Читать книгу That Most Precious Merchandise - Hannah Barker - Страница 10

Chapter 2

Difference and the Perception of Slave Status

Religious difference was the legal and ideological basis of slavery in the late medieval Mediterranean. In theory, anyone might be captured and enslaved by adherents of another religion. In practice, some people were more likely to be enslaved than others, and some were enslaved by coreligionists. Russian Christians, for example, were more likely to be enslaved and less likely to be judicially manumitted than Greek Christians in Christian Italy. Such patterns show that despite the legal and ideological importance of religious difference, it was not the only factor at work in determining slave status.

For medieval jurists, the problem with a system of slavery based on religious difference was the difficulty of proving the religion affiliation of specific slaves. Religious belief was an immaterial quality of the spirit, fully accessible only to God and the individual believer himself or herself.¹ Religious practices like circumcision might leave visible marks on the body, but because most late medieval slaves were women and circumcision applied only to men, their affiliation could not be proven in this way.² Some Mediterranean societies used badges or special clothing to signal religious affiliation, but clothes were easy to change.³ In theory, Christians, Muslims, and Jews followed distinctive dietary laws; in practice, slaves were not asked to eat pork or drink wine as a religious test. Slaves were sometimes able to prove their affiliation by reciting a prayer or creed (the shahāda for Muslims; the Pater Noster, Ave Maria, or Credo for Christians), but they were not able to give binding legal testimony about themselves.⁴ Moreover, newly enslaved people who had not yet learned the language of their masters would struggle to communicate anything about their religious background.

The difficulty of categorizing slaves by religion caused legal and economic problems. People enslaved illegally did have opportunities to challenge their status, as discussed in Chapter 1, and occasionally, they were successful. No master wanted to take on the potential expense and inconvenience of a slave whose status was doubtful. A seller might need to show a contract for his or her initial purchase of a particular slave to verify that slave’s status.⁵ Slave sale contracts often included a clause in which the seller promised to uphold the legality of the sale in court. But there were other ways to categorize slaves and thereby assess the validity of their status. Previous scholarship has shown that medieval people used language, law, customs, descent, and geographical origin along with religion to identify themselves and categorize others.⁶ Of these options, language and race were the most relevant in the context of slavery.

The difficulty of categorizing an individual slave in a reliable way can be illustrated by two descriptions of the same young woman, a slave purchased by Biagio Dolfin in Alexandria in 1419 and sent to Niccolò Dolfin in Venice. Her sale contract, composed in Arabic, described her as “a female slave of Nubian race, called Mubāraka, a Christian woman.” In a letter to Niccolò composed in Venetian, Biagio described her as “a little slave girl, black, Saracen, about fourteen years old.”⁷ These two descriptions are fundamentally inconsistent. Was Mubāraka a Nubian Christian or a Saracen (an Arab Muslim)? The answer is that, in the context of slavery, both racial and religious categories were dictated by the master. Mubāraka was a Nubian Christian in Alexandria because that categorization made her legally enslaveable in Alexandria, and she was an Arab Muslim in Venice because that categorization made her legally enslaveable in Venice.

Before embarking on a detailed discussion of language and race as they operated in the late medieval Mediterranean culture of slavery, however, several caveats are in order. First, although sources in Arabic and Latin used many of the same terms to categorize their slaves, we cannot assume that those terms meant precisely the same thing in both languages. Both Latin and Arabic, for example, used the word Turk to categorize slaves. In Latin sources, it referred to people from Anatolia, but Arabic sources were more likely to call people from Anatolia rūmī.⁸ The term Turk in Arabic sources could be used in multiple ways, but it was associated with nomads, people living in the north, and speakers of Turkic languages.

Second, the use of these terms varied according to genre. In Italy, legal documents and travel narratives used contemporary terms (Tatar, Circassian) for Black Sea people, while literary and scholarly works used classical Greco-Roman terms (Scythian, Sarmatian).⁹ Though it is frustrating for the historian to find multiple terms used to signify one group of people, it serves as an excellent reminder of the cultural construction of race. In the late medieval Mediterranean, racial categories were used inconsistently because different genres constructed them differently.

Finally, it should be noted that few of the slaves originating from the Black Sea were black by either medieval or modern standards. The subject of race in the Middle Ages is a complex one, contested among specialists and frequently misunderstood by nonspecialists.¹⁰ Studies of racism in medieval slavery have generally limited their analysis to black and white rather than engage with this complexity. I argue that the complexity of the medieval framework of race was essential to the medieval framework of slavery. When categorizing a slave by religion did not serve the needs of the master, either because religious affiliation was too difficult to prove in court or because it would lead to the slave’s manumission, masters turned to the much more complex and flexible category of race to justify their ownership. I also find that reexamining the powerful and deeply engrained association between black skin color, race, and slavery in a historical context where slavery was correlated with race but not with skin color helps show the ways in which modern racial thinking is historically contingent.

Language

Language was associated with outsider status in the late medieval Mediterranean culture of slavery. Slaves’ poor command of their masters’ languages, whether or not it was true, was often cited by medieval sources as evidence of their foreign, heathen origin.¹¹ Slaves were also renamed by their masters in ways that marked their status as social outsiders. However, Christians and Muslims perceived the connection between language, outsider status, and slavery in different ways. Christians saw linguistic diversity as a reflection of the diversity of the Christian community and of humanity in general, whereas Muslims saw the Arabic language as a unifying force for the Muslim community.

Muslims placed great weight on Arabic as the language of Islam because it was the language in which God had revealed the Quran. Non-Muslims in disguise could be unmasked through their poor command of Arabic, but Muslims unable to speak Arabic might go unrecognized as believers.¹² Under the wrong conditions, this could lead to their enslavement. For example, two Christian ships arrived in Tunis in 1462 with a group of captives for ransom. A passing traveler, ʿAbd al-Bāsiṭ ibn Khalīl, visited the ships and found one captive left unransomed by the locals, “an excellent Muslim of Turkish race, knowing only Turkish and the language of the Franks.”¹³ Since ʿAbd al-Bāsiṭ spoke Turkish as well as Arabic, he was able to speak with the captive and explain to the locals that he was a Muslim. They hastily ransomed him too and tried to excuse their mistake: “‘By God,’ he said to me, ‘we didn’t know his language at all, we believed that he was an infidel’” because he didn’t speak Arabic.¹⁴ Later in his journey, ʿAbd al-Bāsiṭ was shocked to encounter a group of Berbers who did not recognize him as a Muslim even though he was “addressing them in Arabic speech and confessing the two shahāda.”¹⁵ In fact, the Berbers “didn’t know Arabic at all: their language is Berber, and they don’t distinguish between the language of the Arabs and that of the Franks. They astonished me greatly.”

