I
The millet system, as it developed in the later centuries of classical Islam, owed its specifically Islamic legal bases to the very beginnings of Islam, to the events of Muhammad's Medinan years (622-632), when the Prophet and his followers had achieved majority power in one town at least of Arabia, and were therefore forced to consider the question of the Muslim community's relationship to minorities, in the case of Medina specifically, a Jewish minority.
However, systems for the regulation of minorities such as the Islamic millet system became were by no means alien to the older Near Eastern world, in which hardly any state or empire had ever achieved -- or indeed sought -- an ethnic or religious exclusiveness. Thus the two great empires of Muhammad's time, the Byzantine and the Sassanid ones, had to come to terms with the question of dissenting minorities. Where this dissent took the form of what seemed to be a schismatic or heretical deviation from the prevailing orthodoxy, or where some sort of challenge to the existing socio-religious order was involved, the claims to enforce uniformity of a powerful established church, under imperial patronage, in the case of Byzantium, or of a socially superior priestly caste, in the case of Zoroastrian Persia, could not be gainsaid. Hence Byzantium had to combat strenuously the Paulicians in eastern Anatolia and, later, the Bogomils in the Balkans, just as the Sassanid state and church successfully overcame the challenge in the early sixth century of Mazdakism and also managed to push Manichaeism out of the empire northwards and eastwards into Transoxania and Central Asia. But the existence of other great monotheistic faiths within the borders of the empires was -- somewhat paradoxically in vew of the external support which the devotees of such faiths could often call upon -- felt as a lesser affront to the dominant religious susceptibilities than were the sectarian forms of that dominant faith. Something of respect for the founders of the other great faiths, for Moses and Christ as prophets and holy men, survived in Near Eastern minds, an attitude that was probably to influence Muhammad in his policy toward the dhimmis. In the case of the Jews, there was an ethnic barrier which enabled the cominant castes to despise them for racial exclusiveness as well as for religious purblindness and to adopt something like an attitude of exasperated contempt, as if nothing better could be expected from such a people; and this in fact allowed the Jewish communities to survive. Moreover, the dissenting groups often fulfilled useful administrative, social or economic functions within the empires, thus permitting policies of discrimination against them but not policies of total extermination.
In the pre-Christian Roman Empire, Judaism had been a religio licita, [i.e., the opposite of an illicit religion] but a hardening of attitudes took place after the conversion of Constantine. The Code of Theodosius II (438) restated a body of anti-Jewish legislation which had grown up in the previous decades. Thus chapters eight and nine of book XVI of the Code, De Judaeis, Caelicolis et Samaritanis and Ne Christianum mancipium Judaeus habeat forbade celebration of the feast of Purim, when it was believed that Jews burnt images of the cross, and Jews' holding Christian slaves (as the title of chapter nine implies), and also prohibited the construction of new synagogues and required the pulling down of old, structurally unsafe ones. The Corpus iuris civilis of Justinian I likewise subsumed a host of further discriminatory measures: novella 37 of 535 forbade the practice of Judaism altogether in the North African territories newly reconquered from the Vandals, and novella 146 of 553 actually interfered in Jewish doctrinal matters -- something which was hardly ever to occur under Islam -- by stipulating which Targum was to be used and forbidding use of the Mishnah for understanding the Torah. Already in 383 the council of Byzantine bishops had forbidden apostasy from Christianity, and Theodosius prescribed death for any Jews or Samaritans who persuaded Christians to abandon their faith. Thus the general aim of Byzantine policy towards religious minorities, above all the Jews, was to keep them in a permanently inferior position; in several of the above measures, e.g., in regard to synagogue construction and proselytism, it foreshadowed later Islamic policy aimed at freezing the position of the dhimmi communities, presumably in the hope that over the course of time they would wither away naturally. The question of financial disabilities of the Jews in the Byzantine Empire is somewhat obscure, and specific documentation is lacking. One scholar thought it probable that Jewish communities were liable to a special capitation tax. In the pre-Christian period of the Roman Empire, the Jews had paid a fixed due to the imperial treasury, the aurum coronarium or fiscus iudaicus, and certain Arab geographical writers of the mid-ninth and early tenth centuries aver that the Byzantine state used to take one dinar per annum from every Jew and Zoroastrian.
