Introduction

Rather in the way that the last two chapters formed a diptych, concerning themselves with different but complementary aspects of the production, mobilisation and deployment of surpluses in the late Roman world, this chapter and the succeeding one will form a diptych relating to defining features of the fifth century and its archaeology. This chapter will concern itself essentially with the breakdown of the structures of imperial political, military and fiscal control in the West, on the one hand, and, on the other hand, the appearance of features of the archaeology which may (or may not) be related to the increasing presence and importance of ‘barbarians’ on the territories of the Western Empire. The following chapter will examine the consequence of these developments for wider economic and cultural formations across the West, which ran to a somewhat different tempo.

The archaeology of the fifth century on its own would tell us that there were major changes across the board in the Western provinces in this period; it sees massive alterations over a short timescale. Archaeologically, the fifth century marks an important threshold of development, one that sees important structural changes across the range of the evidence, thus marking a far more important horizon than the ‘crisis’ of the third century. Yet, like the third century, the fifth century was also a crisis, in the technical sense that it was a point at which the ‘patient’ recovered from or succumbed to the ills besetting it; in this case the universal perception is that it succumbed to a virulent attack of barbarians. This is, of course, a view derived from the textual sources, from their narratives, often with a strong moralising agenda, of military, political and administrative collapse and the replacement of the imperial system by the Germanic successor states of the early Middle Ages. In this chapter the traditional discourse of the ‘fall of the Western Roman Empire’ will not be ignored (it hardly can be), but it will be viewed from a different perspective, that of the archaeology, which may give us a rather different range, chronology and causation of events, in particular as regards the place of the various brands of barbarian peoples.

Chronological outline

The traditional narrative and structure for understanding the events of the fifth century has been derived from the historical and other textual sources (for a representative, recent and detailed treatment, see Heather Reference Heather2005 and references). A much abbreviated version of events in the first half of the century is presented here to give a chronology and an outline of events in the short term. It must be emphasised that this is not because this narrative is ‘true’, and still less because the archaeology is simply there to ornament this narrative. It is because the narrative and the events it relates have given both the accepted chronological structure to the fifth century and a version of events. This chronology and these events have for a long time shaped the presentation and the discussion of the archaeological evidence, so that it is necessary to have an appreciation of the traditional narrative, even if the intention is to discard it and replace it with something else more responsive to the nature of the archaeological evidence and its significance(s).

On the last day of what is normally given as the year 406, in fact very probably 405 (Kulikowski Reference Kulikowski2000), large barbarian forces consisting of Alamanni, Alans, Sueves, Vandals and others crossed the frozen Rhine in the region of Mainz and penetrated deep into Gaul, encountering little opposition from imperial forces and heralding fifteen years of war and instability. This provoked the usurpation in Britain of Constantine III, who crossed to Gaul to try to restore order and to deny to the barbarians the passes into Spain. In both of these he was unsuccessful, surrendering to imperial forces at Arles in 411 and being done away with. But by this time the Alans, Sueves and Vandals had penetrated into Spain, where the Sueves carved out for themselves a territory in the north-west of the peninsula. The imperial authorities, under the direction of the very able patrician Constantius (later briefly emperor in 421), manipulated the Visigoths, fresh from their part in the sack of Rome in 410, into south-western Gaul and then into north-eastern Spain to try to defeat the other tribes, before eventually settling them in south-western Gaul in 418, or more probably 419, ceding them rights from Toulouse to the Atlantic. The Visigoths were intervening under Roman auspices in Spain from 422, but a more worrying presage of things to come saw them attacking the imperial seat in Gaul, Arles, as early as 425. In 429, the Vandals and many Alans crossed to north Africa, where they took Carthage in 439, depriving the Western emperors of their richest tax lands. In 433, military command of the West was conferred on Aetius, who tried to stabilise the position by force, defeating the Burgundians and resettling them around Geneva and westwards in 436/7, defeating the Bacaudae (Drinkwater Reference Drinkwater1992) (a local uprising) in Armorica (Brittany) in 437, and in 439 attempting to subdue the Visigoths, who had been attacking Arles and Narbonne, an attempt that was unsuccessful under the walls of Toulouse but at least led to a reaffirmation of the original treaty of 419. The relatively peaceful 440s saw the various contestants with claims over territory in the West circling and manoeuvring; the imperial government under Valentinian III (425–55), represented in Gaul by Aetius, was contesting with the other factions also, the Visigoths, Alans (resettled by Aetius on the Loire in 442), Burgundians in central and southern Gaul, and the nascent Frankish power in the north. In Spain the remnants of the imperial authorities maintained a precarious hold on the Mediterranean littoral and adjacent inland areas, with the Sueves becoming more dominant in the north-west and expanding their territories. In 451 in Gaul, the contending parties sank their differences and united under the leadership of Aetius to face the invasion of Attila and the Huns, successfully facing them down at the battle of the Catalaunian Plains (near Troyes in central Gaul), a battle in which Theodoric I, king of the Visigoths, was killed. Aetius was to meet the fate of several successful late Roman generals by being assassinated, in this case by Valentinian III personally, in 454, Valentinian himself being assassinated in revenge the following year, all of this testament to the poisonous faction fighting on the Roman side, which could affect relations with the Germanic rulers.

As can be seen, two intertwined themes are central to this narrative: the increasing enfeeblement of the unified imperial power in the West and, as both cause and consequence, its lands and power being taken over by a series of kingdoms ruled by dynasties claiming Germanic descent and identity. To turn from this short-term ‘kings and battles’ history to how this all intersected with more medium-term processes, what we need to consider is how the military and political events acted upon the existing structures of the West, and, in particular, how they brought about ‘The end of the Western Roman Empire’. Of course, what is generally meant by this expression is the end of imperial political, administrative and fiscal control, the end of the late Roman state and its structures, and it is that which we shall examine now. What happened to the populations of the West and their political, economic and cultural formations will be the concern of the next chapter. The process of the dissolution of the Roman state control over the territories and peoples of the West and its proximate causes can be fairly readily characterised and understood. As has been stated earlier (p. 19), the political, administrative and fiscal systems of the late Roman Empire depended in the last analysis on the army. It was the army that was there to hold the frontiers against external threat and sought to guarantee internal peace and stability. It underpinned emperors and their reigns over their peoples (or, alternatively, attempted to replace them), and it also underpinned the state’s fiscal system; after all, it was the principal beneficiary of that very system. In order to do this, it had to have ensured sources of manpower, money and materiel. In the fourth century this balance held and the army maintained its manpower (though with increasing difficulty) and was paid and supplied, though, as we shall see, the 390s may well have marked a turning point for the Western armies. But certainly from 406, the Western Empire started to suffer not only military defeat but also, in crucial distinction to the ‘third-century crisis’, permanent, large-scale loss of territory. With territory went recruiting-grounds, taxpayers and resources, enfeebling the army and the state. The incoming Germanic peoples picked up on this weakness and sought to turn it to their advantage by taking further imperial territory. Increasing loss of territory translated into decreasing Roman ability to do anything to restore the situation, a vicious cycle, and by the 450s the once-mighty Western Empire was but one player among many in the campaigns and alliances. By the end of the 470s, it was not even that: military debilitation had resulted in political oblivion. This was the structural crisis of the Western Empire, one from which it did not pull through. Of course, this was not planned or predestined, either by the Romans or by the Germanic peoples. On the Roman side, such things as the settlement of the Visigoths in Aquitaine in 419 were doubtless seen as expedient and temporary – it got the Roman authorities off a particular hook. They could not know at the time that an allocation of land would turn into an independent kingdom. Nor, so far as we can tell from the Roman textual sources, was there any intention on the Germanic side to destroy the empire as such; rather, they wanted to establish their claim to parts of it. If in this process they had to ally themselves with or confront the imperial government, well, that was politics. One could say that the Western Roman Empire was one of the larger of history’s victims of the law of unintended consequences.

The end of the Roman army in the West

Central to the existence of the Roman Empire was its army, which defended imperial territory, safeguarded the person of the emperor and ultimately underpinned the judicial and fiscal systems that sustained the emperor and the empire, and, indeed, the army. Its progressive debilitation and ultimate disappearance are therefore equally central to the study of the fate of the Western Empire, since it can be said that without a Roman army there could be no Roman Empire. The fate of the Roman army in the West and the processes by which it disappeared are difficult to pin down, particularly from the archaeological evidence; nevertheless, the attempt must be made, even if only to demonstrate the problems inherent in the exercise.

In Chapter 2 it was argued that the crucial distinguishing feature of the late Roman army was that it was a standing army, paid, housed, equipped and supplied by the state (the taxpayer), and commanded by officers appointed ultimately by and answerable to the emperor as part of the ‘public’ power of the state. It was a major institution of the Roman state, organised into functional types (comitatenses were the internal field armies, limitanei/ripenses the frontier armies) and into regional commands. Each command (as listed in the Notitia Dignitatum) consisted of a variable number of named units, units that had a long-term existence as organisations independent of the command of which they might form a part or of the soldiers or commanding officers who at any one time made them up (for instance, unit titles listed in northern Britain can be traced back some three hundred years before their appearance in the Notitia). The soldiers were defined by the state authorities through formal processes of recruitment, training, registering in and membership of units; subordination to officers and regional commanders; and receipt of pay, provisions and equipment. This ‘etic’ (external) definition overlapped with the ‘emic’ (internalised) self-definition of the soldiers inculcated and routinised daily through the forts in which they lived, the clothes they wore, the weaponry and armour they used, the oaths they took, the ceremonies they attended, the distinctive military language and laws they used, the unit to which they belonged and its esprit de corps, and their consciousness of membership of the wider ‘imagined community’ of the soldiery, set apart from the wider civilian population of the empire. This definition of the late Roman army has been recapped here, because, central to the argument that follows about how to model archaeologically the demise of this institution, will be precisely the fact that it was a distinctive institution, in particular one that was sustained by the state and was part of the ‘public’ power structures of the state. It will be proposed that crucial for our understanding of what happened to the army in the fifth century will be the idea that it progressively ceased to be sustained by the ever more enfeebled state, and that, as a result and in its place, there came about command over, and expressions of, military power that increasingly were the responsibility of individual commanders rather than the imperial command and control structures: military power increasingly became ‘privatised’.

A key site which we may use as a case study because of its large suite of excavated evidence covering these years is the fort of Krefeld-Gellep (Gelduba) on the lower Rhine (cf. Figure 2.2), more particularly the cemeteries excavated between 1960 and 2000 under the direction of Renate Pirling (for a summary to 1985, see Pirling Reference Pirling1986; for the more recent work, see Pirling et al. Reference Pirling, Siepen, Noeske-Winter and Tegtmeier2000 with bibliography). The majority of the fourth-century inhumation graves exhibited the relatively simple Roman provincial burial rite common across northern Gaul and the Rhineland or Britain (cf. Chapter 2, p. 51) and contained a range of grave goods, most often pottery. Some male graves contained items of dress such as belt suites and crossbow brooches, suggesting that the dead may have been soldiers of the garrison at Krefeld. But the presence in one grave (Gr. 4755) not only of an elaborate, later fourth-century belt suite of Roman manufacture but also a bronze neck ring of a type originating east of the Rhine suggests contact with that area, though this supports the concept of a Mischzivilization with cultural traits borrowed from either side of the river. Likewise, in some female graves of the later fourth century, there were pairs of brooches, including Tutulusfibeln, with both the objects and the way of wearing them reflecting material and practice from east of the Rhine. Here was also one female grave, Gr. 4607, containing a mirror of a type common in the area known to the Romans as Sarmatia. Does this echo the Sarmatae gentiles of the Notitia (cf. Chapter 2, p. 92), or was it just a trinket? At Krefeld there was also metalwork indicative of contacts with Pannonia and some glazed pottery vessels, more common in Pannonia than the Rhineland (Swift Reference Swift2000b: 79–82). The interpretative problem is whether the areas of origin of the objects reflect also and directly the areas of origin of the persons with whom they were buried, in which case they may be used as ‘ethnic’ markers, telling us something about the origins of the garrison of fourth-century Krefeld. Alternatively, of course, the objects may have reached Krefeld and been buried there by means which divorced them from their ‘ethnic’ significance. To complicate matters, Gr. 3007 contained bracelets both of Danubian and of British origin (Swift Reference Swift2000a: 176, Reference Swift2010), demonstrating the difficulties inherent in using objects rather than attributes such as burial rite (or eventually chemical and physical analyses) to determine geographical, let alone ethnic, origins. This was all in the fourth century when the site was a fort, normally thought to have housed a garrison of the regular, standing Roman army: it is a testament to a mix of material culture at one site associated with the late Roman army and thus to the heterogeneity of the personnel of that army.

From the turn of the fourth and fifth centuries, both burial rites and objects start to change, with the first appearance of graves containing weaponry and handmade ceramics of non-Roman origin (e.g. Gr. 2650), though this in itself says nothing about the origins and loyalties of the troops or warriors at Krefeld. But from about the end of the first quarter of the fifth century, we begin to see weapon burials with long swords (spathae), spearheads and knives, along with brooches and buckles of ‘Germanic’ (in the sense of origins east of the Rhine) type. At the same time there are female burials with mainly ‘Germanic’ brooch types and other dress elements. Both male and female burials had glass vessels (including cruder ones of forest glass, compared with the technically superior tradition of the Roman-derived products), but also pottery that was increasingly of ‘Germanic’ forms and handmade. These burials, though in the same cemeteries as the fourth-century ones, tended to cluster in groups. The excavators interpret these changes during the first half of the fifth century, taken together, as representing the arrival of families or kin groups from east of the Rhine, led by male warriors and in some sense supplanting the regular, Roman garrison of the previous century. In the excavators’ opinion these were Franks. While not necessarily accepting such a precise ethnic identification (we shall examine the problems of the archaeology of ‘fifth-century Franks’ below), we can at the very least argue on the basis of the material culture and maybe features of the burial rite (e.g. deposition of weapons) that these burials express much closer links with the material culture and status and gender markers of peoples to the east of the Rhine. Again, in default of physical or chemical analyses, we cannot be certain, but there is a plausible case to be made that these changes represent people as well as objects from east of the Rhine. Given that Roman provincial pottery was still available, and used in some of the burials, the presence of handmade pottery of ‘Germanic’ type would seem to be a persuasive factor in favour of the people, as well as the pottery, being of Germanic origin, especially given the importance of pottery in the burial rites of the Germanic peoples at the time.

What cannot be established from their funerary rites is whether the menfolk buried at Krefeld in the fifth century were, by either ‘etic’ or ‘emic’ definition, ‘Roman soldiers’. Was there a continuing Roman state or government that regarded the men of Krefeld as subject to its control, loyal to the emperor and due some sort of payment or support in exchange? Or were they subordinate to some form of officer who still regarded himself as loyal to the emperor, even if the emperor and his bureaucracy may not have been aware of the existence of either the men or their officers, and thus could be thought of as ‘Roman’ in that sense. Did they regard themselves as in any sense ‘Roman’ soldiers, loyal to the emperor and part of a wider community of the Roman soldiery? Or was their allegiance to an ethnic leader whose loyalties lay to himself and his followers, making it very hard to see Krefeld any longer as a ‘Roman’ military installation with a ‘Roman’ garrison and commander? To date, the Krefeld-Gellep cemeteries provide us with by far the fullest evidence in the West for developments through the second half of the fourth and the first half of the fifth centuries. What this example shows is that, whereas it is possible to demonstrate continued occupation at a site that in the fourth century was a Roman military installation, after the beginning of the fifth century it becomes progressively more difficult to reconstruct from the archaeology what the function and status of these groupings may have been, or their relation (if any) with what remained of the Roman state.

