Sometime in 1815 or 1816, eight-year-old Joseph Marini entered the Bardo Palace to serve the beys ruling the Ottoman province of Tunis. His mother Madeleine had brought him and his siblings from French Corsica to North Africa but then died of plague. Separated from his sister and brother, Joseph converted to Islam in the Bardo palace, becoming the mamluk Salim al-Qursu (Salim the Corsican). The second part of his name, the nisba, indicates that he had not been severed entirely from his past. As an adult, he maintained connections with the French and other Europeans through his sister Thérèse, the mother-in-law of André Mattei, a vice-consul who represented different European states in the port city of Sfax.Footnote 1 A half century after entering the palace, Salim became the head mamluk in the service of the Ottoman governors of Tunis.
Salim's trajectory does not fully conform to the definition of the mamluk in contemporary historiography, which follows the seminal works of David Ayalon.Footnote 2 Unlike the quintessential mamluk, Salim was not brought from Central Asia or the Caucasus but found on the streets of Tunis. Before becoming the head mamluk of the palace, he was not trained to be a soldier or a state servant, but had only been a guard. His life fits more clearly with the kul type of slavery identified by Ottomanists such as Metin Kunt and Ehud Toledano.Footnote 3 Salim was part of a provincial household that replicated the structure of Ottoman imperial households. Like other men and women of his kind, he had to defend and expand the authority of a local dynasty and, to a certain extent, the legitimacy of an Ottoman imperial order.
Scholars have widely acknowledged the lasting nature of the “mamluk phenomenon,” which was in place for over a millennium (from the first Islamic caliphates to the 19th century) and spread from the Middle East to the Delhi sultanate and even to Aceh in Indonesia.Footnote 4 But in contrast to the scholarship on European captives, West African slaves in North Africa, and mamluks in Ottoman Egypt, there is comparatively little written on the mamluks of Ottoman Tunis.Footnote 5 This gap might relate to Ottoman Tunis's relatively small number of mamluks. Whereas more than 2,000 mamluks served in the most important households in Ottoman Cairo on the eve of the 19th century, just fifty to approximately 270 served the governors or beys of Ottoman Tunisia at any one time between the 1730s and 1826.Footnote 6 Yet understanding the processes of categorizing mamluks in Ottoman Tunis provides important insights into three independent but related issues: the relationships between Islamic states and societies, interactions between the Ottoman Empire and its provinces, and forms of military slavery around the globe. Moreover, exploring these processes of categorization allows us to move beyond the a priori definition of the mamluk prevailing in contemporary scholarship. In attempt to articulate a new understanding of the mamluk, this article shows how mamluks actively contributed to their own categorization. Uncovering how these state servants categorized themselves and were categorized by others allows us to see to what extent the mamluk group connected various sectors of state and society, from Ottoman imperial officers to provincial forces in Tunis.
Many historians have seen mamluks as a homogenous group of foreigners who converted to Islam and served an Islamic sovereign. The prevalence of the mamluk phenomenon is interpreted as evidence of a chasm between Muslim rulers and local subjects; if these rulers trusted the locals whom they ruled, the argument goes, they would not have needed to recruit foreigners as mamluks.Footnote 7 Some scholars have even concluded that the existence of mamluks explains the “inability of Muslim sultans to be credibly constrained,” which accounts for why the Muslim world “fell behind.”Footnote 8 In a similar vein, the mamluk phenomenon is said to reflect a lack of “legitimating resources”Footnote 9 or an “inherent dissonance between the political content” of Islamic law “and the vicissitudes of realpolitik.”Footnote 10 Some have drawn on this conclusion as an explanation for authoritarian inclinations or instability in the contemporary Islamic world.Footnote 11
A very different analytical framework results when we consider mamluks a socially heterogeneous group that interacted with other components of Muslim societies. Scholarship on the connections between slave soldiers and free civilians has revealed the mutual shaping of both elite groups within a single society.Footnote 12 Other studies have shown that mamluk groups included slaves as well as free Muslim natives, allowing us to revise and historicize conceptions of freedom and slavery within Muslim societies. Similarly, historians of the Ottoman Empire have discussed extensively whether the equivalent of the mamluk in an Ottoman context, the so-called kul, meant slave of the sultan or, more broadly, servant of the sultan.Footnote 13 Toledano argues that the kul system went from “an all-slave military-administrative establishment in early Ottoman history” to one including “an ever-increasing number of freeborn men.”Footnote 14 Debates about the heterogeneity of military groups and state slavery are now occurring outside Ottoman studies (e.g., in Moroccan studies).Footnote 15 In addition, historians of early modern and modern Islamic societies have demonstrated that in times of social and economic crisis, free subjects were kidnapped in the countryside to be enslaved.Footnote 16 As Toledano observes, “rather than a dichotomy between slave and free … [there is] a continuum of various degrees of bondage.”Footnote 17
