Boyle J.A. The Cambridge History of Iran, Volume 5: The Saljuq and Mongol Periods

Подождите немного. Документ загружается.

INTERNAL

STRUCTURE

OF THE SALJUQ EMPIRE

220

authority

was weakened. Alp-Arslan may well have realized that some

substitute had to be found for the former

tribal

loyalties to which the

Saljuqs

owed their position, and that if he was to keep even a limited

control

over the members of the family, then a standing army loyal to

himself was necessary.

458/1065-7

Alp-Arslan determined to appoint Malik-Shah as his

vali

'ahd, and he had his name included in the khutba

after

his own.

This was probably normal practice. At the same time as this nomination,

Alp-Arslan

divided

part

of the kingdom among his relatives in order to

abate

any

quarrel

with the

succession

of Malik-Shah (see p. 23 5 below).

In spite of this, various members of the family disputed his accession

on the death of

Alp-Arslan.

Qavurd, the

ruler

Kirman,

wrote to

Malik-Shah

stating that it was more fitting for him to succeed, on the

grounds that he was Alp-Arslan's eldest brother while Malik-Shah was

only a young son.

Malik-Shah, who was then aged about eighteen,

replied

that a brother did not inherit if there was a son. This, however,

was Islamic and not

tribal

law. Qavurd made a determined eifort to

seize

the throne, but he was defeated and killed. Tekish, to whom Balkh

and Tukharistan had been assigned, rebelled twice, in

473/1080-1

and

477/1084-5;

on the second occasion he was captured and blinded.

Toghril b. Inal also made an abortive attempt in

482/1089-90

establish his independence in the neighbourhood of

Nasaf.

Syria,

which Malik-Shah had assigned to Tutush in

470/1077-8,

the sultan's

nominal authority

seems

to have been recognized, though he twice had

to intervene in person.

Tutush came himself to pay his respects to

Malik-Shah

in Baghdad in

484/1091.

Malik-Shah

nominated Ahmad, the eldest son of Zubaida Khatun,

as his vali 'ahd, but he died in

481/1088-9.

A new heir was not nomi-

nated, and when Malik-Shah died four years

later,

civil war ensued.

Terken Khatun, his last wife, a Qarakhanid princess, succeeded in

persuading the caliph to have the khutba read in Baghdad in the name

of her son Mahmud, in

return

for which she handed over Abu'l Fadl

Ja'far,

the son of Arslan Khatun and al-Muqtadl, who had been with

Malik-Shah

since his mother had returned from the caliph's court.

(Ja'far

died the following

year;

and Terken Khatun herself died in

487/1094.)

Meanwhile the Nizamiyya

mamluks

(slaves) carried offBerk-

Akhbdr al-daulat

al-Saljuqiyya,

ed. Muhammad Iqbal (Lahore, 1933), p. 56. Sanjar

later

opposed

the

accession

of Mahmud b. Muhammad on the

grounds

of his youth.

See H. A. R. Gibb, The

Damascus

Chronicle

the

Crusades

(London, 1932), pp. 20-1.

INTERNAL

STRUCTURE

OF THE

SALJUQ

EMPIRE

221

Yaruq,

who had

been

imprisoned in Isfahan on Terken Khatun's

orders,

and proclaimed him sultan. He was at this time a boy of thirteen

and in no position to assert himself as the leader of his people.

Isma'il b. Yaquti, his maternal uncle, who was in Azarbaijan, was

persuaded by Terken

Khatun to

side

against Berk-Yaruq. He was

eventually

killed by the Nizamiyya mamluks, as was Taj al-Mulk

Abu'l

Ghana'im, who had collaborated with Terken Khatun. Tutush

and Arslan Arghun, both brothers of Malik-

Shah, also rebelled in

Syria

and Khurasan respectively. The former was defeated in

487/1096,

and the latter three years

later.

Berk-Yaruq's half-brother Muhammad

rebelled in

492/1098-9.

After

many vicissitudes, Berk-Yaruq established

slight supremacy in

497/1103-4

at the

cost

of much disorder through-

out the country and a decline in the prestige of the sultanate. By the

terms

of the peace concluded between them, Muhammad's status was

virtually

that of ah independent

ruler.

