You are on page 1of 95

The problem of nomadic allies

in The roman near easT


*
Ulf Scharrer
To my daughter heloise sibylla,
from whom i learn so much.
Writing about a nomadic subject often causes a wide range of problems, which are
to be found mainly in terminology and in the nature of the evidence. This also ap-
plies to the subject of nomadic confederations in the near east under roman rule
up to the reign of constantine. in this essay i am going to outline some of these
problems, pose some questions and propose possible answers to these, which are
by no means meant to be conclusive, but which are supposed to point to possible
ields of future research. accordingly, i shall irst sketch briely the terminological
range of the concepts of nomadism, ethnicity and the notion of tribe, and against
this background i will try to provide a brief sketch of the problems of nomadism
and ethnicity of nomadic peoples in the roman near east. The second main issue
is the complex of relations between nomadic and sedentary peoples, which includes
friendly and hostile relatations. especially the latter will lead to the debated ques-
tion of a nomadic menace. against this background, the inal and longest section
will be devoted to the problems of nomadic allies themselves.
nomadism in modern scholarship
The irst problem may at irst glance seem to be rather redundant: when we speak
about nomads and nomadism, what do we actually mean by that? Generally speak-
ing, terms belonging to the semantic ield of the word nomadism today are used in
a greater variety of senses than might be expected: they are for example employed
often rather metaphorically as it seems as names of systems of mobile and wire-
less computing;
1
for mobile robots;
2
for the contemporary mobility on the labour-
*
i owe thanks to Ted Kaizer for the invitation to oxford and to the participants of the conference
for a valuable discussion of my paper. i also wish to thank Ted for the careful editing of this
paper. i alone am responsible for all shortcomings of this essay.
1 e.g. bender e.a. (2000); hsi (2003); Kwok and lau (2002); Kylnp e.a. (1996); mackowiak
(2000); makimoto e.a. (2001); mcKnight e.a. (2004); ramachandran (2002), p. 621; sand-
holm and huai (2000); sawhney and schmandt (2000). see also the subtitle of the contem-
porary journal IEEE Personal Communications. The Magazine of Nomadic Communications
and Computing.
2 e.g. Gvozdjak and li (1998).
242 Ulf scharrer
market, be it physical
3
or psychical;
4
in physics for some kind of nuclear vortices;
5

in biological sciences the term sometimes is employed in generic names, e.g. Peri-
clemenes nomadophilia for some sort of shrimps, Rhopilema nomadica for the
mediterranean jellyish or Nomada Scopoli for a certain sort of bees, in addition as
characterisation of certain spiders
6
or generally of polar bears;
7
furthermore for the
mobility of royal courts;
8
for christians following the traces of God;
9
and inally in
contemporary french philosophy and literary theory we ind the concept of no-
madic singularities
10
and the idea of la nomadisation de la raison.
11
although
these examples vary widely in their application of the term nomadism, they all
have in common that the emphasis clearly is on the aspect of mobility.
12
however, in ethnographic contexts the deinitions and applications of the term
nomadism vary to some extent. in this respect there seem to be nearly as many
ideas of nomadism as there are works on the topic.
13
broadly speaking two main
concepts can be distinguished:
14
the irst concept apparently follows the just men-
tioned emphasis on mobility and deines nomadism as a special kind of economy
which is marked, besides mobility, by a certain degree of social coherence and or-
ganization and a close mutual relationship to the sedentary world.
15
since this no-
tion does not necessarily imply a primarily pastoral economy, it also includes non-
pastoral but mobile groups,
16
such as gypsies
17
and hunter and gatherer societies.
18
3 e.g. heute hier, morgen fort in Der Stern (24/2004), p. 17081, esp. p. 17980. see on rus-
sian scientists outside russia: Janositz (1995). see also the somehow strange book of schlgel
(2006), esp. p. 10815.
4 e.g. Welsch (1994), esp. p. 74850.
5 e.g. bishop (1994).
6 e.g. Kunz (2001).
7 rosing and frenz (1999).
8 e.g. briant 1988; leibetseder (2004), p. 127.
9 mironez diez (2001), esp. p. 252.
10 deleuze (1969), p. 12431, with Taureck (1988), p. 16871. The concept of deleuze is em-
ployed e.g. by braidotti (1994) and roy (2003). see also reid (2003); schmitz (2002); Gh-
aussy (1999); pels (1999) and braidotti (1999).
11 borreil (1993). see also the characterisation of m. foucault as nomadic mind by Kohler (1992),
p. 157.
12 see also Guldin (2002), p. 378.
13 see Guldin (2002), p. 379; lyonnet (2004), p. 25; scharrer (2002b), p. 288; scharrer (2002c),
p. 167; scholz 1995, p. 1925.
14 see orthmann (2002), p. 139.
15 see the programmatic title of rao (1987): The other Nomads. This notion is followed e.g. by
streck (1992); streck (2002); leder (2002), p. 1113. see also Gertel (2002), p. 59; Knauf
(1992), p. 634. on similarities see porter (2004), p. 69.
16 see the contributions in rao (1987); mieck (1982), p. 504. see also Knauf (1992), p. 636.
17 e.g. leder (2005), p. 19; marushiakova and popov (2002a); marushiakova and popov (2002b),
esp. p. 356; sigona 2003.
18 e.g. Knauf (1992), p. 634; streck (1992). furthermore see along sega: die heutige regie-
rung ist schlecht, sie zerstrt unser landand sarawak: Unruhe im nomadenland in Pro Re-
genwald News-Letter 18, September 2002, p. 34. This notion is also employed in some hi-
story school books, e.g. abelein e.a. (2004), p. 33; cornelissen e.a. (2004), p. 172; Tatsch and
regenhardt (2004), p. 29. cf. Khazanov (1994), p. XXXViiiiX.
243 The problem of nomadic allies in the roman near east
The second notion of nomadism is more narrow: it implies a mutual depen-
dence between mobility and pastoralism. according to this concept nomadism may
be distinguished from pastoralism by the mobility of ethnic groups or at least sub-
groups, and by the fact that other pastoralists generally are settled. furthermore
nomadism can be discerned from other forms of mobility by pastoralism as the
main economic basis.
19
in this very sense mobility based on pastoralism the
word nomad and related terms are, with only very few exceptions, widely used by
ancient authors from hecataeus of miletus to ammianus marcellinus and beyond.
20

for two reasons in the present context i prefer to use the term in this narrow conno-
tation rather than in the wider sense: irstly this apparently corresponds to the Greek
and latin ethnographic terminology, and secondly the expanded concept of noma-
dism seems to run the risk of becoming too vague and hence in consequence it is
not really useful to deine nomadism as a very special form of economic and cultu-
ral life.
however, following the narrow notion of nomadism there are still a good
number of problems remaining. With respect to the present topic two of these
should be singled out, which are in any case matters of fundamental dispute within
nomadic scholarship generally. first there is the question of which categories are
useful to classify nomadic phenomena. scholarship has suggested different criteria
of classiication,
21
above all the animals that are bred and the geographical zones
and regions. as regards the irst criterium, with respect to the near east the main
distinction often made is that between the herding of sheep and goat on the one
hand and of camels on the other, implying two different degrees of nomadic life.
Whereas nomadic camel-breeders could move into the desert, sheep- and goat-
herding would only be possible in regions that provide suficiant water and graz-
ing-grounds.
22
This leads to the second criterium, namely that a distinction has to
be made according to the natural surroundings. often a distinction is made between
nomadic groups living mainly in desert or steppe regions and those dwelling in
more fertile territories.
23
Within this framework, the arabic term badw, bedouin,
is applied to those nomadic groups whose main economic basis is the camel and
who basically live in the desert or steppe.
24
19 e.g. bar-Yosef and Khazanov (1992b), p. 2; chatty (2002), p. 23; Knig (1981), p. 27; patai
(1951), p. 401; paul (2003), p. 29; p. 33; scholz (1995), p. 2032; scharrer (2002b), p. 288
291, with further references. see marx (1992), p. 255.
20 on the ancient terminology see briant (1982), p. 129; Khazanov (1994), 16; scharrer (2002c),
with further references. see also amm. marc. 31.2.179 on the Halani. see also scharrer
(2002b), p. 289; scharrer (2004) p. 312; stark (2002), p. 364.
21 e.g. scholz (1995), p. 67. see orthmann (2002), p. 138; scharrer (2002b), p. 293.
22 bacon (1954), p. 446; donner (1999), p. 223; parker (1991), p. 500; paul (2003), p. 31;
scholz (1995), p. 606; p. 834; p. 1035. see isaac (1998f), p. 412. see also brentjes (1986),
p. 878.
23 e.g. Khazanov (1994), p. 4069. cf. macdonald (1991), p. 1023. against this view: orth-
mann (2002), p. 138.
24 on the term badw / bedouin see e.g. bacon (1954), p. 44; dhont (2004), p. 14; Knauf (1992),
p. 635; Kuhnen (1991), p. 326; Kupershoek (2001), p. 534; orthmann (2002), p. 140; 1489;
244 Ulf scharrer
all this touches on the second problem singled out here. a widely discussed
question in scholarship is whether nomads could be classiied according to their
level of nomadism, i.e. whether it is possible to speak of pure nomadism or diffe-
rent degrees of semi-nomadism.
25
included in this complex are the aspects of the
relations between settled and non-settled populations, as well as nomadic strategies
to get other goods than those being provided by their animals.
26
With respect to
this, Khazanov argued that in reality pure nomads, in the sense that they are eco-
nomically self-suficient, do not exist and never have existed: in addition to those
goods they get from their animals they are in need of other commodities, which
they can only get from interaction with the sedentary world. for this purpose, no-
madic groups follow different strategies, namely seasonal agriculture of a whole
ethnic group, full scale farming of a part of an ethnic group, trade, services in ex-
change for goods, and inally, as an important economic strategy, the raiding of
settled areas.
27
The close interaction and even symbiotic relationship between no-
madic and settled populations is currently assumed in scholarship,
28
as is the de-
pendence of nomadic groups on the sedentary world.
29
eThniciTY in modern scholarship
i will now turn to the earlier mentioned term ethnic group. When the concept of
ethnicity was developed in the nineteenth century, it denoted primarily a more or
less static group with common characteristics, such as religion, the assumption of
common descent, social hierarchy, common language, common identity etc. as
such, the notion is still widely used today.
30
however, the most signiicant aspect of
the concept that has changed is the idea that ethnic groups are not actually unchang-
EIMW, p. 107 s.v. Bedouin (davis). cf. chatty (2002), p. 3; rets (2003), p. 23; sweet
(1965), 11337; scharrer (2002b), p. 2934, with further references.
25 see Gertel e.a. (2002); hoyland (2001), p. 8990; isaac (1998f), p. 413; robinson (1996),
p. 4312. see also the debate in the time of the french mandate in syria: lange (2002), esp.
p. 157.
26 see scharrer (2004), p. 311.
27 Khazanov (1994), esp. p. XXXi; 3; 345; 6984; 198224, followed explicitly e.g. by dever
(1992), p. 88; scholz (1995), p. 302; stark (2002), p. 3712; staubli (1991), p. 12. see fur-
thermore donner (1999), p. 26; orthmann (2002), p. 16072; paul (2003), p. 356; rothstein
(1899), 1267; scharrer (2002b), p. 292, with further references.
28 donner (1999), p. 246; dostal (1989), p. 3941; isaac (1998f), p. 4123; hoyland (2001),
p. 8990; Khler-rollefson (1992), p. 11; marx (1992), p. 2567; offer (2003), p. 20; orth-
mann (2002), p. 15697; paul (2003), p. 29; 33; robinson (1996), p. 435; stark (2002), p. 366
7; scharrer (2002b), p. 292, with further references; scharrer (2004), p. 311, with further refe-
rences.
29 chatty (2002), p. 1; Gichon (1991), p. 321; parker (2002), p. 77; stark (2002), p. 366; staubli
(1991), p. 14; sweet (1965), p. 113740; Zagarell (1988), p. 357.
30 bloch-smith (2003), p. 4023; brass (1996), p. 856; connor (1994), p. 1003; enloe (1996),
p. 1978; fishman (1996), p. 634; hartwig (1988), p. 1424; hutchinson and smith (1996a),
p. 37; nash (1996); pollard (2000), p. 111; schermerhorn (1996), p. 17; schneckener (1998);
sollors (1996a), p. xi-xii; Tschpe (1988), p. 137; Weber (1996), p. 5660; Wenning (1993),
245 The problem of nomadic allies in the roman near east
ing, but that ethnicity rather is a permanent process, with changing identities and
the genesis of new ethnic groups.
31
some groups can thus merge with other ethnic
groups by taking over their identity, or by forming a new identity together with the
other group, but it is still possible that some groups preserve traits of their identities
to different degrees.
32
Within this complex the concept of tribe needs to be discussed briely, as a
tribal structure is generally regarded as one of the main characteristics of nomadic
societies.
33
There has been written much about tribalism, and some remarks may
fall into the category of tribe and prejudice, one of them being that tribalism pre-
supposes nomadism.
34
Generally the same characteristics that in scholarship are
applied to ethnic groups are regarded as characteristics of the tribe. The most im-
portant aspect, however, is the claim of a common ancestry and thus the supposed
kinship of the members of the tribe.
35
as such a tribe can have a number of clans as
social sub-units.
36
Widely discussed is also the question of the political organiza-
tion of a tribe. on the one hand it is claimed that tribal social structure in general is
homogeneous, with hardly any social stratiication, and that if there is a chieftain,
the latter is without much real force.
37
basically, tribes are regarded not as stable
ethnic units, but also as changeable.
38
although, as i have argued, tribalism is basi-
cally regarded as the characteristic feature of nomadic social life, it is clear that
tribalism is not restricted to nomads and other mobile ethnic groups, but can be
found also in settled communities.
39
To draw a larger picture, two main concepts
have been introduced, namely the notions of segmentary and dimorphic societies.
segmentary societies are marked by different equal social groups without a central
instance of power.
40
according to the concept of the dimorphic society, which has
p. 558. see also barth (1996), p. 296; heckmann (1997), 503; isaac (1998c), p. 25960;
Jenkins (1997), p. 913; p. 1920; Kohl (1998), p. 2702; pohl (1998).
31 barth (1996), p. 30709; 31820; Giordano (1997), p. 605; Kohl (1998), p. 2759. see also
cohen (1996b), p. 374; Jenkins (1997), p. 13; 40; Jenkins (1999), p. 88.
32 bloch-smith (2003), p. 4034; hartwig (1988), p. 1434; Tschpe (1988), p. 1378.
33 Knig (1981), p. 289; scholz (1995), p. 31. see marx (1992), p. 2556; rets (2003), p. 113;
Tanner (1997), p. 7; Zagarell (1988), p. 357.
34 against this view e.g. Kaizer (2002), p. 53, followed by scharrer (2003), p. 734. cf. macdo-
nald (1993), p. 3523; Tanner (1997), p. 78.
35 donner (1999), p. 28; fleckenstein (1988), p. 237; hoyland (2001), p. 1135; Jenkins (1997),
p. 178; Khazanov (1994), p. 140; Knig (1981), p. 289; Kupershoek (2001), p. 23; orth-
mann (2002), p. 2035; paul (2003), p. 28; peters (1999a), p. xiv. on different notions of tribe
see Tanner (1997), p. 56.
36 Giddens (1986), p. 53; Knig (1981), p. 289; fleckenstein (1988), p. 1921; schneider
(2003a), p. 1723. cf. connor (1994), p. 1078.
37 Giddens (1986), p. 53; hoyland (2001), p. 11920.
38 cohen (1996a), p. 8384; marx (1992), p. 258; stark (2002), p. 369; fleckenstein (1988),
p. 2731. on chimpanzees forming new tribal groups, see scharrer and petrischak (2000).
39 Khazanov (1994), p. XXXVi; isaac (1998b), p. 155; Knauf (1992), p. 637; macdonald (1993),
p. 3523; robinson (1996), p. 433.
40 brggemann (1988), p. 7201; dijkstra (1995), p. 13; Khazanov (1994), p. 1447; Kohl
(1998), p. 2734; orthmann (2002), p. 2058; paul (2003), p. 28; sigrist (1994), p. 301. see
246 Ulf scharrer
been developed by rowton,
41
there is an institutional and social interaction bet-
ween sedentary and nomadic worlds, with a chief mostly living in a village or a
town and part of a society being nomads and another part being sedentary. both
concepts are thus to a great extent applicable to the political and social relations
between settled and nomadic groups.
nomadism and eThniciTY in The roman near easT
against this theoretical background i will now turn to the issue of nomadic peoples
and their ethnicity in the roman near east. it must be noted that, when later in this
essay the nomadic relations to the roman authorities will be analysed, the problem
of the nabataeans will not be discussed in the context neither the question to what
extent they may have been nomadic, nor the roman-nabataean relations as both
aspects have been discussed extensively.
42
evidence for nomadism in the roman near east can be found in archaeologi-
cal material, in Greek, latin and islamic authors, and inally in Greek, latin and
semitic inscriptions. each group of sources has speciic problems of interpretation.
regional surveys in north-arabia have revealed countless numbers of stone circles
and enclosures, which apparently are remains of nomadic dwellings. Unfortunately
these remains are dificult to date and can mostly not be ascribed to a speciic ethnic
group. sometimes the building remains are supplemented by ceramics, which re-
veal periods of use.
43
in any case, in most instances conclusions cannot be drawn
with regard to the ethnicity of the relevant groups: most of the ceramics are naba-
taean and this shows that there were contacts, but it reveals nothing about the iden-
tity or structure of the people using the enclosures. fortunately these are often cov-
also hoyland (2001), p. 115. on the history of research on segmentary societies see sigrist
(1994), p. 219. see the critical remarks of porten (2004), p. 70.
41 e.g. rowton (1976). The concept is followed e.g. by dever (1992), p. 86; dijkstra (1990),
p. 903; hauser (1998), p. 5134; sommer (2003a), p. 112; sommer (2005), p. 957. on the
concept see also briant (1982), p. 48; rets (2003), p. 1145; staubli (1986), p. 145; scharrer
(2002b), p. 2956, with further references including citicism of the concept.
42 on the nabataeans see e.g. altheim and stiehl (196469) i, p. 3140; p. 6579; p. 281304;
ball (2000), p. 6073; bowersock (1983), p. 1227 and p. 5775; bowersock (1994b), p. 105
12; briquel-chatonnet (1995b); butcher (2003), p. 968; dijkstra (1995), p. 3474; dussaud
(1955), p. 2170; Gawlikowski (1995b), p. 857; Graf (1997c); hammond (1973); healey
(1989); hoyland (2001), p. 704; levi (1989), p. 36; millar (1993), p. 4008; parker (1986a),
p. 11522; sartre (1982a), p. 12832; sartre (1985), p. 5168; sartre (1989), p. 1423; sartre
(2005), p. 1623, p. 806, p. 2378 and p. 2512. The idea of a nabataean bedouin-state has
been proposed by Knauf (1992), p. 6378, and has in my view reasonably been challenged
by macdonald (1991); see also Kuhnen (1991), p. 3279. on roman-nabataean relations see
bennett (1997), p. 175; bowersock (1983), p. 348 and p. 4956; funke (1989); sartre (1985),
p. 51.
43 see e.g. adams e.a. (1977), p. 369; King (1990), p. 589; parr e.a. (1978), p. 445; Zarins e.a.
(1979), p. 315. on the archaeological interpretation of material remains of nomads, see gene-
rally banning and Khler-roleffson (1992); Guldin (2002); lyonnet (2004); nser (2005).
247 The problem of nomadic allies in the roman near east
ered with rock drawings, which often allow a rough classiication.
44
This is espe-
cially due to the grafiti which in great numbers accompany the drawings and in
still greater numbers stand for themselves. They provide invaluable evidence for
the everyday life, the problems and the social and economic structures of nomadic
groups. still, their interpretation in many resprects is not always as unambiguous as
one might assume at irst glance, as the remarks below attempt to show. as i will
shortly argue in more detail, one of the problems is how the epigraphic and ar-
chaeological material could correspond to the litarary evidence provided by Greek,
latin and islamic authors. These themselves are often problematic: especially much
of the islamic material is interlaced with legends, and often it is not always clear
whether the statements of the authors Greek, latin and islamic apply to the pe-
riod they are writing about, and accordingly to what degree they simply retell evi-
dence from much older sources, and to what degree they follow certain literary
traditions and hence use literary topoi in describing foreign peoples.
45
in literary and epigraphic texts from the irst two centuries of roman rule in the
near east numerous explicit references are made to nomades and scenitae, tent-
dwellers.
46
according to strabo and pliny nomads can be found in mesopotamia,
47

and especially on the right bank of the euphates,
48
from the region of Thapsacus
49

up to the persian Gulf.
50
furthermore nomads are mentioned as dwelling between
the euphrates and coele syria
51
and east of apameia
52
and of the Jordan.
53
finally
nomads are recorded as dwelling in syria
54
and of course in northern arabia.
55
The
literary evidence for the presence of nomads in syria and northern arabia is sup-
44 on rock drawings see anati (1972).
45 on ancient authors on nomadism, see boshnakov (2003), p. 1149; briant (1982), p. 1140;
capomacchia (2004); Graf (1989a), p. 3556; hoyland (2001), p. 967; Khazanov (1994),
p. 69; levi (1989), p. 33; scharrer (2004), with further references; schmitt (2005), p. 417;
stark (2002), p. 364. on ancient ethnography see e.g. briant (1982), p. 101; schneider (1998),
p. 956 and p. 10610; Zst rov (1984); scharrer (2002a), p. 13944; scharrer (2002c),
p. 1689, each with further references. on pliny see macadam (1989), p. 2901; macadam
(1989), p. 2957; rets (2003), p. 332; sallmann (1971); on strabo see brodersen (1993);
drijvers (1998); Winters (2000); scharrer (2002c), p. 197, with further references. strabos use
of earlier sources is exempliied by boshnakov (2003); see also sartre (1989), p. 140. on di-
odorus see mayer (2000); on ptolemy see macadam (1989), p. 3015; Wallace (2000),
p. 1419. see also bowersock (1973), p. 1356; macdonald (2001), p. 251. on byzantine au-
thors see dostlov (1990), p. 1712; Zst rov (1990).
46 on the term skenitai/scenitae see scharrer (2002c), p. 1979. Wolff (2003), p. 135, presents
them wrongly as tribe. on akkadian sources see ephal (1982), p. 101.
47 strabo 16.2.11; 16.3.1.
48 plin. HN 6.125; 143; strabo 16.1.28.
49 plin. HN 5.87; strabo 16.1.8
50 plin. HN 6.125; 148; 151; strabo 16.1.26.
51 strabo 16.1.27.
52 strabo 16.2.11.
53 plin. HN 5.72.
54 strabo 16.4.2; plin. HN 5.65.
55 see e.g. strabo 16.4.2; diod. 2.54.1. it is debated, whether diodorus excurse on arabia is a
quotation from posidonius. see scharrer (2002a), p. 144 with further references.
248 Ulf scharrer
plemented by a couple of Greek inscriptions which explicitly mention nomads,
56

and a latin inscription from Khan Kosseir probably referring to scenitic arabs.
57
an invaluable source of evidence in this context are the so-called Thamudic
and safaitic inscriptions, the former being found mainly in the northern-western
part of the arabian peninsula, the latter in syria east of the Jordan. Two of the main
themes in the safaitic inscriptions are migration and the breeding of animals: many
grafiti mention somebody making a journey,
58
passing through,
59
leaving,
60
arri-
ving
61
or simply being temporarily present
62
or camping
63
at a place, and moving
camp.
64
numerous grafiti testify claims of ownership of camping-places,
65
of
pastures,
66
of ields,
67
of valleys,
68
of water places,
69
of different kinds of buil-
dings
70
and of other places,
71
among them remarks that a certain place belongs to
56 OGIS 421,5 (= IGRR iii 1136; moors (1992), p. 316 nr. 3a); 616 (= IGRR iii 1247); IGRR iii
1254 (= PAAES iii 378; moors (1992), p. 317); PAAES iii 383; PUAES iii a 751 (= moors
(1992), p. 316 nr.3b). These inscriptions are published and commented by sartre (1982a),
p. 1216. see also moors (1992), p. 3156. other inscriptions mentioning nomads: Inv. X 44,4
(= PAT 1378; asad and Yon (2001), nr.16; SEG Vii 139 (Greek section only); Kaizer (2002),
p. 478; schuol (2000), p. 846, with comments; hartmann (2001), p. 523); dussaud and
macler (1901), p. 147 nr. 7; macdonald (1995), p. 99.
57 CIL iii 128.
58 e.g. CIS V 313; 320; Wh 238; 1846; 2023; 20601; 2831; 2066; 2071; 2086; 2502; 2700;
3791; PUAES iV c 353; 701; 1006; 1190; ISB 18; 130; 400; PAAES iV 5,12; 17; 23; SIJ 37; 54;
329; 452. see also Wh 2181.
59 e.g. CIS V 27; 28; 3961; 4824; Wh 149; 741; PUAES iV c 448; SIJ 499.
60 Wh 19; 158; 825; 1200a; 1230; 1241; 1243; 1768; 2124; 2322; 2357; 3237; PUAES iV c 160;
676; SIJ 152.
61 CIS V 62; 99; 112; 278; 329; 1065; 2310; 2405; 2459; 2497; 2579; 2649; 2746; 3131; 3171;
4417; 4757; SIJ 80; 714; 840; Wh 648; 2129; PUAES iV c 7; 217; 245; 254; 257; 299; 434;
712; 719; 720; ISB 58; PAAES iV 5,4.
62 PAAES iV 5,134; SIJ 873; Wh 2242; 3792a.
63 ISB 146; Wh 158; 411; 610; 792; 1096a-b; 1701; 1994; 2052; 2706; 2803; PUAES iV c 29;
146; 161; 162; 169; 183; 184; 189; 258; 361; 369; 379; 396; 553; 568; 571; 574; 635; 641; 644;
658; 659; 753; 1025; 1028; 1109; 1169; 1181; 1182; 1195; PAAES iV 5; 114; SIJ 818. a cam-
ping place (dr) is mentioned e.g. in Wh 69; 1225. see also below, n. 77.
64 e.g. CIS V 307; SIJ 184; Wh 37475; 377; 496; 707; 3136.
65 e.g. ISB 2; 117; 120; 125; 210; 367; 437; Wh 346; 349; 363; 415; 704; 766; 798; 808; 814;
864; 924a; 1065; 1071; 1107; 1119; 1133; 1193; 1230; 1339; 1365; 1747; 1754; 1755; 1797;
1885; 1924; 1966; 2293; 2473a; 2499; 2558; 2539b; 2644; 2733; 2738; 2742; 27423; 2832;
2854; 2859; 2863; 2867; 2900; 2907; 2934; 2937; 2971; 3049; 3262; 3349; 3366; 3381; 3412;
3527; 3636; 3812; 3859; SIJ 370 (?). see also Wh 3708.
66 e.g. Wh 30.
67 e.g. PUAES iV c 1290.
68 e.g. SIJ 712; Wh 1638; 3335; PUAES iV c 534.
69 e.g. Wh 1002; 1503c; PUAES iV c 853; 904.
70 stone enclosure (wd): 1103a; PUAES iV c 219. shelter: SIJ 296; 948; Wh 405406; 564;
2188; 3017; 3050; 3094; 3112a; 3186; 3208a; 3435b; 3442; 3597; 3736b; 3840; ISB 32; 429.
house: Wh 1322; 2456; PUAES iV c 949 ISB 20. halting-place: Wh 3428. building in gene-
ral: Wh 967; 3098. rock: SIJ 964; Wh 1753; 1807; 3037; 3050; 3082a; 3736b; 3840. mound:
SIJ 484.
71 ISB 296; Wh 1411; PUAES iV c 38; 199; 229; 319; 321; 758; 827; 851; 872; 890; 1019; 1163;
1164; 1180; 1214. place in general: SIJ 610; 997; Wh 2805.
249 The problem of nomadic allies in the roman near east
somebody year after year (m fm).
72
The temporary presence could be a stay
overnight
73
or the residence for a whole season.
74
accordingly, some grafiti men-
tion a spring encampment,
75
pitching of tents,
76
or the ownership of a camping-
place year after year.
77
from other inscriptions that mention (or pray for) return
78

it is obvious that most of the migrations were not linear, i.e. not in one direction.
The latter aspect leads to the topic of the breeding of animals. some grafiti
mention the seeking of pastures
79
or the following of animals.
80
sometimes the
animals that are bred or pastured are not speciied,
81
but often they are. mention is
made of the tending of sheep,
82
of goats,
83
of both,
84
of cattle,
85
of horses
86
and of
camels.
87
accordingly, these animals are often depicted in the grafiti.
88
most often
72 e.g. Wh 1916; 2717; 2998; 3757; 3800.
73 e.g. CIS V 230; 404; 1095; 2031; 3568; 4987; 5050; PUAES iV c 350; 595; PAAES iV 5,123;
SIJ 139.
74 spring: SIJ 705; 1008; Wh 181; 325; 403; 790; 1202; 1232; 1271; 3923; PUAES iV c 374;
465; 1103; 1291; lidzbarski (1908), p. 42; PAAES iV 5,5. summer: SIJ 104; Wh 289; 298;
423; 465a; 622/23; 1191; 1765; 1996a; 2399; 3049; 34234 3850; 3787; Khraysheh (1995),
nr. 6. Winter: CIS V 681; 2829; 4810; Wh 1361; PUAES iV c 649; SIJ 252; 1008.
75 Wh 766.
76 e.g. PUAES iV c 344; Wh 641a.
77 Wh 1193; ISB 2; SIJ 171.
78 e.g. SIJ 836; Wh 184; 387; 711; 799; 822; 927; 1021; 1023; 1024; 1373b/c; 1519a; PUAES iV
c 1213.
79 PUAES iV c 379; PAAES iV 5,110; dussaud (1907), p. 101 (dussaud and macler (1901),
nr. 161).
80 ISB 13.
81 e.g. CIS V 34; 155; 207; 269; 272; 275; 362; 433; 586; 652; 686; 892; 1138; 1167; 1172; 1235;
1239; 1667; 2003; 2005; 2166; 2206; 2213; 2299; 2391; 2570; 2758; 2926; 2941; 3131; 3020;
3181; 3234; 3290; 3526; 4027; 4303; 4394; 4435; 4717; 5044; ISB 112; 250; SIJ 170; 218;
473; 673; 744; 1006; Wh 26; 64; 74; 75; 156; 171; 319; 376; 419; 585; 597; 710; 746; 784;
799; 925; 1066; 1120; 1199; 1231; 1361; 1413 (?); 1699; 1771; 1772; 1867a; 2042; 2072;
2114; 2139; 2142; 2149; 2468; 2532; 2654; 3104; 3396; 3636; 3691; ISB 50; 79; 82; PAAES iV
5,135; Khraysheh (1994). slaughtering: PUAES iV c 290; 297.
82 e.g. CIS V 275; 744; 847; 1534; 1993; 2163; 2481; 2552; 3143; 3216; 3577; 4443; 4987; Wh
578; 582; 1023; 1282; 2036; 2808; 3534c; PUAES iV c 326; 331; 344; 355; 562a; 674; 1134;
SIJ 156; 181; 263; 392. on sheep see eissfeldt (1954a), p. 1035.
83 e.g. CIS V 1276; 2286; 2311; 2647; 3235; 3260; 4148; 4772; 4973; 4989; 5050; SIJ 153; 688;
807; Wh 19; 159; 1410; 1673; 1682; 169798; 1725a-b; 2161; 3562; PUAES iV c 157; 351;
360; 419; 722; 729; 1064; PAAES iV 5,122; Zeinaddin (2000), inscr.11. on goats and noma-
dism see scholz (1995), p. 556.
84 e.g. Wh 1141; 1246a; 1900; 2119; 2165; 2802; PUAES iV c 297; 887; 1133.
85 e.g. CIS V 310; 1216; 1871; 2713; 3064; 4392; PUAES iV c 90; 159; Zeinaddin (2000), in-
scr. 4. on cattle and nomadism see scholz (1995), p. 567.
86 e.g. CIS V 3186; 4446; 4855; PUAES iV c 664; SIJ 745; Wh 2803; 3049; 3735; King (1990),
p. 64. see also NSI 109.
87 e.g. CIS V 88; 98; 272; 1338; 1456; 1486; 1783; 1906; 2160; 2167; 2459; 2556; 2943; 3020;
3230; 3961; 3975; 4233; 4815; 4933; 4985; 5089; 5178; 5182; 5288; Wh 42; 161; 238; 325;
348; 581; 806; 1023; 1198; 1308; 1899; 1996a; 2006; 2044; 2145; 3093; 3500; PUAES iV c 4;
146; 179; 357; 406; 448; 613; 656; 698; 701; 968; 1214; 1291; 1873; 3405; 3435a; SIJ 18; 118;
201; 263; 295; 323; 806; ryckmans (1939), nr. 34.
88 drawings of: Goats: e.g. Wh 343. sheep: e.g. PUAES iV c 1179. bull: e.g. Wh 3675b.
250 Ulf scharrer
we ind the drawings of camels
89
and within this category of she-camels.
90
This
corresponds to the fact that she-camels are sometimes speciically mentioned in the
grafiti.
91
as one inscription reveals they provided milk,
92
which may exclear their
predominance amongst the drawings. remarkably, hardly a drawing of a camel ri-
der can be found amongst the safaitic grafiti. on the other hand there are quite a
number of drawings of riders on horses and on donkeys. There are numerous dra-
wings of riders armed with lances, either ighting each other or ighting an animal,
or just being alone.
93
however, horses are not only presented in violent actions:
some inscriptions mention horses following the livestock.
94
The importance of
horse and donkey appears then to be remarkable, especially since the importance of
horses for near eastern nomadic life has sometimes been denied.
95
some inscripti-
ons simply refer to being on the look-out for horses.
96
others refer to the keeping of
donkeys,
97
and quite a couple of drawings depict donkeys and horse in situations
other than warlike.
98
89 CIS V 1005; 1148; 1215; 1552; 1615; 1658; 1852; 1855; 2662; 2773; 3595; 36323; 3659;
4009; 4011; 4631; 4670; 4835; 5042; ISB 367; Wh 57; 370; 424; 472; 47677 (bactrian ca-
mels! see also: SIJ 360); 516; 642; 731; 739; 805; 911; 1228; 1266 (?); 1502; 1587; 1918;
200911a; 2012; 1278a; 1762; 3314; 3760; 3850; 3912b; PUAES iV c 7375; 8082; 312;
420; 1002; 1104; 1294; 1298; PAAES iV 5,5; 14; SIJ 60; 258; 341; 415; 424; 616; 803; Knauf
(1991), p. 95. see littmann (1940), p. 245.
90 e.g. CIS V 187; 235; 257; 519; 533; 594; 1510; 1572; 1627; 1666; 1669; 1738; 1782; 18056;
23512; 2363; 2378; 2391; 2401; 2445; 2484; 2623; 2855; 2877; 2895; 2928; 2998; 3070;
3073; 3137; 3309; 3367; 3420; 3620; 3779; 3807; 4527; 4784; 4830; 4871; 5293; SIJ 341; 307;
473; 654; 415; 424; 473; 654; 731; 780; 987; 1005; Wh 148a; 176; 178; 182; 192; 286; 344;
362; 402; 544; 576; 730; 791; 815; 1236; 1237; 1240; 1248; 1250; 1277; 1655; 170506; 1750;
1778; 1945; 2008; 2018; 2112; 2254; 2442; 2530; 2629; 2639a; 2650; 2673; 2731; 2732; 2737;
2739; 2740; 2748; 2763; 2775; 2783; 2833b; 2942; 2989; 3024; 3157; 3166; 3205; 3211; 3227;
3249; 3276; 3317; 3395; 3418; 3437; 3558a; 3615b; 3627; 3744; 3855; 3903; 3926; PUAES iV
c 169; 210; 28284; 312; 317; 442; 448; 449; 482; 586; 931; 1021; 1039; 1061; 1076; 1159;
1165; ISB 19; 126; 135; 1413; 165; 399; 414; 4236. see also Wh 3663. see the far less dra-
wings of male camels: Wh 2741; 2780; 2782; 3097; 3283; 3811; PUAES iV c 333.
91 e.g. ISB 366; Wh 402; 1234; 2182; 3153; 3197b; PUAES iV c 335; 426.
92 Wh 1031.
93 rider with lance on donkey: e.g. Wh 125. horseman with lance: e.g. CIS V 1282; 1347; 1490;
1630; 2791; ISB 153; Wh 761. horseman lancing a gazelle: e.g. SIJ 63; Wh 368; 431; 627;
2067; 3170. horseman lancing a lion: e.g. PUAES iV c 1292. horseman lancing animals: ISB
154; 166; Wh 2053; 2109. horsemen ighting each other: e.g Wh 782; PUAES iV c 462; 573.
94 e.g. Wh 1677b; 1700a. see also a drawing of a lanced horseman with camels: PUAES iV c
325.
95 e.g. scholz (1995), p. 60. on the importance of the horse in safaitic inscriptions see ryckmans
(1942), p. 131.
96 e.g. Wh 54.
97 e.g. ISB 117; 146; PAAES iV 5,23; Wh 231012; 2870 (she-donkeys); 3092; 3548; 36567a.
98 donkeys: e.g. CIS V 505; 898; 1012; 1357; 1442; 1615; 2040; 2121; 2411; 2840; 2973; 4957;
SIJ 436; Wh 218; 284; 490; 744a; 1648; PUAES iV c 457; 514. see also pictures of donkeys:
e.g. Wh 1832; 2448; 3281; 3388; 3486; 3642; 3767, esp. she-donkeys: Wh 2878; 34234.
horses: e.g. CIS V 159; 439; 498; 766; 1127; 1186; 1228; 1314; 1553; 1777; 1855; 2210;
2916; 4467; 4614; 4619; 4855; SIJ 771; Wh 738; 1457a; 1502; 1743; 1779; 2531; 2571; 3911;
251 The problem of nomadic allies in the roman near east
it is thus apparent that on the basis of the deinition of nomadism presented
above the writers of the safaitic inscriptions are to be considered as nomads, since
the migration with the animals is one of the main aspects in these grafiti.
99
some
authors consider them to have been bedouins, assuming a predominant importance
of the camel.
100
occasionally they are simply termed as shepherds.
101
indeed, some
inscriptions mention the east or the desert (mdbr) as a place of temporary
presence,
102
especially for pasturing.
103
a certain ausn even regards the sight of
the desert as being delightful.
104
i have referred earlier to modern efforts to distin-
guish between different forms of nomadism by the animals being bred and accor-
dingly the regions being dwelled. The safaitic inscriptions, however, refer also to
other animals than camels as being kept in the desert:
105
some grafiti mention mo-
ving towards the desert with sheep
106
and heading eastwards with camels and live-
stock.
107
Keeping sheep and camels together is mentioned too,
108
and hence the
idea of distinct forms of nomadism with respect to safaitic tribes needs to be modi-
ied.
109
it is obvious also from the remarks in the preceding section that it is
inappropriate to refer to the writers of the safaitic inscriptions as bedouins.
110
The other important group of inscriptions are the so-called Thamudic grafiti,
whose motifs are often similar to those of the safaitic texts: there are references to
3912b; PUAES iV c 464; 566; 1136; ISB 91. horseman not being armed: e.g. Wh 811; 990;
PUAES iV c 1297. on the horse in safaitic drawings see macdonald (1996).
99 see the important article of macdonald (1992), followed e.g. by sartre (1990), p. 44. see also
isaac (1998b), p. 154; hoyland (2001), p. 65; parker (2002), p. 77; sartre 1991, 315; 333;
sartre (2005), p. 2345. see furthermore altheim and stiehl (196469) i, p. 280; asad and
Yon (2001), p. 30; contini (1987), p. 37; eissfeldt (1954a), p. 1001; Graf (1989a), p. 3678;
Grushevoi (1985), p. 51; macdonald (2001), p. 252; moors (1992), p. 283; parker (1986a),
p. 118; ryckmans (1942), p. 132; sartre (1997), p. 316; starcky [1956], p. 202 and p. 204;
Wolff (2003), p. 154.
100 e.g. PAAES iV, p. 1068; oxtoby (1968), p. 18, emphasises the predominance of camel. on the
evidence for the use of horses he assumes that some may have been full bedouin, others only
half bedouin. The predominance of the camel is also assumed by isaac (1998b), p. 154; sartre
1991, 315; 333, but without the term bedouin. The writers of the safaitic inscriptions are called
bedouins also e.g. by caskel (1954), p. 42; dirven (1999), p. 27; Gawlikowski (1995a), p. 106;
hartmann (2001), p. 82; Knauf (1991), p. 97; littmann (1940), p. 23; 25; King (1990), p. 63;
schlumberger (1951), p. 131. see also Gichon (1991), passim. cf. isaac (1998b), p. 124.
101 sartre (1990), p. 434.
102 desert: e.g. Wh 742; 927; 2395b; 2816; 2621; 3800; PUAES iV c 719. east: e.g. SIJ 45; 132;
206; 208; 836; 897; 911; Wh 173; 175; 732; 1675; 18501; 1856; 2125; 2128; 2166. Going
eastward to the desert: e.g. Wh 589; 1137; 1139/40; 1855; PUAES iV c 180; 602. Going
southwards: Wh 2110; 2143; 3559a; 3719.
103 e.g. Wh 784; 1066; 1361; 2327.
104 Wh 1604.
105 on being in the desert with camels: e.g. Wh 806; 1198.
106 e.g. Wh 578; 582: rdf hn lmdbr.
107 e.g. Wh 1700a.
108 Wh 1023.
109 see isaac (1998f), p. 412.
110 see macdonald 1991, 10203; macdonald (1993), p. 327.
252 Ulf scharrer
movement,
111
camping,
112
shepherding,
113
and camels.
114
Thamudic rock-drawings
depict mainly camels, and furthermore human igures, sometimes equipped with
bows and spears. some drawings show camel- and horse-riders, sometimes with a
lance, similar to the safaitic drawings.
115
The Thamudic inscriptions raise the question of ethnicity. although it still ap-
pears in scholarly literature rather frequently,
116
it has now for long been clear that
it is inappropriate to speak of the safaites as a speciic ethnic unit, however one
deines it.
117
however, the safaitic inscriptions mention a great number of collec-
tive names, all referred to as l.
118
among them are a few texts which maybe refer
111 e.g. Ph. 238c; 244; 271aj; 271al; 275m; 277a6; 277k3; 279m2; 279al4; 279aw2; 291b4; 297n9;
332g; 351f; 363ag2; 363ai; 366q; 367j; ITham hU135; hU152; hU142; hU261; hU3634;
hU369; hU411; hU425; hU491; hU512; hU545; hU779; hU801; Jsa148; Jsa207; Jsa390;
Jsa489; Jsa505; Jsa513; Jsa580; Jsa605; dgty52,3; Winnett and reed (1970), p. 131 nr. 72.
112 e.g. ITham hU80; hU135; hU259; hU287; hU350; hU361; hU388; hU408; hU691a;
hU776; Jsa211; Jsa241; Jsa243; Jsa343; Jsa4223; Jsa429; Jsa487; Jsa517; Jsa50910; Jsa513;
Jsa624; Jsa643; dgty69; Ph. 237e; 238n; 250b; 257a; 258f; 266ac; 266ag; 266aj; 267k;
271ap; 275a; 276g; 279m1; 279t; 297n2; 299d; 299j; 314h; 328e; 340f; 345l1; 348l; 351s1;
353z; 358ad; 358ao; 366n; 367q2; 369e; 373b; littmann (1940), nrs. 78; 80; Winnett and reed
(1970), p. 137 nr. 100.
113 e.g. ITham hU163; littmann (1940), nr. 82; Winnett and reed (1970), p. 87 nr. 55.
114 e.g. Ph. 266o; 266y1; 270c; 275d-e; 279ap; 292k; ITham eut44; hU103; hU109; hU161;
hU166; hU197; hU226; hU248; hU269; hU284; hU 2914; hU296; hU 299; hU3102;
hU315; hU318; hU336; hU356; hU366; hU397; hU413; hU424; hU4446; hU489;
hU501a; hU524; hU530; hU554; hU556; hU624; hU647; hU765; Jsa254; Jsa326;Jsa233;
Jsa243; Jsa361; Jsa367; Jsa491; Jsa496; Jsa626; Jsa5346; Jsa597; Jsa640; Jsa6501; Jsa661;
Jsa6623; Jsa667; Jsa673; dgty16,4; dgty47; littmann (1940), nrs. 68; 81; Winnett and reed
(1970), p. 85 nr. 45; p. 131 nr.72. she-camels: littmann (1940), p. 668; 71.
115 e.g. Garrard and harvey (1981), p. 142; parr e.a. (1978), p. 48; Zarins e.a. (1979), p. 301;
Whalen e.a. (1981), p. 54; Zarins e.a. (1981), p. 36.
116 e.g. altheim and stiehl (196469) i, p. 280; contini (1987), p. 45; 50; dussaud (1907), p. 2
and p. 21; dussaud (1955), p. 13547; oxtoby (1968), p. 145; Knauf (1991), p. 98; sartre
(1982a), p. 122; sartre (1990), p. 39 and p. 42; sartre (1991), p. 315 and p. 333; sartre (1997),
316.
117 see Graf (1989a), p. 358; macdonald (1993), p. 30410; ryckmans (1942), p. 1324; sartre
(1989), p. 151; sartre (1997), p. 298; sartre (2005), p. 234; Zeinaddin (2000), p. 267.
118 esp. the tribes of: Abd (SIJ 682); Ail (CIS V 66; 305; 321); min (Zeinaddin (2000), in-
scr.2); Amrat (macdonald (2005), p. 120; Khraysheh (1995), nr.12; 56; with Graf (1989a),
p. 360; milik (1980)); Ana (CIS V 347); Arfat (CIS V 1277); Arsat (Zeinaddin (2000), in-
scr.7); Aall (Wh 8); ty (Zeinaddin (2000), inscr.1); Aun (PUAES iV c 160); Alal (SIJ
41; 630; 658); Ar (CIS V 1664); Ara (PUAES iV c 639); Awkat (ingholt and starcky
(1951), nr. 63.2); M (Wh 1771); QS (Wh 2587); Aaqt (CIS V 3192); Aw (CIS V 65;
320; 638; 740; 4394; 4438; 4529; 4568; SIJ 59; 74; 80; 206; Wh 163; 966; 1144; PUAES iV c
325; 342; 361; 540; 1182; SIJ 39; ingholt and starcky (1951), nrs. 60; 63.4; see Graf (1989a),
p. 3613; moors (1992), p. 283); Awwm (CIS V 2481); BD (Wh 866b); bst (macdonald
(2003b), p. 278); Baad (CIS V 2577; 4394; 4447); Bar (CIS V 1758; SIJ 133); Badan (SIJ
237); Bsi (CIS V 5279; Wh 2815; 2875); Baswn (CIS V 102. probably also CIS V 103
(Baws)); Bawgi (CIS V 2955); Bigd (PUAES iV c 1188); Dawyi (CIS V 5175); Daaf (CIS
V 777; 1952; 2544; 2578; 2721; 2949; 4388; Wh 19; see Graf (1989a), p. 363); Dhib (CIS V
4039); aif (CIS V 318; 1573; 1649; 1679; 2839; 2843; 4304; 4439; 5361; SIJ 54; 88; 93; 132;
8234; Wh 17; 1517; 2021; 174; 1673; 1692; 1698; 1700ab; 1727; 2042; 2239; 2606;
253 The problem of nomadic allies in the roman near east
to villages.
119
some inscriptions give collective names with the praeix h- (-ite).
120

