Peter F. Sugar, "Ottoman Social and State Structure" Chapter 2 of his Southeastern Europe under Ottoman Rule, 1354-1804, University of Washington Press, 1977.

1. BACKGROUND

Ottoman social structure was basically the result of a skillful blending of traditional Muslim sociopolitical traditions with Turkic and Byzantine elements. In Ottoman society there were two basic differentiations between individuals. The first distinguished between Muslim and non-Muslim; the second between those who in the broadest possible terms worked for the state and those who did not.

While the Muslim was obviously superior to the non-Muslim by definition, in practice during the early centuries of the empire people of different creeds were often treated almost as equals, in accordance with their profession. Muslims were not allowed to change their religion under penalty of death. Others joined the preferred religious group either by voluntary conversion or by the typically Ottoman practice of forced conversion known as the devsirme (child levy). Later the religious difference became crucial.

Those who worked for the state were mainly Muslims, but by no means exclusively so. The leaders of non-Muslim communities were considered state functionaries responsible for those under their jurisdiction.

The resulting social structure resembled a pyramid-shaped grid (see diagram). The apex of this pyramid, consisting of those who worked for the state, was strictly separated from the base. The base consisted of horizontal layers intersecting confessional lines according to profession. Those working for the state have been aptly called "professional Ottomans" by modern-day scholars, who refer to the rest of the population as the reaya (the flock). This system was originally based on the following concepts.

Sunni Islam's theological, philosophical, scientific, and administrative structures were fully developed by the time the House of Osman began its meteoric rise. While ijma', the concensus of learned men, was still required for any major decision affecting matters of faith and law, ijtihad, the right of interpretation, was in practice limited to the followers of the four different "legal schools." [These four legal schools were: the hanafite, named after the first scholar to expound its principles, Abu-Hanifa (d. 767); the malakite, named after another jurist Malik ibn-Anas (d. ca. 795); the shafi'i school, which got its name from Muhammad ibn-ldris al-Shafi'i (d. 820); and the hanbalites, the followers of the legal interpretation of Ahmad ibn-Hanbal (d. 855).] Each school recognized the others as holding equally valid principles in spite of some disagreements. The Ottomans accepted the most liberal, the hanafite school, but they followed a very strict and almost immutable model.

Equally well established in the Ottoman mind were certain practices introduced by the Muslim-Arab empires, and the Turkic customs that they brought with them from Central Asia. Traditions as well as strongly established principles set the limits within which the Ottomans could organize their realm. It should be stressed from the outset that the organization that they developed covered in great detail every aspect of the communal life and state activity as well as every individual from the sultan down to the lowliest subject; everybody's place, rights, and obligations in society were fixed. The various privileges and duties were not only strictly enforced, at least so long as the state had the power and authority to do so, but they also created something resembling "checks and balances," which limited even the most powerful individuals. A clear border (hadd) defined each individual's position in society. Not only could no one go beyond the stipulated limits, but everyone had the right -- even the obligation -- to resist those who violated his hadd by every possible means. This principle safeguarded the lowly in the social scale so long as the system worked properly, and made Ottoman rule preferable to that of unprincipled nobles to the Balkan peasants.

The Ottoman Empire was "the domain of the House of Osman." So long as the ruling member of this house did not suffer from physical or mental disabilities that, according to the shari'a, made him unfit to rule, and so long as he was a "good Muslim," his power was theoretically unlimited. He owned every inch of his domain and was the absolute master of everybody living on these lands. In this sense the sultan had the rights of the old Turkic tribal chiefs or the gazi leaders. He could appoint and dismiss anyone at will and could even order execution of the highest officials and confiscation of their property. No action could be taken without his approval. It was fortunate for the Ottomans that their first ten rulers were able men.

With absolute power went obligations, which were part of the ruler's hadd. He was to treat his subjects and followers kindly, justly, in accordance with established tradition. Further limitations were placed on him by Muslim law and additional customs, developed originally in Persia and the Arab caliphates.

Below the sultan Ottoman society in general was divided in two different ways. The first way, along religious lines, distinguished between Muslim and zimmi and meant roughly the difference between full and second-class citizen. This made a great difference in the opportunities and positions of state open to members of the two groups, and in taxation.

The second way was the distinction between those who were connected with the state and its institutions, and those who were not. The first group included members of the sultan's household, the entire military establishment, the central and provincial bureaucracy, as well as ulema who manned the legal system, the educational institutions, and the learned professions. From the point of view of the central government, it also included vassals and allies who were supposed to follow the sultan's orders. With the exception of the last mentioned group, whose members always tried to act as independently as possible, these people have been named the "professional Ottomans" by modern scholars. This appellation is fully justified not only because the state structure and machinery rested on their shoulders, but also because these men shared a common set of values and a basic education and considered themselves to be a social group apart from all the others. With very few exceptions, the professional Ottomans were either born or converted Muslims and could be both free men and slaves.

The second group was the rest of the population, well over 90 percent. These were reaya (members of the flock) whose shepherd was the sultan. These sheep, both Muslim and zimmi, were shorn because they were the agricultural and industrial-commercial producers whose labor supplied the goods and taxes that supported the state and the professional Ottomans including the sultan and his household. With negligible exceptions the European subjects of the sultan were all reaya.

This strict system, with its seemingly endless rules and regulations, had great advantages, provided it could be enforced and provided the enemies of the Ottoman Empire did not offer better economic and technological alternatives. When changes became necessary, however, it became an insurmountable obstacle. Every aspect of the system was so closely interwoven with innumerable others that even the slightest change affected and endangered the entire structure. In spite of conservatism and even a certain amount of fanaticism, changes might possibly have been made over the opposition of certain people with special interests but for two facts: the Ottomans simply had no knowledge of and could not envisage any other way of running their state, and they believed that even the slightest change would endanger the entire fabric of state and society. The Ottoman system was the empire's greatest asset from the fourteenth to the sixteenth century and became its greatest liability beginning with the seventeenth century.

The following description of the Ottoman system will touch on practically every aspect of life because the omission of a single aspect would distort the picture. However, some aspects will only be mentioned because they did not influence directly the life of the people living in the empire's European provinces, while others will simply be fitted into the over-all picture superficially at this point and elaborated in more detail subsequently due to their relevance for the people of Southeastern Europe.

2. THE PROFESSIONAL OTTOMANS

Although we are mainly interested in the reaya, a few words must be said first about the professional Ottomans who controlled their lives. We can divide this group, according to its main functions, into four major groups: the mulkiye, the kalemiye, the seyfiye, and the ilmiye. The mulkiye group performed a great variety of duties connected with activities that took place within the imperial palace. Its numbers were in the thousands, its influence was great. According to the rank a member had reached he could be transferred to a corresponding position outside the palace where he performed administrative and military duties.