Our understanding of the mamluk phenomenon is vital to a related debate within Islamic studies, namely, how to place the history of Muslim societies within local and imperial Ottoman contexts. After the Ottoman conquest of the Arab provinces in the 16th century, chroniclers, scribes, and various other social actors in these provinces continued to use the term “mamluk” to categorize different types of slaves, including slave soldiers and servants of governors. Scholars of Ottoman Egypt have fiercely debated this semantic continuity.Footnote 18 In recent debates, historians who find that the practice of keeping mamluks changed little over time insist on the strength and continuity of local histories, which framed prenational and, to some extent, national histories. Another group of historians finds that the social and political use of mamluks at an imperial level changed radically over time. These historians break from a teleological writing of history that finds the origins of a nation through evidence of local continuity. Emphasizing the strong influence of Ottoman practices and structures, they suggest that the mamluks were not “inherent” to Islamic history but rather related to the nature of households.Footnote 19 These households were “complex elite urban units” bringing together “the nuclear (or simple)” and the “extended (or joint) family” with the “clan, tribe, or any other kin-derived formation.”Footnote 20
The mamluk phenomenon also affects our understanding of comparative and global slavery. Historical understandings of the nature of royal and military slavery have often depended on how the role of mamluks has been interpreted. Seeing mamluks mainly as slave soldiers runs the risk of reifying their historical experience. This leads, in turn, to scholars drawing an overly sharp contrast between military slavery “which prevailed in much of Africa and the Middle East” and “less sophisticated schemes for the arming of slaves particularly common to the Americas.”Footnote 21 Such a contrast, which became the basis of so many comparative studies of global slavery, should perhaps be “turned on its head.” Christopher Leslie Brown, for instance, suggests that we should ask why “military slavery, which otherwise prevailed in much of the slaveholding world, fail[ed] to develop in the Americas.”Footnote 22 Historians might also take into account the heterogeneity among and within mamluk groups in the Islamic world.Footnote 23
In each of these debates in the history of Muslim societies and global slavery the issue of how historians categorize slaves and of how slaves are categorized in sources cannot be avoided. Who was acting in the process of labeling these people? Were slaves part of the process? And if so, what do their roles and actions tell us about the state they had to obey and the society to which they belonged? Y. Hakan Erdem notes that the absence of a meaningful abolitionist movement in the Ottoman Empire makes it particularly difficult to recover any kind of slave perspective on the meaning of slave categories.Footnote 24 Another problem, as many historians are aware from the North American case, is that slave narratives were often penned by others.Footnote 25 Consequently, historians often consider sources related to slavery as “mediated documents” that were not the “product of slaves’ own hands.”Footnote 26 Is there another path to piecing together the slave experience from administrative and legal documents? Drawing on evidence from written documents in which slaves appear, Toledano has called for “a viable and credible reconstruction of tentative realities … by resorting to the knowable and verifiable social and cultural context, [and] bridging gaps by carefully allowing a measured use of the educated imagination that, for the historian, brings to life people and communities long gone.”Footnote 27
This article uses a supplementary methodology to understand the categorization of mamluks and therefore to shed light on relations between state and society in Ottoman Tunis. Rather than reading the sources as fragments to be pieced together, I look at written administrative documents and genres (account books, administrative letters, and chronicles) as outcomes of social interactions reflecting various claims within the Tunisian governors’ households.Footnote 28 Considering the archive a subject of inquiry rather than only a source of content, I analyze how the categorization of mamluks changes from one type of historical source to another, and what these changes reveal about the relationships between a provincial society and its subjects.Footnote 29 This methodology discloses social practices that state control or state narratives (such as chronicles) either marginalized or eradicated.Footnote 30 The kinds of claims found in these sources confirm the agency of various social actors, including slaves; they also reveal how members of the Tunisian governors’ households interacted with each other. While relationships between these social actors were asymmetrical, with slaves often dominated by masters, one can still find “mutual dependence between two quite unequal partners.”Footnote 31 Moreover, these relationships involved not just masters and slaves, but also the scribes who had to put relationships into words. The documents I examine date from the beginning of the 18th century, when the dynasty of the Husaynid beys took power in Tunis, to the 1860s, when it became impossible to recruit new slaves in Tunis and when the style and format of administrative material changed dramatically.
In the first section, I contextualize this history by discussing the political and social production of the mamluk category from the late 16th century (when Tunis became an Ottoman province) to the end of the 17th century. I argue that the prevalent use of the word “mamluk” over the Ottoman term kul relates not merely to translation. In other words, the scribes working for local Tunisian authorities did not only choose the Arabic term mamluk because it was somehow a translation of the Ottoman Turkish kul. Rather, the use of the term mamluk reflects a broader process of Ottomanization and localization.Footnote 32 Various social groups, including Ottoman provincial dignitaries, local notables, and Italian converts to Islam, shared the local notion of mamluk, which served as common ground for labelling slaves and servants in households competing for provincial power.