Unfortunately, the term l can have a number of meanings on different levels: it can
signify a people, a tribe, a clan or a family,
121
and it is therefore in most cases
unclear what it actually meant. in bilingual inscriptions (which are very rare) l is
rendered in different ways. The famous safaitic-Greek bilingual from s docu-
ments that the Ubait set up a statue for a certain maleikhat for the building of a
temple. in the Greek version the safaitic term l Ubait is translated as ogo o
2803; 3498; PUAES iV c 360; 687; SIJ 38); Fasrn (CIS V 5287); Farat (CIS V 4037. SIJ 58;
241 (Farith); lidzbarski (1908), p. 45 (Farn)); Fihr (PUAES iV c 1064); FR (Wh 367);
FY (Wh 3562. The name may also be GY); Gdhil (CIS V 2268); Gam (CIS V 2657);
Gabar (CIS V 4332); Gr (SIJ 246); Gurr (CIS V 2155); (ha-)Bagg (ISB 176); ha-Dharr (CIS
V 3663; 3663; PUAES iV c 397; 70; PAAES iV 5,6); Hshim b. Gumair (PUAES iV c 234);
BB (Wh 1607); abb (littmann (1940), nr. 28); fy (Zeinaddin (2000), inscr.3); gg
(CIS V 2823; milik (1960), p. 96); amad (CIS V 1302); Harim (littmann (1940), nr. 10;
PUAES iV c 435); awlat (PUAES iV c 344); an (SIJ 714); ibq (CIS V 4767; Zeinad-
din (2000), inscr. 18); umaiy (PUAES iV c 254; 255); uay (SIJ 295; 319; 320; 323; 342;
3489; 361; 455; 590; 607; 634; 909); JR (Wh 1232); Khil (CIS V 2192; 2318; 2297); Khtil
(PAAES iV 5,113); Kawkab (CIS V 65; 304; 320; Wh 2818); Kawn (CIS V 2843; 4064; 4079);
Khims (PUAES iV c 674); Khumn (PUAES iV c 419); Mis (ingholt and starcky (1951),
nr. 8 ); Mniy (CIS V 2634; 4987); Msikat (SIJ 207; 283; 287; 3101; 314; 644; 684. 6112;
7867; 8345; 6213 (ha-Msik)); Maar (littmann (1940), nr. 29); .M. (Wh 2802); Murib
(ISB 57); Mukabbil (CIS V 1763); Nafr (PUAES iV c 1266); Nabar (CIS V 58; 1368; 2113;
2576; 2577; 3261; 3263; 3648; 4446; PUAES iV c 348; 349; 361; 708; PAAES iV 5,123; 125);
Nasr (ingholt and starcky (1951), nr. 82a); Naaril (Wh 2147); nmr (atallah (1995), nr. 8);
Qadam (PUAES iV c 435); Qamm (CIS V 5050); Qamar (CIS V 8; 1414; 1868; 1870; 1951;
1952; 3757; 4278; 4828; 4845; 4944; SIJ 840; Wh 729; 1003; with Graf (1989a), p. 3656);
Qaiu (PUAES iV a 11); Qam (CIS V 26; 2721; 4441; 4755; Wh 154; 2817; 2820; 3561;
PUAES iV c 707); Qu (Khraysheh (1995), nr.4); Qumair (PUAES iV c 254; 255; PAAES iV
5,127); Rabbil (CIS V 2790); Rahya (CIS V 742; 2670); Raw (CIS V 5162; PUAES iV a 43;
PUAES iV c 1269; littmann (1939), nr.1; with Graf (1989a), p. 364); Raf at (CIS V 4358;
Wh 149); Rubat (?) (dussaud and macler (1901), p. 48, nr.32b); RFT (Wh 3931); RKS (Wh
2837); Saadil (CIS V 556; 781; 4754; Wh 1141); Sad (CIS V 4389); ab (SIJ 59); Salam
(PUAES iV a 44; with Graf (1989a), p. 3645); Salmn (CIS V 1157); Siml (CIS V 1290);
SB (Wh 77); SBT (Wh 1700a); adddat (CIS V 3194); SHWT wa-HBN (Wh 2147, two
tribes?); am (CIS V 743; 847; Khraysheh (1994)); mit (Wh 1867a); DT (Wh 2005);
aar (CIS V 4772); abbat (dussaud and macler (1901), p. 53, nr. 65); ab (Wh 3420);
akhr (PAAES iV 5,51); Slimi (CIS V 4646); awari (CIS V 3686); Taim (CIS V 2555; Wh
711); Taur (SIJ 272); iy (CIS V 2795; 5089); TRM (Wh 587); Ubait (CIS ii 164; V 3262;
Wh 1725a; PAAES iV 5,124; with Graf (1989a), p. 3601; Grushevoi (1985)); Wahabil (CIS
V 74; Wh 1859 (Wahbil)); Waqaril (Wh 2036); Yanat (CIS V 4039); Yahud (CIS V 1270);
Yaurr (CIS V 784; 2156; 4677); YR (Wh 2174); Zaid-l (PUAES iV c 461; SIJ 923). see
Graf (1989a), p. 359.
119 Zeinaddin (2000), inscr. 6 with commentary.
120 e.g. hfy (aiite: CIS V 1341; ISB 424; Wh 1060); hrt (RT-ite: Wh 424 (?); 1236; 3658);
(h)mkbly (MKBL-ite: Wh 400); hly (Aallites: Wh 3134); Ababite (ISB 155). see also the
Greek Aooir (SEG XlViii 1949).
121 see e.g. macdonald (1993), p. 3534; macdonald (1995), p. 97; macdonald (2003b), p. 2789;
van den branden (1966), p. 37. see also SEG XXXVi 1356; macdonald (1999), p. 260; Zein-
addin (2000), p. 276.
254 Ulf scharrer
tev Ooitgvev.
122
Grushevoi argues that the term ogo in Greek inscriptions
from Transjordan is never used to refer to nomads, but exclusively with respect to
city and village dwellers. from this he concludes that the Ubait at the beginning
of the irst century ad had become sedentary and part of the community of s.
123

There does not seem to be further evidence in support of this hypothesis. The
Ubait were clearly closely connected to buildings at s,
124
but their sedentarisa-
tion does not follow necessarily: the Greek version of the inscription is defective,
as the wrong form uaroixooogoovti (instead of the correct uaroixooog-
oovto) shows. it thus might be that the writer of the inscription was not absolutely
luent with Greek and its terminology. at least the bilingual shows that the Ubait
were a social unit more than a clan or a family, and can be regarded rather as a tri-
be.
125
in another bilingual inscription the safaitic term l is translated as ug .
126

There are also some Greek inscriptions mentioning tribes (ug ).
127
other Greek
inscriptions use the form tev in connection with a personal in the genitive
singular,
128
and it seems that these inscriptions rather refer to smaller social units
such as families or clans. finally, names of groups in Greek inscriptions appear just
in the plural form.
129
so far it is not clear to what extent all these groups may be
regarded as nomads.
130
The last aspect leads to the question of ethnic hierarchies. it has become clear
by now that these are hard to determine. however, there is sparse evidence that
some groups ought to be set on a higher level than others.
131
Thus the Aw and the
aif had an own Gad, which is mentioned quite frequently
132
and which could also
be adored by members of other groups.
133
occasionally these two Gadde were ad-
ored by one person,
134
which shows that both ethnic groups seem to have had at
122 CIS ii 164; PAAES iii 428a. The bilingual is also quoted fully by Grushevoi (1985), p. 52.
123 Grushevoi (1985), esp. p. 54, followed e.g. by contini (1987), p. 456; moors (1992), p. 343
n. 6. cf. Graf (1989a), p. 361. on the inscription see also millar (1993), p. 395.
124 see macdonald (2003b), p. 278.
125 see also cass. dio 51.7.1 on demoi and dynasts at the arabian gulf. cf. macdonald (2003b),
p. 279.
126 SEG XlVi 1799,2: ug Xouvgvev.
127 e.g. Aschenoi (PUAES iii a 760); Bitaienoi: IGRR iii 1277; Chauchab: IGRR iii 1269 (see
dussaud and macler (1901), p. 96, nr.263); Mozaidenoi (PUAES iii a 664); Osaisenoi: IGRR
iii 1180; Somaithenoi: IGRR iii 1276; Somaithenoi or Bitaienoi: IGRR iii 1273.
128 e.g. PAAES iii 379: t[ev] Oooiou; 388: tev Boou.
129 e.g. IGRR iii 1132: Arisenoi and Iachphirenoi.
130 on Greek inscriptions see macadam (1983), p. 111.
131 see sartre (2005), p. 235.
132 Gadd-Aw CIS V 853; 857; 860; 893; 994; 1066; 1186; 1292; 1744; 1936; 1955; 2114; 2556;
2617; 2795; 3062; 3143; 4404; 4457; 4646; 4848; PAAES iV 125 (= PUAES iV c 348); PUAES
iV c 306; 574; 640; 1198; 1214; SIJ 1008. see dussaud (1907), p. 14750; oxtoby (1968),
p. 212. Gadd-aif: CIS V 2795; Wh 613; 1725b; SIJ 132; 9112. on the Gadd-Aw see
littmann (1940), p. 108. on the Gad in general see dirven (1999), p. 101 and p. 1045.
133 Gadd-Aw: PAAES iV 125 (= PUAES iV c 348; CIS V 3263) (Udainath from the l of
Nabar); CIS V 4646 (l of slimi); CIS V 4845 (l of Qamar).
134 CIS V 2446.
255 The problem of nomadic allies in the roman near east
least some interaction. Tribal relationship is recorded in a safaitic inscription from
Wadi rm, which commemorates a man, son of bahm, of the tribe uaim, son
of a mother of the tribe aif, which means either that the uaim

were a branch of
the aif or, rather, that there was an intertribal marriage.
135
Graf suggests that the
Aw and the aif were tribal confederations,
136
but this is far from being certain,
since, as stated above, l could design ethnic or familiy groups on different le-
vels.
137
an inscription from the temple of bel at palmyra from ad 132 records the
dedication of a certain obaid/abidu presenting himself as nabataean and
ruhaean.
138
Whereas nothing is known about the ruhaeans, dijkstra interprets
the term rwy as referring to the Raw known from safaitic inscriptions.
139
in this
context it is suficient that the possibilities of different ethnic levels have been
shown.
finally there is the problem of the Thamd. for long it has been clear that wri-
ters of the Thamudic inscriptions do not necessarily belong to the ethnic group of
Thamd.
140
The Thamd have been known as an ethnic group since the eighth cen-
tury bc.
141
in the second century bc agatharchides, quoted by diodorus and pho-
tius, locates them in the north-east of the arabian Gulf, as ptolemy later does too.
142

pliny, probably drawing on iuba of mauretania, places the Thamd further inland
and mentions a town (oppidum) called baclanaza in their territory.
143
besides the
literary evidence, the name of the Thamd is mentioned in some inscriptions, e.g. a
few safaitic texts referring to the Thamd

as l and some Thamudic texts.
144
apart
from these very few traces actually not much is known about the Thamd in the
period of the early roman empire. There even seems to be some sort of uncertainty
whether the Thamd ought to be regarded as nomads or as a sedentary people.
145

some Thamudic inscriptions mention names of social or ethnic groups designated
as hl or l.
146
it is not clear whether at least some of them might belong to the hig-
her ethnic unit of the Thamd, and in most cases this is not very probable.
135 Zeinaddin (2000), inscr.16: mn bn bhm l m bn t l f.
136 Graf (1989a), p. 3634 and p. 366; Graf (1989b), p. 152.
137 on the problem of tribal confederations see schneider (2003a), p. 175.
138 Kaizer (2004a), p. 16970; (= CIS ii 3972; PAT 0319,2; PAAES iV 2,6,2; dijkstra (1995),
p. 108): nby rwy[].
139 dijkstra (1995), p. 1089.
140 caskel (1954), p. 39; Graf (1978), p. 112; Graf (1989a), p. 358; Graf (2002), p. 153; macdo-
nald (1993), p. 335; sartre (2005), p. 234 and p. 237; DNP 12/
1
(2002), col. 240 s.v. Thamu-
disch (Kessler). see Zarins e.a. (1981), p. 36. on the distribution of Thamudic inscriptions see
Graf (2002), p. 1556.
141 starcky [1956], p. 2012. see generally van den branden (1966).
142 diod. 3.44.6; phot. Bibl. 457. see Woelk (1966), p. 2213. ptol. Geogr. 6.7.21.
143 plin. HN 6.157.
144 safaitic texts: Wh 3792a; 3792c. see also Wh 1276. Thamudic texts: ITham hU172; hU637;
dghty51; Jsa280; Jsa339.
145 Thamd as sedentary: ryckmans (1942), p. 132. as nomads: Van den branden (1966), p. 32.
see the mentioning of a village (mdrt) in ITham hU712.
146 e.g. Ar (ITham dghty50,4); Badan (Winnett and reed (1970), p. 79 nr. 18, see also SIJ 87);
-Ata (ITham Jsa409); Gaff (Winnett and reed (1970), p. 86 nr. 48a.); ha-Awal (ITham
hU21); Harim (ITham Jsa450); aad (Winnett and reed (1970), p. 77 nr. 11); Hind (ITham
256 Ulf scharrer
hardly any of the above-mentioned ethnic groups seem to have been noticed
by Greek and roman authors, and this is probably the case because they were too
small and not important enough.
147
apart from the Thamd there is one possible
exception: eratosthenes, quoted by strabo, mentions the Chaulotaioi dwelling next
to the nabataeans.
148
macdonald assumes that they may be identical with the
awlat known from safaitic inscriptions.
149
if this is correct, this ethnic group
apparently lost its possibly predominant position in the period from the third cen-
tury bc to the irst century ad. as a third group, eratosthenes mentions the Agra-
ioi, also noticed by pliny.
150
furthermore, strabo and pliny mention a great number
of ethnic groups in the near east. some of these were clearly named by their city,
151

while others seem to be ethnic names.
152
a number of ethnic groups are mentioned
around mt lebanon,
153
near the nabataean territory,
154
in northern mesopotamia,
155

in inner mesopotamia
156
and in southern mesopotamia and at the persian Gulf.
157
apart from the above-mentioned exception it is thus far impossible to correlate
these ethnic names with any of the known inscriptions. furthermore, the same pro-
blem of possible ethnic hierarchies arises as we have seen with respect to the we-
stern semitic inscriptions: the terms r0vo, yrvo, natio, gens and even populus
could be used throughout Greek and latin ethnography in different degrees. espe-
cially the term r0vo has a wide range of meanings, e.g. an animal multitude and
people, be it small or large units.
158
it has been proposed to translate the word as
tribe,
159
as has been the practice for long in many translations, especially in the
loeb series. There was apparently no clear concept behind the ancient use of all
hU195); Humn (ITham hU260); abb (ITham Jsa607); amaiy (ITham hU498); ubbay
(ITham hU565); I (Ph. 342g); Ilmana (ITham hU5); Maan-l (littmann (1940), nr. 139);
Maay (ITham dgty16,3); Mag (ITham Jsa622 (?)); Namir (littmann (1940), nr. 123); Nimr
(ITham .Jsa596); Nr (ITham hU740); Sabr (ITham eut156); Tinnat (ITham hU27); Wil
(Ph. 370h); Zayyadu (ITham hU262). see also forms with the sufix -y: Aaite (Ph. 292w);
Bbite (Ph. 292v); Nimrite (Ph. 363ac).
147 see Kuhnen (1991), p. 329.
148 strabo 16.4.2.
149 macdonald (2001), p. 252. on the awlat see e.g. PUAES iV c 344.
150 strabo 16.4.2; plin. HN 6.159.
151 strabo 16.2.2; plin. HN 5.812: e.g. the Gazetae, Hemeseni, Epiphanenses on the orontes,
laodicaeans on the lebanon.
152 plin. HN 5.812: e.g. Ituraei, Granucomatitae, Penelenitae.
153 plin. HN 6.142: e.g. Nubei, Ramisi, Teranei, Patami.
154 plin. HN 6.157: e.g. Taveni, Suelleni, Araceni, Arreni, Hemnatae, Avalitae.
155 plin. HN 5.86: e.g. Praetavi (near carrhae), Rhoali (near Zeugma),
156 plin. HN 6.118: e.g. Eladamari, Salmani, Masei, Gurdaiaei, Azoni, Silices; strabo: e.g. Ely-
maioi, Paraikatakenoi, see strabo 16.1.1; 16.1.1718. east of the Tigris esp. plin. HN 6.133:
e.g. the Oxii; Mizaei; see strabo 15.3.12.
157 plin. HN 6.125: e.g. Attali: plin. HN 6.148: e.g. Nochaeti, Zurazi, Borgod, Catarrhei.
158 see e.g. Gschnitzer (2003), 4834; hutchinson and smith (1996), p. 4; Tonkin e.a. (1996),
p. 1920; scharrer (2002c), p. 168, with further references; Ulf (1996), p. 2408. see also e.g.
on old persian dahyu- briant (2001), p. 118; scharrer (2002c), p. 173 n. 22, with further refe-
rences. on indian brahmans as ethnos see diod. 17.102.7.
159 Tonkin e.a. (1996), p. 20; see also Gschnitzer (2003), p. 4703.
257 The problem of nomadic allies in the roman near east
these terms and they appear to be rather interchangable.
160
finally, it remains
unclear to what extent at least some of the peoples mentioned by strabo and pliny
may be regarded as nomads. Unfortunately strabo and pliny give hardly any ethnic
name when they mention nomades and scenitae, which might indicate that nomadic
peoples at the beginning of roman rule in the near east were dispersed in many
rather small groups which may have constituted something like a segmentary soci-
ety. There are only a few exceptions. one of these are the Catarrhei at the persian
Gulf, who are explicitly termed as nomades by pliny.
161
in addition, strabo says
that the mesopotamian Paraikatenoi were more interested in agriculture than other
peoples,
162
which might indicate that many other peoples had other ways of life,
among which were probably nomadic modes.
by the second half of the second century ad the ethnic situation in the near
east seems to have changed to a considerable degree. This is indicated by ptole-
mys lists in the ifth and sixth books of his Geography naming peoples which
mostly cannot be identiied with any of those named by strabo or pliny.
163
The sec-
ond and the early years of the third centuries ad in fact saw the migrations of large
tribal confederations from south and middle arabia into the spheres of roman and
persian inluence. by far the most important of these are clearly the Tankh. The
most important source on this confederation is abar whose account is basically
followed in modern scholarship. according to abar different tribes of the bahrain
united and decided to form a confederation under the name of Tankh. Towards the
end of the period of parthian power this confederation moved northwards to the
iraq.
164
according to abar at this time also other arab groups moved to the iraq.
165

eventually, abar relates, the Tankh and other groups settled at al-rah, and thus
al-rah had three elements of population, its original population, the Tankh and
the alf.
166
as the name Alf (confederation) indicates, this group was consti-
tuted by different tribal groups which apparently did not belong to the Tankh con-
federation.
apparently not all Tankh and other groups settled at al-rah itself. abar
records that the numrah, a tribe apparently allied to the Tankh, settled at nippur
in the realm of the former kingdom of the armn.
167
according to abar the
160 see isaac (1998c), p. 2645; Gschnitzer (2003), p. 4834. There are countless examples in
ancient literature. see e.g. curtius 6.6.36; 7.3.5; plin. HN 5.3637; 5.146; Vell. pat. 2.38.1
39.2; 40.12.
161 plin. HN 6.148.
162 strabo 16.1.18.
163 see potts (1990), p. 222 and p. 2267.
164 abar, Tarkh (History) i 7459. see ball (2000), p. 978; bowersock (1983), p. 1323; bo-
wersock (1994c), p. 1278; brentjes (1970), p. 323; equini schneider (1993), p. 467; hart-
mann (2001), p. 3415; hoyland (2001), p. 2323; potts (1990), p. 2234; rothstein (1899),
p. 2932; shahd (1984a), p. 36972. on different accounts of islamic authors see EIs 9 (1927),
p. 2278, s.v. Tankh (Kindermann); rothstein (1899), p. 29.
165 abar, Tarkh (History) i 822. see also abar, Tarkh (History) i 744 on earlier movements.
on the Tankh in ptolemy see e.g. potts (1990), p. 222.
166 abar, Tarkh (History) i 822. see altheim and stiehl (196469) i, p. 269.
167 abar, Tarkh (History) i 749.
258 Ulf scharrer
Tankh dwelled in tents, living mainly in the euphrates region.
168
from this it may
be concluded that they led a nomadic, if not bedouin life. The character of the sett-
lement of al-rah remains unclear. its name, meaning camp,
169
implies that it
originally was not a fully built settlement, but that at least large parts of it were
tents. The outer appearance of al-rah in the period covered here cannot be traced
reasonably. however, in scholarship the site is regarded as having become the main
political, social and probably also religious centre of Tankh power.
170
abar continues that at the sasanid conquest of the iraq a group of the Tankh,
the Quah, moved to syria and united with another branch of the Quah which
were already there.
171
This westward orientation of segments of the Tankh is re-
lected probably in the famous Greek-nabataean epitaph for fihr, the tutor (to-
ru / rbw) of the lakhmid ruler Gadhima, dated to ca ad 270, from Umm al-
Jeml in the southern hauran, and in the inscription of imraalqays from namra
dated to ad 328.
172
Thus far many points have to remain unclear: to what extent and how were
tribes of the safaitic and Thamudic inscriptions integrated into the Tankh domi-
nion, or did some of them become absorbed during this process?
173
To what extent
was force involved? how was the power of the ruling dynasty, the lakhmids, or-
ganised? some of these questions will be discussed below in the ifth section. ho-
wever, so far the process of ethnic change cannot be traced conclusively. at least
this process must have been considerable, as it is relected by the very different
ethnic groups given by earlier and later Graeco-roman authors.
The Tankh hegemony under the lakhmids has been characterised as a dimor-
phic structure, since the centre of political, social and religious life is said to have
been al-rah, and nomadic elements thus seem to have had close connections to a
168 abar, Tarkh (History) i 74950; 822. see caskel (1954), p. 42.
169 bowersock (1994c), p. 133; shahd (1984b), p. 4908.
170 see bowersock (1975), p. 521; caskel (1954), p. 44; luther (1997), p. 189; potts (1990),
p. 236; rothstein (1899), esp. p. 127. probably al-rah is to be identiied with the city ske-
nai, said to be in persia (steph. byz. s.v. Exgvoi). This reference might also be taken as relating
to the persian realm, i.e. the euphrates region. on skenai see altheim and stiehl (196469) i,
p. 2712; scharrer (2002c), p. 1989.
171 abar, Tarkh (History) i 821. see hoyland (2001), p. 234. on Tankh migrating westwards
at the rise of sasanid power see altheim and stiehl (196469) ii, p. 2512; bowersock (1975),
p. 521; bowersock (1994c), 1278; chapot (1907), p. 29; Graf (1989b), p. 150; hartmann
(2001), p. 346.
172 The epitaph for fihr, nabataean section: PUAES iV a 41; CIS ii 192; Greek section: PUAES
iii a 238
1
; SEG XXiX 1604. on the inscription see de Vries (1986), p. 237; equini schneider
(1993), p. 46; moors (1992), p. 3089; peters (1978), p. 3245. The namra-inscription of
imrualqays: ed. and comm. by dussaud (1902); dussaud (1907), p. 345; bellamy (1985). on
the inscription see also bowersock (1983), p. 1389; dussaud (1955), p. 645; isaac (1992),
p. 23940; millar (1993), p. 4345; moors (1992), p. 3078; shahd (1984a), p. 3153,
p. 5112 and p. 567; Zwettler (1993). on the inscription see also below, n.2347, n. 48891
and n. 4934. see generally bowersock (1975), p. 521; Graf (1989a), p. 37980.
173 e.g. Villeneneuve (1989), p. 138. Graf (1989b), p. 1589 states that with Zenobias defeat the
safaitic inscriptions came to an end. With respect to mesopotamia it seems that indigenous
tribes became absorbed: see abar, Tarkh (History) i 748.
259 The problem of nomadic allies in the roman near east
city.
174
The question is how intense these connections were. so far no conclusive
statement can be made, and it seems more appropriate to apply the concept of seg-
mentary societies to the problem: accordingly, different segments of the Tankh
confederation dwelled in diverse regions of the near east, probably with rather
loose ties to al-rah, and this may be relected in the movements westwards
scatched above.
in this context it is also debatable to what extent the emergence of the term
saracen stands in connection with the arrival of the new ethnic and predominantly
nomadic groups. ptolemy lists an ethnic group of Sarakenoi in direct neigbourhood
to the Thamd, both living in the hejaz,
175
and he mentions the Sarakene, a desert
in the south of the sinai.
176
in the course of the second and especially the third cen-
turies ad the term Saraceni became predominant in the naming and description of
arab peoples by Greek and latin authors, and in mediaeval europe arabs generally
were termed as Saraceni.
177
in scholarship, however, the origin and development
of the term is debated. already pliny mentions the Araceni adjoining the naba-
taeans, which sometimes is regarded as an early reference to the Saraceni.
178
an
inluential explanation is based on the famous Greek-nabataean inscripton from
rawwfa in the hejaz, dating from ad 16769.
179
The inscription states that a
temple was built by the Thamd, explicitly in Greek to tev Ooouogvev r0vo
(a2/c5a) and in nabataean rkt tmwdw (b4). Graf and oconnor suggest that rkt
means confederation, and from this word the ethnographic term Saraceni is said
to be derived: since many other arab groups formed similar confederations, the
word was taken over as a general ethnographic term for arabs.
180
This view has
been challenged especially by shahd, who argues that rkt, otherwise unknown for
174 see Zwettler (1993), p. 112. on the concept of dimorphic societies see above, n. 41.
175 ptol. Geog. 6.7.21.
176 ptol. Geog. 5.16.3. see steph. byz. s.v. Eooxo. see also suda s.v. Eooxgvoi just saying, that
the saracens are an r0vo. There is not much value in this information. see bowersock (1994e),
p. 386; Graf and oconnor (1977), p. 57.
177 see rotter (1986), p. 68; 105.
178 plin. HN 6.157. see e.g. EIs 7 (1927), p. 155 s.v. Saracens (mordtmann); DNP 11 (2001),
col. 52 s.v. Saraceni (Kuhnen); RE ii 2 (1920), col. 2388 s.v. Eooxo (moritz); bowersock
(1994e), p. 386; Graf and oconnor (1977), p. 57. see the remark of eusebius, Onomastikon
p. 118 (ed. Klostermann), according to which Qedar is the region of the saracens.
179 The inscription is published by Graf (1978), p. 910; bowersock (1975), p. 5145 (= dijkstra
(1995), p. 78); altheim and stiehl (19649) V/2, p. 247 (bowersock (1975), p. 514: a disa-
ster). on the date of the inscription see bowersock (1975), p. 515; briquel-chatonnet (1995b),
p. 137; Graf (1978), p. 10; sartre (1982a), p. 28; sartre (1991), p. 334. on the inscription and
the history of scholarship see altheim and stiehl (196469) V/2, p. 24; bowersock (1994b),
p. 1145; bowersock (1975), p. 5134; bowersock (1983), p. 967; dijkstra (1995), p. 7780;
Graf (1978), p. 9; Graf and oconnor (1977), p. 556; sartre (1982a), p. 279.
180 Graf and oconnor (1977); Graf (1978), p. 15; Graf (1989a), p. 354; oconnor (1986), p. 604
5, followed by bowersock (1983), p. 97; bowersock (1986), p. 113; bowersock (1994c),
p. 127; bowersock (1994e), p. 386; Kennedy and riley (1990), p. 38; Kuhnen (1991), p. 329;
sartre (1991), p. 334; Whittaker (1994), p. 1356. see also briant (1982), p. 1478; Gebhardt
(2002), p. 100 n. 5. cf. DNP 11 (2001), col. 52 s.v. Saraka (Toral-niehoff); shahd (1984b),
p. 12831.
260 Ulf scharrer
arab peoples, does not mean confederation, since in this case the Greek transla-
tion would have been ouoio, and thus rkt denotes simply a tribe or a people.
181

he suggests to refer to ptolemy, who is the irst to mention the Sarakenoi, from
which the general term is derived.
182
other important explanations for example re-
gard rq (= east) as the etymological basis of the word.
183
finally, macdonald
stresses an afinity to the arab word sharqiye, which means the annual movement
into the desert. according to the latter views the words rkt