The assumption that all Muslims could speak Arabic was also the basis of a slave market scam practiced in al-Andalus during the thirteenth century.¹⁶ A female slave was offered for sale at a very high price with the claim that she was a fresh captive from Christian territory. After the sale had been concluded and the seller had gone away, the woman addressed the unwary buyer in perfect Arabic. She threatened to complain to the local judge that she was a free Muslim who had been unjustly enslaved, tarnishing the buyer’s reputation as well as costing him the purchase price. Then she would suggest that he resell her to another dupe and split the proceeds with her, perpetuating the scam but reducing his own losses.

In contrast, Christians interpreted linguistic diversity as a sign of religious diversity within the Christian community.¹⁷ Latin and Greek were used as shorthand to represent Catholic and Orthodox Christians. A crusade proposal of 1332 asserted the correctness of Catholicism against the “many Christian peoples of diverse languages who do not walk with us in faith or in doctrine.”¹⁸ Pilgrims and travelers used linguistic diversity to express their amazement at the variety of people they encountered. At Christmas celebrations in Bethlehem, the fourteenth-century pilgrim Niccolò of Poggibonsi explained that “each generation (generazione) celebrates in its own rite, in its own tongue, so that it is a marvel to see so many people thus disguised in tongue and attire (in lingua e in vestimenta).”¹⁹ He mentioned both spoken and written language in his description of the diversity of Cairo, where “one generation is distinguished from another in language and letters and dress.”²⁰ According to Alberto Alfieri, fifteenth-century Caffa was “ornamented by the tongues of its diverse peoples.”²¹ Arnold von Harff used linguistic diversity to indicate the sheer length of his physical journey: “I will, with God’s help and according to my small understanding, now describe [my journey] from country to country, from town to town, from village to village, from mile to mile, from one day’s journey to another, from language to language, from faith to faith, together with all that I have seen and experienced.”²²

What may appear to be a loose association between language and religion was nevertheless used to distinguish groups of people in law and in the courts. In 1224, a plaintiff speaking through an interpreter appeared before the archbishop of Genoa to challenge the status of a slave woman named Maimona. The archbishop ruled that Maimona’s enslavement was legitimate because “she did not seem to him to be from the land of Egypt, rather she seemed to be from the Maghrib on account of her language.”²³ Venice offered to confer citizenship on fifty people in the Black Sea region to boost its presence there, but the offer was open only to those who were “Latin by origin and language.”²⁴ In 1368, Venice forbade its merchants from importing slaves “of the Tatar language.”²⁵ As for Circassians, the Venetian humanist Giorgio Interiano claimed that they were so uncivilized that they did not even have a written form for their language.²⁶ A man named Johan won freedom in Valencia by proving that he was a Hungarian Christian and not a Muslim Turk. His case rested upon the testimony of four Germans who conversed with him in the German and Hungarian languages as well as a doctor who verified that he was not circumcised.²⁷

In Italian documents, the connection between language and religion appeared most often in relation to names. Masters frequently renamed their slaves. Roughly 80 percent of slave women in Genoa were given one of six names (Caterina, Lucia, Maddalena, Margherita, Maria, or Marta).²⁸ These names were also given to freewomen, but the pool of names for freewomen was much larger. Giorgio was a common name among male slaves, far more common than among freemen. Certain names were given only to slaves: Cita (quick), Bona (good), Picenina (little), Benvenuta (welcome), Pucella (handmaiden), Divizia (riches), Melica (musical), and Aspertus or Expertus (experienced).²⁹

Legal documents included both old and new names, often presenting the distinction between them in terms of language and religion. Sometimes language was emphasized: “Caron in Tatar, Paul in our language” or “Chotlu by name, and thereafter called Christina in Latin.”³⁰ Other times religion was emphasized: “called Stoilana in her language, by the grace of baptism Marta” or “not baptized, and is called Achzoach in her language, and in baptism ought to be named Bona.”³¹ Sometimes language and religion were explicitly connected, such as the woman “called Margarita at baptism and in Latin.”³² A Venetian correspondent writing to the Pratese merchant Francesco Datini notified him of the purchase of a new slave and advised him to “have her baptized and give her a name in your own way.”³³

Because the imposition of a Latin name was sometimes linked with baptism, scholars have tended to assume that all slaves with recognizably Latin names had been baptized.³⁴ That is not necessarily true. Some masters gave their slaves Latin names without the sacrament of baptism.³⁵ Also, there were numerous slaves whose old and new names were both associated with Christianity. An Abkhaz girl named Maria was renamed Barbara; a Russian Maria was renamed Marta.³⁶ In such cases, did the conferral of a new name imply a second baptism, or were these slaves renamed without baptism? A woman “called Caterina in her language and Antonia in baptism” was certainly baptized once and may have been baptized twice.³⁷ A Circassian girl “called Serafina in her language but in our idiom Magdalena” seems to have been renamed without a sacrament.³⁸ Baptizing the same individual twice was theologically unsound, even if the two baptisms were performed in different rites. Nevertheless, some Catholic priests seem to have performed rebaptisms, since Pope Martin V threatened to excommunicate anyone who rebaptized Greek slaves.³⁹

Adding race to the set of connections between names, languages, and religions can be misleading too. The name Caracossa might derive from Saragossa in Spain, from Circassia in the Black Sea, from the Greek name Karakouttis, or from a Tatar name.⁴⁰ Each onomastic possibility carries a different set of linguistic, religious, and racial associations. Some slaves had names that reinforced their linguistic or racial categorization, such as Jarcaxa or Jarcaxius for a Circassian.⁴¹ Others had ethnonyms that did not match their linguistic or racial categorization, such as a Tatar woman named Cataio (Chinese), a Circassian woman named Gota (Goth), and a Laz woman named Comana (Cuman).⁴² Nasta, usually considered a Greek name, belonged to a girl categorized as Tatar.⁴³ The name Chotlu or Cotlu was attributed to Alan, Mongol, Russian, and Tatar women.⁴⁴ The result is that names cannot be relied upon to categorize a slave by language, religion, or race.