The second section focuses on financial registers. The daily and biannual accounts of expenses of the Tunisian seraglio from the early 18th century to the 1830s reveal the role of various actors, including the mamluks themselves, in the production of the social category of mamluks as well as a discourse around social and ethnic categories. Within these documents, the mamluk category was used as an intermediate group between Turkish soldiers (at the top of the provincial administrative hierarchy) and Arab tribesmen (at the bottom of this hierarchy despite constituting the largest military force of the province from the 18th to the first half of the 19th century).
The final section examines administrative letters and chronicles, which convey additional ways of understanding and using the term “mamluk.” These two kinds of sources, while reflecting relations between the Tunisian administration and provincial society outside of the governors’ palaces, were less the result of social negotiations than the registers because they were often written by palatial scribes. Administrative letters show prominent mamluks formulating a common belonging to a mamluk corps and seeking to transform provincial subjects into the clients of the governors’ households and of themselves. Chronicles written between the beginning of the 18th century and the 1860s reveal a rich vocabulary of slavery but simultaneously tend to homogenize the historical experiences of mamluks.
A PROCESS OF LOCALIZATION AND OTTOMANIZATION
Before looking at each type of local administrative source in which the term mamluk appears, it is important to consider this term's context of production. Historians working on Ottoman Tunis, Algiers, and Tripoli have mainly preferred the term “mamluk” over the term kul because they typically rely on Arabic and European (rather than Ottoman Turkish) sources. The preference is also part of a broader tendency to portray the early modern history of the Maghrib as a local history of provinces that kept their distance from Istanbul.Footnote 33 However, the prevalence of the mamluk category in Maghribi historiography is not simply due to the mechanical reproduction of a local term. Local writers and chroniclers of the early modern period did not simply choose the term “mamluk” because they sought to claim continuity with a previous local political order or to translate Ottoman concepts and practices into Arabic. In fact, they did not shy away from using Ottoman terms in categorizing local officials. Terms such as agha or dey, which relate to military rank, frequently appear in administrative documents. The term kul was also used, as part of the word kūlūghlī (Ottoman Turkish, kuloǧlu), which referred to the children of Ottoman Janissaries and native women.Footnote 34 The scribes whose Arabic writings I analyze worked alongside colleagues (khūja; hoca) who penned official documents in Ottoman Turkish. The languages these secretaries used—each an imperial lingua franca that could encompass various categories originating from various discourses—significantly overlapped. But the Arabic-language scribes insistently used “mamluk” instead of kul.
The preference for the word “mamluk” in Tunis reflected a broader process of what Toledano terms “Ottomanization/localization.” The process of Ottomanization/localization refers to the encounter of two social groups—on the one hand, Ottoman soldiers, officers, and administrators, and on the other, local wealthy families and urban notables—that worked together within the Ottoman administration.Footnote 35 Rather than reflecting distance between these two groups, the predominance of the mamluk category reflects their closeness. Two crucial factors help explain why the term “mamluk” was considered appropriate. The first of these was the Ottomans’ perpetuation, after the conquest of Tunis in 1574, of local practices of administration and slavery as well as notions of political authority that had been established by the Hafsid sultans who ruled the eastern part of the Maghrib in the 14th and 15th centuries.Footnote 36 The Ottomans deployed the Arabic standard for administrative writings and adopted such local practices as empowering local chiefs, the maḥalla (mobile military camp), and the mamluks.Footnote 37 As was the case in other provinces, the Ottomans needed to rely on local notions and institutions to build legitimacy.
The second major factor in the preference for “mamluk” over kul was that “mamluk” was the commonly accepted term of a number of disparate local groups. In the late 16th and the first half of the 17th century, Tunis, like other major North African port cities such as Algiers and Tripoli, was a new frontier for the Ottomans and a space of conflict between Muslims and Christians (mainly from Spain and Italy). As such, it attracted dignitaries and soldiers from Istanbul, European adventurers, and slaves from the northern shores of the Mediterranean who had converted to Islam. Because each group brought its own conception of slavery and social dependence, all had to rely on the local category of the mamluk.
From the perspective of the fledgling provincial households that wanted to establish authority, the term “mamluk” served as a common ground for labelling slaves and servants.Footnote 38 In Tunis, as well as in Cairo and Algiers (as Jane Hathaway and Tal Shuval have demonstrated), military leaders often gathered followers and slaves from different origins together in military barracks and households.Footnote 39 These military leaders must have categorized some of their men as mamluks. In this context, even Janissaries coming from Istanbul may have used the label to refer to local mamluks as part of maintaining distance from them. Although the Janissaries and mamluks in Tunis had typically followed similar trajectories (before settling down in Tunis, many of them had come from Anatolia or passed through it), no single Janissary was enrolled among the mamluks.
