IN THE MID-890S, the latest nomad menace burst into the heart of Europe. Following in the footsteps of Huns and Avars, the Magyars shifted their centre of operations from the northern shores of the Black Sea to the Great Hungarian Plain. For the most part, the results were everything that past experience of nomad powers would lead you to expect:
[The Magyars] laid waste the whole of Italy, so that after they had killed many bishops the Italians tried to fight against them and twenty thousand men fell in one battle on one day. They came back by the same way by which they had come, and returned home after destroying a great part of Pannonia. They sent ambassadors treacherously to the Bavarians offering peace so that they could spy out the land. Which, alas!, first brought evil and loss not seen in all previous times to the Bavarian kingdom. For the Magyars came unexpectedly in force with a great army across the River Enns and invaded the kingdom of Bavaria with war, so that in a single day they laid waste by killing and destroying everything with fire and sword an area fifty miles long and fifty miles broad.1
The populations of the Great Hungarian Plain and surrounding regions, not least Great Moravia centred in Slovakia, were quickly subdued, and an orgy of equine-powered aggression saw Magyar raiding parties sweep through northern Italy and southern France with a ferocity not seen since the time of Attila, while full-scale Magyar armies defeated their East Frankish counterparts three times within the first decade of the tenth century.
But one element in the usual mix of nomad pastimes is missing from the Magyars’ European tour. Five hundred years before, the two pulses of Hunnic movement westwards – first on to the northern shores of the Black Sea in the 370s, and then on to the Great Hungarian Plain a generation later – had thrown semi-subdued, largely Germanic-speaking clients of the Roman Empire across its frontiers in extremely large numbers. Two hundred years later, the arrival of the Avars west of the Carpathians would prompt the departure of the Lombards for Italy and a widespread dispersal of Slavic-speakers in every direction: south into the Balkans, west as far as the Elbe, north towards the Baltic, and even eastwards, it seems, into the Russian heartland. Destructive as it was in so many ways, the arrival of the Magyars generated no documented population movements whatsoever (apart, of course, from those of the Magyars themselves). Why not? The answer lies in the dynamic interaction between migration and development which had played itself out across the European landscape over the previous thousand years.
The absence of any secondary migration associated with the Magyars is all the more surprising because it is one of the central findings of this study that, contrary to some recent trends in scholarship on the period, migration must be taken seriously as a major theme of the first millennium. This trend has not eliminated migration entirely from accounts of first-millennium history, but certainly incorporates a powerful tendency to downplay its importance. In some quarters even the word itself is avoided wherever possible, because ‘migration’ is associated with the simplistic deus ex machina of the ‘invasion hypothesis’ model of explanation, which was so prevalent up until the early 1960s. In this view, migration meant the arrival of a large mixed group of humanity – a ‘complete’ population: men and women, old and young – who expelled the sitting tenants of a landscape and took it over, changing its material cultural profile more or less overnight. This model was massively overused, and trapped the developing discipline of archaeology into migrato-centric models that crippled creativity. Besides, as many archaeologists have since pointed out, the model didn’t really explain anything anyway, because it never properly addressed the issue of why large groups of human beings might have behaved in such a fashion. This being so, it has been both reasonable and natural for subsequent archaeologists to concentrate upon other possible reasons for material cultural change. And these are legion. Everything from religious conversion to agricultural innovation and social development can have profound effects upon material cultural profiles. A highly suspicious attitude towards migration has also crossed the boundaries between disciplines. Some early medieval historians are now also so convinced that nothing like the old invasion hypothesis could ever have happened, that they are happy to suppose that historical sources must be misleading whenever they seem to be reporting possibly analogous phenomena.
A central aim of this study, however, has been to re-examine the evidence for first-millennium migration with a more open mind, and above all to reconsider it in the light of everything that can be learned about how migration works in the modern world. And from this point of view, one of its key conclusions is that the evidence for migration in the first millennium is both much more substantial and much more comprehensible than has sometimes been recognized in recent years. A deep-seated desire to avoid mentioning migration (a more successful version of Basil Fawlty and the war) has thus been wrenching discussion of some pivotal moments of first-millennium history away from the most likely reconstruction of events, and, in so doing, hampering analysis of the broader patterns of development that were under way.
It is an inescapable conclusion from all the comparative literature that a basic behavioural trait of Homo sapiens sapiens is consistently to use movement – migration (mentioned it again, but I think I got away with it . . .) – as a strategy for maximizing quality of life, not least for gaining access to richer food supplies and all other forms of wealth. The size of migration unit, balance of motivation, type of destination, and other detailed mechanisms will all vary according to circumstance, but the basic phenomenon is itself highly prevalent. In practice, two particular migration models have been retained in even the most minimizing of recent discussions: ‘elite replacement’ for larger-group movement, and ‘wave of advance’ for smaller migration units. Part of the attraction of both has been that they are safely different from the old invasion hypothesis. Elite replacement suggests both that not very many people in total were involved in the action, and that their migratory activity didn’t really have that much effect. If you just replace one elite with another, what’s the big deal? The wave-of-advance model employs mixed migratory units – essentially families – but their colonization of landscapes is piecemeal, slow, by and large peaceful, and decidedly not deliberate – intention being one of the elements of the old invasion model which revisionists find most problematic. How much of first-millennium European migration can be successfully described by employing these models?
