1 MIGRANTS AND BARBARIANS IN APRIL 1994, ABOUT two hundred and fifty thousand people fled from Rwanda in East-Central Africa into neighbouring Tanzania. The following July a staggering one million people followed them into Zaire. They were all running away from a wave of horrific killing which had been set off by probably the most unpleasantly successful assassination of modern times. On 6 April that year, Presidents Juvénal Habyarimana of Rwanda and Cyprien Ntaryamira of Burundi were killed when their plane crashed as it attempted to land at Rwanda’s capital, silencing the two leading moderate voices of the region at one stroke. Other moderate voices in the government, bureaucracy and judiciary of Rwanda were silenced with equal dispatch, and the killing began, not only in the towns but in the countryside as well. The UN estimates that one hundred thousand people were massacred in the month of April alone, and probably about a million altogether. The only escape lay in flight, and in both April and July, men, women and children fled for their lives. Most of the refugees’ possessions were left behind, and with them secure access to good-quality food and water. The results were predictable. Within the first month of the July flight to Zaire 50,000 of the refugees had died, and altogether somewhere close to 100,000 – one tenth – would succumb to cholera and dysentery. Rwanda is only the most dramatic of many recent examples of migration as a response to political crisis. Only slightly later, 750,000 Kosovan Albanians fled to neighbouring countries in a similar response to escalating violence. But large-scale flight from danger is only one cause of migration. More numerous are all the people who use movement to a ‘richer’ country as a strategy for improving the quality of their lives. This phenomenon is found right around the globe. Two hundred thousand people out of a total of three and a half million left the Irish Republic in the 1980s, largely for destinations in economically more dynamic areas of Europe, though many of them have since returned as the Irish economy has boomed, with Ireland itself becoming a major destination for migrant labour. And economic migration is even more prevalent where living standards are poorer. Of different sub-Saharan populations, fifteen million are currently to be found in the Middle East, fifteen million in South and South-East Asia, another fifteen million in North America and thirteen million in Western Europe. The causes of this staggering phenomenon – the numbers are so large as to be virtually unimaginable – lie in massive inequalities of wealth. The average income in Bangladesh, for instance, is one-hundredth of that prevailing in Japan. This means that a Bangladeshi who can get work in Japan at only half the average Japanese wage will earn in only two weeks the equivalent of two years’ income in Bangladesh. Political violence and economic inequality combine to make migration – in its many forms – one of the big stories of the modern world. Nor was it so different in the past. ‘The history of mankind is the history of migration.’ This is a truism, but, as with most truisms, one that is in abroad sense correct. It is a basic implication of the currently available evidence for human evolution that, having evolved in one favourable context on the continent of Africa, different hominid species then used the adaptive skills provided by their extra brain power to colonize most land environments on the planet. The whole world, in essence, is peopled by the offspring of immigrants and asylum seekers. The recorded history of the last millennium, too, throws up many examples of migration, some of them – especially those originating from within Europe – remarkably well documented. The modern USA, of course, is a phenomenon created by immigrants. Up to sixty million Europeans migrated overseas between 1820 and 1940 to destinations worldwide, thirty-eight million of them to North America. Continuing waves of especially Hispanic-speaking immigration mean that the US story has not yet reached any kind of conclusion. Likewise, a quarter of a million people emigrated from Spain to the New World in the sixteenth century, another two hundred thousand in the first half of the seventeenth. In the same centuries, respectively, eighty thousand and half a million British braved the North Atlantic. Moving still further back in time, the documentation becomes scrappy, but migration was certainly a significant phenomenon. In the high medieval period, perhaps two hundred thousand Germanic-speaking peasants moved east of the Elbe during the twelfth century alone to take lands in Holstein, west Brandenburg and the Saxon marches. THE PEOPLING OF EUROPE This book is concerned with a still more distant past: Europe in the first millennium AD. It is a world that hovers between history and prehistory. Some parts of it are studied primarily through written historical sources, others through the material remains that are the preserve of archaeologists. This range of evidence and its combinations pose particular challenges, but there is no doubt that migrants of all kinds were busy within the frontiers of Europe in the thousand years after the birth of Christ. Given the overall part that migration has played in human history, it would be bizarre if they had not been. The first two centuries AD saw Romans move outwards from Italy to bring the joys of town life and central heating to large parts of western Europe. But it is the migration of so-called barbarians from beyond the borders of imperial Europe that has long been seen as fundamentally characteristic of the first millennium. Who were these barbarians, and where and how were they living at around the time that Christ was born in Bethlehem? Barbarian Europe At the start of the first millennium, imperial Europe, defined by the reach of Rome’s legions, stretched out from the Mediterranean basin as far north – broadly speaking – as the River Danube and as far east as the Rhine. Beyond these lines lay Europe’s barbarians, who occupied some of the central European uplands and most of the Great European Plain, the largest of Europe’s four main geographical regions (). The unity of this vast area, however, lies in geological structure, not human geography. While heavy clay soils are characteristic throughout its wide expanses, distinct variations in climate and hence vegetation generate marked differences in its farming potential, both because of the growing seasons and the basic fertility of the soil. Western parts, particularly southern Britain, northern France and the Low Countries, are governed by Atlantic weather systems, bringing mild, damp winters and cooler summers with, again, plenty of rain. Why it was the British who invented cricket, the only game that cannot be played in the rain, remains one of history’s great mysteries. Central and eastern reaches of the plain enjoy a more continental climate, with colder winters and hotter, drier summers. Average winter temperatures fall as you move further east, and summer rainfall declines in a south-easterly direction. Historically, this has had huge effects on farming, particularly in pre-modern eras employing only limited agricultural technologies. In the south-east, even in the famously fertile black-soil region of Ukraine, productivity was limited by low summer rainfall, with settlements clinging to the river valleys. North and east, winter cold imposed serious limitations. Because of the cold, the characteristic deciduous and mixed deciduous/coniferous forests which comprise the natural vegetation of most areas of the plain eventually give way, first to purely coniferous taiga forest and then to arctic tundra. Broadly speaking, the northern boundary of the mixed woodland zone marks the edge of that part of the European landscape where enough humus built up in the soil in the dim and distant past to make normal farming, or an adapted version of it, possible. At the start of the first millennium AD, much of this plain was still heavily wooded, and northern Europe was a long way from developing its full agricultural potential. This was not just because of the trees, but also because of the soil. Potentially highly productive, the thick clay soils of the North European Plain required heavy ploughs to maintain their fertility: ploughs capable not just of cutting furrows but of turning the soil over, so that the nutrients in weeds and crop residues could rot into the soil and be reclaimed for the next growing season. In the mid- and high Middle Ages, this problem was solved by the carruca, the four-wheeled iron-shod plough drawn by up to eight oxen, but at the start of the millennium most of Europe’s barbarians were doing little more – literally – than scratching the surface. So the inhabitants of the European plain were farming at very little, if anything, above subsistence level, and the population was distributed between isolated, cultivated islands amidst a sea of green. Mediterranean commentators were always much more interested in themselves than in the barbarian ‘other’ across the frontier, but even they could see that there were more of these islands of cultivation, and hence a denser overall population, the further west you went. More specifically, they divided the barbarian occupants of the Great European Plain into Germani and Scythians. There had previously been Celts – Keltoi – too, but most of previously Celtic western- and central-southern Europe had been swallowed up by the advance of Roman might. And already at the start of the millennium, these areas were set on a non-barbarian trajectory towards Latin, towns and rubbish collection. The archaeological evidence suggests that the placing of the new boundary of imperial Europe wasn’t just an accident. Pre-Roman Celtic material culture is famous for a distinctive art style, expressed particularly in beautifully crafted metalwork. Celtic settlements of the period also shared a general sophistication in other aspects of material culture: amongst other things, technologically advanced wheel-turned pottery, substantial and often walled settlements (so-called oppida), and the considerable use of iron tools to generate a comparatively productive agriculture. The material remains thrown up by Germanic-speakers in the same period, by contrast, were generally of a much less rich and developed kind. Typical finds from Germanic Europe consist of cremation burials in urns with few or no gravegoods, only hand-worked rather than wheel-made pottery, no developed metalwork style