PROBABLY LATE IN THE summer of 376, the majority of the Gothic Tervingi, the Empire’s main clients on the Lower Danube frontier for most of the fourth century, turned up on the northern banks of the river asking for asylum. They were led by Alavivus and Fritigern, who had broken away from the confederation’s overall ruler Athanaric. The equally Gothic Greuthungi, who had previously lived further from the frontier, east of the River Dniester, soon followed them. Both Tervingi and Greuthungi had been established south and east of the Carpathian Mountains for at least three generations, so it is not surprising that their sudden displacement towards the Danube was associated with a broader wave of regional unrest. After some thought, the east Roman Emperor Valens decided to admit the Tervingi into the Empire, offering them assistance across the Danube, but to exclude the Greuthungi. The latter, however, soon found an opportunity to cross the river without help or permission, and were quickly joined by other uninvited guests: Taifali plus some Huns and Alans in 377, more Alans in 378, and some of Rome’s Middle Danubian Sarmatian clients in 379. Long-established inner clients like the Tervingi, Taifali and Sarmatians, outer clients such as the Greuthungi and Alans, and previously unknown Hunnic intruders were battling it out for control of the zone north of Rome’s east European frontier, and the struggle had spilled over on to imperial territory.
About a generation after 376, the established order beyond Rome’s central European frontier – the Middle Danube basin west of the Carpathians – suffered an equally spectacular collapse. There were probably many smaller-scale participants as well, but four major groupings of barbarians figured in the action. A largely Gothic group, first of all, led by a certain Radagaisus, crossed the Alps into Italy in 405/6. These were followed at the end of 406 by a mixed force of Vandals, Alans and Sueves, who crossed the Rhine into Gaul and cut a swathe of destruction through to Spain. Shortly afterwards, a mixed force of Huns and Sciri crossed into the east Roman Balkans, capturing the fortress of Castra Martis in the province of Dacia. Finally, Burgundians elbowed their way past their western neighbours, the Alamanni, to establish themselves on and over the River Rhine around Speyer and Worms. We don’t know when the Burgundians did this, exactly, but it was sometime between 406 and 413. In fourth-century terms, this again represented a mixture of established frontier clients (Sueves), groups who were occasionally part of Rome’s diplomatic web (Burgundians and Vandals), and complete outsiders to the Middle Danubian region (Alans).1
Nor, from a Roman perspective, was this sequential collapse of its eastern and central European frontiers the end of the misery. The Tervingi and Greuthungi who crossed the Danube in 376 had eventually made a kind of peace with the Roman state in 382, after six years of warfare which, famously, had seen them destroy the Emperor Valens and two-thirds of his field army on 9 August 378. Some of them – how many is a question we must return to – from 395 gathered round the leadership of Alaric and his successors. This force moved first around the Balkans, then into Italy – twice – and finally on to Gaul, where another agreement rooted them more firmly this time, in Aquitaine, from 418. From this settlement eventually emerged the Visigothic kingdom: a first-generation successor state to the western Roman Empire. A similar capacity for continued movement was shown by some of the groups bound up in the central European frontier collapse. Most famously, some of the Vandals and Alans who had ended up in Spain from 409 took ship, twenty years later, for North Africa, where they too eventually established an independent kingdom. And in the meantime the Burgundians too moved on, if in less dramatic fashion. After a heavy defeat at the hands of the Huns, many were resettled by the Roman state around Lake Geneva in the later 430s. From this settlement eventually emerged a third successor state to the old Roman west.
Some of the distances here are extraordinary. The extended trek of the Tervingi and Greuthungi from the north-west corner of the Black Sea to Aquitaine totalled about two and a half thousand kilometres, even just as the crow flies (and as the Goths didn’t). The Vandals went from Slovakia or thereabouts to Tunisia, via Spain and Morocco, not far short of four thousand kilometres, and the Alans who accompanied them even further. Before 376, the River Don marked the western boundary of Alanic territory north of the Black Sea, and from there to Carthage it was a – perhaps literally – staggering five thousand kilometres.
In traditional accounts of the first millennium, these tumultuous events on Rome’s European frontiers and beyond were heralded as the beginning of the great Germanic Völkerwanderung: literally, ‘the movement of peoples’ (even if not all of those involved were Germanic-speakers). The Goths, Vandals, Burgundians and many others who feature in the two chapters that follow were thought of as complete populations of both genders and all ages who had long-standing group identities and deliberately moved in compact groups from one piece of territory to the next. In the process, they destroyed the power of the Roman state in western Europe, and in some accounts of the action this represented the dénouement of a struggle that had begun as long ago as 9 AD when Arminius’ coalition destroyed Varus and his three legions in the Teutoburger Wald. And if this were not a big enough story, the events associated with Roman frontier collapse had, as we have seen, a still bigger role to play in understandings of the creation of Europe. The model they seemed to provide – of entire peoples on the move – was applied wholesale to European pre-history, which was all explained in terms of migration, invasion and ‘ethnic cleansing’. The frontier intrusions of the late Roman period thus provide a crucial test case. Were they undertaken by large population aggregates, mixed in age and gender, or were they not?