In contrast, Mamluk slaves were usually given non-Islamic names. Domestic slaves and eunuchs were given the names of desirable objects or qualities like Amber or Nightingale.⁴⁵ Young mamluks, as well as some of their slave concubines, were given distinctive names composed of Turkic and Persian elements like Aqbirdī or Qarābughā.⁴⁶ Names constructed in that way signaled a slave’s membership in the elite military class, not his or her original language. Of the distinctive mamluk names, some were invented in Egypt, but others had roots in the Black Sea.⁴⁷ Comparing mamluk names with the original names of male slaves in Italian documents can reveal which names were used in the Black Sea. The name Jaqmaq, common among mamluks, also appeared as Zachmach or Iacomacius in seventeen Italian documents from Tana, Venice, Genoa, and Pera.⁴⁸ Thirteen were categorized as Tatar and one as Circassian. Only two mamluks named Jaqmaq were assigned a racial category in the Mamluk sources: one was a Circassian and the other a Circassian or Turkman.⁴⁹ Other male slave names that appeared in both Italian and Mamluk sources were Quṭlūbughā/Cotluboga, Qarābughā/Charaboga, Jarkis/Charcaxius, Kitbughā/Katboga, and Tangrī birdī/Tangriberdi. Female slave names that appeared in both sets of sources were Mughāl/Mogal, Ṭughay/Tochay, and Tulū/Tholu.

Mamluk slave names were marked by another distinctive feature. The names of free people included a chain of patronymics, but slaves and former slaves had only one patronymic, Ibn ‘Abd Allah (son of a servant of God). This naming convention erased the slave’s biological family and signaled his or her foreignness from Mamluk society.⁵⁰ Many Mamluk names also included a nisba or laqab, an adjectival nickname. A single mamluk could have multiple nicknames referring to different facets of his identity, such as his patron (al-Nāṣirī for a client of the sultan al-Nāṣir Muḥammad), his affiliation with a legal or theological school (al-Mālikī for a jurist of the Mālikī school), his physical appearance (al-A’war for a man with one eye), or his place of origin (al-Asqalānī for a man from Ascalon). Nicknames that appear to indicate a place of origin can be misleading, though. Al-Jarkasī, for example, might refer to a mamluk of Circassian origin or to a mamluk with a previous owner named Jarkas.⁵¹

Race in the Late Medieval Mediterranean

We have become used to thinking about slavery as a hierarchy of power built on a racial binary, white over black. This framework does not fit late medieval Mediterranean slavery. First, late medieval Mediterranean slavery was a hierarchy of power built on religious, not racial, difference. Second, neither religious nor racial differences were perceived as binary in the late Middle Ages. Third, racial differences were not articulated in terms of skin color in the late Middle Ages, although skin color was one of many physical characteristics that could carry racial associations. The significance of religion for late medieval Mediterranean slavery was addressed in Chapter 1. The remainder of this chapter addresses the roles played by race, skin color, and other aspects of physical appearance.

To approach race from a new perspective, historians may benefit from conversation with psychologists, anthropologists, and sociologists about how human categories arise and function. The private human self is a consciousness located within a body; it develops by interacting with the world around it. The result is the creation of a social self, “a bundle of perceptions held about an individual by a social world,” in addition to a private self.⁵² Although individuals may try to maintain their social selves in harmony with their private selves, their social worlds may not allow them to do so. This is certainly the case with slaves. While the private selves of slaves are a fascinating and elusive topic of research, the following discussion seeks only to understand certain aspects of their social selves, namely, the perceptions that were relevant to their status as slaves in the minds of the people around them.

A group is a human collectivity, the members of which recognize both its existence and their own membership in it.⁵³ Groups can be constructed based on all sorts of criteria, and individuals normally believe themselves to be members of numerous groups (professors, Star Wars fans, women, Americans) without contradiction. When other people make the decision about how to identify an individual, what they use are categories rather than groups.⁵⁴ This book is chiefly concerned with the categorization of slaves: their identification by others rather than their self-identification as a group.

Ethnicity is a way of categorizing other people based on culture. From an anthropological perspective, it is a common and widespread aspect of humans’ social existence.⁵⁵ Race is a way of categorizing people based on their supposedly permanent, fixed, and inherent differences.⁵⁶ Those differences are usually assumed to be physical, visible, or biological, but they do not have to be. What matters is the belief that they cannot be changed. Race, unlike ethnicity, is not a common aspect of human experience. It appears only in certain societies and historical periods. When race exists in a society as a way of categorizing people but is not associated with a power hierarchy, it can be referred to as race-thinking.⁵⁷ When race becomes linked to a power hierarchy, either through explicit ideologies of superiority and inferiority or through institutional structures that implicitly benefit certain racial categories over others, that is racism.

This definition of race and racism is broad. In specific historical circumstances, it has overlapped with other categories that we now consider distinct from race, including religion,⁵⁸ class,⁵⁹ and language.⁶⁰ Our belief that physical appearance is a meaningful kind of permanent, fixed, and inherent difference is relatively recent. The word race in its current sense, a category of people distinguished from others by certain hereditary physical traits, entered English in the late fifteenth century.⁶¹ The notion that skin color was an important aspect of race emerged in the late sixteenth century, and the ideology of different skin colors as the basis for slavery was not fully developed until the eighteenth and nineteenth centuries.

All of these developments occurred after the period covered by this book. In medieval Latin, the words used to categorize people were gens, generatio, genus, progenies, and natio. In Arabic, the most common terms of categorization were jins, aṣl, naw’, and umma. Gens (pl. gentes) and jins (pl. ajnās) are linguistically related.⁶² They can be translated in several ways, all building on the concepts of kind, type, group, and kinship. Medieval people created and shared both verifiable genealogies and myths of common ancestry to explain their systems of categories and groups.⁶³ Some of the most important mythical genealogies made use of the three sons of Noah, the twelve apostles of Jesus, the Homeric heroes, and the founders of Arab tribes, as well as the gens of Adam, which encompassed all humanity, the tribe of the free (banī al-aḥrār), and the gentes of Christianity, Islam, Judaism, and paganism.⁶⁴

In the late medieval Mediterranean, religious categories were often perceived as permanent, fixed, and inherent. Even after conversion, a convert’s religion of origin remained relevant and could still be used to categorize him or her. This attitude helps to explain why religion could function as the basis of slave status even though slaves were expected to convert. In this sense, then, it is possible to talk about religion as a racial category in the medieval world.

