The slave markets of the Viking world: comparative perspectives on an ‘invisible archaeology’

ABSTRACT Slaving was a prominent activity among raiding and mercantile groups operating across the early medieval world during the Viking Age (c. 750–1050 CE). Historical sources provide explicit descriptions of widespread raiding and slave taking by Viking raiders, as well as a substantial trade in captive peoples. Archaeologists, however, have long-struggled to identify evidence for the transportation and sale of captives in the material record. In order to begin addressing this issue, this study explores the comparative archaeologies and histories of slave markets in order to examine the potential form and function of these sites, and how they might have operated as part of the wider, interconnected Viking world.


Introduction
In 821, the early-medieval Irish chronicle, The Annals of Ulster, states that 'Étar was plundered by the heathens, and they carried off a great number of women into captivity'. 1 This was just one of a number of recorded attacks by Viking raiders on communities living in northern and western Europe during the eighth-eleventh centuriesthe period commonly known as the Viking Age, in which captives were taken. 2 Historical sources indicate that over the course of the period, thousands of people were driven to raiders' ships and taken away to be enslaved and subsequently exploited in various roles across Scandinavia, the Norse colonies of the North Atlantic, or among societies inhabiting lands that lay even further afield.
In light of the explicit descriptions provided by written sources, it is not surprising that the Vikings enjoy a dubious reputation as prolific slavers. While the existence of a substantial trade in captives during the Viking Age has long-been accepted by a number of scholars, many aspects of enslavement and captivity remain enigmatic. 3 We have little knowledge, for example, of how captive groups were acquired and transported, where and how they were sheltered, provisioned, or what conditions they were forced to endure. There is similarly a striking lack of evidence for any kind of infrastructure that supported long-distance networks of transportation and sale. While historical sources cite the role of major settlements as slave trading hubs, evidence for a Viking-Age 'slave market' has yet to be identified archaeologically. Rather than studying the processes by which captives were acquired and sold, scholars have largely chosen to focus more on the roles that enslaved groups may have played among Scandinavian societies themselvesfor example in various household, agricultural, and funerary contexts. 4 As Stephanie Yuhl has argued, this can result in the concept of enslavement becoming 'too easily domesticated into a discourse about paternalism, relationships, community, homes, households, and intact families … the enslaved too easily become "servants" separated from active enslavers'. 5 The study of slave markets is crucial not only to our understanding of enslavement as an institution, but also has implications for our wider knowledge of regional and inter-regional economies, social structures, and everyday life among slaveholding societies.
In order to address these issues, this study will adopt a cross-cultural approach to the study of slave markets in the Viking world, defined here as mainland Scandinavia and those areas of northern and western Europe, the North Atlantic, the Baltic, and modern-day Eastern Europe that lay within the wider sphere of Scandinavian influence through raiding, colonisation, and trading contact. Drawing upon the comparative archaeologies and histories of slave trading, the study will explore the possible form and function of slave markets and how their operation has varied among different cultures. It is important to emphasise, however, that the aim of this study is not simply to superimpose comparative data onto Viking-Age societies. While many elements of captivity and enslavement recur cross-culturally, the application of a comparative approach allows both similarities and elements of variation in these practices to be drawn out of historical data, thereby enabling the evidence from one period to inform our knowledge of another.
Captive taking and enslavement are global phenomena that have been practiced among human societies since prehistory. Among many cultures, the acquisition of captives is intrinsically associated with deracination, violence, and dishonour. 6 The violent ripping of captives from their natal kin groups and communities is often described as a form of 'social death', whereby the captive is robbed of their social identity and rendered as a 'nonbeing' in the eyes of their captors. 7 In some slaving systems, this perception can be reinforced by the commodification of captives as human cargoes, who are transported and sold in the same way as livestock or objects. 8 Later, the enslaved might continue to be treated as a liminal being who is marginalised by their captors as an 'other', an object, or a domesticated animal. 9 Their status as chattel can also be enshrined in legislation designed to cement their captor's ownership over their bodies. 10 Captive taking, sale, and exploitation, however, are multi-faceted processes, and it is clear that the fate of captives has varied cross-culturally. While most if not all captives undergo a process of social death as part of their journey into enslavement, in many cases they can also experience a form of social 'resurrection' or 'rebirth' that marks a degree of reintegration into captor society. In a recent synthetic study of captive taking among small-scale societies, Catherine Cameron has drawn attention to the variable fate of captives acquired as a result of raiding and warfare among a number of indigenous societies in contactperiod North America. While some captives were enslaved, others were singled out for torture and execution, while others still were gifted, traded, sold on to new owners or captured by other groups. Some captives were even adopted into the households of their captors. 11 The imposition of enslavement and the extent to which captives remained 'socially dead' therefore varied according to specific circumstances. Thus the concept of 'captivity' might be considered as a transient state experienced by natally-alienated groups whose fate had not yet been fully determined.
