Slavery and Decline of Slave-Ownership in Ottoman Bursa 1460–1880

Hülya Canbakal; Alpay Filiztekin

doi:10.1017/S0147547920000071

Slavery and Decline of Slave-Ownership in Ottoman Bursa 1460–1880

Published online by Cambridge University Press: 23 June 2020

Hülya Canbakal and

Alpay Filiztekin

Show author details

Hülya Canbakal: Affiliation:
Sabanci University
Alpay Filiztekin: Affiliation:
Sabanci University

Article contents

Abstract
The setting and the sources
Slave-ownership
Slave prices
Conclusion
Footnotes
References

Rights & Permissions

Abstract

The most widely accepted narrative about the long-term history of slavery in Ottoman lands rests on a supply-side story. According to this, military and diplomatic factors reduced the inflow of slaves from the seventeenth century onwards and, consequently, exorbitant prices turned slaveholding into a luxury inaccessible to all but the top elite. Using evidence from probate inventories of the city of Bursa and its hinterland from 1460 to 1880, the present study examines this narrative in light of the incidence of slave-ownership and prices. We observe substantial decline in slaveholding already before the beginning of the government reforms concerning slavery and slave trade in the nineteenth century. We also find a decline in slave prices, both absolute and relative to wages. This is unexpected. Further analysis suggests, on the one hand, that a different supply factor, relative increase in the African slave population due to changes in the global traffic may have been instrumental in these trends, which links Bursa's non-colonial market to world slavery. On the other hand, examination of the consumption/investment preferences of the wealthy suggests that demand for slaves, too, may have declined, we surmise, in response to demographic and social change affecting alternative labor costs as well as cultural change affecting the meaning of slaveholding.

Information

Type: Changing Labor Relations in the Ottoman Empire and Turkey
Information: International Labor and Working-Class History , Volume 97: Changing Labor Relations in the Ottoman Empire and Turkey / Oral History and Indian Labor History , Spring 2020 , pp. 57 - 80

DOI: https://doi.org/10.1017/S0147547920000071 [Opens in a new window]
Copyright: Copyright © International Labor and Working-Class History, Inc., 2020

The study of slavery in the Ottoman Empire has so far followed primarily three courses. One of these involves slaveholding by sultans, the royal household, and the official elite, and emphasizes institutional aspects of the practice, military slavery, household-building, and the harem.Footnote ² Another line of research focuses on the nineteenth century and examines both the last surge and the abolition of slavery in Ottoman lands.Footnote ³ A third and more recent group of works provides a bottom-up perspective exploring the culture of the slaves, micro histories, and the experience of slavery beyond the elite households.Footnote ⁴ Often implicit in the three clusters of research is the assumption of a backbone narrative about the long-term demise of slavery, which rests on a supply-side story, i.e., the number of slaves available. The first key element in this story is wars, successful in the early centuries and yielding ample numbers of slaves. The turn of the tide in the battle field against the Ottomans from the seventeenth century onward reduced slave inflows considerably. Occasional successes in later wars brought in new captives, but with the beginning of a new war diplomacy involving the exchange of prisoners in the eighteenth century, starting with Russo-Ottoman relations and feeding into a new international legal regime, trade became the main source of supply.Footnote ⁵ That, too, was hampered, however, when Russia expanded and came to control the northern routes. The consequent decline in slave supply caused an increase in prices, turning slaveholding into a luxury that was less and less affordable by or profitable for those outside the top economic elite and, geographically, outside the capital or large provincial centers.Footnote ⁶ The trend is thought to have been reversed in the nineteenth century due to a number of factors, including the Greek Revolution (1821–1830), which generated large numbers of captives, the Circassian immigration, involving some 150,000 slaves brought along with their masters in the 1860s, and the possibility of better access to areas of supply in Africa, as the global traffic shifted from the Atlantic to the East. This shift was also matched by rising demand for labor in Egypt and Western Anatolia (as in Iran and further east in central Asia) in response to commercial agriculture destined for global markets, especially during the American Civil War. Ottoman government's policy changes toward equal subjecthood inclusive of the legal status of slaves starting in 1840, and attempts to curtail slave trade in line with international pressures in 1847, were staged against this backdrop and took several decades to get results.Footnote ⁷ Thus, the narrative ends where European abolitionism takes up the center stage as the main agent of change imposed on a reluctant Ottoman elite.

Causal assumptions underlying this long term narrative have not been questioned so far because studies on quantifiable aspects of slavery, such as prices, the extent or intensity of slaveholding, size and characteristics of the slave population, and different types of employment, are limited to detached snapshots or short-term surveys of a single region, with the exception of Austen's studies on trade. Also limited are studies on the link between slavery and labor history, the main reason being that slavery in Ottoman lands, as in Islamicate lands more generally, was characterized primarily by urban domestic (female) servitude and, secondarily, by military/administrative servitude—which declined in importance in the central lands from the seventeenth century onward.Footnote ⁸ In terms of sectoral concentration, in other words, Ottoman slavery was akin to contemporary Eurasian slaveries and different from Atlantic slavery and its Iberian precursors—except, briefly, in the nineteenth century in Egypt and parts of Anatolia.

From the perspective of history writing, prominence of household slavery, military/administrative, or domestic puts the spotlight on integrative processes that assimilated the slave into the master's domain and society at large and can obfuscate subjugation, labor relations, and the economics of slavery. Although new research promises to highlight the labor process anew,Footnote ⁹ as in Mediterranean historiography,Footnote ¹⁰ and add novel instances to the better known cases of agricultural state slaves (ortakçı kullar) and slave weavers and commercial agents of the early period, the view that non-domestic and non-military/administrative slavery was of limited and localized scale, and became rarer from the late sixteenth century onward stands.Footnote ¹¹ We do not disagree with this view but propose a revision. First, ongoing research suggests that regional diversity in patterns of slave employment may turn out to be considerable, and more slaves can be “discovered” in the countryside if we stop looking for large estates as markers of agricultural slavery.Footnote ¹² Second, “domestic servitude” is a highly problematic characterization because it conceals as much it reveals. On the one hand, the line between in-house chores and marketable production of various scale in and around the early-modern house, such as gardening, husbandry, dairying, brewing, and a multitude of cottage crafts, must have been often flimsy.Footnote ¹³ On the other hand, even if it was possible to distinguish between market and non-market activities, the economic importance of home services and production, i.e., of goods for household consumption, has been acknowledged as an important element in value creation for a long time.Footnote ¹⁴ In other words, domestic slaves, especially handmaids, but also concubines, to the extent that the two were separable, were laborers. Therefore, to posit an economic distinction between Middle Eastern slavery as luxury or conspicuous consumption and, for example, Atlantic slavery as capital investmentFootnote ¹⁵ is, at best, misleading. Ottoman non-military slaves performed all that Antebellum field slaves did and vice versa—though functions were fluid and their primacy different.Footnote ¹⁶

In this study, we seek to contribute toward filling some of these lacunae through a long-term study of slaveholding in the region of Bursa. We trace the incidence of slaveholding and prices from 1460 to 1880 using probate inventories. Inventories list slaves of the decedents among other capital goods and personal belongings but do not allow one to identify with any certainty how the owners employed their slaves. Therefore, the analysis proceeds in relation to hypothetical slaves with no sectoral specialization. We then work our way backward from prices, gender, and origin of the slaves toward identifying likely areas of employment. Section one briefly introduces Bursa and our sources. Section two presents descriptive statistics of ownership tracing the incidence of slaveholding among different groups of decedents. Section three examines the prices and explores the pattern of demand in connection with purchasing power, consumption/investment preferences, and alternative labor costs, i.e., wages. Section four concludes.

The setting and the sources

The sub-province of Bursa was one of the most important regions in the Ottoman Empire in a number of ways, particularly in the earlier part of the early modern era. Its administrative center, the city of Bursa, was the first capital of the nascent Ottoman state (1326–1402). With a population fluctuating between twenty-seven thousand and one hundred thousand from the fifteenth to the nineteenth century, the city became and remained a cultural and economic center throughout Ottoman rule.Footnote ¹⁷ In addition to its role in long-distance trade, it had a sizeable silk weaving sector, producing especially more expensive types destined for European markets and the upper end domestic consumers. Like most other parts of the empire, Bursa experienced growth from 1460 to 1600 and from about the mid-seventeenth century to the 1770s, with a crisis in between. Due to the large share of exports in its economy and the possibility of easy access to sea routes, Bursa's economy was open to vicissitudes of the world market and relative strength of the Ottoman economy. Therefore, it faced early deindustrialization, probably starting in the eighteenth century, as well as reindustrialization in the mid-nineteenth century. Interrupted by the Long Depression of 1873–1896, the latter process was accompanied by agricultural growth and continued until World War I.