Several contemporary sources mention the arrival of the Goths on the Danube in 376. All share the same basic view that its ultimate cause was the emergence of a new force on the fringes of Europe: the mysterious Huns (of whom more in a moment). One even puts a figure on the number of refugee Goths gathered on the riverbank: two hundred thousand people of all ages and both genders. Fundamentally, though, our understanding of what was happening depends on one Roman historian in particular: Ammianus Marcellinus. Only Ammianus provides any circumstantial detail at all about the Goths’ defeats and subsequent departure for the Roman frontier. He and he alone, for instance, tells us that at one point there were three separate concentrations of Goths on the Danube’s banks, and that non-Gothic groups were involved in the action too. Likewise, it is only Ammianus’ account that explains how the Greuthungi made their decision to move after the deaths of two kings and how the confederation of the Tervingi split as different factions advocated and won support for alternative responses to the Hunnic menace. Beyond these details, Ammianus, like the other sources, is entirely explicit on two points. First, the Goths came to the river in very large numbers. He never gives a total figure (in fact he says there were too many to count), but he does record that the Emperor Valens gave battle at Hadrianople on the intelligence that he was facing ten thousand opponents, which he understood to represent only part of the total Gothic military force loose in the Balkans at that point. Second, these warriors had come with their wives and children.2
No late Roman commentator ever sat down to draw up a precise description of any migrant group of barbarians – stating that eight out of ten migrant males, say, came with their families – but Ammianus clearly understood the action as being driven by armed migrant males moving with families, their belongings carried in a wagon train, which features at several points as a mobile fortress that could be pulled (like that of the Boers) into a defensive laager, and must have been of enormous size. As noted earlier, historians have often used a multiplier of 5:1 for the ratio of total population to warriors, but that is a guess. But whatever ratio you choose, up to twenty thousand warriors or perhaps even more, plus their families, has to mean many tens of thousands of people in total on the move. And while making it quite clear that not every migrant belonged to one of the major Gothic concentrations, Ammianus does report a striking degree of political coherence among the two main groups of Goths – Tervingi and Greuthungi – who crossed into the Empire in 376. They each negotiated as a body with the Roman state from the banks of the Danube, and continued to act together, for the most part, afterwards.
If we take these main features of Ammianus’ report for 376 together – the Gothic groups’ mixed gender/age makeup, the fact that we are dealing with several tens of thousands of people, that they were on the run from the Huns, and the coherent way in which the immigrants dealt with the Roman state – then you can see what makes modern commentators hesitate. It all adds up to something that looks worryingly like the old invasion hypothesis: one people, one leadership, and one clearly directed move or set of moves with invasion and flight playing a major role. We have seen too that this kind of phenomenon – different again from the flows of predatory migration of the third century and the Viking period – is strikingly absent from modern, better-documented case studies of migration. In the face of both of these problems, can we believe the picture drawn with such clarity by Ammianus?
Establishing the credibility of an ancient historian operating in the classical tradition is never straightforward. Back then history was a branch of rhetoric, and although it aimed at truthfulness, truth did not have to be merely literal. A high degree of artistry was expected, partly for the audience’s entertainment, but this might again be harnessed in the service of bringing out a deeper truth about persons or situations. What we know about Ammianus in particular is deeply intriguing. He closed his History with a memorable and essentially accurate, if limited, one-line self-description: ‘a soldier once and a Greek’ (miles quondam et Graecus). He was born in Antioch in the largely Greek-speaking eastern Empire, and clearly received an excellent education in Greek and Latin language and literature before entering the army, where he rose to mid-staff officer rank as a general’s aide. He faced battle many times and undertook secret missions – behind Persian lines on one occasion and to assassinate a usurper on another – but, as far as we can tell, he never commanded a unit in action. He left the army in the mid-360s on the death of the last pagan Roman emperor, Julian the Apostate, and was himself a non-Christian. Otherwise he doesn’t tell us much about himself or his purposes in writing history, except to mention in passing a few places that he’d visited between leaving the army and eventually moving to Rome in the late 380s, where his History was brought to completion in the early 390s.
There is a huge and growing literature on the historian and his work, from which two points emerge clearly. First, while claiming to be interested in the truth, Ammianus was not averse to deploying literary artistry in the service of what he considered to be true, and sometimes even evasion. The big cultural story unfolding around him in his own lifetime was the progressive Christianization of the Empire, but he deliberately minimized its appearance in his text, and may even have attempted to conceal a personal aversion to it in the guise of favouring religious toleration. And what is true of his treatment of religion may be equally true of his treatment of other matters, where a lack of candour is less obvious.3 But all that said, Gibbon regarded Ammianus as a ‘most faithful guide’. Gibbon was no fool, and the second point reinforces the quality of this judgement. By an extremely wide margin, Ammianus provides the most detailed and informative narrative to survive from the late Roman period (or pretty much any other Roman period, for that matter). What we’ve already seen of his Gothic narrative is true at many other points, as well: the level of circumstantial detail included in his text simply overwhelms, where they overlap, other sources of surviving information. This vast body of knowledge was acquired partly from his own experience (his secret missions get extensive and entertaining coverage, for instance, and Ammianus was also on Julian’s failed Persian campaign), partly from talking to informed participants such as the retired palace eunuch Eucherius, but also from consulting documentary archives. He refers at one point to a ‘more secret’ archive he wasn’t allowed a glimpse of, which makes it plain that there were others that he did see, and at another he lets slip that it was his normal practice to look up the official records of their careers when writing about military functionaries. A French historian has also successfully demonstrated the substantial extent to which Ammianus’ narratives are constructed on his reading of the original dispatches that had gone back and forth between Roman generals and their subordinate commanders.4 In other words, alongside literary artistry and calculated evasion, you have to reckon with Ammianus having engaged in something analogous to modern historical research, without which the degree of detail in his narrative would have been impossible. No simple blanket answer to the question of Ammianus’ reliability is possible, therefore, and passages have to be considered case by case.
