Introduction

What can we say about agricultural slavery in the early Islamic Empire? The answer to this question has been largely defined by legal sources, partly because their theoretical discussions come closest to offering a legal definition of slavery and set of normative rules for how it should work,¹ and partly because the questions recorded in the compendia contain more information about the socio-economic role of enslaved persons than historical narratives offer. Based on the references to employment of slaves in legal texts, historians have tended to view slavery as playing a less important role in agricultural production than other forms of labour and to see large-scale slavery as almost irrelevant to the rural context.

This study uses the western province of Ifrīqiya as an example to show that historians’ reliance on legal sources leads them to miss the relevance of references to slavery in other textual genres, and that other texts suggest not only that slave labour was an important component of agricultural production in the early Islamic Empire, but that large-scale slave labour, involving tens or even hundreds of men at a time, was not as uncommon as supposed. Although legal texts are essential for understanding how slaves were employed in domestic and commercial contexts, I will argue here that they offer less insight into agricultural slavery, because the legal texts that have survived were produced in urban contexts and reflect urban concerns. The jurists whose answers are recorded in the compendia are also unlikely to have been consulted by the very wealthy elite, so that questions relating to large-scale military or agricultural slavery do not feature in these texts. Given the restrictions of their context and function the legal texts should not be seen as representing agricultural slavery comprehensively, meaning that historians should pay more attention to the references in other textual genres. Although the nature of these genres can mean that information is distorted or presented in a misleading fashion, this is less likely to be the case for agricultural slavery because of its peripheral relevance for most narratives. Another important factor for this discussion is the extent of continuity between pre-conquest and post-conquest Ifrīqiya in regard to rural labour and administration. Although continuity has been demonstrated for some provinces of the Islamic Empire, such as Egypt, this is not the case for Ifrīqiya, where depopulation before the conquest, Arab acquisition of rural land and easy availability of slave labour together led to a change in labour forms.

This study will begin with a review of the historiography relating to agricultural slavery in early Islamic Ifrīqiya and then show how the context, or socio-economic circumstances within the province changed after the Arab conquest, which makes it unlikely that the land was managed or worked in the same form. I will then analyse the legal sources that have hitherto been used to understand agricultural labour in early Islamic Ifrīqiya, and the information that they offer about the various forms of labour relevant to this context. I will demonstrate the limitations of these sources before considering information contained in other text genres. On the basis of this information, I will argue that in addition to small-scale slavery playing an essential role in all agricultural production, large-scale slavery was also used for commercial production, albeit rarely. Where it is possible to establish the origin of enslaved people used for agricultural labour, they appear to have come from sub-Saharan Africa and likely arrived in the province via the trading routes operated by Ibadi North Africans.

Like every province of the Islamic Empire, and al-Andalus, IfrÄ«qiya was characterized by particular circumstances that shaped the way that the institution of slavery developed after the Islamic conquest. Not only the Roman background and the nature of the Arab conquest, but also its geographical location between the Mediterranean and the sub-Saharan trade routes influenced the development of slave trade and labour in the province. However, IfrÄ«qiya’s distinctive features were arguably not more distinctive than those in other provinces, and the institution of slavery as reflected in written sources from the first three centuries after the conquest shows remarkable correspondence with slavery in other parts of the Islamic Empire.² Therefore, although the breadth of sources available for understanding slavery in early Islamic IfrÄ«qiya means that this region is the focus of this article,³ the findings and argument are also relevant for other provinces in the same historical period.

1 Defining Agricultural Slavery

It is important to specify at the outset that in this study the term agricultural slavery is used to cover all forms of slave labour relating to agricultural production, including but not limited to parallels with plantation slavery models in other contexts. Plantation slavery, in which large numbers of enslaved people work on raising a single crop such as sugar, cotton or indigo, is most typically known from farms in southern US states, Brazil and the Caribbean.⁴ A similar labour model has also been suggested for Roman latifundia, especially in Italy and Greece between the first century BCE and the second century CE, and to a much lesser extent in North Africa.⁵ Although the model of plantation slavery is relevant for early Islamic IfrÄ«qiya, this study will show that most references to agricultural slavery from this region involve fewer people and non-commercial agriculture, and that there was little differentiation between the tasks performed by enslaved people and by free labourers. However, as the use of slaves in 4th-century Athens and in the northern US colonies have shown, slavery is by no means incompatible with small-scale, near-subsistence level agriculture, nor with the co-existence of an independent peasantry.⁶ References to agricultural slavery in this study are to the use of enslaved persons in agriculture regardless of numbers, work type or production scale.

2 Historiography

Historians do not agree about the extent of slave labour used for agricultural production in early Islamic IfrÄ«qiya, or the early Islamic West more generally. For four decades the most detailed contribution to the topic was an article by Mohamed Talbi published in 1981, in which Talbi argued that slave labour from the Mediterranean played an essential role in the province’s agricultural economy and that the decreasing availability of enslaved people from this source after the mid-4th/10th century was a key factor for the region’s decline.⁷ His article continues to be cited, defended and disputed in contemporary studies, despite the fact that it is now over 40 years since his arguments were first published.

Talbi’s model of agricultural slavery has been repeated in some studies,⁸ but it has been criticized by Kurt Franz and more recently Chris Wickham who both view the evidence as too thin to support his conclusions. They argue that the fact that legal sources make so little mention of agricultural slavery suggests that the concept was not widespread in the early Islamic economy, and that many of the references on which Talbi relies refer to dependent cultivation rather than slavery.⁹ Both Wickham and Franz refer to oppressive sharecropping contracts as the main form of labour production in early Islamic IfrÄ«qiya.¹⁰ If this were the case, the Arab administration of rural land would have been largely a continuation of Byzantine practices before the conquest. As the Arabs who conquered Egypt adopted this approach, changing the Byzantine administrative system incrementally rather than radically,¹¹ it is not unreasonable to assume that they applied a similar practice in IfrÄ«qiya too.