In the Sassanid Empire, there were important Jewish communities in Mesopotamia and scattered across Persia proper as far as Khurasan and Soghdia. Since they were unsupported by any outside power, their presence posed no threat to the security of the state; hence they seem to have survived, as internally self-regulating communities, on their own enduring religious and intellectual resources. It was the Christians of the empire, much more numerous than the Jews (above all in Mesopotamia, but strongly represented in the towns of Persia, too) and at times looking outward to their coreligionists in countries like Byzantium and Armenia, who felt at times the persecuting hand of the Zoroastrian state church, though this seems most often to have fallen on the clergy rather than on the Christian laity. The church of the empire (after the schisms of the fifth century divided into Nestorian and Syrian Monophysite branches, with the former especially strong in the eastern parts) subsisted as a largely autonomous organization under its patriarchs or Catholicoi, as it was to do in Islam. But the church leaders do not seem to have had any fiscal responsibilities as under Islam, where the religious head, Patriarch or Exilarch, collected taxation from his community on behalf of the secular state; the mention that Shapur II (309-379) raised a double capitation tax from the Christians (i.e., double that normally levied on all his subjects, of whatever faith) and that he asked Simeon bar Sabba'e to collect it, implies that this was not the norm. The building of new churches was prohibited as it was to be under Islam, but this regulation was often waived for diplomatic reasons; thus in the peace agreement between Justinian and Khusraw Anushirvan (531-579) of 561, conversely, Zoroastrians in Byzantine territory were to be permitted to build fire temples). Juridically, the Christian communities were left to follow their own laws and customs, such as the Acts of the Synods of Ctesiphon of 410 and 420 for ecclesiastical matters and the Syro-Roman law code, the Leges Constantini Theodosii Leonis, for pesonal status, thus placing them on broadly the same footing as the Christian population within the Byzantine Empire. The question of apostasy from the dominant Zoroastrian faith is not entirely clear in the Zoroastrian legal literature. Certainly, high-born converts to Christianity, from the royal family, the nobility and the upper priesthood, courted death, which was often inflicted, as the Syriac Acts of the martyrs tells us, in a singularly barbarous fashion; and we hear of Christian martyrs as late as 615. Even so, such draconian punishment may not have been generally applied to those below the top social and religious ranks, since we know of several prominent ecclesiastics who were formerly Zoroastrians, such as Sabba (d. 487), from a high Zoroastrian family, and the famous Nestorian Patriarch Mar Abha (d. 552) was formerly a Zoroastrian official; and there are mentioned numerous Christians with Zoroastrian names, often containing theophoric elements, showing that they were not born Christians. Persian imperial policy was more liberal than the later Islamic attitudes to conversion, since Al-Tabari expressly records that Khusrau Parviz (591-628) allowed conversions to Christianity from any faith except Zoroastrianism, whereas under Islam, conversions between the protected religions, e.g., of a Christian to Judaism and vice versa, were forbidden.
In sum, the position of Christians and Jews in pre-Islamic Persia was precarious at certain times, such as those of warfare with Byzantium with respect to the Christians, but the Christians' numerousness, their high level of education and the ability of their faith to appeal even to the highest social classes of the Persians, all allowed them a generally not unfavorable position. Much in their social and legal status corresponds to the position of the Christian community under Islam, but in certain respects it was superior. For one thing, the Persian emperors did not disdain to employ Christian soldiers in their army; the Islamic jizya came to be regarded as in lieu of military service, but dhymmis did not, after the first decades of Islam, have any option of either joining the Muslim forces and sharing in the privileges (primarily financial) of the Arab muqatila or of paying the jizya, till the Tanzimat period of the Ottoman Empire. Thus, in one scholar's words, "the struggle for existence by Christianity in Islam was much harder than in the time of the Zoroastrian Sassinid rule."