In relation to the events of 376, Ammianus’ credibility has recently been attacked on two counts, one profound, the other only slightly less so. Most important, it has been suggested that his account of the events of 376 looks a bit like the old invasion hypothesis because he (and the other authors who write in less detail) couldn’t help but portray the action in that fashion. It was so ingrained in classically educated authors that ‘barbarians’ moved as ‘peoples’ – interrelated ‘communities of descent’ – that they automatically wrote up any example of outsiders on the march on Roman soil along these lines. Deeply ingrained in their heads, in other words, was a migration topos, which made it impossible for them to give an accurate characterization of barbarians on the move. Second, it has been argued that Ammianus’ emphasis on the Huns as the root cause of the Goths’ arrival on the Danube is misplaced. It was in fact Roman action that had destabilized the Gothic client world, allowing the Huns to move into new territories, so that the latter were not quite the ferocious outside invaders that our sources portray.5 These are important critiques, but are they convincing? Has Ammianus misunderstood the significance of the Huns’ role, and did he describe the events of 376 as a mass movement of men, women and children because he lacked the conceptual machinery to do otherwise?
Sometimes, as we have seen, our sources do give good reason for thinking that a migration topos was in operation in their authors’ heads. The sixth-century Jordanes describes third-century Gothic migrations into the Black Sea region as one ‘people’ on the move, when the reality portrayed in more contemporary sources was much more complex. In due course we will encounter another excellent example in ninth-century accounts of fourth- and fifth-century Lombard migrations. But what about Ammianus on the events of 376?
In this case, falling back on the migration topos argument looks deeply unconvincing. To start with, though this is just a footnote, it is entirely unclear to me that Ammianus does envisage either the Tervingi or the Greuthungi as ‘peoples’ in the sense of ‘communities of descent’ in some ideologically reflexive manner. In fact, he does not analyse them at all. What interests him, and this is generally true of ‘imperial’ accounts of barbarians, is the power of these groupings as military and political collectives, and hence any threat they might pose to Roman security. How they worked in detail was not his concern. The Tervingi surely were not a ‘people’ in the classic sense of the word: a closed biologically self-reproducing group whose members all shared pretty much equally in a distinct cultural identity. Social differentiation already existed in the Germanic world at the start of the Roman period, and had grown apace over the subsequent three centuries (Chapter 2). All the Germanic groups of the late Roman period that we know anything about went into battle with two hierarchically arranged groups of fighters, whose investment in their group identities was substantially different. These groups quite probably also incorporated slaves, who were not allowed to fight. That Ammianus does not explore any of this certainly limits our capacity to understand the Tervingi, but that is not the same as saying that he had one simple model for all groups of outsiders on the move. In fact – and this is much the more important point – it emerges clearly from his History that he was perfectly capable of differentiating between different types of mobile barbarian.
In different chapters of his History, for instance, we meet barbarian warbands on Roman soil, engaging in their usual pastime of wealth collection. These groups are always identified as such, their numbers sometimes given in the few hundreds, and Ammianus clearly had no problem in telling a warband from a large mixed body of population. This is perhaps not surprising, given the huge difference in scale between a warband and the Gothic forces in action in 376. For that reason, his account of the battle of Strasbourg, which we encountered in Chapter 2, is still more pertinent. This involved, in Ammianus’ view at least, over thirty thousand Alamanni and their allies, gathered under the leadership of Chnodomarius – and all of them on Roman soil. Despite the size of the opposing force, Ammianus is perfectly clear that this was a military action that had the continued annexation of Roman territory in mind and was undertaken only by males. He also distinguishes clearly the range of recruitment methods that had been used to gather Chnodomarius’ army. Many were the followers of various Alamannic kings present at the battle, but some had overthrown their king to be present, and others were mercenaries hired for the occasion.6 Thus although the action involved very large numbers of barbarians, Ammianus did not suffer from any ‘barbarian army equals people on the move’ reflex.
The point is reinforced if you look more closely at his account of what was going on north of the Danube around the time of the Goths’ appearance on the frontier. Not even all the outsiders who crossed the Danube in the run-up to Hadrianople, for instance, are presented as on the move with families. In the autumn of 377, the Goths found themselves in a difficult situation, trapped in the northern Balkans with food supplies running out. To help lever out the Roman garrisons who were holding the passes of the Haemus Mountains against them, the Goths recruited the help of a mixed force of Huns and Alans, promising them a large amount of booty. The stratagem did the trick. The point here is that, first, Ammianus can identify a force as politically mixed (not a ‘people’) when it was so – this one composed of Huns and Alans – and, second, that although they were mobile barbarians on Roman soil, he made no reference to women and children. For Ammianus, this was merely a mixed mercenary warrior band useful to the Goths in a tight situation.7
Indeed, he does not describe even the Tervingi as a whole ‘people’ moving in untroubled fashion from their old homelands on to the Roman frontier. The Tervingi arrived on the Danube in two separate concentrations in 376 because there had been a split among them. The larger group led by Alavivus and Fritigern was composed of those who had decided to reject the leadership of Athanaric, from the established ruling house, and seek asylum inside the Roman Empire. A second and smaller, though still quite substantial, group later followed them to the river under the command of the old leadership. There, having initially thought of seeking asylum too, Athanaric took an alternative option. Ammianus explicitly describes the Tervingi as a political confederation in crisis, not a ‘people’ taking seamlessly to the road.8 The range of different types even of very large barbarian forces that he was able to describe, and the details of his account of the Goths in crisis, both lead to the same conclusion. Our Greek soldier was both sophisticated enough and well enough informed to describe events on the Danube specifically and accurately. When he tells us that concentrations of Goths came in large numbers, and with their families, this does not look remotely like a cultural topos. In other parts of his History, he described even very large barbarian groups on the move on Roman soil in quite different ways. He chose to present the events of 376 in the way he did quite deliberately, and not because it was the only model in his head. This much now, it seems, is more or less accepted. Even among scholars generally rather suspicious of large-scale migration, only one has tried to discount Ammianus’ account of the numbers involved, and that by asserting the existence of a general migration topos rather than by any more detailed argument.9 On balance, it is highly probable, therefore, that Ammianus knew what he was talking about.
