Empires and Barbarians

by Peter Heather

  Like the Danube crossing of 376, the demographic displacement associated with the collapse of Rome’s central European frontier only partially fits the image of the traditional Germanic Völkerwanderung. The crisis of 405–8 did see massive mixed groupings cross the frontier for reasons that had more to do with factors external to the Empire than anything happening within it. And even if some of these groups were too well organized to resemble the floods of refugees sometimes seen in the modern world, their activities are often explicable in terms of the principles behind modern migration patterns, not least the web of negative and positive motivations driving the migrants, and the massive influence of existing political structures and flows of information. That said, the groupings were complex political associations, not ‘peoples’ in the traditional sense of the term. Some of the groups caught up in the action do seem to have had long histories. Hasding Vandals, for instance, figured in the second-century Marcomannic War. But like all Germanic groups of the late Roman period, they had been through several centuries of dynamic transformation generated by intense interaction with each other and with the Roman state, which meant that they encompassed a wide range of social classes and rights. This internal group complexity was then increased by the inter-group alliances that were forged, such as that between two separate Vandal groups, and Alans and Sueves in order to increase their chances of survival on Roman soil. This added to the picture much greater size, new political ties, and sometimes massive cultural disparity (in the case of the nomadic Iranian-speaking Alans). Even if some of their component units had well-established links, therefore, the entities that crossed the frontier were improvised political alliances, not long-standing aggregates of population.

  We should not wonder, then, that the Roman authorities were able to destroy some of them precisely by targeting the joins in their fabric, notably by attracting away elite military followers of both Radagaisus and Uldin at the expense of the group leader and the less favoured rank and file. But the internal disunity that might naturally be generated by social complexity and improvised alliances is only part of the story. Another striking characteristic of those groups that managed to survive their initial encounters with the Roman state was an apparent capacity to repeat the migratory process.

  FIGHTING FOR SURVIVAL

  The histories of all the major groups who crossed into the Empire at the two moments of frontier collapse followed a similar course. Their initial – overwhelmingly uninvited61 – penetrations of Roman territory were followed by periods of armed struggle. They had to force the Empire to accept that they could not be defeated, and that its normal policies for the subjugation and integration of immigrants could not be imposed upon them. For the Tervingi and Greuthungi, these initial struggles lasted for about six years until the negotiation of a compromise peace agreement with the Roman state, which came into force on 3 October 382. That the Empire was willing to agree to such a deal was entirely due to the Goths’ military capacity, in particular their successive defeats of two Roman emperors – Valens, most famously, at Hadrianople on 9 August 378, then Theodosius in Macedonia in the summer of 380. Other, smaller migrant groups of the period – Taifali, Sarmatians and isolated Gothic subgroups – who failed this initial military test received much harsher treatment, their defeats being followed by total loss of identity, as group members were distributed as unfree labour to Roman landowners.62

  The history of the migrants involved in the crisis of 405–8 is similar. Again uninvited, they had to fight, initially, to carve out new homes for themselves. Some failed. Many of the followers of Uldin and Radagaisus, as we have seen, met with disaster, killed or distributed again as unfree labour, though some elements of each group managed to do a deal with the Roman authorities. Initially at least, the Vandals, Alans and Sueves were more fortunate. After a career of wild violence in Gaul, in 409 or 410 they forced their way over the Pyrenees into Roman Spain, which offered them new opportunities. In 412, six years after their initial crossing, they divided up the bulk of its provinces between them. The Siling Vandals took Baetica, the Hasdings most of Gallaecia, the Sueves north-western Gallaecia, and the Alans, underlining that they were the largest component of the force at this stage, the richer provinces of Lusitania and Carthaginensis (Map 9). There is no evidence that this partition was authorized by the central Roman authorities, but it would seem to represent a more ordered exploitation of economic assets, beyond mere looting.63 The time lag between invasion and eventual settlement, whether we’re talking 376 or 406, is entirely understandable. No large surge of armed, unexpected immigration could ever have come to an immediate modus vivendi with the populations at its point of destination.

  What does need explanation, however, is that some time after these initial settlements – 382 and 412 respectively – both sets of immigrants apparently took to the road again. The Goths settled in the Balkans in 382, as the story traditionally goes, broke into open revolt under the leadership of Alaric in 395 and spent much of the next two years in a Greek odyssey, accompanied by their families and a vast wagon train, which took them as far south as Athens, round the Peloponnesus and then back north again to Epirus beside the Adriatic. After a brief rest, they moved into Italy in 401/2 before returning to the Balkans until 408, when they headed west, spending 408 to 411 in Italy again before taking off for Gaul, where they finally settled down. Likewise, the Vandals and Alans: after a Hispanic interlude which lasted until 429, they took ship across the Straits of Gibraltar and moved east, in two stages, towards the richest provinces of Roman North Africa. They briefly acquired land by treaty in Mauretania and Numidia in 437, before establishing two years later a more permanent home for themselves by capturing Carthage and the cluster of provinces around it.

