns that were some distance from the territory they
had occupied. It is far more likely that the traditional imperatives of barbarian
existence, population pressures and nutrition, propelled the raids on Roman
settlements. On the other hand, one cannot dismiss the possibility that Gal-
lienus' campaign in 256/257, or perhaps the activities of the client he had es-
tablished in the transrhenane zone, may have provoked an Alamannic response.
This much is certain: the Alamannic attack in 259/260 was more debili-
tating than the earlier raids. Whatever one makes of Gregory of Tours'
account (1. 32; cf. Demougeot 1962, 22ff.) of Chrocus, rex Alamannorum,
it is evident that superior commanders were now in charge of their warbands.
No longer content with smash-and-grab raids on peripheral settlements, they
penetrated deep inside the frontier provinces-threatening towns in Switzerland
(Vindonissa-CIL Xill. 5203; van Berchem 1955, 160ff.; Aventicum-Pseudo-
Fredegarius 2. 40), southern France (Arvernus Gregory of Tours 1. 32),
and eventually Italy-and were able to remain in the field beyond a single cam-
paigning season.
The success of these attacks, however, should not be exaggerated. While
it is true that the Romans were forced to surrender for a time their trans-
rhenane forts and settl-ements, nothing indicates that tke legions were expelled
from Mainz and Strasbourg (cf. Pekary 1971, 128ff. ). Because the Alamanni
were not yet able to establish permanent settlements on the territory they had
overrun (Drinkwater 1973, 241ff. ), the reverses suffered in 259/260 did not
force the evacuation of alLthe camps and towns in Superior. Speyer, for
example, survived and shortly thereafter was advanced to colonial status by
Postumus (CIL XIII. 9092). Moreover, many of the forts in the critical Mainz
area were still occupied under Postumus and his successors (Schlinberger 1969,
177ff. ) • In short, the Rhine now became the effective boundary of the province,
protected as before by legionary and auxiliary units. Since the retention of
territory, not the prevention of attack, had always been the primary military
objective, I think it is fair to conclude that the transfer of the vexillations in
258 did not disable the Roman defence of Superior.
The connection between the transfer of the vexillations and the rebellion
of Postumus that some (e.g. de Blois 1976, 6) have discerned is equally pro-
blematic. The frequency of raids in the Rhine zone between 253 and 260 may
well have engendered a sense of foreboding that strengthened the predisposition
to rebellion, but it does not follow that the transfer in 258 persuaded the troops
to challenge Gallienus' authority. If the legions resented the transfer, why did
they not rebel soon after Gallienus returned to Italy? Instead, they delayed
for more than a year, and even then only the units in Inferior participated in
the elevation of Postumus.
The fact that the rebellion occurred in Inferior is significant. It was the
troops in this province that Gallienus had rallied in 256/257 against the Franks.
And it was these same troops who ·had successfully defended the region after

1048
258. What could have motivated these experienced front-line units, who
presumably had received their share of laurels, to defy imperial authority?
The key to the rebellion is unquestionably the 'booty incident" that
Zonaras (12. 24) describes in some detail. He reports that Postumus inter-
cepted a barbarian raiding party somewhere in the Rhine area and recovered
the goods they had stolen from Roman settlements in Inferior. His soldiers
immediately claimed the contraband and asked that it be distributed as booty~
but Silvanus, on learning of their success, demanded that the captured goods
be sent to Cologne. The troops responded by proclaiming Postumus emperor,
marching on Cologne, and assassinating Silvanus and Saloninus.
While the motives of rebels can never be securely identified, Zonaras'
assessment of the incident is persuasive. Nothing in his account contradicts
what we know, or can reconstruct, of military and political conditions on the
Rhine in 259/260. That Postumus should seize the opportunity presented by
the controversy over the booty to challenge the authority of Silvanus, and there-
by remove a rival in the imperial court in Cologne, is hardly surprising.
Nor is it difficult to understand the troops' decision to follow Postumus rather
than accede to Silvanus' demands. Zonaras, in short, provides a credible
account of the rebellion. It is not necessary, therefore, to invoke the transfer
of the vexillations to explain Postumus' actions or those of his troops.
I do not mean to suggest of course, that we should banish all working
hypotheses. But if we are to conduct a proper investigation, following the
sources where they lead, we must be prepared to put existing generalizations
to the test. In this paper I have attempted to demonstrate that we possess
sufficient contextual data on which to base our reconstruction of the events in
question. The appeal to convenient orthodoxies, whatever their value in other
contexts, will add little to an evaluation of the barbarian invasions or frontier
problems in the reign of Gallienus and may serve only to distort further inquiry.

BIBLIOGRAPHY

AlHHdi, A., 1939. 'The Invasions of Peoples from the Rhine to the Black Sea'
in Carob. Anc. Hist. XII, 138-164.
de Blois, L., 1976. The Policy of the Emperor Gallienus. Leiden.
Demougeot, E., 1962. 'Les martyrs imputes a Chrocus et les invasions
alamanniques en Gaule meridionale', Annales du :Midi 74, 5-28.
Demougeot, E., 1969. La formation de l 'Europe et les invasions barbares,
Paris.
Drinkwater, J., 1973. A History of the Gallic Empire of the Third Century
A.D. D. Phil. Diss. Oxford.
Fitz, J., 1976. La Pannonie sous Gallien, Coilection Latomus 148, Brussels.
Hoffiler, V. and Saria, B. , 197 o2 . Antike Insc hriften a us Juga slavien,
Amsterdam.
Mirkovic, M., 1971. 'Sirmium- its History from the I Century A.D. to
582 A. D.', Sirmium I, 5-90.

1049
Pekary, T., 1971. 'Zur Ostgrenze des Gallischen Sbnderreiches im 3. Jr.
n. Chr.' in Roman Frontier Studies, 1967, 128-131, Tel Aviv.
Riese, A., 1892. Das Rheinische Germanien in der Antiken Litteratur, Leipzig.
Sasel, J., 1961. 'Bellum Serdicense', Situla 4, 3-33.
Sa~el, A. and Sasel, J., 1963.
'Inscriptiones Latinae quae in Iugoslavia
inter annos MCMXL et MCMLX re portae et editae sunt ', Situla 5.
Saxer, R., 1967. Untersuchungen zu den Vexillationen des rlJmischen Kaiser-
heeres von Augustus bis Diokletianus, Epigr. Stud. I, Cologne.
Schdnberger, H. , 1969. 'The Roman Frontier in Germany: An Archaeological
Survey', J. Roman Stud. 59, 144-197.
Tudor, D., 19682. Oltenia Romana, Bucharest.
van Berchem, D. , 1955. 'Aspects de la domination romaine en Suisse',
Revue Suisse d'Histoire 5, 145-175.

1050
 72. TYPOLOGY AND LATE ROMAN FORTIFICATION:
THE CASE OF THE 'DIOCLETIANIC TYPE'

James Lander*

Research on late Roman fortification-and thus on Roman frontier history

-has occasionally turned,· and faltered, on problems of typology. This paper
will briefly review some aspects of those problems.
Very few 'fort-types' in late Roman fortification have been identified or
agreed upon, and most forts remain untyped or grouped in very general clas-
ses. But there is at least one fort-type which, in the past, has been widely
referred to: this is the 'Diocletianic type' of castella (which are also in some
cases called guadriburgia).
These days the notion of this 'Diocletianic type' is nearly dead-and I do
not wish to revive it, but would prefer to see the ghost laid firmly to rest;
and, in doing so, to suggest an approach which in the future might allow us to
make some progress toward filling the current typological void with regard
to late Roman fortification.

In the beginning there were two major supports for the so-called 'Diocle-
ti.anic type: one is the fact that a Diocletianic program of fortification-building
across the Empire is attested in our sources (Paneg. Lat. 9, 18, 4; Amm.
Marc. 23, 5, 1-2; Zosimus 2, 34; and Malalas Chron. 12, 308); the other
fac,tor is that, since early in this century, researchers have been rather
startled by the great similarity they have seen in the design of certain castella
in widely separated parts of the Empire. One of the earliest studies compared
the forts of Irgenhausen and Schaan in Europe with such forts as Muhattet el
Haj in modern-day Jordan (fig. 72. I) (Anthes 1917, 138-142; cf. Kornemann
1907' 113) .1
These forts offered a basis for the earliest type-description: they are
relatively small forts, square in plan, with square angle- and interval-towers
which saddle the curtain wall; there is a single defended gate, and the bar-
racks are arranged against the inner face of the curtain wall, creating a
courtyard within the fort (similar descriptions are repeated in Fabricius 1926,
57 8, Poiaebard 1934, 54, Schleiermacher 1951, 173 and Gudea 1974, 179).
To this 'typological core' other forts have been added through the years
by various scholars studying various frontiers. 2 This has caused a slight
loosening of the type-description: for if we include the examples in figure
72. 2-and their chronology makes them suitable candidates-then we must
make allowances for variations in fort-size, in the total number of towers,
and in the disposition of those towers (not only saddling the wall, but also
projecting completely beyond it, as in the case of Dionysias (72.2). (I must
add that certain irregularities ~n the plan of Deir el Kahf (72. 2a) seem to
post-date the original construction.)

1051
 FIGURE 72.1

The Core of the Type

(see note 5)

,---- ....... ,
: i----i '------------·------
1 I I

I L....... .!
'--·I ,.. J
r-------------·..1--
I
I 1
I I
I 1
I I
I I

: !
: : I I
......... i
.----- I r-: ___ ~-j
It 1 I
1 I
I___ .J I
1
I
I
I
1
I
I
I ' 1 ~~- ..... 1 I

.......• I I
..........
,
1
i--I
I
I I
I I
I I
I I
I I
I I
: I
I I
I I

;-,:___ ~ ..... -- ----

'' I'
L::::_r-------
'I

(a) (b)
IRGENHAUSEN SCHAAN

(c)
MUHA 1"fET EL HAJ

0 so
meters

1052
 FIGURE 72.2
,•.-.•J"e•a•a•a•a•a•a•a•i/!.f!!La
·······•···•·••••··•·••••••··•· Additions
to the Type
(a) (see note 6)
DEIR EL KAHF
:~:;
306 ••..
•!(
..•.
•...
.•..
.••.
:·:·
....
·:·:
.•.•
....
:·:·

0 50

meters

.....
••.o.
·····
::::
.•..
(b) ·:·:
::::·
..•.
....,
:-:~
GORNEA
c. 294-303
...·:·:•e••••...,
:-:~
·:·:
:-:.,
··•··
·:-:
:-··
··=~

I
=·=·=·=·=
···•·•···
:::::::::

(d)
DIONYSIAS
c. 306
I

1053
 One feature which is allowed by some scholars (Fabricius 1926, 578,
van Berchem 1952, 11-12 and Atanassova-Georgieva 1974, 170), but not by
others (Poidebard 1934, 52 and Gudea 1974, 179) as an attribute of the 'Diocle-
tianic type' is the rounded tower. We see such towers on Di~nysias (72. 2d),
though recently argued to be Palmyrene constructions (Carrie 1974, 824-850),
and on Khan el Qattar (72. 3d), which, partly for this reason, was excluded by
Poidebard (1943, 52) from his 'Diocletianic type'. But the soundness of
Poidebard's chronological criterion is highly questionable, 3 and van Berchem
(1952, 13-14) tentatively included Qattar as part of the Diocletianic syste:n in
Syria. Castra Martis (72. 3c) in Dacia has been enrolled in the type despite
its round towers (Atanassova-Georgieva 1974, 170), and at least one scholar
(Fabricius 1926, 578) wished generally to include the Saxon Shore forts o.f
Britain, many of which are noted for their rounded towers. However, With
their variations in size and plan, not to mention their peculiar function (Johnson
1976 passim), the Saxon Shore forts as a group can only loosely be associated
with the Irgenhausen-Schaan-Muhnttet type-cluster.
On the general chronological problem, we could possibly change the mis-
leadingly specific name of the 'type' from 'Diocletianic' to something like
'Tetrarchic ', or, more generally still, to 'late 3rd/early 4th century' in order
to include Bourada (72. 3b), a dated fort in Numidia. But even these designa-
tions will not encompass Qa~ el ~allabat (72. 3a), which dates to the early
third century and which appears to have some claim for inclusion in the type
(though, again, these towers are a secondary construction, and we do not
know the original plan of the fort).
One might add that Dr. Gichon-in a memorable presentation to this
congress-referred to one or two structures in Israel which he believes were
built in a period much earlier than Diocletian's reign, and yet which, in the
sum of their attributes, are remarkably similar to forts of the so-called
'Diocletianic type'. The evidence requires further examination.
All in all, there are so many problems with the description and chronology
of this particular group of late Roman forts that it remains difficult to refute
Prof. von Petrikovits' opinion that 'any tendency to date ... Roman fortifications
on typological grounds ... (with a few exceptions) is worse than useless in the
late Roman period' (1971, 203).
And yet we are left with this overall similarity of design-though that
phrase must be qualified-for what these fortifications have in common is the
way they differ from fortifications of the early Roman Empire. And while
'typology' is generally defined as a study of similarity, it is also-and I think,
more importantly-a study of variation; so that even the variations which we
see in the forts already mentioned-and in late fortification generally-should
not be considered a hindrance to typology, but rather a vehicle for it.
A typological study of variation is a tool for examining and explaining
change (which, of course, is what historians do). But a typology is rather
like an analogy: it is suggestive, but does not by itself prove anything. And
if a typology is conceived too abstractly or too rigidly, it can be like a false
analogy: not only uninformative, but positively misleading. Therefore, a
typology should relate to a specific problem of historical change. 4

1054
 FIGURE 7.2 .3

Possible Anomalies
(see note 7)

0 50

meters

(b)
BOURADA
c. 324-330

;:;:;:::;:;:;:;:;:;:;:;:;:~:;:::::::;:;:;:;:;:::;:;:;:;:;:;:;:;:;:::::;:::::::;:;:;:;:;:;:;:::::::::;:::::::::::::::::;:;~~====~=;:

(C)
(\):!..• --(d_)_-F)
CASTRA KHAN EL
MART IS QAT TAR
I!

1055
 Now, I have already mentioned the major features, or attributes, which
have been compared, associated and combined to create the so-called 'Diocle-
tianic type: but for a moment let us consi~er o~e attribute in isolation: namely,
the arrangement of barracks along the inner face of the curtain wall. For this
design-change there have 'been at least four explanations offered in the past,
which I will list briefly.
1. The first view is that the arrangement was a tactical requirement-in
order to avoid in-coming enemy missiles, which would pass over the wall
and, presumably, some distance beyond it (SchiJnberger 1969, 182 and
von Petrikovits 1971, 202).
2. A second opinion is that this barrack arrangement was intended to provide
space in the courtyard -~ither for sheltering livestock and refugees in times
of emergency (von_Petdkovits 1971, 203) or for the location of cisterns
(commented by Dr. Gichon).
3. A third view sees this change in barrack arrangements as being connected
with another change: from the construction of enclosure walls out of turf-
and-timber to their construction out of stone, so that the enclosure wall
is no longer damp earth, and constructing barracks against it is no longer
unhealthy but is actually economical of material (MacMullen 1963, 43f. ).
4. The fourth view, which on the whole I find the most attractive, concerns
the strategic (and also spatial) requirement of accommodating smaller
divisions of man-power in smaller spaces: for as forts become smaller
in area it makes less sense to have barrack blocks with passageways
around them.
One, all, none, or a combination of these views may give a true explanation
for this development, and it is also likely that the rationale would vary from
frontier to frontier. But, in any case, .!!:!!. these ideas could serve as multiple
workiilg-hypotheses for which implications could be deduced and then tested
archaeologically and historically.
No one has yet done this.
However, if the data were. gathered to test these implications, the infor-
mation could then be correlated with data relating to other attributes: the
size of the fort, the presence or absence of rounded towers, the location on
the frontier, the local building traditions, and so on. And from this corre-
lation, meaningful associations would be identified which could prove to be ·
useful types- 'useful' in the important sense that they explain (or rather are
explained by) an historical change.
·This brief paper, arising from work-in-progress on late Roman fortification,
unfortunately affords no opportunity to work out some detailed resolution for
the anomalies encountered in the notion of the so-called 'Diocletianic type':
except to suggest that the examples of fort-designs which have been mentioned
here could be said to belong to two 'categories'-which are certainly unworthy
of being called 'types'.
One category shows a very uniform design possibly imposed by bureau-
cratic writ from the offices close to the emperor-a blueprint which was then
'mechanically reproduced' (to use Prof. van Berchem' s phrase in Bell et al.

1056
1962, 21 ) across the Empire as part of Diocletian's massive re-organization
of frontier defense.
The other category (which I mustadmit is a catch-all) is more flexible
about fort-design, altering to local requirements: and the designs in this
category originate before and continue to evolve during and after the reign of
Diocletian. The course of this evolution has not yet been charted.
The major point I have tried to make is that for even such a simple classi-
fication as the one I have just suggested to be acceptable, it must be tested
with regard to specific problems. There is no fully standardized and univer-
sally applicable type. A type will be 'found' -and will be found 'useful' -""'Only
to the extent that it helps the researcher to resolve some particular problem
concerning the explanation of an historical change.