The mulkiye group broke down into two distinct subgroups; the enderun (inner service) and the birun (outer service). This distinction was based on the physical organization of the imperial palace, which was built around two large courts separated by a wall through which one could pass only by the Gate of Felicity (Bab-i sa'adet), the door that led to the personal presence of the sultan. Those who passed through the gate either spent their entire lives within the confines of the inner palace or were transferred out periodically to other services. The highest ranking enderun functionaries usually were appointed to the most important provincial positions when they left the palace. From there they could be brought back to the capital to serve in the central government, which was part of the birun. Those who belonged to the enderun were all kapi kulus (kapi being another word for gate), slaves of the Porte. After the devsirme system was introduced (see section 5) they came almost exclusively from the European provinces. Their education began in the palace where they moved from the school of pages to higher training centers and duties within the inner courtyard. The best reached the highest position available within the enderun service. When transferred, in accordance with their accomplishments in the inner palace, these men became the most important office-holders in the outer service and in the provinces.

The birun included a great variety of craftsmen, services, schools for these and the various governmental offices, the schools for janissaries, the kapi kulu military units, the offices of the central government, and the various agas (officers) in charge of all these establishments. The people serving in the birun could be slave or freeborn, but all were considered the slaves of the sultan. The most important part of the outer service was the central government proper that functioned through the imperial council (divan-i humayun). Its members, the "pillars of the realm" (erkan-i devlet), held the rank of vezir and served under the presidency of the grand vezir. The most important dignitaries of the empire served in the council: the kadiaskers, the highest judicial functionaries (originally there was one, then there were two for Asia and one for Europe); the beylerbeyis, the chief provincial administrators of Europe and Asia; the defterdars, the chiefs of the imperial treasury (originally two with a third added for Africa); the aga of the janissaries; the kapudan-i derya, commander of the navy; and the nisanci, the secretary of the council who verified its decisions, made certain of the sultan's approval, and attached the sultan's official seal (tugra) to all documents. So long as the system functioned properly most of the council members were indeed kapi kulus who had been educated in the enderun schools.

It is important to realize that in the Ottoman Empire everybody including the members of the imperial council had a fixed place in a table of ranks. Furthermore, whenever multiple offices existed Europe (Rumeli) took precedence over Asia (Anadolu), which in turn came before Africa. Thus, it is easy to list the major dignitaries of the central government according to rank, prestige, and precedence.

Grand vezir
Kadiasker of Rumeli
Kadiasker of Anadolu
Beylerbeyi of Rumeli
Beylerbeyi of Anadolu
Defterdar of Rumeli
Defterdar of Anadolu
Defterdar of Africa
Janissary aga
Kapudan-i derya
Nisanci

While all these dignitaries belonged to the birun part of the mulkiye class of professional Ottomans, their professions clearly placed them, simultaneously, in subgroups. The beylerbeyis were both important military and provincial dignitaries; the janissary aga and the kapudan-i derya were important members of the military establishment; and the defterdars and the nisanci belonged to the bureaucracy proper, although the latter never left the presence of the sultan.

The two chief defterdars were the main officers of the second group of professional Ottomans, the kalemiye (the group of scribes). They were responsible for disbursing funds from the state treasury and indirectly for collecting taxes. The provincial defterdars had direct responsibility for this task until the days of Mehmed II; later they were helped by multezims (tax farmers). Thus, the provincial defterdars and the multezims affected the life of the people directly, and we will deal with them when we discuss provincial administrations. While financial duties were the main function of the kalemiye, members of this group also manned a variety of other institutions that dealt with millions of documents ranging from simple petitions to treaties of state. They were trained in special schools and had their own hierarchy. Because of the special skills and knowledge required to perform its tasks, theirs was the first of the professional Ottoman groups to make membership in its ranks practically hereditary.

The third professional Ottoman category was the seyfiye group, the military men. Strictly speaking, only two types of soldiers belonged to this establishment; the free feudal warriors, the timarli, and the salaried units of the sultan's slave army, the maasli. A third type, the irregulars of various kinds, will also be mentioned in subsequent chapters because of their great importance for our area. Allied armies or special troops, like sappers, are less important for this study because they had almost no effect on the lives of the people.

The timarli were the members of the old "military class," the free-born warriors who originally joined a military leader voluntarily, as did their descendants. They were recruited from the beys and other leading families and from those gazis who had distinguished themselves enough to be granted fiefs and thus enter the "regular" military establishment. This group first and foremost served as sipahis (cavalrymen) in the military, but they also performed fiscal and administrative functions. The latter duties as well as the fact that they derived their income from the labor of the peasantry brought them into direct contact with the majority of the population.

Like everybody else in the Ottoman system, the timarlis were hierarchically organized. Their rank, duties, and income were closely tied to past performances and the reward received for them, land. The basic land unit, called cift, varied in size, but had to be large enough to yield an income sufficient to maintain the holder and his family. One cift constituted the smallest timar whose holder was obliged to maintain his military equipment and horse ready for action. Holders of larger timars were obliged to appear in armor when the troops were mustered, and if the holding yielded amounts larger than the minimum bring retainers to the army. In Europe the smallest timar was supposed to yield 3,000 akces, an amount considered sufficient for maintaining a sipahi. The largest timar brought a revenue of 19,999 akces; for every 3,000 akces over the basic minimal amount, its holder was obliged to maintain a soldier who was called a cebelu and whose major ambition was to earn a timar of his own.

Holdings yielding between 20,000 and 99,999 akces were called zeamet, and those who held them were known as zaims. These feudatories had more important military duties and bureaucratic obligations than the simple timarli, although they had to furnish a cebelu only for every 5,000 akces of income. Holdings yielding over 100,000 akces were known as has and were assigned to the highest provincial officeholders, retired dignitaries, the major officers of the mulkiye class, members of the imperial family, and ladies of the sultan's household. Some were used for the upkeep of the imperial household.

In this manner almost every person working the land was under a landowner who belonged to the professional Ottoman group, if we remember that vakifs were also supported by their labor. Those, however, who were in direct contact with them were the sipahis, the multezims, or the agents of the vakifs. The sipahis of each province obeyed the governor (sancak beyi) of the province in which they lived. He usually carried the honorific title of pasa, held a has, and was subordinated only to the beylerbeyi. The regimental commanders, alay beyis served below the sancak beyi. They were holders of zeamets as were their immediate subordinates, the subasis. These last-named served as the chief police officials in the various kazas, the subdivisions of the sancaks. The lowest ranking officer, a sort of military policeman, was the cerisurucu. He held a large timar and acted as the local police officer in time of peace.

The different types of agricultural holdings were supposed to follow a strictly regulated pattern of division. A "typical" sancak's revenue derived from various agricultural activities was supposed to be assigned as follows:

20 percent for has holding including those of the sultan (Havas-i Humayun)
10 percent for zeamets
40 percent for timars
10 percent for the support of troops manning fortifications
10 percent for vakifs

Part of the remaining 10 percent was reserved for the irregulars (the part set aside for private property will be discussed when landholding and agriculture are covered). The oldest of these fighters were the akincis, volunteers raised in the European provinces among Turks. They performed scouting duties and were feared as raiders. They lived from booty and always had the chance to become timarli, if they distinguished themselves sufficiently. The yaya and piyade, irregular infantry troops, were known only as part of the Ottoman army, but the musellems lived among the Europeans. Settled Turkomans, the musellems were members of the military class and performed regular military duties as cavalrymen. They had small holdings of land on which they paid no taxes and dues, and farmed their holdings themselves. Only one of every three or four served, and those staying behind farmed his plot too. They earned a strictly subsistence income.