This study loosely refers to the “region of Bursa” with a heavy focus on the urban center of the sub-province (sancak). Overrepresentation of urban Bursa in the study is not by choice but reflects respective shares of the rural and urban decedents in the registers. The choice of Bursa for this study is due, above all, to the availability of an uninterrupted series of probate inventories spanning nearly five centuries. Second, Bursa offers multiple perspectives on slavery in Ottoman lands. On the one hand, the city had a major slave market where war captives from the west as well as imports from the southern and the Black Sea routes were gathered, and some of these were employed in silk manufacture in the fifteenth and sixteenth centuries. Both the labor supply and demand for products of the sector over the centuries were tied to a matrix of international and domestic factors. On the other hand, as far as long-term patterns of slaveholding and slave prices are concerned, preliminary findings on central lands of the empire indicate that Bursa was not unique—despite its significance in the history of slavery.Footnote ¹⁸

Database and the probate inventories

The study is based on 4,325 randomly sampled probate inventories that belong to Muslim decedents (see below regarding the exclusion of the non-Muslims). The sample is organized into eleven subperiods, each twenty years long with twenty-year intervals. It starts with the earliest probate records available (1460–1480) and ends in 1860–1880, after which date slaves cease to appear in the sample (Table 1).Footnote ¹⁹ All monetary values presented are real values adjusted for inflation using the Istanbul Consumer Price Index, the only long-term index currently available for the Ottoman Empire.Footnote ²⁰ We take 1740 as the base year.

Table 1.

The sample

We use the raw sample as is and make no attempt to eliminate the demographic biases typically associated with probate inventories or possible long-term changes in the composition of the decedent population. Such biases, best studied so far in European and North American historiography, impair the value of the probates as quantitative indices for the actual population.Footnote ²¹ Ottoman probate records are no exception in that regard. Demographic biases would affect the study of slavery in a number of ways.

One of the limitations of the Ottoman probate records is the underrepresentation of non-Muslims in most places, Bursa being one of them, in most periods. At the same time, Ottomans followed diverse policies regarding slaveholding by non-Muslims,Footnote ²² and neither these policies nor their actual implementation have been studied in any detail. Finally, although sources other than probate inventories indicate the presence of non-Muslim slave owners in Bursa, at least in the fifteenth and sixteenth centuries,Footnote ²³ they must have been much fewer in number. Indeed, we have encountered only one such decedent in the sample of inventories. Therefore, non-Muslims have been excluded in the present study. This implies that our findings overrepresent slave-ownership in Bursa as a whole.

Common to all post-mortem inventory traditions is a male-upper class/status bias in the probate population. To the extent it prevails in Bursa data as well, our estimates, again, overrepresent slaveholding. Similarly, the urban bias among the registered decedents in Bursa registers inflates the incidence of slaveholding as rural slavery in this region was rather limited. Counteracting these upward biases is the absence of centrally appointed top dignitaries, i.e., the largest slave owners, in the probate population. These were kept in Istanbul in registers that belonged to the state elite (Askeri Kassam Defterleri) or together with other financial records.Footnote ²⁴ This situation significantly suppresses the number of slaves one observes in provincial samples, but not the rate of slaveholding since the population share of these dignitaries was negligible. In order to partially remedy these problems, we present separate ownership statistics for identifiable groups of owners, such as males, females, urbanites, rural decedents, people of status (identified by the presence of socio-occupational/honorific titles), and commoners (identified by the absence of the former).

Age bias in probate records is another potential problem that can affect the estimates of slaveholding and the direction of its effect cannot be easily predicted. On the one hand, older people tend to be wealthier and are more likely to own slaves; this skews the incidence of slaveholding upward among the decedents. On the other hand, this is counteracted by manumission. Per religious and ethical stipulations of Islam, not only were non-Muslim slave owners expected/obliged to give up/sell their slaves if the latter converted to Islam, Muslims, too, were encouraged to free theirs on various occasions, especially if the slave converted to Islam. Therefore, an unidentifiable number of slaves escape the purview of the probate records, giving us lower rates of slave-ownership than among the actual population. Two caveats are in order, however. First, low life expectancy in the period studied alleviates the problem. Second, manumission during owners’ lifetime may not have been as common as is assumed.Footnote ²⁵ That all the slaves identified by religion in the Bursa sample (and in many case studies) are Muslim indicates that conversion hardly guaranteed liberation. We assume that the “customary” term of servitude and manumission must have changed over time, like ownership itself, in response to a calculus of charity and economic value. As the payback time of a slave got shorter, as was the case for most slaves in the eighteenth and nineteenth centuries, ceteris paribus, manumission must have become more common.

A parallel mechanism that ended in manumission, and hence curtailed visibility of the slaves in probates, was mükatebe, contracts stipulating manumission after a set term of servitude or payment of a set amount in cash or kind by the slave.Footnote ²⁶ These are also relevant because they involved an intermediate mode of labor between bondage and freedom, and not only because it set a defined term of bondage but it also imparted legal personality to the slave the moment the contract was made, and hence suspended slave status. It is noteworthy that these contracts were the only government recommendation, short of abolishing slavery, to solve the Circassian problem in the second half of the nineteenth century.Footnote ²⁷ It is not clear, nonetheless, if this was a case of “tradition rediscovered” from within the Islamic legal repertoire of slavery in order to tackle the novel challenge of conflicts of unprecedented scale between masters and slaves and mass demand for freedom against the backdrop of authoritarian modernization synthesized with liberal notions of property, citizenship, and freedom, not necessarily of European descent.Footnote ²⁸ The few slaves-on-contract we have encountered in the probates are all from early periods. Other studies, too, suggest that the practice may have become marginal after the sixteenth century.Footnote ²⁹

Slave-ownership

Long-term trend

Probate data point to a high incidence of slave-ownership in Bursa in the early part of the period examined. The peak in slaveholding is the fifteenth century, with an ownership rate of 18.4 percent in 1460–1480 (Table 2a). In the sixteenth century, the overall rate of ownership steadily declines to 14.3 percent by 1600. Following a minor reversal in the seventeenth century—possibly resulting from a change in the composition of the sample—ownership continues to decline until 1820–1840. The eighteenth-century average is 5.5 percent, dropping to 0.1 percent by 1860–1880. The small increase in the rate of ownership observed in 1820–1840 agrees with the existing literature and consists primarily of resurgence in white male slavery. It is plausible that after prohibition of the African slave trade in 1857, slave-owning families became more reserved about their human property and preferred to present them as servants or household dependentsFootnote ³⁰ in which case, one might argue that shrinkage in slave population in 1860–1880 is partly spurious. It may also be spurious to the extent that composition of the decedent population, too, may have changed in the nineteenth century.Footnote ³¹ Yet, the long-term character of the decline in slaveholding marginalizes the significance of this possibility.

Table 2a.

Rate of slave-ownership among different groups of decedents

As to be expected, the wealthier a person was, the likelier it was that s/he owned a slave, and that meant, in general, women were less likely to have slaves than men and commoners than people of status. However, the incidence of slave-ownership declined across all levels of wealth and status (Table 2a and 2b). Thus, the top quintile of wealth, for example, had the highest rate of slaveholding in 1460–1480 (52.5 percent), dropping steadily to 36 percent in 1660–1680 and 0.6 percent in 1860–1880. As for the rest of the decedents, the average rate of ownership was less than 1 percent in the lower two-thirds from the turn of the eighteenth century onwards, and the same is true of those below the top quintile in the following century. Similarly for status: The incidence of slave-ownership among the titleholders was 27.3 percent on average in the sixteenth century, dropped to 18.1 percent in the seventeenth century, and 7.7 percent in the eighteenth century. The revival in slaveholding observed in 1820–1840 is limited to this group only. Among the commoners, only 3.2 percent owned slaves in the eighteenth century and 0.1 percent in the nineteenth century—or none, if we exclude untitled members of important families. It should be reiterated that all these figures overrepresent actual slave-ownership. Reweighting probate population by adding non-Muslims and reducing urban and upper class/status bias would pull them down. It seems important, therefore, to introduce at least a time dimension to sweeping arguments that posit slaveholding as “thoroughly integrated into both middle and upper class households throughout the empire, from Istanbul to Cairo.”Footnote ³²

Table 2b.

Rate of slave-ownership by quintiles of wealth

Since it is the long-term pattern of change in slave-holding that primarily concerns us in this study, we shall cite a few findings about the incidence of slaveholding in other locations just to present a comparative snapshot. For all the prominence of slavery in its early history, the rate of ownership in Bursa in 1660–1680 (15.1 percent among males and 7.9 percent among females) appears lower than in Trabzon, a major center of slave trade on the Black Sea coast (32 percent among males),Footnote ³³ but it is higher than in Manisa, an agricultural region in western Anatolia (14.5 percent among males and 1.3 percent among females). The pattern of decline is shared by all, and the rates of ownership come to converge in three locations around the mid-eighteenth century. There is also sporadic information on the incidence of slavery in some other locations. Pending standardization in sampling criteria and method, Istanbul stands higher than Bursa in the mid-seventeenth century but not Trabzon (two decades later) (22 percent among males and 18 percent among females)Footnote ³⁴ while Edirne, the capital of the Ottoman state after Bursa and before Istanbul, has the lowest incidence of slaveholding among the regions cited (9 percent among males and 3 percent among females) around the same period (Table 3).Footnote ³⁵

Table 3.