A comparable site, this time on the upper Rhine, is Kaiseraugst (Drack and Fellmann Reference Drack and Fellmann1988: 300–12, 411–14). Within the fortification, there was probably a principia, and certainly, and unusually, a major bathhouse, a storehouse and a church. From within the fortification came the major, mid-fourth-century Kaiseraugst treasure (Cahn and Kaufmann-Heinemann Reference Cahn and Kaufmann-Heinemann1984). Some 300 m south-east of the fortress lay a cemetery, of which some 2,000 burials have been excavated and which shows a similar sequence to Krefeld-Gellep. The burials of the second half of the fourth century either had no grave goods or were furnished with pottery and, in a few cases, items of dress, in the standard late Roman way. From the end of the fourth century began to appear burials with ‘Germanic’ grave goods, though, at the same time, there was built a small apsidal structure, quite probably a cella memoriae, so religious change is as evident in this cemetery as any ethnic change there may have been. The cemetery was to remain in use until the seventh century, by which time it held gravestones with Germanic names. But again, for the fifth century, the ethnic identities and the political loyalties of the changing population remain unfathomable, as also their relationship to the Roman state. On the middle Rhine, one may point to sites such as Alzey and to a lesser extent Altrip. The fort at Alzey (Oldenstein Reference Oldenstein1986) had been established under Valentinian I; it was square in shape with projecting towers, and the internal accommodation took the form of buildings along the inside face of the walls, leaving the centre of the enclosure largely free of buildings (Figure 8.1). The fort seems to have been partially destroyed around 400; this was ascribed to the Germanic invasion in 406. Thereafter, the damaged buildings were restored or replaced. The material culture associated with this phase is of types much more closely linked with that from east of the Rhine, and is interpreted by the excavator as the installation of a Germanic garrison, possibly of Burgundians, but still under Roman command. Alzey was destroyed by fire in the middle of the fifth century.

Fig. 8.1 Alzey fort, fourth-century (black) and fifth-century (outline) structures

A common feature of the archaeological sequences at these forts and in their cemeteries is the increase in the amount of ‘Germanic’ material from them after the start of the fifth century. In some of the cemeteries, this material comes from what seem to be male and female graves in restricted areas of burial, and this may well point to family groupings. Both here and in Chapter 2, there has been an insistence that any simple equation between an object and a specific identity or ethnicity, or between the presence of an object in a burial and the identity or ethnicity of the occupant of the grave, should be avoided. These arguments stand. But at the case-study sites and at many others with more partial documentation, what changes in the fifth century is the volume and range of this new material and its increasing dominance of the material culture record, either on its own or in combination with the latest types of Roman-derived material culture, especially belt fittings. Whereas at the level of the individual object or the individual occupation deposit or burial it is possible and desirable to be cautious about ethnic ascriptions, what is different in the fifth century is the aggregate level of these types of material culture, which form a significant proportion of the total by the end of the first half of the century and pretty much the total by the end of the century. The origins of this material culture lay ultimately east of the Rhine and north of the Danube, the ‘Pontico-Danubian’ area, the homelands of the Germanic peoples and of other non-Germanic peoples. It may be objected that this is simply to accept the agenda of Roman writers and their ethnic labelling, but the archaeology makes it abundantly clear that there was in the fourth century and on into the fifth a series of distributions of related material culture across these areas (e.g. the complex often referred to as the ‘Elbegermanen’ [see Drinkwater Reference Drinkwater2007: Ch. 2; Drinkwater and Elton Reference Drinkwater and Elton1992]) that differentiated these peoples from those in lands directly subject to Roman imperial power. It becomes increasingly hard to sustain the interpretation of the large-scale arrival of this material culture at a large number of Roman military sites all along the Rhineland simply in terms of a continuing Roman provincial garrison and/or population as helpless fashion victims with a taste for Germanic goods. The presence of what may well be discrete groupings in the cemeteries with this sort of material, as against the continuing Roman provincial rites, as at Krefeld-Gellep, does look very much like the movement of people, not just of pots or brooches. This, of course, could have coexisted with the indigenous populations starting to redefine their identities and ethnicities in terms of the Germanic incomers, particularly given that these latter may have had a privileged position, one dependent on the martial prowess of the menfolk. So ethnic ascription from material culture remains problematic, and not all burials with ‘Germanic’ material need have been the burials of Germans from across the Rhine. Indeed, what these burials seem to show is the progressive ‘Germanisation’ of the groupings (incomers or indigenous) in aspects of their funerary practice such as, in particular, the preparation of the corpse and, in the case of women, above all by laying-out bodies in forms of clothing which, even if not making claims to specific ethnic identities, were certainly making claims not to follow Roman provincial practices. The increasing presence of triangular ‘Germanic’ combs may suggest the importance of hairstyles alongside the more obvious clothing, suggesting that a situation developed where markers of ethnicity (or at least of not becoming Roman) became more important, particularly at burial, than markers of relation to the Roman state and its army.

So, by the mid fifth century, it would seem that the peoples living and buried at a series of Rhineland forts increasingly used material culture of non-Roman origins, mainly from areas that Roman written sources classed as ‘German’. Whether all these people were from those areas to the east of the Rhine and to the north of the Danube remains unknowable; it may be that some of them were from west of the Rhine but assimilated to this ‘Germanic’ identity (these matters are discussed more extensively later in the chapter). But the question posed in this section of the chapter is that of the processes by which the Western Roman army ceased to exist, or at least to be recognisable in the archaeology. Earlier it was argued that a ‘Roman’ army fulfilled certain ‘emic’ and ‘etic’ definitions as regards the internalised loyalties and practices of the soldiers, the institutional existence of the army and its units, and the political loyalty of the army to the emperor. The changes in the archaeology, particularly the material culture, suggest that during the fifth century the ‘emic’ definitions of the people at these sites moved away from that which had characterised the garrisons of the fourth century towards something which to Roman eyes would appear more ‘barbarian’. But ‘barbarians’ served in the army of the fourth century at all levels, as was seen in Chapter 2, so that in itself is not a sufficient index of no longer regarding themselves as servants of the emperor, or being so regarded by others. What the archaeology at present cannot tell us is where the sense of the community to which individuals belonged lay or where their political loyalties lay. Did they regard themselves as soldiers of the emperor, or did they regard themselves as members of ‘tribal’ or other groupings and followers of individual leaders whose loyalties were negotiable? There is, though, one class of material which may be a strong indicator of an ‘etic’ definition – a definition made by the Roman authorities – and that is the coinage, which was closely tied to the question of army pay and loyalty.

Coinage in the early fifth century

In the previous chapter the striking of coinages by the late Roman state was explicitly tied to the political economy, specifically the payment/clawback system put in place by which precious metal was paid out to discharge the obligations of the state, above all to pay the army, and then recovered through a variety of means including the compulsory changing of gold and silver for base metal by the nummularii. In the West the bulk of state commitments was to the army, to its infrastructure, such as the fabricae, and to the bureaucracy, so this payment/clawback nexus was very closely allied to the army. At the turn of the fourth and fifth centuries, there were major, far-reaching changes to the state production of coinage, a horizon which perhaps has not received due attention (cf. Kent Reference Kent1994). The precious metal coinage had therefore always been central to the state’s meeting its obligations, especially as regarded the payment of the army. Silver coins, principally the siliqua, had been struck at the Western mints, mainly Trier but also Lyon and Arles, through the second half of the fourth century. At the turn of the fourth and fifth centuries, these mints ceased precious metal production, and henceforth the Western authorities struck gold (in small quantities) and more especially silver, principally at Milan. The usurper Eugenius struck a considerable silver coinage there in 393–4, but after his suppression at the battle of the Frigidus in early 395, there was an hiatus in production until 397. The five years between 397 and 402 saw a huge issue from Milan (rev. Virtus Romanorum). Thereafter the principal mint for precious metals in the West was Ravenna, particularly between 408 and 425, with sporadic production at Rome and Aquileia in 407–8 (cf. Guest Reference Guest2005a: 74–6). But these later issues did not penetrate north of the Alps in any quantity. The usurper Constantine III (406–11) did strike in silver, as well as some gold, at mints such as Lyon, both before and after the death of Arcadius in 408, but not in large enough quantities or for long enough to reverse the overall trend. For the coinages in base metal, in the year 395 the Western monetae publicae, the mints supplying these coins, were reorganised, massively changing the scale and nature of the coinage produced and supplied. Trier, Lyon and Arles largely ceased to strike in base metals; in the West, only Rome continued to do so in any volume. This step change was succeeded by another in 402 when the three Gallic mints effectively ceased base metal production at the end of the Victoria Auggg issue in 402, and the succeeding Salus Reipublicae, Urbs Roma Felix and Gloria Romanorum (three emperors) issues of Rome hardly circulated north of the Alps.

So from about 402 the Western Empire was suddenly in a situation where it was no longer producing the coinages, particularly silver and base metal, that had been vital to its revenue and expenditure cycle, the expenditure, of course, directed principally at the army. It is important to recognise that this was developing before the failure of the Rhine frontier and the start of the barbarian land grabs from the end of 405 on, so it cannot be, in origin at least, an effect of these. Also, Italy seems to have been different, since silver was still struck at Ravenna, and bronze continued to be struck at and circulated from the Rome mint. One possible explanation is that 402 marked one of the pauses in the production of coin which are detectable in the fourth century, the intention having been to resume coining, but this never came about outside Italy because of the events of 406 and the following years when a combination of barbarian incursion and the usurpation by Constantine III meant no coins of the later issues would be supplied to such unstable areas. It is noticeable, though, that even after the suppression of Constantine III in 411 and the re-establishment of a measure of imperial political and military control north of the Alps, the large-scale supply of coin to those areas was not resumed. In that case the imperial authorities presumably had to find other means of paying those soldiers and units still loyal to them. A logical, if extreme, response would have been the Western authorities deciding that they no longer needed to support these armies, but this would fly in the face of the fundamental importance of the army in maintaining the empire and the imperial system; it also ignores the textual evidence for something called a West Roman army in the first half of the fifth century. But, less controversially, one might propose that the state had chosen, or had been forced through circumstance, to change the ways in which it sustained and remunerated its armies. The fiscal system, insofar as it related to the armies, was designed in part to produce the wherewithal to pay them, but also the revenues to cover the costs of heads of expenditure such as the construction of military installations, the provision of weapons and equipment (through the fabricae), and the ensuring of foodstuffs and other supplies over and above those directly raised from and transported by the taxpayer. Interestingly, the trend over the latter part of the fourth century had increasingly been to adaerate these obligations – that is, to commute them for coin/bullion payments. If the Western Empire now no longer wished to raise the precious metals necessary for discharging these functions, then, presumably, the armies were to be supported in some other way(s), perhaps by more direct requisitions from the provincial populations, though we have no positive evidence for this. Logically, this would also entail the dismantling of the necessary bureaucracies under the comes sacrarum largitionum (precious metal revenues and payments), the magister officiorum (the fabricae) and the praetorian prefect of the Gauls (billon coinage and materiel). In fact, it is hard to trace these officers, save the praetorian prefect, in the written sources outside Italy much after the start of the fifth century. This picture of what may have happened to the Western armies from the end of the fourth century is provisional and needs more work and thought on the precise chronology of these changes and on the distributions of the various coin issues. But the major changes to the supply and circulation of coinage and the significance of those changes do have to be recognised and pursued. Either it was intended to be a temporary pause, one that was overtaken by events, or it marks a purposive shift in the imperial fiscal system beyond the Alps. Whichever it was, the result was that from the start of the fifth century the Western authorities could not or would not be in a position to maintain a standing Roman army of the traditional form. Some other expedients would have to be resorted to.

The archaeological evidence from military installations suggests that increasingly they were occupied by soldiers or warriors of ‘Germanic’ origin or by locals who were increasingly defining themselves in non-Roman ways. It could be proposed that this is the horizon at which the control of force passes out of the hands of the Romans and into those of the incomers as the latter supplanted the Roman army in what had been its garrison forts – a power and land grab. But the relationship of Roman provincial and Germanic material cultures and burials suggests rather that the two coexisted, at least in the first half of the fifth century. In this case one might propose a scenario whereby instead of a standing army with soldiers whose recruitment, routines of life, identity and ideological commitments were shaped by the Roman army and state, military force in the West was increasingly committed to ‘barbarian’ groupings for whose support the state did not have to take responsibility in the same way as it had for a ‘Roman’ army. In effect, it was an extension of the principle and practice of foederati, tribal detachments who fought for Rome but were not part of her standing armies. Presumably, the Roman state settled them in its forts and allocated them provincial land off which to support themselves on the understanding that they would fight for the emperor under the command of the senior officers (comites rei militaris and the like), who were, as we can see from the texts, still appointed to Gaul and Germany and to a lesser extent to Spain. This would echo the arrangements that we know were made for the settling and support of the Visigothic army and people in south-western Gaul in 419 (see below). Under this scenario the Western authorities would no longer have needed to mint much in the way of coin for the areas outside Italy (and North Africa), and that is precisely what we see.

The progressive loss of the recruiting grounds and the tax base of Germany, Gaul, Britain and Spain from 406 forced the imperial authorities to desperate measures, measures that for the first half of the fifth century seem to have had some success, to judge by the textual accounts of senior Roman generals, such as Constantius and Aetius, in managing to some extent to resist, control and resettle the incoming peoples (cf. p. 359), above all in Gaul, down to the middle of the fifth century. What the texts also show is the increasing importance of the personal charisma and military competence of these generals in persuading an increasingly heterogeneous range of troops and warriors to follow them. The public power of the Roman state was increasingly being supplanted by loyalty to a leader. This is the appearance of the type of retinue of warriors known to the sources as bucellarii (hard-tack men), men who followed a successful military leader because he fed them and his successes yielded booty. After that, the textual sources show the increasing fragmentation of Roman power, or at least the power of those commanders who legitimised themselves by use of the name and aura of Rome, in both northern and southern Gaul (Spain seems by then to have been a lost cause), with the newly emerging barbarian kingdoms (see below) taking over as the possessors of military force. The rulers of these kingdoms, and presumably the leaders of smaller war bands elsewhere in Gaul, were, of course, another expression of this ‘privatisation’ of military force as they jockeyed with each other and the last of the ‘Roman’ commanders for control of people and resources. How different would fifth-century, nominally imperial troops have looked from Germanic warriors of the same period, especially since Germanic troops and units had formed part of the regular Roman army since the fourth century? Were such identities fixed and immutable? The evidence strongly suggests not; individuals and units could segue from one identity to another within an essentially unchanged material culture. For the mixed garrisons of the forts of the Rhine frontier and the interior of the north of Gaul, it was but a short step in the later fifth century to incorporating themselves into the locally dominant ethnic grouping – Franks, Alamanni or whatever. A distant echo of such a process may be the tale related by Procopius (Bellum Gothicum V.12.13–19) of how the last Roman troops on the lower Rhine assimilated themselves to the Franks while keeping their unit identities.