From 2022 more detailed studies of slavery in the rural Maghrib, or Islamic west, approached the question from different standpoints. Khālid Ḥusayn MaḥmÅ«d’s book developed arguments made by Ê¿Abd al-Ilāh Binmalīḥ to highlight the variety of agricultural tasks performed by enslaved people throughout the Maghrib,¹² while Amar Baadj’s article, focusing specifically on IfrÄ«qiya, emphasized the mixed nature of the agricultural labour force, which included hired workers and peasants as well as slaves.¹³ Mohamed Talbi, Khālid Ḥusayn MaḥmÅ«d and Amar Baadj rely (although not exclusively) on legal sources for their analysis and it is noteworthy that Kurt Franz’s criticism of Talbi’s arguments is also based on a concept of slavery drawn from legal sources and refers to legal sources for evidence. Legal texts are an excellent source for understanding how enslaved people were integrated into certain spheres of early Islamic society and for researching socio-economic history in a more general sense. They usually consist of questions relating to a dilemma or conflict that were put to a jurist and which are followed by the jurist’s answer. As such, the questions can offer insights into the interactions, concerns and practices of the society in which the jurist worked, and which are rarely found in historical texts.¹⁴ A reliance on legal texts to approach the question of slavery is therefore a widespread and helpful approach. However, as will be shown below, this approach also has some limitations.

3 Context: Early Islamic Ifrīqiya, ca. 50/670–256/870

It was noted in the introduction that one of the grounds on which Talbi’s argument for agricultural slavery has been rejected is the assumption that rural administration continued after the Muslim conquest in much the same way as it did before the Muslim conquest, meaning that if tenancy farming was the main form of labour in the 6th century,¹⁵ it would also have been the main form of labour in the late 1st/7th and 2nd/8th century too. This argument is not unreasonable, because in the province of Egypt, which the Arabs conquered immediately before invading North Africa, rural administration did indeed remain intact, with only gradual changes. Most of the Arabs remained in the garrison settlement of Fustat and rural land continued to be managed by the monasteries and local inhabitants, who paid a tax to the Muslims and who were not required to convert to Islam.¹⁶ However, the situation was not the same in IfrÄ«qiya. Although land allocations within the garrison settlement of Kairouan were made for some of the troops,¹⁷ it is also clear that others received property in rural regions.¹⁸ It is also clear that North Africans who owned land converted and affiliated themselves with an Arab tribe at the time of the conquest, possibly in order to keep their land.

Arab acquisition of land is indicated by some of AbÅ« Bakr al-Mālikī’s biographies in his Riyāḍ al-nufÅ«s, which describe individuals descended from prominent generals as owning large tracts of land. One example is the religious scholar and judge Ibn Ghānim (d. 191/806),¹⁹ who is described as owning an estate in al-Raydān.²⁰ He had inherited this from his father, who had been one of the leading generals in the Battles of al-Qarn and al-Aá¹£nām (both in 124/741–2).²¹ Other examples are Ê¿AlÄ« b. Aslam al-JabanyānÄ«, whose father, Aslam al-JabanyānÄ«, had been a leading general in the conquest of IfrÄ«qiya and who left numerous land estates (manāzil) outside Kairouan to his son, in addition to land and buildings within the city,²² and Muḥammad b. MashrÅ«q, who owned a large agricultural estate on the way to Sousse. His father MashrÅ«q had been a key military leader under MÅ«sā b. Nuá¹£ayr.²³

Some of the estates, such as that of MashrÅ«q, are described as having been received from the leader as rewards for the generals’ loyalty or prowess in battle.²⁴ However, other land grants were given to encourage settlement in regions vulnerable to reconquest. Along the coast and in border regions tribes were settled in small forts or ribats. These tribes were supplied with inalienable land packages, known as ḥimā for the ribats, which could be used for farming or animals and which were an important source of wealth for their inhabitants.²⁵

There is also evidence that African landowners converted to Islam around the time of the Arab conquest. For example, the scholar and landowner Ê¿AlÄ« b. Ziyād al-TÅ«nisÄ« (d. 183/799) was descended from a local, non-Arab family close to Tunis, as his nisba al-TÅ«nisÄ« indicates. His family is said to have converted two generations before his birth, which would place their conversion around the time of the Arab conquest. The family of the wealthy landowner Ê¿Isā b. MiskÄ«n al-IfrÄ«qÄ« (d. 296/908) were also Romanized North Africans,²⁶ who are described as having converted shortly after the conquest and as having attached themselves to the prestigious Quraysh tribe by establishing a client relationship (walāʾ) with a member of this tribe. Other examples include AbÅ« IbrāhÄ«m Isḥāq al-IfrÄ«qÄ« (d. 266/880), who owned substantial tracts of land and whose family converted in the early 2nd/8th century, or Muḥammad b. Ê¿AbdÅ«s, who is also described as being of North African descent, and owning a small village which brought him considerable wealth.²⁷ Like Ê¿Isā b. MiskÄ«n, Ibn Ê¿AbdÅ«s’ family is described as clients of the Quraysh tribe and it is likely that their conversion and affiliation with Qurayshi Arabs enabled them to keep their property under Muslim rule. There are references to land owned by Arabs having been uninhabited at the time of the conquest,²⁸ and it is likely that many Byzantine landowners took advantage of the proximity of Byzantine territory such as Sicily or Sardinia to leave Africa Proconsularis as the Arabs arrived. The unsettled political and economic situation before the conquest had also led to depopulation and abandonment of land.²⁹ However, in addition to ownerless land, discussions in legal texts also mention examples of land being taken by force.³⁰ Conversion, or affiliation with an Arab general, would have been one means by which wealthy inhabitants tried to avoid this.

It is likely that the end of the 1st/7th and the first half of the 2nd/8th century saw little change in the rural landscape of IfrÄ«qiya, even after the change of ownership. During this period Arab troops made more income from military campaigns in the Maghrib and the Mediterranean than they would have done from farming. Most descriptions of the western conquests focus on the high numbers of people enslaved from the region, with figures such as 100, 000 being used to describe the captures of a single raid.³¹ There are also extravagant descriptions of the captives’ exotic appearance and the high prices that they could reach.³² However, in 122/740 a revolt that began in Tangier spread to other regions in the Maghrib, causing disturbance and eventually leading to the expulsion of the Arab rulers. Even the centre of Arab settlement, Kairouan, was taken by the North African Warfajuma tribe in 140/757.³³ Abbasid troops recovered the city in 144/761 and Arab rule was restored over the province of IfrÄ«qiya. But the rest of the Maghrib remained outside caliphal control. No subsequent attempts to recover control over these regions or to enslave their inhabitants are mentioned by the historical texts.³⁴ The routes that had been used by the Arabs to capture and transport enslaved people continued to function, but they were now largely managed by North Africans who had adopted a version of Islam that had been propagated by opponents of the Umayyad rulers rather than the proto-Sunni version associated with the governing elite. Although the religious landscape at this time was clearly fluid and is difficult to recreate from the sparse references in the sources,³⁵ it appears that most of the North Africans adhered to what came to be known as Sufrid or Ibadi interpretations of Islam and that this shared interpretation strengthened existing bonds between the tribes.³⁶ The bonds between separate Ibadi and Sufrid tribes in North Africa facilitated a long-distance trading network with links into sub-Saharan Africa, which became a major source for slaves.³⁷ Most of the enslaved people were transported east but it is clear that many were also sold in IfrÄ«qiya.³⁸