and Saraceni are not
related etymologically.
184
Whether originally the name of a however deined ethnic group or not, as used
by Greek and latin authors the term Saraceni eventually became, as indicated
above, a collective name of different ethnic groups.
185
latin etymology derives the
name from the scenitae, which is most explicitly stated in some passages of am-
mianus marcellinus.
186
in his famous ethnography of the saracens he describes
them as warlike men having equal ranks, who do not live from any agriculture, but
move around.
187
it is obvious that in this period the term Saraceni basically names
nomadic peoples.
188
remarkable in this context is a passage in the Historia Augu-
sta, which apparently makes a distinction between arabs and saracens.
189
This
might indicate that Arabs, at least in this passage, could be regarded as rather se-
dentary populations in contrast to the nomadic Saracens, and this clearly supports
the above-quoted view of macdonald.
let us return briely to the rawwfa inscription. With respect to the Thamd it
is debatable what the term r0vo actually implies: if it is regarded as designating an
181 shahd (1984b), p. 13840. against this etymology see also macadam (1989), p. 308; macdo-
nald (1995), p. 968.
182 shahd (1984b), p. 12937. similar millar (1993), p. 140. according to macadam (1989),
p. 308 this may have been coincidence.
183 KlP, p. 1548 s.v. Saraka (dietrich); RE ii 2 (1920), col. 2389 s.v. Eooxo (moritz). see bo-
wersock (1994e), p. 386; DNP 11 (2001) col. 52 s.v. Saraka (Toral-niehoff). against this ety-
mology see macdonald (2001), p. 254. on other etymologies see Graf and oconnor (1977),
p. 614; macdonald (1995), p. 946; shahd (1984b), p. 1237.
184 macdonald (1995), p. 945; macdonald (2001), p. 2545.
185 similar was the view of european medieaeval authors on preislamic saracens: see rotter
(1986), p. 105. see also euseb. Chron. 24.14: Abraham ex ancilla Agar generat Ismahel, a
quo Ismahelitarum genus, qui postea Agareni et ad postremum Saraceni dicti. on the syriac
ayye see luther (1997), p. 146.
186 amm. marc. 22.15.1; 23.6.13. on sources see Graf (1978), p. 145. see also shahd (1984a),
p. 23040; 279.
187 amm. marc. 14.4. amm. marc. 14.4.2 states that he already described the saracens in his
now lost history on marcus aurelius. probably the term Saracen was used there anachronisti-
cally. on ammianus marcellinus and the saracens see e.g. bowersock (1994e), p. 385; Graf
(1989a), p. 3546; Graf and oconnor (1977), p. 589; parker (1986a), p. 144; parker (2002),
p. 79; seyfarth (1968); shahd (1984a), p. 23968.
188 see altheim and stiehl (196469) ii, p. 73; bowersock (1994c), p. 1267; bowersock (1994e),
p. 3857; christides (1972), p. 332; Graf (1989a), p. 354; isaac (1998d), p. 278; macdonald
(1995), p. 93; mayerson (1986), p. 36; mayerson (1988), p. 71; mayerson (1994a), p. 283 and
p. 291; seyfarth (1968); shahd (1984a), p. 240; 279; shahd (1984b), p. 2930.
189 sha Tyr. Trig. 30.7. see also festus, Breviarium 3; 16. see bowersock (1994e), p. 388. cf.
eadie (1967), p. 77.
261 The problem of nomadic allies in the roman near east
ethnic group, then the name of the Thamd seems to be on a higher level than
others, which is evidenced by the inscriptions d (nabataean) and e (Greek) from
the temple of rawwfa.
190
The nabataean section mentions that the temple was
built by adat from the rubau, the latter without the term l. The Greek section,
however, states cicOioi O[o]ouogvev ug Poo0ou oixooogoo to rigov
touto (the Sisthioi of the Thamudenoi, of the tribe Rhobathos, built this sanc-
tuary). Whereas the Sisthioi remain enigmatic,
191
it is apparent that there were
many tribes which belonged to the Thamd.
192
The nabataean section b4 records
that the leaders (qdmy) of the rkt had built the temple. it is thus plausible to as-
sume that the Thamd formed an ethnic group on a higher level, probably a confe-
deration, with many ethnic subgroups on the lower level.
193
however, it has been
argued by macdonald that r0vo in this contexts means some sort of a military
unit,
194
and if that is correct not much can be derived from the inscription about the
Thamd as an ethnic group.
it has been argued that the Thamd were absorbed by the saracens, since after
the rawwfa-inscription they do not igure in any records of the history of the ro-
man near east.
195
indeed, abar relates that Tankh groups under the leadership of
d destroyed the Thamd.
196
apparently the Qurn repeatedly narrates the de-
struction of the Thamd on behalf of their sins against the camel of allh sent to
them by the prophet li some time before the rise of islam.
197
in the present con-
text more important is the question to what extent the Thamd of the rawwfa-
inscription are to be regarded as nomadic. a nomadic way of life has been assumed
repeatedly in modern scholarship,
198
but so far there is no direct evidence for this.
While it can basically be assumed that there were both more nomadic and more
sedentary segments of the Thamd, there is a hint in the Notitia Dignitatum mentio-
ning Equites Saraceni Thamudeni in egypt.
199
apparently the Thamd of the unit
are designated as Saraceni, and if it is right that the term Saraceni primarily desi-
gnates nomads, the unit was recruited among nomadic segments of the Thamd. To
what extent the Equites Thamudeni Illyriciani mentioned in the Notitia as being
stationed in palestine
200
were recruited from rather sedentary groups is a matter of
190 published e.g. in Graf (1978), p. 10.
191 cf. Graf (1978), p. 10.
192 see van den branden (1966), p. 23.
193 Graf (1978), p. 112. sartre (1982a), p. 28 translates tribu. see sartre (1982a), p. 130.
194 macdonald (1995), p. 99100, followed by sartre (2005), p. 237.
195 Kuhnen (1991), p. 329.
196 abar, Tarkh (History) i 748.
197 Qurn 7.7176 (destruction by an earthquake (rajfa)); 41.12, 41.16, 51.4345, 69.5 (destruc-
tion by a thunderbolt (iqa)). on the total destruction of the Thamd see also Qurn 11.64
71; 25.4041; 26.141158; 40.32; 53.52; 91.1114. on the sins of the Thamd see also Qurn
9.71; 17.61; 22.43. see also paret (1980), p. 1645; EIs 8 (1927), p. 736 s.v. Thamd (bru);
EIs 7 (1927), p. 1078 s.v. li (bruhl).
198 Graf (1978), p. 12; Kennedy (2004), p. 41; sartre (1997), p. 316; starcky [1956], p. 201; Woelk
(1966), p. 222.
199 Not. Dign. [or.] 28.17. see shahd (1984b), p. 2930.
200 Not. Dign. [or.] 34.22. see shahd (1984b), p. 2930.
262 Ulf scharrer
debate. however, the problems involved with military recruitment shall be dis-
cussed below in the third part of the ifth section.
The emergence of the term Saraceni is in close connection with the process of
bedouinization of near eastern societies that is commonly assumed in scholarship
as having begun in the third century ad.
201
it has to be noted, however, that this
bedouinization is a modern construct for which there is no direct evidence. There
are different explanations for a bedouinization in the roman near east. obviously
most commonly stated is the development of the camel saddle, which is said to
have allowed an expanded use of the camel especially in ighting.
202
other explana-
tions ind the reasons for a bedouinization in roman rule itself,
203
or in its wea-
kness which attracted tribes from southern and middle arabia to migrate nor-
thward.
204
The supposed decline of trade after the fall of palmyra has also been re-
garded as a reason for bedouinization,
205
and inally the fall of client states in gene-
ral is sometimes given as an important reason, since in consequence local control of
nomadic groups is said to have ceased.
206
in the context of the question of bedouinization it is noticeable that with the
Tankh apparently a new and substantial nomadic population element came into
the sphere of persian and roman inluence. it is remarkable that the arrival of rather
nomadic groups coincides more or less with the expanded use of the term Saracen
in Greek and latin texts. apparently the term was applied also to Tankh groups:
abar writes that Tankh fought on aurelians side against Zenobia. in the Histo-
ria Augusta it is said that saracens fought on the roman side against palmyra.
207
at
least there seems to be an interdependency between the migrations of ethnic groups,
a nomadization and the use of the term Saraceni.
finally, the aspect of the term Arab in Greek and latin texts on the one hand
and in indigenous sources on the other needs to be discussed.
208
as it does not need
to be emphasized here, the term has a long history: it already appears in assyrian
201 altheim and stiehl (1972); bernbeck (1993), p. 1678 (on the whole a rather obscure book!);
bulliet (1975), p. 87104; caskel (1953), p. 8; Graf (1989a), p. 3923; Kuhnen (1991); parker
(1990), p. 223; scholz (1995), p. 489; Whittaker (1994), p. 135. cf. Khazanov (1994), p. 101;
oxtoby (1968), p. 1920. Villeneuve (1989), p. 120 assumes the beginning of a bedouinisation
in the ifth and sixth century ad. Kennedy and riley (1990), p. 37 assume a bedouinization
already beginning in the second century ad, which is not to be exaggerated. sceptical about
the concept of bedouinization: lancaster and lancaster (1988).
202 bulliet (1975), p. 87104; caskel (1954); Kuhnen (1991); staubli (1991), p. 198.
203 caskel (1953), p. 8; scholz (1995), p. 489.
204 parker (1990), p. 223.
205 bulliet (1975), p. 104; parker (1986), p. 642; sommer (2003a), p. 43. see butcher (2003),
p. 64.
206 caskel (1954), p. 3940. some scholars argue that after the fall of some client states the popu-
lation became nomadic: on hatra: bernbeck (1993), p. 1678. on palmyra: Graf (1989a),
p. 3923; matthews (1984), p. 169; shahd (1984b), p. 24. on the nabataean kingdom see
cameron (1993), p. 194; Taeschner (1964), p. 301. see also the criticism of caskel by Graf
(1989a), p. 3923.
207 sha Aurel. 28.2.
208 on the discussion see rets (2003), p. 10518. on the arabian terminolgy see orthmann
(2002), p. 14150.
263 The problem of nomadic allies in the roman near east
texts since the ninth century bc and is found in babylonian texts up to the early
parthian period and in the Hebrew Bible.
209
The problem of the etymology is not
solved yet. according to some views it is related to the hebrew word {ar~b~h (=
desert) and thus designates those living in the desert.
210
it thus appears to be a term
applied by others, i.e. sedentary people, to the relevant groups. according to ano-
ther view the hebrew word does not provide a suficient etymological basis, as a
similar word is not known from akkadian or old arabian. it was thus originally a
native term of nomads themselves.
211
also discussed is the use of the term in Greek
and latin authors. Whereas it is often stated that the term implies nomads
generally,
212
other approaches are more distinctive. With respect mainly to Greek
authors up to the period of alexander the Great, macdonald states that Arabia was
primarily a geographic term. The term Arabs was used for different groups with
different ways of life whose common factor was linguistic. Thus the term could be
applied both to nomadic and sedentary populations.
213
This view is supplemented
by the apparent distinction between arabs and nomads made by Greek and roman
authors. Thus, strabo and pliny repeatedly mention Arabes scenitae,
214
which dis-
tinguishes these groups from other arabs. furthermore a passage in strabo explic-
itly makes a distinction between skenitai and Arabs.
215
The nomadic arabs can be
contrasted to sedentary or agricultral arabs who are occasionally referred to in lit-
erary sources.
216
in addition, some hellenistic Greek inscriptions, each mentioning
an arab (Ao) at rhodos, delos and athens, seem to remark on the ethnic origin
of these persons rather than their way of living.
217
furthermore, Greek papyri from
ptolemaic egypt mention arabian villages and towns,
218
thus implying a sedentary
way of life of their inhabitants. finally, the surname of philippus arabs emphasized
the emperors ethnic and geographic origin rather than an earlier nomadism. Thus
the use of the word arab in Greek and latin sources does not at all necessarily
imply nomadic peoples.
209 see briant (1982), p. 1139; ephal (1982); KlP, col. 483 s.v. Arabia (Thomasson); RE II 1
(1895), col. 344 s.v. Arabia (mller); Zwettler (1993), p. 21 n. 21. arabs are mentioned occa-
sionally in the Babylonian Astronomical Diaries: sachs and hunger iii (1996), 329, rs. 2;
125a vs. 21; 124b rs. 20 (= schuol (2000), p. 37 and p. 39 respectively).
210 altheim and stiehl (1972), p. 2968; briant (1982), p. 114; ephal (1982), p. 7; healey (1989),
p. 40; RE II 1 (1895), col. 344 s.v. Arabia (mller); EIMW, p. 51 s.v. Arabia (newby). see also
Tohme (2000), p. 112.
211 ephal (1982), p. 7. see macdonald (2001), p. 232.
212 ball (2000), p. 32; levi (1989), p. 34; millar (1993), p. 221; sommer (2003a), p. 22; sommer
(2005), p. 58, stating that the term is not used homogenously for nomadic, but sometimes also
for sedentary arabs; Villeneuve (1989), p. 124 n. 3. Using the term arabs in the sense of no-
mads: schmitt (2005), p. 4223.
213 dillemann (1962), p. 889; macdonald (2001), p. 2313, p. 2478 and p. 2512; ross (2001),
p. 165 n. 25. on the term in early islamic texts see rets (2003), p. 2495.
214 plin. HN 5.87; strabo 16.1.8; 16.1.26; 16.3.1. see also CIL iii 128.
215 strabo 16.2.11. see chapot (1907), p. 27.
216 e.g. curtius 4.2.24. see macadam (1989), p. 2978.
217 SEG iii 674,34; 523; baslez (1977), p. 101; p. 361; bruneau (1970), p. 4778; IG
2
ii=2
83618362. see also IGRR i 839 from the roman period.
218 e.g. P. Enteuxeis 3,1; 47,1; S.B. 11307,56.
264 Ulf scharrer
in this context shahd states that Greek and roman authors referred to arabs as
a collective ethnic term, which comprehends many single ethnic groups.
219
accor-
ding to an early Greek tradition established by stesichorus and hesiod, arabs stem
from a certain arabus, son of hermes and on his mothers side grandson of be-
lus.
220
The naming of an ancestor of large ethnic units, whose descendants themsel-
ves are regarded as the ancestors of ethnic subunits on different levels, is an ethno-
graphic habit common both to ancient western and eastern consciousness of ethnic
identity. The term arab accordingly could be used by Greek and roman authors
for the inhabitants of Arabia, which is deined geographically, and as an ethnic
term. however, a certain lifestyle is not implied.
221
more problematic are the references to rb in semitic inscriptions and docu-
mentary sources. above the early history of the term is sketched very briely. docu-
ments from the roman near east, however, are currently interpreted ambiguously.
from the second century ad onwards, the word and its terminological ield begins
to be used frequently in documents from hatra, osrhoene and occasionally else-
where. at hatra the terms lord of the arabs (mr rby), majordomo of the arabs
(rbyt dy rb) and king of the arabs (mlk rby) are frequently attested, as well as
other mentions of Arab.
222
similarily some inscriptions from sumatar in the os-
rhoene mention a governor of Arab.
223
it is possible that the syriac term corre-
sponds to the Greek ooor, mentioned in a papyrus from ad 121, which was
found in dura-europus, but written originally in paliga.
224
arabia here is a region
outside the durene territory, since two other papyri from dura-europus refer to
europus toward arabia.
225
furthermore an inscription from the Wadi mukkatep
on the sinai mentions arabs,
226
and inally the famous inscription from namra
entitles maralqays/imraalqays as king of (all) Arab.
227
in the context of this
section it ought to be discussed what could be meant by the term rb, since there
219 shahd (1984b), p. 6. see pollard (2000), p. 1123. on the question of an arabic ethnicity see
von Grunebaum (1963).
220 strabo 1.2.34. RE ii 1 (1895), col. 350 s.v. Arabia (mller). on similar ethnic etymologies see
scharrer (2002c), p. 194. see furthermore on ethymologies of Arabs and Arabia in Greek sour-
ces christides (1972), p. 32931.
221 see peters (1999a), p. xiv; rets (2003), p. 112. see e.g. plin. HN 5.86 and 6.136 on arabian
oppida. see also proc. bell. 1.19.20.
222 lord of the arabs: H78; majordomo of the arabs: H223,2 (= dijkstra (1995), p. 3; 228);
H364,2; king of the arabs: H193,2; H194,12; H195,12 (= dijkstra (1995), p. 228); H196 (?);
H197,2; 4; H198,23; H199,23; H203,12; H287,45 (= dijkstra (1995), p. 2301); H290,2
(= dijkstra (1995), p. 2023); H342 (= Kaizer (2006b), p. 1434), 12 (?);H345,2; H347,2;
H353,2; H370; H373; H375; H379; H1020,5; other mentions: H79,10; 14; H288,d1; H336 (=
Kaizer (2006b), p. 1423), 5; H343 (= Kaizer (2006b), p. 1445), 3 (= dijkstra (1995), p. 183);
H025,3. on inscriptions from hatra see e.g. drijvers (1977), p. 8204.
223 drijvers and healey (1999), as36,1; as47,3; as49,23; as51,1; as52,4; add3,3: ly drb.
224 PDura 20,5. on the document in general see dirven (1999), p. 297; millar (1993), p. 4478,
p. 479 and p. 489; segal (1970), p. 19 and p. 22. The document is also republished fully in
sommer (2005), p. 296 n. 54.
225 PDura 22,3; 25,17: rv Eueae tg ao Aoio.
226 NSI 107. The reading, however, is dubious. see e.g. Graf (1989a), p. 3445.
227 Line 1: mlk lrb (klh). on the inscription see n. 172.
265 The problem of nomadic allies in the roman near east
seem to be some discrepancies about this matter in current scholarship. most often
it is stated that the term refers to the steppe and its nomadic inhabitants,
228
and that
the title of an arabarches refers to an ofice relating to nomadic steppe-dwellers.
229

according to the second view the ofice relates to steppe dwellers, be it nomadic or
sedentary, i.e. villagers.
230
accordingly also the Arab of the hatrean inscriptions
are regarded as the steppe and its nomadic inhabitants
231
or as its nomadic and se-
dentary dwellers, as there are archaelogical traces of sedentary settlements also in
the steppe.
232
in both cases the second view appears thus to be more likely, and
accordingly there is a difference between the term Arab in Greek and latin literary
sources and in the near eastern documents, the irst using the word as an ethnogra-
phic name, the second as a term for people dwelling in the steppe. it is quite pro-
bable that often nomadic arabs are meant, as for example the mishnah speaks of
tents of arabs.
233
however, it was not necessarily a nomadic way of life that seems
to be implied by the documents just quoted.
not only in this context the reading and interpretation of the funerary inscrip-
tion for imraalqays from namra in the eastern hauran is controversial. in the
inscription, the name is given as Maralqays bar Amr, but he is generally identi-
ied with the lakhmid king imraalqays ibn amr ibn ady, known for example
from abar, and thus the inscription is dated to ad 328.
234
it ought to be briely
reconsidered here what the title of king of (all?) Arab implies. The reading is not
clear because of the erosion of the surface.
235
at the end of the opening phrase it is
debatable whether its wording is klh (all) or [w]lqbh (and his title of honour
is).
236
although after a second scrutiny on the stone he accepts the original reading
228 aggoula (1975), p. 197203; ephal (1982), p. 68; dostal (1989), p. 423; Gawlikowski
(1995b), p. 87, followed by Kaizer (2002), p. 57; Kaizer (2003a), p. 286; millar (1993), p. 448;
sommer (2003a), p. 23; p. 38; Teixidor (1964), p. 283; Vattioni (1994), p. 5; p. 49; Wiesehfer
(1982), p. 440. see also the relevant translations of the hatraean inscriptions by beyer, accor-
ding to which bedouins (bedouinen) were implied. cf. sartre (1989), p. 141.
229 drijvers and healey (1999), p. 37; p. 105; Gawlikowski (1995b), p. 87; millar (1993), p. 450;
sommer (2003a), p. 22; sommer (2005), p. 255; Teixidor (1964), p. 283; Tubach (1986), p. 15.
230 ross (2001), p. 25; segal (1970), p. 22; Zwettler (1993), p. 10. also aggoula (1995), p. 75;
dijkstra (1995), p. 253.
231 altheim and stiehl (196469) iV, p. 26971; dijkstra (1990), p. 956; Vattioni (1994), p. 56
and p. 49. see also luther (1997), p. 173; sommer (2003a), p. 38; Zwettler (1993), p. 89. on
the steppe around hatra see e.g. cass. dio 68.31.1.
232 hauser (1998), p. 5123; hauser (2000), p. 1913. also aggoula (1995), p. 745; Kaizer
(2006b), p. 143 and p. 153; Zwettler (1993), p. 12. on the syriac distinction between arabye
as inhabitants of the provincia arabia and arabye, who lived in the steppe see luther (1997),
p. 205. see furthermore orthmann (2002), p. 143.
233 Mishnah VI. Seder. oharot: 2. Ohalot 18.10.
234 on editions and commentaries see n.172. on the identiication and the date see furthermore
altheim and stiehl (196469) iV, p. 34; bellamy (1985), p. 31; bowersock (1975), p. 520;
bowersock (1983), p. 134; bowersock (1986), p. 1134; isaac (1998b), p. 125; parker (1986b),
p. 642; shahd (1984a), p. 32; shahd (1984b), p. 31 n. 38; Zwettler (1993), p. 12. on im-
raalqays see e.g. abar, Tarkh (History) i 834; 845; bowersock (1983), p. 13846.
235 see Zwettler (1993), p. 5 with pl.iiii.
236 on the discussion see dignas and Winter (2001), p. 1989; Zwettler (1993), p. 156. see bel-
lamy (1985), p. 356.
266 Ulf scharrer
by dussaud as klh, Zwettler argues with respect to the inscriptions from hatra
that not all Arab in the sense of all arabs could be meant, since the inscription na-
mes only four or ive tribes ruled by imraalqays. he thus interprets Arab rather as
the entire region over which imraalqays claims to have ruled, without respect to a
certain way of life, be it nomadic or sedentary.
237
in fact, namra was an important
place for the tribes of the safaitic inscriptions.
238
if these inscriptions continued to
be written into the fourth century ad or even later, their writers certainly falled
somehow under the goverment of the lakhmid king
239
, as well as maybe the
Thamd of the rawwfa-inscription.
240
Generally, then, there emerges one problem: if we speak of nomadic allies, it is
in most cases hard to identify them. This problem has two aspects. The irst aspect
lies in the dificulty that it is not always clear whether indeed all relevant groups
should be regarded as nomads, not to speak of bedouins. The second aspect is even
more complicated: if we ind nomadic allies, it is impossible in most cases to iden-
tify clearly deined ethnic groups, as nomadic society appears to be split in many
segments on different levels. it would be a great task to compile an ethnographic
inventory of the roman near east which eventually might lead to an ethnographic
history,
241
but this of course cannot be undertaken here.
nomad-sedenTarY relaTions: anTaGonism and sYmbiosis
i shall now turn to the ield of nomad-sedentary relations. i shall, however, exclude
political relations, as these are the topic of the next section. here, i shall irst dis-
cuss the topic of conlict, and afterwards some forms of symbiosis between no-
madic and sedentary populations in the roman near east. at the end of the irst
section i mentioned raiding as an economic strategy of nomadic groups to gain
goods they could not produce by themselves. in fact, nomadic aggression is a topos
which is to be found throughout the centuries in records produced by sedentary
populations, from akkadian texts up to the present day.
242
in this context it has long
been discussed to what degree nomads should be regarded as a threat to sedentary
populations within the sphere of the eastern roman empire and to the empire it-
self.
243
With modiications, there are basically four positions taken. according to
237 Zwettler (1993). see already bowersock (1975), p. 520. see the reading of dussaud (1902),
p. 4123. see furthermore Weerakkody (2000), p. 113.
238 dussaud (1955), p. 138. safaitic inscriptions from namra or mentioning it: e.g. PAAES iV
5,2; CIS V 241; 523; 1656; 18945; 2732; 2803; 3143; 322348; 3878; PUAES iV c 330;
4067; 426; 532; 539; 675; 1013; ingholt and starcky (1951), nr. 81b.
239 see shahd (1984b), p. 31 n. 38. cf. Knauf (1991), p. 97.
240 see bowersock (1975), p. 522.
241 see SEG XlViii 1909; pollard (2000), p. 171211.
242 Graf (1989a), p. 3512; 354; leder (2005), p. 21; marx (1992), p. 258. see scharrer (2002b),
p. 299300. on Greek and latin texts: beyer (1998), p. 198 n. 89; sartre (1991), p. 332 (=
sartre (1997), p. 3156); see scharrer (2004), p. 3112.
243 on the discussion see Gebhardt (2002), p. 90 n. 1; Graf (1989a), p. 344 n. 8; sartre [1990],
p. 501; staubli (1991), p. 78; Whittaker (1994), p. 137.
267 The problem of nomadic allies in the roman near east
the irst view there was a considerable nomadic threat throughout the period cov-
ered here.
244
The second position taken is that nomads in the irst centuries of ro-
man rule did not constitute a menace, but that by the third century the nomadic
threat became serious.
245
The adherents to the third position suggest that nomads
did actually not constitute a menace,
246
and in this context they sometimes empha-
sise a mutual relationship between the nomad and the sedentary spheres.
247
follow-
ers of the fourth position argue that a distinction is to be made between relations on
a macro- and a micro level. according to this position there could have been con-
licts on the large scale, but on the small scale there could have been peaceful inter-
action too.
248

a central issue within this discussion is the function of the different limites in
the roman near east i.e. the Limes Palaestinae in the negev, the Limes Arabicus,
the Via Nova Traiana and the Strata Diocletiana.
249
The interpretation of the func-
tion of these fortiications depends on the view towards the question of the weight
of nomadic threats. according to the irst view a main purpose of the military in-
stallations was to keep nomads out.
250
Within this ield of arguments, the fortiica-
tions themselves are put forward in favour of a considerable nomadic threat. The
argument, in short, runs as follows: if there are fortiications towards the steppe and
the desert there must have been a threat by desert peoples.
251
The adherents to the
view that there was no considerable nomadic threat argue that the function of a ro-
man limes was not primarily the defence against a large scale attack from beyond
the frontier, but to monitor movements, in the present context nomadic movements.
244 Gichon (1991), p. 3201; Gichon (2002), p. 1868, p. 1901 and p. 193; parker (1990), p. 223;
parker (1991), p. 499503; parker (2002), p. 789; shahd (1984b), p. 223; Whittaker (1994),
p. 136. see also chapot (1907), p. 256; dabrowa (1986), p. 94; isaac (1989), p. 243.
245 hogdon (1989), p. 178; Graf (1989a), p. 34451; hartmann (2001), p. 823; Kuhnen (1991),
p. 32930; leadbetter (2002), p. 768; lewin (2002), p. 91. see also Graf (1978), p. 123;
sartre (1991), p. 332 (= sartre (1997), p. 316); Weerakkody (2000), p. 114. cf. Gichon (2002),
p. 1901.
246 ball (2000), p. 32; Graf (1978), p. 7; hoyland (2001), p. 101; isaac (1989), p. 243; p. 250;
isaac (1992), p. 717; isaac (1998b), p. 1245; Konrad (2003), p. 244; macdonald (1993),
p. 3356; mehl and schmitt (2005), p. 10; sartre (1991), p. 334 (= sartre (1997), p. 318);
sartre (2005), p. 69; Wells (1991), p. 480. see marx (1992), p. 2578; schmitt (2005), p. 421.
247 ball (2000), p. 32; banning (1986); banning (1987); banning (1988), p. 734.
248 banning (1986); banning (1988).
249 on the Limes Palaestinae see Gichon (1991), p. 3224. on the Limes Arabicus see Graf (1978),
p. 13; parker (1990). on the Strata Diocletiana see Whittaker (1994), p. 1367. on military
installations in the roman near east see generally bowersock (1994b), p. 1206; bowersock
(1976); butcher (2003), p. 41520; eadie (1986); Gregory (19961997); isaac (1989), p. 244
56; Kennedy and riley (1990); Konrad (2003); millar (1993), p. 13740; parker (1986a); par-
ker (1986b); parker (1991); parker (2002); pollard (2000), p. 1320; lewin (2002), p. 928;
Kissel (2002); mayerson (1986); sartre (2005), p. 1435; RE Xiii 1 (1926), col. 65660 s.v.
Limes (fabricius). on roman fortiications in northern mesopotamia see dillemann (1962),
p. 195216. on fourth-century fortiications see shahd (1984a), p. 46590.
250 bernbeck (1993), p. 168; dussaud (1907), p. 45; petit (1971), p. 1234; peters (1978),
p. 31820; cf. Graf (1989a), p. 3423.
251 Gichon (1991), p. 320; Gichon (2002), p. 1913; Kuhnen (1991), p. 330; parker (2002), p. 78
9.
268 Ulf scharrer
it is stressed that nomadic populations did not only live outside the sphere of the
roman empire, but also within it. Thus, a limes is regarded not as a hermetically
closed line but as a zone of contact.
252
finally the intensiied extension and
strengthening of the defensive systems in the third century ad, especially under
diocletian, is stressed, which is regarded as an indicator that rome in this period
had to face nomadic peoples being of considerable strength.
253
The interpretation of the limites in the near east with respect to nomadic mo-
vements depends on two underlying concepts, namely the general idea of a roman
limes and the concept of the geographic zone of nomadic dwelling. With regard to
the irst aspect, the old idea (though still occasionally stated) that a roman limes
was a closed borderline between inside and outside the roman empire
254
is dated.
currently roman limites in general, and speciically in syria and arabia, are regar-
ded not as hermetically closed borderlines, but as a military limes, whose purpose it
was to regulate movements including the prevention of smaller raids and the exal-
tation of taxes.
255
accordingly the southern near eastern limites are said to have
monitored nomadic transhumant movements.
256
This leads to the other aspect, the
concept of nomadic dwelling. on the one hand it is sometimes stated that nomadic
populations primarily live in the steppe or the desert, i.e. beyond the zone of an
annual precipation of ca 200 mm.
257
following this concept it could reasonably be
said that the roman limites divide the sphere of the roman empire, which is the
rather fertile land, from the outside regions, and that their main purpose should be
considered as protecting against nomadic raids into the cultivated land.
258
in this
context rowton makes a distinction between exclosed and enclosed nomadism, the
former designating nomadism outside the sedentary sphere, the latter nomadism
within it.
259
as indicated in the third section, the relevant nomadic groups are to be
found within the settled sphere or at least at its fringes. from this it may be conclu-
ded that the roman fortiications in the irst instance had the purpose of monitoring
nomadic and probably other movements rather than of closing the sown from the
desert hermetically.
252 bennett (1997), p. 1778; hauser (2000), p. 194; isaac (1992), p. 74; isaac (1998b), p. 1359;
isaac (1998f), p. 414; mayerson (1986), p. 36; p. 3940; p. 423; mayerson (1988), p. 745;
mayerson (1990), p. 268; parker (1986b), p. 637; Wells (1991), p. 480.
253 parker (1986b), p. 6412; shahd (1984a), p. 145, regarding the limes as a dividing line bet-
ween the desert and the sown.
254 mommsen (1908); RE Xiii 1 (1926), col. 5736 s.v. Limes (fabricius). on the history of
scholarship on the meaning of limes see carri (1995), p. 3441; isaac (1998e), p. 3456.
255 bowersock (1976), p. 2278; Gichon (1991), p. 319; Gichon (2002), p. 18990; mayerson
(1986), p. 44; parker (1987), p. 48. on a similar function of hadrians Wall in north britain see
breeze (1982), p. 846; breeze and dobson (1987), p. 40. cf. carri (1995), p. 416; Graf
(1989a), p. 343.
256 bennett (1997), p. 1778; Gichon (1991), p. 319 and p. 321; parker (1986b), p. 637. see Wells
(1991), p. 480.
257 e.g. butcher (2003), p. 161 and p. 169. see isaac (1998f), p. 4134.
258 Thus stated e.g. by shahd (1984a), p. 145.
259 rowton (1976). see scharrer (2002b), p. 295 with further references to other works by row-
ton.
269 The problem of nomadic allies in the roman near east
With regard to the notion of raiding as a nomadic economic strategy, it has
been pointed out above that raiding is just one of many nomadic strategies of adap-
tion to, and interaction with, the sedentary world. The latter are especially suitable
in regions being inhabited both by sedentary and nomad groups. before coming to
the aspect of symbiosis it is irst necessary to evaluate the evidence on nomadic
raiding and unrest in the spheres of the roman and the persian empires.
There is some evidence of unrest in the regions covered here. literary sources
often mention arab banditry and raids.
260
some authors state in a rather general
way that arabs live from brigandage,
261
which is also said about the nabataeans.
262

already the babylonian Astronomical Diaries mention arab raiding in southern
mesopotamia in the earlier parthian period.
263
Whether the so-called mesene text,
stating that the arabs [] came about the land,
264
refers to peaceful migration or
also to raids is debatable. pliny mentions the Attali, an arabian tribe of brigands,
dwelling near charax,
265
clearly distinguished from the scenitae, who are said to
live beyond them.
266
Thus at least in this case bandits and nomads appear to be two
seperate groups. strabo, however, writes that the skenitai of southern mesopotamia
were brigands and shepherds, migrating according to pasture and booty.
267
pliny, in
another passage, mentions nomads, who harry the territories of the chaldaeans.
268

strabo reports that the people alongside the mountains in northern mesopotamia
were harassed by skenitai and armenians.
269
The mention of the armenians leads to other brigand groups within the par-
thian realm. The most important of these were mountain peoples, who are mostly
described as raiding groups.
270
Whereas some of these peoples are said to be sub-
ject to the parthians, the most prominent group, the Uxians of the Zagrus, seem to
have been rather independent as they were already in the achaemenid empire.
Well known are the heavy ights against them by alexander the Great, who inally
managed to subdue them, although it is not clear for how long.
271
in the present
context there is the question to what degree these mountain peoples are to be regar-
ded as nomads, but unfortunately there are no hints in the evidence about this pro-
blem. at least on the Paraikatenoi strabo reports that they were more interested in
agriculture than other peoples.
272
from this the conclusion might be drawn that
260 e.g. cass. dio 36.17.3; 37.7a.
261 e.g. plin. HN 6.162; Just. Epit. 40.2. see sartre (1989), p. 1437; Wolff (2003), p. 135.
262 diod. 2.48.3.
263 sachs and hunger iii (1996) 124b rv. 20 (= schuol (2000), p. 39).
264 sachs and hunger iii (1996) 125a vs. 21 (= schuol (2000), p. 37):
l
Ar-ba-a-a x x x [x x] ina
KUr i-se-gu-.
265 plin. HN 6.125: Attali latrones, Arabum gens.
266 plin. HN 6.125: ultra quos scenitae.
267 strabo 16.1.26, with scharrer (2004), p. 3156.
268 plin. HN 6.143.
269 strabo 16.1.26.
270 strabo 15.3.12. see strabo 16.1.1; 16.1.1718.
271 on the Uxians generally see plin. HN 6.133. on the Uxians in the achaemenid empire and
their ights with alexander the Great see curtius 5.3.116, with Wiesehfer (2004), p. 189.
272 strabo 16.1.18.
270 Ulf scharrer
most of the mountainous peoples had non-agricultural means of live, and besides
brigandage this may have included animal husbandry.
finally, also herdsmen in the parthian realm could be subject to brigandage.
Josephus tells the story of asinaeus and anilaeus of nearda, Jews within the par-
thian kingdom living from raids and tribute they extracted from herdsmen.
273
more
important in this context are the remarks by abar: While the sasanid ruler shapur
ii was still a child, arabs from bahrain made incursions to frs seizing cultivated
land and local peoples herds of cattle, which was done for need of daily suste-
nance.
274
Why the supply of these groups decayed is not clear, but at least here is a
clear reference to raiding as a nomadic economic strategy. however, abar contin-
ues that shapur, having grown up, made heavy counter attacks, irst in frs itself,
afterwards on a substantial campaign leading him to medina in the south during
which he killed many nomads, took captives and organised deportations.
275
a clear
reference to nomad raiding in the east inally stems from Theophanes. With respect
to ad 335/6 he writes that many assyrians in persia were sold by the saracens,
probably as slaves.
276
With regard to the western regions, especially syria and palestine, banditry is a
topic quite often found mentioned in the relevant literary sources,
277
but only rarely
is reference explicitly made to nomads. occasionally arab raiding is being mentio-
ned.
278
among the bandits mentioned are robbers in general
279
and clearly Jewish
bandits,
280
some of whose actions are said to be anti-roman within the context of
the Jewish war ad 6670.
281
especially within this period we ind religious-politi-
cal groups labelled as brigands by Josephus, as there are especially the zealots and
John of Giala and his followers.
282
in any case, some brigand groups seem to have been connected somehow with
the desert or steppe or with the herding of animals. among the former is an egyp-
tian claiming to be a prophet, who during the reign of nero tried to force his ent-
rance to Jerusalem, coming from the desert.
283
a usurper to the throne after the
death of herod, named athrongeus, who raided the country, is said by Josephus to
273 Joseph. Ant. 18.31471.
274 abar, Tarkh (History) i 833.
275 abar, Tarkh (History) i 8389.
276 Theophanes, Chronographica 33.
277 on banditry in syria, palestine and north-arabia see generally dentzer (1985a), p. 39899;
Graf (1989a), p. 3849; isaac (1992), p. 7789; isaac (1998b); sartre (2005), p. 323, p. 1112
and p. 116; Wolff (2003), esp. p. 13556.
278 cass. dio 36.17.3; 37.7a; iust. Epit. 39.55; 40.2.
279 e.g. Joseph. Ant. 17.285; 18.274; 20.124; Joseph. BJ 1.204; 2.2289; 2.253; 2.271; Minh,
Zeraim, Pea 2.7a.; 2.8a; Bikkurim 1.2b; Moed 2.5a; Tosephta II, Moed 2.12; see IV. Seder:
Neziqim 6.1; 7; luke 10.30; John 10.1.
280 e.g. Minh, Zeraim, Berakot 1.3c. on the sicarii see Joseph. BJ 4.399409; 50913.
281 e.g. Joseph. BJ 2.2645; 6524. see sartre (2005), p. 1137.
282 John of Giala: e.g. Joseph. BJ 2.58594. see BJ 4.845. Zealots: e.g. Joseph. BJ 4.135147.
see rie (2005), p. 101
283 Joseph. BJ 2.2613. see rie (2005), p. 102; sartre (2005), p. 120.
271 The problem of nomadic allies in the roman near east
have been a shepherd (aoig).
284
The Greek term aoig, however, indicates ani-
mal husbandry in a sedentary context, in contrast to nomadic pasturage.
285