When we turn to the Viking Age, it is apparent that many aspects of captive taking and enslavement correspond with those observed elsewhere. Historical sources, for example, clearly demonstrate that these processes were intimately connected with violence and deracination. The concept of dishonour was, similarly, almost certainly associated with enslavement, particularly when applied to the status of captives taken during battle. 12 There is also good evidence to suggest that captives were subjected to a process of social death, with the earliest Scandinavian law codes indicating that the enslaved were perceived as chattel. 13 Textual sources, furthermore, imply that captives could be redistributed by members of the elite to favoured subordinates and retinues. 14 The extent to which captives were commodified for transportation and sale, however, is less clear. Despite good historical evidence for the existence of substantial slave-trading networks, it seems apparent that not all slavers specifically sought captives for sale. In many cases, those taken during Viking raids likely found themselves transported not to slave markets but rather to the home communities of their captors, where they were exploited in many different roles. While at least some of these individuals were undoubtedly condemned to a life of manual labour or sexual abuse, Stefan Brink has argued that the enslaved were incorporated into slaveholding households through the creation of reciprocal relationships that bound captives to captor families, thereby raising the possibility of a potential degree of integration into Scandinavian society. 15 Evidence from literary and legal sources similarly suggests that some captives possessed an altogether more influential status as concubines, wives, or as managers of households and farmsteads. 16 It is therefore possible that any captive's eventual status may have varied according to the particular circumstances of their condition.
As this study is largely concerned with the study of individuals and groups that were acquired through raiding and warfare, it is important to outline the terminology that will be used in the following discussion. In recognition of the potential variability and uncertainty concerning their eventual status, the term 'captive' will be used to describe individuals who were forcibly transported to and sold at markets across the Viking world. For ease of reading, however, the terms 'slaver' and 'slave trader' will be used in reference to individuals and groups who engaged in the acquisition, transportation, and trade of captives. Similarly, the expressions 'slaving', 'slave raiding', and 'slave trading' will be used to describe both the violent and non-violent processes by which captives were obtained and sold. 'Slave market' will also be retained as an umbrella term for the locales where captives were traded or sold, even if this was not the sole or primary function of any given site.

Captive taking and slave trading: the historical and literary evidence
The taking and sale of captives was a fundamental aspect of Viking activity both within and outside of Scandinavia, with historical sources providing a broad picture of slave raiding and trading across an area stretching from the North Atlantic to the Caspian Sea. Given that the Viking Age represents the last period of Scandinavian prehistory, it is important to recognise that the evidence for slaving derives entirely from non-Scandinavian sources. The main sources of data for the activities of Viking raiding groups are historical annals produced by Christian scribes working in the monastic institutions and royal courts of northern Europe. Accounts of Viking raids on the Iberian Peninsula and North Africa as well as Rūs raiding in the Caspian Sea region are provided by Arabic sources, while further references to Rūs slaving and warfare can be found in the Russian primary Chronicle. 17 Other valuable sources of information include the accounts of travellers, saints' lives and the biographies of churchmen who ventured into the northern lands during the Viking Age as well as later ecclesiastical histories. Also of use are Old Norse literary sources, including skaldic poems, the Icelandic family sagas, and Heimskringla -'The Sagas of the Norse Kings'.
All of these sources possess their own limitations and biases. Many historical accounts, for example, served as pieces of political propaganda, selectively whitewashing or augmenting history in order to promote and consolidate the power of royal lineages. As such, they can be considered as providing only an incomplete and highly partial picture of Viking and Rūs activity. 18 Religious biases may have similarly prompted contemporaneous observers and the biographers of churchmen to portray Scandinavian and Rūs societies as overly 'barbaric', thereby aggrandising their own or their subject's morality and achievements. Further uncertainties surround the use of the sagas, whose late date of composition in the twelfth-thirteenth centuries may mean that they reflect aspects of medieval rather than Viking-Age society. 19 Despite this, it is possible to successfully integrate these sources into the study of the Viking Age. While individual historical chronicles or annals should not be taken as providing an exact representation of any particular series of events, the identification of similar patterns of behaviour across multiple sources can arguably provide information on general historical trends. Debates concerning the validity of Old Norse texts are long-running and divisive, but recent years have seen increasing numbers of scholars developing frameworks for their successful integration into archaeological and historical enquiries. There is now some acceptance that these sources can be taken as providing broad, analogous insights into cultural practices and social institutions as they may have existed during the Viking Age. Although it is necessary to acknowledge a measure of uncertainty regarding their use, they offer invaluable perspectives on Scandinavian societies that cannot be extrapolated from material evidence alone. 20 When taken together, these sources provide a relatively comprehensive and consistent overview of slaving activity across the early medieval world. While it is important to recognise that the scale of slaving is almost certainly underrepresented in historic sources given their focus on the fates of kingdoms and their rulers, insular and Continental annals as well as numerous Arabic sources provide ample evidence of both small-scale Viking raiding and what were evidently well-planned, large-scale attacks. They also reveal variability in raiding practices by indicating that some raiders targeted certain demographic groupsnamely women and children, for captivity, as was the case during attacks on Étar in 821 and Seville, Spain, in 844. 21 This may provide some insights into the demands of markets and captor societies, as well as the efforts of raiders to meet these. Although men must have also been taken captive, their role as primary combatants meant that they were more likely to have been killed where raiding groups met with resistance. Those too old or infirm to have been profitably sold or exploited might have been executed.
