In October 1603, a turnacıbaşı[1] from the 73rd regiment of the janissary army entered the city of Bursa accompanied by his men. He was carrying a decree that gave him the power to levy Christian children from the region and bring them to the capital city, Istanbul. As he entered the city, the news of his arrival spread. The turnacıbaşı went to the qadi’s office to register the decree and ask for the collaboration of the qadi to gather the boys of Bursa that were under his responsibility. For the coming two months, the boys would be selected according to clearly defined criteria laid out by the government, then organized under sürüş (batches, lit. herds) of 100 to 150, converted to Islam and transported to Istanbul in order to begin their training according to Turkish customs and the Islamic religion in becoming the sultan’s servants.
The arrival of the janissary officer in the town of Bursa was part of a much larger devshirme process. In the last months of 1603 and the early months of 1604, four different groups of janissary officers were sent to four different areas of the empire to levy children, gathering a total of 2,604 boys that year. The devshirme system was a method used since the fifteenth century to fill the administrative and military ranks of the Ottoman state and army.[2] According to what we know about this system, the officers levied the male children of Christian families (mostly Greeks, Serbians, Bulgarians, and Albanians), as a form of tribute in kind, instead of the head-tax (cizye).[3] These boys were subjected to a second selection process in the capital, after which officers placed promising children in the palace schools to be educated as administrators in various capacities, recording the rest as novice boys (‘acemi oglans), and hiring them out to Turkish villagers in Anatolian towns for a period from between three to five years. In these towns, the children worked as agricultural laborers, learning the Turkish language and Islamic practices. After the agreed period, they were then called back to Istanbul to work for three to five more years as laborers in the city before being enlisted as soldiers in the janissary army.
However, this situation is the “official” version of what happened throughout the devshirme process. In reality, we see varied and different forms and practices of “becoming a devshirme” in Ottoman society, as will be outlined in this article. The devshirme status involved multiple layers: some could become high-level bureaucrats in the Ottoman administration, or turn into successful soldiers who were promoted to high offices in the janissary army. Alternatively, some might end up becoming heavy-duty workers in state workshops such as gunpowder workshops, or workers in the Arsenal - the possibilities were endless. In short, the extent to which one was integrated into the Ottoman system determined one’s possibilities and limitations. Furthermore, reactions to being levied were varied, and this paper intends to show this multiplicity.
The devshirme system has been much-studied by Ottomanists, who have concentrated mostly on questions such as the debates concerning the origin of the practice, whether or not it was legal or not according to Shari’a law, or as a means of understanding how Ottoman bureaucracy was created.[4] Many Balkanists, in contrast, have portrayed the devshirme system as one of the many cruel aspects of the Ottoman invasions that devastated and stripped Balkan cities of their youth who were forced into slavery.[5] Some of these studies were fed by the nationalist ideologies that arose in the nineteenth century and used the image of the barbaric Turk taking children away as a metaphor of the national awakening of the Greeks, Serbians, and Bulgarians. Many of them justifiably pointed out the inhumane nature of the practice.
This article focuses on the process by which children were chosen as devshirmes and what happened to them directly afterwards, not on their status after they became devshirmes. The discussion of whether the devshirmes were slaves or not is, however, still worth debating.[6] On the one hand, the status of devshirme could certainly be regarded as a form of enslavement as it was not usually a voluntary choice. On the other, being a kul of the sultan was far more complex than simply being a slave as we understand it today, since it was accompanied by privileges such as owning land, and even owning slaves. However, such complexities require that the debate on the status of devshirmes be given special treatment in a separate article, and will not be discussed here.
The purpose of this article is to examine the devshirme system from the ‘bottom’ - concentrating on the children as the main actors whose lives were drastically transformed by this system. Who were these children? How were they selected? What were their reactions? The goal is to accumulate as much information as possible about the experiences of levied boys, and to learn more about the dynamics of the selection process. Tracing the stories of the children in the process of being levied is an exciting journey for a historian, despite the limited sources.
Here, I will attempt to reconstruct the experiences of the levied boys by examining the eşkal defter (register of levied children) of 1603-4. This is a unique register that provides detailed information on the children levied during these years. It presents information on the original names of the levied boys, the Muslim names given after they were levied and converted, their parents’ names, and the boys’ physical characteristics and ages. The decree given to the janissary officer who was sent to Bursa to levy children in 1603 is another important document that is used extensively in this study. The decrees concerning child-levy in the mühimme registers, and the kavanin-i yeniçeriyan are also consulted.
This article is composed of three main sections: first, I will present the general outline of the levy of 1603-4 based on the above-mentioned register, such as how many children were levied and from which regions. Second, I analyze the actors who were part of the levy process. My goal is to understand the devshirme system both from the point of view of the Ottoman statesmen as well as to regard it with the eyes of the locals who were subjected to child-levy. In doing so, I claim that child-levy became an important event that directly affected the local political dynamics in the regions where the child-levy took place, and turned the local elites into actors in the selection process. In this part, a section is devoted to the runaways – the boys who escaped being levied. Those children, by rejecting being levied, themselves became actors. Because we have obtained a more complete picture of the process in the Bursa region due to the document (the decree given to the child-levy officer who was sent to Bursa), we can gain a better understanding of how the boys were gathered, the problems that were faced during transportation of the children, and on the resistance methods of the families and the locals. Third, I take a closer look at the levied boys from the areas around Bursa. I will present a map of the regions where the children were levied, their ethnic background, appearance, state of health, and ages. As mentioned above, my goal in this article is twofold: first, to illustrate how the devshirme system worked as an institution, and; second, to trace the journey of the levied children, who they were and what they experienced once they were incorporated into the devshirme system.
1. 1603-4 Child-Levy
According to the 1603-4 eşkal defter, a total of 2,604 boys were taken during the levy of 1603-4. The register that provides information on the 1603-4 child-levy is organized into twenty groups. These groups consist of boys chosen from Rumeli, the Balkans, Albania, Bosnia, and Anatolia. In Anatolia, they were taken from various regions such as the area around Bursa. Four different groups of officers were sent to these areas and each levied independently of each other.