However, medieval people also engaged in something closer to our form of race-thinking, categorizing each other based on physical differences that they supposed to be permanent, fixed, and inherent. The bodies of knowledge through which these racial categories were constructed were ethnography (the study of different kinds of people throughout the world), physiognomy (the study of physical traits associated with different qualities of personality and character), physiology (the study of the functioning of the human body), and astrology (the study of the effects of heavenly bodies on earthly ones). Ethnography, physiognomy, physiology, and astrology were part of a shared medieval Mediterranean intellectual culture: authoritative treatises were written, translated, and exchanged among Christians, Muslims, and Jews.⁶⁵ Medieval scholars debated whether racial categories were truly fixed; for example, whether the children of dark-skinned Ethiopians who moved north would become paler. This debate was largely irrelevant to slavery, though.⁶⁶

In constructing their racial categories, medieval Christians started with the observation that although many human beings resemble one another, none look precisely identical, except for twins.⁶⁷ They interpreted the infinite variety of human bodies as a sign of God’s boundless creativity and natural fertility.⁶⁸ An individual’s nature was taken to refer to both his or her unique traits and the shared traits that made him or her recognizable as one human among many. Medieval Islamic scholarly discourse also made a distinction between the diversity of individual bodies and the shared physical characteristics of groups.⁶⁹ Shurūṭ manuals warned scribes that race (jins) and skin color could not substitute for a full physical description (ḥāliyya).⁷⁰

Medieval discourse on religious difference was largely binary: humanity was divided into believers and unbelievers, the right religion and the wrong ones. The existence of multiple kinds of Christians and Muslims complicated this binary somewhat, but as discussed in Chapter 1, this had only a limited effect on slavery. In contrast, racial difference was perceived as a broad spectrum ranging from the normative body at the center to the monstrous races that peopled the far reaches of the earth.⁷¹ Whether cyclops, cannibals, and blemmyae were real was beside the point; they represented the extremes of human possibility, the furthest ends of the racial spectrum. The medieval theory of climate was used to place races in intermediate positions along spectrum, usually from north to south.⁷² Those living in the frigid northern climate zone were supposed to have pale skin, lank hair, a dull intellect, and a cold temperament. Those living in the tropical southern climate zone were supposed to have dark skin, wooly hair, foolish minds, and a hot temperament. The temperate climate zone in the middle was supposed to produce beautiful, reasonable, and well-balanced people. Since the theory of climate zones was drawn from ancient Greek sources, particularly Hippocrates’ Airs, Waters, Places, it was part of the intellectual inheritance shared by the entire Mediterranean world. Of course, most medieval authors tended to locate the temperate zone close to their own homes, wherever that might be.

The association between race and climate theory explains why notaries and scribes used racial and geographical categories (de progenie russiorum vs. de partibus Russia) interchangeably in legal descriptions of slaves.⁷³ When legislators in Florence tried to define slave status, they struggled to articulate a difference between religious, racial, and geographical categories. Thus they permitted Florentines to own and sell any person “who is not of the Catholic and Christian faith … the aforesaid is understood concerning slaves [who were] infidels by origin of their birth, or born from the race of the infidels, even if at the time when they were brought to the said city, court, or district they were of the Christian faith, or even if at some time afterwards they were baptized.… [A person] is presumed to have been infidel by origin if he or she arose from infidel places and race.”⁷⁴

When Genoa instituted an inspection regime in Caffa in the early fifteenth century to ensure that no Christians were being exported to the Islamic world, the inspectors asked slaves first about their race (natio), apparently considered equivalent to asking about their religion.⁷⁵ In cases of doubtful status, the question was not whether the slave’s race was different from the master’s but how far along the spectrum of racial difference the slave fell. The people of the Black Sea were perceived as distant from the people of the Mediterranean: “certainly if it were not for the Genoese who are there, it would not appear that the people [of Caffa] have any lot with us.”⁷⁶ This is why in Italy, the enslavement of Italian Christians caused outrage, the enslavement of Greek Christians caused discomfort, and the enslavement of Bulgar and Russian Christians was ignored. Although all three categories should have been legally protected from enslavement in Italy, Bulgars and Russians were further along the spectrum of racial difference than Greeks, and therefore it was more socially acceptable to enslave them.

Race and Slavery in the Late Medieval Mediterranean

We have established that racial categories were a factor in determining slave status in the late medieval Mediterranean, even though racial difference was not the ideological basis of slavery. We have also established that late medieval people perceived race not in binary terms but as a profusion of human diversity signifying the endlessly fertile creativity of God in nature. Yet, although the number of racial categories was potentially infinite, only a few of them were strongly associated with slavery in the late medieval Mediterranean. Medieval scholars who produced lists of enslaveable people did not necessarily agree on which races were enslaveable and which were not, but the act of compiling lists demonstrated their belief that it was possible to divide the infinite races of humanity into two categories, the enslaveable and the free.⁷⁷

According to Mamluk-era shurūṭ manuals (collections of model contracts), the list of enslaveable people could be divided into Turks and Sūdān. Turks were supposed to be light-skinned northerners originating anywhere from Europe to China. They could be further subdivided into Qiyāṭ, Naymān, Mongol, Kipchak, Khita’i, Circassian, Russian, Alan, Bulgar, Tatar, Āq, Chaghatai, Georgian, Greek, and Armenian categories, among others.⁷⁸ Thus a Circassian Turk was not biracial but a light-skinned northerner (a Turk in the general sense) originating from Circassia (a Circassian in the specific sense). To make matters more confusing, Turk could also be used in a specific sense as a synonym for Kipchak. Sūdān (literally meaning “blacks”) referred to dark-skinned southerners originating anywhere from Africa to India. They could be subdivided into Ethiopian, Abyssinian, Takrūrī, Nubian, Zaghāwī, Dājūwī, Bajāwī, Indian, Khalanjī, Zanjī, Yemeni, Sarūwī, and muwallad (mulatto) categories, among others.⁷⁹ The two sets of criteria, skin color and geography, that distinguished the Turks from the Sūdān were linked by the medieval theory of climate.

Legally, misrepresenting the racial category of a slave would not invalidate his or her sale, whereas misrepresenting religion or gender would.⁸⁰ According to jurists, this was because religion affected slaves’ legal status and gender affected their function, whereas race did neither. However, Mamluk slave-buying guides included lists of enslaveable races and their stereotypical qualities precisely because it would guide the buyer in choosing the right slave for the right function. Unlike shurūṭ manuals, Mamluk guides for slave buyers divided the list of enslaveable people into three categories: Arabs, ʿAjam, and Sūdān.⁸¹ Sūdān referred to dark-skinned southern slaves, as before. Arabs rarely appeared as slaves during the Mamluk period, but they may have been included for the sake of ethnographic completeness or as a legacy from earlier models of the genre.⁸² ʿAjam could refer to all non-Arabs; to all northern, light-skinned non-Arabs; or to Persians specifically. A fifteenth-century slave-buying manual defined ʿAjam in terms of language: “absolutely everybody who differs from the Arab tongue, such as the Persians and the Turks and the Greeks and the Armenians and the Sūdān and the Berber and the rest of them, although this name specifies the Persian people conventionally.”⁸³

