IN AD 15, THE ROMAN ARMY of Germanicus Caesar, nephew of the reigning Augustus Tiberius, approached the Teutobergiensis Saltus (the Teutoburg Forest, 300 kilometres north-east of Trier). Six years before, three entire Roman legions commanded by P. Quinctilius Varus, totalling with their attendant auxiliary troops maybe twenty thousand men, had been massacred there in one of the most famous battles of antiquity.

The scene lived up to its horrible associations. Varus’ extensive first camp, with its broad extent and headquarters marked out, testified to the whole army’s labours. Then a half-ruined breastwork and shallow ditch showed where the last pathetic remnant had gathered. On the open ground were whitening bones, scattered where men had fled, heaped up where they had stood and fought back. Fragments of spears and of horses’ limbs lay there, also human heads fastened to tree trunks. In groves nearby were the outlandish altars at which the Germans had massacred the Roman ‘colonels’ and ‘senior company-commanders’. Survivors of the catastrophe, who had escaped from the battle or from captivity, pointed out where the generals had fallen, and where the Eagles were captured. They showed where Varus had received his first wound, and where he died by his own unhappy hand. And they told of . . . all the gibbets and pits for the prisoners.1

The massacre was the work of a coalition of Germanic warriors marshalled by one Arminius, a chieftain of the Cherusci, a small tribe living between the River Ems and the River Weser in what is now northern Germany. The ancient Roman sources describing the defeat were rediscovered and passed into broader circulation among Latin scholars in the fifteenth and sixteenth centuries, and from that point on Arminius, generally known as Hermann (‘the German’) – the delatinized version of his name – became a symbol of German nationhood. Between 1676 and 1910 an extraordinary seventy-six operas were composed to celebrate his exploits, and in the nineteenth century a huge monument was constructed in his honour near the small city of Detmold in the middle of what is today called the Teutoburger Wald. The foundation stone was laid in 1841, and the monument was finally dedicated in 1875, four years after Bismarck’s defeat of France had united much of the German-speaking world of north-central Europe behind the Prussian monarchy. The 28-metre copper statue of Hermann is mounted on top of a stone base of similar height, which itself sits on top of a 400-metre hill. The edifice was a reminder that the triumph of modern German unification had its counterpart in the Roman era.

The Hermann monument is actually in the wrong place. The name Teutoburger Wald was first coined for the forested area around Detmold in the seventeenth century, as people began to conjecture where the ancient battle might have taken place. Thanks to some extraordinary finds, part of the actual battlefield has now been identified about 70 kilometres to the north. Just outside Osnabrück, the north German coastal plain is fringed by uplands known as the Wiehengebirge. Since 1987, a large number of Roman coins and various items of military equipment have been recovered from an area about 6 by 4.5 kilometres on the northern fringes of this range, known as the Kalkriese-Niewedde depression. The southern boundary is marked by the Kalkriese Berg, a 100-metre hill, which was heavily wooded in antiquity. At the foot of its northern slope was a strip of sandy soil, part of it so narrow that only four men could have marched abreast. On the other side was a huge peat bog. In AD 9, the Roman army had been moving east–west along the narrow strip led by native guides provided by Arminius – he had convinced Varus that he had Rome’s interests at heart – when it was caught in an ambush between the wooded slopes to its south and the peat bog to the north. As told by our best source, a four-day running battle ensued, during the first part of which the Romans, despite substantial losses, held formation and continued to advance towards safety. By the fourth day, however, it had become clear that the army was cornered and doomed. At this point Varus, having given permission to his surviving troops to do whatever seemed best in the circumstances, committed suicide rather than fall into the hands of his attackers. Few survived to tell the tale.2

The catastrophe reads like a larger-scale version of that suffered by Cotta and his men, betrayed on to similarly impossible ground sixty-three years earlier. The long-term outcome, however, was different. Whereas the Eburones and the Treveri were eventually conquered, and propelled towards learning Latin, wearing togas and building self-governing towns, Arminius’ Cherusci were not. In the late Roman period, the area between the Rhine and the Elbe still remained beyond the imperial frontier, its material culture betraying none of the characteristic marks of Roman civilization. This ancient line in the European sand is still discernible in the modern divide between Romance languages descended from Latin, and Germanic languages. On the face of it, this would explain why the western Roman Empire was to give way, in the fifth century, to a series of successor kingdoms with, at their core, groups of armed Germanic-speakers. Germania east of the Rhine was not swallowed up by Rome’s legions in the conquest period because its inhabitants fought tooth and nail against them, and eventually had their full revenge more than four centuries later in the destruction of the Empire. This was certainly the explanation given by nineteenth-century German nationalists; argued in scholarly circles, it was also brought home to a much wider audience. Felix Dahn, whose great work on Germanic kingship remains a classic, also wrote a famous novel, Ein Kampf um Rom (War against Rome), which went through multiple editions in the late nineteenth and early twentieth centuries.3

The odd thing about all this, though, is that if you had asked any fourth-century Roman where the main threat to imperial security lay, he would undoubtedly have said with Persia in the east. This was only sensible, because in about AD 300 Persia posed an incomparably greater threat to Roman order than did Germania, and no other frontier offered any real threat whatsoever.4 A closer reading of the sources, especially in the light of archaeological evidence that was unavailable to Dahn, suggests a rather different reason for the halting of the legions on the Rhine and Danube in the early first century from stirrings of German nationalism. It also explains why late Romans were much more concerned about Persia than about Germanic tribesmen.

Germania and the Limits of Roman Expansion

IN THE FIRST CENTURY AD, Germanic-speaking groups dominated most of central and northern Europe beyond Rome’s riverine frontiers. The Germani, as the Romans called them, spread all the way from the Rhine in the west (which, before the Roman conquest, had marked an approximate boundary between Europe’s Germanic and Celtic speakers) to beyond the River Vistula in the east, and from the Danube in the south to the North and Baltic Seas. Apart from some Iranian-speaking Sarmatian nomads on the Great Hungarian Plain, and Dacian-speakers in and around the arc of the Carpathians, Rome’s immediate neighbours were all Germanic-speakers: from Arminius’ Cherusci and their allies at the mouth of the Rhine, to the Bastarnae who dominated substantial tracts of territory at the mouth of the Danube (map 2). First-century Germania was thus much bigger than modern Germany.

Trying to reconstruct the way of life and social institutions, not to mention the political and ideological structures of this vast territory, is a hugely difficult task. The main problem is that the societies of Germanic Europe, in the Roman period, were essentially illiterate. There is a fair amount of information of various kinds to be gleaned from Greek and Latin authors, but this has two major drawbacks. First, Roman writers were chiefly interested in Germanic societies for the threat – potential or actual – that they might pose to frontier security. What you find for the most part, therefore, are isolated pieces of narrative information concerning relations between the Empire and one or more of its immediate Germanic neighbours. Groups living away from the frontier hardly ever figure, and the inner workings of Germanic society are never explored. Second, what information there is is deeply coloured by the fact that, to Roman eyes, all Germani were barbarians. Barbarians were expected to behave in certain ways and embody a particular range of negative characteristics, and Roman commentators went out of their way to prove that this was so.

Little survives from inside the Germanic world to correct the misapprehensions, omissions and slanted perspectives of our Roman authors. The Germani did use runes for divinatory purposes for much of the Roman period, and there are other, limited, exceptions to the illiteracy rule, but no detailed first-hand account of life has come down to us from inside Germania. So there is much that we do not know, and cannot ever know, and for most areas of life we have to fall back on information from Roman sources and on more or less informed guesswork. The best that can often be done when attempting to reconstruct social institutions, for instance, is to look at literary sources – especially legal ones – from Germanic-dominated kingdoms of the later fifth and sixth centuries, then try to extrapolate what might also be relevant to earlier eras. Stretching from the Rhine to the Crimea, Germania encompassed many different geographical and economic landscapes, and it is always necessary to consider whether something reported of one group might also be true of another. The literary evidence thus offers us a not entirely palatable choice between the biased testimony of Roman sources and material of a later date. Both can be revealing, but they must be handled honestly and with explicit acknowledgement of their inherent limitations.



To some extent, the lack of first-hand contemporary Germanic sources has been filled by archaeological investigation. This has the priceless advantage of bringing us face to face with contemporary and genuinely Germanic artefacts and contexts, but Germanic archaeology is a subject with a difficult past. As a scientific discipline, it emerged in the late nineteenth century when the Hermann monument was under construction and when nationalism was sweeping through most of Europe. It was generally assumed at this date that the ‘nation’, or ‘people’, was the fundamental unit in which large groupings of human beings had operated in the distant past, and should operate now. Most nationalisms were also fuelled by a strong sense of their own innate superiority. The German nation may have been split up over time into lots of small political entities, but the efforts of Bismarck and others were now, through German unification, successfully restoring the natural and ancient order of things. In this cultural context, Germanic archaeology could have only one agenda: to research the historical origins and homeland of the German people. The first great proponent of such studies, Gustav Kossinna, noticed that the increasing quantities of artefacts then coming to light from excavated cemeteries could be grouped together by similarity of design and burial custom. He built his reputation on the argument that the geographical spread of particular groupings of artefacts and customs represented the territories of particular ancient peoples.5

Such was the quasi-religious fervour surrounding the concept of the nation that politicians were ready to use identifications of the ancient spread of ‘peoples’ as evidence for claims about the present. At Versailles in 1919, Kossinna and one of his Polish disciples, Vladimir Kostrewszki, made rival cases for the positioning of the new German-Polish border on the basis of different identifications of the same set of ancient remains. Things got nastier still in the Nazi period, when high-flown claims about ancient Germania became a basis-cum-justification for territorial demands in Poland and the Ukraine, and an associated sense of ancient Germanic racial superiority led directly to the atrocious treatment of Slavic prisoners-of-war. In the last two generations, however, Germanic archaeology has successfully reinvented itself, and from this have resulted huge advances in our understanding of the long-term social and economic development of the Germani. With the excision of nationalistic assumptions from the interpretation of literary sources, the history of Germanic-speaking Europe in the Roman period can be rewritten in new and exciting ways.

A first gain stems directly from new understandings of the patterns of similar remains which Kossinna was sure could identify the territories of ancient ‘peoples’. While the territory of ancient Germania was clearly dominated in a political sense by Germanic-speaking groups, it has emerged that the population of this vast territory was far from entirely Germanic. In the great era of nationalism, anywhere that threw up plausibly ancient Germanic remains was claimed as part of an ancient and greater German homeland. Analysis of river names has shown, however, that there was once in northern Europe a third population group with its own Indo-European language, located between the Celts and the Germani. These people were under the domination of the other two long before Roman commentators reached the area, and we know nothing about them. Much of ancient Germania was also the product of periodic Germanic expansion, west, south and east from a first, traceable, heartland beside the Baltic. Some early land-grabbing episodes made enough of a stir to register in ancient Greek sources, while others occurred after the rise of Rome and are better known. But this kind of expansion did not annihilate the indigenous, non-Germanic population of the areas concerned, so it is important to perceive Germania as meaning Germanic-dominated Europe. The more one moved south and east through the region during the Roman period, the more likely it is that Germanic-speakers constituted a politically dominant force in very mixed societies.