Migration Modelling
Some of it, certainly. Cheating only slightly in chronological terms, the classic, superbly documented example of elite replacement is the Norman Conquest of England in 1066. In the following twenty years or so, as Doomsday Book shows, an immigrant, basically Norman elite took over the agricultural assets of the English countryside, evicting or demoting the existing landholders. But the overwhelming majority of the indigenous population remained exactly where they had been before the Normans arrived. Likewise, at least some elements of Wielbark expansion in the first and second centuries AD and of its later Slavic counterpart, particularly the spread of Korchak-type farmers through the largely unoccupied central European uplands, probably had a wave-of-advance quality about them. Looking at the millennium as a whole, however, these models are both too simple and too narrow to describe the totality of recorded migratory action.
First, the models themselves need a substantial overhaul. They either collapse different situations into undifferentiated confusion, or are of such limited applicability as to be more or less useless – at least for first-millennium Europe. As currently construed, elite replacement fails to distinguish the particularity of a case such as the Norman Conquest, where the invading elite could fit easily into existing socio-economic structures, leaving them intact, and any broader effects on the total population remain correspondingly small, if not so minimal as those wanting to undermine the importance of migration might think.2 But this kind of elite replacement applies only when the incoming elite was of broadly the same size as its indigenous counterpart, and I strongly suspect, even if I could never prove it, that, over the broad aeons of human history, this will have been true only in a minority of instances.
Certainly the first millennium AD throws up more examples of a different kind of case, where the intruding elite, if still a minority – and even quite a small one – compared to the totality of the indigenous population, was still too numerous to be accommodated by redistributing the available landed assets as currently organized. In these cases, existing estate structures had to be at least partially broken up and the labour force redistributed. As a result of this process, the entire balance between elite and non-elite elements of the population was restructured, and the overall cultural and other effects of the migration process were likely to be correspondingly large. This kind of elite migration could not but have huge socioeconomic consequences, and potentially also much greater cultural ones as the indigenous population came into intense contact with an intrusive elite, which was more numerous than its old indigenous counterpart. It was this intense contact, seen in Anglo-Saxon England and Frankish Gaul north of Paris from the fifth century, and perhaps to a lesser extent the Danelaw after 870, that generated substantial cultural, including linguistic, change, as the indigenous population was forced into modes of behaviour dictated by a new and relatively numerous foreign elite living cheek by jowl among them.3
Different again were cases of only partial elite replacement, particularly common in more Mediterranean regions of the old Roman west in the fifth and sixth centuries. Here there was some economic restructuring to accommodate the intruders – Goths, Vandals, Burgundians and others – but considerable elements of the old Roman landowning elites survived. In the longer term, it was the immigrants in these cases who struggled to hold on to their existing culture, and long-term linguistic change moved in the other direction. That is not to say, however, that this – the most limited form of migration on display in the first millennium – had only negligible consequences for the areas affected. In the first instance, high politics were dominated by the intrusive elites at the expense of their indigenous counterparts, at least when it came to matters like royal succession, and the overall political effect was sufficient to initiate major structural change. The disappearance in the medium to longer term of large-scale, centrally organized taxation of agricultural production, and the consequent weakening of state structures in the post-Roman west, are best explained in terms of the militarization of elite life that followed the creation of those structures at the hands of intrusive new elites.
The wave-of-advance model requires an equally substantial theoretical overhaul. The basic problem with it, even with ostensibly relevant cases such as Slavic Korchak expansion in the fifth and sixth centuries, or Wielbark expansion in the first and second, is that the Europe of the first millennium AD retained few if any uncontested landscapes of the kind that may have existed when the first farmers had been operating four thousand years before. By the year 1000, there were still plenty of forests, and we take our leave of European history at a moment when a further wave of agricultural expansion was in the process of hacking great swathes through them. But farmers had been clearing the landscape for millennia by this date and many of the best spots had long since been claimed. In this kind of context, random, uncontested expansion, even by small groups, was rarely an option. Korchak-type family or extended family groups probably did spread in largely uncontested fashion, but they did so by moving in a thoroughly non-random fashion through less sought-after, more marginal habitats of upland central Europe. And even here, the total subsequent subjugation of landscapes to the Slavic cultural model, combined with the documented aggression of Slavic groups in other contexts, strongly suggests that a degree of coercion might still have been involved. The same was probably also true of earlier Wielbark groups. Early Wielbark expansion seems to have been carried forward by small social units, but adjacent northern Przeworsk communities certainly came into Wielbark cultural line as a result of their activities. This could have been voluntary, but I suspect that examples of small-scale migration from the Viking period give us a more likely model for what was going on.
