IN THE WINTER OF 375/6, rumour reached Rome’s Danube frontier that heavy fighting was under way in eastern Germania north of the Black Sea. Ammianus Marcellinus reports:1 ‘In the beginning the news was viewed with contempt by our people because wars in those districts were not ordinarily heard of by those living at a distance until they were either over, or had at least died down for a time.’ You could hardly blame the imperial authorities for not taking the matter too seriously. The migration of the Goths and other Germani in the midthird century had prompted a political reconfiguration that had led to a hundred years of relative stability in the region. Moreover, the trouble then had come from the north-west (present-day Poland and Byelorussia), not the north-east (modern Ukraine). The last time the north-east had posed a problem was when the Sarmatians had swept all before them in the fifty years either side of the birth of Christ, three centuries earlier. But the Romans quickly learned the error of their ways.

In the summer of 376, a vast throng of people – men, women and children – suddenly appeared on the north bank of the River Danube asking for safe haven in Roman territory. One source, not our best, reports that 200,000 refugees appeared beside the river; Ammianus, that there were too many to count. They came with innumerable wagons and the animals to pull them, presumably their plough-oxen, in the kind of huge procession that warfare has generated throughout history. There were certainly many individual refugees and small family groups, but the vast majority were Goths organized in two compact masses and with defined political leaderships. My own best guess is that each was composed of about 10,000 warriors. One group, the Greuthungi, had already moved a fair distance from lands east of the River Dniester, in the present-day Ukraine, hundreds of kilometres from the Danube. The other comprised the majority of Athanaric’s Tervingi, now led by Alavivus and Fritigern, who had broken away from their former leader’s control to come here to the river.2

If the size of the immediate problem for Roman frontier security was bad enough, the refugees’ identity was even more ominous. Though the first reports had concerned fighting a long way from the frontier zone, the two large bodies of Gothic would-be immigrants camped beside the river were from much closer to home. The Tervingi, in particular, had been occupying lands immediately north of the Danube, in what is now Wallachia and Moldavia, since the 310s at the latest. Whatever was going on in the far north-east was no local skirmish; its effects were being felt throughout the region north of the Black Sea.

The Romans quickly learned what lay behind all the mayhem. Again in Ammianus’ words: ‘The seed-bed and origin of all this destruction and of the various calamities inflicted by the wrath of Mars, which raged everywhere with extraordinary fury, I find to be this: the people of the Huns.’

Ammianus was writing nearly twenty years later, by which time the Romans had a better understanding of what had brought the Goths to the Danube. Even in the 390s, though, the full effects of the arrival of the Huns were far from apparent. The appearance of the Goths beside the river in the summer of 376 was the first link in a chain of events that would lead directly from the rise of Hunnic power on the fringes of Europe to the deposition of the last western emperor, Romulus Augustulus, almost exactly one hundred years later. None of this was even remotely conceivable in 376, and there would be many twists and turns on the way. The arrival of Goths on the Danube marked the start of a reshuffling of Europe-wide balances of power, and it is to this story that the rest of the book is devoted. We must begin, like Ammianus, with the Huns.

From the ‘Ice-Bound Ocean’

THE ORIGINS OF the Huns are mysterious and controversial. The one thing we know for certain is that they were nomads from the Great Eurasian Steppe.3 The Eurasian Steppe is a huge expanse, stretching about 5,500 kilometres from the fringes of Europe to western China, with another 3,000 kilometres to its north and east. The north–south depth of the steppe ranges from only about 500 kilometres in the west to nearly 3,000 in the wide-open plains of Mongolia. Geography and climate dictate the nomadic lifestyle. Natural steppe grasslands are the product of poorish soils and limited rainfall, which make it impossible, in general terms, for trees and more luxurious vegetation to grow. The lack of rainfall also rules out arable farming of any sustained kind, so that the nomad makes a substantial part of his living from pastoral agriculture, herding a range of animals suited to the available grazing. Cattle can survive on worse pasture than horses, sheep on worse pasture than cattle, and goats on worse than sheep. Camels will eat anything left over.

Nomadism is essentially a means of assembling distinct blocks of pasture, which between them add up to a year-round grazing strategy. Typically, modern nomads will move between upland summer pasture (where there is no grass in the winter because of snow and cold) and lowland winter pasture (where the lack of rain in summer means, again, no grass). In this world, grazing rights are as important in terms of economic capital as the herds, and as jealously guarded. The distance between summer and winter pasture needs to be minimal, since all movement is hard both on the animals and on the weaker members of the human population. Before Stalin sedentarized them, the nomads of Kazakhstan tended to move about 75 kilometres each way between their pastures. Nomadic societies also form close economic ties with settled arable farmers in the region, from whom they obtain much of the grain they need, though some they produce themselves. While part of the population cycles the herds around the summer pastures, the rest engage in other types of food production. But all the historically observed nomad populations have needed to supplement their grain production by exchanging with arable populations the surplus generated from their herds (hides, cheese and yoghurt, actual animals and so on). Often, this exchange has been one-sided, with the arable population getting in return no more than exemption from being raided, but sometimes the exchange has been properly reciprocal.

Nomadism, or part-nomadism, has never been the preserve of any particular linguistic or cultural population group. Across the Great Eurasian Steppe many peoples have, at different times, adopted nomadic lifestyles. In the first three centuries AD the western end of the steppe – from the Caspian Sea to the Danube – was dominated by Iranian-speaking Sarmatian and Alan nomads. These had ousted Scythian nomads, also Iranian-speaking, in the last two or three centuries BC. By the sixth century AD at the latest, Turkic-speaking nomads were dominant from the Danube to China, and a Mongol-speaking nomad horde would cause untold devastation in the high Middle Ages. Other population groups, too, took to nomadism. The Magyars who arrived in central Europe at the end of the ninth century spoke – as their Hungarian descendants still do – a Finno-Ugrian language that suggests they may have come from the forest zone of north-eastern Europe, the only other region where such languages are spoken.

Where the Huns fit into this sea of cultural possibilities is unclear. Ammianus Marcellinus knew more about them than did our other Roman sources, but he didn’t know much. His best shot is that they came from beyond the Black Sea ‘near the ice-bound ocean’. They were not literate, so leave us no records of their own to go on, and even their language affiliation is mysterious. Failing all else, linguists can usually decode basic linguistic affiliations from personal names, but even this doesn’t work with the Huns. They quickly got into the habit of using Germanic names (or perhaps our sources preserve the Germanicized versions or Germanic nicknames given them by their Germanic neighbours and subjects), so that the stock of properly Hunnic personal names is much too small to draw any convincing conclusions. They were probably not Iranian-speaking, but whether they were the first Turkic-speaking nomads to explode on to the European scene, as some have argued, remains unclear.4 With such pathetic sources of information, Hunnic origins can only remain mysterious, but a little spice has been added by a famous controversy over whether the Huns were in fact the nomadic Hsiung-Nu, well known from imperial Chinese records.

In the centuries before and after the birth of Christ, the Hsiung-Nu – under the leadership of their Shan-Yu5 – harassed the north-west frontiers of Han China, extracting from it huge quantities of tribute in silks, precious metals and grain. They also contested the control of some of its important western territories, particularly the Tarim Basin where the Silk Road (which started to operate in the last century BC) reaches China. Under pressure from Han armies, they split in AD 48 into northern and southern branches. The southern Hsiung-Nu were subsequently brought into the Chinese orbit, becoming an important force within the imperial system. The northerners remained external, independent and highly troublesome until AD 93, when the Chinese government paid another nomadic group, the Hsien-Pi, to launch a devastating attack upon their homelands. Many northern Hsiung-Nu (reportedly 100,000 households) were absorbed into the victorious Hsien-Pi confederation, but others fled ‘to the west’. That’s the last we ever hear of the northern Hsiung-Nu in the Chinese records.

The Huns we’re concerned with appear suddenly in Roman records in the third quarter of the fourth century. The problem inherent in the superficially attractive equation of these people with the Hsiung-Nu is this: we have gaps between the Chinese and Roman records of nearly 300 years (AD 93 to about 370) and 3,500 kilometres to account for. Moreover, the Huns known to the Romans had a completely different form of political organization from the Hsiung-Nu’s. After AD 48, both branches of the latter had their own Shan-Yu, but the Huns arrived in Europe with a multiplicity of ranked kings and no sign of one dominant figure. The surviving ethnographic descriptions – such as they are – also raise objections. The Hsiung-Nu customarily wore their hair in a long pony-tail; the Huns did not. The two groups used similar weaponry, and bronze kettles are customarily found among their archaeological remains. Given this, there may be some connection, but it clearly won’t do just to say that the Hsiung-Nu had started running west in AD 93 and kept going until they hit Europe as the Huns. The Great Eurasian Steppe is a vast place, but it didn’t, even then, take 300 years to cross. Equally, like most nomadic empires, that of the Hsiung-Nu was a confederation, comprising a smallish Hsiung-Nu core and many other subject groups. The ancestors of our Huns could even have been part of the confederation, therefore, without being ‘real’ Hsiung-Nu. Even if we do make some connection between fourth-century Huns and first-century Hsiung-Nu, therefore, an awful lot of water had passed under an awful lot of bridges during 300 years worth of lost history.6

Roman sources also give us only a very general idea of what brought the Huns to the fringes of Europe. For Ammianus, it was enough just to point out that they exceeded ‘every measure of savagery’ and ‘were aflame with an inhuman desire for plundering others’ property’. The most commonly repeated story in the Roman sources claimed their landing up at Europe’s gates was partly an accident. Some Hunnic hunters, out after game one day, trailed a hind through a marsh into new lands of which they had previously been ignorant. This kind of tale rubbed off on early twentieth-century commentators, who tended to suppose that the Huns had for centuries been engaged in nomadic wanderings in different parts of the Eurasian Steppe, and one year just happened to wander on to the fringes of Europe.7 But this was before anthropologists understood quite so clearly that nomads do not wander around at random, but move cyclically between carefully designated pastures. Given that grazing rights are a key element in nomad subsistence, and guarded so jealously, shifting from one set of pastures to another could never be an accident.