The other salient fact about Germania in the Roman period was its complete lack of political unity. As map 2 (based on Tacitus’ gazetteer) makes clear, it was a highly fragmented world, comprising over fifty small sociopolitical units. There was a variety of ways in which, for brief periods, some of them might be brought together for particular purposes. As we have seen, Arminius mobilized a mixed force of tribesmen in AD 9 to defeat Varus. Half a century earlier, Julius Caesar had encountered another Germanic leader of extraordinary and slightly longer-lived power: Ariovistus, King of the Suebi, who by 71 BC had built up a substantial power-base on the eastern fringes of Gaul and for a time was even recognized as ‘friend’ by the Romans. Caesar eventually picked a fight with him in 58 BC, routing his army in Alsace. One major defeat was enough to break up the coalition. In Arminius’ day there was one other pre-eminent Germanic leader, Maroboduus, who ruled a coalition of various groupings based in Bohemia. Tacitus also records that some Germanic tribes belonged to cult leagues, and pinpoints a moment when one particular prophetess, Veleda, acquired huge influence. But neither cult leagues, nor prophetesses, nor temporarily pre-eminent leaders represented major steps towards Germanic unification.6

As Roman power moved east of the Rhine, different Germanic groups were quite as likely to fight each other as to fight the Romans, and the results could be as brutal as the Teutoburger Wald. There is little discernible cultural difference between these groups, and it was essentially different political identities that divided them, and through which struggles for control of the best lands and other economic assets were fought out. Late in the first century, for instance, a coalition of their neighbours turned on the Bructeri, and invited Roman observers to enjoy the spectacle, reportedly, of 60,000 people being massacred. Tacitus’ Annals also record a fight to the death between the Hermenduri and the Chatti, and the eventual destruction of the landless, and hence troublesome, Ampsivarii: ‘In their protracted wanderings, the exiles were treated as guests, then as beggars, then as enemies. Finally, their fighting-men were exterminated, their young and old distributed as booty.’7

It could hardly be clearer that nineteenth-century visions of an ancient German nation were way off-target. Temporary alliances and unusually powerful kings might for a time knit together a couple or more of its many small tribes, but the inhabitants of first-century Germania had no capacity to formulate and put into practice sustained and unifying political agendas.

Why did Roman expansion fail to swallow this highly fragmented world whole, as it had done Celtic Europe? The halting of the legions’ progress through northern Europe has often been attributed to Arminius’ great victory, but like the destruction of the legion commanded by Sabinus and Cotta in 54 BC, the massacre of Varus’ command was a one-off event for which the Romans duly extracted revenge. Germanicus’ visit to the Teutoburger Wald in AD 15 was part of another major campaign against Arminius’ Cherusci. In the course of it, a second Roman force was ambushed by Arminius’ warriors, but this time the outcome was different. Although hard pressed for a time, the Romans eventually lured their opponents into a trap, with a predictable outcome: ‘TheGermaniwent down, as defenceless in defeat as success had made them impetuous. Arminius got away unhurt [but] the massacre of the rank and file went on as long as fury and daylight lasted.’8 The Romans were assisted by Segestes, a second leader of the Cherusci, who, like many Gallic Celtic chiefs in the time of Julius Caesar, saw considerable advantages in his territory becoming part of the Roman Empire. Not even the Cherusci, let alone the Germani as a whole, were united in their resistance to Rome, and the Teutoburger Wald did not stop the advance of the legions in its tracks. More Roman victories followed in AD 16, and about three years later Arminius was murdered by a faction of his own tribesmen. His son was brought up in Ravenna. Arminus had won one huge, fluke victory, but the underlying reasons for the halting of the legions on the fringes of first-century Germania were altogether different.

LOGISTICS MADE IT likely enough that Rome’s European frontiers would end up on river lines somewhere. Rivers made supplying the many troops stationed on the frontier a much more practical proposition. An early imperial Roman legion of about 5,000 men required about 7,500 kilos of grain and 450 kilos of fodder per day, or 225 and 13.5 tonnes, respectively, per month.9 Most Roman troops at this date were placed on or close to the frontier, and conditions in most border regions, before economic development had set in, meant that it was impossible to satisfy their needs from purely local sources. Halting the western frontier at the Rhine, rather than on any of the other north–south rivers of western or central Europe – of which there are many, notably the Elbe – had another advantage too. Using the Rhône and (via a brief portage) the Moselle, supplies could be moved by water directly from the Mediterranean to the Rhine without having to brave wilder waters.

The real reason why the Rhine eventually emerged as the frontier lay in the interaction of the motives underlying Roman expansion and comparative levels of social and economic development within pre-Roman Europe. Roman expansion was driven by the internal power struggles of republican oligarchs such as Julius Caesar and by early emperors’ desires for glory. Expansion as the route to political power at Rome had built up momentum at a point when there were still numerous unconquered wealthy communities around the Mediterranean waiting to be picked off. Once annexed, they became a new source of tribute flowing into Rome, as well as making the name of the general who had organized their conquest. Over time, however, the richest prizes were scooped up until, in the early imperial era, expansion was sucking in territories that did not really produce sufficient income to justify the costs of conquest. Britain in particular, the ancient sources stress, was taken only because the emperor Claudius wanted the glory.10 With this in mind, the limits of Rome’s northern expansion take on a particular significance when charted against levels of economic development in non-Roman Europe.

Expansion eventually ground to a halt in an intermediate zone between two major material cultures: the so-called La Tène and Jastorf cultures (map 2). Some key differences in the general character of life distinguished the two. As well as villages, La Tène Europe had also generated, before the Roman conquest, much larger settlements, sometimes identified as towns (in Latin, oppida – hence its other common name, ‘the Oppida culture’). In some La Tène areas coins were in use, and some of its populations were literate. Caesar’s Gallic War describes the complex political and religious institutions that prevailed among at least some of the La Tène groups he conquered, particularly the Aedui of south-western Gaul. All of this rested upon an economy that could produce sufficient food surpluses to support warrior, priestly and artisan classes not engaged in primary agricultural production. Jastorf Europe, by contrast, operated at a much starker level of subsistence, with a greater emphasis on pastoral agriculture and much less of a food surplus. Its population had no coinage or literacy, and, by the birth of Christ, had produced no substantial settlements – not even villages. Also, its remains have produced almost no evidence for any kind of specialized economic activity.

In the days when Kossinna’s assumptions ruled, and cultural zones were associated with ‘peoples’, it was traditional to equate the La Tène and Jastorf cultures respectively with Celts and Germani, but such simple equations don’t work. Zones of archaeological similarity reflect patterns of material culture, and material culture can be acquired; people aren’t just born with one set of weapons, pots and ornaments that they retain through thick and thin. While La Tène cultural patterns did originally emerge among some of Europe’s Celtic-speakers, and their Jastorf equivalents among certain Germanic groups, there was no golden rule that made it impossible for Germanic groups to adopt elements of La Tène material culture. And by the time Roman power advanced north of the Alps, some Germanic groups on the fringes of the Celtic world, particularly those around the mouth of the Rhine, had evolved a culture that was much more in line with La Tène than with Jastorf norms.

The Roman advance ground to a halt not on an ethnic divide, therefore, but around a major fault-line in European socio-economic organization. What happened was that most of more advanced La Tène Europe was taken into the Empire, while most of Jastorf Europe was excluded.

This fits a much broader pattern. As has also been observed in the case of China, there is a general tendency for the frontiers of an empire based on arable agriculture to stabilize in an intermediate, part-arable part-pastoral zone, where the productive capacity of the local economy is not by itself sufficient to support the empire’s armies. Expansionary ideologies and individual rulers’ desires for glory will carry those armies some way beyond the gain line; but, eventually, the difficulties involved in incorporating the next patch of territory, combined with the relative lack of wealth that can be extracted from it, make further conquest unattractive. A two-speed Europe is not a new phenomenon, and the Romans drew the logical conclusion. Augustus’ successor Tiberius saw that Germania just wasn’t worth conquering. The more widely dispersed populations of these still heavily forested corners of Europe could be defeated in individual engagements, but the Jastorf regions proved much more difficult to dominate strategically than the concentrated and ordered populations occupying the La Tène towns. It was the logistic convenience of the Rhine–Moselle axis and cost-benefit calculations concerning the limited economy of Jastorf Europe that combined to stop the legions in their tracks. Germania as a whole was also far too disunited politically to pose a major threat to the richer lands already conquered. It was thus entirely appropriate that nineteenth-century German nationalists put the Hermann monument in the wrong place, since they also mistook his real significance. It was not the military prowess of the Germani that kept them outside the Empire, but their poverty.11

As a result, the defended Roman frontier came by the mid-first century AD to be established broadly along a line marked by the Rivers Rhine and Danube. Some minor adjustments apart, it was still there three hundred years later. The consequences were profound. West and south of these riverine frontiers, European populations, whether Jastorf or La Tène, found themselves sucked into a trajectory towards Latin, togas, towns and, eventually, Christianity. Watching from the sidelines as neighbouring populations were transformed by Romanization, Germanic-dominated Europe north and east of this line never became part of the Roman world. In Roman terms, Germania remained the home of unreconstructed barbarians. The same label was used of the Persians in the east. However, this second major group of barbarians posed an altogether different level of threat.

Persia and the Third-Century Crisis

AT NAQS-I RUSTAM, seven kilometres north of Persepolis, lies the burial place of the famous Achaemenid Persian kings of antiquity, Darius and his son Xerxes, whose unwelcome attentions the Athenians and their allies fought off at the battles of Marathon and Salamis in 490 and 480 BC. Here, too, were discovered in 1936, inscribed in three languages on the side of a Zoroastrian fire temple, the proud boasts of a much later Persian king:

I am the Mazda-worshipping divine Shapur, King of Kings . . . of the race of the Gods, son of the Mazda-worshipping divine Ardashir, King of Kings . . . When I was first established over the dominion of the nations, the Caesar Gordian [emperor, 238–44] from the whole of the Roman Empire . . . raised an army and marched . . . against us. A great battle took place between the two sides on the frontiers of Assyria at Meshike. Caesar Gordian was destroyed and the Roman army was annihilated. The Romans proclaimed Philip Caesar. And Caesar Philip came to sue for peace, and for their lives he paid a ransom of 500,000 denarii and became tributary to us . . . And the Caesar lied again and did injustice to Armenia. We marched against the Roman Empire and annihilated a Roman army of 60,000 men at Barbalissos. The nation of Syria and whatever nations and plains that were above it, we set on fire and devastated and laid waste. And in the campaign [we took] . . . thirty-seven cities with their surrounding territories. In the third contest . . . Caesar Valerian came upon us. There was with him a force of 70,000 men . . . A great battle took place beyond Carrhae and Edessa between us and Caesar Valerian and we took him prisoner with our own hands, as well as all the other commanders of the army . . . On this campaign, we also conquered . . . thirty-six cities with their surrounding territories.

This is from the Res Gestae Divi Saporis (The Acts of the Divine Shapur); it encapsulates a strategic revolution, beginning in the third century AD, which transformed the Roman Empire.12

Hitherto, resistance to Rome in the east had been led by the Parthian Arsacid dynasty, which had first established itself about 250 BC. The Arsacids ruled a world that could not have been more different from the forested homes of the north European Germani. The dynasty originated in Parthia and began to spread its dominion more widely over the Near East in the third century BC, quickly bringing under its sway territory from the Euphrates to the Indus. It thus encompassed a huge range of populations and habitats, but the dynasty’s heartland was soon to be Mesopotamia. Again unlike Germania, the history of this region had been punctuated by the rise and fall of great empires, not least the Achaemenid Empire of Cyrus, Darius and Xerxes. They had ruled not only the Near East, but also Egypt, western Turkey and the Fertile Crescent, and had even come within a whisker of swallowing Greece.

The Parthian Arsacids had scored some early victories over the Empire in the late republican period, when Roman power had first penetrated so far east, most famously in the destruction of the army of Crassus, father and son, in 53 BC. But by the second centuryAD the dynasty’s capacity to mount serious resistance to Rome had diminished, and a succession of emperors won major victories on the Persian front. The latest came in the 190s, when Septimius Severus created two new provinces, Osrhoene and Mesopotamia, thereby advancing the frontier south and east. His victories threw the world ruled by the Parthians into crisis. Various members of the dynasty struggled for control, and some outlying regions threw off its suzerainty. As early as 205/6, rebellion began in the province of Fars beside the Indian Ocean. It was led by Sasan, the most important of the regional magnates, and was continued after his death by Shapur’s father, Ardashir I (reigned AD 224–40), the real founder of the Sasanian dynasty. In 224 and 225, he defeated two rival Arsacid rulers and established his control over the other regional magnates who had also broken away from Arsacid domination, before having himself crowned King of Kings in Persepolis in September 226.13

As the Res Gestae Divi Saporis makes clear, the rise to prominence of the Sasanian dynasty was not just a major episode in the internal history of modern Iraq and Iran. Defeat at the hands of a succession of Roman emperors in the second century was a fundamental reason behind the collapse of Arsacid hegemony, and the Sasanians were able rapidly and effectively to reverse the prevailing balance of power. Ardashir I began the process. Invading Roman Mesopotamia for the first time between 237 and 240, he captured the major cities of Carrhae, Nisibis and Hatra (map 3). Rome responded to the challenge by launching three major counterattacks during the first twenty years of the reign of Ardashir’s son Shapur I (reigned 240–72). The results were as Shapur’s inscription records. Three huge defeats were inflicted on the Romans, two emperors were dead, and a third, Valerian, captured. Shapur proceeded to drag Valerian around with him, in chains, as a symbol of his own greatness – an image preserved for posterity in the great carved relief of Bishapur. After his death, Shapur had him skinned and tanned as a permanent trophy. Later in the century, a second Roman emperor, Numerianus, was also captured, but killed immediately: ‘They flayed him and made his skin into a sack. And they treated it with myrh [to preserve it] and kept it as an object of exceptional splendour.’14 Whether this was also Valerian’s fate, or whether he was kept on the floor or the wall, the sources don’t say.

