Abu ʿUbayda (d. 825) was a mawlā (client) of Jewish descent who wrote prolifically about history, religion, and culture. As such, he exemplifies the well-known feature of early Islamic learning that is the Abbasid-era mawlā scholar.Footnote ¹ His grandfather was a freeborn convert, rather than the more common manumitted slave, and it happens that the grandfather's patron—his sponsor, as it were, for admission into Islamic society—was a slave trader named ʿUbayd Allah b. Maʿmar (d. ca. 665). And ʿUbayd Allah b. Maʿmar, on a conservative estimate, had purchased hundreds of slaves from ʿUmar b. al-Khattab, the caliph who, before his assassination by a slave, had presided over the explosive early phases of the Islamic conquests.Footnote ²

It has been suggested that slavery is one of early Islam's most understudied topics. This is surely right and completely understandable.Footnote ³ There are too many topics to research and too few scholars. But it is not just a matter of a labor shortage of Islamwissenschaftler. For one thing, the received wisdom has the late Antique world thronging with holy men and women, war-making monks, crusading emperors and shahs—but few slaves. Because the great age of slaving was presumptively over, the topic has mostly interested historians of the later Abbasid and post-Abbasid periods, when slave armies and slave polities emerged in very different circumstances. For another thing, slaves are taken for granted by the transmitters and authors who narrated the grand political and military dramas of the 7th and 8th centuries; one might say that they are extras in an Abbasid screenplay of an already distant past. Things improve for the 9th century, when our sources fan out: slaves appear in belles-lettres, and law books typically treat the topic under a variety of categories, regulating and sometimes ameliorating slavery conditions and practices; we also have documentary material and other forms of prose.Footnote ⁴ To take a later example, in a chapter on “Traders in Slaves and Animals,” a 12th-century marketplace manual tells us that the slave trader needs to acknowledge what might be called a baseline of human dignity: slaves are not to be sold to unknown buyers; their bodies can be examined, but only within rules that preserve some modesty; mothers are not to be separated from small children.Footnote ⁵ But the evidence has not been added up. For every ʿArib (a celebrated slave courtesan and singer of the Abbasid court, d. 890) there were probably hundreds of thousands of slaves who have no voice at all. We suspect that slaves were integral to Umayyad and Abbasid urban society, but we have not yet demonstrated it.

Slavery is theoretically slippery and sociologically complex, and the challenges are all the greater in earliest Islam. “I am loathe to see the practice of enslaving apply to the Arabs,” ʿUmar is given to opine, but we know that Arabs were sometimes enslaved. We take it for granted that only non-Muslims could be enslaved and that the free could not be sold, but the early evidence tells us otherwise.Footnote ⁶ The complications do not end there. The secondary literature, dependent as it is upon the primary, speaks far more of domestic than agricultural and rural slavery, and my guess is that we underestimate the latter, as I shall suggest. “The currents of the slave trade,” Michael McCormick writes concerning the 8th and 9th centuries, “were broad and powerful.”Footnote ⁷ But trade was not the only source of slaves: especially during phases of increasing or high taxation, debt slavery may have been fairly common, adding to natural reproduction as a source within the caliphate.Footnote ⁸ In what follows, I ignore most of the difficulties of interpretation, and address a single, if thorny, question: how much enslaving took place in the conquests?

Given the late Antique background, we should expect enslaving on some scale. I leave it to Romanists to explain why reports of the 3rd-century death of slavery were premature.Footnote ⁹ What is clear is that the Byzantine–Persian wars generated large numbers of captives; those not executed or ransomed were typically sold into slavery, distributed as booty, or deported and resettled. An example comes in 578 when, according to Theophylact Simocatta (fl. 620s), Byzantine armies captured some 100,000 Persians in lower Armenia.Footnote ¹⁰ According to a late 7th-century Armenian history, the Sasanians returned the favor a generation later (in 602–3), slaughtering countless, and enslaving the inhabitants of some thirty-three villages.Footnote ¹¹ A few years later (in 611), according to an early 9th-century source (Theophanes, drawing on earlier material), “many tens of thousands of captives” were taken when the Persians took Caesarea in Cappadocia.Footnote ¹² Some of the evidence for captives and slaves in the Hijaz was first assembled more than a century ago,Footnote ¹³ but with Hend Gilli-Elewy's contribution to this roundtable, we now have a catalogue of Hijazi slaving in the time of Muhammad. Thanks to Noel Lenski's recent work, we can also see how active Saracen allies and confederates were in slaving and slave trading along the Byzantine–Sasanian frontiers: “[T]hroughout Late Antiquity, both Persia and Rome were aware of the Saracens’ skill in capturing live prisoners.”Footnote ¹⁴ The usual fate of (Arab) captives and prisoners taken during the Prophet's battles seems to have been pardoning, ransoming, and releasing, but there are exceptions (i.e., execution).Footnote ¹⁵ Sometimes the numbers, for what they are worth, are large, such as the 6,000 “women, children and men” of the Hawazin whom the Prophet, in an act of clemency, returned to their tribe.Footnote ¹⁶

