Mawlas: Freed Slaves and Converts in Early Islam
by Daniel Pipes
From Slavery & Abolition: A Journal of Slave and Post-Slave Studies, Vol. 1 (1980): 132 – 177.
Table of Contents
Arabists amuse non-initiates with the exaggeration that every Arabic word has four meanings — a basic one, its opposite, one related to sex, and a final one referring to camels. This variety nearly applies to mawla (pl. mawali), a term which means "Lord, possessor, chief, benefactor, manumitter, protector, lover, follower, charge, cousin, ally, contractor, in-law, slave, freedman, client," as well as religious master, novice slave owner, non-Arabian convert to Islam, and political agent.
Although the Arabic dictionaries list many definitions for this word, they give no clue to its usage. In all but a few of its senses, context makes meaning clear; only some of the social statuses are obscure: freed slave, ally, convert, and political agent. These four usages occur often in the historical sources and it is hard, even impossible, to distinguish them. The sources mention numerous mawlas of these four types, yet the modern reader usually has no way of telling them apart.
Who were most of the mawlas? An answer colors one's understanding of the status of non-Arabian Muslims in early Islam, of military slavery, and it provides an important insight into the nature of the freed slave status.
The sources rarely provide information to identify which mawlas had slave origins. Even though the four meanings at hand appear entirely unrelated and mawlas were everywhere in early Islam, a modern researcher usually cannot distinguish between them. Crone, the foremost student of mawlas, agrees: "As for telling the difference between freedmen and other mawali, most of the time you can't."
Rather than focus on individual mawlas, we do better to study them as a social phenomenon: who were the freedmen, allies, non-Arabian converts, and political agents? What was their place in society? Their positions kept changing during the early period, reflecting the evolution of Muslim society. As the quote at the beginning of this chapter notes, a single term which changes meaning through time causes historians untold anguish; "mawla" is even more confusing, for one kind of mawla stayed constant and the other changed. This study concentrates on developments in the mawla status; each section describes the social circumstances of mawlas in a specific era.
Modern historians have paid mawlas considerable attention. The nineteenth-century Orientalists von Kremer, Goldziher, and van Vloten worked out an interpretation which has been almost uncritically repeated ever since. Unfortunately, few new questions have been asked; in particular, we know little about the identity of mawlas: which of them had slave origins, which were allies, which new Muslims, and which political agents. How did these types differ and what similarities did they share? Some scholars equate mawlas with converts, even specifically Iranian converts, while others assume them slaves. Only a few writers have contemplated the alternatives and tried to work out which kind of mawla prevailed in a given period or locale. Crone alone has tried to understand their social standing.
If our state of ignorance requires further documentation the translations of early Arabic sources into European languages supply it. "Mawla" is indiscriminately rendered as client or freedman. A careful look at this distinction betrays a complete absence of consistency; lacking guidelines for the meaning of mawla, the translator supplies whichever term may appear to make more sense. In fact, he is guessing at the meaning.
It is our misfortune that none of the most important materials for this topic has survived the centuries. We know of at least four books that dealt specifically with mawlas: three called Kitab al-Mawali, by al-Jahiz, al-Kindi, and Abu 'Ubayda and al-Haytham al-'Adi's Man Tazawwaj min al-Mawali fi'l-'Arab. Almost nothing but the titles of these books remains extant; consequently, the research in this study relies on scattered bits of information from historical and legal sources.
A. JAHILI ARABIA
To learn why one word applies to freed slaves, allies, non-Arabian converts to Islam, and political agents, we must begin in pre-Islamic times, whence originates the term mawla. The specific circumstances of Jahili society (i.e. Arabia immediately before the birth of Islam) shaped the mawla status and explain why mawla refers to several little-related social categories; it had a simple basic meaning which could fulfill diverse functions in the changing circumstances of the Jahiliya and early Islam.
The Tribal Organization of Society
Tribes dominated Jahili life, performing many of the functions we commonly associate with government - such as the administration of justice and the prosecution of war - and they provided the only forum for political, social, and economic life. Their overwhelming role implied that tribal affiliation had supreme importance in Arabian life before Islam. In this nearly totally tribal society, the individual had no status or power on his own, but attained these only through affiliation to a tribe.
The paramount function of the tribe lay in protecting its members. In a society without police or other generally recognized civil authority to maintain order, an individual's security derived from the knowledge that his tribe stood behind him; were a tribesman harmed, his people had to inflict a comparable injury on a member of the offender's tribe or else receive compensation. The system worked exclusively along tribal lines; an injury to one member was felt by all; and all members paid for the transgression by one of their fellows. Without the threat of revenge hanging over potential aggressors, a man's life had little value in violent Arabian society; he lived as an outlaw who could be killed and his property plundered without retribution.
Besides assuring security, tribes provided the only political, social, and economic forum; to attain power or wealth, an individual had to work from a tribal base. Affiliation meant enfranchisement; without it, one had none of the good things in life, with it anything was possible. It offered the unique path to well-being. For all these reasons, affiliation to a tribe was the single most important personal possession in Jahili society.
Mawlas as Associate Tribal Members
Most of the population of Arabia acquired membership in a tribe at birth; the child of a tribal member followed a parent's affiliation.
Some persons, however, had no affiliation: despised classes at the bottom of the social order (including vagabonds like the Sa‛alik, itinerant musicians and tinkers) and slaves. Slaves stood outside the tribal system altogether, for they belonged to a master as chattel; they had no status of their own. The slave enjoyed the same protection as, say, a camel; the master revenged an injury to his slave by doing violence to a slave belonging to the offending tribe or by receiving material compensation.
Besides those born into the tribal system and those who lived beneath or outside of it, Jahili society also included persons without tribal affiliation but who needed it. These divide into two broad types, free Arabians and everyone else. Arabians born into a tribe occasionally had to quit it because of a quarrel or disgrace. For whatever reason he left, the refugee Arabian contracted hilf (alliance) with another tribe of his choosing, making a contract which closely paralleled the hilf arranged between two Arabian tribes.
Besides free Arabians, three other types of persons needed tribal affiliation: manumitted Arabian slaves and all non-Arabians, free ones and manumitted slaves. Put another way, all manumitted slaves and all non-Arabians sought affiliation.
The freed slave, whether Arabian or not, needed affiliation upon his manumission; so long as he was a slave, he enjoyed protection through his master, but once free, he had to establish his own affiliation. Thus, upon gaining his freedom, a slave became a person in his own right with his own tribal status. A free non-Arabian who lived for an extended time in Arabian society had to arrange for tribal protection much like a refugee Arabian, though as a foreigner, he had a lower status. The better off he was, the greater his danger without protection from a tribe.
Both of these groups, freed slaves and non-Arabians, acquired the same affiliation, the wala'; both of them became known as mawlas. Though some of them had been slaves and others had always been free, such differences mattered less than that they had all acquired a similar affiliation to a tribe. The wala' formally established them as full members in a tribe, equal in rights and duties to the born members, at least in theory. The wala' took the form of a contract, a solemn agreement binding the mawla, his patron, and the tribe. Ceremonies attended the occasion and in Mecca it could take place at the Ka'ba, lending the ceremony a religious tone. Normally, it did not involve an exchange of money.
Through the establishment of wala', the freed slave or non-Arabian acquired a fictitious filiation to his patron. "The protection which a clan afforded to a man from another tribe, or even to a foreigner, also included the right of kinship." This held especially true for freed slaves, whose patrons often adopted them, with the consequence that "emancipated slaves appear in the genealogical lists without any note of explanation, just as if they had been pure Arabs." The sources take great care to mention the wala', for it identified a person, placing him within the kinship system which defined everyone in Jahili society. Indeed, the sources seem to refer to all freedmen as mawlas, whether or not they had contracted wala'; genealogical neatness appears to have taken priority over strict accuracy. A freed slave did not have to contract wala'; a freed slave without wala' is known as a sa'iba. Yet one hears very little of sa'ibas. Salim, always called the mawla of Abu Hudhayfa, was in fact the sa'iba of Abu Hudhayfa's wife; only subsequently did Abu Hudhayfa establish wala' between himself and Salim, adopt Salim as a son, and marry him to his daughter.
The wala' bound an unaffiliated person to a tribesman backed by a tribe. The patron could be also the whole tribe or even the populace of a city, but these were merely formal differences. Regardless of the patron's legal identity, the wala' involved the same duties and benefits from each of three parties, the mawla, the patron, and the tribe. The mawla gained protection, the necessary prerequisite for an honorable life in Jahili society. With wala' he had relative security and enfranchisement. In return, he pledged loyalty to the tribe and his patron. Although the Arabian kinship differed completely from medieval European feudalism, the mawla's duties did resemble those of a European vassal: fidelitas, auxilium, and consilium (fealty, aid, and counsel).
The patron stood between the mawla and the tribe. He had rights over the mawla's inheritance, should the mawla die without heir. In return, the patron took responsibility either to pay the mawla's blood money or to avenge him. The tribe obligated itself to protect the mawla as it would a born member and gave him a voice in tribal matters. The tribe gained through the addition of a new member, acquiring an extra hand in battle and a new source of funds for the payment of tribal debts such as tribute or blood money. "Such an increase in numbers was probably quite welcome to weak tribes, but it was considered particularly praiseworthy if a tribe could manage without such elements."
In Jahili society each party to the wala' gained from its establishment.
Types of Mawla
Mawlas of servile and free origins have different names. The mawla who has been a slave is a mawla wala', mawla 'atiq, mawla 'itq, or mawla 'itaqa; he who has always been free is a mawla muwalah, mawla 'aqd, mawla hilf, mawla tiba'a, or mawla yamin. While I wish to maintain this distinction in the following discussion, it is necessary to make up new terms to refer to the mawla types because, as it will be shortly apparent, these Arabic terms acquire more than one meaning. The mawla who has been a slave is a "mawla-freedman" or a "slave mawla"; the mawla who has always been free is a "mawla-ally". Although these hybrid Arabic-English words may not be entirely pleasing, they are accurate.
The wala's also have different names: the mawla-freedman has a wala' 'itaqa, wala 'itq, or wala' ni'ma and the mawla-ally has a wala' ni'ma, wala' 'aqd, wala' khidma, wala' tiba'a, wala' luzum, wala' hilf, wala' yamin, or wala' intiqa'. In parallel with the hybrid terms for mawla, these are rendered by "wala' of manumission" and "wala' of alliance," respectively.
Although mawlas had different origins, slave or free, Arabian or not, they stood in similar relationships to their adopted patron and tribe, sharing the same benefits and duties. In other words, mawlas in Jahili society occupied a single social status. Their position was not debased, but they could not hope to be the equals of the sarihs, the born tribal members. They did not threaten the sarihs but had their own place in the social hierarchy. Since the wala' brought benefit to the mawla, patron, and tribe, it found general acceptance.
Looking ahead to the Islamic era, the mawla status had these important characteristics in Jahili society: (a) all non-Arabians, freed or free (but not persons in state of slavery, true slaves) were mawlas; (b) the mawla was a second-class citizen; (c) protection in return for inheritance rights served as the quid pro quo of the wala'. Despite vast social changes, these three points remain valid throughout the Arabian period, until 132/750. The mawla status retains its function as a mechanism to bring outsiders, manumitted slaves or non-Arabians, into society.
B. MUHAMMAD AND THE RIDDA WARS, 1-13/622-34
The mawla status had originally developed to meet the specific needs of Jahili society; one might expect that Islam, which curbed the tribal conditions, would have caused the mawla status to disappear. Instead, the mawla status became incorporated into Muslim society and spread with it. To adapt to the changed circumstances of the Islamic period - and especially to the advent of central political authority - the mawla status acquired new meanings and functions.
Islam left the mawla-freedman status of Jahili society virtually unchanged while fully transforming the mawla-ally status. The different fates of the two types of mawla require us to consider each of them separately. We begin with mawla-freedmen, the less complex of the two.
The wala' of manumission largely retained its Jahili purpose of bringing freed slaves into society. Although Islam created a central authority which took over some governmental functions from the tribes (legislating blood money, arbitrating feuds, making war, and so forth), Islamicate society remained largely tribal in organization. Consequently, the freed slave still needed a wala' of manumission with his patron. The freedman faced perhaps less danger without wala' than in Jahili times, but he still depended on a patron to insure his blood money. Even if the power of the tribes was reduced, early Muslim society was not hospitable to a freed slave lacking patronage.
To provide for this, the Islamic religion strongly encouraged the wala' of manumission. This description is found in two places, an enigmatic Qur'anic injunction and a forceful hadith. The Qur'anic verse condemns the pagan practice of releasing a female camel as a devotion:
Sa'iba here means "a she camel ... let loose for free pasture" which is customarily dedicated "on return in safety from a journey or on recovery from an illness." Figuratively, sa'iba also refers to a slave let loose, that is, one manumitted without wala' to his patron. The Qur'anic condemnation of the camel devotion is therefore also understood as a condemnation of manumission without wala'.
Much more clearly a hadith (a report about the Prophet) quotes Muhammad: "the wala' belongs to the manumitter" (al wala' li-man a'taq). This phrase appears in every discussion of manumission and is understood in two distinct ways: the wala' of manumission can only be established with the manumitter or the manumitter must establish wala' with his freed slave. The second interpretation concerns us more here. The jurists subsequently required wala' to follow every manumission.
The restrictive and compulsory nature of the wala' in Islamic law distinguishes it most sharply from its Jahili antecedent. The flexible arrangements of earlier times (including the voluntary transfer of a wala' of manumission) are no longer legal.
Once instituted in Islamicate society, the mawla-freedman status stayed the same through Muslim history. Regardless of other circumstances, the continued relationship between patron and mawla has remained a constant feature of Muslim society to the present day. It changed so little in early Islam that we need say no more about mawla-freedmen through the subsequent periods.
From Mawla-ally to Mawla-convert
Non-Arabians remained mawlas. Just as non-Arabians living in Jahili Arabia had been mawla-allies and associate members of society, so they remained mawlas and associates in early Muslim society. The mawla status for non-Arabians survived the change from Jahiliya to Islam with just one modification, albeit a major one, the requirement that a free mawla convert to Islam.