  Looked at in this longer term, the immigration pattern of those who fled the Huns thus takes on a distinctly stop-start character. In the past, these narrative gaps were never seen as an obstacle to viewing the secondary migrations as the further history of the same groups that had made the original crossing. More recently, however, it has been suggested that the secondary migrations look much more like the activities of mobile armies than of the mixed population groups that made the original frontier crossings, and were indeed undertaken by what were essentially different groups – warbands on the make – who drew only marginally on manpower from the original migrants. This suggestion has been particularly well received among those sharing the conviction that ancient social units such as the invaders of 376 and 405–8 could never have had a strong enough sense of group identity to hold together through repeated upheavals, over such a long time-scale.64

  So, armies or peoples? And can migration studies help us comprehend the renewable mobility of Vandals, Goths and others?

  The fact that major disagreement can exist on such a basic point of interpretation will tell you instantly that, once more, the sources are not all they might be. They are, however, much fuller for the post-382 history of the Goths, at least for certain years, and they make the better test case. In the Gothic instance, the key initial question is whether those who rebelled under Alaric in 395 really did represent further movement on the part of all or most of the Goths settled under the treaty of 382. This was never doubted in the past, but new expectations that barbarian identity will always have been fluid have fuelled demands in recent years that the correspondence between the Goths who made peace with the Roman state in 382 and Alaric’s rebellious following should be proved. Can it be?

  In simple terms, the answer has to be no. No Roman commentator lists in detail the manpower resources drawn upon by Alaric in 395, or describes exactly how he mobilized support. On the other hand, we are talking about the middle of the first millennium, so this is not surprising, and it is important not to use unsatisfiable demands for an inappropriate degree of certainty as an excuse for denying what is in fact the very reasonable probability that in 395 Alaric did indeed lead a major revolt on the part of the treaty Goths of 382. The argument is not that
all those settled under the treaty necessarily participated in the revolt, but rather that there was sufficient overlap in manpower between those Goths settled under the agreement of 382 and Alaric’s initial followers for the basic point to hold.

  The first plank of the argument is that the better source material indicates strongly that this was in fact the case. Our two earliest, least problematic, entirely contemporary and independent Roman commentators on the rebellion, Claudian in the west and Synesius in Constantinople, describe Alaric’s following precisely as the 382 Goths in revolt. To discredit their testimony, convincing reasons would need to be found for both commentators – writing in separate halves of the Empire, for different audiences and for different purposes – substantially to have misrepresented the action, and none has yet been offered.65 Moreover, this basic observation – powerful in itself – can be strengthened. Synesius and Claudian have sometimes been rejected in recent years on the basis of a passage in the Greek historian Zosimus that reports Alaric as having originally revolted during the Eugenius campaign because Theodosius had only given him the command of some barbarian auxiliaries, rather than a proper Roman command. From this it has been supposed that his ambitions, and hence his revolt in 395, did not originally encompass the mass of the Goths of 382. There are three major problems here.

  First, what was originally the historian Eunapius’ contemporary account of Alaric has become demonstrably mangled at the hands of the sixth-century Zosimus. To put the contemporary Claudian and Synesius to one side on the strength of three lines (literally) of the much later Zosimus, whose account is anyway problematic, with no further argument about why they should both have distorted the action in the same way, is simply unsound methodology.66 Second, rewriting Alaric as having purely Roman ambitions runs into the problem that, just four years after his revolt began, an east Roman general of barbarian origins, one Gainas, took the opportunity of the revolt of some Gothic auxiliary troops to lever himself into power in Constantinople. Alaric à la Zosimus would be an analogous figure, as those who take that route acknowledge, but of the two authors Synesius had no problem in describing Gainas accurately (Claudian doesn’t even mention him).67 Why would either be likely to have misrepresented Alaric when, if certainly hostile to Gainas, he could describe his activities straightforwardly? Third, we can be certain that from the beginning Alaric’s following amounted to a major military force, surely ten thousand-plus warriors, since already in 395 it was able to face down a full Roman field army. If we don’t accept what Claudian and Synesius tell us, that it was the treaty Goths in rebellion that Alaric was leading, we also have to find a large alternative source of military manpower for him. This is not easy, given that the western generalissimo Stilicho had both eastern and western field armies under his command at this point.68

  The second plank of the argument, quite simply, is that it is entirely plausible that the Goths of 382 had maintained sufficient continuity of political identity over the intervening period to mount such a revolt. We are talking only thirteen years. Another generation will have matured in that time, but many adults active in 382 will still have been so in 395. And though woefully ignorant of many of its details, the point of the 382 treaty – for contemporary supporters and critics alike – was that it allowed an unprecedented degree of autonomy to continue among the Goths concerned. Although guilty of rebellion and the death of an emperor, they had not been broken up and widely distributed across the Empire, which is why Themistius, spokesman and propagandist for the Emperor Theodosius, had to work so hard to sell the peace to the Senate of Constantinople. This makes it entirely plausible that the same Goths could have acted in concert again, just thirteen years later.69