Barbarians and breakdown

The part played by the transformation of late Roman military formations and garrisons in the creation of the successor peoples to the Western Empire in the course of the fifth century will now form part of a wider discussion of the archaeological evidence for the settlement in Roman territory of the various Germanic peoples in the course of the fifth century. Before we embark on a consideration of the archaeology, the current state of the debate on using evidence from historical sources to ascribe to aspects of the archaeology, above all burials and the objects from them, a particular ‘barbarian’ identity (Alan, Frank, Vandal, Visigoth, etc.) needs to be outlined. In dealing with the archaeology, rather than try to encompass all the peoples mentioned in the historical sources and all the evidence that has been used, a work which would run to several volumes, we will use the technique of case studies to open up the subject and indicate the range of evidence types and possibilities of interpretation. These case studies will be the Visigoths for the south of Gaul and the Iberian peninsula, and the Franks for the north and centre of Gaul. Other peoples such as Burgundians, Sueves and Vandals will be mentioned as and where appropriate and to give a brief indication of modern studies. First of all, the term that will be most frequently used below to denote these incomers is ‘ethnic’ with its correlates such as ‘ethnicity’. This is currently the standard academic terminology, one that avoids other contentious terms such as ‘tribe’, ‘people’, ‘Volk’, ‘Stamme’ and so on, let alone the loaded concept, ‘race’, terms that lack precision in modern anthropological and ethnographic literature while at the same time giving the impression of a range of different sizes of population groupings in some sort of hierarchical organisation. ‘Ethnic’ and its correlates are used here simply to signify groups of people who were felt at the time or are considered now to be distinguishable from each other not on grounds of age, gender, status or religion but on grounds of having a matrix of attitudes and behaviours that set them apart from other neighbouring groups, and in particular set them apart from the Roman provincial cultures, which thus form a sort of ‘background noise’ against which the different ethnic groups stand out. In the cases we shall be looking at, these are also groupings that were mobile and thus ended up in areas to which they were ‘foreign’ in many senses of the word.

Over the last hundred years and more, the question of ethnic identity, what constitutes it and how it is expressed, has been the subject of intense debate, now conveniently and very accessibly summarised in Halsall Reference Halsall2007: Ch. 2. In the later nineteenth and early twentieth centuries, such identity was held to be ‘primordial’ or ‘essential’ – that is, innate and, indeed, genetic (to be anachronistic) – and expressed through such things as belief in a common descent, a common kingship and nobility, a common religion, a common language, common customs and a ‘national dress’. An important feature of ‘primordial’ ethnic identity was not only that it defined the in-group but that it also defined (usually as inferior) out-groups. There was thus a strong belief in the works of early twentieth-century scholars such as Kossinna in the purity of each racial grouping and that they did not mix with other groupings; thus Alans did not mix with Vandals, for instance, let alone ‘Germans’ with provincial Romans, the latter view supported by many of the law codes issued by the successor kingdoms which distinguished strongly between ‘German’ and Roman provincial to the extent of forbidding intermarriage. Where such ideas about the ‘essential’ nature of ethnic identity and the need to maintain the ‘purity’ of the stock could lead was made catastrophically clear with the Nazis and the ‘Aryan’ identity and the position of that identity in relation to other, allegedly ‘inferior’ identities (Untermensch). Since World War II and in reaction to its ethnically created nightmares, a considerable and lively debate has taken place over the meaning of ‘ethnicity’ and ‘ethnic’ identity, and how to recognise such things – indeed, whether it is possible to recognise such things – in the archaeology. Ultimately, this comes down to whether objects, such as dress fittings, have an inherent ethnic identity and whether particular types of such objects would have been worn only by people of that ethnicity; thus an object betrays ethnic grouping.

Different modern nation states have developed different traditions of exegesis. In the German tradition the ascription of ethnic identities to features of funerary practice, such as deposition and especially object types, has been persistent, and publications regularly identify particular burials with specific ethnic groupings mentioned in the textual sources as being in certain areas at certain dates. This approach has been followed to a considerable degree in Spain (see below), where the influence of German workers has been strong, and to a lesser but still important extent in France, where emancipation of the archaeological evidence from the textual narratives has proceeded relatively slowly, and ethnic identification of burials and of grave goods still forms an accepted part of archaeological publication. The English-speaking tradition, along with some other European traditions, particularly the Dutch, has shown less fealty to the notion of ascribing ethnicities to burial practices and to objects; it has been more influenced by the development of postmodern (post-processual) concepts of the fluidity of identity and its signifiers. To workers in this tradition, objects have no inherent ethnic identity. The people who made them may have regarded themselves as belonging to a particular grouping, and so may the people who used or wore them, but not necessarily the same grouping, and the people who buried someone with such objects may have had their own views on the matter; none of them may have regarded an object or practice as belonging to a specific ‘ethnicity’. There is great doubt over the extent to which any such ‘ethnic’ identities were expressed through particular items of dress, and, of course, individual objects might pass through many hands. In addition, were these objects used only to construct statements about ‘ethnic’ identity, or were other aspects of identity, such as age, gender or status, being signified? Moreover, the ‘ethnic’ identification often rests on an unstated assumption that ‘Alan’ objects were only worn and used by Alans, whereas it is clear from burials that objects from different ‘ethnicities’ can be found in the same grave. In an ethnicist reading, one would have to argue something such as that the deceased was the issue of a mixed marriage: it is equally possible to argue that the objects were chosen because the deceased or their buriers liked them and had no ethnic intention in their deposition. Of course, neither tradition of study exists in isolation from the other, and there has been cross-fertilisation; nonetheless, there remains a difference along the lines sketched above.

Let us summarise what developments in understanding of the mutability of identity have meant. First they showed that it was ‘instrumental’; it could be changed if there was an advantage to do so. This led to the concept of ‘situational’ identity, one taken on as the optimal response to surrounding circumstances. The crucial realisation here was that ethnicity was not innate; rather it could be a result of birth but it could also be opted for and it could be changed. Ethnicity was something that happened in people’s heads, not in their genes: it was ‘cognitive’. This is not to say that this is in any way a weaker form of identity; human beings are capable of believing in such things passionately and to the death (others’ or their own). Moreover, such supposed signifiers of identity as language and religion are nothing of the sort: the modern world contains plenty of examples of speakers of a common language or co-religionists who are very good at hating one another (sometimes bringing us back to ‘the narcissism of small differences’). Halsall (Reference Halsall2007: 40) makes the point that individuals’ identities are also multilayered, and different aspects of them can be emphasised or downplayed in different circumstances. This necessarily means that identity is ‘performative’; the chosen identity must be displayed and acted out to be realised and reified. All this might seem a recipe for a sort of Humpty-Dumpty ethnicity: ethnic identity means what I want it to mean. But there are important constraints. First, and in particular, that individuals have to negotiate their identity with others around them, and this may impose severe constraints on what they can opt for. Second, and relevant for us here, identity can be ascribed to individuals by others (‘etic’) as well as, or instead of, being ascribed by individuals to themselves (‘emic’). An existing group may deny membership to someone wanting to join it, for any of a number of pretexts which seem to them entirely reasonable and compelling. Equally, a dominant group may ascribe to individuals or groups in a less powerful position an identity of its own choosing, one quite possibly not the choice of those thus identified. For the period we are concerned with, this can be particularly important, since it is clear that ‘barbarian’ identities were, at least in part, created as a reaction to how the Romans thought about other peoples (cf. Curta Reference Curta2007), with Roman views instrumental in creating ‘barbarian’ groups’ sense of self-awareness and self-definition (the Goths are a good example). These remarks in turn raise the important point that signifiers such as dress items were not just passive reflections of an existing identity (ethnic, status or other) but were also used actively to construct such identities, including constructing new, different identities that users had chosen to ‘situate’ themselves in and to ‘perform’, thus ‘falsifying’ their original identity as created for them by their parents and their wider social or ethnic grouping.

For late antiquity, work on the ‘ethnic identities’ of the period and what they may have meant and expressed has been particularly associated with the ‘Vienna school’ of Walter Pohl and his co-workers or the ‘Toronto school’ of Walter Goffart and his colleagues, which hold divergent views on how to interpret the nature and settlement of the ‘barbarian’ peoples. Both groups, though, have critically examined the textual sources for the various ethnic groupings of early medieval Europe, in particular the Goths, to see what they tell us about the ways in which ethnic identities were created and made evident, and why individuals or groups made the statements they did. They have shown that such identities were fluid and constantly being manipulated and recreated to fit the current situation, often under Roman influence, direct or indirect. The textual sources do allow us sometimes to approach what members of these groupings thought at the time, or more often what other people thought about them, above all what the Romans thought, since frequently what we have are Roman thoughts (with all their problems of ignorance and stereotyping) about societies that were often not in a position to give their side of the story to a distant Roman commentator working within an established frame of reference about ‘barbarians’. In all these modern studies, great stress has been laid on the concept of ‘ethnogenesis’. If we no longer accept fixed and immutable identities and their transmission down the generations, either for individuals or for groups, then there must be reasons for which and processes by which individuals and especially groups come to differentiate themselves from those around them and to construct ways of doing things that state these differences. Many of these ways of doing things, such as speech, will not be visible to the archaeologist; others, such as dress and appearance, may well be. It should be noted that, following the implicit framework of ancient commentators, the Roman provincial populations are not seen to have an ethnicity as such, other than provincial designations, though these do, of course, go back to perceived ethnic differences at the time of their incorporation into the empire. But by the Late Empire they are to an extent the ‘norm’ against which the ethnic groups are defined. As we shall see, this has led to a situation where their presence in the evidence and thus their contribution to the debate are often underestimated.

In what follows, the more sceptical approaches to ethnicity outlined above will be used in a discussion of the archaeological rather than the textual sources for ‘barbarian identity’. This discussion will therefore concentrate on the material-culture correlates of a range of possible identities, such things as objects, building types and settlement types. Above all, it will consider the funerary evidence, since in preparing a corpse for disposal, the living can make powerful statements about how they see, or would like to see, the identity of the deceased, and this, of course, includes such things as gender, age and status as well as claimed or ascribed ‘ethnicity’. In order not to get embroiled in a seemingly endless range of evidence and possible interpretations, the discussion will focus on the two ‘peoples’ mentioned above, the Visigoths and the Franks, who were crucial for the transformation of what had been the Western Roman Empire into the ‘barbarian’ successor kingdoms. Another major and well-documented (archaeologically as well as textually) people, the Alamanni, will not be dealt with in detail here because their main areas of activity and settlement were either on the periphery of our area of interest, in the Rhineland, or were outside the Rhine frontier altogether and thus fall outside the purview of this work: an excellent and up-to-date introduction to them, focusing primarily on the textual sources but with consideration of aspects of the archaeology, is provided by Drinkwater (Reference Drinkwater2007), and there is a comprehensive introduction to their archaeology by Theune (Reference Theune2004). After a look at the Visigoths and the Franks as case studies, more general conclusions will be drawn as to the role of ethnic identity and interaction in the transformation of the Roman West.

The Visigoths in south-west Gaul and Spain

The Goths are the most intensively studied of the Germanic successor peoples in the Roman West, thanks to a rich documentary corpus including narrative histories, chronicles, letters and saints’ lives written by men from the Roman world who came into direct contact with them or recorded their doings; in addition, and unusually, there are written sources produced by the Goths themselves – for the Visigoths in particular a series of law codes and the acta of a series of Church councils, emanating mainly from the Spanish kingdom in the sixth and seventh centuries. These have given rise to a compendious literature (for starters, see Collins Reference Collins2004; Ebel-Zepezauer Reference Ebel-Zepezauer2000; Heather Reference Heather1991, Reference Heather1996, Reference Heather1999), since the abundance of the written sources has made the Goths the case study par excellence for a Germanic people in the late antique period, the locus classicus for the study of ethnogenesis. This same abundance has often concealed the fact that the archaeological record is much less coherent, and for the Visigoths tells a rather remarkable story.

Let us recap in outline the historical narrative for the arrival and settlement of the Goths in the West. The Goths, or, as they then were, the Greuthungi and Tervingi, were allowed across the Danube in 376 and two years later inflicted on Rome one of her worst military defeats at the battle of Adrianople, in which the Eastern emperor Valens lost his life (for general discussions of the earlier parts of the story of the Goths, see Heather Reference Heather1991: Pt. II, Reference Heather1996: Pts. I and II, Reference Heather1999). In the latter part of the fourth century, they moved westwards, and in 410 under Alaric I they sacked Rome herself, a psychological shock to the entire Roman world. After the sack of Rome, they were manipulated out of Italy and into south-western Gaul, where, along with a group of Alans, they besieged Bazas (Landes), an event best known for causing the ruin of the poet Paulinus of Pella (grandson of Ausonius), before being cajoled into north-eastern Spain to act as imperial agents in the clearing out of other Germanic peoples who had got there in the aftermath of the collapse of the Rhine frontier and the failures of the usurper Constantine III. Finally, in probably 419 (traditionally 418), the patrician Constantius settled them in south-western Gaul from Toulouse down the Garonne valley to the Atlantic, granting hospitalitas rather than direct payment to support them. The terms on which the imperial authorities settled them have given rise to a huge literature concerning the precise meaning and significance of the term ‘hospitalitas’, one of those many late imperial euphemisms for something in reality more brutal; did they receive two-thirds of the land, two-thirds of the tax revenues or something else? (Barnish Reference Barnish1986; Durliat Reference Durliat1997; Goffart Reference Goffart1980: Ch. 4, Reference Goffart2006: Ch. 7; Liebeschuetz Reference Liebeschuetz1997; but see also Halsall [2007: 422–47] for a full review of the debate). This process was christened ‘accommodation’ by Goffart (Reference Goffart1980), in contradistinction to earlier visions of brutal replacement of Roman by German, making the whole process less threatening or violent, and ‘pacifying’ this piece of the past (for a critique of this tendency, see Ward-Perkins Reference Ward-Perkins2005: 5–10). By now the interests of these people clearly lay in the West beyond the Alps, leading them to be described in due course as the Visigoths (western Goths) in contradistinction to the groups which later settled to their east in Italy, the Ostrogoths (eastern Goths).

The settlement of the Visigoths was established under Roman suzerainty and quite possibly as a supposedly temporary expedient; originally, it is very unlikely that either Goths or Romans saw this settlement as implying the creation of an independent political entity rather than simply as a convenient solution to a particular problem. But the developing weakness of the Western Empire meant that gradually the Visigoths came to regard themselves as free agents, and their leaders became kings with their seat at Toulouse. By the mid fifth century, the Visigothic kingdom had taken on a life and identity of its own, and its kings became important players in the political chess of the mid to late fifth century, with Aetius trying and failing to defeat them in 439. But the Visigoths put themselves under his command against the Huns at the battle of the Catalaunian Plains in 451. Afterwards, under Theodoric II, they intervened more deeply into Spain, defeating the Suevic kingdom in 456 and pinning it back into the far north-west; in Gaul they gained Narbonne in 462/3. In 466 Theodoric II was assassinated and replaced by Euric I, who pursued an overtly aggressive policy of expansion, capturing Pamplona, Zaragoza and Tarragona in 473, thus coming to dominate the northern third of Spain, and finally taking Arles and Marseille in 476 (the previous year he had taken Clermont-Ferrand, whose bishop, the author Sidonius Apollinaris, had organised the resistance to the Visigothic takeover, but Sidonius was sold down the river by the imperial authorities in an attempt to hold onto Arles and Marseille – not one of their more successful gambits). The political opposition of the Visigothic kingdom to what remained of the Western Empire was emphasised by the fact that the Visigoths were adherents of the Arian branch of Christianity rather than the Catholic profession of the imperial authorities; Euric I was militantly Arian and anti-Catholic. By his death in 484, the Visigoth Euric I not only controlled south-western Gaul but was also master of much of Spain save the Suevic enclave in the north-west. Under his son Euric II and Euric II’s son Alaric II, the Visigothic kingdom was clearly the major player in the former Western provinces of the Roman Empire. But there was to be one more roll of the dice; in 507 the Franks from the north under Clovis brought the Visigoths of Toulouse to battle at Vouillé near Poitiers and defeated them, killing Alaric II. Thereafter the Visigoths regrouped in Spain; north of the Pyrenees they held only the coastal fringe along the Mediterranean in Septimania. From the early sixth century, the Visigothic kingdom in Spain acquired the trappings of statehood such as law codes, coinage and wars of succession.

