in C. De Vito and A. Lichtenstein eds, Global Convict Labour (Leiden: Brill, 2015), 79-107 Penal enslavement in the early middle ages1 Alice Rio Penal enslavement in the early middle ages is a surprisingly neglected topic. It is often mentioned in passing in studies dealing with slavery during this period, but rarely in more than a paragraph noting its existence. Yet it was a very lasting and ubiquitous practice: examples of it can be found throughout the early middle ages, and even later, in virtually every region of Europe. It had also existed in the Roman world, but one should be wary of taking this as a sign of direct continuity. It was a general European phenomenon, including in areas which had never been part of the Roman empire, such as Ireland and Scandinavia. Even in ex-Roman provinces, it operated with a different logic, and fulfilled different functions in the early middle ages from those it had under Rome. It applied in cases where someone proved unable to pay the compensation owed to someone they had wronged, in effect blurring the line with debt slavery. Compensation payments and servitude were interchangeable alternatives, as opposed to the late Roman situation, where each was formally mandated for different social 1 I am extremely grateful to Alice Taylor for letting me try out chunks of this chapter on her as it was being written, and for discussing it with me; I also give very warm thanks to David Carpenter, Thomas Charles-Edwards, Wendy Davies, Paul Fouracre, Jinty Nelson and Chris Wickham for commenting on drafts, and to Graham Barrett, Daniel Hadas and Jon Jarrett for pointing me to cases I didn’t know about. This chapter was written during research leave obtained thanks to a Philip Leverhulme Prize.  ranks.2 Unlike under Rome, it was no longer intrinsically attached to legal solutions achieved through state involvement, nor did it imply any particular kind of work as its outcome, such as work in the mines or participation in public works. Whereas Roman law had maintained a conceptual distinction between slaves and convicts, early medieval penal servitude was framed in fundamentally the same terms as other forms of unfreedom. It was not just “like” slavery as a result of dehumanisation or violent treatment: it was slavery in a very explicit sense – even if, as we shall see, it amounted to a rather distinctive variety of it in practice. 3 In the specific form it took during the medieval period, penal enslavement therefore amounts to a strikingly new phenomenon. How did such a system come about, and what functions did it serve? The argument I would like to put forward here is that penal enslavement is much more explicable when seen in the broader context of punishment and peacekeeping, rather than simply in the context of slavery. It differed from other means of procuring unfree labour to a notable extent. Capture and sale were one-to-one events: they essentially involved two main parties, enslaver and enslaved. Even when sellers were involved, these events do not seem to have been of much concern to the wider 2 On Roman penal servitude, see Fergus Millar, “Condemnation to hard labour in the Roman empire, from the Julio-Claudians to Constantine,” Papers of the British School at Rome 52 (1984), 124-147, and now the chapter by Groen-Vallinga and Tacoma in this volume. The Visigothic laws are the only post-Roman laws to assign different penalties to humiliores and honestiores, but there is no evidence for actual use of these particular qualifications in the early medieval era. 3 I use the word here to denote unfree legal status; that is not to say medieval varieties of “slavery” were not often very removed from what we might think of as slavery in practical terms. 2  community. Penal enslavement, by contrast, was deeply embedded in local processes of dispute settlement in which communal participation played a vital part, and as a result involved many more distinct interest groups. Thinking about it as just another mode of enslavement means taking an approach too narrow to make sense of the logic of the process as a whole: it was much more complicated than the oppression of a single party by another. Although the successful plaintiff certainly benefited from it, since he or she acquired either a new dependant or the proceeds of their sale, pleasing the enslaver was far from being the only function of this practice. Penal enslavement did not amount to a master plan to provide labour gratis to the rich and powerful; rather, its fundamental aim was social control taken in a wider sense.4 That many people apart from the enslavers themselves accepted and found a use for this practice may be inferred from the complete absence of criticism of it, which would otherwise seem remarkable. Absence of criticism of slavery itself was, of course, the norm during this period; but the crossing of the line between free and unfree, especially when it went in a free-to-unfree direction, was seen as deeply disturbing and worrying, and was often commented on very negatively.5 Carolingian capitularies and other similarly high-minded pieces of legislation abound with 4 Contra Pierre Bonnassie, “The survival and extinction of the slave system in the early medieval West (fourth to eleventh centuries),” in Pierre Bonnassie, From Slavery To Feudalism in South-Western Europe, tr. Jean Birrell (Cambridge, 1991), 1-59: 36. 5 See Alice Rio, “Self-sale and voluntary entry into unfreedom, 300–1100,” Journal of Social History 45:3 (2012), 661–685. 3 admonitions to the rich (potentes) that they should not oppress the free poor (pauperes). On the face of it, one would expect penal enslavement to push all the legislators’ buttons, and to call up many of their bêtes noires: uncertainty and temporariness of status; mixed statuses, free and unfree, within a single family group; the poor slipping into unfreedom in cases where a wealthier and more powerful person would have remained free. What did penal enslavement have to offer that managed to generate so much consensus where there was so much scope for disturbance? What justified it and made it necessary in the eyes of contemporaries? Discussing penal enslavement in early medieval Europe in general is obviously problematic when local practices were so varied, even at the sub-regional level; at the same time, the paucity of evidence precludes a systematic regional comparison. The following is not intended to provide a standard grid applicable to all times, regions and cases within this period, but to present a possible interpretation of some of the core tensions and parameters involved. Compensation and penal enslavement Although details differed geographically, all regions of early medieval Europe shared some important basic traits in their approach to punishment and the process of dispute settlement, insofar as all, to a greater or lesser extent, had very weak structures of enforcement, and faced similar practical challenges. One of these shared traits is the practice of private vengeance or self-help, enlisting the support of friends and family members (I will call this “feud” for short, though some 4  historians prefer to give feud a narrower definition 6). Another is the finely graded system of compensations payable by someone convicted of committing a wrong. The two were profoundly interlinked, and relied to an equally considerable extent on the two most crucial kinds of social grouping available in early medieval Europe: kin and local community. Neither of these, however, should be understood as institutions, nor even as being particularly organised or coherent. “Family” usually involved a small group, though a more extended kin network could sometimes be summoned into existence on particular occasions and for particular purposes. At the outer edges of the widening circles of family relationship, feelings of responsibility and support could be very loose indeed, and solidarity within them was far from an automatic given. Similarly, a “local community” was never a homogeneous whole; throughout this chapter I will use it as shorthand for the collection of varied interest groups, neighbours, friends and enemies who made up such communities – all of whom could play a decisive role in supporting or undercutting any of their other members, whether through action or inaction. Feuding could generate an extraordinary amount of disruption and cost to family members (who might be called up for participation, or indeed become identified as targets themselves), as well as to the rest of the local community. The process of feuding, while internally endless in principle, 6 Guy Halsall, “Reflections on early medieval violence: the example of the ‘blood feud’,” Memoria y Civilización 2 (1999), 7-29. 5  usually came under enormous external pressure urging its resolution, often enabled through the action of third parties. Compensation payments were part of this settlement process. They were not a mandatory replacement for physical retaliation, but an alternative to it. Compensation payments, indeed, required feud as their backdrop: the incentive to pay was linked to the threat of continued hostility in case of default. The clearest statement of this can be found in Lombard laws, which excluded women from receiving compensation payments for the death of their kinsmen, and linked this to their incapacity to pursue feud directly.7 One corollary of this is that those wrongdoers who did not have sufficient power or support to continue to fight their corner through violence would also have been those most easily pressured to offer compensation instead, and to be made to pay the full price for it. They were also, for the same reason, less likely to be able to afford such compensation. The amounts stipulated for compensation payments in the various post- Roman law-codes were, by and large, extravagantly high. Sums to the tune of hundreds of gold coins were routinely listed for homicides (the amount varied according to sex, status and circumstances). It is difficult to see these as anything but unaffordable to the vast majority of the population. On the face of it, such high sums would seem to defeat the whole point of compensation: if the amount prescribed was so high no one could pay it, and the price for default was either death or enslavement 7 Liudprand 13, ed. Friedrich Bluhme, Leges Langobardorum, MGH Leges (Hanover, 1868). 