The survival of the free mawla status comes as a surprise because it seems not only unnecessary but contrary to the spirit of Islam; It seems unnecessary because the appearance of central political authority under Islam should have eliminated the earlier function of the wala' of alliance, the protection of a free outsider. When a religious polity replaced tribal autonomy, the non-tribesmen should have had the same status as tribesmen. Also, this status appears to contradict the spirit of Islam; in its social ideal, Islam is above all the religion of equality, admitting no lines drawn between believers except those of piety. A distinct status for non-Arabians in Islam - i.e. non-Arabian Muslims - contradicts the tenet that all Muslims stand equal before the Lord. Thus the continued existence of the free mawla status for Muslims requires explanation.
Islam demanded the abolition of many Jahili social customs, but few of them in fact fully disappeared. Muhammad called on the Muslims to replace tribal affiliation with an Islamic bond and to reduce an elaborate hierarchy of social statuses to equality, but neither these nor other changes were fully implemented. Indeed, both tribal allegiance and social stratification remain much in evidence among Muslims to this day. The mawla status retained its usefulness because early Islamicate government, even under Muhammad, never attained enough authority to enforce the new order. It could not eliminate the primacy of the tribal social organization, so the mawla status remained necessary.
The free mawla status continued to exist for another reason too; a Qur'anic verse commends the mawla status for non-Arabians:
What does this passage intend? In my opinion, it has more to do with genealogical tidiness (giving even non-Arabians a place in the kinship system) than with social distinction; remembering that only Arabians know the names of their ancestors, this verse urges fictive genealogies for non-Arabians who join the Islamic community in order to identify them. "When non-Arabians entered Arabian life, a name was sought for them. The term "mawla" was given to them in accordance with the Qur'anic verse" quoted above.
Whether or not the Qur'an intended only this, it had much greater consequences. Beyond serving as a genealogical device, the mawla status also preserved the distinction between Arabian and non-Arabian; willy-nilly, it perpetuated the latter's associate status. In theory, the non-Arabian Muslim had complete equality with the Arabian Muslim, but assigning him a distinct status made it more probable that he would occupy a lower position.
The mawla-convert status. For the free mawla status to survive from Jahili to Islamic society, it had to undergo a major modification. Whereas free non-Arabians earlier had affiliated to a tribe by becoming tribal members, in Islamic times they affiliated to the umma (Islamic community) by becoming Muslims. A new requirement for joining Arabian society emerged: conversion. Mawla-allies in Jahili times could belong to any religion, but their counterparts in Islamic times had to be Muslims. (Non-Arabians belonging to a scriptuary religious community could remain non-Muslims and become dhimmis; others had to convert.)
The conversion requirement so changes the nature of the free mawla status that it requires a new name; the mawla-ally of old is gone, replaced by the "mawla-convert." His wala' is the "wala' of conversion." The sources usually refer to these two different free mawla types with the same Arabic term, mawla muwalah; only a few references explicitly distinguish between them. The mawla-convert is sometimes called a mawala islam or "a mawla who converts at the hands of another man." The legal works discuss the wala' of conversion at length, but there are few specific cases on record. We know almost nothing about the manner in which the wala' of conversion took place.
The fact that the sources do not normally identify the mawla-convert as a new kind of free mawla, distinct from the mawla-ally, adds considerably to the confusion surrounding mawlas; in addition to the two entirely distinct types, freedmen and non-Arabians, the latter divide into allies (in Jahili times) and converts (in the Arabian period).
The mawla-convert status resembled the mawla-ally status in several ways: each brought free outsiders into Arabian society through direct affiliation to a member of that society; an Arabian could be neither one nor the other; both share similar inheritance relations with their patrons; and they both gain a fictive filiation. The following passage shows the importance of this filiation for the mawla-convert.
Before Islam, there were no mawla-converts; after Islam, the mawla-allies died off and were not replaced. This leaves a transition period of perhaps fifty years (ca. 1-50/622-70) during which the two types of mawlas existed simultaneously. In this period, we have no precise way of telling whether a free non-Arabian Muslim was a mawla-ally or a mawla-convert. Fortunately, this distinction has only slight importance; the mawlas who joined Islam in its first years, before it became the religion of conquerors, shared in the trying experiences of their patrons and for this they were rewarded in later years. Thus, both kinds of free mawla enjoyed a high status in the years before the conquests began; the ally/convert distinction had little importance.
The very first unambiguous example of a mawla-convert comes from the lifetime of the Prophet. This case holds particular interest, for, if true, it indicates that Muhammad himself inaugurated the mawla-convert status. Manahiya, a merchant from Merv, Khurasan who had heard of the Prophet, took interest in his new religion, travelled to Medina to see him, converted at his hands, changed his own name to Muhammad, became a mawla of the Prophet's, and returned home to Merv a Muslim. The fact that he became a mawla indicates that the Prophet viewed all non-Arabian Muslims as mawlas, even if they did not live in Arabian/Muslim society. Manahiya was free and he lived outside of Arabia so his wala' had to be the wala' of conversion.
Implications of mawla-convert status. In Jahili society the slave and free mawlas had been mutually distinct; a person could not be both a mawla-freedman and mawla-ally. In Muslim society it was possible to be both at once; the mawla-freedman and mawla-convert statuses intermingled. Although a slave need not convert to Islam to gain his freedom, it seems fair to assume that in practice, conversion often preceded manumission, especially in the early period. The freed slave who has converted to Islam is a dual mawla, a mawla-convert-freedman. He is both a slave and a free mawla at once; does this mean he falls into yet another category, or can we consider him either a mawla-freedman or a mawla-convert?
The Shari'a solves this problem by stating unequivocally that the dual mawla is primarily a mawla-freedman:
Since the wala' of manumission precludes a wala' of conversion, the freed slave
Yet a wala' of conversion with the manumitter is redundant and does not occur. Thus, all freedmen who have a wala', whether Muslim or not, are mawla-freedmen. Only converts who never experienced slavery are mawla-converts.
Islam prohibits the enslavement of Arabians; if none may be enslaved, none can be freed; therefore no mawla-freedman is Arabian (except a few survivors from Jahili times). Similarly, the free Arabian who converts to Islam does not become a mawla-convert; consequently, every mawla in Islamicate society, of whatever type, is a non-Arabian. This fact has enormous implications for the subsequent development of the mawla status, for it means that the mawla status provides a means of dividing Arabians from non-Arabians, regardless of religion or social standing. Note, however, that if every mawla was a non-Arabian, not every non-Arabian was a mawla; in special cases, privileged military elements, such as the Persian abna' of the Yemen or the Asawira escaped this status.
Having survived the transition from Jahili times to Islam, the free mawla status went on to acquire a major new role in Islamicate society for the next 120 years.
Most mawlas had slave origins. The many detailed biographies of the Sahaba, the companions of Muhammad, provide information on the proportion of slave to free mawlas. Mawlas constitute about a tenth of the entire Sahaba; of these again about one-tenth, of fifty-six mawlas, can be identified by mawla type. As Tables 1 and 2 show, a large majority of them were of slave origins, perhaps five-sixths of all the mawlas.
A mark in column 1 indicates a mawla-freedman; in column 2, a mawla-ally; and in column 3, a mawla-convert. Any mawla who is also thought to have been a halif (ally) I assume to be a mawla-ally. The names in parentheses are those of patrons.
* The names are listed in (Arabic script) alphabetical order, as in Usd al- Ghaba; many of the names are unusual, so I cannot vouch for the spellings. All but one of these mawlas (namely Abu Qara) have entries in Usd al-Ghaba by Ibn al-Athir; these references are not provided here since they can so readily be found; various other references are given.
C. THE ARABIAN PERIOD, 13-132/634-750
The mawla status for non-Arabian Muslims had been established in Muhammad's lifetime, though it had little social significance at that time, since all Muslims endured the same hardships. After 13/634, however, as the Arabians conquered vast territories, the mawla status became a very important distinction between Arabian and non-Arabian Muslims. Thus, a study of mawla-converts in this period is an inquiry into the relations between Arabians and other Muslims; as such it constitutes the central topic in the social history of early Islam.
The Standard Interpretation
First formulated by von Kremer in 1877, a single view has dominated the study of mawla-converts in the Arabian period. This view has importance for it serves as the keystone to the understanding of the relations between Arabians and non-Arabians in early Islam; therefore, I shall outline its main points before attempting to refute it.
The standard interpretation notes that mawla-converts had enjoyed near equality with Arabian Muslims in Muhammad's lifetime and the changes for the worse occurred shortly after his death. It attributes the decline they experienced either to Arabian racial arrogance or to the assassination of 'Umar I by a Persian slave. Non-Arabian Muslims answered this deterioration with resentment; until 132/750 they struggled with the Arabians to regain an equal position; and the Abbasid take-over represented their successful revolt against oppression.
This argument contains five main points: (1) the Arabians arbitrarily discriminated against mawla-converts; (2) the mawla-converts responded with resentment; (3) they expressed it by agitating against the Umayyads, the Arabians' regime; (4) their efforts ultimately succeeded when they brought the Abbasids to power; (5) the Abbasids came to power with a mandate to make all Muslims, Arabians and others, equal, and they did so, permanently settling this issue.
I do agree that the mawla-converts' position was worse in the Arabian period than either before or since, but I explain these changes differently. The Arabians did not oppress mawla-converts but treated them as any conquerors treat their subjects; the mawla-converts did not resent their position, nor did they fight against the Umayyad regime any more than they fought for them; the Abbasid forces depended more on Arabians than on mawla-converts; and the improvement in the status of non-Arabian converts after 132/750 was a secondary consequence of the Abbasid take-over. This study deals only with points (1) and (2); others are dealt with in my Slave Soldiers and Islam (pp. 117-31, 170-74, 174-77).
The Conquests and Mawla-convert Subordination
The great Arabian conquests dominated Islamicate life from 13/634 until 132/750. The tentative social patterns set in Muhammad's lifetime were completely upset by the transformation of the Muslims from an ordinary group among many into the greatest conquerors of their age. This transformation created a new significance for the mawla-convert status; it served to distinguish between conquerors and conquered among the Muslims and helped maintain the privileges of the former over the latter. Although the mawla-convert status had had no function before 13/634, when being a Muslim in itself brought no benefits, it subsequently acquired an important role as Islam became the religion of rulers.
The new function of the mawla-convert status. Contrary to common belief, jihad (Islamic holy war) should be undertaken to spread not the religion of Islam but its rule. When Muslims make war against pagans, they fight to control territories this often leads to the conversion of vanquished peoples, but not always (e.g. many parts of India, Hungary). The limited goal of jihad lent itself well to the needs of the early Arabian Muslims, for it inspired the conquests and provided elements of cohesion and direction otherwise lacking. Since jihad does not require that Muslims encourage the vanquished peoples to convert, Islam then became a creed of conquerors. For the early Arabians, "it was above all a badge of a united Arabism, the code and discipline of a conquering élite."
The Arabians had no interest in sharing their religion with the subject population and made no efforts to spread it among them.
An important practical reason lay behind the Arabian discouragement of non-Arabian conversion and their subsequent reluctance to treat them as equals. The Arabians received great benefits from their victories, both in booty and in annual pensions; the fewer the beneficiaries and the larger the subject population, the more they received. The Arabian Muslims needed to limit entry onto the diwan, the Register which listed the pensioners. They had to restrict the benefits of conquest to the conquerors alone and deny them to Muslims from the subject populations. If every Berber, Egyptian, Syrian, Aramean, or Iranian who converted received a pension, who would pay? Masses of non-Arabians would convert, the pension system would collapse overnight, and the Arabians would lose all their well earned booty.
The non-Arabian who converted to Islam was, in effect, petitioning to join the society of the conquerors. To prevent this from happening in large numbers, the conquerors had to exclude non-Arabians, whether Muslim or not, from the Register and the fruits of victory. Herein lay the usefulness of the mawla-convert status; it provided an Islamic mechanism whereby the Arabians restricted to themselves the benefits of victory. Wala' in Jahili society had benefited the free non-Arabian, in Muhammad's lifetime it had little meaning for him, and in the Arabian period it served as an obstacle to privilege. This change in Islamic times did not occur either because of antagonisms aroused by Arabian racial pride or 'Umar I's murder, but by the obvious problems of relations between the victors and the vanquished. The change occurred as a result of the conquests; when Islam became a tag of privilege, old Muslims found they needed some new status for new converts in order to prevent them from spoiling the benefits of victory.
A look at taxation policies confirms the Arabian unwillingness to grant non-Arabian Muslims advantages over their unconverted brethren. Although obliged by Islam to release converts from the jizya (poll tax), "throughout the Umayyad period the Arabs were extremely reluctant to relieve people of their poll taxes." In Egypt, for example, they "discouraged conversion and in many instances refused to exempt the convert from his poll tax." The Arabians did sometimes grant converts their rightful benefits, but not without constant arguments. As a result, "whatever the expectations of the convert, in practice, conversion failed to work as a means of tax evasion."
Even without full tax benefits, converts multiplied; already in Mu'awiya I's time Arabians were concerned with the numbers of mawlas. One source (admittedly it is suspected of exaggerations in this regard) states that 20,000 Persians lived in Kufa in 'Abd al-Malik's time and a modern historian thinks that mawlas formed a majority in Iraq as a whole. We do not know how many of these were mawla-freedmen, but if large numbers of non-Arabians converted without advantages, how many more would have with the lure of receiving a pension?
The high status of mawlas who converted before the conquests began stayed high in the Arabian period; their exceptional position confirms my line of argument. Mawlas like Salman al-Farisi and Bilal b. Rabah, who converted before the conquests, formed part of the Muslim aristocracy, despite being non-Arabian, for they had joined the umma before it brought advantages. They shared in the struggles and later joined in the rewards. They did not suffer on account of not being Arabians, but received equal pension pay, occupied high posts, and have been ever esteemed as members of the Sahaba (especially Salman). The prestige of the pre-conquest mawlas belies accusations against the Arabians of racial discrimination; later mawla-converts had inferior positions because they came from the subject populations, not because they were not Arabians.
Arabian and mawla-convert attitudes. Widespread agreement to the contrary, it is not true that Arabians felt prejudice against others. Their discrimination resulted from the inevitable social differences which developed out of the conquests and the sharp lines they drew between victor and vanquished. The situation could not have been otherwise; the mawla-converts came from the conquered peoples and conversion did not change this basic fact. Even when they share religion with their subjects, conquerors do not freely admit the conquered to join their ranks (note, for example the continued low position of Christian converts in British India).
