Nomadic Society and the Seljūq Campaigns in Caucasia

Author(s): Andrew C. S. Peacock

Reviewed work(s):
Source: Iran & the Caucasus, Vol. 9, No. 2 (2005), pp. 205-230
Published by: BRILL
Stable URL: http://www.jstor.org/stable/4030954 .
Accessed: 22/08/2012 21:32

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

BRILL is collaborating with JSTOR to digitize, preserve and extend access to Iran & the Caucasus.

http://www.jstor.org
 NOMADIC SOCIETY AND THE SELJUQCAMPAIGNSIN
CAUCASIA

ANDREW C. S. PEACOCK

Universityof Cambridge

The conquest of much of the Middle East by the Seljiiq Turks from
Central Asia in the eleventh century had a profound impact on the
culture and demography of the region, while the invaders' brutality
horrified contemporary witnesses, both Muslim and Christian. A hadith
current in the period states, "God said, <I have an army that I have
named the Turks and I have settled them in the East. When I am an-
gry with a people, I send against them the Turks>>".'The Turks had
been known in the central Islamic lands before, particularly as soldiers
in the 'Abbasid caliphate's armies. However, it was these invasions of
the Seljiiqs that started the settlement of much of Anatolia, the Cauca-
sus and northern Iran with Turkish populations, although many more
immigrant Turks arrived in the region in the Mongol period, com-
pleting its ethnic transformation. The coming of the Seljuiqs also had
dire consequences for the Byzantine Empire. Although the latter had
defended itself reasonably successfully during centuries of warfare
against Persia and later the Arabs, it was to lose most of its Anatolian
territories to the Turks within a few decades.
The Seljaq conquests were thus extremely significant, not just for
the Middle East but also for Europe. They fatally weakened the Byz-
antine Empire, and the Seljiiq capture of Jerusalem was one of the
immediate inspirations for the First Crusade. Yet the invasions them-
selves remain surprisingly unstudied, even in those areas where they
had the most profound consequences. For instance, very little work
has been done on the material in the Persian and Arabic sources on
the Turkish conquests in Anatolia since Cahen's articles, mostly writ-
ten in the 1960s.2 Indeed, much modern secondary literature relies on

l MahmuldKashghari,DivanuLugat-it-Tiirk,
ed. BesimAtalay(Ankara,1939-1943),I: xvii.
2
See Claude Cahen, Turcobyzantina
et OnensChnstianus(London, 1974), Studies I-VII. Un-
fortunatelyNodar iengelia,Sel6'ukebi
da Sak'art'velo
XI saukuneszi
(Tbilisi, 1968) has been unavail-
able to me.

K Brill,Leiden, 2005 Iran and the Caucasus,9.2

206 ANDREW C. S. PEACOCK

YinanS's study published in 1944,3 which contains very little analysis,

aiming mainly to establish the chronology of the various Seljuiq cam-
paigns.4 In general, research on the invasions has attempted to under-
stand what happened, rather than why it did. In particular, the rela-
tionship between the Seljaqs and their tribal followers, the nomadic
Turkmen,5 has received little detailed attention despite it being the key
to understanding many aspects of Seljiiq rule. The importance of the
Turkmen was noted by the famous Seljiiq vizier Nizam al-Mulk (d.
1092), who commented that "they have a long-standing claim (haqq)
upon this dynasty (dawlat), because at its inception they served well
and suffered much, and also they are attached by ties of kinship".6
This paper is a study of the Seljaq invasions of eastern Anatolia
and southern Caucasia in the mid to late eleventh century. There are
several reasons for focussing on this region. Firstly, the earliest settle-
ments of Turkish tribesmen in the Middle East were concentrated in
it. Secondly, there is a wealth of local sources in Georgian, Armenian
and Arabic to supplement the surviving Arabic and Greek general
chronicles, enabling us to get a much fuller picture of the nature of the
conquests than is possible for many other regions affected by the
Seljuiq invasions. Thirdly, the Seljiiqs themselves were particularly in-
terested in this region, with each one of the first three Seljuq sultans,
Tughril, Alp Arslan and Malikshah, leading major campaigns there. It
is thus natural that Caucasia should form the background for many of
the tales of Dede Korkut, a popular mediaeval Turkish epic which re-

3 MukriminHalil Yinanc, Tisrkiye TarihiI: Anadolu'nunFethi(Istanbul, 1944). More recently,

Yasar Bedirhan,Selguklular veKafkasya(Konya, 2000), adds some details but does not fundamen-
tally reviseour understandingof the conquests.
4 More work, most recently the studiesof S. Vryonis on Michael
Attaliates(see SperosVry-
onis Jr, "A Personal History of the Battle of Mantzikert",ByzantineAsia Minor6th-12thcenturies
(Athens, 1998):225-244) has been done on the non-Muslimsources.Nonetheless,referenceis still
frequentlymade to J. Laurent'sByzanceet les TurcsSeljoucidesjusqu'en 1081 (Nancy, 1913) and R.
Grousset'sHistoirede l'Arrneniedesorigines
a 1071 (Paris, 1947), althoughboth are based almost en-
tirelyon Greek and Armeniansourcesand are ratherdated. See Robert Bedrosian,"The Turco-
Mongol Invasionsand the Lordsof Armeniain the 13-14th Centuries"(unpublisheddissertation,
Columbia University, 1979; available at www.rbedrosian.com),Chapter One, for a survey of
more recentliterature,especiallyin Armenian,on the conquests.
5 In this paper I use the term Turkmen to describe these nomadic Turks, while the sources
describethem variouslyas "Ghuzz","Tuirkmen", and "Turks".It is not clear what (if any) differ-
ence there was in meaning between Ghuzz and Turkmen, althoughoriginallyGhuzzwas used to
refer to non-Muslim Turks. I do not make a strict distinctionbetween Turks and Turkmen for
the purposesof this article(see Cl. Cahen "Ghuzz",EI2, II: 1106-1110, esp. 1106; BarbaraKell-
ner-Heinkele,"Turkmen",EI2, X: 682-685, esp. 682).
6 Nizam al-Mulk, Siyaral-Mulak/SiSasatndma, ed. Hubert Darke (Tehran, 1340): 130; idem,
TheBookof Government orRulesforKi-ngs,
tr. HubertDarke (London, 1960):105.
 NOMADIC SOCIETY AND THE SELJDQCAMPMGNS IN CAUCASIA 207

flects stories of the heroic past of the Turks in the period.7 A Georgian
chronicler records one Turkish leader saying to another, "Why are
you setting out for Greece [Byzantium]? Behold Georgia depopulated
and full of such riches" and apparently the Turks then "changed di-
rection and spread out over the face of the land [of Georgia] like lo-
custs".8
The conventional view of these conquests is best expressed by C.
Cahen, who distinguished four types of campaigns, which he described
as follows: "celles du prince [selgukide] lui-meme, celles que font des
lieutenants plus ou moins autonomes au service de sa politique directe,
celles que font ces lieutenants en marge de cette politique afin d'ali-
menter et de detourner l'appetit de betes et l'instinct de pillager des
Turcomans, enfin celles que font des Turcomans en dehors de toute
intervention selgukide, voir en rebellion contre lui".9 Yet Cahen be-
lieves that the Seljaq sultans were generally more interested in secur-
ing recognition as legitimate sovereigns by the Muslim world, and
their campaigns against Caucasia and Anatolia aimed largely to divert
the Turkmen's attention away from plundering the more important
provinces of Iran and Iraq.10 This view is also adopted by Bosworth
and Lambton, who likewise believe that the aim of the policy of early
sultans such as Alp Arslan was to keep the Turkmen pre-occupied out-
side the ddral-Islam."
This view of the sultans' interventions in Anatolia and Caucasia
more as reluctant-and unsuccessful attempts to control their tribal
followers than as motivated by a desire for conquest, and certainly not
to destroy the Byzantine empire, derives from the depictions of the
Seljuiq rulers in the Islamic sources. Written in Arabic and Persian,
frequently in honour of later sultans, these sources tend to emphasise
the civilised, Islamic aspects of Seljuiq rule, such as the Seljuiqs'role as