Measuring the scale of conquest-era enslaving means making sense of exiguous and vexing evidence. On the non-Islamic side, we have a heterogeneous mix of sources that are immediate and vivid, but have axes to grind. The Islamic sources are later and more comprehensive, but they manifest their own biases. “He [ʿAbd Allah b. Saʿd b. Abi Sarh] met in battle Jarjir [Gregory], who was leading a force of 200,000 in Sbeitla, 70 miles from Qayrawan,” we read in a mid-9th-century history (one of our earliest); “Jarjir was killed and they enslaved and plundered.” “As for the Arabs,” a Syriac chronicler of the late 8th century writes of raids in northern Syria in the 710s, “they captured and plundered everything they found before them.”Footnote ¹⁷ Accounts are not always this laconic and vague, but assembled for the most part by compilers who had no experience in warfare, they generally pay little heed to the experience of the defeated, be they killed, captured, or enslaved. Our authorities tended to focus instead on the concerns of the victors: the status and privileges secured by participants in the conquests, and the precedents for taxation, provincial administration, and communal relations. Let us merely glance at two eastern Mediterranean examples, Caesarea and Cyprus, and extrapolate from there.

It seems that Caesarea first came under attack in the mid-630s, but it only capitulated for good in 640–41, apparently after a long siege. Al-Waqidi (d. 822), who employs numbers more frequently than other 9th-century historians, reports a string of Somme-like whoppers: the defense consisted of 700,000 paid (Byzantine) soldiers, 30,000 Samaritans, and 200,000 Jews; 100,000 men guarded its walls every night. He reports that 4,000 were enslaved.Footnote ¹⁸ Seventy and forty (and their multiples) are about as stereotypical as numbers come, and so might be best translated as “very many.” Apparently independently, two historians have suggested a rule of thumb: numbers below 1,000 should be divided by 10, those over 1,000 by 100.Footnote ¹⁹ That Caesarea was the regional capital—and so, it follows, a bulwark of the Byzantine defense—can explain a concentration of forces and noncombatant populations. Other accounts (in Arabic, Christian Arabic, and Syriac) are also at pains to emphasize the scale of the fighting and mortality; there were 80,000 Byzantine fighting men, some 7,000 were killed, and captives were taken, one reads.Footnote ²⁰

The conquest of Cyprus was apparently effected in stages, starting in 649. Al-Waqidi, apud al-Baladhuri, describes two landings, the second bringing not only a show of force (500 ships), but also a clear resolve to pacify and rule: “He [Muʿawiya] took Cyprus by force, killing and taking captive,” he writes with a vagueness familiar to anyone who reads conquest narrative, “then he confirmed [the terms of their earlier] peaceable capitulation, and sent there 12,000 men, all paid from the dīwān.”Footnote ²¹ (Twelve is another stereotypical number.) The Syriac Chronicle of 1234, drawing upon a mid-8th-century source, is one of several that number the invaders’ ships at 1,700; it is also distinctive in the enslaving it describes: men, women, and children were separated on the island and shipped to Syria and Egypt.Footnote ²² For Cyprus, a late Antique settlement of “unexpected prosperity” on the eve of the conquests, we can also turn to two Greek inscriptions that would have us believe that those enslaved numbered around 200,000.Footnote ²³

We might conservatively estimate that between Caesarea and Cyprus conquering armies enslaved several thousand souls.

Such a modest conclusion is not my point, however. My point is that Caesarea and Cyprus were merely two of the thousands of hamlets, forts, villages, towns, and cities that fell to Muslim armies between the conquests of the 630s and 640s, not to mention those of the 710s. And connecting the two phases was more or less regular raiding on the frontiers, the summer plundering ensuring a flow of captives.Footnote ²⁴ Not all settlements put up the kind of resistance that resulted in mass captivity, of course: in the Sasanian capital of al-Madaʾin, for example, we read that a distinction was drawn between the city folk, who were allowed to return to their homes, and the ruling family and court, which enjoyed no such immunity.Footnote ²⁵ Historical and legal narratives have it that the liability of lives depended on how a given settlement fell.