South-west Gaul

The textual sources therefore clearly present us, in the case of the kingdom of Toulouse, with a Germanic group with a defined identity; a monarchical, aristocratic and warrior society; a defined, stable and expanding territory; in what proved to be the twilight year of the kingdom, a legal system (the Breviarium of Alaric, 506); an ecclesiastical structure (Council of Agde in 506); and, at the level of the kings at least, clear notions of where the interests of the Visigoths lay distinct from those, including the Romans, around them. So how is this reflected in the archaeology of south-western Gaul, its Germanic settlements, structures, burials and material culture? Not at all well – an important exception is James, E. (1977), as is perfectly well known, but often glossed over by omission through concentrating on the written sources. The homelands of the Goths have long been identified by archaeologists, working in a culture-historical paradigm, with the area of the second- to fourth-century Sîntana de Mureş-Chernyakhov culture in the region to the north-west of the Black Sea outside the Roman lower Danube frontier (see Heather and Matthews Reference Heather and Matthews1991 and references), where there was a range of material culture, including brooches, belts and pottery, that is relatively distinctive. This culture was later to be found further west into Roman territories, and is taken to be evidence for the migration of the Goths into the empire in the second half of the fourth century. Reviews of ‘Germanic/Gothic’ material culture in fifth-century south-western Gaul have repeatedly come up with little more than a handful of sites and material (Ebel-Zepezauer Reference Ebel-Zepezauer2000; James, E. Reference James1977, Reference James1991; Périn Reference Périn1991; Rouche Reference Rouche1979 – the blank on Ebel-Zepezauer’s distribution map of ‘Visigothic’ metalwork [2000: Abb. 1] where the kingdom of Toulouse should be is striking – Stutz Reference Stutz2000). There is a small handful of material in the south-west of types deriving from the Sîntana de Mureş-Chernyakhov culture, consisting of four bone combs (Kazanski and Lapart Reference Kazanski and Lapart1995) of a very distinctive form, rectangular with a semicircular projection on the upper side. These items come from sites in a triangle formed by the cities of Agen, Auch and Eauze to the west of Toulouse, namely the villas of Montréal-Séviac and Moncrabeau-Bapteste (Lot-et-Garonne) between Eauze and Agen, and that at La Turraque, Valence-sur-Baïse (Gers), between Eauze and Auch. It should be noted that similar combs have also been recovered from Trier and other sites in the area (Cüppers Reference Cüppers1984: 345), so, though they are of non-Roman type, they are not necessarily solely ‘Gothic’. The combs are of interest, though, not just for the links they provide with the home areas of the Goths, but also for their function; they reflect the importance of hairstyle, a recurrent indicator of barbarian identity among Roman writers. Current research in the région of Midi-Pyrénées is adding to this corpus (J. -L. Boudartchouk, pers. comm.), but it is unlikely to change radically the impression of relatively little ‘foreign’ material culture through the course of most of the fifth century. One other indicator of non-local individuals is the group of four or so burials of people with deliberate cranial deformation carried out in infancy (Crubézy Reference Crubézy, Austin and Alcock1990) from south of the Garonne and, unfortunately, not well dated. Traditionally linked with the Huns (see below, p. 381), such burials in Gaul are small in number and mainly of adult females, suggesting the marrying of individuals whose origins lay in central Europe into populations further west; if so, they are not a good indicator for Visigothic identity.

Telling is the archaeology of fifth-century Toulouse, seat of the kingdom. In terms of material culture, there is a group of six Armbrustfibeln from the city and its environs (Bach et al. Reference Bach, Boudartchouk, Cazes, Rifa and Stutz2002: 534, 536), but this is a type with a wide distribution within the Germanic culture-province, so it cannot be specifically linked to a Visigothic identity. There are also three brooches of the Duratón type (Bach et al. Reference Bach, Boudartchouk, Cazes, Rifa and Stutz2002: 535, 537), a type whose main distribution lies in Spain, where it has traditionally been a marker of ‘Visigothic’ settlement, though such easy equivalence is now under serious question (see below, p. 366). It is only at a horizon datable to the turn of the fifth and sixth centuries, immediately before the Frankish conquest, that more items of apparel of ‘Germanic’ type are recorded from the city; this is a horizon we shall return to below, since it is an important one all across what the texts tell us had become the Visigothic realm. By contrast, there is very little ‘Germanic’ visible in the ceramic assemblages of the fifth century from Toulouse, which remain dominated by utilitarian, local productions in the Gallo-Roman tradition along with DSP of both ‘Atlantic’ and Languedoc types and a tiny amount of African amphorae and ARS (Dieulafait et al. Reference Dieulafait, Boudartchouk and Llech2002), though again recent research has identified a number of pieces of pottery whose parallels lie in eastern Europe, but numerically they are a minute group compared with the overwhelming dominance of Roman provincial ceramics. This is true also of other assemblages in the region, such as that from the fifth-century deposits in the upper town of Saint-Bertrand-de-Comminges (Dieulafait Reference Dieulafait2006), where the ceramics remain resolutely Roman provincial.

When we turn to structures, rather than material culture, recent excavations in and around the walls of the city, combined with re-evaluation of antiquarian observations, have yielded a complex of sites in the north-western area (Figure 8.2). The largest and most striking of these is the site of the Hôpital Larrey just inside the north-western angle of the enceinte (for a summary, see De Filippo Reference De Filippo2002). A building range probably some 90 m long by 30 m wide had a central entrance way with, to either side, large internal courts with, originally, large apsidal terminals against the sides of the entrance. Along the façades of the building were two long galleries or suites of rooms. Because of the later demolition of the building and clearance of the site, the dating evidence was sparse but pointed to the first half of the fifth century. Another major structure of similar date lay a little to the north-west outside the walls, to the north of the later church of Saint-Pierre-des-Cuisines (Cazes and Arramond Reference Cazes and Arramond2002), consisting of a roughly east–west gallery façade with central entrance way. Large-scale constructions of this date were rare anywhere in the West. This has led some to postulate that these structures reflected the new power at Toulouse and to suggest that the Hôpital Larrey site in particular may have been part of the palace of the Visigothic kings. Following on from this suggestion, it was then proposed that the core of the medieval church of Notre-Dame de la Daurade (Cazes and Scellès Reference Cazes and Scellès2002), unfortunately destroyed in 1761 but of which engravings exist (Figure 8.3), was associated with the Hôpital Larrey site, lying as it did only some 250 m to the south-east. This church was a curious polygonal structure with an interior richly embellished by niches framed with marble colonnettes inlaid with gold mosaic, and the niches decorated in gold-ground mosaic with three registers of biblical personages and scenes. Clearly, this was one of the richest and most elaborate programmes of decoration known anywhere in the West at this date (the fifth century). Its function remains unclear. It has been posited to be a royal mausoleum, but its position within the walls makes this less likely. Many prefer to interpret it as the palace chapel of the Visigothic kings (mentioned in Sidonius Apollinaris’s idealising account of his visit to the court of Theodoric II in 455 – Epistulae II.i), and the proposal that the iconography of the mosaics may have reflected Arian theology would fit since the Visigoths were still adherents of that heresy in this period. It should by now be clear that, first, there is no proof positive that these structures had anything whatsoever to do with the Visigothic kings, and that, second, the reason why they are so difficult to claim as ‘Visigothic’ is that in plan, layout and what is known of the decoration they were solidly Roman provincial, with, for example, the best parallel to the Hôpital Larrey plan to be found at the villa of Nérac (Lot-et-Garonne) to the north-west of Toulouse (Balmelle Reference Balmelle2001: 390–3). This echoes the evidence from the countryside for the archaeological near invisibility of the Visigoths.

Fig. 8.2 Toulouse, ‘Visigothic’ sites in the north-western part of the city. A. Saint-Pierre-des-Cuisines; B. Hôpital Larrey building; C. Notre-Dame-de-la-Daurade

Fig. 8.3 Notre-Dame-de-la-Daurade, plans and view of the interior before demolition

Why then this near invisibility for the first fifty years or so (a couple of generations) of the kingdom? This must relate to the experience of the Goths prior to their settlement in Aquitaine and to the circumstances of their settlement and subsequent integration. As has been pointed out on various occasions, prior to their arrival in Gaul, the Goths had spent some forty years touring the central areas of the Roman Empire: cultural influences can flow both ways, and in this case rather than Romans adopting ‘barbarian’ fashions, it appears that the Visigoths were ‘Romanised’. After the sack of Rome in 410, they were clearly under the control of the patrician Constantius, who used them as a proxy for the much weakened Western Roman army in both Gaul and Spain. So by the time they were settled between Toulouse and the ocean, they were thoroughly accustomed to Roman ways (Alaric I’s successor Athaulf had even been married to the emperor Honorius’s sister Galla Placidia during the Goths’ brief sojourn in Barcelona) and must have been using the Roman vocabulary and semiotics of rank and power alongside any of their own. Given that the Roman government formally settled them in the south-west with a treaty and with fiscal provision rather than their invading and taking over, they probably had more to gain from accommodation with the existing system in this very wealthy area, studded, as we saw in Chapter 5, with some of the largest and most splendid villas of the period, housing rich and powerful landowners. It looks as though the first couple of generations of Visigoths adapted to the existing Gallo-Roman style rather than the Gallo-Romans adapting to the Visigothic, so what the well-dressed Visigothic warrior of the mid fifth century wore and fought with is rather a difficult problem for the archaeologist. But as noted above, in the last third or so of the fifth century this started to change, and the ‘Visigoths’ become more visible in the archaeology, and it is this change and the reasons behind it that will be considered below in conjunction with the evidence from Spain (p. 375).

Spain

The Iberian peninsula in the fifth century saw, according to the texts, a whole range of different Germanic peoples, Alans, Sueves, Vandals (both Asding and Siling) and Visigoths, either passing through or carving out for themselves territories at the expense of the Roman state and the Hispano-Roman population, such as the kingdom of the Sueves centred on the late Roman province of Galicia and eventually, in the sixth century, the Visigothic kingdom encompassing the entire peninsula. The historical sources for the period are of varying forms and degrees of narrative reliability (the latter was often not their purpose) and have given rise to a huge literature of possible scenarios for what was going on and why, though it is generally agreed that many of the ‘barbarian’ peoples, such as the Vandals, left next to no trace in the archaeology, usually because they were too transient (see Arce Reference Arce2005a, Reference Arce2005b for a balanced and judicious treatment of the sources and of the history of the fifth century). The Visigothic period in Spain has received a great deal of attention from historians because of its central place in the creation of hispanidad through the forging for the first time of a state comprising the whole peninsula, a state, moreover, that was Christian – indeed, we know most about that state through the decrees of a whole series of Church councils. This state was to be overthrown by the Muslim Arabs from 711, leading to the nearly eight hundred years of the reconquista culminating in the expulsion of the Moors by Los Reyes Católicos in 1492, the crucible in which the Spanish ‘national identity’ was formed (cf. Collins Reference Collins2004: Introduction). In addition, ‘Visigothic’ has also become a chronological and architectural/art-historical style, as well as an ethnic appellation, but that is for a period later than is our concern here. Because of our knowledge of the development of the Visigothic kingdom in the sixth and seventh centuries, there has inevitably been a certain amount of reading back from later conditions into the late fifth century (for instance, of the later legal bans on intermarriage between Goths and Romans, ironically recycling late Roman legislation which saw the problem from the other side) and of the use of the textual sources to condition study of the archaeological material. Here the intention is much more limited; it is to look at the development of what has been interpreted as a ‘Visigothic’ material culture, particularly in the northern part of the peninsula, in the fifth century and to relate this to developments in Septimania (roughly modern Languedoc-Roussillon) and in the Toulouse region.

The Iberian peninsula has yielded a large number of cemeteries that contain burials equipped with items of dress, equipment and personal adornment that were clearly not of Roman provincial types (Figure 8.4). These burials and cemeteries could be dated to the fifth to eighth centuries from the Roman-style objects in them and from parallels to the non-Roman material elsewhere in Europe (for a clear introduction to and summary of the archaeological evidence and the analyses thereof by a range of workers, see López Quiroga Reference López Quiroga2010: esp. Ch. 3). They were concentrated on the Meseta of the central and northern parts of the peninsula, but there were examples of them more or less over the whole peninsula (cf. Ebel-Zepezauer Reference Ebel-Zepezauer2000: Abb. 1) if we include the entire date range. Here our concern must be with the earlier part of the sequence, datable to the fifth century. In the 1940s, the excavation at two sites above all, Duratón (Segovia) and El Carpio de Tajo (Toledo), produced relatively large cemeteries with some graves containing numbers of ‘barbarian’ objects, associated for the most part with female burials and betokening a form of clothing different from that current among the female Hispano-Roman population. The objects were principally brooches, Armbrustfibeln, Bügelknopffibeln and Blechfibeln with some of the rarer Adlerfibeln (in the form of an eagle usually with cloisonné decoration). These were usually worn in pairs at or near the shoulders, suggesting females buried in a two-part tunic or peplos. Associated with these were elements of belts, most usually buckles, often with rectangular plates decorated in a variety of ways but most characteristically with polychrome cloisonné work. Another frequent find was necklaces, generally of glass beads but sometimes of other materials including amber; also occasionally found were earrings. It will be noted that what we have here are female graves with a distinctive, non-Hispano-Roman style of dress. The male graves were much less distinctive and in fact seem to be assimilated to Hispano-Roman traditions of burial (see below). In the intellectual framework prevailing at the time these cemeteries were excavated, particularly that created by German scholars such as Kossinna, these were seen as evidence for the movement of groups of ethnically distinct peoples of Germanic origin across Europe and settling ultimately in Spain. From that it was only a short step to accepting that these were the burials and cemeteries of the Visigoths, who, the texts stated, had moved over the Pyrenees in the latter part of the fifth century. For instance, a Spanish text, the Chronica Caesaraugustana, spoke of the settlement of the Goths between 494 and 497, and this was assumed to have been reinforced by refugees from north of the Pyrenees after the defeat of the kingdom of Toulouse at Vouillé in 507. Since at the time workers lacked independent dating for the burials, this gave a terminus post quem for the appearance of these burial rites and objects.

Fig. 8.4 Spain, ‘Duratón’ grave goods of Ripoll López’s Nivel 2

A major step forward came with the systematisation of the information on these burials accompanied by a re-examination of certain key sites such as El Carpio de Tajo, which was published by Gisela Ripoll López (e.g. Reference Ripoll López1985, Reference Ripoll López1991, Reference Ripoll López1998; English-language summary in Ripoll López Reference Ripoll López1999). She arranged the items of dress and personal adornment into a series of Niveles or levels (corresponding to the arrangement by Stufen of German workers such as Böhme), affording a chronotypology of the material with objects assigned to one or other of her Niveles, though because of the lack of independently datable material, it proved difficult to translate this relative chronology into an absolute one. Ripoll López’s analysis is that there was Visigothic settlement from about 480 in the northern Meseta (corresponding to the expansion of the kingdom of Toulouse under Euric I), the cemetery evidence suggesting the importance of kin and the visible signs through dress suggesting a new, Visigothic identity. Further developments in the material culture and in the cemetery population through the sixth and on into the early seventh century were linked by Ripoll López to the historical record of the Visigothic kingdom in Spain, this burial tradition ultimately dying out in the early eighth century as the Visigothic kingdom succumbed to the Moorish invasions from 711. This approach has been criticised by a number of workers, such as Collins (Reference Collins2004: 174–86), for relying on a priori assumptions deriving from the historical evidence, in particular the absolute dating of the various Niveles by reference to supposed historical events of the fifth to eighth centuries (such as the dates for the entry of the Visigoths into the peninsula) and the relating of aspects of the burial rite to supposed earlier Gothic (or Tervingi/Greuthungi) practice (the Sîntana de Mureş-Chernyakhov culture), which thus restricted discussion of them to a Visigothic identity. Indeed, as he points out, there is little in the metalwork to argue directly that those buried with it were necessarily Visigothic (as we shall see below), and the literary sources attest to the presence of other non-Roman peoples in the peninsula.