Slave-ownership in Bursa: comparison with later capitals

The rate of slave-ownership is paralleled by related changes, such as the average number of slaves per owner, maximum number of slaves held by a single owner, and the share of wealth invested in slaves by slave owners (Table 4). The latter is 27.6 percent from 1460 to 1600. In the following two centuries, this figure is halved, dropping to an average rate of 13.7 percent, with the last decades of the eighteenth century witnessing the lowest figure (7.9 percent) in the whole period examined. In 1820–1840, there is a remarkable increase in the share of the wealth owners invested in their slaves (11.7 percent). However, the sharp reversal here is partly related to the reduction in average wealth in connection with the crisis of the late eighteenth and early nineteenth century. In other words, title-holders, though poorer, held on to their slaves.Footnote ³⁶ In 1860–1880, there are too few incidents of slaveholding for the observations to make statistical sense in this regard.

Table 4.

Intensity of slaveholding

* Based on two observations

Slaves in different sectors

In 1460–1480 when the incidence of ownership is at its highest, all owners are city-dwellers. This may be because the rural sample is rather small in this period. However, even in our best data points for rural observations (1540–1560, 1820–1840, and 1860–1880) (Table 1), slaveholding appears as a heavily urban phenomenon in the region of Bursa. In 1540–1560, the rate of ownership in rural probates is a third of the urban probates (6.1 percent as opposed to 18.3 percent). In the early nineteenth century, the discrepancy is even larger (0.9 percent to 3.6 percent), and finally, in 1860–1880, both of the two slave owners observed are city dwellers (Table 2). As for the average number of slaves per owner, it is lower in the countryside in 1540–1560 (1.5 as opposed 2.1), but higher in 1820–1840 (2.9 as opposed to 1.7)—as is the case with the three intervening centuries. We cannot tell at the moment whether at issue here is a changing trend in labor use in the countryside.

The long-run trend is the same in rural inventories as well, and there is no sign of a revival anytime in the nineteenth century. Agricultural commercialization in the nineteenth century, involving grapes, tobacco, olives, and mulberries in this region, appears to have rested more heavily on free labor supplied by small producers and semi-nomadic seasonal workers coming from nearby regions.Footnote ³⁷ Yet, Circassian immigrants’ slaves are known to have been employed in agriculture in the larger sub-province of Hüdavendigar, encompassing also the district of Bursa, even though they are not visible in our sample.Footnote ³⁸ As Bursa was also among regions where sharecropper slaves were employed in the early centuries of Ottoman rule, further investigation is called for into the interplay between Ottoman laws and variable land-labor ratios and its impact on large estate formation and slave labor.

Preliminary comparison of the components of wealth held by slave owners to those of decedents with no slaves provides further insight into the sectors or lines of activity where slaves may have been used in primary or by employment. According to this, ownership of resources related to textiles appears at least twice as common among slave owners as among the rest (Table 5). The largest discrepancy between slave-owning and non–slave-owning decedents is observed in ownership of textile shops (14.3 fold more common among male slave owners) followed by ownership of textile tools (four fold more common among female slave owners) and ownership of commercial textile stockFootnote ³⁹ (3.6 fold more common among male slave owners). Notably, the difference between slave-owning inventories and those that do not is small in wealth categories “agricultural tools” and “livestock” for both men and women and in category “landownership” for men (but not women who might need extra hands for help with land). While in need of further analysis controlling for wealth and other variables, these findings confirm and add to the foregoing. They suggest that slaveholding was not associated with agricultural activity in Bursa but the presence of some association between agricultural commercial stock and slaveholding calls for more detailed analysis. Association between slaveholding and textile production or trade appears clear. Further research is needed to clarify whether this association rests on direct employment of slaves in the sector, as in the fifteenth and sixteenth centuries, or by-employment of domestic slaves, which is more likely to be the case with female decedents who did not own shops but tools and slaves. The observation that half of the three hundred spinning wheels in Bursa were operated by women in 1678 further supports this.Footnote ⁴⁰

Table 5.

Ownership of means of production

Slave prices

Long-term trends

Contrary to received wisdom about slavery in the Ottoman Empire, the very long-term trend in average slave prices is downward, reaching its lowest in 1860–1880. The latter should be regarded cautiously as there are only eight slaves in probates from 1860–1880 but, it conforms to the pattern of decline that had started already in the mid-eighteenth century. On closer look, one observes two distinct phases in prices from 1460 to 1880. Until the early eighteenth century, prices are on the rise, and afterward, they begin to decline (Table 6). Our preliminary examination of the prices in six other locations in central lands confirms this pattern.Footnote ⁴¹ While an upward price trend is expected due to an increasingly restricted inflow of slaves, its reversal after 1720 is not. Two factors may be at work here: a decline in the demand for slaves and/or oversupply of slaves through trade. The only substantial traffic that could affect the market in the eighteenth century was from Africa. The identity of the slaves in the sample confirms.

Table 6.

Slave prices and wages, akçe, 1740 = 1

* Strong data points are shaded.

A supply-side approach: Gendering and coloring the decline in slave prices

The least common piece of information about the characteristics of the slaves in Bursa probate inventories is their age. Less than a tenth of the slaves in the sample are identified as children, youth, or elderly, and presumably the rest are adults. While the price of child slaves can be as low as a third of the adult prices and the price of the elderly can be much lower, they do not affect the average price levels much as they are few in number.

A nearly universal identity detail about the slaves is their gender. Until the seventeenth century, relative female/male slave prices are erratic, females being 25 percent more expensive in 1540–1560 but 12 percent cheaper in the following period. From the early seventeenth century onward, female slave prices are consistently higher, stabilizing at 10–14 percent difference in the early eighteenth century (Table 6). Thus, gender appears as a relevant price factor but not a very strong one. When the origin of the slaves is juxtaposed onto the gender matrix, a much bigger difference is obtained than between genders alone. A probable increase in the number and/or share of the African slaves correlates with the downward trend in average prices.Footnote ⁴²

In all periods for which we have a reasonable number of observations, the price of African slaves is lower than that of the northern slaves from different parts of Eurasia.Footnote ⁴³ Our highest price observation for an African slave, one of the two so identified before 1700, is in 1460–1480 (21,900 akçe with 1740 as the base period) and may reflect the prices in the exotic slave market at this date. Our best data point to compare the relevance of origins is 1740–1760, where we have forty-three African, twenty west Eurasian slaves, and twenty-two others with unidentified origins. Here, west Eurasian slaves are valued more than double the price of their African counterparts (Table 6). The same is true of 1820–1840, where the sample is slightly weaker. Proximity to places of origin and trade routes obviously mattered, but Bursa was about equidistant from both slave zones (northern/west Eurasian/white, southern/African/black slaves), hence location is an unlikely factor in relative prices.

Price discrepancy between slaves of different origin has been noted in earlier studies as well—though not in comparative light or in terms of its long-term implications for accessibility of slave labor. The discrepancy may have emerged or increased over time. For example, Jennings reports no difference in Cyprus in 1590–1640 while Sak's study suggests that Africans fetched only slightly lower prices in seventeenth-century Konya in central Anatolia. Wilkins finds a similar pattern in contemporary Aleppo while the difference seems larger (64 percent) at the turn of the eighteenth century in Damascus and even larger in Cairo a little later. It was two-to-three fold in Istanbul in the 1850s and 1860s, and could be as high as tenfold.Footnote ⁴⁴ Still, what looks like a time trend indicating increasing price discrepancy cannot be properly understood without considering specific labor needs in each region.