On his deathbed in

498/1105,

Berk-Yaruq

nominated his son Malik-Shah as his vali

ahd, but

although the khutba was read in his name in Baghdad, Muhammad

soon

succeeded

in establishing himself as sultan. During the reign of

Muhammad, his full brother

Sanjar

(whom Berk-Yaruq had sent with

his atabeg Qumach to

Khurasan in

490/1097)

was nominally the

ruler

Khurasan

on behalf of Muhammad, but he was

virtually

independent.

On the death of Muhammad,

Sanjar

defeated Mahmud b. Muhammad,

whom Muhammad had nominated as his vali

ahd at Saveh

^1513/1119,

and he then made himself sultan. He did not, however, move to a more

central

position but reinstated Mahmud in those districts which he had

held in western and southern Persia and

Iraq,

while he himself returned

to Khurasan. Mahmud and his successors are

referred

to in the sources

as sultans, but although they enjoyed a certain measure of independence,

their

status, until the death of

Sanjar,

was only that of maliks. It is not

clear

why

Sanjar

remained in Khurasan: he may not have had any

personal following outside that province, or he may have considered it

unwise to absent himself permanently from the eastern frontiers

since

the tribes of Central Asia were again in a state of unrest and pressing

in upon the Saljuq empire; or it may be that he was influenced by the

wishes of his mother, who was Mahmud's grandmother and who is

alleged to have persuaded him to make peace with Mahmud.

What-

ever

the causes, the arrangement was unsatisfactory.

Sanjar

was

forced to intervene in the western provinces on various occasions,

and was unable either to restrain the increasing ambitions of the

INTERNAL

STRUCTURE OF THE SALJUQ EMPIRE

222

amirs

and atabegs or, ultimately, to prevent the disintegration of the

empire.

Geographically the Saljuq empire was divided into provinces which

corresponded broadly with the provinces that had formed the Sassanian

empire. An elaborate system of roads, which also went back to the

pre-Islamic period, radiated from Baghdad to serve not only the

movement of armies but also that of merchants. The most famous of

these was the great Khurasan trunk road

going

east from Baghdad via

Kirmanshah,

Hamadan, Ray, Nishapur, and

Marv

to the frontier towns

on the

Jaxartes.

Cross-roads branched off from Kirmanshah to Tabriz

and Ardabil; from Hamadan to Isfahan; from Ray to Zanjan and

Tabriz;

and from Nishapur roads went to Tabas, Qa'in, Herat, and

Sistan.

Another route led south-east from Baghdad to Basra, from

whence there lay a road to Ahvaz and Shiraz, where roads from Isfahan

and Ray, from Yazd and Nishapur, and from

Sirj

an, Kirman, and

Sistan all converged.

The administrative divisions of the empire did

not correspond precisely with the provincial divisions, although when

the empire began to fragment it tended to break up into geopolitical

units centred on the major provinces of Khurasan, Azarbaijan, Iraq,

Kirman,

and Fars. Administratively the empire fell into two broad

divisions: an area directly administered by the sultan's divan, and an

indirectly

administered area. In general, the former tended to increase

up to the end of the reign of Malik-Shah and subsequently to decrease.

The indirectly administered area was alienated from the direct control of

the sultan's divan and

assigned

to the amirs and others as iqta's (see

235 below).

These

did not correspond to a fixed area, and they varied

greatly

size.

They tended to be smaller than the geographical

provinces and to be centred on the most important town of a district.

The main centres of the empire under Toghril Beg at the beginning

of his rule were Nishapur and Ray, the latter

being

his capital for a time.

In his final years he spent much time in Isfahan, his chief residence for

some

twelve years, where he expended on public buildings and

improvements a sum

exceeding

500,000

dinars.

Nasir-i Khusrau, who

passed through Isfahan in

444/1052,

stated that it was the most populous

and flourishing city that he had

seen

in Persian-speaking lands.

Royal

See

further,

G. Le Strange, The Lands of

the

Eastern

Caliphate

(Cambridge,

1905).

E. G.

Browne,

"Account of a

rare

manuscript

history

Isfahan",

J.R.A.S. (1901),

pp. 667 ff.

Safar-Ndma, ed. C.

Schefer,

Persian

text

(Paris,

1881), p. 92.

INTERNAL

STRUCTURE OF THE SALJUQ EMPIRE

223

patronage continued under Alp-Arslan, who was greatly pleased with

Isfahan and treated its people with marked favour. Malik-Shah

chose

as his capital and did much building in the city and the neighbourhood.