Small-scale Scandinavian migration units began carving out territories for themselves in northern Scotland and the northern and western isles of Britain from pretty close to the start of the ninth century. In this case, the logistic problem of getting access to shipping imposed constraints that did not apply in the Korchak or Wielbark cases. Hence, as is documented in subsequent Scandinavian expansion into Iceland and Greenland, the migration units, even if small, did have to be organized by jarls or lesser landowners (holds) who had sufficient wealth to gain access to shipping. But whereas Iceland and Greenland were more or less unpopulated landscapes, northern Scotland and the isles were not, and, even if the migrating units were individually small, Scandinavian expansion into these territories was certainly aggressive. Older suggestions that the result was ethnic cleansing are outdated, but the indigenous population was forcibly demoted to lesser status, and, over time, absorbed into the invaders’ cultural patterns. Small-scale migration need not, therefore, necessarily mean peaceful migration. As long as they confronted an indigenous population who did not have larger-scale, regionally based political structures, small migration units could still insert themselves successfully by aggressive means. Alongside a wave-of-advance model for small-scale migration that was random in direction and peaceful in nature, therefore, we need to add small-scale migration flows that were non-random or aggressive, or both. This kind of model is potentially highly applicable to the generally already-occupied landscape of first-millennium Europe, relevant not only to Wielbark, Korchak, and some Viking expansions, but also perhaps to the early stages of eastern Germanic expansion towards the Black Sea in the third century, of Elbe Germani into the Agri Decumates; or of Slavic groups north and east into Russia in the seventh to the ninth centuries.
What also emerges from the evidence is that too clear a line cannot be drawn between wave-of-advance and larger-scale migration. Just because an expansion began with small-scale migration units, does not mean that it stayed that way. The best-documented case here is provided by the Vikings. Initial Scandinavian raiding and settling, in the late eighth and early ninth centuries, were all carried forward by small groups. The earliest recorded violence involved the crews of three ships – perhaps a hundred men – and there is no reason to think that the settlements around Scotland and the isles need have been carried forward by groups much larger than this. But, as resistance and profits both built up, and the desire eventually formed to settle more fertile areas of the British Isles, where larger political structures in the form of Anglo-Saxon kingdoms barred the way, more important Scandinavian leaders became involved in the action, and larger coalitions formed among the migrants. This reached its climax during the Great Army period from 865, when coalitions formed with the idea of carving out settlement areas first in Anglo-Saxon England and then in northern Francia. If the early raids were undertaken by groups of no more than a hundred strong, the series of Great Armies each comprised much more like five to ten thousand men. The water-borne nature of the action in the Viking era always needs to be kept in mind because it imposed logistic problems that did not apply in other cases, but the evolution from raiding parties to great armies nonetheless provides awell-documented model of how – on the back of evident and growing military and financial success – originally small-scale expansion might eventually suck in much larger numbers of participants. The evidence is not so good for some of the earlier expansions, and these were not affected by the problems of water transport. Nonetheless, the expanding momentum of Viking-era migration provides a helpful model for understanding a series of other first-millennium migratory phenomena, not least the second- and third-century Gothic, and fourth- and fifth-century Lombard expansions, which again, it seems, started small, but grew in scale until forces large enough to fight major battles against Roman armies and regional competitors (such as the Carpi), came to be involved. Anglo-Saxon expansion into former Roman Britain can also be partly understood with such a model in mind, and it is potentially applicable to the third-century Alamanni.
Even without venturing into really contentious areas, therefore, the full range of first-millennium evidence suggests some major revisions to now-standard migration models. But in addition to small-scale migration, elite replacements, and migration flows of increasing momentum, first-millennium sources do periodically report large, mixed groups of human beings on the move: 10,000 warriors and more, accompanied by dependent women and children. Not only do such reports arouse suspicion because they seem uncomfortably close to the old invasion hypothesis, but this particular type of migration unit does not figure in modern migratory patterns, where large, mixed groups of migrants are seen only when the motivation is political and negative – when populations are fleeing oppression, pogrom and massacre, as in Rwanda in the early 1990s. This is not what is reported in first-millennium sources, which describe both a more positive motivation and a greater degree of organization among groups intruding themselves in predatory fashion into other people’s territory. Can we believe what the sources seem to be telling us? Should we retain large, mixed and organized groups of humanity as part of the overall picture of first-millennium migration?
