No surviving contemporary source describes the population movements associated with the Marcomannic War in any detail, but we do have one account of third-century Gothic migration preserved in the sixth-century Gothic history of Jordanes, himself partly of Gothic origins. He describes the move of the Goths to the Black Sea:
When the number of the people increased greatly and Filimer, son of Gadaric, reigned as king . . . he decided that the army of the Goths with their families should move from that region [beside the Baltic]. In search of suitable homes and pleasant places they came to the land of Scythia, called Oium in that tongue. Here they were delighted with the great richness of the country, and it is said that when half the army had been brought over, the bridge whereby they had crossed the river fell in utter ruin, nor could anyone thereafter pass to or fro. For the place is said to be surrounded by quaking bogs and an encircling abyss, so that by this double obstacle nature has made it inaccessible. And even today one may hear in the neighbourhood the lowing of cattle and may find traces of men, if we are to believe the stories of travellers, although we must grant that they hear these things from afar. This part of the Goths, which is said to have crossed the river and entered with Filimer into the country of Oium, came into possession of the desired land, and there they soon came upon the race of the Spali, joined battle with them and won the victory. Thence the victors hastened to the farthest part of Scythia, which is near the Black Sea.28
Jordanes is reporting here pretty much a textbook example of the invasion hypothesis in action. One king and one people move en masse to a new home, defeat the indigenous occupants and take possession of the land. How much of this, written down nearly three hundred years later, bears any relationship to third-century realities?
The Flow of Migration
Enough information survives in more contemporary sources to demonstrate that the migration processes of the time were far more complicated than Jordanes’ much later account suggests. For one thing, a whole series of Germanic groups, not just Goths, were involved in the action. More interestingly – and this is a much more substantial departure from his vision – the Goths and other participants in the migration flow did not operate as united, compact entities along the lines of Jordanes’ one king/one people model. This point is best illustrated, in fact, from the Goths, where the more contemporary evidence is fullest. In these sources, Gothic groups are found operating in different ways over a wide geographical area: by land and sea everywhere from the mouth of the Danube, where the Emperor Decius was killed, to the Crimea (a distance of nearly a thousand kilometres) and beyond. Consonant with this highly dispersed action, a whole series of individual Gothic leaders feature: Cniva, Argaith, Guntheric, Respa, Veduc, Thuruar and Cannabaudes. Some appear in alliance with one another, but no overall king of the Goths is ever referred to in reliable contemporary sources, and emphatically not Jordanes’ Filimer.29
The end result of all this disparate third-century activity was also the creation not of one fourth-century Gothic kingdom, which would be the natural consequence of a single well-organized land-grabbing exercise, but several. It is often supposed, again on the strength of Jordanes, that the events we were discussing generated two major Gothic political entities north of the Black Sea: the Visigoths and the Ostrogoths. But Jordanes has retrospectively imposed the Gothic political patterns of his own sixth century on the fourth-century past. Visigoths and Ostrogoths, groupings who formed successor states to the west Roman Empire in the fifth century, were both demonstrably new creations of that century, formed under Roman eyes and on Roman soil, as we will see in due course. No contemporary source gives us a complete survey of the Gothic-dominated north Pontic world of the fourth century; life is never that convenient. But, in the fifty years or so after c.375, at least six major concentrations of Goths appear as independent actors in entirely contemporary historical sources. Each of these is likely to have derived from a politically independent Gothic unit of the fourth century, suggesting that we should be thinking of half a dozen or even more Gothic political entities rather than just two. This is entirely in line with what you would expect from the highly diverse nature of Gothic activity in the third century. And while this point is best documented for the Goths, it applies to the other groups as well. In the great sea raid of 268/9, the participating Heruli divided in two separate groups – one operating with Goths in Attica, the other besieging Thessalonica in Macedonia. Third-century migration was not remotely as simple as Jordanes’ formula – one king, one people, one move – might suggest.30
Third-century patterns also departed from the invasion-hypothesis model on another level, with previously unknown Germanic groups appearing in our sources for the first time. Heruli, Gepids and Taifali all make their historical debuts in third- or very early fourth-century sources. It is possible that they had previously existed and just been overlooked, but, on the simple level of naming names, the first- and second-century listings of Tacitus and Ptolemy seem pretty comprehensive, so that there is some reason to think their silence here significant. Nor is it surprising to find new Germanic groups being created in the course of these tumultuous events. New groups – the Alamanni and Iuthungi – were appearing in the west at precisely this time, and we do know that Germanic groups came and went. Like any human organization, they could be created and destroyed, and the information we have would seem to indicate that the political patterns of Roman-period Germania were pretty fluid. For the first century, for instance, Tacitus describes the creation of the Batavi. Originally a part of the Chatti, they broke away from the main group, acquired their own name and subsequently pursued a separate historical course. His work also recounts the effective extermination at different times of three other groups: the Ampsivarii, the Chatti (showing how sensible it was of the Batavi to have broken away) and the Bructeri. All of this makes it entirely plausible that Gepids, Heruli and Taifali were new units of the third century.31
Even Jordanes, in fact, preserves an echo of this more complex reality. All his accounts of Gothic migration incorporate a strong motif of sociopolitical fragmentation. In the Filimer migration a bridge falls down, parting some of the Goths from the main body. Elsewhere, he tells the story of a previous Gothic migration in three ships from Scandinavia. In that case too, one of the ships lags behind, and out of this separation, so Jordanes tells us, were born the Gepids. Despite the extreme scepticism in vogue in some quarters, there is actually a good chance that both of these stories echo, if at some remove, Gothic oral histories. If so, those histories, while tending to describe migration in terms of kings and peoples, nonetheless preserved something of the deeper reality – that political discontinuity, rather than the uncomplicated transfer of entire pre-existing social units from point A to point B, was a central feature of the action.32
On one very simple level, the archaeological evidence also reflects this basic fact. Although the extensive similarities between the Wiel-bark and Cernjachov systems can reasonably be taken to reflect a substantial transfer of population between the two, the Wielbark system itself did not disappear but continued to exist down to the fifth century in broadly its old haunts to the north-west of what became Cernjachov territory. Archaeologists have also begun to identify a number of intermediate material cultural systems, placed geographically between the two. Discussion continues as to whether to view these as entirely separate from the two main systems, or as local variants of one or the other, and any temptation to identify them instantly with any of the groups named in third- and fourth-century sources needs to be resisted. Material cultural boundaries might reflect political boundaries, but, as we have seen, it cannot just be assumed that they do. However interpreted – and it may be that, under closer scrutiny, the whole Cernjachov system will eventually be recategorized into a series of interrelated regional groupings – the 
group (generated c.180–220 AD) and the Ruzycankan and Volhynian groups (generated c.220–60) make it very clear that the material culture generated by the migrants, echoing the new political order, was distinctly non-monolithic. Not all the Wielbark groups involved in the general move south shared in the same outcome. Some followed one historical trajectory which led to their involvement in the creation of the Cernjachov system and the other new groupings, others continued more or less as before but in a new environment, and some chose not to move at all.33
The third-century migrations were carried out, therefore, not by total population groups but by a series of subgroups, each operating to some extent independently of one another, very much replicating the pattern of many modern migration flows (Chapter 1). Some of the movements associated with the Marcomannic War were probably similar. The attack on Pannonia which opened the war proper clearly did not involve all the Langobardi. Langobardi in large numbers moved definitively into the same Middle Danubian region only some three hundred and fifty years later, in c.500 AD, and most had probably continued to live in the northern Elbe region in between. The same was substantially true of Germanic migrations in the third-century west. Here we have even less narrative evidence, but the archaeology shows very clearly that the Agri Decumates were not occupied in one fell swoop. As we saw in the last chapter, political power remained devolved among the fourth-century Alamanni, and this probably reflects this earlier period when groups moved into the new landscape piecemeal. Some, it seems, moved in soon after the Romans abandoned the territory in c.260, but elsewhere the process was much slower. Elbe–Germanic materials superseded Rhine–Weser materials on the Middle Main, for instance, only in the early fourth century, the best part of two generations later.34 In both east and west, therefore, the third century saw fragmented, diverse flows of migration rather than massive land seizures by ‘whole’ peoples. But how exactly should we envisage the population groups who undertook the migrations?