6  depending on the wishes of the plaintiff, there would have been little incentive for any culprit to go to judgement, and it is hard to see what such a system might have contributed to the settling of disputes. However negotiable such payments may have been in practice, the inescapable consistency with which laws in all areas of Europe stipulated such enormous sums requires explanation. This is only a problem, however, if one assumes that compensation payments were in fact to be met by the person who had been found guilty. In practice, it seems the tendency was the opposite: as with so many other forms of legal process during this period – such as oath-helping, the practice of offering guarantors, and indeed feuding itself –, compensation payments required the activation of networks of support. This support could come from kin. One of the more bizarre clauses of Salic law, for the Frankish kingdoms, explained how a killer who could not pay compensation should go about passing on the debt to his family: apparently he had to swear with twelve oath-helpers that he could not pay, then go and take a fistful of earth from the four corners of his house, then, while standing on his doorstep, throw it over his left shoulder onto his closest relative, and finally jump over his fence with a stake in his hand and wearing only a shirt (perhaps looking silly was the price to pay here).8 But support could also come from elsewhere. A remarkable ninth-century document from Reichenau lists people who apparently clubbed together 8 Pactus Legis Salicae 58, ed. Karl A. Eckhardt, MGH Leges I, 4, 1 (Hanover, 1962); Patrick Wormald, The Making of English Law (Oxford, 1999), 47. The expectation of family participation is also clear from Chilperic’s edict, which orders the man who cannot pay and whose family is unwilling to help to be handed over to the victim: Capitularia regum Francorum, ed. Alfred Boretius (Hanover, 1883-1890), vol. I, no. 4, c. 8. 7  to stand surety for each other’s compensation payments to spread the load.9 This was evidently a tightly knit network of support, though comparable cases in Icelandic sagas show that a large number of people might be persuaded to contribute to a compensation payment, in a bid to encourage and keep the peace.10 Local authority figures, especially if they had brokered the settlement, might also offer to pay a share. The actual amount to be handed over could also be scaled down through compromise, particularly if both sides had engaged in hostile actions which could be taken to cancel each other out. A deal of precisely this kind was made in the best-documented and most frequently discussed early medieval feud, the conflict between Sichar and Chramnesind in late sixth-century Touraine, narrated by Gregory of Tours.11 The conflict started at a Christmas party where everyone had such a good time that a servant ended up dead; things went downhill from there and resulted in a succession of retaliatory acts, including murder, 9 Stefan Esders, “Wergeld und soziale Netzwerke im Frankenreich,” in Steffen Patzold and Karl Ubl, eds., Verwandtschaft, Name und soziale Ordnung (300-1100) (Berlin, 2014), 141-160; ed. Hansmartin Schwarzmeier, “Ein Reichenauer Schuldregister des 9. Jahrhunderts. Ein Beitrag zum Überlieferungsproblem der Reichenau,” in Helmut Maurer, ed., Die Abtei Reichenau: Neue Beiträge zur Geschichte und Kultur des Inselklosters (Sigmarigen, 1974), 17-30, at 20. On sureties, see Wendy Davies, “Suretyship in the Cartulaire de Redon,” in Thomas Charles-Edwards, Morfydd E. Owen and Douglas B. Walters, eds., Lawyers and Laymen (Cardiff, 1986), 72-91. Such networks of support were also expected to keep the person they were supporting to account. The institution of frankpledge, which created groups of people and made them account for the actions of its members, unique to late Anglo-Saxon England, could be seen as a more formalised, and more compulsory, way of dealing with the same concern, in such a way as to leave no one outside some form of local group responsibility; but group solidarity did not automatically follow, since the group were expected to seek out and hand over a culprit rather than necessarily to help them (on frankpledge see Patrick Wormald, Legal Culture in the Early Medieval West (Oxford, 1999), 54-56). 10 William I. Miller, Bloodtaking and Peacemaking: Feud, Law, and Society in Saga Iceland (Chicago, 1990), 276, 278. 8  appropriation of goods and arson. Gregory himself tried to broker a settlement, and offered church funds to help it along. Although this first attempt at settlement failed, church funds remained part of the deal when the dispute was finally settled by judges (judices). Sichar’s payment was also discounted to compensate for Chramnesind’s own actions. Although compensation payments were unaffordable for the vast majority of people who might have to make them, there were thus many layers of safety nets in place. The ability to pay rested on the same conditions as the ability to pursue self-help successfully: kin and community networks of support. Unaffordability, far from diminishing the effectiveness of compensation payments, was very likely their whole point. It was meant to ensure that no single individual could ever pay them on their own, and to avoid anyone getting the bright idea that they could independently and unilaterally decide to engage in disruptive and wrongful behaviour, simply on the basis that they could afford to pay: hardly anyone could, and even those who could certainly could not do it easily. This would have been enough to create an enormous incentive to get on well with one’s family and with at least a substantial number of (preferably elite) local inhabitants. The burden of having to muster interested parties to participate in compensation payments was aimed at creating a real social 11 Gregory of Tours, Histories VII, 47, ed. Bruno Krusch, MGH Scriptores rerum Merovingicarum I, 1, at 367- 8; the end of the story is told at IX, 19. See inter alia J. Michael Wallace-Hadrill, “The bloodfeud of the Franks,” in J. Michael Wallace-Hadrill, The Long-Haired Kings (London, 1962), 121–147; Halsall, “Violence”; Philippe Depreux, “Une faide exemplaire? À propos des aventures de Sichaire: vengeance et pacification aux temps mérovingiens,” in Dominique Barthélemy, François Bougard and Régine Le Jan, eds., La vengeance, 400-1200 (Rome, 2006), 65-85. 9  as well as economic cost for the culprit, who would have had to call in many favours. It was also about spreading responsibility beyond the individual to the collective level – which also implied a greater urgency for internal policing within the support group of its more rash members: a network of support, however helpful it might be in isolated instances, could not, and would not, keep paying compensation on a repeat basis without bankrupting itself. It is worth stressing that such local support does not imply local consensus. One needed only to demonstrate that one had enough support that failure to settle in an acceptable way would lead to very unpleasant consequences for the community. Æthelstan expressed particular worries about the extreme case of the man “so wealthy or of such reputation” that he could not be brought to justice, and ordered that he should be led with his wife and children to a place of exile of the king’s choosing: the expectation was clearly that severing his local ties with the network that had supported him was the only way to remove this unfair advantage.12 Compensation payments were at least partly, then, meant as a test of local and familial standing, as well as of the extent of support backing up any particular individual. Still, if there was a test, an irremediable corollary of it is that some people will have failed it; this is, after all, the point of tests. Some would inevitably fall through all the holes in the safety nets, and the 12 IV Æthelstan 3, ed. and tr. Frederick L. Attenborough, The Laws of the Earliest English Kings (Cambridge, 1922). 10  capabilities and choices of various kinds of third parties were absolutely crucial in allowing this to happen or not. There were several different ways to fail in the compensation payment stakes. One way was simply not having available networks of support in the first place, or having networks of support so socially and economically weak in relation to the wronged party that they could neither put together the amount of the compensation nor put up much of a fight if hostilities were prolonged. For the same reason, these, of course, were also the people least likely to elicit much worry among the local community and authorities over how much trouble they could make if no settlement was reached, and therefore least likely to persuade anyone to share in the payment for the sake of peacekeeping. One case provides a neat contrast to the Sichar and Chramnesind story, and it also involves Gregory of Tours, in a poem addressed to him by his friend Venantius Fortunatus.13 Any process of dispute settlement so profoundly linked to informal negotiation and self-policing by the local community obviously implies that those who can find little material support within it will tend to lose out. The whole process was generally geared towards offering options to third parties over the extent of their participation and support in any particular case, and giving them the capacity to fine-tune their involvement, at various points in the proceedings, according to a wide range of different possible terms. Oath-helping and standing surety as a guarantor, as well as help in paying compensation, all played a part in 13 Fortunatus, Carmina V, 14, ed. Friedrich Leo, MGH Auctores antiquissimi IV, 1 (Berlin, 1881). 