II
The Quranic basis of the later doctrine of dhimma has as its background the numerous references to jihad against the unbelievers, arising out of Muhammad's gradual disillusionment with the Jews and Christians who fialed, as he had originally hoped, to recognize the new faith of Islam as either the valid form of eternal revelation for the Arabs specifically, or as the latest, hence truest in its universal validity, form of God's message for all mankind. Also significant for the prehistory of the idea of dhimma is the document preserved in the Sira of ibn Ishaq and known for convenience as the "Constitution of Medina." It has recently been recognized as in fact a highly composite document, parts of which antedate Badr (2/624) whilst others may postdate the massacre of the third great Jewish tribe of Quraiza (6/627). The main point for our present purpose is that it attempts to regulate inter alia the status of the Jews in Medina (though whether these Jews include the three major tribes or consist of smaller, splinter groups attached to the Arab tribes of the Ansar as confederates is unclear), either as a separate umma (if religion be the main test of an umma) or as part of the general ummah of Yathrib-Medina (if territoriality be an important element in the definition of an umma); in any case, both the Muslims and the Jews are to retain their own din (here almost certainly meaning law as well as religion).
It is here in the Constitution of Medina that we find the term dhimma, for section fifteen states that "the dhimma [here, it seems, "compact guaranteeing security"] of God is one," i.e., all within the umma are equally protected, and all are able to give protection to other members, so that complete solidarity, with everyone in the dual role of protector and protected, is assured for all. The corollary would appear to be that no one outside the umma is to be protected. An increase in Muhammad's claims as religious leader and prophet is seen in his requirement of obedience "to God and His Messenger," paralleled by the fact that in his treaties with the outlying tribes of Arabia who sent delegations to Medina we find mention both of "the dhimma of God" and "the dhimma of Muhammad," here clearly meaning "grant of protection." In some cases, there is mentioned the "dhimma of Muhammad" alone, possibly where tribes were still not yet Muslim in faith but nevertheless in political alliance with Muhammad, hence the Prophet's personal hesitancy about the propriety of extending God's dhimma to pagans. However, Muhammad seems later to have extended the dhimma of God to all tribes and individuals coming within his system of security and alliance; he never showed any compunction about employing non-Muslims in his fighting forces, as with the still pagan Quraish at Hunain, and the concept that only Muslims could share in the advantages of the Muslim umma develops only after the Ridda with the advent to power of the legalistic disciplinarian 'Umar b. al-Khattab. The importance of the possibility of non-Muslim but political and military allies of the Prophet enjoying dhimma in these last years of Muhammad's life is that, in subsequent times, it was to be precisely the non-Muslim scriptuaries dwelling within Islamic society who were accorded this dhimma and became known as the ahl al-dhimma par excellence.
In the actual text of the Qur'an, we find the noun dhimma only twice, in Sura 9:8, 10, coupled with the parallel word ill in the phrase "compact or agreement" (the lexicographers defining ill as "an agreement for mutual protection," "promise of protection," etc). But it is in this same Sura, so filled with the atmosphere of jihad, that we find the Qur'anic crux for the subsequent legal basis of dhimma equals "proection of non-Muslims living within the dar al-Islam." In Sura 9:29, we have "Fight against those who disbelieve in God and the Last Day, who do not account forbidden what God and His Messenger have forbidden, and who do not follow the religion of truth, from amongst those who have been given the Book, until they pay the jizya in exchange for a benefaction granted to them, being in a humiliated position." The general purport of the verse is clear: the People of the Book are exempted from the general sentence of being combatted till death, the inexorable fate of obdurate pagans, but the price of their preservation is to be reduction to a humiliating status in society as second-class citizens, liable to a poll tax -- this last provision being presumably an echo of similar provisions which may have been imposed on minority faiths in the older Near Eastern empires. What was thus envisaged in Islam was a sort of contract promising protection, dhimma, perpetually renewed with the Muslim state in return for acceptance of subordinate status and automatically revoked in the case of nonfullment by the dhimmis (what happened when the Muslim authorities failed to implement the promised protection was left less clear).