Some of the migrating subgroups were warbands – relatively small groups of a few hundred young men under the leadership of a particularly renowned warrior – on the make. The creation of small organized armed groups (such as that immortalized at Ejsbøl Mose) was a characteristic feature of Germanic society in the Roman period, some led by kings and some of them more egalitarian associations. Hence it is no great surprise that some of the archaeological remains hint at the participation of these kinds of group in the third-century action. East of the Carpathians, a few cemeteries have been unearthed from the early Cernjachov era – Cozia–Iasi, Todireni and Braniste – where, contrary to normal Cernjachov and Wielbark practice, the dead were buried with weapons. All the other equipment found would suggest that the groups interred in these cemeteries were Germanic intruders from the north. The presence of weapons, however, suggests that they originated somewhere outside the Wielbark system, probably from within Przeworsk areas further to the south. The cemeteries are not large, and would be entirely in accord with a picture of small armed Przeworsk groups seeking their fortune.35 It would be very interesting to have a full study of the age and gender of the populations found in the ribbon of Wielbark cemeteries that stretch along the Upper Vistula and Dniester. These too might reflect small migratory subgroups similarly skewed in age and gender, rather than a more normal cross-section of humanity. Much of the action in the west is also compatible with this kind of picture, especially since the Agri Decumates were not occupied at one go.
But not all third-century activity is explicable in terms of small groups of a few hundred. The Gothic leader Cniva could not have defeated the Emperor Decius, however restricted the area of his imperial rule, had not the king’s armed following numbered thousands rather than hundreds. The Goths and Heruli defeated around Thessalonica by Claudius are said to have lost several thousand men in the battle. You can obviously doubt the precise accuracy of these figures, but Claudius clearly had a major fight on his hands, and the great sea raid of 268–71, of which this encounter was a part, could not have done so much damage had its component forces been appreciably smaller than losses in the thousands would suggest.36 The evidence from the Marcomannic War is similar. Some of the action can be explained in terms of warbands, but not all of it. There is Dio’s report, for instance, that the Langobardi and Ubii between them mustered six thousand men for their initial attack on Pannonia, and there was a moment when the Quadi, seeking to escape from Marcus Aurelius’ punitive restrictions, were preparing to ‘migrate in a body to the land of the Semnones’ which lay further north between the Elbe and the Oder.37 The Romans prevented this projected move with countermeasures of their own, and we cannot be sure that every single member of the group was about to head off north, but the evidence certainly suggests that Germanic groups numbering several thousand could contemplate hitting the road.
That some groups of migrating Germani, at least, were substantial in number is also indicated by the unfolding pattern of events at their points of destination. The Goths and others who made the trek to the Black Sea, for instance, did not operate there in a vacuum. In 238, after their assault on Histria, the Romans granted the attacking Goths an annual subsidy on condition that they withdrew from the city and returned prisoners. This provoked a howl of protest from the local Carpi, who claimed to be ‘more powerful’ than the Goths. The Carpi, as we have seen, were a group of so-called free Dacians established in the Moldavian hinterland of the Carpathians, semi-subdued clients who had not been brought under formal imperial rule. The expansion into the frontier zone of Goths and other Germanic-speakers brought the migrants into competition with these Dacian groups. And, over time, Gothic power in the region grew directly at the Carpi’s expense. In the end, the Carpi lost out completely. Their political independence was totally dismantled, with large numbers – hundreds of thousands of them, according to Roman sources – being resettled inside the Empire either side of the year 300.38 Again, precise figures can be doubted, but not the overall picture. The Carpi disappear as an independent political force from the early fourth century, and we have explicit evidence that they were resettled south of the Danube. Likewise, there is not the slightest doubt that Germanic-speaking Goths replaced native Dacian-speakers as the dominant force around the Carpathian system.
It was, as we have seen, a well-established imperial response to competition in the frontier zone to thin it out by taking some suitably cowed immigrants into the Empire. The reception of the Naristi had been part of the solution to the Marcomannic War, Constantius had been ready to do the same with some Limigantes in 359, and there is no reason to doubt the reports that clearing out large numbers of Carpi from the frontier zone about the year 300 was part of the solution to the new problems of the third century. Nor were the Carpi the only losers. Further east, Germanic immigrants subdued the Sarmatian kingdoms and the old Greek cities of the Pontus, and an additional effect of their arrival was to make the Empire evacuate upland Transylvania.39 Not all the Carpi were transferred south of the Danube, and much of the indigenous population of Transylvania and the Pontic littoral remained in place. Nonetheless, large-scale resettlements and a total reshaping of the strategic situation in the region are clear signs that Rome’s existing arrangements for frontier security had been undermined by what was a major intrusion into the region on the part of a non-indigenous, Germanic-speaking population. For all of this, we do have to be talking of Germanic-speaking groups who could put several thousand fighting men in the field at any one time. Groups numbering just a few hundreds could never have achieved so much.