11  this: although oath-helpers were not automatically supposed to know anything about the case, it did mean something that one was able to summon them at all, in that it was a mark of support. Finding guarantors to put up securities on your behalf was a similar mark of local standing and insertion in networks of solidarity – which also meant networks of internal policing. Making local support the litmus test of the viability of someone’s case in court necessarily meant that both kin and community had to be offered ways to desolidarise themselves from the accused if they so wished.14 Penal enslavement was therefore a necessary corollary of the structure of punishment during this period, and of the crucial role played by kin and community in the keeping of order. It was implied within that structure, much as the ordeal, typically undergone when the accused was of low status or an outsider to the community, was the flipside of oath-helping. 15 The crucial advantage of this practice was that it offered third parties means of reaching a definitive settlement to the dispute even when they could not, or did not choose to, get involved as far as to offer compensation. Even failure to get a compensation payment together, however, did not mean total failure, or lack of room for negotiation. Penal enslavement, although (like death) certainly at the bottom of the scale of desirability as 14 Outlawry seems to have fulfilled a similar function (for Iceland: Miller, Bloodtaking, 238-239; on penal servitude in Scandinavia: Ruth Mazo Karras, Slavery and Society in Medieval Scandinavia (New Haven, MA, 1988), 52-55). 15 Robert Bartlett, Trial by Fire and Water (Oxford, 1986), 29-33. 12 legal outcomes went, was (unlike death) an adaptable solution: it could happen, or be prevented, on many different terms, and at different stages of the process, leaving room for many possible deals between all parties. It is therefore important to consider not only those cases which eventually ended in penal enslavement for the offender, but also those which might have led to it, but did not, because the parties involved managed to cut a deal. The sequence of options If an offender did fall through all the holes in the various available safety nets (that is, if they could not find oath-helpers or witnesses to clear them; if they could not negotiate an alternative settlement or a discount of their compensation payment with their adversaries; and if, having lost the case, they also failed to secure help from kin, guarantors or local authorities, and the outcome of the case was pronounced to be penal enslavement), even then the matter did not necessarily end there: many different ways of handling the situation still remained. The condemned could offer their property at this late stage, even if it did not amount to the full value demanded, in a bid to dissuade plaintiffs from enacting the full strength of their entitlement. In a strange case from north-west Spain, dated 858, the woman Letasia confessed to having committed adultery with the servus Ataulf, during which affair the couple also managed to consume four cows as well as a quite extraordinary amount of cheese (sixty rounds), none of which belonged to them. She offered all her property to Ataulf’s master in compensation for both the 13  cheeses and the affair, and was not enslaved, as she might have been. 16 In a case from León, from 994, a widow called Cida Aion, who had been caught in adultery with someone else’s husband, was condemned to serve as an ancilla origenale (“as if she had been born a slave woman”). She subsequently managed to avoid this fate by giving all her property (except for her husband’s property, which now belonged to her children). 17 In a Catalan agreement from 988, a certain Sentemir, who came very close to penal enslavement for hiding the testament his brother had made in favour of the monastery of Sant Cugat, managed to get away without even having to give all his property, in exchange for a gift of land and imploring for mercy.18 In these last two cases, no payable compensation is mentioned, and enslavement was read as a literal consequence of statements in the much earlier Visigothic law stipulating it as the punishment for the offences committed. 19 Such stipulations seem to have been especially emphasized in Spain, but could be just as often, it seems, exploited to obtain other valuable things instead: here as in the case of 16 Tumbos del Monasterio de Sobrado de los Monjes, ed. Pilar Loscertales de García de Valdeavellano (Madrid, 1976), no. 75. 17 Colección documental de la Catedral de León, ed. José Manuel Ruiz Asencio, vol. 3 (León, 1987), no. 561. 18 Cartulario de “Sant Cugat” del Vallés, ed. José Rius Serra, vol. 1 (Barcelona, 1945), no. 218. Adam Kosto, Making Agreements in Medieval Catalonia: Power, Order, and the Written Word, 1000-1200 (Cambridge, 2001), at 48-49. 19 The clause for Sentemir’s case is Liber iudiciorum VII, 5, 2, ed. Karl Zeumer, Leges Visigothorum, MGH Leges I, 1 (Hanover, 1902); it in fact stated that enslavement should be the punishment only for humiliores, whereas potentiores would forfeit a quarter of their property. 14  compensation, property and freedom could each be substituted for the other. Sentemir’s contrition was not simply a matter of expected style; it was fully a part of making it up to the monastery for the attempted deception. Beyond the practicalities of the agreement, form – and face – mattered. In another case, this time from late eleventh-century Touraine, a man named Martin Tireuil attacked the prior of St Martin of Tours, Hatto, on his way back from vespers, and stole his horses, forcing him to go home on foot; he then burned down some houses where monks were living. He later presented himself before the abbot, naked, with bare feet, and carrying rods to be beaten with, begging for mercy, and declaring himself ready for whatever punishment the monks thought appropriate. Not having the money to pay compensation for his actions, he agreed to give himself to the monastery as a substitute (cum pauper nimium esset, et nullam substantiam haberet unde tanti dampnum forisfacti recuperare posset, se ipsum pro emendatione illa tradidit). As Dominique Barthélemy has pointed out, the monks then received him not as a serf, but in a looser kind of association, a “half-measure,” with no mention of rituals typical of entry into servitude: his very public act of contrition (in addition, perhaps, to the fact that the monks had earlier been responsible for the death of his son) may have been enough to pressure them into showing mercy.20 20 Loir-et-Cher, Archives départementales, Blois, 16 H 118, no. 11 (= ARTEM no. 2277, http://www.cn- telma.fr/originaux/charte2277/); discussed in Dominique Barthélemy, The Serf, the Knight and the Historian, tr. Graham R. Edwards (Ithaca, NY, 2009), 61-62, with quote at 62. Penal unfreedom would certainly have been an 15  A late eleventh-century case from the Bodmin Gospels similarly highlights the importance of rituals of submission even when the accused was able to redeem himself: Putrael, who was about to be penally enslaved to one Ælfric for an unspecified wrong, had to plead with Ælfric’s brother to intercede in order to allow him to pay in cash and kind instead: in this case allowing him to redeem himself at all was being presented as a great favour, and every effort was made to make him understand that he had had a very narrow escape. His humiliation was witnessed by a correspondingly large number of important people, both lay and clerical.21 Failing the capacity to make an immediate post-judgement deal of this kind, the condemned would be handed over directly to the claimant. It seems to have been largely left up to the latter whether this would mean death, unfree service, or sale on to a third party. Penal enslavement to the offended party could result in very different scenarios. The working and living conditions associated with penal enslavement were as varied as those associated with early medieval unfreedom in general. Its consequences in terms of lived experience partly depended on what the option, as is clear from Le livre des serfs de Marmoutier, ed. André Salmon (Tours, 1864), no. 105 (a. 1062), at 99; for an earlier case (from 945), Recueil des chartes de l’abbaye de Cluny, eds. Auguste Bernard and Alexandre Bruel, vol. 1 (Paris, 1876), at 622-623, no. 669. 21 Ed. Max Förster, “Die Freilassungsurkunden des Bodmin-Evangeliars,” in Niels Bøgholm, Aage Brusendorff and Carl Bodelsen, eds., A Grammatical Miscellany offered to Otto Jespersen on his Seventieth Birthday (Copenhagen and London, 1930), 77-99, at 93, no 33; see David Pelteret, Slavery in Early Mediaeval England (Woodbridge, 1995), 151. 