As for the other line of attack on Ammianus’ credibility – the emphasis he placed on the Huns as the first cause of these population displacements – this derives from a report in the Church History of one Socrates Scholasticus that Athanaric’s confederation split not in 376 in the face of Hunnic attack, but immediately after Valens’ earlier war against the Tervingi which ended in 369. It was after this, according to Socrates, that Fritigern broke with the leadership of Athanaric. On the basis of this, Guy Halsall has recently argued that Valens, not the Huns, was ultimately responsible for the arrival of the Goths on the Danube, in the sense that Valens’ campaigns inflicted defeats on Athanaric and the Greuthungi, thereby destabilizing Rome’s Lower Danubian client states. It was this dislocation that allowed the Huns to move into Gothic territory, and accepting this point undermines the traditional picture of the Huns as outsiders of enormous military power whose migratory intrusion destroyed an existing political order north of the Black Sea.10
Obviously enough, Socrates’ report cannot be seamlessly folded into Ammianus’ picture. The two historians have completely different understandings of when and how the confederation of the Tervingi split. And this, in the end, is the fundamental problem with Halsall’s line of argument. The title of Socrates’ work is accurate, in that most of his work concerns itself with the development of the Christian Church. Only occasionally and tangentially do other events intrude, and then never in very much detail, so that Socrates’ overall knowledge of the fourth-century Goths is much less than that of Ammianus. Furthermore, Socrates was writing in Constantinople in the mid-fifth century, so was not contemporary with the events he was describing. When it comes to politics and military matters, it would be unsound methodologically to correct the contemporary and very specific account of Ammianus on the strength of an isolated report by Socrates, unless there was some compelling reason to do so – which there is not. And in fact, while it is easy on closer inspection to understand Socrates’ as a confused version of Ammianus’ account of Gotho-Roman relations (some of the events are in the wrong order), the opposite is not true, since Ammianus includes much extra material that is not in Socrates’ text. Valens’ war against Athanaric, it is also worth noting, had ended in a stalemate that would arguably have strengthened the Gothic leader’s prestige, since he was invited to a summit meeting on the river with the Emperor and treated with great respect. The conflict would certainly have had much less of a destabilizing effect north of the Danube than the Emperor Constantine’s total victory over the Tervingi in the early 330s, when no Huns appeared.11 So neither of the critiques of Ammianus’ credibility are convincing, and we can reasonably proceed from the premise that large, mixed population groups of Goths were set on the move in the summer of 376 by the aggression of Hunnic outsiders.
That being so, how are we to relate the migratory phenomena Ammianus described to patterns of mass human movement observed in the more modern world? In one sense, the scale and character of the migration flows of 376 are not out of step with modern case studies. For, as Ammianus and all our sources unanimously report, the underlying cause of the Goths’ move to the river was political and negative. The Huns were undermining the stability of the entire north Pontic region, and the Goths were looking to remove themselves to a safer locale. As Ammianus puts it:
[The Goths] thought that Thrace offered them a convenient refuge, for two reasons: both because it has a very fertile soil, and because it is separated by the mighty flood of the Danube from the fields that were already exposed to the thunderbolts of a foreign war.12
In Ammianus’ formulation, the Goths had two motives in mind: the attractions of Roman territory and a desire to escape the insecurity of life north of the Danube.
Taking the second motive first, it is, of course, politically generated migration – in other words, fear – that characteristically sets large, mixed groups of human beings on the move: 250,000 in one month of 1994 in Rwanda, and over a million in another. Given its strongly political motivation, the scale of Gothic migration in 376 is not a problem. Where the action does depart from modern analogies, however, is in the degree of organization shown by at least the three major concentrations of Goths. This is not to deny – quite the opposite, in fact – that much human flotsam and jetsam was at large north of the Danube, but in the midst of it all, the Romans were faced with three fairly coherent groupings: both parts of the now split Tervingi, and the Greuthungi. This is quite different from all modern analogies. Whether one is talking central Europe at the end of the Second World War or Rwanda and Kosovo more recently, floods of political refugees have been precisely that: many unorganized streams of people running for their lives. If the migrants then found themselves in camps, leadership structures and organization have sometimes emerged, but the modern world has never thrown up an example of the kind of ordered evacuation described by Ammianus. Should we believe him?