Finally, it is important to note that the officers in charge of Ottoman Tunis clearly sought fluency in Arabic, which may help explain the preference for the Arabic term “mamluk.” In the first two decades of the 17th century, households competed to hire talented Arabic scribes who could connect them to natives and local literary life.Footnote 40 This was true not only of households whose masters were from Anatolia but also of those ruled by northern Mediterranean converts to Islam. With the emergence of the household of the Corsican Muradid Bey in the 1620s and the increasing hegemony of the household of the Husaynid beys, which was founded by the son of a Cretan janissary and ruled the Ottoman province of Tunis and Tunisia from 1705 to 1957, the use of Arabic in administrative documents grew. At the start of the 18th century, members of the Husaynid household began preserving registers, letters, and chronicles, which is why today we have many of these sources but not their equivalents from 17th-century households. In this provincial context, the prevalence of the category “mamluk” in administrative sources shows that the “mamluks” did not separate state from society. Various social groups—from provincial dignitaries to local notables to Italian converts to Islam—shared the notion of mamluk when attempting to establish provincial authority or to build households that competed for power.
ADAPTING AND REPEATING A MAMLUK CATEGORY IN REGISTERS
Registers are the most useful type of archives for understanding social interactions around the mamluk category. The Tunisian National Archives hold 1,253 registers covering the period from the emergence of the Husaynid Dynasty in 1705 to the implementation of the last Ottoman military, administrative, and legal reforms in Tunis in the late 1850s. There are an additional 2,816 registers spanning the period from the 1860s to the implementation of the French colonial order in the late 1880s.Footnote 41 An impressive amount of detail on mamluks and other groups of servants can be found in the account books, which list daily expenses and incomes for the palaces of the beys. The account books showcase the extent to which mamluks wrote and shaped their own category, but they also show that there were others involved in this process. Mamluks had to deal with economic constraints and ethnic categorizations, and they had to interact with other groups of servants who expressed a sense of belonging to both the Ottoman Empire and local society.
The account books contain evidence for two types of expenses in the Tunisian seraglio: daily and biannual. On a daily basis mamluks were deeply involved in handling their own and others’ financial affairs. They reported expenses in colloquial Tunisian Arabic, which the mamluks had to master.Footnote 42 The appearance of the common Arabic expression ʿalā yād al-mamlūk (on the hand of the mamluk) in the account books is indicative of the key role mamluks played in administering expenses as well. A prominent mamluk often acted as the khaznadār (haznedār), the officer in charge of the state treasury and the palace accounts. Similarly, the registers reflect the power of established mamluks at the time of the enrollment and formal conversion of new mamluks: the bāsh mamlūk, or the mamluk who was in charge of his peers, oversaw and was very close to the newcomers. In these specific moments, these servants also “shaped their future at least partially by their answers when they were interrogated,” which were preserved by the scribes who wrote “down the records.”Footnote 43
In addition to appearing daily in the seraglio, the mamluks were listed twice per year in the biannual registers of the maḥalla’s expenses. In winter the soldiers traveled to the Sahel, the center, and the south to ensure the local population's obedience to the beys. In summer the same soldiers were directed to the west in order to secure northern areas neighboring the Ottoman province of Algiers.Footnote 44 The maḥalla’s expenses suggest that the mamluks had a limited role in these assignments.