In addition to these Muslim auxiliary forces there were some zimmi who performed regular military duties and had the same rights as the musellerns. They were known under various names -- Uskok, Valachs, and Martolos among others -- and were counted as members of the military class because of their occupation. Their origin is not quite clear. They might have been the descendants of Christians who fought as allies and vassals with Murad I and Bayezid I. Although they later became freebooters and bandits, they were part of the military establishment during the first 250 years of Ottoman rule in the Balkans. An article in Islam Encyclopedisi states that they were organized as a military force under Murad II at the beginning of his reign and mentions that of the 3,500-man garrison Suleyman I left at Buda in 1541,1,000 were Martolos. This appears quite possible, given the fact that their number was over 80,000 in 1527. Some of the Christian timarli probably rose to this dignity.

The voynuks are sometimes listed as fighting forces, and so are the Vlachs, people living in Serbia and Macedonia, who at first received certain tax exemptions in exchange for limited border duty. The Vlachs lost their status toward the beginning of the seventeenth century, although they retained their Romance dialect in the midst of their Slav neighbors until modern times. The derbendci, some two thousand families of Rumelia, acted as guards of mountain passes, bridges, and other strategic locations in exchange for tax relief and represent a transition from real fighters to auxiliary forces. They were not required to leave their districts in times of war, and they can be considered as something of a specialized local militia with strictly defined duties.

The voynuks and dogancis were auxiliaries in the narrowest sense of the word. The former were all Bulgarians as were the latter, with a few exceptions. The voynuks raised horses for the palace and other high dignitaries in exchange for land and tax exemptions. The dogancis enjoyed similar privileges, although their duty was to raise hunting falcons. Both groups were considered part of the military and derived their income from land that was left after those with higher claims were satisfied. Finally, we find yuruks listed as soldiers. These nomadic Turkoman tribesmen owned no land, but paid taxes for the right to use various pastures.

The maasli troops were originally members of the sultan's slave army, the janissary infantry and the slave cavalry also known as sipahis. In later periods sappers, artillery men, etc., were also trained from the young men conscripted for janissary duty. All the maasli were kapi kulus.

The seyfiye was the real military group, but all three classes of professional Ottomans listed so far belonged to what had been considered, since Turkic tribal days, as the "military class." The fourth group, the ilmiye and diniye class, was not considered military. Although families from this group might have been slave in origin, its members were most often freeborn Muslims. The double name means cultural and religious class, and the duties of this fourth group, whose members were the ulema, were limited to these activities. The importance of this class becomes obvious when we recall that the legal profession was part of the religious establishment.

Members of the ilmiye and diniye were trained in schools called medreses supported by vakifs. These schools were ranked, and those established by Mehmed II and Suleyman I carried the most prestige. The rank of a given school determined the positions its teachers and graduates received when they left. Each school had only one official teacher, the muderris, although helpers could be appointed. What that teacher taught and believed permanently influenced his students who in turn acted in accordance with the principles they had professed or learned in the various medreses when they were transferred to other positions.

While only one function of the ulema, serving as judges, was known to the people of Southeastern Europe, it must be noted that members of this class had great influence on the central administration and were often appointed to leading positions, and that, after the decline of the state machinery, they performed duties originally assigned to members of the first-mentioned three official groups. The highest ranking of the ulema was originally the kadiasker, but he had to share his power with the seyhulislam after the mufti of Istanbul acquired this title and gradually achieved this supreme position. In a very general way it could be stated that the seyhulislam became something resembling the highest shari'a authority. Although he was appointed and could be dismissed almost at will, his fetva (written ruling) could bring about the deposition of even a sultan. The kadiaskers continued to oversee the court system. Muftis, lower ranking jurisconsults, could also issue fetvas concerning issues that arose in their own territories. They usually did so when a question of proper interpretation of the law was put before them by the sultan or the kadis of the judicial districts (kazas) under their jurisdiction. There were ranks within the judiciary, the highest of which being the kadis of Istanbul, Mecca, Medina, Cairo, Bursa, Edirne, Damascus, Jerusalem, Izmir, and Aleppo. Of greatest importance to the people of all provinces was the fact that wherever they lived there was a kadi and a court that administered both the shari'a and the kanuns. Because a given kadi could not possibly handle all the cases of his kaza, the judicial districts were subdivided into nahiyes, to which kadis appointed naibs (subjudges).

Having given an over-all picture of the four groups of professional Ottomans and their various functions, duties, and obligations, we turn now to the details of the provincial administration. Those professional Ottomans who belonged to the seyfiye and the ilmiye-diniye groups ran the provinces, but it should not be forgotten that many of them came from or returned to the mulkiye class. Nor were those belonging to the kalemiye restricted to the capital. They too had important duties in the provincial administration.

It is almost impossible to establish a clear picture of the number and exact borders of the various provincial administrations because they changed constantly. The administrations themselves were in continual flux, as can be seen from the confusing nomenclature of the top provincial officials. For example, we have seen that in the central government there were two beylerbeyis: one for Rumeli and one for Anadolu. By the end of the fifteenth century nearly forty officials held this title on an honorific basis, although only the two original beylerbeyis performed the tasks of governors general. The administrative districts of the governors general were originally called sancaks. Later they became known as beylerbeyliks, and still later as eyalets with subdivisions referred to as sancaks and later as livas. Furthermore, not all provinces were administered along the same lines. Those directly under the central administration in which timars were established were in the majority, but there were still others that managed their own affairs under traditional leaders (tribal leaders, vassal princes). The latter were known as the hereditary provinces, hukumet sancaks, and their only obligation, besides fealty, was the delivery of a yearly tribute (salyane) to the central government.

According to Inalclk, by the beginning of the fourth quarter of the sixteenth century there were four hukumet sancaks in Europe: Moldavia, Wallachia, Transylvania, and Dubrovnik (Ragusa). There were also six timarli: Rumelia (a province around Edirne not to be confused with Rumeli), Bosnia, Cezar-i bahr-i sefd (Aegean Islands), Buda, Temesvar, and Cyprus. Later Nagyvarad (Varad, Oradea), Eger (Egri), Kanizsa (Kanisa), Silistre, the Morea, Crete, and Belgrade were added, and for a short period Janow and Kamnice (Kamieniec Podolski) in Poland. Several others appeared and disappeared as the provinces were split and reunited. Of the later additions Crete, Nagyvarad, and the Morea had no timars. By the end of the eighteenth century the number of eyalets, already reduced by loss of territory, stood at 25; these were subdivided into 290 sancaks.

Most of the hukumet sancaks in Europe were either relatively small (Crete, Cyprus, Morea), or held for only short times like Nagyvarad and the two Polish provinces. Some (Moldavia, Wallachia, Transylvania, Ragusa) were independent enough to warrant separate treatment. They were in a real sense "peripheral" holdings whose obligations, besides loyalty and the payment of tribute, were limited to admitting kadis to the major cities, and Ottoman garrisons to the larger fortifications. Their independence also explains why the devsirme system affected mainly those of the Orthodox faith and did not touch Romanians and only relatively few Greeks.