NOTES

* I am indebted to Prof. Ronald Mellor and Prof. Michael Speidel for reading
and criticizing an early draft of this paper; and particular thanks are due
to Mr. S. T. Parker for reading two drafts and improving them on several
points. I am also most grateful, for their comments and encouragements
during the congress, to Profs. E. Birley, J. Eadie, M. Gichon, and H.
von Petrikovits and Messrs. C. M. Daniels and D. Kennedy. Of course
none of the above is responsible for the remaining errors in this paper.
1. It is important to note that the oft-cited Diocletianic dating for either
Irgenhausen or Schaan is considered unlikely by Schonberger (1969, 179)
and von Petrikovits (1971_, 186).
2. In North Africa, Bourada (Guey 1939, 208; see my fig. 72. 3b);
Centenarium Aqua Viva (Leschi 1943, 9); Ad Aquas Herculis (Baradez
1949, 222); in Egypt, Qasr-Qarun/Dionysias (Schwartz 1950, 64; see
my fig. 72.2d); in Palestine, Kasr Gehainije (Gichon 1967, 189); and
on the Danube, Ravna (Kondic 1969, 118), Castra Martis (Atanassova-
Georgieva 1974, 169; my fig. 72. 3c), and Gornea (Gudea 1974, 179; my
fig. 72. 2b), among others.
The forts which Poidebard lists (1934, 54f.) under the 'type de Diocletien'
are: Mqehil (pl. 4), Khan as-Samat (pl. 15 and 16 ), Khan at-Trib (pl.
19), Qaryateyn (pl. 19), Khan Aneybe (pl. 27 and 28), Deir et Kahf (pl.
45; my fig. 72. 2a), and Diyate (pl. 50).
The Eastern forts which stimulated Kornemann' s and Anthes' comparisons
with Western forts are to be found in Brilnnow and Domaszewski (1904-
1909), esp. I, 44, fig. 30; II, 8, pl. 41; 38, fig. 62; 49, pl. 43; 62,
fig. 644; 95-97, pl. 44; 265, fig. 86 0-861 and Ill, pl. 53; and in Prin.
Arch. Exp. (1907-1949), esp. II, A, 72 and 146. --
3. Poidebard' s set example for his tower-shape criterion concerned the
legionary fortresses of Lejjiin and Udhrul}., which, because their angle-
towers begin as squares and then round out, Poidebard dated 'perhaps to
Trajan, but certainly not later than Marcus Aurelius' (1934, 52). A

1057
 recent survey indicates that both Lejjun and Udhru9- are, in fact, late
Roman, probably Diocletianic (Parker 1976, 24 and 27).
4. For a recent and excellent discussion of the internal organization of late
Roman fortification (with special reference to Diocletian's Palace in
Split) which relates the 'Sacrum Palatium' to the'Sancta Principia' in the
light of the developing 'Kaiserkult', see Fellmann 1979, 47-55.
5. Irgenhausen: (plan) Anthes 1917, 138. Schaan: (plan) ibid., 141;
(additional description) Beck 1959, 31-33. Muhattet el Haj: (plan)
BrO.nnow and Domaszewski 1904-1909, I, 44, fig. 30.
6. Deir el Kahf: (inscr. dating 306) Prin. Arch. Exp. 1910, III, A, 2, no.
228; (plan) ibid., 1909, II, A, 146; (photo) Poidebard 1934, pl. 45.
Gornea: (stamped tile dating 294-303) Gudea 1974, 175£.; (plan) ibid.
Qasr Bshlr: (inscr. dating 306) CIL III 14149; (plan) BrO.nnow and
Domaszewski 1904-1909, II, 49, pl. 43. Dionysias: (ostr. dating 306)
Grenfell et al. 1900, ostr. 21; (plan) Schwartz et al. 1969, plan II.
7. ~ el HalUibat: (inscr. dating 213) Prin. Arch. Exp. 1910, III, A,
no. 17 = CIL III 144192; (plan) ibid. 1909, II, A, 72. Bourada: (inscr.
dating 324-330) Guey 1939, 214-219; (plan) ibid., 193. Castra Martis:
(plan) Atanassova-Georgieva 1974, 169. Khan el Qattar: (plan) Poidebard
1934, pl. 39.

BIBLIOGRAPHY

Anthes, E., 1917. 'Spa.tr~mische Kastelle und feste sta.dte im Rhein- und
Donaugebiet', Ber. RGK 10, 87-167.
Atanassova-Georgieva, Iordanka, 1974. 'Le quadriburgium de la forteresse
Castra Martis en Dacia Ripensis~ In Actes du IXe Congres international
d'Etudes sur les Fronti~res romaines, ed. D. M. Pippidi; 167-172,
Bucarest.
Baradez, Jean, 1949. Fossatum Africae: Vue-aerienne de !'organisation
romaine dans le Sud-Algerien. Paris.
Beck, David, 1959. 'Das r~mische Kastelle von Schaan', in Limes-studien,
31-33, Easel. (Institut ftlr Ur- und Frfihgeschichte der Schweiz 14).
Bell, H., Martin, V., Turner, E. and van Berchem, D., 1962. The Abinnaeus
Archive, Oxford.
Be re hem, Denis van, 1952. L 'armee de Diocletien et la r.Horme constantinienne,
Paris.
Breeze, David J., 1977. 'The Garrisoning of Roman Fortlets 1 , in Haupt, D.
and Horn, H. G. (eds. ), Studien zu den Milita.rgrenzen Roms II, 1-6,
KOln:Bonn.
Brfinnow, R., and Domaszewski, A. von, 1904-1909. Die Provincia Arabia,
Strassburg.

1058
Carrie, Jean-Michel, 1974. 'Les Castra Dionysiados et l'evolution de l'ar-
chitecture militaire romaine tardive', MEFR 86, 824-850.
Fabricius, Ernst, 1926. 'Limes' in Paulys Real-Encyclopf!die der Classischen
Altertumswissenschaft 13, cols. 572-671.
Fellmann, Rudolf, 1979. 'Der Diokletianspalast von Split im Rahmen der
spf!tromischen Militarachitektur', Antike Welt 2, 47-55.
Gichon, Mordecai, 1967. 'The Origin of the Limes Palaestinae and the Major
Phases in its Development' in Studien zu den MiliUI.rgrenzen Roms, 175-193,
KlHn-Graz.
Goodchild, R., 1950. 'Limes Tripolitanus II', J. Roman Stud. 40, 30-38.
Goodchild, R. and Ward-Perkins, J. B., 1949. 'The Limes Tripolitanus in
the light of Recent Discoveries', J. Roman Stud. 39, 81-95.
Grenfell, B. P., Hunt, A. S. and Hogarth, D. G., 1900. Fayum Towns and
their Papy rii, London.
Gudea, N., 1974. 'Befestigungen am Banater Donau-Limes aus der Zeit der
Tetrarchie' in Actes de IXe Congres international d'Etudes sur les
Fronti~res romaines,ed. D. M. Pippidi, 173-180, Bucarest.

Guey, J. , 19 3 9. 'Note sur le Limes romaine de Numidia et le Sahara au

IVe siecle', MEFR 56, 178-248.
Johnson, Step hen, 1976. The Roman Forts of the Saxon Shore, London.
Kondic, V., 1969. 'Ravna' in Stare Kulture u Derdap. Belgrad, Srpska
Akademija Nauka i Umetnosti, 114-118.
Kornemann, Ernst, 1907. 'Die Neuste Limesforschungen (1900-1906 ),
Klio 7, 7 3-121.
Leschi, L., 1943. 'Le 'Centenarium' d'Aqua Viva', Rev. Afr. 395, 6-22.
Luttwak, Edward, 1976. The Grand Strategy of the Roman Empire from the
First Century A .D. to the Third, Baltimore.
MacMullon, Ramsay, 1963. Soldier & Civilian in the Later Roman Empire,
Cambridge, Mass.
Parker, S. T., 1976. 'Archaeological Survey of the Limes Arabicus: a
Preliminary Report', Annual of the Department of Antiquities of Jordan
21, 19-31.
Petrikovits, Harald von, 1971. 'Fortifications in the North-Western Roman
Empire from the Third to the Fifth Centuries A. D.', J. Roman Stud. 61,
178-217.
Poidebard, A., 1934. La Trace de Rome dans le desert de Syrie. Recherches
aerienne (1925-1932), Bibliotheque archeologique et historique 18, Paris.
Prin. Arch. Exp. 1907-1949. Princeton University Archaeological Expeditions
to Syria in 1904-5 and 1909, ed. H. C. Butler, Leiden.
Schleiermacher, W., 1951. 'Der obergermanische Limes und spatrl:Smische
Wehranlagen am Rhein', Ber. RGK 33, 133-184.

1059
Sch6nberger, H., 1969. 'The Roman Frontier in Germany: . an Archaeological
Survey', J. Roman Stud 59, 144-198.
Schwartz, J., 1950. Qasr-Qarun/Dionysias, Fouilles franco-suisses,
Rapports I, 1948, Cairo.'
Schwartz, J., Badawy, A., Smith, R. andWild, H., 1969. Qasr-Qarun/
Dionysias, Fouilles franco-suisses, Rapports II, 1950, Cairo.

1060
 73. MILITARY DIPLOMATA AND WAR EXPEDITIONS

Slobodan Dusanic

There are two kinds of military diplomata explicitly referring to war ex-
peditions in which their beneficiaries, or their beneficiaries' units, had
participated or were participating at the moment of the document's issue. In
a number of diplomata, that detail seems to have been mentioned only in order
to note that, at the time of their distribution, some of the units enumerated
were stationed outside the province to which the bulk of the troops listed
belonged. 1 The same practice of including, on one diploma, troops which,
even if administratively heterogeneous at the moment of the grant, had
nevertheless had something in common in their previous activity, sometimes
produced a converse formulation of the diplomata, in particular those issued
to the participants in an expeditionary corps constituted from the auxilia of
two (or more) provinces, after their return to their original provinces. 2 Now,
the documents listing auxiliary units, a part of whose personnel had just gone
to serve or was serving at another front, are apparently not thought to give'
privileges more extensive than the 'ordinary' ones. On the other hand, the
documents that refer to auxiliaries from two (or more) provinces having re-
turned from a distant battle may well have represented special grants, even
if their texts do not say so directly. As such, these latter obviously tend to
forn1 a transitional type leading to the other kind of diplomata dealt with in
our paper, i.e. the diplomata resulting from expeditions which brought those
who shared in the action certain extraordinary rights, cited or reflected in
the text m the certificate. As we shall see, the priority in receiving the cer-
tificates played an important role in the whole matter.
To begin with, it is useful to recall briefly these special privileges as
quoted in the diplomata recording, expressis verbis, the soldiers' participa-
tion in a war. Two variants can be noted at once: distinction in war entitled
a soldier, ante emerita stipendia, (a) to the diploma, and (b) to both the
diploma and the honesta missio. Leaving apart specific documents concerning
the Ravennates enlisted to form TI Adiutrix (CIL XVI 11, cf. 1 0), the second
variant appears only in a fragmentary fleet diploma of A. D. 71 (CIL XVI 17),
which refers, after the regular veterans, to the veterani [9H!J ante emerita
stipen[dia eo, quo]d se in expeditione belli fortiter industrieque gesserant,
exauctorati sunt. Whether originating from the circumstances of the Civil
War (Th. Mommsen; J. C. Mann) or, more likely, of the Jewish War (H.
Nesselhauf), this exceptional privilege does not seem to recur. 3 The use of
the former variant, obviously more acceptable to the army administration,
was next used, so far as our evidence goes, in Trajan's Dacian campaign,
when the equites and pedites of the coh. I Brittonum c. R. (CIL XVI 160, of
A. D. 106) were so rewarded, men of whom it is said pie et fideliter expedi-
tione Dacia functi. The same grant has been assumed, not implausibly (Forni

1061
1958, 15-25) in case of the diplomata of the Palmyreni Sagittarii (IDR I 5f. , 8f. ,
of A. D. 120 and 126), though these texts do not cite the qualification ante
emerita stipendia and are silent as to the contribution of the Sagittarii to the
victory over the barbarians. The omission of the ante emerita stipendia tends
to be neutralized by the omission of the formulae emeritis stipendiis and
dimissi honesta missione; on the strength of the examples which will be dis-
cussed later (notably CIL XVI 26 and 72), it may be surmised that the ante
emerita stipendia remained unengraved as the beneficiaries in question came
near to the term prescribed, so that, on this point, the privilege they obtained
was not very great. Perhaps the politico-military situation eventually caused
a delay in the distribution of their diplomata-a delay parallelled by three other
Dacian examples of the early second century4-and required a reformulation
of their constitutio that would leave out the (hypothetical) ante emerita stipendia
from the original version. What appears essential, however, for our purpose
is the two-fold resemblance between the diplomata of the Sagittarii and that
of the coh. I Brittonum: neither grants the conubium or extends the citizen-
ship to the recipients' children, and all of them pertain to one unit only.
These· coincidences admittedly suggest a special, and similar, grant in both
cases. The hiatus of almost ten years between the most recent document of
the Sagittarii series and the Danubian troubles of A. D. 117-9 provides no
strong argument to the contrary; there are analogous instances treated in
this paper, 5 which show that the military authorities, if not ready to award
the successful troops very substantial ahridgments of the term of service,
tended nevertheless to record their merits and acknowledge them in due time,
depending on the auxiliaries' age.
If the nature of IDR I 5f. and 8f. was really determined by the unit's mili-
tary prestige (no explicit reference is made to it), these documents allow us
to search for analogous implicit references to expeditiones belli in other
diplomata. To judge from the documents expressly mentioning such events,
and s'imilar diplomata of the Palmyreni Sagittarii, three criteria can be ap-
plied here. The diploma most likely to reflect its recipients' participation
in an expedition tends (a) to list a small number of auxilia, 6 (b) to list especi-
ally those types of unit suitable for the warfare in question-as a rule, the
alae and the cohortes equitatae were preferred for distant detachments 7-and
(c) to give its beneficiary a certain privilege as to the date of the bestowal of
the diploma.
In fact, at least one example satisfying all the three points may be found.
Opening the series of the diplomata of Type II in the AlftHdy (196 8b, 216,
226)-Mann (1972, 236f.) classification, CIL XVI 26 (of A. D. 80) was issued
iis qui militaverunt ... in alis quattuor et cohortibus decem et tribus ... , quae
sunt in Pannonia sub T. Atilio Rufo, quinis et vicenis pluribusve stipendiis
emeritis dimissis honesta missione, item iis qui militant in alis duabus I
civium Romanorum et II Arvacorum et cohorte vm Raetorum et sunt sub
eodem, emeritis quinis et vicenis stipendiis ... The 3 mounted regiments
cited after item, (VTII Raetorum was a cohors equitata , cf AE 1960, 375; A. D.
129), also occur in the main list of the document, among the alae quattuor
and the cohortes tredecim. The best explanation available to us of this
double occurrence is to assume that a vexillatio of I civium Romanorum, .!!_
Arvacorum and VTII Raetorum each served in an expedition and obtained,

1062
virtutis causa, their diplomas somewhat earlier8 than their coeval commilitones 9
who, though belonging to the same three and the fourteen other units enumera.:..
ted, were not given the chance to acquire that merit; note the contrast between
the militaverunt and the militant-the formula dimissi honesta missione was
naturally applied to the qui militaverunt alone-as well as the significant omis-
sion of the plurave stipendia in the case of those few who presumably in expedi-
tione belli fortiter industrieque se gesserant (to paraphrase CIL XVI 17). There
are actually indications that some vexillationes of the Pannonian (and Lower
German) army took part in the so-called Clemensfeldzug in Upper Germany
£.· A. D. 73-74,10 probably under the two Dornitii as praefecti auxiliorum
omnium adversus Germanos. 11 The auxilia of the Domitii Will have included,
among other units, the Upper German cohort III Gallorum that before long
joined the exercitus Moesiae, a circumstance which helps us understand the
special place of Claudia nova, III Gallorum and V Hispanorum within CIL XVI
28 of A. D. 82, 1 2 and the Pannonian ala II (Hispanorum) Arvacorum-a circums-
tance which supports our interpretation of the appendix to the list of CIL XVI
26.13 The details of chronology remain rather obscure-the war in about 73-
74, the Pannonian grant in 80, the Moesian one in 82-and probably attest,
among other things, the (slight) differences of age between the bulk of the
auxiliaries missi in expeditionem from I civium Romanorum-II Arvacorum-
VIII Raetorum and Claudia nova-III Gallorum- V Hispanorum respectively. As
regards point (c), our suggestion that the privilege of the soldiers referred to
in the appendix of the document of A. D. 80 consisted in receiving their cer-
tificates (shortly) before their honesta missio-which implies that they were
not obliged to wait for their discharge longer than their comrades from the
same list, and tends to confirm the independent importance of the grant of the
diplomata in comparison with the discharge 1 4-finds a corroboration in an
analogous text. The constitution CIL XVI 72, passed in A. D. 127 to men of
the Ravenna fleet, revives "over 20 years after the last known use of that
formula" for the classiarii, the present tense iis qui militant (... qui sena et
vicena stipendia meruer[u]nt),but names, in its only preserved copy, a veteran,
not a serving sailor (Mann 1972, 235). The two irregularities, so far unex-
plained15, seem to reflect a special grant for the crews of the ships distinguished
for industria (to allude once again to the wording of CIL XVI 17 just quoted)
during Hadrian's great tour of A. D. 121-126. 16 Like the recipients of CIL
XVI 26 and, to a degree, those of the diplomata of the Palmyreni Sagittarii
(the privilege of the latter will have been somewhat more marked), the re-
cipients of CIL XVI 72 must have come so close to their honesta missio that
the contradiction between the present-tense formula and the grant to a veteran
was thought tolerable in the official act; perhaps the contradiction merely
shows that the time necessary for correspondence between Ravenna and Rome,
and for engraving the diplomata, sufficed in this case for our beneficiary to
pass from an emeritus to a missus honesta missione. 17
The rather modest priority of the privileged recipients of CIL XVI 26, 72
and, generally, of the similar diplomata classified under the AlftHdy-Mann
Type II, raises the question whether a missic in expeditionem had influenced
some grants which seem to imply no priority in the time of the bestowal at
all, a category including i· g. the whole of the Alft5ldy-Mann Type III. There
have already been such suggestions, especially about the documents which