and no oppida. The general level of agricultural productivity in Germanic-dominated areas was also much less intense. It was precisely because the economy of Germanic Europe produced less of an agricultural surplus than neighbouring Celtic regions, of course, that there was smaller scope for the employment of the specialist smiths and artists required to produce sophisticated metalwork. And while the Romans never took abroad strategic decision to absorb just Celtic Europe, the narratives of attempted conquest indicate that Roman commanders on the ground eventually came to appreciate that the less developed economy of Germanic Europe just wasn’t worth the effort of conquest. Traditional accounts of Rome’s failure to conquer the Germani, as these Germanic-speakers are now often called, emphasize the latter’s destruction of Varus’ three legions at the battle of the Teutoburger Wald in 7 AD. Reality was more prosaic. The defeat was heavily avenged by the Romans in the years that followed, but this couldn’t hide the fact that potential taxes from a conquered Germanic Europe would pay neither for the costs of conquest nor for its subsequent garrisoning. As a result, shortly after the birth of Christ, different Germanic-speaking groups were left in control of a vast tract of Europe between the Rivers Rhine and Vistula (). The primary social and political units of these Germani were characteristically small. Tacitus in the first century and Ptolemy in the second provide an almost bewildering list of group names, which you can only approximately plot on a map. The key point emerges nonetheless with total clarity. There were so many of these political units (‘tribes’ if you like, but that word carries a lot of potentially inappropriate baggage) that, individually, they must have been extremely small-scale. Not all of this area had always, or perhaps even for long, been the preserve of Germani. Graeco-Roman sources document that Germanic Europe had grown in size periodically, even if they provide almost no circumstantial detail about the processes involved. The Germanic-speaking Bastarnae moved south-east of the Carpathians at the end of the third century BC, for instance, to become the dominant force north-west of the Black Sea. Around the turn of the millennium, the Germanic-speaking Marcomanni evicted the Celtic Boii from the upland basin of Bohemia. When we talk of Germanic Europe, therefore, we are really talking about Germanic-dominated Europe, and there is no reason to suppose that the entire population of this truly vast area – some of it militarily subdued in the fairly recent past – was culturally homogeneous in terms of belief systems or social practice, or even that it necessarily spoke the same language. ‘Scythia’ was a catch-all term among Graeco-Roman geographers for inhabitants of eastern parts of the North European Plain, stretching from the River Vistula and the fringes of the Carpathian Mountains to the Volga and the Caucasus (). In Greek geographical and ethnographic tradition, it was often portrayed as a chill wilderness, the archetypal ‘other’, the mirror image of Greek civilization. And to the inhabitants of this world, every imaginable type of uncivilized behaviour was ascribed: blinding, scalping, flaying, tattooing, even drinking wine unmixed with water. In reality, the territory designated by this term encompassed a wide variety of habitats. In the valleys of the great rivers flowing gently south out of the eastern reaches of the Great European Plain good farming country could be found, within, at least, the temperate zones marked by the extent of the forested steppe. To the south lay the much drier landscape of the steppe proper, whose expansive grasslands provided a natural home for the herds of the nomad. Further north and east, less intensive farming regimes gradually faded out, leaving the landscape for the hunter-gatherers of the Arctic Circle. Of these different population groups, nomads will play a major role in our story of the transformation of barbarian Europe in the first millennium, but only an indirect one, so there is no need to explore their world in detail. Suffice it to say that by the start of this period nomad populations had long tended to roam the lands south-east of the Carpathians and north of the Black Sea. Geologically, this landscape is again part of the European plain, but a general lack of summer rainfall makes farming precarious or impossible. East of the River Don, there isn’t enough rain to make farming viable without irrigation, a technology which singularly failed to penetrate these lands in antiquity, and the land retained its natural vegetation: steppe grassland. West of the Don, enough water for farming is to hand in some of the river valleys, but these valleys sit in close proximity to a large swathe of territory, just inland from the Black Sea coast, which is again natural steppe country. Perhaps not surprisingly, therefore, political domination of this landscape in antiquity tended to switch backwards and forwards between nomad and more settled agricultural groups. At the birth of Christ, the Germanic-speaking Bastarnae and Peucini who had moved into the region in the third century BC still retained their domination, but it was about to be overturned by nomadic Sarmatians, who swept through the area in the first century AD. North of the forested steppe, the eastern reaches of the North European Plain are swathed in vast stretches of increasingly coniferous forest. Here, as average winter temperatures fall lower and there is less humus in the soil, conditions become much tougher for farming. It was a world little known to the Mediterranean at the start of the first millennium. In his Germania, Tacitus places the hunter-gatherer Fenni (Finns) in the far north, and another group, the Veneti (or Venethi), between them and the Germanic Peucini on the fringes of the Carpathians: The Veneti have taken a great many customs from the Sarmatians, for in plundering forays they roam through all the forests and hills that rise between the Peucini and Fenni. Still, they are more properly classed as Germani, because they have fixed homes and bear shields and take pleasure in moving fast by foot. Pliny, a little earlier, had likewise heard of the Venedae, as he names them, but reports no detailed information, and even the second-century geographer Ptolemy knew little more about them than a few of their group names. The area was a touch less mysterious than what lay beyond, where people had ‘human faces and features, but the bodies and limbs of beasts’, but only just. Archaeologically, the picture of the inhabitants of these wooded and forested zones of eastern Europe around the birth of Christ is reasonably straightforward. As Tacitus’ comment about permanent settlements implies, it was a world of farmers, but farmers with an extremely simple material culture, less developed even than that prevailing further west in Germanic Europe. The remains of its pottery, tools and settlement are so simple, in fact, that they frustrate any attempt at stylistic or even chronological categorization, being extremely slow to change before the second half of the first millennium AD. This archaeological evidence suggests that it was a world of small, isolated farming settlements, operating at a lower subsistence level than the Germani, with little sign of any surplus, and none of trade links with the richer world of the Mediterranean to the south. The ethnic and linguistic identity of these forest-dwelling Veneti has generated much discussion, in particular regarding their relationship, if any, with the Slavic-speakers who become so prominent in European history after about 500 AD. We will return to this discussion in Chapter 8, but it would appear that the likeliest place to find Slavs – or their most direct ancestors – at the birth of Christ was somewhere among these simple farming populations of the easternmost stretches of the Great European Plain. With only a little simplification, therefore, barbarian Europe at the start of our period can be divided into three main zones. Furthest west and closest to the Mediterranean was the most developed, with the highest levels of agricultural productivity and a material culture that in its pottery and metalwork was already rich and sophisticated. This had long been controlled largely by Celtic-speakers, and much of it had just been brought under Roman rule. Further east lay Germanic-dominated Europe, where agriculture was less intensive, and which consequently lacked the same richness of material culture. Even Germanic Europe practised a relatively intensive agriculture, however, compared with the inhabitants of the woods and forests of eastern Europe, whose material culture has left correspondingly minimal remains. Nothing in this brief survey is really controversial, except, perhaps, where Slavs might be found. What has become highly disputable, however, is the role played by migration in the astounding transformation of barbarian Europe which unfolded over the next thousand years. Barbarian Migration and the First Millennium That some migration occurred within and out of barbarian Europe in the first millennium would be accepted by everyone. The big picture, however, is now very polemical. Before the Second World War, migration was seen as a phenomenon of overwhelming importance in the transformation of barbarian Europe: a spinal column giving the millennium its distinctive shape. Large-scale Germanic migration in the fourth and fifth centuries brought down the western Roman Empire, and established new linguistic and cultural patterns in the north. This was the era when Goths from the northern Black Sea littoral moved over two thousand kilometres to south-western France in three discrete leaps over a thirty-five-year period (c.376–411 AD). Vandals from central Europe went nearly twice that distance and crossed the Mediterranean to end up, again after three discrete moves, in the central provinces of Roman North Africa. This took thirty-three years (c.406–39), including a lengthy sojourn in Spain (411–c.430). It was in these centuries, too, that the history of the British Isles took a decisive turn with the arrival of Anglo-Saxon immigrants from Denmark and northern Germany. Of greater importance still, arguably, was Slavic migration. Slavic origins were always hotly debated, but, wherever they came from, there was no doubting the fact that from relative obscurity in the sixth century Slavic-speakers spread across vast tracts of central and eastern Europe over