These expenses also suggest that, like other servants in the palaces of the beys, mamluks were not recorded as individuals but rather formed a single category, albeit one divided into subroups. Between the early 17th century and the 1830s, the subgroups used in the registers changed. Registers from the 1730s indicate that the eldest mamluks served the bey Husayn ibn ʿAli while other mamluks served his eldest son, Muhammad Bey.Footnote 45 In the 1750s, under Ottoman influence, the bey ʿAli Basha started to distinguish between mamālīk al-kibār, the eldest mamluks, and mamālīk al-bayt (or mamālīk al-ṣarāya, Ottoman Turkish sarāy), the closest servants to the beys within the seraglio.Footnote 46 In 1766, according to register 85 held in the Tunisian archives, mamālīk al-kibār were renamed mamālīk bi-l-sqīfa, the mamluks of the vestibule, a shift that became prevalent in the first half of the 19th century.Footnote 47 Mamluks were thus increasingly categorized according to their roles and locations within Bardo Palace. The seraglio's mamluks were clearly part of the family life of the beys, whereas the mamluks of the vestibule (who were subdivided into four subsections that were interchangeably ʿassa, guards, or dār, house) lived between the palace and the outside world.Footnote 48 The vestibule contained a hammam and an isolation cell. In this area, the mamluks of the vestibule were treated by dedicated physicians.Footnote 49 While maintaining local expressions and mirroring specific roles of the mamluks in the palaces of the beys, this categorization clearly echoed the distinction made in Istanbul between enderūn and bīrūn, the internal and external parts of palaces.Footnote 50 Mamluks who were in charge of their peers, such as the Qabran family, which by the end of the 18th century came to head the mamālīk bi-l-sqīfa, contributed further to this distinction.Footnote 51
Economic constraints as well as cultural and racial thinking influenced hierarchies. From the first half of the 18th to the mid-19th century, the seraglio's mamluks were of Georgian, Circassian, and finally Greek origin, while the mamluks of the vestibule were mainly local Muslims and Italian slaves. This ethnic hierarchy mirrored economic conditions, whether the cost of each type of slave and servant or transformations in the slave trade. Caucasian slaves purchased in Anatolia were the most expensive slaves brought to Tunis and elsewhere in the Arab provinces; they were promoted to high military and administrative positions (generals, viziers, etc.).Footnote 52 As the case of Joseph Marini mentioned at the start of this article suggests, European Christians who converted to Islam, including free men such as Marini, also became mamluks. These Europeans were inexpensive in the 18th century, but they became less easy to purchase and bring to Tunis after the Congress of Vienna in 1815 and the formation of a European military coalition to fight the last North African privateers. The local Muslim society of Tunis also drew mamluks from cities such as Jendouba and powerful tribes such as the Jlass, but these natives were free Muslims who volunteered to serve in the palaces of the beys.Footnote 53
The ethnic divide and price range were deeply connected to a cultural and racial way of thinking that allowed the beys as well as their prominent mamluks and servants to shape cultural discourses. Throughout the 18th and the first three decades of the 19th century, Georgians and then Circassians acted to promote their own ethnic interests and to obtain the best positions in the Ottoman provincial administration of Tunis. They sometimes used their own languages within the palace. These mamluks were either born Muslims or became Muslims before being enrolled in the bey's seraglios. According to local servants of the beys in mid-19th century Tunis, these mamluks claimed that they had converted before their Greek counterparts did, which made them better Muslims and thus better servants for the beys.Footnote 54
At another level, this ethnic hierarchy among the mamluks was a way for these men and statesmen (including their masters, the beys) to make claims of belonging to local Tunisian society, the Ottoman Empire, and the western Mediterranean. By promoting Caucasian slaves as favorite allies, the Tunisian administration again replicated an Ottoman pattern and in so doing articulated an adherence to Ottoman political culture. By recruiting European converts to Islam, the Tunisians could better negotiate with European consuls and mediators: in a context of diplomatic tensions with the Sardinians in 1833, for example, the bey Husayn played upon rivalries between Italians by sending a mamluk originating from Genoa.Footnote 55 Finally, by relying upon Muslim native subjects to join the mamluks, though at lower positions, the Tunisian governors maintained good relations with the local population, transforming subjects into servants involved in the local administration (makhzan).
Even if mamluks were involved in the writing and shaping of their own category, they could not completely control the hierarchies embedded within it. Moreover, they had to find a place among other groups of servants listed in the account books. In the biannual expenses for the organization of the maḥallas, mamluks were set apart and their specific category was listed in an order suggestive of cultural and social hierarchy. The first group mentioned was usually Turkish horsemen (sipahi and ḥawānib, referred to as al-atrāk, the Turks), followed by Arab horsemen and escorts (ḥawānib), mamluks, other officials (such as Islamic judges and scribes), cooks of the beys, and finally the tribesmen who served the state (people of the makhzan tribes).Footnote 56 Turkish soldiers and makhzan tribes were more numerous than mamluks, but as Madeline C. Zilfi argues, the importance of slavery in Ottoman society was “less a matter of quantity than of social resonance,” which meant that “slaveholding was living proof of societal hierarchy.”Footnote 57
The location of mamluks between Turkish soldiers and native makhzan tribes, an order repeated in the parade of the maḥalla as well as in listings in account books,Footnote 58 mirrored the internal mamluk order where Caucasians were promoted to the top ranks and natives were in the lower positions. Here again the Tunisian dynasty of governors signaled its acceptance of an Ottoman order while incorporating native tribes into the government of the Tunisian province. And while the tribesmen's listing in the accounts and location in the parade articulated their subaltern position, they were the largest servant group, receiving two-third of the wages (paid biannually) distributed to all the groups that participated in the maḥalla from the 1750s to the late 1830s.Footnote 59
The intermediary position of the mamluks between Turks and local tribesmen is suggestive of the social order envisioned and enacted within the palaces of the beys. In this social order, mamluks were similar to kul officials who, according to Ze'evi, “had the advantageous position of being equally distant and segregated from all groups, yet situated at their approximate convergence point; an elite ideally poised to govern such a diversified empire.”Footnote 60 Within these parameters, the mamluks of the beys were a transitional and thus central element of a social order that brought together “Turks” and natives, Janissaries and tribesmen. The mamluk category did not divide state and society, as various historical interpretations suggest. On the contrary, mamluks were an intermediate social force between the imperial and local officers of a provincial authority.