Why do historians expect Arabians to have acted differently from other conquerors? The reason may lie in the belief that the Arabians originally intended to convert the conquered populations and subsequently they shied away from granting the converts their full rights, retreating into selfish domination. If conversion had been their intention, then indeed, they deserve reproach for succumbing to the material benefits of excluding non-Arabians, but conversion was not the purpose of the conquests (as indeed, it is never the legal purpose of jihad); we should not begrudge the Arabians their well earned rewards.
The Arabians had to draw lines between themselves and the peoples they had vanquished; they were not prejudiced, nor did they feel an overweening racial pride, but they protected their interests in a predictable and reasonable way. Mawlas accepted these reasons and did not resent their lesser status among Muslims. As conquered peoples, how could they expect treatment as equals? Not only would this be presumptuous, but it defies all common sense. Perhaps historians expect to find resentment because they project backwards the well-known feelings from later centuries, so eloquently expressed in the Shu'ubiya literature; it is perhaps assumed that they are valid also for mawla-converts in the Arabian period, however different that earlier situation was.
Confirmation for my interpretation comes from the military record; mawlas fought at least as much for the Umayyad regime as for its enemies (documented in Chapter 4 of Slave Soldiers and Islam); this indicates that they did not resent the social order it represented.
All Mawlas Shared a Single Social Status
Indications of a single status. Although different in origins, mawla-freedmen and mawla-converts shared much in the Arabian period; their similarity was so great that they can even be regarded as a single group. They had the following characteristics in common:
(3) Non-tribal. Muslim society had shed some of the tribal structure of Jahili days, but tribes still counted and mawlas remained outsiders (only Arabian tribal affiliation counted; it didn't help to belong to an Iranian or Berber tribe). All mawlas and almost no one else participated in Arabian/Islamicate society without having been born into it with tribal affiliation. This made all mawlas in need of a patron to sponsor and protect them.
(4) Low social status. As both non-Arabians and members of the subject populations, slave and free mawlas alike suffered social discrimination. This fact held true with little difference for slaves (who were forcibly brought in) and for converts (who attached themselves voluntarily).
(5) Disruption and isolation. All mawlas found themselves cut off from their own peoples and not fully accepted into Muslim society. This is self-evident in the case of the mawla-freedman; enslavement disrupted his life totally, uprooting him from his people and then thrusting him into alien surroundings. By the time of his manumission, he had probably lost all ties to his homeland, so he remained as a freedman in the country of his captivity. As a freed slave, he depended heavily on his patron, yet remained an outsider, isolated from the mainstream of society.
The disruption and isolation experienced by a mawla-convert are less obvious, but they were nearly as great as the mawla-freedman's. The mawla-convert's background explains this dependence. Which individuals in the subject population were most likely to convert to Islam? Bulliet notes that as a social process, conversion takes place primarily to maintain or improve social status; therefore, given the low social status of the mawla-convert, we may assume that he was yet worse off before conversion to Islam. He had been perhaps a landless peasant, a prisoner of war, or an oppressed city dweller. Non‑Arabians converted when they had nothing to lose and something to gain - perhaps social advantage more than economic.  Conversion to Islam amounted to "a voluntary entry into clientage [for] peasants in search of a patron." The conversion of individual non-Arabians without roots in their own communities "was the only important pattern of conversion to Islam in the first century of the hijra"; in later times, converts did not experience the same disruption.
Thus, the mawla-convert was most likely a person from the bottom of the social ladder in his own society, someone who took the chance of joining the Arabians on any terms. Once a Muslim, he did not differ markedly from other non-Arabians who had been slaves.
(6) The merging of wala'. Mawla status in Arabian times passed on to subsequent generations; the children of mawlas were almost all mawlas, as the following marriage combinations indicate:
A mawla mother could have non-mawla children, but if the father was a mawla, his children had to be mawlas too. These male mawla children in turn had mawla grandchildren, and so on indefinitely.
The continuation of the mawla status through multiple generations has a hidden but enormous significance for the argument that all mawlas share a single social status. As time passed and the original act of manumission or conversion receded, the status of the descendants became increasingly indistinguishable. Later generations of mawlas began life neither as slaves nor as non-Muslims; they were born as free Muslims. Since they personally did not become free nor did they convert, they bore the characteristics of neither a mawla-freedman nor a mawla-convert. Through the generations, their ancestors' wala' type was forgotten and lost importance.
Plenty of examples show that the wala' was inherited for many generations. Abu Hanifa, the founder of the Hanafi law school, was the grandson of a mawla-freedman and he was also known as a mawla. Abu'l-'Atahiya, the poet, was the great-grandson of a mawla and still called a mawla. Tahir b. al-Husayn, the Tahirid ruler until 207/822, was a mawla, though in his case also a great-grandfather had established wala'. Similarly, al-Fadl b. Rabi', a major Abbasid political figure, had an ancestor four or five generations back who became a mawla, though we also hear that his own father was freed from slavery. The unclear wala' of al-Fadl b. Rabi' typifies the confusion and complexity in tracing wala' genealogies.
As the wala' passed from one generation to the next, the numbers of mawlas grew, without any new manumissions or conversions. Of course, manumission and conversion did not stop, but relative to the increasingly undifferentiated mass of mawlas by inheritance, these new mawlas formed an ever-smaller percentage of the total mawla population. With time, nearly all mawlas became indistinguishable, being neither mawla-freedmen nor mawla-converts, but just plain mawlas.
(7) The ambiguous status of conquered peoples. The mawla question is terribly complicated by the unclear status of peoples vanquished by the Arabians. In theory, all the inhabitants of a region conquered 'anwatan became slaves. 'Anwatan is usually translated "by force," but recent research shows that it really signified in this context "without treaty." A small locality or a town might be taken by force, with or without treaty, but not whole countries such as Egypt or the Sawad (i.e., Iraq), for within those countries both treatied and treatiless conquests took place; nonetheless, both Egypt and the Sawad are said to have been conquered 'anwatan. What can one make of this? The 'anwatan description fits the facts of the conquests very poorly; Noth suggests that they were not contemporary accounts of the conquests but developed some seventy years later, the creation of Umayyad jurists justifying then current taxation practices.
If the 'anwatan conquest has little meaning, we have to look at the details of a region's settlement terms in order to assess the status of its population. Did the Arabians enslave en masse the inhabitants of regions conquered by force of arms? Did they enslave just the soldiers and their retinues (asirs) or entire populations (sabis)? If entire populations fell into slavery, then that region would later have to include massive numbers of mawla-freedmen and few mawla-converts. The astonishing ambiguity and complexity of this matter leads me to think that no one could tell what kind of mawla the subjects of certain regions were. The situation in the Sawad illustrates this complexity with a flourish.
After the conquest of the Sawad, 'Umar I "established the revenue of the Sawad as a waqf (mortmain) and included the peasants as part of that waqf, even though they were not slaves." What then was the status of a population that formed part of a waqf grant but was not enslaved? The sources call them ahl al-ard, riqqs and 'abd qinns, since the Muslim legal experts were not familiar with the concept of the adscript (serf), they floundered about for a term to call this status. The difficulty arose because:
The peasants of the Sawad were legally free yet called slaves; if this confused the jurists, how can it help but bewilder everyone else? The following two passages show the attempts made to understand the status of the Sawad population:
Confusing the situation yet further, an Arabian told al-Mukhtar in 67/687 that mawlas "were received as booty from God" when the Sawad had been conquered. Thus, this population cannot unequivocally be assigned either a slave or a free status. When they converted to Islam, they became either mawla-freedmen or mawla-converts and no one could tell which.
Another example may show that the Sawad was not an isolated instance; the lawyers attained even greater heights of sophistry when they explained the peace settlement between the Arabian conqueror 'Amr b. al-'As and a Berber tribe. The Berbers agreed to send their children as part of their jizya payments. From this fact, an eminent legal scholar inferred: "had [the Berbers] been slaves, this would not have been required of them." In other words, the requirement that the Berbers send their children into slavery implied the parents' freedom! This unblushing analysis does everything but tell us whether the conquered peoples were slaves or free.
A Muslim jurist from a later century, az-Zayla'i (d. 743/1342), reasoned out his own solution of this problem:
Az-Zayla'i admits to not knowing what kind of mawla these persons were and presents his own reasons for thinking them mawla-freedmen. Although a reasonable solution, his proposal does not solve the basic question: who was enslaved in the Arabian conquests? The term "mawla" obscures the answer, for "many of the literary and historical sources apply it to every non-Arabian Muslim convert, whether or not he was a slave." Mawla came to mean "non-Arabian Muslim;" although "apparently rejected by the lexicographers as untechnical [this meaning is] commonly found in the sources."
(8) Indications in the sources. Although no contemporary writer explicitly states that all mawlas shared a single social status, several hint at this. They do so in the following ways:
(a) Lack of interest. The chronicles rarely pay attention to mawla type; few individuals are referred to specifically as mawla-freedmen or mawla-converts. The legal books often make this distinction because it had legal importance; perhaps the chronicles ignored it because it had no historical importance.
(b) Confusion. Possibly the chroniclers said nothing about mawla type because it is assumed that the reader can distinguish between them on his own? To the contrary, not only could the reader not recognize mawla types, but even authors were sometimes confused. Abu Muslim is the prominent case; the sources disagree on his being a mawla-freedman or mawla-convert. Even contemporaries had to ask about mawla type; when Muslim b. Dhakwan went on a mission for Yazid III, his identity came under question: he called himself "a mawla of Yazid"; when asked, "a mawla-freedman or a mawla-convert?" he answered "a mawla-freedman," to which the reply came, "good, that is preferable." These and other examples betray everyone's confusion about the mawla status; this implies that no major differences separated the two types.
The Shari'a takes such confusion into account; if the mawla claims to be of one type and the patron says he is of the other, ash-Shaybani gives more weight to the mawla's statement.
(c) A single category. Al-Jahiz in Manaqib al-Atrak recounts mawla pride in the great mawlas of early Islam. The list includes mawla-freedmen of the Prophet (Zayd b. Haritha, Shaqran, and Anasa), prominent mawlas in the Abbasid movement (including Abu Muslim and the mawla-freedman Abu Salama), and the Khurasaniya, an Abbasid military unit recruited in Khurasan. That al-Jahiz lumped mawla-freedmen and mawla-converts together indicates that mawlas themselves saw mawlas as a single category.
Modern authorities agree with these conclusions. Crone remarks that we repeatedly "meet the mawla as the client pure and simple with the free and the freed client as the two major variations." Jacob Lassner concurs: "The legal distinction between free born and manumitted clients is generally of no consequence."
The fact that all mawlas of the Arabian period shared a single status has two implications: that the ratio of mawla-freedmen to mawla-converts is untraceable; and that this ratio had little importance because all mawlas resembled each other.
All mawlas were unfree. What social standing did mawlas have? Legal differences aside, all mawlas were non-Arabian Muslims outside the tribal network occupying positions of low social status and isolated from their own people; thus, they shared a disjointed and vulnerable status. Mawlas depended so much on their patrons that they were unfree.
The slave in Muslim society of the Arabian period was removed from his home to serve an Arabian. As a slave, he had almost no resources of his own but depended on his master. His circumstances hardly changed with manumission for the freed slave could do little on his own; whether he needed financial aid, legal protection, or political patronage, the mawla-freedman usually could not escape continued dependence on the patron for the good things in life. He could, no doubt, depart, perhaps even return to his own people, but this seems to have been rare; once part of the conquerors' society, if only at its fringes, the freed slave found it hard to leave.
The same held true for the mawla-convert. He too lost ties to his people, though he did this voluntarily and in return for social and economic benefits. The mawla-convert usually had a humble social origin (some aristocrats also converted in this period to preserve their holdings, but that is another matter) and had nothing to lose when he converted to Islam. He hoped to gain by it; attaching himself to the society of the rulers, he gained an entree to the privileged world of the Arabians. He gave up community, religion, and independence of action to serve an Arabian patron. Although never formally enslaved or manumitted, the service he rendered resembled that of a freed slave. Whether the patron was a landowner, townsman, soldier, or caliph, the mawla-convert depended on him for access to wealth, legal protection and sometimes even power. Falling into disfavor meant the abrupt end to a career; all hopes of advancement and well-being lay in the patron's good will.
Mawlas of the period 13-132/684-750, both slave and free, depended on the favor of their Arabian patron, subsuming themselves and their interests to him. Enslavement or conversion both led mawlas to a position of dependence which I call "unfree."
D. THE FIRST ABBASIDS, 129-205/747-820
The Abbasid take-over produced the last major transformation in the free mawla status; henceforth it remained the same. Before 132/750 it had developed in response to the special requirements of Jahili and early Islamicate society; after this date it acquired a definition which was to remain the same regardless of changes in social circumstances. In this respect, it came to parallel the mawla-slave status which had always remained constant. At the Abbasid take-over, the mawla-convert status disappeared and a new kind of mawla, an abstract political type, emerged. Although few in number, these held high positions and figured prominently in Abbasid military and political life.
As the Abbasids consolidated their power, Muslim political unity ended. It is therefore inaccurate to imply that the Abbasids represented the only government by Muslims; yet their rivals are little known, even the Spanish Umayyads. The few bits of information we have on the other dynasties give reason to believe that the free mawla status went everywhere through the same changes that took place in the Abbasid territories. This section by necessity concentrates on materials from the caliphate but its conclusions should be valid for independent regions as well.
Mawla-freedmen and Mawla-converts
Modern histories give the impression that mawla-freedmen disappeared and mawla-converts remained in the first Abbasid period; in fact, the opposite occurred.
Mawla-freedmen remained. Although often forgotten, the slave mawlas continued quietly to exist. Many references to the capture of mawlas, their prior slavery, or their manumission attest to their presence. Simple references to them may be found in the Appendix; here follow some of the more striking facts:
(1) A discussion about one of al-Mansur's mawlas states that he was most likely a mawla-freedman and not a free mawla, indicating the continued difficulty in distinguishing between mawla types.
(2) Muhammad b. Sulayman, the governor of Kufa 147-55/764-72, had 50,000 mawlas, of whom 12,000 were manumitted slaves ('itaqa).