Invasion
Even when employing the most up-to-date methods – DNA or steady-state isotope analysis – the kind of evidence that archaeological investigation brings to this debate is at best only a blunt tool. It remains hotly disputed whether much usable DNA will ever be recovered from human remains of first-millennium vintage laid down in the damp and cold of northern Europe. And too much has happened in demographic terms since the first millennium for the percentage distributions of modern DNA patterns to give much clear insight into the relative proportions of their progenitors 1,500 years ago, except perhaps in the highly exceptional case of Iceland, where there was no human population before the Viking era.4 Steady-state isotopes, likewise, only reveal where someone came to dental maturity. The children of two immigrants will have fully indigenous teeth, and this kind of analysis will always carry an in-built tendency to underestimate the importance of migration. Arguments based on more traditional types of archaeological investigation – the transfer to new regions of items or customs originally characteristic of another – are also unlikely to be any more conclusive.
The reasons are straightforward. By the birth of Christ, most of Europe had been settled and farmed, after a fashion, for millennia. And since even the most aggressive and dominant of immigrants usually still had a use for indigenous populations as agricultural labour, migration did not tend to empty entire landscapes. Furthermore, as all comparative study has emphasized (and modern experience shows), when migrants move into an occupied landscape, the result – in material and non-material cultural terms – is always an interaction. There are only a relatively few items in any particular group’s material cultural profile that are so loaded with meaning that they will be held on to, for good or ill, in the longer term. Everything else is open to change under the stimulus of new circumstances, so you can hardly expect migration to involve the complete transfer of an entire material culture from point A to point B in normal first-millennium European conditions. There will always be some elements of continuity in the material cultural profile of any region subject to migration, and this makes it entirely possible, if you are so inclined, to explain any observable change in terms of internal evolution. Goods and ideas can move without being attached to people, and if what you observe archaeologically is no more than a limited transfer of either, it will always be possible to explain it in terms of something other than population displacement. But the fact that it will always be possible to do this does not mean that it will necessarily be correct to do so, and the inherent ambiguity of archaeological evidence is sometimes misinterpreted. Ambiguity means exactly that. If the archaeological evidence for any possible case of migration is ambiguous – which it usually will be – then it certainly does not prove that migration played a major role in any observable material cultural change – but neither does it disprove it. What all this actually amounts to is that archaeological evidence alone cannot decide the issue. It is important to insist on this point because there has been a tendency in some recent work to argue that ambiguous archaeological evidence essentially disproves migration, when it absolutely does not. Overall, of course, this forces us back on to the historical evidence. How good a case can be made from historical sources for the importance of large, organized and diverse groups of invaders on the move in the first millennium?
The answer has to be complex. There are some clear instances where a migration topos, a misleading invasion narrative, has been imposed on more complex events. Jordanes’ account of Gothic expansion into the northern Black Sea region in the late second and third centuries is a classic case in point, as is the picture of the fourth- and fifth-century Lombard past to be found in Carolingian-era sources and beyond. But in other cases, the historical evidence in favour of distinct pulses of large-scale migration involving 10,000-plus warriors and a substantial number of dependants is infinitely stronger: the Tervingi and Greuthungi who asked for asylum inside the Roman Empire in 376, for instance, or the movement of Theoderic the Amal’s Ostrogoths to Italy in 488/9. In both these cases, attempts have been made to undermine the credibility of our main informants, respectively Ammi-anus and Procopius, but they lack conviction. Ammianus described many different barbarian groups on the move on Roman soil in the course of his historical narrative and only on this one occasion does he refer to very large mixed groups of men, women and children. The idea that he was infected by some kind of migration topos in this instance, but not elsewhere, takes a lot of believing. Likewise Procopius: he is not in fact the only source to describe Theoderic’s Ostrogoths on the march to Italy as a ‘people’ in a quasi-invasion-hypothesis sense of the term (a large, mixed grouping of men, women and children). One contemporary commentator even described them as such in person to Theoderic and other actual participants gathered at his court. You wouldn’t want to hang anyone in a court of law on this kind of evidence, but its credibility is pretty much as good as anything else we get from the first millennium. To reject it on the basis of a supposed migration topos is arbitrary.5
Not quite in the same category of solidity, but still well within the usual limits of first-millennium plausibility, likewise, is a range of evidence indicating that moves of a similar nature were made by large, organized Vandal and Alanic groups on to Roman soil from 406, and by Radagaisus’ Goths in 405.6 And while more argument is certainly again required, by far the likeliest reconstruction of Alaric the Visigoth’s career indicates that it was founded on mobilizing the Tervingi and Greuthungi of 376, settled in the Balkans by treaty in 382, into a series of further moves from 395 onwards. These are all instances of large, mixed-group movement that pass muster on all the normal rules of first-millennium evidence. There are also enough of them to require us not to dismiss too quickly a series of other cases, where the evidence is a notch or two weaker: in particular the population movements associated with the rise and fall of the Hunnic Empire, which saw the gathering-in of armed, largely Germanic groups, and their subsequent departures from the Great Hungarian Plain as competition built up among them in the era of Hunnic collapse. Here the evidence for large group migration is either partial (the cases of the Rugi or Heruli), or implicit rather than explicit (those of the Sciri, Sueves, and Alans). Although we can find some convincing cases where the action has been mistakenly cast in the form of an invasion-hypothesis-type population movement, therefore, there are many others where there is no good reason to think that this has happened. And, in fact, even the Goths and Lombards are worth a closer look.