The pattern in the west was not quite the same. There are no records of conflict on such a large scale, and because the Alamanni were taking possession of abandoned territory in the Agri Decumates, they did not face the same imperative of having to oust sitting tenants. But that still doesn’t mean that all the action was very small-scale. Outside the Agri Decumates, the Alamanni did impose themselves over other indigenous Germani, such as the Rhine–Weser groups who eventually lost out to them on the Middle Main. This may well have required more consolidated group action. Likewise, as we will explore in more detail in a moment, the Burgundians came in sufficient numbers to preserve their own, distinct east Germanic dialect. Additionally, Burgundians and Alamanni periodically competed with one another in the fourth century, and this could easily have begun already in the third. If so, it will have been another factor pushing the Alamanni into more concerted, group, action – something that is entirely in line with the evidence we have for the whole emergence of their group dynamic.
On the one hand, the Alamannic confederation was the result of a long-drawn-out political process. When we first meet Alamanni in the third century, for instance, the Iuthungi were not part of the confederation. But by the mid-fourth, they were: one among the several cantons among whom protocols of under- and overkingship seem periodically to have operated (Chapter 2). This process had begun early in the third century. At one point, it was trendy to argue that the first convincing mention of the Alamannic confederation could not be dated before the 290s. It was then natural to argue that the third-century raiding and land-grabbing on this sector of the Rhine had been conducted by independent warbands, who started to form larger group structures only after seizing theAgri Decumates. But this dating was much too late. The Emperor Caracalla was already fighting Alamanni as early as 213. And while the Alamannic confederation did not at this point incorporate all the subgroups who would be part of it in the fourth century, this does suggest that major political reconfiguration was already under way right at the beginning of the third century, which in turn makes it necessary to think of the Alamanni as more than a collection of warbands even at this point.40 This being so, the action in the west was probably quite similar to that unfolding simultaneously east of the Carpathians, involving some larger-scale groups as well, certainly, as warbands.
There is probably a more general logic, in fact, to how such patterns of armed expansion tend to unfold, because the eastern evidence is reminiscent in some key respects of the better-documented flow of Norse expansion into western Europe in the ninth century. Here too, the action started small. The earliest recorded incident involved just three boatloads of Norwegians causing trouble on the south coast of England around the year 790. It stayed small-scale for about a generation and a half, but then grew as larger confederate bands began to operate in western waters from the 830s, some of them led by ‘kings’ or ‘jarls’, men who were already important in Norse society. The confederative tendency then reached a climax from the 860s in the great army era, when several of the larger groupings began to combine in new ways to achieve ends that required the application of still greater levels of force. In the case of the Vikings, the end in sight was defeating the armies of Anglo-Saxon and Frankish kingdoms. All this strongly recalls the patterns of Germanic expansion visible in the third century. It may well have begun with small-scale raiding, but sacking Roman cities, defeating Roman emperors and appropriating the assets of existing frontier clients all required a much greater level of force, leading, as in the Viking case, to the evolution of new confederations among the migrants so as to generate forces of appropriate size for the new ventures.41
The extent to which groups of Germanic immigrants incorporated women and children at different stages of the expansionary process still requires detailed study. But one striking contribution of the Wielbark system to the Cernjachov was precisely in the field of female costume (at least, female burial costume). As noted earlier, in both, women’s clothes were held with two brooches (fibulae) of similar style, one on each shoulder, and the same styles of necklaces and belts appear too. This mode of dress is not found among Dacian-speaking groups of the Carpathians before the third century. It is hard to believe that such a striking transfer could have occurred without substantial numbers of women – and therefore children too – having trodden the road south. The point is confirmed by the fact that the Goths, at least, among these migrants maintained their Germanic language over several generations from the mid-third to the later fourth century. If, as with the Scandinavian intrusion into Russia in the ninth and particularly the tenth centuries, we were looking at a phenomenon accomplished largely by small groups of armed men, as this latter example clearly was, then we would expect, as happened in Viking Russia, that the immigrants would quickly take on the language of their indigenous hosts. But as Ulfila’s Gothic Bible so spectacularly demonstrates, this was not the case following the third-century migrations. Ulfila was working among the Gothic Tervingi in the mid-fourth century, up to a hundred years after the immigrations to the Black Sea began, and the immigrants’ language remained at that point unambiguously Germanic.42 Without Gothic mothers to teach the language to their children, this could never have happened.
The evidence for the other migration flows is more sparse. Even for the Marcomannic War era, however, there is a limited amount of evidence that some of the migrant groups included women and children. It comes, fortunately, from Dio rather than from theHistoria Augusta, and is therefore that much more credible. The blocked attempted move of the Marcomanni and Quadi into the land of the Semnones, for instance, specifically included the people as a whole (Greek, pandemei). Even more explicit, the Hasding Vandals negotiated at one point to leave their women and children in the safe keeping of a local Roman commander, while they attempted to take control of lands which had previously belonged to a free Dacian group called the Costoboci. This latter piece of evidence, in particular, makes it impossible to believe that the action of the Marcomannic War was carried forward entirely by young men on the make.43
The historical evidence for the third-century west, unfortunately, is entirely non-explicit on this front, leading one recent commentator to assert that it is ‘commonsense’ that the action there was carried forward by small male warbands. But some early female and child burials from the Agri Decumates incorporate intrusive Elbe–Germanic materials, and I would be cautious about making such assertions. It is not clear either that warbands would have been able to exert a sufficient level of force, or, even early on, that we shouldn’t be reading any appearance of the label ‘Alamanni’ as itself significant of a confederative political entity. More positively, the Burgundians showed a similar capacity to the Goths to hold on to the particularities of their language over the long term. The evidence for the east Germanic nature of their dialect is unequivocal, but actually dates from the very end of the fifth century, from the independent Burgundian kingdom that emerged in the Rhône valley out of the process of west Roman imperial collapse. Hence the Burgundians managed to retain their distinctive dialect despite two hundred years of living in the west. Like the linguistic patterns of the Gothic Tervingi, this is inconceivable without at least some ‘complete’ social groups, involving women and children, having made their way to the Main from east of the Oder.44 It is only reasonable to be more guarded when responding to the general dearth of migration descriptions from the third-century west. To say that it is ‘commonsense’ to think in terms only of warbands, however, is as much an assertion as always applying the invasion hypothesis, especially given the clear evidence from the contemporary east of more diverse patterns of migratory activity. In their different ways, the Vandal, Burgundian and Gothic evidence all give us excellent reason to think that second- and third-century groups of migrant Germani sometimes included women and children. That being so, I would hesitate to assert the contrary in the case of the Alamanni.