In view of the foregoing, the onset of a decline in average prices in the eighteenth century begins to make sense. Since it was the west Eurasian supply that shrank, particularly after the Russian expansion in the north, the price of west Eurasian slaves continued rising, and that of women peaked in 1820–1840, when Russia captured Georgia and Circassia and came to control all traffic from Circassia. It must have been this trend that occasioned a ban on slave exports to Egypt in 1786.Footnote ⁴⁵ Therefore, one has to turn to the relative size of the African population among slaves to understand the decline in average prices.Footnote ⁴⁶

Ethnic origin is rarely stated in the probates before the eighteenth century, but studies based on other sources indicate that grand majority of the slaves in Bursa in the fifteenth and sixteenth centuries were from the north, i.e., war captives from Eastern Europe or people trafficked from northern Black Sea Region and the Caucasus. African slaves began to arrive in central lands of the empire already in the sixteenth century but appear heavily outnumbered by northern slaves at this time.Footnote ⁴⁷ The African traffic is noted to have increased in the eighteenth centuryFootnote ⁴⁸ but there is little quantitative detail available on this matter. In fact, estimates of the volume of traffic toward the Ottoman lands from the seventeenth century to the last quarter of the eighteenth century may be taken to indicate stability in absolute figures, a matter in need of further investigation.Footnote ⁴⁹ Be that as it may, geo-ethnic origin gets recorded in more than half of the cases from the mid-eighteenth century onward while the total number of slaves in the probate inventories continually shrinks. This may be an indication of increased diversity within the slave population as well as vigilance about slaves’ origin, lest war-time prisoners get lost in a web of old style slavery, contrary to new rules of international law of war and ideas of subjecthood in sovereign states.Footnote ⁵⁰

While changes in the volume of traffic from the two supply zones must have fostered the opposite movement of the prices of African and Caucasian slaves, the presence of price differentials in the first place is where demand begins to enter the picture. These differences reflected long established patterns of employment, Africans in menial jobs and domestic service as opposed to west Eurasian slaves in military, administrative jobs, and domestic partnership (concubinage) as well as domestic service. These patterns drew on a cultural system that attributed different qualities and aptitudes, and thus labor roles and market value to slaves of different gender and origin.Footnote ⁵¹ Hence the higher value of fair-skinned slaves and desirability of such women all the way from the Indian Ocean to the Mediterranean and colonial America.Footnote ⁵² As there was no legal boundary between domestic and sexual service for an unmarried female slave in Islamic law; one might speculate that all things being equal, Africans would have been unfavorably priced even as household slaves as they were less versatile. Educated expressions of such cultural preferences can be found in Ottoman physiognomy, an heir to its Islamic precursor, itself an heir to Greek scholarship. While currently there is no comparative account of the scholarly views on and popular perceptions of color and ethnic origin in different parts of the Ottoman world, it is conceivable that less than gracious views of the notable bureaucrat/scholar Mustafa Ali (1541–1600) of Africans were not uncommon.Footnote ⁵³

A demand-side approach: Slave prices as labor cost

While a shift from a predominantly white body of slaves to a predominantly African body of slaves may account for the unexpected downturn in average slave prices in the eighteenth century, the drop in the price of Caucasian slaves, too, in 1860–1880 is noteworthy. Further research is needed to verify whether this decline is related to a short-term supply boom due to Circassian immigration and Russia's relaxation of controls over the traffic routes or a decline in the demand for domestic slaves or concubines as well.Footnote ⁵⁴ In this section, we investigate the possibility of a decline in demand for slaves in general, not particular types of slaves, as the areas of employment cannot be identified with any certainty. For this purpose, first, the connection between slave prices and alternative labor costs will be considered, and then, investment preferences of the wealthiest group of probates will be examined.

An early insight into the question of alternative labor costs can be found in Baer (1967), who argues that the emergence of a cheap and free labor market as a result of proletarianization in the last decades of the nineteenth century were factors without which top-down measures to end slavery in Egypt would have failed.Footnote ⁵⁵ To proletarianization proper, of both men and women, one should add the likely emergence of other labor forms, for example, semi-free labor of debt peons or, even more pertinent, child domestics.Footnote ⁵⁶ Corvée and sharecropping arrangements resulting from indebtedness in the countryside are observed in various regions in Anatolia and the Balkans in the nineteenth century but the topic has been understudied as peasant alienation is thought to be limited due to high land-labor ratio.Footnote ⁵⁷ As for female labor, it is known to be quite common in important textile centers like Bursa and Ankara, at least, in the seventeenth century, and the number of non-domestic female laborers as well as domestics may have further increased in the nineteenth century, if not before.Footnote ⁵⁸ Finally, a better alternative to unskilled adult domestics were children, and studies on nineteenth-century Istanbul indicate that first, foster/maid children kept in wealthy and middle class households gradually became wage laborers and, second, their numbers kept increasing and, third, so did their wages.Footnote ⁵⁹ All these processes can be associated with the decline in slave prices if one can demonstrate them to be at work in Bursa from the eighteenth century onwards, i.e., when prices took a downturn. In the absence of quantitative information about any of them, especially before the nineteenth century, we turn below to adult male wages to consider the relationship between free and bonded labor, assuming long-term agreement in the labor market.

Throughout the period studied, slave prices and wages move incrementally in the same direction but diverge in the very long run. Unlike absolute prices, which peaked in the early eighteenth century, slave prices relative to wages peaked about a century earlier, measuring 545 days of skilled labor and 740 days of unskilled labor in 1620–1640 (Figure 1). From then on, they declined steadily going below the fifteenth-century level by 1860–1880. Adding the cost of feeding, clothing, and lodging a slave would raise the workday equivalent of the slave prices perhaps by a factor of 1.75 (assuming that three fourths of the daily wages were physical reproduction costs). Thus, in 1820–1840, for example, anyone who anticipated to employ the same (unskilled) laborer for a year and had ready cash for advance payment would have been better off buying a slave as the workday equivalent of the average slave price was 354 at the time, and with the living expenses added, it amounted to less than two years of unskilled wages.Footnote ⁶⁰ There is also some evidence indicating that wages in Bursa were lower than Istanbul,Footnote ⁶¹ which would raise the relative slave prices some more and the workday equivalent up to two full years perhaps. From an economic point of view, this may still seem rather appealing but increasingly favorable relative costs do not trigger an increase in slave-ownership.

Figure 1.

Slave prices relative to wages

One may argue that this is to be expected if slave prices were already far beyond the means of all but the few wealthiest as indicated by the increasing discrepancy of wealth between slave owners and the rest (Table 2b). Further, “wealth elasticity” of the demand for slaves might be limited—unless new venues of employment emerged. Looking closer at the purchasing power of the wealthiest would, thus, be revealing.

Slave purchasing power of the top quintile in the probate population, i.e., the number of slaves an average estate could buy in 1460–1480 was 19.5 slaves and the rate of ownership was 52.4 percent. Two centuries later, in 1660–1680, when their purchasing power was up by a third (33 slaves), the rate of ownership was down by a third (36 percent) (Figure 2). Starting in the early eighteenth century, mean wealth in the top quintile steadily declined as did its slave equivalent. Still, the eighteenth-century purchasing power average was 26.6 slaves, i.e., higher than in 1460–1460 while the incidence of ownership was half of what it had been then (22.9 percent). The divergence between the purchasing power and actual ownership continued in the nineteenth century. In other words, the relationship between the two changed over time, and more slave purchasing power did not automatically lead to a higher incidence of slaveholding. This raises the possibility that the decline in slave prices was also related to withering demand as well as an increase in the slave labor supply.

Figure 2.

Slave purchasing power and ownership in top quintile of wealth

Conclusion

This study confirms received knowledge about a steady decline in slave-ownership in the Ottoman Empire, long preceding the first government attempts to curb slave trade in the nineteenth century. The decline is observed across all economic and social strata with a minor revival in the early nineteenth century among title-holders. However, unlike what is commonly assumed in the relevant literature, rising prices may not have been the only factor in this process. That they increased parallel to shrinkage in slave supply is validated up to the first quarter of the eighteenth century but not thereafter. After 1700–1720, prices and slaveholding both decline.

These observations call for a revision of the narrative about the decline of slavery in the Ottoman Empire, at least, in the region of Bursa and, possibly, the rest of the central lands. Price variation among slaves of different origin and increasing preponderance of African slaves appear instrumental in the decline of the average prices. Therefore, changes in the volume of African traffic, a factor underestimated in the study of slavery in central lands before the nineteenth century, emerge as a key element to be studied further, and inextricably link the history of slavery in this zone to globalization—which set the direction of the African traffic and created pockets of export-oriented production where slave labor proved profitable. The latter was not the case in the region of Bursa.

Decline in slave-ownership despite declining prices calls for an investigation of the demand factor. For this purpose, we have examined slave prices in connection with wages and found that after 1700–1720, slave prices declined relative to wages, too. Thus if withering of slaveholding signals a switch to alternative forms of labor, it does not appear economically rational if we consider male wage-earners alone. While female and child labor may have been more competitive—and needs to be investigated—it is not likely that their wages moved differently from male wages, hence increasing divergence with slave prices is expected to prevail.

The main axis of further research into why demand for slaves may have changed would involve labor needs of regional economies responding to domestic or global demand (expanding, but rather domestic in the case of Bursa after the seventeenth century) and effects of wars, epidemics, and immigration on regional demography. Labor demand and supply, free or bonded, for domestic or other employment, must have varied considerably in Ottoman lands from 1740 to 1880, a period that saw consecutive episodes of growth and crisis. In global perspective, however, the long run of early modernity was marked by demographic and economic growth—which Stanziani further characterizes as a “common wave of capitalism,” arguing that labor institutions interacted across Eurasia and despite demographic growth, labor processes, and coercion intensified.Footnote ⁶² As the historiography of the global turn has placed the Ottoman lands ever more firmly in Eurasian patterns of change, there is basis for asking how the interplay between the increase in African slave population and little known processes of free and intermediate forms of dependent labor were regional manifestations of this broad trend.