Marv,

too, enjoyed a special position, as the capital of Chaghri Beg

Da'ud and later of

Sanjar;

and Malik-Shah also built a suburb there

called Panj Dih. Toghril Beg built a new quarter at Baghdad on the

Tigris, which included a Friday mosque and bazaars, and was sur-

rounded by a wall. There Malik-Shah built his mosque known as

Jami

al-Sultan, the foundations of which were laid in

485/1092.

The sultans spent much of their time on military campaigns and

travelling

about their empire. The

dargdb,

or court, did not remain in

the capital but was to be found wherever the sultan was. Government

officials

such

as the vizier, accompanied him, and presumably the

"private"

treasury often went with him as well. To what extent a

distinction was made between the sultan's private treasury and the state

treasury

is not clear, nor do we know what monies were paid into the

former.

The vizier was in charge of the state treasury, but there appears

to have

been

a special treasurer directly under the sultan in charge of the

private

treasury. Malik-Shah is said to have built a fortress outside

Isfahan, known as Shahdiz (this the Batinis subsequently captured), in

which his treasury, armoury, and young slaves in training were kept

when he was absent on campaigns.

Alp-Arslan

and Malik-Shah also had stores of money in strongholds

in different parts of the kingdom, chiefly to facilitate their military

expeditions. Alp-Arslan is said to have had a store in the fortress of

Giv

near Farahan so that if he was travelling between Khurasan and

Iraq

he could obtain from there anything he needed. On one

occasion

when he reached Farahan on his way to Anatolia, he is reported to have

taken from it one million dinars for the

expenses

of the expedition.

Malik-Shah

had similar strongholds: one

such

was the fortress of

Quhandiz near Nishapur.

Sanjar's

treasury was apparently kept in

Marv,

his capital. When he was absent fighting the Qara-Khitai in

532/113

7-8,

the

Khwarazm-Shah,

Atsi'z,

captured

Marv

and took

Sanjar's

jewels.

Subsequently

Sanjar

invaded Khwarazm and recaptured them.

The sultan's wives and womenfolk also frequently accompanied him

Ravandi, Rabat

al-sudur,

ed. Muhammad Iqbal (Gibb Memorial

Series,

new

series,

vol.

11,

London, 1921), p. 156.

Nasd'ib-Ndma,

attributed

to Nizam al-Mulk; MS. in my

possession,

fol. 30,

cols.

a-b.

Cf. Akhbar

al-daulat

al-Saljuqiyya,

Bundari, p. 257.

INTERNAL

STRUCTURE OF THE SALJUQ EMPIRE

224

on expeditions. In

536/1141,

when

Sanjar

was defeated on the Qatvan

steppe by the Qara-Khitai, his wife Terken Khatun was captured, but

she was subsequently released. Some of the sultan's wives wielded

unusual influence. Toghril

Beg's

wife, who died in Zanjan in 452/

1060-1

and was buried in Ray, was, according to Ibn al-Jauzi, a wise

woman to whom To

ghril Beg entrusted his

affairs.

Terken Khatun

played an active

part

in the struggle for the

succession

after her husband

Malik-Shah's

death. Some of the wives had their own divans and

viziers,

and were women of substance;

Sanjar's

mother, for instance,

had her own mamluks.

It appears to have

been

usual for the sultans'

wives to have personal estates. Simirum was

part

of the pension of

Guhar

Khatun, Muhammad b. Malik-Shah's wife. There are also many

instances of a sultan's granting iqta's to Saljuq women and to women

who married into the Saljuq house. For example, when Toghril Beg

married

the caliph's daughter in

454/1062,

assigned

to her Ba'quba

and all that his late wife had held in Iraq.

After

putting down the

rebellion of Qara-Arslan in Kirman and reinstating him in the province,

Alp-Arslan

allocated to Qara-Arslan's daughters, in response to their

father's

request,

100,000

dinars, iqta's, garments, and money for the

expenses

of marriage celebrations.

Marriage

alliances played an important

part

in Saljuq policy. In ad-

dition to the marriage alliances made with caliphs, there were numerous

marriages

between Saljuqs and members of former local ruling

houses

such

as the 'Uqailids, the Buyids, Kakuyids, Mazyadids, and others.

Members of the bureaucracy, the religious classes, and learned men

frequented the court, but except for the sultan's private household, it

was essentially a military court, composed, from the reign of Alp-

Arslan

onwards, mainly of amirs and members of the standing army

('askar).