(3) A governor of al-Amin's enjoyed the loyalty of his mawlas because he had manumitted them, raised them out of poverty, and given them everything that they ever had. Consequently, they showed him immense gratitude.
(4) Al-Mu'tasim's Turkish soldiers are referred to in the sources indiscriminately as 'abds, bandes, ghulams, mamluks, mawlas and raqiqs. This interchangeability points to the slave origins of these mawlas.
Mawla-converts disappeared. Mawla-converts gained full social standing as a consequence of the Abbasid take-over. This occurred due to changes in the organization of the Muslim armies; briefly put, the Umayyad regime had always to maintain the distinction between victors and vanquished, between the Arabians and mawla-converts. The Abbasids, to the contrary, had a more egalitarian approach and had no use for the mawla-convert status as a means to distinguish between their soldiers. Accordingly, after the Abbasid take-over, the mawla-convert status dropped out of common use.
With privilege and power no longer exclusively associated with Arabians, "it was presumably no longer necessary for the Persian or Aramean convert to become the client [mawla-convert] of an Arab tribe." Non-Arabian Muslims could now participate fully in Muslim military and political life; the mawla-convert status lost its import, became "socially neutral," and then disappeared.
Although the mawla-convert status is prescribed by the Qur'an and should exist in all Muslim societies, regardless of social conditions, it is easy to understand that it fell into desuetude when mawla-converts no longer had a distinct social position. True, this status existed in the short period before the Arabian conquests had begun, although it served no function then, but those were the first years of a new religion and one might expect to find religious injunctions more carefully executed then than subsequently.
While the mawla-convert status lost its significance almost overnight, the persons called mawla-converts probably kept this title until their deaths. Persons who converted after 132/750 did not become mawlas unless they were slaves, but it took decades (till the 170/790s, say) for the mawla-convert status to disappear. Accordingly, references to mawlas until forty years into the Abbasid period might still refer to mawla-converts.
While the Abbasids allowed the mawla-convert status to disappear, they created a new type of free mawla, known in Arabic either as mawla istina' or mawla amir al-mu'minin; in English, I shall call him a "mawla-agent." The wala' involved here is wala' istina' or the "wala' of agency."
Information on this new type of wala' is hard to find. Hilal as-Sabi' and Ibn Khaldun hint at its purpose but the law books which provided most of our knowledge about earlier developments in the free wala', say nothing about the wala' of agency. They are silent because, unlike all the other types of wala' discussed, the wala' of agency had no legal basis; it was a political, not a legal relationship. The mawla-agent received his title as an honorific; it implied nothing at all about his social origins or his current status. He could be Muslim or not and of slave or free origins.
The wala' of agency suggests an abstraction of wala', since it made a legal relationship into a political bond. The political bond could have had a different name, for it shared none of the essential qualities of a literal wala'. Why then was it called a wala' relationship? To confuse historians? No, there was a better reason:
By its nature, this political wala' only took place between men with political power, not between ordinary persons. The patron did not always have to be the ruler, but he did have to be a leading personage. When the caliph was patron, the mawla-agent took the title mawla amir al-mu'minin.
It appears that some mawla-agents existed already in Umayyad times; at least, so one might surmise whenever a wala' relationship did not fit the legal categories. Two important examples both involved (perhaps not coincidentally) Inner Asian rulers. In 93/712 Qutayba b. Muslim referred to the free, non-Arabian, non-Muslim of Samarqand, Tarkhun, as his mawla when he said, "I shall avenge Tarkhun's blood, for he was my mawla and under my protection." In a similar manner a few years later,
On the basis of this evidence, Crone concludes that "if he calls himself the mawla of Jarrah, it means that all the non-Muslim rulers of Khurasan styled themselves mawali of the changing Arab governors.
Aside from the possible use in protocol to standardize titulature, the abstraction of wala' served no purpose in the tribal society of the Arabian period. Stray examples to the contrary do not disprove this.
The real need for an abstract, political wala' came in the first Abbasid period. When the Abbasids undermined the tribal basis of Muslim society, they needed new forms of allegiance to replace it; and their need for new bonds remained acute throughout their first century. In the mawla-agents they developed (according to Omar),
The term mawla amir al-mu'minin emerged during al-Mansur's reign, indicating the need for a new bond. These mawla-agents served the caliph in many capacities: as Sahib ar-Rukkab, collector of kharaj, director of the arsenal, mayors of the quarters of Baghdad, vizier, and governors. A coin minted in 157/774 at Qinnasrin by "Musa, mawla amir al-mu'minin" provides the first numismatic confirmation of this title. Other early coins with this term date from 184/800 and 185/801.
These politically important mawla-agents must not be confused with the larger numbers of mawlas who served al-Mansur and his successors in the army. The sources sometimes make this distinction explicit, in 199/815 al-Ma'mun's governor in the Hijaz collected the slaves and mawlas of the Abbasids to fight an 'Alid rebel and then split them, weakening his forces; in 213/828 al-Ma'mun sent mawlas but no ghulams against the rebel Babak.
Although Omar states (in the above quote) that even Arabians could become mawla amir al-mu'minins, Crone correctly points out that "there is no unambiguous example of an Arab being granted the title." Several other nationalities are specifically mentioned, however, such as Berber, Bukharan, and Tabaristani. Could a literal mawla (i.e., a mawla-freedman or a mawla-convert) become also a mawla-agent? Yes, several prominent examples indicate this possibility:
A mawla amir al-mu'minin could keep this title while serving various caliphs. For example, one held positions under al-Mansur and Harun ar-Rashid. Five or seven mawla amir al-mu'minins witnessed the agreement signed in 186/802 between al-Amin and al-Ma'mun dividing the Abbasid empire. A papyrus indicates that a mawla amir al-mu'minin could himself be a patron to a mawla.
Al-Ma'mun tried to induce the ruler of Tabaristan, Qarin b. Shahriyar (or Wandad), to convert to Islam by offering him the title of mawla amir al-mu'minin, but Qarin refused. Subsequently, Qarin's son Mazyar converted and received the title mawla amir al-mu'minin. He became governor in Tabaristan and the surrounding districts for the caliph. Later he went on to lead a major rebellion against al-Mu'tasim, in 224/839. Other accounts say that Mazyar converted already in al-Ma'mun's time in order to gain the caliph's support in his quest to wrest control of Tabaristan from his uncle. Mazyar's brother al-Fadl became governor of Homs, but his unpopular rule led to his death at the hands of the populace of that city.
The story of Mazyar contains two curious aspects: in most accounts, Mazyar is called mawla amir al-mu'minin, but twice at least he calls himself instead muwali amir al-mu'minin. Historians translate muwali as "ally" (in contrast to the mawla who is a "client"), but I prefer to translate this term as "client". The use of the term muwali instead of mawla confirms the political nature of Mazyar's relationship with the caliph. Secondly, all accounts agree that Mazyar's father Qarin was not a Muslim, but one story tells of Mazyar receiving advice from his father's mawla. How could a non-Muslim living in Dar al-Herb be patron to a mawla? This usage hints at the ever more figurative and less legal meanings that the term mawla came to acquire in subsequent eras.
At all times, two different types of mawla existed, one formerly slave, the other always free. The changes which the status underwent may be summed up as follows:
Because it changed with developments in Muslim society, the free mawla status has received more attention both in other studies and in this one; this should not make us forget slave mawlas, but it often does. For example, both the quotations at footnotes 145 and 146 ignore that the term mawla always retained the meaning of "freed slave." Other analyses imply that all mawlas in the first Abbasid period were political agents of the caliph and ignore the legions of ordinary mawla-freedmen affiliated to shopkeepers and landowners.
It is a mistake to disregard mawla-freedmen, for through the early Islamic period, from Muhammad to al-Ma'mun, most mawlas either were mawla freedmen or they experienced a comparable servility. This holds true for each of the three Islamic eras studied here, Muhammad's time, the Arabian period, and the first Abbasid. In Muhammad's time, the sampling provided by Sahaba mawlas shows that mawla-freedmen far outnumbered free mawlas. In the first Abbasid period, mawla-converts disappeared and mawla-agents numbered only a few individuals, so most mawlas were freedmen. The Arabian period presents a more complex situation; I have tried to show that slave and free mawlas during this time shared the same social status ("All Mawlas Shared a Single Social Status") and that this status resembled slavery ("All mawlas were unfree"). Although free mawlas never underwent formal enslavement, they shared a position similar to that of a freed slave. Thus nearly all mawlas in early Islam experienced slavery or something akin.
The Arabic sources often use the terms "slave" and "mawla" interchangeably; the two had so much in common that contemporaries - and hence we too - can treat them nearly as a single social category. The similarities of their real circumstances far outweighed the differences of their legal status; both depended on a master/patron and were subject to his control. Al-Jahiz summed up the situation in the following passage about secretaries (katibs), most of whom were mawlas:
This precisely describes the mawlas: subordinate and servile, even if not legally enslaved.
This view of early mawlas goes contrary to the assessments of most modern historians; they often assume (without explicitly saying so) that the great importance of mawlas in the military and political life of the first two Islamic centuries meant they could not have been slaves. Mawlas constantly showed up as bodyguards, soldiers, courtiers, officials, and administrators; for example, nearly 10 per cent of the Muslims on record at Muhammad's first battle, Badr, were mawlas; and nearly every single hajib (chamberlain) from Abu Bakr's on was a mawla. Doubting the servility of these men makes no sense in light of subsequent Islamic history, for slaves regularly filled key positions. Virtually every Muslim dynasty employed men of slave origins in the highest public positions; it is not hard to imagine that the Rashidun, Umayyads, and Abbasids did so too. Indeed, I argue elsewhere that the use of mawlas foreshadowed the institution of military slavery (the systematic use of slaves and freedmen as soldiers) and was vital to its development. In short, the importance of mawlas is no reason to assume that they had free origins.
APPENDIX: IDENTIFYING MAWLA-FREEDMEN
The historical sources very rarely state explicitly whether an individual mawla was a freedman, ally, convert, or agent. In many primary sources, I have found only a few specific identifications (e.g., Mawla-freedman: KM 251; ar-Rashid 222; Tanbih 346 = Jah 301; Combe 1.102. Mawla-convert: KM 162). To identify other individual mawlas by type requires some ingenuity. This discussion begins with the most simple and certain indications of mawla type and then proceeds to the most complex and speculative.
(1) Synonyms. Mawlas were sometimes referred to by other names which indicated that they had slave origins:
(2) Specific mention of slavery, manumission, or the return to slavery.
slavery (MDh 3.421; ITB 1.272.)
manumission (KM 231, 234, 252,255, 270; T2.649, 3.854; FM 120, 156; Ibn Khallikan 5.189; TYa' 2.413; Jah 101; AM 3, 25; ar-Raqiq 132; ITB 2.106.)
return to slavery (T2.1648, 1813.)
In one discussion, Ibn Khaldun appears to refer to the enslavement of mawlas (1.245).
(3) Father. If the mawla's father were a free (i.e., never enslaved) Muslim, then the mawla could not be a slave or a mawla-freedman. Were the father ever a non-Muslim or a slave, then the mawla could be of any type. Even if the father was known as a pious Muslim, we can say nothing about his son.
(4) Marriage guardianship. Only mawla-freedmen "are to be understood when it is reported that the mawla did not have the right to marry off his daughter. A marriage was only viewed as valid when the permission of the patron, the former master, had been given." If a mawla made this decision on his own, without consulting his patron, he had to be free.
(5) Inheritance. Slave and free mawlas differed both in their inheritance from the patron and bequeathing to him. The mawla-freedman inherited ahead of mawla-converts if a waqf had been set up for them by the patron. The patron ordinarily inherited from his mawla-freedmen but not from free mawlas.
(6) Religion. A non-Muslim mawla could not be a convert to Islam, therefore he had to be a mawla-freedman, mawla-ally, or mawla-agent.
(7) Regional origins. A mawla born free within Dar al-Islam could only be a free mawla, for no dhimmi or free Muslim might be enslaved. If, to the contrary, he was born or acquired as a slave, then he had to be a mawla-freedman. A person born free in Dar al-Harb could become either type of mawla, depending on whether he was enslaved or not.
In Jahili times, an Arabian could only be a mawla-freedman, since the Arabian equivalent of the mawla-ally was called a halif.
While the preceding indications reliably distinguish slave mawlas from free ones, they do so in only a few cases. Further indications, though less certain, may sometimes prove useful.
(8) Patron. Free mawlas had only Arabian patrons; a non-Arabian patron indicates a slave mawla. Since the wala' of conversion had the purpose of affiliating an isolated individual to a tribe, no free person established wala' with a non-Arabian who also lacked a tribal affiliation of his own. Hence, non-Arabians were patrons only to their own mawla-freedmen; upon manumission their slaves had no choice but to establish wala' with the patron, regardless of his own status.
It is unclear whether female Arabians became patrons to free mawlas too; while az-Zayla'i quotes an hadith which says not, ash-Shaybani writes that it is possible. In any case, women patrons were not common; indeed, except for one from Jahili Arabia, nearly all of them belong to the caliph's family: the wives of 'Uthman, al-Mansur; al-Mahdi, the granddaughter of al-Mansur. In one account, an Umayyad governor of Ifriqiya in Sulayman's time is said to have been the mawla of a woman, though other versions disagree. The fact that the wife of a patron could be called mawlah (patroness) confuses the situation considerably.
Anyone could be patron to a slave mawla - even a slave, if he had his master's permission. Mawlas often became patrons; the mawla's mawla was a mawla-freedman and he had a special name: maqit (pl. muqqat), though it was hardly ever used. Maqits were held in extremely low regard; al-Farazdaq said this about the eminent grammarian 'Abdallah b. Ishaq (d. 127/745): "If 'Abdallah were a mawla, I would describe him, but he is only the mawla of a mawla [and so not even worthy of contempt]!" Maqits were fairly common; they included in their number several men of historical importance such as one of Hayyati an-Nabit's cavalry leader; Musa b. Nusayr, the conqueror of Spain (a maqit according to one account only); and a governor of Ifriqiya whose patron, a mawla, had also held that same position.