Unfortunately, we can only guess at the motives behind the Huns’ decision to shift their centre of operations westwards. The story of the hind concludes with the hunters telling the rest of the Huns of the wonders of the new land they’d found, and Ammianus, too, picked out the motive of economic gain. The idea that it was the wealth of the northern shores of the Black Sea that attracted Hunnic attentions is perfectly plausible. While less extensive, the grazing lands of the western steppe are rich, and have attracted many a nomad group over the years. The area north of the Black Sea was occupied by client groups of the Roman Empire, who benefited economically from different relationships with the Mediterranean world, and there is no reason to doubt that Huns also felt its call. At the same time, in the case of some later nomad groups for whom we have more information, a move on to the western edge of the steppe was often associated with the desire to escape a more powerful nomad confederation operating towards China. The Avars, who would have much the same kind of impact on Europe as the Huns, but two centuries later, were looking for a safe haven beyond the reach of the western Turks, when they appeared north of the Black Sea. At the end of the ninth century, likewise, the nomadic Magyars would move into Hungary because another nomad group, the Pechenegs, was making life intolerable for them further east. In the case of the Huns, we have no firm indication that a negative as well as a positive motivation was at work, but we can’t rule it out. Further east, in the later fourth century, the Guptas were pushing on to the Silk Road from northern India, and by the early to mid-fifth century the Hephthalite Huns were ruling the roost somewhere between the Caspian and Aral Seas. As early as the 350s, this reconfiguration of the balance of power was reverberating further east on the steppe, causing the Chionitae to move into the fringes of the Persian Empire, east of the Caspian Sea.8 It may also have played a role in the Huns’ decision to shift their grazing lands westwards.

Mysterious as the Huns’ origins and animating forces may remain, there is no doubt at all that they were behind the strategic revolution that brought the Goths to the Danube in the summer of 376. It is normally assumed that at that time they were fleeing from Huns who had suddenly exploded en masse on to the northern Black Sea littoral. It is further assumed that these Huns were virtually breathing down the Goths’ necks as they scrambled for the Danube in the hope of securing asylum inside the Empire, and that, once the Goths had reached Roman territory, the Huns immediately became the dominant power in the lands adjacent to the river. This is what you will find stated more or less explicitly in most modern accounts: Huns arrive suddenly (375/6); Goths leave in panic for the Empire (376); Huns become dominant beside the Danube (from 376).

This pattern is based on the account given by Ammianus, who paints a highly convincing picture of Gothic panic: ‘The report spread widely among the other Gothic peoples that a race of men hitherto unknown had now arisen from a hidden nook of the earth, like a tempest of snows from the high mountains, and was seizing or destroying everything in its way.’ We need to look past the rhetoric, however, at what Ammianus is actually telling us. After first subjugating the Alans, the Huns then started attacking the Gothic Greuthungi. The resistance of the Greuthungi was led by Ermenaric, who eventually gave up and seems to have allowed himself to be ritually sacrificed for the safety of his people.9 Ammianus’ wording is a little vague, but the reflex, documented among several ancient groups, to hold their political leadership responsible for the fate of the group, is an interesting one. When times got tough, it was seen as a sign from the gods that the old leader had offended them and needed to be sacrificed in propitiation of the offence. Ermenaric was succeeded by Vithimer, who carried on the fight but was eventually killed in battle.

At this point, control of the Greuthungi passed to two military leaders, Alatheus and Saphrax, who ruled in the name of Vithimer’s son Vitheric. Having decided to retreat to the banks of the River Dniester, they were met there by a force of Tervingi under Athanaric. But Athanaric was now attacked from the rear by some Huns, who had found an alternative ford over the river, and retreated back to his heartlands closer to the Carpathian Mountains. There he attempted to stem the Hunnic tide by constructing a fortified line against them. In my view, this was probably the old Roman walls on the River Olt, the Limes Transalutanus.10 But the plan came to naught. The Tervingi were harassed by more Hunnic attacks as they worked on the defences, which damaged their confidence in Athanaric’s leadership. Most of the Tervingi broke with him at this point, and under new leaders, Alavivus and Fritigern, came to the Danube to request asylum inside the Roman Empire. The Greuthungi of Alatheus and Saphrax opted for a similar strategy, following the Tervingi to the river (map 5).11

Some of these events unfolded very quickly. From the death of Vithimer in battle, the action is pretty continuous down to the arrival of both Tervingi and Greuthungi on the banks of the Danube. Even in its entirety, this sequence needn’t have occupied any great length of time. If, as seems likely, the Goths arrived sometime in late summer or early autumn 376, then Vithimer’s death need be placed no more than a year before. In principle, even a few months would have been sufficient for the intervening events, which would place Vithimer’s death between mid-375 and early 376. Given that a good time for agriculturalists to move on is after they’ve taken in the harvest, it was perhaps most likely late summer or early autumn 375 that the Greuthungi took to the road.12


This somewhat breathless last act, however, followed a more measured drama. It is impossible to date precisely, because Ammianus gives us only vague indications of time; but what he does tell us is suggestive. He states, first of all, that Ermenaric resisted the storm brewed up by the Huns ‘for a long time’ (diu). We also hear that Ermenaric’s successor Vithimer fought ‘many engagements’ (multas . . . clades) against the Huns until he was killed in battle. There is obviously no way to be sure how long all this took, but the swift denouement which followed Vithimer’s death clearly ended a longer struggle, and it was the Greuthungi’s decision to move that precipitated the final crisis. How far back in time the preceding struggle might have gone on is a matter of judgement, but the nature of Hunnic operations does have a bearing on the argument.

To secure their entry to the Empire, first of all, Gothic embassies left the banks of the Danube to seek out the emperor Valens and put their case. Valens, however, was in Antioch – which meant a round trip of over 1,000 kilometres; even so the ambassadors were not deterred. Once they reached Antioch, the two parties had to confer and decisions had to be made, then communicated back to the Roman commanders on the Danube. All of this must have taken well over a month, during which time the mass of Goths continued to sit beside the river, more or less patiently, waiting for the green light to cross. There is no record of any Hunnic attacks upon them during this period. Furthermore, the Huns who attacked Athanaric came in small groups, sometimes weighed down by booty:13 raiders, therefore, rather than conquerors. The Huns’ political organization at this date didn’t run to an overall leader but comprised a series of ranked kings with plenty of capacity for independent action. When he was trying to fend off the Greuthungi’s Hun-generated military problems, for instance, Vithimer was able to recruit other Huns to fight on his side.14 In 375/6, there was no massive horde of Huns hotly pursuing the fleeing Goths: rather, independent Hunnic warbands were pursuing a variety of strategies against a variety of opponents.

What was happening, then, was not that a force of Huns conquered the Goths in the sense we normally understand the word, but that some Goths decided to evacuate a world that was becoming ever more insecure. As late as 395, some twenty years later, the mass of Huns remained further east – much closer, in fact, to the northern exit of the Caucasus than to the mouth of the Danube.15 And it was other Gothic groups, in fact, not the Tervingi or Greuthungi, who continued to provide Rome with its main opposition on the Lower Danube frontier for a decade or more after 376. The Romans had to deal with a heavy assault on the same front launched by a second force of Greuthungi under one Odotheus in 386; and still more Goths – perhaps the leftover Tervingi who hadn’t followed Alavivus and Fritigern to the Danube – were operating somewhere in the Carpathian area at much the same time.

The Golden Bow

NONE OF THIS MAKES the arrival of the Huns any less revolutionary. While small-scale trouble was endemic to the Danube frontier, as everywhere else, strategic revolution was rare. Roman imperial history had seen only two such moments in the northern Black Sea region. A varied climate and ecology is one of the area’s chief peculiarities. Between the Carpathians and the Don there is enough water, particularly in the river valleys, to support arable agriculture, but east of the Don grain cannot be grown without irrigation. At the same time, the southern part between the Carpathians and the Don, just beyond the Black Sea coastal strip, is dry enough to generate steppe conditions. In this fringe of Europe, adjacent areas are ecologically suited, therefore, to nomads and agriculturalists and, in antiquity, the region was dominated by first one type of population group and then the other. Alongside the Scythian nomads, Germanic-speaking agriculturalists, Bastarnae and others, had thrived in the last few centuries BC. Their domination was broken by nomadic Iranian-speaking Sarmatians around the year zero. Two hundred years later agricultural Goths pushed south and east around the Carpathians, extending their domain as far east as the Don, subduing those Sarmatians who remained. What was it about the Huns, then, that allowed them in the later fourth century to redress the military balance in favour of the nomadic world?