The first group of officers worked under Serseksoncu Mustafa in Rumeli, where seven groups of boys were gathered and sent to Istanbul separately. The Rumeli group is the best-recorded in terms of dates, and we can trace the route of the officers clearly. Surprisingly close to the capital, they started levying from the area along the Marmara Sea, west of Istanbul along to the Gallipoli peninsula: Silivri, Rodoscuk, Migalkara, Kavak, and Gelibolu.[7] From there, they sent 109 boys to Istanbul, and took another 105 children from Midillu Island. They continued towards the area of Ilmiye, İnöz, Keşan, İpsala, Megri, Firecik, and Dimetoka to conscript 104 children, moving further west to Gümilcine, Yenice-i Karasu, Taşyüzü, Baraketlü, Pravişte, Draman, Kavala, and Zihna to take another 168. The fifth group comprised 127 children from Siroz, Timurhisar, Selanik, Avrethisar, Yenice-i Vardar, Vodane. The officers then moved to the south along the Aegean Sea to the towns of Karaferye, Alasonya, Dominik, Tırhala, Yenişehir, Fenar, and Serfiçe. Finally, 128 children were collected around Izdin, Modenç, Salona, Atina, and Agrıboz. The officers levied the first group in January, and when they reached their final destination it was August.
Another group of officers levied children from around Bosnia. Four groups of children were taken almost every three months from approximately the same regions. Unlike in Rumeli, where the officers moved systematically along a given route and departed each location after selecting the boys, the levy-officers in Bosnia wandered around the same region and collected groups to send to the capital every three months in a process that took almost a year. It is safe to assume that the presence in the region of the levying officers for this length of time probably gave town dwellers a chance for negotiation. This might be done both by preventing some boys from being levied and sneaking others into the levied groups. Perhaps the officers, as in the case of Bursa, were sent to the area in the autumn, and it took them three months to select the first group of children. In the following period, the officers selected children approximately every three months from the towns of Hersek, Mostar, Nüvesin, Balagay, Yeni Pazar, Imoçka, Foça, Vişegrad, Bosna, Saray, Gabala, Teşene, Çayniçe, Taşlıca, Tuzla, Mostar, and Çelebipazarı.
It should be noted that of the groups sent from Bosnia, unusually, 410 children were Muslims, and only 82 were Christians. This was due to the so-called ‘special permission’ granted in response to the request by Mehmed II (r. 1451-1482) to Bosnia, which was the only area Muslim boys were taken from. These children were called poturoğulları (Bosnian Muslim boys conscripted for the janissary army). They were taken only into service under bostancıbaşı, in the palace gardens.[8]
The third group - that of the liva of Avlonya, was sent to present-day Albania, where the officers began to select boys from the south of the region. The actual levy process lasted four months, but the selection might have started earlier than this. They levied 122 boys from Merdak, Eregri-kasrı, and Pogonya, and then they moved on to Premedi to take 130 boys. They conscripted 194 children from the kazas of Avlonya, Müzakiye, and Belgrad, and finally, from Ilbasan and Işpat, they took 122 boys. One of the groups from the Manastır and Pirlepe area in the province of Rumeli seems to have been levied by the Avlonya group. This batch is recorded as a levy from Rumeli, but the style of the scribe, and the route of the officers denote that this batch was more likely to have been collected by the officers appointed to the Avlonya area, where 145 boys were taken.
In the final group, four batches were taken from Anatolia, having been selected from the Christian villages around Bursa, Biga, and Kocaeli. This group of officers did not date the levy in detail, but noted down that it was during the month of January and February. The children were picked from around Kocaeli, Iznik, Lefke, Akhisar, and Yenişehir in the first month, and in the next, from around Bursa, Mihaliç, Manyas, Bilecik, and Biga.
2. The Levied Children of Bursa and the Local Politics of ChildLevy in the Towns
If we assume that the levy of children from a certain town would have invariably been moments of crisis for the inhabitants, the documents can be looked at from another perspective - from the eyes of the locals who were subjected to the devshirme system. We must remember that, whether a devshirme was a slave or not, this levy was not a choice. Those who had been selected by the officers would be taken, and those who had not been selected would be left. The villagers, therefore, found other ways to interfere with the selection, and this raises the question of whether it was always a smooth process. Was the system applied exactly as it was formulated? How did the local power relations and political ties intervene in the selection process? Or, more from a more humane perspective, what did those parents who did not want to give their children away do?
One aspect that needs consideration is the question of time. The total length of time it took for the officers to select boys as devshirmes would have directly affected events during the selection process. In Ottoman history books, the levy has been portrayed as almost a spontaneous event where janissary officers stormed into villages and selected whomever they clapped eyes on. The documents, however, suggest a different reality. For instance, the turnacıbaşı arrived in Bursa and registered his decree of child-levy in the qadı’s office in October 1603. The decree ordered fathers to bring every boy aged between fifteen and twenty years from the villages, counties (bilad), waqf lands, and fiefs (tımar), to the turnacıbaşı.[9] From then on, the officers commenced the selection of boys in a process that lasted for two or three months, with the groups of boys being formed in January and February 1604. This levy in Bursa lasted for almost five months after the arrival of the officers in the town. The levy of 1603-4 in other regions lasted even longer than the Bursa group: in Albania four months, in Rumelia eight months, and in Bosnia an entire year. It can be assumed that the process in other regions was similar to the Bursa case; the levy-officers arrived at their destination several months before the actual levy began. Taking this information into account, then, the devshirme should be seen as a phenomenon that occupied the towns and villages involved for quite a while - the selection of children took a long time. It was a more bureaucratic process, rather than being an occasional event in the town square.
It is worthwhile here to further consider the issue of the length of the levy-taking. Why did it take so much time? What was going on in the towns during this period? Perhaps the levies were lengthy affairs because of the meticulous work of the officers. They might have taken their time to find the groups of children from which they would make their selection. This could be a quite difficult task since there was a common tendency to hide children away from the officers, as the Bursa decree proves. It declares that the punishment for those who attempt to hide children from the officers is execution. Thus, it was not always easy to find and examine the children of the town.