In both instances, we are dealing with highly retrospective miscastings of the action. Jordanes was writing about events that happened three hundred years before his own lifetime, and the Lombard authors in the ninth century and beyond about migratory activity that was then four to five hundred years in the past. On one level, it is easy to see why mistakes might have crept in, but there is more to say here than just that. For in neither case was it complete fantasy to be thinking in terms of migration of some kind. The totality of the evidence for both the second- and third-century Goths and the fourth-and fifth-century Lombards does indicate that substantial population displacement played a major role in these eras of their respective pasts.
The evidence is better for the Goths. Here we have contemporary accounts locating Goths in northern Poland in the first and second centuries, but north of the Black Sea from the mid-third. There was also a major material cultural revolution north of the Black Sea in the third century, in the course of which a whole series of customs and items became prominent in the region, which had not previously been part of its characteristic profiles. Some of the more distinctive among the new features, moreover, had been well-established aspects of life and death in first- and second-century Poland. These archaeological indications cannot prove that Gothic migration took place from the Baltic to the Black Sea regions, but, taken in conjunction with the contemporary historical evidence, they amount to a very serious argument to that effect. And while that historical evidence clearly indicates, as we have seen, that, even if there were many separate groups involved in the action rather than one ‘people’, and that some of them were perhaps originally numerically challenged, this did not remain the case throughout the migratory process. The third-century Goths provide, in fact, an excellent case of a migration flow of increasing momentum, which didn’t really stop until the Gothic Tervingi had displaced the Carpi as the dominant grouping between the Danube and the Carpathians in the decades either side of the year 300. Though much less detailed, the Lombard evidence is similar.
Lombards are well attested in the Lower Elbe region, just south of modern Denmark, in the first and second centuries AD. In this case, there are no contemporary historical indications at all of any major population displacements from this region in the Roman period, and what archaeological evidence there is might suggest only a series of relatively small ones, like the first Gothic flows towards the Black Sea. Yet again, however, Lombards were present on the Upper Elbe in sufficient numbers by the 490s to move in and destroy the hegemony of the Heruli in the western half of the Great Hungarian Plain through main force. Whatever its earlier forms, therefore, Lombard expansion towards the Danube, like that of the Goths towards the Black Sea, eventually took the form of much larger pulses of population. Neither Jordanes nor our Lombard sources invented the concept of large-scale migration from nothing, therefore, even if they miscast its form. And, just to push their rehabilitation one stage further, subsequent Gothic and Lombard migrations, occurring between these initial flows and our sources’ composition, had taken the form of large, composite group moves, both in the direction of Italy: the Ostrogoths in 488/9 and the Lombards some eighty years later.7
When examined more closely, therefore, neither Jordanes nor the Lombard sources give us reason to deny the reality of the large-group migrations recorded in other sources. That said, it is very important to recognize that even our echt examples of large-group first-millennium migration do not conform exactly to the old invasion-hypothesis model. Not even the largest groups were whole ‘peoples’ moving from one locality to another untouched by the process. They could both shed population and gain it. This was presumably even more true of drawn-out migration flows, such as the second- and third-century Goths and fourth- and fifth-century Lombards, but the pattern is only explicitly documented for some of the large-group moves. Decisions to move on such a scale were never lightly, and often caused splits. The Tervingi who crossed into the Empire in 376 left behind them north of the Danube a significant minority of the old group’s membership who adhered to the old leadership. Theoderic the Amal’s father caused another split when he moved the then Pannonian Goths into the Roman Balkans in 473, and Theoderic himself left behind at least some elite Goths who were absorbed into the military-political hierarchies of the east Roman state. When it comes to gathering recruits, the Lombards were joined by a mixed group of 20,000 Saxons for their move to Italy, together with descendants of much of the flotsam and jetsam left over by the post-Attilan struggles for power in the Middle Danube. Theoderic the Amal, likewise, added a body of Rugi to the Gothic following built up over two generations by his uncle and himself. Similarly, the relationship between the two Vandal groups and the Alans who crossed the Rhine together became much tighter in the face of Roman counterattack in Spain, so that, by the time they invaded North Africa in 429, the survivors, united behind the Hasding monarchy, were much more of a cohesive political unit than the loose alliance they had been twenty-three years before. As much recent work has emphasized, there was as much of the snowball to these migratory movements as the billiard ball.