Well documented are the measures undertaken by herod the Great against bri-
gands ravaging syrian and palestine districts. apparently most important in this
context are the Trachonitis, the auranitis and batanaea, which were added to
herods realm by augustus with the purpose of securing these regions against bri-
gands, whose raids seem to have been directed especially against damascus.
286
of
these brigands Josephus reports that they lived in caves together with their cattle
and even robbed each other.
287
eventually they were forced by herod to turn to
agriculture.
288
While these bandits were not titled as nomads, there might be some
striking parallels to the raiding activities mentioned in the safaitic inscriptions,
which are analysed below. probably some of these groups lived a nomadic way of
live,
289
others lived rather from brigandage. Within the context of herods cam-
paigns against brigands in Judaea Josephus reports that there were families,
290

though it is not clear whether these had a tribal structure or just have to be regarded
as clans or families. describing the successful campaings of herod against bri-
gands ravaging the syrian frontier, Josephus mentions that the syrian villagers wel-
comed the end of the bandits and especially of their head ezekias.
291
This remark
shows that whole villages could suffer from brigandage.
The most prominent brigands of the region apparently were ituraean groups.
strabo reports that mountains in the lebanon were occupied by brigand arabs and
ituraeans, who had fortresses as operational bases from where they raided the
peasants in the clears and robbed merchants from arabia felix. according to strabo,
after the successful campaigns of pompey against the ituraeans security from raids
had been established in the region by roman soldiers.
292
in any case an inscription
from the augustan period states that Quirinus had fought against an ituraean
fortress in the mountains.
293
apparently the ituraean threat to the peasant popula-
tion seems to have ended after this period.
294
in modern scholarship the ituraeans
284 Joseph. Ant. 17.27884, esp. 278; BJ 2.605, esp. 60.
285 on the distinction see bar-Yosef and Khazanov (1992b), p. 2; sartre [1990], p. 44; scharrer
(2002b), p. 28990. on shepherds in syria-palaestine see butcher (2003), p. 1612; herz
(2005), p. 1934; scharrer and Zangenberg (2005a), p. 108; Villeneuve (1985), p. 1256. see
e.g. Mishnah IV. Seder: Neziqim 6.1.
286 Joseph. Ant. 15.344348; BJ 1.398400. see also strabo 16.2.20, with isaac (1989), p. 242.
see Gebhardt (2002), p. 247; isaac (1992), p. 626; sartre (1991), p. 634 (= sartre (1997),
p. 55); Weber (2002); Weber (2003); Wolff (2003), p. 1379.
287 Joseph. Ant. 15.346; 348.
288 Joseph. Ant. 16.271.
289 on references in ancient authors to nomads living in caves see scharrer (2002c), p. 2001;
scharrer (2004), p. 314. a nomadic way of live is assumed e.g. by dentzer (1985a), p. 399;
sartre (1989), p. 145.
290 Joseph. Ant. 14.4157; 42030; BJ 1.30913.
291 Joseph. Ant. 14.15960; BJ 1.2045.
292 strabo 16.2.1820.
293 ILS 2683,1113.
294 see dabrowa (1986), p. 94; isaac (1998b), p. 12640; macadam (1999), p. 280; sartre (1991),
p. 320.
272 Ulf scharrer
are titled mainly both as brigands and nomads,
295
and only seldom it is stated that
they were not.
296
There is actually no evidence at all that the ituraean groups who
lived from raiding had a nomadic way of life. curtius, describing alexander the
Greats siege of Tyrus, mentions arabian farmers on the lebanon ighting against
the macedonians.
297
if atkinsons assumption, that these could be ituraeans,
298
is
correct, already with respect to this early time the ituraeans should not be regarded
as nomadic. of course it is possible that curtius wrote from the perspective of his
own time, when the ituraeans had been paciied. apparently, roman rule at the be-
ginning seems to have provided security against robbery: strabo states that pompey
had freed the farmers in the clears from robbers of the mountains, and that roman
soldiers had established security for merchants from arabia felix against robbe-
ry.
299
The evidence discussed so far reveals that banditry, when mentioned in the
literary sources, does not necessarily imply nomadism.
300
later on in the period covered here, brigands are mentioned only occasionally.
Thus the Historia Augusta mentions that aurelian on his way to palmyra suffered
from mischief by the brigands of syria.
301
in this context it is very probable that
these brigands were not ordinary robbers, but that their actions were directed by an
anti-roman and pro-palmyrene motivation. hartmann assumes that these brigands
were bedouins of the region, originally ighting on the side of palmyra.
302
This to-
pic will therefore be dealt with in the ifth section.
some evidence for conlicts between nomadic groups and the roman authori-
ties might be found in the safaitic inscriptions. Unfortunately, however, the inter-
pretation of much of the epigraphic evidence is far from unambiguous. a couple of
inscriptions refer to an l rm. a matter of some dispute is the question whether l rm
refers to the people of rome or an indigeneous tribe rm.
303
There are a couple of
arguments in favour of both views. first of all rm is designated as l, a designation
generally not used by the writers of the safaitic grafiti for larger political units,
especially not the nabataeans. furthermore, two inscriptions mention the escape or
the expulsion of rm.
304
if rm is an indigenous tribe, these inscriptions would not
295 e.g. sartre (1990), p. 49.
296 e.g. Gawlikowski (1997), p. 41. see altheim and stiehl (196469) i, p. 3147 and p. 3512,
and Winkler (1993), p. 197, who write that the ituraeans originally were a nomadic tribe which
had become sedentary. according to Jones (1931), p. 269 in the roman period the ituraeans
had become sedentary.
297 curtius 4.2.24.
298 atkinson (1980), p. 300.
299 strabo 16.2.18; 16.2.20, with isaac (1989), p. 242.
300 sartre (2005), p. 233.
301 sha Aurel. 26.1.
302 hartmann (2001), p. 382.
303 l rm referring to rome: butcher (2003), p. 409; eissfeldt (1954a), p. 94; Graf (1989a), p. 375
6; isaac (1992), p. 73; moors (1992), p. 311. l rm referring to a tribe rm: Khraysheh (1995),
p. 411. see hartmann (2001), p. 84; sartre (1985), p. 68. on the problem see macdonald
(1993), p. 3289.
304 escape of rm: pUaes iV c 675. expulsion of rm: CIS V 12. see the critical remarks of mac-
donald (1993), p. 330.
273 The problem of nomadic allies in the roman near east
cause any problems, as their wording is that of similar events with other tribes
discussed below. if rm stands for rome, it is a matter of question whether the in-
scriptions refer to either the defeat of a roman force by nomadic groups or by ano-
ther force, for example the palmyrenes or persians. assuming that the defeat was
caused by nomadic groups, this would require a strong military organization of
these. a inal conclusion on this issue is presently beyond reach. a further inscrip-
tion, similarly using the wording of tribal raiding, mentions rm raiding the naba-
taeans.
305
again, if rm means rome this inscription probably refers to the roman
annexation of nabataea in ad 106, which, however, appears to have been without
violence.
306
if rm refers to an indigenous tribe, this inscription would not cause any
problems either, since the nabataean realm is occasionally referred to as object of
raids in the safaitic inscriptions.
307
however, no inscription has been found so far whose writer identiies himself
as belonging to the l rm, and there is no ethnic group mentioned in literary texts
which would it the safaitic name. furthermore, some safaitic inscriptions mention
rebellions against rm, a notion not found in texts referring to intertribal conlicts.
Thus, a certain abd states that he had escaped from a group of warriors (mrdt) in
the year the nabataeans revolted against rm.
308
Unfortunately, nothing is known
about such a revolt, but it can hardly be imagined that the nabataeans revolted
against a group that was so small that Greek and latin sources did not take any
notice of it. Therefore a revolt against the power of rome, probably after ad 106,
might be assumed. probably the group of warriors which is mentioned in the in-
scription appeared in the course of this revolt, although it is also possible that they
were warriors of another tribe without any connection to the nabataean revolt at
all.
similarily, some safaitic inscriptions mention rebellions of other ethnic groups
against l rm, especially by the Aw.
309
other inscriptions refer to people being
being distressed by rm or escaping from rm.
310
Whereas statements of these kinds
are not to be found in inscriptions mentioning intertribal conlicts, the wording of
other references to conlicts with rm are, for example, ights with rm, the rm attak-
king, the rm killing the tribe of Salmn, and inally the tribe of Baad being robbed
by another group, probably rm.
311
305 CIS V 4866.
306 on the annexation of the nabataean kingdom see bennett (1997), p. 1756; bowersock
(1994b), p. 1128; bowersock (1983), p. 7584; butcher (2003), p. 445; dijkstra (1995),
p. 37; Gebhardt (2002), p. 87105; millar (1993), p. 926; sartre (1985), p. 6872; sartre
(2005), p. 867; 1334; shahd (1984b), p. 1921.
307 see below.
308 Wh 2815.
309 Aw: CIS V 1292; 4438 (= SIJ 39). others: Khraysheh (1995), nr. 6; SIJ 78. see also SIJ 707.
see Graf (1989a), p. 3768; Graf (1989b), p. 1523; sartre (1982a), p. 1278.
310 being distressed: e.g. PUAES iV c 640. escape: e.g. PAAES iV 5,59; CIS V 1713; 3688;
3721; 3776; 37878; PUAES iV c 87; 94; 157; 406; SIJ 352.
311 attack on a certain mtum: CIS V 4439. Salmn being killed: CIS V 1151. Baad being rob-
bed: CIS V 4447.
274 Ulf scharrer
one inscription mentions rm wintering near abila.
312
if rm was an indigenous
ethnic group this would not cause any problem, as it could stand for a nomadic
group spending the winter at that place; if rm stands for rome, the inscription can
only be understood as a reference to roman troops being on, or returning from,
campaign who used abila as winter quarters. finally, some inscriptions mention a
war of rm.
313
so far, the phrase war of is only found with respect to greater
political and ethnic units such as the nabataeans, the Jews or dedan in a Thamudic
inscription.
314
macdonald convincingly states that the references to a war of the
nabataeans does not necessarily imply that the writers of the inscriptions were in-
volved in the struggles.
315
The same can be said about the references to a war of
rm.
There are other inscriptions which, with caution, could be interpreted as evi-
dence for roman-nomadic conlict. an inscription states that the emperor (hmlm)
ined the tribe of Aw.
316
The writer of this inscription declares to be on the look-
out for his imprisoned fellows. if the roman emperor is meant and if rm stands for
rome, the ining of the Aw, and probably the imprisonment of some members of
this ethnic group, could possibly be connected to the above-mentioned revolt of the
Aw against rm. it is debatable, however, whether the emperor mentioned is the
roman one. he is not designed as king (mlk), but as emperor. in this context in-
scriptions referring to a revolt of a certain damas need to be discussed briely. The
irst inscription, written by a certain Khair of the l of Msikat, mentions the revolts
of dama and of a certain murib. The writer states that he is on the look-out for
the enemy and asks allt and -ar for securitiy and existence.
317
another in-
scription, referring only to the revolt of dama, is written by a certain magd of the
l of aif.
318
in an important article Winnett interprets these inscriptions as a revolt
against the nabataean king rabbel ii (ad 71106), against the background of the
weakness of the nabataean empire.
319
While Winnetts arguments are certainly stri-
king, his interpretation is of course not conclusive. at least one other inscription
records that murib revolted against the emperor.
320
if this is the roman emperor
and not the nabataean king there are three ways to interpret the evidence: firstly,
dama revolted against the nabataean king and in the same time murib revolted
against the roman authorities. secondly, both revolted against rome, and no revolt
against the nabataean authority is implied. Thirdly, both revolted against rome
and the nabataean kingdom, regarding a rebellion against an ally of rome as a re-
312 CIS V 1868.
313 PAAES iV 5,45; CIS V 533; 2815; 4448.
314 War of the nabataeans: PAAES iV 5,45; CIS V 220; 3680; Wh 2113. according to sartre
(1985), p. 68, these inscriptions refer to ad 106. War of the Jews: SIJ 688. War of dedan: Ph.
266.
315 macdonald (1991), p. 111.
316 PUAES iV c 644.
317 SIJ 287.
318 SIJ 823.
319 Winnett (1973), followed e.g. by bowersock (1983), p. 1556.
320 SIJ 281.
275 The problem of nomadic allies in the roman near east
volt against the roman emperor himself. so far, none of the three ways of interpre-
tation can be favoured.
Two inscriptions written by two brothers from the tribe of Daaf might also
refer to quarrels with the roman sedentary world: both were written in the year in
which the tribe of Qamar attacked the government or empire (sln).
321
in addition
some inscriptions refer to an escape or protection from the government or em-
pire.
322
Unfortunately it is not clear whose government is meant, be it roman, of
another tribe or of a local or regional power like the nabataeans. a certain malkat
in an inscription near namra asks -ar for protection against the government,
and a certain unain states that he escaped from the namrat of the government
or empire (nmrt hmln) to the tribe of Aw.
323
at least in these cases the roman
government can be assumed, and it seems that the Aw provided some sort of
protection against it. in this context it is noticeable that there was a roman castel-
lum at namra.
324
The inscriptions should probably be read in the context of the
revolt of the Aw against rm. in any case, it is still debatable whether sln always
refers to the roman authorities, as is commonly assumed.
325
besides the inscriptions discussed above, others also mention captured per-
sons.
326
it is possible, though certainly not imperative, that these were imprisoned
by the roman authorities. at least some of these inscriptions might also refer to
persons being captured by members of other tribes, and the same can be said about
inscriptions mentioning a guard arresting members of a tribe:
327
hartmann assumes
roman police actions,
328
but it cannot be decided clearly whether in each case the
guard mentioned is roman, nabataean, Jewish, palmyrene or a local one.
There are other inscriptions of which the interpretation is ambiguous. one in-
scription refers to a rebel against the king.
329
another one mentions a year in which
a tower of the king in the high region of Yarkhn perished.
330
This inscription has
been interpreted as a reference to the extermination of a roman tower and its garri-
son by arabs.
331
a grafito mentions the escape from Qar naqat, about which
nothing is known so far.
332
and inally some inscriptions refer to a king (mlk) wa-
ging war against certain tribes.
333
it seems debatable whether all these texts are to
be read in a direct roman context: it is not conclusive at all that the king mentio-
ned in the inscriptions is a roman one. it could also be a reference to a nabataean
321 CIS V 1952; 2802: snt trq l qmr hsn.
322 PUAES iV c 424; 1013.
323 PUAES iV c 539; 1013.
324 see moors (1992), p. 345; CIL iii 1114.
325 e.g. ryckmans (1942), p. 135.
326 CIS V 2113; 2172; 2292; 2552; 4698; PUAES iV c 183; 643; Wh 2999; Wh 1565; 1675;
ITham hU451.
327 Wh 1253.
328 hartmann (2001), p. 823.
329 ISB 163.
330 CIS V 3064: snt tsp r hmlk bsrn yrkn.
331 ryckmans (1942), p. 135, followed by Graf (1978), p. 5.
332 PUAES iV c 653.
333 Wh 1700a: against the SBTT. PUAES iV c 1065; SIJ 705.
276 Ulf scharrer
or perhaps even a Jewish king. it has furthermore been stated that mlk could also
designate a local tribal leader,
334
but for this view there is no support in the safaitic
inscriptions. The places mentioned could therefore also have been nabataean or
Jewish posts.
a popular topic of the inscriptions is being on the look-out (tr), though it is
seldom speciied what is actually being watched for.
335
it seems that look-outs
could be special places or structures, since some grafiti mention the ownership of
a look-out.
336
some inscriptions do specify what is being watched, for example
animals, enemies, more rarely nabataeans, rm and other peoples, the own men or a
saddle back.
337
finally, Tanner has interpreted a Greek inscription from the area of
isma, according to which romans always win, as referring to nomadic incursions
during the Jewish war.
338
of course this interpretation is possible, but not neces-
sary: there is no reference to nomads at all in the inscription, and it might equally
refer to the above-mentioned unrest within palestine.
however ambiguous the interpretations of the safaitic texts remain to be, there
seems to have been at least some nomadic unrest directed against the roman autho-
rities, especially the rebellion of the Aw: bringing the evidence on their revolt
together, it appears most probable that the roman authorities were indeed faced
with them. furthermore, besides the somehow contradictionary evidence on rm, on
the whole it still seems to be the case that the term designated rome. if so, at least
some evidence of conlict between the roman authorities and nomadic groups can
be found. however, there is, at least in the safaitic inscriptions, no evidence at all
of a large scale attack of a nomadic group against the roman empire, and we should
rather assume some sorts of raids.
339
in the case of the Aw, the breaking with
rome seems to have been purely onesided, as rome apparently answered the at-
334 on mlk as nabataean king: Graf (1989a), p. 368. on mlk as a tribal leader shahd (1984b),
p. 31 n. 38.
335 e.g. CIS V 5; 412; 456; 4851; 545; 1345; 220; 235; 294; 341; 344; 351; 361; 406; 434;
SIJ 87; 135; 235; 240; 263; 323; 534; 590; 681; 699; 718; 729; 764; 776; 781; 809; 814; 851;
Wh 29; 66; 17475; 18788; 224; 267; 271; 317; 319; 334; 350; 357; 380; 407; 426; 580; 599;
60102; 612; 701; 705b; 898; 968; 982; 1031; 1116; 1127; 1148; 1591a; 1601; 1763; 1906;
1940a; 1940c; 1974; 1978; 1983; 2137; 2255; 2328; 2381; 2534; 2806; PUAES iV c 184; 354;
424; 613; 660; 895; 1010; 1069; ISB 87; 90; Zeinaddin (2000), inscr. 9; 11. see also being on
guard (r): SIJ 435; Wh 608.
336 Wh 245; 270; 318; 1516; 1612; 166364; 1916; 1961b; 2554.
337 animals: e.g. Wh 54; 359. camel: Wh 1996a; horse: Wh 54. mount: Wh 2837. sheep: SIJ
752 (?). donkeys: SIJ 784. live-stock: Wh 3724. enemies: e.g. SIJ 420; 808; 8534; 946;
Wh 155; 290; very probably 390; very probably 613; 624; 947; 2539; 3286; PUAES iV c 210;
469; 576; 606; 618; 698; 701; 708; 1263. nabataeans: e.g. SIJ 855; Wh 157. romans: e.g.
PUAES iV c 709. saddle-bag: Wh 372. rain: Wh 72; 112224; 1215; 2005; 2256; 2278;
3401; 3533b. son and daughter: Wh 577. father: PUAES iV c 709; Wh 1901. fellow: ISB
365a; PUAES iV c 1056; SIJ 816; 837. rebell: ISB 66. firewood: Wh 1902. lions: SIJ 14.
see also Winnett and reed (1970), p. 131 nr. 73; PUAES iV c 210; 400; ISB 15. awlat: SIJ
716.
338 SEG Xl 1524 (= AE 1990, 1016), with Tanner (1990), p. 1858.
339 see butcher (2003), p. 409; Gebhardt (2002), p. 247 n. 5; isaac (1998b), p. 124; macdonald
(1993), p. 3334.
277 The problem of nomadic allies in the roman near east
tempts simply with ining the group and probably with taking some of its members
hostage.
in the third century ad there are some references to campaigns by roman em-
perors against arabs and saracens. an important campaign is the one by septimius
severus. according to eutropius and the Historia Augusta he defeated the parthi-
ans and the inner arabians (Arabes interiores) and hence received the title Arabi-
cus, among others.
340
This campaign is sometimes viewed as testifying to an in-
creased nomadic menace.
341
however, according to cassius dio, severus also
campaigned against hatra, during which siege he lost many soldiers since he was
repeatedly attacked by arabian cavalry.
342
Thus, following the notion of Arabia in
the Graeco-roman authors, as discussed above, the arabian campaigns of septi-
mius severus were not necessarily directed against nomadic groups, although ights
with some of these groups could have been involved.
Towards the end of the third century ad there are occasional references to the
campaigns by roman emperors against the saracens. The Panegyrici Latini men-
tion that maximian in ca ad 290 subdued the saracens.
343
finally, the reorganiza-
tion of the eastern frontier under diocletian is often regarded as a response to an
increase of nomadic pressure, as is the emperors successful campaign against the
Saraceni.
344
indeed, John malalas records saracen incursions during diocletians
rule.
345
saracen raiding is also testiied by eusebius: with respect to ca ad 250 he
writes that leeing christians were sold into slavery by the saracens.
346
apart from the rather ambiguous literary sources on the West some inscriptions
too are dubious for the uncertainty of their readings. a latin inscription from a ro-
man fort near Khan Kosseir, northeast of damascus, dated to ca ad 193, states that
it was built with respect to public security and the terror of the scenitic arabs.
347

in addition, an inscription from the sinai dated to ad 18991, which i have quoted
above, records that arabs devasteded the land.
348
if the semitic term Arab means
desert or steppe dwellers, this inscription would testify to some nomadic raiding on
the sinai.
an important document is the Greek section of an honorary inscription from
palmyra for oglu from ad 199.
349
it records how oglu was honoured because
340 eutr. 8.18.4; sha Sev. 9.910. The titles given by eutropius are conirmed by the epigraphic
evidence: see e.g. bernand (1999), nr. 14.
341 see hartmann (2001), p. 823.
342 cass. dio 76.11.2. see luther (1997), p. 173 n. 284.
343 Pan. Lat. 11.5.4, with nixon and rodgers (1994), p. 89. see also millar (1993), p. 434.
344 leadbetter (2002), p. 868; parker (1986b), p. 643. on diocletians campaign against the sa-
racens paneg. lat. 11 [3].5.4. see Graf (1989a), p. 3467; isaac (1992), p. 73; (1998a), p. 125
n. 12
345 malalas 12.38.
346 euseb. Hist. Eccl. 6.42.4.
347 CIL iii 128: in securitatem publicam et scaenitarum Arabum terrorem. The reading of the
inscription is not conclusive: see Gebhardt (2002), p. 24950.
348 NSI 107. see sartre (1982a), p. 127. Kuhnen (1991), p. 32930 interprets the inscription as
referring to conlicts between rome and nomads, but this is not at all sure.
349 Inv. X 44,4 (= PAT 1378; asad and Yon (2001), nr. 16; SEG Vii 139 (Greek section only);
278 Ulf scharrer
he took part in frequent campaigns against the nomads.
350
Unfortunately the pal-
myrenean section does not have an equivalent for the term. in any case it is obvious
that palmyra had to protect its caravans against assaults, at least some of which
were nomadic.
351
The threat posed by nomadic groups should not be exaggerated,
since the oglu-inscription is the only one which explicitly mentions nomads.
Whereas it is sometimes stated that nomads were the greatest danger to palmyrene
caravans,
352
the evidence for this is rather sparse and ambiguous. a palmyrene in-
scription from ad 144 explicitly mentions robbers (line 7: gyy) under the com-
mand abdallat the ahitaia (line 15: [a]ooo0E eri0gvE / line 7: bdlt
yhty).
353
however, the robbers mentioned in the inscription do not seem to have
been nomadic. it is stated explicitly that abdallat brought the robbers together (line
15: toi uaoutou ouvo0rioi / line 7: dy kn ). it is not clear either whether Ahitaia
is a tribal or a place name.
354
finally, it ought to be emphasized that on the whole
there is not much direct evidence in the palmyrene inscriptions for caravans in dan-
ger of being plundered.
355
and there is no evidence either that palmyra had treaties
with nomadic groups to secure caravan trafic.
356
There does not seem to be any good evidence in the safaitic texts for raiding of
palmyrene territory. only one inscription might be interpreted as a reference to
some sort of conlict: a certain imrn states that he came to Tadmor and killed a
man called ha-Mlik bn FTNY.
357
according to Winnett and harding, the editors
of the inscription, the name FTNY, which so far is not attested in safaitic inscripti-
ons, could be of palmyrene origin. however, two points have to be stressed: irstly,
the conlict might have been purely between individuals, and secondly, the attribu-
tion of names to certain ethnicities often appears to be quite dubious.
358
in this
context the assumption of the title Arabicus Maximus by the palmyrene Augustus
Vaballathus needs to be mentioned briely.
359
hartmann suggests that Vaballathus
in ca ad 270 assumed this title after his victories against arab nomads who raided
schuol (2000), p. 846, with comments; hartmann (2001), p. 523). see altheim and stiehl
(196469) i, p. 346; ingholt (1976), p. 129; matthews (1984), p. 168; millar (1993), p. 332.
350 lines 34: xoi oio to / xoto tev voooev ototgyio ouvoorvov.
351 bowersock (1989), p. 79; hartmann (2001), p. 427; maraqten (1996), p. 2302; sartre (1991),
p. 332; sartre (2005), p. 238. on palmyrene caravans being in danger of being harassed see
e.g. the inscriptions PAT 0197 (= Kaizer (2002), p. 602; schuol (2000), p. 845); Kaizer
(2002), p. 633. see generally dirven (1999), p. 37; Gebhardt (2002), p. 2856; hartmann
(2001), p. 64; matthews (1984), p. 1678; millar (1998b), p. 1323. dirven (1999), p. 37
n. 144 interprets PAT 0197 as referring to nomadic assaults, but this is not conclusive.
352 e.g. maraqten (1996), p. 232. see also matthews (1984), p. 164.
353 Kaizer (2002), p. 623. see maraqten (1996), p. 232; millar (1998b), p. 133.
354 see dirven (1999), p. 378 n. 144. according to maraqten (1996), p. 232, the robbers mentio-
ned in the inscription were nomads.
355 see maraqten (1996), p. 231.
356 Thus stated e.g. by maraqten (1996), p. 232; matthews (1984), p. 164.
357 Wh 2833a, with comm.
358 macdonald (1999), esp. p. 2546; contini (1987), p. 278; Kaizer (2002), p. 57; Kaizer
(2004a), p. 170 n. 32. see generally scharrer (2006), p. 3525, with further references.
359 ILS 8924. on Vaballathus see hartmann (2001), p. 24259.
279 The problem of nomadic allies in the roman near east
palmyrene territories and some of whom might have been in persian service.
360

again, it cannot be stated conclusively to what extent sedentary arabs too might
have been referred to by this title.
i have mentioned above some possible references in the safaitic inscriptions to
the nabataean kingdom. in scholarship nabataean fortiications are commonly as-
sumed to have had the purpose of controlling and warding off nomads.
361
indeed,
besides the possible evidence discussed above in the context of roman-nomadic
conlicts, there is occasionally some direct evidence in the safaitic inscriptions with
regard to the nabataean territory as an object of raiding.
362
a certain JFN states that
he has become poor and he asks ru for retribution (qbt) from the nabataeans.
363

apparently this JFN regards the nabataeans as the cause of his poverty, but the
speciic conditions of this remain in darkness. maybe JFN had to pay tribute, or
perhaps he was plundered by nabataeans. furthermore, a certain mulaikat asks
ru for help to get booty from enemies, from the edessenes (? RHY), the naba-
taeans and the awlat.
364
at least in the case of the (probable) raid into edessene
territory the god helped him, since he states in another grafito to have carried off
booty from the edessenes.
365
finally an inscription asks for rus protection against a certain mliku.
366

This inscription could be interpreted as referring to one of the nabataean kings of
this name, but of course the request could equally have been directed against the
leader of a nomadic ethnic group, and it could even belong to the context of an in-
dividual feud. The same can be stated about an inscription which records that a
certain rithat took a Wakat prisoner:
367
either this inscription refers to a naba-
taean king aretas or again the inscription refers to a rather individual problem.
The inscriptions by mulaikat reveal that, apart from the nabataeans, other local
powers could also be subject to nomadic raiding. some inscriptions cannot be inter-
preted conclusively, as some of them for example mention rebelling (rg) and re-
bels (rgy), generally without speciing what or whom is rebelled against.
368
fur-
thermore an inscription mentions the conquest of DB (qhr db).
369
Unfortunately it
is not clear whether DB is a region or an ethnic group, and it cannot be known eit-
her who was the conqueror, whether it was one of the local powers or a nomadic
ethnic group. another inscription mentions the obliteration of the tribe of RFT.
370

in this case it is not clear either whether this tribe became obliterated by roman or
360 hartmann (2001), p. 2668.
361 e.g. bowersock (1983), p. 1545; Gichon (2002), p. 188; parker (1990), p. 222. see sartre
(1991), p. 64 (= sartre (1997), p. 55); shahd (1984b), p. 22.
362 e.g. the rahya raiding the nabataeans: CIS V 2570. see also Wh 3925. see bowersock
(1983), p. 154; Gichon (2002), p. 188; parker (2002), p. 78.
363 Wh 3747.
364 Wh 3736a.
365 Wh 3605.
366 CIS V 2172.
367 SIJ 296.
368 e.g. CIS V 303; 4119; ISB 66; SIJ 39; 289.
369 SIJ 851.
370 Wh 3931.
280 Ulf scharrer
regional powers or by another tribe. finally a certain iddn boasts to have raided a
large populace (nm mm kbr).
371
The term for the people in this inscription is not
l, but m, so it seems improbable that another nomadic ethnic group is referred to
and more likely that a sedentary community was the object of the raid. however, it
cannot be stated conclusively what the term large (kbr) implies, be it a larger vil-
lage, even a city, or the territory of a local power. The same has to be stated about
occasional references to oppressions:
372
unfortunately it is unknown who oppressed
whom.
besides the raiding of local powers there is some occasional evidence in the
safaitic inscriptions for raids into the sedentary and agricultural sphere in general.
however, much of this evidence is rather ambiguous. Thus, the reading of an in-
scription mentioning raiding in arable lands is uncertain.
373
another inscription sta-
tes that a certain otherwise unknown place RL was destroyed by the tribe of
BB.
374
Unfortunately it is unclear whether RL refers to some sedentary settle-
ment or a region of nomadic dwelling. also the interpretation of an inscription
which mentions a raid into pastures is unclear:
375
the pastures in question could
have been used in sedentary contexts or by nomadic groups.
indeed, raiding and the desire for booty is a prominent topic in the safaitic and
in some Thamudic inscriptions. many inscriptions include a request to the gods for
booty, or (afterwards) a thanking them, most often ruay/ru and allt, but also
other deities.
376
some inscriptions express a general desire for, or aquiring of, boo-
ty.
377
it is remarkable that, at the same time, these gods are occasionally asked for
peace and security.
378
apparently the writers of these inscriptions did not see any
contradiction in their attitudes, and especially the inscriptions asking the gods for
successful raids and peace at the same time therefore reveal that plundering has
371 ISB 390.
372 CIS V 223 (?).
373 ISB 438.
374 Wh 2255.
375 CIS V 3345.
376 To allt: CIS V 8; 70; 882; 1371; 1859; 2207; 2481; 4262; 4332; 4986; ISB 18; 76; 167;
PUAES iV c 418; 1196; 1252; 1263; SIJ 37; 45; 56; 78; 88; 105; 114; 658; 700; Wh 24; 53;
153; 159; 167; 173; 177; 181; 393; 397; 575; 577; 599; 711; 732; 947; 999; 1019; 1022; 1030
31; 1070; 1105; 1666; 1697; 1725a; 1850; 2125; 2818. To baal: CIS V 3180. To baal-samy:
CIS V 3149; PUAES iV c 259. To ruay / ru: e.g. CIS V 1084; 1556; 1757; 1961; 2266;
2315; 2514; 2682; 2765; 3085; 3121; 3156; 3382; 3450; 3598; 3811; 4055; 5047; ISB 78; SIJ
125; 184; 224; 253; Wh 1906; 2135; 2187; 2375a; 2812; 2975; 3731. To Yitha: CIS V 785;
2636; 3168. ru and allt: PUAES iV c 160. on the pantheon of the safaitic inscriptions
see e.g. dussaud (1907), p. 11656; dussaud (1955), p. 1407; oxtoby (1968), p. 205. see
also sartre (2005), p. 301
377 e.g. CIS V 2519; 4978; Wh 689; 1796a; Ph. 292z
378 inscriptions asking for peace and booty from allt: PUAES iV c 619; 1196; 1252. allt
being asked for security and booty: SIJ 80; 261 (?); 729; 840; Wh 947; PUAES iV c 1263.
request for security and booty to -ar: SIJ 59. request for peace and booty to -ar:
ITham Jsa658. request for security and booty to -ar and allt: Wh 2837. request to
ru for relief from enemies while having departed for plunder: SIJ 816. request for booty
and protection from evil to rua: Wh 3730.
281 The problem of nomadic allies in the roman near east
indeed to be regarded as a common feature of the nomadic life of the writers of the
inscriptions, which is relected in their religious mentality.
Whereas it is not clear in each case whether a raid was directed against seden-
tary or nomad communities,
379
much evidence from the safaitic inscriptions points
to intertribal conlicts, which has occasionally been discussed in modern scholar-
ship.
380
some inscriptions state that a tribe expelled (br) another,
381
two inscrip-
tions mention that the tribe of Qumair did injury to the tribe of Chumaiy,
382
and one
inscription mentions the Aw warring with the Baad.
383
The Baad are also men-
tioned as making peace. The writer of this inscription stems from the tribe of
Nabar.
384
however, it must remain open with whom the Baad made peace. pro-
bably this inscription refers to a peace agreement between the Baad and the Aw,
referred to explicitly in another inscription.
385
a conclusion of a peace is also men-
tioned in another inscription, according to which the tribe of Msikat made peace
it has to remain open with whom.
386
another text mentions the Aw ighting
with the abh, apparently with the purpose of getting booty.
387
The Aw as well
as the aif could be subject to raids, as is clear from an inscription that states that
there were incursions against the pastures of both groups.
388
a certain Taymo asks
for the protection of the Gadd-Aw and the Gadd-aif against the tribe of yi
who had apparently killed or captured some of his companions.
389
in this context
an inscription mentions that the aif were cut to pieces (btr), unfortunately without
saying by whom.
390
finally in some inscriptions gods are asked for protection from
the tribe of aif.
391

i have already mentioned an inscription which asks for booty from the naba-
taeans, probably the edessenes and the awlat. another inscription mentions the
ighting against awlat, and in this case it is unclear too whether a region or the
ethnic group itself is meant,
392
since the term l is not used. finally a certain
adddat boasts to have expelled (br) the awlat and to have held them back
(bs).
393
as well as the awlat, the Thamd could be involved too in conlicts
with the writers of the safaitic inscriptions.
394
379 besides the requests to gods quoted above see e.g. CIS V 4276; SIJ 905.
380 sartre (1989), p. 1656; Wolff (2003), p. 155. on intertribal conlict for economic reasons at
the beginning of the islam see orthmann (2002), p. 3358.
381 PUAES iV c 435; Wh 1231. probably CIS V 786.
382 PUAES iV c 2545.
383 CIS V 2577.
384 CIS V 4446.
385 CIS V 4394.
386 CIS V 787.
387 SIJ 59.
388 CIS V 2446.
389 CIS V 2795.
390 Wh 1849.
391 Wh 1698: request to -ar; Wh 1029: request to ru.
392 ISB 365a. With oxtoby (1968), p. 96.
393 Wh 1231.
394 Wh 1276; 3792a; 3792c. see Graf (1997b), p. 180.
282 Ulf scharrer
a few inscriptions mention the seizing of animals,
395
which shows that raids
could also occur between rather nomadic groups. accordingly a certain Y[]
writes that he was poor and went on a raid.
396
in another inscription allt is asked
for granting plunder to somebody whose supply had decreased.
397
it is clear that in
this case raiding is considered as some sort of supplementary economic strategy. in
a further inscription a certain man states that enemies drove him to despair. in
the same inscription he also says that his goats bore young.
398
it can be assumed
that his enlarged herd was his enemies object. in this context, many safaitic and
some Thamudic texts mention somebody having fought (zz) and the killing of sin-
gle persons, and some inscriptions refer to raids, mostly without giving further de-
tails.
399
a further inscription mentions the loss of camels by robbery during migra-
tion.
400
other grafiti state that somebody has been robbed, deceived, is being per-
secuted, all as well without further details.
401
finally, the safaitic inscriptions occa-
sionally mention rescue actions by members of a group to free a companion.
402
as
stated above, it is not clear whether some of the captured ones were imprisoned by
roman or nabataean authorities. in the same way references to slaves
403
are difi-
cult to interpret: in most cases it cannot be said whether slaves were captured in the
course of raids against other tribes or sedentary peoples, or whether they were
bought in the context of commercial enterprises (see below).
some inscriptions record their writers fear of enemies.
404
basically there were
three ways of reaction to such mischief. The irst one was revenge (tbl). in the sa-
faitic and some Thamudic texts most often blood revenge is mentioned, which is
frequently asked from the gods.
405
in addition revenge is taken when one has been
395 e.g. donkeys: Wh 218. camels: Wh 179; 676. animals not speciied: CIS V 4848.
396 Wh 651.
397 ISB 76.
398 Wh 3562.
399 fighting: e.g. CIS V 1534; ISB 69; 154; Wh 53; 65; 123 (?); 128 (?); 290 (?); 782; 983; 995;
1147; 1253; 1591d; 1601; 1662; 2113; 2854; PUAES iV c 462; 573; 1261; SIJ 976 (?); Zein-
addin (2000), drawings 23; ITham Jsa26; Ph. 270e; 297m. see also PAAES iV 5,120; SIJ 131.
Killing: e.g. CIS V 16; 743; 2113; 3149; 4404; 4423; Wh 595; 607; 1198; 1200a; 3000; PAAES
iV 5,134; PUAES iV c 2356; 297; 579; 587; 589; 590; 653; 658; 669; 1157. raiding: e.g.
CIS V 3095; 4305; Wh 94; 1661; 2818; 3731; PUAES iV c 160; 742. War: CIS V 2209. on
enemies in general see e.g. Wh 1019; lidzbarski (1908 and 1915) i, p. 42.
400 CIS V 2552.
401 being robbed: CIS V 1868; 2657; 3360; 4919; 5099. The tribe of Khil being robbed: CIS V
2192; 2297; 2318. being deceived: Wh 2584; 3134; 3393. being persecuted: CIS V 5088;
5121.
402 e.g. Wh 1518; 575; 610; 619a; 1027.
403 e.g. PUAES iV c 14243; Wh 346; 865.
404 e.g. PUAES iV c 306; 330. enemies in general: CIS V 57; 70; 420; 924; 994; 1362; 1432;
2023; 2683; 2816; 3062; 3754; 4261; 4284. see also Wh 1974.
405 request for blood revenge to ruay/rau: e.g. PUAES iV c 460b; SIJ 825; Wh 80. To allt:
e.g. Wh 367; 1220; 17812; 2911; PUAES iV c 242; 305; 385; 518; 679; 751; SIJ 750 (?). To
allt and -ar: PUAES iV c 461. other mentionings of blood revenge: e.g. PUAES iV c
233; 1266; Wh 1354; lidzbarski (1908 and 1915) i, p. 44. see also Wh 3596a. on blood re-
venge at the beginning of the islam see orthmann (2002), p. 30215.
283 The problem of nomadic allies in the roman near east
wronged or robbed.
406
sometimes vengeance in general is asked from the gods,
again occasionally combined with a request for peace.
407
apparently, these requests
were no pure threats: a certain LF states that he has killed someone and is now
afraid.
408
The second reaction was escape. besides the inscriptions mentioning es-
cape from rm, which have been discussed above, there are other texts mentioning
hiding or leeing from enemies, other tribes and their territories and occasionally
even from ones own companions.
409
The third way was to pray to the gods for se-
curity and protection. i have already mentioned the joint requests for both security
and booty. The gods who were asked for plunder were similarly begged for security
and peace in general, as well as more speciically for protection from enemies.
410
in
addition some inscriptions express a general desire for peace or security without
naming any gods.
411
Whether in these contexts the enclosures found had the pur-
pose of protection from raids is debatable.
412
With respect to booty sartre argues that also proit may be meant by that.
413
in
this context the interpretation of a Greek epigram from sane is interesting. it states
that a certain Kaianos erected a monument r ioig ototig, and that Kaianos was
406 being wronged: request to allt: e.g. PUAES iV c 1234. request to ruay: e.g. Wh 80.
being robbed: Wh 179.
407 request for vengeance to allt: rykmans (1939), nr. 31; 39G. To ru: ITham hU80;
hU789,8; littmann (1940), nr.129. request for revenge to nahy: ITham hU130. To Khil:
ITam hU760. To attarsamm: ITham hU 645; Jsa 388. request for peace and vengeance to
allt: PUAES iV c 518. see also ITham hU800; littmann (1940), nr. 136.
408 Wh 3364.
409 hiding or leeing from enemies: e.g. ISB 10; Wh 390; 398. escape in general: e.g. CIS V 92;
295; 617; 1742; 2076; 2820; 3097; 3111; 3931; SIJ 156; 183; ITham Jsa 547; Wh 153; 390;
1605; 1862; 2561; 3342; 3896. request to allt for escape: e.g. Wh 135. fleeing from ano-
ther tribe: Wh 153. fleeing from the own companions: Wh 785.
410 requests for peace/security: To allt: e.g. PUAES iV c 90; 155; 156; 161; 189; 237; 245;
257; 326; 331; 353; 436; 469; 578; 595; 602; 606; 613; 618; 635; 649; 660; 664; 675; 678; 698;
708; 709; 718; 719; 7278; 753; 895; 1133; 1182; 1195; 1198; 1297; Wh 1849; 2139; 3562;
SIJ 134; 240; 263; 284; 2956; 321; 350; 534; 681; 806; 837; 844; 897; Wh 1899; 19012;
1994; 2022; 2110; 2127; 2129; 2145; 2168; 2184; 2284; 2302; 2339; 2802. To baalsamy: Wh
54; PUAES iV c 258; 260; 722. request for security to -ar: SIJ 296; 745; Wh 61; 1698;
1771; 2706. request for peace to Gadd-Aw: PUAES iV c 1214. request for security to
Gadd-aif: SIJ 132; 9112. request for peace/security to ruay / ru: ISB 10; 65; 72; 75;
ITham hU767; SIJ 714; 716; 746; PUAES iV c 286; 293; 303; 734; 1095; Wh 75; 91; 1219;
1672; 1772; 2833a; 3286. To ai-ha-Qaum: PUAES iV c 417; Wh 633; 1978. To Yitha:
PUAES iV c 262. request for security to allt and -ar: SIJ 208; 296; 300. request for
protection to allt and Gadd-Aw: PUAES iV c 306. request for peace to allt and ruay:
PUAES iV c 353. request for security/protection from enemies: To allt: CIS V 1278; 1871;
2003; 2194; 4355; PUAES iV c 17980; 184; 210; 330; 374; 404; SIJ 37; 715; Wh 173; 2125.
To baalsamy: CIS V 2129. To Gadd-aif: SIJ 912; Wh 613; 1725b. To ru: e.g. CIS V 834;
879; 1077; 1548; 1930; 2187; 2315; PUAES iV c 495; Wh 2163. To ai ha-Qaum: CIS V
586. To Yia: Wh 2539. To allt and Gadd-Aw: e.g. PUAES iV c 184; 306.
411 e.g. dussaud and macler (1901), p. 367, nr. 11; p. 612, nr. 96; p. 689, nr. 124; dussaud and
macler (1901), p. 1012, nr. 299; p. 117, nr. 398. General desire for protection against ene-
mies: e.g. CIS V 31; 4251; PUAES iV c 427; Wh 973.
412 This has been suggested by eissfeldt (1954b).
413 sartre (1991), p. 333 (= sartre (1997), p. 317). see also sartre [1990], p. 43.
284 Ulf scharrer
aougrio.
414
merkelbach and stauber suggest that this inscription was set up by
an arab in roman military service, who retained his nomadic mentality by calling
the proit by his service booty.
415
actually, aougrio does not have the meaning
of booty, as it is interpreted by merkelbach and stauber, but just means rich of
proit. This inscription is therefore not at all for certain written by a former nomad
and does not support sartres view. however, a Thamudic inscription mentions
booty by a caravanier,
416
and in this context it is supposed to be a reference to pro-
it. in any case, when viewing the safaitic evidence together it is still very likely
that substantial booty was gained by raids.
finally, there is also evidence that Tankh groups raided other tribes. The irst
ights obviously happened when Tankh who had come to mesopotamia met with
groups which were already settled there.
417
abar occasionally mentions Tankh
raids, thus at imyar.
418
most important in this context, at least for abar himself,
is the Tankh king Gadhima, whom he presents as an effective raider of other tri-
bes, at least sometimes with the purpose of getting livestock.
419
however, there are also other supplementary economic strategies to be found
in the literary and especially the epigraphic evidence, which rather point to peace-
ful interaction. in between symbiosis and conlict is the extraction of tribute from
traders crossing nomad territory. strabo reports that on the route from syria to me-
sopotamia there were halting places with water brought from elsewhere. The skeni-
tai who dwelt there were moderate in their exaction of tribute, whereas the chief-
tains of the euphrates region were not.
420
according to Wolff the skenitai here
ought to be regarded as brigands,
421
but this must certainly be an exaggeration: as
sedentary communities could exact taxes, so could nomad groups.
certainly the most prominent example in the near east for collection of taxes
is the famous tax-law of palmyra from ad 137.
422
among other topics it refers to
grazing rights outside and inside the palmyrene territory and the import of wool,
and this has reasonably been interpreted as evidence for the presence of nomads
and their local trade with palmyra.
423
indeed, there are some safaitic inscriptions
which mention going to palmyra,
424
and some safaitic texts have been found in the
414 peek (1955), nr. 375; merkelbach and stauber (2002), nr. 22/39/01.
415 merkelbach and stauber (2002), p. 414.
416 ITham hU657.
417 abar, Tarkh (History) i 747.
418 abar, Tarkh (History) i 749.
419 abar, Tarkh (History) i 7502. see also his nephew and successor amr b. ad as effective
raider: abar, Tarkh (History) i 768.
420 strabo 16.1.27. see altheim and stiehl (196469) i, p. 2701.
421 Wolff (2003), p. 135.
422 CIS ii 3913; NSI 147; IGRR iii 1056 (Greek and latin sections only). on the tax-law in general
see e.g. bowersock (1989), p. 802; brodersen (1987); Kaizer (2002), p. 52; matthews (1984);
scharrer and Zangenberg (2005b), p. 112.
423 bowersock (1989), p. 71; p. 82; bowersock (1991), p. 42930; butcher (2003), p. 170; matt-
hews (1984), p. 173. see also dijkstra (1995), p. 85.
424 CIS V 1649; 16645.
285 The problem of nomadic allies in the roman near east
palmyrene.
425
Two inscriptions may be directly related to some sort of trading with
palmyra, since they mention the driving of pack-animals and camels to the city.
426
Towards the east, dura-europus and its surroundings seem to have been fre-
quented by groups of those who wrote the safaitic inscriptions.
427
Whether dura
can be regarded because of this evidence as a centre for nomadic groups remains
however the subject of debate.
428
at least it may be assumed that there was some
sort of peaceful interaction between the people at dura-europus and the nomadic
groups arriving there.
429
The same can be said about namra, which also was fre-
quented by nomadic groups. many safaitic inscriptions testify to the presence of
their writers at namra itself and its territory.
430
besides the above-mentioned tak-
ing of plunder, there were apparently also peaceful actions: some inscriptions men-
tion pasturing and slaughtering of camels in the region.
431
Quite striking is an in-
scription by a certain Ward which mentions the seeking of green fodder at the ex-
pense of the community.
432
Unfortunately it is not clear what exactly is meant by
this: is it the community of namra on whose expense the fodder is going to be
acquired? if so, does the inscription imply some sort of stealing? Whereas these
questions can so far not be answered properly, it is quite likely that pasturing in the
territory of namra was at least tolerated by its authorities, especially since there
was a roman fortress. in general some nomadic movements elsewhere appear to
have been rather peaceful, as some inscriptions mainly mention driving camels to
and from the hauran and to rm.
433
furthermore a Greek inscription from the syrian
village Zebireh, dated to ad 213, records that the Arisenoi and the Iachphirenoi
lived near the Habibenoi, the inhabitants of habiba.
434
probably the Arisenoi are to
be identiied with the Arsat, known from a safaitic inscription.
435
occasionally there are references to trading activities. pliny remarks that many
peoples of western arabia were engaged in trade.
436
With respect to nomadic trade
this can be supplemented by some epigraphic evidence. Thus, rock drawings occa-
sionally depict caravans.
437
a Thamudic inscription mentions a caravanier.
438
besi-
des the inscriptions mentioned above, in the context of nomadic peaceful interac-
425 ingholt and starcky (1951), nrs. 2.4; 8; 21.2; 34.3; 54b; 60; 63.2; 63.4; 8082b.
426 CIS V 663; PUAES iV c 718. see King (1990), p. 60; schlumberger (1951), p. 131.
427 e.g. CIS V 517580. see millar (1993), p. 445 and p. 450; macdonald (2005); moors (1992),
p. 283.
428 Thus stated by dirven (1999), p. 6.
429 see millar (1993), p. 450.
430 inscriptions from namra itself: CIS V 34483641. inscriptions from the territory: CIS V
322348. see also the references to coming from namra (CIS V 241; PUAES iV c 675) and
going there (PAAES iV 5,2; CIS V 523; 1656; 18945; 2803; 3143; 3878).
431 pasturing: PUAES iV c 406; 426. slaughtering: PUAES iV c 532. see also the driving of ca-
mels to the region: PUAES iV c 4067; 426.
432 PUAES iV c 407.
433 hauran: e.g. Wh 161; 238. Rm: PAAES iV 5,31.
434 IGRR iii 1132.
435 Zeinaddin (2000), inscr. 7.
436 plin. HN 6.162.
437 e.g. Zeinaddin (2000), drawings 6; 9.
438 ITham hU657. on Thamudic caravans see van den branden (1966), p. 43.
286 Ulf scharrer
tion with palmyra, there are other inscriptions referring to driving pack-animals.
439