16  guilty party had to offer. Free smallholders, with land to hand over along with themselves, seem to have mostly continued to occupy and cultivate whatever land they had previously owned, with the difference that they could no longer dispose of this property freely and would have had to hand over a share of their crop. But this minimal-disruption model did not apply in all cases: the young girl from Fortunatus’s poem, who had little to offer but herself, was more likely to become a domestic slave than anything else, and it is clear that some penally enslaved persons were bought and sold away from their communities (indeed, their removal was sometimes the whole point).22 Plaintiffs, for obvious reasons, were not always keen to keep an offender in their own service. Gregory of Tours tells the story of a priest whom Bishop Aetherius of Lisieux had to redeem twice: he gives more details on the first case, which had involved his running away with a woman from a good family, and trying to pass her off as a man by cutting her hair short and dressing her in men’s clothes. Her family caught up with him, and were prepared to kill him unless someone bought him (he got out of this lightly in comparison with the woman, who was burned alive). Gregory, though he does not go so far as to say so explicitly, seems to indicate it might have been better if they had killed him (the same priest went on to try to murder his benefactor, which explains Gregory’s limited sympathy), and he comments in a rather judgemental way on the woman’s 22 See below, 50, 53, 59. Lombard law only allowed iudices to sell thieves abroad if they had been caught three times: Liutprand 80 (ed. Bluhme, Leges Langobardorum). 17  family being motivated by hunger for gold (sicut cogit auri sacra famis).23 Accepting compensation, or a redemption price, was not always up to the job of restoring honour in particularly egregious cases. Sale to a third party also offered different possible consequences. It could (and probably most often did) involve becoming that person’s slave instead. But third parties, like Bishop Aetherius, could also choose to restore those they purchased to full freedom. One document from Vic in Catalonia, dated 5 April 933, shows a priest, with the charming name of Nectar, redeeming a certain Felix from a woman called Adalgis, to whom he had been enslaved by judgement after killing her son, for 30 solidi.24 It is unclear how long after the enslavement this redemption took place: perhaps immediately, perhaps not. It may be no accident that the redeemer was a priest, an especially appropriate kind of person to take charge of the redemption of captives more generally. He is also said to have come with “good men” (boni homines) to make his request before Adalgis: this shows some sign of support for the condemned man, even if it did not stretch so far as to help him pay. In such cases, the practical outcome looks very similar to getting help with the compensation payment, but important differences remained. Crucially, the intervention had a different meaning: although it did amount to a gesture of kindness to the offender, it did not amount to a 23 Gregory of Tours, Histories VI, 36, at 306-7; the expression comes from Vergil, Aeneid III, 56-57. 24 Catalunya Carolíngia IV: els comtats d’Osona i de Manresa, ed. Ramón Ordeig i Mata (Barcelona, 1999), part 1, no. 392. For another case of a woman transferring her son’s murderer to a third party, from 953, see Catalunya Carolíngia IV, part 2, no. 706. 18  public act of support for his actions, or to solidarity with him in the prosecution of the case itself. It was, instead, a charitable act. The purchase price may have also been lower than the price of the compensation payment: wherever amounts are stated, they are a long way off the sort of scale involved for direct compensation – as with the 20 solidi it took to redeem the errant priest in the Gregory of Tours story above, or the 30 involved in the case of Felix from Vic. In a case from the church of Santillana, dated to 15 May 1062, a woman named Gatea, who had had a thief as her lover but failed to denounce him, was condemned to pay the staggeringly high sum of 300 solidi as his accomplice, which she predictably found herself unable to pay. She was enslaved by the judges and bought by a priest called Felix for the price of two pieces of white linen canvas, together amounting to 33 cubits – a long way off the compensation payment (whether Felix was as nice as Nectar and freed her afterwards is not said).25 Purchase of the penally enslaved by a third party was, therefore, a way of settling the case which still went some way towards compensating the plaintiffs even if the offenders found no one to help pay compensation, without having either to kill them or keep them directly in their own service. It also allowed sympathetic bystanders who would not, or could not, go so far as to offer direct support during the court case to intervene with a lower level of financial and social commitment – either by freeing 25 Documentos para la historia de las instituciones de León y de Castilla (siglos X-XIII), ed. Eduardo de Hinojosa y Naveros (Madrid, 1919), doc. XV; cited in Charles Verlinden, L’esclavage dans l’Europe médiévale, vol. 1 (Bruges, 1955), 137. 19  them, or, if keeping them, by at least removing them from the direct control of those whom they had harmed, and thus helping to ensure they would not be put to death. Needless to say, they would also have been procuring labour at the same time, but this did not necessarily detract from the sense that buying penally enslaved persons was a good and charitable action, even if the purchase did not immediately lead to redemption and freedom. At any rate, Frankish formulae (documentary models), which preserve several texts in which penally enslaved persons address their third-party buyers in order to acknowledge their new relationship with them, show that praising the pietas of one’s buyer was seen as good form.26 Temporariness and compliance The condition of the penally enslaved may indeed have been more likely to be temporary than that of other unfree persons, including in cases where plaintiffs kept the new dependants in their own service. Lombard law distinguished according to the scale of the crime: while all those ‘prodigal or ruined’ persons who could not pay compensation had to be handed over as slaves to the plaintiff, those who owed under 20 solidi were explicitly transferred only temporarily, until they could pay off their 26 Formulae Andecavenses 3, ed. Karl Zeumer, Formulae Merowingici et Karolini Aevi, MGH Leges V (Hanover, 1886), and tr. Alice Rio, The Formularies of Angers and Marculf: Two Merovingian Legal Handbooks (Liverpool, 2008), 51-52); Formulae Marculfi II, 28 (tr. Rio, 211); Formulae Arvernenses 5. On this type of text, see Alice Rio, Legal Practice and the Written Word: Frankish Formulae, c. 500-1000 (Cambridge, 2009). 20  debt.27 Elsewhere, there are also hints that they could expect preferential access to manumission, especially on the death of their new master. This phenomenon is especially visible in Anglo-Saxon England. 28 The woman Wynflæd, in a mid-tenth century will, listed a number of men and women to whom she wished to grant freedom after her death, and went on to ask her children to free for the good of her soul any other penal slaves (witetheow) whom she might have left out, going out of her way to say she had enslaved them herself.29 Highlighting her own part in their fate could be interpreted as a mark of contrition, if she felt she should free first those whom she had had a direct hand in enslaving. On the other hand, the list of names is so long that she could not possibly have enslaved them all as a successful plaintiff herself, even allowing for an exceptionally litigious personality. It is possible, therefore, that she was highlighting her role in their enslavement not out of guilt, but, on the contrary, because she was rather proud of it – if she had acquired them as a third party, as a charitable gesture. Reading this as an act of charity would not necessarily have conflicted with her decision only to free them at her death. The choice to redeem at one’s death people who had been enslaved specifically as a result of guilt, as opposed to those who had simply been born into unfreedom, may also have carried an added religious 27 Liutprand 152; see also Aistulf 22 (ed. Bluhme, Leges Langobardorum). 28 Pelteret, Slavery, 120-123. 29 Anglo-Saxon Wills, ed. Dorothy Whitelock (Cambridge, 1930), no. 3, at 10-13; = Sawyer no. 1539 (http://www.esawyer.org.uk/charter/1539.html). 21  significance: it mirrored Christ’s redemption of a guilty humanity through his own death, thus opening to it the possibility of salvation – a salvation which testators were of course hoping would be granted to themselves as well.30 Several other Anglo-Saxon wills from the mid-tenth to early eleventh centuries, both lay and ecclesiastical, similarly go out of their way to highlight the freeing of penal slaves in particular.31 The expectation of manumission was explicitly a part of the negotiations in one extraordinary case preserved in the archive of Otero de las Dueñas in Northern Spain, dated June 1022.32 A man named Enego, with the help of his mother Auria and his brother Velasco, absconded with a woman named Midona, a chambermaid (cubileira) of Count Fruela Muñoz. The document says they “stole” her by raptus (a general category which could include anything from rape to consensual elopement). The count’s men 30 The concept of Christ’s redemption in those terms was particularly developed by Augustine: e.g. De Trinitate XIII.xiv.18; original sin had enslaved humanity to the Devil, and full freedom could be restored only through Christ: De civitate Dei, XIV, 11. The link between manumission and salvation could be made through Luke 6:37, Dimittite et dimittemini (“Forgive, and you shall be forgiven,” or in a closer translation, and more to the point, “Release, and you shall be released,” cited in a manumission document in Formulae Salicae Lindenbrogianae no. 9, ed. Zeumer, Formulae, 273), echoed in the Pater Noster (Dimitte nobis debita nostra sicut et nos dimittimus debitoribus nostris). I am grateful to Thomas Charles-Edwards for this point. 