Nothing could better symbolize the new world order. The rise of the Sasanians destroyed what was by then more or less a century of Roman hegemony in the east. Rome’s overall strategic situation had suddenly and decisively deteriorated, for the Sasanian superpower, this new Persian dynasty, despite Rome’s best efforts in the middle of the third century, would not quickly disappear. The Sasanians marshalled the resources of Mesopotamia and the Iranian plateau much more efficiently than their Arsacid predecessors had done. Outlying principalities were welded more fully into a single political structure, while the labour of Roman prisoners was used for massive irrigation projects that would eventually generate a 50 per cent rise in the settlement and cultivation of the lands between the Tigris and Euphrates. This certainly began under Shapur, if not his father, and the resulting increase in tax revenues was coordinated by a burgeoning bureaucracy and directed to the maintenance of an at least partly professional army. Not for nothing did Shapur revive, in his diplomatic posturing towards the Romans, a claim to the old Achaemenid Empire in its entirety: he wanted not only Iran and Iraq, but Egypt, the Fertile Crescent and western Turkey as well.15


Previously, Rome had operated in a context of dominance across all its frontier zones. Opponents might achieve local successes, but defeats could easily be reversed by mobilizing the Empire’s existing resources. Now, the rise of a rival superpower was a huge strategic shock. It had reverberations not just for the eastern frontier regions, as Shapur’s record of sackings and captures recounts, but for the Empire as a whole. Not only did a much more powerful enemy have to be confronted on its eastern frontier, but the defence of all the other frontiers still had to be maintained. For this to be possible, a major increase in the power of the Roman military was necessary. By the fourth century, this had produced both larger and substantially reorganized armed forces.

As we saw in Chapter 1, the Roman army of the early imperial era was divided into legions, each a small expeditionary army of 5,000-plus men, recruited exclusively from existing Roman citizens, with auxiliary units (infantry cohortes and cavalry alae) recruited from non-citizens. By the fourth century, the legions had been broken down into a greater array of smaller units. In some ways this formalized actual Roman practice, for individual cohorts of 500 men had often operated separately from the main body of their legion. In addition, the different classes of unit had been reorganized. Instead of legions and auxiliaries, the late Roman army was composed of frontier garrison troops (limitanei) and mobile field forces (comitatenses) gathered behind the three main frontiers: the Rhine, the Danube and the east. The field forces were more heavily equipped and a touch better paid, but the garrison troops were formidable too, not part-time soldier-farmers as they have sometimes been portrayed. For particular campaigns, they were often mobilized alongside the field forces. There was also much more specialization at the unit level: regiments of mounted archers (sagitarii), heavy artillerymen (ballistiarii) and plate-armoured cavalry (clibanarii, ‘boiler boys’) were employed. Overall, where Caesar had relied almost exclusively on the legionary footsoldier, there was now a greater emphasis on cavalry. Some of the heavy cavalry units were developed in direct imitation of their Persian counterparts, who had played a major role in the defeats of Gordian, Philip and Valerian. Nonetheless, in terms of numbers the late Roman army continued to be dominated by infantry, especially the mobile field forces; infantrymen, not dependent on the availability of animal fodder and able to move long distances while still fighting effectively, were actually more mobile in strategic terms.

The size of the late Roman army remains a hotly debated topic. We have a pretty good idea of the paper strength of the army under the Severan emperors of the early third century, immediately before the rise of the Sasanians. It consisted of thirty legions of 5,000-plus men each, and a similar number of auxiliaries, making a grand total of around 300,000 troops. But attempts to calculate total army size for the late Empire on a similar basis, even though we have a pretty complete listing of its units dating from aboutAD400 in a source called the Notitia Dignitatum (see pp. 246–8), founder on the fact that the theoretical strength of the various types of unit generated by the reorganization clearly varied, and we are not sure how big some of them were. Argument has turned, therefore, on two reported global figures: one of 645,000; the other, referring specifically to the time of the eastern emperor Diocletian (reigned AD 284–305), of 389,704 plus 45,562 in the fleets, making a total of 435,266 men. Both estimates are problematic. The first figure is given by the historian Agathias, writing in the early 570s, in a passage which favourably contrasts the 645,000 with a military establishment in his own day of 150,000 as a means of criticizing contemporary emperors. It was entirely to his point to exaggerate the past figure. The figure of 435,266 commands, a priori, greater credibility both because of its precision and the fact that the context is non-polemical. But it, too, is given by a sixth- rather than a fourth-century writer, in a text composed over two hundred years after Diocletian’s death, which is hardly ideal. We also know that substantial military reorganization continued after Diocletian’s reign, with the distinction between comitatenses and limitanei becoming fully formalized under Constantine. Even if we accept the figure as accurate, then, there is every reason to suppose that the army continued to expand subsequently; historians’ estimates have therefore ranged between 400,000 and 600,000. Even a lower estimate would suggest that between the early third and mid-fourth centuries the 300,000-strong Roman army increased in size by at least one-third, and quite possibly by substantially more.16

That there was such a large increase is placed beyond doubt, to my mind, not only by the evolving strategic situation – the fact that Rome now faced a rival superpower in the east – but also by the fact that the late third and early fourth centuries saw a major financial restructuring of the Empire. The largest item of expenditure had always been the army: even an increase in size of one-third, a conservative estimate, represented a huge increase in the total amount of revenue that needed to be raised by the Roman state. Ask a modern state to fund an increase of 33 per cent or more in its largest budgetary item, and you would soon see bureaucratic hair turning grey. It is entirely consistent with both the size of the new Persian threat, and of even the lowest estimates for late Roman military expansion, that the fiscal patterns of the Empire had to be radically transformed as part of its response to the rise of the Sasanians. Much of the Roman response to third-century crisis has often been attributed to the emperor Diocletian. But while his reign saw many reforms brought to completion or substantially advanced, most of these changes were long-term processes rather than the product of a single mind. This was certainly the case with the reorganization and expansion of the army, and equally true of the financial reforms required to fund it.

The first fiscal response of third-century emperors to the onset of crisis was to lay claim to all existing sources of revenue. Sometime in the 240s–60s, first of all, long-established city revenues – the proceeds of endowments, local tolls and taxes – were confiscated by the state. City officials continued to have to raise the monies and administer the endowments, but the proceeds could no longer be spent locally. This change has often been blamed on Diocletian, but none of our sources for his reign, many of them hostile to his financial reforms, mentions it. It was one of the easiest possible, and hence probably one of the first, responses to financial crisis. These new funds were not inconsiderable, and in the fourth century a portion was sometimes returned to the cities by emperors who wished to curry local favour.17

This income was not enough, however, to cover the entire cost of the new army, and in the late third century emperors also pursued two further strategies. First, they debased the coinage, reducing the silver content of the denarii with which the army was customarily paid. Those of Gallienus (reigned 253–68), for example, were essentially copper coins, containing less than 5 per cent silver. This strategy produced more coins, but the inevitable result was massive price inflation. Diocletian’s Prices Edict of AD 301 fixed the price of a measure of wheat, which in the second century had cost about half a denarius, at no less than a hundred of the new, debased denarii. Comparative evidence suggests that you have about a month before merchants realize that the new coins are even worse than the old ones and put the prices up, so each debasement bought hard-pressed emperors a brief breathing space. Debasement and price-fixing were no long-term solution, since merchants just took their goods off the shelves and operated a black market instead. In the longer term, the only remedy was to extract a greater proportion of the Empire’s wealth – its Gross Imperial Product – via taxation. This too was instigated in the depths of the third-century crisis, when, at particular moments of stress, emperors would raise extraordinary taxes, in the form of foodstuffs. This bypassed problems with the coinage, but, by the nature of its unpredictability, was very unpopular. Finally, under Diocletian, a new regular tax on economic production, the annona militaris, was fully systematized.18

The sudden appearance of a Persian superpower in the east in the third century thus generated a massive restructuring of the Roman Empire. The effects of the measures it took to counter the threat were not instant, but the restructuring eventually achieved the desired outcome. By the end of the third century, Rome had the strategic situation broadly under control: enough extra troops had been paid for to stabilize the eastern front. In AD 298 the emperor Galerius, Diocletian’s co-ruler, won a major victory over the Persians, and from that point on Persian power was more or less effectively parried. There were some further, occasionally dramatic, Roman losses in the next century, but also some victories, and generally speaking the new military establishment was doing its job. Roman–Persian warfare now centred on periodic sieges of massively defended fortifications rather than seriously damaging campaigns of manoeuvre such as Shapur’s invasions of Syria. Fortresses might occasionally be lost, such as Amida in 359, but this kind of defeat bore no resemblance in scale to the disasters of the third century. The strategic clock could not be put back, but both military and fiscal structures had expanded appropriately to cope with the greater Persian threat.19

It is important to recognize, however, just how much of an effort the Romans had to make in order to achieve this state of affairs. Confiscating city revenues and reforming general taxation were not easy matters. It took over fifty years from the first appearance of the aggressive Sasanian dynasty for Rome to put its financial house in order, and all this required a massive expansion of the central government machine to supervise the process. As we saw in Chapter 1, from AD 250 onwards there was a substantial increase in the number of higher-level imperial bureaucratic posts. Military and financial restructuring, therefore, had profound political consequences. The geographical shift of power away from Rome and Italy, already apparent in embryo in the second century, was greatly accelerated by the Empire’s response to the rise of Persia. And while multiple emperors co-reigning had not been unknown in the second century, in the third it was the political as well as administrative need for more than one emperor that cemented the phenomenon as a general feature of late Roman public life.

As a string of emperors was forced eastwards from the 230s onwards to deal with the Persians, this left the west, and particularly the Rhine frontier region, denuded of an official imperial presence. As a result, too many soldiers and officials dropped out of the loop of patronage distribution, generating severe and long-lasting political turmoil at the top. In what has sometimes been called the ‘military anarchy’, the fifty years following the murder of emperor Alexander Severus in AD 235 saw the reins of Roman power pass through the hands of no fewer than twenty legitimate emperors and a host of usurpers, between them each averaging no more than two and a half years in office. Such a flurry of emperors is a telling indicator of an underlying structural problem. Whenever emperors concentrated, at this time, on just one part of the Empire, it generated enough disgruntled army commanders and bureaucrats elsewhere to inspire thoughts of usurpation. Particularly interesting in this respect is the so-called ‘Gallic Empire’. When Valerian was captured by the Persians in 259, civilian functionaries and military commanders on the Rhine frontier put together their own regime under the leadership of a series of generals, who ran Gaul for the best part of three decades. It was an entirely non-separatist, properly Roman regime: quite simply a way of making sure that a satisfactory slice of the imperial cake was distributed in their own corner of the Empire.20


IN THEIR DIFFERENT WAYS, then, both of the Empire’s most dangerous neighbours profoundly influenced its development. The relatively low level of economic achievement prevalent among the warmongering and politically fragmented Germani imposed a barrier to imperial expansion beyond which it was too unprofitable to continue. As a result, Rome’s European frontiers came broadly to rest along the Rhine and Danube. The Near East, home to an equally martial population, had a stronger history of political cooperation and an economy capable of supporting a larger population in a wider variety of activities. The catalyst provided by the Sasanian dynasty turned the region into a rival superpower, whose appearance on the scene forced the Roman state to take stock. Army, taxation, bureaucracy and politics: all had to adapt in order to meet the Persian challenge. The one aspect of the Empire that did not adapt was its ideological world view, and, within that, the position it allocated to all these ‘barbarians’.