The real "core" provinces were those in which sipahis lived on timars. Supreme command of each of these provinces rested with the governor. Every governor maintained smaller versions of the central government in his seats of residence and had to divide his authority and privileges with those officials who served on the provincial councils. The governors were first of all military leaders responsible for controlling the sipahis, fighting smaller engagements such as border skirmishes, and contributing their forces to major campaigns. Occasionally, in accordance with custom, they held court and decided a few disputes. These officials also governed directly the territory that surrounded their capitals and formed a subunit within the larger whole. The other subunits were administered by beys. All of these dignitaries were in charge of military officers whose major functions have already been mentioned.

While few of the provincial councils were "complete," all consisted of at least the most important officials and mirrored the division in the central administration and the already mentioned Ottoman "checks and balances." Foremost of these were the defter eminis (custodians of the records), who kept registrations of fiefs in order. They belonged, as did their superior the mal defterdari (property registrar), to the kalemiye. The mal defterdari was the province's treasurer. He received all the taxes, paid the province's expenses, and remitted the surplus to the central treasury. The defterdars, like the governors, were appointed by the central government. They acted independently of each other and could even check and report on each other's activities. Finally, the kahya of the pastures (the chief of the order executors) was an important functionary of the provincial assembly. It was the duty of the cavuses to execute orders, mainly those concerning punishments, that resulted either from the governor's verdict or from that of the various kadis' courts. Another functionary called the mefkufcu served under the mal defterdari. Though probably not a member of the local council, he levied the taxes on immovable property in the province and administered those fiefs that were momentarily vacant for the benefit of the provincial treasury.

In addition to the representatives of the seyfiye and kalemiye, those of the ilmiye-diniye group were also present in each province. They consisted of the muderrises and their students in the provincial medreses, but most importantly of the muftis, kadis, and naibs who manned the courts. These courts administered both the shari'a and the kanuns, which differed already from province to province.

Kanuns, being based on the sultan's urf, were valid only for his lifetime. Although they had to be "in conformity" with the principles of the shari'a, they consisted mainly of local laws and regulations in force before the Ottoman conquest, which the sultan confirmed. If the sultan failed to do so, the region had no kanun. Before proper authority was established in central Hungary, for example, already appointed kadis had to force zimmi litigants to travel long distances to their traditional judges, their feudal lords who had left the Ottoman occupied lands' to get rulings simply because the kanuns for the region were either not yet established or not yet known to them. Each sultan promptly reconfirmed the kanuns of his predecessors, and the laws were repeatedly issued in complete collections for each province, the kanunnames. However they were too variegated and complicated for any jurist to handle. For this reason most judges spent their entire professional life either in Rumeli, Anadolu, or Africa, and if they took their duties seriously and were honest men they became real experts in the laws of a given region. A good, experienced kadi was an asset for any province and its inhabitants.

3. THE REAYA

The reaya, both Muslim and zimmi, constituted the overwhelming majority of the population and supported the professional Ottoman establishment. So far as the Muslims were concerned, the Ottomans faced the same dilemma that the Arabs had had to face when they extended their sway during the first century after the hijra: their main duty was to spread the true faith, but since all Muslims were supposed to be equal, if the privileges of the first Muslims were extended to all there would be no one left to produce or pay taxes. The Ottomans inherited the Arab solution including the harac, the land-use tax everybody had to pay regardless of religion. The problem, however, arose in a new form, the best example of which being the Turkoman tribesmen. Members of the "military class," potential gazi material, they were too numerous to be absorbed among the professional Ottomans and too unproductive when considered as reaya. As we have seen, the situation was resolved by transforming these Muslims into musellems or yuruks. In this manner they were not "degraded," remained members of the "military class," and became producers. The yuruks even paid certain taxes.

More vexing was the fact that until the conquests of Selim I (1512-20) the majority of Ottoman subjects were non-Muslims, several of whom, especially in Anatolia, showed a willingness to convert. Conversion was supposed to be the ultimate aim of a good Muslim state, yet mass conversion would have produced economic chaos and ruin. The extreme leniency, unusual even for a Muslim state, with which the first Ottoman rulers handled their zimmi subjects, and their great emphasis on kanuns -- both compatible with the treatment of the "people of the book" contained in the shari'a -- were clearly designed to keep the reaya satisfied with the position they occupied as producers. This also contributed to the ease with which European peasants accepted Ottoman rule and influenced the conversion pattern that will be discussed in the next section. After the conquests of Selim I, when the empire acquired a Muslim majority, the Ottomans became much less generous in their handling of the zimmi. By then, however, the patterns were set and the Ottomans' power was sufficient to keep the system going.

The reaya were grouped first informally and after 1453 formally into millets. Both the term and social organization according to religion had precedents. In early medieval Muslim literature the expression millet (milla) refers only to Muslims and differentiates them from the zimmi. There were other precedents, too, for this treatment of minorities. Religious leaders of minority groups held positions similar to the leaders of the Ottoman millets in Sasanid Iran, and some of Justinian's edicts concerning the Jews resemble the later Ottoman millet system. Curiously enough, the Ottomans used the term basically for non-Muslims, although occasionally a reference to a Muslim millet does appear. Officially, from the beginning -- that is, since Mehmed II's relevant legislation in 1453 -- there were only two millets, the Orthodox and the Armenian. Although the Jewish millet was not "officially recognized" until 1839, it had a head (millet basi) and functioned very well from 1453 on.

Very little has to be said in this connection about the Muslims, although numerous inhabitants of the European provinces followed Islam because their relationship with the authorities depended either on their being subject to the shari'a regulations of their morality and private lives or on their occupations. So far as the first of these two considerations is concerned, the legal system has already been presented and a few words have been said concerning the basic religious duties of Muslims. The shari'a also regulated marriages, divorces, inheritances -- in general all relationships within the family and community -- and so long as the Muslims adhered to these and paid the required fees their lives followed a well-defined and regulated pattern.

The largest of the millets was the.Orthodox. There can be no doubt that Mehmed II had several good reasons for creating a Christian millet limited to the adherents of the Eastern Rite. After he had conquered Constantinople and renamed it Istanbul, he considered himself the legitimate successor of the emperor and wanted to retain his position and prerogatives in dealing with the Orthodox church. He fully realized that the great majority of his Christian subjects were Orthodox, and that most Orthodox were opposed to the reunification of the churches to which the Emperor John VII agreed at the Council of Florence in 1439, making them potentially loyal to a ruler whose major enemies were Roman Catholics. Furthermore, he was determined to rebuild his new capital and make it once again a great trading and manufacturing center. To accomplish this goal he needed loyal Christians and he trusted only the Orthodox. He therefore wanted to give them a special organization.

Not only did he create an Orthodox millet, but he also picked as its first head (millet basi) the monk Gennadius, an outspoken enemy of the Council of Florence, and elevated him to the patriarchate. Gennadius, born George Scholarios, was a well-known theologian and qualified by his learning as much as by opposition to Roman Catholicism to become the highest ranking dignitary of a reunited Orthodox church. Because of the serious problems this reunification created in the seventeenth and especially in the eighteenth centuries, a few words must be said about it before turning to the rights of the Orthodox millet.