1063
fully satisfy our criteria (a) and (b). 1 8 An example, which, in my opinion,
proves this point has recently been discovered: the soldiers of no more than
three Upper German mounted cohorts were given in A. D. 65 a Type I diploma
which for several reasons indicates a special grant for their participation,
together with certain alae and the leg. 1111 Scythica, in Nero's Eastern cam-
paign, £.· A. D. 57-64 (Dusanic 1978, 469-75). Thorough analysis of this docu-
ment is impossible in the present paper, short as it must be, but a brief
word should be said on wider problems.
One thing must be stressed first. If the above interpretation of CIL XVI 26
and 72, and IDR I 5-9, is accepted, many transitional cases may be adduced
linking the ' normal" grants to our "crypto-special" grants, with regard to
both the three criteria1 9 spoken of here and the formulation of the diplomata,
notably its evolution from the AlftHdy-Mann Type II to Type Ill. 20 It is dif-
ficult to take this assimilation of the 'normal' and the 'crypto-special' class
simply as a sign that the presumed participation in a war expedition need not
have always brought the participants certain extraordinary privileges, and
was implicitly noted only for the sake of practical book-keeping. 2l Such a
hypothesis would i. a. run contrary to the fact that (a) diplomata listing all the
units of a province are virtually unknown, 22 and (b) that there exists a category,
enigmatic it is true, of the veterans xwp l <; xa"-xwv (Kraft 1951, 132-8; Degrassi,
1971, 265). It seems that even the recipients among the dimissi honesta mis-
sione quinis (senis) et vicenis pluribusve stipendiis emeritis were specially
privileged in comparison with some of their veteran comrades. Their privi-
lege may have consisted, again, in receiving the diplomata with a shorter
delay after discharge than that 23 imposed on missicii 24 ~erhaps labelled
o\JE'tpavol. oL ywp't< xat..xwv ot vuv in CIL XVI App. 525). Much how-
ever might be said in favour of the old hypothesis (Domaszewski 1967, 1, 75
n. 2) that virtually every military diploma represented a special grant ob vir-
tutem, denied for ever to the (rather numerous ?) veterani sine aeribus. 26
Parallel to the gradual diminution and standardisation of the term of military
service, the development of the auxiliary and fleet diplomata as a genre cer-
tainly attests a tendency for them to become less and less exclusive a re-
ward-even the discrimination between the serving soldiers and the missi
honesta missione in the AlftHdy-Mann Type 11 may sometimes have been de-
termined by the soldiers' merits in peace rather than war. All the same, we
should never underestimate the inclination of the imperial administration to
make exceptions :in order to promote the efficiency of the exercitus Romanus. 27

1064
 NOTES

1. AE 1968, 513 (A. D. 105) recording cohorts I Hispanorum and I Thebae-

orum, 'extranslatae in Iudaeam'; the transfer is to be connected with the
Arabian campaign (H. -G. Pi1aum) or, perhaps, with some troubles in
Palestine (cf. Syme 1958, 222 n. 5); CIL XVI 61 (A. D. 114) recording
ala I F1avia Aug. Britannic. o0 c. R. missa in expeditionem (probably the
Parthian War, cf. Nesselhauf 1936, ad loc. ); on CIL XVI 28 see below,
nn. 12, 21. The recipients of the diplomata need not have departed with
their younger comrades but obviously remained, in the expectation of
diplomata, under their former command (cf. Nesselhauf 1936, ad CIL XVI
28; below, n. 5).
2. CIL XVI 99 (A. D. 150); cf. IDR I 7 with 7a (A. D. 123). The latter list
reveals that the regiments concerned had fought together in Trajan 1 s
Parthian War (not in some Dacian operations of A. D. 117-118 as I. I.
Russu thought); cf. Dusanic and Vasic 197 4, 425. The distance in time
(1) between the War and the honesta missio (per Marcium Turbonem), and
(2) between the honesta missio and the grant (more than three years at
any event, cf. e. g. Reynolds 1976, 191), presents no obstacle to identifica-
tion with the campaign against Parthia-we may note i. a. the shortage of
troops under Trajan and in the early part of Hadrian's reign (cf. Dusanic
1977, 182 n. 27)-and finds a parallel (for (2) ) in CIL XVI 160 (the grant
fell formally in A. D. 106, but in practical terms in A. D. 110) or in the
diplomata (IDR I 5f. , 8f. ; CIL XVI 26, 28) (see infra) . Obviously, the
veterans discharged by the first two governors of Dacia were retained in
service for the expedition in the East, and some of them even longer, as
voluntarii (AE 1969-70, 583: the end of the inscription should read missus
voluntarius honesta missione a Terent[io Scau]riano consulare [exerci]tus
provinciae nov[ae Daciae ... ], instead of the editor's provinciae nov[ae
?Mesopotamiae ... ], for several reasons, not the least the date of the
legionary's enrolment- "sometime before A. D. 85"-which postulates a
date of the honesta missio within Scaurianus' Dacian, not Eastern,
command) or missicii, if there was any tangible difference between these
categories.
3. Kraft 1951, 134, thinks of an honourable missio causaria.
4. CIL XVI 160; IDR I 7+7a (see above, n. 2).
5. See infra, on CIL XVI 26 and 28.
6. CIL XVI 160 (one ala), 60 (one ship), 61 (ala I F1avia Aug. Britannic.);
IDR I 5f. , -8f. -(one numerus sagittariorum); CIL XVI 114 (?one
mounted numerus); AE 1968, 513 (two cohortes equitatae-I Hispanorum
and I Thebaeorum); CIL XVI 26 (two alae-I civium Romanorum and II
Arvacorum, and a cohors equitata-VIll Raetorum), 99 (five alae from
two provinces); IDR I 7+7a (two alae and ?one cohor s equitata), etc.

1065
 -
7. See the preceding note; IDR I 7+7a have been connected with ILS 2732,
which mentions a praef. vexillation. eq. Moesiae Infer. et Daciae
(Dusanic and Vasic 1974, 413 n. 38, 425) ..
8. Perhaps approximately a year, to judge from the largely analogous case
of.CIL XVI 30 (Sept. 3, 84) and 31 (Sept. 5, 85), on which cf. AlftHdy,
1968b, 226 n. 30.
9. Mann' s comment (197 2, 236) on the formulation of the diploma rightly
draws attention to the emeritis qu in is et vicenis stipendiis (as contrast-
ing with the quinis et vicenis p 1 u rib us v e stipendiis from the main text)
but neglects the fact that these words qualify the service of soldiers from
the three units only.
10. On that expedition, cf. Lieb 1967, 96f. , with the sources and bibliography.
11. ILS 990f., cf. Merlin 1944, 527f. The dating of the two praefecturae has
been controversial: A. D. 70 (Al:ftlldy), 73-7 4 (Bormann, Zangemeister),
or 77-78 (Groag), cf. AlftHdy 1968a, 131-135. I prefer the middle solu-
tion, which relates them to Clemens' expedition.
12. All three figure in the Upper German diploma of May 21, A. D. 7 4 (CIL
XVI 20), which presumably marks a major success of Clemens against
the Germans. It is beyond doubt that at least III Gallorum came from the
Rhine to the Danube before the Moesian diploma of April 28, A. D. 75,
was issued (Mirkovic 1968, 177), a fact fatal to the conjecture that the
occurrence of this and the other two regiments in the item clause of the
list on CIL XVI 28 was due to a recent transfer (so the scholars cited by
Mirkovic 1968, 179 n. 18, who excludes, with good reason, the alterna-
tive of two homonymous cohorts III Gallorum. For a different view, cf.
Visy 1978, 37ff. ). The correct solution had been surmised by Wagner
(1938' 138 ).
13. A fragment at Augusta Rauricorum (Lieb 1968, 129-132 = AE 1971, 277)
records an anonymous prefect of the ala Moesica torquata who at the same
time commanded, as a praefectus or a praepositus, a vexillatio of an ala
Hispanorum. Though the post has usually been dated to the epoch (before
the middle of the first century A. D.) during which both the alae seem to
have belonged to the exercitus Germaniae &lperioris (cf. AlftHdy 1968a,
24), a date .9.· A. D. 73-74, and an extraordinary command uniting the two
cavalry units detached from other provinces, seem preferable for the
following reasons. (1) The ala Hispanorum was present not in full
strength, but in part only, which would well accord with an expeditionary
corps analogous to the vexillationes equitum Moesiae Inferioris et Daciae
·sent to the East by Trajan (ILS 2732, cf. above, n. 7)-the participation
of the whole of the ala Moesica would be comprehensible in that case as
this unit had a shorter distance to come (from neighbouring Germania
Inferior), whereas IT Hispa.~orum had been stationed in the very south
of Pannonia (Wagner 1938, 47f. ). (2) The presence of the combined regi-
ment at Augst is perfectly explicable Q. A. D. 73-7 4, when that place
served as something of a base for the war (Lieb 1967, 96f. ) , less so in
pre-Flavian times (cf. Lieb 1968, 132: "Einen brauchbaren Grund, irgend.
in dieser Zeit Krafte nach Augst zu legen, sehe ich nicht ... "). Be it noted

1066
 that the ala II Hispanorum Arvacorum (epigraphically, both the ethnics
may occur, or only one of them), not only I Hispanorum Arvacorum, is
cited in some early inscriptions without the numeral (e. g. in CIL III 3286,
Mursa).
14. Cf. (e. g. ) Forni 1958, 17 n. 38 (on ILS 8888 and the like).
15. Nesselhauf (1936, ad loc.) treats the form of the recipient's rank (partially
preserved: ex [---]; a different reading, like .ex[acto], is improbable)
as a slip (which "eo facilius irrepere potuit, quod iam per annos viginti
solis exequitibus et expeditibus auxiliorum constitutiones dari solebant");
Mann 1972, 235, adds that "nothing in the formula suggests that this was
a special grant".
16. The emperor's satisfaction with his fleets found expression also in the
reverses of his coins, with the representation of a warship and the inscrip-
tion Felicitas Augusti (cf. Kienast 1966, 76 n. 100: "ausgesprochene
Anlassprltgungen" ).
17. The interval may have been shorter than two months or so in some cases
(Degrassi 1962, 62f. ).
18. CIL XVI 60 (with the editor's note; Forni 1958, 16, allows the possibility
of a grant ante emerita stipendia here too); IDR I 7+7a (above, n. 2). Cf.
CIL XVI 23 (AI:ruldy 1968a, 155f.) and the documents referred to infra,
n.19. -
19. For instance, in an analysis of the diplomata of the later second century
issued for Pannonia Inferior, Fitz (1959, 440) has arrived at the conclusion
that 'nur die Veteran en der krutlpfenden Truppen entlassen wurden 1 •
20. The preponderance of veterans among the recipients after A. D. 106 (Mann
1972, 237) and the omission of the aut plura/plurave after A. D. 117
(Alfflldy 1968b, 224) are also characteristic in this respect.
21. Which might have demanded, at the moment of the grant, the grouping of
units whose emeriti formed an entity some time before, a grouping which
would be parallel to that in AE 1968, 513 and CIL XVI 61 (cf. above, text
and nn. 1-2). But even CIL XVI 28, which comes near, in that respect,
to AE 1968, 513 and CIL XVI 61, seems to represent a special grant re-
flecting the beneficiaries' merits in war (above, n. 12). On the other
hand, the practice of uniting the auxilia of two provinces, if linked together
through a period of common service, was not followed constantly. See,
for an exception, CIL XVI 67 (A. D. 120): the document does not unite
the veterans of the Macedonian regiment with their ex-comrades from
Moesia Superior, the province in which the former served £: A. D. 100
(cf. Nesselhauf 1936, ad loc. , nn. 1, 5). The date of the transfer from
Moesia to Macedonia is uncertain, probably because the former and the
latter did not participate in a sufficiently important common enterprise.
22. For some references on that problem, Dusanic and Vasic 1974, 419
85-89; Dusanic 1978, 469 nn. 71f. We must also note the absence of the
diplomata issued to troops not employed on the limes (e. g. the auxiliaries
permanently stationed in Spain, Gallia, Dalmatia, Macedonia, Thrace,

1067
 Asia Minor, etc.; the documents CIL XVI 34, 38, 40, 67, 128 and the
like represent only apparent exceptions.
23. The possibility of such delays (beside the 'technical' ones, cf. supra, n.
17) should be reckoned with when explaining the occurrence of the names
of two legates on CIL XVI 43 (cf. Nesselhauf 1936, ad loc.; CIL XVI
SuppL, p. 215) and 69, as well as the chronological discrepancies between
the imperial titles and the dating formulae in many other diplomata.
24. On them see e.g. Domaszewski 1967, 78 n. 11 (+ add.); cf. 8peidel 197 0,
151 with nn. 101-104; above, n. 2.
25. The reading 01, vuv and the interpretation of the entire phrase, are
admittedly uncertain (cf. (e. g.) Kraft 1951, 132). I follow Wilcken and
Lesquier in that matter .
26. Otherwise, it would not be easy to understand i. a. why several diplomata-
which, in all likelihood, reward the units distinguished for bellica virtus-
themselves attest to substantial delays of the grants (above, nn. 2, 23).
27. Cf. H. Wolff's just observations (1974, 479-510; esp. 510, on the select-
ive 'Vergtlnstigungen' which do not answer wholly or permanently the 'ob-
jective BedUrfnisse der Soldaten') concerning the decurions' and centurions'
privileges cited in AE 1960, 103 and CIL XVI 132.

BIBLIOGRAPHY

Al:ftildy, G. , 1968a. Die Hilfstruppen der rtsmischen Provinz Germania

Inferior, Dusseldorf.
AlftHdy, G. , 1968b. 'Zur Beurteilung der Milit9.rdiplome der Auxiliarsoldaten',
Historia xvii, 215-227.
Degrassi, A., 1962, 1971. Scritti vari di Antichita, Vol. 1 (1962), IV (1971),
Venezia-Trieste.
Domaszewski, A. von, 1967. Die Rangordnung des r~mischen Heeres. 2nd
edn. revised by Dobson, B. , K~ln-Graz.
Dusanic, S. , 1977. 'Two Epigraphical Notes' (Serbian), Ziva Antika xxvii
179-190.
Dusanic, S. , 1978. 'A Military Diploma of A. D. 65', Germania lvi, 461-475.

Dusanic, s. and Vasic, M. R., 1974. 'Fragment of a Military Diploma from

Moesia Superior', Germania lii, 408-425.
Fitz, J. , 1959. 'Die Militardiplome der Auxiliartruppen aus Pannonia Inferior
in der zweiten ffiUfte des 2. Jahrhunderts', Acta Antig. Hung. vii 421-442.
Forni, G. , 1958. 'Contributo alla storia della Dacia Romana', Athenaeum
xxxvi, 3-29.
ILS, (!nscriptiones Latinae Selectae ed. Dessau, H., 1892-1916, Berlin.

1068
IDR Inscriptiile Daciei Romane, ed. Russu, I. , Vol. I, Bucharest.
Kienast, D., 1966. Untersuchungen zu den Kriegsflotten der r~mischen
Kaiserzeit, Bonn.
Kraft, K. , 1951. Zur Rekrutierung der Alen und Kohorten an Rhein und Donau,
Bern.
Lieb, H., 1967. 'Zum Clemensfeldzug' in Studien zu den MiliUlrgrenzen Roms,
Koln-Graz, 94-97.
Lieb, H., 1968. 'Truppen in Augst', in Schmid, E., Berger, L. and
Buergin, P. (edd.) Provincialia (Festschrift fttr Rudolf Laur-Belart),
Basel-Stuttgart, 129-132.
Mann, J. C., 1972. 'The Development of Auxiliary and Fleet Diplomas',
Epigraphische Studien ix, 233-241.
Merlin, A. , 1944. Inscriptions Latines de la Tunisie, Paris.
Mirkovic, M. , 1968. 'Die Auxiliareinheiten in M~sien unter den Flaviern',
Epigraphische Studien v, 177-183.
Nesselhauf, H., 1936. Corpus Inscriptionum Latinarum Vol. XVI: Diplomata
Militaria, Berlin.
Reynolds, J. M., 1976. 'Roman Inscriptions, 1970-75', J. Roman stud.
lxvi, 174-199.
Speidel M., 1970. 'The Captor of Decebalus, a new Inscription from Philippi 1 ,
J. Roman Stud. lx, 142-153.
Syme, R., 1958\ Tacitus, Oxford.
Wagner, W. , 1938. Die Dislokation der r~mischen Auxiliarformationen in
• den Provinzen Noricum, Pannonien, Moesien und Dakien von Augustus bis
Gallienus, Berlin.
Visy, Z. , 1978. 1Der Beginn der Donau- Kriege des I>omitian 1 , Acta Arch.
Hung. xxx, 37-60.
Wolff, H. , 197 4. 1 Zu de:n BUrgerrechtsverleihungen an Kind er von Auxiliaren
und Legionaren', Chiron iv, 479-510.