Generation after generation of mawlas could become patrons in turn, thus creating chains of maqits which extended even five mawlas in length: A the mawla of B the mawla of C the mawla of D the mawla of E. Most of these chains returned ultimately either to the Prophet or to a caliph. A few returned to tribes. In one unusual case, two or three mawlas involved in a maqit relationship were da'is (non-Arabians who claimed Arabian descent).
A woman could participate in a maqit relationship; in one case, a woman was in the middle, both a mawla and a patron simultaneously.
(9) Transfer of wala'. The slave mawla could contract wala' with his manumitter and with no one else. He might choose to terminate this wala' but he could not contract a new wala' with anyone else. The patron of a slave mawla also had little freedom of action; he might not make manumission conditional on not having a wala' between them nor could he dispose of the wala' in any way. Neither the slave mawla nor his patron had much latitude.
The free mawla and his patron did have more room to maneuver. The mawla could establish wala' with any person or tribe of his choosing; later he could dissolve the wala' without the accord of his patron and contract a new wala' with a new patron. He was restricted only if the patron had paid blood money on his behalf; then he had to remain with his patron. The free mawla lacked, the automatic protection of his slave counterpart, but had the freedom in which to make new arrangements. The right of a free mawla's patron unilaterally to terminate a wala' is disputed, however. The voluntary nature of the wala' of conversion gave it a political complexion absent from the wala' of manumission.
This distinction implies that the mawla who changed patron was free. Legal regulations, however, did not always hold. Did it become easier or harder with time to transfer wala'? Modern historians disagree. Goldziher believes it became easier:
Crone, to the contrary, thinks it became harder; before ca. 153/770 lawyers had recognized the transfer of wala' as legal but subsequently they forbade it on the basis of its affiliative (nasab) character.
In Jahili Arabia a woman once manumitted a slave, then transferred his wala' to her husband; a slave who died in 170/787 purchased his freedom from a woman and then another woman purchased his wala'. These two examples indicate that the slave mawla could transfer his wala' at both the beginning and the end of our period. Thus, the transfer of wala' did not indicate a free mawla with any certainty.
(10) Names. Certain names appear to have been used especially by slaves, so a mawla bearing such a name (e.g., Masrur, Salim, Kaysan) was most likely of slave origin.
(11) Tribal and personal wala'. Slave mawlas usually had a personal wala', that is their patrons were specific individuals; free mawlas more often had tribal wala'. There was no fundamental difference between these two, for every wala' included both a specific patron and his tribe, yet these distinctions do reveal different emphases. The freedman had a specific patron who manumitted him, so he was most likely to be called the mawla of his manumitter and took on the manumitter's genealogy. When 'Amr b. al-'As manumitted Wardan, Wardan was henceforth known as "Wardan the mawla of 'Amr b. al-'As." 'Amr himself was more visible than the tribal affiliation he provided.
The free mawla, to the contrary, affiliated less to an individual and more to a whole tribe. He need not have had an earlier relationship with any individual tribal member. The tribal affiliation might well overshadow his relations with a titular patron. Consequently, he was more likely to be known as the mawla of a tribe: "Abu Hanifa, mawla of the Banu Qufl."
This distinction did not always hold true. For example, Musa b. Nusayr, a captive, was known as the mawla of several different tribes.
(12) Fictive filiation. Slave mawlas lost their own genealogies more than did free mawlas; therefore they usually adopted their patron's. The sources usually list after the slave mawla's own name (his ism) that of his patron's and then the patron's ancestors': "Salim, the mawla of Muhammad b. Abi Yusuf b. Muhsin b. ..." The free mawla tended to retain his own family's genealogy, so two or more of his ancestors' names were recorded after his own: "Nu'man b. Siyyar b. Yassar, the mawla of Ibn 'Amir." Unlike the wala' of manumission, that of conversion "did not establish a tie of kinship, but accorded a privilege in return for a service."
If (11) and (12) always applied, then the following rule would hold: personal wala' and a single name go together and indicate a slave mawla; tribal wala' and a proper genealogy go together and indicate a free mawla. These rules hold usually, but not always. The names probably provide a more certain guideline than the wala'.
(13) Era. The mawla status passed from one generation to the next, even though the later generations experienced neither slavery nor conversion. With time, then, the percentage of first generation mawlas decreased; the percentage of slave mawlas dropped even more precipitously, since slaves became scarcer with the decline in Umayyad fortunes and termination of Muslim conquests. Increasingly, mawlas were the descendants of persons who had been enslaved or who had converted. The later the mawla, the more likely that he himself did not experience slavery.
(14) Employment. The mawla in the entourage of a caliph or high official was more likely to have slave origins than other mawlas, for the rulers had access to large numbers of captives. Mawlas not in their service but living in the cities or countryside were probably less often slave mawlas.
(15) Region of origins. Regions which resisted the Arabian conquerors suffered more battles and raids which produced slaves (e.g., Khurasan, the Daylami mountains, Transoxiana). Further, in some cases, such as the Sawad, it is possible that the entire civilian population was enslaved. Slaves and mawla-freedmen were more likely in lands which experienced the most warfare and enslavement.
(16) Context. The mention of mawlas together with slaves suggests that the mawlas too were slaves; when mentioned with an ethnic or political grouping, mawlas were probably non-Arabians, both slave and free. The first group of the following terms when used in connection with mawlas indicates slave mawlas, the second group indicates free mawlas.
Group I (Slave Mawlas)
Group II (Free Mawlas)
(17) Convention and consistency. Mawlas are erratically mentioned in the sources; one author alone may identify a person as a mawla, all but one may identify him as such, or they may divide more evenly. Our knowledge about mawlas reflects these vagaries. For example, the rebel leader Abu's-Saraya is called a slave in only one source; one account calls the military leader Hammad al-Barbari a slave mawla and the other does not even indicate he was a mawla; Abu 'Awn, an Abbasid governor of Egypt, is a mawla in only one of three accounts; most histories of the Muslim conquest of Spain state that Tariq b. Ziyad conquered Gibraltar with "mostly Berbers, with only a few Arabians," while one source specifies "mostly Berbers and mawlas, with only a few Arabians." Many other similar examples can easily be found.
Even within the same chronicle, there are inconsistencies. For example a bodyguard leader, Salma b. Abi 'Abdallah, is first mentioned by at-Tabari in 116/734; he is identified as a mawla in an account twelve years and four hundred pages later. Also, not every person who one expects to be a mawla is called one (e.g., a Berber leader converted to Islam, but nothing indicates he became a mawla.)
I suspect that writers more consistently identify mawlas who were of slave origins and who were closest to manumission or conversion, because for them the mawla status had more significance than for free mawlas and those of later generations. But one cannot draw any conclusions; the inconsistencies follow no pattern. For example, Wahshi, whom other writers described as a slave in many ways, is not once called a slave by al-Baladhuri, either in AA 1.322, 328, 363 or FB 89.
 M. Bloch, The Historian's Craft (New York, 1953), trans. P. Putman, p. 34.
 Ibn Manzur, Lisan al-'Arab (15 vols., Beirut, 1374-76/1955-56), 15.409; Tha'lab, Sharh Diwan Zuhayr b. Abi Sulma (Cairo, 1363/1944), p. 349.
 See various other dictionaries, such as: (a) Ibn Durayd, Kitab Jamharat al-Lugha, ed. M.S. Salma et al. (4 vols. Hyderabad, 1345-1351), 1.188. (b) al-Jawhari, as-Sahah, ed. A.'A. Gh. 'Attar (6 vols. Cairo, 1377), pp. 2529-30. (c) Ibn Faris, Mu'jam Maqayis al-Lugha, ed. A.S.M. Harun (6 vols. Cairo, 1366-71), 6.141.
For an elaborate treatment of some of the meanings of the term 'mawla' in the Qur'an, Qur'an commentaries, and the hadith, see al-Baltistani, an-Nahj as-Sawi.
 Mawla often means 'true slave' as well as 'of slave origins.' The Appendix lists examples of mawla meaning the same as other Arabic words for slave. FM 156 mentions the manumission of mawlas. Slaves who were subsequently manumitted are commonly referred to as mawlas, probably more so than slaves who never received their freedom.
Mawla in the sense of 'patron' can also be obscured, but when ambiguity exists, the patron is usually indicated as an 'upper mawla' (mawla min fawq, mawla a'la). On this, see az-Zayla'i 5.180; al-Jahiz, Manaqib 13; al-Marghinani 517; Crone, 'Mawali' 155; Biddle 54-55. For an example, UG 1.345.
The patron can also be known as sayyid: TAS 2.2253; Ibn al-Qutiya 39.
Mawla as slave owner: of a 'abd, MDh 3.187; of a jariya, MDh 3.325; of a khasi, Nawadir al-Makhtutat 1.297; of a ghulam, T3.1223.
Forand has suggested that 'the expression mawlaya ("my mawla") is never used in the sense of my "freedman," but always refers to the superior party' ('Development' 114). Though this is often correct, mawlaya means 'my lower mawla' in many cases, e.g.: IS 5.330; TMaw 62; ADA 269=T2.1852; Din 343; UH 3.314 (=T3.716), 339; T2.1249; 3.928, 1094; Ibn al-Qutiya 24; Jah 147; 'Iqd 4.400; Imama 2.99; IS 3.1.30.
 Ashtor 33 agrees: 'In the lands of the Fertile Crescent and in the adjacent countries of the Near East, the problem of the mawali was the crucial question.'
 Ayalon agrees: 'The importance of thorough and exhaustive examination of the term mawla as used under the Umayyads and early 'Abbasids should be particularly emphasized. ... The key to the beginnings of the mamluk institution might be found there. ... The correct understanding of the meaning of that term [mawla] is essential for the study of the slave institution in Islam' ('Preliminary Remarks' 47 and 48, fn. 4; see also 'Reforms' 4, 41).
 The Appendix deals with this problem.
 Personal communications, 11 June 1977; see also her 'Umayyad Clientage' 1 and 'Mawali' 99.
 Von Kremer 2.154-59; Goldziher's Volume 1, Chapter 3; van Vloten's Chapter 1. Azizi's Chapter 2 and Sharif passim take this viewpoint to its extreme statement. A discussion of this view follows below, 'The Standard Interpretation.'
 E.g., H. Lammens, La Syrie, Précis historique (2 vols.; Beirut, 1921), 1.61; Ashtor 29; Shaban, Islamic History 1.95; Hodgson 1.222; in many of W.M. Watt's writings.
 E.g., CHI 4.36; Spuler 34. S.I. Kashif, Misr fi Fajr al-Islam (2nd ed: Cairo, 1970), p.77 seems to think them non-Arabians.
 E.g., Ayalon 'Reforms' 1; al-Kharbutli, al-Mukhtar 280; Haas passim (despite the disclaimer in footnote I on p. 7); al-'Ali, at-Tanzimat 60.
 Besides Crone, 'Mawali' 96, 99-118; az-Zabidi 76-77; A.A. ad-Duri, Judhur at-Ta'rikhiya li'sh-Shu'ubiya (Beirut, 1962), p.18.
 Almost any translation of an early Arabic source into a European language provides numerous examples. Some at random: (a) G. Wiet, Les Pays (Cairo, 1937), pp. 15-22, 45, 61. (b) E. Fagnan, Al-Bayano' l-Mogrib (2 vols.: Algiers, 1904), 2.61, 130. (c) F.C. Murgotten, The Origins of the Islamic State (New York, 1924), p. 251 implicitly understands 'lower mawla' where 'upper mawla' is meant (cf. Ayalon, 'Preliminary Remarks' 45). (d) Combe 1.194-95.
 Sezgin 2.99; 'Iqd 3.416-7, 6.77.
 Sezgin 1.358; R. Guest, 'Introduction' to al-Kindi, Kitab al-Wulah wa Kitab al-Qudah, p.10; Ibn Hajar al-'Asgalani, Tandhib at-Tandhib (Hyderabad, 1325-27), 1.30; al-Maqrizi, al-Mawa'iz 2.137, 202, 250.
 Sezgin 2.98; Ibn an-Nadim 80.
 Ibn an-Nadim 145-6; Ibn Khallikan 6.106.
 The following discussion of the mawla status generally ignores the complex and arcane possibilities which take up so much space in the law books, for they account for only a small fraction of all the mawlas. The problems of mawlas with non-Muslim patrons, of manumission in Dar al-Harb, of apostates, foundlings, and so forth involve but few persons.
Also, we shall ignore perplexing references such as one to a 'mawla of God and his Prophet' (al-Maqrizi, al-Mawa'iz 2.137).
 But not the meaning of a political agent, which is not a legal status and has no precedent in Jahili Arabia.
 E.A. Belyaev, Arabs, Islam, and the Arab Caliphate in the Early Middle Ages, tr. A. Gourevitch (Jerusalem, 1969), p. 62; Tyan 1.23.
 W. Robertson Smith Kinship and Marriage in Early Arabia (2nd ed. rev.: London, 1903) deals with this intricate matter at length.
 The literature on this subject does not, as I do, restrict hilf to free Arabians and understand wala' as applying to all other persons; indeed, Tyan considers the two terms identical in meaning (1.25-26 and 'Hilf' in EI²).
I make this distinction on the basis of information from the biographies of the Sahaba, from the following points:
(a) The scores of halifs that appear in Usd al-Ghaba all appear to be Arabian, for they all have their own proper genealogies. A genealogy normally indicates an Arabian.
(b) All mawla-allies appear to be non-Arabians (they are listed in Table 1); note also KM 141.
(c) A person who is thought to be either a halif or a mawla-ally usually has disputed ethnic origins. Since the biographers are unsure of his background, they cannot assign him either a hilf or wala' relation with certainty. The problem arises in the following biographies: Habib b. Aswad; Khabab b. al-Aritt; Sa'd b. Khawla; Samra b. Jandab; Suhayb b. Sinan; 'Ammar b. Yasir.
I cannot explain the occurrence in Islamic times of a wala' between tribes (T 2.491) or between a tribe and an individual patron (TMaw 91) even if he is the caliph (Ibn al-Qutiay 25). See also al-Jahiz, Mu'awiya 26.
 Goldziher 1.107.
 IS 4.1.77.
 Tyan 1.24-25.
 Belyaev, Arabs, Islam and the Caliphate, p. 61.
 Robertson Smith, Kinship and Marriage, p. 52.