When Mehmed created the Orthodox millet two other sees were acting as quasi-partriarchates: Ohrid (Ochrid, Lichnida) and Pec (Ipek). A third see also existed at Turnovo, until 1394 when the city was conquered by the Ottomans and its bishop recognized the Patriarch of Constantinople as his superiors. These sees presided over Orthodox churches whose liturgical language was Church Slavonic, not Greek. They stood for the national identity of Bulgarians and Serbs, although theologically there were no differences between them and Constantinople. To Mehmed II, who was a good Muslim and recognized only theological differences, it made perfectly good sense to create only one Orthodox millet under the leadership of one partriarch who served as the millet basi. For the Christian communities of the Balkans this measure was to create serious problems.

Mehmed himself installed the new patriarch, following the ceremony established by the emperors. The only important innovation was that he handed Gennadius a berat, an imperial warrant confirming him in his new dignity. The issuance of this document was in perfect keeping with Ottoman usage, but in later years it made the sale of berats, and therefore the partriarchal dignity, possible. With the berat of Mehmed II, Gennadius became patriarch, millet basi of the Rum (Orthodox) millet, a high ranking pasa entitled to three tugs (horse tails) of the Ottoman Empire, undisputed master of a reunited church, and the official who was responsible for the behavior and loyalty of all the Sultan's Orthodox subjects.

Besides full ecclesiastical powers and jurisdiction, the partriarch acquired legal powers in those cases, such as marriage, divorce, and inheritance, that were regulated by canon law. Concommitant with these legal powers went certain police powers that even included a patriarchal jail in Istanbul. Naturally, the church was permitted to collect the usual ecclesiastical dues, but it was also made responsible and was often consulted in the assessing and collecting of taxes due the state. Finally, ecclesiastical courts had the right to hear and decide cases in which all litigants were Christians, provided they voluntarily submitted their cases to church courts rather than to the kadi.

In this manner the traditional hierarchy, from the patriarch down to the lowly parish priest, was invested with a great number of administrative and legal functions in addition to the traditional ecclesiastical ones. A bureaucracy developed that was on a lower and more limited level than the official bureaucracy, but nevertheless paralleled it. In this manner the church became the only "national" institution with which the Orthodox could identify. The fact that, unlike the Ottomans who considered them to be members of one millet, the Orthodox Christians did not feel as members of one "nation" made the subordination of the Slavonic Rite churches to the Oecumenical Patriarch a significant move that created serious difficulties for him, the central government, and the Balkan people especially in the eighteenth century.

For all practical purposes the new powers conferred on it by Mehmed II made the church a state within a state. As time passed the hierarchy began to think of itself both as the de facto ruler and protector of the Christians and as the sole body to deal with the central authority. It performed these double duties with great skill, and the lives of the Balkan Christians depended as much on the efficiency of the ecclesiastical authorities as on that of the Ottomans. This explains why differences among them, which have already been referred to in connection with the Ohrid and Pec metropolitanates, were fought out to a large extent within the ecclesiastical realm. The church that survived the destruction of state had to become the institution on which a revived nation-state could be rebuilt.

Before we leave the Orthodox millet, a few words should be said about Cyprus, although this island was not conquered by the Ottomans until 1570-73. After conquest its inhabitants were enrolled into the Orthodox millet, and some Turkish garrisons placed on the island totally restructured the ethnic-ecclesiastic situation. Herein lies the origin of the modern problems of Cyprus.

Cyprus had had an autocephalous church since the Council of Ephesus in 431. This church consisted of twenty bishoprics under an archbishop. After the great schism the island was solidly Greek Orthodox. During the Third Crusade King Richard I of England captured Cyprus, in 1191, and sold the island to Guy de Lusignan whose successors ruled the island until 1475, when it became a Venetian possession. Thus, for something like four hundred years preceding the Ottoman conquest, Cyprus was ruled by Roman Catholics. During this period the archiepiscopal see became Roman Catholic as did sixteen of the twenty bishoprics. Some of the people became Catholics, but the majority remained Orthodox. When the Ottomans came, the Roman Catholic hierarchy and part of their flock left the island. Turks who had come to the island originally as occupation troops subsequently became settlers. The remaining inhabitants became members of the Orthodox millet, and while the number of bishoprics remained four, the archbishopric was re-established. In this manner the Turkish conquest was also, ironically if we think of the present situation, the Orthodox "reconquest" of Cyprus made possible by the millet system.

The millet basis of the other two millets, the Haham basi (chief rabbi) of Istanbul of the Jews and the Gregorian Patriarch of Istanbul, had the same rights, privileges, and duties that the Orthodox patriarch enjoyed. Therefore, only those aspects of their offices and subordinate organizations will be mentioned that differ significantly from that of the Rum millet.

Although, as already mentioned, the Jewish (Yahudi millet) was not officially recognized until the nineteenth century, Mehmed II appointed the first Haham basi, Moses Kapsali, in 1453. At the same time he declared that Jews were permitted to settle in Istanbul. Kapsali was given protocolar precedence over the partriarch, and beginning with the rule of Suleyman I, the Jewish community was the first to be given the privilege of appointing an agent (kahya) to represent it before the central government. While the Haham basi needed the sultan's berat to assume his office, all of Moses Kapsali's successors were freely elected by their coreligionists.

There were several reasons for the preferential treatment given to the Jews. While Mehmed II looked at the Orthodox as potentially the most loyal of his Christian subjects, he was certain of the fidelity of the Jews. No other state in fifteenth century Europe had treated them better than had the Ottoman. Since the days of Murad II they had served the sultans well, mainly as court physicians. Furthermore, they possessed several valuable skills including the knowledge of languages (besides Turkish, Arabic, and Persian) that were occasionally needed. Moreover, the Jews were the original "people of the book."

In the eyes of the Ottoman government, the Jews, like the Orthodox, were one millet. The Jews did indeed present a united front toward Muslims and "gentiles," but they were anything but united within their own community. First, there was a marked difference between those Jews who had been living in the land at the time of Ottoman conquest and the emigrants who, persecuted elsewhere and learning about the tolerance of the Ottomans, came in great numbers in the fifteenth and sixteenth centuries. Furthermore, both of these groups were further divided into two sub-groups. The "original" Ottoman Jews belonged either to those who followed the Talmud (the rabbinical Jews), or to those who did not (the Karaites). The latter group formed during the rule of the 'Abbasids in Baghdad, first appeared in Palestine and then spread all over the lands of the caliphs. Only a few of the Karaites reached those parts of Europe under Byzantine rule. There, Greek-speaking Talmudic Jews known as Romanios or Gregos were in the majority.

The emigrants were either Sephardic or Ashkenazi Jews, depending on their land of origin and their mother tongue. Of these two groups the Sephardic Jews are the more important for two reasons. First, they were the most numerous Jewish settlers in Istanbul and the Balkans; second, they became the most influential group and dominated communal life and the election of the Haham basi for centuries. Besides Istanbul, where they became the best known armament makers and performed the already mentioned functions for the central government, Salonika, Edirne, Nikopolis, and to a slightly lesser degree Sofia and Sarajevo became important Sephardic centers in the Balkans where important Jewish cultural developments occurred during Ottoman rule.

While some of the Ashkenazi emigrants settled in the European lands of the Ottomans and in Istanbul, most moved to Palestine. They came from Germany, Austria, Hungary and the Romanian Principalities. In the Holy Land they soon mingled with rabbinical Jews and became the traditional Jewish population of Palestine.