  That original treaty had also left unresolved two big issues in Gotho-Roman relations, and it was these that came to a head in Alaric’s revolt of 395. First, the Romans had recognized no overall Gothic leader in the peace of 382. This was in line with established Roman policies for limiting the political cohesion of groups they perceived as potential threats – it was standard policy towards Alamannic overkings, as we have seen, in the fourth century. Also, it was facilitated by developments within the confederations of the Tervingi and Greuthungi themselves. In both, the decision to move into Roman territory had been accompanied by political turmoil at the top and the removal of established leaderships, whether by death in battle or political overthrow.70 In the run-up to the battle of Hadrianople, Fritigern had tried to fill the gap, and there is good evidence both that the struggle had continued after 382 and that Alaric, too, had had to overcome rivals for the overall leadership of the Goths. It is certainly possible that his position was evolving in 394/5. Although attributable more to Zosimus’ garblings of Eunapius’ original, it may be the case that Alaric originally had ambitions for a more Roman career. But in the event he chose the Gothic option, and there is one excellent – if indirect – piece of evidence that he elbowed at least one rival out of the way to do so. Alaric’s later career, and that of his brother-in-law and successor Athaulf, were dogged by the interventions of a Roman general of Gothic origins by the name of Sarus, who waged a one-man war with a view to undermining any peace deal the two were looking to negotiate with the west Roman state, into whose service Sarus had moved. What’s so interesting here is that Sarus’ brother Sergeric eventually organized the coup in which Athaulf and his immediate family were killed, and made himself – briefly – ruler of Alaric’s Goths. So Sarus clearly came from a family grand enough to compete for the overall leadership of the Goths, and his unrelenting hostility suggests that Alaric’s rise was responsible for his departure for Roman service.71

  Alaric’s broader political success among the Goths, moreover, was intimately linked with the line he took on the second unresolved issue of the treaty of 382: the military obligations owed by the semi-autonomous Goths to the Roman state. As noted earlier, it was normal Roman policy in peace agreements imposed upon outsiders to extract drafts of young males for its armies. This may well have happened in 382, creating Gothic auxiliary units in the regular Roman army. But as had previously been the case with the Tervingi north of the Danube from 332, the treaty stipulated in addition that the Goths should provide irregular military service in the form of larger, autonomously led contingents for specific campaigns. Contingents from the Tervingi had fought on four occasions for Rome against Persia, between 332 and 360, and similar demands of the treaty Goths were made by the east Roman Emperor Theodosius I for his two civil wars against the western usurpers Maximus and Eugenius.72

  There is compelling evidence that this military service was resented by the Goths. On each of the campaigns against the usurpers, the participation of the treaty Goths was accompanied by revolts of some kind. Theodosius’ decision to seek assistance on the second occasion prompted a vicious quarrel among the Gothic leadership over how they should respond to his request.73 The fate of the Gothic forces on the second expedition also shows precisely why it was problematic. At the battle of the Frigidus in September 394, they found themselves in the front line on the first day and suffered heavy casualties. One contemporary Roman historian commented that the battle saw two victories for Theodosius: one over the usurper Eugenius and a second over the Goths. Given that the Goths’ semi-autonomy was tolerated by the Roman state only because they couldn’t be properly defeated, there was a real danger that such casualties would change the balance of power sufficiently to allow the Romans to rewrite the terms of the treaty. It is not in the least surprising, therefore, that almost as soon as they got home from the Frigidus campaign, sometime in winter 394/5, the treaty Goths rose in revolt under a leader committed to rewriting the terms of 382.74

  Much of what we would like to know about the treaty, and the pattern of Gotho-Roman relations it dictated, is beyond recovery. But as with so many diplomatic agreements, it was clearly a working compromise that left some of the more contentious issues to be resolved later. But it is entirely reasonable to suppose that Alaric’s revolt of
395 was of the nature that our two contemporary commentators describe. He was the leader of the bulk of the 382 Goths in revolt, the treaty having left them autonomous enough to be capable of rewriting their terms of agreement by collective action, and losses at the Frigidus had given them a real reason for discontent. This interim conclusion then prompts another set of questions. Why did the Goths’ rebellion in search of better terms involve further migration? It is, after all, perfectly possible to revolt without picking up the family and taking to the road again, lock, stock, and two smoking barrels.75

  The fact that they had an established migration habit has to be one element in the explanation. As its history shows, this was a population grouping prone to solving its difficulties by moving on to pastures new. The descendants of those who had moved from Poland to the Black Sea in the third century and into Wallachia in the early fourth, who had attempted to migrate west of the Carpathians in the 330s and who eventually crossed the Danube in 376, were a population group that knew a great deal about the practicalities of large-scale, long-distance movement, and had shown themselves ready to use it as a strategy for solving their problems. And, of course, some of those who crossed the Danube in 376, would certainly still have been alive in 395. But even groups with well-established migration habits do not move without excellent reason, and the travels of Alaric’s Goths, after the revolt, played a specific role in an unfolding strategy aimed at rewriting the unsatisfactory elements of the treaty of 382.