The negotiation process that the locals initiated with the officers is another factor to consider. Quite naturally, the locals would have wanted to have a say about who would be selected or spared. The question of how many and who would be selected became a zone of negotiation between the state representatives and town dwellers. With the arrival of the officers in town, local politicking would have commenced. During the extended stay of the Istanbul delegation, somebody was probably paying to house and feed them, and this provision did not always come from state funds. This factor might have contributed to the negotiations as well.
While the state’s priority was the number of children to be taken primarily according to its own needs at the center, there were limitations on how many boys could be levied. In a pre-modern agricultural economy, children were seen as human resources from whom the best use should be derived. Those cultivating the land or working in the mines, for example, were not seen as suitable for levy in the regulations. The state did not want to exploit the human resources of an area to the extent that an economic drawback resulted. This opened the door for negotiation between the locals and the state representatives, and provided the locals with an opportunity to prevent some of the children from being taken away.
Wealthy landowners always tried to protect Christian children living and working on their lands from being levied. These landowners saw the young population as assets that supported production in their lands. The hass, zeamet, or waqf lands they owned were granted exemption-right papers (muafnames) from certain taxes, or if the population was mainly Christian, from devshirme. [10] In several cases, the trustee of waqfs acted as the spokesperson of villages that belonged to the waqf against the state for not giving boys as devshirme. In 1646 (1056), the trustee of the waqf of Grand Vizier Mustafa Pasha, requested from the authorities that the devshirme officers should not disrupt the locals of the villages in Talanda, since they belonged to the Mustafa Pasha waqf and were exempt from devshirme.[11] In the same year, the sister of Sultan Ibrahim I (r. 1640-48), Ayşe Sultan, petitioned that the officer responsible for levying devshirmes collected money from the villages that she owned as a fief conferred on the royal women ( paşmaklık hass) in Yanya. The officer collecting children was warned by a decree and threatened with severe punishment.[12] In 1573 (981), some villages in Filibe were exempted from giving akıncıs,[13] since they were owned by the soup-kitchen waqf of the Sultan in Üsküdar.[14]
Protecting children from being levied through presenting muafnames may have reached a point in the early seventeenth century where it generated tension between landowners and the state. This might be a reflection of the emergence of relatively stronger tax-farmers, and stronger families with large waqf lands in the seventeenth century, who then wanted to exert greater control over their labor force. The local landowners and waqf administrators of the seventeenth century might have reached to a power that could challenge or limit the extent of the devshirme levy. This struggle between the state and the landlords over the exemption rights of the villages is reflected in the 1603-4 levy in Bursa. The decree mentions that there had been villagers who did not want to give their children away, claiming that they were the re‘aya of waqf lands.
Being a village on waqf land granted them the right to be exempt from taxes and the child-levy. Harsh language was used in the decree, making it clear that no exemption would be made for those villagers, even if they had held an exemption right in the past.[15] A similar claim from Yenipazar in 1559 (967) argues that they possessed a decree exempting them from giving children away as devshirmes, and it was again rejected by the state.[16] The state found a solution in refusing the given privileges altogether and taking the children anyway.
Renouncing the given exemption rights on the waqf lands can be seen as an indicator of the tension between the state and rich landowners, especially if we compare it with the tolerance shown to the exempted villages that were considered state lands. When the villages with the exemption rights were state lands with obligatory duties, the state was more tolerant compared to the waqf villages. During the 1603-4 levy, the villagers of a Christian village called Eğerciler, in Bursa, declared that they were responsible for providing sheep to the capital, and the children of the village were very much needed as shepherds. They asserted that even though they were not obliged to give any children for the army, the officers took some anyway, and that they should be returned. The villagers’ claim that it was in tremendous need of future shepherds was taken seriously by the state, and a decree commanded the return of the children.[17]
Beyond the tension between the state and the landowners, poor villagers attempted to find legal grounds for not giving children up as devshirmes. The local community developed strategies to avoid devshirme, or at least to control who would be taken through negotiations. As was debated in the previous section, the length of time it took to levy children — from almost six months to a year— provided a suitable environment for negotiations. Leaders of the community such as voyvodas, [18] qadis, and subaşıs [19] formed lobby groups and negotiated with the levy-officers to prevent their children from being taken, especially if the land lacked exemption rights. This seems to be a common situation that the state encountered, since the levy-officer was warned against these lobby groups in a written decree in the case of Bursa.
The same warning against such actions of voyvodas, judges, and subaşıs was made in the decrees ordering levying devshirmes from Kocaeli, Bolu, Kastamonu, Çorum, Samsun, Amasya, Sinop, Malatya, Karahisar-ı şarki, Arapkir, Cemişkezek, Sivas, Maraş, Erzurum, Diyarbekir, Kemah, and Bayburt in 1622 (1032). Similarly, a warning about these groups is found in the 1621 (1031) levy decree, although the regions were not indicated.[20] But lobby groups could sometimes be effective in reaching a deal with officers. The villagers in the kazas of Karaman, for example, managed to keep their children by collaborating with the appointed officers during the 1574 levy.[21] At times when there was no collaboration, the rejection of the childlevy reached the level of rebellion. In 1540, for example, a village in Iskenderiye (Alexandria in Albania) attacked and wounded the officers who came to levy boys.[22] In 1558, villagers around Ilbasan refused to give children to the officers and rebelled against the state, and were ordered to be severely punished.[23]
In addition to the local power groups, individual attempts were taken to avoid service. Villagers sometimes tried to prevent the levy of village boys by falsifying baptism registers, circumcising them or declaring them married.[24] Some parents went a step further to get their children back. In 1564, for example, villagers from Sis came to Istanbul and kidnapped their children back.[25]
Runaways
As the land system shifted away from tımars to tax-farms during the seventeenth century, the devshirme system became more appealing to a growing number of landless youth. These tax-farming policies were the knots that tied the centralized model of the empire of the fifteenth and sixteenth centuries to the early-nineteenth-century modern state. Renting tax-farms to contractors for life through revenue contracts (iltizam) was the result of the privatization of fiscal policies that began in the seventeenth century and reached its peak during the eighteenth century.[26] Renting out the tax-farms to wealthier groups resulted in the emergence of some commoners who lacked lands large enough to provide for them. It is possible that this transformation in the land distribution policies affected the areas populated by Christian communities, and that the youth who suffered from the lack of land sought other options in order to survive. One of these was becoming a devshirme. In the 1603-4 levy, 41 percent of the boys were 18 or above (up to 20), reflecting a possible new source of child-levy. Those who wanted to sneak into the levied groups became another problem that the officers had to cope with.