As to the total number of migrants involved, it is impossible to say. We have few figures for this, and anyway could make only a wild guess at the size of the various indigenous populations affected by the Marcomannic War and third-century Germanic expansions. But it is precisely when faced with this kind of evidential impasse that the qualitative definition of mass migration used in comparative studies becomes helpful. The ‘shock to the political systems’ at the receiving end of each of the migration flows could hardly be clearer. Especially in the third century, the Roman Empire abandoned Transylvanian Dacia, many of the Carpi were pushed out of long-established homes into the Empire, and whatever remained of them, together with the independent Sarmatian kingdoms and Greek cities of the north Pontic littoral, were all eventually subdued into a new political system created by incoming Germanic-speaking immigrants. The domination of this region by Germanic-speakers, so evident from c.300 AD, was the result of an armed migration flow certainly to be numbered in thousands, and very probably tens of thousands. Using a qualitative rather than a numerical type of definition now commonly adopted in migration studies, this was straightforwardly a ‘mass’ migration. So too, of course, in the west. The arrival of Alamanni and Burgundians, the evacuation of the Agri Decumates, the overturning of the domination of Rhine–Weser groups, and even the earlier emigration of the Naristi, cannot be thought of as anything other than serious events for the areas affected.45
They were also significant for the migrants themselves, amongst whom were generated in the course of migration both new political structures and even some entirely new groupings. Roman sources naturally concentrate on the violence that spilled over on to Roman soil: sackings of cities, raids over the Black Sea and displacements of groups like the Carpi. But even once the old indigenous groups had been overcome, another, periodically violent, process was let loose, as the various immigrants set about establishing a new political order among themselves. Roman sources from around the year 300 refer in passing to competition between the immigrant groups now established in Dacia.46 One of the fruits of this process was probably the confederation of the Tervingi in which, as we saw in the last chapter, a series of kings were subordinate to a ruling ‘judge’. I suspect that these kings were the descendants of originally separate migrant groups who came, by whatever means and for whatever reasons, to accept the domination of the Tervingi ruling dynasty. Jordanes’ simple picture of the transfer of a whole people from the Baltic to the Black Sea entirely fails to capture these different levels of complexity.
There is, of course, much we will never know about these second-and third-century migrations. They clearly were flows of population, not the single pulses envisaged by the invasion-hypothesis model, and some of the action, especially in the early phases, was probably carried forward by warbands. But much larger forces than this were required for the more ambitious activities that were also a documented part of the process, such as permanent land annexations and fighting the bigger battles against the Roman Empire. The fact that the migrations also in some cases led to such substantial transfers of linguistic and material cultural patterns indicates that some of these larger groups consisted not just of men, but of women and children too. In part, then, the action does show some characteristics reminiscent of the old invasion hypothesis, especially in the extension of the domination of Germanic-speakers over the Black Sea region. Their armed arrival also eventually caused the exodus of not insignificant numbers of the indigenous population. Not as simple as the old invasion hypothesis, and not as antiseptic as an elite transfer, the Germanic takeover of the Black Sea region hovers somewhere between the two. It could perhaps be seen as a modified invasion-hypothesis model, where the migrants came in a flow that built momentum over time rather than in a block, where much of the indigenous population remained in place, but where large and mixed groups of migrants asserted themselves vigorously as the new political masters of the landscape.
This much, however, is only an interim conclusion. We need to ask further questions if we are going to arrive at any real understanding of the action. Why did these migration flows occur when they did, and why did they all tend in the direction of the Roman frontier? The conclusion that they consisted on occasion of mixed population groups raises another question of huge importance. Some modern migration flows do consist of large mixed groups, when political motives predominate. But these take the form of unorganized floods of refugees, without political leadership or direction. Before the evidence for large, mixed and, above all, organized migrant groups can be accepted, it is also necessary to counter the ‘commonsense’ objection that expansionary activity was usually undertaken just by all-male warbands. Is there a cogent reason why some of the Germanic migration flows should have included women and children alongside armed men when it came to asserting their domination over a new landscape?
Inner and Outer Peripheries
The direction of the migration flow – broadly towards the Roman frontier, but with a much more dramatic outcome in the east in terms of how much territory the migrants took over – begins to make sense if we think about it in the light of two factors that play a central role in all modern migration flows: ‘fields of information’ and the general context created by political structures. The field of information operating in the west doesn’t require much comment because the Alamanni were moving over such a short distance into the Agri Decumates. We can take for granted that they knew or could quickly come to know about their destination. But the epic treks of the eastern Germani need much more explanation. How much did the migrants actually know about the northern Pontus?
The most direct route from the Wielbark and Przeworsk areas of central and northern Poland to the Black Sea ran around the outer arc of the Carpathians, making use of the Upper Vistula and Dniester valleys. As we have seen, a ribbon of Wielbark cemeteries of the right date suggests that this was the route taken. Not only is this one of central-eastern Europe’s natural thoroughfares, but traffic along it had been particularly busy in the few centuries either side of the birth of Christ, when it was being exploited as one of the two main axes of the Amber Route uniting the Mediterranean world with the amber-producing shores of the Baltic Sea. As mentioned earlier, the solidified gum of submerged trees was highly prized for jewellery in the Mediterranean world, and one of the chief exports out of Germania. Substantial numbers of merchants were thus regularly treading the Goths’ main migration route during this era, and some Wielbark populations were actually involved in the trade, constructing and maintaining a complex series of wooden bridges and causeways close to the Baltic.47 Late second- and third-century Germanic-speaking migrants from north-central Europe could thus draw on substantial reservoirs of information about lands south and east of the Carpathians, and possible routes towards them. Thanks to the amber trade, they knew the route well and had some understanding of the societies and contexts that lay at the other end. I strongly suspect, though, that a closer analysis of the archaeological evidence would show that this east Germanic migration flow initially operated in a channelled fashion, the migrants clustering at only a relatively few destinations within the north Pontic zone, until the full regional situation became better known to them.