Those

amirs who held "administrative" iqta's (see p.

233

below) often found it necessary to have their agents at court to keep them

informed of current developments and to watch over their interests.

In addition, there were numbers of hostages taken from various tribal

groups—Turkmen,

Kurds,

Shabankara, and others—and from former

ruling families, and they remained at court as sureties against the

rebellion of their relatives.

The amirs were mainly

mamluks

(slaves) or freedmen. Their status was

Al-Munta%am

vol. vni, p. 218.

Al-Muqarrab Jauhar had

originally

been

her mamluk and was

transferred

to Sanjar on

her

death

in 517/1123 (Bundari, p. 250).

INTERNAL

STRUCTURE OF THE SALJUQ EMPIRE

not originally equal to that of the free-born, but they could and did

attain

to the highest positions in the state, and then they themselves

aquired

mamluks. In some cases both slaves and freedmen married into

the

royal

house. The atabegate is

further

proof that no social stigma

attached to the slave or freedman from the time of Malik-Shah on.

Whether

this was the case in the early days of Saljuq expansion, how-

ever,

is not clear. Isra'il b. Saljuq, when he sent a message to his family

urging them to fight for the kingdom of Mahmud b. Sebuk-Tegin, is

alleged to have spoken of him with contempt as the son of a freed slave

(mauld).

By Islamic law the

possessions

of a mamluk escheated to his

master;

in practice, however, on the death of a

royal

mamluk the sultan

often granted his

possessions

to one of the mamluk's descendants.

Freedom was bought by the mamluk, given by the sultan, or else

usurped;

and if a mamluk attained to a position of power he became

virtually

free.

Many of the later sultans, who came to the throne in

extreme youth, were dominated by the amirs. Even

Sanjar,

if Bundari

is to be believed, fell under the influence of successive mamluks, whom

he singled out for special favour. At least two of them are alleged to

have

been murdered on his orders

after

they fell from favour.

The amirs were divided broadly into three groups: those at the

sultan's court; "landed" amirs, i.e. those holding "administrative"

iqta's;

and "wandering" amirs, who owed no permanent allegiance to

anyone, but moved about the empire serving different leaders and taking

possession

of districts as opportunity arose. The grouping of the amirs

was not constant, however. Those at court frequently changed, and the

distinction between a "landed" and a "wandering" amir was not

fixed; thus a wandering amir who usurped

possession

of a district and

was then confirmed in his

possession

of it by the sultan became a

landed amir. As a class the amirs had no community of interest.

Apart

from

rare

instances their jealousies prevented close co-operation among

them except for a limited time.

Just

as the sultan used marriage alliances

to strengthen his position, so the amirs by intermarriage among them-

selves and with the ruling house sought to strengthen their own

position. Their power, relative to that of the sultan on the one hand

and the bureaucracy on the other, increased noticeably

after

the reign

of Malik-Shah. They were quick to resent any attempt to reimpose

control

over them, and their disobedience

after

the accession of

Mahmud b. Muhammad was a marked feature of the times.

Ravandi, p. 91.

5 вен

INTERNAL

STRUCTURE

OF THE

SALJUQ

EMPIRE

The

chief

official of the court until towards the end of the reign of

Muhammad b. Malik-

Shah was the

vakJl-i

dar,

who appears to have

been

a kind of intermediary

between

the sultan and the vizier. According to

Bundari his

position

was more privileged than that of the

hdjibs

(chamberlains). He had to be "eloquent and able to triumph in diffi-

cult situations in matters of

speech,

independent in

establishing

proof if

necessary; [his words] free from banality and

distinguished

by grace,

and able to understand the different

moods

and characteristics of the

sultan".

His precise relation to the vizier, the head of the divan, is not

entirely clear. Ibn Balkhi states that in his time the vakil-i dar was the

vizier's deputy.

Towards the end of Muhammad b. Malik-Shah's

reign the vakil-i dar was replaced by the

amir

hajib,

who was a member

of the military

classes

and not of the bureaucracy. Although this

change

was apparently provoked by the

incompetence

of a certain Zaki,

who had

been

appointed vakil-i dar by Sa'd al-Mulk Abu'l Mahasin

Avaji,

the vizier of Muhammad b. Malik-Shah, it may

signify

decline

in the

position

of the vizier, as well as the increased militarization of

the state.