1069
 74. BRITISH AUXILIARY REGIMENT8-0RIGINS AND
EARLY NOMENCLATURE

D. B. Saddington

The origins of the auxiliary regiments of the early principate are wrapped
in obscurity (Saddington 1975). Those of the British regiments might seem
easier to trace in view of the fact that there is a fixed terminus post~-
A. D. 43. The earlier British units may now be listed as a preliminary to a
discussion of their origins and developing nomenclature.
The literary and epigraphical evidence other than that of the diplomas may
be considered first. Initially no distinction will be made between Britannica,
Britannorum and Brittonum in the titulatures. A Coh. or Cohh. Britannorum
were involved in the Vitellian attack on North Italy in 69 (Tac. Hist. I, 70).
Later in the same year an Ala Britannica was in Italy (Hist. m, 41). Agricola
used "Britanni" from_ the long pacified South at the battle of Mons Graupius in
84 (furr. 29). The Coh. m Britannorum on an Italian stele can probably be
dated to 69 (Dessau 1916, 2560). 1 It was described as being part of the
"exercitus Raeticus". The deceased recorded on the stone was only 25, with
6 stipendia, which means that the regiment cannot be dated before 63. When
it was raised is not clear, but it must have been transferred from Britain at
some stage under Nero for it to have formed part of the Raetian army by 69.
The inscription T. F. Draccus eges alae I F. D. Brit. m. c. R. ciuis Sequanus
an. XXXXV stupendiorum XXII may be included as the next datable evidence.
The abbreviations can be expanded from another inscription from Pannonia
Superior which reads: Ca(e)lius Saco(nis) f. an. XXX h. s. e. milis ala o0
Flauia Domitiana ciui. Romana Britanica (Dessau 1916, 9140; Kraft 1951, no.
202)2 (which apparently lacks the numeral I). Since "D" must represent
"Domitiana", T. Flavius Draccus must have acquired his citizenship under
Domitian. The bestowal of the imperial name and the grant of citizenship
were probably made during one of Domitian's Danubian Wars, that is, in either
86 or 92/3. The inscription comes from Vindobona (Vienna), where the regi-
ment is recorded on the tombstones of 2 other T. Flavii, but with Augusta
replacing Domitiana, presumably due to the damnatio memoriae of Domitian.
The Ala I Flavia Augusta Britannica C. R. can be brought down to 74, if
Draccus died in the last year of Domitian' s principate. Should he have died
in either of the wars mentioned above, it can be brought down as early as 70
or even 64. Presumably it was Milliaria from the outset. Whether it owed
its title Flavia to Domitian or to Vespasian is not clear. The inscription
"---JVirssuccius [E ?] si eq. imag. coh. I Brit. tur. Monta. ann XXXV sti.
XV h. s. e. Bodiccius imag. et Albanus h. p." gives the appearance of being
early and may seem to record a Coh. I Britannorum (Dessau 1916, 2581;
Kraft 1951, no. 1231; Wagner 1938, 104). 3 The names Bodiccius and Virssuccius
are Celtic, if not actually British. But since the regiment comes from Pannonia,

1071
where a Coh. I Britannica is recorded in 80 and 84, Britannica must be the
correct expansion of "Brit. ". Presumably the soldiers concerned were Britons
drafted into a Coh. I Britannica that was transferred to Pannonia in the Flavian
period. A Coh. I Fl. Brittonum (CIL III 2024; Kraft 1951, no. 1261) occurs
in Dalmatia possibly in the Flavian period. 4 In the Flavian period tiles were
fired in Lower Germany by a Coh. II Brit. (at Vetera- Xanten) and Coh. II Br.
M. E. (at Fectio - Vechten) and a Coh. VI Br. (at Vetera). These two regi- ·
ments were Brittonum (CIL xm 12424. 1-2, 12425, 12423; Alfl:Hdy 1968, 195). 5
Early British regiments known from the diplomas are entitled as follows:

Regiment Year Reference6

Ala I Britannica C. R. 7 110 163

Ala I Flavia Britanniciana Milliaria C. R. 102 47
(Ala I Flavia Aug. Britannica ooC. R. 114 61)
Coh. I Britannica 80 & 84 26 & 30
(Coh. I Britannica Milliaria 85 31
Coh. I Britannica oOC. R. 105 49)
Coh. II Britannor.ao C.R. P. F. 110 163
Coh. Ill Britannorum 107 55
Ala I Brittonum C. R. 8 123 t-21
Coh. I Brittonum Milliaria 85 31
(Coh. I Brittonum Milliaria Ulpia Torguata P.F.C.R. 106 160
Coh. I Brittonum ooUlpia Torquata C. R. 110 163)
Coh. II Brittonum 107 56
Coh. II Brittonum ooC. R. P. F. lOO 46
Coh. II Flavia Brittonum 99 45
Coh. Ill Brittonum 100 46

If the minimum period of service of 25 years be subtracted from the dates

given, the Coh. I Britannica can be brought down to 55 and the Coh. I Brittonum
Milliaria to 60. The other regiments remain Flavian. The lengthening titula-
tures and the use of the sign oo for Milliaria on the later examples should be
noted. The Ala I Flavia (Augusta) Britannica o0 C. R. of 102, 110 and 114 also
occurred on the stelae discussed above (Britanniciana in 102 does not seem to
be a significant variant). The Coh. I Britannica has been discussed above, as
well as the Coh. Ill Britannorum, which must have returned to Raetia after the
civil wars of 69, since that is where the diplomaof 107 locates it.
The nomenclature of the British regiments causes difficulties, especially
when abbreviations occur. There are 3 categories. Firstly, there is an Ala I
and a Coh. I Britannica. Secondly, there is a series of Cohh. Britannorum.
Those numbered II and Ill have been cited: a Coh. I Britannorum appears in
164. 9 The third and largest group of cohorts is that of the Brittonum whose
highest rum er al is VI. There would also appear to be some duplication of
units: the Cohh. I and II Flavia Brittonum of Dalmatia and Moesia Inferior

1072
respectively must be different to the polyonymous Cohh. I and II Brittonum of
Germany and the Danubian front. Besides the Cohh. , there was an Ala I Brit-
tonum C. R. which may go back to 123, as noted. As far as British soldiers
serving in these regiments are concerned, 2 almost certain examples were
noted in the Coh. I Britannica in Pannonia: another is recorded on the diploma
of Moesia SUperior in 105.10 Twu Britons received the diplomas of the Coh. I
Brittonum Milliaria Ulpia Torquata P. F. C. R. of Dacia issued in 106 and 110.
Accordingly Britons occlir in 2 of the 3 categories: there is no reason why they
should not be found in the Cohh. Britannorum.
Cheesman (1914, 171, n. 1) has remarked that "it does not seem possible
to make any distinction, chronological or otherwise, between the titles
Britannica, Britannorum, and Brittonum". 11 But the title Britannica wuuld
normally mean "coming from" or "stationed in Britain", not "consisting of
Britons". This is clearly shown by Tacitus calling the Batavian cohorts that
had served in Britain Britannica auxilia and by his assignation of the term
Britannica to legions. 12 Fabricius has distinguished the Brittones as "die zu
Beginn der Flavierzeit noch unabh1tngigen Briten und die in den folgenden
Dezennien aus den Neuunterwurfenen gebildeten Truppen, in Gegensatz zu den
Britanni, den frUheren r~mischen Untertanen der Insel, und den alae und
cohortes Britannicae, den aus den britannischen Provinzialen ausgehobenen
ttlteren Auxilien" (see stein 1932, 245; largely accepted since then). But as
noted, the Coh. I Brittonum Milliaria goes back to 60, just before the revolt
of Boudicca under Nero, and the other Cohh. Brittonum listed above go back
to 75-8, i.e., before the gTeat thrust north in the middle Flavian period.
Therefore Fabricius' contrast between Brittones and Britanni does not appear
tenable. There seems no reason for interpreting the term Britannica in the
senses proposed by either Cheesman or Fabricius. If Fabricius' "Provinzalien"
means "Roman citizens in Britain", the names quoted above do not bear his
contention out. It is true that Britons appear to have been recruited into the
Coh. I Britannica before it was transferred to Pannonia, but it was usual for
the peregrine inhabitants of a province to be drafted into any regiment in the
province. The Ala I Britannica is significantly named the Britanniciana on
the first diploma in which it is mentioned: this termination for units stationed
in a province, both legionary and auxiliary, is found even in the literary re-
cord.13 There also appear to have been only one Ala and one Coh. Britannica
as distinct from the many regiments Britannorum arrl Brittonum. There
seems no reason for not preserving the traditional meaning in their case. As
regards the ethnica Britannorum and Brittonum, at present no real distinction
can be traced. The 2nd century diplomas reveal a disconcerting inconsistency
in spelling, as do other epigTaphical documents, such as the Trajanic dedica-
tion on which the Coh. VI Britt(onum) Eg. P. F. is translated as ans CpT)c;
~· Bpt.'ta:vvt.xTic; l.nnt.xTic; E~ae:~ouc; nt.a'tt.xfic; (AE 1972, 573).
The evidence for British auxiliaries cannot be dated earlier than the
principate of Nero. But already by then a regiment had been transferred to
Raetia. The dilectus (Tac. ~- 15) was operating in Britain by the revolt of
Boudicca: it seems likely that British units were raised soon after the invasion
under Claudius but that the vagaries of the literary record and the fact that the
habit cf erecting stelae was not adopted immediately have resulted in a lacuna
in the evidence.

1073
 NOTES

1. For problems in dating auxiliary inscriptions, cf. Saddington 1973. I

hope to present elsewhere the detailed arguments for the dates proposed
in the cases discussed.
2. The dating and interpretation here is based on Wagner 1938, 20ff. Cf.
AE 1940, 5 = Kraft 1951, no. 201; CIL III 4575 = Kraft 1951, no. 203
which reads T. Fl. Bardi ueter. ala-;! Fl. Aug. Brit.oo c. R. ~ and CIL
III 4576 = Kraft 1951, no. 204 which reads T. F. Verecund. Mag[ .•. J
eques alae I Fla. Aug. Brit.oo C. R. turma Italici an. XXXX s. XIX h. s.
es. Pro(culus ?) et Priscinus uex(illarii) et Ingenus hered ..
3. Cf. Kraft 1951 on the names; for the reading "Esi", cf. Dusanic 1967,
69.
4. For the date, cf. Al:Wldy 1962, 267.
5. Cf. the discussions f?Y Bogaers 1965 18; 1967 65; 1969 33ff., 42ff.
6. The figures refer to humbers in CIL XVI: figures preceded by t-
Roxan 1978.
7. Nesselhauf (1936, 163 n. 2) has identified the Ala I Britannica C. R. (in
Dacia) with the Ala I Flavia Augusta Britannica Milliaria C. R. listed
below of 110 and 114. But, as Radnoti and Barkoczi pointed out (1951,
195) they cannot be the same, as the latter regiment is recorded in
Pannonia on the same day in 110 as the former appears in Dacia.
Although expressing doubts, Nesselhauf (1936, 61 n. 4) and they suggest
that the latter regiment may be the same as that recorded on CIL m
6748, which appears to read ---al]a I Flauia [---I ---Au]g. Britton[--- I
-t--] c. R. his to[r lq]uata ob u[irtu/te]m. However if "Britton." is cor-
rect, this should probably be identified with one of the British alae on
Roxan 1978, 21 of A. D. 123: the first, the ala I Brit[ton. ? c. R.] (dis-
cussed below), was in Dacia, the second, the [ala] I Britann. c. R. (spelt
ala Briton. c. R. on the outer face of the diploma), was in Pannonia
Inferior.
8. The name depends on an emendation by ~rincz 1977, 281ff. adopted in
Roxan 1978, 21. The emendation gains support from CIL XVI 179180 of
A. D. 148 which record an Ala I Brittonum C. R. (cf. XVI 17 5 of A. D.
139 and 11213 of A. D. 160 as emended in CIL XVI Suppl. p. 216).
9. Roxan 1978, 64 of Dacia, recording the regiment as I Britannor. Equit.
However, it is not certain that this was the correct title. On other Dacian
diplomas of the same date (Roxan 1978, 63; CIL XVI 185) the ethnicum is
abbreviated in such a way that it could be expanded to either Britannica
or Britannorum. Nesselhauf (1936, 185) (without of course knowing
Roxan 1978, 63 and 64) equated the regiment with the Coh. I Britannicaoo
C. R. of Dacia, known from CIL XVI 163 of 110 (cf. Roxan 1978, 21 of
A. D. 123 and 35 of A. D. 133; previously the unit was in Pannonia in 80
and 84, entitled I Britannica, and in 85, entitled I Britannica Milliaria,
and in Moesia Superior in 10317 and 105 as I Britannicaoo C. R .. ). How-
ever, in view of the absence ofoO and the presence of Equitata in the

1074
 titulature of the regiment in Dacia in A. D. 164, it is probable that we
are dealing with a different regiment, i.e., a Coh. I Britannorum
Equitata as distinct from a Coh. I Britannica oo C. R.
It is not clear what the number of the Hadrianic Coh. [---] Britannorum
of Dacia Porolissensis (Roxan 1978, 31) was, but presumably it wasj.
10. Kennedy 1977, 253f., seems over-cautious in suggesting that only Lucco's
mother, not Lucco himself, was British.
11. The Thesaurus Linguae Latinae and the Oxford Latin Dictionary do not
distinguish the senses.
12. Tac. Hist. IV.15. 1; IT. lOO, 1; cf. the uex(illaiio) Brit(annica) of the
Flavian inscription Dessau 1916, 2515. Kennedy 1977, 251 accepts this
interpretation of the word.
13. Cf. Suet. Gal. 20, 1. For the form Britannicianus, cf. the pedites
singularesfuitanniciani of CIL XVI 54 of A. D. 103/7. One will agree
with Davies 1976, 144, that the word means "drawn from the army of
Britain", but not with the contrast he suggests with Britannicus.

BIBLIOGRAPHY

AlftHdy, G., 1962, 'Die Auxilartruppen der Provinz Dalmatien', Acta Arch.
Hung. xiv, 259-296.
AlfOldy, G. , 1968, Die Hilfstruppen der rOmischen Provinz Germania inferior,
DUsseldorf.
Bogaers, J. E., 1965, 'Romeins Nijmegen', Numaga xii, 1Q-37.
Bogaers, J. E. , 1967, 'Die Besatzungstruppen des Legionslager von Nijmegen
im 2. Jahrhundert nach Christus', in studien zu den Milit:trgrenzen
Roms, KOln-Graz, 54-76.
Bogaers, J. E., 1969. 'Cohortes Breucorum', Bericht R. 0. B. xix, 27-50.
Cheesman, G. L., 1914, The Auxilia of the Roman Imperial Army, Oxford.
Davies, R. W. , 1976, 'Singulares and Roman Britain', Britannia vii, 134-144.
Dessau, H. , 1916, Inscriptiones Latinae Selectae, 4 vols. 1892-1916, Berlin.
Dusanic, s., 1967, 'Bassianae and its Territory', Arch. rug. viii, 57-81.
Kennedy, D. , 1977, 'The ala I and cohor s I Britannica', Britannia viii, 249-
255.
Kraft, K., 1951, Zur Rekrutierung der Alen und Kohorten an Rhein und Donau,
Bern.
LOrincz, B. , 1977, 'Zur Ergnnzung und Datierung dazischer Militttrdiplome',
Acta Arch. Hung. xxix, 281-5.
Nesselhauf, H., 1936, Corpus Inscriptionum Latinarum. Vol. XVI: Diplomata
Militaria, Berlin.

1075
Radnoti, A. and Barkoczi, L. , 1951, 'The Distribution of Troops in Pannonia
Inferior during the 2nd Century A. D. ', Acta Arch. Hung. i, 191-226.
------- -

Roxan, M. M., 1978, Roman Military Diplomas 1954-1977, London.

Saddington, D. B., 1973, 'Towards the Dating of Early Inscriptions Recording
Roman Auxiliary Regiments', Vestigia xvii (Akten des VI. Internationalen
Kongresses fltr Griechische und Lateinische Epigrafik, 1972), 538-540.
Saddington, D. B., 1975. 'The Development of the Roman Auxiliary Forces
from Augustus to Trajan', in Temporini, H. (ed. ), Aufstieg und Nieder-
gang der rDmischen Welt, Berlin-New York, ll. iii, 176-201.
stein, E., 1952, Die kaiserliche Beamten und TruppenkOrper im rDmischen
Deutschland, Vienna.
Wagner, W. , 1938, Die Dislokation der rOmischen Auxiliarformationen in den
Pr~vinzen Noricum, Pannonien, Moesien und Dakien, Berlin.