The Romans quickly came to appreciate where the military strength of the Huns lay. Ammianus describes no Hunnic battle in detail, but leaves us this general description that gets straight to the point:

[The Huns] enter battle drawn up in wedge-shaped masses . . . And as they are lightly equipped for swift motion, and unexpected in action, they purposely divide suddenly into scattered bands and attack, rushing about in disorder here and there, dealing terrific slaughter . . . They fight from a distance with missiles having sharp bone, instead of their usual points, joined to the shafts with wonderful skill; then they gallop over the intervening spaces and fight hand to hand with swords.

Zosimus, a sixth-century writer drawing on the account of the fourth-century historian Eunapius, is equally vivid: ‘[The Huns] were totally incapable and ignorant of conducting a battle on foot, but by wheeling, charging, retreating in good time and shooting from their horses, they wrought immense slaughter.’16 These Roman commentators leave no room for doubt. The Huns were cavalry, and above all horse archers, who were able to engage at a safe distance until their opponents lost formation and cohesion. At this point, the Huns would move in for the kill with either bow or sabre. The essential ingredients in all this were skilled archery and horsemanship, the capacity to work together in small groups, and ferocious courage. As many have commented, and as was demonstrated repeatedly in antiquity and in the Middle Ages, the Eurasian pastoralist’s life was a hard one, and the kinds of skills, not to mention the magnificent horses, a nomad required for everyday existence set him up equally well for battle.

But this was true of all Eurasian nomads, and doesn’t really explain why the Huns were particularly successful. As well as the Germanic Goths, they were also able to defeat fellow nomads, such as the Iranian-speaking Alans. What gave them the edge? Both were renowned horsemen, but they fought in different ways. Whereas the Huns, as relatively lightly equipped horse archers, set a high value on manoeuvrability, the Alans, like the Sarmatians in general, specialized in heavy cavalry – cataphracts, as the Romans called them. Both rider and horse were protected; the rider’s main weapon was the lance, supplemented with a long cavalry sabre, and the lancers operated in a compact mass. This narrows the question down further. For the Scythians, whom the Sarmatians replaced as the dominant power north of the Black Sea in the early imperial period, had been horse archers, just like the Huns, and employed very similar tactics – but at that point, lance had prevailed over bow. Why, three centuries later, did the balance tilt in favour of the bow?

The answer doesn’t lie in the basic construction of the bow the Huns used. Both Huns and Scythians used the so-called ‘wonder weapon of the steppe’. When we in the West think of bows, we usually have in mind ‘self’ bows, made of a single stave of wood and assuming a simple concave shape when put under tension. Steppe bows were completely different. To start with, they were composite. Separate sections of wood provided a frame for the other constituent parts: sinew on the outside that would stretch, and bone plates on the inside that would be compressed, when the bow was tensed. Unstrung, these bows also curved in the reverse direction: hence the weapon’s other name, the recurve bow. Wood, sinew and bone were glued together with the most powerful adhesive that could be concocted from fishbone and animal hide, and when fully seasoned the bow’s hitting power was tremendous. Remains of such bows (usually the bone plates) have been found in graves from the Lake Balkhash region dating back to the third millennium BC. So by the fourth AD, it was hardly a new weapon.

The key to Hunnic success seems to lie in one particular detail whose significance has not been fully recognized. Both the Huns and the Scythians used the composite bow, then, but whereas Scythian bows measured about 80 centimetres in length, the few Hunnic bows found in graves are much larger, measuring between 130 and 160 centimetres. The point here, of course, is that size generates power. However, the maximum size of bow that a cavalryman can comfortably use is only about 100 centimetres. The bow was held out, upright, directly in front of the rider, so that a longer bow would bang into the horse’s neck or get caught up in the reins. But – and here is the answer to our question – Hunnic bows were asymmetric. The half below the handle was shorter than the half above, and it is this that allowed the longer bow to be used from horseback. It involved a trade-off, of course. The longer bow was clumsier and its asymmetry called for adjustments in aim on the part of the archer. But the Huns’ asymmetric 130-centimetre bow generated considerably more hitting power than the Scythians’ symmetrical 80-centimetre counterpart: unlike the Scythians’, it could penetrate Sarmatian armour while keeping the archer at a safe distance and not impeding his horsemanship.

Some idea of what it was like to use the recurve, or reflex, bow can be derived from trials with composite ‘Turkish’ bows in the early modern and modern periods. These were generally about 110 centimetres in length, but symmetric, since they were made for infantry rather than cavalry. They were also the product, of course, of a further millennium of development, outperforming larger Chinese and Asian bows of the same basic design. Their performance certainly startled Europeans, used to ‘self’ bows. In 1753 the best shot before the modern era, Hassan Aga, launched an arrow a grand total of 584 yards and 1 foot (roughly 534 metres). He was a renowned champion, but distances of well over 400 metres were commonplace. This bow’s power, too, is awesome. From just over 100 metres’ distance, a Turkish bow will drive an arrow over 5 centimetres through a piece of wood 1.25 centimetres thick. Because of its asymmetry and the fact that infantry archers can plant their feet firmly, unlike their mounted counterparts, we need to knock something off these performance figures when thinking about what they tell us about the fourth century. The Huns didn’t have stirrups, but used heavy wooden saddles which allowed the rider to grip with the leg muscles and thus create a firm firing platform. Nonetheless, Hunnic horse archers would probably have been effective against unarmoured opponents such as the Goths from distances of 150 to 200 metres, and against protected Alans from 75 to 100 metres. These distances were more than enough to give the Huns a huge tactical advantage, which, as Roman sources report, they exploited to the full.17

The bow wasn’t the Huns’ only weapon. Having destroyed the cohesion of an enemy’s formation from a distance, their cavalry would then close in to engage with their swords, and they often used lassos, too, to disable individual opponents. There is also some evidence that high-status Huns wore coats of mail. But the reflex bow was their pièce de résistance. Carefully adapted, by the mid-fourth century it could face down the challenge of the Sarmatian cataphracts. The Huns, as you might expect, were well aware of their bows’ uniqueness, as slightly later sources, dating to the fifth century, attest. The historian Olympiodorus of Thebes tells us that in about 410 Hunnic kings prided themselves on their archery skills,18 and there is no reason to suppose that this was not already the case in 375. On the night that the greatest Hun of all – Attila – died, the Roman emperor Marcian dreamt that ‘a divine figure stood by him and showed him the bow of Attila broken that night’.19 And the archaeological record confirms, likewise, that the Hunnic bow was a symbol of supreme authority. In four burial sites the remains of bows entirely or partly encased in engraved gold sheet have been found. One was entirely symbolic: only 80 centimetres long, it was covered with so much gold that it could not have been flexed. The other three were full length, and it’s possible that here we are looking at real weapons with gold casings.20 Thus embellished, the source of the Huns’ military dominance became a potent image of political power. It also allowed them to dominate the western edge of the Great Eurasian Steppe.

Ammianus Marcellinus was right. It was the Huns who were behind the military revolution that had brought the Tervingi and Greuthungi to the Danube sometime in the late summer or early autumn of 376. At this point, the rise of Hunnic power ceased to be a problem for the peoples of the northern shores of the Black Sea exclusively. It now presented the eastern emperor Valens with a huge dilemma. Tens of thousands of displaced Goths had suddenly arrived on his borders and were requesting asylum.

Asylum Seekers

WITH A RARE UNANIMITY, the vast majority of our sources report that this sudden surge of would-be Gothic immigrants wasn’t seen as a problem at all. On the contrary, Valens happily admitted them because he saw in this flood of displaced humanity a great opportunity. To quote Ammianus again – but most other sources tell a similar story:

The affair caused more joy than fear and educated flatterers immoderately praised the good fortune of the prince, which unexpectedly brought him so many young recruits from the ends of the earth, that by the union of his own and foreign forces he would have an invincible army. In addition, instead of the levy of soldiers, which was contributed annually by each province, there would accrue to the treasury a vast amount of gold.

Thus soldiers and gold both at the same time – usually you got one or the other. No wonder Valens was pleased.

Most of the sources also give a broadly similar account of what went wrong after the Goths crossed the river (probably at or around the fortress of Durostorum (map 6). The blame for what happened next is placed mostly on the dishonesty of the Roman officials on the spot. For once the immigrants started to run short of supplies, these officials exploited their increasing desperation to run a highly profitable black market, taking slaves from them in return for food. Unsurprisingly, this generated huge resentment, which the Roman military, especially one Lupicinus, commander of the field forces in Thrace (comes Thraciae), only exacerbated. Having first profited from the black market, then having made the Goths move on to a second camp outside his regional headquarters at Marcianople (map 6), he made a botched attack on their leadership, at a banquet supposedly given in their honour. This pushed the Goths from resentment to revolt.21 So the story goes, and so it has often been repeated by historians. Blaming Valens for his stupidity in agreeing to admit the Goths, the local Roman military for their greed, and the Goths – just a bit – for resorting to violence makes for a perfectly coherent account. Considered in all its details, however, it is not the whole truth.