Finally, all of the above concerns slaves as laborers, domestic or otherwise, whose possession and employment rested on considerations of profitability. Yet, the value of slaves did not derive solely from the economic value they produced. First, slaves were also prestige goods whose possession alone generated utility in the form of power and reputation accorded to an ideal lifestyle—of which an ideal household and sheer display of wealth were a part. Concern for prestige cut across slave systems, domestic and plantation slavery alike, but mattered for the wealthiest and the noblest above all.Footnote ⁶³ Looking at the Bursa sample from this perspective would point to socio-cultural change affecting conspicuous consumption preferences of those who could afford to buy slaves but no longer did. Such change needs to be explored in connection with new expressions of distinction that the eighteenth century may have brought as well as patterns of household formation—not only among the grandees of the land mimicking the sovereign but also humbler folk whose notions of respectability may have involved or come to involve different ideals.Footnote ⁶⁴ Therefore, it may be more appropriate to speak of the decline of slavery in the plural to mark different the dynamics among different classes.

Second, labor process, too, was culturally defined and built on ideas about entitlements, human nature, and servitude. One cannot rule out, in view of the foregoing, that a climate ill-disposed to slaveholding emerged several decades earlier than is commonly presumed—except among the top elite and the bureaucracy, the bulwark of resistance to abolitionism,Footnote ⁶⁵ and suppressed slave purchases and shortened the “customary” term of service before manumission. As far as state policy is concerned, Smiley insinuates a trickle-down effect from war diplomacy concerning the captives to slavery at large, reflection of a shared new turn in state-making, and the global order in the eighteenth century. Similarly, Clarence-Smith's reference to possible “undercurrents” of antislavery sentiments and ideas that may have survived the mainstream approval of slavery in Islamic historyFootnote ⁶⁶ is relevant for eighteenth-century thought, an important component of which was rereading orthodoxy in diverse ways.Footnote ⁶⁷ This is not to suggest that cultural change was the driving force but, rather, to recognize that profitability considerations alone may have been nowhere sufficient to end slavery.Footnote ⁶⁸ It may turn out that even the cultural demise of slavery was a “connected” process and not because of imperialist abolitionism alone but of deeper parallels.

Footnotes

We thank Terrence Walz and Ehud R. Toledano for generously sharing their extensive knowledge on the topic and Y. Hakan Erdem for his contribution to an early version of this study. We also thank our anonymous readers for very valuable suggestions.

References

NOTES

2. Kunt, Metin, The Sultan's Servants: The Transformation of Ottoman Provincial Government, 1550-–1650 (New York, 1983)Google Scholar; Peirce, Leslie Penn., The Imperial Harem: Women and Sovereignty in the Ottoman Empire (New York, 1993)Google Scholar; Hathaway, Jane, The Politics of Households in Ottoman Egypt: The Rise of the Qazdaglis (Cambridge, 1997), 37–46Google Scholar; Ergin, Nihat, Osmanlı Devlet'inde Kölelik (İstanbul, 1998)Google Scholar; Fay, Mary Ann, Unveiling the Harem : Elite Women and the Paradox of Seclusion in Eighteenth-Century Cairo (Syracuse, NY, 2012)Google Scholar; Nalçacı, Nida N., Sultanın Kulları, Erken Modern Dönem İstanbul'unda Savaş Esirleri ve Zorunlu İstihdam (İstanbul, 2015)Google Scholar; Argıt, Betül İpşirli, Hayatlarının Çeşitli Safhalarında Harem-i Hümayun Cariyeleri, 18.Yüzyıl (İstanbul, 2017)Google Scholar.

3. Toledano, Ehud, The Ottoman Slave Trade and Its Suppression (Princeton, NJ, 1982)Google Scholar; Erdem, Y. Hakan, Slavery in the Ottoman Empire and its Demise, 1800–1909 (London, 1996)CrossRef Google Scholar; Walz, Terence and Cuno, Kenneth M., eds., Race and Slavery in the Middle East: Histories of Trans-Saharan Africans in Nineteenth-Century Egypt, Sudan, and the Ottoman Mediterranean (Cairo, 2010)Google Scholar; Montana, Ismael M., Abolition of Slavery in Ottoman Tunisia (Gainesville, FL, 2013)CrossRef Google Scholar.

4. Faroqhi, Suraiya, “From the slave market to Arafat: Biographies of Bursa women in the late fifteenth century,” Turkish Studies Association Bulletin, 24 (2000): 3–20Google Scholar; Toledano, Ehud R., As If Silent and Absent: Bonds of Enslavement in the Islamic Middle East (New Haven, CT, and London, 2007)Google Scholar; Sonbol, Amira El Azhary, ed., Beyond the Exotic: Women's Histories in Islamic Societies (Syracuse, NY, 2005)Google Scholar; Powell, Eve M. Troutt, “Will That Subaltern Ever Speak? Finding African Slaves in the Historiography of the Middle East,” in Middle East Historiographies: Narrating the Twentieth Century, eds. Gershoni, I., Singer, A. and Erdem, H. Y. (Seattle, 2011)Google Scholar; Powell, Eve Troutt, Tell This in My Memory: Stories of Enslavement from Egypt, Sudan and the Ottoman Empire, (Stanford, CA, 2012)CrossRef Google Scholar.

5. Smiley, Will, From Slaves to Prisoners of War: The Ottoman Empire, Russia, and International Law (Oxford, 2018)CrossRef Google Scholar; Şakul, Kahraman, “What happened to Pouqueville's Frenchmen? Ottoman Treatment of the French Prisoners during the War of the Second Coalition (1798–1802),” Turkish Historical Review 3 (2012): 168–95CrossRef Google Scholar.

6. İnalcık, Halil, “Servile Labor in the Ottoman Empire,” in The Mutual Effects of the Islamic and ]udeo-Christian Worlds: The East European Pattern, eds. Ascher, Abraham, Halasi-Kun, Tibor, and Kiraly, Bela K. (New York, 1979), 42Google Scholar; Faroqhi, “From the slave market,” 3–20; idem, “Quis Custodiet Custodes? Controlling Slave Identities and Slave Traders in Seventeenth and Eighteenth Century Istanbul” in Frontiers of Faith, eds. Andor, Eszter and Tóth, István Györgi (Budapest, 2001), 119–34Google Scholar. idem, “Manumission in 17th-Century Suburban Istanbul,” in Mediterranean Slavery Revisited (500–1800) – Neue Perspektiven auf mediterrane Sklaverei (500–1800), eds. Hanß, Stefan and Schiel, Juliane (Zurich, 2014), 387Google Scholar.

7. Erdem, Slavery, 26, 53–61; Toledano, The Ottoman Slave Trade, 90; Austen, Ralph A., “The Trans-Saharan Slave Trade: A Tentative Census,” in Uncommon Market: Essays in the Economic History of the Atlantic Slave Trade, eds. Gemery, H.A. and Hogendorn, Jan S. (New York, 1979), 23–66Google Scholar; idem, “The Mediterranean Islamic Slave Trade out of Africa: A tentative Census,” Slavery & Abolition 13 (1992), 221–22, 240Google Scholar; Ricks, Thomas M., “Slaves and Slave Traders in the Persian Gulf, 18th and 19th Centuries: An Assessment,” Slavery and Abolition 9 (1988): 60–70CrossRef Google Scholar. For a summary of trade figures, see Fisher, Alan, “Chattel Slavery in the Ottoman Empire,” Slavery and Abolition 1 (1980): 32CrossRef Google Scholar. Manning, Patrick, Slavery and African Life: Occidental, Oriental, and African Slave Trades (Cambridge, 1990), 136–40Google Scholar; Compare Ze'evi, Dror, “Kul and Getting Cooler: The Dissolution of Elite Collective Identity and the Formation of Official Nationalism in the Ottoman Empire,” Mediterranean Historical Review, 11: 2 (1996), 191CrossRef Google Scholar.

8. Zilfi, M. C., Women and Slavery in the Late Ottoman Empire: The Design of Difference (New York, 2010)Google Scholar seeks to historicize the primacy of domestic female servitude.