Again, I think broadly that we should. The observable contrasts between the events of 376 and modern mass migrations do make sense in the light of some of the basic differences in context. Part of the explanation for the oddity of the action, for instance, lies in the nature of the Hunnic threat facing the Goths in 376. The Goths have generally been portrayed in modern accounts as panic-stricken refugees desperately fleeing masses of Huns who were hot on their trail. The primary authorities provide plenty of justification for this view, since they surround the Goths’ arrival on the Danube with an aura of panic and defeat. The historian Zosimus can stand for many others:
By wheeling, charging, retreating in good time and shooting from their horses, [the Huns] wrought immense slaughter. By doing this continually, they reduced the [Goths] to such a plight that the survivors left their homes which they surrendered to the Huns, and fleeing to the far bank of the Danube begged to be received by the emperor.13
Narrative details preserved by Ammianus, however, suggest a significantly different picture. The Huns first attacked the Alans, Iranian-speaking nomads, who lived to the east of the Goths beyond the River Don. Having joined some of the Alans to themselves, they then attacked the Greuthungi. After a considerable struggle and the death in battle of two Greuthungi leaders – Ermenaric and Vithimer – the group decided to retreat westwards. This brought them into the territory of the Tervingi confederation. Its leader Athanaric advanced to the River Dniester, alarmed no doubt in equal measure by reports of the Huns and the fact that a large body of alien Goths was now camped on his borders. A surprise Hunnic raid then forced him back towards the Carpathians, where he attempted to create a defensive line to protect his domains. From Ammianus’ geographically elusive description it is possible to deduce that this may have been improvised out of an abandoned line of Roman fortifications that had been used to protect old Roman Dacia north of the Danube, the limes transalutanus. But more Hunnic raids undermined the collective confidence of the Tervingi in Athanaric’s leadership, and caused the ‘majority’ of them both to abandon him and to seek refuge inside the Roman Empire. They were joined in this enterprise by the still retreating Greuthungi, who seem to have adopted the asylum idea from the Tervingi.14
How long had this all taken to unfold? The Huns’ attack on the Goths is usually written up as ‘sudden’, and, implicitly or explicitly, the events compressed into a timeframe of little more than a year. But some of the narrative details suggest otherwise. Of the two kings of the Greuthungi, Ermenaric resisted the Huns for ‘a long time’ (diu), and Vithimer ‘for some time’ (aliquantisper), a resistance which included ‘many engagements’ (multas clades). These are indefinite chronological indicators, but a ‘long’ resistance is more likely to be measured in years than months. Moreover, the Huns were still not breathing down the Goths’ necks even when the latter reached the Danube. They were able to sit patiently by the river, while an embassy was sent to the Emperor Valens to transmit the request for asylum in person. But Valens was about fifteen hundred kilometres away in Antioch, and, travelling by land, the embassy will have taken well over a month. None of this suggests that Huns were present in large numbers close to the Danube in 376, even if the Tervingi had just suffered from two substantial raids at their hands.
The point finds general confirmation in subsequent events, which show that many Huns were still operating well to the north-east of the Black Sea as late as 400 AD. Most modern reconstructions have tended to picture them sweeping as far west as the Carpathians and even beyond, in 376 or immediately afterwards. In 395, however, when the Huns mounted a huge raid on the Roman Empire, their first on anything like such a scale, they went through the Caucasus Mountains, not across the Danube. This has been seen as a cunning plan, with the Huns dragging their horses thousands of kilometres around the northern shores of the Black Sea from Danubian bases – but this is absurd. Horses and men would have been exhausted even before the attack began. What this raid really shows is that, as late as 395, most of the Huns were still well to the east of the Carpathians, perhaps located in the region between the Volga and the Don (Map 7). This is confirmed by other reliable evidence, namely that more Goths (other than the Tervingi and Greuthungi of 376) and other non-Huns provided the main opposition to the Roman Empire across its Lower Danube frontier certainly as late as 386, ten years after the initial Gothic emigration, and quite probably beyond.15 Although the Huns certainly started the revolution north of the Black Sea which manifested itself in the arrival of the Goths on the Danube in 376, they did not themselves come so far west in large numbers at that point. In other words, the Tervingi were not facing an immediate deluge of Hunnic arrows and did have the opportunity to make a more measured response to the mayhem unfolding around them than is generally envisaged.16
But if the Tervingi had the time to organize the kind of orderly evacuation Ammianus describes, is it plausible to suppose that they did? This would imply that they possessed a decision-making body of sufficient strength and coherence to formulate and push through such a plan, raising related issues about political capacity and about the strength of their group identity. That the leadership of the Tervingi could formulate ‘big’ decisions is clear enough from other evidence. As we saw in Chapter 2, the confederation managed to sustain coherent policies towards the Roman state, and, in particular, with regard to the degree of subjection that they, as clients, were willing to tolerate. This even stretched to the ambitious policy of organizing the persecution of Gothic Christians, because the new religion was associated with the Empire’s cultural domination. There is nothing implausible per se, then, in the idea that the Tervingi might have had sufficient strength of identity to respond as a group to the new threat posed by the Huns.