The exact date of both the text and the contents of DedeKorkut is a matter of some contro-
versy, and given the work's originally oral character, one unlikely ever to be resolved conclu-
sively. Some elementsprobablydo reflectthe periodjust after the conversionof the Turks,while
other parts of the frontiernarrativesseem to recall the wars between the AqquyunuluTurkmen
and the Georgians in the fifteenth century (see TheBookof DedeKorkut, tr. Geoffrey Lewis (Har-
mondsworth,1974): 10-12, 18-19).
8 R. W. Thomson, Rewriting CaucasianHistory:theMedievalArmenian Adaptationof theGeorgian
Chronicles.TheOriginalGeorgianTextsandtheAnnenian Adaptation(Oxford, 1996):310.
9 Claude Cahen, "Lapremierepenetrationturque en Asie Mineure",Byzantion XVII (1946-
1948), 12 (Reprintedin idem, Turcobyzantina et OriensChristianus).
10 Idem, Pre-Ottoman Turkey:A GeneralSurveyof theMaterialandSpiritualCultureandHistogyc.
1071-1330, tr.J.Jones-Williams(London, 1968):23.
" C. E. Bosworth,"The Politicaland
Dynastic History of the IranianWorld (1000-1217)":
43; A. K. S. Lambton, "The Internal Structureof the SaljusqEmpire":246, both inJ. A. Boyle
(ed.) TheCambridge HistogyofIran,vol. V: TheSaljuqandMongolPeriods (Cambridge,1968).
208 ANDREW C. S. PEACOCK

defenders of Sunni Islam and builders of madrasahs. Yet they were gen-
erally written by authors of Arab or Iranian, not Turkish, descent, and
reflect an ideal of Islamic kingship more than the reality. Moreover,
none of these works were written until long after the initial conquests.'2
Thus it is unsurprising that they rarely linger on the question of the
relationship between the sultan and the Turkmen.'3 Yet, it is clear
from chancery documents issued in the twelfth century that even then
the Turkmen retained an important role in the Seljuiq state.'4 It is also
evident that many of the so-called raids of the mid to late eleventh
century were in fact led by leading members of the Seljiiq family, in-
cluding the sultans themselves. Indeed, even under Malikshah (d.
1092), when it is generally thought that traditions of Perso-Islamic
government had reached the peak of their influence partly through the
offices of the sultan's famous Persian vizier Nizam al-Mulk, it is evi-
dent that such pillaging and seemingly mindless destruction continued
on border areas, apparently with the participation of the sultan him-
self.'5
This paper will consider why Caucasia and eastern Anatolia were
so attractive to the Seljiiqs and the Turkmen, suggesting that the im-
portance of these regions derived not from their place outside the ddr
al-Islam-for much of the area was in fact Muslim already-but rather
to the existence of conditions ideal for a nomadic lifestyle. It will be
argued that nomadic society profoundly influenced the nature of both
the Seljiiq conquests and Seljuiq rule. First, however, it is necessary to
understand the political situation in Caucasia on the eve of the Turk-
ish invasions.

There had been a Muslim presence in the region long before the arri-
val of the Seljuiq Turks. Arab tribes had first settled in Caucasia in the
eighth century, mainly for military reasons.'6 Those around Derbend
were to man the Caliphate's defences against its great enemy, the
Khazar Empire in southern Russia. Some were settled around Lake
Van to ensure the continued obedience of the Armenian lords who

12 See Julie Scott Meisami, Persian to theEnd of theTwelfthCentugy

Historiography (Edinburgh,
1999): 141-145; C. Hillenbrand,"Some reflectionson Seljuqhistoriography",Antony Eastmond
(ed),EasternApproaches toByzantium(Aldershot,2001): 74, 83, 86.
13 A. K. S. Lambton, "Aspectsof Saljuiq-Ghuzz settlementin Persia",D. S. Richards (ed.),
PapersonIslamicHistogy,III IslamicCivilization
950-1150 (Oxford, 1973):107-109.
14 Ibid.: 109-110.
'5 Thomson, Rewriting CaucasianHistogy:307.
16 See Aram Ter Ghewondyan, TheArabEmirates in Bagratid
Arnenia,tr. N. G. Garsoian(Lis-
bon, 1976):29ff.
 NOMADICSOCIETYANDTHESELJOQCAMPAIGNS
IN CAUCASIA 209

had a tendency to revolt against their overlords, whoever they were,

for so independent-minded was the Armenian nobility that in an ear-
lier age it had successfully petitioned the Persians to abolish the Arme-
nian monarchy itself.'7 Other Arabs came to take part in the ongoing
war against Byzantium. In addition to these Arab tribes, in the tenth
century Kurdish dynasties became important, especially the
Shaddadids based in the fertile province of Arran.'8 Although Arabic
remained a spoken language in parts of Caucasia until at least the
twelfth century, to the surprise of travellers from elsewhere in the
Middle East,19the Arab tribes, who had by now established themselves
as independent dynasties, were strongly influenced by the culture of
the region. Both Kurdish and Arab dynasties started to adopt ancient
Iranian names, particularly those typical of the Caspian region of
Daylam, which exerted a strong cultural influence on the territories to
the north of it. Thus among the Sharwanshahs, descended from the
Arab governors of Sharwan, by the beginning of the eleventh century
Arab names had been supplanted by Iranian ones such as Maniichihr
and Farlburz. Some dynasties, meanwhile, such as the Rawwadids,
who had probably also been of Arab origin, started to Kurdicise them-
selves, adopting typically Kurdish names such Mamlan and Ah-
madil.20Nonetheless, the literary language of the Muslims of Caucasia
remained Arabic, or sometimes Persian.
Until the early eleventh century, Armenian and Georgian king-
doms and principalities had existed among the patchwork of Muslim
states. At least among the ruling classes, intermarriage between Mus-
lims and Christians was common, and among members of the Shad-
dadid dynasty typically Armenian names such as Ashot are found.2'
However, this should not necessarily be taken to indicate the devel-
opment of a friendly symbiosis: the Shaddadid ruler Abu-'l-Aswar was
quite happy to invade and occupy the lands of his Armenian neigh-
bour, David of Tashir, despite being married to his sister.22 Most of
these Armenian kingdoms had been annexed by Byzantium in the

17 Artexias IV was removed as king in 428 by the Sasanians

at the request of the nobility:
Nina Garsoian, "The ArsakuniDynasty",Richard G. Hovannisian(ed.),Armenian PeoplefromAn-
cientto ModemTimes,Vol.I. TheDynasticPenods:FromAntiquity to theFourteenthCentury (New York,
2004):93.
18 See Ahmad Kasravi,Shahniydrdn-i
Gumndm (Tehran, 1377):226ff.
'9 Vladimir Minorsky,A Historyof Sharvan andDarbandin the10th-lith centuries (Cambridge,
1958): 170-172.
20 C. E. Bosworth, TheNew IslamicDynasties: A Chronologicaland GenealogicalManual(Edin-
burgh, 1996): 140-142, 150-152.
21 V. Minorsky,Studiesin Caucasian
Histogy(Cambridge,1953):22, 24.
22Ibid.:51-52.
210 ANDREW C. S. PEACOCK

early eleventh century, and their rulers or their descendants had been
pensioned off with estates elsewhere in Anatolia.23 In place of these lo-
cal rulers who had played an important role in supporting Byzantine
power in the region, for its eastern defences the empire came to rely
on a system of forts^which has been compared to a Maginot line, that,
if breached or circumvented, became useless.24 The Byzantines had
further weakened their own border by replacing levies of local troops
with mercenaries, and had made themselves distinctly unpopular with
the local Armenian population by persecuting it on religious grounds.25
Together, these policies would contribute to the fall of Anatolia to the
Turks.
Apart from the handful of petty principalities that had survived the
Byzantine cull such as Tashir,26 only two significant Christian powers
in the region survived, the two Georgian Kingdoms of Abxazet'i-K'ar-
t'li and Kaxet'i-Heret'i. Tiflis, which did not become capital of a
united Georgia until 1122, was a city-state in Muslim hands at this
point,27 and other Georgian lands such as the provinces of Tao-Klar-
jet'i and Savset'i in modern Turkey were under Byzantine control.
Nonetheless, the early eleventh century marks the beginning of a uni-
fied Georgia for the first time since the Roman period, with its two
principal regions, Kart'li and Abxazet'i, joined together under the rule
of a branch of the Bagratid family.28
Such, in outline, was the situation in Caucasia on the eve of the
Seljuiq invasions. These Muslim dynasties of Kurdish or Arab origin
sometimes competed with Byzantium and Georgia, and sometimes
were allied to them. Byzantium's desire to extinguish the Armenian
principalities occasionally led it to call on neighbouring Muslim dy-
nasties like the Shaddadids for assistance; conversely, sometimes fear-
ful of the Muslims becoming too powerful, it would strike directly at
the Shaddadids in their principle cities, Ganja and Dvin.29 As a result,
a new front line between Byzantium and the Muslim world was drawn
up in the Caucasus, beyond traditional Byzantine territory. The Shad-
dadid capital, Ganja, became a magnet for religious warriors who