Nizam

al-Mulk describes the functions of the ha

jib

in terms of

those

of a court official. But

since

the court was a military court, the amir

hajib was normally a Turkish amir and the men under him military

slaves. He was therefore concerned with the maintenance of military

discipline

as well as with court ceremonial. He

became

the

most

important official at court, while the

amir

haras

(chief

of the guard)

and the

jdnddr

(chief

executioner) ranked after him. Bundari states that

the amir hajib (or

amir4

bar)

regulated the

audiences

of the sultan and

transmitted the sultan's

commands

to the vizier.

'Ali b. 'Umar, the

amir hajib of Mahmud b. Muhammad, eventually

became

'aridal-jaish

(muster-master) of the army. Many of

those

who

held

the office of

amir hajib were

among

the

most

powerful amirs of the day.

There were a series of other officials at the court,

such

as the

dkhur-

sdldr

(master of the horse), who

looked

after the royal stables.

Sanjar

appears to have had

extensive

herds of horses, as probably did many of

the other sultans; and there was

also

khwdn-sdldr,

who was in charge

of the royal kitchens. The latter was no small charge, for tribal

custom

demanded

that the

ruler

should

keep

open

table, and this involved

the daily provision of

food

for large numbers. Alp-Arslan was stated

to have fifty head of

sheep

slaughtered daily, which, together with other

Bundari, p. 86.

Fdrs-Ndma,

p. 91.

P. 107.

226

INTERNAL

STRUCTURE

OF THE

SALJUQ

EMPIRE

227

15-2

food, were eaten by the amirs and the poor. Under

Sanjar

the Ghuzz

paid an annual tribute of

24,000

sheep to the

royal

kitchen.

Among the sultan's most important functions—and in this the

traditions

of the steppe and of Islamic government coincided—was to

judge.

Both traditions required that he should be accessible to his

people. His chief medium as a judge was the

mazalim

court, which,

according to Islamic theory, it was his prerogative to hold. This court

was also the main channel of contact between the sultan and his subjects,

though its effectiveness as

such

was limited. The procedure of the court

and its rules of evidence differed from those of the shar'l courts: i.e. it

was necessary that the man who presided over the mazalim court

should

possess

the power to exercise his functions and to apply the

rules

of justice, which was not the case with the qadi, who presided

over

the shar'I courts. The

first

reference to a mazalim court held by a

Saljuq

sultan is to the one held by Toghril in Nishapur in 429/1038.

Malik-Shah

is also alleged to have given justice personally.

Nizam

al-Mulk

maintained that it was indispensable for the

ruler

to hold a

mazalim court twice a week to exact redress from the unjust, to dispense

justice, and to listen to the words of his subjects without an inter-

mediary.

This recommendation, however, was based on expediency

rather

than on a love of justice. "Always", states

Nizam

al-Mulk,

"there

will be many persons at the court demanding redress for

injustice, and if they do not receive an answer they will go away, and

foreigners

and envoys who come to the court and see this complaining

and disturbance will think that great tyranny takes place at this court."

The majority of the Saljuq sultans, however, in all probability

delegated their function of presiding over the mazalim court to the

vizier

or the qadi and in the provinces to the great amirs and Saljuq

maliks who held large areas as " administrative" iqta's and often in

turn

delegated their functions to their representatives. Thus under the

Saljuqs

as under earlier

rulers

the mazalim court became, not an ex-

ceptional appeal to the sultan in person, but an everyday application to

his representative to be dealt with according to a settled practice.

The

general tendency was for the mazalim jurisdiction relative to the shar'l

courts

to extend. Many of the

cases

coming

before the mazalim court

were

probably concerned with the collection of taxes and general

Siyasat-Nama,

p. 10.

Ibid.

p. 207.

C£

F. Amedroz,

The

Mazalim

Jurisdiction in the

Ahkam

Sultaniyya

Mawardi",

J.R.A.S.

(1911),

p. 655.

INTERNAL

STRUCTURE OF THE SALJUQ EMPIRE

228

litigation. It also dealt with

cases

against government officials. Its

decisions

were carried out by the shahna and his officials, or by the

military.

Another

of the sultan's important functions was to defend his

country

and his people in war. In this, too, the tradition of the steppe

and of Islamic government coincided. The first three sultans led the

army

in person.