With the end of Arab expansion in the region, and trans-African slaving now under the control of non-Arab tribes, farming became the main source of income for the Arabs who had settled in the province. This was especially the case after the Aghlabid rulers, who governed the province from 184/800 onwards, decommissioned many members of the jund or Arab military, meaning that they no longer received payments from the dÄ«wān or administrative office of the ruler. Denser settlement and more reliance on farming is reflected in the archaeological evidence, which indicates higher production and better connectivity between rural regions in the 3rd/9th century.³⁹

Muslim ownership of rural land did not automatically mean a change in how the land was worked. But it did have administrative consequences. The first of these was the altered tax status of the land. Technical terms for different categories of land ownership, such as qatÄ«Ê¿a, ḍayÊ¿a and ḥimā, are mentioned for IfrÄ«qiyan estates in the later sources, such as the al-Nawādir of Ibn AbÄ« Zayd al-QayrawānÄ« (d. 386/996) and al-Dāwudī’s (d. 402/1011) Kitāb al-Amwāl. They also appear in the 8th/14th-century historical chronicles of Ibn Ê¿IdhārÄ« and other authors. However, the earliest surviving legal compendium, the Mudawwana of SaḥnÅ«n b. SaÊ¿Ä«d, does not use these terms and instead refers to all estates with the more general term qarya. This may indicate that the more precise definitions were not developed until later, in the context of the debates caused by land redistributions under the Fatimids.⁴⁰ However, it is clear from the discussions in the 3rd/9th-century sources that rural estates under Muslim ownership were subject to the lower Ê¿ushr tax rate and that those under non-Muslim ownership were subjected to the higher kharāj tax. The estates owned by Muslims generated less income for the state and more income for the owner.⁴¹ The wealth of the owners, that could translate into political power, can be seen as a factor for political instability in the 3rd/9th century.

Another consequence of Muslim ownership was that the property was now subject to Islamic inheritance laws, meaning that it had to be divided among the heirs of the deceased. The dilemmas that this division could cause are mentioned often in the legal compendia and reveal a lot about the size of estates and the kind of products that were grown on them.⁴² They also show that the successive division of an estate could lead to the impoverishment of the land, as the heirs attempted to derive a living from diminishing property parcels.⁴³

There were legal means for maintaining the integrity of a deceased person’s property in Islamic law. They included the right to pre-emptive purchase, known as shufÊ¿a, which other heirs could assert if one of them decided to sell his or her share after inheritance. Another was the owner’s right to bequeath a portion of his or her property to an individual while still alive. The bequest (waṣīya) could not be more than a certain proportion of the entire wealth of the owner. Another mechanism was the waqf, or the ḥabs, by which the family could no longer claim ownership of the property, but instead shared the rights to the income obtained from it.⁴⁴ References to owners establishing a ḥabs or waṣīya for their landed properties are rarer in the legal compendia than ḥabs for other forms of property such as clothing or slaves and so it is difficult to assess how far these mechanisms were used to maintain the integrity of estates.⁴⁵ However, it seems likely that many large estates were divided into smaller estates in the course of the 3rd/9th century, which would have reduced possibilities for large-scale commercial farming and may have affected the kinds of crops that were grown.

A further consequence of Muslim ownership of these lands is that the proprietors had access to a wider variety of slaves than their non-Muslim contemporaries did. Islamic law prevented non-Muslims from owning Muslim slaves and also imposed restrictions on non-Muslims purchasing slaves that could be owned by Muslims.⁴⁶ Although most of the slaves arriving from the Mediterranean region were not Muslim, people from the regions to the south of IfrÄ«qiya often were, which would have made their purchase by non-Muslims difficult.⁴⁷ It is likely then, that Muslim landowners who wanted to use slave labour had an advantage over non-Muslim landowners, because the latter were only permitted to own Jewish or Christian slaves and these were not as widely available. In general it appears that there was a higher supply of slave labour after the Arab conquest then there had been under Byzantine rule, due to the raids that the Arabs conducted in the west and then the management of the slave trade routes by North African tribes.

4 What do the Sources Tell Us about Agricultural Slavery in Early Islamic Ifrīqiya?

Of all the literary sources, legal texts contain the most information about slaves and slave labour and have consequently been invoked more often than other text categories to understand agricultural slavery in IfrÄ«qiya. Most of the texts were compiled later than the period under study but some, such as the Mudawwana of SaḥnÅ«n b. SaÊ¿Ä«d, are held to have been compiled in the 3rd/9th century while others contain rulings and teaching attributed to persons from this historical context.⁴⁸

Most of the discussions about slaves and slave labour in the legal texts refer to domestic or commercial slavery. Topics such as sexual relations with slaves, the emancipation of slaves and commercial transactions involving individual slaves occur frequently, but there are fewer references to slaves performing agricultural labour or being purchased for these tasks. When these references do appear, they indicate that slaves were part of a varied labour force and worked alongside free labourers. Employers do not seem to have distinguished between the kinds of tasks assigned to enslaved or free labourers, or to have regarded slave labour as better or worse than hired work.⁴⁹ For example, in a query to SaḥnÅ«n about a free man hired to watch sheep who passes the task to one of his slaves,⁵⁰ neither the questioner nor SaḥnÅ«n distinguishes between the status of the two labourers. The issue at stake is that the contract specified a specific shepherd, who was then substituted by another. Slaves are described as assisting with ploughing and irrigating, overseeing orchards,⁵¹ transporting wood, guarding fields, maintaining irrigation systems,⁵² and watching sheep⁵³ but free men are also described as carrying out these tasks, in some sources.