23 See Nina Garsoian, "The Byzantine

Annexation of the ArmenianKingdoms in the Elev-
enth Century",Hovannisian(ed.),Armenian Peoplefrom
AncienttoModernTimes,vol. I: 187-198.
24 Jean-Claude Cheynet, "La
conception militaire de la frontiere orientale (IX' -XIII'
siecle)",in Eastmond,Eastern Approaches:
63
25 Garsoian,op.cit.:193, 196-197.
26 Ibid.: 192-193.
27
V. Minorsky,C. E. Bosworth,"al-Kurdj"EI2, V: 489.
28
W. E. D. Alien, A Histogyof theGeorgian
People(London, 1932):83-85.
29 Minorsky,Studies: 52-64.
 NOMADIC SOCIETY AND THE SELJDQCAMPAIGNSIN CAUCASIA 211

wished to fight the Byzantine threat. Kayka'fis b. Iskandar, later ruler

(1049-1087) of the Ziyarid state on the south Caspian, remarks in the
Qa,biusndmah, a mirror for princes dedicated to his son, that he spent
much of his youth engaged injihdd, firstly in India, and subsequently at
the Shaddadid court in Ganja, where he stayed for several years
fighting the Byzantines.30
Thus, when the first Seljiiq Turks arrived in Caucasia, probably in
the late 1020s, the region was deeply fragmented politically. Yet, it
was by no means territory outside of the ddral-Islam, but a patchwork
of Muslim and Christian states. Seljuiq attacks were not, however, re-
stricted to the Christian territories of Caucasia, but affected the Mus-
lim ones at least equally. Nor were these attacks on Muslim lands nec-
essarily carried out by irregular groups of Turkmen, but rather were
often led by the sultan himself, as will be demonstrated below. Thus,
the Seljuiqs' interest in Caucasia cannot have been prompted by a de-
sire to find lands outside of the ddral-Islim for their followers to pillage,
because many of these lands not only formed part of the Muslim
world, but had also, as Kayka'iis the Ziyarid indicates, gained a repu-
tation as a centre of the jihdd. Attacking the Muslim centres of the jihdd
does not easily lend itself to being interpreted as part of a strategy of
promoting an image of the Seljiiqs as protectors of Islam. We must
seek elsewhere an explanation for the Seljaq conquests in the region,
and thus we will turn to a detailed examination of the conquests.
However, as Cahen correctly states, there were a variety of different
types of campaigns, ranging from those conducted by rebellious
Turkmen to those led by the Seljuiq sultan himself. The former were
often quite small in scale, sometimes involving as few as twenty men,31
while the latter were usually much more considerable affairs. Here, we
shall focus only those campaigns against Caucasia led by members of
the Seljuiq family themselves. These are naturally by far the best
documented campaigns and are of the greatest interest as they illus-
trate the priorities of the sultans themselves.

The exact causes of the migration of the Seljuiq Turks westwards are
obscure, although it seems in part they were obliged to leave Central
Asia by a combination of famine and pressure from the rulers of the
area. When they arrived in the Middle East, they were only recent

30
Kay Ka'usb. Iskandar,Qa-busndmah, ed. Ghulamriqla Yflsufi(ehran, 1371):41 (Chapter7).
31 Sibt ibnu 'I-Cevzi,Mir'dti 'z-ZemanfiTarihi'I-Ayan,ed. A. Sevim (Ankara,1968) (hence-
forth, Sibt, Mir'dt):21.
32
On the early Seljuiqssee C. E. Bosworth,"Politicaland DynasticHistory":11-23.
212 ANDREW C. S. PEACOCK

converts to Islam, and many still clung to their nomadic ways, al-
though Tughril, the first Seljiiq sultan, adopted this Islamic title in
1038, shortly after his conquest of Khurasan.33 However, attacks on
the region by Turkmen connected to the Seljiiqs had started before
this, although Tughril did not appoint his cousin Qutlumush b. Isra'il
b. Seljiiq over the provinces of Armenia and Azerbajian until this
date.34Unfortunately, the sources for this period only contain passing
references to Turkish attacks and it is difficult to reconstruct the se-
quence of events with any accuracy. Enough, however, survives to
make an attempt possible, and a rough chronology of known major
attacks is presented in an appendix to this article.
The Rawwadid territories in what is now Iranian Azerbaijan were
one of the first regions to be settled by Turkmen fleeing Central Asia,
and were used by them as a base to extend into Armenia and eastern
Anatolia. The earliest credible reference to Turkish attacks in these
areas is in an inscription recording the death of Vasak Pahlavuni,
prince of the local Armenian Artsruni dynasty, in 1029.35 This is con-
firmed by Ibn al-Athir, who mentions an attack on Armenia by
Turkmen based in Urmiyah in Azerbaijan in this year.36 Cahen be-
lieves there was then a lull until 1035, when more attacks occurred.
Already by 1038 the Turkmen had penetrated so far and were so dan-
gerous that the Georgians were forced to abandon their attempt to
capture Tbilisi from the Muslim Ja'farid dynasty for fear of them.37
However, the main wave of attacks does not seem to have started until
the early 1040s, and was probably prompted by Vahsuidan the
Rawwadid's massacre of Turkmen in Tabriz, which resulted in a mi-
gration further westwards.38A Seljuiq attack is recorded sub anno 434
A. H./ 1042-3 by al-'Az4mi (d. late 12th century), who states that Turks
commanded by the amirs Biiqa, Anasughli, Bektash and Mansiir at-

3 Ibid.:23.
34 $adr al-Din al-Husayni, Akhbdral-Dawlatal-Sajiqiyyah,ed. Muhammad Iqbal (Beirut,
1404/1984): 17 (references throughout to this edition unless otherwise specified). Armenian
sources indicate that the first Turkish attackson the Van region may have occurred as early as
1015, and this date is still frequentlycited in modern secondaryliteratureas the date of the earli-
est Seljuiqincursions.However, as Cahen has demonstrated,it must be rejected,and the earliest
Turkish attacksprobably did not occur until 1029, followed by a lull until 1035: C. Cahen, "A
propos de quelques articlesdu KoprlilAnnagant", journalAsiatique242 (1954): 275-279 (reprinted
in idem, Turcobyzantina et OriensChristianus).
35 Grousset,op.cit.:551, n. 4, citingAlishan,Shirak(Venice, 1879): 148.
36 Ibn al-Athir,al-Kamil, IX: 383.
3 Ibid., IX: 457.
38 Ibid., IX: 384-385; Bosworth, "Politicaland dynastic history":41; Kasravi, Shah7iydrdn-i
Gumnam: 170-173.
 NOMADIC SOCIETY AND THE SELJCQCAMPAIGNS IN CAUCASIA 213

tacked first Armenia and Azerbaijan, and then the Kurdish Marwanid
dynasty in the Diyar Bakr.39The first significant Seljiiq victories ap-
pear to have come in the wake of this campaign with the capture of
the Byzantine constable of Ercis on the northern shores of Lake Van
by forces possibly commanded by Qutlumush in 1044 or 1045.4? For
the next few years the situation is extremely confused, although it is
clear that Seljuiqs maintained an interest in the area. Especially in-
triguing are the activities of Qutlumush, who is found one year fight-
ing off a Byzantine attack on Shaddadid Dvin and the next is himself
besieging the Shaddadids in Ganja, a siege which apparently lasted a
year and a half, scarcely the behaviour of nomads interested solely in
plunder.41
Around 1048-9, another cousin of Tughril, Ibrahim Yinal, led a
major campaign in eastern Anatolia, which in 1054 was followed by
one commanded by Tughril himself. More or less annual attacks on
one part or other of the region ensued, culminating in the great cam-
paign against Georgia and Eastern Anatolia led by Alp Arslan in 1064,
shortly after his accession to the sultanate. Alp Arslan led three cam-
paigns in the region, a second one against Georgia, Arran and Ana-
tolia in 1067, and thirdly and most famously, the campaign that cul-
minated in the defeat of the Byzantines at Manzikert in 1071. Unfor-
tunately, the region is virtually completely ignored by the Islamic
sources during the reign of Malikshah (1072-1092), although it is given
detailed attention by the Georgian chronicles, although it is clear that
Malikshah himself appeared on campaign in Caucasia in 1080.42
Seljiiq attacks continued, although towards the end of this period the
political situation in Caucasia changed to their disadvantage with the
emergence of a strong Georgian kingdom, which expanded at their

39 AzimiTarihi:Selfuklularlailgiliboliimler(H.
430-538=1038/9-1143/4), ed. Ali Sevim (Anka-
ra, 1988), Arabic text: 3. 'Az.imi'saccount is partlyconfirmedby his contemporaryIbn al-Azraq
al-Fariqi,who mentionsan attackon the Marwanidsin the same year, led by the amirsBiuqaand
[A]na*ughli,sent to the Diyar Bakrby Tughril who had granted it as iqta'to them (Ibn al-Azraq
al-Fariqi, Ta'ritkhal-Fdriqi:al-dawlahal-Marwadnyyah, ed. Badawl 'Abd al-Latif 'Awad (Beirut,
1974): 160).
40 AzimiTarihi,Arabic text: 4, s. a. 435/1042-3; ArneniaandtheCrusades, Tenthto Twe#fhCen-
turies:TheChronicleofMatthewof Edessa,tr. Ara Edmond Dostourian(Lanhamand London, 1993)
(henceforth,Matthew, Chronicle): 74. (1045);these two authorsattributethe captureof the consta-
ble of Ercisto the amirsBuiqaand Anasughlion their returnfrom the Diyar Bakr.However,Jean
Scylitzes, Empereurs de Constantinople,tr. Bernard Flusin (Paris, 2003), Greek text: 446/tr.: 371;
NicephorusBryennius,Histoiarum LibriQusattuor,
ed. P. Gautier(Brussels,1975):96-99; and Zona-
ras, EpitomeHistorion,ed. M. Pinder (Bonn, 1841-1897),III: 636-7, state that it was carriedout by
Qutlumushin the wake of his defeat at Mosul.
41 AzimiTarihi,Arabictext: 6.
42 al-Husayni,Akhbdr 63.
214 ANDREW C. S. PEACOCK