More recently still, there has been further reconsideration of the date and significance of these burials. This was prompted partly by the excavation to modern standards of further cemeteries of this type, such as la Olmeda (Palencia), or sites in the province of Madrid such as Cacera de las Ranas or Gózquez de Arriba (the excavations in the 1940s were much more summary and major questions remain over how accurate the groupings of objects apparently from one burial actually are); partly by an improved knowledge of similar burials and material elsewhere in Europe; partly by the application of the more recent thinking about ‘identity’ and ‘ethnicity’ (including increased recognition of the other ‘Germanic’ peoples recorded as present in the peninsula); and partly by the recognition that these burials often form a minority in the cemeteries concerned (López Quiroga Reference López Quiroga2010: 199–268).

The dating of the objects from these burials has been greatly refined, partly by taking more account of the ‘Roman’ material from the graves and partly because of improved chronologies for their parallels elsewhere in Europe. It is clear that the earliest of this material dates to the first half of the fifth century and belongs to the horizon sometimes termed ‘Pontico-Danubian’ after its distribution in the area west of the Black Sea (Pontus) and in the lower Danube basin, linking through to the area of the Sîntana de Mureş-Chernyakhov culture. This material includes gold-and-garnet belt fittings from Portugal at Beja (cf. von Rummel Reference von Rummel2007: 342–53) and the elaborate gold necklace from the Blanes site at Mérida. Another highly distinctive necklace, from Beiral (Algarve), has parallels in rich, non-Roman burials elsewhere in fifth-century Europe, as well as at Valleta del Valero (Lleida) in north-eastern Spain. A comb of a distinctive form, rectangular with a semicircular upper projection, from Castro Ventosa (León) again has parallels in the Sîntana de Mureş-Chernyakhov complex and also parallels those from north of the Pyrenees in the kingdom of Toulouse. By contrast, a necklace from Vigo (Pontevedra) has beads of amber of a distinctive mushroom form, otherwise known from the Elbe region (normally seen as the homelands of the Alamanni). These and a number of other objects without good archaeological provenance (including a sword from Beja) constitute what López Quiroga (Reference López Quiroga2010: 112–32) terms his Nivel 1A, which he dates to the first half of the fifth century. In it he sees a strong military element, but notes that it is impossible to give any precise ethnic ascription to these objects, or, indeed, to tell whether they may have been worn by ‘barbarians’ in what was left of the late Roman army of the period.

In the second half of the fifth century, the presence of non-Roman material becomes more noticeable and widespread; this is the period of the early phases of classic cemeteries such as Duratón and El Carpio de Tajo – the Nivel 1B of López Quiroga (Reference López Quiroga2010: 133–49). As well as the Armbrustfibeln and Bügelknopffibeln already well established from Nivel 1A, Nivel 1B is the heyday of the Blechfibeln of ‘Smolin’ or ‘Kosino-Gyuilavan’ type (named after sites in the Pontico-Danubian region) with expanded, silvered head plates and large foot plates. Other elements of the material culture, such as other brooch types and belt fittings, also derive from the area of the Sîntana de Mureş-Chernyakhov culture, showing the east European area of influence, an influence also present in the north of Gaul (Kazanski et al. Reference Kazanski, Mastykova and Périn2008; Kazanski and Périn Reference Kazanski and Périn2006), a region not usually regarded as a major focus of Visigothic settlement. The focus of distribution of burials containing this material remains on the Meseta of the centre and north of the interior of the peninsula, but extends into other regions also, as far as Andalusia in the south (cf. Ebel-Zepezauer Reference Ebel-Zepezauer2000: Abb. 35), with the later examples of this material well evidenced in Septimania, the Languedoc-Roussillon littoral, where the Visigoths held out against the Franks after the defeat at Vouillé. This discussion has focused on the material culture because that is the most widely distributed and most intensively reported and studied aspect of these sites and the one that has been most important in constructing a Gothic identity for these burials. But a small number of burials demonstrate a practice that is also a strong link with the Pontico-Danubian region – the placing of the corpse(s) in niches cut into the side of the grave pit, a practice observable across much of eastern Europe, which has been linked with the ‘Alano-Sarmatian’ culture. Be that as it may, it was certainly alien to Hispano-Roman funerary practice (López Quiroga Reference López Quiroga2010: 151–6). It is not difficult to see why these burials were linked with Visigothic identity both in the 1940s and subsequently, since their distribution corresponds with important areas of the sixth-century Visigothic kingdom such as the Meseta and Septimania, with a peninsular distribution more generally, and the objects are clearly of non-Roman, ‘Germanic’ derivation. But it is now necessary to reassess the archaeological evidence and its possible meanings independently of the written sources and their pre-existing agenda.

There are a number of points about these burials and cemeteries that need to be made (I am most grateful to Gisela Ripoll for discussion on these points). First of all, in some of the cemeteries that have been excavated and recorded with sufficient care, from Duratón and El Carpio de Tajo on, the burials with distinctive rites and/or material culture normally form a minority of the total cemetery population. For instance, burials with dress elements form just under a third (31.57%) of the burials excavated at El Carpio de Tajo; the other two-thirds lack such material. At Cacera de las Ranas, the proportion was similar, 51 out of 145 recorded burials (35.17%) had dress fittings. Others, though, had much higher figures; 34 of the 52 burials (65.4%) from Herrera de Pisuerga contained grave goods. But much discussion of these cemeteries and the material from them has tended to concentrate on the contents of the furnished burials and to ignore the unfurnished ones, decontextualising the ‘Visigothic’ burials from much larger numbers of burials in the Hispano-Roman tradition and thus risking giving a false impression of a dominance in the funerary record that these burials simply do not have; they are a minority of known fifth-century burials in the Iberian peninsula. We shall return to the question of the relation with other burials. Of the furnished burials considered as ‘Visigothic’, the great majority seem to be female, though this is usually argued on the basis of the objects rather than of the osteological evidence, clearly running the risk of a circular argument. It is worth noting that there is very little pottery of ‘Germanic’ type from these cemeteries – we are dealing essentially with the evidence of dress. The most common elements are the pairs of brooches and belt buckles; more rarely, there are necklaces or strings of beads (which probably ran between the brooches), as well as earrings. Particularly notable are the more elaborate Blechfiblen and Adlerfibeln, whose disposition on the body was part of a suite of dress adornments. Comparison with other cemeteries of the period and with what is known of ‘Roman’ and ‘barbarian’ dress in the period (von Rummel Reference von Rummel2007) strongly suggests that this was a female form of accoutrement, related to the fastening of a distinctively non-Roman dress form.

The stress placed on the form of dress of females in these cemeteries suggests both that they were of particular social importance and that their non-Romanness was important; a very specific set of statements about their identities was being constructed by the ways in which this material was being deployed. Given the links of the dress elements to the Pontico-Danubian region, it would seem overcritical to deny that non-Roman-provincial material culture was being used to make statements about these burials, though to move from this to specific ‘ethnic’ interpretations is almost certainly not possible. Both the Pontico-Danubian region and the Iberian peninsula contained a mixture of different ‘ethnic’ or sub-‘ethnic’ groupings, to judge by the textual sources (Alans, Goths, Sarmatians, Sueves, Vandals, etc.) and to judge archaeologically by the mixing of elements from different antecedent culture groups (e.g. Przeworsk, Sîntana de Mureş-Chernyakhov and Wielbark), rather than the racially pure Stammen of early twentieth-century views. The predominance of female accoutrements in these graves has caused sight to be lost, to a considerable extent, of what constituted a male grave (cf. Ebel-Zepezauer Reference Ebel-Zepezauer2000: 130–2 – less than two pages in all). Unlike northern Gaul, it is not possible to identify a tradition of weapon burial in the Iberian peninsula (there is one weapon burial from the ‘Visigothic’ cemeteries). With the lack of this or the deposition of other grave goods that might be held to express a masculine identity, it would seem that either the menfolk of these women were not buried with gender-specific accoutrements, or these graves may comprise some of the simpler burials with just belt fittings, though it would be useful to know whether, for instance, the Armbrustfibeln and Bügelknopffibeln in many of these graves had retained the male gender significance their antecedents in the late Roman world had had. This will come with more research and publication that genders burials through osteology rather than objects. The visibility and presumably importance of females in the funerary world is very much at variance with the texts, which depict an almost exclusively male world, one with a strong military or warrior ethos (for similar problems of differential gender visibility in the burial record of fifth- and sixth-century ‘Ostrogothic’ Italy, see Barbiera Reference Barbiera2010). So there remains then the question of why these particular female burials were invested with such distinctiveness. Since the emphasis here is on the non-Roman identity of the deceased through her dress, it may be that what we are seeing is an important matrilineal element in the social structures of these people(s), with the wives and mothers confirming the separateness of the identity of their husbands, sons and daughters. It may be no coincidence that, on the textual side, the law codes of the Visigothic kingdoms, from the Breviarium of Alaric in 506 down to the end of the sixth century, were exercised by the problem of intermarriage between Romans and (Visi-)Goths.

Another point worth making about these burials relates to the questions of social hierarchy and cemetery location. The social hierarchy expressed by these burials, especially from the middle of the fifth century, would seem not to be very developed, to judge both by the quantities of grave goods and by their materials (largely copper alloy with glass and paste insets). As yet there are no ‘princely’ burials from the peninsula what might betoken the upper reaches of a hierarchised society. Yet the picture presented unanimously by the texts is of a society ruled by a king with a nobility. The current absence of high-status burials is therefore worth remarking on. It may be that since the various incoming peoples were Christian we should be looking for such burials in and around the principal churches of cities such as Barcelona and Braga (and eventually Toledo) or at other forms of high-status settlement, about which we presently know little (see Chapter 9, p. 441). In this case, what we may be seeing in the rural cemeteries under consideration here are non-elite segments of the society. The increasing appreciation of the context of these burials in cemeteries where the majority may not follow this particular female burial rite also raises the question of the contribution of the indigenous, Hispano-Roman population. In particular, are the distinctive burials those of an element of the population ‘foreign’ to the cemetery, or might they represent the adoption by local families or kins of the self-representation of an incoming and important new strand in the population of the peninsula? Do we have here what Brather (2008: 429) characterises as ‘politisch und nicht ethnisch gotisch’, with groups within the indigenous communities choosing, presumably for reasons of self-advancement, to define themselves as these political and not ethnic Goths? It is also worth pointing out, following, for instance, Fuentes (cited in von Rummel Reference von Rummel2007: 53), the coincidence of the heartlands of these ‘Visigothic’ cemeteries with those of the earlier, fourth-century, ‘Duero’ burial tradition, the ‘cementerios tardohíspanos con ajuares’ (p. 88), which had some similarities in its use of dress to distinguish certain individuals or groups; were the fifth-century traditions to an extent a development of already existing trends in Hispano-Roman burial practices in the Meseta region? On the other hand, it is worth remembering the small number of highly distinctive burials with the corpses in lateral niches; this is a central- and eastern European practice, not an Iberian one.

There is then the question of the relationship of the material culture from these Spanish cemeteries to that north of the Pyrenees, the area from which the Visigoths are reported to have arrived in the last third or so of the fifth century. As we have seen above, the visibly ‘Visigothic’ archaeology of the kingdom of Toulouse remained weakly developed through much of the fifth century. Some of the material, such as the small number of combs, does have the occasional parallel south of the mountains (e.g. the comb from Castro Ventosa), but this just demonstrates the weakness of the comparanda; one would not construct a story of a Visigothic kingdom in south-western Gaul which invaded the Iberian peninsula on the basis of four combs to the north of the Pyrenees and one to the south. There is little in the archaeology of the kingdom of Toulouse which could be said to be antecedent to, and the source of, developments south of the Pyrenees through the fifth century. In fact, recent research, as implied above, on the antecedents for the material in Spain points rather to the Pontico-Danubian region, well to the east rather than the north, and such distinctive items of ‘Visigothic’ dress as the Adlerfibeln seem to originate from there, possibly by way of Italy and the Ostrogoths rather than the Visigoths. It will be clear from the above that it is not possible to identify archaeologically the Visigothic settlement of the Iberian peninsula in any way that resembles that depicted in the various textual sources; and the same goes for the Sueves, Vandals and others. This is not to say that the texts recording Visigothic and other Germanic invasions, migrations and settlement are ‘wrong’; they were representing something that seemed real (or at least plausible) to them at the time they were written. But they may have been partial in their view; they may have oversimplified a much more complex situation; or, if written outside the south-west of Gaul or the Iberian peninsula, they may not have had a clear appreciation of what actually went on. There is also the problem that most of the authors, if not all, were from the Roman tradition or even ‘side’ and that, therefore, precise ethnographic and chronological accuracy was not their primary concern. That there were Visigoths settling in the Iberian peninsula in the course of the later fifth century cannot be doubted. What can be doubted is the ability to recognise them in the ways traditionally employed.

The dating of this material now looks to start considerably earlier, by at least a generation, than the historical dates for the settlement of the Visigoths in Spain. It is, moreover, very difficult to identify any ‘type fossils’ in the material culture that point to a specifically Visigothic identity; the nearest one can get is the presence of material closely linked to the Sîntana de Mureş-Chernyakhov complex. Even so, this material is very much linked to the aspects of life (or death) to do with personal appearance; other aspects such as those represented by ceramics are very rare both north and south of the Pyrenees. So it would seem that what has been called ‘Visigothic’ identity was being constructed in very specific ways and in very specific contexts. Given that in the cemeteries burials with this type of material were often in a minority, it may be that this was just one of a range of possible strategies for making statements about the dead, and those statements may not be to do with what we term ‘ethnicity’ but rather with gender and status; this type of burial is heavily gendered. Other statements could have been made elsewhere in the funerary cycle, ones that are not so archaeologically visible; for instance, this may be where the affirmations of maleness were made but do not show up with the deceased in the grave. It is also worth recalling the question posed by Theuws (Reference Theuws2009) of whether ‘ethnicity’ was something that mattered when preparing the dead for the tomb in antiquity, or is it just something that matters to us? Might the material thought of as ‘Visigothic’ – or even just ‘Germanic’ – simply have been useful for making statements about the dead that were not concerned with their ethnicity? There is another methodological problem that is not really acknowledged. There does not seem to be a complementary non-‘Germanic’ suite of dress and other items; there does not seem to be a Hispano-Roman Tracht, or, at least, not one that is visible to the archaeologist. If those burying the dead wanted to make certain statements about the deceased that involved using dress as a signifier, did they have an alternative to using what we style ‘Visigothic’? So the positive visibility of ‘Visigothic’ markers may be being emphasised by the lack of competing markers.