A comparison of the daily and biannual listings of the mamluks suggests that the mamluk category came into formation in two overlapping stages. The daily listings reflected various adaptations to the arrival of new men enrolled as mamluks, while the biannual listings for the maḥalla were crucial moments in which new mamluks could find a place in the category and among other servant groups. While mamluks clearly participated in the shaping of the daily listings, they had to interact more actively with other servants and followers of the beys in the production of the biannual censuses. Moreover, while daily listings were expressed according to the expenses that needed to be reported, the maḥalla was ordered through repetitions of the same categories and formulae.
More broadly, the registers clearly confirm that the processes that contributed to the creation of the mamluk corps within the households of the Tunisian beys involved the government, local society, Istanbul, and various segments of Mediterranean societies. As a result, and as was the case in Cairo, the mamluks populating households of the beys originated from different places.Footnote 61 Similar to other areas of the Ottoman Empire, these men found their place through different paths: captivity, kidnapping, being purchased as slaves, kinship with people in places of power, patronage in exchange for service, and the creation of kinship ties through adoption.Footnote 62 The social stratification of mamluks is yet further evidence of their heterogeneity. Some were integrated into other services of the beys’ palaces and administration. In the daily and biannual expenses, they could be cooks in the palace kitchens, craftsmen, carpenters, or even physicians.Footnote 63 They might escort the beys’ guests in the countryside, command military troops, or serve as principal advisers (viziers) of the Tunisian governors.Footnote 64 Registers convey an ethnic and social diversity that defies any simplistic definition of mamluks.
CIRCULATION OF A WORD IN LETTERS AND CHRONICLES
Administrative letters and chronicles, the other kinds of Arabic-language sources that referred to mamluks, reveal other ways of understanding and using the term “mamluk.” These sources were less the outcome of broad social interactions than registers because fewer actors were involved in their production. Administrative letters and chronicles were mainly written by scribes who worked for either prominent mamluks or the beys. The trajectories of scribes were different than those of mamluks. Scribes did not usually have to find their way to Tunis because they were already rooted in local society as members of the local administration. As in Istanbul, Tunisian scribes were often the offspring of dynasties of secretaries. When replicating categories and social orders in administrative writings, Tunisian scribes not only cultivated formalism and repressed individual initiative but also adapted the richness of social reality to the needs of a social and political provincial order.Footnote 65
It was the scribes who transported the term “mamluk” from the registers to other sources. While they were involved in the production of the registers, scribes conveyed other meanings of the mamluk category in their administrative letters and chronicles. The administrative letters in particular were deeply related to the registers even as the two types of documentation were kept in different collections of the Tunisian archives due to their divergent formats. Letters and registers contained similar standards, formulae, and content, all related to administrative business. They also contained public and private information, including fresh news concerning the beys’ relatives and households.
Whereas a single register was a collective document produced by a number of individuals from various social groups, the writing of administrative letters involved only two individuals: a scribe and a prominent mamluk. The latter, who had neither the time nor the skill for such administrative exercises, would sign the letter. The administrative letters in the Tunisian archives thus offer a glimpse of the social world that mamluk dignitaries and their scribes experienced. In their correspondence mamluk dignitaries confirmed their own social status as slaves and servants by expressing their loyalty to their masters, the governors of Tunis, and by articulating their identity as mamluks. In one example from 1832, the Greek mamluk Muhammad Khaznadar defined his situation in colloquial Tunisian in a letter to his Circassian master Shakir Sahib al-Tabiʿ, who by then was the vizier or main adviser to the bey: “I am your mamluk … I have neither mother nor father. You are my mother. You are my father. I have only you, my lord.”Footnote 66 In another letter, sent from France to Mustafa Khaznadar in 1839, Farhat al-Mamluk expressed his satisfaction at the good health of “our brothers the mamluks.”Footnote 67
There is a broader sense in which the letters complement the registers. The registers were written and kept inside the palaces or the houses of scribes, whereas the letters reported on events outside the palaces, in various social scenes and spaces in the province and beyond. In the letters, scribes reported on mamluks acting as leaders of maḥallas or governors of local tribes, towns, or rural areas. Until the 1850s, scribes closely followed mamluks in their travels. In reports of political missions to Istanbul from 1780 to 1850, scribes mentioned them as crucial political actors and reported on how they presented themselves in front of other interlocutors. Letters often described prominent mamluks as brokers between the palaces of the beys and local or European subjects. The dignitaries sought to transform provincial subjects into clients of the governors’ households and their own clients, and in so doing they spread the master/client relationship and the mamluk category that they experienced in the palaces of the beys. For example, in the mid-19th century, the mamluk Ismaʿil Sahib al-Tabiʿ, who supervised the grain silos of Tunis, received dozens of letters from individuals seeking financial help. These individuals claimed to be the faithful servants of Sahib al-Tabiʿ.Footnote 68 In sum, the registers and letters constitute two different narratives about the relationship between state and society, and about local, imperial, and global senses of belonging. What mattered when writing the registers was time: the category of mamluk was adapted every day and twice a year according to an administrative hierarchy. What mattered when writing letters was space: the same mamluk category and the authority of masters were displayed, confirmed, and expanded in different contexts and social fields.