 Zaydan 4.21.
 KM 139; Umm 4.58.
 Zaydan 4.21. Perhaps the mawla of a tribe or a city was in fact the mawla of its leader?
 F.L. Ganshof, Feudalism, trans. P. Grierson (3rd ed. rev.: New York, 1964), pp. 84, 87, 92. The mawla's relations with his patron do not meet the definition of patronage established by E. Gellner, 'Patrons and clients,' in E. Gellner & J. Waterbury (eds.), Patrons and Clients in Mediterranean Societies (London, 1977), pp. 1-2.
 Crone, 'Mawali' 157, 229-30, n. 10.
 al-'Ali, at-Tanzimat 66.
 Goldziher 1.106, translated by S. Stern (1.102).
 Tyan 1.24; al-Marghinani 513; Ibn Khaldun 1.245 also calls them mawla riqq or mawla istina' wa hilf.
 Tyan 1.24; Goldziher 1.106; Forand, 'Relation,' pp. 59-60; Biddle, p. 55.
 Crone, 'Mawali,' p. 155;'Umayyad Clientage,' p. 9; and the dictionaries cited in notes 2 and 3.
 The slave still remains outside the social system in this respect. Cf. E. Tyan, 'Diya' in EI².
 Qur'an 5.106.
 Yusuf Ali's fn. 809 on p. 275 of his translation of the Qur'an.
 Umm 6.189 indicates the manner of this transition. For brief discussions, see R. Brunschvig, 'Abd' in EI²; T.P. Hughes, A Dictionary of Islam (London, 1885), p.555.
 A ubiquitous hadith; for discussion see Umm 4.52ff, 7.208. Umm 4.56 adds, 'and it is never transferred.'
 Umm 6.187.
 Umm 4.52, 56, 7.209 paraphrase this: 'The wala' is to no one but the manumitter' (al-wala' la yakun bi-hal illa'1-mu'tiq). Umm 4.52 illustrates this with examples from the Prophet's life.
 Umm 6.187.
 az-Zayla'i 5.181. For a rare mention of sa'ibas in (possibly) Islamic times, see al-Fakihi 36.
 This is a controversial matter. Cf. Goldziher 1.141; J. Schacht, The Origins of Muhammadan Jurisprudence (Oxford, 1950), p.173; Crone, 'Mawali,' p. 159 and 'Umayyad Clientage,' p. 4.
For examples of the sale of wala' in (probably) Islamic times, see Ibn Khallikan 6.89; KM 253.
 Umm 4.52.
 In Africa, the wala' relationships apparently still exists: J.S. Trimmingham, The Influence of Islam Upon Africa (London, 1968), p. 93.
 Qur'an 49.13.
 Qur'an 49.14.
 Qur'an 33.5.
 az-Zayla'i 5.177 recognizes this explicitly.
 ash-Shayib, pp. 265-66. My conclusions are drawn largely from Umm 4.51. The Qur'an commentaries say nothing about the social implications of this verse; I have searched in more than a dozen of them without profit.
 Amin 107.
 Umm 4.52, 56.
 The 'Constitution of Medina' refers to Jewish mawlas (Abu 'Ubayd 263-64=IH 1.503-04). See also Ibn Khallikan 5.189.
 Umm 4.56; Crone, 'Mawali' Chapter 4.
 Umm 6.186; al-Qaurdi, p. 88. The legal books go out of their way to point out that the mawla-convert does not acquire mawla status on the grounds that he comes to life in Islam just as a slave comes to life with manumission (as-Sarakhsi, al-Mabsut 8.92).
 An example: Tayfur 132.
 Some details may be found in ADA 254=Ta'rikh al-Khulafa (Moscow, 1967), pp. 523-24.
 Asl 4.1.194 explains why; an Arabian already has a tribal affiliation.
 Umm 7.121.
 This hadith, like the one discussed in note 44, appears in nearly every consideration of mawlas. For some interesting variants and remarks, see az-Zubayr, p. 333; Ibn al-Qutiya, p. 76; Abu 'Ubayd, p. 311.
 az-Zayla'i 5.177.
 The dhimmi status contains vestigial elements of the wala' of alliance. Cf. Asl 4.1.194, lines 8-10.
 The mawlas who joined the Muslims before Muhammad's death retain their high status even when new mawlas occupy inferior positions. They stand apart from the changes which follow.
 UG 4.317-18. This story is open to grave questions: it seems not to appear in the earlier accounts of Muhammad's life. It may well be a later fabrication.
 Asl 4.1.249-53 provides regulations for manumitting non-Muslim slaves.
 Cahen, 'Histoire economico-sociale,' p. 203; Zaydan 4.45, 47. Since non-Muslims ought not to fight for Islam, it seems yet more likely that mawla-freedmen who did fight were converts.
Wala' occurs even when patron and mawla are of different religions (Sachau, p. 141).
 Or a mawla-freedman-convert, for conversion may follow manumission and this changes nothing. Cf. Asl 4.1.252-53.
 Asl 4.1.175.
 az-Zayla'i 5.177.
 al-Marghinani, p. 518.
 Asl 4.1.253.
 Asl 4.1.251-53 provides information on regulations for non-Muslim mawlafreedmen.
Crone, 'Mawali,' pp. 90-100 suggests that a mawla-freedman can sometimes be called a mawla-muwalah, but I have found no support for this statement.
 Abu 'Ubayd 177, 181; Zaydan 4.46.
 Asl 4.1.177, 194.
 The abna' of the Yemen: UG 2.129, 4.186, 227. Asawira: FB 373 tells how they allied (halafa) with an Arabian tribe.
 Kremer 2.154-62. Some of the repetitions and elaborations of his account include the following: Goldziher 1.139; Zaydan 2.29, 4.87, 119; Azizi Chapter 2; Sharif [German] 17; F. Loekkegaard, Islamic Taxation in the Classic Period (Copenhagen, 1950), p.128; Spuler 35; al-Kharbutli, al-Hukm, pp. 162-69; al-Kharbutli, al‑Mukhtar, pp. 288-90; az-Zabidi, pp. 83-84; ar-Rawi, pp. 209-210; CHI 4.34-38; Ashtor, p. 29; B.Turner, Weber and Islam (London, 1974), p. 86; A.A. Dixon, The Umayyad Caliphate 65-68/684-705 (London, 1971), p.48.
 Arabian racial arrogance: CHI 4.42; Zaydan 4.89.
'Umar I's assassination: Azizi 44-45; al-Kharbutli, al-Mukhtar 282; M.T. an-Najjar, al-Mawali fi'l-'Asr al-Umawi (Cairo 1368/1949), p.24.
 Kremer 2.156; M. Canard, 'L'expansion arabe: le problème militaire,' L'Occidente e l'Islam nell'alto medioevo (Spoleto, 1965), p.41; Hodgson 1.199. Islam adding cohesion: Donner, p. 191.
 Hodgson 1.223.
 Ibid. 1.226; Crone, 'Umayyad Clientage,' p. 15.
 Azizi, p. 33; Crone, 'Umayyad Clientage,' p. 14.
 Crone; 'Umayyad Clientage,' p. 17.
 Dennett, Conversion, p. 88.
 Ibid., p. 115.
 Converts did gain exemption from the poll tax on occasion in the Sawad (ibid., p. 42), Mesopotamia (ibid., p. 48), Syria (ibid., p. 62), and Egypt (ibid., p. 83). Nobles often retained their high status through conversion (Hodgson 1.242).
 Crone, 'Mawali,' p. 101.
 'Iqd 3.413.
 Din, pp. 288, 293.
al-Kharbutli, al-Mukhtar, p. 287.
 Equal pension pay: FB 455. Salman al-Farisi was a slave at the Battles of Badr and Uhud (UG 2.330), yet he received 4,000 dirhams yearly at 'Umar's special orders and 'Ammar b. Yasir received 6,000 (Abu 'Ubayd 301).
High posts: Salim, the mawla of Abu Hudhayfa, is mentioned by 'Umar I (as 'Umar is dying) as a candidate to succeed him as caliph; Salim, however, was already long dead (T1.2776-77). Shaqran, a slave of the Prophet's, took charge of the division of the booty at al-Muraysi' (IS 3.1.34).
Salman al-Farisi: his symbolic importance in the Islamic tradition is explained in L. Massignon, Salman Pak et les prémices sprirituelles de l'Islam iranien (Tours, 1934). Note also the extensive bibliography, pp. 47-52.
 Crone, 'Umayyad Clientage,' p. 14; Biddle, pp. 55-56; B. Lewis, Race and Color in Islam (New York, 1971), p. 23.
 For example, Azizi, p. 52.
 Zaydan 4.44 thinks that the Arabians viewed the conquests primarily as a way to acquire slaves.
 Mawla resentment of the Arabians: Zaydan 4.87-88; Grunebaum 73; Hodgson 1.283; Ashtor 23.
Zaydan 4.58 suggests that non-Arabians accepted the primacy of Arabians out of respect for their founding a new religion.
 al-'Ali, at-Tanzimat 67; az-Zabidi 74-75; and Crone, 'Mawali' 96 recognize aspects of their similarity.
 Differences are discussed in the Appendix.
 See above, 'Implications of the mawla-convert status.'
 See above, 'Implications of the mawla-convert status.'
 Bulliet, p. 33. Also Dennett, Conversion, p. 39. Ashtor, p. 66 points out that Islamicate rule brought 'a deterioration in the conditions of peasant life.'
 Cahen, 'Histoire économico-sociale,' p. 205. They lost economically in one sense, for the Muslim may not inherit from non-Muslim relatives. Cf. Umm 4.53, lines 21-22; J. Schacht, 'Mirath' in EI¹; N.J. Coulson, Succession in the Muslim Family (London, 1971), p.187.
 Crone, 'Mawali,' p. 101 and 'Umayyad Clientage,' p. 17.
 Cahen, 'Histoire économico-sociale,' pp. 208-09.
 ash-Shaybani, Asl 4.1.193, lines 12-13 discusses the legal continuation of wala' beyond two generations.
 (a) Asl 4.1.176. Note also al-Marghinani 515. (c) Asl 4.1.192. (e) ibid. 4.1.176f; Goldziher 1.127-33; az-Zabidi, pp. 78-79.
An adult son may choose his own patron (Asl 4.1.196). This may explain how a father and son had different patrons (e.g., al-Jahiz, Fakhr 1.179).
For the variety in the children of a single woman, note Sumayya: one child a Qurashi, one an Arabian, and one a mawla! ('Iqd 6.133). ADA 191 tells of a mawla's son accepted as a saliba.
 KM 248; J. Schacht, 'Abu Hanifa al-Nu'man b. Thabit' in EI².
 A. Guillaume, 'Abu'l-'Atahiya' in EI².
 KM 214; Ibn Khallikan 2.517, 3.88.
 Ibn Tiqtaqa, p. 177-78; Jab 125; Sourdel 1.88; KB 241 mentions his father as the mawla of al-Mansur.
 Abu 'Ubayd, p. 103; Noth, p. 155. For a general analysis of the conquests' legal implications, see Schmucker.
 Noth, p. 153.
 For Egypt: Noth, pp. 151-52; Hill, pp. 34-53: R. Brunschvig, 'Ibn 'Abdalh'akam et la conquête de l'Afrique du Nord par les Arabes: Etude critique,' Annales de l'Institut d'Etudes Orientales, 6 (1942-47), pp.115-19.
For the Sawad, see the following note.
Umayyad jurists: Noth, pp. 161-62.
An Armenian historian states that all Armenia was taken into slavery by the Muslims in the Armenian year 300 (Stepanos Tarontsi 1.135).
 Hill, pp. 99-118; Schmucker, pp. 96-134; Donner, pp. 100-137; Noth, p. 152.
 Forand, 'Status,' p. 30. For a similar situation, note the encomienda system which flourished in 16th century Spanish Latin America; free persons had to undertake servile labor. See C. Gibson's summary in Spain in America (New York, 1966), pp.48-67.
 Forand, 'Status,' pp. 36, 32-34.
 Abu 'Ubayd, p. 185.
 Abu 'Ubayd, pp. 240, 193, 238; FB 225.
 Quoted in Amin, p. 106, fn. 2. I have been unable to find this pasage in az-Zayla'i's published works.
 ash-Shayib, p. 266; also Amin, p. 103-07; A. von Kremer Culturgeschichtliche Streifzüge auf dem Gebiete des Islams (Leipzig, 1873), pp.10-11; Zaydan 4.47; Haas, pp. 6-7; al-Kharbutli, al-Mukhtar, p. 280.
 Crone, 'Mawali,' p. 96
 Some who are will be found listed in the Appendix.
 Many historical sources provide an opinion on the subject of Abu Muslim's mawla type. The following call him a mawla-freedman (some by implication): T2.1726, 1916; Din 337-38; al-Maqdisi 6.61-62, 92-93; TYa' 2.367; Ibn Tiqtaqa, p. 139; Theophanes, p. 66. T2.1769 indicates he was a mawla-freedman, contrary to Abu Muslim's own claims. Some other sources include: MDh 3.254; UH 3.18283; TMaw 53; KM 214 (calls him Kurdish). Abu'l-Fida 1.208; ADA 253-67; Ta'rikh al-Khulafa (Moscow, 1967), 523-24.
 Asl 4.1.277.
 Zayd b. Haritha: V. Vacca, "Zaid b. Haritha" in EI. Shaqran: UG 3.2; IS 3.1.34; AA 1.478-79. Anasa: UG 1.132; AA 1.478; IS 4.1.33.
 Abu Muslim: see note 129. Abu Salama: S. Moscate, 'Abu Salama Hafs b. Sulayman al-Khalal' in EI²; Sourdel 1.65-73.
 Khurasaniya: Omar, 'Composition,' pp. 165-68. The quote from al-Jahiz comes from Manaqib al-Atrak, p. 14-15.
 Crone, 'Umayyad Clientage' 8.
 Personal communication, 7 September 1977. See also CHI 4.40.
 Between 132/750 and 205/820, a region broke away fom Abbasid control approximately every five years. These can be traced in E. von Zambaur, Manuel de Généalogie et de Chronologie pour l'histoire de l'Islam (Hanover, 1927).
 The existence of mawlas can be established but more detailed information about their status is hard to find.