How exactly these decisions were taken, and by whom, depends on the spread of social power in Gothic society at this date. In particular, the degree of social stratification and the extent of the ‘gaps’ between strata would dictate who was involved – and in what ways – in the decision-making process. At the top of the social scale, leaders such as Athanaric, Alavivus and Fritigern – called ‘judges’ and ‘kings’ in our texts – would have been actively advocating particular policies, but, as emerged in Chapter 2, a broader (freeman?) group would have enjoyed some kind of collective veto on suggestions made by their superiors, and hence would have played at least a passive part in the process. Elements of Ammianus’ narrative do indicate that this was so. The discussion about the decision to enter the Roman Empire was drawn out. Ammianus’ comment is diuque deliberans: they were ‘considering for a long time’. And I strongly suspect it was a heated exchange, too. Likewise, once south of the Danube, the new leadership of the Tervingi is repeatedly found ‘urging’ and ‘persuading’ its rank and file towards specific lines of policy, not simply issuing orders.17
This does not mean, of course, that the entire population of the zones dominated by the Tervingi was involved in the decision-making. The archaeological remains and historical sources both tell us that this was a culturally complex world. It had been created by the military power of Germanic-speaking immigrants, who remained its dominant force. But, despite the evacuations of the Carpi on to Roman soil around the year 300, substantial elements of the old indigenous populations – Dacian-speakers, Sarmatians and others – remained in place under Gothic domination. The hardest question of all to answer, in fact, is what was the relationship between the incoming Germanic-speaking elites brought there by the migration processes of the third century and the residual indigenous population? Largely because you cannot easily tell them apart in the archaeological evidence, the current assumption seems to be that the two groups quickly mingled in sociopolitical terms as well as geographically. But this is neither a necessary assumption, nor even likely. Given that identity is fundamentally subjective, located internally in the self-consciousness of individuals and their relationships with one another, then material cultural similarities are neither here nor there. The idea that material culture might reflect group identity has found some support from comparative studies, but all the reported cases have involved a specific item or two ascribed symbolic significance, not broad regional assemblages of artefacts. And to know for certain which particular items are significant, you need precise ethnographic information.18 The fact that the remains of the Cernjachov system are broadly similar right across the board does not mean that there were not distinct group identities within it.
It is extremely important, moreover, not to forget the general historical context. The Goths and other third-century Germanic immigrants into the Black Sea region won their place by right of conquest, and had come to enjoy the riches of the frontier zone. Given that background, it is unlikely that differences in identity between themselves and those they subdued would have broken down quickly, even if there weren’t the same differences in physical characteristics that helped keep Boers and their new neighbours apart in an analogous situation after the Great Trek. Germanic identity, because of the conquest, meant higher status, and letting indigenous groups across that status divide potentially threatened the immigrant’s privileged position. We are, in short, looking at a quasi-colonial context, where the intrusive elite had real reason to protect their privileges against indigenous groups who might wish to erode them. That the fourth-century Gothic world did indeed operate in this fashion is suggested by the way in which Roman prisoners captured in the previous century seem to have been treated. From among their number came Ulfila, and a Christian Church was clearly allowed to operate amongst the prisoners’ descendants over several generations. When Ulfila was expelled from Gothic territory in 347/8, furthermore, many of these descendants went with him, implying very strongly that they formed a distinct, and presumably inferior (or they would not have left), community within the Gothic realm.19 This kind of subjugated autonomy is found, as we shall see in the next chapter, in other complex barbarian state formations of this era.
This is not to say that no indigenous individuals or even groups of individuals managed the leap to a more integrated higher status among the incoming Goths. The need to recruit military manpower might well have led to some alliances that were more equal, like that made between the Goths and some Huns and Alans on Roman soil in 377. It is also possible that some indigenous groups would have been allowed access to the intermediate status of lower-grade fighter (freedmen?), where individuals were allowed to fight, and had considerable advantages over slaves, while being nonetheless personally dependent upon particular freemen. Overall, however, since identity was linked to status, integration could never have been automatic.
Thinking about the events of 376 in this light, Tervingi decision-making would certainly have involved the freeman class, since the advocates of particular policies needed to win its support. The evacuation presumably encompassed both freemen and freedmen, since, between them, these social classes provided the military capacity of the group, and lower-grade warriors are encountered in other Gothic groups on the march.20 Even so, this would still leave many indigenous groups on the outside, I suspect, who were involved in neither process, and such, it seems, is suggested by both the literary and the archaeological evidence. One historical source refers to ‘Carpo-Dacians’ north of the Danube after 376, when the Tervingi who dominated the Carpathian region had already left, and there is no sign that all Cernjachov settlements and cemeteries came to a grinding halt at that date.21 My own best guess is that the complex sociopolitical world of the Tervingi comprised a dominant Germanic-speaking Gothic elite, most of them able to trace their origins back to third-century immigrants, with dependent freedmen and slaves of various origins closely tied in with them. Alongside this world of the Goths ‘proper’, as it were, also existed many communities descended from the older indigenous populations of the region. They had certainly been subdued by the Goths, and may well have paid various kinds of tributes, but were probably largely autonomous on a day-to-day basis, and that much less likely to have participated in the evacuation of 376.
In short, what we can reconstruct of the confederation of the Tervingi – in particular its military, political and cultural capacity to sustain itself in the face of Roman power – is broadly consonant with the idea that its leading political groups – ‘kings’ playing to an audience of militarized freemen and perhaps also, to a lesser extent, freedmen – could have engaged in a decision-making process of the kind Ammi-anus reports. Given the circumstantial detail he reports, and the fact that there is nothing inherently implausible in the action as he describes it, then his account should broadly be accepted. There is certainly not a big enough problem here to justify setting his narrative aside because of a priori assumptions about the limitations of group identity in the Germanic world. These doubts are based in part upon a one-sided reading of recent debates about group identity, and the broader run of evidence does generally indicate that the top echelons of the Tervingi shared at least a strong enough sense of political identity to make Ammianus’ account of their decision-making perfectly plausible.