Nevertheless, even without the textual evidence, the distribution of this material by the turn of the fifth and sixth centuries over most of the Iberian peninsula, with a concentration in the northern Meseta but starting to extend further – for instance, into the south (the former Baetica) and the north-east, with a marked concentration north of the eastern end of the Pyrenees in Septimania (the old Narbonensis) – would suggest an archaeological culture province with a certain unity in its material culture and funerary practices, if not in other aspects of life. Why did the production of such material and its deployment as signifiers in burials become so visible from the later fifth century? An explanation that may have value is that of political and social stress. In periods of uncertainty and change, self-definition and self-representation become more important as a means of staking out both ‘emic’ and ‘etic’ identities and thus the place of an individual or group in relation to others, and as a means of maintaining the internal cohesion of the group. Bearing this in mind, we may consider the material culture in relation to the historical narrative, reading the texts not for a precise depiction of ‘what happened’ but more for a general context within which different groups were operating. It was argued above that the relative archaeological invisibility of the Visigoths of south-western Gaul was due to the circumstances of their settlement and subsequent deployment; it suited them better to integrate with the existing Roman provincial culture than to stand out against it. But in the later fifth century circumstances seem to have changed to ones where identities mattered, though those identities were not necessarily ‘ethnic’. South of the Pyrenees, they were confronted by groups such as the Sueves in the north-west of the peninsula or the Hispano-Roman populations with the remains of other Germanic peoples of passage. As noted above, even in the heartlands of the kingdom, at Toulouse itself, it is in the late fifth century that distinctive brooches and buckles begin to appear in any number, just at the point that the threat from the Franks became clear. It would seem unlikely to have been a coincidence that it was in the same year, 506, the year before the climactic campaign for the existence of the kingdom of Toulouse, that a set of laws defined Visigothic royal jurisdiction (the Breviarium of Alaric) and the first Church council reuniting all the bishops of the Visigothic lands at Agde was held. Surely Alaric II was trying to define ‘Visigothic-ness’ as never before (and, as we shall see below, it was precisely at this period that the ‘Franks’ also forged a series of archaeologically visible correlates to their proclaimed identity). In an instrumentalist reading of the use of material culture in constructing ethnicity, it can be argued that it was only in the later fifth century that the Visigoths, both as an ethnic group and as a political entity, had the need to make ever more public statements of their affiliations. To the archaeologist, this may appear above all through the medium of dress, in particular the Tracht in which the womenfolk of those who defined themselves as Visigoths were buried, making it clear where their ethnic loyalties (which were, of course, not necessarily the same as their ethnic origins) lay. On the other hand, the circumstances of the Visigoths’ intrusion into the lands south of the Pyrenees and the fact that there they discovered a number of discrete ethnic groupings who contested their intrusion probably induced another kind of situational stress. The Visigoths found themselves in a milieu where in the mid to later fifth century they needed to define and display their identity in a way they did not need to north of the western half of the Pyrenean chain. The other area where, perhaps even more concentratedly than in Spain, they proclaimed their identity loud and clear was Septimania. This may be because the capture of Septimania had been in the face of bitter opposition from the moribund imperial authorities, and after the fall of Toulouse in 507 the Visigoths of the region were now the only ones of their people north of the Pyrenees who had retained political autonomy and were facing the growing power of the Franks. So both in the kingdom of Toulouse towards the end of the fifth century and in Spain from a couple of decades earlier, we may be able to see reasons why a hitherto archaeologically shy grouping increasingly found it necessary to make public statements about their ethnic affiliations, real or imagined. From the early sixth century, the rulers of Spain deployed Visigothic identity as a political tool to help legitimate their power, create an identity for their new state, and promote its cohesion (cf. Collins 2006: Ch. 9), albeit under Ostrogothic hegemony for much of the first third of that century. A united kingdom comprising all the peninsula was not achieved until the 570s. Whether this identity is the explanation for the dress of some female burials in rural cemeteries has, for now, to remain an open question.

The Franks and northern Gaul

The end of the preceding discussion of the Visigoths has introduced the Franks as another major ‘barbarian’ power player by the turn of the fifth and sixth centuries, and the study of their archaeology provides us with a salutary case study. The literary sources present us with a coherent if sketchy account of the early history of the Franks from the later third to the later fifth centuries (for discussions of the textual evidence for this period, see Feffer and Périn Reference Feffer and Périn1987: Chs. 1–3; James, E. Reference James1988: Ch. 2; Rouche 1997). From the beginning, they seem to have comprised a number of different groupings such as the Chamavi, Bructeri or Salii, giving the impression of a confederation rather than a single ethnic identity: their name (recorded, of course, only from the Roman side), Franci, seems to be a Germanic word signifying ‘fierce’ or ‘bold’. In addition, to the north of the lower Rhine, the Roman sources also name peoples, such as the Frisians and the Saxons, distinct from the Franks. The Franks are recorded by the Romans as making trouble for them in the later third century and again in the middle of the fourth. On the other hand, the Notitia Dignitatum lists several units of Franks serving in the Roman army, as far from home as Egypt, and some Franks rose to senior commands under the emperor, the magister militumArbogast even creating an emperor, the usurper Eugenius. During this period they took over and settled the area of northern Gaul bounded to the north by the Rhine, to the west by the North Sea and to the east by the Maas/Meuse, and lying north of the Cologne–Bavai–Boulogne fortified road (cf. p. 47), the area known as Toxandria, which became the base of the Salian (‘salt’) Franks. We hear little of them in the late fourth and early fifth centuries, and it is not until the middle of the fifth century that we hear of them again, expanding southwards but also on good terms with Aetius and supporting him against the Huns in 451. Historically, the crucial period is the reign of the two kings Childeric (d. c. 481), who arguably forged a Frankish identity, and his son Clovis (c. 481–511), who defeated Syagrius, the last soi-disant sub-Roman commander in northern Gaul. In 486 at Soissons, Clovis converted to Christianity and was baptised at Reims – this is usually said to have happened in 496 – and he then went on to defeat the Visigoths of Toulouse at the battle of Vouillé in 507, thus gaining command of almost all of Gaul. Clovis has been a figure of immense importance in the construction of the identity of France as the king of the Franks who unified much of the later French national territory and converted his people to Catholic Christianity, cementing both by the calling of a Church council at Orléans in 511, the year of his death. His places of baptism and burial, Reims and Saint-Denis near Paris, were to become the royal coronation and mausoleum churches of medieval and later France (cf. Rouche Reference Rouche1997: vol. II). Not surprisingly, the scanty historical record has been much studied from this perspective of hindsight. Since the discovery of what is generally accepted as Childeric’s tomb at Tournai in 1653, Frankish material culture has also been drawn on (famously, Napoleon I had golden bees modelled on those in Childeric’s tomb sewn onto his imperial coronation robe of 1804).

But in fact, as with the Visigoths, the archaeology is far more equivocal and stands as a good case study of the relationship between what was happening on and in the ground and the external perspective of Romans at the time or of later chroniclers and historians such as the late sixth-century bishop, Gregory of Tours, in his Historia Francorum. If we look at the archaeology of the Rhineland, on both sides of the river, in the fourth century, we see a picture of a wide range and a mixture of different groupings, expressed especially through their material culture – what we have already encountered under Böhme’s (Reference Böhme1974) label of a Mischzivilization – a culture province characterised by a mixing of Roman provincial and ‘Germanic’ elements, with the Roman army playing a central role in its creation. This is echoed by Whittaker’s characterisation of Roman frontiers as a zone rather than a line, where those in the zone had more in common with each other than with their ‘parent’ cultures to the rear (Whittaker Reference Whittaker1994: Ch. 4). As well as the Roman provincial population and the units of the Roman army (cf. Chapter 2), there were two major Germanic groupings, the Franks on the lower Rhine and the Alamanni on the middle and upper Rhine. In fact, the Alamanni seem to have been as much a confederation of smaller groupings as were the Franks, and it is very difficult to identify anything in the archaeology east of the middle and lower Rhine that was distinctively ‘Alamannic’ rather than generically Germanic and being influenced by Roman provincial practice, in particular the huge culture grouping between the Elbe and the Rhine long characterised as the Elbegermanen (for discussion and a bibliography, see Drinkwater Reference Drinkwater2007: Ch. 2), but whose extent overlaps with the presumed homelands of the Franks as well as of the Alamanni (Drinkwater Reference Drinkwater2007: Ch. 3; Fuchs et al. 1997: 20–110; Theune Reference Theune2004: 57–196). The confederate nature of the Alamanni agrees with the proposition that proximity over time to the Roman Empire encouraged socio-political change among the peoples along the frontier, especially the deepening of the hierarchy and the extension of control over a number of hitherto autarchic groups or peoples, and the names of some of the groupings within the Frankish complex are known as those of independent ‘tribes’ in the first and second centuries. This was in part promoted by the development of rulers or elites with access to Roman symbols of power (often buried with the dead) along with control of goods traded with or looted from imperial territory. The funerary evidence (Drinkwater Reference Drinkwater2007: Ch. 3; Steuer Reference Steuer1997) shows that a warrior identity was important on the male side, so warfare either with Rome or with neighbouring peoples, to bring them into subjection or to take tradable goods such as slaves, would have contributed to the socio-political modifications in train. This is, of course, a variant of the core-periphery and prestige-goods models proposed some time ago for later Iron Age north-western Europe (cf. Woolf Reference Woolf1990), but they seem to have value in this context also.

As for the Franks, the following discussion will avoid that precise ethno-cultural label, preferring instead to talk of the evidence for Germanic (and other) settlement, since, as we shall see, if we did not know that some of the inhabitants of these areas of northern Gaul were called Franks, it might be impossible to identify them archaeologically as a distinct group. Traditionally, the main way of trying to tie the historical Franks to the archaeology has been the presence in burials of a type of throwing axe with a long, curved blade, which has been linked with the textual mentions of the francisca as the defining weapon of the eponymous Franks. The problem is that the written sources for this weapon are all relatively late and sometimes distant geographically as well – for instance, the early seventh-century bishop of Seville, Isidore – and they have great difficulty in establishing whether the francisca was just a securis, axe, or a bipennis, a double-headed axe (which, of course, would not fit the archaeological ‘francisca’ at all) (cf. Pohl Reference Pohl and Reimitz1998: 33–6). The distribution map of the single-bladed axe type usually labelled francisca (Hübener Reference Hübener1977) shows a spread across northern Gaul, particularly between the Seine and the Rhine, and eastwards into the Rhine–Danube re-entrant, thus corresponding broadly with the heartlands of the historical kingdom of the Franks but also with a large area historically labelled as belonging to the Alamanni. Unfortunately, the dating of these finds is not always secure, so it is not certain which may be of sixth-century date and thus part of the conscious creation of a Frankish identity (see below). So instead of accepting a predetermined textual narrative and agenda, we shall look at the evidence from the fourth and fifth centuries for the spread of an archaeological complex whose origins lay east and north of the Rhine and which it is thus reasonable to label ‘Germanic’. The evidence will principally be that of settlement and building types, on the one hand, and material culture, particularly from funerary contexts, on the other hand.

Settlements characterised by the presence of buildings of north German rather than Roman provincial type begin to appear in the Roman Rhineland and northern Gaul in the course of the fourth century (for the sites from a recent survey, see Lenz Reference Lenz2005, with references; cf. Vermeulen Reference Vermeulen2001: 53–63), principally in Toxandria, but some are also now known further south. A detailed study of part of the lower Rhine (Van Es and Hessing Reference Van Es and Hessing1994) shows a classic Mischzivilization with Roman-style military sites and elements of material culture coexisting with structures of Germanic type along with metalwork and pottery, with grave goods belonging to both traditions and showing again the importance of Roman official issue metalwork in Germanic graves. Perhaps the most diagnostic building type was the ‘long house’ (Wohnstallhaus) with opposed entrances in the long sides, and people housed at one end and animals at the other (cf. Hamerow Reference Hamerow2002: Ch. 2); this type was already present in the third and fourth centuries at classic sites beyond the Rhine in Germany and the Netherlands such as Flögeln (Kreis Wesermünde) and Wijster (Drenthe) respectively. These ‘long houses’ are known from Toxandrian sites of the fourth century such as Gennep near Nijmegen (Gelderland) and Neerharen-Rekem (Limburg), but also, more unexpectedly, from the 1990s excavations on the site of Saint-Ouen-du-Breuil (Seine-Maritime) in Normandy north of Rouen (Van Ossel and Ouzoulias Reference Van Ossel and Ouzoulias2000: 149–51) (cf. p. 280). At these three sites and other similar ones, the ‘long houses’ were accompanied by satellite groups of Grubenhäuser/fonds de cabane/sunken featured buildings (SFB), often thought of as Germanic (especially by archaeologists in Britain, where they are seen as diagnostically ‘Anglo-Saxon’), but in fact they are of much less definite origin (cf. Hamerow Reference Hamerow2002: 31–5). In addition to these two very distinctive types, there were also rectangular, post-built structures, again usually with opposing doorways in the long sides, at a variety of sites, including late Roman fortifications, such as Krefeld-Gellep, or villas in northern Gaul, such as Marolles-sur-Seine (cf. p. 430) and Limetz-Ville – again accompanied by Grubenhäuser – or the Rhineland at Köln-Müngersdorf (Lindenthal, Nordrhein) and Voerendaal (Limburg). The number of sites with such structures continued to increase through the fifth century, in Toxandria (Thoen and Vermeulen Reference Thoen and Vermeulen1998) but also in the parts of Gaul from the Seine valley northwards (Lenz Reference Lenz2005: 415–21; Ouzoulias and Van Ossel Reference Ouzoulias and Van Ossel2001; Van Ossel Reference Van Ossel1992). With the exception of the Grubenhäuser, the origins of these building types and of the arrangement of the settlements lie north and east of the Rhine and can be seen chronologically to cross the lower Rhine frontier into Toxandria in the fourth century (and perhaps further south as at Saint-Ouen-du-Breuil), becoming established in northern Gaul in the course of the fifth century. As we have seen, the Germanic settlement in Toxandria was to an extent established with the connivance of the Roman authorities, and Germanic settlement further south in the course of the second half of the fourth century may well have been under Roman auspices as laeti or gentiles. As we shall see, whether these people all thought of themselves as Franks, even in Toxandria, is quite another matter.

The burial rites and material culture are even more ambiguous, particularly in terms of Roman-described ethnicities. Recent studies of fourth- to fifth-century metalwork have increasingly shown great variation in stylistic origins. This is in great measure due to the groundbreaking series of publications by Michel Kazanski and his co-workers, which demonstrates a mastery of the material not only from Gaul but also from a huge swathe of Europe from the Caucasus westwards, and also of publications in a range of modern languages. This has enabled the Gaulish material to be set in a much wider context and links to be made across much of temperate Europe; it is worth noting, though, that the exegetical framework for this major contribution remains strongly ‘ethnic’ within an essentially culture-historical framework. In addition, studies of pottery have also shown the presence of several types of handmade vessels in ‘Germanic’ rather than Roman provincial traditions: few if any of them can be or, even more interestingly, have been confidently labelled as ‘Frankish’ (though some are labelled ‘mérovingien’, simply serving to confuse the issue). As we have already seen both in this chapter and in Chapter 2, even in the fourth century the literary evidence for northern Gaul and the Rhineland, such as the Notitia Dignitatum, refers to peoples such as the Sarmatians (cf. Lebedynsky Reference Lebedynsky2002) and other ‘Oriental’ peoples (Kazanski Reference Kazanski1986; cf. Kazanski and Legoux Reference Kazanski and Legoux1998; Kovalevskaya Reference Kovalevskaya1993) as well as Franks or Alamanni. With the events of the opening decade of the fifth century, a whole range of Germanic peoples, Alamanni, Alans, Burgundians, Franks, Gepids, Goths, Heruli, Saxons, Sueves and Vandals are mentioned (cf. esp. Jerome, Ep. CXXIII), and, by the middle of the fifth century, Huns also. There do appear to be indications in the material culture of changes in northern Gaul at this period; for instance, Swift (Reference Swift2000a: 213–19) has argued that a distinction is visible to east and west of the Maas/Meuse–Sambre line, with perhaps Germanic peoples to the west and a new line of defence along these rivers. Because of the literary sources for fifth-century Gaul, modern scholars of the archaeology have sought to assimilate elements of the funerary archaeology and material culture to the various historically attested peoples.