Once in the Black Sea region, further fields of information quickly came into play. The migrants soon learned of the attractive possibilities for gain presented by raiding the economically more developed Roman Empire, and the different routes by which such raids might be mounted. Some of these may have been partly known already, since Gothic troops were serving with Roman armies against Persia even before the attack on Histria in 238. An inscription dated thirty years before that attack records the presence of what were probably Gothic soldiers in the Roman army of Arabia. But before they arrived, the immigrants can have had no knowledge of the geography of the eastern Mediterranean and its hinterland, or that the rich coastlands of northern Asia Minor lay just across the Black Sea. That information soon became available. From the mid-250s they were mounting sea-borne raids across the Black Sea, and there is no doubt where the necessary intelligence came from. Historical sources are explicit that the maritime expertise of the populations of the Greek cities on the northern shores of the Black Sea provided ships and sailors for these expeditions. It is no stretch to suppose they were also the source of the information that lucrative raiding possibilities awaited anyone who could find some way to cross the two hundred kilometres and more of open sea that separate the northern and southern shores of the Pontus.48
But if the choices of third-century migrants were being influenced by information flowing along the Amber Route, they were also shaped, perhaps more decisively, by the political structures of the world around them. In the modern world, migration flows are interfered with by states who attempt to channel, encourage, or limit them by means of passports, border controls and immigration policies. Ancient state structures were much less sophisticated, but the Roman Empire did operate immigration policies when it came to group admissions, and its broader effect on the Germanic migration flows of the third century is obvious.
Expansion out of Germanic north-central Europe from the middle of the second century was not just limited to Wielbark groups. Langobardi and Ubii from the mouth of the Elbe, further west, delivered the wake-up call that marked the onset of the Marcomannic War. And non-Wielbark Germania, as we have seen, remained active into the third century. The migration both of Alamanni from the Elbe triangle and of Burgundians from further east played a major role in reconfiguring the political geography of the Rhine frontier region at exactly the same moment as other assorted Germani were expanding east of the Carpathians. In the course of the third century, therefore, the strategic situation was being restructured in broadly similar ways all along Rome’s Germanic-dominated European frontiers. And in both east and west, Rome’s strategic position was altered for the worse. In both, an element of migration was operating alongside broader political transformations. In the west, political restructuring predominated and is the element which most immediately catches the eye: its result, the new Frankish and Alamannic confederations. In the east, by contrast, migration predominated, because Germanic expansion took in such huge tracts of territory and effected major cultural changes. In their new world north of the Black Sea, the migrants created larger and more complex political structures than anything that had existed among them in north-central Europe.
But if the broad mix of components was similar, the overall effect was entirely different as between east and west. East of the Carpathians, vast new territories were brought under the control, or at least hegemony, of Germanic-speakers, and a large number of new political units came into being, as the migrants spread out over the landscape. In the west, the geographical expansion of Germanic-dominated territory was limited merely to the Agri Decumates, and political transformation ran more straightforwardly towards confederation than the greater diversity of political process visible north of the Black Sea. The explanation for these fundamentally different outcomes lies in the fact that Germanic expansion in the west ran head-on into the military and political structures of the Roman Empire. These were at times during the third century in considerable disarray, mostly because of the rise of Sasanian Persia, which stretched Roman resources and demanded substantial redeployments to the east. This presented expansion-minded Germanic groups with many opportunities for short-term profit in mid-century, but in the long run Roman imperial structures proved durable. After a long period of modification (not least a substantial increase in taxation), enough resources were found both to parry the Persian threat and to limit the possibility of large-scale Germanic expansion to the west. Put simply, because of the strength of the Roman army and its fortifications, Germanic expansion in the west was confined to small amounts of new territory, and most of the energy was channelled into internal political restructuring and short cross-border attacks. There could be no repeat in the west of events north of the Black Sea, where the more fragmented power structure of Rome’s clients allowed the Germanic migrants to create a new hegemony over vast tracts. The Roman Empire may not have had the bureaucratic capacity to issue passports, but its frontier structures played a major role in shaping the contrasting outcomes in eastern and western Europe of exactly the same explosive mixture of migration and political reorganization.49
Looked at closely, then, the routes and varying results of these migrations make perfect sense in the light of modern migration studies. But two big issues remain. What caused such a substantial number of Europe’s Germanic-speakers to take to the road at precisely this time? And how are we to account for the apparently anomalous nature of the migration flow, involving, as the evidence suggests it did, some large mixed social groups?
More than anything else, the lack of first-hand information hampers our understanding of migrant motivations. The discussion cannot be conducted on the basis of individual case studies, as any more recent analogue would be, because none exist. Nonetheless, as we have seen, migrants’ motivation is nowadays generally modelled using a matrix with economic and political motives on one set of axes, correlating with voluntary and involuntary movement on the other. The general expectation is that all four parameters will apply in virtually all cases, if in dramatically varying combinations, so that for some migrants voluntary economic motivations will predominate, whereas for others involuntary political ones. And even if we can’t do the job in the kind of detail we would like to, adopting this kind of approach remains a highly productive way forward.
Looked at in the round, the evidence strongly suggests that there was an involuntary political element to at least part of the migration flow. We’re clearly not talking in terms of a politically motivated flood of refugees on anything like the scale of Rwanda in the early 1990s. But a striking feature of late second- and third-century Germania, as we saw in the last chapter, is the evidence for an increase in violent political competition, as shown by the relatively dense cluster of weapons deposits turning up in Danish and other bogs. And there is no reason to suppose that the proto-Danes of the third century were particularly quarrelsome. The smart money must be on the bogs having allowed both more and better evidence of increased violence to be preserved there than elsewhere. At least, the contemporaneous emergence of the new political confederations, with their growing martial ideologies of leadership, are hard to envisage without the same kind of increased violence. Against this backdrop, it would be odd, in fact, if migration into new areas had not come into the equation as one response to the heightened dangers of living in old Germania. Growing competition for control of the same assets has always been one major cause of relocation.50 But if escalating political competition partly helps explain why so many Germani were on the move in the first place, a much more positive economic motivation helps explain its main geographical direction.