 Aghani 10.61.
 ar-Rashid, p. 222.
 Pipes, p. 196.
 Ibid., pp. 174-81.
 W.M. Watt, Islam and the Integration of Society (London, 1961), p.168; Hodgson 1.305.
 Bullet, p. 34. Note that both this and the following quote are referring only to mawla-converts, not to all mawlas.
 CHI 4.309; see also Grunebaum, p. 80; G. Rotter, Die Stellung des Negers (Bonn, 1967), p. 113; D. Ayalon, "Aspects of the mamluk phenomenon" Der Islam 53 (1976), p. 205, fn. 17.
 Also, the Muslims in those first years were much closer to the Jahili period, when the mawla status had an important and beneficial role.
This raises the question: did subsequent dynasties founded by religious movements reinstate the mawla-convert status for pious reasons?
 For an example, see Aghani 10.61.
 Hilal as-Sabi' 104, 122-23, 131; Ibn Khaldun 1.245-46, 332-33.
 Shaban, Islamic History, p. 2.9.
 Crone, 'Mawali,' p. 140.
 Shaban, Islamic History 2.9. Also O. Grabar, The Coinage of the Tulunids (New York, 1957), footnote 5 on pp. 39-40.
 Crone, 'Mawali,' p. 135.
 P. Balog, "Pious invocations probably used as titles of office or as honorific titles in Umayyad and 'Abbasid times," Studies in Memory of Gaston Wiet, ed. M. Rosen-Ayalon (Jerusalem, 1977), p. 62. Also Balog, Umayyad, p. 4.
 Crone, 'Mawali,' p. 96-97, referring to Sogdiiski Sbornik (St. Petersburg, 1934), p. 55.
 Ibid., pp. 96-99 documents a number of irregular wala's dating from the Arabian period; they are only a small fraction of the total number of mawlas that we know of.
 Omar, 'Composition,' p. 169. Biddle, p. 66 states that mawlas in the Abbasid period 'were full blooded, highly placed Arabs,' but he cites no specific proof.
 KB 245-53. On the governors, see KB 248 and T3.405 and TYa' 2.372, 384. Many of the mawla governors mentioned on TYa' 2.384 must have been mawla amir al-mu'minirs.
 In an unusual passage, Ibn al-'Adim (d. 660/1262) mentions having seen this ancient coin and records its inscription on 1.60. The British Museum has three of them (Lane-Poole 1.197). The Bibliothèque Nationale only has two (Lavoix, pp. 444-45), Leningrad has one (Tiesenhausen, p. 92, with references).
 184/800: Miles, p. 76, with a discussion of the mawla's possible identity on p. 77. 185/801: ibid., p. 78; Lane-Poole, pp. 211-12.
 Against the 'Alid: T3.982. Against Babak: TMaw 386.
 Crone, 'Mawali,' p. 136.
 Berber: Ibn Rusta, p. 75. Bukharan: al-Kindi, pp. 193 and 445. Tabaristani: Mazyar (on whom, see text at fns. 170-74).
 As well as the major figures who follow, note also the person mentioned on Combe 1.102.
 TYa' 2.387, 409.
 Five: al-Qalqashandi 14.89. Seven: al-Azraqi, p. 165-66.
 Grohmann, From the World, p. 141.
 Ibn Isfandiyar, pp. 146-47.
 FB 339.
 T3.1268-1300 has the fullest account of this rebellion.
 TYa' 2.476-77; also Ibn al-Faqih, p. 309.
 FB 134.
 Both times by al-Ya'qubi: KB 277 and TYa' 2.477.
 Historians: V. Minorsky, 'Mazyar' in EI¹; G. Wiet, Les Pays (Cairo, 1937), p. 81; E. Marin The Reign of al-Mu'tasim (833-842) (New Haven, 1951), p. 107, n.499.
I base my translation on A. de Biberstein Kazimirski, Dictionnaire arabe-français (2 vols.: Paris, 1860), 2.1609; H. Wehr, A Dictionary of Modern Written Arabic, ed. J.M. Cowan (Wiesbaden 1961), p.1101.
 TYa' 2.477. T3.1583 calls Qarin b. Shahriyar a mawla amir al-mu'minin.
 Shaban, Islamic History 2.9-10 is the worst offender; Omar, 'Composition,' p. 169.
 There is one figure I cannot account for; of 50,000 mawlas in about 151/768, only 12,000 less than a quarter of the total, were freedmen ('itaqa), according to ar-Rashid, p. 222. Who then were the other mawlas? They may have been converts who still continued to be known as mawlas, although their status had disappeared with the Abbasid take over some twenty years earlier. The same explanation may account for the many mawlas of the Hashimi family who received khums from Harun ar-Rashid (al-Maqdisi 6.101).
For subsequent usage of mawla, see al-Qalqashandi 6.31-32 and Ayalon, 'Eunuchs,' p. 280, both discussing its use in the Mamluk Kingdom of Egypt. Cf. Note 51.
 See the reference in Note 141.
 Dhamm Akhlaq al-Kuttab, p. 190.
 Shaban 2.63-66 states this most strongly. Cf. Pipes, pp. 14-15.
 Materials on slaves and mawlas in warfare are collected in Chapter 4 of Slave Soldiers and Islam. Their role in politics will be published eventually as 'Government Slaves in Early Islam.' [DP note: I did not publish that study.] Biddle's lists provide some information on this; Haas, pp. 9-27 is more complete. Most of Haas' study deals with slaves' and mawlas' cultural role (in religion, arts and crafts, music and poetry, grammar, and learning). On the hajibs see Ibn Habib, pp. 258-60.
 Chapter 6 of Slave Soldiers and Islam.
 Synonyms in Usd al-Ghaba are omitted here and may be found among the entries listed in Table 1.
 KM 270, also KM 233.
 KM 241.
 Von Kremer 2.155; Sachau, p. 141; az-Zabidi, p. 79; Crone, p. 158. 'Iqd 3.413.
 al-Khassaf (d. 261/874), Kitab al-Awqaf (Bulaq, 1904), pp. 115-116, quoted in Forand, 'Relation,' pp. 64-65. Also Umm 7.121. Actual cases: FM 135; T3.1236.
 For an example, FM 99.
 For example, KM 225,
 Asl 4.1.196, 273.
 az-Zayla'i 5.178; Asl 4.1.193.
 Jahili Arabia: KM 139. Caliph's family: KM 235, 261; Tayfur, p. 299; Ibn al-Faqih, p. 284.
 BM 2.25 and Khalifa, p. 430 in contrast to BM 2.27; TYa' 2.294.
 MDh 3.335.
 Asl 4.1.194.
 One example: Ibn Yahya b. Mus'ada in Nawadir al-Makhtutat 1.288. For a discussion, Zaydan 4.85-86.
 Quoted by Abu'l-Fida 1.208; for three further references, see Crone, 'Mawali,' note 103 to Chapter 4.
 Hayyan an-Nabati's mawla: T2.1582. Governor of Ifriqiya: BM 1.51 and 2.29.
 IS 5.211 = KM 288 = Ibn Rusta 228.
 To the Prophet: FM 93 and 100; Aghani 17.9 (this goes only as far back as Abu Bakra but he was a mawla of Muhammad's; cf. UG 5.151).
To a caliph: KM 112 = 258, 245 (the last has two maqits since Humran [on the first line] was the mawla of 'Uthman b. 'Affan); ITB 2.40 (to a caliph's son).
 Tanbih 346 = Jah 301; KM 245; T3.21 and 75, 76.
 Aghani 17.9.
 KM 245.
 Legal references in note 46; FB 344 gives a specific case and Jah 270 an exception.
 No conditions: Asl 4.1.202; al-Marghinani, p. 514. No disposal: Asl 4.1.200-205.
 Asl 4.1.196-97; al-Marghinani, p. 518.
 Asl 4.1.182. AA 4a.62 tells of an example of bloodmoney paid for a mawla.
 al-Marghinani, p. 518 says he may; Sachau, p. 126 says he may not.
 Crone, 'Mawali,' p. 159 and 'Umayyad clientage,' p. 4, quoting J. Schacht, The Origins of Islamic Jurisprudence (Oxford, 1950), p. 173.
 Jahili Arabia: KM 139. Slave who died in 170/787: KM 253.
 Other examples of wala' transfer: AA 1.485, 11.254 and 260; T2.834, 1809 and 1913, 3.927 and TMaw 252; Din 286. Ibn Khallikan 6.89.
 On these names, see the index to at-Tabari, vol. 1. Hrbek, 'Die Slawen im Dienste der Fatimiden,' Archiv Orientálni 21 (1953), pp. 557, 566 fn. 151 (Hrbek has not published the intended study on slave names, however.) See also Ayalon, 'Eunuchs,' pp. 274-79.
 Forand, 'Relation,' p. 65.
 FB 247; T1.2064; al-Kindi, p. 52; Khalifa, p. 103; al-Maqqari 1.141, 156.
 Crone, 'Umayyad clientage,' p. 6. See also Forand, 'Relation,' p. 62. Ibn Khaldun 1.246 appears to contradict this (as Forand points out); the difference is resolved by remembering that by Ibn Khaldun's time no mawla-converts had existed for centuries, so when he wrote 'mawlas' he meant only mawla-freedmen.
 Ayalon, 'Reforms,' p. 41 has a similar listing of mawlas in combinations.
 Abu's-Saraya: Bal'ami 4.499. Hammad al-Barbari: TYa' 2.412-13. Abu 'Awn: al-Kindi, p. 101; not FM or ITB. Tariq b. Ziyad: al-Maqqari 1.142, 159; AM 6.
 ITB 1.158.
An asterik (*) indicates that I have relied on a translation.
Abu'l-Faraj al-Isfahani. Kitab al-Aghani. 20 vols. Bulaq 1285/1868-69. Index: I. Guidi, Tables alphabétiques du Kitab al-Agani. Leiden, 1306.
_______. Maqatil at-Talibiyin. Ed. S. A. Saqr. Cairo, 1368/1949.
Abu'l-Fida. al-Mukhtasar fi Akhbar al-Bashar. 4 vols. Cairo, 1325.
Abu 'Ubayd. Kitab al-Amwal. Ed. M.K. Ibrahim. 2nd ed. Cairo, 1395/1975.
Abu Yusuf. Kitab al-Kharaj. Ed. Q.M. al-Khatib. 5th ed. Cairo, 1396.
Agapius [=Mahbub]. Kitab al-'Unwan. Ed. and tr. A. Vasiliev in Patrologia Orientalis, 5(1909), pp. 557 -692; 7 (1911), pp.457-591; 8 (1912) pp.399-550.
Akhbar ad-Dawla al-'Abbasiya. Ed. A.A. ad-Duri and A.J. al-Muttalibi. Beirut, 1971.
Akhbar Majmu'a fi Fath al-Andalus. Ed. E. Lafuente Y. Alcantara. Madrid, 1867.
al-Azdi, Abu Zakariya Yazid. Ta'rikh al-Mawsil. Ed. 'A. Habiba. Cairo, 1397/1967.
al-Azdi, Muhammad b. 'Abdallah. Ta'rikh Futuh ash-Sham. Ed. A.A. 'Amir. Cairo, 1970.
al-Azraqi. Akhbar Makka.Ed. F. Wüstenfeld. Leipzig. 1858.
al-Baladhuri. Antab al-Ashraf.
vol. 1. Ed. M. Hamidullah. Cairo 1959.
vol. 2. Ed. M.B. al-Mahmudi. Beirut, 1394/1974
vol. 4a. Ed. M. Schloessinger and M. Kister. Jerusalem, 1971.
vol. 4b. Ed. M. Schloessinger. Jerusalem. 1938.
vol. 5. Ed. S.D.Goitein. Jerusalem, 1936.
vol. 11. Ed. W.G. Ahlwardt. Greifswald, Germany, 1883.
_______. Futuh al-Buldan. Ed. M.J. de Goeje, Leiden 1866. Also : Ed. S. Munajjid. Cairo, 1956.
*Bal'ami. Tarjama-yi Tarikh-i Tabari. Lucknow 1291/1874. Translation: H. Zotenberg, Chronique de Tabari. 4 vols. Nogent-le-Rotrou, 1867-74.
Balog, P. Umayyad, 'Abbasid, and Tulunid Glass Weights and Vessels Stamps. New York, 1976.
Combe, E. et al. Répertoire chronographique d'épigraphie arabe. 16 vols. Cairo. 1931-64.
adh-Dhahabi. Tajrid Asma'as-Sahaba. Ed. S.A.H. Sharaf ad-Din. 2 vols. Bombay, 1389-90/1969-70. References to this book are by biography numbers, not pages.
ad-Dinawari. al-Akhbar at-Tiwal. Ed. A.M. 'Amir and G. ash-Shayyal. Cairo, 1960.
Diyab al-Iklidi. A'lam an-Nas bima Waqa'a li'l-Baramika ma'a Bani al-'Abbas. Cairo, 1280.
al-Fakihi. Ta'rikh Makka. [A segment] Ed. F. Wüstenfeld. Leipzig, 1859.
Grohmann, A. Arabic Papyri in the Egyptian Library. 6 vols. Cairo, 1936-62.
_______. From the World of Arabic Papyri, Cairo, 1952.
Hilal as-Sabri'. Rusum Dar al-Khilafa. Ed. M. 'Awwad. Baghdad, 1383/1964.
Ibn 'Abd al-Hakam. Futuh Misr wa Akhbarha, Ed. C.C. Torrey, New Have, 1922.
Ibn 'Abd Rabbih. al-'Iqd al-Farid. Ed. A. Amin et al. 7 vols. Cairo, 1940-53.
Ibn al-'Adim. Zubdat al-Halab min Ta'rikh Halab. Ed. S. ad-Dahhan. 3 vols. Damascus, 1951-68.
Ibn al-Athir. Usd al-Ghaba fi Ma'rifat as-Sahaba. 5 vols. Cairo, 1280.
Ibn Habib. Kitab al-Muhabbar. Ed. E. Lichtenstädter. Hyderabad, 1361/1942.
Ibn Hisham. as-Sira an-Nabawiya. Ed. M. Saqa et al. 2 vols. 2nd ed. Cairo, 1375/1955. Translation: A. Guillaume. The Life of Muhammad. Lahore, 1967.