Another significant departure from the old invasion model is the fact that, when looked at closely, these large groups were mixed not only in age and gender, but also in status. Visions of the Germanic Völkerwanderung produced in the great era of nationalism in the nineteenth and early twentieth centuries had in mind large invasion groups of free and equal warriors with their families in tow. But within the larger groups, two separate status-categories of warrior are documented, and there is reason to suppose that non-militarized slaves also participated in at least some of the moves. It is only the higher warrior class that fell into the ‘free’ category, and the fact that they were by definition an elite class suggests that this group was some kind of minority. The key decisions about migration in the period were being taken, therefore, only by a minority of the participants, with lesser warriors and slaves having little if any influence. Recognizing the reality and significance of these status distinctions also imposes clear limits on the extent to which currently fashionable ideas about the freedom with which group identities could be chosen and discarded can really have applied in practice. What kind of idiot would have chosen to be of lesser-warrior or slave status if group identity was entirely a matter of individual choice? By extension, this also indicates that we need to be careful as to how far we suppose snowball-type phenomena to have operated. Since much of the population of barbarian Europe was not in control of its own destiny, the right to join or not join in large-group migration must have been exercised only by certain, more elite elements among participating populations.8
The final modification that must be made to the old invasion-hypothesis model of large-group migration concerns its supposition that large-scale intrusions drove out existing populations. There are several good examples of large-scale invasion in the first millennium, but none where the evidence suggests mass ethnic cleansing. Indigenous populations were often faced with a choice between accepting subjugation or moving on, a choice which would have felt particularly brutal to indigenous elites who had most to lose from the arrival of a new set of masters. But there is no convincingly documented case where the response to this choice led to the complete evacuation of an extensive landscape. At the very least, indigenous populations supplied good agricultural labour, and many of our immigrant groups anyway had lower social-status categories into which newly subjugated indigenous populations could easily be slotted.
These alterations are important, but they remain modifications rather than denials of the basic proposition that the evidence for large, mixed, and organized migrant groups from the first millennium is, ostensibly, periodically convincing. Nationalist visions of whole ancestral ‘peoples’ clearing out new landscapes for themselves to enjoy can be consigned to the recycle bin of history. The groups documented in our sources were political entities, which could grow or fragment, which contained individuals occupying lesser- and higher-status categories, and which inserted themselves in correspondingly complex ways into new but already thoroughly inhabited environments. Though fair enough on the basis of the available historical evidence (and not denied by the archaeological), can this proposition still be maintained in the face of the non-appearance of such phenomena in modern migratory patterns? The answer to this question is bound up, in my view, with that to a far larger one: Why did European migration take the forms it did in the first millennium? Answering this question requires us to set the observable patterns of demographic displacement between the birth of Christ and the year 1000 against all that comparative study can teach us about migration as a general human phenomenon.
Migration Mechanics
There are a myriad detailed ways in which the mechanics of first-millennium migration correspond to what has been observed in better-documented case studies of early modern and more recent migration. Not least of these, the crucial importance of active fields of information in dictating precise destinations is just as prominent in the first millennium as in later eras. Germanic expansion towards the Black Sea in the third century was clearly exploiting information about the region which had built up through the operations of the Amber Route. Slavic groups first came to know the Roman Balkans as raiders before exploiting that knowledge to turn themselves into settlers as and when political conditions permitted. Scandinavian expansion to the west in the Viking era likewise operated on the back of intelligence acquired by participation in the emporia trading networks of the eighth century, while those working to the east took a generation or so to find their way down the river routes of western Russia to the great centres of Islamic demand for northern goods, having originally opened up the eastern hinterland of the Baltic to feed western markets. To these entirely uncontroversial examples, I would also add some others. A major contributory factor to the apparently odd stop/start migratory patterns of some of the groups entering Roman territory either side of the year 400 was the need to acquire information about further possible destinations before hitting the road again. The Goths, especially the Tervingi who entered the Empire in 376, already knew about the Balkans, for instance, but not about Italy and Gaul, to where they moved on in the next generation. It took twenty years (and their participation in two Roman civil wars that took some of them lengthy distances in that direction) before they were ready to take the next step. Likewise the Vandals and Alans: Spain marked the end of their original migratory ambitions, and it again took twenty years and some exploratory sea raids before they were prepared to venture across the Straits of Gibraltar to North Africa. More generally, the whole broader phenomenon of migration flows of increasing momentum is clearly a product of growing knowledge. It was precisely the fact that exploratory expansionary ventures into a new region produced profitable outcomes for the pioneers that encouraged others to participate. In some modern cases, such as the spread of the Boers northwards from the Cape, the pioneers were deliberately recruited scouts, sent to check the viability of larger-scale expansion. The same effect could also be achieved, however, by a less formal grapevine.