Even a brief glance at the discipline of migration studies requires us to ask a more precise range of questions, however, if we are really going to understand the action. Why did the Tervingi and Greuthungi of 376 respond to the crisis generated by the Huns, first, by moving at all, and second, by deciding to move across the Roman frontier? Ammianus gives us no further details, so that we cannot hope to recover everything discussed in that highly charged meeting north of the Danube. But what can be learned from migration studies about the kinds of factors that play upon migrant decision-making, suggest a few observations of importance.
The fact that the Tervingi should have responded by moving is not in itself that surprising. We knew that its dominant political class was largely descended from Germanic-speaking migrants who had carved out their position in the Black Sea region as recently as the third century. Comparative migration studies have demonstrated repeatedly that a migration habit tends to build up within population groups. As noted earlier, older generations who have themselves moved pass on to their offspring the expectation that, if necessary, one might move in search of better conditions. And the ructions chiefly associated from a Roman perspective with the third century had carried on well into the fourth in lands beyond the frontier. Only after 300 AD, did the Tervingi take full control of the territories between the Carpathians and the Danube that had previously been the preserve of Carpic groups. Several such groups were transported south of the Danube by the Romans between c.290 and 310, and it was this that had allowed the Tervingi to move in. But even as late as the 330s, the Tervingi were still on the move. In 332, they started to move west of the Carpathians into the territory of some neighbouring Sarmatians, but Roman military action forced them to return to the Lower Danube region. Some of those who had participated in the events of the early 330s will still have been alive in 376, so that the possibility that one might solve life’s big problems by migrating was certainly a living tradition amongst the Tervingi elite.22
Another recurrent theme of migration studies, the importance of an active field of information, also played a central role in the decision to seek out the new territory they wanted inside the Empire rather than anywhere else. The Tervingi, of course, knew a great deal about their powerful neighbour; they had been semi-subdued Roman clients since the 320s. This must have influenced their choice of destination, once they had decided that they needed to up sticks.23 The advantages they perceived in this Roman option require, though, a bit more thought. Ostensibly, the Goths presented themselves to the Empire as refugees, offering it military service in return for sanctuary. But the Empire had well-established policies for the settlement of would-be immigrants, and of these the Tervingi were, again, well aware. They had witnessed at first hand the resettlements of Carpi around the year 300, and further resettlements of Sarmatians in the 330s. The terms of these resettlements were not necessarily punitive – they could range from the seriously unpleasant to the generous – but all resettlements were made in the context of overt Roman military domination. This precondition did not apply, however, in 376. When the Tervingi requested asylum, the Emperor Valens found himself in the middle of a long and complicated dispute with Persia, which he had initiated, and all his striking forces were tied up in the east.
This makes the issue of motivation on both sides, Roman and Gothic, significantly more complicated. A variety of sources are unanimous that Valens was extremely happy to see the arrival of the Goths on the Danube, viewing them as a ready source of military recruits. But it was a key feature of Roman imperial propaganda that no emperor should ever have policy dictated to him by barbarians, and this reported joy has to be seen as the propaganda it undoubtedly was. Only an idiot would be happy to see the total breakdown of political stability on one of his two major frontiers when he was already engaged in hostilities on the other, and, though many things, Valens was no idiot.24 Indeed, absence of overwhelming joy is confirmed by the careful policy he formulated. Rather than letting in all the unsubdued Goths requesting asylum, he admitted only the Tervingi of Alavivus and Fritigern, while posting all available troops in the Balkans to exclude the Greuthungi of Alatheus and Saphrax. Faced with not having enough troops to exclude all the Goths, he was making the best of a bad job.25
As for the Tervingi, it is a fair presumption that they were well aware of Valens’ situation. Frontier clients were adept at interpreting Roman troop redeployments – from the Danube to the Euphrates in preparation for the hostilities with Persia, for instance – and one basic fact of life in the frontier contact zone was that information leaked through it like a sieve. Ammianus tells one famous story of the Alamanni, who first began to suspect that trouble was brewing further east on the Danube in 376 because troops were being moved away from their front, then had their suspicions confirmed by a Roman guardsman of Alamannic origins returning home on retirement.26 But even if it seems unlikely that the Tervingi were second-guessing Valens from the start, we have two strong indications that they had something a bit more ambitious in mind than accepting the submissive role they knew the Empire usually assigned to immigrants. As Ammianus tells us, first, their request was for ‘part of Thrace’ not just as an escape route from the Huns, but also because its fields were fertile. Immigrants into the Roman world, as we have seen, were usually broken up into small groups and went wherever the Roman state chose. The Tervingi, however, had a more proactive choice in mind.
In seeking to understand this, it is important to factor in the general patterns of economic development operating in and around the Roman world. The Goths and other Germanic migrants of the third century had moved into the Black Sea region because it was part of a more developed inner periphery around the Roman Empire, with many economic attractions. And while these migrants were benefiting from that greater wealth, the Roman Empire was operating at a still higher level of development, with still greater economic surpluses. This wealth was immediately visible to outsiders in the Empire’s frontier zones in the form of towns, fortifications, armies, even villas, all of which, as we have seen, regularly attracted cross-border raiders. Ammianus’ account of Gothic motives – that Roman wealth had entered their calculations – makes perfect sense, therefore, and also recalls modern case studies, where it is rare for economic motivations to be absent from immigrants’ calculations, even when their thinking has a strong element of the political and involuntary about it. It also meant, of course, that the Goths were not just refugees in 376, since any ambition to share in Roman wealth was bound to bring them into conflict, in the longer term, with the Roman state, even if Valens was currently too preoccupied with Persia to put up much of an argument.