Slaving was also a by-product of more formalised warfare and conflict, with the Annals of Fulda, for example, recording that a large number of Frankish captives were taken following their defeat in battle by a Viking force during the year 880. 22 Conflict between expansionist polities and kingdoms in Scandinavia would have also provided a potential domestic source of captives who could be sold or exploited by the victors. Outside of Scandinavia, Viking groups regularly engaged in conflict with each other, either as mercenaries or in order to further their own political ambitions, while Adam of Bremen noted that pirates operating in the seas around Denmark during the eleventh century would regularly attack shipping in order to take captives. 23 Given that participation in warfare and mercantile activity were primarily male occupations, the vast majority of captives taken in these circumstances would have been men. It is worth noting the work of Leïla Blili, who has drawn attention to French records of captives taken by Barbary pirates in the Mediterranean during the sixteenth century. 24 While women represented half of the captives taken during raids on settlements, the overwhelming majority of captives taken at sea were men. During the Viking Age, the targeting of shipping lanes similarly may have yielded significant numbers of male prisoners, potentially offsetting the apparent gender imbalance in captives as suggested by accounts of attacks on settlements.
What became of captives once they had been abducted is difficult to say, but it is likely that many found themselves funnelled into the interlocking trade networks that spanned the Viking world. Given that the taking and the holding of captives make different demands on a captor's organisational skills, time, and resources, it was probably not uncommon for them to be quickly sold on to new owners. 25 This can be seen in the Life of St. Findan, in which the Irish captive Findan passes through the hands of three owners soon after being abducted by a Viking group during the ninth century. 26 In Óláfs saga Tryggvasonar, the young Óláfr is similarly sold twice in the time immediately following his capture by Estonian pirates. 27 We cannot rule out the possibility, however, that some raiders possessed a more specialised knowledge of market conditions and the expertise to sustain captives, allowing them to play a double role as slave traders.
For many of those captives taken in western Europe, especially the British Isles, their eventual destination might have been the Norse colonies of the North Atlantic. Others might have been taken to the Continent; in 1047, for example, The Anglo-Saxon Chronicle records a Viking raiding fleet as selling a cargo of English captives in Flanders before sailing home. 28 It is also possible that captives were taken to Scandinavia in order to be sold or exploited among their captors' communities, as suggested by the lives of Anskar and Rimbert, which mention captives living in and being transported through the major market centres of Birka and Hedeby during the ninth century. 29 In Laxdaela saga, we similarly find a number of women, including an Irish captive named Melkorka, being sold by a Rūs trader at a market associated with a public assembly in the Brenno Islands. 30 Evidence for the long-distance trafficking of captives along the eastern trade routes of the Viking world is provided by numerous sources, including the tenth-century Byzantine source, De Administrando Imperio, which describes Rūs merchants transporting captives by boat along the Dniepr river. 31 Also of note is the famous account of the Arabic diplomat Ah  mad ibn Fad  lān, who encountered a group of Rūs slavers and a number of female captives on the Volga river in 922. 32 Neither source provides any indication of where the captives originated, and while we cannot rule out the possibility that at least some might have been conveyed from western Europe or Scandinavia through the hands of multiple traders, it is more likely that the vast majority were taken captive during internecine warfare or Rūs raiding in the Slavic regions of the southern and eastern Baltic. 33 Archaeological evidence for slave trading in the Viking world While written sources provide a relative wealth of information on captive taking and sale during the Viking Age, slaving is often considered to be an archaeologically 'invisible' practice. As such, it is not surprising that the material record allows only fragmentary insights into this activity. Perhaps the most evocative evidence for slave trading is a corpus of what appear to be iron shackles and collars, most of which have been recovered during excavations at urban centres thought to be associated with the slave trade, such as Dublin, Birka, and Hedeby (see Figure 1). Four examples are known from Birka, six have been recovered at Hedeby, and at least one is known from Dublin. Other examples are known from sites such as Skedala, Sweden, Trelleborg on the island of Zealand, Denmark, Neu Nieköhr near Rostock, Germany, Winchester, England, and several crannog sites in Ireland. 34 A small number of shackles have also been identified in Khazaria, a region that is now part of southern Russia, where several large slave markets are known to have existed. These served primarily to facilitate the sale of captives to the Abbāsid Caliphate and the Sāmānid Emirate, both of which were major consumers of slave labour. 35 These finds are ultimately ambiguous and can be interpreted in different ways; it has been argued, for example, that they may have been used for restraining animals instead of humans. 