Whereas in these cases it is not clear whether people traded with goods or with the
animals themselves, thus securing the supply of animals for the great trading cen-
tres and their entrepreneurs, other inscriptions mention goods in general or the tra-
ding with salt speciically.
440
besides economic interaction there was also some symbiosis in the sphere of
worship. While the classiication into nomad and sedentary cults is rather dubious,
441

there was nonetheless some religious interaction between the different spheres.
Thus, some safaitic inscriptions can be interpreted as referring to some sort of no-
madic pilgrimage to the nabataean sanctuary of baalshamin at s in the hauran.
442

The suggestion that there was a roman military presence at s in order to control
nomads and their interaction with sedentary peoples
443
may be sound, but so far
this hypothesis has not been veriied. other inscriptions testify to nomadic worship
of gods in the palmyrene and in palmyra itself.
444
finally, in one inscription allt
and baal are asked for safety, and the writer states to be on his way to palmyra.
445

since the worship of baal is seldom found in safaitic inscriptions, it can be assu-
med that the palmyrene bel is meant. in the present context it has to be pointed out
that pilgrimages and acts of worship such as the ones mentioned presuppose at least
some degree of mutual accommodation between nomad and sedentary parts of the
population in the region.
446
Thus, the overall picture of nomadic activities in the roman near east is di-
verse, and sometimes it even seems to be inconsistent. first of all, there is no liter-
ary or epigraphic evidence at all that there existed something like a nomadic me-
nace to the roman and persian empires which threatened their security and exi-
stence as a whole. The irst real crisis to the roman empire in the east caused by
nomads might have been the campaigns of the saracen queen mavia in ad 378,
and even in this case the extent of the threat she caused to the empire remains deba-
table.
447
it is not likely that the military installations in the region served the pur-
pose of parrying large-scale nomadic attacks on the empire itself. Whereas the main
439 CIS V 3661; SIJ 363.
440 Goods: SIJ 195. Trading with salt: Wh 24; 479. on trade of the writers of the safaitic inscrip-
tions see sartre (2005), p. 263
441 Kaizer (2002), p. 567.
442 macdonald (2003b). see dentzer (1985a), p. 4045; pollard (2000), p. 2056; on the sanc-
tuary at s see dentzer-feydy e.a. (2003); hammond (1973), p. 62.
443 pollard (2000), p. 97; 2056.
444 schlumberger (1951), inscr. nrs. 2.4; 21.2; 34.3; 54b; 60; 63.2; 63.4; 8182b. dirven (1999),
p. 81.
445 CIS V 1665.
446 see Villeneuve (1989), p. 135. on the central function of cities for the religious life also of the
country see scharrer and Zangenberg (2005b), p. 113.
447 on mavia see differing views e.g. of altheim and stiehl (196469) ii, p. 32832; ball (2000),
p. 98100; bowersock (1994c); Graf (1989a), p. 3489; isaac (1998b), p. 125; mayerson
(1980); millar (1993), p. 388; parker (1986a), p. 1456; sartre (1982a), p. 1404; schmitt
(2003); shahd (1984a), p. 14258, p. 1834 and p. 188201. on later nomadic incursions see
e.g. isaac (1998b), p. 1417.
287 The problem of nomadic allies in the roman near east
function of the stronger northern fortiications obviously was the defence of the
empire against persian campaigns, the installations in the south as well as the na-
bataean posts seem to stand in the context of monitoring, controlling and securing
nomadic movements.
448
of course it cannot be denied that there were nomadic raids against sedentary
populations, although sometimes a clear distinction between nomadic groups and
simple bandits cannot be drawn.
449
isaac even states that ancient nomads did not
have an aggressive ideology: both the victims and the raiding nomads, for opposing
reasons, exaggerated the spoil being taken.
450
This view seems to play down the
raiding activities of nomadic groups.
451
it has been stated above that booty and
plunder as well as peace or security were asked from the deities at the same time.
Whether this can reasonably be called an aspect of an ideology is doubtful. howe-
ver, it shows that raiding apparently was a somehow integral part of nomadic life.
nomadic raiding seems to have been on a rather lower scale there is no evidence
of large-scale raids, but only looting by a few persons is attested.
452
in this context it is even debated whether the military installations served the
purpose of protecting the sedentary population against nomadic attacks. sartre sta-
tes that nothing is known about what rome actually did in this respect.
453
he inter-
prets even the military post at namra as serving the purpose of keeping order
among the nomads rather than of protecting sedentary people.
454
This view dimi-
nishes the function of the military installations.
455
earlier in this section i have argued that there are no clear borderlines between
nomadic and sedentary zones of dwelling, and that the roman military installations
cannot be regarded as such.
456
indeed, besides those texts discussed above, there
are some other texts which attest movement across the borders: a safaitic inscrip-
tion mentions someone coming from rm, another someone leaving the empire, a
third someone spending the summer in rm.
457
one text even mentions the pasturing
of the border (fr) along the canals (qnwt).
458
The same can be said about the naba-
taean territory: a certain binarih states that he camped [returning] from the naba-
taeans or because of the nabataeans (ll mnb).
459
of course the differences of
448 see dabrowa (1986); Gichon (1991), p. 319. cf. shahd (1984b), p. 246. on nabataean forti-
ications see parker (1986a), p. 1168. on nabataean fortiications against neighbouring peop-
les (ad repellendos gentium vicinarum) see amm. marc. 14.8.13.
449 see butcher (2003), p. 40910.
450 isaac (1998f), p. 412.
451 see robinson (1996), p. 435.
452 on the distinction see Whittaker (1994), p. 138.
453 sartre (2005), p. 69.
454 sartre (1991), p. 333 (= sartre (1997), p. 3167).
455 see parker (1991), p. 5023
456 see Graf (2002), p. 1535; mayerson (1986), p. 36. on later nomads migrating between the
syrian coast and mesopotamia see dillemann (1962), p. 734.
457 coming from rm: SIJ 351. leaving the empire: CIS V 66. spending the summer in rm: Wh
1996a.
458 Wh 1199.
459 Wh 158.
288 Ulf scharrer
the readings are important with respect to the question of peaceful interaction.
460
in
this context, however, the inscription attests to nomadic movement to and from the
nabataean territory. This is further supported by the references to writers of the
safaitic inscriptions from within the nabataean realm.
461
apparently there were
seasonal migrations between the steppe and the agricultural regions,
462
which were
not hindered by military installations.
Quite striking is the fact that antagonism and symbiosis could occur in paral-
lel.
463
Thus, all sorts of interaction must have taken place on a very local level.
With respect to nomadic groups, whether speciic members of them raided and tra-
ded at the same time and with the same opponent or partner remains unknown due
to a lack of evidence. probably different segments of the nomadic societies, inde-
pendently from each other and at different times, followed varying economic stra-
tegies with which the sedentary populations and other nomad groups had to cope.
The realization of the very different ways of interaction between nomadic and se-
dentary populations, and among nomadic groups themselves, forms a basis for the
question of nomadic allies.
nomadic allies
in the preceding sections i have discussed the problem of nomadism itself and the
dificulties in drawing conclusions from the evidence with respect to roman rule in
the near east. subsequently, a rather complex picture of interactions emerged. in
this section i will attempt to apply the results of the preceding sections to the prob-
lem of nomadic allies themselves, which is the actual topic of this essay. first i
shall deal with the problem of nomadic political organization. as a contrast i shall
then proceed to discuss the political and social relations of nomad and sedentary
groups on the regional level. i will then discuss the evidence for nomads as allies of
the parthian and roman empires. This discussion shall be followed by a short sec-
tion on the changes in the third century ad, which eventually leads to a sketch of
the relation between the Tankh and the sasanid and roman powers up to
imraalqays.
nomadic political organization
in the previous sections i already discussed the concepts of dimorphic and segmen-
tary societies. Unfortunately we know hardly anything about nomadic social orga-
nization in the roman period. apparently there were some sort of leaders already
460 Unfortunately hardly anything is known about peaceful interaction between the writers of the
safaitic inscriptions and the nabataeans: see parker (2002), p. 78.
461 PUAES iV a 11; 434.
462 see Gichon (1991), p. 321.
463 against the simple reduction of roman period nomad-sedentary relations to antagonism, see
also sommer (2005), p. 2067.
289 The problem of nomadic allies in the roman near east
before lakhmid rule: according to strabo the arabian skenitai in the region bet-
ween the euphrates and mesopotamia were divided into small sovereignties
(ouvootrio).
464
a little earlier he mentions phylarchs on both sides of the euphra-
tes.
465
finally he writes about an alchaedamnus being king (ooiru) of the
rhambaeans, who were nomads dwelling west of the euphrates.The king can pro-
bably be identiied with the arabian dynast alchaudonius who is referred to by
cassius dio.
466
it is debatable whether the Greek terminology reveals anything
about the organization of these groups. The term dynasteia implies that there was a
leading family. With respect to the phylarch and the basileus this is not so clear, but
it cannot be excluded. finally it is questionable whether alchaedamnus as king had
a higher position than dynasts or phylarchs. There are other references to dynasts of
arabians in the classical texts. against the background of what has been said in the
third section, it cannot be known whether nomadic, sedentary or mixed people are
meant in each case. flavius Josephus, for example, refers to kings and dynasts.
467
if
alchaudonius and alchaedamnus are indeed identical, not too much can be drawn
from the terminology. finally abar mentions arab princes of tribes in babylonia
before the arrival of the Tankh.
468
some safaitic inscriptions might be interpreted as referring to leading men:
thus a raid of horseriders was done under a leader (mr),
469
but it is not clear whe-
ther this was just a temporary leader for the raid itself or a leading man of an ethnic
group. another inscription states that a certain Yaslam was a powerful chief
(hm).
470
other inscriptions mention an overseer (nr): a certain ausn asks Yia
for help by driving away mnJm the overseer.
471
in another inscription only an
overseer is mentioned.
472
again, it is not clear whether a nr was a leading man of
an ethnic group or an ofice installed by regional authorities with the purpose of
controlling nomadic groups. The irst of the inscriptions cited here could be viewed
as supporting the latter interpretation, but so far no conclusions can be drawn. Thus
there are at least three terms for leadership in the safaitic inscriptions, apart from
the word king (mlk), for which references have been cited above in the context of
nabataean and roman hostile relations to nomadic groups.
473
other safaitic in-
scriptions probably point to leading men without using one of these terms: two in-
scriptions mention a year in which Ym warred with the l of Thamd.
474
fur-
thermore, a certain adddat boasts to have expelled the awlat and to have held
464 strabo 16.3.1.
465 strabo 16.1.27.
466 strabo 16.2.10; cass. dio 36.2.5. The identity is implied by rets (2003), p. 348, and sartre
(2005), p. 36.
467 Joseph. BJ 2.67.
468 abar, Tarkh (History) i 747.
469 CIS V 4276.
470 Wh 2006.
471 Wh 2163.
472 Wh 2163.
473 see n. 3334.
474 Wh 3792a; 3792c.
290 Ulf scharrer
them in check.
475
another inscription refers to a year in which rayatum brought
back aif into power.
476
finally, in the fourth section i have refered to single
people revolting against rm or the nabataeans. a single person could certainly not
on his own wage war against a whole ethnic group, or control one, or bring a whole
tribe back into power. The persons mentioned in those inscriptions are therefore li-
kely to be leading men. however, the level of their leadership cannot at all be ixed:
were the leading men from the safaitic inscriptions leaders of a whole tribe, of a
segement, or even of a confederation of tribes about which we know nothing from
the safaitic inscriptions? against the background of the evidence, so far there is no
answer to these questions.
in the third section i have discussed some of the problems of the interpretation
of the rawwfa-inscription. The main problem is what exactly the term to tev Oo-
ouogvev r0vo / rkt tmwdw implies, be it the Thamd in the sense of an ethnic
group or of a military unit of the roman army. against this background the impli-
cations of the nabataean term qdmy (b4), mostly translated as leaders, erecting
the sanctuary appear to be controversial. presuming that the Thamd of the inscrip-
tion are a confederation, it has been stated that there was no king over the group,
but some sort of council of elders and leaders of subtribes.
477
according to some
views the inscription thus reveals the rise of tribal confederations and their leaders
to increased power.
478
if interpreted in this way, the inscription from rawwfa
could be regarded as revealing a glimpse of a development from rather small to
large nomadic ethnic units, at the end of which emerged the rule of the Tankh.
however, if the temple and the inscription were set up by a military unit instead, no
such conclusions could be derived from the evidence. in that case the qdmy were to
be interpreted as military commanders.
something more is known about the Tankh. it seems that at irst there were a
couple of chieftains, for example al-qr, one of the irst to move with his group
into mesopotamia.
479
according to abar, the Tankh Gadhima of the Banu Wa-
bar was the irst to rule there. abar relates that other tribes joined him, that dele-
gations were sent to him and that tribute was paid to him.
480
by getting a son amr
by a woman of the lakhmid he became the founder of the lakhmid dynasty.
481
it is
debatable to what extent the lakhmid actually were a tribe rather than a clan. The
title ooiru / mlk is also attested for him in the bilingual epitaph of his tutor fihr
from Umm al-Jeml.
482
When i discussed the migrations of the Tankh in the third
475 Wh 1231.
476 CIS V 3661.
477 see Graf (1978), p. 16. on this concept in general see Giddens (1986), p. 53.
478 bowersock (1994c), p. 134; isaac (1992), p. 239; isaac (1998f), p. 417; Kuhnen (1991), p. 329.
479 abar, Tarkh (History) i 747.
480 abar, Tarkh (History) i 750.
481 see abar, Tarkh (History) i 755. on the dynasty and problems of the historic tradition see
rothstein (1899), p. 3860; ball (2000), p. 97. on genealogical problems of the irst lakhmids
see hartmann (2001), p. 341. on Gadhima see equini schneider (1993), p. 46.
482 nabataean section: PUAES iV a 41; CIS ii 192; Greek section: PUAES iii a 238
1
; SEG XXiX
1604. but see millar (1993), p. 4334, stating that the Gadhima of the inscription and of abar
need not be the same.
291 The problem of nomadic allies in the roman near east
section, i mentioned their orientation westwards. it is debatable whether the inscip-
tion for fihr also attests to the extension of lakhmid power to the west.
483
at least
Gadhima expanded into the Jazirah, and for this purpose he had to have the ruler of
this territory killed.
484
abar continues that after Gadhima the Tankh were split
into two groups, one under amr b. abd al-Jinn, the other under amr b. ad, the
son of his sister, but that they eventually managed to come together.
485
Unfortuna-
tely abar does not tell us how this happened.
apparently amr b. ad became the next lakhmid to rule the Tankh. of him
abar relates that he was the irst lakhmid ruler to settle at al-rah. more impor-
tantly, he did not recognise local princes of the iraq. finally, according to abar he
ruled the lands from iraq to the hejaz.
486
Thus, the title king of (all) Arab of the
epitaph of imraalqays, son and successor of amr b. ad, gets more sense. in the
third section i discussed the term saying, with Zwettler, that probably not all arabs
were meant, but certainly a wide stretch of the northern arabian steppe.
487
besides
the problem of the reading of the irst line, as discussed above, there are other un-
certainties due to the erosion of the epitaph. The discussion about the reading of
line 2 and the irst part of line 3 in the present context is not that important, as the
sense seems to be basically the same: imraalqays subdued tribes and their kings.
488

however, the reading and the sense of the second part of line 3 and of line 4 is quite
controversial.
489
The most important alternatives are that either imraalqays distri-
buted the tribes among his sons (nzl bnyh lwb),
490
or that he dealt gently with
the nobles of the tribes, and appointed them viceroys (nbl bnbh lw wwklhm).
491

Thus, the way the king organised his power over the subdued tribes is a severe mat-
ter of debate, and both alternatives seem to be entirely possible, be it the exercise of
indirect power with leaving the leading men in their positions, or the direct integra-
tion of the tribes into the Tankh confederation under the leadership of the lakhmid.
if one does not take the term sons too literally, the irst view could be supported
by a remark of procopius, according to which a certain Kaisios was chief of the
Maddenoi, one of the subdued tribes mentioned in the namra inscription. This
Kaisios probably is to be identiied with Qais b. salama, a cousin of imraalqays.
492

however, it is obvious that imraalqays exercised his power over a wide territory
and its inhabitants. according to abar imraalqays ruled over the frontier region
483 millar (1993), p. 431.
484 abar, Tarkh (History) i 756. see hartmann (2001), p. 335.
485 abar, Tarkh (History) i 761; 768.
486 abar, Tarkh (History) i 7689.
487 on the inscription and its editions see above, n. 172 and n. 2347.
488 on this see bellamy (1985), p. 3740. on the tribes being subdued see shahd (1984a), p. 38
43. dussaud (1902), p. 415 and p. 417. on the campaigns see also altheim and stiehl (1964
69) ii, p. 3223.
489 see bellamy (1985), p. 35. on the discussion see bellamy (1985), p. 405; rets (2003),
p. 46970.
490 dussaud (1902), p. 412; p. 4178.
491 bellamy (1985), p. 405, with the translation on p. 46, followed by isaac (1992), p. 239 n.118.
on other readings see rets (2003), p. 46970.
492 proc. bell. 1.20.9 with Grunebaum (1963), p. 20. on the Maadd see shahd (1984a), p. 43.
292 Ulf scharrer
of the arabs of rabah, muar and the rest of the tribes of the deserts of iraq, he-
jaz and Jazirah.
493
if the reading of the end of the opening phrase (discussed in the
third section) really is lrb klh (all Arab) this is, at least from imraalqays per-
spective, hardly an exaggeration.
494
eventually he was succeeded in this regency
by his son, amr b. imraalqays.
495
With respect to nomadic power we then face a development from smaller eth-
nic units to a really large political and probably even ethnic unit, beginning at least
with the expansion of Tankh power. This power was enlarged by the process of
absorbing existing tribes (see the third section) by confederacies, and by con-
quest.
496
certainly many tribes kept their ethnic identity, but this was broadened by
the integration into the large confederation of the Tankh.
497
nomads and sedentaries: political interactions before ca ad 300
in this context and seemingly in contrast to the aspect of expansion of nomad po-
wer, i will now analyse nomad-sedentary political and social interaction in connec-
tion with the question of roman allies. in the second section i very briely sketched
the mondern concepts of segmentary and dimorphic societies, which are marked by
a close economic and socio-political interaction of nomad and sedentary groups.
Whereas in the third section i discussed mainly the economic interaction between
nomads and sedentary peoples, with the topic of socio-political interdependency i
now turn to a ield crucial for the understanding of the problem of nomadic allies in
the roman near east.
first of all, there was apparently no attempt by the roman and parthian au-
thorities to settle nomadic groups, a proceeding exercised by many empires and
states up to the present day.
498
There are only two references to the settling of no-
mads, but this was not attempted by either empire. The irst reference is from the
beginning of the chronological scope of this paper: plutarch writes that the arme-
nian king Tigranes, actually a parthian client king, removed the scenitic arabs to a
nearby settlement for the purpose of trade.
499
Unfortunately, neither can the loca-
493 abar, Tarkh (History) i 834. see bowersock (1983), p. 1412; hartmann (2001), p. 348;
sartre (1982a), p. 138.
494 see dussaud (1955), p. 65; isaac (1992), p. 240; sartre (2005), p. 362; shahd (1984a), p. 35
6; Taeschner (1964), p. 312.
495 abar, Tarkh (History) i 845. see on the rule over nomadic confederations Giddens (1986),
p. 54.
496 on alliances see abar, Tarkh (History) i 748.
497 see dussaud (1955), p. 69.
498 sommer (2005), p. 52 on a roman control and encouragement of the sedentarization is too
speculative. against such views with respect to the Decapolis see e.g. Wenning (2005), p. 145.
see however the poetical account of some sort of sedentarization exercised by roman authori-
ties on the Dahae living north of the caspian sea luc. 7.429: Nec vetitos errare Dahas in
moenia ducat. on examples of modern programs of settling nomads see e.g. Ganzer (2002);
martin (2002); offer (2003); scholz (1999).
499 plut. Vit. Luc. 21.4. see rets (2003), p. 3489. on Tigranes see sartre (2005), p. 2730; KlP,
col. 8278 s.v. Tigranes 1 (duchesne-Guillemin).
293 The problem of nomadic allies in the roman near east
tion of this settlement be traced, nor is it certain that this settlement was a new
foundation. by this move, Tigranes obviously intended to get control over trade.
The other reference is rather from the chronological end of this essay. in the fourth
section i mentioned that shapur ii campaigned extensively in the arabian penin-
sula against nomadic groups raiding sasanid territories. in this context abar re-
cords that the persian king had many of them deported and settled at different
places in his empire.
500
apparently, this coincides with practices of empires from
the assyrian and achaemenid periods, and this practice is not speciically dealing
with nomadic groups.
501
by his deportations shapur obviously tried to gain control
over hostile elements. against the background of the absence of a nomadic threat to
the eastern roman empire it is actually not astonishing that the roman authorities
did not, as far as we know, take any steps to sedentarise nomadic groups.
however, there is some controversion on the question to what extent nomadic,
especially safaitic, groups became sedentary in villages or cities. in the third sec-
tion i have discussed the problem of the Ubait. Whereas the evidence on a sed-
entarization of this ethnic group might be regarded as ambiguous, there is some
debate about whether other writers of the safaitic inscriptions became settled per-
manently in cities and towns. The view that they at least partly did so is founded on
two bases.
502
firstly, a few safaitic and Greek inscriptions naming ethnic groups
have been found in sedentary contexts.
503
secondly, a couple of personal names
found in village or city contexts such as bosra are regarded as safaitic names, and
thus as pointing to a nomadic origin of their bearers who by then had become sed-
entary.
504
on the other hand it has been stressed, in my view reasonably, that none
of this evidence points to a considerable wave of sedentarization. first, the pres-
ence of nomadic groups at a certain place does not prove and not disprove either
them settling down,
505
but it points to some sort of interaction, as the examples
from namra and the palmyrene which were discussed in the fourth section reveal.
furthermore, as has been currently stated, the attribution of personal names to cer-
tain ethnic groups is a rather ambiguous matter.
506
in any case, as i have argued, it
cannot be disproved either that nomadic individuals or groups settled down, but the
evidence on this is really sparse: only one safaitic inscription explicitly mentions
500 abar, Tarkh (History) i 839.
501 see e.g. briant (2001), p. 168; scharrer (2006), p. 34950. on deportations in the ancient
world see generally Kehne (2006). see also plut. Vit. Luc. 26.1 and plin. HN 6.142 on the po-
licy of deportation of Tigranes of armenia. on the deportation of the inhabitants of mesopota-
mian singara to persia under shapur ii see amm. marc. 20.7.7.
502 This view has been suggested e.g. by beyer (1998), p. 198 n. 89; littmann (1940), p. 234;
moors (1992), p. 284304; sartre (1982b); sartre (1989), p. 1589; sartre (1991), p. 333 (=
sartre (1997), p. 317); Villeneuve (1989), p. 1345. With respect to bosra: de Vries (1986),
p. 234; sartre (1982b).
503 sartre [1990], p. 41 and p. 456. cf. Graf (2002), p. 154.
504 sartre [1990], p. 467. see e.g. PUAES iii 4; 56; 244; 741; 744; 792; 800
2
; 800
7
. bosra: e.g.
PUAES iii 554
1
6; 567; 5802; 5936; 599.
505 see sartre [1990], p. 50.
506 see n. 358. important with respect to the current problem: macdonald (1999), p. 2568; 2627.
294 Ulf scharrer
the settling (qrr) at BRKT.
507
somewhat dificult to interpret is a safaitic inscrip-
tion from the palmyrene set up by a certain msik from the people of namrat
(l nmrt).
508
probably here l is being used to describe a civil community, maybe a
xoivov,
509
in this case the inhabitants of namra and its territory. it is not certain
whether msik lived at least periodically a settled life at namra or whether he was
a nomad being socio-politically bound to the town.
one of the problems in this context is that of ethnic designations. There are
basically three formulations of expressing ethnic identity. firstly, there are tribes
(ug ) mentioned in village contexts:
510
an inscription from aerita dating from the
time of commodus mentions a tribe of the Osainenoi
511
and another one mentions
the tribe of the Dabanenoi.
512
a further inscription from dr il-leben is set up by a
certain aurelius Glakos of the village rma of the tribe of the Chasetenoi.
513

Whereas not much is known about these groups, the tribe of Chauchab that is
mentioned in a fragmentary inscription from rma is identical with the Kawkab
known from safaitic inscriptions.
514
Therefore there seem to have been socio-poli-
tical relations between village and steppe dwellers at least occasionally. secondly,
the identity probably of a clan or a family could be expressed by the name of its
head or ancestor. Thus an inscription mentions that a house was renewed by a cer-
tain bachros of those of bachros (tev Boou).
515
Thirdly, local identities could
be termed just with the sufix -os, and thus in many cases it cannot be known con-
clusively whether a place- or an ethnic name is implied.
516
To mention only a few
examples, there are mentions of a Nouaitenos, or the Maneinenoi.
517
sometimes it
is clear that a sedentary community is implied, e.g. in the case of the koinon of the
Seeienoi or the Sauroi.
518
interesting in this context are two inscriptions from Zorava, a metrokomia.
519

The irst one mentions the Saamenoi of/among the Zoravenoi.
520
it remains a que-
stion whether the Saamenoi are identical with the am known from safaitic in-
scriptions.
521
if so, this inscription would testify that at least a part of this ethnic
group was integrated into the village society of Zorava. if not, this inscription
507 Wh 2411.
508 ingholt and starcky (1951), nr. 81b.
509 on the term see e.g. Grushevoi (1985), p. 534.
510 references are collected by moors (1992), p. 29399 and p. 3416. see also sartre (1999),
p. 201; Villeneuve (1989), p. 137.
511 IGRR iii 1180.
512 PUAES iii a 786
6
.
513 PUAES iii 765
11
: xeg Pgro ug Xoogtrvev.
514 IGRR iii 1269: ug Xouo. on safaitic references to the Kawkab see n.118. on safaitic
groups in village contexts see e.g. sartre [1990], p. 40.
515 PAAES iii 388. see also PUAES iii 756.
516 references are found in moors (1992), p. 3047.
517 Nouaitenos: cis V 2824. Maneinenoi: IGRR iii 1172.
518 Seeienoi: e.g. PAAES iii 428b. Sauroi: e.g. IGRR iii 1143.
519 on Zorava being a metrokomia: IGRR iii 1155 (= PAAES iii 432g); sartre (1999), p. 201. on
metrokomiai in southern syria see sartre (1999); sartre (2005), p. 209.
520 IGRR iii 1156.
521 for references on the am see n. 118.
295 The problem of nomadic allies in the roman near east
shows at least that different sub-identies existed within the community of the vil-
lage. in this context the second inscription from Zorava is interesting. it states, that
the farmers (yroyoi) of or among the Zoravenoi set up a nike for m. aurelius and
caracalla.
522
Whereas this inscription is an example of some sort of emperor wor-
ship in rural societies, it is more important to note that only the farmers of the vil-
lage set up the statue. Therefore there must have been other groups, besides crafts-
men also herdsmen. as stated in the fourth section, herdsmen are also known from
rather sedentary contexts. if, however, the Saamenoi are the safaitic am, it might
be that they did the shepherding on behalf of the village community.
523
of course
all this is quite speculative and can thus only be taken as a hypothesis.
With respect to village societies not much can therefore be stated clearly. in the
second section i have argued that tribalism in general is not restricted to nomad
societies, and this has also been stated reasonably with respect to near eastern vil-
lage communities in the roman period.
524
This actually makes it quite dificult to
interpret the evidence with respect to socio-political ties between sedentary rural
communities and nomadic groups. apparently there are different interpretations,
which do not exclude each other from case to case. firstly, connections between
nomadic and rural sedentary societies have been stressed.
525
secondly, according to
moors it may have been possible that nomadic tribes lived sedentary lifes in vil-
lages or that some parts of ethnic groups were sedentary and others nomadic, thus
being examples of segementary social organizations.
526
moors, with respect to the
Aw, furthermore states that in springtime they pursued tillage and in the other
seasons they migrated with their animals, thus being a typical example of semi-
nomadic life.
527
Unfortunately the latter view cannot be veriied from the epigraphic
evidence. The third view is quite sceptical about the question of nomadic groups
becoming sedentary in villages. especially macdonald argues that there is no evi-
dence on writers of the safaitic inscriptions having become sedentary.
528
sartre is a
bit more speciic on that matter: the inscriptions in question do not say whether
their writers have become sedentary in villages. furthermore it cannot be traced
from the evidence to what extent their writers actually were integrated into the vil-
lages social structures.
529
only some light may be shed on this problem by an in-
scription mentioning a syndikos nomadon, an advocate of the nomads, of the
522 IGRR iii 1154.
523 on differerent occupations of tribes in roman time near eastern villages see macadam
(1983), p. 1112. on nomads and villagers being one community divided into agriculturists
and shepherds see butcher (2003), p. 170.
524 e.g. isaac (1998b), p. 155; macdonald (1993), p. 3523; sartre [1990], p. 47. on ancient near
eastern villages in general see e.g. beyer (1998); sartre (1999); sartre (2005), p. 22433; Vil-
leneuve (1985), p. 75113; p. 1216; scharrer and Zangenberg (2005a) with further references.
on villages in the palmyrene see schlumberger (1951).
525 dirven (1999), p. 21. see also staubli (1991), p. 118.
526 moors (1992), p. 3367. The second view has been suggested also by sartre (1991), p. 3334.
see also Villeneuve (1985), p. 1167.
527 moors (1992), p. 283. see also above, section 2.
528 macdonald (1993), p. 313.
529 sartre [1990], p. 478 and p. 52.
296 Ulf scharrer
name Theodoros.
530
according to moors, this was an oficial whose function it was
to be in contact with nomads.
531
apparently the inscription stems from a village
rather than from a military context, but inal statements about this cannot be made.
especially at villages many oficials who were syndikoi were found.
532
in the pre-
sent context noticeable is also an inscription from a village in the hauran which
remembers a certain iulius malchus, a syndikos of the Aouorenoi.
533
Unfortunately
this ethnic group is not known from the safaitic inscriptions, and furthermore it is
not designed as phyle. obviously this is an indigenous group, but its character is
not clear. so far moors view appears to be reasonable. Thus, again there seems to
have been some socio-political interaction between steppe and village dwellers. it
cannot be said, however, to what extent the nomads of the inscription were really
integrated into the village society, i.e. whether they were an integral part of it.
against this background of evidence and discussion, at this point it is not reaso-
nable to speculate about the possibilies of dimorphic or segementary structures in a
rural context. of course, in some instances this might have been the case, but an
overall unequivocal picture cannot be traced. at least it seems quite likely that there
was some social interaction, whatever this may mean exactly for each particular
case.
mostly villages were socially and politically dependent on towns or cities.
534