the next two hundred years. Substantial parts of this landscape had previously been dominated by Germanic-speakers, so the rise of the Slavs represented a huge cultural and political shift. It created the third major linguistic zone of modern Europe alongside the Romance and Germanic tongues, and the boundaries between the three have remained little altered since they were first created. Scandinavian migration in the ninth and tenth centuries then completed a millennium of mass migration. In the Atlantic, entirely new landscapes were colonized for the first time in Iceland and the Faroes, while Viking migrants in western Europe established Danelaw in England and the Duchy of Normandy on the continent. Further east, other Scandinavian settlers played a key role in creating the first, Kievan, Russian state, whose limits established and delineated the boundaries of Europe down to the modern era. No single view of any of these migrations and their significance ever won universal acceptance. Many of the details, as we shall see in the chapters that follow, have always been and will remain highly controversial. But the conviction that barbarian migration played a hugely formative role in the history of Europe in the first millennium was a distinctive feature of all European scholarly traditions up to 1945. It was true of history on the very grandest scale. Here first-millennium migrants were seen as establishing the main linguistic divides of modern Europe: between its Romance-, Germanic-, and Slavic-speaking populations. But migration was given a critical role at more intimate levels too. Particular sets of migrants were considered to have laid the foundations of such long-lived and geographically widespread political entities as England, France, Poland and Russia, not to mention all the Slavic states who clawed their way to independence from Europe’s multinational empires in the nineteenth and twentieth centuries. In the interwar era, the proportion of modern European nation states who traced the origins of their distinctiveness back to first-millennium migrants was staggering. This shared vision of the past is what more recent scholarship has come to call a Grand Narrative. Argument never ceased over the details, but that didn’t really matter. The important point was that so many of the population groupings of modern Europe considered the roots of their own distinctiveness to lie in a continuous history stretching back to a migratory moment somewhere in that specific thousand years. An integral part of this narrative was a particular vision of the nature of the population units doing the migrating. Many of these moves were not well documented in the historical sources, some not at all. But what historical information there was did sometimes talk of large, compact groups of men, women and children moving together in highly deliberate fashion from one habitat to the next. This information struck a chord. Since the migrant groups were seen as the start of something big – entities with a long future of continuous distinctiveness ahead of them leading inexorably to the nations of modern Europe – it was natural to apply this vision to them all. Thus all the migrant groups of the first millennium – documented or not – came to be viewed as large, culturally distinctive and biologically self-reproducing population groupings which moved, happily unaffected by the migratory process, from point A to point B on the map. These distant ancestors had to be numerous and distinctive enough to explain the existence of their many and now politically self-assertive descendants in the modern era. A good analogy for the migration process envisaged might be billiard balls rolling around the green baize table. Something might make the balls roll from one part of the table to another – overpopulation at the point of departure was the usual suspect – but any one ball was straightforwardly the same ball in a different place when the movement had finished. This view was applied particularly to Germanic groups involved in the action of the fourth to sixth centuries, but also to a considerable extent to Slavs and Scandinavians as well. Modern Slavic groupings such as Serbs, Croats and Slovenes, for instance, all traced their history back to coherent migratory populations of the first millennium. This first-millennium narrative was itself part of a still grander narrative accounting for the whole peopling of Europe in prehistoric times. The birth of Christ marked the moment when written historical information started to become more or less available for large parts of Europe north of the Alps. Reconstructing the more distant past relied entirely upon archaeological evidence and tended to be written up – before 1945 – in terms of a sequence of ‘more advanced’ population groups succeeding one another as the dominant force in the European landscape. The first farmers of the late Stone Age arrived from the east to displace the hunter-gatherers, the copper users did the same for the stone users, the bronzesmiths for the copper users, until eventually we reached the Iron Age and the first millennium AD. The details of this bigger picture do not concern us, but to understand what follows it is necessary to realize that a migration model taken from some first-millennium texts – one where coherent groups of men, women and children moved intentionally to take over landscapes – was imported back into the deeper past, wholesale, to explain the developing patterns of archaeological remains from prehistoric Europe. It was because of what people thought they knew about first-millennium migration that the first farmers, then subsequently those who worked copper, bronze and iron, were all viewed – successively – as outside population groups moving in to take over the European landscape. Within this grandest of all grand narratives about the peopling of Europe, our period represented an end and a beginning. It saw the last in the sequence of major migrations by which the whole history of the continent had been shaped since the last Ice Age, and marked the start of a Europe peopled by entities with a continuous history – that’s to say, groupings largely untouched by further migration – down to the present. It also provided the migration model by which all of this European history was ordered. Its sheer prevalence is the key to understanding the virulence of subsequent intellectual response. THE GREAT MIGRATION DEBATE Since 1945, so many key elements of this migration-driven narrative of the European past have been challenged that the old certainties have been eroded. In some parts of Europe, the narrative continues broadly to hold sway, but particularly in English-speaking academic circles, migration has been relegated to a walk-on part in a historical drama that is now largely about internally driven transformation. This intellectual revolution has been so dramatic, and its effects on more recent accounts of first-millennium migration so profound, that none of what follows will make sense without some understanding of its major outlines. A key starting point is the completely new understanding, which emerged in the postwar era, of how human beings come together to form larger social units. Identity Crisis It may seem strange that the first port of call in thinking about migration should be group identity, but the old grand narrative of European history has ensured that migration and identity are inextricably linked, at least when it comes to the first millennium AD. This is for two basic reasons. First, the billiard ball model of migration that powered this narrative assumed that human beings always came in compact groupings of men, women and children who were essentially closed to outsiders and reproduced themselves by endogamy (marrying someone who was already a member of the group). Second, in what is essentially the same view of group identity played out over the long term, it was presumed that there was a direct and tangible continuity between immigrant groups of the first millennium and similarly named nations of modern Europe. Thus the Poles were the direct descendants of the Slavic Polani, the English of Anglo-Saxons, and so forth. National identities were ancient, unchanging ‘facts’, and their antiquity gave them a legitimacy which overrode the claims of any other form of political organization. Where they did not prevail as the prime mode of political organization, then some other power structure (such as the old multinational empires of central and eastern Europe) had in the meantime erected itself by the illegitimate use of force, and needed to be overturned. Both assumptions have been shown to be flawed. Nazi atrocities played a key role in stimulating historians to think again about the presumption – generated at the height of European nationalism in the late nineteenth and early twentieth centuries – that nations had always existed, and were the fundamentally correct way to organize larger human communities. In Nazi hands, these ideas led straight to claims for Lebensraum, based on how much of Europe the ancient Germani had once controlled, and, with the added dimension of claimed German racial superiority, to the horror of the death camps. Historians would probably have got there anyway at some point, but the excesses of runaway nationalism provided a powerful stimulus to corrective reflection. On closer examination, the assumption that ancient and modern speakers of related languages somehow share a common and continuous political identity has proved unsustainable. The kinds of national identities that came to the fore in nineteenth-century Europe were created in historical time, and did not represent the re-emergence of something fundamental but long submerged. Without the kind of mass communications that became available in the eighteenth and nineteenth centuries, it would have been totally impossible to bind together numerically huge and geographically dispersed populations into national communities. Group identity simply did not function in the same way in earlier eras without canals, railways and newspapers, a world where ‘country’ meant ‘county’, for instance, for the vast majority of the British population. The creation of modern nationalism also required the conscious input of intellectuals, who created national dictionaries, identified national costumes, and collected the dances and folktales which were then used to ‘measure’ ethnicity (I’ve always thought of these men as looking a bit like Professor Calculus out of Tintin). These same individuals then