The Jews were almost never involved in agricultural pursuits, but their tax loads and obligations were no lighter than those of the other reaya because taxes were attached to their professions. In order to insure the unity of this millet, Selim I abolished the office of nagid, the highest ranking Jewish office in the Near East, when he conquered Cairo in 1517. Because religion and "nationality" coincided for the Jews in the early modern period, the millet system suited them perfectly. Under that system they were able to live for centuries better than anywhere else.

If the Jewish millet was a case of theory coinciding with practice, the Armenian millet, which lacked both ethnic and religious homogeneity, was a prime example of theory diverging from practice. Although some Armenians maintained a connection with Rome, the great majority belonged to a monophysite church, which had been established some time in the third century by Gregory the Illuminator. Once a strong church spreading beyond the borders of Armenian-inhabited territories, it had lost some of its influence by the time the Ottomans appeared. Its strongholds were outside the realm of Mehmed II in Greater and Lesser Armenia, and the seat of its head, the Catholicos, was also outside Ottoman territory, in Erivan (Echmiadsin). Yet there were several people in the Ottoman Empire who did not fit into any of Mehmed II's millets, and these not only included Armenians but also an Armenian bishop who resided in Bursa. He was clearly the highest ranking religious dignitary who could not be fitted into either of the two other millets in the lands ruled by the Ottomans. This bishop, called Horaghim, was appointed Armenian Patriarch of Istanbul by the sultan, and his millet was recognized in 1461.

This millet is so special because its head, who had all the rights, honors, privileges, and duties of the other millet basi, officially presided over those subjects of the sultan who could not be subordinated to the leaders of the other two millet groups. While he was not responsible for the few Roman Catholics, who were considered members of the Orthodox millet, he was in charge of the splinter sects and heretics living in the Ottoman Empire. In Europe the most important of these were the Paulicians and Bogomils, although by the time of Mehmed II their numbers and importance were declining. Only a small group, the Pavliniki, survived into modern times near Sofia. It was from these European heretical groups that the first major groups converted.

4. THE PROBLEM OF CONVERSION

It was, of course, strictly forbidden under penalty of death for a Muslim to leave his faith. The Ottomans did not pay any attention if zimmis converted to a religion other than Islam because this did not affect their status in the Ottoman Empire. What was at issue was conversion to Islam. When the Ottomans first established a permanent foothold in Europe in the middle of the fourteenth century, there was a small Muslim colony in Constantinople and a few small settlements scattered throughout the Balkans proper. By the beginning of the 1970s there were almost seven million Muslims living either in European Turkey or in those parts of Southeastern Europe that had once been under Ottoman rule. Of these about 2.5 million lived in Turkey, the same number in Yugoslavia, an estimated 1.2 million (no official figures are available) in Albania, 0.5 million in Bulgaria, and the remainder either in Greece or Romania.

Ethnically, the Muslim population of the Balkans is either Turkish or Slav, Greek, Albanian, etc. The Turks came as settlers, and the others were converted. This early and important transformation in the ethnic and religious composition of the Balkan population deserves a few words.

The problem of the Turkomans and the attempt of the Ottoman authorities to settle them has already been mentioned. Their influx into Ottoman lands was especially strong in the fourteenth and early fifteenth centuries, but, like the emigrants who came to the United States, most of these nomads settled in the eastern provinces of the empire. In the fourteenth century the Ottomans moved some of them to Europe, particularly into the lands of today's European Turkey, where they were either settled as musellems or remained nomadic for a longer period as the already mentioned yuruks. As we know, after this first massive settlement of Turks most of those who were brought over from Asia Minor were settled along major military roads, at fortified points, or in the major cities. The only Muslims in the countryside were the sipahis and a few officials, the descendants of whom can still be found in these strongly Christian regions.

In 1478 the Ottomans conducted a census in Istanbul, and in 1520-30 for the provinces. They counted taxable hearths (households), not individuals. The Muslim and zimmi hearths paid different taxes so they were counted separately. These figures were analyzed and published by the excellent Turkish historian, Omer Lutfi Barkan. His findings confirm in detail the general picture. In Europe 18.8 percent (194,958) of the hearths were Muslim, 80.7 percent (832,707) were Christian, and .5 percent (4,l34) were Jewish. Of the Muslim hearths 85 percent -- the great majority -- were concentrated in ten of the then twenty-eight European kazas. In only four of these kazas, those of Vize (in today's European Turkey), Silistra and Chirmen (both in Bulgaria today), and Gallipoli, were the Muslims in the majority. Outside of the eastern Balkans there were also significant Muslim settlements in Macedonia, Bosnia-Hercegovina, and Thessaly, regions that present a special problem.

Equally important is the fact that even in those regions where the Muslims were numerous they were heavily concentrated in the major cities. The figures given by Barkan and analyzed further by Vryonis for twelve major cities are highly relevant. The urban nature of the Muslim population is well demonstrated by the figures given in table 1 and is quite understandable. The settlement of nomads and the granting of fiefs occurred in the countryside, but the major centers of provincial administration, and most importantly the points of interest to the trade guild-connected akhis, were the cities, which also controlled major roads that the Ottomans wanted in safe hands.

Of the Muslim hearths listed only 19 percent were nomadic, the rest being families settled either in the cities or in the countryside. Vryonis argues convincingly that about 30 percent of the additional Muslim hearths listed in the 1520-30 census can be attributed to Turkish immigration, leaving about 96,500 hearths that were Muslim as the result of conversion. Several examples of forced conversion exist dating from the period when the empire became more conservative than it was in the earlier centuries, less tolerant and even fanatical. The best known example is the conversion drive in the Rhodope Mountain region beginning around 1666. In the period predating the census on which these figures are based, however, only the devsirme can be considered forced conversion. Yet, because those taken under this system were removed from their native land, this form of forced conversion did not influence the religious composition of the Balkans revealed in these figures. The conversion of the Albanians is also a seventeenth century phenomenon; earlier even the majority of the timarli were Christians. Thus, these earlier conversions must have been "voluntary. "

Vryonis lists as the major reasons for earlier conversions: economic and legal advantages, the influences of the medreses and other Muslim institutions, fear, and the adaptability of folk religion. He is right, but I and other scholars believe that the fourth reason is the most important. Folk-Islam and the role of the akhis has been mentioned already in the first chapter. The role of Seyh Bedreddin both under Musa and later as the leader of a popular revolt of 1416 indicate that similar beliefs must have existed among the majority of his followers, the Balkan Christians. According to anthropological and sociological studies among both the Greek and Slav Orthodox in the Balkans a phenomenon paralleling that discussed in connection with Islam existed. Here, too, lack of thorough religious training, the resulting misunderstanding of basic dogmas, the survival of pagan rites connected with fertility, health, etc., in an environment that did not change with the acceptance of Christianity, the willingness of the clergy to accommodate their flocks, the great division between higher and lower clergy, and the discontent with authorities who were closely identified with the "established" church led to the inclusion of enough pagan rites, customs, superstitions, and beliefs to create a folk culture and religion whose remnants are still observable in the folk customs of today's Balkan people.