9. Nur Sobers Khan, “Slaves ‘without’ shackles: Forced labour and ‘manumission in the Galata court registers, 1560–1572,” (PhD dissertation, Cambridge University, 2012); Mustafa Akkaya, “Osmanlı Dönemi Batı Anadolu'da Tarımsal İşgücünde Kullanılan Köleler,” in Z. Güneş Yağcı, F. Yaşa and D. İnan eds., Osmanlı Devleti'nde Kölelik (Istanbul, 2017), 249–68; Hayri Gökşin Özkoray, “Work, Labor and Local Belonging,” paper presented at the IGK Work and Human Lifecycle in Global History conference, 14 June 2013, cited in Witzenrath, Christoph, “Introduction: Slavery in Medieval and Early Modern Eurasia: An Overview of the Russian and Ottoman Empires and Central Asia,” in Eurasian Slavery, Ransom and Abolition in World History, 1200–1860, ed. Witzenrath, C. (New York, 2016), 4CrossRef Google Scholar; Özbay, Rahmi Deniz, “Osmanlı İmparatorluğu'nda Köle Emeğinin İstihdamı ve Mükâtebe Yöntemi,” Kocaeli Üniversitesi Sosyal Bilimler Enstitüsü Dergisi 17/1 (2009): 148–63Google Scholar. On the link between military slavery and manual labor see Yılmaz, Gülay, “Becoming a Devshirme: The Training of Conscripted Children in the Ottoman Empire,” in Campbell, Gwyn, Miers, Suzanne, and Miller, Joseph C. eds., Children in Slavery Through the Ages (Ohio, 2009), 119–34Google Scholar.

10. Martínez, Ivan Armenteros and Ouerfelli, Mohamed eds., L’économie de l'esclavage en Méditerranée médiévale et moderne (Aix-en-Provence, 2016)Google Scholar; Blumenthal, Debra G., Enemies and Familiars, Slavery and Mastery in Fifteenth-Century Valencia (Ithaca; London, 2009)Google Scholar; Kolchin, Peter, “The Big Picture: A Comment on David Brion Davis's ‘Looking at Slavery from Broader Perspectives’,” The American Historical Review 105/2 (2000): 469–70CrossRef Google Scholar.

11. Agricultural slaves were employed on farms owned by the royal family, other grandees, and waqfs. Common during the early Ottoman expansion, the practice disappeared after the sixteenth century. Ö. Barkan, L., ‘Kulluklar ve Ortakçı Kullar,” Istanbul Üniversitesi İktisat Fakültesi Mecmuası I (1940), 98–245, 397–406Google Scholar; idem, Süleymaniye Camii ve İmareti İnşaatı (1550–1557), (Ankara, 1972–79), vol.1, 13, 93, 132, 135; İnalcık, “Servile Labor,” 29–37; Akkaya, “Osmanlı Dönemi.”; Sahillioglu, Halil, “Slaves in the Social and Economic Life of Bursa in the Late 15^th and Early 16^th Centuries,” Turcica 17 (1985): 83–86CrossRef Google Scholar; Fisher, “Chattel Slavery,” 37; Seng, Yvonne, “A Liminal State: Slavery in Sixteenth-Century Istanbul,” in Slavery in the Islamic Middle East, Princeton Papers: Interdisciplinary Journal of Middle Eastern Studies, ed. Marmon, Shaun, 7 (1999): 25–42Google Scholar.

12. For example, agricultural (male) slavery may have been the dominant mode in Manisa in western Anatolia until the nineteenth century. Hülya Canbakal and Alpay Filiztekin “Slaveholding and its demise in Ottoman central lands: Bursa-Manisa-Trabzon, 1460–1880,” unpublished paper submitted at the Fifth World Congress of Middle Eastern Studies, Sevilla, July 16–20, 2018.

13. See Blumenthal, Enemies and Familiars, 80–121 for the use of domestic slaves in a variety of productive activities including agriculture in fifteenth-century Valencia, and Özbay, “Osmanlı İmparatorluğu'nda Köle Emeğinin İstihdamı,” 155, for possible shifts in slaves’ work assignments.

14. Kuznets, Simon, “National Income, 1929–1932,” NBER 49 (1934): 1–12Google Scholar. A recent study on the US economy, for example, concludes that incorporating home production into national accounts increases US GDP by 39 percent in 1965 and 26 percent in 2010. Bridgman, Benjamin et al. , “Accounting for Household Production in National Accounts, 1965–2010,” Survey of Current Business 92 (2012), 23–36Google Scholar. Also see Sarti, Raffaella, Bellavitis, Anna and Martini, Manuela (eds.), What is Work? Gender at the Crossroads of Home, Family, and Business from the Early Modern Era to the Present (New York, 2018)Google Scholar for new approaches to domestic labor.

15. Austen, “The Trans-Saharan Slave Trade,” 43.

16. Johnson, Walter, Soul by Soul, Life inside the Abtebellum Slave Market, (Cambridge, MA, 1999)Google Scholar.

17. Behar, Cem, Osmanlı İmparatorluğu'nun ve Türkiye'nin Nüfusu 1500–1927 (Ankara, 1996), 25, 27, 33Google Scholar; Gerber, Haim, Economy and Society in an Ottoman City: Bursa, 1600–1700 (Jerusalem, 1988)Google Scholar; Çizakca, Murat, “Price History and the Bursa Silk Industry: A Study in Ottoman Industrial Decline, 1550–1650,” The Journal of Economic History 40 (1980), 533–50CrossRef Google Scholar.

18. Canbakal and Filiztekin, “Slaveholding and its demise.”

19. The data come from the larger project Hülya Canbakal, “Distribution of Wealth in the Ottoman Empire,” Turkish Agency for Scientific and Technological Research (TUBITAK), Project No. 108K034, 2008–2012. For detailed information, see Hülya Canbakal and Alpay Filiztekin, “A database and an Assessment of Ottoman Probate Records,” work in progress.

20. Pamuk, Şevket, İstanbul ve diğer kentlerde 500 yıllık fiyatlar ve ücretler, 1469–1998 (Ankara, 2000)Google Scholar; Özmucur, Süleyman and Pamuk, Ş., “Real wages and standards of living in the Ottoman Empire, 1489–1914,” The Journal of Economic History, 62, 2 (2002: 293–321)CrossRef Google Scholar.

21. McCants, Anne E.C., “Inequality among the poor of eighteenth-century Amsterdam,” Explorations in Economic History 44 (2007): 1–21CrossRef Google Scholar; Keibek, A. J. and Shaw-Taylor, L., “Early modern rural by-employments: a re-examination of the probate inventory evidence,” Agricultural History Review 61 (2013): 244–88Google Scholar. Arkell, Compare Tom, “Interpreting probate inventories,” in When Death Do Us Part: Understanding and Interpreting the Probate Records of Early Modern England, eds. Arkell, Tom, Goose, Nigel and Evans, Nesta (Oxford, 2000), 14–37Google Scholar.

22. Fisher, “Chattel Slavery,” 35.

23. Sahillioğlu, “Slaves,” 83–86.

24. Öztürk, Sait, Askeri Kassama Ait Onyedinci Asır İstanbul Tereke Defterleri: Sosyo-Ekonomik Tahlil (İstanbul, 1995)Google Scholar; Özcan, Tahsin, “Muhallefat,” DVİA (Istanbul, 2005), 30: 406–07Google Scholar.

25. Brunschvig, Robert, “Abd,” Encyclopaedia of Islam, (Leiden, 1960): 24–40Google Scholar; Erdem, Slavery, 152–53; Nalçacı, Sultanın Kulları, 99; Clarence-Smith, William G., Islam and the Abolition of Slavery (Oxford, 2006), 67–70Google Scholar; Fisher, Alan, “Studies in Ottoman Slavery and Slave Trade, II: Manumission,” Journal of Turkish Studies 4 (1980): 49–56Google Scholar.

26. Atar, Fahrettin, “Mükatebe”, DVİA 31 (2006): 531–33Google Scholar; Engin, Nihat, “Köle,” DVİA 26 (2002): 246–48Google Scholar. For similar forms of promised manumission in Renaissance Italy, see McKee, Sally, “Domestic Slavery in Renaissance Italy,” Slavery and Abolition 29: 3 (2008): 312CrossRef Google Scholar.

27. Ehud Toledano, The Ottoman Slave Trade, 164–65.; Karamürsel, Ceyda, “Transplanted Slavery, Contested Freedom, and Vernacularization of Rights in the Reform Era Ottoman Empire,” Comparative Studies in Society and History 59/3 (2017): 690–714CrossRef Google Scholar. Russia implemented a similar policy of self-purchase in the Caucasus for the same purpose in the early nineteenth century. Clarence-Smith, William G., “Slavery in Early Modern Russia,” in Mediterranean Slavery Revisited (500–1800) – Neue Perspektiven auf mediterrane Sklaverei (500–1800), eds. Hanß, Stefan and Schiel, Juliane (Zurich, 2014), 137–38Google Scholar.

28. Another reinvention was imposition of the slave tax in the eighteenth century to distinguish rebel non-Muslims from non-rebels, also driven by redefinition of subjecthood. Smiley, From Slaves to Prisoners of War, 177. See also Clarence-Smith, Islam and the Abolition of Slavery, 129–50; Freamon, Bernard K., “Straight, No Chaser: Slavery, Abolition, and Modern Islamic Thought,” in Indian Ocean Slavery in the Age of Abolition, eds. Harms, Robert, Freamon, Bernard K., and Blight, David W. (New Haven, CT, 2013), 61–80CrossRef Google Scholar.