Barbarians and the Roman Order

IN THE SUMMER OF AD 370, a group of shipborne Saxon raiders slipped out of the River Elbe, and headed west along the north coast of continental Europe. Avoiding the defended Roman frontier, they eventually disembarked in northern France, probably somewhere west of the Seine. The Romans quickly brought up enough troops to force them to negotiate. As Ammianus Marcellinus, our best fourth-century Roman historian, reports it:21

After a long and varied discussion, since it seemed to be in the interest of the state, a truce was agreed upon, and in accordance with the conditions that were proposed the Saxons gave as hostages many young men fit for military service, and then were allowed to depart and return home without hindrance to the place from which they had come.

But things were not what they seemed. While negotiating, the Romans secretly placed heavy cavalry, together with some infantry, between the Saxons and their ships:

The Romans with strengthened courage pressed upon the Saxons from all sides, surrounded them, and slew them with their drawn swords; not one of them could again return to his native home, not a single one was allowed to survive the slaughter of his comrades.

Ammianus continues:

Although some just judge might condemn this act as treacherous and hateful, yet on careful consideration of the matter he will not think it improper that a destructive band of brigands was destroyed when the opportunity at last presented itself.

As far as Ammianus was concerned, when it came to despatching barbarians, double-dealing wasn’t a problem.

Killing barbarians still went down extremely well with the average Roman audience. Roman amphitheatres saw many different acts of violence, of course, from gladiatorial combat to highly inventive forms of judicial execution. A staggering 200,000 people, it has been calculated, met a violent death in the Colosseum alone, and there were similar, smaller, arenas in every major city of the Empire. Watching barbarians die was a standard part of the fun. In 306, to celebrate his pacification of the Rhine frontier, the emperor Constantine had two captured Germanic Frankish kings, Ascaricus and Merogaisus, fed to wild beasts in the arena at Trier. He also made very sure that a wider audience around the Empire heard of his triumph.22 If no barbarian kings were available, there were always alternatives. In 383 our old friend Symmachus, then Urban Prefect of Rome, wrote to the emperor Valentinian II to say how much the Roman audience had enjoyed feasting their eyes on the spectacle of some rank-and-file Iranian-speaking Sarmatian being slaughtered by gladiators. What’s striking is Symmachus’ commentary:

Rumour does not conceal the splendid outcome of your wars, but a victory gains greater credence if it is confirmed by sight . . . We have now seen things that surprised us when they were read out to us: a column of chained prisoners . . . led in procession, and faces once so fierce now changed to pitiable pallor. A name which was once terrifying to us [is] now the object of our delight, and hands trained to wield outlandish weapons afraid to meet the equipment of gladiators. May you enjoy the laurels of victory often and easily . . . let our brave soldiers take [the barbarians] prisoner and the arena in the city finish them off.23

For him these deaths symbolized that civilized Roman order would continue to prevail over the barbarian forces of chaos.

The antipathy towards barbarians so uninhibitedly expressed in the arena rested, for articulate Romans, on much more than mere hatred. At more or less the same moment as the Saxons were being ambushed on Rome’s north-west frontier, the orator and philosopher Themistius, employed as an imperial spin-doctor, was standing in front of the Senate of Constantinople to justify the policies of his employer, the emperor Valens. The speech contains one particularly telling remark: ‘There is in each of us a barbarian tribe extremely overbearing and intractable – I mean temper and those insatiable desires which stand opposed to rationality as Scythians and Germans do to the Romans.’24

Barbarians had their own well defined place in this Roman universe, based on a specific vision of the cosmos. Human beings, Romans argued, consist of two elements: an intelligent, rational spirit, and a physical body. Above humankind in the cosmos there exist other beings who, although endowed with greater and lesser powers, all share the characteristic of being formed purely of spirit. Below humankind are animals, encompassing pure physicality. Humanity is unique in combining both spirit and body, and from this flowed the Roman vision of rationality. In fully rational people – such as elite Romans, of course – the rational spirit controlled the physical body. But in lesser human beings – barbarians – body ruled mind. Barbarians, in short, were the reverse image of Romans loving alcohol, sex and worldly wealth.

Barbarian irrationality showed up in other ways too. As far as a Roman was concerned, you could easily tell a barbarian by how he reacted to fortune. Give him one little stroke of luck, and he would think he had conquered the world. But, equally, the slightest setback would find him in deepest despair, lamenting his fate. Where Romans would calculate probabilities, formulate sensible plans and stick to them through thick and thin, hapless barbarians were always being blown all over the place by chance events. Barbarian society was also collectively inferior: a world where might equalled right, and where those with the largest biceps triumphed. Barbarians thus provided the crucial ‘other’ in the Roman self-image: the inferior society whose failings underlined and legitimized the superiorities of the dominant imperial power. Indeed, the Roman state saw itself not as just marginally better than those beyond its frontiers – but massively and absolutely superior, because its social order was divinely ordained. This ideology not only made upper-class Romans feel good about themselves, but was part and parcel of the functioning of Empire. In the fourth century, regular references to the barbarian menace made its population broadly willing to pay their taxes, despite the particular increases necessitated by the third-century crisis.25

Although the strategy worked well enough, casting their neighbours beyond the frontiers as the antithesis of Roman order while using them as a peg on which to hang the burden of taxation was not without its own costs. The image of the barbarian made anyone from outside the Empire seem a threat, and also, by definition, a lesser human being belonging to a benighted society. The overwhelming implications of this attitude were, first, that conflict should be the normal state of relations between Roman and non-Roman; and second, that the Roman Empire should be victorious in whatever it aspired to. What did divine favour mean, if not security against defeat at the hands of those lacking that divine favour? The supreme imperial virtue – again often represented pictorially on coinage as a deity awarding a crown of laurel leaves as this suggests, was one of victory. And any failure to deliver it could be taken as a sign that the current incumbent of the purple was not the right man for the job.26

Imperial spokesmen faced the task, therefore, of angling their accounts of events on the frontier to maintain the required image of imperial invincibility. In early 363, for instance, the emperor Julian took a huge military gamble, leading his army 500 kilometres across Persian soil right up to the outskirts of the capital, Ctesiphon. The Persian King of Kings, Shapur, had let him advance, then sprung a trap. The Romans were forced into a fighting retreat all the way back to home territory. By the end of June, when Julian was killed in a skirmish, the situation was hopeless. The Roman army still had 250 kilometres to go, had more or less run out of supplies, and was managing to retreat only about five kilometres a day because of Persian harassment. Julian’s successor Jovian – elected on the campaign – had no choice but to negotiate a humiliating peace. The Roman army was allowed to depart, but surrendered to the Persians two major cities, Nisibis and Sangara, a host of strongpoints and five border provinces (map 3). But so pressing was the expectation of victory, especially at the start of a reign when the seal of divine approval needed to be particularly evident, that Jovian could not afford to acknowledge defeat. His coinage proclaimed the Persian peace a victory and Themistius was trundled out to reinforce the point. The spin-doctor’s discomfort is only too evident. The best he could come up with was this: ‘The Persians showed that they were voting for [Jovian as emperor] no less than the Romans by throwing aside their weapons as soon as they became aware of the proclamation, and shortly after were wary of the same men of whom before they had had no fear.’ He followed up with the quip, based on a famous story about the election of the Achaemenid King of Kings Darius in 522 BC, that the – obviously irrational – Persians chose their rulers according to the neighing of horses.

Not a bad effort at a brave front, perhaps, but this was one spin that no one was buying. By January 364, Jovian had already faced protests from eastern cities complaining about the surrender and, tellingly, in a speech to the Senate that lasted at least three-quarters of an hour Themistius devoted only about a minute to the Persian question before moving on smartly to more promising matters.27 In this case, policy could not be made to square with expectations of victory, and Themistius, shortly after, was on much safer ground when he could admit it. Jovian died in February 364, and, at the end of the year in a first speech for his successor Valens, Themistius seized upon Jovian’s early death, after only eight months in power, as a clear sign that his rule had not been divinely sanctioned. In this way, the loss to the Persians could be satisfactorily explained, and a nasty dent in the Roman self-image removed.28

But such catastrophic losses even to the Persians were now rare, as we have seen, and Rome held an overall military advantage on its European frontiers. With just the odd white lie, expectations of victory could usually be satisfied and inconvenient reality prevented from scrambling the key message: the barbarian on the other side of the frontier had no place in the Roman order, and was being duly and regularly destroyed. Indeed, violent confrontation was a significant element in Roman foreign policy on all its frontiers, but reality – as much on the Rhine and Danube as in the east – was much more complicated than was implied by the simple ‘them and us’ view.

To explore this reality in more detail, we can narrow the focus to one corner of Rome’s European frontier, the lower reaches of the River Danube separating the Roman diocese of Thrace from the Germanic-speaking Goths who, in the fourth century, dominated lands between the Carpathians and the Black Sea.

Thrace: The Final Frontier

IN 369, THE SAME YEAR that Symmachus’ embassy presented the emperor Valentinian with crown gold (p. 22), a summit meeting took place in the middle of the River Danube, close to the fortress of Noviodunum. Valentinian’s brother, the emperor Valens, ruler of the eastern Empire, pushed off from the south bank in a magnificent imperial barge. From the north bank he was joined by Athanaric, leader of the Tervingi, the Germanic Goths settled closest to the frontier. Athanaric had been at war with Valens for the best part of three years. For once, we have an eyewitness account of the event, penned by Themistius for the Senate of Constantinople. He had attended the meeting as the head of a senatorial embassy to the emperor. As Themistius tells it, Valens managed thoroughly to perplex his enemy:29

Valens was so much cleverer than the man who spoke for the barbarians that he undermined their confidence in him and rendered the verbal contest [on the boat] even more hazardous than the armed [contest of the previous three years]. All the same, having thrown his opponent he then set him on his feet once more, stretched out his hand to him in his confusion and made him a friend before witnesses . . . And so [Athanaric] went away highly contented, in the grip of contrary emotions: at once confident and fearful, both contemptuous and wary of his subjects, cast down in spirit by those aspects of the treaty in which he had lost his case but exulting in those in which success had fallen to him.

Athanaric’s followers were in pretty poor shape too:

[They] were dispersed in groups along the bank in docile and amenable mood, a horde defying enumeration . . . Looking at both banks of the river, [I saw] the [Roman one] glittering with soldiers who in good order looked on with tranquil pride at what was being done, the other burdened with a disordered rabble of suppliants cast down upon the earth.

Athanaric and his Goths thus played their parts perfectly, according to the traditional Roman script. The details of the peace agreement mentioned by Themistius only confirmed Valens’ domination. The emperor now discontinued the annual gifts that the Goths had been accustomed to receive, confined cross-border trade to only two designated centres, and inaugurated a programme of defensive building to ensure that Gothic raiders would have no opportunity for causing further trouble. Expectations of Roman dominance over pathetically inferior barbarians had been magnificently fulfilled.

But looked at more closely, the story as told by Themistius doesn’t quite add up. Hostilities had not been opened by Valens, but by Athanaric. In 364/5, Roman intelligence reports were already indicating that the Goths were becoming restive, and Valens had sent reinforcements to the Danube front. When, in 365, those reinforcements were bribed by Procopius, the uncle of the former emperor Julian, to kick-start his usurpation, Athanaric sent the would-be usurper a contingent of three thousand Goths. If the Goths had been happy being paid to keep the peace, as Themistius reports, why had Athanaric behaved so aggressively? Valens also failed, despite three years of campaigning, actually to defeat the Goths in battle. In 367 and 369 his armies ranged at will in Gothic territories, looting as they went. And they were only kept at bay in 368 by a premature melting of the Alpine and Carpathian snows. The flooding Danube made it impossible for the Romans to string up the pontoon bridges by which they customarily moved their heavy equipment across the river. Through strategic manoeuvre – running away – Athanaric managed to avoid being cornered. By the time peace was made, the Goths were massively inconvenienced and suffering major food shortages, but they were never trapped into total submission in the way that they had been some thirty years earlier, in the time of the emperor Constantine, who had forced their unconditional surrender. Since the Romans had not so decisively defeated them as Themistius would have us believe, it seems odd that the treaty of 369 enforced harsher terms upon them than that of 332.