Still, even if we grant that modes of capitulation could produce modes of immunity, the remaining evidence cannot be wished away. According to one of the earliest accounts of the conquests, assembled in Khuzistan in the 660s or 670s, the Persian commander offers any amount of tribute to stop conquest armies from taking captives and warring. Another, written in about 687 in northern Iraq, describes the loss of Christian favor among the Muslims: fueled by an apocalyptic fervor, he reduces Islamic rule to punishment, captivity, and deportation. Al-Baladhuri's conquest reports of Nubia speak of regular and large-scale enslaving,Footnote ²⁶ and those who hesitate to give any credence to such a late text will have to account for the mid-8th-century documentary evidence showing that even when the caliphs’ armies failed to conquer, clients were expected to deliver slaves in the hundreds.Footnote ²⁷ To read conquest accounts means occasionally coming across the names of those who, taken captive in or after a battle, eventually made good (or got lucky) and found a patron and so freedom; to read the biographical compilations of scholars who lived in the 7th and early 8th centuries is to come across such figures more or less all the time. ʿAyn Tamr is as good an example as any: specialists in Prophetic biography will recall that among that conquest's spoils was one Yasar, who, having changed hands once or twice, was manumitted and had a son named Ishaq, who in turn bore a son of his own, Muhammad (d. 767), who would turn into an Abbasid courtier and celebrated biographer of the Prophet. There were many others taken at ʿAyn Tamr, and not all are lost to oblivion.Footnote ²⁸

In sum, conquest accounts describe large-scale enslaving and any doubt about that scale does not survive reading in the Islamic historical tradition. As is well known, the legal category of slavery in early Islamic society does not align with any single occupational or economic role.Footnote ²⁹ Most slaves presumably went into mostly invisible forms of labor, especially domestic service, manual labor, crafts and manufacturing, and working for owners and proprietors who included freeborn (Arab or otherwise) and mawālī by manumission. Slaves appear frequently as scribes on early state documents.Footnote ³⁰ Too little has been said of their role in the agricultural economy, where our sources’ coverage is so spectacularly poor. Chained together to prevent flight, they were put to work digging canals in southern IraqFootnote ³¹ ; they fled anyway. In the late 650s, ʿAli could claim 8,000 mawālī and slaves (ʿabīd) enrolled on the dīwān in Kufa. A generation or so later, al-Mukhtar caused outrage by enrolling freedmen (mawālī) and runaway slaves in his rebellion; a leitmotif of these accounts is a world out of balance, where lowly slaves arise and humiliate their masters.Footnote ³² A few years later, we read about a slave named Maymun who took the nisba al-Jurjami, having been amongst the slaves belonging to Christian cultivators who had temporarily joined the Jarajima (Mardaites); ʿAbd al-Malik thought so highly of his skills that he “requested his masters to manumit him, which they did.” He was then put in command of 1,000 men and joined Maslama b. ʿAbd al-Malik in campaigning against Tyana.Footnote ³³

Sayf b. ʿUmar has an eager participant in the conquest of Iraq recount breathlessly how in the wake of the battle he happened upon a beautiful woman, luminous as a gazelle, even; he claims her a slave and she eventually bears him a child.Footnote ³⁴ At least in part, such stories of plunder had appeal because they presented the wealth of the discredited Old World as reward for Arabians stoked by the hot fires of nascent Islam. They presumably were edifying to read. We should be more attentive to underlying processes. One was political. In its appeal to low-status non-Arabs, al-Mukhtar's movement was a precursor to the Hashimiyya of the middle of the 8th century. The Arabs’ monopoly on political power would come to an end because the powerfully activist message of the Prophet's legacy was spreading among the conquered. To some, the conquered and the enslaved were one and the same; now the conquered and enslaved were taking their world back. Another process—closely related, of course—was social. The case of Muhammad b. Ishaq reminds us that clientage was turning slaves into Muslims, and so small islands of Arab-Muslim settlement were absorbing enormous volumes of cultural capital—especially late Antique learning. In time, mawālī would dissolve as a social category, Arab and non-Arab having assimilated towards each other; Islamic society was being born. This society would include slaves, and they were not always quiescent.Footnote ³⁵