also generated the educational programmes that solidified the elements of national culture that they had identified into a self-reproducing cultural complex which could be taught at school, and by that means reach a still larger body of humanity in an era when mass primary education was rapidly becoming – for the first time – a European norm. The emergence of nationalism is a great story in itself, and has rightly attracted a lot of attention in the last generation or so of scholarship. The point for us, though, is straightforward. Europe has not been peopled since the first millennium by large blocks of population conscious of distinct nationalist affiliations which fundamentally shaped their lives and activities. Nineteenth- and early twentieth-century affiliations cannot be imposed on the deeper past. Feeding into this reconsideration of the nationalist phenomenon was an equally revolutionary set of conclusions emerging from the work of social scientists studying exactly how, and how strongly, individual human beings are ever attached to any kind of group identity. In this field, the world was turned upside down in the 1950s by an anthropologist called Edmund Leach, who investigated how identity worked in the hills of northern Burma. Leach was able to show that an individual’s group identity does not necessarily vary with measurable cultural traits, whether material (types of houses or pottery, for example) or non-material (shared social values, belief systems and so on). People sharing the same set of measurable cultural traits (including language: the great symbol of group identity in the nationalist era) can think of themselves as belonging to different social groups, and people with different cultures can think of themselves as belonging to the same ones. Fundamentally, therefore, identity is about perception, not a check-list of measurable items: the perception of identity the individual has inside his or her head, and the way that individual is perceived by others. Cultural items may express an identity, but they do not define it. A Scotsman may wear a kilt, but he remains a Scotsman even if he doesn’t. As a great deal of further work has confirmed, this suggests an entirely different view of the bonds that create human group identities from that which prevailed before the Second World War. Up to 1945, identity was viewed as an unchanging given, a defining aspect of any individual’s life. But studies inspired by Leach’s work have shown both that an individual’s group identity can and does change, and that a particular individual can have more than one group identity, sometimes even choosing between them according to immediate advantage. In our post-nationalist world, this seems less surprising than it might have done sixty years ago. My sons will have both American and British passports, where before 1991 they would have had to opt for one or the other at eighteen (at that point you could be a joint American citizen only with Israel and Ireland – an interesting combination); EC citizens have both their home-national and a European identity. And instead of being seen, as used to be the case, as an overriding determinant of life choices, group identity is now sometimes relegated to a much more minor role. Particularly influential in first-millennium studies, for instance, has been a set of essays published by the Norwegian anthropologist Fredrick Barth in 1969. The collective view emerging from these papers portrays identity as no more than a strategy for personal advancement. As circumstances change, making first one group identity then another more advantageous, the individual will vary his or her allegiance. As Barth famously characterized it in the introduction to these essays, group identity must be understood as an ‘evanescent situational construct, not a solid enduring fact’. This work transports us a million miles from the expectation that individuals will have one fundamental identity that defines them for life, a notion that not only seemed unchallengeable in the era of nationalism, but was also the basic assumption behind the migration model that drove the Grand Narrative of European development in the first millennium (and, indeed, the deeper past as well). The billiard-ball view of migration absolutely assumed that migrants moved in complete social groups that were closed to outsiders, that replicated themselves by endogamy and that possessed their own culture, which was identifiably different from that of any other group they might encounter on their travels. This vision rested in part, as we’ve seen, on some historical texts, but mostly on prevailing assumptions about how human groupings were organized, since the historical texts were actually few and far between. Once nationalist assumptions about group identity had been undermined, it was open season on the old Grand Narrative that had rested so firmly upon them. The New Millennium? The lead in thinking again about the deep European past from a post-nationalist perspective has been taken by archaeologists. Traditional approaches to European archaeology worked by mapping patterns of similarity and difference in archaeological finds of broadly the same date across a given landscape, so that defined sub-areas – called ‘cultures’ – came to be marked out. Originally such definitions tended to be based almost exclusively on pottery types, since pottery fragments are both indestructible in themselves and relatively easy to find, but any kind of similarity, whether in burial customs, house types, metalwork or whatever, might have been used in principle, and has been since. The empirical fact that boundaries can sometimes be drawn between areas of archaeological similarity and difference emerged quickly in the nineteenth century with the rise of archaeology as a scientific discipline. In that intellectual and political context – again we’re talking the height of European nationalism – it proved irresistible to equate the cultures depicted on the maps with ancient ‘peoples’, who were, after all, each presumed to have had their own material (and non-material) cultures. If you were very lucky, and were working on a late enough period, you might even be able to name the bearers of the culture you had found in the ground on the basis of information from a historical text such as Tacitus’ Germania. Now often called ‘culture history’, the development of this approach is particularly associated with the German scholar Gustav Kossinna, who was active from the late nineteenth into the early twentieth century. His approach was a touch more sophisticated than is sometimes appreciated. He did not say that all areas of archaeological similarity should be equated with independent ancient peoples. This was only true, he argued, in cases where sharp boundaries could be drawn between different archaeological areas, and where the similarities within the bounded area were marked and distinct. But terms such as ‘sharp’, ‘marked’ and ‘distinct’ were always made to be argued over, and the fundamental assumption of archaeological investigation in this era was that you would normally find your remains neatly packaged in distinct ‘cultures’, and that these cultures were the remains of ‘peoples’. The key point for us is that Kossinna’s culture history underpinned much of the Grand Narrative. Thinking of archaeological cultures as ‘peoples’ carried within it a powerful tendency to explain major archaeological change in terms of migration. Where particular and distinct assemblages of material remains – archaeological ‘cultures’ – were each equated with ancient ‘peoples’, who were also viewed as the basic unit of human social organization, it was only natural to think any change to an existing pattern of remains represented the impact of a new ‘people’. Given that each people had its own ‘culture’, when you suddenly found a new ‘culture’ on top of another, you then might well think that one ‘people’ must have replaced another. Migration, particularly in the form of the mass replacement of one population group by another, thus became the characteristic means by which observable changes to archaeological remains were explained. In modern parlance, although the term had not yet been coined, the peopling of Europe was envisaged as being driven forward by one massive episode of ethnic cleansing after another, in what has been evocatively dubbed the ‘invasion hypothesis’ view of the past. The impact of new understandings of group identity on this old intellectual structure has been profound. Once the assumption was removed that the material remains of the past would present themselves in neatly packaged ‘cultures’ left by ancient ‘peoples’, it became much less clear that they did. As more material has come to light and existing finds have been subjected to closer scrutiny, many of the boundaries between supposedly distinct cultures have started to blur, while the identification of important local variants has often undermined the homogeneity of supposed cultures from within. Equally, and perhaps even more important, while patterns of similarity do nonetheless sometimes exist, and, where they do, usually mean something important, it has also become clear that no simple rule (such as ‘cultures’ = ‘people’) can be applied universally. The precise significance of any particular pattern of similarity and difference will depend in fact on exactly what is similar and different about it. An observable archaeological ‘culture’ might represent the physical remains of anything from an area of general social or economic interaction, to an area of shared religious belief (where, for instance, funerary rites are similar), or even, in some cases, an area of political association (as Kossinna essentially supposed). A good way to summarize the difference in approach, it seems to me, is that Kossinna thought of archaeological cultures as the remains of entities – ‘peoples’ – but modern archaeologists regard them as the remains of systems of interaction, and the nature of that interaction does not have to be the same in every case. Rethinking the nature of cultures in this way has allowed archaeologists to demonstrate that even major material cultural changes can have causes other than outside invasion. Since patterns of observable archaeological similarity can be generated for a variety of reasons – trade, social interaction, shared religious belief or anything else you can think of – then changes in one or more of any number of these areas might be responsible for an observable change. Changes do not have to reflect