29. İnalcık, “Servile Labor,” 25–52; Halil Sahillioğlu, “Onbeşinci Yüzyılın Sonu ile Onaltıncı Yüzyılın Başında Bursa'da Kölelerin Sosyal ve Ekonomik Hayattaki Yeri,” Gelişme Dergisi (1981): 76–80; Sobers Khan, “Slaves ‘without’ shackles.” For seventeenth century examples, see Özbay, “Osmanlı İmparatorluğu'nda Köle Emeğinin İstihdamı.”

30. Ferguson, Michael and Toledano, Ehud R., “Ottoman Slavery and Abolition in the Nineteenth Century,” in The Cambridge World History of Slavery, Volume 4, eds. Eltis, D. and Engermen, S. L. (Cambridge, 2017), 211Google Scholar.

31. Canbakal and Filiztekin, “A database.”

32. McKee, Sally, “The Familiarity of Slaves in Medieval and Early Modern Mediterranean,” in Mediterranean Slavery Revisited (500–1800) – Neue Perspektiven auf mediterrane Sklaverei (500–1800), eds. Hanß, Stefan and Schiel, Juliane (Zurich, 2014), 503–04Google Scholar.

33. The female sample is too small in this period.

34. Istanbul estimates are based on Öztürk, Askeri Kassama Ait, 359–66, which uses probate registers of the state elite (askeri) alone. The author reports to have selected 1000 probates out of 2670 that are in the first twenty registers. The criteria of selection are not stated. Öztürk, ibid. 105, 138. Compare Yvonne J. Seng, “The Üsküdar Estates (Tereke) as Records of Everyday Life in a Ottoman Town, 1521–24,” Unpublished PhD Dissertation, Chicago University, 1991, 274, where the rate of slave ownership in a mixed sample of 89 probates is 4.5 percent and all of the slave owners are askeri.

35. Selma Kuşu, “Şer'iye Sicillerine göre H.1065-1079/M. 1655–1669 Tarihleri Arasında Edirne'de Sosyo-Ekonomik Hayat” (MA Thesis, Trakya Üniversitesi, 2009). Nearly all studies on Ottoman slavery quote and use Barkan (1966) for Edirne, which is based on probate registers of the state elite alone and points to a much higher level of slave-ownership (44 percent). We do not think the study is useful for statistical purposes because the sample includes “interesting” cases only, as explicitly stated by the author himself. Ömer Barkan, L., “Edirne Askeri Kassamı'na ait Tereke Defterleri (1545–1659),” Belgeler 3 (1966): 1–473Google Scholar. See also Colette Establet and Jean-Paul Pascual, La gent d’État dans la société ottomane damascène. Les ‘askar à la fin du XVIIe siècle (Damas, 2011), http://books.openedition.org/ifpo/1678 and Grehan, James, Everyday Life and Consumer Culture in Eighteenth-Century Damascus (Seattle, 2007), 173Google Scholar. The former, based on both elite and non-elite probates, reports 13 percent slaveholding at the turn of the eighteenth century; the latter, based on registers of the commoners, reports 2 percent slave-ownership in mid-eighteenth century.

36. On slaveholding as a status symbol among the official elite (askeri), see Establet and Pasquale, La gent d’État, 113. Also Baer, Gabriel, “Slavery in Nineteenth Century Egypt,” The Journal of African History 8 (1967): 422CrossRef Google Scholar; Sahillioğlu, “Slaves,” 47.

37. Keyder, Çağlar, “Türk Tarımında Küçük Köylü Mülkiyetinin Tarihsel Oluşumu ve Bugünkü Yapısı,” Toplumsal Tarih Çalışmaları (Ankara, 1983), 221–74Google Scholar; idem, “Introduction: Large-Scale Agriculture in the Ottoman Empire?” in Landholding and Commercial Agriculture in the Middle East, eds. Ç. Keyder, and Tabak, F. (Albany, NY, 1991): 1–15Google Scholar.

38. Kasaba, Reşat, The Ottoman Empire and the World Economy: The Nineteenth Century (Albany, 1988), 62–64, 84Google Scholar. Compare, Saydam, Abdullah, “Tanzimatçıların Ağalık ve Beylik Kurumunu Kaldırmaya Yönelik Çabaları”, Toplumsal Tarih 1 (1994): 9–13Google Scholar.

39. To differentiate commercial amounts of stock from stocks for home use, we have set as threshold the median value of the relevant goods in each period and considered larger stocks as commercial.

40. Faroqhi, “Guildsmen and handicraft producers,” 343.

41. Hülya Canbakal and Alpay Filiztekin, “Slaves and Slavery among Muslims in Ottoman Bursa (1460–1880),” Unpublished paper, Early Modern Economic and Social History Seminars, University of Cambridge, May 2016 considers slavery in Bitola, Manisa, Kayseri, Trabzon, Antep, Diyarbekir.

42. Compare Zilfi, Madeline, “Thoughts on Women and Slavery in the Ottoman Era and Historical Sources,” in Beyond the Exotic: Women's Histories in Islamic Societies, ed., Sonbol, Amira El Azhary (Syracuse, NY, 2005), 133Google Scholar.

43. African slaves in Bursa inventories are identified as “Arab,” “Mağribi (northwest African),” and “Habeşi (Ethiopian).” West Eurasian slaves are “Rusi (Russian),” “Çerkes (Circassian),” “Gürci (Georgian),” “Kazak (Cossack),” “Tatar,” “Bosnevi (Bosnian),” “Macari (Hungarian),” “Frenk (west Europian),” and “Arnavud (Albanian).” Color-based designations are “zenci (black),” “ak kul,” or “beyaz gulam” (white slave). The latter are from Manisa inventories. Manisa Court Register Nos. 6, 99; 12, 106; 230, 40–45.

44. Jennings, Ronald C., “Black Slaves and Free Blacks in Ottoman Cyprus,” Journal of the Economic and Social History of the Orient 30 (1987): 294–295CrossRef Google Scholar; Sak, İzzet, ‘Konya'da Köleler (16. yüzyıl sonu- 17. yüzyıl),” Osmanlı Araştırmaları 9 (1989): 164–66Google Scholar; Establet and Pascual, La gent d’État, 113; Wilkins, Charles. L., “A Demographic Profile of Slaves in Early Ottoman Aleppo,” in Eurasian Slavery, Ransom and Abolition in World History, 1200–1860, ed. Witzenrath, C. (New York, 2016), 231, 237–38Google Scholar; Walz, Terence, Trade between Egypt and Bilad as-Sudan, 1700–1820 (Paris, 1978), 208Google Scholar; Toledano, The Ottoman Slave Trade, 64–67.

45. Murray Gordon, Slavery in the Arab World (New York, 1992), 81; Erdem, Slavery, 19–23, 40–49.

46. On the increasing share of the African slaves, see Toledano, Ehud R., “Shifting Patterns of Ottoman Enslavement in the Early Modern Period,” in The Ottoman Middle East, Studies in Honor of Amnon Cohen, eds., Ginio, Eyal and Podeh, Elie (Leiden; Boston, 2014), 201–20Google Scholar.

47. Sahillioglu, “Slaves,” 98–104; Compare Zübeyde Güneş Yağcı, “The Black Sea Slave Trade According to the Istanbul Port Customs Register, 1606–07, in Eurasian Slavery, Ransom and Abolition in World History, 1200–1860, ed. C. Witzenrath (New York, 2016), 212–13.

48. Ergin, Nihat, “Köle,” Diyanet Vakfı İslam Ansiklopedisi (Istanbul, 2002), vol. 26: 246–48Google Scholar.

49. Austen, Ralph A., “The Islamic Red Sea Slave Trade: An Attempt at Quantification,” in Proceedings of the Fifth International Conference on Ethiopian Studies, ed. Hess, Robert L. (Chicago, 1979), 461–64Google Scholar; idem, “The Trans-Saharan Slave Trade,” 34–35, 62–63; Wright, John, The Trans-Saharan Slave Trade (London; New York, 2007), 168CrossRef Google Scholar. Trans-Saharan, east African traffic as well as Red Sea traffic appears on the rise from the 1780s. Ricks, “Slaves and Slave Traders,” 63–67; Montana, Abolition of Slavery, 43–44; Lovejoy, Paul E., Transformations in Slavery: A History of Slavery in Africa (Cambridge; New York, 2011), 60–61CrossRef Google Scholar.

50. Smiley, From Slaves to Prisoners.

51. Johnson, Soul by Soul, 155; Davis, David Brion, “Looking at Slavery from Broader Perspectives,” The American Historical Review 105/2 (2000), 457–59, 463CrossRef Google Scholar.

52. McKee, “The Familiarity of Slaves,” 506–07; Campbell, Gwyn, The Structure of Slavery in Indian Ocean Africa and Asia (London, 2004), xviiiCrossRef Google Scholar; Johnson, Soul by Soul, 138–61. Baer, “Slavery in nineteenth-century Egypt,” 418; Lewis, Race and Slavery, 88–89, 137 fn.24. As far as legal marriage with black slaves was concerned, social suitability rules (kafaa) in Islamic law applied.