Such a solution was based on sound pragmatic grounds. Even within the Arabian peninsula, where the vestiges of ancient indigenous paganism soon crumbled and Islam became the majority faith, there were in the last years of Muhammad's life and during the caliphate of Abu Bakr, communities of Zoroastrians in Bahrain and Oman and of Christians in Najran and other parts of southern Arabia who stood firm over their own faiths but were friendly toward the Prophet and his new creed, as is shown for example by the treaty which Muhammad made with the Christians of Najran, by whose terms the people of that town were to aid the Prophet in time of war with weapons and beasts. Outside the peninsula, some sort of modus vivendi with the vast populations soon to be subjugated by the generals of 'Umar and 'Uthman had to be hurriedly worked out. The Arab warriors were for long a dominant minority which had to cncentrate itself on the amsar and in certain other urban centers; the broad consent, if only tacit, of the governed owuld have to be eventually secured. Moreover, one or two of the minorities formerly oppressed by the old regimes, like the Samaritans in central Palestine and perhaps some of the Jewish communities, had given positive assistance to the incoming Arabs. We do not have to picture the masses of Christian population across the Fertile Crescent as actually welcoming the conquerors -- the age-old antipathy of sedentaries for Bedouins from the deep desert must have caused profound fear and disquiet at their approach -- but the alienation of these Semitic pouplations, on social and religious grounds from the Greek Melkite ecclesiastical hierarchy and the Byzantine Greek ruline elite of soldiers and officials was not conducive to wholehearted resistance against the invaders in most cases. Even where towns and regions did resist, and submitted eventually through force of arms rather than a peace treaty, mass enslavement or extermination (as had been possible in the limited confines of the Medinan oasis with the Banu Quraiza) was clearly not feasible.
Hence the dhimma system came into existence almost inevitably, but in a somewhat informal way; the elaboration of a tight legal system here was to be the work of later, systematizing jurists, above all in the Abbasid period. The first requirement was definition of the terms ahl al-dhimma and ahl al-kitab. The Qur'an was somewhat vague. In three places (2:59/62; 5:70/69; 22:17) we have the Jews, Christians and Sabi'un linked together, but with the addition in the last verse, from surat al-hajj, of the Zoroastrians (al-majus), and traditional exegesis came to distinguish these from the polytheists. The position of the Jews and Christians as scriptuaries is clear enough. Whoever the original Sabi'un may have been -- and they must have been some vestigial sectarian group on the fringes of Arabia -- the designation later became an umbrella one under which various groups, like the so-called "sabians" of Harran, devotees of later classical paganism, were to claim kitabi status. The position of the Zoroastrians, somewhat uncertain because of their single mention only in the Qur'an, required very urgent clarification, for by 652 the last fugitive Sassanid emperor, Yazdagird III, was dead, and apart from remote regions like Dailam, the Caspian provinces and Sistan, all Persia lay under Muslim control. In any case, there had been Zoroastrian groups established in eastern and southern Arabia in pre-Islamic times, conceivably including some native Arab converts. A hadith attributed to the Prophet subsequently asserted that Muhammad expressly extended the terms of 9:29 to the Magians; some Persophile jurists were to maintain that the Zoroastrians had formerly been genuinely in possession of written scriptures, but these had been destroyed at a date before the advent of Islam. In this fashion, a neat solution could be suggested for the difficulty that both the written Avestan originals of the Zoroastrian sacred texts and the Middle Persian translations seem to have been lost in the disruption of the Zoroastrian religious institution by the Arab invasions, and were only committed to writing again during the Zoroastrian literary renaissance of the ninth and tenth centuries.
When the Muslims first acquired a foothold in the Indian subcontinent through the conquest of Sind in the early eight century, a situation arose similar to that of Persia; the teeming populations there oculd not be slaughtered en bloc, but how, in the absence of any Qur'anic nass could the Pagan Hindus be assimilated to dhimmi status? On the evidence of one Arab account of the conquest of Sind, there were certainly massacres in the towns of Sind when the Arabs first arrived; but in some places, there was no resistance, expecially, it seems, amongst the Buddhist minority, with their traditions of nonviolence. Peace treaties were made with the local communities on what had been standard conditions during the overruning of the Fertile Crescent and Persia, e.g., the requirement of giving hospitality to Muslims, and the commander Muhammad b. al-Qasim al-Thaqafi is recorded as accepting the submission of the fortress and town of Ror after several months' siege in return for tribute and as conceding that their idol temples should not be disturbed because, he proclaimed, "Idol temples are exactly like the churches and synagogues of the Christians and the fire temples of the Zoroastrians."