Occasional Mediterranean imports aside, archaeological investigation has uncovered, as we saw in Chapter 1, a Germanic world at the dawn of the Roman era possessed of only a simple material culture: handmade ceramics, little in the way of precious metals, and relatively few mechanisms for expressing status in material form. Much of this changed over the next few centuries, as the Germanic world opened itself to contacts with the more developed economies of the Mediterranean. The result was to generate many new flows of wealth across the frontier – from the profits of new trade links, diplomatic subsidies and cross-border raiding. A key point about all this new wealth, however, was that the profits were not evenly distributed in social terms: particular classes benefited disproportionately. And nor were they equally distributed geographically. Most of the wealth-generating contacts with the Roman Empire distinctly favoured Germanic groups established in the immediate frontier zone.
Diplomatic subsidies were paid only to groups established close to the frontier. It’s hard to tell, for instance, exactly how far beyond the frontier Constantius’ rearrangement of frontier politics, examined in the last chapter, would have stretched: certainly more than just a few kilometres, but probably no more than two or three days’ march for his legions, so maybe a hundred kilometres. Cross-border raiding, while not confined to those right on the frontier, was certainly also much easier for them. The same is true of trade in agricultural produce and other raw materials. Transport logistics made it much easier for those in the frontier zone to supply their products to the Roman soldiers – the basic source of demand. The point should not be overstated. Villages such as Feddersen Wierde could profit from Roman demand because of easy water transport, and some high-value exchange networks, like those of slaves or amber, stretched far into the interior of Germania. It was presumably predatory interior Germani, for example, who did the initial raiding for slaves, and the Wielbark causeways show that someone in northern Poland was benefiting hugely from the Amber Route. Nonetheless, most of the wealth flows benefited, solely or unequally, the frontier zone, and even longer-distance trade had to work its way eventually through middle-men – kings taking tolls, if nothing else – right on the border. It can’t have been only Vannius king of the Marcomanni who saw the wealth-generating potential of making Germanic traders bring their goods to Roman merchants on his soil, so that he could charge tolls. Monopoly is such a beautiful way to make money, and it was in order to achieve precisely this, presumably, that trading arrangements figured so strongly in diplomatic agreements between client states and the Roman Empire.
As much recent work on frontiers in general, and the Roman frontier in particular, has underlined, it is important to view installations like Hadrian’s Wall as the centre of a zone of cultural and economic contact stretching out for some distance either side of it, rather than as a preclusive defensive line. One result of this has been a tendency in recent analysis to underestimate the amount of violence and confrontation that we should be expecting along and across the frontier line. There is a significant element of truth in this, but it is not the whole truth about the frontier situation. For all that populations either side of it engaged in a whole range of contacts with one another, they were not just sitting in their own little frontier comfort zone, blithely unaffected by the wider world.
The rhythms of frontier coexistence could be undermined from one direction, for instance, by a Roman emperor’s need for prestige. In the later 360s, Valentinian I wanted to show his landowning taxpayers that he was tough on barbarians. He therefore reduced the annual gifts to the kings of the Alamanni – with disastrous results. They recycled these gifts to their followers in order to sustain their own prestige, so that Valentinian’s economies threatened their authority. The result was a wave of violence that destabilized the Alamannic sector of the Rhine frontier.51 Much more fundamental in my view, however, was the inherent tendency of the frontier zone to be destabilized from the non-Roman side. The reason why this was so follows on directly from what we have just observed. Its intense contacts with the more developed Roman world opened up the frontier to a whole series of new wealth flows. Their overall effect was to create a two-speed Germania whose economy and society worked at higher and more intense levels of development the closer you got to the Roman frontier, and vice versa. As a result, marked differences in wealth quickly built up between this frontier zone and the Germanic interior. In my opinion, this inequality was a crucial further component in the motivation behind the migration flows of the second and third centuries. They are an entirely logical consequence of the broader phenomenon of unequal development. Armed groups from the less rich outer regions were looking to seize by force a share of the attractive opportunities available closer to the Rhine and the Danube.
Such a tendency had begun to manifest itself as early as the first century. We have already encountered the first-century client king Vannius who prospered long and happily in the first half of the century on the basis of Roman subsidies and the wealth he derived from Roman merchants residing in his kingdom. This happy state of affairs eventually came to an end in 50 AD, however, when his wealth was ransacked by a group of Germani from outside the frontier zone, who put together an expedition of sufficient strength to seize his assets.52 This same basic motive – to seize the wealth available in the frontier zone – is evident or deducible in all the events of the third century. The Black Sea region, for instance, was rich in visible and highly moveable potential booty. Fabulously wealthy individual burials, full of precious metals, are characteristic of the archaeology of the Sarmatian kingdoms of the north Pontic shore in the early centuries AD. There is every reason to suppose that so much wealth acted as a magnet to Goths and others who had got wind of it through the regular passage of people and information up and down the Amber Route. And once they had moved in, the Germanic immigrants acted in other ways that were all designed to maximize their access to its riches.53
The first armed Goths to appear in the Black Sea zone not only sacked one of its cities, Histria, in 238, but also promised to keep the peace if they were granted an annual Roman subsidy.54 This group clearly grasped what kind of regular income could be forthcoming from a more intimate relationship with the Roman world. So too with the raiding: the motive, whether in the Balkans or across the Black Sea, was to acquire moveable wealth in all its available forms, human and otherwise. The remains of the new Germanic-dominated kingdoms established there, visible in the Cernjachov cultural system, illustrate the point succinctly, especially when compared with Wielbark remains. Precious metals are much more often found. In Cernjachov remains of the late third and fourth centuries, silver fibulae are reasonably common; they are rarely found in Wielbark burials of the first and second. Roman pottery is also extremely common in Cernjachov settlements and burials, both fine dinner services and the remains of amphorae that originally held wine or olive oil. Despite the absence of any internal or private Gothic account of the immigrants’ motives, I am confident, therefore, that they organized themselves into armed groups precisely to gain access to the wealth of the frontier zone. Its wider range of contacts with the Roman world had not only made the frontier much richer than outer Germania, but by the same token had made it a natural target for groups from the interior, who organized themselves to seize their own shares.