1076
 tt
75. FU.NDE VON ROMISCHEN MEDAILI.DNS ZU BEIDEN BEITEN
DES RHEIN-DONAU-LIMES

Jerzy Wielowiejski

Die vergleichende Forschung lber die Verteilung r8-

misoher FundmUnzen zu beiden Seiten des Dona~Limes
(W1elowiejsk1 1973 und 1977) inspirierten auch zu Unter-
suchungen Uber die Verbreitung der Funde von Medaillons.
Dieses Problem war bereits vor 10 Jahren Gegenstand der
Betrachtungen ia Rahmen des ramisohen laisertums UDd
des n8rdl1chen Barbariwa.s (Wielowiejski 1970). Jfan hat
dabei die Unrichtigkeit der Ansichten einiger Forscher
erwiesen, die das Erscheinen von Kedaillons hauptdch-
lich beziehungsweise ausschliesslich in bestimmten Ge-
bieten gesehen haben, ohne den aktuellen Forschungs-
stand zu berUcksichtigen. Heute werden wir diese Punde,
bereichert ua neue Entdeckungen und Puhlikationen, in
engerem ~erritorialrahaen, d.h. in den Rhe~ und Do-
nauprovinzen einerseits und in den Gebieten 8stl1~
des Rhein und n8rdl1ch der Donau andererseits, betrach-
ten.
~la Kedaillons werden Gelegenheitsemissionen betra-
chtet, die zuaeist eine Vervielflltigung gangbarer
M&nzen waren. In der Praxis aber wurden bisweilen tur
lledaillons auch in Ualauf befi.ndliche llul tip le, TOr
allem goldener KGnzen angesehen. Die in der llteren
Literatur ziemlich oft gebrauchte Bezeicbnung "Medail-
lon" erlabt bei nicht aehr vorhandenen
,
Exemplaren kei-
ne Feststellung, ob dies wirklieh Gelegenheitsemissio-
. nem waren. Aus diesem Grund.e ziehe ioh hier auch Multi-
ple aus den Schatzfunden der Limeszone in Betracht.
Berflcksichtigt wurden auch Med.a.illons, von denen Jl8ll
nicht mit Bestimmtheit sagen Jrann, ob sie r8mischen
Mt!nzstltten entstamJ:aen, oder aber eine lfachahmung ori-
ginaler Medaillons sind.
Die Medaillonfande k8nnen - unabblngis von ibrer
Tauglichkei~ bei der Untersuchung der Emissionapolitik

1077
einselner Kaiser und der r8aischen Ikonographie, die sie
den Medaillons mit unbekan»ter Fundstelle gleichstellen
- als Zeugnisse TOn Eretgn1ssan unterschiedlieheD Charak-
ters ausgewertet werden:
1) vereinzelter Ereignisse ait beschrlnkter Reicbweite;
2) Ereignisse von allgemeinea Charakter und breiter ter-
ritorialer und zeitlicher Reichweite.
Zeugnisse vereinzelter Ereignisse sind vor allea
Schltse, die 1a allgemeinen ausser den Medaillons auch
Gold- und SilbermUnsen sowie allerlei Kostbarkeiten ent-
halten. Da jtlngste lledaillon oder die splteste llflnse er-
lauben - auf Grund der Kenntnis allgemeiner politischer
Ereign1sse - die Zeit der BerguDg des Schatzes zu beat~
men. So wurde z.B. der iD. :bona (L~ubl~B.Da) entdeolt't;e
Solidiscbatz Jlit :multiplen lllnzen Ende August 352, beTor
die Stadt von den Truppea des Constantius II besetz-t;
wurde, deponiert_ (Jelocm1k_ 1967). AUCh der SCbats aus
dem Kastell Augusta Raurica (Ira.iseraugat) ait 1? Silber-
aedaillons, Geflssen und lltlnSen 1st wahrscheinlich 'YOB
einea Mitglied der Besatzung des Megnentius vor daa E~
marsoh der Truppea des ConstaD:tiua II Tersteckt worden
(Instinslq 1971, 156). Die SchAltse ait lledaillons bese-
ugen also konkrete l'riegs:bsndlUDgeD. in den grensnahen
Pro'Yinsen. 1'erautlich sind auch die in Europe. ausserhalb
des Liaea entdecltteD Sohatsflmde llli't lledaillons infolge
der Kriegsbandlungen, TOll denen w1r ~ecloch zuaeist nichta
wissen, :l:D. die Erde geraten.
Ia Gegensats zu den Schlltzea, lass en sich clie hflufi-
ger in Siedlungen, selteDer in Grlbern, aeisteu ~edoch
vereinsel t gefundenen lledaillons nicht 111.1; una beJcsnnten
Ereignissen in 'ferbindu:ng bringen• Dies bedeutet ~edoch
nicht, dass diese :Funde _fUr die historische Interpre1;a-
tioD bedeutungalos aind. Ihr ErkeuntDiswert kann aber
erst naCh der Anwendung der Zone~Tergleicha.ethode he-
rausgearbeitet werden.
Ua die EigenschBften des Auftretens r8mischer Meda-
illonfunde in den RheiD- UD4 DoDS.upro'Yi.nsen des Kaiae-

1078
reichs einerseits und in den Llndern des Nordens ande-
rersei ts hervo.rzuheben, wurden im Ra.hmen eines jeden
diese.r grossen Gebiete zwei Zonen ausgesondert, von de-
nen die erste Gebiete zu beiden Seiten des Limes, die
ande.re die weiteren Gebiete umfasst. Ein Vergleich des
Auftretens von Medaillons (Tabelle 1. und 2.) beweist,
dass sie in den Seh!tzen vie 6fter in grenzfernen Ge-
bieten anget.roffen werden. Dies sind Gebiete, die in
Europa ausserhalb des Limes, nahe der Rheinlinie und
oft sehr weit von der Donau entfernt liegen. In den an-
deren Funden gibt es jedoch dazu keine Analogie. W!hrend
sie in der Limeszone sehr oft auftreten, werden sie in
den n8rdlichen L!ndern meistens in grosser Entfernung
vom Limes gefUnden. Alles in allem ist die Ve.rdichtung
der Funde in der Limeszone vie gr8sser, als in den
grenznaben Gebieten des Barbaricums. Man k8nnte hierbei
eine Widerspiegelung der Tatsaehe sehen, dass gerade am
Rhein und an der Donau die Hauptzentren von Militlr, Ver-
waltung und Wirtschaft der Grenzprovinzen lagen. Eine
Reihe der Hauptzentren der DBrdliehen T8lker dagegen.
ausgeno:mmen die Rhein- und einige Donaugebiete, bef'an-
den sich weit entfernt ~n den Grenzen des Kaiserreichs.
Eine noch geringere Anzahl von Medaillons, die nach
ihrea Erzgehalt und der Chronologie bezeichnet sind
(Tabelle 3.) weist ebenfalls auffallende Analogien und
Unterschiede in den beiden er8rterten Gebieten auf. Die
Mt.lnzen aus deu Scbatzfunden flberwiegen tlberall ganz ent-
schieden Exemplare aus anderen Fundarten, wobei in bei-
den Gruppen die Goldemissionen dominieren. Anders jedoch
sieht die Sache aua, wenn wir den Erzgehalt der Funde in
zwei chronologischen Abschnitten untersuchen, ~n denen
der erste den Zeitraum bis 312, der zweite die Zeit bis
zur Mitte des 5. Jh. uafasst. la ersten Zeitabschnitt
sind die uns aus den Schltzen bekannten Medaillons, so-
wohl in den grenznahen Provinzen, als auch in den nar-
dlichen Naehba.rl!ndern., ausschliesslich aus Gold, wlh-
rend in anderen Fundarten in dem umrissenen Teil des

1079
Kaiserreichs ausschliesslich bronzene Medaillons, 1a
Barbaricua dagegen Kedaillons aus allen drei Metallen
auftreten. Die Eigenart der beiden Fundgruppen unter-
streieht die Tatsache, dass in ihnen in der Regel Emis-
sionen verschiedener Kaiser auftreten. All dies fUhrt
una zu der Schlussfolgerung, dass die M:edaillons aus den
Siedlu.ngs-, Grab- und Einzelfunden intolge ganz anderer
Prozesse in die Erde gerieten als die Medaillons, die
thesauriert wurden. Vermutlich waren dies analoge Pro-
zesse zu denen, die verursacht haben, d.ass nichtthesa-
urierte Mtlnzen in die Erde kam.en.
Eine gans andere Erseheinung beobachten wir zur Herr-
schaftszeit der konstant1n1schen Dynastie und ihrer Nach-
folger. Zu beiden Seiten des Limes Uberwiegen zwar wei-
terhin Goldmedaillons, doch flberwiegen sie n8rdlich der
Grenze bei weitem sowohl in den Schltzen (100 ~) wie
auoh in den anderen Fundarten ( 94.1 %) gegentrber der
r8mischen Grenzprovinzen (57. 5 ~ in den Schl!tzen un4
66.7 ~ in den restliChen Funden). Der verhlltnismlssig
geringe Prozentsatz von Goldmedaillons in den Sehltsen
aus den besagten Pro~inzen 1st das Ergebnia der Entde-
ckung einer grlfsseren Anzabl "VVD silbernen Exemplaren
. "
in Kaiseraugst. Die J.hnlichkeit der lletall- und Chrono-
logiestruktur l!sst vermuten, da.ss ein Teil der Exeapla-
re aus den Einzelfunden aus den verstreuten Schatzfl.m-
den herrtlhren.
Es muss unterstichen werden, d.ass die Metall- und
Chronologiezusammensetztmg der .lledaillons in den er~r­
terten Gebieten von der Zusammensetzung dieser Ob~ekte
in den restliehen Gebieten des Kaiserreichs abweicht.
"
Dort besteht weder ein grosses Ubergewicht von goldenen
Exemplaren, noch geprlgte Emissionen a us der Zeit von
312 bis zur Hlllfte des 5. Jh. Das geht sowohl aus Haupt-
katalogen hervor die auch Fundmedaillons umfassen (IDbi-
tschek 19091 Gnecchi 1912; Dressel 1973) wie auch aus
den Zusammenstellungen der Funde von Medaillons (Regling
1912, 238; Toynbee 1944; Gumowslti 1960; Wielowiejski
1970, Tab. I!).

1080
 Es ist zu erwlgen, ob und inwiefern die beobachteten
Analogien und Differenzen den wahren Stand der Dinge im
Altertum widerspiegeln. Dies ist nicht leicht, denn der
Grossteil r6mischer Medaillons, die sich heute in den
Museums- und Privatsammlungen befinden, verfttgt Uber
keine Fundortangaben. Der heutige stand der Forsehung
tlber .Medaillons erlaubt anzunehm.en, dass in die Hlnde
der Forseher ein gr8sserer Anteil von Exemplaren ait
bekannten Fundort in den Grenzprovinzen und in Gebieten
ausserhalb des Limes geriet, ala in Italien selbst oder
in anderen Mi ttelaeerprovinzen des Kaiserreichs. Deshalb
scheint eine Vergleichsuntersuchung von Medaillons zu
beiden Seiten des Rheins und der Donau fUr die Festle-
gung historisch begrftndeter Analogien und Differensen
zweckmlssig zu sein (Wielowiejski 1970, 134-136). H1nge-
gen mUssen die Erwlgungen Uber die Quantitlten der Fun-
de sich in Anbetracht grundlegender Unterschiede ia
Stand der Registrierung der Entdeckungen von Medaillons
in den Rhein- und Donauprovinzen gegen dem in den n8r-
dlichen Bachbarl!ndern einerseits UDd des Mittelaeer-
raums andererseits, nur auf Bemerkungen allgemeinen
Cbarakters Uber die Metall- und Chronologiezusammense-
tzu.ng, die wir vorher pr!lsentiert haben, bescbr!nken.
Aus Rtlcksicht auf' die verhll tnis.rdlssig geringe Anzahl
von Funden und Exemplaren, da.rf hierbei jedoch nieht
die Methode zur Erforschung der Reprlsentativitlt der
Funde angewandt werden, die bei der Erforschung Ton
Schltzen r8mischer MUnzen und der Struktur ihres U~a­
ufs zu beiden Seiten des norisch-pannonischen Limes
bereits erste Resultate erbracht hat (Wielowiejski
19?7; Wielowiejski, Matuszewski 1979).
Um die Ursachen der grossen Konzentration von Medail-
lonfunden in den Grenzprovinzen des Kaiserreichs und in
einigen Llndern des Barbaricums zu verstehen, muss man
an ihre F\mktion im r8mischen Staat denken. An.f!lnglich
haben sie die Kaiser aus verschiedenen Metallen, insbe-
sondere aus Bronze, anllsslieh des Neuen Jahres sowie

1081
verschiedener Jubillumstage gesehlagen und den h8heren
Hofbeamten vergeben (Cl~ 1976). Oft gaben die Kaiser
die Medaillons naeh der Thronbesteigung aus, um diese
Tatsache zu bekunden und die Person des neuen Herrschers,
dessen Bild sich auf dem Avers befand, weithin zu propa-
gieren. Sp!lter, insbesondere im. 3. und 4. Jh., wurden
die Medaillons auch zum Gedenken an wichtige histori-
sche Ereignisse, z.B. Siege Uber die Feinde, gesehla-
gen, was in entsprechenden .Aufschriften (z.B. GLORIA
ROMANORUM) und bisweilen in Bildnissen seinen .Ausdruck
fand. Der Kaiser vergab diese Medaillons an Hofpers8n-
lichkeiten, ausgesuchte Ratsherren, Provinzstatthalter,
h8here Offiziere sowie Herrscher der Nachbarv8lker.
Diese Pers8nlichkeiten wurden ihrem Stand und Ansehen
nach beschenkt, wovon die grosse Unterschiedlichlteit
in der Gr8sse und ia Wert der Medaillons zeugt (Seeck
1898). Bit der ~erteilung von Medaillons befasste sieh
zur Zeit des splten Kaiserreichs ein spezieller Beamte
fUr Angelegenheiten der Kaiserspenden (comes saerarua
largitionua)(Seeck 1898, 21f.).
Die .Austeilung von lledaillons war nicht nur ein In-
strument der Prestigepolitik der Kaiser, sondern auch
ein wichtiger Faktor der Aussenpolitik des splten Kai-
serreichs, deren Hauptziel die Gewlhrleistung der Si-
cherkeit der Grenzen war. Die Verteilung der Medaillon-
funde aus dam 4.-5. Jh. zeigt zwei Seiten dieses Pro-
blems. Ih1•e Konzentration in der Limeszone weist auf
ein hlufiges Beschenken hBhere Offiziere und Beamten
mit kostbaren Andenken in den immer mehr angriffsbe-
drohten Grenzabscbnitten hin. Diese Geschenke, die eine
Art Auszeichnung und Gunstbezeugung seitens des Kaisers
darstellten, sollten wahrscheinlich zur Hebung der Mo-
ral der Armee und der VerwaltungsbehBrden beitragen.
Das Auftreten von Medaillons in den Siedlungsaggloaera-
tionen n8rdlichen T8lker dagegen darf mit diplomati-
"
schen Geschenken und Tributen, die Uberflllen auf das
r8mische Territoriua verbeugen sollten, in Zusammenhang

1082
gebracht werden. Sie wurden den FUrsten der Germanen,
Hunnen und anderen St!mme von kaiserlichen Gesandten
Ubergebene Davon kann das Auftreten von goldenen Meda-
illon.s in den Solidisch!.tzen, meisz zusa.mmen m.it ver-
schiedenen kostbaren Schucksachen, zeugen. Man, hat der
Jhe wachsenden Schwere von goldenen Medaillons
Aufmerksamkeit geschenkt (Wielowiej 1
138f@), wobei die gr8ssten unter ihnen ein Gewicht von
200-750 g erreichten. Unbestreitbar scheint die Folge-
rung ttber die gegenseitige Abh!ngigkeit zwischen der
ungestUm wachsenden Bedrohung seitens der Nordv8lker,
insbesondere der Goten und Hunnen, und dem Gewicht der
Medaillons, die in ihren Beistz Ubergingent zu
Die Medaillonfunde in den L!ndern des Bar-
baricums sind nieht nur ein Ausdruck diplomatischer Kon-
takte und Ansamml ung der von den R6mern fUr die Nicht-
aggreasion erhaltenen Kostbarkeiten. Die vom Kaiser
selbst kommenden Geschenke mUssen zweifellos den Ehr-
geiz und die Eitelkeit n8rdlieher Fttrsten befriedigt
sowie 1hr Anaehen gehoben haben. Es ist anzunehmen~
dass so wie der FrankenldJnig Chilperich (Honumenta Ger-
maniae Historica, Scriptores Rerum Merovingicarum, I,
1885~ 245f.) sich auch die frUheren AnfUhrer verschie-
dener V6lker des Nordens durch die Geschenke in Form
von Medaillons geehrt fUhlten, ganz abgesehen von ihrem
materiellen Wert. Von ihrer zielbewussten Zuschaustel-
lung zeugt auch die Tatsache, dass, die auf Reichsge-
biet gefundenen Medaillons zumeist keine zus~tzliehen
Verzierungen aufweisen, w!hrend die von ausserhalb sei-
ner Grenzen stammenden Exemplare sehr oft in Rahmen
eingefasst und die manchmal mit reicher Ornamentik ver-
" in Form. einer T6J.le verzehrt
ziert und mit einer Ose
sind, die an der Rahmen angel6tet wurde. Um den Hals des
Ftirsten oder Anf8hrers b!ngend muss diese Dekoration
sicherlich auf die Zuscbauer grossen Eindruck gemacht
haben. Es wird angenommen, dass die Medaillons als Aus-
zeicbnung,en ( "Orden"), Zeichen des Reichtum.s, sicherlich

1083
-aber auch der Macht angesehen wurden (Brajcevskij 1953,
53). Nach Art dieser Medaillons wurden auch Aurei und
Solidi zum Aufh!ngen adaptiert (Konik 1964; Wielowiej-
ski 1970, 140-144).
Unsere Erw!gungen haben bewiesen, dass trotz der ver-
h!ltnism.M.ssig immer noch geringen Anzahl von Medailon-
"
funden, die auffallenden Ahnlichkeiten sowie einige Un-
terscniede in ihrer Verteilung und Zusammensetzung in
den rBmischen Grenzprovinzen und in den n6rdlichen Nach-
barl~dern keinesfalls zufallig sind. Sie spiegeln - !hn-
lich den vergleichsmHssig behandelten Funden r8mischer
Mftnzen - die wichtigsten Merkmale der damaligen Wirklich-
keit wider, die nicht nur der unmittelbaren Nachbar-
schaft, sonder.n auch den verschiedenartigen Kontakten
entsprangen, von denen sowohl schrieftliche, wie auch
archltologische und numism.atische Qnellen ze~en. Die
Medaillonfunde sind demnach eine wertvolle Quelle zur
Erforschung der historischen Rolle des Limes, vor allem
in der spltramischen Zeit.

LI:l!ERA TUR

Brajeevskij, M.J. 1953: Arheologi6ni svideennja ueasti

Shidnin Slovian u politi~nih podijah rimskoj istorii
III-IV st. n.e., Arheologija (Kijev), VIII, 45-55.
Clay, C.L. 1976: Roman Imperial Medallions: The Date
and Purpose of Their Issue, Actes du 8 e Co~es
' Inter-
. -
national de Numismatique, New York-Washington, septe~
bre 1973 (Paris-Basel), 253-265.
Dressel, H. 1973: Die r8mischen Medaillons des Mftnzka-
binetts der Staatlichen Museen zu Berlin bearbeitet
von K. Regling, Dubl~ZUrich.
Gnecchi, F. 1912: I medaglioni romani, I-III, Milano.
Gumowski, M. 1960: Medaliony rzymskie na ziemiach pol-
skich, Meander, XY, fasc. 11-12, 586-603.