The recognized subdivisions of ʿAjam shifted over the course of the Mamluk period. In an anonymous thirteenth-century slave-buying guide, ʿAjam included Persians, Turks, Kurds, Rūmī,⁸⁴ Armenians, Franks, Alans, Indians (al-hind), and Berbers. In the fifteenth century, al-‘Ayntābī added Circassians, Daylamites, Zaranj, and more Indians (al-sind). Circassian slaves had served in Egypt since the thirteenth century or earlier, but their addition to the list of enslaveable races in the fifteenth century was probably a result of their rise to political power in the late fourteenth century. The inclusion of Indians, al-hind and al-sind, among the ʿAjam is also notable because the shurūṭ manuals tended to categorize them as Sūdān.⁸⁵

Contemporary observers noticed certain trends in the racial composition of the Mamluk slave population. Circassian, Rūmī, Kipchak, and Turk were the most common categories used for slaves. Tatar, Mongol, Turkman, Kurd, Armenian, Cypriot, Frankish, Indian (hind), and Ethiopian (ḥabashī) categories were used less frequently. Chinese, Russian, Samarqandi, and West African (takrūrī) categories were represented by single individuals.⁸⁶ One notable trend, according to medieval observers, was the shift in the mamluk population from Kipchak Turks to Circassian Turks at the end of the fourteenth century, as discussed in Chapter 5. Another notable trend was the preference of most sultans and amirs for slaves of the same race as themselves. The perception that political factions were based on racial solidarity was widespread in Mamluk sources, even though modern historians have shown that factions presented as racial often included individual mamluks of various races.⁸⁷ The two trends were linked by the suggestion that Sultan Barqūq precipitated the shift away from Kipchaks by favoring Circassians like himself.⁸⁸ Barqūq’s wife, Ird, a Turk, was said to have warned him against this course: “make your army a variegated one of four races, Tatar, Circassian, Anatolian and Turcoman, and then you and your descendants can rest easy,” because no single racial faction would be able to dominate Mamluk politics.⁸⁹

Medieval Christian philosophers explored the idea of enslaveable races via Aristotle’s theory of natural slavery discussed in Chapter 1. Searching for examples of natural slavery in their own societies led them in a variety of directions. Citizens serving a ruler, artisans, peasants, and barbarians were all identified by medieval philosophers as natural slaves.⁹⁰ Medieval philosophers knew that the word barbarian signaled linguistic difference: “note that barbarians, according to certain people, are said to be those whose language differs entirely from Latin. Others indeed say that whoever is a foreigner is a barbarian to every other foreigner.… But according to he who speaks more truly [i.e., Aristotle], barbarians proper are said to be those who are strong in the strength of the body, are lacking in the strength of reason, and are almost without laws and without the rule of law.”⁹¹ In other words, although barbarians were superficially distinguished by language, they were truly set apart from civilized people by their physical strength, lack of reason, and lack of law, all qualities associated with natural slaves.

Where medieval philosophers diverged from Aristotle was their emphasis on barbarian races, characterizing the customs and bodies of entire groups of people as bestial.⁹² For example, according to Albertus Magnus, “we call barbarians those who neither law, nor civility, nor the rule of any other discipline disposes to virtue, whom Tullius called forest men in the beginning of Rhetorica, conversing with the wild forest beasts in the manner of wild beasts, who are not Greeks or Latins, who are disciplined and fed by a lordly and paternal rule. For such bestial people eat raw meat and drink the blood of humans, they delight to eat and drink from the skulls of humans, they find new kinds of tortures by which they delight to kill people.”⁹³ Such bestial people were located at the far end of the racial spectrum and could be justifiably enslaved by civilized people. In this intellectual context, the long association between Greeks and civilization may have been another reason why medieval Catholics were less comfortable enslaving Greeks than Bulgars or Russians, even though all were Orthodox Christians.

Because the influx of Black Sea slaves to Italy did not begin until the late thirteenth century, Thomas Aquinas and the other great scholastics of the thirteenth century did not link Aristotle’s slaves with the Tatars serving in the homes of wealthy Italians.⁹⁴ Aristotle himself had identified the Scythians of the ancient Black Sea as a barbarian people prone to natural slavery on the basis of their climate, and both Thomas Aquinas and Albertus Magnus had repeated the connection between Scythians, barbarians, and natural slaves. But it was Italian humanists who equated Scythians and Tatars with natural slaves. According to Giovanni Gioviano Pontano,

As I heard from the ancients, the custom was that Thracians and also Greeks who inhabit the Black Sea be sold: who, lest they be in the service of the barbarians, merchants sailing the Black Sea, having redeemed them from the Scythians, were offering them for sale. For it seemed more honorable to serve them for a short time, while they repaid the money paid per head, than to be the plunder of barbarians and submissive to perpetual servitude, also with the greatest disgrace of the Christian name. Because today also are saved towards those whom he calls Bulgars and Circassians.⁹⁵

In other words, the Orthodox Christians of the Black Sea (Thracians, Greeks, Bulgars, and Circassians) could be enslaved by Catholics to save them from the Scythians (Tatars), the archetypical barbarians.

Other humanists also liked to use anachronistic classical terms for the contemporary population of the Black Sea. Racial categories formulated a thousand years ago in quite different historical circumstances reappeared not only in private letters but also in official government correspondence and notarial documents.⁹⁶ In 1416 and 1417, for example, the Genoese notary Giuliano Canella categorized five different slaves as “Gepids or Zichs.”⁹⁷ Zich was a widely recognized racial category in the fifteenth century, and it would be possible to find individuals who self-identified as members of that group. Gepid was a racial category dating from the fifth and sixth centuries, a group with no self-identifying members in the fifteenth century.

Being able to categorize slaves by race mattered especially to notaries because it was required by Roman law. The Justinianic Code stated that “those selling slaves should declare their race (natio) when making the sale; for the slave’s race may often induce or deter a purchaser; therefore, we have an interest in knowing the race; for there is a presumption that some slaves are good, coming from a race with no bad repute, while others are thought bad, since they come from a notorious race.”⁹⁸ As a result, Italian notaries included race along with gender, age, and name in slave sale contracts and other types of documents. Because race had a predictable place in the boilerplate language for slave sales, it is easy to compile a list of racial categories for slaves commonly used by notaries.