It should be mentioned that there were always willing families and boys who wanted to take their chances within the devshirme system. A written source of the time confirms that there were parents who were happy to have their sons chosen, thinking that they would escape from poverty, and have the possibility of a career.[27] When we reach the seventeenth century, the problem appears to become more chronic — probably due to the new group of older boys who were willing to enroll in the system. The turnacıbaşı responsible for levies in Bursa was warned in a written decree against accepting anyone into the devshirme system who did not meet the criteria.
An interesting note in the 1603-4 levies from Bosnia and Albania implies that there were attempts of such youths to include themselves amongst those selected: the records, most likely written after the arrival of the children in Istanbul, draw attention to some children as possibly being Jewish (şekine-i arz-ı yahudi). Jews were not allowed in the janissary army, and so in suspected cases, the entire batch would be sent to the Arsenal as indentured laborers. This note shows not only that the selected boys were still screened closely to prevent Jews from infiltrating the system at the beginning of the seventeenth century, but also that the system was appealing for some youths who were trying to enroll themselves as devshirme despite the regulations.
There were many, however, who did not want to become devshirmes. When the local politicking was unsuccessful, or the parents could not help them, such children took their chances in escaping by themselves. There are documents revealing that some levied children escaped back to their homelands and converted back to Christianity. The escapes generally took place during transportation, or after they were placed in their obligatory service after arriving in the capital.
The transportation of levied children to Istanbul was a serious matter that the levy-officer had to arrange properly. To prevent escapes or kidnaps during the transfer, all the boys were dressed in red clothing (kızıl aba) and a conical red hat (külah).[28] Rather harshly, the cost of clothing and transportation was charged to the families of the levied children.[29] The characteristics of each selected boy were carefully recorded in two registers, so that if they tried to escape they could be distinguished and brought back. Also, forgery was prevented through comparison of the two registers, one of which stayed with the devshirme officer, while the second was sent to Istanbul with the sürücü (the officer who brought the boys to Istanbul). The Bursa decree warned the officers that during transport to Istanbul, the boys should be guarded closely, and that they should not camp at the same place twice nor accept any food from the locals.[30] These precautions attest to the generally involuntary nature of the procedure, and the difficulty of preventing escape.
Detailed information on the transportation routes of levied children from different regions is unfortunately lacking. One document, however, reveals that the levies that were taken from Mihaliç in the Bursa region were transported through the Dutlimanı port of Bandırma to İstanbul in 1567 (975).[31] We learn from the document that while the sürü was resting at the port, some boys escaped to the Marmara and Mirali Islands where Christian villages were the majority. The devshirme register that we are examining in this article also has the potential to provide us with information about runaways. In the 1603-4 register, the levies from the Bursa region were not dated fully, but the dates that are noted down indicate that those levied first were from the Kocaeli region. Then, it is probable that the officers moved to first Bursa, Mihaliç, and Manyas, and then to Biga. Finally, the levy-officers transported the children from Dutlimanı again, from where six children were also levied.
As mentioned above, there were two registers for each levy. It is understood that what we have is the combination of all the registers brought by the transportation officers to Istanbul, the register of the Chief of Janissaries (Yeniçeri Ağası). Four different groups of officers went to four different regions to levy, and each group’s scribe noted down their own levies. These registers were probably bound together after the groups of children were brought to Istanbul.
Questions such as when they arrived in the capital, whether there were examined all together as they arrived in the city, or whether this was an extended time period just like the levy in the field, remain a mystery. What we do know is that the sürü was brought to the capital, where the children were allowed to rest for two to three days. In order to ease the culture shock and language barrier, chosen Christian families hosted the children in Istanbul. According to a unique document - a decree that reiterates the regulation for 976 deported non-Muslims residing in 14 neighborhoods of Istanbul - the Christian residents of those particular neighborhoods were responsible for feeding the boys from the time they were brought to the capital to the time they were registered.[32]
After resting and spending some time with these Christian families, their first experience in Istanbul was a roll-call and check-up. They were stripped in the presence of the Chief of Janissaries, examined for bodily defects, and circumcised. We do not have enough information about when the circumcision actually took place - this could have been either at their original destination, or upon their arrival in the capital. But the most likely scenario is that it was done after the children arrived in Istanbul, as the regulation denotes. Perhaps they spent time healing after the operation near the same Christian families.
What is of particular interest in the devshirme register, is that during these rollcalls, the condition of the boys after arriving in the capital was noted down on the corners of each entry. One annotation already mentioned, is the expression of “suspected Jews” (şekine-i arz-ı yahudi). We know that if there was any suspicion that a Jewish child was present, then the entire batch was sent to the Arsenal. Some boys from the Avlonya and Bosnia batches in the register were noted as being şekine (suspicious). The other notes that we come across are ill (hasta), and dead (merhum). Yorgila, later Abdülrahmet, from the village Kelemur in the sub-district of Mihaliç was recorded as ill. Sinan, later Ali, from the village of Kebir in Mihaliç in Bursa, was recorded as dead. Another mark that was consistently used in the register is a letter “م “.It is not explained what it denotes, but is presumably an abbreviation for ‘present’ (mevcud). If this is the case, there were eight children’s names lacking this sign.