Over the succeeding centuries, it became accordingly normal practice to concede to the Hindus and other non-Muslims the status of ahl al-dhimma. Here on the remote fringes of the Islamic world, it was obviously difficult to enforce the strict regulations of the Shari'a regarding dhimmis. Territory not infrequently changed hands along the Hindu-Muslim frontier, and Muslims found themselves living under Hindu rulers (where in the case of the Rashtrakutas of the Deccan, their Muslim subjects in the Bombay coastal region enjoyed much favor and religious and legal freedom); and in this syncretistic environment, Islam came to adapt itself to the immemorial caste system of India, rather than attempt to change it. An interesting instance of what seems to be the adoption of an ancient Hindu discriminatory custom towards an inferior minority by the Muslim conquerors has been suggested by one scholar as the explanation of a puzzling passage in an Arab history about the activities of an Abbasid governor in Sind during the caliphate of al-Mu'tasim (833-42). IN this, the governor is dealing with the community of Jhats, considered an unclean group in the Hindu social system; he took the jizya from them, "sealed" their hands (i.e., the practice of giving a receipt for payment of poll tax in a particularly humiliating way, attested for Iraq and others of the central Islamic lands from the Umayyad period onwards), and ordered that every Jhat when he went out should be accompanied by a dog. The apparent explanation is that the dog is an unclean animal to both Hindus and Muslims, and the requirement that every member of the despised Jhat group should be distinguished by having him with a dog would thus be a peculiiarly local form of ghiyar, distinguishing feature, analogous to the zunnar and the colored patches of the dhimmis in the lands of the caliphate further west.
III
The treaties made by Muhammad and the commanders of the first Muslim armies as they overran the arabian peninsula and then expanded over the whole Near east seem to have been ad hoc conventions, made by Muslim leaders on the spot with the chief or headman of the town or region in question. The historical sources providing details, such as Ibn 'Abd al-Hakam for Egypt and baladhuri for the conquered lands in general, date from up to two and a half centuries after the agreements were made and the more detailed and circumstantial are the conditions of surrender, the more suspect are the documents. The original agreements must in fact have been quite vague ones, often in the first place oral ones in an age when literacy, at least on the Arab side, was very rare, and when hardly any Coptic, Aramaic, or Persian native speakers could be expected to know Arabic anyway. It is certainly impossible to deduce a neat dichotomy, as later jurists were to do, between voluntary surrender according to a peace treaty (sulhan) and conquest by force of arms ('anwatan). However, there are some broadly common features in these agreements. The subject peoples are above all required to pay the poll tax or jizya as the price of the dhimma extended to them. In addition, various services, express of humiliation, are demanded. They have to act as guides through unknown terrains for the Muslims, and give Muslim travellers shelter for between one and three nights and days: they have to keep up roads and bridges; they have to supply the Muslims with basic foodstuffs like corn, oil and honey and raw materials like pitch and timber for building purposes; and they must undertake not to provide aid or comfort for the Muslims' enemies. The dhimma received in exchange comprised freedom of person and freedom of legal status, i.e., continuance under personal law codes with access to personal courts; freedom to retain property (only houses abandoned by fugitive owners were to be taken over by the Muslims); and freedom of conscience and the exercise of monotheistic religion (only churches abandoned by their Christian congregations were to be turned into mosques). In the case of certain privileged groups who had rendered special services to the incomers (like the Samaritans who had acted as guides) or who lived in strategic areas on the borders of the Muslim empire and had to be specially conciliated (like the Mardaites or Jarajima of the Amanus region of northwestern Syria, bulwarks against a Byzantine revanche; the Arab tribes of Taghlib and Tanukh in the upper Jazira, adjoining the Byzantine frontier, who might have been tempted to go over to the Greeks; and the kingdom of Nubia, remote and protecting the Islamic province of Egypt from the unknown potential threats from inner Africa), exemptions from these general requirements were made often involving financial concessions or the privilege of fighting at the side of the Muslim warriors and thus charing in the captured plunder.