31 These are the wills of Bishop Ælfsige of Winchester (Anglo-Saxon Wills no. 4, at 16-17; = Sawyer no. 1491); Bishop Ælfwold of Crediton (Sawyer no. 1492); Archbishop Ælfric (Anglo-Saxon Wills no. 18, at 54-5; = Sawyer no. 1488); the woman Ælfgifu (Anglo-Saxon Wills no. 8, at 20-21; = Sawyer no. 1484); the ealdorman Ælfheah (Anglo-Saxon Wills no. 9, at 24-25; = Sawyer no. 1485); and the Ætheling Æthelstan (Anglo-Saxon Wills no. 20, at 56-57; = Sawyer no. 1503). The Synod of Chelsea of 816 had ruled that any English people enslaved during the lifetime of a bishop were to be freed on his death (Councils and Ecclesiastical Documents relating to Great Britain and Ireland, eds. Arthur W. Haddan and William Stubbs (Oxford, 1869-78), III, 583, chapter X, cited in Pelteret, Slavery, 83). The practice was obviously not systematic, and some penally enslaved persons could be transferred with land they were living on: e.g. Sawyer no. 1285. 32 Colección documental de Otero de las Dueñas, eds. José Antonio Fernández Flórez and Marta Herrero de la Fuente, vol. 1 (León, 1999), nos. 150-151. 22  pursued them, raising the cry of raptus, and caught them, after which Enego recognised his guilt. Proceedings were then adjourned so the judges could look up what Visigothic law had to say on the matter. This, it turned out, was that if a free man took a free woman by raptus, he should be enslaved to her and her family, but that under no circumstances should the issue be settled by marriage.33 Three days later, however, when the court convened again, Midona asked not to be separated from Enego, with the result that the couple were both enslaved to the count and his wife Amuna. Crucially, a provision was added to the charter to the effect that if the count and his wife died before them, they and their future children would regain their full freedom. This provision would be annulled if they tried to escape, in which case they and their children would become exactly like the “slaves by birth” (servi originales) of Fruela Muñoz and Amuna, and would remain so forever. This case is remarkable on a number of counts, but is most telling for our purposes in the clear sense it gives that there were servi and servi, and that such nuances in the social practice of unfreedom could be framed as both incentives and sanctions, securing at one stroke a more acceptable outcome for the enslaved and greater assurances of their future compliance for their new lords. The need to find means to control the recent penally enslaved through incentives must have been felt particularly keenly: after all, these were, by definition, people who had already shown themselves willing to play 33 The text quoted in the charter is from Liber iudiciorum III, 3, 1, but the given reference is to III, 3, 2, and this latter clause, which deals with a couple who wish to stay together, is indeed more to the point. 23  fast and loose with existing norms of behaviour. It would have been especially crucial as, once someone passed into unfree status, they fell from then on under the legal responsibility of their new lord rather than their family. This, on the face of it, gave the enslaved tremendous potential scope for resistance against their enslavers – though given the punishments applicable to the unfree, it might come at great cost to them too. The issue was clearly serious enough that Charlemagne included it among the items to be added to Salic law; the same clause was reiterated in Charles the Bald’s Edict of Pîtres, and again, further afield, in the Leges Henrici in England. This law gave the lord of someone who put himself into unfree service as a pledge for debt or for failure to pay compensation, and then went on to cause damage to someone else, the stark choice of either paying compensation, in which case he got to keep his dependant, or bringing him to be tried in the public assembly, thereby treating him as a free man and losing any claim over him.34 A similar practice is hinted at in a clause of the seventh-century Anglo-Saxon king Ine, which deals with the case of a man eager to avoid going to trial by ordeal but lacking the wherewithal to get himself out of it. He could hand himself over to anyone willing to pay his pledge, but on the understanding that this same person would then be responsible for him in any future cases; if he was 34 Capitularia I, no. 39, c. 8; II, no. 273B, c. 34; Leges Henrici Primi 89.3, ed. Leslie J. Downer (Oxford, 1972). What legislators insisted lords absolutely must not do was to treat their dependants sometimes as slaves and sometimes as free, according to what suited them: II Cnut 20.1, ed. and tr. Agnes J. Robertson, The Laws of the Kings of England: From Edmund to Henry I (Cambridge, 1925); Leges Henrici Primi 78.2b. 24  later claimed on different charges and the same person failed to give pledges, the latter would lose any rights over him and the debt.35 The expectation of future release, as well as its timing, may well have constituted a crucial incentive for compliance. The key to achieving this was to find a way to make penal enslavement the end of the particular dispute, while still retaining enough flexibility in outcome to give the penally enslaved incentives to keep on the straight and narrow, and to continue to comply with the requirements of their new service. This could be done through the hope of manumission, as in Anglo-Saxon England or the case of Enego and Midona. In Francia, the temporariness of penal enslavement was treated in a different way, though it too was aimed at instilling a hope of future release for good conduct. This involved a reading of penal enslavement as a form of pledging, using one’s own status as collateral for debt (the unpaid compensation) rather than renouncing it definitively. This is already in evidence in the law of Charlemagne cited above, and several other Carolingian capitularies, when outlining enormous fines to be paid for wrongdoing (in particular involving the bannus, the fine to be paid for failure to obey the king’s command, such as refusal to join the army or to accept new coinage), insist that if the wrongdoer could not pay, he should give himself as a pledge, and recover full freedom once the fine was paid. 36 This is exactly the situation envisaged in the case of one unlucky burglar whose case was 35 Ine 62 (Attenborough, Laws). 25  preserved in a formula. He had been caught ransacking the cellar of a monastery, and had to agree to put your arm on my neck and cause myself to be handed over to you by the hair of my head before witnesses, in such a way that until I am able to return your solidi, I must be in your service and do whatever tasks you yourself or your subordinates order, and if I seem negligent or slow in this, I promise that you may order the same discipline to be inflicted upon my back as on your other servi.37 This is a very different way of approaching the issue, but one equally concerned with controlling behaviour by maintaining the enslaved person’s stake in sticking to the dispute settlement. Pledging was in theory temporary – though it was probably less so in reality, since one has to wonder how often people did end up repaying their debt. Using free status as a loan security seems to have been a fairly standard practice in other, more straightforward instances of debt as well.38 In the case of penal self-pledging, though, it had a striking additional twist, in that it 36 Capitularia I, no. 20, c. 19; no. 41, c. 3; no. 70, c. 3; no. 74, c. [1]; no. 139, c. 2; II, no. 201, c. 2. Also Concilia aevi Merovingici, ed. Friedrich Maassen, MGH Leges III, 1 (Hanover, 1893): Concilium incerti loci, c. 14, at 195. 37 Formulae Pithoei fragmenta 75 (= Formulae Bignonianae 27) (Zeumer, Formulae). Temporariness, as well as the difficulty of repayment, is again emphasised in Bavarian law, which stipulated that no Bavarian should lose his life or inheritance except for rebellion, but, if he could not compensate his victim, should serve until he had earned enough “over months and years”: Lex Baiwariorum, ed. E. von Schwind, MGH Leges I, 5, 2 (Hannover, 1926), II, 1, at 292-3; translated in Carl Hammer, A Large-Scale Slave Society of the Early Middle Ages: Slaves and Their Families in Early Medieval Bavaria (Aldershot, 2002), 78. 38 Rio, “Self-sale,” 671-2. 26 confirmed the permanence of the debt beyond the point of enslavement: all was not made square through this arrangement – indeed, rather than cancelling out the debt, it had the effect of making it ongoing, fixed in a holding pattern. It also allowed the possibility of conditional reinsertion within the local community at large. More than the hope of manumission, the possibility of future release from penal self-pledging would have encouraged and pressured the pledge-giver not only to play by the rules with respect to their new lord, but also to continue to seek support and approval from the rest of the locality, in a bid to succeed eventually where they had failed initially – namely, in securing some sort of financial backing and help from third parties in order to pay compensation. If they failed, it would presumably never be paid, and their unfree status would become in fact, even if not in principle, permanent. This made the issue of timing much less urgent, and gave everyone involved, whether third parties or plaintiff, the option to intervene at any point of their choosing during the duration of the pledge-giver’s service. The expectation that penal enslavement might be only temporary could therefore play an important part in ensuring that the agreement would be stuck to, even by those who were most disadvantaged by it. The fact that this possibility was conditional on good behaviour might have brought the practice closer to rehabilitation as it might be understood in later periods. In some cases, indeed, penal enslavement does seem to be connected with 27  penitential concerns in a Christian sense. Charlemagne, in his Admonitio generalis, famously took issue with people who wandered about in his kingdom naked, wearing irons, and generally making a spectacle of themselves on the pretext of doing penance: these people, he said, if they had committed a crime, would do much better to stay in one place, and “work and serve and do penance according to what canonical penalty has been imposed on them.”39 There is scant evidence for penal enslavement, or service in general, as a canonical penalty, but it does feature in the Penitential of Theodore, which stated that a layman’s penalty for carrying off a monk from his monastery “by stealth” was to “either enter a monastery to serve God or subject himself to human servitude.” 