The next step in the levy process was the distribution of the children to different locations according to their abilities and looks. Those with the most potential were selected for special training at the Palace School (Enderun). This minority group was expected to become the administrators and governors of the Ottoman state. The rest were recorded as “acemi” boys and were hired out to Turkish villagers for a period of approximately three to eight years.[33] The regulation states that if the boy was conscripted from Rumeli, he would be sent to Anatolia, and vice versa. The reason for this was to place them in locations far from their villages in order to prevent them from fleeing.[34] This indicates that their participation was usually not voluntary. Many other sources from histories to travelogues verify the prevalence of this practice. Koçi Bey, an Ottoman historian, mentions vaguely that they were sold to “Turkistan” for two flori for four to five years.[35] Nicolay Nicolas, a traveller, also gives the duration of their placement with Turkish peasants as four years, and specifies the rural areas around Bursa and Karaman.[36] Evliya Çelebi says that boys were distributed to Turks for half an akçe, and a yearly amount of çuha. The outstanding (güzide) ones were placed in state workshops, and the rest were placed near the shoe-makers in Istanbul.[37] Hoca Saadettin Efendi notes that these children were given to those willing to take them, especially those who were state officials (devlet hizmetinde).[38]
One decree also shows us two cases where Andrea, the son of Davud from Livadya, was hired out to a mü’ezzin, and Berata from Avlonya, placed at a pasha’s farm.[39] Koçi Bey mentions a register that was kept to follow up on the children sent to rural areas, and says that call-backs would be made every four to five years according to this register. However, no such register has been located in the archives so far.[40] Also, the regulations mention that every year, the ‘acemi oğlanı kethüdası (chief of the novice boys) sent someone to the areas where these boys were placed in order to check up on them.[41]
Some entries selected from Bursa court records show that the children were followed up quite closely: an ‘acemi oğlanı placed near İsak bin Hamza for service in Karaman – a village of Bursa – was recorded as having died from the plague. Davud, who served near Emir Isa bin Mahmud in Çavuş village of Bursa; Hızır and İlyas, serving the grocer Mehmed bin Hızır; and Hüseyin, serving kethüda Hacı Halil, all died from the same plague in Bursa.[42] We have these records because when an ‘acemi boy died, the family took responsibility for him were obliged to record his death in the courts in order to be able to prove the cause of death to the authorities. It should be kept in mind that the families were being paid in return for this service. One striking point in all these stories is that Bursa emerged as a center both for levying the boys and training the newly conscripted ones from Rumelia. It would be interesting to see how this was coordinated, but we have yet to discover documents detailing this process.
Runaways were an ongoing problem even after the boys were placed in their obligatory service, including their time training as soldiers. An interesting report lists the names of 404 children, ‘acemi oğlans’, who went missing in 1626.[43] This is quite a high number of boys to suddenly disappear. Unfortunately, the document does not provide information on how the escape took place, nor the reasons. The list was probably prepared for the officers who were responsible for tracing the boys back to the capital. This list is not the only document reporting runaway boys. 72 boys who were placed as gardeners (bostancı oğlanı) in Edirne were reported to be either dead or runaway in 1567(975).[44] Another document states that 467 gardener boys also escaped from Edirne.[45]
Where did these runaways go? It seems that they tried to go back to their places of origins. Since the state meticulously tracked down runaways, we can witness some of their stories in the documents. For example, in a decree sent to the beg of Aksaray, it was stated that the levied Christian children who completed their service with Turkish villagers as agricultural laborers ran away to the villages of their origin and converted back to Christianity. The decree ordered that they should be recorded in the registers and then killed.[46] Those who could not make it back to their places of birth hid in other locations. A group of runaways was protected and hidden by the locals of Mir Ali Island and Marmara Island while the batches of children were being transported from the port of Dutlimanı in Bandırma in 1567.[47] Some people, particularly non-Muslim locals, seemed to resent the devshirme system, and protected the runaways from the state. The document proves that there was a certain solidarity in this region against the levies. The priests and the kethüda of the village were called for interrogation but did not even show up.
The policy toward runaways was not always to kill them, but to return them to the system - the same goes for those kidnapped.[48] The children who were kidnapped en route to Istanbul were levied from Limni and captured by the enemy. Sotiri, later Hızır, for example, was found in a ship (kadırga) that Rodos Beg and Kaya Beg seized from the enemy, and was placed among the state captives (miri esirs) to do penal servitude on ships. It was ordered that the child be sent to Istanbul.[49] The experience of captivity of the converted devshirme boys was not limited to forced labor on ships. Sometimes these devshirme boys were regarded as renegades and taken to the Inquisition court to be punished because of their conversion to Islam.
Bartolomé and Lucile Bennasars’ book on the Christians who were converted to Islam during the sixteenth and seventeenth centuries reveals 297 cases of slaves who were caught mainly from territories close to the Ottoman lands, who were then brought to the Inquisition mostly in Sicily and Venice.[50] These slaves were Slavs, Bosnians, Croatians, Bulgarians, Russians, Poles, Hungarians, Greeks, Malteses, Albanians, and Armenians. At their court defense, ten slaves argued that they were taken captive during battle - 29 at sea - and 135 of them argued that they were taken captive during raids by the Tatars or Turks. 41 of them did not provide any explanation for their conversion, 27 of them claimed the captivity through diverse ways and 11 accepted a voluntary conversion.
Among these captives, 43 men argued that they were taken as devshirmes by the Ottoman authorities. These 43 men were levied between 1561 and 1615, and were represented at the Inquisition courts between 1577 and 1637. Out of these, 36 of the devshirmes were taken to the tribunals of Sicily.[51] One of them was Hungarian, three Ragusan, two Armenian, and 37 Greek.[52] One of the Armenians declared that he was originally from Adana, and the other one, Jean Chiriaco, was from a small town Azow, where the river Don meets the Black Sea.[53] Jean Chiriaco and Alexandre Cunel declared that they were voluntarily given by their families to the janissary officers to become janissaries, whereas Dimitri from Sofia, and Jean-Georges, considered their levy as a bad luck.[54]
In their declarations at the Inquisition tribunals, one can see the tendency to declare that they were levied very young. Thus, they could not have resisted conversion, and they did whatever possible to remain a Christian. Of course one cannot disregard the fact that they were trying to prove their innocence in court. Still, the stories they reveal are extremely interesting. A twenty-five-year-old Hungarian Jean de Guaro (?) for example, explained to the Inquisition that he was taken as a devshirme at the age of five or six, brought to Istanbul, circumcised, and converted to Islam. Then he was sent on pilgrimage along with a janissary around the year 1600. He was then given to a janissary and an English convert who was arrested at Gandia.[55] Nicolas from Gura claimed that he was four when he was brought to the Palace in Istanbul where he was educated in the Muslim religion. One day during festival celebrations, he approached the Porte with two other Greek captives and became a palace gardener. Soon after, he embarked upon a journey towards Cairo with the hope of finding a French ship going towards Venice. He then came across the Tuscan galleys and escaped.