References to slave numbers are quite rare in the queries, but where they do appear, it seems that some landowners owned several slaves and did not attach much value to individuals, whereas people without land held fewer slaves and did attach more value to them. For example, a man who has used his slave to rent land asks what to do if he now needs his slave back, suggesting that he is not able to do without his slave’s help.⁵⁴ But when discussing the facilities that should be provided by a landowner SaḥnÅ«n describes a slave and work animal as “something small and easy to provide” (in contrast to a bucket),⁵⁵ which suggests that the landowner held enough slaves and animals to regard one or two as expendable.

The equivalence between the slave and the farm animal in the previous example is quite typical for discussions about slaves employed in agricultural contexts. The objectification of the enslaved person is also reflected clearly in discussions about pawns or pledges, in which slaves are treated as units of value, like land and work instruments.⁵⁶ Even when enslaved people performed similar roles to free people, working alongside them, the legal sources make the fundamental difference in their status clear.⁵⁷

Some of the discussions in the legal texts refer to landowners using sharecropping arrangements to work their land. The blanket term for these arrangements in legal compendia from IfrÄ«qiya is musāqa, and other terms signifying different forms of the relation tend not to be used.⁵⁸ The principle of the sharecropping agreement was that one party would provide land or other resources while the other party provided the labour. The product would then be divided depending on the terms of the contract.

The terms of the musāqa contracts mentioned in the legal compendia are not the same as the sharecropping agreements with bound tenants or coloni evidenced from pre-Islamic IfrÄ«qiya. These latter were usually long-term arrangements and coloni who left their land could be punished or returned to their original estates.⁵⁹ By contrast, most of the musāqa contracts specify a duration of one to four seasons and a specific land form (orchard, field etc.), making them more suitable for small and medium-size farmers who were renting out some of their land on a short-term basis. They were also concluded between two individuals and tailored to the capacities of both parties, making the musāqa contract an impractical means of managing a large estate with several farmers. Another indication is that the musāqa contract could be made far from the land to which it applied; one of the rulings in the Mudawwana specifies that if the contractor describes the land accurately, then the musāqa contract remains valid.⁶⁰ This question and specification does not fit well to the notion of peasant tenants who were bound to an estate.⁶¹

Although a minority of the musāqa agreements refer to inhabited land allocations, where the residents would do the work,⁶² most of the discussions stipulate that the person taking on the land had to bring his own labourers, or work the plot himself. This might suggest that musāqa contracts were used to overcome labour shortage problems arising from the Arabs’ cultivation of new land or land that had been abandoned by tenant farmers in the Vandal and Byzantine period. It is notable that the contracts are for the short term and do not provide for permanent or semi-permanent use of the land.

In addition to slave labour, the legal texts also reflect the importance of hired labour to farm the land. Hired labour was widespread in IfrÄ«qiya before the Arab conquest and was provided by both local workers and men who entered the region on a seasonal basis.⁶³ In the early Islamic period waged labourers (sing. ajÄ«r, plu. ujarāʾ) provided domestic and commercial services,⁶⁴ but many of the queries mention agricultural tasks such as irrigation or ploughing,⁶⁵ suggesting that these labourers were particularly important for the agricultural sector.

In some cases the discussions in the compendia indicate that the hired labour was performed by a slave rented out by his owner.⁶⁶ But in most cases the identity of the ujarāʾ is not clear. In his description of the revolt of the Kutama against the Arabs in Belezma, al-NuwayrÄ« (d. 733/1333) mentions that the Kutama accused the Arabs of treating them as slaves,⁶⁷ which suggests that tribes living in the regions surrounding Arab territory also provided labour at some points in the year, and that their treatment or payment could be a source of conflict. It is also interesting to see the term slave being used in a metaphorical sense by al-NuwayrÄ« to mean exploitative or abusive treatment by the employer, at least in the 7th/14th century.

In some cases Arab landowners also appear to have worked their land themselves, as well as helping one another on an informal basis.⁶⁸ Other queries indicate that men could contribute labour for agricultural tasks without payment. This is reflected by a query put to Ibn SaḥnÅ«n about the employer who provides food to men helping him with the harvest; should the food be distributed differently between the waged and the unwaged labourers?⁶⁹ Ibn SaḥnÅ«n’s answer suggests that sometimes food was used as payment, but that it was not uncommon for men to help with the harvest without a formal arrangement for compensation.

Labour of the landowner and his family, reciprocal help and the informal assistance of neighbours would have been more relevant for smaller land plots which did not require constant and intensive cultivation. Indeed, most of the discussions in the legal compendia seem to refer to quite small plots of land, such as orchards or single fields. It is likely that the queries to which they respond relate to Kairouan’s “green belt” or the land held by the city’s residents just outside the city, which they used for subsistence or low-level commercial farming.⁷⁰ This land was made more fertile by the investment in irrigation works by the Aghlabid rulers and played an important role in sustaining the population of the city.⁷¹ The irrigational infrastructure still existed in the 8th/15th century, when some of its sub-sections were described by al-ḤimyarÄ« (d. 900/1495) in his al-Rawḍ al-Miʿṭār.⁷² The legal sources show well how these smaller agricultural properties were integrated into the life of the city, and how their cultivation could be combined with other activities such as scholarship or trade.

However, the green belt around the city was not the main agricultural engine for the province as a whole, and its small farms were unlikely to have been able to supply the city with all of its needs. A significant proportion of agricultural production was located in rural regions further away from the city, even if the urban system determined the form and extent of this production. Geographers who visited IfrÄ«qiya extolled the small and medium-sized farms peppering the countryside and the historical texts refer to large estates owned by the elite, which were well outside the reaches of the city. In addition to the examples mentioned previously, the estate owned by Manṣūr b. Naá¹£r al-ṬunbudhÄ« is a good illustration of the wealth and political power that agricultural property could bring. al-ṬunbudhÄ« owned a large property known as Ṭunbudh close to Tunis and used this as the base from which he organized a revolt against the Aghlabid ruler Ziyādat Allāh, in 823. Another example of a wealthy landowner is ʿĀmir b. NāfiÊ¿, who joined al-Ṭunbudhī’s revolt but eventually turned against him.⁷³ The references to these estates show that they were very large and it is possible to speculate, as Talbi did, about whether they were the same estates or villas that had adorned the countryside in the 4th and 5th century under Roman rule.⁷⁴ The sale of their produce to the cities would have provided considerable wealth to the owner as well as food security for the cities.⁷⁵ However, despite their size and economic significance, there are no queries in the legal compendia referring to estates of this nature, and neither do the queries refer to practices that would have been relevant to the medium-sized farms outside the green belt.