A good example of this is the Alans (Kouznetsov and Lebedynsky Reference Kouznetsov and Lebedynsky2005: esp. Ch. 7), who crossed the Rhine with the other peoples in 405/6, and within ten years the texts mention them in the Rhineland, Aquitaine and Spain. Then silence for the Alans in Gaul until, in 440, some were settled by Aetius in the middle Rhône valley round Valence, and two years later in larger numbers in the Loire valley round Orléans, at the same time that Aetius was setting the defeated Burgundians to the east in Savoy (for their history and material culture, see below) as part of a programme of stabilising the situation against the Visigoths. Not surprisingly, attempts have been made to identify the area of Alan settlement. One means has been to use ‘Alan’ and its derivatives as a toponym element (e.g. Alainville) indicating Alanic settlement, but, given that none of these names are attested to before the ninth century and most much later, this is not a reliable method (Alemany Reference Alemany2006). More convincing is the linking of a small amount of burial evidence with sites more provably ‘Alan’, such as the objects from the rich female burial at Airan (Calvados) (Kazanski Reference Kazanski1982) with parallels in the Ukraine, or at another Normandy site, Saint-Martin-de-Fontenay (Calvados) (Pilet Reference Pilet1980; Pilet et al. Reference Pilet, Alduc-Le Bagousse, Buchet, Helluin, Kazanski, Lambart, Martin, Pellerin, Pilet-Lemière, Vipard, Clet-Pellerin and Van Vliet-Lanoë1994), where, in a large rural cemetery in use from the fourth to the seventh century, there were a small number of burials of people with their crania deformed by binding in infancy, a practice associated by classical writers with the Alans and the Huns (cf. Bóna Reference Bóna2002: 25–6). It is one that is also found sporadically across eastern and central Europe from the Pontico-Danubian region westwards. The burials do contain some objects with links to the Caucasus, but do not contain objects such as cauldrons that were the hallmark of the Hunnic male burial (cf. Buchet Reference Buchet1988). It has been pointed out (Hakenbeck Reference Hakenbeck2009) that such burials (and similar ones found south of the Garonne and noted above [p. 361]) are very small in number and, where the necessary osteological analysis has been undertaken, they are overwhelmingly those of adult females. It is again central and eastern Europe, the ‘Pontico-Danubian’ region, that is the focus of this practice for fifth-century Europe (its origins lie further east and earlier), suggesting that the outliers in Gaul may be women married out of their own ‘culture’ (Hakenbeck Reference Hakenbeck2009: 77–9), so these individuals should not necessarily be taken as proxies for the movement of larger groupings. In fact, the archaeology of the peoples ancient sources chose to call the Alans, Huns and Sarmatians is in many respects very similar and can be traced back to their areas of origin in southern and south-western parts of the Russian Federation and into the Ukraine; it is normally related to culture provinces such as Przeworsk and Wielbark. By the time they get to Gaul, certainty over precise ethnicity is very difficult for us (if not for them). Even so, intentional cranial deformation would seem to be a very precise indicator of a set of beliefs about personal appearance and would require a detailed knowledge of how to perform the practice. It is a practice with no Roman provincial antecedents but with antecedents outside the empire, and thus one of the ways of proclaiming non-Romanness.

Just to complicate the picture in northern Gaul still further, there is a group of late fifth- to early sixth-century Armbrustfibeln and Bügelknopffibeln, principally in the basins of the middle Seine, the Marne and the Somme with a scattering further east (Kazanski et al. Reference Kazanski, Mastykova and Périn2008: esp. Abb. 19; Wieczorek Reference Wieczorek1996: 353). These are labelled ‘Visigothic’ since their parallels lie in south-western Gaul and to an extent in Spain. This, of course, is perilously close to ascribing an ethnic identity to objects, something only their makers and users can have, objects being used to construct and reflect such identities. In this case, do we have to postulate Visigoths far from home, or simply the last resting place of much-travelled brooches? Moreover, further north, in Toxandria and along the Gallic side of the Channel, there is material, particularly pottery, whose closest parallels lie in south-eastern Britain. This has led workers – for instance, at the fifth-century cemetery of Vron (Somme) (Seillier Reference Seillier1989) – to characterise this as ‘Anglo-Saxon’, a descriptor that has also been applied to other pottery from the Netherlands down along the North Sea littoral (Vermeulen Reference Vermeulen2001: 67–8). This is because the literary authorities fix the Anglo-Saxons in Britain, and they are mentioned on the Gallic side of the Channel, so similar material from the Gallic littoral is ascribed a similar ethnicity. What the people who actually made and/or used this pottery thought of themselves as could have been very different. Of course, many of the other burials in the same cemeteries as the burials and objects noted above, and in other cemeteries without ‘immigrant’ rites and/or objects, continued the established, late Roman rite of inhumation with few or no grave goods; presumably, these were in large measure burials of the descendants of the Gallo-Roman population, but also perhaps of immigrants who had ‘Romanised’. The purpose of this presentation of archaeological evidence that has been tied to a number of different ethnic groups, as described in the literature, has been to give some indication of how confused the archaeological picture for fifth-century northern Gaul, in the regions that were to become the heartlands of the Frankish kingdom, actually is.

So, as noted above, there is very little in the archaeology which can be firmly labelled ‘Frankish’ or ‘Merovingian’ rather than generically ‘Germanic’, or what is found can be related to other, non-Roman peoples, with some more specific ascriptions (but always to peoples other than the ‘Franks’). It is not until one gets to the horizon represented by the burial of Childeric (d. c. 481) at Tournai (Périn and Kazanski Reference Périn and Kazanski1996) that it becomes possible to talk of an increasingly coherent and distinctive set of burial practices in northern Gaul and to label them ‘Frankish’ if one so wishes. This burial was discovered, possibly under a barrow, in 1653 near the later church of Saint-Brice across the river from the late Roman fortifications. High-quality drawings of the principal contents were published two years later by a local doctor, Chifflet, though the objects themselves were mostly stolen and destroyed in 1831. The drawings show that it was a warrior burial with sword (in an elaborately decorated scabbard), spear and francisca. There was a large quantity of gold-and-garnet jewellery, some of it probably from horse harness, but most of it related to the occupant’s clothing. The presence of a signet ring engraved with the name and image of Childeric identified the occupant of the grave and dates the burial to somewhere in the latter part of the fifth century. The form of the items of jewellery and the use of gold and garnet both link the burial to ‘Germanic’ material culture, but the presence also of an elaborate, openwork, gold crossbow brooch of latest Roman type and a purse of over one hundred gold and more than two hundred silver Roman coins shows another side to the burial. The crossbow brooch can be read, along with the gold-and-garnet dress accessories and the weaponry, as a version of late Roman military apparel, showing the importance of Roman military symbolism and the symbolic language of late Roman dress to Childeric as legitimating his position; indeed, some textual sources have him as some sort of Roman-approved governor as well as Germanic king. This may also be seen in his signet ring, suggesting the need to seal written documents in a late Roman official manner. More recent excavations (Brulet Reference Brulet1996) have reinforced the ‘Germanic’ aspects to the burial, with the discovery in the vicinity of the proposed barrow over the grave of three burials each containing several horses, twenty-one in all, as well as human burials. Horse burials of early medieval date are known from across central and western Europe (cf. Müller-Wille Reference Müller-Wille1996), particularly between the Rhine and the Elbe (Figure 8.5), linked with high-status ‘princely’ or ‘royal’ burials, presumably because of the martial qualities of the horse and probably also because of its links to deities in the Germanic pantheon. The Tournai burial thus can be interpreted as a Germanic royal burial, yet it lies on former imperial lands reinterpreted in a Roman way which would have been comprehensible to the population of Gallo-Roman descent (cf. Halsall Reference Halsall2001). Childeric’s son Clovis, who presumably oversaw the creation of his father’s tomb, was to take the process to the next stage by converting the bulk of his subjects to Christianity. So it is really only from the late fifth-century horizon of the tomb of Childeric that we can begin to see the creation of a material culture in northern Gaul that is sufficiently homogeneous to be labelled ‘Frankish’ (Wieczorek et al. Reference Wieczorek, Périn, von Welck and Menghin1996), one that blended Roman provincial elements with a range of Germanic elements in terms of both material culture and the development of the characteristic Reihengräberfelder (cemeteries with graves in rows), though the Germanic elements continued to share many aspects with areas to which the literary sources ascribe different ‘ethnic’ labels such as Alamanni or Thuringians. The microregional study of the region of Metz (Halsall Reference Halsall1995: esp. Chs. 2 and 8) shows well that what is usually thought of as a Frankish or Merovingian cemetery with its clear structuring of grave goods by gender and by age developed there from the early sixth century, and not earlier (it is also worth noting the prevalence of grave goods in the burials of what was, nominally at least, a Christian population). The early sixth century was also the horizon when the Frankish law code (‘Salic law’) seems first to have been codified and when Frankish kings such as Theudebert (not a Roman-style name) started to strike coin in their own name rather than that of the emperor, suggesting that now they had confidence in their kingdom’s own separate political identity both internally and by contrast with other powers.

Fig. 8.5 Distribution of early medieval horse burials in Europe

The example of the north of Gaul shows that archaeology offers a very different range of evidence from the historical narrative, one which can be read only with great difficulty in terms of some sort of linear ‘ethnogenesis’ of any particular people, least of all the Franks as identified in the texts. From the later fourth century and on through the fifth century, the archaeology, especially of burial, can be read to demonstrate the presence of a variety of different groups, to which ethnic labels may be attached if desired. But this is to read it from a very particular perspective; it can and should also be read along axes such as status, gender, age, family and kin, as much as, or in conjunction with, ethnicity. If we did not have the textual evidence, it would probably be impossible to recognise ‘the Franks’ as a geographically and culturally distinct grouping in the archaeology; rather there would be a generic, ‘Germanic’ background in the burial rites and material, itself imposed on a more extensive Roman provincial rite, against which ‘oddities’ in the material culture or in practices, such as cranial deformation, would stand out. This is in no way to seek to deny that there were people who defined themselves, through their language and customs (habitus), through their religious beliefs, and through their beliefs about the genesis of their people and their ruling family or families, as ‘Franks’ and were recognised as such by contemporaries. Indeed, it might be argued that it was their lack of distinctiveness which was ultimately the key to their success; they had the flexibility to accept and be accepted by other groups, probably particularly the Gallo-Romans, when other groupings more jealous of their own identity died out or were subsumed from the end of the fifth century by the burgeoning Frankish political, cultural, religious and ethnic identity.

On their way to dominance, the Franks were to conquer and subsume the kingdom of the Burgundians, a people with a considerable presence in the historical record and therefore worth brief consideration here. They appear in a ‘first’ kingdom from 406 to 436 on the middle Rhine around Speyer and Worms, until it was suppressed by Aetius. Seven years later, he settled them in their ‘second’ kingdom in east-central Gaul between Lyon and Geneva, the region known as Sapaudia (Savoy), perhaps resuscitating them as a political and presumably military entity to act as a counterweight to the ambitions of the Visigoths. By the end of the fifth century, they had consolidated their hold over much of eastern and south-eastern Gaul, until conquered by the Franks in 534. The historical sources tell us of their kings, queens and other royal and noble personages. The kingdom issued coins, promulgated law codes and was clearly a recognised and recognisable political entity. Archaeologically, it might as well never have existed. From the areas it controlled comes only a handful of objects of ‘Germanic’ type, mainly brooches, some of which are of ‘Alamannic’ type (Escher Reference Escher2005: vol. II). There is nothing there that can be said to be distinctively ‘Burgundian’, as opposed to ‘Alamannic’ or ‘Visigothic’ or, indeed, ‘Frankish’, apart from the coins, where gold was struck in the name of the emperor, and silver and copper in the name of the king (Grierson and Blackburn Reference Grierson and Blackburn1986: 74–7). The region appears as one in which there was a scatter of non-Roman objects and burials, but the majority of the archaeology remains of Roman provincial type. It would seem that though an ethnic self-definition was important to the ruling elite, it was not to the great majority of their subjects. To what extent, then, was the second Burgundian kingdom and its ‘barbarian’ identity one imposed by the Romans (‘etic’) rather than one internally generated (‘emic’) that did not spread far beyond the ruling dynasty and warrior elite, who used it as a tool to legitimate their rule over a population that was overwhelmingly Gallo-Roman?

Discussion: ethnicity, archaeology and history

It is worth briefly drawing out some of the arguments that have appeared in the discussions above of the archaeology of ‘barbarian’ or ‘Germanic’ settlement on the territories of the West in the fifth century and its relationship to historically described ethnic groupings and the origins of the early medieval kingdoms. The Visigoths and the Franks are the two groupings who were ultimately the most ‘successful’ of all the Germanic peoples recorded as taking over the territories of the Western Empire that are our concern here. Each was to create a large, powerful and stable kingdom on either side of the Pyrenees, encompassing essentially the lands of Hispania and Gallia respectively, with royal dynasties ruling over a people of mixed Germanic and Roman provincial descent, with civil and ecclesiastical structures (to a great extent taken over from the empire), taxes, laws (among other things distinguishing between ‘German’ and ‘Roman’ subjects), coinages and the apparatus of a functioning state, even if they had much less actual power than the late Roman emperors had had (cf. Wickham Reference Wickham2005: Ch. 3). Along the way they had defeated and incorporated Germanic and other peoples and kingdoms, sometimes relatively powerful ones such as the Sueves in Spain or the Burgundians in Gaul, as well as absorbing smaller groupings such as the Alans. This, of course, is essentially a historical narrative.

The archaeology presented above can now be seen to be far less clear-cut than used to be thought when its interpretation and use were controlled by the historical narrative. This is above all in the case of the Visigoths, whose archaeological ethnicity in fifth-century Spain was long held to be self-evident; on the other hand, discussion of the archaeological identity of the Franks has long had a tendency to take the burial of Childeric as its starting point and to avoid looking too hard at what came before. It is worth reviewing how this came about and what the archaeology has to contribute to the debate. Clearly, the ‘traditional’ narrative and the identification of the ethnicity of practices and, above all, artefacts were derived from the written texts because of their chronological and intellectual primacy. Archaeology was a johnny-come-lately, in its turn shaped by the pre-existing textual categories. The weaknesses of this approach are now widely recognised and acknowledged. But there are features of how the archaeological debate has developed that perhaps require further consideration, especially the domains of archaeological evidence within which the debate has been conducted. The ‘ethnicity’ debate has principally been sited within the domain of funerary archaeology linked with artefact studies, or, to put it another way, the preceding debate has almost entirely been about burials and grave goods; there are huge areas of the archaeological record not much implicated in or entirely untouched by the ethnicity question. In part this dominance of the funerary record can be attributed to the fact that artefacts tend to be well preserved in graves; in part to the fact that their disposition around the body can be read off in terms of dress, dress that is different from that of the numerically superior indigenous populations. After one has established differences, it has been but a short step to having recourse to the texts to label those differences. We have seen above how slippery such categorisations in fact are once ‘essentialist’ or ‘primordialist’ notions of ethnicity are discarded and replaced by ‘situationalist’ or ‘performative’ notions of ethnicity and other identities. We have also seen that often we are also dealing with strongly gendered representations, with female burials outnumbering male in certain regions such as Spain, suggesting, perhaps, that aspects such as descent and lineage were more important considerations than a simple category of ‘ethnicity’. Leading on from this, it is also worth noting that the identification of ‘ethnic’ identity has often eclipsed other possible readings of the burial evidence such as status, age and, of course, gender. So even within the domain of funerary archaeology, the selection of evidence and approaches has been very partial. We may recall Theuws’ (Reference Theuws2009) disturbing question as to whether ‘ethnicity’ was something that people at the period actually felt was worth inscribing in the burial record. Or has our penchant for reading the funerary evidence in this way simply been a reflection of hugely important and contentious twentieth- and twenty-first-century political and cultural categories?