40 The Penitential of Finnian also stated that an adulterous woman who had left her husband and gone to live with another man, but was now repentant and wished to atone for her sin, should do so by becoming the slave of her husband. 41 Both cases are notable in that they turned penal enslavement into a way for offenders to 39 Capitularia I, no. 22, c. 79. 40 Penitential of Theodore I, 3, 1, ed. Hermann Joseph Schmitz, Die Bussbücher und die Bussdisciplin der Kirche (Mainz, 1883), 527; tr. John T. McNeill and Helena Gamer, Medieval Handbooks of Penance, new edn. (New York, 1990), 186. 41 Penitential of Finnian 43-44, ed. Ludwig Bieler, The Irish Penitentials (Dublin, 1963), tr. McNeill, Penance, 95-96. Other cases cited by McNeill as prescribing service as penance in Welsh and Irish texts (e.g. at 252, 254) do not in fact deal with unfree status, but with serving “in place of a son” the parents of the man whom one had slain, to compensate them for the filial duty they had been deprived of. Filial duty could be extremely binding in its own right, as is made clear from story of Librán in Adomnán of Iona’s Vita Columbani, II, 39, eds. Alan O. Anderson and Marjorie O. Anderson (Oxford, 1991). 28  make things right beyond the direct requirements of dispute settlement: in this penitential context, they could opt not to go through with it, although presumably only at great cost to their integration within the community. The understanding of human servitude as a direct consequence of sin was thus not only, or not always, metaphorical.42 The process of penal enslavement therefore allowed enormous room for manoeuvre, both before and after the event. Throughout the process, several options were given for varying degrees of desolidarisation from the accused, while simultaneously leaving wide possibilities for reinsertion, at the discretion of interested parties. The notion (however unlikely in reality) that penal enslavement should be only a temporary stage also presumably cushioned its impact on the offender’s immediate family. The extent to which families should be held responsible and suffer, or not, for their kinsmen’s behaviour was a difficult issue, and one which occupied kings and legislators throughout the period. The limits of responsibility: laws and family strategies Out of all the parties connected with a wrongdoer, one might expect kin to have been the most supportive. For some higher-status families, the prospect of unfreedom for one of their members must have been unthinkable, and its prevention worth any cost. Kin groups’ response thus no doubt depended partly on status and self-image. Families higher up the social scale, however, were also those for which this outcome was least 42 On slavery as a consequence of sin, see e.g. Augustine, De civitate Dei XIX.15. 29  likely in any case, since they would also have been those best placed to negotiate a special arrangement or discount on the settlement. Kin groups had a basic duty of solidarity, but this does not necessarily mean that they backed all of their members with equal assiduousness. Given a finite amount of resources, families might have to make difficult choices regarding whether, or how far, to help with compensation payments. The prospect of a family member becoming unfree might well, of course, be deeply worrying, not least because others risked being tainted by association, making them more open to accusations of being unfree themselves. But families could take a distinctly unsentimental approach to this problem too, as is made clear from a Frankish capitulary forbidding anyone from killing an unfree family member out of fear of being categorised as unfree themselves.43 If anything, the importance of kin solidarity, arguably the most fundamental cornerstone of early medieval society, encouraged a highly disciplined approach to family relationships. The fate of the errant priest’s lover in the Gregory of Tours story cited above is enough to show that family members could be the most ruthless of all possible enforcers. Kin groups, for a start, were far from being strictly horizontal organisations. Some members of the same family could be vastly more powerful than others, and their relationship with weaker members more akin to patronage than to solidarity between equals. That more distant kinsmen could exact a high price for their intervention is evident from the case of the Irishman Librán in Adomnán of Iona’s Life of Saint Columba: 43 Capitularia I, no. 39, c. 5. The penalty was death for the culprit and enslavement for his children. 30  when he was convicted of a killing and risked being put to death, a family member redeemed him, but on the condition that Librán become his unfree dependant, blurring the line between kin and other possible third parties.44 Family members’ involvement could be based on just as much strategising and calculation as any other party’s. Much of this calculation must have been based on the impact of the penal enslavement of one family member on the others. In general, legislation mostly tried to shield families from the consequences of the enslavement of one of their members. Many laws on the subject were especially concerned with protecting the unity of the married couple, which might be jeopardised under the circumstances. While most law-makers took a very dim view of mixed-status marriages in principle, in the case of enslavement of one member of an already married couple, ensuring the permanence of the marriage took precedence: the free wife of an enslaved man was encouraged to stay with her husband, and her freedom as well as that of their children was protected – though penal enslavement still 44 Vita Columbani, II, 39. This was almost certainly a member of his own family rather than the victim’s, though the text unfortunately does not make this completely unambiguous. On this story, see Thomas Charles- Edwards, Early Irish and Welsh Kinship (Oxford, 1993), 317-319; on penal enslavement in Ireland in general, Fergus Kelly, A Guide to Early Irish Law (Dublin, 1988), 97-98 and 215-6. For a particularly mysterious case, see David A. Binchy, “The Saga of Fergus Mac Léti,” Ériu 16 (1952), 33-48, at 39; cited in Robin Chapman Stacey, Dark Speech: The Performance of Law in Early Ireland (Philadelphia, PA, 2007), 69. For Wales, see the example of the thief in Vita Sancti Cadoci 33, ed. Arthur Wade-Evans, Vitae Sanctorum Britanniae et Genealogiae (Cardiff, 1944), 94-97, and two cases from the Llandaff charters, no. 127a and no. 218 (from 955), ed. J. Gwenogvryn Evans, The Text of the Book of Llan Dav (Oxford, 1893), 127 and 218-21; cited in Wendy Davies, Wales in the Early Middle Ages (Leicester, 1982), 64; Wendy Davies, An Early Welsh Microcosm: Studies in the Llandaff Charters (London, 1978), 43, 167 and 180-1. 31  counted as one of the few legitimate grounds for divorce, at least in the earlier Carolingian capitularies.45 But the expectation of solidarity and commonality of interests in a marriage could cut both ways: for instance, a wife and children might be seen as having directly benefited from a theft, bringing up the question of how far they should also be involved in the punishment. A law of Ine ruled that a thief who stole without the knowledge of his wife and children should pay 60 shillings, but that, if the wife and any children above the age of ten knew of his actions, they should all be enslaved. In the early eleventh century, Cnut gave more precise directions for ascertaining knowledge: the wife was guilty only if the stolen goods were found in places to which she kept the keys, such as her storeroom, her chest and her coffer, but not in the rest of the cottage, on the understanding that a wife could not stop her husband from bringing home anything he liked. Cnut also tried to limit the exposure of very young children to penal enslavement with their parents, saying that “those who have never tasted food” should not count as having benefited from the theft.46 However harsh this legislation might seem, and however concerned with not letting anyone get away with anything, it is striking 45 Capitularia I, no. 16, c. 6; no. 157, c. 4; no. 158, c. 1; II, no. 201, c. 15 in ms. Paris BnF 4613; no. 252, canones extravagantes 2 expressly forbade enslavement as a cause for divorce. Earlier, see Concilium incerti loci, c. 14, ed. Maassen, Concilia aevi Merovingici. 46 Ine 7.1 (Attenborough, Laws); II Cnut 76.1-3 (Robertson, Laws). On young children, see also Wulfstan, Sermo Lupi ad Anglos, ed. Dorothy Bethurum, The Homilies of Wulfstan (Oxford, 1957), at 262, lines 45-48; tr. Dorothy Whitelock, English Historical Documents, vol. 1, 2nd edn. (London, 1979), 930. Wulfstan later had a hand in drafting Cnut’s legislation. 