It is clear, then, that the reactions to the practice of child-levy varied: as we have seen, there were fortune seekers who wanted to become devshirmes, against whom the state had to warn the officers, there were also those who resisted giving the children away in every possible way. Among those who resisted, sometimes a collective action was developed against levies where community leaders negotiated with the levy-officers. When the negotiation attempts failed, occasionally desperate families rebelled against the state. More interesting is the attempts of the children themselves to avoid becoming devshirmes. As can be seen in the lists reporting child runaways, the children sought ways to return home even after they were integrated into the system - after they became soldiers, some used battles to desert.[56]
Among those who were integrated into the Ottoman system and promoted to the highest posts of the Ottoman government, some never forgot their roots. For example, Sokollu Mehmet Pasha renewed the Serbian Orthodox Church by declaring the restoration of the Peć Patriarchate during his third vizierate (1561-1565). Koçi Bey, in accordance with his will, was buried in his birthplace of Gümülcine (in the Thracian region of present-day Greece).[57] Two declarations in the Inquisition court records suggest that their Christian origins haunted some devshirmes. Marian Zalee from Negrepont and Georges of Michelis from Preveza, two Greek devshirmes, in the years of 1561 and 1567 respectively, were taken to the palace of the sultan at the same time. Marian practiced all the prayers as a Muslim and improved himself as a cook in the palace kitchens. He was a novice soldier among the ‘acemi oglans of the palace gardens. After nine years of his apprenticeship, at the age of eighteen, he became a janissary and was sent to fight against the Venetians in the battle of Cyprus in 1570, to Tunus where the Turkish army besieged the port of La Goulette in 1574, and to Persia. Interestingly, in 1589, at the age of thirty-seven, he presented himself to the commissariat of the Saint-Office of Messine and declared that he had never forgotten his Christian origins, and that he wanted to return to his birthplace.
Likewise, Georges of Michelis was also taken to the Palace of the Sultan, this time as an iç oğlan. He stayed in the Palace until he was thirty-two years old and worked as a gardener and a barber. He then became the Chef of Gardens (bostancı başı), who was responsible for all surveillance, espionage, and policing within Istanbul. When he retired from the position at the age of forty-eight, he sailed to Sicily, found the commissioner of the Inquisition court of Messina and declared that he wanted to confess the reality that he was a Christian in origin and want to cleanse his conscience.[58] Further similar stories will no doubt be uncovered when more historians start examining the issue from the perspective of the devshirme boys.
It is critical to remember that the devshirme system was not based on choice; children were sometimes selected involuntarily, yet sometimes parents encouraged their children to be enlisted as devshirmes. In either situation, what they would become or how they felt about being a devshirme was entirely personal, based on unique experiences upon integration into the Ottoman system.
3. A Closer Look at the Boys from Bursa according to the 1603-4 Register
Local Origins
Who were the children selected for the obligatory service of the Ottoman state? From where were these children chosen? 530 boys were levied as devshirmes from the Bursa region within six months. As can be seen in Table 1, 80 of them were from the center of Bursa or from the villages directly connected to the town. 236 of them were from the sub-district (kaza) of Mihaliç; 27 from Manyas, 24 from Enescik(?), and 20 from the villages of Yenişehir. From Biga, which was a separate liva in the region, 48 children were levied. Finally, from the liva of Kocaeli, 90 boys were taken. These 90 were gathered from the kazas of İznikmid, Karacabey, İznik, Lefke, Akhisar and Yalova.
The children were levied from 141 villages in total. It is hard to elaborate on whether the selection-ratio system was applied as stipulated in the regulations — levying one boy out of every forty households — because we do not have any information about the population of these villages. If we look at some examples, it appears that there was an even distribution of the levy-duty on villages in general. There were, however, cases where this rule of one-in-forty ratio was clearly not followed. In Biga, for example, the children were gathered from 38 different villages. From the villages of Kemer, Karabiga, and Hoca, and Gökpınar(?) the officers took four, three, three, and three children, respectively. From among the rest (35 villages), only one child was levied from each. The levy was evenly distributed in Biga. In some cases, however, a high number of children was levied from certain villages. For example, 37 boys were taken from Balıkesir, sixteen of them from the village of Anahor, another sixteen from Göbel, four from Eleksi, and only one from Hakirpınarı.
The village of Filedar in Mihaliç provided the highest number of children – 33 boys. Even if the one-in-forty household ratio was applied, which does not seem possible for Filedar, this would not be enough to prevent devastation of a village that gave thirty-three children at one time. Another explanation of this rather high number of levied boys from Filedar might lie in the fact that the village had experienced a recent plague outbreak. Some of the boys levied from Filedar were recorded to have bubonic plague marks (hıyarcık yaresi) on their faces. This suggests that there had been a recent outbreak in the village, and it is likely that many villagers had died in this outbreak - including adults and children – an event that might have left the town with a large number of orphans. Of course, according to the levy regulations, the levying of orphans was forbidden, but there may have been cases where the regulation was not followed, such as the case of Filedar.
As can be seen on the map, the levies took place mostly around the villages close to Bursa’s center. According to the names that were noted down in the register, the children were mostly Greek. Some names such as Karagöz and Alagöz, however, point to the possibility of Armenian levies. As mentioned before, Bursa was a center for both levying children for devshirme and for hiring out the levied boys to Turkish families. As the map shows, villages such as Hamamlıkızık, Karaman, Balıklı, Armud, Serme, Adaköy, Katırlı köy, and Akçapınar were where the boys did obligatory service after they became devshirmes and before they were called back to Istanbul. These are most probably not the only places that the boys were distributed, but the only ones that were found in the Bursa court records to date.[59] In addition to these, we know that these children were also placed with people in the center of Bursa. Interestingly, the locations where these two actions took place were not far from each other. In some cases, the villages that give out levies and hired devshirmes intersected, for example in Akçapınar, Balıklı, or Katırlı köy. How this was practically worked out is unknown.