expense under David the Restorer (1189-1125).43 Qutb al-Din Isma'l

b. Yaqiuti, the sultan's cousin, was appointed over the entire area of
Arran and Azerbaijan, and the Shaddadids, who had proved to be de-
cidedly unreliable allies, were deposed, although the Sharwanshahs
retained power as Seljuiq vassals.4 Insufficient details survive to con-
struct a reliable chronology for the campaigns under Malikshah.
Forty years separate the first major Seljiiq intervention in the re-
gion, Ibrahim Yinal's campaign against Ruim in 1048-9, and the start
of the Georgian recovery in the north with King David's accession.
However, many of the campaigns during this period share certain
features. Firstly, the Seljuiqs show a definite tendency to return to
places they have already attacked (see map). Thus, the mountainous
province of Tao was attacked by Ibrahim Yinal, Tughril, Alp Arslan
and Malikshah, while Somxit'i was attacked by both Alp Arslan and
Malikshah, and Xordzean was attacked in 1048 and 1057. Manzikert,
most famous as the site of the crucial battle between Alp Arslan and
the Byzantine emperor Romanus Diogenes in 1071, had previously
been subjected to two major attacks, the first by Ibrahlm Yinal in
1048-9, and the second by Tughril in 1054-5. Standard Seljuiq tactics
were for the army to divide into several wings, which is why so many
attacks on several places in the same expedition are recorded.45
Secondly, it is interesting to note that many of the places attacked
by the Seljiiqs were heavily fortified. Manzikert, for instance, was al-
ways strongly defended, but the sources state that Tughril besieged the
town in a determined and professional manner, first of all sending in
sappers to undermine its walls, and when this failed using mangonels.46
Nor was this an isolated event, for Alp Arslan used similar tactics and
siege machinery against Ani in 1064.47 The accounts by al-Husayn
and Ibn al-Athir of Alp Arslan's campaign of 1064 indicate that the
principal targets were in fact fortresses and fortified cities. It might be
argued that this campaign, under the command of the sultan, cannot
be assumed to be representative of the Turkmen attacks in the region.
However, lesser known campaigns, such as Qutlumush's year and a

4 Allen, op.cit.:95-100; Thomson, ReriitingCaucasianHistogy:309ff.

44 Bosworth,"Politicaland Dynastic History":94-5. Minorsky,Histogyof Sharvan andDarband:
68, 120; al-Husayni,Akhbdr73; al-Bundari,Zubdatal-.Afu,srah, ed. M. T. Houtsma (Leiden, 1889):
140. A branchof the Shaddadidssurvivedat Ani.
45 See M. Brosset,Histoire dela Georgie
(St Petersburg,1849),I, 347, n. 4; Allen, op.cit.:91.
4 Matthew, Chronicle: 86-88; Aristakesde Lastivert,Ricitdesmalkeurs de la nationarninienne,
tr.
Marius Canard and Haig Berberian(Brussels,1973):89ff (referencesto the page numbersof the
classicalArmeniantext).
47 Ibn al-Athir, X: 40-41.
al-Kamil,
 NOMADIC SOCIETY AND THE SELJUQCAMPMGNS IN CAUCASIA 215

half long siege of Ganja around 1047, and the attacks of 1056-7 on
well-fortified Kemah and Koloneia, the latter described by Bryer and
Winfield as "the most formidable of the Pontic natural strongholds",48
indicate that Turkmen acting in smaller groups were equally ready to
attack strongly defended locations. Aristakes specifically mentions that
among the targets of Ibrahlim Yinal's campaign of 1048-9 were "the
castles of Tao and Arsarunik"'.49 Far from avoiding the "Maginot
line" of Byzantine forts, the Seljiiq Turks, whether acting under the
command of their sultan or not, deliberately attacked it.
Furthermore, it is clear from Arabic, Armenian and Georgian
sources that when the Seljuiqs had conquered a town or fort, they
would often then systematically destroy it, most often by fire. Thus fire
was used against Ani and against Axalk'alak'i,50and there was a unit of
naffittn (naphtha-throwers) in the Seljuiq army, whose job seems to
have been to burn anything that hindered the Seljiiq advance.5' The
anonymous author of the Georgian chronicle known as 7TheHistogyof
David, Kingof Kingscomments that "fire, an unexpected adversary, con-

48 Anthony Bryer,David Winfield, TheByzantine Monuments andTopography of thePontos(Dum-

barton Oaks, 1985),1: 146.
49 Aristakes,Ricit 67. Arsarunikis the region to the south west of Ani.
50
al-Husayni,Akhbdr38-39; Ibn al-Athir,al-Kdmil,X: 39. I follow Bunyatov's reading of
Axalk'alak'ifor the place mentioned by these two authors as JY JUl8(or similarforms),which is
not, however, entirely certain. See the discussionin Marius Canard, "La campagne armenienne
du sultan selguqide Alp Arslan et la prise d'Ani en 1064" REArm. 2 (1965): 241 (reprinitedin
idem, L'expansion arabo-islamique
etsesripercussions
[London, 1974]) and Sadr ad-Din Husayni,Akh-
bhrad-daulatas-SeldZukyya, ed. and tr. Z. M. Bunyatov (Moscow, 1980): 180, where the editor
identifiesit with Axalk'alak'i.PhoneticallyAxalk'alak'iis the obvious choice, and other evidence
confirmsthat it was taken by the Seljulqs,most importantlya Georgian inscriptionat the nearby
village of Mirashkhanirecording the death of Parsman,lord of Mesxet'i, in Axalk'alak'iduring
the Seljiiq conquest (publishedby V. Silogava, "Ert'iepigrap'urijegli t'urk'-seljukt'asemosevis
sesaxeb XI saukunissak'art'velosi",Aspinja.Samc'xisa dajavaxet'issazgvarze(Axalc'ixe,2000): 201-
244). However, the descriptionof the town given in the accounts of al-Ijusayni and Ibn al-Athlir
(which draw on the same source)clearly does not refer to Axalk'alak'i.Firstly,it is describedby
the Arabic sourcesas having walls 100 dhira'sin length, whereas the Georgian sourcesrelate that
Axalk'alak'iwas inadequately defended by its walls at this point (Thomson, Rewriting Caucasian
Histo?y:299). Secondly, the city is describedas being surroundedby a mountain on all sides but
the south, and with many fortsnearby on smallermountainsand with a great riverthe size of the
Oxus running in front of the walls. Axalk'alak'iin fact is situatedon a high plateau with moun-
tains nearbybut not surroundingthe city, and has only a very small river.It seems likelythat the
name of the place given by al-Husaynt and Ibn al-AthIrdoes indeed recall Axalk'alak'i.How-
ever, the source for this report in his descriptionhas confused Axalk'alak'iwith another city in
the region. At any rate, the account is clearly exaggeratedas there is no river even remotely as
large as the Oxus inJavakheti.
51
al-Husayni,Akhbdr39, 44; Minorsky,Studies in Caucasian
Histoy:65.
216 ANDREW C. S. PEACOCK

sumed all dwellings".52 Occasionallya castle was allowed to stand un-

touched after its conquest and used by the Seljiiqs, as in the case of
Kaputru near Ararat.53The Arabic sources mention one such occa-
sion duringAlp Arslan'sGeorgiancampaignof 1064, which witnessed
the destruction of Axalk'alak'iand Ani. Alp Arslan's son, the prince
Malikshah,was accompanyingthe expedition,and intendedto destroy
a castle they had conquered but was prevented from doing so by the
vizier Ni-am al-Mulkwho wished it to become "a frontierfortification
for the Muslims"(thughr However, even when such forti-
li-'l-muslimin).
ficationswere preserved,the Seljiiqsusuallypreferredto grant them to
a local Muslim lord than to garrisonthemselves:in this case the castle
was given to the amir of Nakhichevan.i4 The defeat of Christian
kings-whether the Georgian rulersof Kaxet'i-Heret'ior Abxazet'i-
did not result in any attempt to annex their kingdoms,but rather the
imposition of poll-tax, leaving the monarchs in place (althoughAkh-
sartanof Kaxet'i-Heret'iconverted to Islam, presumablyto save him-
self from having to pay the jizyah).55Moreover, we are frequentlyex-
plicitlytold by the sourcesthat the Seljiiqsdid not stay permanentlyin
many of these places. This is perhaps most vividly illustratedby the
conquest of Ani, one of the most importantcities in the region and an
ancient Armenian capital. This is portrayedby the Muslim sources as
a great victoryfor Islam, greetedwith celebrationsin Baghdadand the
congratulationsof the Caliph.56After such a prestigiousconquest, it is
surprising to find that the Seljiuqssold the city to their unreliable
Shaddadidvassala few years later.57Thus, the objectiveof these expe-
ditions was not the annexation of these territoriesin any conventional
manner.
It is also interestingto note that these campaigns that led to such
destructionwere conducted by senior members of the Seljiiq family
themselves. Scholarship has tended to emphasise that much of the
damage done to Anatoliaby the Turkswas a resultof the depredations
of bands of nomads acting without the permission or even tacit en-
couragementof the Seljuiqgovernment. It is certainly true that some
atrocities were indeed committed by individual bands of Ttirkmen.
Sometimes these might be very small groups indeed, such as the loot-