Ibn 'Idhari. al-Bayan al-Mughrib.
vol. 1. Ed. R. Dozy et al. Leiden, 1948.
vol. 2. Ed. R. Dozy et al. Leiden, 1951.
vol. 3. Ed. É. Lévi-Provençal. Paris, 1930.
vol. 4. Ed. I. 'Abbas. Beirut, 1967.
*Ibn Isfandiyar. Tarikh-i Tabaristan. Tehran, 1942. Translation: E.G. Browne. An Abridged Translation of the History of Tabaristan. Leiden and London, 1905.
Ibn Khaldun. al-Muqaddima. Ed. E. Quatremère. 3 vols. Paris, 1958.
Ibn Khallikan. Wafiyat al-A'yan. Ed. I. 'Abbas. 8 vols. Beirut, [1968-72].
Ibn an-Nadim. al-Fihrist. Cairo. 1948.
Ibn Qutayba. Kitab al-Ma'arif. Ed. F. Wüstenfeld. Göttingen, 1850. Also: Ed. T. 'Ukasha. Cairo, 1969.
(pseudo-) Ibn Qutayba. al-Imama wa's-Siyasa. 2 vols. Cairo. 1355-56/1937.
Ibn al-Qutiya. Ta'rikh Iftitah al-Andalus. Ed. J. Ribera. Madrid, 1868.
Ibn Rusta. al-A'laq an-Nafisa. Ed. M.J. de Goeje. Leiden, 1892.
Ibn Sa'd. Kitab at-Tabaqat al-Kabir. Ed. E. Sachau et al.9 vols. Leiden, 1905-40.
Ibn Taghri Birdi. an-Nujum az-Zahira. Ed. A.Z. al-'Adawi et al.16 vols. Cairo, 1348-92/1929-72.
Ibn Tiqtaqa. al-Fakhri. Beirut, 1386/1966.
al-Jahiz. Fakhr as-Sudan 'ala'l-Baydan, in Rasa'il al-Jahiz. Ed. A.S.M. Harun. 2 vols. Cairo, 1384/1964. 1.177-226.
_______. Fi Dhamm Akhlaq al-Kuttab in ibid., 2.183-209.
_______. Fi Mu'awiya wa'l-Umawiyin. Ed. 'I.A. al-Husayni. [Cairo], 1364/1946.
_______. Manaqib al-Atrak in Majmu'at Rasa'il. Ed. M. as-Sasi al Maghribi. Cairo, 1324, pp. 2-53.
al-Jahshiyari. Kitab al-Wuzara' wa'l-Kuttab. Ed. M. Saqa et al. Cairo, 1357/1938.
Khalifa b. Khayyat. Kitab at-Ta'rikh. 2 vols. Ed. S. Zakkar. Damascus, 1967-68.
al-Kindi. Kitab al-Wulah wa Kitab al-Qudah. Ed. R. Guest. Leiden, 1912.
Lane-Poole, S. and R.S. Catalogue of the Oriental Coins in the British Museum. 10 vols. London 1875-90.
Lavoix, H. Catalogue des monnaies musulmanes de la Bibliothèque Nationale. Khalifes orientaux. Paris, 1887-96.
al-Maqdisi. Mutahhar b. Tahir. al-Bad'wa't-Ta'rikh. Ed. C. Huart. 6 vols. Paris, 1899-1919.
al-Maqrizi. al-Mawa'iz wa'l-l'tibar [al-Khitat]. 2 vols. Bulaq, 1270.
al-Mas'udi. Muruj adh-Dhahab. Ed. M.M. 'Abd al-Hamid. 4 vols. 5th ed. Cairo. 1393/1973.
_______. at-Tanbih wa'l-Ashraf. D. M.J. de Goeje. Leiden, 1894.
Miles, G.C. The Numismatic History of Rayy. New York, 1938.
Nawadir al-Makhtutan. Ed. A.S. Harun. 2 vols. 2nd ed. Cairo, 1392-93/1972-73.
al-Qaduri. Al-Matn. 3rd ed. Cairo. 1377/1957.
al-Qalqashandi. Subh al-A'sha. 14 vols. Cairo, 1913-22. Index: M.Q. al-Baqli. Faharis Kitab Subh al-A'sha. Cairo, 1972.
al-Qur'an. Text and translation: A.Y. Ali. The Holy Qur'an. N.p., 1946.
ar-Raqiq al-Qayrawani. Ta'rikh Ifriqiya wa'l-Maghrib. Ed. M. al-Ka'bi. First part. Tunis, 1968.
ar-Rashid b. az-Zubayr. adh-Dhakha'ir wa't-Tuhaf. Ed. M. Hamidullah. Kuwait, 1959.
as-Sarakhsi. al-Mabsut. Ed. M. Radi. 30 vols. Cairo, 1324-31.
ash-Shafi'i. al-Umm. 7 vols. Cairo 1321-25.
ash-Shaybani. Kitab al-Asl. Ed. M. 'Abd al-Mu'id Khan. 4 vols. (incomplete). Hyderabad, 1966-. Partial translation: M. Khadduri. The Islamic Law of Nations. Baltimore, 1966.
*Stepanos Taronts'i. Patmut'ium Tiezerakan. Ed. Malkhasiants'. St. Petersburg, 1885. Translation: E. Dulaurier, F. Macler, Histoire Universelle: 2 vols. Paris, 1883-1917.
at-Tabari. Ta'rikh al-Rusul wa'l-Muluk. Ed. M.J. de Goeje et al. 3 vols. in 15 parts. Leiden, 1879-1901. Also : Ed. A.F. Ibrahim. 10 vols. Cairo, 1960-69.
Ta'rikh al-Khulafa'. Ed. P.A. Gryaznevitch. Moscow, 1967.
Tayfur. Kitab Baghdad. [A segment] Ed. H. Keller. Leipzig, 1908.
*Theophanes. Chronographia. Ed. C. de Boor. Leipzig, 1883-85. Translation : L. Breyer. Bilderstreit und Arabersturm in Byzanz : das 8. Jahrhundert (717-813) aus der Weltchronik des Theophanes. Graz, Austria, 1957.
Tiesenhausen, V. Monetii Bostochnovo Chalifata. St. Petersburg, 1873.
al-'Uyun wa'l-Hada'iiq. Ed. M.J. de Goeje. Leiden, 1869-71. Part 3 of this work is published in a volume entitled Fragmenta Historicorum Arabicorum.
al-Waqidi. Kitab al-Maghazi. Ed. M. Jones. 3 vols. London, 1966.
al-Ya'qubi. at-Ta'rikh. 2 vols. Beirut, 1379/1960.
_______. Kitab al-Buldan. Ed. M.J. de Goeje, Leiden, 1892.
al-Yaqut. Mu'jam al-Buldan. Ed. F. Wüstenfeld. 6 vols. Leipzig, 1899-73.
az-Zayla'i. Tabyin al-Haqa'iq: Sharah Kunz ad-Daqa'iq. 6 vols. Bulaq, 1313-15.
az-Zubayr b. Bakkar. Akhbar al-Muwaffaqiyat. Ed. S. M. al-'Ani. Baghdad, 1972.
al-'Ali, S.A. at-Tanzimat al-Ijtima'iya wa'l-Iqtisadiya fi'l-Basra fi'l-Qarn al-Awwal al-Hijri (The Social and Economic Institutions of Basra in the First Century Hijra). Baghdad, 1953.
Amin, A. Fajr al-Islam (The Dawn of Islam). Vol. 1, 3rd ed. Cairo, 1933.
Ashtor, E. A Social and Economic History of the Near East in the Middle Ages. London, 1976.
Ayalon, D. "The eunuchs in the Mamluk Sultanate." Studies in Memory of Gaston Wiet. Ed. M. Rosen-Ayalon. Jerusalem, 1977. Pp. 267-95.
______. "The Military reforms of the Caliph al-Mu'tasim-their background and consequences." Jerusalem, 1963 (mimeographed).
______. "Preliminary remarks on the mamluk military institution in Islam," WTS, pp.44-58.
Azizi, M. La domination arabe et l'épanouissement du sentiment national en Iran. Paris, 1938.
al-Baltistani, M.A. an-Nahj as-Sawi fi Ma'na al-Mawla wa 'l-Wali (The Proper Path to the Meaning of the Terms "Mawla" and "Wali"). Najaf, 1388/1968. Discusses these terms only in their religious context; no consideration of the social meanings.
al-Basha, H. al-Alqab al-Islamiya (Islamicate Nomenclature). Cairo, 1957.
Biddle, D.W. "The development of the bureaucracy of the Islamic empire during the late Umayyad and the early Abbasid period." Ph.D., University of Texas, 1972.
Bulliet. R.W. "Conversion to Islam and the emergence of a Muslim society in Iran." In Conversion to Islam. Ed. N. Levtzion. New York, 1979, pp. 30-51.
Cahen, C. "Histoire économico-sociale et islamologie: Le probléme préjudiciel et l'adaptation entre les autochtones et l'Islam," Colloque sur la sociologie musulmane. Ed. A. Abel. Brussels, 1962, pp.197-215
Cambridge History of Iran. Vol. 4 : From the Arab Conquest to the Saljuqs. Ed. R. Frye. Cambridge, 1975.
Crone, P. "The Malawi in the Umayyad period." Ph.D., University of London, 1973.
_______. "Umayyad clientage." Conference on the Economic History of the Near East. Princeton, 1974 (photostat).
Dennett, D.C., Jr. Conversion and the Poll Tax in Early Islam. Cambridge, Mass., 1950.
Donner, F.M. "The Arab tribes in the Muslim conquest of Iraq." Ph.D., Princeton University, 1975.
Encyclopaedia of Islam. 1st ed. 4 vols. Leiden, 1913-38.
Encyclopaedia of Islam. 2nd ed. 4 vols. (incomplete). Leiden, 1954-
Forand, P.G. "The development of military slavery under the Abbasid caliphs of the ninth century A.D. (third century A.H.), with special reference to the reigns of Mu'tasim and Mu'tadid." Ph.D., Princeton Univeristy, 1962.
_______"The relation of the slave and the client to the master or patron in medieval Islam," International Journal of Middle East Studies, 2 (1971), pp.59-66."
_______The status of the land and inhabitants of the Sawad during the first two centuries of Islam," Journal of the Economic and Social History of the Orient, 14 (1971), pp.25-37.
Goldziher, I. Muhammedanische Studies. 2 vols. Halle, 1889-90. Translation: Ed. S.M. Stern, Muslim Studies. 2 vols. London, 1967-71.
Goto, A. "An aspect of Arab society of the early seventh century: non-related people in the male descendants' groups," Orient (Tokyo), 12 (1976), pp.75-88. An analysis of mawla and halif.
Grunebaum. G.E. von. Classical Islam: a History 600-1258. Chicago, 1970.
Hass, S.S. "The contributions of slaves to and their influence upon the culture of early Islam." Ph.D., Princeton University, 1942.
Hill, D.R. "The role of the camel adn the horse in the early Arab conquests." WTS, pp.32-43.
Hodgson, M.G.S. The Venture of Islam. 3 vols. Chicago, 1974.
al-Kharbutli, A.H. al-Mukhtar ath-Thaqafi, Mir'at al-'Asr al-Umawi (al-Mukhtar ath-Thaqafi: Mirror of the Umayyad Age). Cairo, .
_______. Ta'rikh al-'Iraq fi Zill al-Hukm al-Umawi (The History of Iraq in the Shadow of Umayyad Rule). Cairo, 1959.
Kremer, A. von. Culturgeschichte des Orients unter den Chalifen. 2 vols. Vienna. 1875-77.
Noth, a. "Zum Verhältnis von Kalifaler Zentralgewalt und Provinzen in Umayyadischer Zeit: die 'Sulh'-'Anwa' Traditionen für Ägypten und den Iraq," Welt des Islams, 14 (1973), pp.150-62.
Omar, F. "The composition of 'Abbasid support in the early 'Abbasid period, 132/749-169/785," Bulletin of the College of Arts and Sciences, Baghdad. 11(1968), pp.158-82.
Pipes, D. Slave Soldiers and Islam: the Genesis of a Military System. New Haven, 1981.
ar-Rawi, T.I. Al-'Iraq fi'l-'Asr al-Umawi (Iraq in the Umayyad Period). Baghdad, 1965.
Sachau, E. Muhammedanisches Recht nach Schafiitischer Lehre. Stuttgart & Berlin, 1897.
Schmucker, W. Untersuchungen zu einigen wichtigen bodenrechtilichen Konsequenzen der islamischen Eroberungsbewegung. Bonn, 1972.
Sezgin, F. Geschichte des arabischen Schrifttums. 5 vols. (incomplete). Leiden, 1967-.
Shaban, M.A. Islamic History: a New Interpretation, 2 vols. Cambridge, 1971-76.
Sharif, M.B. Beiträge zur Geschichte der Malawi-Bewegung im Osten des Chalifenreichs. Basel, 1942. [Also enlarged Arabic edition: as-Sira' bayn al-Malawi wa'l-'Arab. Cairo, 1954.]
ash-Shayib, B. Ta'rikh ash-Shi‛r as-Siyasi ila Muntasif al-Qarn ath-Thani. 5th ed. Cairo, 1976.
Sourdel, d. Le vizirat 'abbaside de 749 à 936. 2 vols. Damascus, 1959-60.
Spuler, B. Iran in früh-islamischer Zeit. Wiesbaden, 1952.
Tyan, E. Institutions du droit publique musulman. 2 vols. Paris, 1954-56.
az-Zabidi, M.H. al-Hayah al-Ijtima'ya wa'l-Iqtisadiya fi'l-Kufa fi'l-Qarn al-Awwal al-Hijri (Social and Economic Life of Kufa in the First Century Hijra). Cairo, 1970.
Zaydan, J. Ta'rikh at-Tamaddun al-Islami (The History of Islamic Civilization). 5 vols. Cairo, 1902-06.
Jacob Lassner, The Shaping of 'Abbasid Rule (Princeton, 1980) appeared after this article was submitted for publication. Each of us has analysed the status of mawlas without knowledge of the other's conclusions.
Comment on this item
Support Daniel Pipes' work with a tax-deductible donation to the Middle East Forum. Daniel J. Pipes