Take, to begin with, normal Roman policy towards asylum seekers. Immigrants, willing or otherwise, in 376 were a far from new phenomenon for the Roman Empire. Throughout its history, it had taken in outsiders: a constant stream of individuals looking to make their fortune (not least, as we have seen, in the Roman army), supplemented by occasional large-scale migrations. There was even a technical term for the latter: receptio. An inscription from the first century AD records that Nero’s governor transported 100,000 people ‘from across [north of] the Danube’ (transdanuviani) into Thrace. As recently as AD 300, the tetrarchic emperors had resettled tens of thousands of Dacian Carpi inside the Empire, dispersing them in communities the length of the Danube, from Hungary to the Black Sea. There had been a number of similar influxes in between, and while there was no single blueprint for how immigrants were to be treated, clear patterns emerge. If relations between the Empire and the would-be asylum seekers were good, and the immigration happening by mutual consent, then some of the young adult males would be drafted into the Roman army, sometimes forming a single new unit, and the rest distributed fairly widely across the Empire as free peasant cultivators who would henceforth pay taxes. This was the kind of arrangement agreed between the emperor Constantius II and some Sarmatian Limigantes, for instance, in 359.22 If relations between the Empire and migrants were not so good, and, in particular, if they’d been captured during military operations, treatment was much harsher. Some might still be drafted into the army, though often with greater safeguards imposed. An imperial edict dealing with a force of Sciri captured by the Romans in 409, for instance, records that twenty-five years – that is, a generation – should pass before any of them could be recruited. The rest, again, became peasant cultivators, but on less favourable terms. Many of the Sciri of 409 were sold into slavery, and the rest distributed as unfree peasants (coloni), with the stipulation that they had to be moved to points outside the Balkans, where they had been captured. All immigrants became soldiers or peasants, then, but there were more and less pleasant ways of effecting it.23

There is, however, another common denominator to all documented cases of licensed immigration into the Empire. Emperors never admitted immigrants on trust. They always made sure that they were militarily in control of proceedings, either through having defeated the would-be immigrants first, or by having sufficient force on hand to deal with any trouble. Constantius and the Limigantes provide a case in point. In 359, something went badly wrong. True to form, Ammianus puts it down to bad faith on the part of the Sarmatians, but the causes may have been more complex. Be that as it may, all hell broke loose at a crucial moment:

When the emperor was seen on the high tribunal and was already preparing to deliver a most mild address, intending to speak to [the Sarmatians] as future obedient subjects, one of their number struck with savage madness, hurling his shoe at the tribunal, shouted ‘Marha, marha’ (which is their warcry), and the rude crowd following him suddenly raised a barbarian banner and with savage howls rushed upon the emperor himself.

What happened next is very revealing:

Although the attack was so sudden that they were only partly armed, with a loud battlecry [the Roman forces] plunged into the bands of the savages . . . They butchered everything in their way, trampling under foot without mercy the living, as well as those dying or dead . . . The rebels were completely overthrown, some being slain, others fleeing in terror in all directions, and a part of them who hoped to save their lives by vain entreaties, were cut down by repeated strokes.

The Limigantes’ acceptance on to Roman soil had been carefully negotiated before Constantius showed himself, so all should have been well. But when it wasn’t, there were plenty of Roman troops to hand and it was the Limigantes who were wiped out.24

This highlights a key element in the generally accepted account of what happened in 376 that just doesn’t ring true. Valens, we are told, was filled with joy at the Goths’ arrival on the Danube. But in 376 the Roman army was demonstrably not in charge of the situation, and when things started to go wrong after the crossing, order could not be restored. Lupicinus, whatever his personal culpability for the Goths’ revolt, simply didn’t have enough troops on hand. After the banquet, he immediately rushed his available forces into battle against the rebellious Goths and was soundly defeated.25 In the absence of total military superiority, which was central to normal Roman receptiones, it is just not credible that Valens was anything like as happy about the arrival of the Goths on the Danube as the sources claim.

The shortage of Roman troops in the Balkans had a simple enough cause. In the summer of 376, Valens was deeply embroiled on his eastern front, and had been for some time. As we saw in Chapter 3, he had ended his war against Athanaric in 369 with a compromise, because he was needed in the east to deal with Persian ambitions in Armenia and Iberia. After 371, taking advantage of Persia’s difficulties in its own far eastern territories, Valens had made some important gains, managing to put Roman nominees in control of these Caucasian territories. By 375, though, Shapur, Persian King of Kings, was back. Determined to hold firm, Valens sent three aggressive embassies in the summer of 376, which told him to back off or expect a fight. Such diplomatic posturing required appropriate military preparations, so that not only had Valens made haste to Antioch, the regional headquarters for Persian campaigns, but the vast majority of his mobile striking forces was in the east as well. When the Goths arrived on the Danube, therefore, Valens was already fully committed to an aggressive policy in the east, and it was bound to take him at least a year to extract his forces diplomatically, or even just to turn them around logistically.26

For a while Valens probably still hoped that the Danube crisis could be managed in such a way as to allow him to pursue his Caucasian ambitions, perhaps even with the addition of some extra Gothic military manpower, as the sources report. Given how far the Danubian situation departed from normal Roman expectations of control, however, we might also expect him to have been treading very carefully, wary of potential problems. And the available evidence shows that he was. As we noted earlier, one thing is clear: of the two Gothic groups who arrived at the Danube, only the Tervingi were admitted.27 The Greuthungi were refused permission to enter the Empire, and such troops and naval craft as were available in the Balkans were placed opposite them to keep them north of the river.28 Valens did not, then, rush to accept every Goth he could find so as to build up his army and fill the treasury’s coffers at one and the same time.

Let’s also have a closer look at his relations with the Tervingi. No source describes the terms agreed with this group in any detail, and, thanks to the rebellion, they were never fully implemented. The new relationship was certainly presented to the Roman public as a Gothic surrender – deditio – but that in itself tells us little; both Constantine’s and Valens’ earlier treaties with the Tervingi were described as such when they involved quite different relationships (see pp. 72-6). Everything suggests that the agreement of 376 incorporated some unusual features, highly favourable to the Goths. To start with, they exercised an unusual degree of control over their place of settlement. In normal circumstances, the emperor decided where to place immigrants, tending to spread them out. In 376, it was agreed that the Tervingi should be settled only in Thrace, and this was their choice. The details of how the settlement was to be organized are unclear; in particular, we are left in the dark on the crucial issue of whether they were to be settled in clusters large enough to preserve their political and cultural identity. This would again have been highly unusual, but, given that they were able to choose their own settlement area, may well have been part of the agreement. Otherwise, we know only that hostages were taken, and an immediate draft of young men for the regular Roman army; and that the agreement envisaged the Goths possibly serving en masse as auxiliaries on particular campaigns, much as they had between 332 and 369. There were also some confidence-building measures. In particular, the Tervingi leadership declared itself willing to convert to Christianity.

The fact that the agreement was sold to its Roman audience as a surrender must not confuse the issue. In both its military and its diplomatic details it departed from Roman norms. The Tervingi extracted much better terms in 376 than those usually granted even to immigrants being treated as friends. Lacking sufficient military clout on the Danube, Valens was forced to depart from tried and trusted Roman methods. We might expect him to have been wary about admitting even the Tervingi, therefore, and there are, in fact, strong hints that he was.29

As we’ve seen, the main cause of the Tervingi’s revolt was food shortages and black-marketeering beside the Danube. The Goths, it seems, spent autumn and part of winter 376/7 beside the river, and only moved on to Marcianople sometime in late winter or early spring. Even when the revolt got under way, they still had difficulty in finding food, because ‘all the necessities of life had been taken to the strong cities, none of which the enemy even then attempted to besiege because of their complete ignorance of these and other operations of the kind’. This relates to the summer of 377, but long before that year’s crops had ripened. The Romans, it would seem, had deliberately moved the harvest of 376 to fortified strongpoints which the Goths lacked the military technology to take. Feeding the hungry Tervingi was anyway a formidable task for the Roman state, given its bureaucratic limitations. It had to plan carefully enough for major military campaigns when its own troops needed feeding. The Goths, of course, had no means of growing their own food at this point, since the agreement hadn’t yet got as far as land allocations. Once their stocks had been consumed, securing all other food supplies gave Valens a lever of control over them.

The emperor was also quick to negotiate military assistance from his western colleague, the emperor Gratian, son of his brother Valentinian I. Probably in 377 our old friend Themistius, orator, philosopher, senator of Constantinople and a close confidant of Valens, visited Rome. There he delivered his thirteenth oration. This speech, derivative and uninspired – perhaps delivered on the tenth anniversary of the emperor’s accession, which fell in 377 – celebrated Gratian as the Platonic ideal of a ruler. Much more interesting than the speech is the fact that Themistius was present in the west at such an important moment. And, as he makes clear, his journey from Syria had been made at breakneck speed:

. . . my course was almost equal to the course of the sun, from the Tigris to Ocean [the Atlantic; i.e. the west]; it was an urgent journey, a flight over the surface of the earth, just as you [Socrates] say Eros once hurried, with sleepless days following the nights. I lived my life on the road and under the open skies, sleeping on the ground and out of doors, with no bed to lie on and no shoes to put on . . .30

The pace he described here is much faster than you’d think the rather run-of-the-mill contents of the speech would demand, which suggests that his embassy had another, more urgent aim. The presence of some western troops, already available to the east for campaigning in the Balkans in summer 377, gives the clue. Such campaigning would have required prior negotiation sometime during winter 376/7, possibly even before the revolt of the Tervingi had broken out. It was this necessity that drove Themistius and his companions so relentlessly across land and sea. The ambassadors were charged with negotiating a joint imperial response to the Gothic problem that had suddenly appeared on Valens’ doorstep.