Why is the representation of agricultural slavery in the legal texts incomplete? Why do the queries and answers in the legal texts tell researchers so little about labour and slavery in the rural areas outside the green belt of the large cities?

Part of the answer lies in the nature of the texts. As mentioned at the beginning of this study, the queries in legal texts are usually seen as reflecting the social context in which the jurist worked. Most of the jurists whose teaching has been transmitted until today worked in an urban setting, where there was a higher proportion of free Muslim men with the leisure to study than in rural regions. It is also because the archives for storing texts were usually in the large mosques of cities, such as the famous archive of Kairouan,⁷⁶ and so texts produced in these environments were more likely to survive than teaching records made in rural environments.

Because the surviving legal compendia were produced in an urban environment, the questions that they contain reflect urban concerns. With regard to the topic of slavery, the texts give a valuable insight into the ways that enslaved people contributed to the social and economic fabric of the city and its green belt, but they do not show how enslaved people were integrated into rural communities, because these communities are less well-represented in the legal sources. Rather than travelling to an urban jurist to ask his advice, people living in rural communities would have consulted a local authority figure, whose responses, if written, have not survived. Therefore, it is difficult for us to establish how enslaved people were employed in rural regions or what role they played in rural society.

Another reason why legal compendia are less helpful for understanding agricultural slavery, or labour more generally, is because of the social role and status of the jurists. Most jurists in early Islamic IfrÄ«qiya were men of Arab descent, some owned considerable properties, and those close to the ruling circles had significant political and social influence. But their social role, insofar as it is reflected in the compendia of their teaching, consisted in solving the personal dilemmas of free Muslims and mediating in inter-personal conflicts. A typical question is that of the man who buys jewelry and who is then contacted by another man who claims that the jewelry belongs to him.⁷⁷ The jurist’s rulings were not legally enforceable and so for more serious cases the parties would have addressed the qāḍī or judge, whose rulings were enforceable. There are accounts of the elite visiting religious scholars as students but these visits served above all to demonstrate their piety and there is no evidence that they asked the scholars for advice about their estates. Although historians describe the dilemmas and squabbles of the ultra-wealthy in some detail, none of their accounts are echoed in the legal compendia and it is rare to find a context of great wealth for any of the legal queries; either the elite’s queries went unrecorded or they used an alternative channel to solve their disputes. Therefore, the legal compendia cannot give readers an insight into the dilemmas of the ultra-wealthy elite of early Islamic IfrÄ«qiya.

The limitations of legal compendia as a source for understanding agricultural labour are also shown by the information contained in other text genres, which often departs from that provided by the compendia. One example is a reference in a collection of scholarly biographies. In his account about the scholar Muḥammad b. MashrÅ«q, the author al-MālikÄ« describes Ibn MashrÅ«q visiting the estate that he inherited from his father and being greeted by the inhabitants, who lined the paths along which Muḥammad proceeded and called out to him, “we are your slaves, and everything that you see in this village is yours”.⁷⁸ Although similar expressions of loyalty by slaves to their new masters are known from some 19th-century accounts, it seems more likely that this account describes the servile relationship between the tenants bound to these estates and the owner than the behaviour of people who were actually enslaved, as there are no other examples of slaves behaving in this way from the early Islamic context.⁷⁹ Therefore, whereas the legal queries relating to sharecropping cannot be seen as unambiguous proof of bound tenancy, this biographical account does seem to indicate the continuation of the institution after the Arab conquest.

The study of Noel Lenski on agricultural labour forms in Roman North Africa finds that before the conquest, peasant tenancy was more significant than slave labour for agricultural production, except in the 5th–6th century when the Vandal rulers likely neglected the laws of tenancy farming and profited from a surplus of slave labour.⁸⁰ However, many coloni had left their bound tenancies during Vandal rule,⁸¹ and although the Novels of Justinian indicate a concern to strengthen the institution of tenant farming after the reconquest in 533 it is not clear that these measures were successful. It also seems that many tenant farmers left their land after the Muslim conquest. In Ibn Ê¿Idhārī’s explanation of why one Umayyad governor of IfrÄ«qiya was murdered in 102/720 he attributes it to the governor’s forced repatriation of dhimmÄ«s, or the conquered non-Muslims, who had converted to Islam and left their farms for a life in the city.⁸² If this interpretation is correct, then the governor’s practice here can be paralleled with the approach of al-Ḥajjāj b. YÅ«suf to the same problem in Iraq, where the institution of bound tenancy appears to have lost its cohesion after the Arab conquest, perhaps due to the rapid urbanization in these regions.⁸³

Another example, from a historical text, appears to refer to large-scale slavery on rural estates. In Ibn Ê¿Idhārī’s account of al-Ṭunbudhī’s rebellion against the Aghlabid ruler Ziyādat Allāh he describes one of the co-rebels, ʿĀmir b. NāfiÊ¿, as collecting 1000 black men (alf aswad) armed with shovels and picks, to fight for him.⁸⁴ The term for slave (Ê¿abd) does not appear in this account but recent research by Rachel Schine has shown that terms for blackness and enslavement were used interchangeably even at this early date.⁸⁵ Even if the men referred to by Ibn Ê¿IdhārÄ« were free men, or hired labourers, it seems likely that they were the descendants of people who had been brought to IfrÄ«qiya as slaves from a region south of the province. However, the description of ʿĀmir commanding this large number of men to fight for him at short notice makes better sense if we assume that readers were familiar with the notion of very large numbers of slaves working in an agricultural environment who could potentially use their tools for other purposes.