the arrival of a new social group but might be caused by any substantial alteration in the system that originally created it. Indeed, it was deep dissatisfaction with the intellectual limits of the invasion hypothesis, overemployed as a monolithic model of change, as much as the impact of the new understandings of group identity, that drove a whole generation of archaeologists in the English-speaking world to reject its tenets in the 1960s, and in many other parts since. For very good reasons, therefore, archaeologists have increasingly looked beyond the invasion hypothesis to other types of explanation altogether, since the 1960s. These new approaches have been highly fruitful, and in the process undercut much of the broader sweep of the old Grand Narrative. Up to about 1960, European prehistory was envisaged as one population group after another using their new skills – in farming technology or metallurgy – to establish dominance over the landmass and expel their predecessors. Nowadays, much of the evolution of central-western European society between the Bronze Age and the Roman Iron Age (roughly the last two millennia BC) can be convincingly explained without recourse to mass migration and ethnic cleansing. Instead of one set of invaders after another overthrowing each other, the European past is now peopled with human beings who could learn new skills and, over time, develop new economic, social and political structures. There is one further element to this intellectual revolution that has had a huge impact on more recent approaches to the story being explored in this book. In the process of freeing themselves from the undoubted tyranny of culture history and the invasion hypothesis, certain (particularly British and North American) elements of the archaeological profession have come to dismiss migration almost entirely as an agent of significant change. Such has been their collective sigh of relief at escaping from Kossinna’s conceptual straitjacket that some have resolved never to have anything to do with migration again. For these archaeologists, migration is associated with a previous, less advanced era in the intellectual development of their discipline, when in their view archaeology was subordinated to history. The billiard-ball migration model found some of its justification in historical sources, as we have seen, and when cultures were thought of as ‘peoples’ it was possible to write about prehistoric archaeological transformation as a quasi-historical narrative, with people X succeeding people Y, and so forth. As a result, a basic equation has grown up in the minds of some archaeologists between any model of the past involving population movement, and simple-mindedness. As a recent introduction to early medieval cemeteries put it, avoiding migration in explanations of archaeological change ‘is simply to dispose of an always simplistic and usually groundless supposition in order to enable its replacement with a more subtle interpretation of the period’. Note the language, particularly the contrast between ‘simplistic’ and ‘groundless’ (the world dominated by migration) with ‘more subtle’ (any other kind of explanation). The message here is loud and clear. Anyone dealing with the geographical displacement of archaeologically observable artefact types or habits, who wants to produce an account of the past that is at all ‘subtle’ or ‘complex’, should avoid migration at all costs. The tables have turned. From a position of overwhelming dominance before the 1960s, migration has become the great Satan of archaeological explanation. Such a major intellectual U-turn was bound to have a profound impact on the way historians approached the first millennium, where archaeological evidence was always of vital importance, and, of course, historians had in the meantime been thinking about the significance of the great identity debate for themselves. The consequential landmark of change in historical thinking, the starting point for all subsequent approaches to identity and hence first-millennium migration, was a book published in 1961 by the German scholar Reinhard Wenskus. Entitled Stammesbildung und Verfassung (The Generation and Bonding of Tribes), it showed that you don’t have to read far even in the pages of the first-century Roman historian Tacitus to find some Germanic groups being totally exterminated, and other entirely new ones being created. And when you get to the great migrations of the fourth to the sixth centuries, the evidence for discontinuity only multiplies. As we will explore in more detail later, all the Germanic groups at the heart of the successor states to the Roman Empire in this era – Goths, Franks, Vandals and so on – can be shown to be new political units, created on the march, many of them recruiting from a wide range of manpower sources, some of which were not even Germanic-speaking. The political units formed by the Germani in the first millennium were thus not closed groups with continuous histories, but entities that could be created and destroyed, and which, in between, increased and decreased in size according to historical circumstance. There has been much discussion since of the details of how group identity might have worked among first-millennium Germani, and on its likely strength, and we will need to return to these arguments in due course. But all subsequent discussion has accepted and started from Wenskus’s basic observations. These observations have had a profound knock-on effect upon understandings of Germanic migration. Under the old view of unchanging closed group identities, if group X was suddenly encountered in place B rather than in place A, it was only natural to conclude that the whole group had moved. Once it is accepted that group identities can be malleable, then in principle only a few – maybe even a very few – of group X need have moved to provide a core around whom a population from disparate sources then gathered. The billiard-ball view has thus come to be replaced by the snowball. Instead of large, compact groups of men, women and children moving with determination across the landscape, many now think in terms of demographic snowballs: originally small groupings, probably composed largely of warriors, who, because of their success, attract large numbers of recruits as they travelled. Such post-nationalist readings of the historical evidence for barbarian Europe in the first millennium had similar but independent roots to the new dawn that was sweeping simultaneously through archaeology. But the vehemence of the archaeologists’ new mindset has added further momentum to the evident potential for rewriting the story of barbarian migration from historical sources. So convinced now are some historians that large, mixed migration units could never have been a feature of the past that they have started to argue that the handful of historical sources that apparently report the opposite – the source of the invasion-hypothesis model of migration – must be mistaken. Graeco-Roman sources, it has been suggested, are infected with a migration topos, a cultural reflex that made Mediterranean authors describe any barbarians on the move as a ‘people’, whatever the real nature of the group. A European history composed of long distance, large-scale population moves is being replaced by a history of small-scale mobile groupings, gathering in followers as they went. Migration – though the word is now scarcely used – remains part of this story, obviously, but with the scaling-down of the numbers of people envisaged as participating in those journeys, the key historical process is no longer the movement itself but the gathering-in of new recruits afterwards. There is a beautiful symmetry here. The old Grand Narrative subdued archaeology to the demands of history, with archaeological cultures that were understood as ‘peoples’ and a migration model derived from first-millennium historical sources which ordered the progression of these cultures into a historical narrative punctuated by episodes of large-scale migration and mass ethnic cleansing. Now, the credibility of these same historical sources has been undermined by a reaction against migration which started with the archaeologists’ ferocious rejection of culture history and the invasion hypothesis that was its natural corollary. History used to lead archaeology; now archaeology is leading history. In the process, a vision of early European history driven by outside emigration has given way to another characterized by few immigrants but by many people adapting to whatever stimuli were provided by the few who did move: a story largely of internal development. This is in its own right a beautiful pattern. We have now reached a point that is the mirror image of where we were fifty years ago. But while this is satisfyingly symmetrical as an intellectual progression, is it convincing history? Should migration be relegated to such a minor, walk-on part in the history of barbarian Europe in the first millennium AD? MIGRATION AND INVASION The invasion hypothesis is dead and buried. No longer would we even want to litter prehistoric and first-millennium Europe with a succession of ancient ‘peoples’ carving out their chosen niches via a lethal cocktail of large-scale movement and ethnic cleansing. Arguably, such a cocktail should never have existed. At least the ethnic-cleansing element of the old Grand Narrative finds little support that I know of in the sources. The demise of the invasion hypothesis does not mean, however, that migration has entirely disappeared from the story. Nor could it. Even if you accept that a migration topos operated among Mediterranean authors, their cultural fantasies would still have had to be underpinned by population movements of some kind, and some of the archaeological evidence is likewise suggestive of humanity somehow on the move. Two alternatives to the invasion-hypothesis model of mass migration have consequently come into use. The first is the ‘wave of advance’ model. Applicable to small migration units, it provides an alternative view of how a group of outsiders might take over a landscape. It has been applied in particular to the spread across Europe of its first proper farmers in the Neolithic period, and shows how, even with individually undirected moves, farming populations might nonetheless have come to dominate all suitable points in that landscape. According to this model, Neolithic farmers did not arrive en masse and oust the hunter-gatherers in an invasion. Rather, the farmers’ capacity to produce