A note of caution on the eastern emperor’s part too is suggested by the most mysterious of all the events that were unfolding at this time beside the Danube. As food shortages worsened, and the Goths’ hostility grew, Lupicinus moved the Tervingi on to his regional headquarters at Marcianople, as we noted. But to supervise the process, he was obliged to use the forces that had previously been keeping out the Greuthungi. The Tervingi did eventually move, but the redeployment of the Roman forces allowed the Greuthungi to cross the river on to imperial territory. Lupicinus, as commander, must have been getting desperate – clearly, the situation was spiralling out of control. Ammianus reports that, to cap it all, the Tervingi moved only slowly towards Marcianople, so as to allow the Greuthungi to catch up with them. (The Greuthungi may have crossed the Danube slightly more to the east than the Tervingi, at Sacidava or Axiopolis (map 6).) When the Tervingi were about 15 kilometres from their destination, Lupicinus invited their leaders to dinner. Ammianus describes the party:

Having invited Alavivus and Fritigern to a dinner party, Lupicinus posted soldiers against the main body of the barbarians and kept them at a distance from the walls of the town . . . Great wrangling arose between the inhabitants and those who were shut out, which finally reached a point where fighting was inevitable. Whereupon the barbarians . . . killed and despoiled a great troop of soldiers. When Lupicinus learned by a secret message that this had happened . . . he put to death all the attendants of the two leaders, who as a guard of honour and to ensure their safety were waiting for them in front of the general’s quarters. When the [Goths] who were besieging the walls heard this news, in their resentment they gradually increased their number to avenge their kings, who, as they thought, had been detained by force . . . And since Fritigern was quickwitted and feared that he might be held with the rest as a hostage, he cried out that they [the Romans] would have to fight with heavy loss of life, unless he himself were allowed to go out with his companions to quiet the people . . . And when this request was granted, they all departed.31

It is difficult to know precisely what happened. On the face of it, the botched attack was the result of misunderstanding and panic, but banquet hijacks were a standard tool of Roman frontier management.

Removing dangerous or potentially dangerous leaders was an excellent means of spreading confusion amongst opponents. Ammianus describes four other occasions over a span of just twenty-four years when Roman commanders made dinner invitations an opportunity for kidnap. One of these four was the unauthorized initiative of a local commander, but the other three resulted from direct imperial orders. In one case, a commander on the Rhine was given a sealed letter, which he was not to open unless he saw the Alamannic leader in question on the Roman side of the river. When this happened, and he did, he was instructed to shunt him off to Spain. Lupicinus, I suspect, was in receipt of similarly contingent orders. Valens, still at Antioch, could not be consulted at every turn – requests for orders from his Danubian commanders would have had a turn-around time of weeks. So Lupicinus’ instructions with regard to the Tervingi must have left considerable room for personal initiative; all the same, I don’t believe that he would have been let loose on the Gothic problem without careful guidance about what to do in a variety of foreseeable scenarios. The arrival of a huge number of unsubdued Goths in Roman territory at a point when the main Roman army was mobilized elsewhere, was much too potentially dangerous not to have been thought through. Lupicinus had been told, I suspect, that if things looked as if they might be getting out of hand, then he should do what he could to disrupt the Goths – and hijacking enemy leaders, as already mentioned, was a standard Roman reflex. But it was Lupicinus’ call. In the event, he went for that worst of all possible worlds: first one thing, then the other, with neither stratagem whole-heartedly pursued. Instead of a continued if uneasy peace or a leaderless opposition, he found himself facing an organized revolt under an established leader.32

Both common sense – would you be pleased to see chaos descend on a second front while you are heavily engaged on a first? – and comparison with other cases of licensed migration into the Roman Empire make it clear that Valens could not have been nearly so pleased to see huge numbers of Goths arrive on the Danube as our sources, however unanimously, report. As we have seen, imperial ideology required all barbarians to be shown to be subservient, and whatever the panicking going on behind the scenes in 376, the emperor’s policy had to be presented to his taxpayers as a freely chosen strategy that would benefit the Empire. Ammianus offers us a strong hint here. His account refers to the input of ‘learned flatterers’ (eruditis adulatoribus) into Valens’ Gothic policy.33This immediately brings to mind Themistius, who did such a good job for Valens on the peace of 369. He was with the emperor in Syria in the summer of 376, before his sudden dash westwards, and I suspect that a speech such as that of 369 was one of the ways whereby he convinced the east Roman court that, contrary to all appearances, letting in a horde of untamed Goths was actually a jolly good idea. The unanimity of our sources, then, reflects the propaganda that the emperor used to justify his policy, not the real reasoning behind it.

The Huns had thrown the Roman Empire and a large number of Goths into a new and unprecedentedly close relationship. The emperor certainly didn’t desire this relationship: not, at least, in the form it took. The Goths too had their doubts and hesitations. Their decision to seek asylum inside the Empire was not taken lightly. When the majority of the Tervingi broke with Athanaric, they had done so at a large gathering where the issues were debated at length.34 You can understand their wavering. Moving into the territory of such a powerful neighbour was no easy decision. Given the efficiency of the crossborder telegraph, they probably knew that Valens was currently overstretched on the Danube because of the war with Persia. The emperor might be willing to grant concessions for the moment, but there could be no guarantee that his attitude might not harden later. It’s hardly surprising, therefore, if the Goths were trying to think ahead: to prepare themselves to deal with the power of the Empire in the longer term as well as now.

Although the Romans treated them quite differently, the Tervingi and Greuthungi remained closely in touch. Hence, as already noted, when the Tervingi were forced by Lupicinus’ troops to move on to Marcianople, they were already aware that the Greuthungi had crossed the river and so slowed their pace.35 The Tervingi were entering the lion’s den and, even if apparently receiving more favourable treatment than the Greuthungi, they had every interest in forming a united front with as many Goths as possible against the Empire’s overwhelming superiority in both manpower and resources. By so doing, of course, they broke at least the spirit of their agreement with Valens. But if the emperor could find ways of rewriting the agreement of 376 for the longer term, then so could the Goths.36

And this, it seems to me, is the real story. Both the Goths and the Romans had been thrown by the Huns into a new and more intense relationship. Neither side trusted the other, and neither was totally committed to the agreement negotiated – when both were under duress – in 376. That this initial agreement failed to hold cannot really have surprised anyone. The way was now clear for a test of military strength, upon whose outcome would hang the nature of a more durable settlement between the immigrant Goths and the Roman state.

The Battle of Hadrianople

HOSTILITIES OPENED ON the morning after Lupicinus’ fatal banquet. The return of Fritigern and the violence of the night before prompted a first round of pillaging in the immediate vicinity of Marcianople. In response, Lupicinus gathered what men he could and advanced to the Gothic camp, about 15 kilometres outside the city. His force was quickly overwhelmed – few, apart from Lupicinus himself, managed to escape. Sometime in late winter or spring of 377, war began in earnest and was to last no fewer than six campaigning seasons before peace was restored on 3 October 382.37 The action of the first two years, up to the battle of Hadrianople, can be followed in considerable detail in the narrative of Ammianus Marcellinus (which is not to say that he tells us everything that we want to know). After the battle, the sources become thinner. What is very clear, however, is that the entire war – all six seasons’ worth – was confined to the Balkan provinces of the Roman Empire. This is a landscape that has been fought over many times in history, and its very particular geography has always dictated the nature of the action.

The northern part of the peninsula is roughly rectangular, broader to the north than the south, and to the west than the east (map 6), its salient physical feature being its mountains. To the east, the Stara Planina (or Haemus Mountains) rise to rounded summits averaging 750 metres; the highest peak reaches 2,376, while the more rugged Rhodopes are a touch higher with many peaks at over 2,000 metres. Further west, running north–south, are the Dinaric Alps. Over time, their limestone has eroded into sharp crags and pockmarked hillsides, often covered with prickly, unpleasant scrub: the characteristic Karst landscape of the western Balkans. Alongside the mountains lie three wide plains: the Danubian Plain to the north, the Thracian in the south-east, and the Macedonian between the Rhodopes and the Dinarics. Another characteristic feature of the peninsula is its many alluvial upland basins, where rainwater and snowmelt erosion have built up layers of fertile soil in pockets between the mountain peaks.

The nature of this landscape has shaped the region’s history. Most obviously, the plains and upland basins define discrete sections of cultivable land, where there are likely to be concentrations of population. Many of the mountain zones are extremely rugged, which, especially combined with the region’s harsh winters, has limited longdistance communications to only two main routes. North–south, the key highway runs through the Morava and Vardar river valleys connecting the Danube via modern Skopje (the Roman Scupi) to the Aegean at Thessalonica. North-west to south-east, a second important route starts again at the Morava valley, but turns left at Nis? (the Roman Naissus) to work its way through fertile upland basins past the Bulgarian capital Sofia (the Roman Serdica), then over the Succi Pass to connect with the rich upland plain of the Sredna Gora and on to the Thracian Plain. In the Roman period, this was a military trunk road. Landscape also dictates communications more locally. The Rhodopes are extremely difficult to cross from north-east to south-west, for instance, and movement north and south through the Haemus mountains is channelled through just five major passes: the Iskar valley in the west, the Trojan and Shipka Passes in the centre, and the Kotel and Riski further east.