52
Thomson, Rewriting Caucasian Histoty...:311
5 Matthew, Chronicle:78-79.
54 Ibn al-Athir,al-Ka:mil,
X: 38.
55
al-Iusayni, AAhbdr.44.
56 Ibn al-Athir,al-Kamil,X: 41; al-Husaynm, Akhbdr.
35-6.
Minorsky,Studies:81.
 NOMADIC SOCIETY AND THE SELJDQCAMPAIGNS IN CAUCASIA 217

ing of a monastery and the killing of all its monks by a group of Turks
twenty-strong recorded by Sibt b. al-Jawz!.58However, some of the
campaigns in Caucasia under discussion here clearly fall into a differ-
ent category: they were led by the sultan himself and senior officials in
the Seljulq state. In Alp Arslan's campaign of 1064, the best docu-
mented of all the attacks on Caucasia, there participated not only Alp
Arslan's son, the future sultan Malikshah, but also the governor of
Khurasan, the prosperous heartland of the Seljuiq empire, and the vi-
zier Nizam al-Mulk himself. Nizam al-Mulk was not, of course, a
Turk, but rather a highly cultivated Persian who came from an urban
milieu totally alien to that of the nomadic Turks. Even if it may be ar-
gued that Alp Arslan and his son maintained enough of the traditions
of the steppe to enjoy pillaging expeditions-although it is hard to see
why these would be necessary for the masters of some of the richest
areas of Muslim world such as Baghdad and Khurasan-it certainly
cannot be suggested that Nizam al-Mulk was motivated by any such
Turkic traditions. Rather, the presence of the most senior figures of
the Seljiiq family and their government is testament to the great im-
portance with which these campaigns were held by the Seljiiq family.
So we are presented with the slightly perplexing situation of the
Seljiiq state engaging in a war of apparently mindless destruction
against territories it had seemingly had no intention of holding perma-
nently. No simple explanation for this situation is given by any of the
sources. The Armenian and Georgian chroniclers lament the massa-
cres, the burning of their towns and villages, and the depopulation of
their homelands. Arabic sources, on the other hand, merely present us
with a list of conquered cities, with equally little suggestion as to what
exactly the Seljuiqs were trying to achieve. However, some less direct
indications in the sources suggest some reasons why the Seljaqs were
so interested in the region, which I shall now turn to examining.

According to al-Husayni, Caucasia was granted to Qutlumush, the

cousin of Tughril, in 1038, even before it had been conquered.59

58 Sibt, Mir'dt:21; Bar Hebraeus, The

Chronography, tr. Earnest A. Wallis Budge (London,
1932),I: 210.
'9 Other sources do not agree with al-Husayni, suggestingthat Qutlumushwas granted the
Caspianprovincesof Iran while Yaqilitwas sent to Azerbaijan.However, as Cahen has correctly
demonstrated("Qutlumushet ses fils avant l'Asie Mineure",DerIslam39 (1964): 14-27, esp. 20
(reprintedin Turcobyzantina)),
al-Husayni's account is corroboratedby the references in other
sources to Qutlumush's involvement in the campaigns against Vaspurakan, Ganja and Kars.
However, Caucasia was not the sole area, in which Qutlumush was involved, for he is also re-
ported as fighting the 'Uqaylids of Mosul, while it was in Northern Iran, where he made last
218 ANDREW C. S. PEACOCK

Azerbaijan had, however, already gained a substantial Turkmen

population, estimated by Kasravi at around 40,000 individuals.60Yet,
neither Qutlumush nor the Turkmen of Azerbaijan had a happy rela-
tionship with Tughril, who considered himself their leader, yet was not
necessarily recognised as such by the Turkmen. In these complicated
relationships lies an important clue to understanding Seljuiq activities
in Caucasia. The roots of these tensions are to be found in the story of
the Seljaq family's rise to prominence in Central Asia in the late tenth
and early eleventh centuries.
The first prominent member of the family whose career is reasona-
bly well attested is the elder son of Seljiiq, Arslan Isra'il, who in the
late tenth century is found fighting on the side of the Samanid rulers of
Central Asia. In the 1020s, economic circumstances, in particular se-
vere famine, seem to have obliged Arslan's followers to flee to the rich
province of Khurasan where they settled. Whether or not this was
with the permission of the government of the Ghaznavid ruler
Mahmid, as later writers claim,6' they soon made a nuisance of them-
selves through plundering. Arslan was eventually captured by
Mahmuid and imprisoned, and his nomadic followers fled further west
towards Azerbaijan.62 Thus, the first Turks to arrive in the region, in
1029, were the followers of Arslan, known as the 'Irdq7yyah Turkmen
due to the settlement of many of them in 'Iraq-i 'Ajam, i. e. western
Iran. It has already been noted that many of these settled in Azerbai-
jan, where initially the Rawwadid ruler Vahsuidan welcomed them de-
spite their plundering.63
Meanwhile, with Arslan in prison, another branch of the Seljaq
family, Arslan's nephews Tughril and Chaghri, rose to prominence as

stand againsthis cousin Alp Arslanin 1064. Other sourcesindicateTughril grantedhim the Cas-
pian provincesof Gurgan and Damghan, and he is mentioned campaigningas far away as Isfa-
han (for summaryof his career, see Cahen, op.cit.).These differencesin the sources reflect the
nomadic basis of the Seljuiqstate (if it can even be called a state at this date) and are not neces-
sarily contradictory.Qutlumush may have been allotted territoriesto north of Iran and/or in
Caucasia by Tughril, but that does not mean that he would have felt constrained to remain
within them, particularlyas they were still unconquered. As Lambton <InternalStructure>>...:
218) states, "[the Seljuiqs]probably thought that their rule extended wherever their people
roamed in searchof pasture,and not, in any case at first,that it was tied to a given area"(foran-
other example of Tughril granting unconquered territoriesto his followers, see Ibn al-Azraq,
Ta'rJkh al-Fdriq:160, where the Marwanid lands of Diyar Bakr are granted to Bulqaand Ana-
sughli).
60
Kasravi,Shahniydrdn-i
Gumndm: 159.
61
E.g. Bundari,Zubdatal-Nu4rah: 5.
62
On these developments, see Bosworth, "Politicaland Dynastic History": 11-23; Ibn al-
Athir, al-Kdmil,IX: 377-8.
63
Kasravi,Shahnyadrdn-i
Gumndm: 156-8;Ibn al-Athir,al-Kamil,IX: 381-383.
 NOMADIC SOCIETY AND THE SELJQQCAMPAIGNSIN CAUCASIA 219

chiefs of other groups of Turkmen whom they led out of the famine-
stricken lands of Central Asia to the province of Khurasan so recently
evacuated by the followers of Arslan. There were thus two distinct
main waves of Seljuiq migration westwards. The brothers Tughril and
Chaghri seized control of the main towns of Khurasan, so that with his
capture (albeit temporary) of Nishaptir in 1038, Tughril was able to
proclaim himself sultan, indicating a desire to be recognised as more
than just a nomad chief. At this point the territories that had fallen to
the Turks, so far along with those yet unconquered, were divided up
between the members of the family of Seljtiq.64Division of territories
among members of the ruling dynasty was a characteristic of Turkish
states, which shied away from investing one individual with absolute
65
sovereignty. Qutlumush received Caucasia, and possibly the Caspian
provinces of Iran.
The sources contain few references to relations between Qutlu-
mush and Tughril during this period. Sometimes they cooperated, for
Qutlumush's expeditions against the Byzantines in Dvin and against
the 'Uqaylids in Mosul seem to have been conducted at Tughril's re-
quest. To judge from the scanty information that survives about his
activities, he was not particularly successful from a military point of
view. Neither did his long siege of the Shaddadids in Ganja nor his
intervention against the 'Uqaylids and their ally al-Basasirl, the Turk-
ish soldier who was Tughril's rival for Baghdad, bear fruit.66His loy-
alty to Tughril was probably more a matter of expediency than any-
thing else, for he does not seem to have shown any enthusiasm for the
sultan when his cousin Ibrahim Yinal rebelled.67 Nonetheless, he
doubtless commanded too much support among his own Turkmen
followers to be suppressed easily.
The death of Tughril in 1064 brought to the fore the problems of
succession that bedevilled mediaeval Turkish states. According to the
traditional Turkish conception, authority rested in the Seljaq family as
a whole rather than in one particular member of it. Nonetheless,
amongst the members of the family, there might be one who held a
senior position, through having won it by feats of arms and the recog-
nition of the tribesmen. As Rudi Lindner explains, "tribesmen sup-