32  how concerned legislators were to think through the implications, and to judge the limits, of collective responsibility. How far the inheritance of property might be affected was also a thorny issue, and some laws went out of their way to protect the interests of heirs in cases of penal enslavement.47 One Carolingian capitulary dealing with a man unable to pay the heribannus owed to the king for having failed to join a military campaign said he should give himself as a pledge to the fisc until he was able to repay his debt, and expressly stated that if he died before the debt was repaid, his heirs should still receive their inheritance as normal, and would not lose their freedom themselves.48 In Visigothic law, the guilty parties in cases of rape, adultery or going back on a betrothal were to be handed over to those whom they had wronged, but a distinction was made between those wrongdoers who did not have children, whose property was to be transferred to the plaintiffs along with their bodies, and those who did have children, who would be transferred without their property.49 One clause on divorce and bigamy included a long list of provisions regarding how the guilty party’s property might be 47 A genealogy of unfree dependents of Santa Fiore in Arezzo, dating from around 1100, mentioned as an ancestor of living servi a certain Lupolus, who had been enslaved for a theft in another place; but the subsequent unfreedom of his descendants does not seem to have come from this, but from the fact that his son, Dominicus Scaramella, subsequently moved to Santa Fiore to work as a cook, swearing an oath to the abbot Rodulf “sicut servus domino”, and married an unfree woman there. The list probably included Lupolus and Dominicus only to ram home the point that all previous generations had been unfree one way or another, thus strengthening the monastery’s hold on its current dependents. Documenti per la storia della città di Arezzo nel medio evo, ed. Ubaldo Pasqui, vol. 1 (Florence, 1988), no. 293, at 401. 48 Capitularia I, no. 74, c. [1]. 33  divided among heirs; they themselves were to be taken entirely out of the equation by being sent into perpetual exile or given as slaves to whomever the king chose.50 Penal enslavement was also expressly said to make all debts square: even if it did not cover the full extent of the debt owed, for instance in the case of multiple crimes, liability nevertheless stopped with the wrongdoer himself, turning enslavement into a way of limiting the impact of the debt and preventing overspill on other family members.51 Legislators therefore seem to have gone out of their way to try to shield the rest of a kin group from the penal enslavement of one of its members, and to minimise its implications for their own status and inheritance. 52 Paradoxically enough, this would have resulted in a situation where penal enslavement could be comparatively less disruptive to a kin group as a whole than collective participation in a compensation payment, which could represent an enormous investment. The property of other kin members was secure even in cases where the punishment of the individual was at its harshest, as when penal enslavement entailed being 50 Liber iudiciorum III, 6, 2. 51 Liber iudiciorum V, 6, 5. 52 Ribuarian law is an exception, and extends the debt of a man too poor to pay compensation to his children for up to three generations (Lex Ribuaria 12, 2, eds. Franz Beyerle and Rudolf Buchner, MGH Leges I, 3, 2 (Hannover, 1954), 78). This clause, however, features only in the context of exceptionally high fines of 600 solidi, which may have constituted a special case (that is, actions too serious to be made up for by one single person’s forfeiting of freedom). 34  sold “across the sea”, in an early Anglo-Saxon law.53 Even when it seems clear that those condemned were to be transferred along with their property to the person they had harmed, at least no one else’s property within the kin group was affected as a result of their conviction – nor would it be in the future, since the wrongdoer was at one stroke also placed under somebody else’s legal responsibility. This presumably had a strong impact on families’ decision-making processes, and wrongdoers must at times have come under terrible pressure from their own kinsmen to bite the bullet for the sake of the family. Less valued family members may well have found themselves losing out at this game, even if their kin were not in abject poverty. Families could certainly help if they calculated that it was worth it; on the other hand, they could also decide to cut their losses in the case of particularly difficult individuals, or those more marginal to family interests. Some laws suggest that family groups, left to their own devices, might prefer to hedge their bets. Timing again played an important part here. A law of Ine insisted that kinsmen could only claim compensation for the killing of an enslaved relative within one year from their enslavement; if he was killed after that and they still had not redeemed him, they would receive no compensation.54 There was thus a limit to how long one could 53 Wihtred 26 (Attenborough, Laws). 54 Ine 24; see also II Edward 6, also denying wergeld to a penal slave’s relatives (Attenborough, Laws). 35  expect to keep some sort of stake in enslaved relatives: while a year seemed a reasonable amount of time to allow a kin group to get their act together, a longer period implied that they were stalling and waiting until it became clear that the relative did indeed have a useful role to play within the group. The “wait-and-see” approach is likely to have applied particularly to the young. A law of Æthelstan insisted that thieves under fifteen years of age should not be killed, but that their relatives should stand surety to the full amount of their wergeld on the assurance they would not commit any other crimes; if their relatives failed to redeem them, the thieves were to swear they would not commit any other crime and to go into penal servitude until the wergeld was paid. 55 How long such a boy might have to wait may well have depended on how many other heirs might be produced in the future. All this suggests that families could be readier than one might expect to accept penal enslavement as a temporary measure while making up their minds whether this particular kinsman was worth the money. The reference to their keenness to collect wergeld when their enslaved relatives were killed shows that their ideal scenario was one where they got to keep a level of association with their enslaved relatives for some purposes (collecting compensation) but not others (paying out compensation). Going a step further, it is also possible to see families using penal enslavement as a roundabout and rather extreme method for policing 55 VI Æthelstan 12.2 (Attenborough, Laws). 36  themselves, and detaching themselves from some of their own members, if the latter’s crimes had significant repercussions on the honour of the kin group or constituted a serious transgression within it – and, at least in Spain and Italy, this sometimes meant enlisting agents of the king as enforcers. This could happen through either inaction (failing to help pay compensation) or active participation. Inaction seems to have applied, for instance, in a case from Vic from 987, in which a man was enslaved by the official in charge for having sneaked into his mother’s house at night and murdered his wife, even though (the document makes sure to tell us) she was entirely blameless.56 Active participation seems to have been involved when a kin group found itself unable to curb deviant behaviour among its own. This seems the most plausible way, at least, of making sense of some at first sight extraordinarily interventionist legislation stipulating penal enslavement, especially in cases of sexual misconduct. Gender clearly played an important part here. Women were a point of special vulnerability in a family’s honour, and could devalue it significantly by engaging in inappropriate conduct.57 They come up correspondingly frequently in legislation stipulating penal enslavement. While such measures could be taken as signs of kings’ aspirations to regulate family life, they may also have been responding to a demand on the part of families. In some cases 56 Catalunya Carolíngia IV, part 3, no. 1517. 57 Janet L. Nelson and Alice Rio, “Women and laws in early medieval Europe,” in Judith Bennett and Ruth Mazo Karras, eds., The Oxford Handbook of Women and Gender in Medieval Europe (Oxford, 2013), 103-117. 37  this is entirely explicit, as when the family group itself was the beneficiary of the enslavement of one of its female members: this is the case, for instance, in a Visigothic law ruling that if a woman whose rapist had been enslaved to her later went on to marry him, she was to be herself delivered as a slave to her own heirs, along with all her property. 58 In Lombard law, a free woman who married a slave was to be enslaved by her relatives, who had the right either to kill her or sell her out of the country, with the right to “do what they wish with her property”. 59 Some families, then, seem to have been keen to enlist the authority of the law to intervene in the highly sensitive matter of sexual honour, and to use penal enslavement for this purpose. Indeed, the state could be expected to continue to police behaviour and protect family honour even after the act of enslavement: two Carolingian capitularies for Italy reiterate existing Lombard law enslaving an adulterous couple to the woman’s husband, and further add that if the couple were sold to a third party and continued their affair as slaves, the palace would confiscate them both and put an end to it.60 Clearly kings had their own reasons to wish to stop adultery, which increasingly became a major public concern in the course of the early 58 Liber iudiciorum III, 4, 14 (Ervig). 59 Rothari 221 (ed. Bluhme, Leges Langobardorum). If they failed to do this, the king’s agents were to place her with the palace’s slavewomen. See further Liutprand 24 and 98. Later Carolingian legislation transferred them to the master of the male slave they married: clearly they were not interested in using this legislation to acquire more slaves of the fisc (Capitularia I, no. 159, c. 1; II, no. 201, cc. 13-14 in ms Paris BnF 4613). 60 Capitularia I, no. 157, c. 3; no. 158, c. 2. 