The Physical Aspects and State of Health of the Boys
The state set a high criteria for the selection of the children. Physical competence was one criteria; the boys’ social and psychological states was another. According to the regulations, they should be unmarried, rural dwellers, with no artisan skills. The state was looking for candidates that could be easily assimilated into the system, submissive to authority, and who could be more easily trained. Strong social ties such as marriage, or skills that would give a boy economic independence, were a handicap. Boys who had traveled to Istanbul and had returned home were not wanted, for they would be too vigilant (“çok yüz görmüş ve bi haya olur”); orphans were also not accepted because they were believed to be greedy, and lacking a proper upbringing.[60] However, as we mentioned in the case of Filedar, the regulations may not have been followed to the letter in every case. Similarly, in the 1603-4 levy, there were several boys levied from the centers of the towns and were recorded as from the center (nefs).
The regulation on the selection of boys who would be levied highlighted important physical characteristics that the boys had to preserve. They should be able-bodied and good-looking. Tall (tavilü’l-kame) boys should not be taken since they would be goofy (ahmak), nor short ones (kasır), since they would be obstinate ( fitne).[61] Such an opinionated character analysis was not specific to the child-levy - the Ottomans had developed a method of character analysis by looking at the physical characteristics of people, the details of this which were written down in kıyafetnames.[62]
The examination of the Bursa levy in the register indicates that 67 percent of the boys were middle height (orta), and 27 percent were tall (uzun), and only 2 percent were recorded as small (küçük). The height of 4 percent of the boys was not determined because the register is damaged in certain parts. Prof. Hedda Reindl-Kiel’s examination of a sample of 601 boys from the same register reveals that 40 percent of the boys were tall, and 60 percent were of medium height.[63] There were no short ones and two were unidentified. These percentages indicate that the tendency was indeed to levy boys of middle-height. The tall ones, however, were also sometimes levied. It is possible to argue that the aim was to select the children that were physically capable of bearing the hard conditions of the training period and of being a soldier. The percentages from both samples reflect that short or small children were not preferred. The officers appointed for child-levying were most likely professionals with some recognized experience. The ability to categorize the height of these children as tall, medium, or short, appropriate to their age groups required experience in this area, as well as some familiarity with child development.
The description of the children was not primarily aimed at providing information for character analysis, but more for a kind of security check. These descriptions functioned almost like pre-modern photographs. They were used for identifying the children during controls. This was necessary in order to maintain accuracy, and to trace the boys before and after the levy. Therefore, very detailed information on the skin-color, eye color, and the color and shape of the eyebrows of the children was gathered as well. The categories used were: dark-skinned (kara yağız), brown (gendum gün), blonde (saruşın), and fair (ak bağırlu) for the skin color; for the eye color: dark (kara), hazel (ela), brown (koyun ela), blueish-hazel (gök ela), and blue (gök).
According to the register, most of the boys had brown hair, brown or hazel eyes, and black or auburn eyebrows. 18 percent were dark-skinned, 69 percent brownhaired, 10 percent blonde, and 2 percent fair (chart 2). Eye color was: 15 percent black, 41 percent brown, 31 percent hazel, 7 percent blueish-hazel, and only 5 percent blue (chart 3). Finally, the colors of eyebrows were: 53 percent dark, 38 percent auburn, and 8 percent blonde (chart 4). Eyebrows were also defined in terms of shape: 54 percent had open (açık), and 27 percent had frowning (çatık) brows. The rest were only indicated by color.
Another physical characteristic that was recorded was any kind of mark on their faces, heads, or hands. We read detailed descriptions of these marks in the register. Birthmarks, marks from injury or accident, or any kind of mark due to disease, were noted down. Again, differentiating these required a certain expertise on the officers’ part. The examination of the information on these marks provides valuable information on the history of diseases that the children most commonly had – at least those that left marks on the skin.
One of the diseases that badly affected the Bursa region during the early modern era was plague. Bubonic plague was an acute epidemic accompanied by high mortality. In Europe, there were recurrent waves of bubonic plague between 1348 and 1720. The disease was passed to humans by the bite of infected fleas, and within six days of infection, approximately 60 percent of the infected died.[64] Our devshirme register reveals evidence of plague outbreaks in Bursa at the end of the sixteenth century. In the 1604 child-levy, some children were recorded as having marks from plague, especially the bubonic plague. Six percent of the boys had marks as a result of the bubonic plague (hıyarcık yaresi) and 1.5 percent had marks of ta’un (plague). 33 boys recovered from bubonic plague in 1603-4, which suggests that many more died, since the disease had a very high mortality rate.
Another type of mark that was noted down in the register was from smallpox, an infectious inflammatory disease that left pockmarks on the face. The disease was defeated after the discovery of the smallpox vaccine by Edward Jenner (1749-1823). Prior to this, however, it was often fatal. Among the sultans of the Ottomans, Ahmed I (r. 1603-17), Ahmed III (r. 1703-30), and Abdülmecid (r. 1839- 1861) had smallpox.[65] Seven percent of the children levied from the Bursa region had this disease and recovered from it, bearing its marks on their faces (yüzünde çiçek alameti). Another common mark written on the registers was furuncle marks, or boils (çıban yaresi). This seems to have been a common skin disease in the seventeenth century. Ten percent of the children had boils on their faces. Ottoman surgeons were effective against this disease.[66].
All these disease markers show that health conditions were poor in early seventeenth-century Bursa. At least 25.5 percent of the children experienced one of the conditions mentioned above. Other than disease marks, scars were also noted down. Among 531 boys, 329 (62%) had scars on their faces or hands. Prof. Hedda Reindl-Kiel’s examination of a sample of 601 boys reveals a similar picture - 395 boys (66%) had scars. Reindl-Kiel attributes this to the fact that the officers chose boys that had a tendency to fight. The reasons for some of the injuries were indicated: seven had resulted from a reaping-hook (orak), six from a knife (bıçak), and nine from a stirrup (üzengi). Since the indicated injuries were so low in number in our sample, Prof. Reindl-Kiel’s observation makes sense. The high percentage of scars could point to the aggressive nature of the boys. These findings reveal something about the violence of early modern life in general.