This possibility is also demonstrated by another reference of Ibn Ê¿IdhārÄ«. The historian refers to IbrāhÄ«m b. al-Aghlab confiscating the horses and the slaves belonging to the estates of prominent members of the jund or settled Arab military,⁸⁶ probably because of the threat that their manpower constituted to his rule. IbrāhÄ«m’s measure and Ibn Ê¿Idhārī’s observation that this provoked a rebellion indicates that many members of the jund held large slave forces and were prepared to use these for political ends.⁸⁷

Large-scale slavery is never mentioned in the legal compendia and for this reason historians tend not to see it as widespread in the early Islamic Empire. However, the laconic tone in which the historical author mentions the presence of hundreds of enslaved people on a single estate indicates that his readers would not have been surprised by his account. Like most Islamic historians, Ibn ʿIdhārī’s main focus is on the political narrative of the Islamic Empire and the free Muslim men that constituted its main actors. Except for the exceptional ones that rose to political power, slaves feature only as background information. However, their peripheral role in the narrative is more likely to ensure a realistic portrayal as the historian is unlikely to create a context that would not be credible to his readers.

The historians’ lack of interest in slavery and agriculture is also illustrated by the accounts of the Zanj revolt.⁸⁸ The Zanj rebellion, or rebellions, occurred in southern Iraq between 255/869 and 270/883 and was or were serious enough to challenge caliphal authority for a short period. There has been some debate about whether all of the rebels were slaves and Ghada Hashem Talhami, among others, has argued that many participants came from a wider social stratum.⁸⁹ However, the question of whether other demographic groups were also involved in the revolt is less relevant for this study than the sources’ consensus that it began among slaves that are described as zanj or having an east African origin and who were employed in removing nitrous topsoil so that the land beneath it could be used for farming.⁹⁰ The sources do not give figures for the total number of rebels, but in his account of the revolt al-ṬabarÄ« refers to one holding of 15,000 slaves working on the al-Ahwaz side of the River Dujayl, to over 800 slaves joining the rebel movement,⁹¹ to the defection of 5,000 “black and white” rebels to the Abbasids,⁹² and to the rebel leader Ê¿AlÄ« b. Muhammad selecting 5,000 of the bravest troops, “Zanj and others”, to assault the government forces.⁹³ The repetition of the number 5,000 when referring to troops, rebels and slaves suggests that it was intended less as a precise figure than as a rounded representation of large numbers of men.⁹⁴ Nonetheless, it does convey an impression of how the sources used by al-ṬabarÄ« assessed the numbers of the rebels and how many slaves they thought it reasonable to assume were potential followers of the rebel movement.⁹⁵ Nothing in the way that al-ṬabarÄ« mentions the numbers of labourers or their tasks suggests that he saw anything unusual or interesting about the presence of thousands of enslaved people working in the rural lands outside Basra. The people are only relevant insofar as the rebellion in which they participated threatened the stability of the Abbasid Empire.

To put the point differently: if it were not for the fact that they participated in a rebellion of political significance, the fact that thousands of people were enslaved in East Africa and brought to southern Iraq where they were kept in isolation from other population groups and worked solely on preparing saline land for agriculture would not be known. Similarly, if it were not for the fact (or anecdote) that ʿĀmir b. Nāfiʿ collected an army of agricultural slaves to defend his position in a rebellion, this example of large-scale slavery in early Islamic Ifrīqiya would also have gone unrecorded.

Because of the rarity of such references in historical texts and because there are no references to large-scale slavery in legal texts, on which they usually rely, historians have tended to see the slavery behind the Zanj revolt as an anomaly. But if they take account of Islamic historians’ aversion to mentioning both slaves and rural environments, and the legal texts’ inability to fully represent agricultural society or the wealthy elite, then historians might usefully pay more attention to these incidental references when defining or discussing slavery in the early Islamic Empire.⁹⁶

Military slavery is another example of a form of slavery that historical texts indicate was widespread in the early Islamic Empire, but which is never mentioned in legal queries. Well before its transformation into a state institution under the Mamluks, the practice of employing slaves in a private army was widespread in the Islamic Empire, including IfrÄ«qiya.⁹⁷ The men who worked in these armies were purchased as slaves, even if it became customary to emancipate them after their training, and due to their influence on political events they often feature in historical narratives. However, military slaves and the phenomenon of military slavery are never discussed under the category of slavery in the Islamic compendia.⁹⁸

As with agricultural labour, part of the reason why legal compendia do not mention this category of slavery is the relation between the jurists who compiled the compendia and the people who would have owned the enslaved soldiers, who in this case were caliphs, regional rulers or possibly local governors. As mentioned previously, jurists could wield significant social influence and often played an important role in legitimizing a ruler or his actions. However, the rulings given in legal compendia never relate to a ruler’s decisions. Although scholars developed the institution of siyāsa sharÊ¿iyya as a means of providing guidance for, or legitimizing, state policy, the discourse of this field had little in common with the answers to individual questions asked on a local level.⁹⁹ This distance was compounded by the fact that religious scholars ostensibly mistrusted political power and avoided consorting publicly with its representatives. It would not have helped a scholar’s reputation among his peers had he been seen to be seeking the ruler’s favour or openly offering his friendship. Not only the scholars’ reserve in relation to leaders, but also the leaders’ distance from the circles that consulted jurists, means that if a leader did have a dilemma concerning his ownership of thousands of enslaved men for military purposes, he would probably not have consulted a jurist for advice, or at least not the same jurist that the inhabitants of the larger cities nearby were consulting. Even if a leader had asked a jurist about a sensitive political or economic question the meeting would likely have taken place privately and the details of the question would not have formed part of the teaching corpus that was transmitted to students and which survives in compilations today.

Another reason why large-scale slavery rarely features in legal compendia is because early Islamic jurisprudence was not well-suited to deal with queries of this nature, due to its dependance on notions conceived in the intellectual context of the Qurʾan and hadith. The existence and practice of slavery is well attested in these texts and it is clear that slaves played an important role in pre- and early Islamic Mecca and Medina.¹⁰⁰ But the references in these sources indicate that enslaved people were held in small numbers, and that households with over ten slaves were very rare. Enslaved persons were acquired through war,¹⁰¹ and usually formed part of the household after their acquisition. They are described as performing a variety of jobs, including domestic labour, military assistance and agricultural tasks, usually alongside or in close contact with their owners.¹⁰² Emancipation, which is given a high moral value in the Qurʾan, was a significant act for the owner, given the small number of enslaved persons in a household and the corresponding labour loss that one less slave would entail.