1084
Instinsky, H.U. 1971: Der sp!tr6mische Silberschatz von
Kaiseraugst, Akademie der Wissenschaften und der Lite-
ratur Ma.inz, Abhandluns;en des Geistes- i.md Sozialwis-
senschaftlichen Klasse, Nr. 5, 141-156.
Jelo~nik, A. 196?z Les multiples d'or de Magnenee deco-
uverta a' Emona, Revu.e Num.ismatigue, Ser.a VI, vol. IX,
209-235.
Konik, E. 1964: Roman Suspension-Coins found in Sile-
sia, !2!, Lii 2, 30?-314.
Kubitschek, w. 1909: Ausgewlhlte r8mische Medaillons
der kaiserlichen MUnzsemmlupg in Wien, Wien.
Regling, K@ 1912~ R8mischer Denarfund von Fr6ndenberg,
Zeitschrift fUr Numismatik, XXIX, 189-253®
Seeck, Ou 1898: Zu den Festmftnzen Constantins und sei-
ner Familia, Zeitschrift fUr Numismatik, XXI, 1?-65.
Toynbeefl' J .M.c. 1944: Roman Medallions, Nl.UUismatic
Studies Nr~ 3, New York.
Wielowiejski, J. 1970: Uwagi o rozmieszczeniu znale-
zisk i funkcji rzymskich medalion6w oraz monet adap-
towanyeh do zawieszanja, Wiadomosci Numizmatyezne,
XIV 3, 129-145.
- 19?3: Analogien und Unterschiede in der Verteilung
und der Chronologie der rBmischen MUnzsch!tze zu bei-
den Seiten der Donau, Aetes du VIIIe Congr~s Interna-
~ional des Sciences Pr~historiques et Protohistori-
gues, Beograd 1-15 septembre 1971, t. III, 257-262.
- 197?: R8mische Schatzfunde zu beiden Seiten des no-
risch-pannonischen Limes, Studien zu den Milit!rgren-
zen Roms, II, Vortr!se des 10. Internationalen Limes-
kopgresses in der Germania Inferior, K81n, 41?-42?.
Wielowiejski, J., Matuszewski, A. 1979: Anwendung der
statistischen Methode zur Erforschung der Umlaufs-
struktur von r6mischen Mftnzen in Mitteleuropa, Stu-
~ien zu FundmUnzen der Antike, 1, Berlin, 265-280.
-
1085
 Tabelle 1. Anzahl der Funde von r6mischen Medaillons
in den Rhein- und Donauprovinzen
(in Klammern Exemplara.nzahl)

F====================================ft===================
Ubriges Zusam-
Limeszone Provinzial- men
Provinzname gebiet
Scha tz- Andere Scha tz-. Andere
funde Funde funde Funde
~================================~=========~=====~=-·==·=
German:ia und
Gallia Belgiea1 1(17) 4 (4) 2(18) 5 (5) 12(44)
Noricum2 1 (1) 1 (1)
Pa.nnonia3 1 (4) 13(13) 3(24) 4 (4) 21(45)
Moesia. und4 10(10) 2 (2) 12(12)
Thracia

1 Beaurains-lea-Arras, Dalheim, Helleville, Ghent,

Kaiseraugst, K81n, Reil, Saarbrtlcken, Saasenheim.,
Trier
2 Zell am Ybbs
3 Bad Deutseh-Altenburg und Petronell, Du.uaujvS.ros,
Iza-Le~, Ljubljana, Mitro"Vica (U11gebung),
o. sz&ny, Petri~anec, Ptu~, Sisak, Sopron, Szar-Feher,
Wien
4 Adamklisi, Bojana, Bregovo, Celeiu, Dobrudscha,
Eisernes Tor, Gaurene, Zatonje

1086
 Tabelle 2. Anzahl der Funde von r6mischen Medaillons
in den Llndern des n8rdlichen. Barbaricums
(in Klamm.ern Exemplara.nzahl)

===============- ============== ============== ======

Nahe der Weit von der
Limes zone Limes zone Zusam-
Land
Schatz- Andere Schat Andere :men
fun de Funde funda Funde
================ ======= ======- ======
Niederlande 1 1 (6) 1 (6)
Bundesrepublik
Deutschland 3 (3) 3 (3)
Tschecho-3
slowakei 1 { 1) 1 ( 1)
Jugoslawien4 1(15) 1(1
Rnm!njen·5 3 (3) 1(13) (4) 6(20)
Polen6 1 (4) ? ) 11)
UdSSR7 5 (6) 3 (3) 8 (9)
Dl.ne:m.ark8 3 (3) 3 (3)
No:rwegen9 2 (2) 2 (2)

Zusammen 2(21) 6 (6) ?(23) 20(20) 35(70)

===============- ======= ====== ======= -======= ===:~~==

1Velp
2 Mf!nchhof, Neuwied, Wtlrzburg
3 Frjdek-Mistek
4 Bor~a
5 Arad, SiebenbUrgen, stareova, Szilag;y-Somly6,
f
Temesvar
6 Kur6w, :Uloteczno, Opoki, Przem,ysl (Umgebung),
Slupia Nowa, Trzebicko, Zag6rzyn, Zb6jna
? Alytus, Boroczyce, BrarJ, Kapsu.ka.s, Kijev, Rajki,
Sala~oliai, "Scbatzfun.d aus der Eremitage"
8 Alles8, Faxe, Trtlnderup
9 Zwei Grabfunde aus unbekannten Ortschaften

1087
 3® Chronologische Verteilung rljmischen
Medaillons den Rhein- Donauprovinzen

====~=========~===============

Zeit Me tall Schatzfunde Andere Funde

Anzahl %

AE 17 100.0
!nsgesam.t • 16 100.0 ' 17 100.0

2.3 12

40 100.0 18 100.0

AU 4
1
.;12 I 4

Insgesamt 5 100.0 9 100.0

AU 42 100.0 16
Nach. AR

Insgesamt 100.0 100e0 .

~~===~~~~=================L==============-===============j

1088
 Summary

The comparative method of analysis of 46 sites, where med-

allions were found (102 objects) within Rhine and Danubian provinces
of the Roman Empire, and 35 sites (70 objects) in Europe north of
limes, has shown occurrence of medallions more often in hoards,
found on the outlying territories. Finds not in hoards are grouped
in the northern lands mainly far from limes, whereas in the.provinces
close to the limes they are found mostly in the borderlands. The
majority of the medallions on both sides of the border are dated to
the period from A.D. 312 to the middle of 5th century A.D. and, in
the contrast to finds from 1-3 centuries A.D., they are mainly gold.
The density of medallion finds of 4-5 centuries in the limes area
indicates that they were presented to high officials and military com-
manders to improve the morale of army and administrations on the
border still threatened by invasions. While the occurrence of med-
allions in sites of barbarian tribes in the North should be interpreted
as diplomatic gifts from Roman emperors to chieftains of German,
Hun and other peoples, and also as tribute to prevent the invasions
of Roman territories. In the 4th century the correlation between the
growing danger of invasion by northern peoples and the weight of med-
allions that they possessed can be observed. These medallion finds
are now a valuable source for the interpretation of the historical role
of the limes towards the end of its life.

1089
 0 L D A T E N I N D E R R ~ M I S C H E N K U N S T

Zur unterschiedlichen Darstellungsweise des Milit~rs

in Rom und den Provinzen

Gotz \vaurick

In zahllosen Darstellungen der r~mischen Kunst sind ge-

rlistete Krieger iiberliefert. Sie treten uns in Gestalt
von Gottern, als mythische Helden, als irdische Herr-
scher aber auch als Offiziere und einfache Soldaten ent-
g;egen Bei einem Vergleich ihrer Riistungen mi t den Ori-
ginalfunden wird man meistens groBe Unterschiede fest-
stellen. Bemerkenswert sind vor allem die zahlreichen
griechischen Rustungsstiicke und Waffen in der romischen
Kunst, obwohl doch Legion~re und Auxiliarsoldaten in
Wirklichkeit ganz anders ausgeriistet waren.
Erl tern l~Bt sich das etwa am Beispiel des wohl be-
rtihmtesten geschnittenen Steins aus dem Altertum, an der
soge Gemma Augusta in Wien. Dieser 19 cm hohe Onyx-
Cameo darf als ein Hauptwerk romischer Staatskunst be-
zeichnet werden. Er zeigt im oberen Fries den Kaiser
Augustus, der den siegreich zurtickkehrenden Tiberius
empf~ngt. Auf welches historische Ereignis die Szene
anspielt, kann nicht mit Sicherheit gesagt werden: u.a.
wurde mit recht guten Grtinden die erfolgreiche Nieder-
schlagung des pannonischen Aufstandes durch Tiberius im
Jahre 9 n@ Chr. in Betracht gezogen. Der im oberen Friei
ge:feierte Sieg wird in der unteren Halfte des Cameo durch
Soldaten, Gefangene und ein Tropaion direkt angesprochen.
Uns interessieren jedoch nur die Typen der Riistungen und
Waffen: Helm, Panzer, Schild und Schwert.
Zwei Helme geh~ren zum attischen Typ mit halbrunder Ka-
lotte, herabgezogenem Nackenschutz und vor allem mit dem
visierartigen Schild tiber der Stirn, der an den Seiten
oft in Voluten endet. Dieser Typ hat in der griechischen

1091
Kunst eine jahrhundertealte Tradition. Ein Urkundenrelief
aus dem Jahre 403 v. Chr. z. B. zeigt Athena mit dem
attischen Helm. Originale dieser Form haben sich bisher
nicht gefunden. Es gibt aber Varianten, die eindeutig vom
attischen Helmtyp abgeleitet sind. Ein hellenistischer
Bronzehelm aus r.:elos irn Louvre ist gut mit den Darstel-
lungen der Gemna Augusta zu vergleichen. Der gleiche
Helmtyp erscheint auf dem Fries des Siegesmonumentes von
Aemilius Paullus in Delphi, und zwar nicht bei den RB-
mern, sondern bei den t-;akedonen,. Er ist dami t als grie-
chischer Helm im 2a Jahrhundert v. Chr. belegt.
Ein dritter Helm der Gemma Augusta kann ebenfalls mit
griechischen Originalfunden verglichen werden; ein frag-
mentierter Bronzehelm von der Tamanhalbinsel in SlidruB-
land hat nicht nur die spitze Form, sondern auch den
Stirnschild 1 der seitlich in Voluten endete Der Fundort
licgt im Gebiet griechischer Kolonien.
Obwohl wir die tibrigen Helme nicht rnehr besprechen kBnnen,
ist folgendes festzustellen: soweit sich die Helme der
Gemma Augusta einordnen lassen, gehoren sie zu grie-
chischen Typen. Vergleicht man sie in der Gegenprobe mit
den aus Originalfunden bekannten romischen Helmtypen, so
bestehen keinerlei Ubereinstimmungen.
Als Korperschutz erscheint auf der Gemma der Nuskelpanzer,
der die Form des mannlichen Korpers bis zu den Leisten
wiedergibt. Als Vergleichsstlick aus der griechischen
Kunst zeigen wir das attische Grabmal des Aristonautes
aus dem 4. Jahrhundert v. Chr. Die dargestellten Panzer
sind also ebenfalls ursprunglich griechische Rustungs-
stucke. Allerdings wurden sie - wie wir von anderen
Denkmalern wissen - in der romischen Armee noch van Offi-
zieren getragen.
Von den Schilden besi tzen wir z1vei in der kreisrunden
Form, wie sie fur griechische Riistungsstticke in der
hellenistischen Kunst geradezu typisch ist. Auf dem be-
rei ts erwahnten Aemilius-Paullus-i'1onument tragen die
Makedonen kreisrunde Schilde,die Romer dagegen Lang-

1092
schilde. Im romischen Heer der Prinzipatszeit jedoch
kommt der Rundschild noch ganz vereinzelt vor: dort ist
er auf die kleine Gruppe der Standartentrager beschrankt.
Von den iibrigen Schilden gehort die Pelta am Tropaion
ebenfalls in den griechischen Bereich; nur das Scutum zu
FiiBen der Roma ist ein typisch romisches Riistungssttick.

Das Schwert des Soldaten, der gerade das Tropaion auf-

richtet laBt sich ebenfalls klar einordnen. Bezeichnende
Merkmale sind die hohe Parierstange und das ausladende
Ortband, das die Scheidenbreite betrachtlich tibertrifft.
Genauso werden seit langer Zeit in der griechischen Kunst
die Schwerter dargestellt, wie an dem im 2. Jahrhundert
v. Chr. entstandenen kleinen Fries von Pergamon. Das
Schwert aber, das die Romer zur Zeit der Gemma Augusta
verwendeten, sah vollkommen anders aus: es war der Gla-
dius, dessen hervorstechendes Merkrnal die in einer langen
Spitze zulaufende Klinge ist. Entsprechend sind die zu-
gehorigen Scheiden an der Spitze geforrnt: das Ortband
besteht nur aus einem kleinen Knopf - nicht aber aus
einem breit ausladenden Gebilde wie bei der Darstellung
auf' dem Cameo.
Eine kurze Bilanz nach diesen Vergleichen lautet also,
daB fast alle Riistungsstticke und Waffen auf der Gemma
Augusta griechisch oder zumindest griechischen Ur-
sprungs sind. Soweit wir sie mit Originalen vergleichen
konnen, handelt es sich bei diesen stets um griechische
Fundstiicke, wie sie im romischen Heer nicht mehr ver-
wendet wurden.
Der Ktinstler des Cameo, der seinen eigenen Kaiser sowie
mehrere Mitglieder aus dem zeitgenossischen Kaiserhaus
darstellt und einen Sieg des romischen Heeres verherr-
licht zeigt also Riistungsstiicke und Waffen, mit denen
dieser Sieg gar nicht errungen wurde.
Ehe wir diese Feststellungen an einigen anderen Kunst-
werken tiberpriifen, wollen wir uns noch einmal die
charakteristischen Bestandteile der griechischen Rtistung
einerseits und der romischen andererseits betrachten:

1093
griechisch ist der Muskelpanzer, r~misch der Ketten-,
Schuppen-und Schienenpanzer. Griechisch der Rundschild,
romisch sind das halbzylindrische Scutum, der Oval- und
der Sechseckschild. Griechisch sind der korinthische und
der attische Helm sowie hellenistische Formen, r~misch

u.a. Typ Hagenau und Typ Weisenau.

Auch bei vielen anderen romischen Kunstwerken kommen
griechische Waffentypen vor, die im r~mischen Heer nicht
mehr verwendet wurden. So haben die R~mer ihre mytholo-
gischen Gestalten fast alle aus der griechischen Kunst
libernommen und damit auch deren Rustungen. Diese Waffen
aus der "mythischen Welt" wurden auch fur alle symboli-
schen und allegorischen Themen verwendet oder wenn man
den Kaiser in einer heroischen Sphare darstellen wollte.
Aber selbst bei realistischen Themen, die sich auf histo-
rische Ereignisse beziehen, vermiBt man oft die zeit-
gen~ssischen romischen Waffentypen. Auf den Bildwerken
der stadtr~mischen Kunst und der groBen Reichskunst
tragen die Soldaten bis weit in das 1. Jahrhundert n.
Chr. hinein irnmer wieder griechische Rustungsstucke. Zu
den altesten historischen Reliefs gehort das Census-
relief aus spatrepublikanischer Zeit. Zwar sind Lang-
schilde und Kettenpanzer der Soldaten r~misch, die Helme
aber gehoren eindeutig zu hellenistischen Typen. Der
Helm des Reiters laBt sich mit einem Originalfund in
Turin vergleichen. Dieser Helm - mit griechischer In-
schrift - hat dieselbe hohe, spitze Form und die fur
hellenistische Helme charakteristischen Falten in dem
breiten Rand.
Auch das spatrepublikanische Grabmal der Julier in St.
Remy liberliefert hellenistische Helmtypen. Einer hat halb-
kugelf~rmige Kalotte, breiten Rand uber der Stirn und
eine flache, zipfelartige Bekronung. Das Vorbild k~nnte

ein Originalfund aus Makedonien sein, der in griechischer

Schrift einen illyrischen K6nigsnamen hellenistischer
Zei t tri::~gt.

Auf einem spatrepublikanischen Grabmal in Rom tragen die

1094
Krieger ebenfalls hellenistische Helme. Besonders die Vo-
luten und die Falten im Rand konnen wir mit einem Ori-
ginalfund aus Kazanlak in Bulgarien vergleichen. Der-
selbe Helmtyp findet sich auf einem Relieffragment in
Rom wieder, das schon in claudische Zeit gehort.
Auf den romischen Bildwerken his weit in das 1. Jahr-
hundert n. Chr. hinein tragen die Soldaten demnach
griechische Helme, wahrend die wirklichen Helme der
romischen Soldaten - also Typ Hagenau oder Weisenau
in den Reliefs nicht zu finden sind. Die Erklarung dafur
mussen wir in dem starken EinfluB der hellenistischen
Kunst auf die Werkstatten Italiens und Sudgalliens
suchen. Auf diesem Wege sind auch die in der griechischen
Kunst ublichen Waffen und Rustungen den Bildhauern im
romischen Reich bekannt geworden. Als Beispiel kann der
Pergamonaltar genannt werden, wo etwa der Volutenhelm mit
dem breiten gefalteten Rand vorkommt, wie wir ihn in der
stadtromischen Kunst immer wieder finden konnen.
Der EinfluB der hellenistischen Kunst in Rom, Italien und
Sudgallien war noch in der fruhen Kaiserzeit so stark,
daB die wirkliche Ausrustung des Soldaten als Vorbild fur
den Bildhauer keine Rolle gespielt haben kann. Eine ge-
naue Wiedergabe der romischen Rustung ohne historisie-
rende Tendenzen setzt sich erst spater durch. In Rom wird
sie flir uns nicht vor der spatflavischen bzw. trajani-
schen Zeit faBbar: die 113 n. Chr. geweihte Trajanssaule
gibt den Legionssoldaten mit Schienenpanzer und Scutum
ganz genau wieder oder den Auxiliarsoldaten mit Ketten -
bzw. Schuppenpanzer und Ovalschild; alle tragen Helme,
die in ihrem Aussehen zwar variieren, in ihrer Grundform
aber dem Typ Weisenau entsprechen.
Die entscheidenden Impulse flir eine so genaue Wiedergabe
der Soldatenausrustung kommen nicht aus der stadtromischen
Kunst selbst, sondern aus den Provinzen, wo die helle-
nistischen Einfllisse wohl keine Rolle gespielt haben.
Zu der Zeit als die groBe Stadt-und Reichskunst noch eng
der hellenistischen Tradition verpflichtet war, erschie-

1095
nen die Soldaten auf ihren Grabsteinen.in den Provinzen
bereits so, wie sie wirklich ausgesehen haben. Wir ver-
gleichen an vier Beispielen die Darstellung auf den
Grabsteinen mit Originalfunden:
1. Ein Grabstein aus Aquileia zeigt genau den Helm vom
Typ Hagenau, wie er z.B. in NeuB gefunden wurde.
2. Auf rheinischen Grabsteinen wird h~ufig der Reiter-
helm mit einer als Haarschopf gestalteten Kalotten-
oberfl~che wiedergegeben, z. B. am Kolner Grabmal des
Durises. Das Frgamnnt eines solchen Helmes hat sich
in Koblenz gefunden.
3. Ein Schuppenpanzer wie auf dem Banner Grabstein des
Vonatorix kommt als Originalfund z. B. aus dem Kastell
Straubing.
4. Als Schwert tragt Hyperanor auf seinem Grabstein aus
Bingerbrtick den spitz zulaufenden Gladius mit reich
verzierter Scheide. Das Fragment einer ~hnlichen

Scheide wurde in Wiesbaden gefunden.