When the Goths crossed the Danube in AD 376, they entered a Roman world that had imposed itself on this landscape for over 300 years in the north, and nearer 500 in the south, where by 146 BC Macedonia had been conquered and turned into a Roman province. In large measure, the Romans worked with the landscape, rather than against it, but there was one main exception. Aside from the two natural axes of long-distance communication, they forced two additional east–west routes through the Balkans. In the south, and constructed as early as 130 BC, the famous Via Egnatia followed the Aegean coastline from Constantinople to Thessalonica – an easy enough route – but then struck determinedly through the peaks and troughs of the Dinarics to reach the Adriatic at Durres (the Roman Dyrrhachium). Further north, at the end of the first century AD, Roman military engineers carved a road through sheer solid rock at the Iron Gates, where the River Danube cuts through the southern extension of the Carpathian Mountains, to connect the Lower and Middle Danube regions. The Balkans was the junction between east and west, and the Empire didn’t skimp on its highways. Even as late as 376, the Balkans’ prime function, viewed from a central imperial perspective, was to provide a bridge between the two halves of Empire; and many resources were devoted to maintaining the roads, and the towns and way-stations along them. These both protected travellers and provided the logistic support that made possible the high-speed connections recorded in the papers of Theophanes (see pp. 104-7).


The imperatives of Empire also dictated that central funds be spent in two other areas of the Balkans. The Danube Plain north of the Haemus Mountains had been an imperial frontier for three centuries by the time the Huns were creating mayhem north of the Black Sea. Early on, major legionary bases had been established at Oescus and Novae. By the fourth century, the regional headquarters at Marcianople, whose walls enclosed an area of 70 hectares, oversaw the operation of the frontier zone, and a series of larger and smaller fortresses guarded the river line and studded the countryside behind it. Many of the larger civilian settlements were also walled by this date, and had subsidiary military functions. Further south, political rather than military imperatives dictated expenditure. In the south-east of the peninsula, the emperor Constantine refounded the ancient Greek polis, or city-state, of Byzantium as Constantinople, which, by the third quarter of the fourth century had become in every respect a new imperial capital. Endowed with mighty walls and beautiful public buildings, the city had also seen massive investment on infrastructure: harbour facilities and granaries that could deal with grain fleets from Egypt, and aqueducts that drained the hills over 100 kilometres away to service the burgeoning population of a naturally rather arid site. It was a huge centre of economic demand, and, in addition to all the imperial funds spent on it, had many inhabitants with money to burn. The rich needed both houses inside the city and cooler retreats in the country, as well as services of all kinds. In the fourth century, the south-eastern Balkans were booming as never before, and Constantinopolitan cash spilled over into the nearby communities of the Thracian Plain.

The Balkans were also host to other Roman communities, whose Romanness was the product of a more organic, long-term development. Some Roman cities sat on ancient foundations. Many of the communities of the Adriatic coast had a long pre-Roman past, and this was even more true of Macedonia and the Black Sea littoral, where cities like Thessalonica, Philippopolis, Anchialus and Odessus had classical Greek roots. These areas boasted both proper Roman cities complete with the standard repertoire of public buildings, and a flourishing countryside, cheerfully exploited to good effect by a landowning class living in luxurious villas. ‘Proper’ Roman life could also be found in other parts of the peninsula. In the fourth century, the Danubian Plain was still dotted with Roman towns and villas. In part, these communities can be viewed as a spin-off from Roman defence spending. Many of the town councils of the region were populated with the descendants of legionary veterans, and many villa estates had their origins in the land grants the state customarily made to retired soldiers. Many fortunes were made servicing the consumer demand triggered by soldiers’ pay. But Roman life in the region had generated its own momentum, and its monuments are too substantial to be explained solely by state spending. The same was true of the central corridor from Philippopolis through the Sredna Gora and Serdica into the Morava valley. Here again, state spending had certainly kick-started things, but the Pax Romana had allowed an authentic Roman life to develop, and in most of the upland basins as well. The twin obstacles of mountain and climate that had resulted in far fewer cities and a correspondingly lower percentage of intensely worked land than in many other areas of the Empire, had not prevented the Balkans from developing into a properly Roman world.38

This was the panorama that faced the Goths at the outbreak of war. Everything suggests that the Greuthungi joined in the hostilities immediately.39 Established at this point in the vicinity of Marcianople, they found themselves in the middle of the belt of Roman military installations that guarded the Danube line. Some layers showing damage, datable to the war years, have been found in the remains of smaller forts, but both written and archaeological evidence confirm that Ammianus was right to emphasize that the Gothic leader Fritigern ‘kept peace with walls’.40 It would have been suicide for the Goths to assault these Roman frontier forts, many of which had been reequipped at the start of the fourth century with huge U-shaped bastions designed to carry the brutally effective Roman wall artillery. The garrisons were pretty numerous: twenty-three units in the province of Scythia and twenty-seven in Lower Moesia, with particular concentrations at Noviodunum, Axiopolis, Troesmis, Transmarisca, Durostorum and Novae (map 6). These garrison troops, however, were primarily trained to patrol and deal with small-scale raids, not to provide mobile forces for large-scale field operations, and Lupicinus had anyway drawn off much of their manpower to create his scratch force. In defeating Lupicinus, therefore, the Goths had already neutralized the only mobile Roman force in the region, and the remaining garrisons faced certain destruction if they ventured out piecemeal. These installations posed no immediate threat to the Goths and could be safely ignored.41

Besides, the Goths had more immediate concerns. They had, of course, plenty of scores to settle. As we noted earlier, a winter in the open on the Danubian Plain, where even average daytime temperatures do not climb above zero in January and February, combined with the Romans’ black-marketeering, had infuriated them. There was also the pressing need to secure food supplies. The Goths may well have brought with them at least some of the harvest of 376, and the Romans had been supplying them with a certain amount of food in the meantime, but there was no possibility of planting crops for the current year. After plundering easy targets in the immediate vicinity of Marcianople, therefore, Gothic eyes turned to the great highways running from the Danube towards the metropolitan splendour and economic boom that was the south-east Roman Balkans.

Goths next appear in the vicinity of Hadrianople, already south of the Haemus Mountains, and some two hundred kilometres south of Marcianople. The total defeat of Lupicinus’ force there had robbed the Romans of any chance, at this point, of holding the Haemus barrier against them. A much smaller force of Goths was stationed at Hadrianople. Led by Sueridas and Colias, it had long been part of the Roman army. When news of the revolt further north reached the city, trouble broke out between the citizens and these Goths, and they threw in their lot with Fritigern. It was at this moment, Ammianus records, that Fritigern ‘advised them to attack and devastate the rich and fruitful parts of the country, which were still without protectors and could be pillaged without any danger’. The outcome, from the Roman point of view, was frightful:

[The Goths] advancing cautiously spread over every quarter of Thrace, while their prisoners or those who surrendered to them pointed out the rich villages, especially those in which it was said that abundant supplies of food were to be found . . . With such guides, nothing that was not inaccessible and out of the way remained untouched. For without distinction of age or sex, all places were ablaze with slaughter and great fires; babies were torn from the very breasts of their mothers and slain, matrons and widows whose husbands had been killed before their eyes were carried off, boys of tender or adult age were dragged away over the dead bodies of their parents. Many older men, lamenting that they had lived long enough after losing their possessions and their beautiful women, were led into exile with their arms pinioned behind their backs, weeping over the glowing ashes of their ancestral homes.42

The Goths were hungry and had many resentments to burn off; the people of the Thracian Plain suddenly found themselves in the front line, and paid the price for everything that had happened during that winter on the Danube. Note, too, the willingness of some of the Roman population to assist the Goths in their plundering. Some perhaps helped them out of fear, but there was many an oppressed peasant with his own scores to settle. The Pax Romana did not benefit all Romans equally.