Thus, if there existed social interaction of whatever kind between village dwellers
and nomadic groups it can be asked whether this also was extended within the con-
text of relations between villages and cities. Unfortunately, so far no answer can be
given to that problem. however, quite often it has been stated in scholarship that
also in the cities of the decapolis, namely at bosra, at least parts of nomadic tribes
became sedentary.
535
again this view is based on the existence of names being at-
tributed to the writers of the safaitic inscriptions,
536
which, as stated already twice,
is rather problematic. furthermore mentions of safaitic ethnic groups and other
tribes contributed to that picture. macdonalds and sartres above-quoted views
with respect to nomadic dwelling at villages are to be applied to the problem of
towns and cities: the references to safaitic tribes do not necessarily imply that they
had become sedentary or formed an integral part of the citys social structure, alt-
hough this is still possible.
furthermore, it has been stated repeatedly that also cities were structured tri-
bally. one of the main problems especially of near eastern cities is the question of
whether the relevant tribes preserved the tribal structures of pre-urban societies or
whether they ought to be regarded as artiicial. Whereas this problem will rise again
530 PAAES iii 383 (= sartre (1982a), p. 126 nr.8; moors (1992), p. 315 nr. 4b): Orooeou ouvoi
xou voooev.
531 moors (1992), p. 3312.
532 moors (1992), p. 396401 and p. 4179.
533 PAAES iii 401a (= moors (1992), p. 315 nr. 4a).
534 see scharrer and Zangenberg (2005b) with further references. on the problem of deining a
city see scharrer (2002b), p. 2835.
535 e.g. contini (1987), p. 49; de Vries (1986), p. 234; moors (1992), p. 284. see n. 504.
536 sartre (1985), p. 141 and p. 1478.
297 The problem of nomadic allies in the roman near east
below with respect to palmyra, it is also a matter of debate with respect to bosra
and other cities of the decapolis.
537
among the tribes known from city contexts are
the Somaithenai and the Bitaieni, both from the polis soada, the name of the latter
maybe having the semitic bit as a compound. With respect to these tribes it is noti-
ceable that for both of them the ofice of an episcopus, an overseer, existed at
soada.
538
one of these was cornelius palma, who under Trajan created the provin-
cia Arabia, and who is also designated in the relevant inscription as antistrate-
gos.
539
Unfortunately the exact relations between palma and the Bitaieni or So-
maithenai is not clear. furthermore it cannot be said whether these tribes were civic
tribes or, if not, whether they were nomadic groups at all. it therefore has to remain
a matter of debate to what extent the ofice of an episcopus might resemble that of
the overseer (nr) known from safaitic inscriptions.
540
it is possible that at least some of the inscriptions ought to be seen rather in the
context of economic interaction, as discussed in the fouth section. Thus, against the
background of the evidence so far no statement can be made about the extent to
which the nomadic groups were socially and politically bound to these cities, and
hence it is not clear whether it is reasonable to speak of dimorphic or at least seg-
mentary societies.
i have already argued that, apart from very few examples quoted, there is no
evidence of attempts to settle nomadic groups in cities or villages. based on the
interpretation of tribes and names at bosra, de Vries suggests that in the irst cen-
tury ad local tribes settled at bosra under political control and with encourage-
ment of the nabataean authorities.
541
so far, this hypothesis cannot be supported
by any unambiguous evidence.
The relation between the nabataeans and other ethnic groups, especially no-
mads, is a question which has actually not been pursued in detail so far. much about
the nabataeans themselves, besides much evidence on their culture, is still unclear,
especially the question to what extent the nabataeans themselves are to be regarded
as nomadic or at least of nomadic origin.
542
in any case, it has occasionally been
stated that some of the groups known from the safaitic inscriptions and some wri-
ters of Thamudic texts lived under the political inluence or control of the naba-
taean kingdom.
543
Graf bases his notion mainly on three observations: irstly, the
nabataean god -ar is being worshipped by writers of Thamudic and safaitic
inscriptions steming from the nabataean realm; secondly, some of the writers ad-
opted names of nabataean kings and queens; and thirdly, the idea is based on lin-
537 Tribes artiicial at bosra: e.g. macadam (1983), p. 111. Tribes preserving indigenous structu-
res: e.g. bowersock (1991), p. 427; sartre (1982b), p. 85. on urban tribes of the decapolis with
references: moors (1992), p. 28793 and p. 33841. on the distinction of different sorts of
tribes see moors (1992), p. 2856; sartre (1982b), p. 878.
538 Somaithenai: IGRR iii 1276. Bitaieni: IGRR iii 1277. either of these tribes: IGRR iii 1273.
The Somaithenai are also mentioned in IGRR iii 1213.
539 IGRR iii 1273.
540 Wh 2163. see above, n. 4712.
541 de Vries (1986), p. 234.
542 see the references in n. 42.
543 e.g. Graf (1989a), p. 36870; de Vries (1986), p. 2345.
298 Ulf scharrer
guistic grounds, mainly the orthography of safaitic texts.
544
Whereas this evidence
certainly reveals some sort interaction between the writers of safaitic and Thamu-
dic inscriptions and nabataean groups, it cannot not be interpreted conclusively
with respect to political and social interaction between nabataeans and nomadic
tribes in the region. is it really appropriate to regard the worship of -ar as an
example of political loyalty to the nabataeans? or does the use of nabataean phra-
ses and orthography really reveal more than some sort of cultural inluence? even
the very few safaitic-nabataean bilingual inscriptions do not necessarily have to be
interpreted as evidence for the socio-political afiliation of their writers.
545
maybe
the use of nabataean kings and queens names could be regarded as evidence of
loyalty, but still it is not really clear whether the use of these personal names might
have followed just a fashion and whether they at all had the nabataean names as a
model.
This evidence certainly reveals some interaction between the nabataeans and
other ethnic groups. in the third section i briely discussed the palmyrene inscrip-
tion of obaid/abidu who identiied himself as a nabataean and a ruhaean/
Raw.
546
as said above, dijkstra reads rawwaha, thus referring to a group
known from safaitic inscriptions. his reading is supported by some nabataean in-
scriptions mentioning the safaitic tribes of Raw, Salam

and Qaiu.
547
all these
inscriptions come from the nabataean realm itself, the inscriptions mentioning the
Raw and the Salam from Umm al-Jeml. interesting in this context is the hypoth-
esis put foward by de Vries, that, against the background of the inscriptions and the
fact that in the early roman stratum of the town there is a no distinctive nabataean
pottery to be found, Umm al-Jeml mainly was not dwelled by nabataeans them-
selves, but by nabataeanised arabs, thus implying some sort of sedentarization
of the tribes just mentioned. from this he draws the conclusion that there was some
kind of alliance between these groups and the nabataean authorities.
548
at least
with respect to the Raw there seems to have been some, as obaid/abidu felt
both nabataean and probably Raw. Thus certainly at least some nomadic groups
formed an integral part of the nabataean empire.
549
however, in the fourth section i have discussed nomadic raids against the na-
bataean empire. Therefore, while some nomadic groups might have lived in peace-
ful coexistence with the nabataean empire, others apparently had rather hostile re-
lations. in that section i also discussed the revolt of damas. if this was a revolt
against the nabataeans, in this case nabataean rule was felt at least by some writers
544 Graf (1989a), p. 36870.
545 safaitic-nabataean bilinguals: e.g. Zeinaddin (2000), inscr.13. see macdonald (1993), p. 348
51.
546 see above, n. 1389.
547 Raw: PUAES iV a 43. Salam: PUAES iV a 44. Qaiu: PUAES iV a 11. on further refe-
rences to these groups see n. 118.
548 de Vries (1986), p. 2345.
549 see dijkstra (1995), p. 41, sartre (2005), p. 2378 and macdonald (1991), p. 1068 on the fact
of different ethnic groups within the nabataean kingdom. on a tribal structure of the naba-
taeans see e.g. Goodman (1997), p. 260.
299 The problem of nomadic allies in the roman near east
of safaitic inscriptions to be something foreign in any case they did not regard
themselves as nabataeans in the way that obaid/abidu did. Thus, probably some
segments of the nabataean kingdom can be regarded as having a dimorphic struc-
ture. in addition, some sort of peaceful interaction must have existed also on the
socio-political level besides probable antagonistic tendencies of some groups to
which i have already referred.
somehow dubious with respect to the question of dimorphic societies is the
emesene society. according to strabo, its irst rulers in the irst century bc, samp-
sigeramus and his son iamblichus, were both phylarchs of the ethnos of the Emese-
noi.
550
according to cicero, iamblichus simply was a phylarch of the arabs (phyl-
archus Arabum).
551
first, the term ethnos with respect to the Emesenoi is striking,
since it is questionable whether it refers to the people of emesa or to a native ethnic
group. emesa itself was a seleucid foundation. With respect to this it has to be
stressed that seleucid cities seemed to have been settled to a considerable degree
with indigenous people already in the hellenistic period.
552
it can therefore not be
stated conclusively whether the dynasts of emesa emerged out of the population of
emesa in the course of the inal seleucid decline in the irst century bc, or whether
by some unknown process an ethnic group of a similar name like emesa the
Greek Emesenoi occupied the city coming from outside.
553
against the background of the wide use of the term ethnos which was dis-
cussed in the third section it is possible that the population of emesa at this time
was a rather homogeneous civic group, though without the claim to common de-
scent. in any case, the main seat of the dynasty at this time was arethusa to the
north of emesa.
554
Thus, the phylarchs of the Emesenoi controlled some territory.
in this context ciceros phrase should not be overinterpreted: as native arabians
and ruling over cities probably settled at least by some arabs, from a roman point
of view it seems quite appropriate to stress their arab identity. again it has to be
emphasised that no speciic way of life of sampsigeramus, iamblichus and their
subjects can be concluded from the phrase.
555
it has occasionally been stated that the emesene dynasts indirectly controlled
desert tribes,
556
but there seems to be no direct evidence for this. Unfortunately not
much is known about emesa and its relations to the surrounding territories in the
hellenistic period, and we do not know anything either about nomadic groups li-
ving in the vicinity of the emesene territory, although there might have been some.
Therefore, not too many conclusions from the rather scarce evidence ought to be
drawn.
550 strabo 16.2.10. on emesa see also the paper by Kropp in this volume.
551 cic. Fam. 15.1.2.
552 see scharrer (2006), p. 342, p. 3468 and p. 362, with references; sommer (2005), p. 228
n.15.
553 on the problem see sartre (2005), p. 334.
554 ball (2000), p. 34. see strabo 16.2.10.
555 according to rets (2003), p. 3545, they were nomads.
556 e.g. ball (2000), p. 34. sommer (2005), p. 59, counts emesa among those centres with a di-
morphic structure.
300 Ulf scharrer
maybe the process at emesa was somehow similar to the rise of the lords and
kings of edessa.
557
like emesa, edessa was a seleucid foundation, possibly also
largely populated with indigenous groups. furthermore, the process which in the
second century bc eventually led to the establishment of abgarid rule cannot be
traced in detail any more.
558
in this context there are mainly two approaches to ex-
clear the genesis of edessene identity and rule.
559
according to the irst approach,
nomad groups from outside settled in the city and established their rule.
560
in an-
other approach sommer suggests to apply rowtons model of a dimorphic society
to edessa: since the macedonian and indigenous settlers at edessa had to live from
shepherding or trade they had to make arrangements with surrounding nomadic
groups, and by ethnogenesis nomads and city dwellers became a group of common
identity with edessa as the administrative centre.
561
so far none of these approaches
can be proved by any evidence, and we should now have a further look at the evi-
dence from the roman period.
like the rulers of emesa, the lords of edessa were originally called by Greek
authors phylarch, which likely corresponds to syriac mry (lord) known from
syriac sources.
562
The Greek term in the irst instance reveals that at the beginning
there was some sort of tribal organization at edessa. furthermore, festus calls the
early edessene ruler phylarchus Saracenorum.
563
although this reference is
anachronistic,
564
it is a matter of debate whether edessene rule at least at this time
implied nomadic groups. however, it is not clear whether by some process of eth-
nogenesis all edessenes and maybe even surrounding groups ought to be regarded
as one tribe. furthermore the term, as well as the syriac word, does not imply the
extent to which groups outside the city fell under the dominion of the ruler of
edessa.
33 bc is the terminus post quem the rulers of edessa became kings, probably
in the context of alliance with the parthian empire.
565
The irst edessene king to be
named in roman literature is abgar V Ukkm, who is presented by Tacitus in ad
49 as rex Arabum.
566
While this designation certainly implies the rule over a larger
557 see sommer (2005), p. 22835. on the beginning of edessene kingship see luther (1999b).
see also the paper by sommer in this volume.
558 see sartre (2005), p. 6; sommer (2005), p. 22830. on different families see luther (1999b),
p. 4478.
559 on these approaches see sommer (2005), p. 230 n. 23.
560 drijvers (1980), p. 10; p. 17; sartre (2005), p. 33; segal (1970), p. 16.
561 sommer (2005), p. 2301.
562 see luther (1999b), p. 44852, followed by sommer (2005), p. 232 n. 32; sommer (2003a),
p. 392.
563 festus, Breviarium 14.
564 bowersock (1994e), p. 389.
565 luther (1999b), p. 4512, followed by sommer (2003b), p. 392; sommer (2005), p. 232.
566 Tac. Ann. 12.12.2. see luther (1999b), p. 173 and p. 451. luther also quotes Tac. Ann. 12.14,
but the relevant passage (12.14.1) does not name abgar explicitly as king, but just says that
izates left with the adiabenian and soon abgar with the arabian army (Izates Adiabeno, mox
Acbarus Arabum cum exercitu abscedunt). on the context of the passage see sommer (2005),
p. 234.
301 The problem of nomadic allies in the roman near east
territory beyond edessa itself,
567
again sedentary and nomadic arabs may be im-
plied, as it certainly is by the frequent designation of the nabataean king ruling
over sedentary and probably nomadic subjects as king of the arabs in Greek and
latin literature.
568
Unfortunately there is no equivalent to this designation either in
nabataean or in osrhoenean inscriptions, though it is found in the hatrean inscrip-
tions discussed below.
in this context the small corpus of inscriptions from sumatar, some 70 km sou-
theast of edessa and dated to ad 1656, is of fundamental importance.
569
a couple
of these inscriptions mentions a ly drb.
570
While in the third section i have briely
discussed these inscripions with respect to the term Arab, here i will discuss the
ofice itself. The exact purpose of the ofice of ly drb is not clear,
571
and different
interpretations have been suggested. mostly the term ly has been translated as
governor, and thus a common view is that the holders of the ofice simply con-
trolled the bedouins of the osrhoene
572
or acted mainly as intermediaries between
edessa and its arab sorroundings.
573
other interpretations suggest that the holders
of the ofice were instead integrated in the native arab society. so luther seems to
interpret them rather as native arab rulers being subject to the edessene king.
574

more explicit in this context is sommer, who interprets personal names constructed
with bar combined with a reference to children, brothers and father in the inscrip-
tion of Tiridates as relecting the integration of the governor of Arab into a tribal
structure. furthermore no settlements are found in the near surroundings, and thus
nomadic groups are to be assumed. sommer thus pledges for a social dimorphism
in osrhoene.
575
in this context it is currently assumed that the osrhoenean ly drb resembles
the ofice of an arabarches mentioned in a papyrus from dura-europus from ad
121,
576
which has been briely discussed with regard to the term Arab in the third
section. apparently, this function is also known from ptolemaic papyri.
577
further-
more cicero in a letter from 59 bc asks atticus to spy on a certain Theophanes to
experience the attitude of an Arabarches towards him.
578
in this context Arabar-
ches probably is an ironic designation for pompeius, and by this cicero either wants
567 on the problem of the extension of the kingdom of edessa see sommer (2005), p. 235.
568 e.g. Joseph. Ant. 14.15; 370; 16.220; 20.77; BJ 1.159; 181; 360; 487; cass. dio 37.15.1; proc.
bell. 1.19.20. on the nabataeans as arabs see e.g. Joseph. BJ 1.187. on Petra in Arabia see
e.g. Joseph. Ant. 14.80; BJ. 1.267; 4.454.
569 on the inscriptions and sumatar itself see ross (2001), p. 245; segal (1953), p. 97107; segal
(1970), p. 23; sommer (2005), p. 2523.
570 drijvers and healey (1999), as36,1; as47,3; as49,23; as51,1; as52,4; add3,3.
571 see sommer (2003a), p. 223.
572 sartre (2005), p. 147.
573 ross (2001), p. 25.
574 luther (1997), p. 173.
575 sommer (2005), p. 2545. The inscription of Tiridates: drijvers and healey (1999), as36 (=
dijkstra (1995), p. 253).
576 PDura 20,5; see n.224. on the resemblance see sommer (2005), p. 254.
577 see rets (2003), p. 4101; KlP, col. 4823 s.v. Arabarches (colpe).
578 cic. Att. 2.17.3.
302 Ulf scharrer
to express pompeius authority over the arabs or to stress ironically an arab iden-
tity of pompeius, calling him somewhat like the commander of the arabs.
579

Therefore the exact meaning of ciceros use of the loan-word cannot be stated un-
ambiguously. The ofice of an arabarches mentioned in the dura papyrus itself is
being held by a certain manesos, son of a phraates. besides being arabarches ma-
nesos also was strategos of mesopotamia and parapotamia and had the function of
a aooaotg, a term in this context dificult to translate, since the function of this
ofice is not clear.
580
furthermore, manesos seems to have had different oficials
under his direct command, such as the lender in PDura 20, the eunuch phraates.
finally, manesos quite likely is of iranian, i.e. parthian origin, as his own and his
fathers name indicate. Thus, manesos appears to have been a high iranian ofical
at dura-europus.
581
so far the exact administrative or military function of the durene ofice is not
clear.
582
segal writes that the main function of the ofice was the organization of
protection of roads against arab attacks and the exercise of jurisdiction.
583
in con-
trast to that, sommer, in analogy to his interpretation of the osrhoenean all
deArab, suggests that the durene arabarches too indicates a dimorphic social
structure at dura-europus. he bases his arguments on two observations: irst, se-
leucus, a durene oficial known from a irst-century bc inscription, is known to
have been genearches and strategos of the city. sommer regards the term genear-
ches as resembling the term phylarchos, a sheikh. although he states clearly that
seleucus and his successors claimed to belong to the macedonian leading class of
dura, sommer suggests without further argument that seleucus, by being genear-
ches, was also integrated into tribal contexts, assumingly nomadic surroundings of
the city. sommer continues that as strategos of mesopotamia and parapotamia ma-
nesos had control only over the territory of dura-europus, not a very large district,
and he concludes that the strategies of seleucus and manesos had basically the
same territorial extension. again, sommer continues, that seleucus and manesos
likely had the same functions, and that both were integrated into tribal structures as
tribal leaders, seleucus as genearch, manesos as arabarch.
584
sommers second
observation is that one of the witnesses of the treaty in PDura 20, metolbaissa, was
the grandson of the military commander and parthian ofical menarnaios. sommer
irst interprets the names as semitic and thus steming from nomadic or local peasant
milieus, and argues that the local oficials were ethnically and culturally distinct
579 The second view has been brought forward by rets (2003), p. 410.
580 on the problem see sommer (2005), p. 296, who interprets the ofice as a kind of Hofsene-
schal, actually a term from medieval europe since the merovingians. on this ofice see e.g.
halphen (1995), p. 143; lebecq (1990), p. 68; mitteis and lieberich (1992), p. 72; schneider
(1995), p. 55.
581 see sommer (2005), p. 2989.
582 see millar (1998a), p. 477.
583 segal (1970), p. 223.
584 sommer (2005), p. 297300. see also sommer (2003a), p. 22.
303 The problem of nomadic allies in the roman near east
from rather macedonian urban elites. both assumptions he interprets as showing
the merging of tribal and bureaucratic structures.
585
of course, sommers hypothesis is not to be dismissed entirely, but it seems to
be based on a number of assumptions which are at least debatable. first, it is not
compulsory that a genearches was the same as a phylarchos. basically, both phyl-
archos and genearchos could also have been ofices with regard to the political
structure of a city itself. furthermore, it is by no means clear that the arabarches
had the same function as the genearches beforehand, as sommer argues, namely
that seleucus and manesos basically had the same scope of ofices, by rather dimi-
nishing the title of strategos of mesopotamia and parapotamia. also, the names
mentioned in PDura 20 are not necessarily of semitic origin. What the document
shows, however, is that all those mentioned were integrated into the parthian elite.
Thus it seems that sommer tries to prove his theory of a dimorphic structure at
dura-europus by applying the necessary assumptions.
586
i therefore tend to the far
less sophisticated view that the function of the arabarches might have been only
the exercise of authority over sedentary and nomadic steppe dwellers,
587
or even
more carefully said, that the holder of the ofice established relations of whatever
kind with the neigbouring Arab populations. maybe he dealt with members of sa-
faitic groups which as i said in the fourth section 4 are found quite often at
dura, maybe he exercised some sort of military control, and maybe he did both. it
also could be the case that the ofice was created within the context of a roman
retreat from mesopotamia after the campaigns of Trajan,
588
thus in presumably a
period of some turmoil. We do not seem to know any details for certain. at least
manesos, the holder of the ofice, seems to have been a high parthian oficial, which
might stress the importance added to dealing with Arab at least in this period.
from the limited evidence it is not clear to what extent the durene arabarches
really resembles the ofice of the osrhroenean all deArab.
589
indeed, so far at
least one other all is known from osrhoenean inscriptions, a governor of birta,
known from an inscription dated to ad 67.
590
sommer seems to suggest that birta
and Arab are to be regarded as two distinctive administrative units: whereas he re-
gards the governor of Arab as being integrated into a nomadic tribal society, the
governor of birta, coming from completely different natural surroundings, was a
governor of a certain district of administration.
591
it is then quite reasonable to as-
sume that Arab in the inscriptions from sumatar does not refer to a clearly deined
district of administration, but to the steppe and its dwellers.
585 sommer (2005), p. 3001.
586 see the brief criticism of this way of proceeding in Kaizer (2002), p. 264, followed by scharrer
(2003), p. 75.
587 see rets (2003), p. 411.
588 see rets (2003), p. 439. on the eastern campaigns of Trajan see bennett (1987), p. 191203;
dillemann (1962), p. 27386; halfmann (1994), p. 5817; sartre (2005), p. 1467.
589 see aggoula (1995), p. 756; ross (2001), p. 25.
590 drijvers and healey (1999), as55.
591 sommer (2005), p. 2535.
304 Ulf scharrer
it is also certainly not that surprising that Tiridates, one governor of Arab, set
up an altar among others for the lives of the king and his children, thus expressing
his loyalty to the royal house at edessa.
592
of course it may be asked whether Tiri-
dates might have been a native inhabitant of birta, but a deinite answer to this
question is not to be found. furthermore, another governor of Arab, a certain
barnabar, set up a statue for a freedman of antoninus ([n]twnyns) caesar, aurelius
afsay (wrylws fsy).
593
probably the inscription dates to the years ad 17480,
as the caesar, being called lord and benefactor in the inscription, is likely to be
marcus aurelius antoninus,
594
which also is indicated by the freedmans name.
aurelius apsay seems to have been an important igure in osrhoene in that period,
since he is honoured by two more inscriptions from sumatar.
595
apparently, by ho-
nouring a roman freedman, barnabar at the same time expressed his loyalty to
rome, very likely in the time that edessa was ruled again by the pro-roman king
manu Viii, after the interruption by the pro-parthian prince Waels seizure of po-
wer in ad 162.
596
in fact, the ofice of a governor of Arab seems to have been in
some connection with these events, since, as far the evidence is concerned, the of-
ice was introduced in the context of Waels usurpation.
597
maybe the opponents
wanted to secure Arab support. Thus, barnabar, with his expression of loyalty to
rome, also demonstrated his support of manu Viii. so far there is nothing known
about the role played by the Arab in this context: most likely it seems that they fol-
lowed the political alliance of their all.
in the context of the change of alliances a passage from cassius dio needs to
be discussed briely. With respect to the events in ad 2123 dio writes that abgar,
the king of osrhoene, rose to rule over kindred tribes (oouev) and that he was
very cruel to their leaders (aorovtoi) and changed them to roman customs.
598

Very probably dio refers in this passage to abgar iX severus, who in ad 2123
ruled according to dio for just nine months, before edessa was turned into a colony
by caracalla in the same year.
599
The question in the present context is what could
have been implied by the terms oouev and aorovtoi. There are mainly three
different interpretations: either this passage may be regarded as evidence of a
somehow dimorphic or maybe even segmentary society. if read in this way, abgar
forced the leaders of different tribes in the osrhoene and at edessa itself under his
rule, the tribes being related by social ties with each other. it could also be read so
that abgar asserted his power against outstanding people of his own or related fa-
592 drijvers and healey (1999), as36. see dijkstra (1995), p. 253; sommer (2005), p. 253. on the
phrase for the life of , see generally dijkstra (1995).
593 drijvers and healey (1999), as49 with comments. see ross (2001), p. 25
594 sommer (2005), p. 255.
595 drijvers and healey (1999), as48; as50. see sommer (2005), p. 255.
596 on the problem of Wael, see ross (2001), p. 40.
597 see drijvers (1980), p. 12234; drijvers and healey (1999), p. 195; ross (2001), p. 401;
sommer (2005), p. 253 n. 92.
598 cass. dio 78.12.1a.
599 sartre (2005), p. 149 and p. 344; sommer (2005), p. 2423.
305 The problem of nomadic allies in the roman near east
miliy or tribe. finally, the terms might be interpreted simply as relating to the city
of edessa itself and in this context maybe especially to civic tribes.
Unfortunately none of these hypotheses can be proven. and it remains a matter
of discussion why the osrhoene was turned into a colony. sommer suggests that the
nomadic society of the osrhoene was undergoing a process of sedentarization and
that rome therefore turned to the strategy of direct rule instead of indirect pow-
er.
600
apparently, there is no evidence at all for an ongoing process of sedentariza-
tion in the osrhoenean realm although in fact, as discussed briely in the third
section, the beginning of a bedouinization process of the near east is assumed for
this period. The (irst) annexation of the osrhoene probably stands in the context of
roman and sasanid policy of turning the last client kingdoms into colonies, maybe
in connection with the rise of sasanid power. so far there is no evidence that noma-
dic groups in this context appeared in roman strategic thinking.
so the evidence on nomad-sedentary socio-political bonds from the osrhoene
is quite ambiguous. for certain, the inscriptions from sumatar reveal the associa-
tion of the governor of Arab with the edessene king, be he pro-roman or pro-
parthian.
601
The question is, however, whether the edessene kings just exercised
indirect control over the Arab by a subordinate ofice,
602
or whether edessa was the
centre of a tribal state within a dimorphic society.
603
if the notion of a dimorphic
society is narrowed to the concept of steppe populations being controlled by a city
centre, the evidence from sumatar is unambiguous with that respect, as the os-
rhoenean steppe dwellers were controlled from the city. Whereas there certainly
were socio-political relations,
604
it has to remain a matter of debate whether the
osrhoene was an integrated dimorphic tribal society, as it seems to have been at
hatra and its surroundings, to which i will turn now.
it has been suggested that the osrhoenean ly drb resembles the hatrean rbyt
dy rb, currently translated as majordomo of Arab,
605
or even the hatrean king of
Arab.
606
The early urban development of hatra still remains enigmatic, as it is the
case with most of the cities covered here. most commonly the sedentarization of
segments of the nomadic population of the Jazirah is assumed as the reason for
hatras urban growth.
607
Within this context, as at edessa, the development of titles
is of some importance. first of all, like at edessa, the title of a mry (lord) became
substituted by the title mlk (king), probably after ad 1512 and probably in the
context of the campaigns of lucius Verus in ad 1616.
608
600 sommer (2005), p. 242.
601 see ross (2001), p. 25.
602 drijvers and healey (1999), p. 105.
603 sommer (2005), p. 59. see the references above.
604 see dijkstra (1995), p. 251.
605 dijkstra (1995), p. 253. on evidence see above, n. 222.
606 ross (2001), p. 25 n. 26.
607 see e.g. dijkstra (1995), p. 175; sommer (2005), p. 368.
608 see altheim and stiehl (196469) iV, p. 2657; dijkstra (1990), p. 967; dijkstra (1995),
p. 17785; hartmann and luther (2002), p. 164; hauser (1998), p. 5023; hauser (2000),
p. 189; Kaizer (2006b), p. 1478; luther (1999b), p. 44850; segal (1986), p. 604; sommer
306 Ulf scharrer
so far, the title of the lord of the arabs (mry rby) is attested only once in the
hatrean inscriptions,
609
besides many other references to lords in general.
610
in
this context cassius dio refers to a certain mannus, who in ad 115 was phylarch
of arabia near edessa.
611
There are some uncertainties about this mannus: sartre
seems to regard him as an osrhoenian phylarch from sumatar under abgar Vii
(reg. ad 10917).
612
however, the most prominent view is that the mannus of cas-
sius dio was a hatrene lord, probably to be identiied with the lord man known
from hatraean inscriptions.
613
if this is correct, the lord of hatra was regarded as
some sort of tribal leader at least in Graeco-roman eyes, and this view is also sta-
ted occasionally in scholarship:
614
dijkstra assumes that there may have been more
than one lord of hatra at the same time, being tribal leaders.
615
according to som-
mer the hatrene lord was some sort of primus inter pares in the elite of tribal pa-
triarchs settling at hatra.
616
he stresses the existence of elders (qy), interpreted
by him as evidence for some sort of tribal council, and the lord could also be an
elder at the same time.
617
so far, neither view can be proven conclusively. at least
before the introduction of kingship at hatra elements of a segmentary society can
be found.
in this context it should be stressed, that none of the known inscriptions menti-
ons something like a lord of hatra: With the one above-mentioned exception,
only the title of a general lord has come down to us. it is to some extent similar
with the kingship at hatra: only the titles of the king or king of Arab/the arabs
are known.
618
The titulature has been interpreted repeatedly as expressing a com-
mon identity of hatra and its steppe surroundings.
619
This may be supported by a
passage from abars History. The author writes that at hatra a certain sirn of
the Qudaah from the clan of the Tazd b. ulwn ruled in the Jazira over many sub-
tribes of the Qudaah. his kingdom in the time of shapur i is said to have extended
as far as syria.
620
if this is the last hatrene king, sanatruq ii,
621
this would indeed
(2003a), p. 2933; sommer (2003b); sommer (2005), p. 3704; 3813; Tubach (1986),
p. 1969, p. 231 and p. 2467. see also dijkstra (1995), p. 3.
609 H78.
610 see the references in the works cited above, n.508. see also the references in hartmann and
luther (2002), p. 164 n. 15.
611 cass. dio 68.21.1.
612 sartre (2005), p. 146.
613 see sommer (2003b), p. 3889 and p. 392; sommer (2005), p. 3734.
614 e.g. generally by hartmann and luther (2002), p. 1634.
615 dijkstra (1990), p. 97; dijkstra (1995), p. 181.
616 sommer (2003b), p. 3934.
617 sommer (2005), p. 379.
618 see aggoula (1994), p. 160. King of Arab / the arabs: H193,2; H194,12; H195,12 (= dijk-
stra (1995), p. 228); H196 (?); H197,2; 4; H198,23; H199,23; H203,12; H287,45 (= dijk-
stra (1995), p. 2301); H290,2 (= dijkstra (1995), p. 2023); H342 (= Kaizer (2006b), p. 143
4), 12 (?); H345,2; H347,2; H353,2; H370; H373; H375; H379; H018,5.
619 e.g. hauser (2000), p. 191; millar (1998a), p. 495; sommer (2003a), p. 22; sommer (2005),
p. 379.
620 abar, Tarkh (History) i 827.
621 on the identiication see Zwettler (1993), p. 10.
307 The problem of nomadic allies in the roman near east
indicate strong tribal ties between the ruler of hatra and steppe-dwellers, and thus
the title king of Arab/the arabs would indeed express a common identity.
in addition, repeatedly H79
2
has been considered as an important document,
dating from probably right after ad 238. it testiies the erection of a statue for sa-
natruq ii by two brothers, Yahbararyan and alkud, who name their genealogy
down to the fourth generation. The inscription furthermore states, that the dedica-
tors trust upon nara and the Gad of Arab, that mana, son of sanatruq, may not
suppress them by force or, in another possible translation, that they will not do any
harm to mana. in addition, the inscription mentions the possessions or the people
of the dedicators outside and inside and the hope for remembrance in hatra and
Arab.
622
Their ancestor amibarak is said by dijkstra to be the same person as a
majordomo (rbyt) of this name mentioned in H336 and H343, as well as in H25.
These inscriptions record a decision of the hatrenes and the arabs, that anyone
stealing within or outside the citys wall shall be stoned. What could these inscrip-
tions show? dijkstra reasonably interpretes them against a theoretical background
of a dimorphic and segementary society: all persons mentioned thus represent
people inside and outside hatra, that is the city and its Arab hinterland.
623
howe-
ver, as dijkstra himself writes, H79
2
also records the end of a conlict between
hatra and the steppe dwellers.
624
The division between hatra on the one hand and
Arab on the other might allow the conclusion that at least at this time both parts
where not a single political and social entity,
625
although the title of a king of Arab
apparently tries to enforce that. in this context it is probable that the Gad of Arab
referred to in the inscription is worshipped both by city- and steppe-dwellers.
against this background it appears as rather dubious that cassius dios reference to
the arabian cavalry as having assailed the army of septimius severus at hatra in ca
ad 200 just implies nomads from outside hatra.
626
in the preceding section i already mentioned the ofice of a majordomo
(rbyt). it has been debated whether the translation majordomo is appropriate:
sommer stresses the literal meaning to be great of the house actually the literal
meaning of major domus and thus suggests rather to translate administrator or
principal.
627
it is a matter of question whether the function of the hatraean rbyt
may have been similar to that of the safaitic nr, currently translated as overseer.
628

however, two hatrean inscriptions explicitly mention a majordomo/principal of
Arab (rbyt dy rb).
629
dijkstra suggests, that also the rbyt in H336 and H343 is a
622 on the inscription in detail and its datation see caquot (1963), p. 25; dijkstra (1990), p. 81
90 and p. 934; Teixidor (1964), p. 2804. see also altheim and stiehl (196469) iV, p. 2435.
623 dijkstra (1990), p. 905 and p. 97. see also dijkstra (1995), p. 1838; hauser (1998), p. 513
4; hauser (2000), p. 1901; sommer (2003a), p. 368 and p. 401; sommer (2005), p. 3769;
Zwettler (1993), p. 101.
624 dijkstra (1990), p. 93 and p. 97. see also sommer (2003a), p. 38.
625 see altheim and stiehl (196469) iV, p. 245; segal (1986), p. 60.
626 cass. dio 76.11.2, with the interpretation of sommer (2005), p. 369. on the emperors siege of
hatra see Tubach (1986), p. 21623.
627 sommer (2005), p. 3767.
628 see above, n. 4712 and n. 540.
629 H223,2 (= dijkstra (1995), p. 3; 228); H364,2.
308 Ulf scharrer
majordomo of Arab,
630
but this is not compulsory, as there could have been other
realms of the ofice of the rbyt.
631
however, H336 and H343 do reveal that the
rbyt could chair councils and probably act as an intermediator between conlicting
parties. perhaps this was also one of the main tasks of the rbyt dy rb. apparently,
the ofice of a rbyt could be exercised also by people dwelling in the steppe, as the
reference to amibarak having been rbyt shows.
632
Thus, if the concept of a di-
morphic society is appropriate at all, it certainly is with respect to hatra.
This again is a matter of debate with regard to palmyra. an ethnic identity bet-
ween Tadmoraeans/palmyrenes and nomadic groups has been assumed repeatedly
for different periods of palmyrene history. Whereas with respect to the second mill-
ennium bc this view is mainly based on an unlikely complement of a fragmentary
tablet from mari and on an overinterpretation of a passage in the annals of Tiglat-
pilesar i,
633
the idea of a somehow nomadic origin of the palmyrenes of the roman
period is also quite common in scholarship.
634
This notion is based on different
evidence. first, appians narrative on a raiding campaign of marcus antonius
against palmyra in 41 bc is often interpreted in this direction, as the light of the
palmyrenes carrying with them their removable possessions is regarded as expres-
sing some sort of nomadic habit.
635
first the historicity of this passage seems to be
doubtful.
636
in addition the whole narrative does not reveal anything about nomadic
or semi-nomadic groups.
637
furthermore the worship of gods regarded as typical
steppe deities has been interpreted as showing a rather recent sedentarization.
638

however, the division of nomadic and sedentary deities appears to be rather doubt-
630 dijkstra (1990), p. 90.
631 segal (1986), p. 645; sommer (2005), p. 379.
632 see sommer (2005), p. 379. on the ofice of majordomo see also Kaizer (2006b), p. 143.
633 The relevant passages of the mari tablet and of Tiglat-pilesars annals are quoted and discussed
with references by scharrer (2002b), p. 30711 and p. 3168. sommer (2005), p. 150 n. 36
apparently misreads me when he states that in scharrer (2002b), esp. p. 3189, i suggest a di-
morphic structure at second-millennium bc Tadmor on a thin basis of evidence ( der (ebd.
319) die bronzezeitliche oase als urbanen mittelpunkt einer dimorphen Gesellschaft cha-
rakterisiert, freilich auf sehr dnner Quellenbasis.). in fact on p. 319 i am saying that against
the background of the sparse material no far-reaching conclusions are to be drawn. in this con-
text it is stated there that it cannot be said to what extent there was a dimorphic or polymorphic
structure at Tadmor (die fragen nach dem Komplex polymorpher bzw. dimorpher struktu-
ren knnen aufgrund des zugnglichen Quellenmaterials nicht beantwortet werden.). on the
concept of a polymorphic society see scharrer (2002b), p. 296 with references. maybe my
discussions of terminonolgy and approaches, in the context of second-millennium bc Tadmor
probably a bit extensive (scharrer (2002b), p. 283301), somehow stimulated sommers inter-
pretation of my views.
634 e.g. altheim and stiehl (196469) ii, p. 270; funke (1996), p. 222; Gawlikowski (2003),
p. 89; matthews (1984), p. 169; sommer (2005), p. 180; staubli (1991), p. 118.
635 app. B. Civ. 5.9. The passage is quoted fully also in sommer (2005), p. 152 n. 39. on the inter-
pretation see e.g. seyrig (1970), p. 89; sommer (2005), p. 180 n.156. see furthermore the refe-
rences in hekster and Kaizer (2004), p. 71. see also millar (1993), p. 321; scharrer (2002b),
p. 308.
636 hekster and Kaizer (2004).
637 see scharrer (2002b), p. 3089.
638 seyrig (1970), esp. p. 92.
309 The problem of nomadic allies in the roman near east
ful.
639
finally the existence of different sorts of tribes at palmyra is seen as sho-
wing at least a tribal origin of the palmyrene society, often equated with nomadic
roots.
640
This view has been modiied by Kaizer, who convincingly argues that tri-
bes were not restricted to nomadic groups. he furthermore points to the integration
of palmyra and the villages of its hinterland and thus suggests as a hypothesis that
through ecological conditions the city of palmyra developed in the context of an
interplay with the surrounding villages.
641
however, the ecological circumstances
and probable changes which may have led to a rather sudden urbanization are not
clear,
642
and i would thus like to modify Kaizers hypothesis by stressing probable
economic and political developments, which might have led to an urbanization of
palmyra. however, so far the concept of a rather prompt sedentarization of no-
madic tribes at palmyra seems to stand on uncertain grounds.
in this context it is noticeable that none of the palmyrene tribes are known from
other regions. sommer stresses, however, that the palmyrene bny tym are also
known from inscriptions from hatra and the hauran. although he admits that tym
and tymw are quite common names and the term bny in some cases could also refer
to families, he suggests family ties between hatra and palmyra. as evidence for
this he also interprets palmyrenean inscriptions from hatra. based on this quite
ambiguous evidence sommer states that there were strong ties between nomads
and sedentaries in the palmyrene. although he clearly says that the idea of a
palmyrene militia or desert police is a modern presumption, which is not based on
any direct evidence, he assumes its existence by the presence of palmyrene archers
at the euphrates and the fact that caravans were protected. against the background
of these two assumptions sommer states that the palmyrene militia were recruited
among nomadic tribal relatives of the palmyrenes, and as a support for this hypo-
thesis he refers furthermore to the dislocation and equipment of the militia a mi-
litia, whose existence is not really certain. sommer continues with assuming that
the strategoi of the militia probably were recruited among the nomad elites. he re-
gards the strategoi as having a double function, being nomad cheikhs and urban
oficers. from all these assumptions he draws the conclusion that similarily to ha-
tra there was a dimorphic society at palmyra.
643
in addition, the safaitic inscriptions from the palmyrene have been interpreted
along similar lines: according to seyrig the writers of these inscriptions were part
of the sedentarised population.
644
altheim and stiehl conclude from the inscrip-
639 see Kaizer (2002), p. 567.
640 Gawlikowski (2003), p. 89; hartmann (2001), p. 467; matthews (1984), p. 169; sommer
(2005), p. 1801. following the idea of tribal origins, but stating clearly that this does not
mean necessarily nomadic roots: dirven (1999), p. 227. see also the critical discussion in
Kaizer (2002), p. 4355. against this view see also scharrer (2003), p. 73.
641 Kaizer (2002), p. 513. on the palmyrene see schlumberger (1951); millar (1993), p. 299.
642 see scharrer (2003), p. 73.
643 sommer (2005), p. 155, p. 17881, p. 207, p. 2123, p. 216, p. 218 and p. 222. The idea that a
part of the palmyrene urban elite was made up of bedouin chieftains is also stated by dirven
(1999), p. 21, and staubli (1991), p. 118. already in sommer (2003a), p. 43, palmyrene society
is characterized as dimorphic.
644 seyrig (1970), p. 82.
310 Ulf scharrer
tions that nomads were in the army of palmyra.
645
as said in the fourth section, the
evidence from the palmyrene and palmyra itself in the irst instance shows that
there were some relations, appararently mainly economic. Thus the assumption of
a dimorphic society at palmyra, as at dura-europus, is mainly based on presumpti-
ons which are not at all relected unambiguously in the evidence. besides, as i have
also discussed in the fourth section, the Greek section of the bilingual in honour of
oglu refers to campaigns against the nomads.
646
sommer writes correctly that the
inscription falls into the period of the parthian campaign of septimius severus.
647