In his speech, Themistius ‘forgot’ to mention, however, one crucial extra detail. Halfway through Valens’ Gothic campaign, all hell had broken loose on a corner of the Persian front. Having made major gains in Mesopotamia through the treaty with Jovian, the Persian King of Kings Shapur now turned his attention to Caucasia. In 367/8 he ousted the rulers of Armenia and eastern Georgia, who had been Roman allies, and replaced them with his own nominees. Safeguarding the Persian front was much more important to Valens than reducing the Goths to total submission, so that this new threat exerted huge pressure on him to extract his forces from the Balkans and redirect them eastwards. But Valens had already mobilized on the Danube and his taxpayers were expecting victory. He also had the Goths’ support of Procopius to avenge. He thus kept the war going into 369, but when total victory again proved elusive, he needed to make a compromise peace. That the meeting between Valens and Athanaric did generate a compromise is clear. Themistius notes that the Goth was ‘exulting in those [aspects of the treaty] in which success had fallen to him’. The same point is made, interestingly, by the location of the summit meeting. Roman emperors normally paraded their standards triumphantly on barbarian soil, and forced barbarian kings to submit to them there. Only one other waterborne summit is recorded in fourth-century sources, this time on the Rhine – again, a Roman emperor (Valentinian) needed to secure one frontier to tackle a problem on another. That peace was also a compromise.30

The real task facing Themistius in selling the Gothic peace to the Senate now comes into focus. He presented the discontinuation of annual gifts to the Goths as a great gain to the Roman state. In fact, it was a rather small one. The state had used gifts for centuries to build up the position of client kings. We would call it ‘foreign aid’. The great loss to the Romans – which Themistius doesn’t mention – was the right, now rescinded, to call on Gothic military assistance against Persia. What emerges particularly clearly is the slickness of Themistius. A vivid scene of Gothic submission was conjured up for his audience, with Valens all-powerful at the peace-making. And the orator’s bravado performance seems to have done the trick, since two contemporary sources describe the peace as a reasonable end to the war. Valens’ face had been successfully saved.31

For our purposes, however, there is a shadowy but much more important point lurking behind Themistius’ smoke-screen. It is impossible to know everything that Athanaric had in mind, since his precise aims are not recorded by our Roman sources, but he was clearly no mere stock barbarian of the Roman ideological ‘other’. He and his fellow Tervingi had been in receipt of Roman gifts for thirty years but were willing to put them at risk to avoid having to fight for the Empire. The same went for the trading privileges inherent in the open frontier established by their earlier treaty with Constantine. That these privileges were real and enjoyed by the Goths is visible in the archaeological record. Fourth-century Gothic sites are littered with the pottery sherds of Roman amphorae, most of them broken wine containers (by the sixth century biberunt ut Gothi – ‘drinking like Goths’ – had become proverbial). Despite this, Athanaric had a determined agenda to extract the Tervingi from the least acceptable constraints of Roman domination. He was able to rally support for this stance from among his Goths, and then used sophisticated strategies to achieve his ends. At first he had been ready to fight the Empire outright, but when Procopius’ plans for usurpation offered him the opportunity to fiddle in internal Roman politics instead, he took this route – hoping, presumably, that a successful Procopius would grant willingly what the Goths would otherwise have had to extract from Valens by force.

Here, reality contradicts Roman ideology in substantial ways. The usurpation of Procopius saw one Roman allying with a barbarian against another Roman, although, admittedly, Athanaric was no more than a junior ally. Nor was he an aimless barbarian intent only on the nearest bit of plunder. He had, rather, pursued a variety of means to renegotiate the bundle of obligations and privileges that Constantine had imposed on the Tervingi after his great victory of the 330s. Constantine had also tried – in a stock Roman diplomatic manoeuvre – to impress upon the ruling house of the Tervingi the benefits of Roman civilization. One of the hostages sent to Constantinople as part of his treaty was the son of the then ruler. Such hostages could be, and were, executed if the terms of peace were broken. But, more generally, they were used to convince the next generation of barbarian movers and shakers that hostility to Rome was pointless, and that they would be much better off embracing it. Sometimes the strategy worked; in this case it didn’t. The prince of the Tervingi sent to Constantinople was Athanaric’s father, and even though they put up a statue to him behind the Senate house, he was not won over (maybe they should have tried putting it in front). When handing on power in due course to his son, he forbade Athanaric ever to set foot on Roman soil, and Athanaric continued to press for as much separation as possible.32 The shipborne setting of his summit meeting with Valens implicitly acknowledged the Goth’s sovereignty over lands beyond the Danube, and, in the aftermath of the new agreement, Athanaric found himself free to persecute Gothic Christians. Christianization had been promoted among the Goths by previous emperors, as we shall see in a moment, so here was another deliberate rejection of Roman ideologies. No low-level barbarian, Athanaric was a client king with a coherent agenda for renegotiating his relationship with the Roman Empire.

Little Wolf

IF THE REAL PROFILE of Athanaric can be partially recovered from the distorting mirror of Themistius’ speech, two astonishing manuscripts give us much more direct access to the Gothic world of the fourth century. The first is one of the greatest treasures to survive from antiquity: the Codex Argenteus. Now housed in the Uppsala University Library in Sweden, it is a luxury copy of a translation of the four Gospels into the Gothic language. Transcribed in Italy in the sixth century, the book originally comprised 336 pages. Only 187 survived at Uppsala, but much excitement accompanied the discovery of one more, in 1970, in a long-forgotten hiding-place for relics in the cathedral at Speyer in south-west Germany. The text is written in gold and silver ink on purple-dyed parchment of an exceptional fineness – it was made from the skin of newborn (or even unborn) calves. Ink, dye and parchment all mark this out as a colossally expensive book commissioned by an individual of the highest standing, quite likely Theoderic the Amal, Ostrogothic king of Italy in the sixth century. The second manuscript is more modest but, in its own way, equally extraordinary: a plain and quite badly damaged fifth-century text prosaically known as Parisinus Latinus 8907. Most of it is devoted to an account of the Council of Aquileia in 381, when Bishop Ambrose of Milan, a stalwart of what was just about to become Christian orthodoxy, defeated his opponents, and to the first two books of Ambrose’s most famous work, the De Fide (On the Faith). Written into the margins of the De Fide is another work, known only from this battered manuscript: a commentary on Aquileia by Bishop Palladius of Ratiaria, one of Ambrose’s opponents there. This commentary includes a letter written by Auxentius of Durostorum, which, together with the Codex Argenteus, illuminates the extraordinary achievements of one of Athanaric’s humblest subjects: Ulfilas, the Little Wolf of the Goths.33

Born at the beginning of the fourth century, Ulfilas was the offspring of Roman prisoners living among the Tervingi. They were part of a substantial community of captives taken by Goths during the late third century. At this point, Goths were launching seaborne attacks across the Black Sea from southern Russia into Roman Asia Minor. Ulfilas’s family was taken from a small village called Sadagolthina near the city of Parnassus in Cappadocia, located on the northern shores of what is now Lake Tattu in central Turkey. His name, meaning ‘Little Wolf ’, is unequivocally Gothic, showing that the captives adapted linguistically to their new situation; but they continued to use their own languages too. In addition to Gothic, Ulfilas grew up literate in both Latin and Greek, and Greek was probably his language of preference. That he had these accomplishments implies a great deal about the captives’ living conditions. They probably formed a largely autonomous body of farmers, required to hand over a substantial portion of their produce to their Gothic masters but otherwise left more or less to their own devices. Quite a lot of them were firm Christians. Ulfilas, we are told, grew up and matured in his faith in this decidedly polyglot setting, becoming a junior clergyman with the rank of lector in the exiles’ church. This kind of subject community is known to have existed in other barbarian kingdoms in late antiquity, and some were able to preserve a sense of difference over several generations. In the case of Ulfilas, the relatively obscure life of a second-generation involuntary immigrant was about to be transformed by the fact that the Tervingi happened to be the group of Goths settled closest to the Roman frontier at a moment when the Empire was busy converting itself to Christianity.

In the early 340s the emperor Constantius II decided to raise the stakes in the hostage situation in which Athanaric’s father was currently ensnared. Flexing his political muscles in the way he was about to do was only possible, of course, because of the military dominance that Constantius’ father Constantine had established over the Tervingi in the 330s. As one of several initiatives designed to show off his Christian piety, Constantius attempted to boost the fortunes of his fellow Christians living under non-Christian rule. He thus arranged for Ulfilas, already prominent among the prisoner community, to be ordained bishop ‘for the Christians in Gothia’, bringing him to Constantinople for the purpose in 341 as part of an embassy. Ulfilas then went back north of the Danube and for the next seven years ministered happily to his flock. But something went wrong and, in the winter of 347/8, when he found himself at the centre of a diplomatic crisis in Gotho-Roman relations, he was expelled from Gothia by his Tervingian masters, along with a large number of his fellow Gothic Christians. Historians have guessed that he may have spread his message beyond the prisoner community to other Goths, but there was also a wider context. By 348, Constantius wanted to draw another military contingent from the Tervingi for the latest bout of Roman–Persian warfare, and accepting that his Christianizing initiative should stop was perhaps the price he had to pay for it. Nonetheless, Constantius went to the Danube and greeted Ulfilas ‘as if he were Moses himself’.34

It might have seemed like the end, but it was only the beginning. Ulfilas and his followers were settled around the city of Nicopolis ad Istrum, close to the Danube frontier and still in contact with what must have been the many Christians who remained in Gothic territories. It was here that Ulfilas produced the Gothic Bible translation preserved in the Codex Argenteus. His method was simple – he gives a word-for-word rendering of a standard fourth-century Greek Bible text – and his translation owes more to Greek grammar and syntax than to that of the Goths. It was a prodigious feat. According to tradition, Ulfilas translated everything except the Old Testament Book of Kings, which he thought would only have encouraged the Goths to become even more warlike than they already were. A low-status subject member of the Gothic Tervingi had produced the first literary work in any Germanic language.35

This was one part of the Ulfilas story. The other is told in Auxentius’ letter so uniquely preserved in Parisinus Latinus 8907. Constantine’s conversion brought about extraordinary transformations within Christianity. Amongst other things, it became imperative for Christians, who no longer lived in communities mainly isolated from one another by the hostility of the Roman state, to define a set of doctrines. The process began at the Council of Nicaea in 325, where the relationship of God the Son to God the Father was defined as homousios: ‘of the same substance/essence’. But this was just the start of the argument. The Nicene definition of the Christian faith only became fully accepted, after much argument, following the Council of Constantinople in 381, and for much of the intervening fifty-six years official Roman Christianity held to a much more traditional position, describing Christ as ‘like’ (homoios) or ‘similar in substance/essence to’ (homoeusios) God the Father.

Much effort in the interim had gone into constructing coalitions between different Churchmen, many of whom had hitherto simply assumed that they believed the same things. They were now being forced to decide which of a range of theological positions best expressed their understanding of the faith. Into this arena, sometime after 348, strode Ulfilas. Auxentius’ letter contains the statement of belief that Ulfilas left as his last will and testament, and succinctly explains the reasoning behind it. Ulfilas was one of the more traditional Christians: he found the Nicene definition unacceptable because it contradicted the scriptural evidence and seemed to leave little room for distinguishing God the Father from God the Son. In Auxentius’ account:

In accordance with tradition and the authority of the Divine Scriptures, [Ulfilas] never concealed that this God [the Son] is in second place and the originator of all things from the Father and after the Father and on account of the Father and for the glory of the Father . . . holding as greater [than himself] God his own Father [John 14:28] – this he always made clear according to the Holy Gospel.

What’s more, people listened. Again, in Auxentius’ words:

Flourishing gloriously for forty years in the bishopric, [Ulfilas] preached unceasingly with apostolic grace in the Greek, Latin, and Gothic languages . . . bearing witness that there is but one flock of Christ our Lord and God . . . And all that he said, and all I have set down, is from the divine Scriptures: ‘let him that readeth understand’ [Matthew 24:15]. He left behind him several tractates and many commentaries in these three languages for the benefit of all those willing to accept it, and as his own eternal memorial and recompense.