The study of modern migration also devotes much effort to the key issue of why some people from any particular community choose to move, whereas others in more or less identical circumstances stay put. Tackling this complex issue fully requires the kind of detailed information which is simply unavailable for the first millennium, but it is worth pointing up the relevance of the issue. In the cases of large-group migration reported in any detail in our sources, there is no instance where the decision to move did not generate some kind of split among the affected population group. The same is true, only more so, of the more extended migration flows. For all the Germani of Polish origin who ended up by the Black Sea in the third century, there were many others who stayed behind, shown by the fact that the Wielbark and Przeworsk cultural systems continued to operate. Likewise, many Angles and Saxons did not relocate to England in the fifth and sixth centuries, and Scandinavia was not emptied in the Viking period. Such divergences of response were only natural, of course, given the magnitude of the decisions involved, and first-millennium populations clearly felt the same stress of migration as modern ones, even if we can’t explore their reactions in detail.
Stress also manifests itself in the modern world in the phenomenon of return migration. Looked at closely, all modern migration flows see substantial numbers of immigrants returning to their original homelands. Again, the level of information is not sufficient to allow us to discuss this topic properly for the first millennium, but aspects of the Viking period emphasize that it, too, needs to be recognized as a real phenomenon. The initial phases of Scandinavian expansion were all about gathering wealth, whether by raiding or trading, or both. Having gathered their wealth, different individuals then made different choices about how to invest it. Some chose, even early on, to stay put at their points of destination in the east and west (as shown by the early settlements in northern Scotland and the isles), whereas others chose to take their new wealth back home to Scandinavia, eventually prompting a massive shake-up in Baltic politics. With this example in mind, I (as others) would be happy to believe reports that some incoming Anglo-Saxons also eventually chose to return to the continent.9
Closely related, too, to the stress of migration, but this time something we can explore in greater detail, is the highly significant influence on patterns of movement of an ingrained migration habit. In modern migration flows, an existing tradition of mobility often plays a vital role in dictating which individuals within a particular group of people will decide to move. Individuals who have moved once are more prone to move again within their own lifetimes, but, less intuitively obvious, the habit is also passed on between generations. The children and grandchildren of migrants are much more likely than the average to move again themselves. A tradition of personal or familial mobility clearly generates a greater propensity to attempt to solve life’s problems, or look for greater opportunity, by moving to new localities. Anyone might move if the stimulus to do so is large enough, but the required stimulus is smaller for those with established migration habits.
The effects of this factor can be seen at work on at least two different levels in the first millennium. First, at least two of the broader population flows, those of the Wielbark and Przeworsk Germani in the second and third centuries, and of the early Slavs three hundred years later, involved populations whose farming techniques were then insufficient to maintain the fertility of any individual piece of arable land for more than a generation or two. A general, periodic local mobility was simply a fact of life for these populations, and there is every reason to suppose that this facilitated the eventual transformation of a more random wave-of-advance-type expansion into a channelled migration flow when information began to filter back about the opportunities available at an entirely new set of longer-distance destinations. Second, a more specific tradition of distinct, longer-distance relocation clearly built up among some particular first-millennium populations. The fourth-century Gothic Tervingi are probably most famous for the fact that a majority of them decided to seek asylum inside the Roman Empire in 376. That decision was greatly facilitated, however, by active memories of recent migrations. This same Gothic group had taken possession of their existing lands in Wallachia and Moldavia between the Lower Danube and the River Dniester only in the decades either side of the year 300, and a generation or so later, in the 330s, had attempted to move bodily to new locations on the fringes of the Middle Danube region. It was the children of those who had moved into Moldavia and Wallachia who were on the move again in the 330s, and their children and older grandchildren who decided to seek a new life inside the Roman Empire in 376. Similar observations apply to many of the other groups caught up in the rise and fall of the Hunnic Empire, both those who fled inside Roman borders in the crises of 376–80 and 405–8, and those who moved first to the Middle Danube under Hunnic influence and/or duress, and then out of it after Attila’s death. The willingness of some Norse to move on to Iceland and Greenland in the later ninth century was likewise facilitated by the fact that they were the immediate descendants of Viking immigrants to Scotland and the isles. In fact, examples like the Goths or Slavs demonstrate how moves that were initially generated by general traditions of local mobility could then spawn the more specific traditions of larger-scale mobility that underlay the move of many Tervingi on to Roman soil in 376, in the same way that internal migrants within the European landscape provided many of the recruits for the trek to North America in the nineteenth century.