36 It is not unreasonable, however, to assert that some or all of these objects were suitable if not specifically intended for use on humans, and in this the utilitarian design of the Viking-Age finds conforms to one that remained relatively consistent from ancient times through to the early modern period. 37 Unfortunately, details regarding the specific contexts of these objects are often lacking. 38 Given the scale of slaving as described in historical sources, the paucity of shackles and restraints is striking. Several major Viking-Age trading sites, including Ribe and Kaupang, have yet to produce such finds, and they are also unknown at market centres such as Sebbsersund and Fröjel. In this, the Viking-Age finds are unexceptionalonly a few hundred of these objects, dating broadly to the First Millennium CE, have been found across Europe to date. 39 There are several explanations for this dearth of finds. Given their relative expense, metal shackles and chains may have been susceptible to being melted down for reuse when no longer needed. It is also possible that metal restraints were not primarily intended for the mass transportation of captives. In the early medieval world, the 'binding' of hostages and captives represented a fundamental aspect of elite power. 40 Numerous references to this practice are made in Irish sources, and the significance of binding as a motif is seen in the Anglo-Saxon Chronicle, which in 942 describes the population of northern England (at this time under the control of the Scandinavian kings of York) as being in 'heathen's captive fetters'. 41 The use of metal shackles was therefore perhaps mainly reserved for certain occasions, for example to emphasise the subjugation of an adversary. A much more likely reason for the dearth of metal finds is that slavers preferentially used restraints made from organic materials such as wood or rope. If maritime raiding and piracy represented one of the primary means of obtaining captives, then the availability of ship's cordage would have made this a cost-effective alternative to metal restraints. Indeed, recent excavations in the harbour area at Birka have yielded substantial quantities of Viking-Age rope, emphasising the ubiquity of its use in maritime contexts. 42 Despite the small number of shackles, their presence in nodal trading centres such as Dublin, Birka, and Hedeby suggests that captives were inhabiting or moving through these sites. Indeed, it has been argued that the developmental trajectory of Dublin (established by Viking raiders during the 840s) was driven by slaving activity. 43 Exactly where captives were held and sold in these locations is a matter of speculation. Given the importance of maritime transportation in slave-trading networks, the fact that five of the six shackles from Hedeby were recovered during excavations of the harbour might indicate that captives were sold on the wharves or directly from slavers' ships. As archaeological investigations within major Scandinavian marketplaces have thus far failed to reveal large compounds or open spaces that might have functioned as public markets, it is also possible that captives were held and sold from private properties. During excavations at Temple Bar, Dublin, archaeologists identified a small (c. 3 × 2.5 m) post and wattle outhouse that seemed to have been used primarily as an animal pen during the eleventh-twelfth century. The presence of a hearth and human fleas within the structure were tentatively taken to suggest human occupation during its early stages of use, perhaps by lower-status or captive individuals. 44 This raises the possibility that captives were held, transported, and sold in the same way as livestock, suggesting an overlapping function for animal pens situated at trading sites. The sale of Melkorka in Laxdaela saga noted above also indicates that structures associated with slave trading did not need to be substantial or even permanent, given that she was held alongside 11 other women in a tented booth at a seasonal marketplace. 45 In addition to finds of shackles, silver (especially Arabic dirham coins) has often been taken as a proxy for slave trading across the Viking world. 46 While silver was being used in economic transactions in the North Sea emporium of Ribe during the late eighth century, large quantities of silver in the form of Abbāsid dirhams began to make their way into Scandinavia via eastern trade routes in the years around 800. 47 The first significant influx of dirhams occurred during the early ninth century, and it is around this time that silver began to be used for payment in other major marketplaces such as Hedeby, Kaupang, and Birka. 48 A more substantial influx took place during the tenth century, at which time the use of silver seems to have become more widespread. 49 The economic value of these coins lay primarily in their use as bullion, which could comprise whole or chopped up pieces of jewellery, ingots, and coinage (a form of material collectively referred to as 'hacksilver'), and it is likely that large quantities of dirhams would have been melted down for this purpose. Finds of silver bullion and the weighing equipment used in transactions have been made not only at major trading centres, but also at a number of camps established by Viking fleets operating in Europe during the mid-late ninth century, indicating the use of this material among itinerant raiding and mercantile groups. Not all imported dirhams, however, were melted down as demonstrated by the hundreds of thousands of coins (which must themselves represent only a tiny fraction of those in circulation) that have been recovered across the Viking world to date. That so many intact dirhams, imitations, and even forgeries have been found implies that their intrinsic value was widely recognised both as a motif and a medium of exchange.