Many of these rites and beliefs were universal. People everywhere were interested in rain at the right time of the year, in fertility, and in other factors basic to primitive rural life, and were continually seeking to ensure that everything would go right by creating spirits, demons, etc., who if handled properly would behave in a manner favorable to those who depended on their good will. If one keeps in mind that the customs connected with these basic beliefs, fears, and habits were only in part Turkish or Islamic, and remembers that part was learned from the Greek inhabitants of Asia Minor during the long centuries of frontier life, the similarities between certain aspects of folk religion as practiced by the Muslims and Christians will not come as a surprise.

The Christians, of course, had certain rituals and beliefs that the Muslims did not -- or rather were not supposed to -- have. Among these the belief in saints, the use of icons, and baptism were the most important. As we know, the akhis were able to accommodate everybody on the frontier with ease. By the time that the Ottomans began to make serious gains in Europe, the akhis had become pretty much 'urbanized," almost a part of the "establishment," and were basically nothing more than guild organizations with broad, folk-religious codes that permitted them to settle in the towns of Europe, organize and lead the various guilds, and dominate town administrations. Together with the medreses and other religious-charitable institutions they probably account for conversions, possible mainly on a family basis, that occurred in the cities.

The role that the akhis once played in the Anatolian countryside now devolved to the dervish orders which were responsible for the important conversions that occurred prior to the 1520-30 census in Macedonia and Bosnia, and for the later ones in Albania. Although some of these dervish orders followed the "road" of the akhis -- and became urban and well established -- the most important of these were the Bektasi and the Mevlevi -- most of them remained basically rural and retained their folk character. It is not quite clear how many different dervish "orders" and "suborders" existed at any given time and how many operated in the European provinces where the two groups just mentioned made a lasting early impact. We do know that their numbers were considerable.

Like the akhi, the dervishes had their tekkes, and their code or path (tarikat) describing the proper life leading to a mystic understanding of Allah. And like the akhi, the dervishes built on the long tradition of sufism and independent search for the right life. Yet, there were marked differences. Soon after his death the founder of a given dervish order, around whose home or tomb the central tekke was usually located, became venerated much as a Christian saint and was often credited with miraculous powers. These "saints" were known as the friends (evliyas) of God and could be dead or living. The ease with which the patron saint of a European village could be established as one of the evliya recognized by a given dervish order requires little explanation.

In order for this identification to occur, the dervishes had to move around and reach the various localities of the Ottoman Empire. Unlike the akhi, the dervishes wandered almost constantly, preaching and practicing their tarikat and numerous related ceremonies. They were the babas, a sort of combination of holy man, miracle worker, medicine man, etc., and were often regarded as living saints. Their eclecticism and pragmatism knew practically no bounds. Given the numerous similarities between folk-Christianity and folk-Islam, they had little difficulty in fitting local customs into their tarikats. Furthermore, what they preached had certain advantages. The old formulae that ensured "good fortune" were broadened by the addition of customs that they had brought with them, and by the not negligible circumstance that those who followed them passed from the zimmi to the Muslim group. What emerged was a curious variety of European, or rather Balkan, "folk-Islam," which included icons, baptism to prevent mental illness, and many other basically non-Muslim features.

It was not difficult for Christians whose faith was of the superstitious folk variety to pass over to a similar but more secure folk version of Islam. I believe that this explanation of the early mass conversion, advanced by several scholars, is more believable than the equally popular interpretable lion that attributes such conversion either to the wish of the population to retain its landed possessions or to the desire of previously persecuted heretics (mostly Paulicians and Bogomils) to become the master of theirs oppressors.

There were people like the Christian timarli in Albania elsewhere in the Balkans, too, in the early period. Furthermore, a Christian family could retain a basic cift, although it could neither increase its size nor subdivide it. To these considerations can be added the example of Bosnia, where in the rural regions Bogomils were numerous at the time of the Ottoman conquest.

Table 1 shows that by 1520-30 Sarajevo was 100 percent Muslim. While, this city had been built by the Ottomans on the site of a small village, the ratio of Muslims in other towns and cities was equally impressive. The same table indicates that the kaza of Bosnia as a whole was 46 percent Muslim by this time. For this kaza there is also data from a census taken about thirty years earlier, in 1489, indicating that at that time the Muslims made up only 18.4 percent of the district's population. Obviously, a mass conversion occurred during that thirty-year period and was strongest in the urban centers where the Bogomils were the weakest. This problem has not yet been resolved, nor have similar cases elsewhere been fully explored. It might be perfectly true that with time the great majority of the Bogomils, Paulicians, and other "heretics" turned to Islam, but it appears unlikely that they were the spearheads of conversion.

One final argument supports the theory that conversion was really only an easy transition from one folk level to another. Religious social movements, which took newer forms with time, not only remained a constant phenomenon leading to several serious disturbances to be discussed later, but were also so strong that they even affected the Jewish millet. There are many reasons why conversion to Islam was much less extensive in the European than in the Anatolian provinces of the Ottoman Empire, but it is not necessary to investigate them in the context of this study. What appears to be important is that while numerous socioeconomic factors did exist that would have made conversion appear desirable, the Ottoman's rapid regulation of the situation in the various European provinces made those factors less impelling than they were in constantly troubled Anatolia. In Europe conversion was limited pretty much to certain elements who had really never understood or practiced their faith correctly and for whom, therefore, apostasy was less a question of belief than of convenience.

5. THE DEVISIRME SYSTEM

If voluntary conversions were convenient for those who changed their faith, the devsirme system, the best known form of forced conversion, was convenient for the sultans. In this system were contained all of the issues discussed in this chapter: the professional Ottomans and their institutions, the reaya, and conversion.

The verb devsirmek can be translated to mean either to collect or to enroll. From the point of view of those whose children were periodically taken, the devsirme certainly was a collection of a very special due or levy. For the Ottomans it represented an enrollment of new recruits into their well-established institutions. It is important to realize that the slave institutions of the sultans, including the janissary corp (yeni ceri), antedate the establishment of the devsirme. It has already been mentioned that not only was slavery a long-established institution in the Muslim world by the time the family of Osman became important, but attention has also been directed to the nature of Muslim slavery and to the slave body guards, mostly Turkish, of the 'Abbasid caliphs in Baghdad.

The Muslim ruler, like every other person with sufficient means, could buy slaves, but he had also been entitled since the earliest Muslim days to what the Turks called pencik, the possession of one-fifth of all prisoners of war who could be enslaved because they resisted. It is traditionally accepted and probably true that Murad I (1360-89) organized the first janissary units from the slaves he owned. When the devsirme system began and when it turned into the major source for recruits is not certain, although tradition credits Murad with this innovation as well. Our earliest written reference comes from a Christian sermon of 1395. That slaves were used early not only as cannon fodder is attested by the fact that when Bayezid I, at the beginning of his rule, subdued various Turkish principalities in Anatolia, he appointed slaves as their governors. Thus, by the early 1390s a slave system was attached to the Ottoman court. The devsirme recruits also served this double function. Despite the uncertainty it appears likely that the devsirme was introduced sometime during the last quarter of the fourteenth century.