The scribes working for the households of the beys also wrote the major chronicles on the history of Ottoman Tunis.Footnote 69 Unlike the Mamluk sultanate of Egypt, where the awlād al-nās, or offspring of mamluks, authored such historical narratives, in Tunis mamluks and their progeny had no such responsibility.Footnote 70 Similar to other Islamic governments, the beys commissioned these narratives from their scribes,Footnote 71 who wrote them for an audience within the palace as well as erudite scholars in the capital.Footnote 72 The few scribes who authored chronicles were clearly aware of their enormous power to narrate the history of Tunis. In the second half of the 18th century, the scribe Hammuda ibn ʿAbd al-ʿAziz used his chronicle to settle a score with his main enemy, the Georgian mamluk Mustafa Khuja, who advised the same bey, Hammuda Basha.Footnote 73 In the first half of the 19th century, another secretary, Ibn Abi al-Diyaf, reflected on his own conception of history and political authority in his chronicle. Unlike registers and letters, chronicles did not shape and reshape the meaning of the mamluk category every day or every six months, within or outside the palace. The chroniclers who wrote them were framing broad narratives in a longue durée to insert the dynasty they served into a broader Islamic history.
Similar to terms such as köle, cariye, and waṣīf (pl. wuṣfān), all of which grew in importance in the Ottoman Empire and in the administrative documents of Alawi Morocco, the mamluk category was central in registers and letters.Footnote 74 In the chronicles, two very different interventions into the mamluk category can be discerned. On the one hand, scribes took into account the diversity of slavery in Ottoman Tunis. On the other hand, when including the beys’ mamluks in their narratives, scribes tended to focus on mamluk elites, which could homogenize and reify the category of the beys’ mamluks.Footnote 75
Reflecting the richness of the slavery lexicon during this period, chroniclers used “mamluk” in relation to many other terms that described slavery as a legal status and a relation of social dependence. If a slave was said to be an ʿilj, he had converted to Islam.Footnote 76 A ghulām was a young mamluk trained in the entourage of the Tunisian governor.Footnote 77 The mawālī (sing. mawlā), in contrast to the ghulām, were well advanced in the state cursus honorum; they were manumitted slaves and interacted with their masters through a relationship of legal patronage (walāʾ).Footnote 78 Moreover, chroniclers applied the mamluk category to various groups of slaves or servants, such as Christian captives and West African slaves. For instance, both African slaves and free men were labeled as mamluks when they were forcibly enrolled in the new regular army (niẓām al-jadīd; nizām-ı cedīd) in 1836–37.Footnote 79 Chroniclers also reported on the speeches of Muslim free men who described themselves as mamluks. By comparing themselves to slaves and, implicitly, to a slave of God and his Will, these free-men sought “to indicate how humble” they were.Footnote 80 Under the reign of the governor Husayn ibn ʿAli Bey in the first three decades of the 18th century, a Muslim scholar, al-shaykh Ibn ʿAshur, reportedly claimed, “I do not own any mamluks because I am a mamluk.”Footnote 81
At the same time, chroniclers cast mamluks as being close to the beys or, more often, emphasized the political roles of prominent mamluks who had come from Greece, Circassia, or Georgia. References to the beys’ mamluks rarely included lower-status mamluks, such as Tunisian natives, those originating from Italy, or even the sons and grandsons of mamluks who were sometimes listed in the registers. In their attempts to inscribe a Tunisian local narrative within a broader Islamic history, the chroniclers assimilated the Tunisian authorities into ancient Islamic practices and government institutions. As such, in their portraits of the beys’ entourages, chroniclers made reference to their own knowledge of the past, quoting excerpts from the Qurʾan, Ibn Khaldun, or Nizam al-Mulk.Footnote 82 In this sense, compared to registers and administrative letters, chronicles—which many modern historians consider the best source for gaining a broad understanding of the historical situation of Ottoman provinces—can be misleading insofar as they erased various levels of agencies and social interactions that were at play between state and society in the shaping and reshaping of the mamluk category.