The largely voluntary economic motivation of our third-century migrants thus also had a strongly political dimension. This was not the negative political motivation that underlies modern trails of refugees, but a predatory motivation of a more (for want of a better word) ‘positive’ kind. The wealth building up among the mainly Germanic societies on the fringes of the Roman world could not be tapped into by individuals just turning up and requesting a share of it, in the way that migrant labour might seek employment in the industrial or service sectors of a modern economy. The new wealth was not being generated in factories which needed large quantities of labour. On the contrary, it was located at the courts of client kings, who redistributed the profits derived from their various transactions with the Roman state to their key supporters. It was these kings who initially received the subsidies, toll income, payments for military service and, quite probably, a cut from the cross-border raids as well. A few individual immigrants presumably managed to work their way into royal retinues, but this represented no mass demand for immigrant labour. Retinues were not very numerous and required only military specialists anyway. For larger groups of immigrants, the only way to liberate any of the new wealth was to turn up armed and in sufficient numbers to replace a client king and take control of his income. In the third century many immigrants grasped the opportunity, in both east and west, and client kings were replaced in such numbers as to rearrange the political geography all along Rome’s European frontiers.
Germanic Voortrekkers?
This important point explains why, although mainly economic and voluntary in nature, Germanic population flows in the later second and third centuries sometimes involved large migration units. This stands in stark contrast to similarly voluntary flows in the modern world, where the units of migration tend to be tiny: the individual or a few companions. What seems to be a contradiction is explained by fundamental differences in the economic context. Modern migration flows are actually dictated by the type of economic opportunity available – a mass demand for individual workers. The same principle applied in the Roman period, but the nature of the economic opportunity was different. In the modern world an immigrant can access a reasonable share of the wealth being generated by economic development by getting work in factories or service industries. In the second and third centuries, the path to success lay in being the leader, or part of the military elite, of a client state occupying a profitable corner of the Roman frontier world. Here, the appropriate migration unit – even though the population flow was voluntary and largely economically motivated – sometimes had to be large to succeed. From the days of Rome’s earliest advance to the Rhine and Danube, no attractive spot along the frontier was ever unoccupied. If you were an outsider wanting to become part of the profitable frontier system, your only option was to move in with sufficient force to oust a sitting tenant. Predatory activity from outside the frontier zone may have begun with smallish-scale raids such as the first-century attack on Vannius (although even that looks pretty substantial), but if you wanted to move into a region permanently, military manpower in the thousands, not hundreds, was called for.55
This explains, if in slightly paradoxical fashion, the other apparent anomaly in these ancient migration flows: that women and children sometimes made their way towards the frontier zone alongside their warrior menfolk. Why this was so follows on from the scale of military force required to take over one of the revenue-generating positions close to the frontier. In Roman-era Germania it was easy enough, as we saw in the last chapter, to put together warbands of a few hundred men, but forces of this size, while fine for raiding, could never have effected the kind of structural change we see happening right along Rome’s European frontiers in the third century – everywhere new sets of immigrant clients were replacing the incumbents. If we consider this problem in the light of the degree of development then prevalent in Germania, in order to assemble forces of the necessary size for higher-order military activities such as conquest, kings would need to convince not just their retainers but also large numbers of armed freemen to take part in the expedition. As we have seen, society was not yet so dominated by kings and their retainers that mobilizing these latter alone would provide enough men for the task in hand, any more than assembling just Alamannic kings and their retinues would have given Chnodomarius a chance at Strasbourg.
This observation is central to the seeming peculiarity of third-century Germanic migration. Retinue sizes were structurally delimited by the scale of available economic surplus. So large numbers of freemen had to be involved, and this greatly increased the likelihood that at least some families would also participate in any given expedition. When these expeditions were long-distance, one-way trips, as were those of the Goths and other Germani from Poland to the Black Sea, this was unavoidable, as with the Vandals in the Marcomannic War.56 Because of the massive overuse of the invasion hypothesis in the past, there is great resistance now, particularly among archaeologists, to the idea that mixed groups might ever move in force, deliberately to take over a new landscape. This negative reaction – that such a vision of any past events must be a myth, even if it is reported in contemporary and generally reliable sources – is so well entrenched that it is worth pointing out that analogous phenomena have been observed in the modern world.
By about 1800 AD, there were around forty thousand Boer settler families farming within the confines of the original Dutch settlement in the hinterland of the Cape of Good Hope, first established in 1652. Most of them were interconnected by marriage. But as the fiscal and cultural pressure of British imperialism started to build up in the early nineteenth century, they began to look for new lands. The Boers’ group organization did not run to a state structure but was sufficiently established for a commission (theCommissie) to send out scouting parties to check out the agricultural potential of neighbouring territories. One party brought back disappointing news of what is now Namibia, but a second – consisting of twenty-one men and one woman – made its way over the Zoutspansberg Mountains and found that the northern Transvaal and Natal offered more promising opportunities. As a result, individual parties began to assemble and make their way north at a rate of ten to fifteen kilometres a day, at first in groups of about fifty to a hundred families, each accompanied by their livestock and with all their worldly goods crammed into a wagon pulled by oxen. In February 1836, Hendrik Potgieter set out with two hundred people and sixty wagons, closely followed by other groups of similar size: Johannes van Rensburg with nineteen families, Louis Tregardt with seven (including the eighty-seven-year-old Daniel Pfeffer to teach the thirty-four children in the group), Andries Pretorius with sixty wagons, and Gert Maritz and Piet Retief with one hundred each. All of these groups consisted of men, women and children of all ages.
Aside from the quality of the grazing, the Boers had been attracted by the scouts’ reports that unclaimed land was plentiful. This proved mistaken. There were two militarily powerful kingships in the target areas, the Matabele of Mzilikazi and the Zulu of Dingane, who were not about to let the Boers take whatever they wanted. After initial attempts at negotiation, one of which led to the famous death of Piet Retief at the hands of Dingane clutching a supposed agreement over land grants, and the deaths in a subsequent night raid of five hundred trekkers including fifty-six women and one hundred and eighty-five children, the Boer leaders decided that the power of these kings had to be broken. So they reorganized themselves to create larger striking forces, which ruthlessly smashed the power of their enemies. The trekkers enjoyed a major technological advantage: five-foot-long flint-locks they could fire several times a minute from horseback. Hence relatively small Boer forces could wreak havoc. Even when attacking Mzilikazi’s main political centre, a few hundred men killed three thousand Matabele at no cost to themselves and burned the king’s kraal to the ground. Dingane’s Zulus, too, proved powerless in the face of firearms. These military successes encouraged more trekkers to move away from British rule, and twelve thousand of them eventually headed away from the Cape.