Data about the racial categorization of slaves in Genoa and Venice are presented in Figures 14 and 15 in Chapter 5. Russians, Caucasians (Circassians, Zichs, Abkhaz, Mingrelians), and Tatars made up the great majority of slaves in both cities. The label of “Other” hides a very diverse population. The following racial categories were assigned by the notaries to ten or more slaves: Albanian, Black, Bosnian, Bulgar, Canary Islander, Cuman, Ethiopian, Greek, Hungarian, Jewish, Moor, Saracen,⁹⁹ and Turk.¹⁰⁰ Racial categories assigned by the notaries to fewer than ten slaves were Alan,¹⁰¹ Armenian, Berber, Catalan, Goth,¹⁰² Laz,¹⁰³ Libyan, Majar,¹⁰⁴ Meskh,¹⁰⁵ Mongol, Ruthenian, Sarmatian, Serb, Slav, Spanish, Uighur, and Wallach. A few racial categories (dovagus, raamanus, and cevia) do not seem to have equivalents in modern English. Indian and Chinese racial categories appeared only in Caffa and Tana.¹⁰⁶

When Christian observers categorized mamluks, they used many of the same racial categories as the Genoese and Venetian notaries used for slaves held locally. Mamluks were categorized as Abkhaz, Albanian, Abyssinian, Bulgarian, Circassian, Greek, Mingrelian, Russian, Tatar, Turk, and Wallach.¹⁰⁷ But Christian travelers also noticed mamluks from Germany, Hungary, Catalonia, Aragon, Italy, and Sicily. In 1482–1483, the Dutch traveler Joos van Ghistele met Nāṣir al-Dīn, a mamluk from Danzig, who was a treasury official of the sultan.¹⁰⁸ In 1480–1483, Felix Fabri and Paul Walther de Guglingen met a mamluk named Sefogul, a German from Basel whose relatives Felix knew.¹⁰⁹ Two mamluks, Conrad of Basel and an unnamed Dane, guided the traveler Arnold von Harff in Cairo.¹¹⁰

Unfortunately for both Italian notaries and Mamluk scribes, their racial categories were not adequate to describe the complexity and diversity of the people living around the Black Sea. Their hesitation in the face of a human reality that did not fit into neat categories is evident in the sources.¹¹¹ Most authors of Mamluk biographical dictionaries did not mention race. When they did, they preferred the broad categories of Turk, Circassian, Rūmī, and Tatar. In a few cases, they disagreed: the amir Bahādir al-Minjakī was either Rūmī or Frankish, and the amir Jaqmaq al-Arghūnshāwī was either Circassian or Turkman.¹¹² The early fourteenth-century sultan Baybars al-Jashankīr might have been a Turk, or he might have been the first Circassian ruler.¹¹³ Sultan Khushqadam was consistently identified as Greek, but there was debate over whether he, Lājīn, or al-Mu’izz Aybak was the first Greek sultan.¹¹⁴

Some Italian notaries left a blank space where race should have appeared in their slave sale contracts.¹¹⁵ Others put the burden of categorization on the seller, as with a slave “of Goth origin, as she seems to that same Iohannes [the seller] to be from Gothia.”¹¹⁶ The same was true for a slave “who is said to be from Russia”¹¹⁷ and a slave “of the race of the Russians, as it is asserted by the said slave woman, and whom I sell to you as being of the race of the Russians.”¹¹⁸ The Genoese notary Antonio di Ponzò borrowed the phrase “as is” (talis qualis est), a formula for disclaiming responsibility for a slave’s health, and repurposed it to disclaim knowledge of a slave’s race. A woman whom he could not categorize was sold “of race as is” (de proienia talis qualis est).¹¹⁹ Another notary hedged by describing a slave as “of the Abkhaz or another race.”¹²⁰

Notaries also made mistakes in racial categorization. Sometimes they wrote one racial term, then crossed it out and replaced it with another one. Russians and Tatars were most often mistaken for each other in this way.¹²¹ In other cases, notaries gave two racial categories for a single slave: Tatar Russians, Tatar Alans, Tatar Turks, Tatar Circassians, Greek Circassians, Greek Russians, Greek Walachs, Russian Armenians, Russian Bulgars, Russian Circassians, Russian Bosnians, Wallach Bulgars, Bulgar Turks, Bulgar Tatars, Saracen Ethiopians, and Berber Moors.¹²² Some combinations can be explained by the dual significance of the term Greek as race or religion. A Russian Greek might therefore be of Russian race and Greek religion. What a notary meant by a Tatar Alan is harder to explain.

Slavery and Physical Appearance

Black Sea slavery was not white slavery. It was not black slavery either. Although both Latin and Arabic sources occasionally described slaves in terms of color, what they intended to convey was not necessarily the color of the slave’s skin. To grasp the significance of color in late medieval descriptions requires knowledge of late medieval theories of physiognomy and physiology as well as ethnography.

In Europe, state and ecclesiastical authorities first began keeping lists of people and their descriptions in the late fourteenth century.¹²³ The initial purpose of the lists was to track undesirables, such as heretics and criminals, who moved from town to town, but they were quickly adapted for tax collection too. Because names were not sufficient to identify wandering heretics and bandits, the list makers added brief descriptions of their clothing, badges and symbols that they wore, marks on their skin, and their color.¹²⁴ One of the most notable early lists was a register of slaves created in Florence in 1366.¹²⁵ The Florentine register described slaves in terms of color, stature, and marks on the skin.

When late medieval texts referred to a person’s color, whether that person was free or a slave, they meant the color of the body rather than the color of the skin. According to the Galenic theory of humoral medicine, all living bodies were composed of a mixture of cold, hot, wet, and dry elements.¹²⁶ In the human body, these elements mingled in the form of four fluids: blood, choler (yellow bile), phlegm, and black bile. Complexion (complexio) was the state created by a specific mixture of the four fluids. It had a range of meanings. Complexion could refer to an individual’s humoral state at a particular moment in time or to an individual’s innate and characteristic humoral balance. It could also be used to characterize groups: sex, age, race, climate zone, and astrological sign were all believed to affect complexion.

From a medical perspective, each human being was believed to have a unique personal complexion shaped by both nature and habit. This personal complexion could be affected at any given moment by many factors, including air, exercise, sleep, diet, excretion, and emotion. Each organ within the body also had its own complexion. Physicians sought to determine the personal complexion of each patient, the balance of elements and fluids that was normal for that particular body. Then they could intervene in various ways to restore and maintain the patient’s health by restoring and maintaining the correct balance of humors for that patient. The best complexion, from a medical perspective, was a well-balanced one.