The socio-economic situation reflected in the Qurʾan changed rapidly after the Islamic conquests of the 1st/7th century, with the change most relevant for this discussion being the increased number of enslaved persons now entering Muslim possession. The military victories of the Muslim armies resulted in unprecedented numbers of slaves and land entering the emerging Islamic Empire and beneficiaries of the conquests could own over 1,000 slaves.¹⁰³ The increased accessibility of slave labour, alongside the vast wealth that some Muslims now owned, inevitably changed both the roles of enslaved people and their relationships with their owner. More distance between the owner and his or her slaves, a propensity to see enslaved people as dispensable, and organization into mass labour projects such as construction,¹⁰⁴ military or farming, are some of the changes arising from the influx of so many enslaved people into the empire.

However, these changes are not reflected clearly in Islamic legal discourse.¹⁰⁵ By the early 2nd/mid-8th century Islamic legal texts show sophisticated engagement with certain questions relating to slavery, most prominently the technicalities of manumission and the norms of sexual relations with slaves.¹⁰⁶ But their discourse remains focused on slaves employed in relatively small numbers in households. This is reflected not only in the chapters dedicated specifically to slavery-related issues such as the sections on mukātaba or walāʾ, but also in the questions about slaves scattered throughout the compendia, which usually refer to the purchase of a single person or to relations with less than five slaves. The questions and answers recall both the strong relationship between owner and slave evoked in the Qurʾan and the high moral value of manumission, which must have been less significant for owners of several hundred people. Thus, from an early point we see a discrepancy between the relationships evoked by the legal texts and the development from the late 1st/7th century of mass slave import and forced labour. It is possible that the lack of references to large-scale military or agricultural slavery is also due to the jurists’ reluctance to move outside the scope of slavery evoked by the Qurʾan. Although the development of legal discourse about slavery between the Qurʾanic period and the 2nd/8th century clearly drew on other sources besides these texts,¹⁰⁷ perceptions of what Islamic law could say about slavery may well have limited the kinds of questions that were put to jurists or the kinds of questions that they answered.

Unlike large-scale agricultural slavery, large-scale military slavery is so widely represented in historical texts that it is accepted as fact by contemporary historians and its omission in legal texts is rarely commented on.¹⁰⁸ But if researchers are able to recognize references to other forms of large-scale slavery in historical texts and to appreciate the reasons why legal compendia ignored these forms, why should the case be different for agricultural slavery? This is particularly in view of the fact that manual labour—in contrast to warfare—was not highly valued in early Islamic society; rather, Arab nobles are said to have disdained physical labour and to have avoided it when they could.¹⁰⁹ It would seem logical that the settled Arabs used the high numbers of enslaved people entering their ownership to perform essential tasks that they themselves disliked.

Rather than adapting our notions of Islamic slavery to the discourse of Islamic jurisprudence, it seems more constructive to accept the limitations of the legal sources and to recognize that Muslims engaged enslaved people in ways that were not mentioned by the jurists, in addition to ways that were.

5 What Can We Say about Slavery in Early Islamic Ifrīqiya?

In view of this analysis of what sources do and do not say about slavery, what conclusions can be drawn about agricultural slavery in early Islamic Ifrīqiya? The first observation is that slavery was an important component of agricultural production on both a subsistence level and in a more large-scale sense. Although other forms of labour were also important, sources indicate that landowners relied on enslaved people working alongside other labourers wherever necessary.

Most references to agricultural slavery in legal sources are to the small farms around Kairouan, and tend to refer to slaves in single figures. However, SaḥnÅ«n’s reference to slaves as dispensable and easy to acquire for a landowner suggests that larger slave holdings were standard for some urban landowners. This variation, between large and small slave holdings, is relevant for the smaller estates in more rural regions. Here the system of bound tenancy was continued but changing socio-economic circumstances and the cultivation of new or abandoned lands meant that mobile and short-term labour also played an important role. Enslaved people did not constitute short-term labour in the same sense as waged labourers and their labour was not necessarily cheaper, as they had to be clothed, fed and housed the whole year round.¹¹⁰ But slaves could be exchanged or hired out to other landowners at short notice, and were not tied to a specific land area in the same way that tenant farmers were, which could be an advantage.

There is evidence that large-scale slave forces were used for commercial agricultural production on some of the bigger estates beyond the urban green belt. This is not discussed in the legal sources but it seems likely on the basis of references to large numbers of slaves in historical texts, the influx of enslaved people into IfrÄ«qiya after the Arab conquest, and the shortage of agricultural labour due to migration caused by political disturbance, cultivation of new land, and the low status of manual labour in early Islamic society. However, it is unlikely that large-scale slave labour was the norm in rural regions. The number of large estates is likely to have declined in the 9th century and there are references to Aghlabid rulers confiscating land and slaves owned by the elite because of the political threat that they constituted. The slaveowner’s obligation to feed and clothe his slaves all year round was costly and so large-scale slave labour would only have been tenable in a stable enterprise, where the production required intensive labour throughout the whole year.¹¹¹ In estates such as these, or for rendering dead land farmable, a large holding of enslaved people could have been a preferable option to tenancy farming.

Contrary to Mohamed Talbi’s assertion that the backbone of agricultural slavery was provided by enslaved people drawn from the Mediterranean, it seems more likely that the large-scale slave holdings consisted of people with a sub-Saharan origin. Clearly, significant numbers of enslaved people were entering IfrÄ«qiya via raids and commercial exchange across the Mediterranean. Raids of Mediterranean islands had been a regular phenomenon since the beginning of the Arab conquest and intensified after the Arabs lost their access to the Maghrib.¹¹² Most raids, such as those on Rome, Brindisi, and Ischia, tend not to be mentioned in the Arabic historical texts if they did not lead to permanent settlement,¹¹³ and historians say little about border settlements such as Bari whose relationship to the caliphate was unclear.¹¹⁴ However, the few references that do exist suggest that raids from IfrÄ«qiya and border settlements in the Mediterranean were important sources for slaves, even accounting for the profitable ransom business that led to many enslaved people returning to their homes.