Apart from their technological superiority, which meant that relatively few Boers were required to fight even major battles, what happened here is identical to that suggested by reports of what went on north of the Black Sea in the third century (and, indeed, in the Viking west in the ninth). Small groups of wealth-seeking intruders reorganized themselves into larger groups when it became apparent that the acquisition of capital wealth – control of the land – required the destruction of major political obstacles. The way that an initially peaceful migration flow quickly turned itself into deliberate armed predation is also a salutary reminder. Homo sapiens sapiens is perfectly capable of organizing itself into armed groups with sufficient capacity to seize the assets of others, and does sometimes do so using migration as the vehicle. Equally important, and despite the overtly military element to their activities, the Boer migration units always contained women and children as well as men, just as the third-century materials indicate was the case with at least some of the Germani. This not only shows that armed mixed groups are an a priori possibility (which – so strong is the rejection of the invasion hypothesis – some have come to doubt), but also reinforces the reason why this will tend to happen. Where the military capacity of a land-grabbing group depends either only partly or not at all upon professional soldiery, but rather on owner-farmers who also fight, then any of those farmers who join the migration stream will bring their families with them. Young Boers were taught to ride and shoot from an early age – so, too, the women, who were far from helpless in battle even without their men – and it was this military capacity that subdued the Matabele and the Zulus. As we know, second- and third-century Germania had some military retinues, but they were not huge, and since they did not have a massive military advantage such as firearms over the Carpi and Sarmatians, the Germanic groups who forced their way into the northern Pontus needed to be much larger than a Boer commando. They had to draw, therefore, on the larger cross-section of freemen fighter-farmers in Germanic society, and these men naturally brought their families with them.
To have a chance of success, would-be expedition leaders had to couch their recruiting drives in broad enough terms to attract freeman warriors. No description of one survives from this early era, but these few words depicting the Gothic leader Theoderic putting together his first major military expedition in c.470 AD nicely evoke the likely process:
Now Theoderic had reached man’s estate, for he was eighteen years of age and his boyhood was ended. So he summoned certain of his father’s adherents and took to himself his friends from the people, and his retainers almost six thousand men.57
This expedition wasn’t a one-way trip, so there was no reason to take families, but it shows that, even in the fifth century, mobilizing a sizeable force meant looking beyond the retinues and towards a broader tranche of Germanic society. For a complete explanation of the second- and third-century phenomena, however, and particularly of what made freemen and their wives open to persuasion that joining an armed expedition to the Black Sea was a good idea, we also need to bring in one further factor, which again figures strongly in modern case studies of migration: inherent mobility.
The populations of both the Przeworsk and the Wielbark zones – like the inhabitants of the rest of Germania in our period – practised a mixed agriculture. Cows, as Tacitus reports and as is borne out in some of the settlement archaeology, were a status item by which wealth was measured, but grain was the staple diet, and its production the cornerstone of economic activity. The Germani were not nomads in any real sense of the word; they did not cycle their herds between designated blocks of summer and winter pasture, as some contemporary steppe nomads did. But in the early centuries AD many Germanic societies, and certainly those of Wielbark areas, lacked the necessary agricultural expertise to maintain the fertility of their arable fields over more than a generation or so. Viewed in anything but the short term, therefore, their settlements tended to be mobile. As the fertility of one set of lands was exhausted, the population would move on, constructing new settlements as they went. Consonant with this, in the Wielbark world cemeteries seem to have provided a much more stable focal point for life as well as for death. They were much longer-lived – that at Odry remained in use for the best part of two hundred years, during which time many settlements came and went – and perhaps even functioned as centres of communal life. A striking characteristic of Wielbark cemeteries before 200 AD, for instance, is a large stone circle, containing no burials but sometimes equipped with a post in the middle. Archaeologists have plausibly suggested that these circles may have marked out communal space for meetings. Be that as it may, the Wielbark population clearly expected regularly to relocate itself.58
This is highly relevant because comparative studies have repeatedly demonstrated that migration is a life strategy more readily adopted by populations who are already mobile. The point even applies across generations. Statistically, the children and grandchildren of immigrants are much more likely than the average to move on. Another reason why population groups comprising men, women and children were ready to trek from the Wielbark and Przeworsk areas to the Black Sea is that their inability to maintain long-term agricultural fertility meant that they were already pre-programmed to use relocation as a strategy for achieving greater prosperity. In one sense, to direct that strategy in a coherent move over a relatively long distance represented no more radical a departure, say, than the seventeenth-century English peasant who, having made it out of the countryside and into the town, then decided to take ship for the Americas. In another, of course, it was.
Up to about 200 AD, perhaps on the strength of a slight population increase – to judge by the number of settlements in use in each generation – relocation on the part of Wielbark groups took the form of a steady if unspectacular drift southwards into previously Przeworsk areas. This phase of Wielbark expansion corresponds quite well with what we might expect from a wave-of-advance model, the drift south being the product of random individual choice as the population slowly increased, rather than a large-scale flow of directed migration. Movement north was constrained by the Baltic Sea, and in any case soils improved as you moved away from the sandy, rocky deposits left on its southern shores by ancient glaciers. The subsequent trek to the Black Sea was a totally different kind of enterprise. The distances involved were much greater, and the moves took place over a shorter time. Second-century expansion spread out three hundred kilometres or so in a south-easterly direction over something like fifty to seventy-five years. Its third-century counterpart covered well over a thousand kilometres in an equivalent time. So this second flow, or second stage of the same flow, was obviously much more directed, and it had to be.
Steady settlement drift from a perhaps slightly expanding population had now become deliberate armed intrusion, for financial gain, into an alien political locale. And, again, the parallels with the history of the Boers are striking. Between the first settlement of 1652 and 1800, individual settler families drifted outwards from the Cape over the eight hundred kilometres separating it from the Orange River, which marked its original boundary, as population expanded (Louis Tregardt, for instance, had seventeen children by four wives). This, too, would fit a wave-of-advance model. Movement across the river in response to the negative political, economic and cultural impetus provided by the British was built on the back of this tradition of movement, but directed and accelerated into a quite different phenomenon. Migration units became larger, and, as we have seen, the population flow rapidly evolved into military predation when it was resisted. Likewise, too, in the case of the third-century Germani: their shift in destination with regard to the northern Pontus required careful planning. Individual Germanic families from the north drifting into the Black Sea region would have got precisely nowhere, assuming that they had it in mind to annex land. Establishing military hegemony in a new world required careful planning and a mass of population, even if this mass was organized in a number of separate expeditionary forces rather than the one ‘people’ envisaged by the old invasion hypothesis.