Unfortunately, the tractates and commentaries haven’t survived. Ulfilas ended up on the losing side of doctrinal debate and his works, like those of so many of his party, were not preserved. But we do know from Auxentius and other sources that he was heavily courted not only by Constantius but also by the eastern emperor Valens, and did eventually sign up to the doctrinal settlements they put forward in, respectively, 359 and 370. He also built around himself an influential group of non-Nicene Balkan bishops, who were a major force within the Church. Auxentius was one of these, and Palladius of Ratiaria another. The last image we have is of Ulfilas riding into doctrinal battle yet again, at the age of seventy, at the Council of Constantinople in 381. This was his last hurrah, and the Council’s decisions effectively consigned him and his followers to the footnotes of history. But that was not how it was in his own lifetime. This Gothic subject of humble origins was a major player in the doctrinal debates of the mid-fourth century.36

AGAIN, REALITY confounds image. Viewed through a Roman lens, barbarians were utterly incapable of rational thought or planning; sensualists, they lacked motive, apart from an overwhelming desire for the next fix. But our two fourth-century barbarians were neither stupid nor irrational. At the pinnacle of Gothic society, Athanaric and his councillors were faced with the brute reality of devising ways of coping with overwhelming Roman power. They could hope neither to defeat it in open conflict nor to insulate themselves from it. They could, however, formulate and pursue agendas designed to shape their relations with the Empire in the way that best suited them, while minimizing those aspects of Roman domination they found most oppressive. They could also be desirable allies in wartime and in civil conflict, and could sometimes manipulate matters for their own benefit. Lower down the social scale were communities literate in Greek and Latin who transmitted enough of current Christian culture to generate a man like Ulfilas.

The reality of Roman-Gothic relations was not, therefore, the unremitting conflict between absolute superior and inferior that Roman ideology required. The Romans still held themselves aloof, the dominant party, but Goths could be useful. The periodic conflict between them was part of a diplomatic dance that saw both sides taking steps to maximize their advantage. Barbarians weren’t what they used to be. Even if cast firmly as junior members, the Goths were part of the Roman world.

Client Kingdoms

THIS DIDN’T APPLY exclusively to the Goths on the Danube, even if most fourth-century Germanic societies are not so well documented as the Tervingi. Small-scale raiding into imperial territory was endemic. The Saxon raid of 370 was perhaps more serious than some, but it wasn’t just spin on Themistius’ part that he ends his account of the Gothic war of 367–9 with a vignette of Valens fortifying those parts of the Lower Danube frontier that other emperors hadn’t reached. He and his brother were active in both building fortifications and providing garrisons. But in the fourth century, major conflicts occurred only about once in a generation on Rome’s European frontiers. One of the first acts of the emperor Constantine, in the 310s, was to undertake a major pacification of the Rhine frontier – the lands of the Franks and Alamanni (map 4). We know of no further serious conflict in this region until the early 350s. The trouble that broke out again in 364/5 was to do with a change in Roman policy (a unilateral cut in the foreign aid budget); otherwise, nothing of note occurred here before the end of the 370s. Further east, the Middle Danube frontier facing the Sarmatians, Quadi and Marcomanni saw a major Roman military intervention under the emperor Constantine, but much later in his reign, in the 330s. The next outbreak of violence there came in 357, and another in 374/5. On the Lower Danube, home to the Goths, the settlement of the 330s gave, as we have seen, more or less thirty years of peace.


In each of these campaigns, the Romans – with greater or lesser difficulty – established their military dominance, sometimes just by pillaging widely enough to force submission, sometimes by victory in a set-piece battle. In 357, for instance, the emperor Julian led a Roman force of 13,000 men into action near the city of Strasbourg on the Roman side of the Rhine, against the assembled kings of the Alamanni. He won a stunning victory. Of the 35,000 opponents led by their pre-eminent overking, Chnodomarius, some 6,000 were left dead on the battlefield and countless others drowned trying to flee across the river, while the Romans lost a grand total of 243 soldiers and four high-ranking officers.37 The battle is an excellent example of the continued effectiveness of the remodelled Roman army of the late imperial era. From the massacring of Saxon raiders in northern France to Constantine’s subjugation of the Tervingi, this type of military dominance was the norm at all levels on Rome’s European frontiers.

In one respect, such victories were an end in themselves. They punished and intimidated, and certainly the historian Ammianus considered that it was necessary to hit barbarians regularly to make them keep the peace. On another level entirely, however, military victory was the first act in constructing broader diplomatic settlements. After Strasbourg, Julian spent the next two years on the other side of the Rhine making separate peace treaties with various Alamannic kings, just as his co-emperor Constantius II was doing with other groups on the Middle Danube.

As we have seen, to the Roman public these treaties were all presented as following essentially the same pattern: the barbarians surrendered themselves completely (called in Latin an act of deditio) and were then graciously granted terms in a treaty (Latin,foedus), which made them imperial subjects. In reality, however, the details varied dramatically, both in the degree of subjection enforced and in the practical arrangements. Where the Romans were fully in control of the situation, as Constantius was on the Middle Danube in 357, they might well interfere in their opponents’ political structures, dismantling confederations that appeared overly dangerous and promoting pliant sub-kings to independent authority as seemed to best suit Rome’s long-term interests. The Romans also extracted recruits for their army as part of most agreements, sometimes stipulating as well that larger bodies of men should be provided for particular campaigns. In 357/8 the emperor Julian also made the Alamanni pay reparations for the damage they had caused. These often took the form of grain supplies, as in this instance, but, where this was impossible, labour, wood for construction and cartage were demanded. Giving hostages, as happened with Athanaric’s father, was also quite standard, and sometimes brought greater success. One Alamannic prince was so impressed with the Mediterranean religions he encountered on Roman soil that on his return he renamed his son Serapio in honour of the Egyptian god Serapis. Where the Romans were less in control, labour, raw materials and manpower might have to be paid for, and political structures that had evolved independently given the stamp of approval. Either way, beyond the defended frontier itself lay a belt of largely Germanic client kingdoms that were firmly part of the Roman world.38

This is not to say that these states were entirely under Roman control, or necessarily happy about being junior members of the Roman world order, as we have seen in the case of Athanaric. If other priorities got in the way, then barbarians could find themselves prospering, sometimes temporarily, sometimes more permanently. The early 350s, for instance, saw a rash of usurpations in the western half of the Empire, beginning with the murder of Constans, brother of the then eastern emperor Constantius. Constantius made it his priority to suppress the usurpers, and it was this which allowed Chnodomarius to build up the Alamannic army that would face Julian at Strasbourg. Once the usurpers had been put down, however, the Romans reined in, then utterly defeated, the Alamanni in two years of campaigning. Chnodomarius had been too aggressive, even seizing territory on the Roman bank of the Rhine, for the Romans to contemplate doing a deal. About a decade later, however, a new, pre-eminent leader of the Alamanni appeared: Macrianus. Valens’ brother Valentinian spent half a decade trying to curb his power, making a number of kidnap and murder attempts. But, unlike Chnodomarius, Macrianus never let his ambitions stray on to Roman territory, so that when trouble brewed on the Middle Danube Valentinian could invite him, without too much loss of face, to a shipborne summit on the Rhine of the kind at which Valens had entertained Athanaric on the Danube. There he gave Roman approval to Macrianus’ pre-eminence, and Macrianus proved a reliable Roman ally as long as he lived. These client kingdoms also had political agendas that didn’t involve Rome. Political life among the Alamanni had its own pattern, with kings regularly inviting each other to feasts. We hear too of wars between Alamanni and Franks, and between Alamanni and Burgundians, but nothing of their causes and consequences.39

Overall, then, Rome’s relations with its fourth-century European frontier clients didn’t fit entirely comfortably within the ideological boundaries set by the traditional image of the barbarian. The two parties now enjoyed reciprocal, if unequal, relations on every level. The client kingdoms traded with the Empire, provided manpower for its armies, and were regularly subject to both its diplomatic interference and its cultural influence. In return, each year they generally received aid; and, sometimes at least, were awarded a degree of respect. One striking feature is that treaties were regularly formalized according to norms of the client kingdom as well as those of the Roman state. The Germani had come a long way from the ‘other’ of Roman imaginations, even if the Empire’s political elite had to pretend to Roman taxpayers that they hadn’t. What has also become clear in recent years, is that this new order in Roman–German diplomatic relations was based on a series of profound transformations in Germanic society.

The Transformation of Germanic Europe

THE WRITTEN EVIDENCE does contain some important clues that fundamental changes had occurred in the three and a half centuries separating Arminius from Athanaric. In the mid-third century, the west Germanic tribal names famous from the works of Tacitus suddenly disappear from our sources. Cherusci, Chatti and so forth were replaced by four new ones: Franks and Alamanni on the Rhine frontier, and Saxons and Burgundians further to the east (map 4). South-eastern Europe north of the Black Sea also now saw major political changes. By the fourth century, a huge swathe of territory from Rome’s Danube frontier to the River Don was dominated by Gothic and other Germanic-speaking groups, making late Roman Germania even larger than its first-century counterpart.

The new situation beyond the Black Sea was generated by the migration of Germanic groups from the north-west, largely from what is now central and northern Poland. In a series of independent, small-scale initiatives, between about AD 180 and 320, they had advanced around the outer fringes of the Carpathian Mountains. North of the Black Sea, the migrating groups were competing against each other, against indigenous populations such as the Dacian-speaking Carpi and Iranian-speaking Sarmatians, and against Roman garrison forces. The process was, not surprisingly, violent. The Empire decided to abandon its north Danubian province of Dacia in 275, and large numbers of Carpi were eventually resettled on Roman soil around the year 300. The violence spilled over on to Roman soil in regular raids, and it was during one of these that Ulfilas’ parents were captured. The result was a series of largely Gothic-dominated political units, of which Athanaric’s Tervingi were closest to the Danube. Beyond them, to the north and east, was an unknown number of others.40 We have no idea of relative percentages, but the populations of these units were certainly mixed, with large numbers of Dacians and Sarmatians, not to mention Roman prisoners, living under the political umbrella of immigrant Goths and other Germani. The dominance of the Germanic immigrants is clear, however, from both Roman narrative sources and the linguistic evidence of Ulfilas’ Bible.41

The significance of the name changes on the Rhine frontier and in its hinterland has been hotly disputed. Again, in all probability, immigration was involved. Burgundians do appear in Tacitus’ account of first-century Germania, but significantly to the north-east of the region inhabited by their fourth-century namesakes. It is likely enough that some kind of migration was behind this shift of locale, but, as in the east, it probably did not take the form of a total replacement of the existing population.42 Otherwise, we know that beneath the umbrella of the new names, some of the old groups continued to exist. Bructeri, Chatti, Ampsivarii and Cherusci are all reported in one source as belonging to the Frankish confederation of tribes, and detailed contemporary evidence shows that among the Alamanni several kings always ruled simultaneously, each with his own largely autonomous domain. At the battle of Strasbourg, for instance, Julian faced seven kings and ten princes.

At the same time, however, Alamannic society was by this date consistently throwing up an over-king: an individual in each generation who wielded more power than his peers. Chnodomarius, defeated by Julian at Strasbourg in 357, was one of these, as were Vadomarius at whose rising power Roman policy was next directed, and Macrianus whom Valentinian was eventually forced to recognize in 374. It was not a hereditary position, and it is not recorded either how you became an over-king, or what benefits it brought you. Our Roman sources weren’t interested enough to tell us. The chances are, however, that it involved some financial and military support upon demand, a development of some importance suggesting that the name changes of the third century had a real political significance. In Alamannic territories, a new superstructure had invaded the world of small independent political units characteristic of the first century. It is perfectly possible, although there is no evidence either way, that contemporary Franks and Saxons had developed similarly unifying institutions and habits. Further east, on the Danube, the Gothic Tervingi certainly had. Athanaric ruled a confederation that contained an unknown number of other kings and princes.43

But it wasn’t merely in political structure that fourth-century Germania differed from its first-century counterpart. A range of archaeological evidence has shed new light on the deeper social and economic transformations that brought the world of Athanaric into being. The story begins in the muddy fields just east of the northern sector of Rome’s Rhine frontier. In the early 1960s two small rural sites – Wijster in the Netherlands and Feddersen Wierde in Germany – were excavated. The findings were revolutionary. Both turned out to be farming settlements whose occupants practised mixed arable and pastoral agriculture, and both originated in the first century AD. The revolutionary aspect was that, for most of their history, these had been village communities with large numbers of houses occupied simultaneously: more than fifty in the case of Wijster, thirty at Feddersen Wierde. Furthermore, the settlements were occupied until the fifth century. The importance of this lies in what it implies about agricultural practice.