it is not stated in the text that oglus campaigns took place in this context, but it
rather seems the case that the campaigns were on behalf of palmyra itself. howe-
ver, the lack of a palmyrenean equivalent to nomads in the bilingual is interpreted
by asad and Yon as showing an intention on the palmyrenean side not to speak
about victories over nomads in a language they could understand.
648
Unfortunately
it is not known which nomads exactly were fought against, and therefore the extent
of antagonism between nomadic groups and palmyra with respect to ca ad 200 is
not clear. sommer states that the palmyrene caravan inscriptions throughout draw a
misleading picture of nomads as foreigners and enemies.
649
as i have said in the
fourth section, there is only the oglu-inscription which explicitly draws such a
picture. however, palmyrene relations with nomadic groups can certainly not be
reduced to simple antagonism.
With respect to the question of a common ethnic and social identity of palmy-
renes and nomadic groups, it has been stressed already that there is no equivalent
known from palmyra to the hatraean king of Arab/the arabs. on the contrary, the
honorary title Arabicus Maximus of queen Zenobias son Vaballathus from ca ad
270 expresses a socio-political distinction, if not antagonism to arab steppe dwel-
lers.
650
The mention of this year, actually the fall of palmyra, leads to the problem
of its famous queen Zenobia. Whereas her ancestry is wideley discussed,
651
the is-
lamic tradition, and here especially abar, presents her under the name of al-Zabb
as stemming from nomadic origins, and consequently abar describes the war
against palmyra and its eventual fall completely as a tribal conlict: her father amr
b. arib, ruler of a territory from the Jazirah to the fringes of syria, was slain during
an attack by Gadhima. Zenobia revenged her father by killing Gadhima, after she
had offered him in vain to join their lands.
652
of course it is not clear to what extent
abars narrative is to believed literally:
653
while Zenobia after the defeat of pal-
myra was brought to rome by aurelian, abar writes that she was killed in re-
645 altheim and stiehl (196469) ii, p. 2723.
646 see above, n. 349. see furthermore millar (1993), p. 3323.
647 sommer (2005), p. 206 n. 249.
648 asad and Yon (2001), p. 63.
649 sommer (2005), p. 179.
650 ILS 8924. Kaizer (2002), p. 578. see also Yon (2003), p. 15. Generally with respect to palmy-
rene inscriptions: sommer (2005), p. 131; hartmann (2001), p. 634.
651 see e.g. ingholt (1976), p. 1367.
652 abar, Tarkh (History) i 75661.
653 abars narrative about Zenobia is taken literally e.g. by ball (2000), p. 7885. see also roth-
stein (1899), p. 37. on abars narrative see equini schneider (1993), p. 4552.
311 The problem of nomadic allies in the roman near east
venge for the murder of Gadhima.
654
furthermore, he omisses completely the ro-
man part in the fall of palmyra. against this background also Zenobias nomadic
descent is rather questionable. if there is some truth in this, some sort of dimor-
phism with respect to Zenobias rule might be assumed.
however, besides the legendary traces abars narrative may be relevant irst
with respect to the arab allies of Zenobia.
655
according to the Historia Augusta
Zenobia was allied with the saracens and the armenians.
656
Whereas the Historia
Augusta does not say anything about the exact ethnic identy of Zenobias saracen
allies, abar writes that her army in the ight against Gadhima consisted of rem-
nants of the amliq and proto-arabs, furthermore the Tazd, the li and some
Quah.
657
apparently, some of these represent groups already dwelled in the re-
gion before the arrival and rise of the Tankh. Therefore some sort of ethnic con-
lict may have been involved in the war of rome against palmyra.
658
however, it
seems to be rather inappropriate to reduce this to a tribal conlict.
659
it is apparent
that palmyrene rule at least at this time extended also over nomadic groups, as
seems relected by a remark of procopius calling Zenobias husband odaenathus
ruler of the saracens living in the region.
660
While this remark together with
abars narrative on the origins of Zenobia/al-Zabb might be interpreted as indi-
cating some sort of a dimorphic society at palmyra, it rather seems that both the
Historia Augusta and abar present saracens or, as the case may be, arab groups
as allies of palmyra,
661
which maybe were won by aurelian, to which i shall turn a
bit later.
in this context the role of safaitic tribes needs to be discussed briely. in the
fourth section i have sketched probable economic and religious relations of safaitic
people with palmyra, and it was said there that there is hardly any evidence on con-
lict. Graf suggests intimate connections and he states that at least some safaitic
tribes, such as the important group of the Aw, might have been allied with pal-
myra and its queen against rome. This suggestion is based on the assumption that
the safaitic inscriptions equate media with palmyra, and Graf thus interprets the
references to a roman-median war as evidence for palmyras struggle with rome,
stating that at least some writers of the safaitic inscriptions were friendly to Zeno-
bia.
662
indeed, besides general references to medes in some safaitic inscriptions,
663

one inscription mentions a rebellion against rm in the year the medes came to
654 abar, Tarkh (History) i 766. on Zenobia in the triumph of aurelian at rome see sha Aurel.
34.3.
655 on problems of abar see hartmann (2001), p. 34950; millar (1993), p. 4323.
656 sha Aurel. 27.4.
657 abar, Tarkh (History) i 757.
658 on the li see bowersock (1994c), p. 135; hartmann (2001), p. 342; sartre (1982a), p. 146
9; shahd (1989), p. 2447.
659 Thus done apparently by ball (2000), p. 78. on the conlict see bowersock (1983), p. 1337.
660 procop. Bell. 2.5.5.
661 see hartmann (2001), p. 276; sommer (2005), p. 169.
662 Graf (1989b), followed by ball (2000), p. 79, and partly by equini schneider (1993), p. 68. see
also Graf (1989a), p. 392.
663 e.g. SIJ 78; 88.
312 Ulf scharrer
bosra.
664
a further inscription states that a certain ain of the aif acted as guide
in the year caesar ejected the medes.
665
finally, an inscription refers to booty
being made in the year of the war of rm against the medes.
666
The latter inscrip-
tion can actually be used only as evidence of nomad raiding. Whether the war of rm
and the medes provided good opportunities for this may be a matter of debate.
more interesting are the references to the rebellion and the guide. it is quite pro-
bable that the campaign of the medes provided a good opportunity for rebellion
against roman rule. The acting as a guide by ain is more dificult to interpret,
since it is not clear for whom he did this, whether it concerned roman or median
forces or whether it took place in a very different context, maybe that of caravan
trade. however, Grafs equation of the safaitic medes with palmyra appears to be
rather speculative.
667
What can be read from the inscriptions is that apparently in
periods of roman weakness or trouble some safaitic groups tried to escape roman
rule. The mentions of medes could also refer to persian campaigns, e.g. in 40 bc,
ad 260 or even ad 614, the middle date being the most likely.
668
as indicated by abar there was some conlict between palmyra and the
Tankh already before the roman campaign against palmyra. probably there were
some palmyrene ights with the Tankh in the hauran, likely in the context of the
westward extension of Tankh power under Gadhima and the palmyrene occupa-
tion of arabia,
669
and it is likely that already odaenathus struggled with arab no-
mads.
670
i have suggested above that Zenobias arab/saracen allies probably consi-
sted of groups which eventually became incorporated into Tankh power, and thus
their alliance might have been directed at least to some extent against the growing
power of the Tankh, actually the opponents of palmyra. in this context it is hence
a matter of debate who exactly the saracens were that aurelian, according to the
Historia Augusta, could bring on his side by force and diplomacy.
671
it is not clear,
whether among these were groups mentioned speciically as allies of Zenobia by
abar. probably this passage refers to an alliance between aurelian and the Tankh
based on the common interest of war against palmyra. it seems as if aurelian could
more or less easily win these saracens. on the other hand aurelian in his triumph
at rome after his victories had among other prisoners of war Saracens being led.
672

Unfortunately it is not clear who these Saracens exactly were. it might be assumed
that they stemmed from those groups supporting Zenobia, probably some of the
tribes mentioned by abar.
664 SIJ 78.
665 SIJ 88.
666 CIS V 4448.
667 see hartmann (2001), p. 2767.
668 cf. Graf (1989b), p. 153; moors (1992), p. 311. according to dussaud (1955), p. 140, the in-
scriptions refer to the persian war of ad 614.
669 see equini schneider (1993), p. 47 and p. 689; hartmann (2001), p. 27881 and p. 349;
moors (1992), p. 309.
670 hartmann (2001), p. 194.
671 sha Aurel. 28.2.
672 sha Aurel. 33.4.
313 The problem of nomadic allies in the roman near east
To what extent the other saracens, probably the Tankh, were decisive in the
inal fall of palmyra is a matter of debate.
673
however, in this context maybe the
title of Vaballathus as Arabicus Maximus becomes clearer: The title may either re-
fer to Vaballathus conquests in the provincia Arabia,
674
or his campaigns were di-
rected against those arab groups hostile to palmyra, probably those being allied to
rome or the Tankh. less likely then is the idea that Vaballathus just campaigned
against nomads raiding palmyrene territory, probably being in persian service.
675

Thus at palmyra, at least in Zenobias time, antagonism can be found and, if not
really symbiosis, at least an alliance which became shattered eventually by the fall
of palmyra.
from all which up to now has been said, certain aspects of the problem of no-
madic allies have become obvious: nomadic groups existed in interaction with the
sedentary populations and especially to regional powers, i.e. the roman and per-
sian client states.
676
besides economic symbiosis there is also strong evidence for
socio-political bonds. as i have tried to show, however, it is rather inappropriate to
apply the concept of a dimorphic society to all regional powers alike. Whereas ha-
tra certainly can be described as such, the application of the concept appears to be
debatable in the cases of edessa and emesa, and quite unlikely with respect to pal-
myra. against this background it can for example not be said for certain to what
extent the ofices of the hatrean rbyt dy rb, the osrhoenean ly drb and the du-
rene arabarches resemble each other.
677
at least it seems unlikely, that the docu-
ments which have been discussed here resound the control over Arab as a hostile
element, as occasionally has been stated.
678
in contrast, the steppe seems to have
been more or less integrated in regional political structures, the degrees of integra-
tion spanning from a dimorphic society at hatra to alliances of palmyra. being so-
cially and politically more or less bound to regional powers, many nomads and
other steppe dwellers apparently followed their political centres in their alliances
with the great powers, rome and persia: up to aurelians successful efforts to win
saracen allies of Zenobia, there are no alliances or othere political actions known,
which nomadic groups might have pursued independently from their relevant cen-
tres, be it emesa, edessa, hatra or palmyra.
in this context the question may be raised to what extent an explicit roman
strategy of indirect control of steppe dwellers by regional powers, i.e. client king-
doms, actually existed, a view which has been stated repeatedly with respect to
673 That the saracen allies of rome were decisive has been suggested e.g. by bowersock (1983),
p. 137; Graf (1989b), p. 150. against this view e.g. hartmann (2001), p. 382 n. 83. see ball
(2000), p. 801 and p. 85.
674 Thus stated by Teixidor (2005), p. 2023.
675 Thus suggested by hartmann (2001), p. 268.
676 see Kennedy and riley (1990), p. 36; shahd (1984b), p. 31.
677 in favour of a resemblance see sommer (2005), p. 379, stating, that the hatraean rbyt dy rb
resembles the arabarches at palmyra. since the ofice is not known from palmyra, this is pro-
bably just an accidental error. see also above, n. 576, n. 589 and n. 6056. against the re-
semblance of the ofices see e.g. dijkstra (1995), p. 253.
678 millar (1998a), p. 512. see also sommer (2003b), p. 396; Teixidor (1964), p. 283. against this
view e.g. hauser (1998), p. 512.
314 Ulf scharrer
different regional powers, especially palmyra.
679
against the background of the
fourth section this view appears to be debatable: as a considerable threat to the east-
ern roman empire posed by nomadic groups seems not to have existed, the control
over steppe dwellers is not likely the main purpose of the institution of client king-
doms, the more so when, if isaac is right, nomadism in the near east was actually
never the dominating culture.
680
if in addition probable tribal bonds between no-
mads and villages and possibly some cities on the very local level are considered, it
looks as if socio-political control already existed to a considerable degree.
681
Thus
the main purpose of the client kingdoms certainly was the function of buffers and
allies against the parthian and to some degree to the sasanid empires.
682
alliances before ca ad 300
against this background it is not surprising that there is actually not so much evi-
dence on nomadic groups directly allied either to rome or parthia before the rise of
Tankh power. The irst references on arab groups as allies in roman contexts are
to be found in the evidence on the eastern campaigns of lucullus.
683
according to
plutarch, Tigranes of armenia gathered among others arabs against lucullus.
684

probably after roman successes against Tigranes, kings of the arabs are reported
to have offered lucullus their support against Tigranes.
685
eventually plutarch
states that lucullus could gain the territory up to the red sea by the support of the
arabian kings.
686
Unfortunately it is not clear whether all the arabs mentioned
by plutarch belonged to nomadic groups. however, it seems probable that at least
some of them did. Whereas plutarch does not give any details on the arabs, at
least one arab ally is known from other sources. cassius dio reports that the
arabian dynast alchaudonius in 69 bc offered his alliance to lucullus.
687
if, as
suggested above, he is to be identiied with strabos alchaedamnus, the king of the
nomadic rhambaeans, he would be the irst nomad ally of rome explicitly known.
luckily, there is further evidence on alchaudonius / alchaedamnus conduct
after his alliance with lucullus, and this its well with the remark just made, na-
mely that lucullus victories were decisive for the local dynasts offering their sup-
679 With respect to palmyra: hartmann (2001), p. 85, p. 98 and p. 427; petit (1971), p. 123; sartre
(1991), p. 69; shahd (1984b), p. 224. With respect to palmyra and the nabataean kingdom:
bowersock (1983), p. 155. With respect to the nabataean kingdom: Gichon (1991), p. 320;
sartre (1991), p. 64 (= sartre (1997), p. 55). With respect to emesa: ball (2000), p. 34. With
respect to hatra: sommer (2003b), p. 396. Generally: sartre (1991), p. 31 (= sartre (1997),
p. 27); sommer (2005), p. 64.
680 isaac (1998f), p. 4112.
681 on a coexistence see also shahd (1984b), p. 31.
682 see Wiesehfer (1982), p. 43940.
683 on lucullus eastern campaigns see dillemann (1962), p. 2638; sartre (2005), p. 357.
684 plut. Vit. Luc. 25.6.
685 plut. Vit. Luc. 29.5.
686 plut. Vit. Comp. Luc. 3.2.
687 cass. dio 36.2.5.
315 The problem of nomadic allies in the roman near east
port: cassius dio explicitly states that alchaudonius attached himself always to
the stronger party. corresponding to that he revolted when crassus parthian cam-
paign in 53 bc turned out to be a disaster.
688
according to strabo alchaedamnus is
to be found among the allies of caecilius bassus, a follower of pompey during the
civil War. strabo relates that originally he was a friend of the romans, but at some
point regarded himself to be treated unjustly by rome and thus retired to mesopo-
tamia and became a mercenary in bassus service.
689
if alchaedonius and achae-
damnus are identical, the supposed unjust treatment might have been a consequence
of his foresaking of cassius. at least already at this early time it is clear that tribal
leaders, and among them probably nomadic chiefs, tried to obtain their advantage
by attaching themselves to the apparently strongest power. similarily might have
been the conduct of other groups of which the nomadic element is not clear: Thus
cicero in 51/50 bc states that the emesene phylarch iamblichus is to be regarded
as a good friend of our state, and this he showed at least by reporting to cicero, at
this time governor of cilicia, that pacorus, the son of the parthian king orodes, had
crossed the euphrates.
690
in the civil War iamblichus and his father, together with
other phylarchs, were allied to caecilius bassus, and this was actually one reason
why bassus could stand a long siege being encircled at apamea.
691
With respect to allies, strabo remarks that the phylarchs and their tribes along
the euphrates are selfwilled. he continues that some phylarchs there attach them-
selves to the parthians, others to the romans. among these, nomadic tent-dwellers
(skenitai) near the euphrates are rather not attracted by rome, whereas the tent-
dwellers towards the west are.
692
Unfortunately no more details are known. it was
probably one of the functions of the arabarches at dura-europus, as discussed
above, to deal at least with some of these groups. at least parthia, so far as the evi-
dence goes, in contrast to rome, did have an ofice designed explicitly to deal with
relations with the steppe dwellers.
again, the inscription of rawwfa is problematic. in the third section i have
discussed the question of whether a military unit or a Thamd confederation is re-
ferred to. furthermore the inscription states that antistius adventus had made peace
among the Thamd (wrmhm).
693
according to bowersock, the phrase wrmhm
refers to intertribal conlicts, and thus the governor of the province apparently acted
as peacemaker among the tribes.
694
parker argues that this has to be seen in the
context of security troubles caused by nomadic tribes on the fringes of the empire,
and besides the threat of roman forces there was also diplomacy, at rawwfa the
securing of peace.
695
Graf furthermore writes that by the introduction of emperor
688 cass. dio 40.20.1.
689 strabo 16.2.10.
690 cic. Fam. 15.1.2 (amicum esse rei publicae nostrae). see Grouchevoy (1995), p. 110.
691 strabo 16.2.10.
692 strabo 16.1.278. see funk (1996), p. 2234.
693 inscription (see n. 179) b4.
694 bowersock (1975), p. 517. see also Graf (1997b), p. 172 and p. 178; isaac (1992), p. 239;
sartre (1982a), p. 130; sartre (1997), p. 317.
695 parker (2002), p. 78. see also Graf (1978), p. 11.
316 Ulf scharrer
worship at rawwfa, a central place of the Thamd, they became foederati of ro-
me.
696
especially after the parthian campaigns of lucius Verus a couple of inscrip-
tions, especially in the syrian steppe, attest to the worship and honouring of mar-
cus aurelius antoninus and lucius aurelius Verus.
697
The inscription from
rawwfa is the only bilingual of these so far.
The latter points might to some extent be put in contrast to macdonalds above-
quoted argument,
698
namely that the inscriptions and the shrine from rawwfa
were set up by a Thamd military unit: if so, why was it a bilingual? how could the
role of antistius adventus as peacemaker be understood? if macdonald is right,
however, the rawwfa inscription could not be used as evidence for nomadic al-
lies, since irstly, as stated above, it is not clear to what extent the Thamd themsel-
ves were nomadic, and secondly the relation to rome would not be exactly ob-
vious: it cannot be said whether the Thamd have to be regarded as a conquered
people, as allies or simply as a rather independent group from which auxiliaries
could be recruited. if the irst interpretation of the inscription is correct, it still can
hardly be stated conclusively whether rawwfa was a part of the roman sphere,
i.e. of the provincia Arabia, or not.
699
following the notion of limes as a zone of
contact, as discussed in the fourth section, this question seemingly is not that im-
portant, since rawwfa was certainly to be found within this zone. in this case at
least roman inluence on and diplomatic relations to the steppe would be attested
once more.
The problem of the rawwfa inscription leads to the aspect of probable auxi-
liaries in the roman and parthian armies. it has repeatedly been stated that on the
regional level nomads were employed by the roman and probably parthian autho-
rities for the protection of the frontier, especially against raids of other nomadic
groups or other enemies.
700
in this context the idea of client tribes has been brought
into discourse, the most important of which are stated to be the Aw.
701
in addi-
tion butcher suggests that dominant tribes furthermore had the task of collecting
tribute from other tribes and in their areas of inluence on behalf of rome.
702
The
notion of the Aw being a client tribe of rome is more or less explicitly based on
696 Graf (1978), p. 112. on rawwfa as a central place of the Thamd see issac (1998a), p. 125
n. 10. see also sartre (1982a), p. 35.
697 CIS V 2314; IGRR iii 1113; 1125; PUAES iii a 4; 11; 803; SEG XVi 1641; bowersock
(1994d). on the campaigns of lucius Verus see e.g. angeli bertinelli (1976), p. 2530; christ
(1992), p. 3346; eadie (1967), p. 141; Gebhardt (2002), p. 1269, p. 1527 and p. 1623;
Goodman (1997), p. 73; millar (1993), p. 1113; ross (2001), p. 379; sartre (2005), p. 146
8; Ziegler (1964), p. 1134.
698 see n. 194.
699 bowersock (1986), p. 112; Graf (1997b), p. 17982; Kennedy (2004), p. 41; millar (1993),
p. 140; shahd (1984b), p. 29. on the discussion see bawden e.a. (1980), p. 73.
700 e.g. dabrowa (1991), p. 3656; dussaud (1907), p. 5; Graf (1978), p. 45; nicolle (1996),
p. 92; rothstein (1899), p. 127; sartre (1991), p. 69; shahd (1984b) p. 31 and p. 33; Wells
(1991), p. 480.
701 on client tribes in general see Wagner (1985), p. 28. on the Aw as probably being an impor-
tant client tribe see sartre (1991), p. 333 (= sartre (1997), p. 316); shahd (1984b), p. 31.
702 butcher (2003), p. 64.
317 The problem of nomadic allies in the roman near east
a Greek inscription from rma set up for a certain odaenathus, strategos of the
Aw and phylarch (ototgygoovti Aouiogvev xoi uogoovti).
703
accor-
ding to sartre, odaenathus had a double function being chief of the Aw and head
of a military contingent of this tribe.
704
since the inscription was set up by his wife
and his father, it might be assumed that the early death of odaenathus could have
been in the context of his military duties, but this of course is by no means certain.
Unfortunately the date of the inscription is not clear, and thus it cannot be stated
conclusively whether odaenathus and his contingent were in nabataean, roman or
other service. its exact function, however, is not clear. There is no evidence that the
Aw or other ethnic groups collected tribute, etc. furthermore, in the fourth sec-
tion i have sketched probable revolts of the Aw against rome. it is not clear
whether the tribe or its leader(s) changed their attitude towards rome because of
political circumstances or whether only a segment of the group revolted. Unfortu-
nately the importance of the Aw as allies is not clear, since the evidence just
stems from members of this group themselves. below i shall deal in some more
detail with the problems of the terms phylarch and strategos, which will show that
it might also be a matter of debate whether the term strategos necessarily suggests
a military function.
however, there is quite a lot of evidence on arab troops in the roman army. in
the irst instance there are dromedary units. a number of latin and Greek inscripti-
ons attest the presence of dromedarii in syria, as there are for example the Ala I
Ulpia Dromedariorum and the Ala Valeria Dromedariorum.
705
an inscription from
bostra refers to equites singulares exercitus Arabici, item dromedarii.
706
Unfortu-
nately these texts do not tell us where the dromedarii were recruited. dabrowa
suggests, that many dromedar riders might have been recruited among arab tribes
under roman rule.
707
bowersock suggests that the personnel of the dromedary
units mainly were nabataeans.
708
indeed there are some nabataean inscriptions,
which testify to members of a nabataean guard in the hejaz. however, these texts
are probably from the pre-annexation period.
709
Graf suggests that the dromedary
units were installed after the fall of the nabataean kingdom and recruited mainly
from nabataeans.
710
furthermore, there are also some palmyrene forces of drome-
darii.
711
Therefore, the recruitment of nomads in the dromedarii seems not to have
been a dominant feature. however, this does not mean that there were no nomads in
these units. Two inscriptions from namra refer to a dromedarios, one to a Thaim,
703 OGIS 617 (= sartre (1982a), p. 124 nr. 5; moors (1992), p. 316 nr. 2d).
704 sartre (1982a), p. 124. on the problem of the inscription see millar (1993), p. 42930.
705 Ala I Ulpia: CIL XVi 106. Ala Valeria: ILS 2541 (= CIL iii 123 = CIL iii 14160
1
). see brown
(1941), p. 1489. on other dromedarii see Graf (1997b), nrs. 2; 5; 6.
706 CIL iii 93 (= PUAES iii a 528).
707 dabrowa (1991), p. 365. on the problem of recruitment see pollard (2000), p. 11142.
708 bowersock (1983), p. 157.
709 Graf (1997b), nr. 21. see Graf (1997b), p. 1756. a veteran of the army of an arethas is also
known from PUAES iii a 211.
710 Graf (1997c), p. 269.
711 CIL iii 123 (= CIL iii 14160,1); CIL iii 93; CIL XVi 106. see dabrowa (1991), p. 364.
318 Ulf scharrer
son of sidmos, the other to a Gaddos.
712
in the fourth section i have suggested that
namra probably was at least an ecomonic centre for nomadic groups. it is there-
fore possible that both were members of nomadic groups dwelling in the region,
but it is also possible that they rather lived a sedentary life. it thus seems inappro-
priate to regard the dromedarii as units mainly drawn from nomadic allies. This
may be supported by the statement in the third section, that at least writers of the
safaitic inscriptions rather used the horse for ighting.
The other important group of auxiliaries to be considered here are bowmen, the
sagittarii. shahd states that most often the desert warrior, i.e. the eques sagittarius,
was used for patrolling and controlling the frontier.
713
of course this seems to be
quite probable, as bowmen are quite often depicted in the safaitic grafiti. Quite
often eastern cohortes sagittarium are attested in latin inscriptions without any
details on recruitment.
714
at bingen near the rhine the Cohors I Sagittarium was
stationed. among its soldiers two syrians are found, one crete and one sidonian,
715

and thus this unit was ethnically mixed. other inscriptions attest to the ethnic ori-
gins of some units, and here are to be found mainly syrians, furthermore ascaloni-
tes and apamaeans.
716
so far, among the sagitarii there is no recruitment explicitly
among nomadic groups known, although this remains possible.
of some importance are a number of Greek inscriptions which attest the ser-
vice of natives in the army, as may be judged by the semitic names of their writers
or their relatives.
717
among these is also at least one native of the Legio III Cyre-
naica, a certain sulain, whose son arabianus is said to have been a member of the
governors staff (oixioiou tg gyrovio).
718
furthermore, a certain ai
(Ero), son of cassianus, who was magister, is recorded to have erected a gateway
712 Thaim: IGRR iii 1259; Gaddos: PUAES iii a 747.
713 shahd (1984b), p. 33.
714 e.g. the Cohors I Sagittarium miliaria Gordiana in dacia (CIL iii 6279; 8018); the Cohors I
Aelia Caes(ariensis) miliaria sagittariorum equitata from Klosterneuburg (roman cannabi-
aca) (CIL iii 564546; XVi 76; 96; 97); the Cohors I Sagittarum in syria (AE 1995, 1569); the
Cohors III sagittarium at apamea and later in latium (CIL iii 335; CIL XiV 3955); the Cohors
I Ulpia sagittarium (equitata), which took part in the parthian war of ad 1616 (CIL iii 600,
with brown (1941), p. 230. on these units see brown (1941), p. 22834.
715 CIL Xiii 751215; 11962a. syrians: CIL Xiii 7512; 11962a. crete: CIL Xiii 7513. sidonian:
CIL Xiii 7514.
716 e.g. the Numerus Syrorum sagittariorum in dacia (CIL iii 8032; 12601a); the Cohors I
Syrorum sagittariorum miliaria in mauretania (CIL Viii 21038; XVi 73); the Ala II Syrorum
sagittariorum miliaria civium Romanorum (CIL XVi 73); the Cohors I miliaria nova Seve-
riana Surorum sagittarium (CIL iii 363839; 10581; 15170); the I Ascalonit(arum) sag(ittaria)
in syria (CIL XVi 106.); the Cohors I Apamenorum Sagittariorum equitata (BGU 462,56;
ILS 2724). furthermore see the numeri Syrorum in moesia (CIL iii 7493) and in mauretania
(CIL Viii 9381; 9962; 9964; 10469). on syrians in the roman army and these units see brown
(1941), p. 24455.
717 IGRR iii 1304; 1307; 132931; 1337; PUAES iii a 284; 785
1
; peek (1955), nr. 375 (= merkel-
bach and stauber (2002), nr. 22/39/01. on this inscription see above, section 3). Legio III Cy-
renaica: PUAES iii a 349; 795
6
; 797
4
.
718 PUAES iii a 795
5
.
319 The problem of nomadic allies in the roman near east
and a upper hall at Umm al-Jeml.
719
These inscriptions, however, do not allow the
unequivocal conclusion that their writers had nomadic origins. There are some
more inscriptions referring to soldiers bearing roman names who were married to
women with semitic names, two of which were fround at namra.
720
While nothing
can be said about the ethnic background of the soldiers, it may be possible that their
wives stem from the nomadic surroundings of namra.
more important in the present context are some inscriptions mentioning troops
recruited from certain tribes. Thus, an inscription from ad 208 records the building
of a monument for Guthth(l)a(t), commander of the native troops enrolled among
the Mothani.
721
furthermore the tribe of the Mozaidenoi set up a statue for their
patron (aotevo) aurelius antoninus sabinus as a mark of gratitude.
722
it re-
mains unclear whether this patron is to be identiied with aurelius sabinus, a vete-
ran of the tribe of the Aschenoi (ug Aogvev)
723
. if this is the case, this would
reveal intertribal contacts which extended to the army. at least the reference to a
patron of a tribe shows connections between tribes and members of the roman
army. however, it is unfortunately not clear whether all these tribes are to be regar-
ded as nomadic, since there are no equivalents known from indigenous inscripti-
ons.
above all, the inscriptions referred to here apparently reveal that indigenous
members of the roman army used roman names, at least for their sons. Unfortuna-
tely there is not much known about the change of names or double names, but for-
tunately there is a stele set up for a Galba, who [is] also called Golanes [= Gauln],
son of abdalgos.
724
besides inscriptions from military contexts, there are also
some roman names to be found in villages.
725
problematic in this context is an in-
scription mentioning people with roman names, probably from the clan of the
Amtare (tev Atog).
726
in this case it cannot be said either whether there is a
nomadic background. With respect to the inscriptions mentioned here, a military
context cannot be excluded, but that is not certain. The same can be said about the
very few roman names to be found in safaitic inscriptions.
727
one inscription, for
example, mentions a certain harp b. claudius b. Kaammih (rb bn qlds bn
kmh).
728
at least the bearing of roman names seems to show some fashion, maybe
connected with loyalty to rome.
against the background of what has been said in the third section it is also a
matter of debate to what extent arab contingents of the roman and parthian ar-
mies, as referred to frequently by Greek and roman authors, were recruited from
719 PUAES iii a 244.
720 PUAES iii a 704. inscriptions from namra: PUAES iii a 7556.
721 PUAES iii a 223: Iou00o aoiaooitou yrvtiiev rv Mo0ovoi ovororvev.
722 PUAES iii a 664. on the inscription see millar (1993), p. 429.
723 PUAES iii a 760.
724 PUAES iii a 56. on the problem of double names with respect to seleucid babylonia and pa-
lestine see scharrer (2006), p. 3524, with further references.
725 e.g. PUAES iii a 765
11
; 787.
726 PUAES iii a 758.
727 on these see e.g. King (1990), p. 58.
728 Wh 837.
320 Ulf scharrer
nomads.
729
among arab auxiliaries we ind troops supplied by the emesene ruler
iamblichus to caesar in the alexandrian war,
730
and an arab, probably a nabataean
troop in the roman civil war.
731
probably at least some of these troops were recrui-
ted from the nomadic subjects of these regional powers, similarily to the arab ca-
valry of hatra and the auxiliaries of Tigranes, which at least partly were so.
732
fi-
nally, Josephus reports that after the death of herod the Great Varus, in order to
relief sabinus, marched to ptolemais with auxiliary troops furnished by kings (o-
oiriev) and dynasts (ouvootev).
733
Unfortunately it cannot be said who these
kings and dynasts exactly were. probably there were some nomadic troops among
these auxiliaries, but this is not certain.
apparently the problem of identifying arab warriors and distinguishing them
from others sometimes existed already in antiquity. Thus herodian, describing the
parthian troops of artabanus in ad 2178, presents them as ighting with long
lances from the backs of their camels.
734
it cannot be said to what extent his refe-
rences to heavy armed camel and horse riders in the parthian army and his general
statement on eastern barbarians riding on horse and camel refer at all to arabian
troops.
735
furthermore, cicero reports that parthians who were recorded to have
invaded syria were indeed arabs being equipped like parthians.
736
This makes it
quite dificult to draw conclusions from armament on ethnic origins.
only seldom do we ind unambiguous references on nomads serving in the
roman and parthian armies, although military service has been regarded in scholar-
ship as a supplementary economic strategy of nomadic groups.
737
one reference is
found in the safaitic inscriptions. a grafito states that a certain arj was in com-
mand (yrb) during a siege, and the grafito is accompanied by a drawing of a woo-
den covering, which is protecting bowmen.
738
Thus, although nomadic groups
themselves generally are not considered as effective in the storming of cities,
739

here we have even a commander during such a siege, very probably with nomadic
roots. however, unfortunately the context of the siege is not clear, and it cannot
even be said in what army arj fought, be it (likely) a roman, a parthian or ano-
ther army. Unclear is an inscription referring to a horseman in the unit of the l
729 e.g. parthians and arabs move in direction of antiochia: cic. Fam. 15.4.7. arabs in the army
of pompey at the battle at pharsalus: luc. 7.514, with rets (2003), p. 3967. arab auxiliaries
in the parthian army: Tac. Ann. 6.44. arabs in the army of Titus: Joseph. BJ 5.290; Tac. Hist.
5.1. see rets (2003), p. 395.
730 Joseph. Ant. 14.129.
731 Joseph. Ant. 11.277.
732 Tigranes: cass. dio 36.37.7a. hatra: Tac. Ann. 12.15.1; cass. dio 76.11.2.
733 Joseph. BJ 2.67.
734 hdn. 4.14.3.
735 hdn. 4.15.1; 4.15.3.
736 cic. Fam. 3.8.10. see isaac (1992), p. 237.
737 paul (2003), p. 38; p. 556.
738 Wh 2502a.
739 e.g. proc. bell. 2.19.12, with Graf (1989a), p. 389. on dromedarii being not useful in battles
see dabrowa (1991), p. 3656. see generally on tribal warfare schmitt (2005).
321 The problem of nomadic allies in the roman near east
Amrat.
740
This is probably a reference to a tribe serving as some sort of auxiliaries
or police for rome. That there were some sort of patrols is revealed by a further
safaitic inscription referring to a patrol of nm.
741
Unfortunately it is not clear,
whether this patrol surveyed some territory on behalf of rome or of the nabataean
kingdom. furthermore it cannot be stated whether its members were dromedarii or
not. These reference do not allow the conclusion that the mentionings of troops (j)
in safaitic inscriptions
742
imply nomadic groups at roman, nabataean or persian
service, since according to one grafito a troop followed an enemy in order to recu-
perate camels,
743
thus referring to tribal conlict.
in the context of a discussion of nomadic allies of fundamental importance are
a couple of Greek inscriptions found in the hauran area which mention strategoi
and maybe an ethnarchos of nomads. i have already hinted at the problem of the
interpretation of an inscription of a certain odaenathus of the Aw being phylarch
and strategos.
744
interpretations of these designations vary to a large extent. accor-
ding to the irst interpretation the titles refer to ofices being held by nomads them-
selves and being positions within a certain group.
745
according to the second inter-
pretation these titles held by nomads imply the nomads acknowledgement by the
roman authorities.
746
slightly different from the latter is the third interpretation,
according to which the titles were given to some nomad chiefs by rome.
747
finally,
a pure local context of the inscriptions has been stressed.
748
it is therefore appropri-
ate to discuss these texts in some more detail.
The irst inscription to be dealt with here comes from arb, a village in the
hauran, which records the erection of a statue for an unknown legatus Augusti pro-
praetor and antistrategos by those of an ethnos of nomads because of [his] up-
rightness (oi oao r0vou voooev oyvrio oiv).
749
according to sartre the no-
mads were soldiers recruited among nomadic people.
750
macdonald argues that the
nomadic groups of the hauran never called themselves nomads, and that this de-
signation appears to be a rather vague identiication of the donors of the monument.
referring to possible meanings of the word r0vo other than ethnographic, he assu-
mes that in this context the term is meant to designate a military unit, as he suggests
with respect to ethnos of the Thamd of the rawwfa inscription.
751
Whereas this
hypothesis appears to be quite convincing, other contexts than military ones are
still possible. at least this inscription reveals the honouring of a high roman of-
740 Quoted by macdonald (2005), p. 120.
741 Quoted by macdonald (2005), p. 124.
742 e.g. Wh 23.
743 PUAES iV c 146.
744 see n. 703.
745 bowersock (1983), p. 131; Kuhnen (1991), p. 330.
746 isaac (1992), p. 238; sartre (1991), p. 333 (= sartre (1997), p. 316), followed by hartmann
(2001), p. 83 n. 76; shahd (1984b), p. 31.
747 chapot (1907), p. 2930; dussaud (1902), p. 413.
748 millar (1993), p. 430.
749 IGRR iii 1254 (= PAAES iii 378; sartre (1982a), p. 124 nr. 4; moors (1992), p. 317).
750 sartre (1982a), p. 124.
751 macdonald (1995), p. 989. see above, n. 194.
322 Ulf scharrer
icial by nomadic groups and his dealing with them which was regarded as being
fair.
from a clearly military context stems a broken inscription from the djebel el-
druz. it was set up in the reign of agrippa ii by a certain charetos, eparch of the
Cohors Augusta and strategos of nomads (rao[o ] oarig au [youotg xoi
ototgy]o voooev).
752
if the reading of the inscription is correct, charetos had
two functions, one being the oficer of a cohors, the other being strategos of no-
mads. according to sartre, charetos was the military leader of nomadic troops in
the service of agrippa ii.
753
While this may be possible, the ofice of the strategos
nomadon is debated. a strategos with respect to nomads is furthermore mentioned
in two other inscriptions. The irst one is a fragmentary text mentioning a strategos
of camps / squadrons of nomads (ototgyo aoroev voooev).
754
The se-
cond one, dating from the reign of hadrian, records the erection of a monument for
an ethnarch [and] strategos of nomads (r0voou ototgyou voooev).
755
The ofice of a strategos of nomads has been interpreted in different ways. it
has been regarded as an ofice in charge of the command of nomadic units
756
or
with the function of controlling transhuman movements and defence against noma-
dic incursions.
757
furthermore it has been debated whether the holders of this ofice
were nomad chiefs themselves or rather other people.
758
indeed, the ofice of a
strategos is a complex problem, as many strategoi are found in different contexts,
which are themselves often discussed. Thus at palmyra the scope of the ofice has
been regarded as mainly a military function in- and outside the citys territory.
759