Auch an dem in der ersten Halfte des 1. Jahrhunderts n.
Chr. entstandenen Bogen von Orange erscheinen viele
Einzelheiten von der wirklichen Ausrtistung des romischen
Soldaten, z. B. der Schuppenpanzer, den die hellenistische
Kunst nicht dargestellt hat, aber auch die geschwungenen
Panzerschliefien, die wir als Original aus romischen
Kastellen kennen. SchlieBlich sei bier noch der schon
oft gezogene Vergleich zwischen den Reliefs vom sog.
Mainzer Oktogon und den Originalfunden wiederholt. An
diesem Monument aus flavischer Zeit kommen die typischen
Rtistungsstlicke der romischen Soldaten vor: der Helmtyp
Weisenau, das halbzylindrische Scutum der Legion~re und
das Schwert in der ftir den Gladius charakteristischen
Form.
Zusammenfassend konnen wir also sagen, daB die anti-
quarisch genaue Darstellung der romischen Rtistung in den
Provinzen seit Beginn der Kaiserzeit selbstverst~ndlich

war, wahrend sie sich in der sp~tromischen Kunst erst

viele Jahrzehnte sp~ter durchgesetzt hat. Dieser er-

1096
staunliche Unterschied erkl~rt sich aus zwei Grtinden:

erstens spielte der Einflufi der hellenistischen Kunst

(~it ihren griechischen Waffentypen) in Rom bis weit ins
1. Jahrhundert n. Chr. hinein eine grofie Rolle, wahrend
er die nordwestlichen Provinzen nicht mehr erreichte.
Zweitens waren in Rom und im ~utterland praktisch keine
Truppen stationiert, wahrend die Bildhauer in den kaiser-
lichen Provinzen Soldaten und deren Rtistung immer vor
Augen batten.
Wahrend die meisten kulturellen und ktinstlerischen Im-
pulse ihren Ursprung in Rom haben, und von der Haupt-
stadt in die Provinzen wirken, ist es bei der anti-
quarisch getreuen Darstellung des Soldaten in der ro-
mischen Kunst gerade umgekehrt. Sie war fur die Pro-
vinzen schon jahrzehntelang selbstverstandlich, ehe sie
sich auch in der Hauptstadt durchsetzen konnte.

Anmerkung:
Diese Fragen sind Teil eines Aufsatzes liber "Histori-
sierende Waffen bei den Romern", der in ausfiihrlicher
Form im "Jahrbuch des Romisch-Germanischen Zentralmuseums
Mainz" erscheinen wird.

1097
 Summary

Numerous representations in Roman art portray armed soldiers.

On comparing their armour with actual finds there are often substan-
tial differences. The artists prefer mostly Greek types of arms
which are no longer used by the Roman troops. This is readily ex-
plicable for certain themes. The Romans took over their mythologi-
cal figures mainly from Greek art and with it their arms. These
weapons "from the mythical world" were also used for all symbolical
and allegorical themes or whenever the emperor was to be presented
in a heroic sphere. Even in realistic themes, relating to historical
events, contemporary types of arms are often missing. On reliefs
of Roman urban art, Roman soldiers wear Greek single pieces of
armour until the 1st century A.D. This may be explained by the
strong influence of hellenistic art on the Roman, Italian and South
Gallic workshops.
An accurate representation of Roman armour is not attainable
in Rome before the late Flavian/Trajanic period. The explanation
for this fact lies in the provinces, where Hellenistic influence had
been of no importance. Until the middle of the 1st century A.D. and
later on Greek weapons were mostly represented in Rome, while in
the provinces soldiers on monuments were depicted in a realistic
manner. This distinction appears comprehensible, as there were
scarcely troops stationed in the motherland, while the sculptors in
the emperor's provinces always had within sight the soldiers and
their armour.

1098
77. ZWIEBELKNOPFFIBELN ALS KENNZEICHEN VON SOLDATEN

AUF SPATROMISCHEN DENKMALERN

Heinrich Zabehlicky

Zwiebelknopffibeln haben als chronologisch gut faBbare

Fundgruppe besonders in der Erforschung spatromischer
Graberfelder in den letzten Jahren groBe Beachtung
gefunden. Zur Frage des Tragerkreises und der Laufzeit
des Fibeltyps sollen hier einige Beobachtungen an den
Darstellungen solcher Fibeln auf spatromischen Denk-
malern vorgelegt werden.

In einem jeweils weiteren Zusammenhang wurden die Tra-

ger unseres Fibeltyps schon wiederholt definiert.
Belaev (1928, 1o4) hat die Zwiebelknopffibel als Aus-
zeichnung und kaiserliches Gnadengeschenk bezeichnet.
Delbrueck (1929, 39 und 41) sieht sie als zugehorig
zur Chlarnys der Beamten an und meint, daB Offiziere
meist eine Scheibenfibel t~ugen •• Patek (1942, 146ff.)
nimmt die Produktion in militarischen Fabriken an und
htilt Soldaten ftir die Trager dieser Fibeln. Auch Heur-
gon (1958, 23) schrankt den Tra~erkreis auf Manner,
besonders Soldaten ein, er betont aber auch, daB alle
kaiserlichen Funktionare die Zwfebelknopffibel tragen
konnen, da die offentliche Funktion, auch die mit mehr
zivilern Charakter, im 4. Jahrhundert dem Militardienst
angenahert wird. Zu ahnlichen Schltissen kommt Ftilep
(1968, 4o7ff.), der hochrangige Soldaten und administr~
tive Beamte als Trager nennt. DaB die unterschiedlichen
Materialien Gold, Silber und Bronze sowie ver~oldete

Bronze auch eine Differenzierung des sozialen Status

der Trager bedeuten, hat Keller··(1971, 171) betont. Er
meint aber, daB die Fibeln nicht auf Manner im Staats-
dienst beschrankt sein mtissen. Als Beamtenfibel wird
unser Typ von Wessel (1971, 346f.) bezeichnet. Noll

1099
(1974, 283f.) behandelt goldene "Kaiserfibeln", die ei-
ne Auszeichnung ftir hohere Beamte und besonders Offi-
ziere sein konnen. Jobst (1975, 94) betont die Herkunft
aus der militarischen Uniform der diocletianischen Zeit
und schlieEt, wie andere, den Kaiser als Trager der
Zwiebelknopffibel ein. Eine Abkehr von der Deutung als
Militarfibel enthalt die Arbeit von.Vag6 und Bqn~ (197~
167) sowie deren Rezension (Lanyi 1978, 261). Hier wer-
den die Zwiebelknopffibeln als Abzeichen der zw erbli-
chen Diensten verpflichteten staatlichen Arbeiter al-
ler Berufsgruppen aufgefaEt, immerhin auch der halb
bauerlichen, halb militarischen limitanei.

Die Entstehungszeit des Typs ist mit den letzten beiden

Jahrzehnten des 3. Jahrhunderts recht eng festgelegt
worden (Keller 1971, 26f£.; Jabst 1975, 95). Beim End-
datUm ergibt sich eine deutliche Diskrepanz zwischen
den Farschern, die van den Fibeldarstellungen ausgehen,
und denen, die kankrete Fibelfunde bearbeiteJ. Obwahl
die letzteren die bildlichen Darstellungen van Zwiebel-
knapffibeln sehr wahl kennen und auch anftihren, nehmen
sie das Ende des Typs mit der Mitte des 5. Jahrhunderts
oder mit dem Ende des westromischen Reiches an. Die
Darstellungen sind aber auch im 6. Jahrhundert noch
haufig und begegnen vereinzelt sagar im 7. Jahrhundert.

Zur Frage nach den Tragern der Zwiebelknapffibeln fallt

freilich die Probe an den realen Fibelfunden enttau-
schend aus. Wahl sind die meist in Mannergrabern ge-
fundenen Zwiebelknapffibeln in Garnisanen hafiger als
in zivilen Siedlungen, doch ist in den Kastellarten
auch die Gesamtzahl der Graber groBer 2 • AuBerdem be-
gegnen tatsachlich auch in Frauen- und Kindergrabern
Fibeln unseres Typs 3 • Dabei ist natlirlich an die Ver-
erbung der Fi bel zu denken. Sagar wenn sie :orimar nur
an Soldaten ader Beamte ausgegeben wurden, hat sicher
niemand die Einkleidung des Leichnams tiberwacht. Auch
muEten Veteranen keineswegs an ihren Dienstorten besta~

1100
tet werden. So sind fUr die Frage nach den Tragern der
Fibel zunachst die Darstellungen heranzuziehen.

Aus der gar nicht gerin~en Anzahl der Denkm~ler, die

,_,. -· ,_

Zwiebelknopffiheln zeigen, kannten nur einige Beispiele

aus,g;ewa.hl t werden, cUe die f!enannten Prap-en beleuchten.
sallen.

1. Medaillandarstellunp des Galerius van einem Tarbo~en

im Palast des Galerius in Saloniki. Saloniki, Arch.
Museum. Nach 298 n. Chr.
Rlisch 1969, 195 Nr. M 2o, Abb. 1o9. - Laubscher
1 g 7 5 , 1 1 9 • - Br e nk 1 9 7 7 , 1 8 7 , A h b • 1 6 Lt a •
Das Medaillon zeigt den Kaiser mit einer Bdelsteinge-
schmlickten Zwiebelknopffibel an der rechten Schulter.
Die Verzierung mit Edelsteinen war der Fibel des Kai-
sers vorbehalten (Delbrueck 1929, 4o).

2. Christlicher Reliefsarkaphag aus Ram. Vatikan, Museo

Pio Cristiano Inv. Nr. 222. 1. Drittel 4. Jh.
Deichmann 1~67, Nr. 22.
H~ufig ist auf christlichen Reliefsarkophagen des 4.

Jahrhunderts die Gefangennahme Petri dargestellt. Die

Soldaten, die den Anostelflirsten festnehmen, tragen den
i

Mantel mit der Zwiebelknopffibel. Ihre Kopfbedeckung

ist der nill~_p~ann_Q_p_t_cu.§_, die- Fellmtitze (Ubl 1976),
ein charakteristischer Teil der militarischen Uniform
diokletianischer bis konstantinischer Zeit. Die Kombi-
nation von Fellmlitze und Zwiebelknopffibel ist ftir die
Soldaten in dieser Szene die Regel. In naiv anachro-
nistischer Weise wird Petrus zusammen mit Soldaten in
der Uniform der Herstellungszeit dargestellt. Bier
spiegeln sich gewiB die Ereignisse der nur wenige Jahre
zurtickliegenden diokletianischen Christenverfolgung.
Die ausfiihrenden 0r£ane dieser Verfolgung trugen also
Zwiebelknopffibeln. In Rom selbst waren das die cohor-
.:t~§- YE-..:9_a_n_§.e ( Freis 1967, 23ff.). Ihr Kamrne.ndant, der
,E!_'0~eJ~_q_t~E)~-~J'Jl_i~?, ist in den Martyrerakten der prami-
nenteste Christenverfolger, als Ausflihrende werden die

1101
Dienstgrade centurio und _9ommentarienses. genannt. In
den beiden M~nnern, die Petrus verhaften, wird man also
die getreue Darstellung von Soldaten der cohortes Erba-
na.e erblicken dtirfen, wie sie dem Betrachter des fri:ihen
4. Jahrhunderts aus dem r5mischen StraEenbild vertraut
waren.

3. Goldglas mit Familiendarstellung aus Intercisa

(Dunajvaros). Budapest, Nationalmuseum. 1 •. Halfte
4. Jh.
Fiilep 1968, 4o1 ff., Taf. 1/1.
Der Innocentius benannte IVlann tragt eine Zwiebelknopf-
fibel und das Segmentum. Im Zusammenhang mit den aller-
dings zweihundert Jahre jiingeren Mosaiken aus S. Vitale
in Ravenna deutet J:tulep ihn als Hofbeamten, vielleicht
silentiarius. Diese Deutung ist wohl zu eng gefaEt. ·In
den Donaul~ndern sind Goldgl~ser eher selten, fiir ihre
Besitzer kann jedenfalls ein gehobener sozialer Status
angenommen werden. Intercisa war nicht autonome Stadt,
aber Garnison einer Auxiliartruppe. Der Dargestellte
gehorte also wahrscheinlich nicht zur l\1unizipalaristo-
kratie, sondern zum Offizierskorps.

4. Wandmalerei im Cubiculum M der Katakombe an der Via

Latina, Rom. Konstantin - Theodosius.
Ferrua 196o, 74, Taf. 72, 73/1. - K5tzsche-Breiten-
bruch 1976, 42ff., bes. 45, Taf 1 a.
Die Kontroverse um den Inhalt der Darstellung- nach
F'errua ( 196o, 7 4) wiirfeln Solda ten urn das Gewa:rid
Christi, nach Kotzsche-Breitenbruch (1976, 42ff.) wird
auf den Beruf des Grabinhabers als Rennstallbesitzer
hingewiesen - ist hier nicht von Bedeutung. Nach der
Fellmiitze der rechten Figur sind sicher Soldaten dar-
gestollt und die Zwiebelknopffibeln best~tiqen das. In
Stad trom kann wohl wieder an die Truppe der coho,Ete~.

}l..!'!>EE.9-~. p;edach t ·vJerden.

5. Wandmalerei im Mausoleum von Durostorum (Silistra).

Ende 4. Jh.

1102
 Grabar 1967 a, 159, Abb. 164.
Auf einer weiteren sepulkralen Wandmalerei ist der Grab-
inhaber des Mausoleums mit einer Zwiebelknopffibel dar-
gestellt. Wie schon bei dem gezeigten Goldglas hat er
das Segmentum, das auf ravennatischen Mosaiken Wtirden-
trager charakterisiert. Ihn mit der in Durostorum sta-
tionierten Legion in Verbindung z~ bringen ist ver-
lockend, aber unbeweisbar.

6. Brautkasten von Proiecta und Secundus. Silberschatz

vom Esquilin. London, British Museum. Urn 38o n. Chr.
Volbach- Hirmer 1958, 65 f., Abb. 116. - Grabar
1967 b, 3oo, Abb. 345. - Buschhausen 1971, 21o ff.
Nr. B 1, Taf. B 21 - 26. - Dinkler-von Schubert 197~

218.
Die Identifizierung des Mannes, der eine Zwiebelknopf-
fibel tragt, mit L. Turcius Secundus, der urn die Mitte
des 4.
Jahrhunderts Corrector von Picenum und Flaminia
war, wird bestritten 4 • Wegen der Kostbarkeit des Kast-
chens und wegen der in Gold angegebenen Fibel wird man
jedenfalls an einen hohen Wlirdentrager denken dtirfen.

7. Elfenbeindiptychon des Stilicho, Tafel B. Domschatz

1
in Monza. Urn 395 n. Chr.
Volbach- Hirmer 1958, 57, Abb. 62. - Grabar 1967 a,
16, Abb. 12 rechts. - Volbach 1976, 55f. Nr. 63,
Taf. 35.
Die Tafel zeigt den Sohn Stilichos, Eucerius, neben
seiner Mutter Serena. 'die sein Vater auf Tafel A des-
selben Diptychons tragt der Knabe eine Zwiebelknopffi~

bel, die dem Typ 6 nach Keller (1971, 26ff.) entsprich~

Dieser Typist im Fundbestand seltener als andere Type~

aber meist vergoldet, was ihn wohl als Rangabzeichen

kennzeichnet. Die Ubereinstimmung: der Fibeln von Vater
und Sohn hat Lanyi (1978, 261) als Bestatigung daftir
genommen, daB die Zwiebelknopffibeln Kennzeichen here-
ditarer Xmter waren. Gerade liber die Stellung der hier
dargestellten Personen wissen wir aber gut Bescheid.

1103
Stilicho war als magister utriusoue· militiae der rang-
hochste Soldat. In dieser hohen Machtposition und durch
seine Verschwtigerung mit dem Kaiserhaus hatte er die
Moglichkeit, sein~m Sohn im Kindesalter eine eintragli-
che Ehrenstellung zu verschaffen. Nach Zosimus 5, 34, 7
war Eucerius tribunus et notarius. Diesen Titel tragt
der zweite Beamte der schola notarioru~ (EnBlin 1937,
2453f. s.v. tribunus (11)), die die kaiserlichen Gehei~
sekretare umfaBte. Er hatte Rang und Einktinfte wie die
pro~~sules, was eine sehr ausreichende Versorgung be-
deutete. AuBer der Protokollfiihrung gehorten auch Ge~

sandtschaften und militarische Auftrage zu den Aufgaben

dieser Beamten. Zusammen mit dem Titel tribunus ist da-
mit wohl eindeutig eine Einordnung des Eucerius in den
militarischen Bereich gegeben, ja iiberspitzt ausge-
driickt zeigt das Diptychon genau den Titel des Knab~n:
Die Fibel bedeutet "tribunus" und die Schreibtafel ''no-
tarius".