The use of silver in slave trading is logical not only because of its portability and durability but also due to it being a mutually recognisable and trusted form of payment. It may be for this reason that we find silver first being used for transactions in major, permanent marketplacesthose inter-regional trading nodes through which captives are most likely to have passed. 50 Despite this, the vast majority of dirhams known from Scandinavia (some 67,000 coins) have not been recovered from major marketplaces but rather on the Baltic island of Gotland, where numerous Viking-Age hoards have been found. The largest of these alonethat from Spillings (see Figure 2), deposited some time after 870/1, contained over 66 kg of silver including more than 14,000 dirhams. Why so much silver was amassed on Gotland is a matter of debate, but Dagfinn Skre has recently suggested that the Gotlanders were quick to take advantage of the eastern silver routes that opened up in the early years of the Viking Age by establishing themselves as important intermediaries within long-distance slave-trading networks. 51 Thus far, no direct evidence for slave trading has been identified on Gotland, and it is possible that the island functioned primarily as a base (and financial repository) for merchants who plied their trade in foreign markets.
Although silver could be used as a means of exchange in any transaction, there is some written evidence to suggest its specific use in the purchase of captives. In Laxdaela saga, Hoskuldr purchases Melkorka for three marks of silver (although the merchant also has a number of other captive women for sale at one mark each), and an enslaved man named Thrand is purchased for the same amount in Egils Saga. 52 In early medieval Ireland, a female captive functioned as a unit of valuethe cumal, which equated to around three ounces of silver, eight to ten cows, and a unit of land. 53 It is worth emphasising, however, that captives could have been exchanged for many other forms of payment, some of which are unlikely to have survived in the archaeological record. In Óláfs saga Tryggvasonar, for example, Óláfr and his foster-brother are traded for 'a rather good goat', and Óláfr is later exchanged for a cloak. 54 Other objects may have been used in more systematic trading. According to the Jewish emissary Ibrāhīm ibn Ya'qūb al-Tartushi, the main currency used in Prague (a major slave market) during the mid-tenth century was squares of embroidered cloth that were valued at one-tenth of a silver qinshār. 55 A tenth-century peace treaty between the Byzantines and the Rūs stipulated that Rūs merchants whose captives escaped while in Byzantine territory were to be recompensed with a predetermined quantity of silk per head. 56 While few Byzantine coins are known from northern lands in contrast with the huge number of dirhams, silk is widely found in high-status burial contexts across Viking-Age Scandinavia, possibly indicating that Rūs merchants sought silk (perhaps in addition to other goods, such as glassware or wine) in exchange for captives. 57 Another important commodity was salt, something that has been noted as being traded alongside and in exchange for captives in other early medieval contexts. 58 Stephen Lewis has argued that Viking raids on Noirmoutier in Brittany during the ninth century may have been motivated by the desire to obtain both captives and the salt that was a major product of the region. 59 Tentative evidence for an association between salt, silver, and perhaps slave trading can be found in context of the Spillings hoard, part of which had been buried in a bag that had probably once contained salt. 60

Comparative archaeologies and histories of slave markets
Recent decades have witnessed significant advances in the study of enslaved peoples, as seen in the increasingly nuanced examination of unfree labour and exploitation, the agency of enslaved groups, and the diverse lives of captives in various contexts. Archaeological studies of slave markets, however, are demonstrably lacking. Although a number of putative sites in Roman Italy, North Africa, and Anatolia have been discussed by Elizabeth Fentress and her colleagues, their findings were heavily critiqued by Monika Trümper, whose repudiation of the concept of purpose-built slave markets led her to conclude that Graeco-Roman slave markets are an 'archaeological fiction'. 61 While Trümper's conclusion was perhaps overly negative, her work nonetheless highlights the inherent difficulties associated with the study of these sites. These difficulties are further illustrated in a survey of nineteenth-century slave trading in northern Ghana by Benjamin Kankpeyeng, who noted that the only surviving evidence for the slave market at Salaga (the largest and most important in the region) comprises the mud roof of a warehouse where captives were apparently held during inclement weather. The only other known evidence for slave trading in Salaga are the iron shackles that can still be found within many of the former merchant homes, attesting to the human cargoes that once moved through the town. 62 The situation regarding the identification of slave markets is slightly more favourable in early modern, Western contexts, where it is possible to obtain historical documentation testifying to their existence and operation. Perhaps the most well-known slave markets are those of the antebellum American South, located in slave-trading entrepôts such as Richmond, Virginia, Charleston, South Carolina, and Savannah, Georgia. Even in instances where the location of these sites is known, identifying archaeological evidence for slave trading itself remains a challenging task. Recent excavations at the nineteenth-century Lumpkin's Jail site in Richmond, for example, identified several areas of the slave-trading complex, including a cobbled courtyard, brick drain features, a kitchen and two other buildings spread across two terraces as well as over 16,000 finds dating from the 1830s to the twentieth century. While remnants of the jail building and an enclosing compound wall were identified, these interpretations were heavily dependent on intensive documentary research and nineteenth-century photographic evidence, with heavy disturbance preventing the artefactual evidence from providing detailed insights into daily life and slave trading within the site. 63 When we turn to the historical record, however, it seems that specialised slave-trading sites like Lumpkin's Jail may be exceptional. In many cases, captives were often sold alongside a variety of commodities in multi-functional market sites, reflecting the likelihood that few merchants in history would have dealt only in captives. 