64
al-Husaynm,Akhbdr...:17; Bundarl,op. cit.:8; Zahir al-Din Nishapuiri,TheSaljziqndma,
ed.
A. H. Morton (n.p., 2004): 14.
65 Lambton, "InternalStructure":218. This attitudepersistedeven much later in the Seljtuq
period (see Bosworth,op.cit.:105).
66
Cahen, "Qutlumushet ses fils":20-21.
67 Ibid.: 22.
220 ANDREW C. S. PEACOCK

ported not the eldest son, but the candidate who best represented their
interests, for their welfare and survival depended upon their chief's
ability to represent them in the search for pasture, plunder or a modus
vivendi with stronger neighbours".i8 While Tughril seems to have at-
tempted to make himself the superior Seljuiq, he still had to recognise
his brother Chaghri, ruler of the eastern part of the Seljiiq empire,
who does not seem to have ever acquiesced in Tughril's pretensions to
superiority.69 On the death of Tughril, whom Chaghri had prede-
ceased, succession to his position as sultan was inevitably disputed.
Initially, his obscure nephew Sulayman became sultan, but he was
swiftly removed in favour of Alp Arslan, another son of Tughril's
brother Chaghri. The accession of Alp Arslan did not go unchal-
lenged, and his cousin Qutlumush now declared himself sultan claim-
ing as his justification that "my father was the best and most senior of
the tribe".70
Alp Arslan successfully defeated Qutlumush, who died while at-
tempting to flee. It is clear, however, that Qutlumush had attracted
substantial support from the Turkmen.7' According to Sibt b. al-Jawzl,
the horsemen alone supporting him numbered some 50,000,72 and in
the eyes of the Turkmen his claim to the sultanate would have been at
least as good as Alp Arslan's, and possibly better. While the 'Iraqyyah
were nominally subjects of Tughril, they had always proved most re-
luctant to accept any of his attempts to exert his authority over them.73
Certainly, amongst the descendants of the followers of Arslan, the
most senior Seljiiq until his imprisonment and death at the hands of
the Ghaznavids, one can imagine that the claim of the son of their late
leader would have had much appeal. It was these followers of Arslan,
the 'IrdqyyahTurkmen, who populated Armenia and Azerbaijan, the
provinces where Qutlumush was active.

68
Rudi Paul Lindner,"Whatwas a Nomadic Tribe?",Comparative Studiesin Societ_
andHistogy
24 (1982):693.
69
RichardW. Bulliet,"NumismaticEvidencefor the Relationshipbetween Tughril Beg and
Chaghri Beg", Dickran K. Kouymjian (ed.), N'earEasternNumismatics, Iconography,Epigraphyand
Histoy:StudiesinHonorof George C.Miles(Beirut,1974):239-6.
70 Rashid al-Din, Cimi'al-Tavdrih: II. Cild,5. Ciiz,ed. A. Ates (Ankara,1960):28.
71 Sibt, Mir'dt77: "wa-indamma ilayhial-Turkndn wa-'l-Atrdk".
72 Ibid.: 110. This figure is doubtlesssomewhat exaggerated,as on Kasravi'scalculations,it

would mean Qutlumushhad a greaternumber of cavalryalone than the entire 'Irdqivyah popula-
tion of Azerbaijan,which is difficultto believe even if we accept that most nomads would have
been mounted. Nonetheless, Sibt's figurescan be understoodas reflectinggreat number of Turk-
menjoined with Qutlumush.
7 Ibn al-Athir,al-Kamil, IX: 386, 389, 507.
 NOMADIC SOCIETY AND THE SELJUQCAMPAIGNS IN CAUCASIA 221

In light of this, the aims of Alp Arslan's campaign of 1064, the year
of his accession, become more comprehensible. The reason the sultan
evinced so little interest in making permanent territorial gains is be-
cause he was concerned less with conquering the Georgians than in
gaining the loyalty of the 'Irdqiyyahalready settled in Caucasia. Like-
wise, it was because this campaign was central to establishing the sul-
tan's authority over the tribesmen that the expedition was accompa-
nied by key figures in the Seljuiq state such as Nizam al-Mulk. That
this Caucasian campaign is related more to Seljiiq internal politics
than disorganised pillaging can be illustrated by considering what is
known of patterns of Turkish settlement in Caucasia in the Seljuiq pe-
riod. Virtually nothing is said about this by the Islamic or Armenian
sources, but fortunately the Georgian Chronicles do give us detailed,
although not complete, information as to where Turks settled.
The Turks that came to Caucasia were, of course, nomads. The
nomadic lifestyle is characterised by transhumance between summer
pastures, called yayldqs, and winter pastures, called qishldqs. Summer
pastures tend to be in the cool highlands, and winter pastures in the
warm lowlands.74 The Georgian Chronicles allow us to trace these
movements in Caucasia with some accuracy (see map). Summer pas-
tures were apparently in the foothills of Ararat, Somxit'i and in the
mountainous province of Tao-Klarjet'i and Savset'i.75 On one occa-
sion apparently, the Turks made use of Tao in winter as well, seeking
refuge from attacks by Georgian armies in this remote and inaccessible
province.76 Tao, however, cannot have been a regular winter pasture
as conditions are far too cold to allow this, but rather just an emer-
gency place of refuge, known to the Turks as a summer pasture.77
Winter pastures, the Chronicle tells us, were in Gac'ianni, around
Samsvilde and along the banks of the Kur and Iori rivers. The chroni-
cler has left a vivid description of the behaviour of these Turks:
"Theywouldsettle... in all thosebeautifulwinterquarters,
wherein winter,as
in the seasonof spring,grassis mowedand wood and waterare foundin
abundance.A multitudeof all kindsof gameexiststhereand thereis every
sortof recreation.In thoseregionstheywouldsettlewiththeirtents;of their

74 Th7eMissionof FriarWilliamof Rubruck, ed. and tr. P. Jackson and D. Morgan (London,
1990):72.
75 Thomson, Rewriting CaucasianHistory:310, 323. According to the Chronicle (: 310), the
Turks spread as far north as Kut'aisi and Kart'li , and their settlement certainly included the
yayldqsof Samc'xe-Javaxet'i,i. e. the region near Axalk'alak'i.
76
Ibid.:324.
77 For the location of some yayldqsin Tao-Klarjet'i, especially near Barhal, see Wakhtang

Djobadze, EarlyMedievalGeorgian Monasteries in HistoricTao,Klaiet'iandgavset'i(Stuttgard1992):

158, 178, 190.
222 ANDREW C. S. PEACOCK

horses,mules,sheep,and camelstherewas no reckoning.Theyled a blessed

existence;theywouldhunt,relax,taketheirpleasure,andtheyexperiencedno
lackof anything.Theywouldengagein commercein theircities,but would
invadeourbordersfortheirfill of captivesandplunder.In springtheywould
ascendthe mountainsof Somkhitiand Ararat.Thus duringsummerthey
wouldhaveeaseandrecreationon the grassandpleasantfields,withsprings
and floweringmeadows.So greatwas theirstrengthand multitudethatyou
could say:<<All
Turksof the whole world are here>>".78

The Georgian Chronicle thus allows us to locate some of the pastures,

but due to its own regional perspective not all of them. However, de-
tailed information about the pastures used by the Mongols and other
Turkmen tribes survive, for Caucasia was favoured by later nomads
too for its summer and winter pastures. It is likely that exactly the
same pastures were used by the Turkmen immigrants of the eleventh
century. Thus, it is probable that Aladag north of Lake Van, a well-
known Mongol yayldq,79was also used as summer pasture by the
Seljuiqs, possibly as an extension of the Ararat pasturelands. Another
popularyayldq was in southern Chaldia, a pasture used by the Aqqu-
yunlu Turkmen in the fourteenth century.80
A comparison of the maps of the campaigns and the pastures is in-
structive (see map). As is evident, the vast majority of military opera-
tions take place around the summer and winter pastures. Areas that
were not pastureland escape comparatively lightly, or even completely.
For instance, the southern shore of Lake Van was a rich region con-
taining many monasteries that one might have thought would have
been an obvious target for plundering. In addition, the major towns of
the old Armenian kingdom of Vaspurakan, only recently abolished,
Van and Vostan, were located here.8"Yet, there is scarcely any record
of Seljuiq attacks on the area despite its history being well documented
in this period. This may perhaps be explained by the fact that much of