38  middle ages. But this does not necessarily mean that they were imposing their agenda forcefully on unwilling families, or that theirs was an unwanted intrusion: they were, at the same time, offering families the option of jettisoning their more problematic members (and women especially). This seems to be the case, for instance, for enslavement following marriage to an unfree person, which features in virtually every early medieval law-code; such laws were in theory applicable to both sexes, but women are discussed in them much more prominently.61 If families wished to, they could always arrange a deal with the man’s lord, presumably in exchange for a counter-gift, and formularies show many examples of this.62 If they did not wish to do this, however, they could decide to let the lord’s claim stand, and sever the ties of kin responsibility. This is not unlike the function of penal enslavement among Mudejars in late medieval Catalonia, as analysed by Mark Meyerson: penal enslavement, while securing cheap labour for Christian lords and the king and emphasising the subjection of the Muslim population, was at the same time used pragmatically by the Mudejar in-group to sever ties with 61 Pactus Legis Salicae 13.8 and 25.4; Lex Ribuaria 61, 14–18, ed. Franz Beyerle, MGH Leges I, 3, 2 (Hanover, 1954); Liber iudiciorum III, 2, 3; Liber constitutionum 35, ed. Ludwig R. von Salis, Leges Burgundionum, MGH Leges I, 2, 1 (Hanover, 1892); Lex Alamannorum 17, eds. Karl A. Eckhardt and Karl Lehmann, MGH Leges I, 5, 1 (Hanover, 1966); Capitularia I, no. 142, c. 3. 62 Formulae Andecavenses 59; Formulae Marculfi II, 29; Cartae Senonicae 6; Collectio Flaviniacensis 102; Formulae Salicae Merkelianae 31; Formulae Salicae Bignonianae 11; Formulae Salicae Lindenbrogianae 20; Formulae Morbacenses 18 and 19; Formulae Augienses Coll. B 41 (ed. Zeumer, Formulae). Rio, Legal Practice, 216-23. 39  its more undesirable elements. Unsurprisingly, these undesirable elements more often than not also involved adulterous women.63 Even enslavement mandated by law for particular crimes is thus likely to have corresponded to a social demand, and to have played into the hands of at least some families. This seems to be confirmed by the fact that laws could be much more lenient when dealing with infringements of the king’s own rights, which is the opposite of what one would expect if the state had been taking on a highly proactive policy regarding penal enslavement. In the Edict of Pîtres, in a clause stipulating payment of the bannus for rejecting good denarii, Charles the Bald instructed his representatives that, if some culprits could not pay the full sum, they should write a report and leave it up to the king to determine what punishment was to be meted out, “so that men are not weighed down unduly or beyond measure; for, as scripture says, ‘we do not require the amount, but the fruit’; that is, we do not demand dishonest profit, but only what is given to the kingdom for punishment.”64 Carolingian kings, while careful to leave room for penal enslavement as a legitimate form of dispute settlement, never seem to have found much of a place for it in their own arsenal of “public” punishment. Charlemagne, indeed, explicitly ruled against the use of enslavement to resolve cases of theft: 63 Mark Meyerson, “Slavery and the social order: Mudejars and Christians in the kingdom of Valencia,” Medieval Encounters 1 (1995), 144-73. 64 Capitularia II, no. 273B, c. 22. This is in contrast with I, no. 20, c. 19 and no. 74, c. [1], which had ruled the defaulter should hand himself over to the fisc as a pledge. 40  theft, like other major crimes, was deemed too serious to be settled through compensation, and was to be punished instead through mutilation or death.65 These other forms of punishment isolated the criminal in a non- negotiable, final way. Like penal enslavement, they left the rest of the kin group undisturbed and restricted the impact of punishment exclusively to the individual; to this extent, they fulfilled a largely similar function. But unlike penal enslavement, such methods also denied the kin group the option of showing solidarity by redeeming a kinsman. Penal mutilation would become an ever more dominant feature of punishment in the later middle ages. It may in fact have picked up where penal enslavement left off; at any rate, cases of penal enslavement after the twelfth century seem much more limited in number, despite more plentiful surviving documentation, and certainly more limited in their application.66 In its early medieval form, penal enslavement was too flexible, too open to negotiation, too embedded in personal loyalties and solidarities to amount to a reliable form of state punishment, and it did not develop into one. Its legitimation by kings in law was not ultimately driven by any state agenda, but by the desire to respond to the needs of kin and local politics. Conclusion 65 Capitularia I, no. 77, c. 15; no. 20, c. 23. 66 In later medieval Spain, it applied only to Muslims, not Christians (Meyerson, “Slavery”). A case cited by Hyams for thirteenth-century England stipulates that if a tenant tried to leave his land or failed to pay his rent, he and his family were to become unfree: the case itself only presents a hypothetical way of dealing with a future default, and is much more exclusively concerned with lords’ labour strategies (Paul Hyams, Kings, Lords and Peasants (Oxford, 1980), 183, citing PRO E326/12169). 41  Early medieval penal enslavement was very unlike Roman or modern penal servitude in that the needs and aims it fulfilled were not only, nor even primarily, the state’s. Penal enslavement is much easier to document for periods and places where it formed part of a penal system controlled by a big interventionist state – as for instance under Rome, or in China, where it seems to have constituted a major source of slaves. 67 But it is important to remember that this was not the only form it could take. In small-state societies it responded to a fundamentally different logic, as the outcome of largely private arrangements struck between disputing parties or between their kin groups. The evidence for such processes, however, tends to be much scarcer than for state-controlled ones. Enslavement for crimes, for instance, is known to have been widespread in pre-colonial Africa, but detailed evidence for how it worked is hard to come by, because of the lack of written records before the Atlantic slave trade (which skewed its internal dynamic by encouraging its use as a pretext for sale to European merchants), and also because such practices were not subsequently recognised in colonial laws.68 Penal enslavement in early 67 Orlando Patterson, Slavery and Social Death (Cambridge, MA, 1982), 126-129. 68 See Allan Fisher and Humphrey Fisher, Slavery and Muslim Society in Africa: The Institution in Saharan and Sudanic Africa, and the Trans-Saharan Trade (London, 1970), 71-76; Carol MacCormack, “Wono: institutionalized dependency in Shrebro descent groups,” in Suzanne Miers and Igor Kopytoff, eds., Slavery in Africa: Historical and Anthropological Perspectives (Madison, WI, 1977), 181-203, at 195-196; in the same volume, see also James Vaughan, “Mafakur: a limbic institution of the Margi (Nigeria),” 85-102, at 91; Gerald Hartwig, “Changing forms of servitude among the Kerebe of Tanzania,” 261-285, at 269-270; Svend Holsoe, “Slavery and economic response among the Vai (Liberia and Sierra Leone),” 287-303, at 290 and 294; Martin Klein, “Servitude among the Wolof and Sereer of Senegambia,” 335-363, at 343. For the example of the Igbo, see Jack S. Harris, “Some aspects of slavery in Southeastern Nigeria,” Journal of Negro History 27 (1942), 37- 42  medieval Europe could constitute an important case study for such processes: while its approach to penal enslavement was certainly more characteristic of the small-state model, it was not, in contrast to Africa, affected by colonial laws, nor by the existence of a mass demand from an external market.69 Most crucially, participants there did (however unsystematically) generate a written record, so that the specifics of particular cases are not irretrievably lost to us. Penal enslavement in early medieval Europe found its place in a legal system where family and group solidarities were paramount, both for the keeping of order, through the internal policing fostered by collective responsibility, and for the purposes of settlement once a dispute had begun. Such a system required the existence of inverse mechanisms for dissolving solidarity in those problematic cases which stretched the limits of what kin and community were prepared to do to help. This was precisely what penal enslavement achieved. The point of it, in contrast to compensation, was to sever the individual from both kin and community for the purposes of punishment – not only for the particular dispute at hand, but also for future disputes, by placing the offender under the responsibility of a new lord rather than that of kin and community. In this sense, it could be a very effective form of coercion. It was also very 54, cited by Patterson, Slavery, 128. Most of these examples relate to a pre-1800 situation. 69 Regardless of the view one wishes to take on the scale of European slave exports to Muslim regions, no evidence exists that it involved penally enslaved persons in any significant way (for a maximalist reading, see Michael McCormick, The Origins of the European Economy (Cambridge, 2001); also Jeffrey Fynn-Paul, “Empire, monotheism and slavery in the greater Mediterranean region from Antiquity to the early modern era,” Past and Present 205 (2009), 3-40). 43  flexible, and allowed many possibilities for reinsertion on different conditions, and at different stages of the process. This highly strategic deployment of “social death” can help us to see more clearly the profound conditionality of kin and community networks of support, which, in order to mean anything at all, needed to leave room for selection and discrimination.70 70 On “social death”, see Patterson, Slavery. 44