Because complexion was an internal rather than external state, medical training required physicians to learn how to read their patients’ internal state through external signs. But because humoral balance affected the mind as well as the body, nonphysicians were also interested in reading the external signs to learn about internal qualities of personality and character. This was physiognomy. Over the course of the fourteenth to seventeenth centuries, as both ecclesiastical and state authorities placed increasing importance on listing and identifying people, the dominant meaning of complexion shifted from the internal, concealed blend of humors to the external, visible signs.¹²⁷ The emphasis also shifted away from the unique humoral balance of the individual to the categories of complexion associated with categories of people. Strong emphasis on complexion as a permanent group state rather than a transient individual state emerged in the late sixteenth and early seventeenth centuries, at the same time as the new word race became widespread in the sense that we use it today.¹²⁸

The principal colors attributed to the human body in Galenic discourse were white, red, yellow, and black, but a healthy body should be mixed in color, ideally a mixture of red and white.¹²⁹ The poet Petrarch described his own color as “between white and dark brown” (inter candidum et subnigrum), while King Ludwig of Bavaria was compared to the biblical King David, “white and red in color” (colore candidus et rubicundus).¹³⁰ Healthy young mamluks in training were likewise described as “white and red” (candidus et rubecundus).¹³¹ When a slave sale contract mentioned color, notaries tended to place it between the physical description and the health warranty, because it pertained to both clauses. Mixtures such as “whitish brown” (bruna quasi blanca), “brown between two colors” (brunus inter duos colores), “olive-brown” (brunam olivegnam), “blackish olive” (seminigrum seu ulivignum), “mixed color” (coloris lauri), or “medium color” (medio collore) signaled humoral balance and therefore good health.¹³² Otherwise, the colors attributed to slaves were black (nigra, nera), brown (bruna, bruneta), olive (olivastra, olivegna), red (rubera, rosa), and white (alba, blanca). Blackness has received the most scholarly attention because of its implications for the Atlantic trade in African slaves, but blackness meant something subtly different to medieval notaries than it did to modern slavers.¹³³

Although the color of slaves is of great interest to us today, it was not particularly interesting to medieval Italian notaries. No more than 3 percent of slave-related documents produced in Venice mentioned slaves’ color. In Genoa, thirteenth-century notaries recorded slaves’ color more consistently than those in the fourteenth or fifteenth centuries.¹³⁴ During the thirteenth century, 38 percent of slave-related documents from Genoa mentioned color. During the fourteenth century, 30 percent of slave-related documents mentioned color, but during the fifteenth century, the figure was only 2 percent. Although the fourteenth-century figure seems significant, 82 percent of those references came from just two notaries, Bartolomeo Gatto and Giovanni Bardi.¹³⁵ In the thirteenth century, references to color came from a much larger proportion of notaries. The reason for the shift from general interest in color to interest on the part of just a few notaries to general disinterest to renewed interest in the sixteenth century is unclear.¹³⁶ The shift suggests that the relationship between color and slavery did not develop in a linear way.

The Galenic system of humors was equally fundamental to medical theory in the Islamic world.¹³⁷ Indeed, it was through Arabic translations of Greek texts that humoral theory reached Latin physicians. As a result of the Galenic emphasis on mixture and balance, the terminology of color in Arabic was complex. The following passage is drawn from a Mamluk shurūṭ manual, a guide to writing legal documents. It comes from a chapter explaining how to compose a physical description of the parties to a contract, including their age, stature, forehead, eyebrows, eyes, nose, cheeks, jaw, lips, mouth, teeth, neck, and distinguishing features. For color, it offers the following array of possibilities:

If a man is very aswad (black), he is called ḥālik (pitch black). If his black is mixed with red, he is called daghmān. If his color is pure, he is called asḥam. If the black is mixed with yellow, he is called aṣḥam. If his color is muddy, he is called arbad. If it is purer than that, he is called abyaḍ. If it is fine yellow and leaning towards black, he is called ādamī in color.¹³⁸ And if it is below arbad and above adama he is called very adama. If it is pure adama, he is called shadīd al-samra (very brown). And if it is purer than that, he is called asmar in color. And if it is purer than that, he is called raqīq al-samra (fine brown). If it is purer than that and leaning towards white and red, he is called ṣāfī al-samra ta’aluhu ḥamra (pure brown rising to red), and is called raqīq al-samra bi-ḥamra (fine brown with red). If his color is very pure he is called ṣāfī al-samra (pure brown) and is not called abyaḍ (white) because bayāḍ is leprosy. If he is completely white, he is called anṣaḥ. If there is paleness in his whiteness, he is called ashqar (pale). If he is paler than that, he is called ashkal. If there is nevertheless increasing red, he is called ashqar. If there are nevertheless freckles, he is called anmash (freckled). If his color is pure and leaning towards yellow and he is not ill, he is called asḥab in color.¹³⁹

Because this passage is about describing the parties to a contract, its range of colors is meant to apply to free people as well as slaves.

With such a rich variety of colors available, restricting scholarly discussion of Mamluk slavery to black and white is misleading. Nevertheless, it has frequently been claimed that the terminology of slavery in Arabic reflects a binary division between black and white. Derivatives of the root mīm-lām-kāf (mamlūk) supposedly referred to white slaves, whereas derivatives of the root ‘ayn-bā-dāl (‘abd) supposedly referred to black slaves.¹⁴⁰ The root rā-qāf-qāf (raqīq) supposedly applied to both white and black slaves. Color-based definitions of these terms, however, tend to come from nineteenth-century dictionaries such as E. W. Lane’s An Arabic English Lexicon and not Mamluk dictionaries such as Ibn Manẓūr’s Lisān al-ʿarab.

When medieval sources are read without the aid of nineteenth-century dictionaries, there are numerous exceptions to the color-based definitions of mamlūk and ʿabd. The eleventh-century Ḥanafī jurist al-Sarakhsī discussed a hypothetical case in which an ‘abd was sold as a Turk but was actually a Greek or an Indian (sindī).¹⁴¹ According to the rule, an ‘abd should be black, but Turks and Greeks did not fall under the Sūdān category, and Indians could be either Sūdān or ʿAjam. Al-Asyūṭī’s fifteenth-century shurūṭ manual refers to the exchange of “a white or black ‘abd for a female slave,” even though, according to the rule, it should have been a white mamlūk and a black ʿabd.¹⁴² Another fifteenth-century jurist, al-Suyūṭī, called a group of thirteenth-century amirs ‘abīd of the treasury, although as military commanders and former slaves, one would expect them to be white mamālīk.¹⁴³ Yet not all mamluks were white. Five mamālīk were described as brown (asmar) in Mamluk biographical dictionaries.¹⁴⁴ There was at least one Ethiopian mamlūk whose color was not given and whose brother was also enslaved as a eunuch.¹⁴⁵ A fourteenth-century marriage contract between two slaves referred to the husband, a Nubian (nūbī), as a mamlūk without describing his color.¹⁴⁶ Color was relevant to how Mamluk masters used their slaves, as discussed in Chapter 3, because it was associated with stereotypes about physical health and temperament. But during the Mamluk period, color was not relevant to determining slave status or the terminology of slavery.