The Arabs in IfrÄ«qiya could also purchase slaves from across the Mediterranean via the trade operating in Europe during this period.¹¹⁵ Latin and Frankish sources, such as the letters of Pope Adrian I (d. 795), the Liber Pontificalis,¹¹⁶ and the account of Bernard the monk,¹¹⁷ refer to significant numbers of people being sold by European traders to Arabs or Muslims, while the Mudawwana of SaḥnÅ«n b. SaÊ¿Ä«d mentions Byzantine merchants selling enslaved people in the markets of IfrÄ«qiya.¹¹⁸ Some of the queries in the Mudawwana describe conversations between the merchants and their customers, suggesting a common language and a mutual understanding of how slave sale in the Muslim province worked.¹¹⁹ Geographical texts such as Ibn Khurradādhbih’s (d. after 911) Masālik also describe the arrival of northern slaves via ship from al-Andalus.¹²⁰ These textual sources indicate that large numbers of enslaved people were entering Africa via the Mediterranean and that although many were sold eastwards, many also stayed in the region. Therefore, it is clear that slaves from the Mediterranean were an important part of the slave trade economy.¹²¹

However, literary references to Byzantine or Italian (rÅ«mÄ«) slaves suggest that they were more often employed for domestic and commercial contexts than for agricultural labour. Al-WisyānÄ« mentions that BulukkÄ«n b. ZÄ«rÄ« (r. 972–984) had a Byzantine slave who managed his date orchards,¹²² but the emphasis in this account is on the privileged relationship that the slave had with the ruler rather than his agricultural work. In his legal compendium Ê¿Abd al-Wahhāb al-BaghdādÄ« (d. 1031) mentions the ruling that a Byzantine slave (rÅ«mÄ«) used for farming tasks may be exchanged for two Byzantine slaves used for other tasks.¹²³ This suggests that Byzantine slaves employed for agricultural work were comparatively rare and that their work was highly valued. Both of these examples are from a later context than the focus of this article and al-Baghdādī’s reference is from the eastern Islamic empire; literary sources from early Islamic IfrÄ«qiya do not mention Byzantine slaves in the context of agriculture at all.¹²⁴

However, they do mention enslaved people from sub-Saharan Africa. Most of the references to labourers in an agricultural context refer to them as aswad, indicating a sub-Saharan background. For example, the agricultural slaves collected by the rebel ʿĀmir b. NāfiÊ¿ are described as aswad, and the agricultural slaves, or descendants of agricultural slaves, noticed by Ibn Ḥawqal (d. 978) on his travels to Cap Bon are also described using this term.¹²⁵ Ibn Ḥawqal does not specify whether the people whom he sees are free or enslaved, but even free inhabitants would have been the descendants of enslaved sub-Saharan people, who were brought to work the rich agricultural area in this region.

Although they are only two examples and not necessarily representative, these references to sub-Saharan men within large-scale slave forces correspond to other evidence showing that enslaved people from sub-Saharan Africa could be acquired in large numbers and for a low price.¹²⁶ This is also indicated by the fact that IbrāhÄ«m b. al-Aghlab was able to build up a private army consisting of 5,000 men from sub-Saharan Africa at the beginning of his reign.¹²⁷ If enslaved people from sub-Saharan Africa were less expensive and easier to obtain than people from across the Mediterranean, then this could have been a reason why they were employed in environments requiring large numbers of slaves, whereas more expensive slaves, which constituted more prestigious possessions, were used for household slavery, trade and diplomacy. This logic assumes that the obligation for providing food and clothing for the enslaved people entailed similar costs for both population groups.¹²⁸

Conclusion

A higher reliance on legal sources than on other kinds of written texts has led researchers to underestimate the relevance of slavery to agricultural production in early Islamic Ifrīqiya. Because of the context in which the legal texts were produced and the social role of the jurist, the queries in the legal compendia do not represent the rural population comprehensively and neither do they represent the wealthy elite who owned large estates. It is likely that most of the references to agricultural production in the legal compendia refer to the green belt around Kairouan which, although a region of genuine agricultural production, was tied to the urban population and cannot be seen as representing rural society in a wider sense.

By inferring from discussions about green belt farming, and using other literary sources such as historical and biographical texts it is possible to recognize slavery as important for agricultural production in rural regions beyond the green belt. However, enslaved people formed part of a variegated labour force and there was no distinction between the tasks that they performed and those of free people. Examples of large-scale slavery, in which enslaved people were used for commercial agricultural production in a manner similar to their employment in plantation slavery, are present in written sources but are comparatively rare. Large-scale slave holdings probably became rarer as Islamic inheritance law and rulers’ policies led to large estates being broken into smaller land packages. However, it appears that the practice was more relevant to early Islamic Ifrīqiya than has been acknowledged until now. More understanding of irrigation techniques in the early Islamic period, and of the crop varieties grown on different estates would give a better picture of how and when manpower was required, which would help to nuance our understanding of forced labour in this period.

These observations, like the source criticism on which they are based, are also relevant for understanding agricultural production in other parts of the early Islamic Empire. IfrÄ«qiya was distinctive because of the confluence of Mediterranean and sub-Saharan slave traffic in the province, and because Arab conquerors acquired land here quite quickly after the conquest, but the presence of agricultural estates is evidenced in other parts of the Empire, both as a continuation of pre-Islamic culture and as a result of land being made suitable for agriculture after the Islamic conquest.¹²⁹ The circulation of Bryson’s TadbÄ«r al-manzil, which refers to the use of slaves for manual labour, indicates the importance of agricultural production on these estates.¹³⁰ It would seem possible, then, to extend the methodology of this study to understand agricultural slavery in other Islamic provinces.

Another question to which the findings of this paper relate is the continuity between the pre-Islamic and post-Islamic way of life in rural IfrÄ«qiya. The study has shown that most information about the socio-economic history of the region is derived from sources produced in urban contexts, which were more heavily dominated by a Muslim population than rural regions.¹³¹ Less information is available for rural regions but it is evident that some pre-conquest structures such as tenant farming continued. Rather than a reproduction of earlier patterns, the persistence of these structures might be better referred to as altered continuity due to the influence of post-conquest innovations and more gradual developments from the Vandal conquest. For example, while there are some overlaps between the social roles and hierarchies depicted in late Roman mosaics and Islamic written sources from this region, there are also striking differences, particularly in relation to race and gender.¹³² In the case of agricultural production, which was the focus of this article, the classical heritage of the province was undoubtedly an important factor for the use of slave labour, but its geographical location, post-conquest history and the economic-political structures of the early Islamic Empire played an equally definitive role.