The Roman response to these disasters came in the form of a first consignment of troops from the east. Valens sent one of his chief advisers, the general Victor, to sue for peace with Persia on whatever terms he could get; and in the meantime he detached some troops from Armenia under the generals Trajanus and Profuturus, who reached the Balkans in the summer of 377. Their impact was substantial. The Goths quickly withdrew north of the Haemus Mountains. At this point, too, the first fruits of Valens’ hasty diplomacy materialized. A smallish force from the western Empire under the command of Richomeres hastened over the Succi Pass to join Trajanus and Profuturus. Reinforced, the Romans advanced north of the Haemus range as far as the Gothic wagon laager, which, Ammianus tells us, was situated at a place called Ad Salices, ‘town by the willows’ (map 6).43 The Romans decided to risk battle; and the Goths were up for a fight, once the last of their foraging parties returned. Only Ammianus describes the encounter, and his account is far from graphic. About half of it is devoted to a rhetorical description of the dead and dying, and he tells us nothing of the numbers or dispositions of the two sides. In overall terms, however, the battle was clearly a bloody draw. At one point, the Roman left wing gave way, but reserves rescued the situation and the fighting ended at nightfall. The Romans had suffered grievous losses, but so too had the Goths, and afterwards they stayed inside their wagon circle for an entire week. Summer was at this point giving way to autumn, so we are probably in September 377.44

The Romans made excellent use of the respite. The battle had cost them dear, but for the moment they had retaken the initiative, for the first time since Lupicinus’ defeat. Heavily outnumbered as they were, the available forces had no prospect of defeating the Goths; so instead, quick to exploit one of the features of the Balkan landscape, they fortified the passes through the Haemus Mountains. Marcianople itself commanded the eastern end of the range, so presumably a substantial garrison was left there. The rest of the troops were dispersed to block the five main routes south. The plan was simple, as Ammianus explains: ‘They doubtless hoped that the dangerous mass of enemies, crowded together between the Hister [Danube] and the waste places, and finding no way out, would perish from lack of food.’ It was also well laid. Some of the passes through the Haemus Mountains are quite broad, but they are all high. Exactly 1,500 years later, in the Russo- Turkish war of 1877, the Russians sent a flying column south from the Danube to seize the Shipka Pass, which leads through the central Haemus range to Hadrianople and the main road to Constantinople/ Istanbul. They successfully captured it, but weren’t reinforced, and for five days (21–25 August) 4,400 Russians had to face the assault of 30–40,000 Turks under Suleiman Pasha. At the end of the battle there were three and a half thousand Russian casualties, but they had held the pass, and over 10,000 dead Turks littered the hillside. For two months after the encounter at ‘the town by the willows’, the Romans were as successful as the Russians would be:

Since everything that could serve as food throughout the lands of Scythia and Moesia [the two Roman provinces north of the Haemus] had been used up, the barbarians, driven alike by ferocity and hunger, strove with all their might to break out . . . After many attempts, they were overwhelmed by the vigour of our men, who strongly opposed them amid the rugged heights.

The Romans were desperately trying to buy time, hoping that the arrival of winter would bring the campaigning to an end and give Valens and Gratian time to bring reinforcements to the Balkans by springtime.

Their hopes, however, were misplaced. ‘Just as autumn was turning to winter’,45 intelligence reports came in that the Goths had found new allies. A force of Huns and Alans had been recruited to the Gothic cause by promises of booty. When he heard this, the Roman commander decided that the passes could no longer be held. As soon as one pass was forced, the soldiers holding the others would be cut off and stand little chance against the numerically superior Goths. He lost no time in pulling back his troops. For the most part the retreat worked, but one Roman detachment was caught in the open at a major crossroads near Dibaltum south of the Haemus Mountains, and seems to have been exterminated.46 The Goths, now with Hunnic and Alan allies (who need not have been very numerous to swing the delicate balance of power back in the Goths’ favour), were free again to rampage south of the Haemus Mountains. They did so, to telling effect, in dispersed groups throughout the winter of 377/8, ‘filling [as Ammianus tells us] the whole country as far as [the province of] Rhodope and the strait which separates the two great seas [the Hellespont] with a most foul confusion of robbery, murder, bloodshed, fires, and shameful violation of the bodies of freemen.’

This time the raiding spread wider and lasted longer, but there was plenty to occupy the Goths on the rich Thracian Plain, and the damage extended no further west than the eastern slopes of the Rhodope Mountains. Ammianus treats us to another lengthy account of Roman misery rather than giving any precise details, but other sources tell us that the Goths came close to the walls of Constantinople, where they were finally driven off by Arab auxiliary forces in Roman service. The Arabs’ habit of drinking the blood from the slit throats of their dead opponents discouraged the Goths from pursuing the argument further, but there were not enough Roman troops or allies available to take broader countermeasures. Until reinforcements started to arrive from the east, the Goths had plenty of time for some productive looting. Some of the damage shows up in the archaeological record. All the main excavated late Roman villas of the region, north and south of the Haemus Mountains, were abandoned at this point, most of them showing an extensive destruction layer.47

Sometime early in 378, the bulk of Valens’ field forces began to arrive from the east. The army gathered slowly in the vicinity of Constantinople, as its units arrived from Mesopotamia and the Caucasus. It is probably wrong to imagine this happening very early in the year, since a Roman field army, like its counterparts everywhere until recent times, could not begin operations until the grass was growing sufficiently to feed the animals pulling its baggage and heavy equipment. Valens himself didn’t arrive in Constantinople until 30 May, and this was probably more or less the point at which large-scale operations first became feasible. He received from the capital’s population a far from warm welcome, and there was some rioting. Constantinople had been a hotbed of resistance to Valens during an attempted usurpation at the start of his reign, and there were also religious issues afoot. In addition, of course, many of the richer citizens would have recently suffered financial and other losses in the Gothic raiding. Once assembled after the long march from the east, his army rested in preparation for battle. Valens was an emperor with a great deal to prove.

THE ROMAN PLANS for 378 were well laid. By granting major concessions in the Caucasus, Valens had bought peace from the Persians and could shift most of his mobile forces back to the Balkans. Negotiations had continued with Gratian: the western emperor had promised to come in person to Thrace, bringing with him the western field army. The best troops from both halves of the Empire were thus gathering in order to put the Goths in their place. No source defines the precise aim of the joint campaign, but it is pretty easy to guess. The emperors were assembling enough troops to win a resounding victory; then it would be business as usual. Imperial invincibility would be seen to be re-established, and of those Goths who remained on Roman territory some would die in amphitheatres across the Empire, some would be drafted into the army, and the majority widely distributed as unfree labour.

But in the fourth century, as in any other, ‘no plan survives first contact with the enemy’. In this case, the enemy took an unexpected form. As Gratian was collecting his expeditionary army in the west, it became obvious, from the other side of the frontier, that gaps were appearing in the Roman defensive line-up on the Upper Rhine and Upper Danube. The news was confirmed by a Roman soldier of Germanic origin returning home to his people the Lentienses, a branch of the Alamanni, who inhabited the Alpine foothills on the frontiers of Roman Raetia (modern Switzerland). In February 378, when Gratian had already sent many troops east to Pannonia in the Middle Danube region for the upcoming campaign, the Lentienses crossed the upper reaches of the frozen Rhine. This initial assault was repulsed, but Gratian received intelligence that it was merely an opening gambit, and that much more substantial attacks, by many thousands of Alamanni, were being planned. The emperor and his advisers decided that the Goths would have to wait. Part of the expeditionary army was pulled back west from Pannonia and more troops drafted in from Gaul, enough to allow Gratian to launch a strong pre-emptive assault. He was determined to secure his rear before turning east, and pressed home the assault to the point of a lengthy siege against the chief group of suspects, who were holed up on a mountain top. Slowly but surely the campaign ground on until the Lentienses surrendered and the ex-Roman soldier was punished.48

All of this made perfect sense from Gratian’s perspective, but placed Valens in an impossible situation. He had arrived in Constantinople on 30 May and left the city twelve days later, advancing to an imperial villa at Melanthias, 50 kilometres further into Thrace, where his troops were concentrating. Pay and supplies were distributed and attempts made to bolster the troops’ morale in preparation for the campaign. But Gratian failed to appear. And while Valens waited, the Goths were far from idle. Their foraging parties continued to operate and their main forces were distributed between Nicopolis and Beroea, thus controlling both ends of the strategic Shipka Pass. The Goths, it would seem, were keeping their options open: they might move on north, or south through the Haemus Mountains. At this point, Valens’ generals got wind of a detached Gothic raiding party in the vicinity of Hadrianople, and rushed a column forward to ambush it. The night attack was a success, and prompted Gothic countermeasures. Fritigern called in all his raiding parties and moved the entire main body, wagons and all, south of the Haemus Mountains to Cabyle – then further south still, on to the Thracian Plain proper, to avoid the danger of further ambushes. The endgame was fast approaching. The mass of Goths were now north of Hadrianople on the main road from Cabyle. Valens was south of Hadrianople, with his army collected and rested. Gratian, however, was still nowhere to be seen, and summer was dragging on.

Valens joined his army outside Constantinople on 12 June. But July came and went, and still no Gratian. The eastern army had been sitting around for the best part of two months, and nothing had happened except for the ambush of one Gothic raiding party. The troops were becoming restive and morale was ebbing away. Then, instead of Gratian’s army, a letter arrived minutely detailing the victories the western emperor had won over the Alamanni. He was, he promised, still coming; but it was already August, late on in the season, and Gratian’s successes touched a nerve. Valens’ patience was fast approaching breaking-point. Then came news of the Goths’ advance south towards Hadrianople. Intelligence reports put the Gothic numbers at only 10,000 fighting men, many fewer than Valens was expecting. This figure was based, I believe, on the misconception that only Fritigern’s Tervingi, and not the Tervingi and Greuthungi combined, were nearing Hadrianople at this point. Jealous of Gratian’s success, Valens was deeply tempted. Was this an opportunity to win a morale- and esteem-boosting victory over a significant number of the enemy? Opinion among his generals was divided. Some urged boldness; others counselled waiting for Gratian. Provisionally, the hawks won. Trumpets sounded the advance, and Valens’ army moved in battle order up to Hadrianople, then constructed a defended marching camp (temporary earth ramparts) outside the city.