on the other side the palmyrene strategos is said to have been the supreme magi-
strate of the city and not necessarily a military commander.
760
furthermore it seems
not always to be clear whether palmyrene strategoi were appointed by rome.
761

differing views exist also with respect to the role of the nabataean strategos
(strtg):
762
on the one hand the ofice is regarded as a purely military function,
mainly associated with the protection of caravans,
763
on the other hand it is said to
have been rather a civil and administrative function.
764
finally it has been sugge-
752 OGIS 421 (= IGRR iii 1136; sartre (1982a), p. 123 nr. 1; moors (1992), p. 316 nr. 3a).
753 sartre (1982a), p. 123.
754 PUAES iii a 751 (= sartre (1982a), p. 124 nr. 3; moors (1992), p. 316 nr. 3b).
755 OGIS 616 (= IGRR iii 1247; sartre (1982a), p. 123 nr. 2; moors (1992), p. 315 nr. 1a).
756 sartre (1982a), p. 124; asad and Yon (2001), p. 63, assuming a nomadic desert police; hoy-
land (2001), p. 99, with respect to PUAES iii a 751.
757 Graf (1978), p. 16; Villeneuve (1985), p. 117.
758 nomadic chiefs: Grouchevoy (1995), p. 1146; sartre (1982a), p. 124. local people: millar
(1993), p. 430. people from outside: macdonald (1993), p. 3756.
759 ingholt (1976), p. 1247; matthews (1984), p. 169; moors (1992), 3269; sommer (2005),
p. 173. see e.g. PAT 1063 (= Kaizer (2002), p. 467).
760 dirven (1999), p. 262.
761 dirven (1999), p. 2623.
762 on the problem see Graf (1997c), p. 2759.
763 hammond (1973), p. 109; maraqten (1996), p. 229.
764 briquel-chatonnet (1995b), p. 136.
323 The problem of nomadic allies in the roman near east
sted that the nabataean strategos could also be a tribal chief.
765
in addition there is
the problem that some inscriptions also testify to the strategos as a village ofical.
766

of course, the ofice of a strategos could be held by indigenous people, as an in-
scription in honour of marcus aurelius and lucius Verus shows.
767
still, the scope
of the strategos of nomads is far from unambiguous.
The interpretation of the term determines and depends on the reading of the
other terms. Thus it is a matter of debate whether the term aorog voooev is
to be translated as squadron of nomads or just as nomadic camp.
768
it may be a
topic of further research to what extent the term strategos of camps/squadrons of
nomads might have been used in analogy to the nabataean chief of camp (rb
mrt) as is known from an inscription from ad 44,
769
and to the latin praefectus
castrorum.
of considerable importance in this context is the implication of the term eth-
narch, which has also been interpreted differently. The adherents of the irst view
suggest that the title, as well as that of phylarch, was given to tribal chiefs by the
roman authorities, or at least that they were acknowledged as such, in order for the
chief and his tribe to control other tribes.
770
according to the second, similar, inter-
pretation, the strategos of nomads and the ethnarchos were roman titles taken over
by nomad chiefs, and ought to be regarded as primer of the late antique phylar-
chat.
771
in contrast to this, it has been stated that in the safaitic inscriptions there
are no equivalents, that ethnarchs and phylarchs are therefore not necessarily no-
madic chiefs, and furthermore that it is not cogent that they were appointed by the
roman authorities.
772
in fact, there are other references to ethnarchs. Two Greek inscriptions were set
up by an ethnarch. The irst one is by a certain annelos, son of samethos.
773
against
the background of his interpretation of the function of the ethnarch, sartre suggests
that annelos might have been the chief of a safaitic tribe.
774
The other inscription
refers to a certain philippos, whose patrimonial name unfortunately is illegible.
775

sartre suggests that the name of the father might have been semitic. he further-
more takes into account, that at least some of the ethnarchs may have been naba-
taeans.
776
again the scope of the ofice is not clear. if an ethnos was, as macdonald
suggests, at least in some cases a military unit, the ethnarch may have had military
765 moors (1992), p. 3216.
766 e.g. PAAES iii 150. see moors (1992), p. 3906; Villeneuve (1985), p. 82. see also sartre
(1991), p. 2412 (= sartre (1997), p. 228).
767 PUAES iii 803.
768 squadron of nomads: macdonald (1995), p. 99100. nomadic camp: sartre (1982a), p. 124.
769 Quoted in Graf (1997c), p. 2789.
770 dussaud (1902), p. 413; moors (1992), p. 3189; sartre (1982a), p. 124; sartre (1990), p. 47
8. sartre (2005), p. 235 and p. 359.
771 Kuhnen (1991), p. 330.
772 madonald (1993), p. 3702.
773 sartre (1982a), p. 125 nr. 6.
774 sartre (1982a), p. 125.
775 PUAES iii a 675 (= sartre (1982a), p. 125 nr. 7).
776 sartre (1982a), p. 1256.
324 Ulf scharrer
functions. on the other hand, ethnarchs are also known from very different con-
texts, for example as an oficial title of the herodians.
777
Thus an ethnarch may
well have been an indigenous chief, though the ofice does not necessarily imply a
nomadic origin. however, there are no references to whether he was appointed by
the roman authorities or whether he was just acknowledged by them.
Unfortunately the term phylarch is not found in the inscriptions referring di-
rectly to nomadic groups. Whereas since the fourth century ad the phylarchat ap-
parently became institutionalized as an expression of some sort of client relation,
778

there are no clear references with respect to earlier roman rule, and accordingly
there are basically two different views on the ofice: on the one hand phylarch has
been interpreted as a roman administrative term for tribal leaders with relation to
rome.
779
on the other hand it has been argued that before the fourth century ad a
phylarch was the head of a tribal group.
780
in this context Grouchevoy stresses that
the term phylarch, at least as used by strabo, could design the head of either a no-
madic or a sedentary group.
781
it seems indeed that the term phylarch did not have a clearly deined meaning.
already cicero used the term as a loan-word calling the emesene ruler iamblichus
a phylarchus Arabum, and accordingly strabo designates sampsigeramus and his
son iamblichus each as phylarch of the Emesenoi.
782
as stated in the second part of
this section, it is not that clear whether the term Emesenoi could also imply inhabi-
tants of emesa who did originally not belong to the ethnic group of sampsigeramus
and his son. Whether in the context of rival claims to the seleucid throne sampsi-
geramus was conirmed as phylarch by the ptolemies
783
is not clear from the evi-
dence. in this respect the hatrene lords, apparently translated as phylarchs in Greek
texts, are also problematic: first, like with the emesene rulers, it is obvious that
they did not rule just one tribe. furthermore, before the installation of kingship at
hatra it cannot be stated conclusively whether the lords were installed by the par-
thian authorities or at least acknowledged by them, and the same can be said about
the edessene lords. besides these rulers of larger and at least partly sedentary
groups, strabo mentions phylarchs at the euphrates who attached themselves to the
stronger party, at least some of whom were nomadic.
784
Thus a phylarch could ap-
parently also rule over smaller ethnic groups. furthermore, it does not seem to have
been the case that they were installed, dismissed or acknowledged by the roman or
parthian authorities.
777 hammond (1973), p. 37.
778 Grouchevoy (1995), p. 11931; isaac (1992), p. 2409; hoyland (2001), p. 81; isaac (1998f),
p. 4145; shahd (1984a), p. 1924, p. 51418 and p. 521; shahd (1989), p. 5012; Whittaker
(1994), p. 136. see also generally isaac (1992), p. 2409; peters (1999a), p. xviii-xxv. see also
schmitt (2005), p. 431.
779 moors (1992), p. 3189.
780 mayerson (1994c), p. 3434.
781 Grouchevoy (1995), p. 1102.
782 cic. Fam. 15.1.2; strabo 16.2.10.
783 so ball (2000), p. 34.
784 strabo 16.1.278. on phylarchs in strabo see also Grouchevoy (1995), p. 1101.
325 The problem of nomadic allies in the roman near east
Unfortunately there is not much epigraphic evidence for the period covered
here. The context of one inscription mentioning a phylarch is not clear,
785
and we
are not told whether there are connections with nomadic groups or not. however, i
have already discussed the inscription of odaenathus, who was strategos of the
Aw and phylarch.
786
While this odaenathus was probably the head of a military
contingent, it cannot be drawn from the inscription whether he was installed or ack-
nowledged as phylarch, or whether he was rather a tribal leader who was chosen by
his fellow tribesmen or who had inherited the position. many details about nomadic
alliances with rome and parthia up to the third century remain therefore unclear.
There is evidence that nomadic groups fought in the armies and policed the steppe
regions, but there is no clear evidence on the issue of whether something like cli-
ent tribes existed.
changes in the third century ad
expecially in the third section, but also in some other contexts throughout this es-
say, i have referred to the problem of bedouinization. i have quoted the most promi-
nent view, namely that the fall of local powers and especially that of palmyra,
which led to a ceasing of control over nomadic groups, has been regarded as the
main reason behind a process of bedouinization.
787
besides bedouinization, there
are different views on the consequences of the end of client states. firstly, it has
been stated that the lack of control, following the defeat of palmyra as the last re-
gional power, brought about a shift from nomadic involvement in trade to brigand-
age.
788
however, so far there appears to be no reliable evidence for an increase of
nomadic brigandage. as stated in the fourth section, also the strenghtening of the
frontier fortiications does not necessarily imply more raids than before: they pre-
sumably ought to be understood in the context of the rise of the sasanid empire
instead. furthermore, isaac argues convincingly that we do not know anything
about the effect of the decline of trade on nomadic groups.
789
This remark is the
more important since, as was discussed in the fourth section, there is not much evi-
dence on nomadic involvement in trade. finally, there is an important, but somehow
obscure remark in the Historia Augusta, according to which the later usurper in
egypt, firmus, as a native of a city called seleucia had relations with the sara-
cens.
790
Unfortunately we do not know exactly which of the many seleucias is
being referred to. however, since firmus is known to have been a trader, there
might have been some commercial relations.
785 IGRR iii 1095.
786 see n. 703.
787 see n. 206.
788 funke (1996), p. 22830; parker (1986b), p. 641. see also bernbeck (1996), p. 406; hogdon
(1989), p. 178.
789 isaac (1992), p. 71.
790 sha Quad. 3.3. on firmus see Teixidor (2005), p. 224.
326 Ulf scharrer
secondly, sommer assumes centrifugal energies of nomads set free after the
breakdown of the dimorphic centres palmyra and hatra.
791
This may be supported
by archaeological evidence, as according to bernbeck there was a decline in rural
settlements in the surroundings of hatra, palmyra and dura-europus after the third
century ad.
792
it is not clear, however, whether there existed or exists something
like centrifugal energy of nomadic groups: as i have argued in the irst section,
current scholarship stresses a dependence of nomadic groups on the sedentary
world. it is furthermore not necessary that nomadic groups should move elsewhere
after their centre was destroyed. finally, as discussed in the second part of this sec-
tion, the interpretation at least of palmyra as a dimorphic society is rather proble-
matic.
Thirdly, it has been argued that the defeat of palmyra eventually led to the esta-
blishment of the Tankh rule in the steppe regions between the sasanid and roman
empires, that they as substitute for the more sedentary client kingdoms became an
important political factor in the region, and that the Tankh thus became important
as allies for rome and persia.
793
indeed, as discussed in the second part of this sec-
tion, there seems to have been a conlict between the growing powers of the Tankh
and palmyra in the second half of the third century ad.
794
in this context the idea
of a vacuum after the fall of the client kingdoms has been put forward a vacuum
which was eventually illed by the Tankh rulers.
795
i will not discuss the problem of bedouinization here, but try to bring these
aspects together. Graf simply states that Zenobias rise and fall disrupted traditio-
nal social life of north arabia.
796
so, to broaden this statement beyond palmyra:
what are the implications of this disruption which followed the end of edessa, ha-
tra and palmyra, but also of other regional powers? in the second part of this sec-
tion i have sketched a picture of nomadic afiliations of different degrees to client
kingdoms, be it as part of a dimorphic society as at hatra, be it as inhabitants of a
kingdoms territory with political relation to the centre as at emesa and edessa, or
be it as allies to a regional power as at palmyra. all these socio-political bonds
were cut from the middle of the third century ad. at the same time the near east
faced the immigration and rise of the Tankh which expanded over the whole
steppe region. The exact process cannot be traced in detail, but it seems to have
been a mixture of conquest (as described in the irst part of this section) and ethno-
genesis (as sketched in the third section). This is well illustrated by abars ac-
count of the conlict between Zenobia and the Tankh king Gadhima: as we have
seen in the second part of this section, Zenobias father, the amalekite chief amr
b. arib, was according to abar the ruler of the Jazirah before he was slain by
791 sommer (2003a), p. 43.
792 bernbeck (1996), p. 403.
793 ball (2000), p. 97; bowersock (1975), p. 521; de Vries (1986), p. 237; sartre (2005), p. 358;
shahd (1984b), p. 312. see isaac (1992), p. 71.
794 see also with respect to hatra: Wiesehfer (1982), p. 446.
795 bowersock (1975), p. 521; dussaud (1955), p. 634; hartmann (2001), p. 100 and p. 351;
sartre (2005), p. 360; Zwettler (1993), p. 13.
796 Graf (1989b), p. 159. see also shahd (1984a), p. 3901.
327 The problem of nomadic allies in the roman near east
Gadhima.
797
Unfortunately not much is known about this king. however, it seems
that he illed the political vacuum in the Jazirah caused by the fall of hatra, and
there is no evidence that he gained his power by force. his murder integrated the
Jazirah into the realm of Gadhima. i therefore suggest that the socio-political ties
which were cut after the fall of the client kingdoms were substituted by the Tankh
network of alliances and dependencies.
798
it is this process which might be called
bedouinization: the establishment of new socio-political relationships which in
contrast to the earlier period were not between nomads and sedentary powers, but
between different nomadic groups that turns nomadic groups into an important
political factor in the near east.
nomadic allies at the beginning of the fourth century ad
The last part of this section shall be devoted to the new alliances between the noma-
dic groups under the hegemony of the Tankh and the roman and sasanid empires.
The beginnings are certainly marked by the rise of the sasanids and their policy
towards the new ethnic groups. With respect to ardashrs conquest of the iraq,
abar remarks that a large part of the Tankh did not want to stay in his kingdom
and become his subjects, and so some groups moved westwards to syria.
799
Thus,
at the beginning of sasanid rule at least some Tankh groups were not inclined to
become vassals of the sasanid king. it is not clear, however, what sort of relations
these groups had with the roman authorities in syria.
800
apparently there was some change in the middle of the third century ad, which
is to be found in the rise of Gadhima and the policy of shapur i towards the Tankh
and other groups. in the irst part of this section i have sketched how that Gadhima
was the irst to establish the supreme rule of his family, soon to be known as the
lakhmids, over numerous ethnic subgroups. in this context abar remarks that he
did not recognise the regional princes of iraq and that this continued up to the rise
of shapur.
801
indeed, abar continues saying that, after Gadhima, his nephew
amr b. ad also became ruler in the surroundings of the iraq and the hejaz,
802

and this implies that Gadhima had already ruled over these regions. in the present
context abars statement that they, i.e. Gadhima and amr, became rulers be-
cause the persian kings employed them for this purpose, relying on them to keep
the adjacent arabian kings under control is important.
803
shapur seems to have
been the irst sasanid ruler who tried to establish some sort of client relation be-
797 abar, Tarkh (History) i 756. see equini schneider (1993), p. 143; hartmann (2001), p. 335
and p. 347; Tubach (1986), p. 2367.
798 on the network see potts (1990), p. 236.
799 abar, Tarkh (History) i 821. see also above, section 3.
800 according to altheim and stiehl (196469) ii, p. 2512, in this context Gadhima went over to
rome, but this is not certain.
801 abar, Tarkh (History) i 768.
802 abar, Tarkh (History) i 769.
803 abar, Tarkh (History) i 769.
328 Ulf scharrer
tween the Tankh and his own authority by supporting the rise of one of their lead-
ing families. apparently the Tankh by this time seem to have become a conside-
rable military force which was not only useful for small scale raids: according to
abar, Gadhima even raided standing armies.
804
however, as stated in the second
part of this section, the role of the Tankh as a factor in the inal defeat of palmyra
remains a matter of debate. it is in any case clear that shapur i deliberately em-
ployed nomad groups at the western frontiers of his empire and beyond from a
passage in abar, according to which the Yemenite tribe of the rabah sent some
of its members to the iraq who, after a correspondence with shapur, were settled at
al-rah with the purpose of securing the frontier.
805
as stated in the irst part of
this section, at the same time al-rah became the centre of lakhmid rule.
around the same time there are some hints of saracens serving in the roman
army: so Valerian seens to have assigned saracen auxiliaries to aurelian and pro-
bus.
806
in the second part of this section i have sketched the problems concerning
the change of sides of saracen groups from Zenobia to aurelian and the relevant
background. These seem to be the only important references to saracens ighting
on roman sides up to ca ad 300. more often there seem to have been roman cam-
paigns against saracen groups, as sketched in the fourth section. Unfortunately the
evidence on saracen-roman relations is quite sparse for this period, and it seems
that the sasanids were the irst to establish some sort of client relationships: in the
Notitia Dignitatum, compiled in the fourth century ad, some saracen units are
mentioned as roman auxiliaries.
807
it has sometimes been argued that some layers
go back to the pre-byzantine period,
808
but this is not certain.
it is furthermore debatable whether all the saracens who are mentioned as au-
xiliaries of the roman army did indeed belong to the Tankh confederation. as
stated in the second part of this section, the Tankh seem to have supported aure-
lian against palmyra. it is striking that the persian authorities do not seem to have
tried to restrain the Tankh from doing so, although they had supported Zenobia by
sending troops.
809
on the other hand, Zenobias husband had heavily defeated
shapur in ca ad 260 on his return from his syrian campaign,
810
and here sasanid
politics appears to be rather obscure. We are faced again with the problem of chan-
ging sides, which i have described with respect to earlier small groups in the third
part of this section, and which will be discussed in more detail below in the context
of imraalqays. in the present context not too many conclusions can be drawn from
the burial of Gadhimas tutor at Umm al-Jeml, i.e. in the roman sphere.
811
The
804 abar, Tarkh (History) i 750.
805 abar, Tarkh (History) i 7701.
806 sha Aurel. 11.3; sha Prob. 4.1.
807 Not. Dign. [or.] 32.278. on the Notitia Dignitatum see shahd (1989), p. 46074. on saracens
in the Notitia see shahd (1984b), p. 2830 and p. 545. on fourth-century papyri see mayer-
son (1988), p. 734; mayerson (1994b), p. 3223.
808 rets (2003), p. 511; shahd (1984b), p. 27, p. 30 and p. 51.
809 sha Aurel. 27.4.
810 hartmann (2001), p. 12940.
811 see above, n. 482.
329 The problem of nomadic allies in the roman near east
epitaph just shows that he was present there, but it does not tell us why. maybe he
was on some diplomatic mission, but this is pure speculation. at least the inscrip-
tion does not enforce the conclusion that Umm al-Jeml was in Gadhimas sphere
of inluence, or that Gadhima changed to the roman side.
812
as stated in the second
part of this section, the Tankh or at least parts of them went over to aurelian under
Gadhimas successor, his nephew amr b. ad.
as we have seen, at this time amr b. ad was ruler in the iraq by the employ-
ment of the sasanid kings. This is clearly supported by a passage in the famous in-
scription from paikuli which was set up by the sasanid king narseh. The inscrip-
tion mentions a couple of dynasts and kings who stayed by our advice and
counsel.
813
among these is amru King of the lakhmids,
814
who is probably to
be identiied with amr b. ad. What is important is that there seems to have been
some sort of client relationship,
815
which is indicated by the fact that he placed
himself under the advice and counsel of the sasanid king. appararently many
groups were placed under his rule: according to abar, his son imraalqays succee-
ded him and became the governor over the frontier region of the arabs of rabah,
muar and the rest of the tribes of the deserts of iraq, the hejaz and the Jazirah.
816

it is not really clear whether amr b. ad had already ruled over all these groups.
in the present context it is interesting that the rabah, who, as i have said, settled
at al-rah with the encouragment of shapur i, were placed under lakhmid rule.
With respect to amr b. ad, there is no evidence at all that he placed himself un-
der roman authority in the course of the war against palmyra, and his support for
rome against Zenobia apparently did not bring on any troubles from the sasanid
side.
With regard to amr b. ads son imraalqays position with respect to rome
and persia, quite divergent views are based on different interpretations of the ac-
count by abar and of his epitaph from namra. according to abar, imraalqays
succeeded his father and became governor over a wide stretch of territory from the
iraq to the hejaz, and it is stated clearly that he was still a sasanid vassal under
bahrm ii and shapur ii.
817
in this context imraalqays epitaph from namra is
quite striking
818
the more so since there are divergent readings of the passages
which are relevant in the present context. i have briely discussed his campaigns
and his claim to rule over (all) arabs, but there are more considerable problems
with regard to his status, his rule and his alliances.
812 so stated by ball (2000), p. 85.
813 NPi 93: middle persian (blocks h5,056,05), line 46: [PWNW pndy W pdy]sy ZY LNE
YKOYMWNd. The parthian version is too fragmentary.
814 NPi 92: middle persian (blocks h16.031,04), line 44: m[rw] lhmdyn ML(KA) / parthian
(blocks g15,051,06), lines 4142: W mrw lhmyn MLKA W mrw pgrnn [MLKA (?)].
815 see hartmann (2001), p. 344, hoyland (2001), p. 235.
816 abar, Tarkh (History) i 8334.
817 abar, Tarkh (History) i 8334.
818 on the inscription see above, n. 172, n. 2347, n. 48891 and n. 4934.
330 Ulf scharrer
The irst problem to be dealt with here is the interpretation of line 4, which has
been read mainly in three different ways:
819
irst it has been suggested, that
imraalqays placed them [the tribes he conquered] as cavalry under the service of
the romans (wklh lfrs wlrwm).
820
according to the second view the king appoin-
ted them [his sons] as deputies to the persians and the romans (wklh lfrs
wlrwm).
821
finally, bellamy reads it such that the nobles of the subdued tribes as
viceroys became phylarchs of the romans (wklm frsw lrwm).
822
The main pro-
blem, therefore, is the appropriate meaning of frs in this context, be it in the plural
form cavalry, phylarchs, or be it persia (Fris).
823
according to the adherents of the irst and third interpretations, the inscription
shows that by the time of his death imraalqays was a vassal to rome, which is
presumably supported by his burial place in roman territory.
824
as client to rome
he was thus installed to rule the nomads of the syrian desert, which is said to be
implied by his title king of all Arab, given to him by the roman authorities.
825

accordingly he is assumed to have changed his allegiance from persian to roman
sides.
826
although he follows the irst reading, bowersock remarks that the inscrip-
tion does not say anything about a foedus of imraalqays with rome, and he is
therefore rather sceptical about the idea that imraalqays was some sort of a roman
client king.
827
he is also sceptical about the kings adoption of christianity, which
is stated by abar
828
and is often regarded as a reason for his change of alliance:
instead bowersock suggests that in this context there are two kings of the name
imraalqays, one in abar and one in the inscription.
829
Whereas this has to remain
a matter of debate, it seems quite improbable that imraalqays as king of all Arab
ought to be regarded as a roman client king who got this title by rome. first there
is no evidence at all that he was regarded as disloyal by the sasanid authorities, and
after him his son amr was appointed governor over the arabs by shapur ii.
830
fi-
nally, procopius states that in ca ad 530 Justinian considered the Ghassanid king
arethas worthy of rule over all saracens allied to byzantium, and the author re-
819 on the discussion see bellamy (1985), p. 413; rets (2003), p. 46970. see also moors
(1992), p. 30910.
820 dussaud (1902), p. 418; bowersock (1983), p. 13940.
821 bowersock (1975), p. 522; hartmann (2001), p. 3489 n. 304; sartre (1982a), p. 136.
822 bellamy (1985), p. 425 with the translation on p. 46, followed by isaac (1992), p. 239 n. 118.
823 on the problem see also shahd (1984a), p. 434.
824 bowersock (1983), p. 140; isaac (1992), p. 734; parker (1986b), p. 642; potts (1990), p. 239;
shahd (1984a), p. 46, p. 58 and p. 60.
825 dignas and Winter (2001), p. 199; bellamy (1985), p. 45 n. 7; dussaud (1955), p. 139; shahd
(1984a), p. 5267. according to shahd (1984a), p. 513, imraalqays ruled the territory of the
provincia Arabia.
826 altheim and stiehl (1957), p. 1423; altheim and stiehl (196469) ii, p. 258 and p. 31620;
altheim and stiehl (196469) iV, p. 2 and p. 280; hartmann (2001), p. 3479; nagel (1998),
p. 2; parker (1986b), p. 642; shahd (1984a), p. 324, p. 457 and p. 374, followed by moors
(1992), p. 310; Winter and dignas (2001), p. 199. see also isaac (1992), p. 240.
827 bowersock (1986), p. 114. see also peters (1978), p. 326.
828 abar, Tarkh (History) i 834.
829 bowersock (1986), p. 115.
830 abar, Tarkh (History) i 845.
331 The problem of nomadic allies in the roman near east
marks that this had never happened before in the roman sphere of inluence.
831
if
procopius remark is correct, it is a strong argument against the view that imraalqays
was installed as client king over all arabs by rome.
The view that imraalqays changed his allegiance is regarded as possible also
by butcher, who in addition considers that the king died while he was raiding ro-
man territories on behalf of the sasanid king.
832
in this case it would have been
quite improbable that imraalqays assigned troops to the roman authorities. how-
ever, those adhering to the view that the king handed over tribal warriors both to
persia and to rome come up with different explanations. according to the irst
view, imraalqays could have had treaties with both rome and persia.
833
it has also
been suggested that the king ruled over groups in both the sasanid and the roman
sphere of inluence.
834
finally, the idea has been put forward that imraalqays was
not a client of either empire, but that he was independent instead.
835
in this context the readings and interpretations of line 1 of the namra-
inscription appear to be controversial too. There are two main readings of the
line.
836
according to the irst one, imraalqays assumed the diadem (sd lty).
837

according to the other, the king says that his title of honour was master of asad
and maiy (lqbwhw w sdn wmyn).
838
Whereas the second reading does not
pose too many problems, the interpretations which follow the irst reading differ to
some extent. The main question is whether imraalqays got the diadem from rome
or from persia, since that would express his client relationship to the respective
power.
839
however, there is no evidence to support this view. if this reading is cor-
rect, it seems that imraalqays assumed the diadem himself, most probably in the
consequence of his victorious campaigns. by this he would actually follow the hel-
lenistic tradition, according to which royalty, expressed by the diadem, was based
on military success and victory.
840
i have already shown that rome does not seem
to have given supreme power over saracen allies to a single ruler before Justinian.
Thus, if imraalqays had assumed the diadem, it is most likely that he assumed it
himself, together with the title king of (all) Arab, both in consequence of his suc-
cessfull campaigns.
841
This may be supported by the kings statement in line 5 of
831 proc. bell. 1.17.47. see funke (1996), p. 233. on the Ghassanids see e.g. ball (2000), p. 101
5; balty (1989), p. 189; butcher (2003), p. 701.
832 butcher (2003), p. 65. see hoyland (2001), p. 79.
833 dussaud (1907), p. 378; peters (1999a), p. xii.
834 sartre (2005), p. 362.
835 bowersock (1975), p. 5212; Graf (1978), p. 16; sartre (1982a), p. 1389.
836 on the discussion see bellamy (1985), p. 356; potts (1990), p. 239; rets (2003), p. 46970.
837 dussaud (1902), p. 4145; bowersock (1975), p. 522. see shahd (1984a), p. 367.
838 bellamy (1985), p. 356 with the translation on p. 46, followed by isaac (1992), p. 239 n.118.
839 stated e.g. by dussaud (1902), p. 414.
840 on the hellenistic tradition see austin (1986); austin (1999), p. 1312; mehl (1999), p. 156,
p. 32 and p. 367; scharrer (2000), p. 1079, with further references. see also the references in
scharrer (1999), p. 96.
841 The view that imraalqays assumed the diadem himself is stated by bowersock (1975), p. 522.
The view that he assumed the title king of (all) Arab himself is suggested by Zwettler (1993),
p. 145.
332 Ulf scharrer
the namra-inscription, which states that no king equalled his achievements.
842

further speculation is possible. so it remains the question to what extent imraalqays,
by assuming the diadem, imitated constantine, who added the diadem to the royal
ornate in ca ad 3256.
843
if he did, it can be asked whether this was a sign of loy-
alty to the christian emperor. also the eastern tradition may be a subject of de-
bate. against the background of the lack of further evidence and especially of the
problematic reading of the namra-inscription it is obvious that answers to these
questions must remain speculative for the moment.
The main question in this context is therefore that about the nature of the rela-
tions of the Tankh rulers with the roman and especially the sasanid empire. as is
stated repeatedly above, it is obvious that there is some evidence that they were
installed, and regarded as some sort of client kings, at least by the persian kings.
This did not hinder them to ight on the roman sides, as was the case with aurelian
against palmyra, or as can be drawn from the namra-inscription to deliver
troops to rome or to place tribal leaders under roman rule. however, there is no
unambiguous evidence that rome at this time installed Tankh rulers as client
kings. furthermore, the idea of a change of alliance by imraalqays has turned out
not really convincingly.
some critical remarks ought to be made in this context about the position of
al-rah. it has most commonly been designated as capital or residence of the
lakhmid kings,
844
and this seems to be supported by the fact that according to
abar the centre of the lakhmids was indeed al-rah, a view followed by schol-
arship.
845
occasionally the concept of dimorphic societies has been applied to al-
rah.
846
all these views do imply that the Tankh rulers, the lakhmids, had their
seat at al-rah and from there ruled a more or less deined nomadic empire, a phe-
nomenon known from other historical periods and regions.
847
as stated repeatedly
above, the sources clearly deine al-rah as belonging to the sasanid realm. so
how are the movements of the kings, especially of Gadhima and imraalqays, to be
explained? following the concept of al-rah as capital or residence, there are two
main interpretations: irstly, the kings ruled over other segments or groups of the
Tankh confederation,
848
and secondly, the kings changed their allegiance.
849
The
notion of capital or residence is problematic in this context as it implies that the
rulers mainly stayed at this place. especially the term capital follows a rather mo-
dern concept, and it has been reasonably questioned especially with regard to the
seleucid empire:

since the seleucid kings were most of the time on journey or cam-
paign they stayed at different cities, and the idea of a single capital of their empire
842 bellamy (1985), p. 46 with the remarks on p. 45.
843 christ (1992), p. 747.
844 e.g. contini (1987), p. 55; luther (1997), p. 189; nagel (1998), p. 2.
845 bowersock (1994c), p. 133. see also the irst part of this section.
846 sommer (2005), p. 97.
847 see paul (2003), p. 567.
848 see e.g. ball (2000), p. 98. on different groups of the Tankh see schmitt (2005), p. 429.
849 With respect to Gadhima, see sartre (2005), p. 360.
333 The problem of nomadic allies in the roman near east
appears therefore anachronistic.
850
not only the seleucid kings, but also other ru-
lers seem to have been more or less permanently on the move, for example the
achaemenids or the arsacids.
851
also the Tankh rulers were frequently on cam-
paign, and their movement is the less astonishing if one accepts the idea of the
Tankh empire being nomadic. it is therefore not convincing to conclude that a
king who left al-rah to the west changed his alliance.
in order to describe the nature of the relation between the early Tankh rulers
and the great powers, i think it appropriate to introduce the concept of frontier
warriors. according to this notion nomadic groups who dwelt between two major
powers are employed to defend the frontiers while at the same time enjoying a high
degree of autonomy. There is often a balance at the frontiers between alliance and
revolt.
852
With respect to the empire of Gadhima and imraalqays it has been ar-
gued that it was some sort of an independent buffer between the roman and persian
empires.
853
against the background of the notion of frontier warriors this is not
quite correct. according to this concept, the speciic nature of the relation between
nomads and major powers lies in the dificulty to describe dependence and reliance:
on the one hand the nomadic frontier warriors are not really subdued, while on the
other they are not really autonomous either.
854
if one applies this to the Tankh ru-
lers, the contradiction between their employment by the sasanid kings and their
seemingly quite independent actions becomes, if not dissolved completely, at least
explicable: the early Tankh rulers were formally allies of the sasanid kings and as
such had mainly the task to defend the frontier; at the same time they were indepen-
dent enough to be on friendly relations with rome and especially in the case of
the struggle against Zenobia to follow their own interests.
it seems that the conduct of smaller nomadic groups was similar to this. accor-
ding to ammianus marcellinus, saracen dynasts (reguli) offered the crown to the
roman emperor and honoured him as lord of the world.
855
apparently these dy-
nasts must have been more or less closely connected with the Tankh lords of the
region. however, at the same time they followed their own policy, seemingly quite
independently from the lakhmid kings.
856
furthermore, as ammianus records, the
saracens at dura-europus, who had been allied to Jovian, became enemies of Ju-
lian, because they did not receive proper payment and taxes from him.
857
it is ob-
vious that all this is the expression of a conduct similar to that of the earlier noma-
dic groups which were described by strabo and which i have sketched in the third
850 austin (1999), p. 15860
851 on the movement of the achaemenid kings see briant (1988); briant (2001), p. 1067; schar-
rer (1999), p. 1101. on the movements of the arsacids see hauser (2005), p. 1812. see on
the movements of the frankish kings e.g. also schneider (1995), p. 52.
852 beck (2003), p. 12930. This concept is not to be confused with the elaborate, but too contri-
ved semiotic concept of cross-border commuters (e.g. sommer (2005), p. 107 and p. 4028).
853 With respect to Gadhima: hartmann (2001), p. 347. With respect to imraalqays: bowersock
(1975), p. 522.
854 beck (2003), p. 131.
855 amm. marc. 23.3.8.
856 on the reguli see schmitt (2005), p. 42930.
857 amm. marc. 25.6.10.
334 Ulf scharrer
part of this section it seems, therefore, that also on the lower level the concept of
frontier warriors may be applied, and that the political behaviour of the smaller
groups did not change between the beginning and the end of the period covered in
this paper.
conclUsion
The notion of frontier warriors is quite appropriate to summarise the problem of
nomadic allies in the roman near east. it seems that the relations with the roman
and persian empires of most, if not all, of the nomadic groups covered here can be
reasonably described by this concept: it can be applied to the Aw and the aif,
who may have changed their alliances according to the political circumstances, as
did nomadic groups at the euphrates according to strabo. furthermore, on the
higher level the Tankh seem to have enjoyed some degree of autonomy while at
the same time being vassals to the sasanid kings. besides the changes sketched in
this paper, this seems to be a major constant of nomadic alliances with the great
powers.
apart from explaining the nature of nomadic alliances, the idea of frontier
warriors at the same time points with its vagueness to the problem with regard to
clear statements on nomadic groups, and on their economic and socio-political con-
duct: many of the groups covered throughout this essay are discernible as having
seemingly contradictory relations to the sedentary world, be it some sort of econo-
mic exchange or raiding, be it revolting or allying. against the background of the
theoretical approaches which i have sketched in the irst section, all these features
might fall in the category of speciic nomadic strategies of survival, depending on
the state of the physical and socio-political surroundings. While this may be an ex-
planation, at the same time nomadic groups cannot really be ixed to a certain con-
duct.
Whereas these problems may be found also in other historical and geographical
contexts of nomadic life, it is clear that with respect to the roman near east there
is the additional problem of reading and interpreting the diverse material, and so-
metimes it is possible only to speculate which conclusions should be drawn from
speciic readings and interpretations of the evidence, as the examples of the Thamd
at rawwfa, the Greek inscriptions from the hauran and the epitaph of imraalqays
have shown. There certainly is further evidence, not covered in this essay, which
may contribute to sharpen or abandon some of the views presented here.
besides the attempt to be cautious, there are surely preconceptions in this essay
which enter some interpretations of evidence brought forward here. one of these
preconceptions is that the notion of client kingdoms, itself a modern concept,
which has been viewed not without criticism.
858
against the background of this
essay it seems that the idea of client kingdoms is not fully appropriate to describe
858 on the concept see e.g. Kaizer (2003a), p. 291; Kehne (2000), including criticism. see also the
introduction to this volume.
335 The problem of nomadic allies in the roman near east
many features of the nature of alliances between nomadic groups and the major
powers. even in the case of the Tankh, whose rulers apparently were installed by
the sasanid kings from shapur i onwards, this concept is shattered by the more
appropriate notion of frontier warriors.
it thus seems that nomads by their very nature often escape the sources and our
structuralizing mind, conceptually splitting into a good range of singularities. alt-
hough it refers metaphorically to metaphysical entities, a quotation of Gilles de-
leuze seems to me most appropriate to summarize the problems of this essay: de-
leuze writes that les singularits-vnements correspondent des sries htro-
gnes qui sorganisent en un system ni stable ni instable
859
following deleuze,
braidotti states that writing is the process of undoing the illusory stability of ixed
identities
860
at least writing this essay has been such a process.
859 deleuze (1969), p. 125.
860 braidotti (1994), p. 15.

You might also like