8. Diptychon des·Rufius Probianus, Tafel B. Urn 4oo n.

Chr.
Volbach 1976, 54f. Nr. 62, Taf. 34.
Als Stellvertreter des praefectus praetorio hat der vi-
carius urbis Romae wieder einen militarischen Rang. Flir
den "notarius 11 links, der ebenfalls die Zwiebelknopffi-
bel tragt, gilt wohl dasselbe wie ftir den soeben ange-
ftihrten Eucerius. Weniger laSt sich tiber die "patricii"
im unteren Feld aussagen, die auch mit Zwiebelknopffi-
beln ausgestattet sind.

9. Aeneasdarstellung im Codex Vergilius Vaticanus, Pic-

tura 32. Urn 42o n. Chr.
de Wit 1958, 155, Taf. 4o/1.
Aus der Buchmalerei stammt dieses Beispiel, das den op-
fernden Aeneas im Purpurmantel, d.h. in der Tracht des
Kaisers, zeigt. Wegen des kleinen Formates ist die gol-
dene Zwiebelknopffibel nur sehr undeutlich angegeben,
aber als solche erkennbar. Der Kaiser hat als oberster

1104
Feldherr, wie schon b~i Galerius festgestellt, diese
Fibel getragen.

1o. Wandmalerei in der Katakombe S. Gennaro, Neapel.

Friihes 6. Jh.
Brenk 1977, 136, Abb. 67.
Diese Ausmalung einer Grabkammer zeigt den Familienva-
ter Theotecnus mit Chlamys und Segmentum, wozu offenbar
untrennbar die Zwiebelknopffibel gehort, di~ hier ein
besonders gut ausgepragtes goldenes Exemplar des Typs 6
bei Keller (1971, 26ff.) ist. Auch wenn die Inschrift
nichts naheres aussagt, kann man Theotecnus wohl nach
seiner Tracht als hohen Beamten ansehen.

11. Mosaik ins. Apollinare Nuovo, Ravenna. 1. Viertel

6. Jh.
Deichmann 1958-1976, Tafelband Taf. 2oo; 1, 316;
2/2, 176£.
Auf dem Mosaik mit der Vorftihrung Christi vor Pilatus
erscheint Pilatus und einer seiner Begleiter im Dienst-
kosttim des 5. bzw. 6. Jahrhunderts, das wieder durch
das Segmentum und durch den Mantel, der von der Zwiebel-
knopffibel gehalten wird, charakterisiert ist.

12. Codex Rossanensis, fol. 16. Rossano. 6. Jh.

Volbach- Lafontaine 1968, 183f., Abb. 52.
Auf eine weitere Pilatusdarstellung sei verwiesen, wo
das kleine Format der Buchmalerei wieder nur eine sehr
undeutliche Wiederg~be erlaubt hat.

13. APsismosaik in s. Cosma e Damiano, Rom. 526 - 53o

n. Chr.
Brenk 1977, 132, Abb. 37, 37. - Wilpert 1916, Taf.
1o6f. - Grabar 1967 b, 136, Abb. 146.
Mit dem hl. Theodor begegnet ein Heiliger mit Zwiebel-
knopffibel. Wie paBt dieser fromme Mann in die Reihe
martialisch-militarischer Gestalten? Der hl. Theodor
erlitt das Martyrium als Soldat im Heer des Kaisers
Maximianus (Weigert 1976, 447ff. s.v. Theodorus)! Er

1105
tr~gt, wie auch andere "Heilige Soldaten••, diese Fibel
also durchaus zu Recht.

14. Silberteller aus Lambousa, Nikosia, Cyprus Museum.

613 -63o n. Chr.
Rice 1959 Abb. 73. - Grabar 1967 b, 3o6 und 3o8,
Abb. 352, 355. - Vermeule 1976, 115ff. V 1, 7.
Dargestellt ist die Hochzeit Davids mit der T~chter des
Konigs Saul, vor dem die Zeremonie stattfindet. Sowohl
der Konig als auch der erfolgreiche F'eldherr David tra-
gen Zwiebelknopffibeln, die allerdings nur mehr recht
fltichtig angegeben sind.

15. Mosaik in H. Demetrios, Saloniki. 7. Jh.

Volbach - Hirmer 1958, 86f. Abb. 217. - Grabar
1967 b, 113 Abb. 143. - Volbach - Lafontaine 196?,
Farbtaf. III.
Der hl. Demetrios, der hier zwischen den Grtindern sei-
ner Kirche steht, hat noch die schwache Andeutung einer
goldenen Zwiebelknopffibel, die allerdings von der ebe~

falls goldenen Gloriole fast "verschluck~wird. Sie

kommt ihm als 11 Heiligem Soldaten 11· ikonographisch zu
(Myslivec 1974, 41ff. s.v. Demetrius). Beachtenswert
ist,;daB der Praefectus Leontius, bei dem man als hohem
Beamten ebenfalls eine Fibel erwarten wtirde, nicht die
typische Tracht mit Fibel und Mantel tr~gt.

16. Die Heiligen Sergios und Bacchos auf einer Ikone

vom Sinai. Akademie in Kiew. 6. - 7. Jh.
Grabar 1967 b, 186 Abb. 2o1. - VoJ.bach- Lafontaine
1968, 179 Abb. 42.
Die beiden Heiligen Soldaten sind durch die goldenen
Halsringe - "Manikia" - und die .Segmenta als BP.amte
oder Offiziere bezeichnet(Weigert 1976, 329f. s.v. Ser-
gius und Bacchus). Die drei Punkte bei der rechten
Schulter mochte ich als den ikonographischen Rest der
Zwiebelknopffibel ansehen, da diese Position genau der
'11rageweise unserer B'i beln entspricht. Verstanden wurde
dieses "Attribut Heilisrer Soldaterl1 aber nicht mehr.

1106
Zusammenfassend laBt sich also feststellen: Die schon
an den realen Fibelfunden erkannte Entstehung des Typs
(Jobst 1975, 95) amEnde des 3. Jahrhunderts im milita-
rischen Bereich zeigt sich auch an den Darstellungen
auf den christlichen Sarkophagen aus dem Anfang des 4.
Jahrhunderts. Soweit sich die Trager naher definieren
lassen, ist eine Einordnung in das Militar ~~glich. Ei-
nige Gra.bsteine aus der Zeit der Tetrarchie aus Aqui-
leia zeigen das noch deutlicher, z.B. der des cent~ri~

Flavius Augustalis (Rebecchi 1976, Kat.Nr. 7; Ubl 1976,

218 Nr. 2.1.2, Abb. 3), der als Offizier die charakte-
ristischen Ausrustungsstticke des Soldaten diokletianisch
-konstantinischer Zeit, Fellmlitze und Zwiebelknopffibel,
tragt. Besonders in der 1. Halfte des 4. Jahrhunderts
begegnen die Zwiebelknopffibeln fast ausschlieBlich bei
~

Personen nachweisbar soldatischen Charakters. Spater

kommen auch Darstellungen vor, wo eine solche Zuweisung
weniger sicherv aber jedenfalls m~glich ist. Die promi-
nentesten Beispiele, die Mosaiken mit Justinian und
Theodora inS. Vitale in Ravenna, zeigen Begleiter aus
.dem engeren Hofstaat. Welche Titel den Fibeltragern im
einzelnen zukommen, ist kaum festzustellen. Das Beispiel
dep tribunus et notarius Eucerius zeigt, daB militari-
sche Titel nicht unbedingt auf Angehorige der Truppe
beschrankt sein mtissen. Die scholae am Kaiserhof waren
z. T. militarisch organisiert und ihre Angehorigen tru-
gen eben militarische Dienstgrade bzw. Titel. Und mehr
zu diesen Titeln als zum militarischen Charakter der
Tatigkeit gehorte wohl die "Distinktion'' der Zwiebel- .
knopffibel.

Fur die chronologische Eingrerizung der Zwiebelknopffi-

beln ergibt sich eine wichtige Beobachtung. Nicht nur
im 4. und 5., sondern auch noch im 6. Jahrhundert wird
dieser Fibeltyp wiederholt dargestellt und zwar so klar
und deutlich, daB man anneh~en kann, die Ktinstler kann-
ten ihn aus eigener Anschauung. Erst die Darstellungen
aus dem 7. Jahrhundert zeigen den Typ unklar und ver-

1107
waschen. bzw. auf vollig unverstandene Reste reduziert.
Danach ware anzunehmen, daB die Fibeln noch im 6., aber
nicht mehr im 7. Jahrhundert getragen wurden. Gegentiber
den bisherigen Datierungen (Keller 1971, 26f.; Jobst
1975, 95) bedeutet das eine wesentlich langere Lebens-
dauer des Typs. Diese Diskrepanz erklart sich daraus,
daB sich die bisherigen Typologien und Chronologien
doch fast ausschlieBlich am Fundbestand an realen Fi-
beln aus den Rhein- und Donauprovinzen orientieren. In
diesen Gebieten ist aber schon aus historischen Uberle-
gungen nach dem 5. Jahrhundert nicht mehr mit reichs-
romischen Militars und Funktionaren zu rechnen. Nur an
gut datierten Fibelfunden aus dem byzantinischen Osten
und aus Italien lieBe sich das Auftreten der Fibel noch
im 6. Jahrhundert nachweisen, doch ist dieses Materi?l
in Mitteleuropa kaum bekannt. So mochte ich aus den
Darstellungen schlieBen, daB Zwiebelknopffibeln ein
Teil der Uniform-bzw. des Dienstkosttims des 4. bis 6.
Jahrhunderts waren.

Anmerkungen:
1 Konkrete Funde: Keller 1971, 26ff. (Stidbayern). -
Jobst 1975, 91ff. (Lauriacum). - Vag6 - Bona 1976,
167 (Intercisa, Stidostfriedhof). Darstellungen:
Belaev 1928, 112. - Wessel 1971, 347.
2 Das wurde an Fibelfunden aus Grabern in Stidbayern
(Keller 1971) und im norischen Teil Niederosterreichs
(Zabehlicky 1976, 462ff., bes. 469) festgestellt.
3 Besonders diese Tatsache hat Vago - Bona 1976, 167
bewogen, die Bedeutung der Zwiebelknopffibel als
militarischen Uniformbestandteil zu bestreiten und
einen groBeren Tragerkreis anzunehmen.
4 Nach alterer Literatur nimmt Buschhausen 1971, 21off.
Nr. B 7 diese Identifizierung und eine Verwandt-
schaft mit einem gleichnami·gen Konsul an. Dinkler-
von Schubert 1977, 218 bestreitet die Moglichkeit,
-den Gatten der Proiecta genau zu ·identifizieren.

1108
BIBLIOGRAPHIE:

Belaev 1928: N. Belaev, Die Fibel in Byzanz. Seminarium

Kondakovianum 3, 1928, 49ff. Dt. Zusammenfassung
1o2 - 114.
Brenk 1977: B.Brenk, Spatantike und frtihes Christentum.
Propylaen Kunstgeschichte, Suppl. Bd. 1, 1977.
Buschhausen 1971: H. Buschhausen, Die spatantiken Me-
tallscrinia und frtihchristlichen Reliquiare. 1. Teil:
Katalog. Wiener byzantinische Studien 9, 1971.
Deichmann 1958 - 1976: F. W. Deichmann, Ravenna, Haupt-
stadt des spatantiken Abendlandes. 1, Geschichte und
Monumente(1969). 2/1, Kommentar 1(1974) .. '2/2, Kom-
mentar 2(1976). Tafelband = Frtihchristliche Bauten
und Mosaiken von Ravenna(1958).
Deichmann 1967: F.W.Deichmann(Hsg.), Repertorium der
christlich-antiken Sarkophage. 1.Bd. Rom und Ostia.
Bearbeitet von G. Bovini und H. Brandenburg.
Delbrueck 1929: R. Delbrueck, Die Consulardiptychen und
verwandte Denkmaler. Studien zur spatantiken Kunst-
geschichte i~ Auftrag des Deutschen Arch. Inst. 2,
1929.
Dinkler-von ,Schubert 1977: E. Dinkler-von Schubert, Re-
zension von Buschhausen 1971. JbAntChr 2o, 1977,
215 - 223.
EnBlin 1937: W. EnBlin s.v. tribunus(11) in RE 2. Reihe
VI A 2, 2453f.
Ferrua 196o: A. Ferrua ,s.J., Le pitture della nuova
Catacomba di Via Latina. Monumenti di antichit~
Cristiana. 2. seria, 8, 196o.
Freis 1967: H. Freis, Dle cohortes·urbanae. Beih. BJb
21. Epigraphische Studien 2, 1967.
Ftilep 1968: F. Ftilep, Early christian gold glasses in
the Hungarian National ~!iuseum. ActaAntHung 16, 1968,
Llo1ff.
Grabar 1967 a: A. Grabar, Die Kunst des frtihen Christe&
turns. Universum der Kunst. (Deutsch) 1967.

1109
Grabar 1967 b: A. Grabar, Die Kunst im Zeitalter Justi-
nians. Universum der Kunst. (Deutsch) 1967.
Heurgon 1958: J. Heurgon, Le tresor de Tenes. 1958.
Jobst 1975: \'!. Jobst, Die romischen Fi beln aus IJauria-
cum. Forschungen in Lauriacum 1o, 1975.
Keller 1971: R. Keller, Die sptitromischen Grabfunde in
.Siidbayern. lViiJnchner Bei tr. zur Vor- und Ji'riihgeschich-
te 14, 1971.
Kotzsche-Brei tenbruch 1976: J,. Kotzsche-Brei tenbruch,
Die neue Kata~o~hF an der Via Latina in Ro~.

JbAntChr Brg.Bd. 4, 1976.

Lknyi 1978: V. L~nyi, RPzension von V~g6 - B6n~ 1976.
ActaArchHung 3o, 1978, 259ff.
Myslivec 1974: J. ~yslivec s.v. Demetrius 41ff. in
Lexikon d. ohristl. Ikonographie Bd. 6, 1974.
Noll 1974: R. Noll, :2~ine goldene "Kaiserfibel" aus
Niederemmel vom Jahre 316. BJb 174, 1974, 221ff.
Patek 1942: E. von Patek, Verbreitung und Herkunft der
romischen Fibeltypen in Pannonien. DissPann II/19,
1942. Deutscher ~ext 146ff.
Rebecchi 1976: F. Rebecchi, Le stele di eta tetrarchica
al Museo di Anuileia. Anuileia Nostra 47, 65 - 13o.
Rice; 1959: D. T. -nice, Kunst aus Byzanz. Ausstellungs-
katalog Mtinchen 1959.
Riisch 1969: A. Rtisch, Das kaiserzeitliche Portrait in
Makedonien. Jdi 84, 1969, 59 - 196.
Ubl 1976: H. Ubl, Pilleus Pannonicus, die FeldmUtze
des sp~tromischen Heeres. Festschr. Pittioni 2 =
ArchAust Beih. 14, 1976, 214- 241.
V~g6 - B6na 1976: E. B. V~g6 und I. B6na, Die Gr~ber­
felder von Intercisa 1. Der sn~tromische Slidost-
friedhof. 1976.
Volbach - Hjrmer 195R: ··,!. F. Volbach und M. 3irrr'er,
FrUhchristliche Yunst. Die Kunst der Sntitantike in
West- und Ostrom. 1958.
Volbach - Lafontaine 1g~s: ~;. F. Volbach una J. Lafon-
taine-Dosogne, Byzanz und der christliche Osten.
Prapyl~en Kunstgeschichte Bd. 3, 1968.

1110
Volbach 1976: W.F. Volbach, Elfenbeinarbeiten der Sp~t­
antike und des frlihen Mittelalters. 3. Aufl. 1976.
Weigert 1976: c. Wei~ert s.v. Ser~ius und Bacchus 329f.
und s.v. Theodorus Tiro von Euchaita 447ff. in
Lexikon d. christl. Ikonographie Bd. 8, 1976.
Wessel 1971: K. Wessel s.v. Fibel 546ff. in Reallexikon
zur byzantinischen Kunst Bd. 2, 1971.
de Wit 1958: J. de Wit, Die Miniaturen des Vergilius
Vaticanus. 1958.
Wilpert 1916: J. Wilpert, Die r~mischen ~osaikPn und
Malereien der kirchlichen Bauten vom 4. bis 13.
Jahrhundert. 1916.
Zabehlicky 1976: H. Zabehlicky, Die spatantiken und
v~lkerwanderungszeitlichen K~rpergraber aus dem
norischen Teil Nieder~sterreichs. Diss. Wien 1946.

Summary

Bulbous fibulae frequently appear in the finds and are represented

on many monuments. Their chronology and typology are fixed to some
degree. The group of the wearers and the time when they went out
of use is considered by examining some examples of representations
of these fibulae. Mostly they are seen as characteristic of military
personnel, and more generally of officers. Recently this association
with the military has been refuted. The earliest representations from
the end of the 3rd and first half of the 4th century A.D. show, almost
exclusively, soldiers of different rank (eg. on Christian sarcophagi).
From the later 4th to the 6th century A.D. magistrates and officers
of the court appear, so that a distinction betwP.en military and civil
service is not always possible. In addition, so.:..called 'saint soldiers'
also appear with bulbous fibulae. The predominance of soldiers in
. the earliest representations and the attribution to the 'saint soldiers'
affirm the origin of the bulbous fibula in the military uniform of the
time of Diocletian. Representations of the 7th century A. D. show
these fibulae in a rough and hardly recognisable manner. Probably
by this tlme these fibulae were no longer worn and the artists had to
copy ancient models without really understanding them.

1111