Moreover, color was not correlated with race in the context of late medieval slavery. The theory of climate zones could have been used to link color and race via geographical location, but in practice, one could not predict slaves’ color based on their race or their race based on their color. Greek slaves were described in Arabic sources as pale, red, or brown and in Latin sources as white, olive, or brown.¹⁴⁷ The Greek sultan Khushqadam was “light in complexion with a beautiful golden yellow dominating it,” according to Ibn Taghrī Birdī, but red in complexion according to Ibn Iyās.¹⁴⁸ Tatars might be white or olive.¹⁴⁹ Circassian slaves were white, olive, brown, or red.¹⁵⁰ Saracens could be white, black, olive, or mixed.¹⁵¹ In the Florentine slave register of 1366, Moors were more likely to be described as white (seventeen cases) than as black (thirteen cases), and there were also olive and mixed-color Moors.¹⁵² Whiteness was attributed to Abkhaz, Bulgar, Circassian, Russian, Saracen, Slavic, Tatar, and Turkish slaves in the Latin sources; in the Arabic sources, it was attributed to Circassians, Greeks, and Turks.¹⁵³ Brownness was attributed to Circassian, Laz, Saracen, and Iberian slaves in Latin; in Arabic, it was attributed to Greeks and Tatars.¹⁵⁴ Blackness was used as a color and a racial category in both Latin and Arabic sources. In addition to Blacks, black color was attributed to Iberian, Ethiopian, Canary Island, Indian, Moorish, and Saracen slaves. In Arabic, it was attributed to Ethiopians, Nubians, Zanjis, Zaghāwis, Bujawis, and Qandaharis.¹⁵⁵

Finally, color was not the only aspect of physical appearance relevant to slavery. Florentine and Pisan sources frequently mentioned slaves’ stature.¹⁵⁶ Some notarial descriptions included hair color.¹⁵⁷ Tatars in Italy (but not in the Mamluk kingdom) were distinguished by the shape of their faces. One of the chief characteristics of the Tatar face was broadness and flatness, including flatness of the nose.¹⁵⁸ Sculpted Tatar heads on two fourteenth-century capitals in the south and west porticoes of the Doge’s Palace in Venice offer contemporary images.¹⁵⁹ A Venetian merchant described a Tatar slave as having “a face like a board,” that is, flat, and implied that she was ugly.¹⁶⁰ Franco Sacchetti, a Florentine poet, gave a favorable description of a slave woman who “doesn’t have a very Tatar face.”¹⁶¹ Felix Fabri judged the Tatar face negatively, as well as the Tatar hairstyle, which reminded him of idiots (stulti) in Germany.¹⁶² Pero Tafur claimed that among the Tatars, “the most deformed are of the noblest birth.”¹⁶³

The poet and humanist Petrarch linked the perceived ugliness of the Tatar face directly to its association with slavery. In a 1367 letter from Venice to his childhood friend Guido Sette, at that time archbishop of Genoa, he wrote, “Already, a strange, enormous crowd of slaves of both sexes, like a muddy torrent tainting a limpid stream, taints this beautiful city with Scythian faces and hideous filth. If they were not more acceptable to their buyers than they are to me, and if they were not more pleasing to their eyes than to mine, these repulsive youths would not crowd our narrow streets; nor would they, by jostling people so clumsily, annoy foreign visitors, who are accustomed to better sights.”¹⁶⁴ Although the features of a Tatar face did not change much when the Tatar being described was free, the associations were less negative.¹⁶⁵ Travelers characterized Mongol women as unattractive but hardworking, able to fight and hold power alongside men. Marco Polo described Tatars at the court of Kubilai Khan as noble and beautiful people. Illustrators of his text portrayed them with the same colors as European nobles.

Descriptions of the faces of Turkish slaves appear in both Italian and Mamluk sources as well. The Italians judged Turks, like Tatars, to be ugly: “[they] have short faces, broad in the upper part and narrow below. Their eyes are very small and very similar to those of that small beast [weasel], which by instinct hunts rabbits in their warrens and underground holes. Their noses are rather like those of the Indians [Ethiopians], and their beards closely resemble those of cats.”¹⁶⁶ Mamluk descriptions of Turks portrayed them more positively. According to al-Asyūṭī, a legal description of a Turkish mamluk should mention “whether he has sprouted a moustache, that he is white of complexion, with a prominent forehead, with big deep-black eyes, long lashes, and lids painted with kohl, with a low-bridged nose, flat jaws, ruddy cheeks, red lips, well spaced teeth, with a small mouth, a long neck, of full stature, with small feet.”¹⁶⁷ A female Turkish slave should be described as follows: “a young woman, white tinged with red in color, with a prominent forehead, as in the previous description but in feminine form.”¹⁶⁸

Conclusion

Although language and race did not form the ideological basis of late medieval Mediterranean slavery, they played an important role in determining who was and was not enslaved in practice. When it proved difficult or disadvantageous for masters to use religion to categorize their slaves, they turned to language and, especially, to race as substitutes. Yet the strength of race as an intellectual framework is also its weakness. Race subsumes many kinds of difference (ancestry, geography, culture, climate, humoral complexion, physical appearance, astrological sign, etc.) into a single category, and it promises us that this category is fixed, inherent, permanent, natural, and therefore reliable. Racial shorthand was useful to Italian notaries and Mamluk scribes who needed to categorize individual slaves in a few brief but meaningful words.¹⁶⁹ Racial shorthand was also useful for slave buyers who needed help in choosing the right slave for the right purpose and avoiding slaves whose status was doubtful. Cases in which medieval racial categories broke down could usually be ignored because of the power differential inherent in slavery: if the notary or scribe said that a slave was Circassian, the slave’s ability to refuse that category was severely limited. Notaries and scribes were also able to gloss over paradoxical dual categories, such as Tatar Alans. Some slaves challenged their racial categorization in court, such as the Greeks who proved they were not Bulgars in Chapter 1. But race was used to confirm the status of others, such as Russian Orthodox slaves in Italy and Turkish Muslim slaves in Egypt, even though they should have been freed on religious grounds.

The fact that medieval people understood race as a spectrum rather than a black–white binary helps explain why some slaves were more successful in petitioning for freedom than others. It also helps explain why modern historians have struggled so much to understand how medieval racial categories worked in the context of slavery. We know that race is culturally constructed and historically contingent, as is slavery, and religion, and color. Nevertheless, today we still find race to be a convenient way to collapse many kinds of difference into a single one. Racial categories are so convenient that we sometimes struggle to understand how others could use them differently than we do. To explain the logic underlying medieval Mediterranean slavery, we have to acknowledge that the racial categories with which we are familiar are not permanent, fixed, inherent, or reliable. In fact, no matter what analytical framework we use, the complexity of human experience means that there will always be individuals and cases that do not fit neatly into our categories.