The second indication that the leadership of the Tervingi had higher-order ambitions in mind, and was well aware of the likely consequences, emerges in their reaction to Valens’ eventual decision to admit them, but not the Greuthungi. Instead of just rejoicing at their own good fortune, they continued, as Ammianus tells us, to maintain contact with the Greuthungi, with a view to joint action.27 This strongly suggests that the Tervingi’s leaders had formulated a more ambitious agenda, one that might well require concerted action on the part of both groups to realize. As to the precise nature of these ambitions, one can only guess. But the elite of the Tervingi were directly descended from third-century migrants who had witnessed a Roman withdrawal, under pressure, from the old province of Transylvanian Dacia. This deeper perspective, drawing on a longer-term field of information, as well as their own more immediate experience of Roman clientship, may have powered the hopes that made them turn their eyes towards the Empire in the summer of 376. Behind their self-presentation as refugees may well have lain the hope that they could make the Empire withdraw in due course from part of Thrace as well, and thus gain possession of a fertile landscape whose economic development was generally higher even than that of the inner periphery.
No wonder the discussions were lengthy . . . Moving on to the territory of the Roman state, especially if your ambitions strayed beyond the bounds of total submission, was a manoeuvre fraught with danger. Valens’ army may have been fully occupied in the summer of 376, but it was not going to be so for ever, and the Tervingi had first-hand knowledge of its power – from the 330s when it had forced them out of the lands of their Sarmatian neighbours, from the service of their own auxiliary forces within it between times; and from the 360s, when their only mechanism for avoiding outright defeat at its hands had been to run away. What all this emphasizes, of course, is that seeking asylum inside the Empire, despite its obvious economic attractions, was a stratagem that could only work if the migrants were able to field a significant military force. Without it, they would have not the slightest hope of fending off the Roman military counteraction, which was bound to follow in due course. The power of the Roman state supplied, therefore, a fundamental reason why the migration unit had to take the form it did, and this is entirely in line with another key point underlined by comparative migration studies.
Existing political structures are always a key determinant of the nature of migratory activity. Because of their relatively low economic development, fourth-century Germanic kings could support specialist military forces numbering only in the few hundreds. Forces of that magnitude stood no chance of facing up to a Roman emperor complete with a field army intent upon restoring ‘normal’ patterns of immigration. The best a small immigrant military force might hope for was to find employment as a reasonably well-treated auxiliary unit in the Roman army, and some Gothic groups of this kind who had entered the Empire at other times, it seems, followed precisely this trajectory.28 But for the Goths’ more ambitious enterprise of 376 to stand a chance of success, the leadership of the Tervingi needed to involve the broader militarized element of Gothic society: its freemen with their dependent freedmen – if my identification of the two warrior status groups is correct. The exact terminological identifications do not really matter, though. The key point is that large numbers of warriors were required, and just as in the third century, this meant that recruitment had to look beyond the world of specialist military retinues.
As a result, and again as in the third century, it was entirely natural that the migration units should encompass women and children alongside the warriors. The Goths of 376, like the third-century immigrants from Poland to the Black Sea, were set on a one-way trip, and the option of leaving families behind them on a what-if ? basis did not exist. Families left at home would have been much too vulnerable to predation from the Huns. And as already noted, the women as much as the men had had the migration habit firmly entrenched among them by the remembered life choices of their immediate ancestors. On the immediate everyday level, Germanic economic development could not support enough unencumbered specialist warriors to take on the Roman state unaided.
Looked at closely, then, the move of the Tervingi in 376 becomes less like the old invasion hypothesis in action than it might at first appear. The decision to move split the confederation, and, given the patterns of third-century history that had established their domination of their corner of the Black Sea region, a decision on the part of the Germanic-speaking elite to move on would not have emptied the landscape. As we have seen, this was a society with a considerable degree of social stratification, distinguishing between maybe four different social levels: free, freed, slaves integrated into ‘Gothic’ households, and, perhaps, largely autonomous tribute-payers as well. The kings and the broader (freeman?) elite were the dominant group within this culturally complex world, and many elements of its total population were not necessarily tied closely enough into their sociopolitical structures to be caught up in the migratory tide.29 But neither, all that said, is there any reason to doubt Ammianus’ basic premise that this Tervingi elite amounted to a large mass of individuals, numbering several tens of thousands. Not only is the account coherent in itself and confirmed by other sources, but it also makes sense in the light of the principles that underlie observable patterns of human migration.
This conclusion is important in itself, but there is a bigger point here too. Given the higher level of documentation provided by Ammianus Marcellinus, the events of 376 provide an important test case, illustrating what might also have been possible in other less well-documented instances involving Germanic groups of the late Roman period. It cannot just be assumed that all migratory phenomena of this era took the same form, and some certainly did not. But if we think of 376 as round one of the traditionalVölkerwanderung, then it is reasonable to think that it saw a large-scale movement being undertaken not by a single ‘people’, but by a coherent mass of population. And the picture is drawn for us by a well-informed contemporary who was evidently not slave to an ideological blindness about barbarians on the move. It also makes good sense given both the broader history of the Gothic world as itself the product of a migration into the Black Sea region, and the spread of political power and military capacity in contemporary Germanic society. To the predatory migration flow building up from small-scale activities into much larger forces, which was characteristic of the third century, we can add a second form of predatory (or, in the Goths’ case partly predatory) migration: the massed, mixed group. This is an important interim conclusion to keep in mind when considering the second stage of Roman frontier collapse.