As has been mentioned, the selection of these boys took six months to a year in the 1603-4 levy, and most of this time was spent on composing the pool from which the final levies would be made. In the final stage, however, these boys were gathered, probably at the center of the villages of towns, to be recorded and prepared for transportation. At this point, the officers investigated these boys to record them in the registers. All the physical characteristics mentioned above were examined. This must have been one of the most tragic moments in the children’s lives, when they felt the power of the state, here represented by the turnacıbaşı and his assistants, over their lives and their bodies.
The children were investigated very carefully at this stage. We understand this from the very detailed descriptions of their scars: round (değirmi), small (küçük), big (azim), long (uzun), unimportant (hurd), spread (cabeca) or several (evferi) scars. The exact location of the scars was also detailed: behind the right ear (sağ kulağının ardında), on top of the head inside the hair (başının depesinde saçının içinde), on the index finger of his left hand (sol elinin baş parmağında), or following his left eyebrow (sol kaşının kuyruğunda). Not only were the scars written down, but any beauty spots or moles were recorded. It was also the time that these children were converted and given Muslim names. This was probably the initial moment of proper enslavement for them. The register contains the names of the parents of the children, the original name of the children, and the Muslim names given to them after conversion.
The Ages of the Boys
The register also maintains information on the age of the levied boys. The distribution of ages is quite striking since the average age of the levied children was higher than originally assumed. Childhood according to Islamic law was considered to finish at the age of 15.[67] As it is understood from the Pençik Kanunnamesi (the regulation stipulating one-fifth of war captives taken by the state), the goal of the state was to levy children who had reached puberty. They could thus be defined as gulams.[68] In the regulations it was mentioned that boys should not have beards. In order to be considered gulam (a child who had reached puberty), a boy needed to age between 12 and 15 years. The data that we have from the conscriptions of the 1490s is found under the classification of müteferrik defterler (miscellaneous registers) in the Prime Ministry’s Archives, and reveals that the classical application of the system was in accordance with this definition of gulam. The children aged 12 to 15 were levied and the average age was 13.5 in the late fifteenth and early sixteenth century.[69]
In the 1603-4 register, however, this age group of gulam is extended up to 20 years. The children conscripted in the early seventeenth century were, on average, aged sixteen-and-a-half. 1,066 boys (85 percent) were aged 16-20. 42 percent of the boys in the register were 18 years old and above. Less than 1 percent of the boys were under 12 years old. The decree that was sent to Bursa by the janissary officers shows that this was no coincidence, as it ordered that boys between the ages of 15 and 20 should be conscripted.[70] The average age of the 530 boys levied from Bursa was almost 16. The most frequent age was 18 (20 percent); then ages 16 and 17 (16 percent each).
Why did the age of the levied boys increase? One reason for seeking older boys was perhaps to reduce the number of casualties during conscription. The high risk of becoming sick during the transfer to Istanbul, the extreme work conditions of ‘acemis’, and being more vulnerable to kidnapping might be additional reasons for the state to choose older boys during levies. Indeed, the levies from Bursa in 1603-4 show that there were only two children recorded as sick and another as dead. This might be related to the maturity and the strength of the boys. Of course the availability in a given area was also crucial.
The goal of levying boys without devastating the economy of the region could have resulted in the tendency of selecting the older boys who were unemployed and landless. The shift from timars to tax-farming in the early seventeenth century caused an increase in the population of landless young peasants in rural areas. The state might have hoped to place these new social groups within the devshirme system, while the system itself became more appealing to those young men, since it provided exemption from taxes and guaranteed an income.
Finally, the invention of new weaponry and warfare techniques at the end of the sixteenth century created a need for a new type of soldier. Firearms, which could be mastered after a short period of training, became the main weaponry used by the janissary armies. During this time, not only the weaponry changed, but the entire army organization, war technology, and the size of the army altered. The janissary army almost tripled in size during this period.[71] As has been recently revealed, however, other sections of the army such as sipahis, and the armies of powerful Ottoman grandees, also grew.[72] In order to be proficient at using firearms, a short period of training was more than enough. The janissaries who had been levied at the ages of 12 to 15 trained as professional warriors during their novice years, and started to become a burden to the state.
Conclusion
Our story began with the arrival in Bursa of a turnacıbaşı in the janissary army with a decree to levy Christian boys in October 1603. The entire process lasted six months and four batches of boys were selected from the livas of Bursa, Biga, and Kocaeli. All these groups were ready for transportation in January and February of 1604. We have seen specifically in our case study of Bursa, but also in other parts of the empire where boys were levied as devshirmes, that this was quite a complicated process, which generated its own local politics. This paper has attempted to understand the specifics of the process, how bureaucratized it was, and how it triggered the local political groups and dynamics in the selection process.
The experience of being a devshirme in the Ottoman Empire must be placed within a multi-layered context. One could climb high up the social ladder, or alternatively only just manage to survive in poor conditions. Equally, you could become a high-level bureaucrat in the Ottoman state, a soldier, a worker, or perhaps never get promoted from being an ‘acemi oğlan to be a janissary, while some simply died en route even before reaching their destination of Istanbul. And in similarly complex ways, the reactions of boys and families to being levied varied from kidnapping children back or fully grown adults returning to their homes, to others actually trying to slip into the groups selected by the officers for transport to the capital.
The aim of this paper was to investigate the personal stories of devshirme boys and their reactions to being ‘wanted’ by the Ottoman state. It was the stories of those who resisted that I find more compelling. To follow them along the path life had set them and to see how they coped with the seventeenth-century authorities and the obligations the state set on them was an exciting personal journey as an historian. More importantly, to develop a sense of what it might mean to be a Christian child in the early modern Ottoman world was crucial for this work -who these children were is critically important. Therefore, the paper also concentrated on where these children were selected, their age, looks, and health as registered in the documents. It was found that these boys were mostly Greek and Armenians and mostly had brown hair. They were strong and aggressive boys who were growing up subject to very poor health conditions.
This is the first time that such a register listing the devshirme boys has been studied, and it is probably the first time that we can access such detailed information on the boys and how they were levied. I hope that further similar research will be conducted so that we can better understand the impact of one of the most important institutions of the Ottoman Empire, the devshirme system, on the daily life of its subjects, and particularly on the children whose lives it transformed.