food in much greater quantities meant that their population numbers grew so much more quickly that, over time, they simply swamped the hunter-gatherers, filling up the landscape from the points nearest the first farming sites, as individual farmers grew to maturity and sought their own lands. It is a model for small-scale, family- or extended-family-sized moves and unintentional takeover, which, by virtue of these qualities, also allows for the possibility that some of the indigenous hunter-gatherers might have learned farming skills for themselves as the process slowly unfolded. What could be more attractive for scholars trying to free themselves from a world of mass moves and conquest? Even more popular among archaeologists, because of its greater range of potential applications, is the ‘elite transfer’ model. Here, the intrusive population is not very large, but does aggressively take over a territory by conquest. It then ousts the sitting elite of the target society and takes over its positions of dominance, while most of the underlying social and economic structures which created the old, now expelled or demoted, elite are left intact. The classic example of this phenomenon in medieval history is the Norman Conquest of England, where, because of the astonishing wealth of information surviving in Doomsday Book, we know that a few thousand Norman landholding families replaced their slightly more numerous Anglo-Saxon predecessors at the top of the eleventh-century English heap. Again the vision of migration suggested by this model is much less dramatic than that envisaged under the invasion hypothesis. It retains the latter’s intentionality, and some violence, but because we’re talking only of one elite replacing another, with broader social structures left untouched, this is a much less nasty process than the ethnic cleansing that was central to the old model. And because it is merely a question of swapping a few elites around, the outcome is likewise much less dramatic and in one sense less important, since all the main existing social and economic structures are left in place, as they were in England by the Norman Conquest. The intellectual response to the oversimplicity of the invasion hypothesis has thus taken the form of developing two models which in different ways minimize the importance of migration, whether by cutting back on the likely numbers involved, the degree of violence, the significance of its effects or, in one of the two, the extent to which there was any real intent to migrate-cum-invade at all. These models are obviously much more compatible than the invasion hypothesis with those visions of group identity that deny that large, compact groups of humanity could ever intentionally move as a cohesive block from one locality to another. But while these models are certainly more sophisticated, and are to that extent a step in the right direction, they do not yet, even in combination, add up to a satisfactory overall approach to migration in first-millennium Europe. Confining discussion to a framework supplied by just these two models involves three specific problems, and one much more general one. Mistaken Identity? The first problem stems from the fact that in their excitement that human beings do not always organize themselves in self-reproducing, closed population groups (and, I think too, in their determination to banish for ever the abominations of the Nazi era), historians and archaeologists of the first millennium have tended to concentrate on only one half of contemporary discussions of identity in the social-scientific literature. At the same time as Leach, Barth and others were focusing on group behaviour and observing individuals swapping allegiance according to immediate benefit, a second group of scholars turned their attentions to the close observation of individual human behaviour. These have sometimes been called ‘primordialists’, because they argue that group affiliations have always been a fundamental part of human behaviour. Some of these studies seemed to come up with different conclusions from those generated by Leach and Barth in that they showed that, in some cases, inherited senses of group identity apparently cannot be manipulated at will, but constrain individuals into patterns of behaviour that go against their immediate interests. Differences in appearance, speech (whether language or dialect), social practice, moral values and understandings of the past can – once they have come into existence – act as formidable barriers to individuals who might wish, for personal advantage, to attach themselves to a different group. The two lines of research have sometimes been held to contradict each other, but in my view they do not. They actually define the opposite ends of a spectrum of possibility. Depending upon particular circumstances, not least past history, inherited group identities can exercise a more or less powerful constraint upon the individual, and provide a greater or lesser rallying cry to action. Again, this is firmly in line with observable reality. In terms of larger group identities now, the rhetoric of Britishness strikes a much stronger chord in the United Kingdom in contemporary debates about the EU, for instance, than does, say, Luxembourgeoisness in its home corner of Europe, neatly located between Germany, France and Belgium. And so too at the level of the individual: individual members of any larger group show marked differences in their levels of loyalty to it. Accepting the fact that group identity is sometimes a stronger and sometimes a weaker force in people’s lives does not, I would stress, really contradict what Barth had to say (even though he might think it did). His famous aphorism is that identity must be understood as a ‘situational construct’. Fair enough, but a crucial point is that all situations are not the same. Influenced in part by the old Marxist dogma that any identity that is not class-based (as group identities will not be, unless every member has the same status) must be ‘false consciousness’, and partly by the fact that he was primarily reacting against a world dominated by nationalist ideologies, Barth stressed, and was most interested in, the kinds of situations that produced weak group affiliations. But even the logic of his own phrasing implicitly allows that there might be other situations that produced stronger types of group affiliation, and the so-called primordialist research has explored some of them. Two entirely different types of constraint can act as barriers. On the one hand, there are the informal constraints of the ‘normal’, whether we’re talking food, clothing, or even moral values. Research has suggested that the individual picks up many of these group-defining characteristics in the earliest years of life, which helps explain, of course, why they might sometimes have a profound effect, making individuals feel so uncomfortable outside the norms of their own society that they cannot happily live anywhere else. On the other hand, and sometimes operating alongside such senses of discomfort, there can also be much more formal barriers to changing identities. As an individual, you can in theory claim any identity you want to, but that doesn’t mean it will be recognized. In the modern world, group membership usually means having the appropriate passport, and hence the ability to satisfy the criteria for obtaining it in the first place. In the past, of course, passports didn’t exist, but some ancient societies monitored membership carefully. Rights to Roman citizenship were jealously guarded, for instance, and a whole bureaucratic apparatus was set up to monitor individual claims. Greek city states had earlier followed similar strategies. Such bureaucratic methods relied on literacy, but there is no reason why non-literate ancient societies might not also have controlled membership closely in certain conditions. There can also be degrees of group membership. America and Germany, in the modern world, have more and less officially accepted large groups of foreign workers without necessarily giving them full citizenship rights, and herein lies the key, in my view, to a total understanding of the identity question. When full group membership brings some kind of legal or material advantage – a set of valuable rights, in other words – we should expect it to be closely controlled. The underlying conclusions to emerge from the identity debate are more complex, therefore, than has sometimes been realized. For individuals born into all but the simplest of contexts, group identity comes in layers. Immediate family, wider kin, town, county, country, and these days international affiliations (such as citizenship of the EU), together with their own life choices – the desire, for instance, to live somewhere else entirely – all provide the individual with possible claims to membership of a larger group. But any claim he or she makes does have to be recognized, and, according to context, these possible affiliations might exercise a more or less powerful hold upon them. Essentially, Barth’s famous aphorism sets up a false contrast. All group identities are ‘situational constructs’ – they are created, they change, they can cease to exist entirely – but some are more ‘evanescent’ than others. From this follows a first potential problem in current approaches to migration in the first millennium. They are predicated on the supposition that large-group identity is always a weak phenomenon, but this is only a half-understanding of the identity debate. If a position on identity is adopted a priori – whether it is viewed as strong (in the era of nationalism) or weak (in the currently emerging consensus) – then evidence to the contrary will be ignored or argued away. To my mind, it is important to be willing to re-examine the evidence for migration in the first millennium without assuming that the population groups involved will necessarily have been bound together so weakly as some of the current half-understandings of the group identity issue would suppose. The second problem emerges when the virulent rejection of migration as a possible agent of past change among some English-speaking archaeologists is set against the kinds of archaeological reflection of migration that is likely to turn up in practice. It is not usual in the modern world for entire social groups to move in a block, and, as we shall see in the chapters that follow, this was also true of the period being explored in this study. There is in fact little or no evidence of first-millennium