64 Wakālat al-Ğallāba, the most prominent slave market in sixteenth-to nineteenth-century Cairo, for example, was used for the sale of various goods in addition to captives destined for general sale, while more valuable captives were kept and sold in merchants' houses. The market was presided over by a sheikh who was responsible both for its management and for ensuring that merchants paid their sales taxes. 65 In nineteenthcentury Moroccan cities such as Marrakesh, Fez, and Rabat, market squares were given over to the sale of captives for certain periods of time, several days a week, while the market space at Salaga was used for the sale of foodstuffs in the morning and for captives later in the day. 66 The allotment of specific times for slave trading may again suggest that the operation of these sites was overseen by external parties who monitored and taxed the sale of goods. The transient presence of captives within such sites means that these groups would leave little to no trace in the archaeological record, and for those scholars who cannot rely on detailed oral narratives or documented histories of slave trading, the challenges associated with identifying slave markets are greatly magnified. Are slave markets as Trümper argues, however, an 'archaeological fiction'? 67 While Trümper is surely correct in highlighting the pitfalls associated with the identification of slave markets, her findings are determined to some extent by her very definition of these sites as 'all squares, buildings, and structures where slaves could be sold'. 68 Trümper's focus on the use of a discrete place may be overly-categorical and to more fully understand the nature of slave trading, we should look beyond the use of specific spaces and places for the sale of captives and consider other contexts within which this activity took place.
In circumstances where slave traders were transporting captives by sea, slave sales might have taken place not at formal marketplaces but rather at the harbourside or even from ships themselves. In colonial Charleston, the sale of enslaved Africans as part of the transatlantic trade often took place within merchants' yards, warehouses, and on wharves. 69 In early nineteenth-century Rio de Janeiro, captives were similarly kept in and sold from houses on Valongo Street, located in the immediate vicinity of the wharf that was the disembarkation point for newly arrived slave ships. 70 Not all marketplaces, however, would have been located in major coastal ports, nor would they have been used permanently or even regularly. In colonial South Carolina, from 1723 four-day fairs held at the hinterland settlements of Dorchester, Ashley Ferry Town, and Childsbury provided opportunities for buyers to purchase both captives and other merchandise. 71 During the nineteenth century, Balangingi slavers operating in the 'Sulu Zone' would similarly sell their captive cargoes at seasonal markets and trading posts in addition to major, permanent marketplaces. In other cases, captives were sold directly from the raiders' ships, usually to creditors who provided supplies for raiding voyages. 72 When considering the potential morphology of slave market sites, it is important to note that, in many contexts, the number of captives purchased at any one time was usually small. Citing sale lists for the slave ship Hare, Kenneth Morgan has shown that the majority of captives purchased during two sales in Charleston, in March 1755 and July 1756 were sold singly or in pairs. In regions where slave trading was less-centralised, such as the Chesapeake, the number of single sales was even higher, with more than a third of captives being purchased singly by the early eighteenth century. 73 Zephyr Frank and his colleagues have similarly highlighted the polyfocal and small-scale nature of slave trading in Rio de Janeiro through an analysis of 408 slave sales made during February-May 1869. They found that rather than being confined to a formal marketplace, sales of individual captives took place across every neighbourhood in the city. 74 The decentralised nature of slave trading indicates that the concept of the 'slave market', in many cases, might be expanded to encompass the entire urban fabric of a town or city.
Given that many Viking-Age captives were acquired through raiding and warfare, it is finally necessary to explore slave trading during conflict. Perhaps the most obvious comparative context for this is the Classical world, especially the extensive trade in captives that resulted from the consolidation and expansion of the Roman Republic and early Empire. The enslavement of defeated enemies during military campaigns provided a major source of captives for the Roman state, and the legions were often accompanied by merchants (mercatores) who would purchase prisoners from the army following battles. 75 The mercatores operated outside of the military's control and, with few exceptions, were not afforded any protection while in the field, but their symbiotic relationship with the army is clear. While the mercatores provided the soldiery with a rapid source of extra income and the opportunity to unburden themselves of large numbers of captives, the taking of large numbers of prisoners would have created favourable conditions for merchants who could purchase them at low prices. 76 Given the lucrative potential for expansionist warfare to generate quantities of prisoners who might be sold into captivity, both temporary camps and more permanent fortifications occupied by armed groups might have served as epicentres for slave trading in numerous ancient and historical contexts.