In the last few centuries BC, an extensive (rather than intensive) type of arable agriculture had prevailed across Germanic Europe. It alternated short periods of cultivation with long periods of fallow, and required a relatively large area of land to support a given population. These early Iron Age peoples lacked techniques for maintaining the fertility of their arable fields for prolonged production, and could use them for only a few years before moving on. Ploughing generally took the form of narrow, criss-crossed scrapings, rather than the turning-over of a proper furrow so that weeds rot their nutrients back into the soil. Ash was the main fertilizer.

This is where the settlements of Feddersen Wierde and Wijster differ. For early in the Roman period, western Germani developed entirely new techniques, using the manure from their animals together, probably, with a more sophisticated kind of two-crop rotation scheme, both to increase yields and to keep the soil producing beyond the short term. For the first time in northern Europe, it thus became possible for human beings to live together in more or less permanent, clustered (or ‘nucleated’) settlements. Further north and east, the muck took longer to spread. In what is now Poland, the territories of the Wielbark and Przeworsk cultures, Germanic settlements remained small, short-lived and highly dispersed in the first two centuries AD. By the fourth, however, the new techniques had taken firm hold. Settlements north of the Black Sea, in areas dominated by the Goths, could be very substantial; the largest, Budesty, covered an area of thirty-five hectares. And enough pieces of ploughing equipment have been found to show that populations under Gothic control were now using iron coulters and ploughshares to turn the earth properly, if not to a great depth. Recent work has shown that villages had emerged in Scandinavia too. More intensive arable agriculture was on the march, and pollen diagrams confirm that between the birth of Christ and the fifth century, cereal pollens, at the expense of grass and tree pollens, reached an unprecedented high across wide areas of what is now Poland, the Czech Republic and Germany. Large tracts of new land were being brought into cultivation and worked with greater intensity.44

THE MAIN OUTCOME of all this was that the population of Germanic-dominated Europe increased massively over these Roman centuries. The basic constraint upon the size of any population is the availability of food. The Germanic agricultural revolution massively increased the amount available, and the increase in population shows up in the cemetery evidence. Cemeteries in continuous use throughout the Roman period all show dramatic rises in burial numbers from the later period.

Other sectors of the economy were also transformed. It is impossible to construct any kind of global overview, but iron production in Germania increased massively. In Poland, production at the two largest centres (in the Swietokrzyskie Mountains and in southern Mazovia) generated in the Roman period 8–9 million kilos of raw iron. This was much more than could have been consumed by local Przeworsk populations, and plenty of smaller extraction and smelting sites have also been recovered, such as the fifteen or so fourth-century smithies clustered on the bank of a river at Sinicy in the Gothic-dominated Ukraine. Similarly with pottery: at the start of the Roman period, the Germani made all of theirs by hand, and for the most part, apparently, on a local and ad hoc basis. By the fourth century, this kind of pottery was being replaced by wheel-made wares, fired at much higher temperatures and hence more durable and more sophisticated. These were the products of highly skilled craftsmen. Whether Germanic potters could make a living just from their pottery is unclear, but economic diversification was certainly under way. The change was most marked in areas of production geared to elite consumption. Grave goods show that glass was a treasured item among Germanic populations in the early centuries AD. Up to about the year 300, all glass found in Germanic contexts was imported from the Roman Empire. This is presumably why it was so valued – rather like Italian handbags are now. In the 1960s, however, at Komarov on the outer fringes of the Carpathians, excavators unearthed a fourth-century glass foundry. Such was the quality of its widely distributed products (all the way from Norway to the Crimea) that they had previously been thought to be Roman imports. The glass factory, complete with moulds, left no doubt that they were made in Germania.

A similar story can be told about precious metals. Up to the birth of Christ, very few indigenously produced items of precious metalwork have been identified in Germanic settings, and in the first two centuries AD the vast majority of decorative items were still being produced in bronze only. By the fourth century, intricate silver safety-pins (fibulae) of a number of types had become familiar items of Germanic dress; and a few examples survive of work on a grander scale, notably one of the silver dishes of the famous treasure unearthed at Pietroasa in Romania in the late nineteenth century. How some, at least, of this ware was produced is suggested by evidence from the village of Birlad-Valea Seaca (in modern Romania), which probably fell within the territory ruled by Athanaric of the Tervingi. A characteristic grave good of Gothic territories north of the Black Sea is a composite comb made from deer antler. Combs had great cultural importance. Hairstyles were used by some Germanic groups to express either groupaffiliations (as in the famous Suebian knot) or status (the long hair of the Merovingian rulers of the Franks). At Birlad-Valea Seaca, excavators unearthed nearly twenty huts containing combs and their constituent parts at different stages of production. Clearly, the entire village was devoted to manufacturing combs.45

There is much more that we’d like to know. Were these combs being produced commercially and exchanged, or was this some kind of subject village from which so many combs were demanded annually as tribute? Whatever the answer, there is no mistaking the extent and importance of the economic revolution that had transformed Germanic Europe by the fourth century. New skills were being developed, and goods were being distributed over far wider areas. Some of this production may have been non-commercial, goods being destined as gifts from one ruler to another, for example, but we know that the Tervingi traded extensively with the Roman world, as did peoples on the Rhine frontier. And although no coinage was produced in Germania, Roman coins were in plentiful circulation and could easily have provided a medium of exchange (already in the first century, Tacitus tells us, the Germani of the Rhine region were using good-quality Roman silver coins for this purpose).

ECONOMIC EXPANSION was accompanied by social revolution. Dominant social elites had not always existed in Germanic Europe, or, at least, their presence is not visible in the cemeteries which are the main source of our knowledge. For much of the first millenniumBC, central and northern Europe was marked by a near-universal adherence to cremation as the main form of burial rite, and grave goods were pretty much the same everywhere: some tatty handmade pottery and the odd decorated pin. Only in the third century BCdid richer burials (the grandest among them often referred to by their German term, Fürstengraber, ‘princely graves’) begin to appear, and they were few and far between. Once again it was first in the Roman imperial period that strikingly disparate quantities of goods began to be buried with different members of the same Germanic communities. In the west, rich graves cluster chronologically, with one group at the end of the first century AD and another at the end of the second. But it is extremely unlikely that ‘princes’ existed only at these isolated moments, so that there is no easy correlation between wealthy burials and social status. Further east, numbers of grave goods built up similarly over the Roman period, but other means, such as huge mounds of stones, for marking out special status were first explored by second-century Germani. Unusually rich or grand burials say most, of course, about the pretensions and claims of those doing the burying, and it has been suggested that rich burials mark moments of intense social competition rather than moments of particular wealth.46

Fortunately, we have some less ambiguous evidence, some of it written, to help us interpret their longer-term significance. Although there is little sign in the first century of the hereditary transmission of political pre-eminence, and leadership even within small groupings was often multiple rather than individual, in the fourth century leadership among the Tervingi was handed down across three generations of the same family: in reverse order, Athanaric, his father the hostage, and the leader of the Tervingi who negotiated with Constantine. The best-informed of our Greek and Latin sources consistently label these leaders ‘judges’, but we don’t know what title ‘judge’ translates. There is every reason to suppose that the power of the second stratum of kings and princes, beneath these overall leaders, was also hereditary. A similar pattern prevailed among the Alamanni. The position of over-king was not hereditary, as we noted earlier, not least because the Romans tended to remove those who achieved that status; but the status of Alamannic sub-kings clearly was. Mederichus, the high-status Alamannic hostage who changed his son’s name to Serapio in honour of the Egyptian god, was the brother of Chnodomarius who led the Alamanni to defeat at Strasbourg in 357. Serapio also ruled as a king, and commanded the army’s right wing in the battle – a sign, perhaps, that he was not overly enamoured of his exotically Mediterranean name. Succession may not have passed straightforwardly from father to son, but Chnodomarius, Mederichus and Serapio represent a royal clan with the ability to pass its power across the generations. The same was probably true of other Alamannic kings. When the Romans eliminated the over-king Vadomarius, considering him to pose too great a threat, they also removed his son Vithicabius from the scene, suggesting that the father’s power was at least potentially heritable.47

Archaeological evidence, too, has shed important light on the fourth-century Germanic elite. Archaeologists have managed to identify, dotted across Germania, some of the centres and dwellings from which it exercised dominion. On the fringes of the Rhine valley, in prime Alamannic country, excavations on the hill known as the Runderberg at the town of Urach have revealed a massive fourth-century timber rampart surrounding an ovoid area of seventy by fifty metres. Inside, several buildings were constructed, including a large timber hall, and smaller buildings dotted the hillside below. The hall was very much the kind of place where Alamannic leaders would have hosted gatherings for each other, and no doubt also feasted their retainers. Whether the smaller dwellings were occupied by retainers, craftsmen or ordinary Alamanni is not yet clear (the excavation has not been fully published). Further east, in Gothic-dominated territories, a few fortified centres, such as Alexandrovka, have been identified and partially explored. On most sites north of the Black Sea, Roman pottery sherds account for between 15 and 40 per cent of the total findings. At Alexandrovka, Roman, largely wine, amphorae sherds amount to 72 per cent; clearly a lot of entertaining went on here. What would appear to be the villa of another Gothic leader has been found at Kamenka-Antechrak. Consisting of four stone buildings with annexes and a courtyard, it covered an area of 3,800 square metres. Its extensive storage facilities and its above-average quantity of Roman pottery (over 50 per cent, this time consisting of both wine amphorae sherds and fine table wares) indicate that it, too, was a major consumption centre. At Pietroasa in Romania, finds of pottery and storage facilities show that a fourth-century Gothic leader reused an old Roman fort for similar purposes. This kind of separate elite dwelling was a new phenomenon.48

Clearly, therefore, the new wealth generated by the Germanic economic revolution did not end up evenly distributed, but was dominated by particular groups. Any new flow of wealth – such as that generated by the Industrial Revolution, in more modern times, or globalization – will always spark off intense competition for its control; and, if the amount of new wealth is large enough, those who control it will erect entirely new authority structures. In Western Europe, for instance, the Industrial Revolution eventually destroyed the social and political dominance of the landowning class who had run things since the Middle Ages, because the size of the new industrial fortunes made the amount of money you could make from farming even large areas look silly. It is hardly surprising, therefore, that Germania’s economic revolution triggered a sociopolitical one, and other archaeological finds have illuminated some of the processes involved.

In antiquity, much of the Jutland peninsula was dotted with pools and extensive peat bogs, now largely dried out by modern land reclamation projects. Recent excavations have shown that, because of their capacity to swallow even very large items, these and similar parts of the North Sea coastal hinterland have long been used by nearby populations as depositories for their sacrifical goods. Individual items – from chariots to gold dishes – datable to a variety of periods have been unearthed. In the Roman period, from the later second to the fourth century AD, a series of weapon sacrifices were made, many of which have emerged from bogs and pools across the area: Vimose, Thorsbjerg, Nydam near Ostersotrrup, and Ejsbøl Mose. Many comprised the arms and equipment of large retinues – even whole armies – which were ritually mutilated as part of the sacrificial act. The most astonishing set of finds of the third century, made at Ejsbøl Mose in southern Jutland, gives us the profile of the force to which the weapons originally belonged. In this excavation archaeologists found the weapons of a small army of two hundred men armed with spears, lances and shields (at least sixty also carried swords and knives); an unknown number of archers (675 arrowheads were excavated) and twelve to fifteen men, nine of them mounted, with more exclusive equipment. This was a highly organized force, with a clear hierarchy and a considerable degree of military specialization: a leader and his retinue, not a bunch of peasant soldiers.49

From this we can begin to see how leaders could so distance themselves from their peers as to make their power hereditary. In the Germanic world of the first century, power ebbed and flowed quickly. But if one generation of a family could use its new wealth to recruit an organized military force of the kind found at Ejsbøl Mose, and then pass on both wealth and retainers, its chances of replicating power over several generations were considerably increased. Organized military forces provided the enforcement by which the claims aired in rich burials were asserted in practice. By the fourth century, retinues were a crucial attribute of the powerful. Chnodomarius, the Alamannic leader defeated by Julian at Strasbourg, had a personal retinue of 200 warriors,50 inviting comparison with the Ejsbøl Mose deposit.