After

the death of Malik-Shah, with the

accession

young boys to the throne, this was not always the case. As a result the

ties between the sultan and the army were weakened, and the soldiers

tended more and more to give their loyalty to their immediate com-

manders

rather

than to the sultan. Desertions in the later period became

relatively

frequent. Moreover the jealousies and quarrels of the amirs

detracted from the efficiency of the army as a fighting force. The core

of the standing army was formed by mamluks and freedmen. On

campaigns they were joined by contingents furnished by the amirs and

by tribal auxiliaries. The armies of the amirs were also in many

cases

composed

round a nucleus of slave troops. The mamluks and freedmen

of the amirs and others, if they did not pass on the death of their

masters to their heirs, were sometimes incorporated into the royal

'askar,

forming a division known by the name of their late master.

Throughout the Saljuq empire there were also bodies of unemployed

soldiery, who were ready to join the standard of any leader in the hope

of plunder. Their existence facilitated the assembling of an army at

short

notice, but it also made easy the rebellion of discontented maliks

and amirs.

For

the most

part

the mamluks of the standing army were Turks who

had

been

captured or bought on the eastern frontiers of the Islamic

world.

A number of Georgians, Greeks, and Armenians, who had

been

captured on the western frontiers—or, like

some

of the Turks,

were

the children of

such

captives—were also enrolled in the standing

army.

Many of the mamluks were carefully trained to fulfil their

various

functions, which were administrative as well as military.

The most important group in the army after the Turks, however, were

the Dailamites, who were chiefly infantry, whereas the Turks were

cavalry.

Nizam

al-Mulk advocated that the army should be "mixed",

half

Turkish and half Dailamite. In practice, although the army was

composed

of different elements, the Turkish greatly predominated.

Nizam al-Mulk

states

that

the

system

training

that

had

prevailed

under

the

Samanids

had

fallen

into

disuse

by his day (Siydsat-Ndma, p. 94).

INTERNAL

STRUCTURE

OF THE

SALJUQ

EMPIRE

229

Tribal

auxiliaries were provided mainly by Turkmen. There were

also

Kurds,

Arabs,

and Shabankara. Quite

apart

from the family ties

between Saljuqs and Turkmen, it was a matter of moment for the sultan

treat

them well, because, in addition to their

being

a useful reinforce-

ment to his army, they were a potential source of strength for the

enemy. They were extremely mobile and could assemble in a short

space of time. Their guiding motive was plunder. Tribal auxiliaries

were

some

cases

employed by the amirs, and by local

rulers

such

the Mazyadids.

The chief weapon of the army was the bow and

arrow.

This the

cavalry

used from the saddle, shooting without dismounting or halting.

Spears,

of which the khatti spear was the most renowned, swords,

clubs, shields, and a kind of horse armour, were in use. The Khurasan

army

under

Sanjar

had elephants. Greek

fire

was used and a

siege

engine known as the

manjaniq.

This latter was not particularly effective,

and if the provisions of the

besieged

town held out the defenders were

usually able to withstand the besiegers. A more deadly

siege

weapon

was the mine, which was used by the Saljuq armies in

Syria.

The army

was in

some

cases

accompanied by a field hospital. The one attached to

Mahmud b. Muhammad's army was equipped with instruments,

medicines, and tents, and staffed by doctors and orderlies

(ghulams);

for

its transport it had two hundred Bactrian camels.

The cavalry manoeuvred with

speed

and flexibility: one of their

favourite

manoeuvres was the feigned

retreat.

Warfare

was necessarily

seasonal. Most of the forces assembled by the amirs tended to disperse

harvest time; and they were generally reluctant to be absent from their

iqta's for long periods, lest a

rival

should attack their

possessions

during

their

absence or the sultan should

assign

them to

some

other amir.

is difficult to

arrive

at accurate figures for the

size

of the standing

army.

The numbers given by the sources, which are probably only

broad

approximations, often include the armies of the amirs and

tribal

auxiliaries

which joined the sultan when he went on a campaign.

Toghril

Beg's

standing army was probably

fairly

small. Alp-Arslan at

the time of his assassination was accompanied by

2,000 slaves. Malik-

Shah,

when he was vali

ahd, had

15,000

soldiers attached to him; as

sultan,

he is said to have had 40,000 horses always in his service. In

473/1080-1

he is alleged to have dismissed 7,000 Armenians, who then

Bundari, p. 124. Ibn al-Khallikan, however (vol. n, p. 82), gives the figure of camels

needed to transport it as forty.