ethnic cleansing. First-millennium migration almost always consisted, therefore, of moving part of a population from point A to point B, with at least some of the latter’s indigenous population remaining in situ, the only exception being Iceland which was unoccupied when the Norse arrived there in the ninth century. This being so, you can never expect to find the complete transfer of an entire material culture. Rather, only certain elements of the old material culture would be likely to be brought to point B: those invested with particular meaning, perhaps, for the subgroup of the migrant population actually involved in the migration process. At the same time, some or much of the indigenous material culture of point B would probably continue, and some entirely new items or practices might be generated by the interaction of the migratory and host populations. The archaeological reflections of many first-millennium migratory processes, in other words, will often be straightforwardly ambiguous in the sense that you could not be absolutely certain, just on the basis of the archaeology alone, that migration had occurred. So far so good: if the only archaeological evidence for a possible migration is ambiguous rather than definitive, so be it. Better that than populate European history with a series of phantom invasions. Where this does become a problem, however, is when migration is viewed as ‘always simplistic’ and ‘usually groundless’. If you approach the issue in this frame of mind, then the ambiguity of the evidence will not be treated in an even-handed fashion. Where you’re looking at some archaeological transformation which might or might not represent the correlates of a migratory process, then it is important to say exactly that – no more and no less. But because archaeologists have just gone through such a nasty divorce from migration, some have a strong tendency (at least in Britain and North America) to want to write it out of their accounts of the past entirely. It is now enough in some quarters to show that an observable transformation might have been generated without migration for this to be taken as a proven fact. But since the archaeological reflections of many migration processes will only ever be ambiguous, the basic fact that just about every kind of archaeological transformation can, with sufficient intellectual ingenuity, be explained in terms other than of migration, doesn’t mean that it should be. The right answer is not to say that, because there is ambiguity, migration has been disproved, but to accept the ambiguity and see if anything else – especially historical evidence where appropriate – helps resolve it. It is not safe, then, either to build your estimate of the potential scale of first-millennium migration on the presumption that group identities were always weak, or to dismiss its existence and importance if you find only ambiguous archaeological evidence. These two observations in turn generate the third problem. The concept of a migration topos – the idea that Mediterranean writers were led by a cultural reflex to see any barbarians on the move as a ‘people’ – has sometimes been used to dismiss historical evidence for large, compact and mixed migration groups. Up to this point, however, its supposed prevalence is based on assertion rather than on any properly argued demonstration that it really existed. As a concept, it has gained a priori plausibility from the idea that group identities could never have been strong enough to generate the kind of large-group migration that the sources seem to be reporting, and from the fact that, as already noted, the archaeological reflections of migration are often ambiguous. But if archaeological ambiguity is only to be expected, and it is unsafe just to assume that all first-millennium group identities were necessarily weak, this obviously undermines the support these points have been supposed to provide for the supposed existence of a migration topos. So it will be necessary in what follows to examine on a case by case basis whether the historical accounts of large-group migration can really be dismissed so easily. Even by themselves, these three problems would be sufficient to warrant a re-examination of migration in the first millennium. But there is also a fourth, and much broader, reason why current treatments of the topic require a thorough overhaul. Migration and Development The comparative study of human migration has a lengthy pedigree. Like many other fields, it has proceeded from originally simple models to more complex and interesting ones, particularly in the last scholarly generation or so. Interest originally focused upon economic motives as the paramount factor in explaining population movements, with a landmark study arguing pretty successfully that immigration to the United States was positively correlated with its business cycles. The quest to understand first-millennium migration has seen some engagement with this rapidly developing field. When thinking about causation, for instance, the concept of ‘push’ and ‘pull’ factors – things that were bad about a point of departure and attractive about the destination – has long been part of the scholarly vocabulary. The importance of accurate information in shaping migration flows, and the fact that larger-scale migration is sometimes preceded by pioneering individuals (‘scouts’) whose experiences add momentum to what follows, are likewise part of the landscape. But these ideas are no more than the tip of the comparative-migration iceberg and, in general terms, the literature has been little explored by those studying migration in the first millennium. This is a strange omission because the comparative literature offers a wide range of well-documented case studies against which to compare the first-millennium evidence, with an obvious potential to expand the range of possible migration models beyond the limits of wave-of-advance and elite-transfer. Amongst other examples, more recent history gives us economically driven flows of migrants, who are unorganized in the sense that all are making individual decisions. Nonetheless, they can over time, and especially when allied with population increase among those who have already reached the point of destination, fill an entire landscape: even one as big as the United States. The twentieth century has also underlined the importance of another basic cause of migration: political conflict. Individual refugees fleeing persecuting regimes are extremely common, but political disturbances can also generate much more concentrated migration flows. The most horrific example from recent years is Rwanda, where this chapter began. But there are many others: ethnic cleansing in the former Yugoslavia, the expulsion in just three months of eighty-eight thousand foreigners from Saudi Arabia in 1973, the movement of twenty-five million refugees in central and eastern Europe at the end of the Second World War, the flight and continued plight of Palestinian refugees. Aside from expanding the underlying intellectual frame of reference, the comparative literature also indicates that it is necessary to ask more detailed questions of any migratory process than has customarily been done in first-millennium studies. Early modern and modern case studies have thrown up no instance where the entire population of place A has moved en masse to place B. Migration has always turned out to be an activity confined to certain subgroups, and a particularly fruitful line of questioning has stemmed from this observation. What leads some individuals to stay at home, when their fellows in more or less identical circumstances move? Work directed at understanding this phenomenon has identified some interesting patterns. Economic migrants tend – certainly in the first instance, at least – to be younger, often male and, in terms of their own societies, relatively better educated. Migration also tends to be undertaken by the already mobile. On closer inspection, half of the Dutch migrants to what became New York turn out to be people who had already migrated once before, from other parts of Europe to the Netherlands. Likewise, many of the ‘Irish’ participating in the early stages of the colonization of North America came from Scottish families, which, just a generation before, had moved to Ireland. Longer-distance migration flows have always to be understood, therefore, against established patterns of internal demographic dislocation. Participants in the latter will have a greater than average likelihood of providing manpower for the former. Even within these variegated patterns of participation, however, the decision to migrate does not turn simply upon what you might term rational economic calculation. Other factors complicate the individual’s thought process. Information about both projected destinations and the routes to them is one key variable. Large-scale migration flows to a new destination only begin once the pros and cons of the route, and of the potential new home, become generally understood. Before that stage, ‘channelled’ migration is correspondingly common. Under this pattern, population groups from relatively restricted departure areas end up clustered together again in specific areas at the point of destination. This seems to be caused both by limitations on the amount of available information, and by the kind of social support that can be provided by a host population from the migrants’ point of departure. Transport costs, not surprisingly, also intrude into a potential migrant’s calculations, and psychological costs are important too. The strangeness of life in a new place and the disruption to emotional ties binding the individual to family and friends affect decisions to move, as well as subsequent decisions about whether to remain. A substantial flow of return migration is thus a significant feature of all well-documented population displacements. Over and above all these factors, potential migration flows can be interfered with by the political structures in existence at either the point of departure or that of arrival, or both. Since the 1970s, Western European countries have more or less brought to a halt the flows of legal migrant labour from particular parts of the Third World, which had been a regular feature of life since the Second World War. This decision was motivated by political rather than economic considerations, since industry still wanted the relatively cheap labour that migrants provide, but governments were concerned to pacify the hostility towards migrant