In all these agreements, the provisions are often summary in form and by no means rigorous. There is, for instance, no mention as yet of the ghiyar or shi'ar, the requirement of distinctive dress for dhimmis; the Muslim warriors, a small minority in the conquered lands in any case, were in these early years perfectly distinctive through their Bedouin dress and mounts and thus in no need further to distinguish themselves from what eventually became in most parts of the empire a non-Muslim minority only. The neat systematization of the social and legal rights and disabilities of the dhimmis dates from a later time, as Islamic religion began to permeate for the first time the general fabric of Near Eastern society, whereas previously older attitudes of the jahiliyya had been still strongly held. Above all, this definition is a reflex of a general urge in all walks of life towards the erection of legal frameworks, in which each individual's rights and duties vis-a-vis the shari'a, in the case of the Muslims, and vis-a-vis the Islamic state in the case of all peoples within the dar al-islam, Muslims and kitabis alike.
The classic formulation of the general status of the dhimmis was to be that of the so-called "Covenant of 'Umar." A forerunner of this is the document known as "the Prophet's edict to all the Christians" and then "...to all mankind," preserved by two oriental Christian sources, the anonymous Nestorian chronicle of Si'irt and the Jacobite Barhebraeus's Ecclesiastical Chronicle. This edict is said to have been originally made by Muhammad with the Christians of Najran. Various monasteries and other institutions of the Christian Orient later claimed to possess genuine copies of this document, confirmations of which were connected with various historical figures like the caliph Mu'awiya and the Nestorian Catholicos Isho'yahb II, and one historian has noted that "une immense fortune etait reservee a ce document"; "authentic" copies of it have continued to turn up till the present century, e.g., in 1909 at the Armenian Catholic Patriarchate in Istanbul. Nevertheless, it is a patent fabrication, probably the work of some Nestorian priest or monk.
The "Covenant of 'Umar" itself exists in extenso only in authors of as late as the end of the eleventh century and the twelfth and then, with numerous variations, in later authors. It is true that the early Abbasid jurist Abu Yusuf (d. 798) attributes in his Kitab al-kharaj certain of the conditions of the "Covenant" as it was later constituted to 'Umar b. al-Khattab's general Abu 'Ubaida. Turtushi's version claims to be an agreement made with the Christians of Syria at the time of their capitulation, without specifically mentioning any city like Damascus, Hims, Jerusalem, etc., although Ibn 'Asakir does state that the "Covenant" was in fact Abu 'Ubaida's capitulation treaty with the people of Damascus and all Syria, elsewhere further citing Khalid b. al-Walid's imposition of a capitulation treaty on the dhimmis of Syria whose conditions resemble closely those of the "Covenant."
The essentials of the "Covenant," said to have been promulgated by 'Umar I in reply to a request for aman by the Christians of Syria and related on the authority of the Companion 'Abd al-Rahman b. Ghunm (d. 687), are as follows: the Christians undertake not to erect any new churches, monasteries or monks' hermitages, and not to repair those falling into ruin; to give hospitality to Muslim travellers for up to three days; not to shelter spies or harm the Muslims in any way; not to teach the Qur'an to their children; not to celebrate their religious services publically; not to prevent any of their kinsfolk from freely embracing Islam; to show respect for the Muslims in various ways, such as rising in their presence; not to imitate the Muslims in matters of dress or hairstyle, to use their manner of language and their patronymics; not to use riding-beasts with saddles, or to bear any arms; not to have seals engraved in Arabic characters; not to sell alcoholic drinks; to shave the front of the hair and to wear the distinctive girdle or zunnar; not to parade the emblem of the cross publicly in Muslim quarters and markets, or to beat the naqus (wooden clappers used instead of bells to summon the faithful to worship) or to chant loudly; not to conduct public processions on Palm Sunday and at Easter; not to bury their dead in the same neighborhoods as Muslims are interred; not to keep slaves who have been the property of Muslims; and not to build houses which might overlook those of Muslims. The Christians agree to observe these conditions, in exchange for protection, and to regard contraventions as absolving the Muslim state from its obligation of dhimma.