CONCLUSION
Seeing administrative primary sources as the products of social interactions is fruitful in many ways. By focusing on records of actions, claims, and self-representations, this article reveals the heterogeneous voices that contributed to the shaping of a slave category and different levels of agency among mamluks. Registers were sources that documented the daily lives of most mamluks, translating their daily social integration in the palace into words and conveying their biannual reordering into groups and subgroups of servants. The letters mostly reflected the discourse of mamluk elites, their scribes, and their clients outside the beys’ palaces in different spaces and social milieus. Chronicles gave a broad overview of the political situation of Ottoman Tunis (making them the easiest documents for historians to access). However, privileging chronicles in formulating our understanding of the mamluk phenomenon is problematic. While these narratives can display different meanings of slavery and social dependence, they can also reproduce a homogenized concept of mamluks that obscures the heterogeneous nature of the category. The scribes, mostly originating from Muslim families, appeared as crucial mediators who enabled the circulation of a category and its reshaping in different media. Whereas scholarship on Ottoman officialdom has provided a sense of the skills, works, and cursus of scribes, this study, which approaches archives as the products of social interactions more than as containers of content, facts, and data, reveals creative practices of writing involving more than scribes.
Moreover, seizing the many uses of the term “mamluk” at the provincial level confirms the extent to which provincial concepts were not simply tools of localization but notions through which a province such as Tunis was Ottomanized. With this point in mind, the mamluk category, at the intersection of various attempts to establish households in the 17th century, enabled the Ottomanization of the beys’ palaces during the following century. The beys’ mamluks were divided into two subgroups (the mamluks of the seraglio and those of the vestibule), a divide that replicated the structures of the main households in the imperial center and ultimately of the Ottoman imperial household.
In terms of a global approach to slavery, the study of the state and military slave phenomenon is enriched by an approach that, rather than assuming an a priori definition of slavery, explores how slaves and servants contributed to their own definition. This method deepens our knowledge of the circulation of such experiences within and beyond the Islamic world. If we conceive of the mamluk experience as a series of complex processes revealing not only a deep interaction between state and society, but also, and more importantly, “a continuum of various degrees of bondage rather than a dichotomy between slave and free,”Footnote 83 then the sharp distinctions drawn between societies that used military slaves and societies that relied mainly on mercenaries (or temporarily hired armed slaves) must be reconsidered.Footnote 84 Such an approach also allows us to consider the end of the mamluk phenomenon and of kul-type slavery in the second half of the 19th century as the outcome of global changes.
In Tunis, the categorization of mamluks changed dramatically after the 1850s for two related reasons: Ottoman state reforms and the end of the slave trade (which, in the case of the mamluks of Tunis, centered largely on Georgians and Circassians).Footnote 85 The suppression of slavery and the increasing difficulty of purchasing slaves on the northern shores of the Mediterranean and in the Caucasus forced the Tunisian beys to reconstruct their households. As Toledano notes, when slave labor became unavailable, the number of free servants increased.Footnote 86 When prominent Greek, Georgian, and Circassian mamluks got old, the beys of Tunis began to promote native servants among the sons and grandsons of mamluk dignitaries and among Muslim freemen born in local families. This process was reinforced by the implementation of Ottoman state reforms. As in other provinces, the beys of Tunis needed to enroll native subjects in the new regular army alongside the Janissaries, members of makhzan tribes, and mamluks. They also had to train their officers in European military knowledge, sometimes by sending them to European capitals. These changes required the beys' mamluks to adopt new forms of literacy and therefore new ways of perceiving themselves. The Circassian mamluk and former Tunisian general and minister Husayn, for instance, corresponded with the American consul in Tunis about the issue of slavery in the context of the American Civil War.Footnote 87 At the start of the 1860s, he was also in charge of the first official Tunisian journal, al-Raʾid al-Tunisi (The Tunisian Guide), which not only depicted the mamluks of the Abbasid caliphate but informed its readers of the fate of Husayn's homeland and people, Circassia and the Circassians, then struggling against Russian military conquest.Footnote 88
In addition to extending their scholarly knowledge, some of the beys’ mamluks drew connections to historical mamluk figures. They even started to write personal narratives. A mamluk such as Husayn was aware of the slavery issue in both the United States and the Mediterranean region. He could have been among those slaves who, according to J. Amelang, connected the “two autobiographical worlds of Mediterranean and transatlantic slavery.”Footnote 89 Facing these dramatic changes, Joseph Marini, alias Salim Qursu, the Corsican Muslim referred to in the introduction of this article, was not looking for mamluk counterparts. By the beginning of the 1860s, Salim felt lost and marginalized within the beys’ palaces. Sent to Alexandria, he ended up in Malta, where he sought asylum from the French consul. But the consul denied Salim's right to legal protection even though he was the son of a French Corsican.Footnote 90 Salim then had only one option: return to Tunis. There he became one of the last heads of the beys’ mamluks.