78 Thomson, Rewriting Caucasian History:323.

79 T. A. Sinclair,EasternTurkey: AnArchitectural
andArchaeological
Survey(London, 1987),I: 273;
Charles Melville, "The Itinerariesof Sultan Oljeitu, 1304-16", Iran28 (1990): 56 and fig. 1: 58;
John Masson Smith, Jr "Mongol Nomadism and Middle Eastern Geography: Qishlaqs and
Tumens", R. Amitai-Preiss,D. Morgan (eds.), TheMongolEmpireanditsLegacy(Leiden, 1999):42-
3, 47.
80 The Aqquyunlu soon spread out occupying much of
Armenia as ayayldqand the Diyar
Bakr as a qishldq,and conducted campaigns against Georgia. Their expansion thus replicates
many patternsof the Seljulqinvasions.Their originalcentre appearsto have been Sinir near Bay-
burt in Chaldia, where one of their earliest chiefs, Qutlu, is buried, and where he founded a
mosque. See John E. Woods, TheAqquyunlu: Clan,Confederation,
Empire(Salt Lake City, 1998):34,
56. For reasons of space, I have not considered the numerous Seljiiq campaigns in the Diyar
Bakr,but it is likelythat they were likewiseinspiredby a need for the qishldqs
of the area.
81 Sinclair,op.cit.:222ff.
 NOMADIC SOCIETY AND THE SELJDQCAMPAIGNSIN CAUCASA 223

the land on the south of Lake Van were unsuitable for pastures, but
rather are steep, barren, waterless mountains and valleys.82 For this
reason, the Seljtuqswere not attracted to the region. A local Armenian
historian of the Middle Ages strikes an uncharacteristically positive
note when discussing this area, remarking of Amiuk and the island of
Aghtamar, the most important fortress and monastery there, that in
this period of attacks, "For them was accomplished the saying of the
inspired psalmist David: 'The islands shall be happy and all the in-
habitants therein'; they rejoiced in delight according to Soloman's ex-
hortation".83 However, to the north of the lake, on the other hand,
towns such as Ercis, Berkri and Manzikert, all near the southern fringe
of the Aladag pastures, were regularly attacked.
Likewise, both Alp Arslan's campaigns of 1064 and 1067 were di-
rected at areas of knownyayldqs and qishldqs:the summer pastures of
Tao and Somxit'i, and the winter pastures in southern Kart'li and in
Kaxet'i, along the banks of River Kur. Indeed many of the areas at-
tacked in 1064 feature in the Turkish epic, the Bookof Dede Korkut:we
read there of the Turkmen pitching their tents around Surmeli and
Agcakale near Ani, two of the fortresses destroyed by Alp Arslan.84
Earlier campaigns in 1048, 1054-5 and 1056-7 had also concentrated
on areas around the pastures of Tao and Chaldia. The same is true of
the campaigns under Malikshah, which attacked Tao and Samsvilde,
although these campaigns around the pasturelands were accompanied
by an attack on the heart of the Abxazian kingdom, raiding its capital
Kut'aisi. This was probably motivated mainly by a desire for plun-
der.85
Information about eastern Caucasia, meanwhile, is much scantier,
and we are reliant on a few references preserved in an abridged ver-
sion of a local history written in Arabic and preserved in Munec-
cimbasi's eighteenth century compilation, the Jdmi' al-Duwal. As early
as the 1040s the rulers of Sharwan, the Sharwanshahs, had built a wall
around their capital Yazidiyyah to keep out the Turkish tribesmen,
and in 1066 the Seljiiq commander Qarategin attacked Yazidiyyah

82 E. g. ibid.:224.
83 Thomas Artsruni,Historyof theHouseofArtsrunik,tr. Robert W. Thomson (Detroit, 1985):
371-2. According to the same anonymous continuatorof Thomas Artsruni,Van was indeed at-
tacked once. Given the importanceand prosperityof the city, this is surprisinglylittle, compared
with the repeatedattackson Ercisand Manzikert,for example,in the north.
84 Bookof DedeKorkut106. The identificationof the Surmariof Ibn al-Athirand al-Husayni
with Surmeli seems uncontroversial.In identifyingSabidshahrwith Agsakale, I follow Bunyatov
in Sadr ad-Din Husayni,Akhbdr ad-daulat ed. and tr. Z. M. Bunyatov:180.
as-Se1dZukfyya,
85 Thomson, Rewriting Caucasian
History:310.
224 ANDREW C. S. PEACOCK

and Baku.86 Other operations, particularly under Malikshah, were

conducted by his commander Savtegin, who was responsible for much
of the conquest of the region, which had been granted to him as far as
the Bab al-Abwab (Derbend) as a fief (iqtd') by Alp Arslan.F The in-
adequate nature of the evidence makes firm conclusions are hard to
draw here, but certainly the area stretching south of Yazidiyyah, the
Mughan, and adjacent territories in Arran, were important qishldqsin
Mongol times, so it is unsurprising that the Turks should have been
attracted to the area in the Seljuiqperiod.88
The significance of the connection between these military cam-
paigns and the location of pastures is not made explicit by the sources.
However, as discussed above, one of the main duties of the tribal chief
was to secure plunder and pasture for his followers. An account of one
of Ibrahim Yinal's campaigns against Caucasia shows this leading
member of the Seljuiq family functioning in exactly the same way. Ac-
cording to this report
"A large group of Turkmen from Central Asia came to Ibrahim, and he said
<<Myterritoriesare too small for you and to support your needs. The best
thing is to go and attackAnatolia, fight in God's path, plunder, andI will come
inyourwakeandasszstyou>>".

Thus, the role of Ibrahim was to assist these nomads obtain the
lands they needed to be able to survive. Analogously, in later Seljiiq
times, the provision of pasturage was one of the principle tasks of the
shi4naor government official appointed over the Turkmen.90 If we
consider the Seljuiq campaigns in Caucasia in the light of such exigen-
cies of nomadic society, they become much more comprehensible than
the picture of apparently mindless violence and destruction presented
by the sources. Although, as Ibrahim's promise suggests, there was a
place for plundering in these campaigns, they were to a large extent
concerned with securing pasture land for the Seljiiqs' nomadic follow-
ers. This explains why the Seljuiqsdestroyed so many fortifications and
cities without ever intending to occupy them permanently: they were
interested not in the cities but in the pastures around them, but to se-
cure control of the pastures they had to ensure the cities could not be
used by anyone else to threaten their control of the surrounding

86
Minorsky,A HistoyofSharvan andDarband: 33, 35-37, 53-55.
87 Minorsky,Studies in Caucasian
Histoy:24-5.
88 Smith, "MongolNomadismand Middle Eastern
Geography":45-8.
89
My italics;Ibn al-Athir,al-Ka-mil,
IX: 536.
90 A. K. S. Lambton, "The Administrationof Sanjar'sEmpireas Illustratedin the 'Atabat
al-
kataba",BSOASxx(1957):382.
 NOMADIC SOCIETY AND THE SELJICQCAMPAGNSm CAUCASA 225

countryside. The Seljaqs' policy of burning down settlements leaving a

wake of destruction behind may have been barbarism, but it was not
mindless barbarism. A passage in Dede Korkutrelates such an event,
telling of a peaceful hunting trip in Samc'xe during which the Turkish
huntsmen were attacked by "sixteen thousand black-mailed infidels"
from Axalc'ixe.9' The details may be legendary, but it reflects the dan-
gers of the Christian towns to the nomadic Turks.
As Ibrahlim indicates, it was not enough for the tribal leader in
this case members of the Seljuiq family-to give permission to their
followers to seek new pastures, but rather he had to assist them ac-
tively. As Lindner describes, "[t]he tribe owed its success to the chief's
talents for finding pasture or providing plunder".92 Thus such senior
members of the Seljiiq family led these campaigns because to do so
was necessary to secure their positions with their followers and to un-
derline their credentials as leaders of the tribe, often against competi-
tion from rival Seljuiqs. This suggests that the Seljuiq dynasty main-
tained much closer links to its tribal roots than is often imagined, and
that the campaigns in the crucial region of the Caucasus were dictated
by the requirements of a nomadic, tribal society the need for pas-
tures and disputes over primacy within the leading family. Plunder, of
course, also played a major role in these expeditions: Ibn al-Athir tells
us that Armenia was raided so frequently that the price of a beautiful
slave-girl fell to the low price of five dinars.93
Nonetheless, the Seljiiqs did from time to time conquer and keep
towns in Caucasia, and required local rulers to pay tribute to them.
Indeed, with the settlement of these large numbers of Turks in the re-
gion, Caucasia was a strategically vital area, although as indicated
above the Seljuiqs often maintained the established Muslim dynasties
of the region as their vassals. Muhammad Tapar, Seljiiq sultan in the
early twelfth century, had Ganja as his capital before he successfully
established control over Iran and Iraq. Fortunately, we have a first
hand view of what it was like to live in Seljaq Caucasia in the writings
of one of the sultan's officials, Mas'uid b. Namdar, who also served the
local dynasty of the Sharwanshahs. Mas'uid was a mustawfi(revenue of-
ficial) in Baylaqan, a town about a seventy miles south east of Ganja,
Muhammad's capital. He depicts the chaos of life in early twelfth
century Caucasia, with two competing Turkish commanders trying to

9' BookofDedeKorkut: 90.

92
Lindner,"Whatwas a Nomadic Tribe?":700.
93 Ibn al-Athir,al-Kamil,IX: 389.
226 ANDREW C. S. PEACOCK

take charge of the town, one with a document of investiture for it from
the Sharwanshah, the other presumably from the Seljuiq sultan;94the
constant rivalry between Christian andJewish groups;95and the huge
power of "the rabble" (al-ghawghd')as Mas'ad describes the masses and
their leaders.96Nowhere is there any sign of the Seljaq authorities be-
ing remotely interested in intervening in the chaos, and much of
Mas'ud's writings are a lament at his helplessness in the face of the
rabble. The fact that the Seljaqs were unable or rather uncon-
cerned to establish order in a major city so close to the sultan's
capital is deeply revealing of their priorities, which remained more
deeply rooted in Turkish tribal society for rather longer than is often
assumed. Of course, I do not wish to suggest that tribal life and kinship
disputes were the sole concern of Seljuiq rulers by any means and ex-
plain everything about the Seljuiq state and its formation. Numerous
other factors played a part, such as war with the Fatimids and the
need for the Seljuiqs to legitimise themselves to the Persian and Arab
populations over which they ruled. However, tribalism defined Seljuiq
society and often determined the policies of the sultans.