communities that had grown up in some quarters of their own societies. Migration flows from the old sources have continued, in fact, but in the greatly modified form of family reunification, not new migrant workers, and there has followed a corresponding shift in gender and age patterns among the migrants. Flows of women and the relatively elderly, wives and dependent parents of the original migrants, have replaced the procession of young men. This is but one example of the general rule that political structures will always dictate the framework of available options within which potential migrants make their decisions. Migration studies also offer new ways of thinking about the effects of migration, of how to form some estimate of whether to rate it a more or less important phenomenon in any particular case. Thanks to the legacy of the invasion hypothesis, these kinds of argument in the first-millennium context are now often wrapped up with the issue of migrant numbers. Are we looking at ‘mass migration’ or at a smaller phenomenon, something more like elite transfer? – with estimates of a migration flow’s importance being adjusted up or down according to the numbers involved. But since first-millennium sources never provide unquestionable data on numbers, even when there’s any at all, it is hardly surprising that such arguments often become deadlocked. Of potentially wide application, therefore, is the relative, rather than statistical, definition of mass migration generally adopted in the comparative-migration literature. For what, in fact, constitutes a ‘mass’ migration? Is it the arrival of an immigrant group that numbers 10 per cent of the population at the point of destination? – 20 per cent? – 40 per cent? – or what? And a migration flow needs in any case to be considered from the viewpoint of all its participants. Theoretically, a flow of migrants might amount to a small percentage of the population at its point of destination, but represent a large percentage of the population at its point of departure. What is elite transfer from the host population’s perspective, therefore, could be a more substantial demographic phenomenon for the immigrants themselves. To encompass this variety of situations and avoid numerical quibbling, migration studies have come to define ‘mass’ migration as a flow of human beings (whatever the numbers involved) which changes the spatial distribution of population at either or both the sending and the receiving ends, or one ‘which gives a shock to the political or social system’, again at either end or both. This is not just to assume that information and insights from more modern eras are automatically applicable to the first millennium. Migration studies have generally been working with twentieth-century examples, observed more or less contemporaneously, or with the European settlement of the Americas, either North and South in the first phase from the sixteenth to the eighteenth century, or just the North in the case of the huge immigration waves of the later nineteenth and early twentieth. There are major structural differences between any of these worlds and first-millennium Europe. The latter’s economy was overwhelmingly agricultural in nature, and at subsistence, or not far above, in its levels of output. It had no mass production, so that nineteenth- and twentieth-century patterns of migrant labour being sucked first from agricultural to industrial Europe and then from outside Europe altogether simply do not apply. The population of first-millennium Europe was also smaller than its modern counterpart to a quite astonishing degree, and even as late as 1800 governments of European countries tended to control emigration much more than immigration. The governmental and bureaucratic capacities, likewise, of first-millennium states (to the extent that there were any) were also much less developed, so that they clearly did not have the same capacity to make and enforce immigration policies as their more modern counterparts. Similarly with transport and the availability of information. Both existed in the first millennium, but transport costs were huge compared with the modern world. Perhaps the most famous economic statistic from the ancient world is the report in the Emperor Diocletian’s Edict on Prices (from c.300 AD) – that the cost of a wagon of wheat doubled for every fifty miles it was carried. Where transport remained expensive, as it did down to the later nineteenth century, this posed substantial problems to would-be migrants, although these could sometimes be obviated by state assistance. Information in a pre- or non-literate world also circulates over very different (that is, shorter) distances, and in an entirely different fashion from a world with mass media, again making it more difficult for would-be migrants to gather information about possible destinations. In the high Middle Ages, this was sometimes countered by designated agents mounting recruiting drives, but the limitations that would have affected information flows in the first millennium are obvious. Nonetheless, and at the very least, modern migration studies generate afresh range of issues and more detailed questions to move the study of first-millennium migration well beyond the old invasion-hypothesis model and even beyond current responses to that model. It is on the issue of what causes migration, however, that the modern world has most to teach those of us who grapple with the first millennium. On the level of the individual migrant, comparative analysis has moved far beyond drawing up lists of push-and-pull factors. There are two basic drivers behind migration: more voluntary economically motivated migration, and less voluntary political migration. But a hard and fast distinction between economic and political migration is usually impossible to maintain. Political reasons may come into a decision that appears economic, since political discrimination may underlie an unequal access to resources and jobs. The opposite is also true – that economic motives can be bound up in an apparently political decision to move, if not quite to the extent that a sequence of British Home Secretaries have sought to maintain. In any case, economic pressures can be as constraining as political ones. Is watching your family starve to death because you have no access to land or a job an economic or a political issue? These complexities mean that a potential migrant’s decision-making process now tends not to be analysed in terms of push-and-pull factors, but modelled as a matrix whose defining points are on one axis economic and political, and on the other voluntary and involuntary, with each individual’s motivation usually a complex combination of all four elements. In general terms, would-be migrants can be understood as facing a kind of investment choice. The decision to migrate involves various initial costs – of transport, of lost income while employment is sought, of the psychological stress of leaving the loved and familiar – which have to be weighed against possible longer-term gains available at the projected destination. Depending upon personal calculation, the individual might choose to leave or to stay, or to leave temporarily with a view to making enough of again to render a return life in the home country much more comfortable (another major cause of return migration). All this is enlightening and challenging in pretty much equal measure, but at the macro level, migration studies have a still more profound lesson to offer. Not least because politics cannot be easily separated from economics anyway, economic factors remain one of the fundamental triggers of migration. Disparities in levels of economic development between two areas, or in the availability of natural resources, have been shown repeatedly to make a migration flow between them likely, so long, of course, as the immigrant population also values the commodity which is more available at the point of destination. This is a fundamental conclusion of so-called ‘world systems theories’, which study relations between economically more developed centres and less developed peripheries, where some migration between the two often proves to be a major component of the relationship. This key observation tells us two things. First, a satisfactory study of migration in any era will require a combination of more general analysis (such as the basic economic contexts making migration likely) with the answers to a series of precise questions: who exactly participated in the flow of migration, why, and how exactly the process began and developed? Second, and even more important, it emphasizes that there is a profound connection between migration and patterns of economic development. Because of the legacy of the invasion hypothesis, it is traditional in first-millennium studies to draw a clear dividing line between internal engines of social transformation, such as economic and political development, and the external effects of migration. For a generation and more of archaeologists since the 1960s, internal transformation has been seen as locked in a death struggle with migration when it comes to explaining observable changes in the unearthed record of the past. Given this particular intellectual context, the most fundamental lesson to be drawn from migration studies is that such a clear dividing line is misconceived. Patterns of migration are caused above all by prevailing inequalities in patterns of development, and will vary with them, being both cause and effect of their further transformation. In this light, migration and internal transformation cease to be competing lines of explanation, but two sides of the same coin. Old ways of thinking about the first millennium generated one Grand Narrative of how a more or less recognizable Europe emerged from the ancient world order of Mediterranean domination on the back of a thousand years of invasion and ethnic cleansing. New information and, not least, new understandings of both group identity and migration have effectively demolished that vision, and it is time to replace it with something new. It is this central challenge that Empires and Barbarians will attempt to take up, arguing above all that migration and development need to be considered together, not kept apart as competing lines of explanation. They are interconnected phenomena, which only together can satisfactorily explain how Mediterranean domination of the barbarian north and east came to be broken, and a recognizable Europe emerged from the wreck of the ancient world order.