Aside from the emotional costs of migration, financial ones were also a major factor in any migrant’s calculations. Most first-millennium migration that we know anything about was a question, more or less, of walking and wagons. It involved no major transportation costs, apart from wear and tear to animals, peoples and wheels, and participation was consequently open to many. It nonetheless involved many indirect costs, above all the potential food shortages that were bound to result when movement disrupted normal agricultural activity. As a result, food stocks had to be maximized before moving, unless circumstances were completely overwhelming, and this meant that autumn was the classic moment to make a move – just after the current year’s harvest had been gathered and while there was still a chance of some grass growing to feed the oxen pulling the wagons and other animals. Alaric’s Goths moved into Italy in both 401 and 408 in the autumn, Radagaisus’ Goths in autumn 405. The Vandals, Alans, and Sueves who crossed the Rhine at the very end of 406 likewise presumably began their trek from the Middle Danube in the autumn of that year.10
As usual, there is little information about the impact of migration costs beyond this very basic point, but logistic problems do show themselves from time to time in the available data. Above all, extended periods of movement left groups particularly vulnerable in economic terms. Flavius Constantius was able to bring Alaric’s Goths – now led by Athaulf and Vallia – to heel by starving them out in 414/15. By that date, they had been living off the land without planting crops for six or seven years. Later in the fifth century, similarly, after the collapse of the Hunnic Empire, the surviving sources give us just a little insight into the logistic strategies adopted by Theoderic the Amal. His grouping journeyed around the Balkans with wagonloads of seedcorn in the 470s, and one dimension of its diplomatic negotiations with the Roman state involved providing it with agricultural land. Even on the march, noticeably, this group always sought to establish more regular economic relationships with Balkan communities, rather than merely robbing them. This meant that the communities could keep on farming and producing surpluses, from which the Goths could siphon off a regular percentage, whereas destroying them by pillage would only have fed Theoderic’s followers once.
Logistic factors had a still bigger impact upon population flows requiring more than land transport. Mass access to sea transport did not even become a possibility until the advent of steerage class in the enormous transatlantic liners of the later nineteenth century. Before that point, travel costs necessarily limited participation in any kind of maritime-based expansion. Again, the Viking period provides the best-documented first-millennium example. Ships were highly expensive, and even specialist cargo ships could carry only limited numbers of people and their goods. Thus Viking raiding required the less well-off to come to some kind of joint arrangement for funding the purchase or hire of a ship (though how many shipowners, I wonder, would be willing to hire out shipping for raiding ventures?), or to attach themselves to a leader of higher status.11 Logistic limitations figured even more strongly when it came to the settlement phases, when so many more types of people and a wider range of bulky farming equipment were required. The relevant shipping costs alone make Stenton’s suggested phase of large-scale Norse peasant settlement in the Danelaw pretty much inconceivable. Who would have bothered to pay for this when there was a subdued, low-status Anglo-Saxon labour force already in situ? We also see the effects of logistics on the Icelandic settlements, where each immigrant unit was headed by a higher-status individual who could presumably cover the large initial investment costs of transportation. Logistics may also have limited the number of Scandinavian women who participated in the Norse migration flow compared to the other land-based movements of our period. Modern DNA patterns suggest that only one-third of immigrant women to Iceland came all the way – directly or indirectly – from Scandinavia, with the rest moving a shorter distance from the British Isles. This may reflect the fact that it was too expensive for more than a minority of warriors to bring their Scandinavian sweethearts with them. On the other hand, given that the men involved were relatively wealthy, and that they were pagan and polygamous, it may be that women outnumbered men early on, with each Scandinavian male bringing with him not only his Scandinavian sweetheart but a couple of British or Irish babes besides.
As frustratingly limited as all the Viking-period evidence is, our other major first-millennium examples of maritime movement – Anglo-Saxon moves across the North Sea and the expedition of Vandals and Alans across the Straits of Gibraltar – are not even illuminated in this much detail. The impact of logistic demands must, though, have been similar. Possibly the Vandals and Alans were in a position to use intimidation to requisition some transport, but I doubt that they were able to avoid having to meet most of their costs themselves. And there is reason to think that the fact that they could only move a relatively few people at a time dictated their initial choice of landfall: Morocco, far away from the better-defended heartland of Roman North Africa, which was their ultimate destination. The logistic impossibility of moving many people at once was probably also a major factor in the drawn-out nature of Anglo-Saxon migration into southern Britain.
So far, so good: there are a whole series of ways in which the comparative literature sheds light on first-millennium migration, even given the limited data set available. But this still leaves some big questions to answer. Above all, can we in fact accept what the sources appear to be telling us: that the first millennium occasionally saw large, mixed, and organized population groups take to the road? And, if so, how are we to explain this phenomenon both in its own context, and the fact that it has not been observed in more modern and better-documented eras? Again, modern migration studies, in my view, can help answer these questions, but a fully satisfactory explanation also requires us to explore first-millennium migration patterns against the backdrop of a much wider set of transformations that were unfolding simultaneously in barbarian Europe.