Clearly, such provisions as the requirement of distinctive clothing, the prohibition of building houses overlooking the private quarters of Muslim houses and the forbidding of overt Christian ceremonial in public places where Muslims might be affronted, point to a society where Muslims and dhimmis are inextricably mingled, and there is no indication of anything like the Jewish ghettos of mediaeval Europe or even of the specifically Jewish quarters of mellahs of later Islamic North Africa. The chronologically piecemeal character of the "Covenant" is apparent, and little of it seems to fit the conditions of the early conquest period in Syria; its attribution to the second caliph does however accord with the picture which grew up of 'Umar b. al-Khattab as the great lawgiver and organizer of the military and administrative bases of the Islamic state. One may also suspect some confusion with the later Caliph 'Umar b. 'Abd al-'Aziz (717-20), the "Hezekiah of the Umayyads" as one scholar has called him, who does seem to have endeavored to apply the letter of Sura 9:29, about humiliating the dhimmis. The Chronicle of Si'irt states that 'Umar II was more hostile than previous rulers to the Christians, that he imposed hardships on them and on the celebration of their liturgy, and that he established the rule that the killing of a Christian by an Arab (i.e., a Muslim) did not necessitate the exaction of a life in retaliation but only a monetary payment. Michael the Syrian records similar policies, including a worsening of the financial means of support for Christian ecclesiastical institutions and increased incentives for Christians to become Muslims. It is indeed quite probable that 'Umar continued the policies of Arabization begun twenty years previously by his uncle 'Abd al-Malik, extending them from the administrative sphere to the personal religious one. The period of quiescence in Muslim-Christian relations during the decades immediately followint the conquests was now coming to an end; to it there succeeded in the eighth century a sharpened consciousness of the intellectual and religious divisions of the three great monotheistic faiths and their communities, seen in the increasing appearance of Muslim-Christian polemical literature and the apologetic role of scholars like St. John of Damascus (d. 749?). The increasingly influential religious institution of 'ulama and fuqaha' wished now to formalize what had earlier ben a matter of varying local, somewhat informal agreements, and inevitably, the dominating secular power worked in the direction of rigidity and of worsening the dhimmis' legal status. Hence one scholar may well be right to some extent when he states that "it owuld seem that it (the Covenant) was an exercise in the schools of law to draw up pattern treaties."
IV
To recount the course of subsequent relations of the Muslims with the Jews and Christians would be a lengthy task, and is one which needs to be approached thorugh monographs on specific communities at particular periods, as has in fact ben done by Mann for the Jews in Fatimid Egypt, by Ashtor for the Jews in Mamluk Egypt and in Muslimj Spain, by Galante for the Jews of Istanbul, and by Simonet and de las Cagigas for the Mozarabs in Spain. [you can look for any of these works in our library under the author mentioned above] Here, only a few of the general trends in these historical processes can be noted.
Further spasms of anti-dhimmi restrictions followed on 'Umar II's increasingly severe measures during the early Abbasid period, especially under Harun al-Rashid and under the strongly orthodox Sunni and traditionalist al-Mutawakkil. Especially interesting is the requirement of distinctive dress, which Abu Yusuf traces back to 'Umar II. The zunnar or girdle is always mentioned, but 'Umar II is said to have prohibited the dhimmis from wearing the typically Muslim qaba or cloak, taylasan or gown of the learned classes, and the sirwal or trousers held round by bands of cloth. In the specially discriminatory senses of the two words, the ghiyar "mark of distinguishing" and shi'ar "mark of recognition" make their appearance under the two Abbasid caliphs mentioned above. Al-Tabari records under the year 807 that Harun al-Rashid ordered one of his officials in Baghdad to see that the dhimmis there were differentiated from the Muslims in matters of dress and mounts; this outburst of zeal was connected with an intensification of warfare with Byzantium, and the destruction of Christian churches along the frontier zone of the Taurus Mountains was also enjoined. Then in 853 al-Mutawakkil enjoined that the dhimmis should wear two honey-colored (i.e., yellow) durra'as and that they should ride only mules and donkeys, not horses. We now have the concept of a band or patch of a specific color being compulsorily placed over the shoulders of the dhimmis garments. One Christian source says that al-Mutawakkil required the Christians to wear honey-colored taylasans together with the zunnar, and in subsequent times, patches of red and blue are also mentioned. The favored Muslim colors at this time were probably white (the official color of the Umayyads), black (that of the Abbasids) and green (that of the 'alids), but there does not really seem to have been any deep rationale behind these stipulations on color beyond that of pure differentiation. Only in Mamluk times, when hostility towards the dhimmis had intensified considerably, do we find for instance that the Jews in Egypt had to wear a yellow turban, Christians a blue one and Samaritans a red one, and in 1354 the requirement of similarly-colored garments was made (in the Fatimid period, however, the eccentric caliph al-Hakim had decreed black turbans for the dhimmis, because black was the color of the detested Abbasids); another source states that the Jews and Christians of Egypt had to wear yellow leather boots with one red and one black garter. In Muslim Spain, where the numbers of the Mozarab Christians must have been not much less, and conceivably more, than those of the Muslims, regulations for the distinctive dress of Christians and Jews seem to have been imposed only with the advent of the Almoravids in the later eleventh century and then under the Almohads; the secretary of the Almoravides, Ibn 'Abdun (d. 1134), demanded that such measures be imposed in Seville, but they do not appear actually to have been put into practice until Almohad times.
