In this project, both ethnicity and race are necessary components of the research methodology. Race and ethnicity are different but overlapping strategies for human categorization that operate both within social practice and within scholarship. Disentangling them from one another is difficult but necessary, especially in order to translate into each other different idioms of scholarly discourse. Race and ethnicity are at one and the same time categories of social and political practice and categories of social and political analysis (Brubaker and Cooper 2000: 4). Both of them have been historically mobilized in order to provide foundations for discrimination, positive or more often negative, between human groups – race much more often than ethnicity (but see Mamdani 2001). In scholarly discourse after the Second World War, ethnicity has often been employed in place of race, in an understandable reaction to the horrors of the Holocaust. At the same time, ethnicity has found its own field of application, as an overarching concept that covers a particular class of cultural notions of difference between groups of people, in anthropological studies from the sixties of the 20th century onwards. Unlike race, ethnicity is typically used to refer to strategies of boundary-making between groups that consider themselves different from each other in terms of descent and ancestry, but in practice recognize each other thanks to distinctive cultural traits (Wimmer 2008 for a summary of problems and approaches). Also, ethnic boundaries tend to oppose group to group, and can vanish from perception in daily practice until specific historical processes confer saliency upon them. In recent years, scholars have increasingly devoted thought to defining and delimiting the respective areas of application of race and ethnicity in the study of antiquity (Mac Sweeney 2021) and the Middle Ages (Jordan and Reimitz 2021). The path is open for a mutually advantageous use of these two strategies of categorization and of their respective scholarly discourses to the study of the Mediterranean world in the pre-modern age.

Meanwhile, the study of systems of enslavement in the pre-modern Mediterranean world has itself progressed in very important ways. The notion of the fundamental difference, in this respect, between the Greeks and Romans on one side and the rest of ancient cultures on the other, central to the seminal work of M. I. Finley (Finely 1998 for a condensed version) has been questioned by D. M. Lewis, who has been able to show that slavery was in fact widespread all across the Eastern Mediterranean and in Mesopotamia during the same centuries in which the enslavement systems of the Greeks were taking their shapes (Lewis 2018). At the same time, Finley’s fundamental distinction between slave societies and societies with slaves, which again isolated Graeco-Roman antiquity in a category including the Antebellum South of the United States and Brazil, has been interrogated and replaced with a more analytic approach that recognizes the differences between the economic function and the cultural importance of slave systems in different historical contexts (Lenski and Cameron 2018, Vlassopoulos 2021). In a comparable change of perspective, the old notion of the transformation of ancient slavery into medieval serfdom has been completely discarded and scholars have been able on the one hand to offer a new and different explanation for the (numeric) decline of the Roman imperial system of enslavement (Harper 2011) and on the other to demonstrate the development and enduring significance of slavery in the western and eastern Mediterranean world (Rio 2017, Rotman 2009). It would be fair to say that we are now in a position to investigate the cultural underpinnings of unfree labor in the ancient and post-antique world on a new footing.

In the Greek poleis of the Classical and Hellenistic periods, from the 5th century BCE to the 2nd CE, the majority of enslaved people came from regions of the Eastern Mediterranean and from the Black Sea. They would have been clearly set apart by their speech patterns, even if they had mastered the language of their masters. As many as 20-40% of the population of an average Greek polis would have been composed of enslaved non-Greeks and their descendants (Canevaro and Lewis 2022). While Greeks were by no means averse to enslaving fellow Greeks, signs of a cultural sanction against this practice appear to cluster around the 4th century BCE; given the widespread opposition to enslaving people from one’s own ethnic group or race that is met in many different cultures, this pattern might be just a function of the chronological distribution of our evidence – a question worth further exploration. Be that as it may, whenever our evidence for individual enslaved persons gets granular, with cases starting from the late 5th century BCE, their origin turns out to be from outside the world of the Greek poleis. While Greek literature does not show much in terms of racial stereotyping of enslaved persons, it is the case that Aristotle’s theory of the natural slave leans heavily on the opposition Greek-barbarian (or barbaros, the Greek word for whoever did not speak Greek), and on the supposed proclivity of the latter to give up claims to personal freedom (Garnsey 1996). Surprisingly, though, enslavement does not really feature in the abundant bibliography on the cultural construction of the opposition Greeks-barbarians, and Aristotle’s theory is never taken as a potential indicator of more wide-spread ideology of enslavement.

On the other hand, the diversity of systems of enslavement in the Greek world created space for the use of notions of otherness of a different sort. A number of poleis during the classical period exploited self-reproducing groups of enslaved people rather than depend on the continuous inflow from the outside. In Sparta, in the poleis of Thessaly, in Crete, these groups tended to be identified and categorized in ethnic terms. Even though their true origin was likely diverse and included enslaved outsiders akin to those encountered in other poleis, these groups of unfree were construed as members of previously free communities that had been enslaved in the very distant past. In the best documented case, that of Sparta, this projection of unfree ethnicity was supported by practices of ritualized contempt that were meant to instill in the enslaved the persuasion of their inferiority. The ethnic identity of the enslaved was a double-edged sword, however, and went on to form the foundation of their final emancipation (Luraghi 2008).

At the other end of the chronological spectrum, in the Islamic world from the 10th century onwards there emerged a genre of treatises dedicated to the purchase and trafficking of enslaved people. In these treatises, a taxonomy based on racial, ethnic and regional origins was developed, indicating which groups were best suited to which tasks. This ethnography of enslavement, with roots in medical literature going back all the way to antiquity, focused on the peoples of the Caucasus, the Turkish peoples of the Eurasian steppe and of inner Asia, and the black peoples of eastern as well as of sub-Saharan Africa (Müller 1980, Lewis 1992). In practice, ethnic definitions were omnipresent in the market for enslaved people in the Medieval Mediterranean (Barker 2019). Ethnonyms like ‘Turk’ or ‘Circassian’ were employed to refer to their putative origin, and ended up, as is so often the case, becoming synonym for ‘slave’. Again, rather than the categorical, latently genetic discourse of race, what appears to underpin the otherness of the enslaved in Medieval Islamic culture is a Mediterranean ethnography that was at the same time a kind of practical knowledge for enslavers.