53. Mustafa ‘Ali, The Ottoman Gentleman of the Sixteenth Century: Mustafa Âli's Mevâ’idü’n-nefâ’is fî Kavâ’idi'l-mecâlis; “Tables of Delicacies Concerning the Rules of Social Gatherings”, trans. Douglas S. Brookes (Cambridge, MA, 2003), 28–30, 102–05. Ali also distinguishes between Ethiopians and other Africans. The distinction was alive and reflected in prices as late as the nineteenth century. Also Hoyland, Robert, “Physiognomy in Islam,” Jerusalem Studies in Arabic and Islam 30 (2005) 361–402Google Scholar; Sobers-Khan, Nur, “Firasetle nazar edesin: Recreating the Gaze of the Ottoman Slave Owner at the Confluence of Textual Genres,” in Well-connected Domains: towards an Entangled Ottoman History, eds. P. Firges et al. (Leiden, 2014), 93–109; Lewis, B., Race and Slavery in the Middle East, An Historical Inquiry (Oxford; New York, 1990), 43–49Google Scholar, 54–61, 92–98; Gordon, Slavery, 56–57, 61; Fisher, “Chattel Slavery,” 41; Baer, “Slavery in Nineteenth Century Egypt,” 419.

54. Seven out of eight slaves observed in Bursa in 1860–1880 are Circassian while there are none so identified in the preceding period. BCR No. C-122 MŞH.ŞSC.d 3144, 84b-87b; C-112 MŞH.ŞSC.d 3134, 72a-73b. Gordon, Slavery, 81. For increase in concubinage in Egypt due to abundance of African slaves in the same period, see Kenneth M. Cuno, “African Slaves in Nineteenth-Century Rural Egypt: A Preliminary Assessment,” in Race and slavery in the Middle East, 93. On the decline of polygamy as a parallel process, see Duben, Alan and Behar, Cem, Istanbul Households: Marriage, Family, and fertility, 1880–1940 (Cambridge; New York, 1991), 149–50Google Scholar.

55. Baer, “Slavery in Nineteenth Century Egypt,” 438–39.

56. A very similar combination of cheap female and child labor and indenture formed the background of the decline of slavery in Renaissance Italy. See McKee, “Domestic Slavery,” 320.

57. Aytekin, E. Attila, “Cultivators, Creditors and the State: Rural Indebtedness in the Nineteenth Century Ottoman Empire,” The Journal of Peasant Studies 35: 2 (2008): 292–313CrossRef Google Scholar; Kaya, Alp Yücel, “On the Çiftlik Regulation in Tırhala in the Mid-Nineteenth Century: Economists, Pashas, Governors, Çiftlik-holders, Subaşıs, and Sharecroppers,” in Ottoman Rural Societies and Economies, ed. Kolovos, Elias (Rethimnon, 2015), 333–79Google Scholar; idem, “Were peasants bound to the soil in the large estates (çiftliks) of the XIXth century Balkans? Reappraisal of the Question of new/second serfdom in the Ottoman historiography,” 3rd International Conference of Greek Economic History Association in Economic and Social History, May 2017. See Stanziani, Alessandro, “Introduction: Labour, Coercion, and Economic Growth in Eurasia, Seventeenth–Early Twentieth Centuries,” in Labour, Coercion, and Economic Growth in Eurasia, 17th–20th Centuries, ed. Stanziani, A. (Leiden; Boston, 2013), 19–20CrossRef Google Scholar regarding lack of correlation between labor scarcity and coercion.

58. Gerber, Haim, “Social and Economic Position of Women in an Ottoman City, Bursa, 1600–1700,” International Journal of Middle East Studies, 12, no. 3 (1980): 231–44CrossRef Google Scholar; Faroqhi, Suraiya, “Guildsmen and handicraft producers,” in The Cambridge History of Turkey: Volume 3: The Later Ottoman Empire, 1603–1839 (Cambridge, 2006), 343Google Scholar; Quataert, Donald, “Ottoman Manufacturing in the Nineteenth Century,” in Manufacturing in the Ottoman Empire and Turkey, 1500–1950, ed. Quataert, Donald (Albany, NY, 1994), 87–121Google Scholar.

59. İrfan Kokdaş and Yahya Araz, “In Between Market and Charity: Child Domestic Work and Changing Labor Relations in Nineteenth-Century Ottoman Istanbul,” ILWCH, in this issue; idem, “The Changing Nature of the Domestic Service Sector in Nineteenth-Century Istanbul: Female Slaves and Wage-Earning Girls,” unpublished paper. We are grateful to the authors for sharing their work in progress with us. On foster-maids, see also Ginio, Eyal, “Living on the Margins of Charity,” in Poverty and Charity in the Middle Eastern Contexts, ed. Ener, Mine, Singer, Amy and Bonner, Michael (New York, 2003), 165–84Google Scholar; Gürer, Ahmet Şamil and Bay, Abdullah, Osmanlı Toplumunda Beslemelik Kurumu (Istanbul, 2013)Google Scholar; Maksudyan, Nazan, “Foster-Daughter or Servant, Charity or Abuse: Beslemes in the Late Ottoman Empire,” Journal of Historical Sociology 21/4 (2008): 488–512CrossRef Google Scholar.

60. Compare Ehud R. Toledano, Ottoman Slave Trade and its Suppression, 1840–1890 (Princeton, NJ, 1982), 62–66, according to which the pay-back period for an African slave in Istanbul in the 1860s and 1870s was 6–7 years at 2,000 guruş per slave.

61. Faroqhi, “Guildsmen and handicraft producers,” 336–54; Pamuk, İstanbul ve Diğer Kentlerde 500 Yıllık Fiyatlar ve Ücretler (Ankara, 2000), 77–82.

62. Stanziani, “Introduction,” 1–26.

63. Blumenthal, Enemies and Familiars, 154–93; Johnson, Soul by Soul, 109–15.

64. Artan, Tülay, “Mahremiyet: Mahrumiyetin Resmi.” Defter 20 (1993): 91–115Google Scholar; Quataert, Donald, ed., Consumption Studies and the History of the Ottoman Empire, 1550–1922: An Introduction (Albany, NY, 2000)Google Scholar; Suraiya Faroqhi, “Consumption and Elite Status in the Eighteenth and Nineteenth Centuries: Exploring the Ottoman Case,” in Stories of Ottoman Men and Women, ed. Suraiya Faroqhi (Istanbul, 2000); Grehan, James, “Smoking and ‘Early Modern’ Sociability: The Great Tobacco Debate in the Ottoman Middle East (Seventeenth to Eighteenth Centuries),” The American Historical Review 111 (2006): 1352–78CrossRef Google Scholar; Hamadeh, Shirine, The City's Pleasures: Istanbul in the Eighteenth Century (Seattle & London, 2007)Google Scholar; Sajdi, Dana, The Barber of Damascus: Nouveau Literacy in the 18^th-Century Ottoman Levant (Stanford, 2013)Google Scholar.

65. Toledano, Ehud R., Slavery and Abolition in the Ottoman Middle East (Seattle, 1998), 129–34Google Scholar; idem, “Late Ottoman Concepts of Slavery (1830s-1880s),” Poetics Today 14 (1993): 477–506CrossRef Google Scholar; idem, “Abolition and Anti-slavery in the Ottoman Empire: A Case to Answer?” in A Global History of Anti-slavery Politics in the Nineteenth Century, eds. Mulligan, William and Bric, Maurice (London; New York, 2000), 117–36Google Scholar.

66. Smiley, From Slaves to Prisoners, 171, 181, 236; Clarence-Smith, Islam and the Abolition, 47–65, 151–76; Wilkins, “A Demographic Profile,” 228 suggesting a link between the substantial increase in manumissions in seventeenth-century Aleppo and the Kadızadeli movement, which points to the complex ways by which sunni “orthodoxy,” even a version of salafism, could work to limit slavery.

67. Levtzion, Nehemia and Voll, John O., Eighteenth-century Renewal and Reform in Islam (Syracuse, NY, 1987)Google Scholar; Schulze, Reinhard, A Modern History of the Muslim World (London; New York, 2000)Google Scholar; idem, “Is there an Islamic modernity,” in The Islamic World and the West. An Introduction to Political Cultures and International Relations, ed. Hafez, Kai (Leiden, 2000), 21–32Google Scholar; Küçük, Harun, “The Case for the Ottoman Enlightenment: Natural Philosophy and Cosmopolitanism in Eighteenth-Century Istanbul,” Perspectives on Europe 42 (2012): 108–10Google Scholar.

68. Eltis, David and Engerman, Stanley L., “Dependence, Servility, and Coerced Labor in Time and Space,” in The Cambridge World History of Slavery, vol. III, AD 1420-AD 1804, Eds. Eltis, D., Engerman, S. L. (Cambridge, 2011), 20–21CrossRef Google Scholar.