Viking-Age slave markets in cross-cultural perspective
This discussion has focused on the nature and scale of slave trading as it has varied cross-culturally and through time, and there are several trends that might be applied to the study of the Viking Age. First, given the balance of the evidence explored here, it seems that sites associated with historic slave trading were rarely self-contained or dedicated solely to this purpose. In many cases, the spaces used for selling captives were also used for selling an array of commodities. The sites of slave markets, furthermore, might have been used for this purpose only periodically or sporadically. Second, the evidence suggests that slave market sites varied in scale, ranging from large public forums to private properties. As such, slave trading could have been ingrained into the very fabric of any settlement, with the sale of captives potentially taking place in stores or merchant's yards, in private properties, or even on the streets. Lastly, it is important to consider that slave markets were not necessarily permanent or even terrestrial sites. The use of temporary or seasonal markets for slave trading implies that the sale of captives in many contexts did not require the establishment of any permanent infrastructure. In circumstances where ships or other vessels were used in slave taking and transportation, the sale of captives might have taken place on wharves or even aboard ship. We also should not rule out the possibility that captives could be sold from one ship to another while at sea.
Given the general dearth of archaeological evidence for slave markets, it is perhaps not unreasonable to question whether we should realistically expect to identify these sites based on the study of material evidence at all. Rather than actively attempting to locate the sites of slave markets themselves, archaeologists might do better to instead focus on 'proxies' for slave trading within broader mercantile contexts, which can in turn provide information on systems of captive exchange and sale. In light of the evidence presented here, it is clear that the developing urban centres of the Viking world must be considered as focal points within slave-trading networks. Archaeological evidence indicates that these sites attracted merchants from far and wide, providing access to distant markets. Bringing captive cargoes to these locations may have offered traders significant financial benefits. In addition to centralising the slave trade and allowing merchants to become familiar with market conditions and the demands of buyers, the monitoring of these sites by elites may have served as a stabilising influence that facilitated commerce. A number of major Viking-Age port towns, including Hedeby, Birka, and Kaupang, seem to have been associated with royal estates located a short distance from the settlements themselves. 77 The ninth-century Vita Anskarii also describes Birka as being administered by a royal bailiff, who lived in close proximity to the town and who was probably responsible for both the defence and economic administration of the port. 78 While this might have led to the imposition of taxes on merchants wishing to do business, these costs would presumably have been offset by the prospect of a more regulated trading environment, making these sites attractive as trading locales.
It is important to stress that we should not expect to find purpose-built slave markets at these locations. While major settlements such as Dublin, Birka, or Hedeby might well have possessed spaces that functioned as public marketplaces, these probably would not have been used solely for slave trading. This in part reflects the likelihood that most Viking-Age merchants would not have invested their entire capital in captives; as noted by ibn Fad  lān, the Rūs slavers that he encountered on the Volga river brought with them not only captives but also a cargo of furs and other goods. 79 It is also possible that captives were held in the properties of merchants. If we again turn to the account of ibn Fad  lān, we see that the captives accompanying the Rūs merchants lived alongside their captors and were sold from the houses that they built for themselves on reaching their destination. 80 Though open to interpretation, the evidence from Dublin might similarly suggest that animal pens could have been used to temporarily house captives. Of further note are the shackles from Hedeby, recovered during excavations of the harbour complex. Sven Kalmring has suggested that Hedeby's wharves functioned as the town's major marketplace, and these finds could imply that captive cargoes were held and sold on wharves or directly from ships in a manner similar to that seen in colonial South Carolina and the Sulu Zone. 81 If so, then private properties, harboursides, and ships themselves might be understood as performing the function of slave markets.
It is also necessary to look beyond nodal port settlements if we are to identify the multiplicity of locations where slave trading might have taken place. The evidence indicates that seasonal or temporary marketplaces and fairs were important sites within interconnected slave-trading networks. This is especially relevant within the context of the Viking Age, given that only a very small percentage of the Scandinavian population lived in large, urban centres. Slave markets might have emerged, even temporarily, in littoral or riverine contexts where itinerant merchants could beach their ships, with the arrival of the merchants themselves perhaps serving as a catalyst for the establishment of a shortterm marketplace. 82 Regular public assemblies and fairs, furthermore, provided important opportunities for rural communities to gather in order to settle legal issues, socialise, and to establish temporary marketplaces in order to trade. 83 Possible evidence for trading activity within the context of an assembly site has recently come to light in the Roma district of Gotland, location of the Gotlandic public assembly, where metal weights of the kind used in bullion transactions and fragments of dirham coins have been found. 84 Given the potential for trade in urban environments to have been regulated by elites, it might not be unreasonable to expect slave trading at markets associated with major assembly sites, and it is once again worth noting that the sale of Melkorka in Laxdaela saga took place at an assembly attended by members of the aristocracy and rulers. 85 If assembly sites did indeed serve as slave markets, then this argument also has significant implications for the sale of captives in the North Atlantic colonies such as Iceland, where a lack of nucleated settlements meant that regular assemblies provided an impetus for dispersed populations to come together and trade.