If we turn away from the study of certain aspects of the funerary record and certain classes of material culture to other domains of the archaeological record, the ‘ethnic’ reading of the evidence becomes decreasingly frequent, often because decreasingly possible. We may turn first to a domain where such identifications have been attempted – settlements and structures. As noted above (p. 379), certain types of structure, principally the ‘hall house’ and the SFB, have traditionally been assimilated to ethnic interpretations; they are the preferred residential and ancillary buildings of Germanic peoples. Now it is perfectly true that the hall building in its archetypal form of the Wohnstallhaus with accommodation for humans and for animals under the one roof does seem to have its origins in the regions of north-west Europe outside the Roman frontiers along the Rhine. It appears in increasing numbers south of the lower Rhine in the fourth century, often associated with the spread of the Franks into Toxandria. It also appears at some sites well to the south of Toxandria in the fourth century (e.g. Saint-Ouen-du-Breuil). But at such sites, though there may be a small amount of non-local (‘Germanic’) artefactual material, the great majority of the material culture is of Roman provincial types, producing a dissonance between any ethnic interpretations based on the structures and those based on the artefacts. In time, essentially from the sixth century on, rectangular, timber structures became widespread across Gaul north of the Loire (cf. Peytremann Reference Peytremann2003: Ch. 5), often grouped into nucleated settlements or ‘villages’. But by this time they are seldom interpreted in ethnic terms but more in terms of changing building technologies, developing structures of rural society and exploitation of agricultural resources. This raises the point that originally such structures in their builders’ homelands outside the empire were not a statement of ethnicity but of particular forms of social structure and agrarian practice with their accompanying ideologies (cf. Roymans Reference Roymans1996a, Reference Roymans1996b). It can be envisaged that as people with their origins in the regions that employed this form of structure settled within northern Gaul, they might have constructed such buildings, but this may have been because it was what they were used to and it expressed their social formations and their agricultural regimes, not because they were thereby making an ethnic statement.

Indeed, the classic north German/Low Countries Wohnstallhaus is rare in northern Gaul; instead it is rectangular, post-built, timber structures (occasionally on sill beams) that predominate from the fourth century on, a form much less distinctive than the Wohnstallhaus and with antecedents in the Roman provincial building stock (Peytremann Reference Peytremann2003: 290–7; Van Ossel Reference Van Ossel1992: Ch. 7). To judge by the activities attested to at such sites, they could fulfil a range of agrarian and artisan functions, so they might best be characterised as general-purpose agricultural structures. It has also been noted that the sunken-featured buildings, long thought to be ‘Germanic’, are in fact of much more equivocal origin since they are not a feature of the regions outside the Rhine frontier, and their origins cannot be tied to proven ‘Germanic’ settlements (cf. Hamerow Reference Hamerow2002: 31–5). Furthermore, when we pass south of the Loire, let alone south of the Pyrenees, these supposed markers of Germanic ethnicity fade out to almost nothing, despite the historical attestations of the long-term presence of peoples of Germanic origin in these regions. We have already remarked upon the lack of ‘Germanic’ material culture in the realms of the kingdom of Toulouse, and this goes also for building types; indeed, it was noted that this region conserves the tradition of elite Gallo-Roman residences, apparently even for the Visigothic court. South of the Pyrenees, the few fifth- to sixth-century settlement sites we have, either installed within former Hispano-Roman villas or on new sites, as we shall see in the next chapter (p. 441), have little or nothing in their plans, building techniques or layout that suggests roots in northern or central Europe. There is hardly any archaeology of fifth-century ‘Visigothic’ structures and settlements. So over most of the area that concerns us and through the fourth and fifth centuries, there is little that we can read as betraying any ethnic identity. Whether that is how such structures were read at the time is even more debatable.

Modern archaeological identification of ‘Germanic’ and other non-Roman peoples, as we said above, has been largely done through burials and the grave goods from them. Certain forms of material culture, principally dress fittings, have been read in terms of proclaiming particular ethnic identities. The problems with doing this have been dilated upon at length above. Here the intention is just to point out that most of the archaeological domain of artefacts and material culture in this period is not susceptible to such interpretations. Let us take the most widespread artefactual material: pottery. Certain types of pottery have traditionally been interpreted as ‘Germanic’, based on their form and decoration, and sometimes their fabric. The classic manifestation of this is vessels that are handmade, in a limited repertoire of forms; this is principally the carinated, biconical ‘urn’, sometimes undecorated but over time produced with an increasingly elaborate range of incised and/or stamped decoration, sometimes with bosses pushed out on the surface. This is utterly unlike the Roman provincial repertoires in terms of technology and form; its fabric and decoration also are distinctive. Its antecedents lie outside the empire in the same regions of western Germany and the Low Countries that were home to the Wohnstallhaus; it is not difficult to see why this has been treated as the ceramic facies of incoming Germanic peoples. Such pottery appears within the empire in the fourth century – for instance, at Saint-Ouen-du-Breuil, where ceramics of this type (interpreted in terms of ethnicity) occur in small quantities alongside much larger quantities of Roman provincial pottery (interpreted in terms of trade). But it is in the fifth century that it becomes more widespread across northern Gaul and more common on individual sites. Many of these sites are cemeteries, and therefore these ceramics have been assimilated to ethnic identifications of the people buried with them and with other ‘barbarian’ material, as in the identification of ‘Anglo-Saxons’ buried on the Gallic coast of the Channel, partly on the basis of the forms and decoration of the vessels in the graves (Soulat Reference Soulat2007). Such interpretations, of course, are subject to the same strictures on the ‘ethnic’ identification of objects as are classes of material such as brooches and belt fittings; objects do not in themselves have an ethnicity. The cemeteries in which these types of ‘Germanic’ pottery occur are inhumation cemeteries, rather than the cremation cemeteries of the Free German homelands (or, indeed, of large areas of fifth-century eastern Britain), so the form of burial corresponds to indigenous Roman provincial practice. Some of the other material in the graves is of Roman provincial manufacture (decreasingly so through the fifth century), but again such material is not ascribed an ethnicity in the analyses of modern workers. Some of the graves discussed by Soulat (Reference Soulat2007) also contain objects he characterises as ‘Merovingian’ (Frankish?) along with the ‘Anglo-Saxon’ material, raising problems for any neat ethnic interpretation. But comparable ceramics are also found on settlement sites, where interpretations are often as much in terms of functionality as of ethnicity; are these vessels for cooking, for the serving of food (communally? individually? and therefore related to social structure and practice?) or for other purposes? Are some of the vessels found in cemeteries specifically funerary, thus making them part of a rather different discourse about funerary practice? Is their increasing dominance through the fifth century not the consequence of any ‘Germanisation’ of material culture, but more a reflection of the breakdown of the large-scale manufacture and distribution of Roman provincial material, especially pottery, and the consequent move to simpler, more local production in styles that modern workers have called ‘Germanic’? Nevertheless, it is undeniable that ceramics of these distinctively non-Roman types and of non-Roman antecedents did become widespread in Gaul north of the Loire during the fifth century (as they did also in eastern Britain). As with the dress accoutrements from Spain traditionally labelled ‘Visigothic’, while it may not be possible to follow such precise ethnic ascriptions, what this does seem to show is that some people in life and in death wished to present themselves in ways that were non-Roman. What this meant in terms of precise ethnicity and whether such people originated from east and north of the Rhine or from among the Roman provincial population are not questions that we can as yet pronounce upon with any degree of certainty.

South of the Loire and on into the Iberian peninsula, ceramics of ‘Germanic’ origin remain extremely rare. We have seen (p. 361) that there is a small quantity from Toulouse, seat of the Visigothic kingdom, through the fifth century. Interestingly, this all seems to be from occupation contexts and not from funerary ones, and this is true across the south-west of Gaul, where fifth-century burials, even those containing objects derived from the Sîntana de Mureş-Chernyakhov culture, have yet to yield any such pottery. The same largely holds true for fifth-century burials to the south of the Pyrenees, where the graves containing ‘Visigothic’ dress items very rarely contain ‘Germanic’ pottery. And, as with building types, so with ceramic types, fifth-century settlements to north and south of the Pyrenees, be they villas in the Roman provincial style or timber structures, were associated with Roman provincial products or local fabrics. As we shall see in the next chapter (p. 405), the fifth-century pottery of these southern regions is discussed essentially in economic terms as an indicator of commerce and of the integration, or not, of areas into wider exchange networks, even for areas that the texts tell us had passed under ‘barbarian’ dominion. Whatever the funerary or other significances of non-Roman forms and decorations of pottery to the north and the east of the Rhine and north of the Danube, these had been completely lost over the time and distance separating these ‘homeland’ regions from southern Gaul and Spain.

Another class of material with a wide distribution across the West throughout our period is glass. The chrono-typology of this material is now much better understood thanks to major recent publications (for an overview, see Foy Reference Foy1995; for a more detailed study of north-eastern Gaul, see Feyeux Reference Feyeux2003). Technologically and typologically, the Roman provincial production centres continued to operate through the third, fourth and fifth centuries. The forms characteristic of fifth-century and later cemeteries in northern Gaul were overwhelmingly concerned with the service and consumption of liquid. These comprise a limited range of flagons, but more especially a series of cup and beaker forms from the relatively simple hemispherical to the more elaborately decorated types, most famously ‘claw beakers’ and glass imitations of drinking horns. Again because of their presence in graves that have been labelled ‘Germanic’ in view of the other material culture in them, especially items of dress, these have been thought of as expressions of Germanic practice; indeed, the more inventive forms, in particular the ‘claw beakers’, have been seen as expressions of a ‘Germanic’ aesthetic as opposed to the more restrained canon of Roman provincial glass forms. In fact, the ‘claw beaker’ (and related forms) clearly developed out of and elaborated late Roman products; if one looks also at the volumes these vessels would have held, especially the flagons, they would seem more appropriate to the consumption of wine rather than any imagined Germanic beer drinking. And again, this ethnic association of late glass forms has always been a feature of the analysis of northern Gaul, with its cemeteries with a range of offerings, rather than of southern Gaul or Spain, where glass is rare in burials but overwhelmingly a find from domestic sites, where it is seen in functional terms with no ethnic overtones.

Currently, there are powerful new methods of physical and chemical analysis of human (and animal) bone being refined and becoming more widely available, and these should, in time, allow a whole new series of data sets to be interrogated. These, of course, are techniques involving the extraction and characterisation of such things as ancient DNA (including mitochondrial DNA), blood types and stable isotopes. Currently, the most promising of these seems to be stable isotope analysis. For our purposes here, there are two caveats that need to be entered. Suppose the analysis of a later fifth-century skeleton from a grave on the Spanish Meseta showed that the individual had spent his childhood in the Pontico-Danubian region. This would be of considerable interest but would not in itself say anything about his particular ethnicity, be it Alan, Goth, Hun, Suevic, Vandal or other, since all of these are recorded by historians as having been present at one time or another in this region. Given that the texts tell us that the Visigoths entered Spain from south-west Gaul, it would be very hard to argue that the individual was by birth a Visigoth, though, of course, he could have redefined himself as such at some stage in his life. There is also the ‘second-generation problem’ – the offspring of putative incomers to Spain from eastern Europe would have grown up in central Spain, and, if buried in the same cemetery as their parents, would have an entirely ‘Spanish’ isotopic signature, whatever they or others may have considered their ethnic identity to have been, and which might have been pronouncedly different from that of the indigenous population (an obvious modern parallel is that the children of Pakistani immigrants to Birmingham would have an English isotopic suite but be in several important ways culturally distinct from the majority English population).

Let us take a cautionary tale from an area within the compass of this book. Excavations in the 1960s on the fourth-century cemetery at Lankhills, Winchester, identified a number of burials whose rite differed from that of the Romano-British population and which the excavator identified as having originated in the area of Pannonia, on the middle Danube (Clarke Reference Clarke1979: Ch. 4, Sect. 2). Publication of work on a further tranche of the cemetery incorporated a programme of isotopic analysis on both ‘local’ and ‘intrusive’ burials (Booth et al. Reference Booth, Simmonds, Boyle, Clough, Cool and Poore2010: Ch. 5). In brief, there were individuals with local isotopic signatures and individuals with non-local ones. Of the groups defined by burial rite, some of the ‘local’ burials had non-local isotopic signatures, and others had local signatures: some of the ‘intrusive’ burials had local isotopic signatures, some had non-local ones. But the non-local ones suggested origins either in other parts of Britain, or, if from mainland Europe, the signatures suggested origins from the area of the western Mediterranean basin, not Pannonia. Only one burial had an isotopic signature suggesting an origin in central Europe, and the burial rite of the grave was not of the classic ‘intrusive’ type. Clearly, there was a significant mismatch between the groupings identified through burial rite and those identified through stable isotope analysis. It is only too likely that similar results will be replicated elsewhere, so this approach will not be a ‘magic bullet’ that will clarify the situation; more likely, it will add a further set of variables to analysis of the funerary archaeology.

‘Barbarians’ is a topic where there is a wide divergence between the textual and the archaeological evidence. The texts are specific about named peoples, named areas and named dates. The archaeology simply does not see these peoples in the way historians, either at the time or in later analyses, do. This is not to say that the events the ancient sources describe did not happen; the events, or at least something approximating to them, probably did happen, albeit that the sources probably greatly simplify much more complex chains of events, selecting among them for their particular moralising and other agendas. This is, though, to say that the material culture and other correlates, which have for so long been taken as more or less self-evident expressions of the presence of peoples of named ‘barbarian’ origins at certain times and in stated places, simply are not susceptible to being read in this way. In some areas at some times, we find building types and material culture types that are clearly not derived from Roman provincial practice and in most cases can be linked to antecedents either across the mid and lower Rhine or in the ‘Pontico-Danubian’ area. Clearly, these represent in some way the appearance on Roman territory of individuals or groups who defined themselves in some important ways through signifiers of ‘non-Romanness’. This, of course, is an essentially negative definition and thus of limited use. It would be an exercise in hypercriticality to claim that such material culture and other aspects of the archaeology were not in some way making statements about particular forms of non-Romanness, and about particular identities proclaimed especially through dress (and probably other forms of bodily presentation such as hairstyles). These identities were in part to do with matters such as gender, status and age. But alongside this, there were very probably also claims to particular ‘ethnic’ identities. But as we have seen, such identities were not fixed and immutable; they were changeable according to situation and utility. Likewise, the archaeological material does not persistently recur in limited and bounded sets of associations of a type that would support the identification of particular ethnic identities in an old ‘culture-historical’ way; this material, too, is fluid and mutable, suggesting that the material correlates of these identities could be adapted, perhaps to suit the preconceptions of others. Rather than lamenting the loss of the old certainties, we should perhaps see this as an opportunity to refashion our perceptions of what was happening across the Western Empire in this period at both the macrolevel and the microlevel of the individual cemetery or settlement and at the subregional level, particularly in the northern part of Gaul, where ‘non-Roman’ signifiers both in life and in death were clearly important. But equally there remain large areas of the West, particularly in southern Gaul and Iberia, where the historical sources attest to the long-term presence of ‘barbarian’ peoples, probably in significant numbers, but they remain hard to pin down in the archaeology; clearly, it was possible to be a ‘barbarian’ while living in Roman-style buildings and using Roman-style material culture. Presumably, there were signifiers such as elements of bodily presentation or perhaps more especially language, nomenclature and a consciousness of belonging to a particular community defined by political loyalties and shared ideas, such as origin myths and lineages, or, indeed, particular types (‘heresies’) of Christianity, but these are signifiers that leave little or, more often no, trace that we are able to recognise in the archaeology.