Other sources emphasize that such retinues had plenty of uses outside of battle. The persecution of Christians which the Goth Athanaric launched after partially extracting the Tervingi from Roman domination in 369 generated a document of particular vividness, the Passion of St Saba, the story of the persecution and death of the Gothic martyr of that name. Saba was a ‘proper’ member of the Tervingi, not the descendant of Roman prisoners. The Passion was written on Roman territory, where the saint’s body was found after his death. Among the many precious details we are given is that intermediate-level leaders among the Tervingi had their own retinues and used them to enforce their will. It was a pair of heavies sent by a certain Atharid who eventually did Saba to death by drowning.51

Retinues also help explain the nature of fourth-century seats of power. They were built and functioned, as we have seen, as centres of consumption (like the Runderberg, or Pietroasa in Romania). From the early medieval texts we learn that generous entertaining was the main virtue required of Germanic leaders in return for loyal service, and there is no reason to suppose this a new phenomenon. It required not only large halls, but also a regular flow of foodstuffs and the means to purchase items such as Roman wine, not produced by the local economy. As the existence of specialist craftworkers also emphasizes, Germania’s economy had developed sufficiently beyond its old Jastorf norms to support a far larger number of non-agricultural producers.

The bog deposits make another crucial point. As sacrifices to the gods, they were probably thank offerings for victory: the Ejsbøl Mose deposit celebrates the destruction of the 200 men whose weapons were consigned to its depths. There’s no way of knowing exactly who they were. Were they the army of one small Germanic group defeated by that of another? Tacitus offers a revealing commentary on some Chatti and those who defeated them, a group of Hermenduri, in a struggle over salt deposits: ‘Both sides, in the event of victory, had vowed their enemies to Mars and Mercury. This vow implied the sacrifice of the entire beaten side with their horses and all their possessions.’52 The ritual sacrifice of defeated enemies was clearly not new. Just one of these small first-century tribal groupings could have put more than a couple of hundred men in the field, so that the Ejsbøl Mose deposit may celebrate the destruction of a bunch of rootless warriors on the make – possibly defeated while raiding south Jutland for booty, or in order to establish the kind of dominance that would have allowed them to extract tribute and foodstuffs more regularly. Either way, the find shows that while new flows of wealth usually end up being distributed unequally, this never happens without conflict.

Another feature of much of Germania in the Roman period was a marked increase in the number of weapons burials. Military retinues were not only the result of sociopolitical revolution, but also the vehicle by which it was generated, and large-scale internal violence was probably a feature of the Germanic world from the second to fourth centuries. The hereditary dynasts who dominated the new Alamannic, Frankish and Saxon confederations probably established their power through aggressive competition. The same was true, in a slightly different context, of the Gothic world further east. There, a much larger element of migration was involved, but to create the confederations such as Athanaric’s Tervingi, indigenous populations had to be subdued and new hereditary hierarchies established. In both east and west, the growing wealth of the region generated a fierce struggle for control, and allowed the emergence of specialist military forces as the means to win it. The outcome of these processes was the larger political confederation characteristic of Germania in the fourth century.

The Beginnings of Feudalism?

SOME SCHOLARS have concluded that, already in fourth-century Germanic society, it was only a small aristocratic class, well equipped with armed retainers, that mattered. There are, however, many third- and fourth-century burials, apart from the richest, that containsome grave goods: males with weapons and females with quite sophisticated arrays of personal jewellery. These burials are far too numerous to belong just to kings and a feudal nobility. Later, written evidence offers strong hints as to whose they might have been. In the late fifth and early sixth centuries, Germanic successor states to the western Roman Empire produced large numbers of legal texts. These consistently portray Germanic (and Germanic-dominated) societies at this later date as comprising essentially three castes: freemen, freedmen and slaves. Unlike its Roman counterpart, where the offspring of freedmen were completely free – and thus freemen – freedman status in the Germanic world was hereditary. Intermarriage between the three castes was banned, and a complicated public ceremony was required for an individual to jump across any of the divides. This mode of legal categorization is widely found – amongst Goths, Lombards, Franks and Anglo-Saxons, for instance. A relatively large freeman class, rather than a small feudal nobility, is also visible playing important political and military roles in the Ostrogothic Italian kingdom, and important political, military and landowning roles in the Frankish and Lombard. Freemen were probably also the subjects of the weapons burials of fifth- and sixth-century Anglo-Saxon England, which were clearly used to claim status rather than merely to signal that the individual had been a warrior.53

Given that much more wealth had flowed into Germanic society between the fourth and the sixth centuries, as various Germanic groups took over parts of the Roman Empire, I don’t believe that political participation could have been any less in the fourth century than in the sixth. If anything, it ought to have been broader. So if a relatively numerous freeman class still existed in the sixth century, it surely did two hundred years before. In other words, a quasi feudal warrior aristocracy did not yet dominate Germania in the late Roman period. And Roman sources, despite their lack of interest in the inner working of Germanic societies, provide just enough evidence to confirm the point. Fourth-century Gothic kings couldn’t just issue orders, for instance, but had to sell their policies to a relatively broad audience, and Gothic armies of about AD 400 contained large numbers of elite fighters – freemen, in other words – not just a few warrior aristocrats. These elite fighters had their own fighting dependants; the later law codes state that freedmen (but not slaves) fought, presumably alongside the freemen whose dependants they were.54 This is not to say that all freemen were equal: some were much richer than others, especially if high in royal favour. But social power was not yet confined to a small nobility.

How kings and nobles, complete with their retinues, interacted with the rest of freeman society is not something that archaeology can shed much light upon. Nor are the Roman sources much help. But, to be able to feed and reward them, every figure with a substantial armed retinue – all the Alamannic kings, and the ‘judges’ and kings of the Tervingi – must have established some rights to economic support from freemen and their dependants. In fourth-century Germania there is no sign of the bureaucratic literacy necessary for large-scale taxation, but agricultural produce must have been regularly exacted. Again, therefore, the situation had clearly moved on from the first century, when contributions were occasionally made to distinguished chiefs on a voluntary basis (as Tacitus tells us in his Germania). Obviously, kings were responsible for representing their subjects in any negotiations with outside powers – such as the summit meeting between Athanaric and Valens – and for formulating ‘foreign policy’. They must also have had the right to require military service of their subjects, as foreign policy often involved little more than deciding whom to make war upon. The job description also included some kind of legal function. At the very least, kings will have judged disputes between their grander subjects. Whether they had the right to make general laws, as opposed to decisions in specific cases, is more doubtful. Law-making in the Germanic kingdoms of the post-Roman west looks like a new function, and, even then, was exercised only in the context of consensus. When a law code was devised, it was at assemblies of the great and the good, and issued in the name of all.55

Fourth-century Roman sources shed little light on how precisely kings and their retinues intersected with this freeman caste, but the Passion of St Saba does get us a bit closer. The persecution of Christians among the Tervingi was a policy decision of the overall leadership, involving the sub-kings as well as Athanaric himself. Enforcement, however, was largely in the hands of local village communities, retainers unfamiliar with local circumstances being sent round from time to time to check on progress. In the case of Saba’s village, this gave the locals every chance to frustrate a policy with which they were clearly out of sympathy. Faced with the order to persecute, they swore false oaths that there weren’t any Christians amongst them. This village, at least, clearly wanted to protect its Christians from Athanaric’s persecution, and there was nothing his retainers could do about it. They had no idea who might or might not be a Christian; it was because Saba wouldn’t go along with the deception that he was martyred.56

Germanic society remained, then, a broadly based oligarchy with much power in the hands of a still numerous freeman elite. It had some way to go before it reached the feudal state of the Carolingian era.

Rome, Persia and the Germans

OUR EXPLORATION OF the changes that remade the Germanic world between the first and fourth centuries clearly shows why Roman attention remained so firmly fixed on Persia in the late imperial period. The rise of that state to superpower status had caused the massive third-century crisis, and Persia remained the much more obvious threat, even after the eastern front had stabilized. Germania, by contrast, even in the fourth century, had come nowhere close to generating a common identity amongst its peoples, or unifying its political structures. Highly contingent alliances had given way to stronger groupings, or confederations, the latter representing a major shift from the kaleidoscopic first-century world of changing loyalties. Although royal status could now be inherited, not even the most successful fourth-century Germanic leaders had begun to echo the success of Ardashir in uniting the Near East against Roman power. To judge by the weapons deposits and our written sources, fourth-century Germani remained just as likely to fight each other as the Roman state.

That said, the massive population increase, economic development and political restructuring of the first three centuries AD could not fail to make fourth-century Germania much more of a potential threat to Roman strategic dominance in Europe than its first-century counterpart. It is important to remember, too, that Germanic society had not yet found its equilibrium. The belt of Germanic client kingdoms extended only about a hundred kilometres beyond the Rhine and Danube frontier lines: this left a lot of Germania excluded from the regular campaigning that kept frontier regions reasonably in line. The balance of power on the frontier was, therefore, vulnerable to something much more dangerous than the periodic over-ambition of client kings. One powerful exogenous shock had been delivered by Sasanian Persia in the previous century – did the Germanic world beyond the belt of closely controlled client kingdoms pose a similar threat?

Throughout the Roman imperial period, established Germanic client states periodically found themselves the targets of the predatory groups settled further away from the frontier. The explanation for this is straightforward. While the whole of Germania was undergoing economic revolution, frontier regions were disproportionately affected, their economies stimulated not least by the presence nearby of thousands of Roman soldiers with money to spend. The client states thus tended to become richer than outer Germania, and a target for aggression. The first known case occurred in the mid-first century AD, when a mixed force from the north invaded the client kingdom of one Vannius of the Marcomanni, to seize the vast wealth he had accumulated during his thirty-year reign.57 And it was peripheral northern groups in search of client-state wealth who also started the second-century convulsion generally known as the Marcomannic War. The same motivation underlay the arrival of the Goths beside the Black Sea. Before the mid-third century, these lands were dominated by Iranian-speaking Sarmatian groups who profited hugely from the close relations they enjoyed with the Roman state (their wealth manifest in a series of magnificently furnished burials dating from the first to third centuries). The Goths and other Germanic groups moved into the region to seize a share of this wealth.

The danger posed by the developing Germanic world, however, was still only latent, because of its lack of overall unity. In practice, the string of larger Germanic kingdoms and confederations – now stretching all the way from the mouth of the Rhine to the north Black Sea coast – provided a range of junior partners within a dominant late Roman system, rather than a real threat to Rome’s imperial power. The Empire did not always get what it wanted in this relationship, and maintaining the system provoked a major confrontation between senior and junior partners about once every generation. Nonetheless, for the most part, the barbarians knew their place: none better than Zizais, the leader who approached the emperor Constantius for assistance in 357:

On seeing the emperor he threw aside his weapons and fell flat on his breast, as if lying lifeless. And since the use of his voice failed him from fear at the very time when he should have made his plea, he excited all the greater compassion; but after several attempts, interrupted by sobbing, he was able to set forth only a little of what he tried to ask.58

An inability, at first, to speak, then a little quiet sobbing and the stuttering out of a few requests, did the trick. Constantius made Zizais a Roman client king and granted him and his people imperial protection. Woe betide the barbarian who forgot the script.

The later Roman Empire was doing a pretty good job of keeping the barbarians in check. It had had to dig deep to respond to the Persian challenge, but it was still substantially in control of its European frontiers. It has long been traditional to argue, however, that extracting the extra resources needed to maintain this control placed too many strains on the system; that the effort involved was unsustainable. Stability did return to Rome’s eastern and European frontiers in the fourth century, but at too high a price, with the result that the Empire was destined to fall – or so the argument goes. Before exploring the later fourth and fifth centuries, it is important to examine the Empire of the mid-fourth more closely. Was it a structure predestined to collapse?

If you find an error please notify us in the comments. Thank you!