The regulations concerning distinctive dress were only applied sporadically, as discussions in the Muslim sources of their frequent renewal show, and fell into desuetude toward modern times. Often it was on the fringes of the Islamic world, or in conservative and isolated regions of it, that discriminatory measures lasted longest. Thus until the establishment of the Protectorate in Morocco (1912), Jews there had to go on foot in their own quarters and had to take off their sandals or slippers when in the presence of Muslim dignitaries and when passing Muslim holy places. Shi'i areas of Islam were often more rigorous toward dhimmis than the mainstream Sunnis. Thus in the Zaydi highland zone of Yemen, the Jews' houses were limited in height and were further restricted in the use of external decoration. In the district of Yazd in central Persia, the Zoroastrian community, much oppressed within the Muslim environment, built low, fortress-like single-story houses, with covered-over courtyards for easier defense, until affairs began to improve for them in the later decades of the nineteenth century.
Although protected by the contract of dhimma, the dhimmis were never anything but second-class citizens in the Islamic social system, tolerated in large measure because they had special skills such as those of physicians, secretaries, financial experts, etc., or because they fulfilled functions which were necessary but obnoxious to Muslims, such as money-changing, tanning, wine-making, castrating slaves, etc. A Muslim might marry a dhimmi wife but not vice versa, for this owuld put a believing woman into the power of an unbeliever; for the same reason, a Muslim could own a dhimmi slave but not a dhimmi a Muslim one. The legal testimony of a dhimmi was not admissible in a judicial suit where a Muslim was one of the parties, because it was felt that infidelity, the obstinate failure to recognize the true light of Islam, was proof of defective morality and a consequent incapability of bearing legal witness. In the words of one jurist, "the word of a dishonest Muslim is more valuable than that of an honest dhimmi." On the other hand, the deposition of a Muslim against a dhimmi was perfectly valid in law. It was further held by almost all schools of Islamic law (with the exception of the Hanafi one) that the diya or blood money payable on the killing of a dhimmi was only two-thirds or a half of that of a free Muslim.
It is surprising that, in the fact of legal and financial disabilities such as these outlined above, and of a relentless social and cultural Muslim pressure, if not of sustained persecution, that the dhimmi communities survived as well as they did in mediaeval Islam. Under the stigma of worldly subjugation to another faith, inevitably viewed by many as a manifestation of divine displeasure, and cut off from ready access to such centers of spiritual and cultural life as Rome and Constantinople, the standards of ecclesiastical discipline and clerical literacy amongst the Eastern Christian churches inexorably declined. Whenever a dhimmi was in difficulty under the laws of his own community, he could generally escape by a timely conversion to Islam, thereby putting himself outside the reach of his own legal system. In such a fashion, members of the highest clergy, such as bishops or Catholicoi of the Nestorian and Jacobite churches, not infrequently escaped the consequences of serious crimes like simony, unnatural vice and even murder. Hence in several regions of the Islamic world, such as North Africa and Nubia in the west; Circassia, Daghistan and other parts of the northern Caucasus; and Persia and Central Asia in the east, Christianity disappeared completely by the fourteenth or fifteenth centuries. The mainly urban, though not exclusively urban, Jewish communities were more tenacious and Judaism survived even in a hostile enrivonment like Yemen, where until the exodus of 1949 a high male literacy rate and a fervent attachment to the practices of rabbinical Judaism preserved the historic community there largely intact.