Equally uncertain are the frequency of the levy, the number of youths enrolled, and the date of abolition of the devsirme. It appears most likely that there was no fixed time interval between the levies, nor was there a specified number of recruits to be furnished. The levies were probably ordered when recruits were needed, first infrequently, then frequently, and finally again rarely, and in each case the authorities fixed the number of youths according to their needs. I am one who believes that the total of young Christians taken "officially" from their homes by the devsirme amounted to approximately two hundred thousand during the roughly two-hundred-year period when it was practiced. Traditionally, Ahmed II (1691-95) has been credited with the final abolition of the system. This is not certain, but as a regular measure it ceased toward the end of Murad IV's reign (1623-40), and continued on an occasional basis until sometime toward the end of the seventeenth century.

Even when it was the major source for the recruitment of kapi kulus, the devsirme was by no means applied to all zimmi living in the Ottoman Empire. Once again, numerous details remain without satisfactory explanation. The acquisition of slaves by the sultan was certainly not only permissible but clearly "legal" according to the shari'a under circumstances defined by the pencik or by purchase. The devsirme affecting "protected people," however, appears to have been "illegal," although Paul Wittek argues that according to the shafi'i legal school only those who were "people of the book" prior to 622 were entitled to zimmi status. This might have explained why relatively few Greeks and Armenians were drafted had the Ottomans followed shafi'i doctrines; they did not. Whatever the explanation, in practice the devsirme exempted youths residing in cities, married young men, voynuks, dogancis, and people in other special occupations such as mining and in villages responsible for repairing passes and roads. There were other temporary exemptions, and in practice a large number of those subject to the child levy were Slavs from the Orthodox millet. Yet, there are too many exceptions to allow any clear pattern to emerge.

From the point of view of the sultans the kapi kulu system made excellent sense. It followed the established tradition of slave body guards, put the various slaves to good use, and created a group of men entirely dependent on the ruler. Bayezid I's appointment of slaves as governors was not without precedent in the Muslim world. What was new, besides the devsirme, was the ever broadening use to which slaves were put. When eight months after the conquest of Istanbul Mehmed II appointed a kapi kulu, Mahmud Adeni, grand vezir, the displacement of free-born Turks in the administration was completed, although occasionally a few of them achieved high office even before the kapi kulu system disintegrated some two centuries later. The reasons for this switch to slaves should be obvious from even the few facts of Ottoman history that have already been presented.

The various princes of allied Turkish states were always trying to regain independence, and even after they became leading families of the Ottoman upper class, they stuck to the key tradition and remained a force that could challenge the power of the ruler. The most remarkable example of this behavior presented thus far was the deposition of Mehmed II and the re-enthronement of Murad II engineered by Halil Cenderli in 1446. No wonder, then, that it was Mehmed II who completed the steadily growing tendency of the rulers to rely only on their slaves and that it was he who gave the slave training schools their final form.

The reason for establishing a slave-ruled administration does not explain the innovation of the devsirme system. There are no facts but there are three rather good arguments for the introduction of the devsirme. The first is simple and obvious: while the purchase of slaves was expensive, the recruitment of devsirme youths was free, although certain slaves continued to be purchased in the Crimea and the Caucasus. Furthermore, if the last decades of the fourteenth century are accepted as the time when the devsirme was introduced, this innovation coincides with a period in which slaves were not only very expensive, but in very short supply. In Europe the conquered lands were, as a rule, attached as vassal states to the Ottoman Empire and could not supply pencik manpower in sufficient quantity. In Central Asia, Persia, and Iraq, the rising Mongol Empire was cutting off the slave trade routes and reserving slaves for itself, while in Anatolia even direct conquests could not supply enough slaves because most of the conquered were Muslim Turks. It was at this time of slave shortages that Murad I and Bayezid I, who had great difficulties with the begs, extended more and more the use of slaves in the administration and military of the Ottoman Empire increasing the demand for slaves. The coinciding of these factors might possibly explain the introduction of the devsirme way of acquiring kuls.

The reason for extending the imperial slave system by collecting devsirme recruits served not only to recruit cannon fodder as janissaries, sipahis of the Porte, and later as sappers, artillery men, and other special troops. While the soldier-slaves were in the majority and were assigned to the leading men or the acemi oglani (foreign sons) training establishments of the birun services, the most promising among them went to the various enderun schools as icoglanis (inside sons-pages), and, as previously indicated, became the masters of the state on graduation.

The fact that most of the masters of the state for two-hundred years were of Balkan-Slav-Christian origin raises the last set of problems to be discussed before this section, and with it the introductory part of this volume, can come to a close. What did the devsirme mean to those whose children were affected by it? Did they profit from friends and relatives in high places, or was the experience entirely negative?

Quite apart from the inhuman features of a system that separated children forever from their parents, the negative aspects of the devsirme are obvious, if we remember that the recruiting ground was relatively narrow (mainly the Slav Orthodox), that the ablest youngsters were taken, and that the number of those removed was greater than the previously offered estimate of "offficial" devsirme recruits because the yaya-basis (janissary officer in charge of selecting young people), in cahoots with the local kadis and sipahis, made a profit by illegally gathering and selling extra boys as slaves. The ethnological and economic damage resulting from this practice has never been accurately measured or scientifically estimated, but it must have been considerable. Although these people were forced to convert to Islam and were often fanatical upholders of the Ottoman system and their new faith, it is very unlikely that they completely forgot their origin and native language. Some of the fairly good relations certain provincial officials had with their charges and the rapidly developing modus vivendi between some sipahis and the peasantry living on their fiefs can probably be explained by these memories.

There is better proof that the Ottoman officials of devsirme origin did indeed remember their childhood. Ibrahim pasa of Greek origin, who held the office of grand vezir under Suleyman I for thirteen years prior to his execution in 1536, was accused of many wrong doings while in office, including the charge that he feathered his relatives' nests. Mehmed Sokollu, grand vezir from 1564 to 1579, not only established contact with his family, but also helped all the Serbs by successfully arguing for the re-establishment of the Metropolitanate of Pec in 1557 with his brother as archbishop even before he had achieved the highest dignity of the state. While the deeds of the most famous are the best known, it is more than likely that lesser grandees acted in a similar manner. If this had not been the case why were the yaya-basis frequently bribed not only to overlook one's children, but to pick them even if they did not meet the standards of recruitment? Finally, a striking example exists in the case of the Bosnian Muslims who insisted that their children be taken to Istanbul. As early as 1515, at the request of the Muslim community, one thousand young Bosnians were enrolled in the schools of the enderun without the customary selection process that divided the acemi from the icoglanis, and the process was repeated several times later in the sixteenth century. Both those Christian parents who paid bribes to have their sons enrolled as well as the Muslims in Bosnia and elsewhere who tried to extend the devsirme to their children must have expected certain advantages not only for the youngsters but for themselves as well. Otherwise their behavior would make no sense at all.

In spite of these possible advantages, the disadvantages in human, ethnological, and economic terms were overwhelming and, on balance, far more important than the beneficial aspects of the devsirme system. Together with a use of slaves that was far more extensive than in earlier Islamic states, the devsirme that made this slave system possible was a unique Ottoman innovation. It is also the most prevalent example of forced conversion in the years predating the seventeenth century.