The brief consideration of the relationship between slave traders and the armies of the Roman Republic and Empire also has implications for another potential type of slave market. Historical sources indicate that large, migratory Viking fleets operating in northern and western Europe often established temporary bases from which to raid, and there are numerous instances where Viking groups operating out of such locations are recorded as taking captives. 86 While these captives may have been taken in order to provide various services within the camps themselves, it is also possible these sites served as central slave trading nodes. 87 As noted above, potential evidence for trade in the form of silver and weighing equipment has been found at or in the vicinity of a number of known and purported Viking camps that were established in Ireland and England from the mid-ninth century. 88 Many of the sites established by Viking groups operating on the coast and inland waterways of Irelanddescribed in contemporaneous sources as longphuirt (sing. longphort, or 'shipbase')were established only as temporary encampments, but some such as that at Dublin evolved to become major towns. Questions relating to the subsistence of large Viking fleets have been raised by a number of scholars, given that at least some of these are now thought to have comprised hundreds of ships and thousands of individuals. 89 Could a trade in captives, gathered by Viking groups engaging in raiding and mercenary activity, coercion, or trade, have contributed to this? The location of encampments and longphuirt would have been widely known, and their situation on major riverine routes and the coast would have provided numerous opportunities to engage in mercantile activity. Like the later West African 'slave castles' of the transatlantic slave trade, emergent markets such as that at Dublin might have functioned as places where captives and other goods could be concentrated and sold to merchants for transportation elsewhere. 90 This could have allowed large Viking forces to contribute to their own subsistence while in the field.

Conclusion
This study has explored the comparative archaeology and history of Viking-Age slave markets. The discussion was intended to shed light on the means by which captives were distributed through extensive slave-trading networks in order to fulfil the demands of the markets that are known to have existed across the early medieval world. Given that slave markets are an essential feature of many slaving systems, the study of these sites has important implications for our understanding of the processes associated with the acquisition, transportation, and sale of captives as well as of the institution of enslavement itself.
The study explored the evidence for slave trading and slave markets in a number of comparative contexts. From this, it is clear that slave markets and slave trading practices have varied across time and space. While in some contexts slave dealers constructed purpose-built compounds and structures to facilitate the sale of captives, the selling of captives in most cases required no formal infrastructure to support it. Indeed, in many societies, it seems that almost any public space, marketplace, private dwelling, or place of business might have served as a slave trading locale. This potentially opens up the concept of the Viking-Age slave market to include not only the marketplaces of major trading settlements but also individual farmsteads and halls, the booths of traders who frequented public assemblies, harbours, wharves, beach markets, ships, and the camps established by Viking groups operating in the field. The archaeological ephemerality of slave trading, however, means that captive groups are unlikely to be well-represented in the material record. The evidence for slave trading may therefore often lie unnoticed amongst the 'noise' of archaeological data, especially in contexts where this took place on a relatively small scale alongside other commercial activities.
Given that relatively little detailed research into Viking-Age slaving practices has taken place to date, this study highlights several avenues of research that require further consideration. First, it is clear that detailed study of the locales used for trade and exchange is necessary in order to identify potential proxies for slave trading and the activities that accompanied it. This would provide invaluable insights into the construction and maintenance of the trade networks through which captives were being transported and sold. A second significant issue concerns the acquisition and transportation of captives themselves. Given the importance of seaborne raiding and maritime transportation during the Viking Age, further work is necessary to understand the organisation and operation of piratical raiding fleets as well as their role in bringing captives to market. More detailed comparative studies of early-modern pirate groups, such as the Barbary Corsairs and the Balangingi slavers of the Sulu Zone, may provide useful insights into the acquisition and transportation of captives by Viking groups as well as the relationships that raiders maintained with terrestrially-based merchants and elites. Related to this, there is a need to better understand the conditions and motivations that underpinned the acquisition, trade, and exploitation of captives, not only from an economic perspective but also in terms of the socio-political mechanisms that drove raiding groups to engage in slaving activities.
Although captivity and slavery remain an enigmatic aspect of daily life among Viking-Age Scandinavian societies, this study has demonstrated the potential for comparative research to provide insights into this practice and the ways by which it was enacted. While the general archaeological invisibility of captive peoples will always render it difficult if not impossible to identify these groups in the material record, it is, as Elizabeth Fentress has asserted, 'high time that we started looking'. 91 It is hoped that additional cross-cultural studies will allow archaeologists to shed further light on the institution of enslavement as it manifested in both the distant and more recent past, and in doing so to contribute to a greater understanding of how this practice continues to impact contemporary societies today.