The Seljaqs' policies towards Caucasia underline the deep suspicion

with which we must treat the Persian and Arabic accounts that extol
the Seljuiqsultans as heroic fighters for Islam and monarchs in the Per-
sian tradition. Rather, while they may have been happy to be por-
trayed as such to some audiences in order to legitimise themselves,
much of their mindset remained that of the tribesman. After all, for all
that writers sought to portray the sultan Tughril as an Islamic mon-
arch who had saved the Islamic world from Shl'ism, one must re-
member that the same Tughril was happy to offer to surrender his
prize of Baghdad to his Shi'ite rival al-Basasiri in exchange for the
mention of his name in the khutbahat prayers.97The bombastic style
that much of the historical writing of the Seljuiq period employs seeks
to disguise the unpleasant reality that the new masters of the eastern
Islamic world were driven by very different preoccupations from the
ideals of justice and Perso-Islamic rule that their servants extolled. Nor
was their interest in areas such as Anatolia and Caucasia inspired sim-
ply by a desire to divert their less civilised followers away from the

94 Mas'uldibn Namdar, Majmi'atqisaswa-rasd'il wa-ash'dr/Sbomik pisemi stixov,ed.

rasskazov,
B. M. Beilis(Moscow, 1970),text: 111ff/f. 107bff.
95 Ibid., text: 112/f. 108b.
96 Ibid., text: 47ff/f. 232aff.
97 M. Canard, "al-Basasi-ri" EI2, : 1073-1075, esp. 1074.
 NOMADIC SOCIETY AND THE SELJQQCAMPAIGNSIN CAUCASIA 227

central Islamic lands. Rather Caucasia was crucial to enabling the

sultans to fulfil their traditional role as tribal chiefs by providing pas-
ture and plunder for the Turkmen, who, in Nizam al-Mulk's words,
had a claim on the Seljuiq dynasty. The numerous interventions in
Caucasia by sultans and leading members of the Seljiiq dynasty illus-
trate clearly the way, in which nomadic priorities remained at the
heart of the Seljuiqstate.

APPENDIX

Chronology attackson Caucasiato thereignofMalikshah98

of majorSeljiuq

1029 'Iraqiyyah Turkmen penetrate Azerbaijan, using Urmiyyah

as a base to attack Armenia (Ibn al-Athir, al-Kdmil, IX:
383).
1038 Tughril, who declares himself sultan in this year, sends
Qutlumush to take charge of Armenia and Azerbaijan (al-
ilusayni, Akhbdr...:17). Georgians forced to abandon siege
of Tbilisi due to Turkmen advance (Ibn al-Athir, al-Kamil,
IX: 457).
1040 Massacre of 'Iraqiyyah Turkmen by Vahsiidan the Raw-
wadid, prompting further migrations (Ibn al-Athir, al-Kdmil,
IX: 384-385).
1042-3 Turkish amirs attack Azerbaijan and Armenia; probably a
reference to the events of c. 1045 (AzimiTarihi,Arabic: 3).
c. 1045 Turks (according to some sources, commanded by Qutlu-
mush) capture the commander of Erci? and take him to
Khoy where he dies (Matthew, Chronicle:74; Artistakes, Re-
cit: 63ff; Bryennius, Historiarum:96-99; Scylitzes, Empereurs:
447/tr. 372; Zonaras, Epitome, III: 636-7). Sharwanshah
fortifies Yazidiyyah against Turks (Minorsky, A Histoiy of
SharvanandDarband:33).
1046-8 Two Byzantine attacks on the Shaddadid city of Dvin, ap-
parently beaten off with Seljuiq assistance after Tughril sent
Ibrahim Yinal and Qutlumush to the rescue (Azimi Tarihi,s.
a. 437/1045-6 and 438/1046-7, Arabic: 5-6).

98 The references given below are not

intended to be comprehensive, but rather are indica-
tive of the main sources, concentrating on the Islamic sources.
228 ANDREW C. S. PEACOCK

1047 Qutlumush besieges the Shaddadid capital, Ganja, for a

year and a half (AzimiTarihi,Arabic:6).
1048 Campaignled by IbrahimYinal againstAnatolia, reaching
to within fifteen days' march of Constantinople.Attackson
Chaldia, Tao, Xordzean, Taron, Trebizond, Manzikert,
Erzurum (Ibn al-Athir,al-Kdmil,IX: 546; Aristakes,R6cit:
67ff. According to Matthew of Edessa (Chronicle: 76), this
campaigntook place in the followingyear and Ibrahimwas
accompaniedby Qutlumush.
c. 1049 War between Tughril and Ibrahim Yinal supported by
Qutlumush leading to the defeat of Ibrahim. Byzantines,
aided by Georgianmarcherlord LiparitOrbeliani,advance
eastwardsrecapturingKaputru castle near Ararat. Liparit
falls into the hands of the Turks. (Ibn al-Athir,al-Kamil,IX:
557; Matthew, Chronicle: 89ff; Thomson, Rewriting Caucasian
Histogy: 294).
1053-4 Qutlumush attacks Kars, massacringits population (Azimi
Tarihi,Arabic:10;Aristakes,Recit:83-4).
1054-5 Tughril's campaign against Anatolia and Caucasia, cap-
turing Tabriz and gaining allegiance of Shaddadids.Cap-
ture of Berkriand Erci?,while Manzikertand Erzurumat-
tacked, and expeditions to Chaneti, Mt Barhal, and two
unidentifiedplaces in Caucasia, "the Abkhaz fortress",the
foot of the Caucasus,as well as an expeditionsouthwardsto
the Anti-Taurus(Ibn al-Athlr,al-Kamil,IX: 597; Matthew,
Chronicle: 86-88; Aristakes,Recit:84-88).
1056-7 Seljulq attacks on Kemah, Koloneia, Ispir, Khordzean,
Chaldiaand Chaneti(Aristakes,Ricit: 107, 116-117).
1058-9 Attackson Taron and Sasun(Matthew,Chronicle: 93).
1059-60 Attacksby Seljuiqamirs on Armenia, reaching as far Sivas
(Matthew,Chronicle: 94-97).
1063 Byzantines defeat a Seljaq army at Erzurum, killing its
commander,Yuisuf(Matthew,Chronicle: 101).
1063 Death of Tughril,temporaryaccessionof Sulayman.
1064 Rebellion of Qutlumushand its suppressionby Alp Arslan.
Alp Arslan'scampaign against Georgia, attackingSuirmeli,
Sapidshahr(A- akale?),Marmashen,Axalk'alak'i,Ani and
other unidentified locations (Matthew, Chronicle:101-104;
Aristakes,Recit: 134-136; Ibn al-Athir,al-Kdmil,X: 37-41;
 NOMADIC SOCIETY AND THE SELJOQCAMPAIGNSIN CAUCASIA 229

al-Iusayni, Akhbdr:34-40; Thomson, RewritingCaucasian

History: 298-299).
1066-7 Turkish commander Qarategin attacks Sharwan, Yazidly-
yah and Baku(Minorsky,Histogy:36).
1067 Alp Arslan'ssecond Caucasiancampaign,subjugatingking-
dom of Kaxet'i-Heret'i,the king of which convertsto Islam.
Attemptto penetrateinto CaucasusMountainsthwartedby
winter, while Tiflis is sold by Alp Arslanto the Shaddadids
(al-Husayni,Akhbdr...:43-6; Sibt, Mir'at: 136; Thomson,
Rewriting CaucasianHisto?y:300-301).
1071 Defeat of the Byzantines by Alp Arslan at Manzikert (for
Muslim sources see FarukSumer and Ali Sevim, IslamKay-
naklarzna goreMalazgirtSava,z(Metinler ve Qevirileri)
(Ankara,
1988).See also Vryonis,"PersonalHistory").
1072 Assassinationof Alp Arslanin CentralAsia and accessionof
Malikshah.
1075 Seljuiqsabolish dynasty of Shaddadidsof Ganja (Minorsky,
Studies:25.).
1080 Malikshah's campaign against Georgia (Al-Husayni,Akh-
bdr...:63; al-Bundarl,Zubdatal-Nugrah: 140; Mohammed en-
Nesawi, Histoiredu SultanDIelaled-DinMankobirti, Princedu
Kharezm, ed. and tr. 0. Houdas (Paris,1895), I:175, II: 291;
Thomson, Rewriting Caucasian History:309ffW.
1086 Campaign in Caucasia led by amir Buizan.Bosworth("Po-
litical and Dynastic History":95) indicates that Malikshah
appearedin the region in person, but there is nothing in the
sources he cites to support this. (Ibn al-Athir,al-Kamil,X:
287; Matthew,Chronicle: 155).
230 ANDREW C. S. PEACOCK

.<~~C fC~ ~~~~~~~~~~~~

12i

U,

co(

4VX1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1
0~~~~
mv~~~~~~~
uj~~~~~~~~~~~~~~~(

/ (

C/) 7m ~ ~ ~