Now more letters arrived from Gratian. He was on the move, and his advance guard had kept open the vital Succi Pass between the Haemus and Rhodope Mountains, so that he could move straight down the great military road to Hadrianople. Some of Valens’ generals continued to argue for delay, therefore, but as Ammianus reports, ‘the fatal insistence of the emperor prevailed, supported by the flattering opinion of some of his courtiers, who urged him to make all haste so that Gratian might not have a share in the victory which, as they represented, was already all but won.’

On the night of 8/9 August, with the two sides now in close proximity, Fritigern sent a Christian priest to Valens as a peace envoy, but the emperor would have none of it. At dawn, the Roman army hastened on to the north of Hadrianople, leaving its baggage and a suitable guard in the marching camp; the imperial treasury and other more valuable items were left inside the city walls. All morning the Romans marched north, until, at about two in the afternoon, the Gothic wagon circle (‘as if turned by a lathe’, as Ammianus puts it) came into view. As the Roman army deployed, two further sets of Gothic peace envoys arrived. Valens dithered. He was in the process of arranging an exchange of hostages when two regiments on the Roman right wing, without having been ordered to do so, surged forward to attack. After months of waiting, battle had finally begun in earnest.49

Accounts of ancient battles are never all you would like them to be. Ancient audiences wanted to hear about great deeds of derring-do, not military science. In the case of Hadrianople, in fact, Ammianus presents us with one of his best efforts at battle depiction. The Goths had drawn up their wagons in a circle to reinforce their battle line; the Romans deployed with a mixture of cavalry and infantry on each wing, and the bulk of the heavy infantry in the centre. Although the left wing had not fully formed when the battle began, it seemed, at first, to be making the most progress. It pushed the oncoming Goths right back to their wagon circle and was on the verge of carrying even that by storm, when disaster struck. As the Roman left wing surged forward, Gothic cavalry under Alatheus and Saphrax, combined with some Alans (presumably the ones with whom an alliance had been made the previous autumn), ‘dashed out as a thunderbolt does near high mountains and threw into confusion all those whom they could find in the way of their sudden onslaught and quickly slew them’. With both Tervingi and Greuthungi confronting him on the battlefield, Valens was now exposed to a far larger enemy force than he had imagined. He had given battle on mistaken intelligence, and the Goths had achieved complete tactical surprise.

Ammianus is not absolutely clear about what happened next, but the Gothic cavalry seems to have smashed into the Roman left wing. It was certainly from the left wing that the disaster unfolded. First, its cavalry support was dispersed and then its main force was overwhelmed – caught, perhaps, between the defenders of the wagon circle and the onrushing Gothic cavalry. The destruction of the left wing in turn exposed the Roman centre to a massive flanking attack. Since the Romans were in their customary close order – in the fourth century they often still operated the wall-of-shields formation – the effect was calamitous:

The foot-soldiers thus stood unprotected, and their companies were so crowded together that hardly anyone could pull out his sword or draw back his arm . . . arrows, whirling death from every side, always found their mark with fatal effect since they could not be seen beforehand nor guarded against . . . and in the press of ranks no room for retreat could be gained anywhere, and the increased crowding left no opportunity for retreat.

Indeed, the heavy Roman infantry regiments of the centre were so closely pressed together that they had no hope of manoeuvring to bring the weight of their weaponry to bear. Their normal tactical advantages in arms, armour and training now counted for nothing.

The troops were also reaching exhaustion point. Valens had pushed them into battle, without rest or food, after an eight-hour march in the August sun; on the Thracian Plain, the average midday temperature at this time of year approaches 30 degrees Celsius. The Goths had turned the temperature up even further by taking advantage of a favourable wind to light huge fires, which were now pouring smoke and heat down on their opponents. After fierce fighting, the main Roman battle line eventually broke and fled. The result, as always in such circumstances, was a massacre. Army and emperor perished together. What exactly happened to Valens, nobody knew for sure. His body was never found. Some said that, wounded, he was taken to a farmhouse which the Goths surrounded and burned to the ground when arrows were fired at them from an upper window, and that one of his attendants escaped to tell the story. Ammianus doesn’t seem to have believed this account, although it is widely reported. Perhaps the emperor was stranded and simply cut down in anonymous fashion somewhere on the battlefield.

Valens’ gamble had failed. The emperor himself was dead, and the Goths, against all expectations, had won a stunning victory, destroying in the process the best army of the eastern Roman Empire. How many Roman troops died that day is hotly disputed. Ammianus tells us that thirty-five officers of tribune rank (approximately equal to regimental commander) died, along with two-thirds of the troops. From a complete listing of the eastern army dating from about 395, about twenty years after the event, we also know that sixteen elite regiments suffered such severe losses that they were never reconstituted. But none of this gives us a total figure, since we don’t know the size of the original army and a number of the dead tribunes will have been staff officers rather than unit commanders. Some historians think that Valens brought with him upwards of 30,000 men – 20,000 dead at Hadrianople, then. Even given the peace deal with Persia, however, the emperor could not afford to denude the east of all its troops and we have to remember that he was expecting reinforcements from Gratian. My own opinion is that Valens brought more like fifteen thousand men to the Balkans in 378, and was looking for a similar number from Gratian. Between them, these forces would have enjoyed a 1.5:1–2:1 advantage over the Goths, which ought to have been more than enough. But because of the faulty intelligence report, Valens gave battle at Hadrianople, in my view, with perhaps a slight numerical disadvantage instead of, as he thought, a 1.5:1 advantage over just the Tervingi. His force was undone by the Goths’ extra numbers, but above all by the huge tactical surprise they brought off. If I’m right, Roman losses on 9 August will have been more in the region of 10,000 men.50

But in an important sense, the quarrel over numbers is academic. The central point is that Valens’ jealousy of Gratian, and his impatience, had undone the Empire. In Ammianus’ view, the Romans had known no such defeat since the battle of Cannae in 216 BC, when Hannibal had annihilated a whole imperial army. Victory left the Goths masters not only of the battlefield, but of the entire Balkans. Roman military invincibility had been overturned in a single afternoon, and Gratian could only look on helplessly from the other side of the Succi Pass, about 300 kilometres distant, as the triumphant Goths rampaged through the southern Balkans. Against all the odds, and despite their opponents’ advantages in equipment and training, the Goths had triumphed and the path to Constantinople lay open. As Ammianus reports, ‘From [Hadrianople] they hastened in rapid march to Constantinople, greedy for its vast heaps of treasure, marching in square formations for fear of ambuscades, and intending to make many mighty efforts to destroy the famous city.’

Valens was dead, his army destroyed; the eastern Roman Empire was there for the taking.

‘Peace in Our Time’

I’VE NEVER QUITE known whether to believe the vignette with which Ammianus, on almost the last page of his history, takes his leave of the Gothic war. Having shown us the victorious Goths preparing to besiege Constantinople, he then feeds us the following image:

[The Goths’] courage was broken when they beheld the oblong circuit of the walls, the blocks of houses covering a vast space, the beauties of the city beyond their reach, the vast population inhabiting it, and the strait nearby that separates the Black Sea from the Aegean. So they destroyed the stores of military equipment they were preparing . . . and spread everywhere across the northern provinces.51

It is almost too good to be true: a perfect metaphor for the entire war. And you have to remember that, by the time he was writing, in the early 390s, Ammianus knew the outcome of the war even if he chose to end his account in 378. Victory over Valens at Hadrianople was just enough to give the Goths a glimpse of the prize that was Constantinople; but that in turn was enough to convince them that they hadn’t the slightest chance of capturing it.

The Goths faced three overwhelming disadvantages that made it impossible for them to defeat the Roman Empire outright. First, even if, taking the maximum conceivable figure, we reckon that there were 200,000 of them in all, with the capacity to produce an army of 40–50,000 men – although I do think this figure too high – this would still have been rather paltry compared with the grand sum of imperial resources. The Empire’s army totalled, as we’ve seen, 300–600,000, and its population was in excess of 70 million (a minimum figure). In a fight to the death, there could be only one winner, and the cannier Goths – some of whom among the Tervingi had travelled the breadth of Roman Asia Minor to fight in the Persian wars – were perfectly well aware of this. Fritigern’s peace overtures to Valens before Hadrianople show that he, for one, never lost his sense of perspective. He told Valens that, if the imperial army put on a decent enough show of martial intimidation, he would be able to persuade his followers to reel in their military ardour and make a compromise peace.52 The quid pro quo that Fritigern had in mind for himself, interestingly enough, was that Valens should recognize him as king of all the now allied Goths, thus cutting out Alatheus and Saphrax, not to mention all his other would-be rivals among the Tervingi. As it turned out, the imperial army failed to deliver its part of the deal, perishing virtually to a man. But, a bit like Pearl Harbor, when there is a fundamental mismatch in resources and capacity one shock victory at the beginning of a struggle can’t change its course.

To this fundamental problem we can add two more. First, there is no record of the Goths taking any major fortified imperial centre during the six years of war. Conditions clearly became fraught in the Roman Danubian communities that were cut off from the centre for extended periods; we don’t know, for instance, if and when they were able to plant crops. But no city was ever taken by siege.53 This meant that the Goths were unable to get their hands on stocks of weapons and supplies, or to set themselves up in a defended stronghold of their own. The second problem arrived on the back of the first. The Gothic force at large south of the Danube between 377 and 382 wasn’t just an army, but an entire population group: men, women and children, dragging themselves and their possessions around in a huge wagon train. With no secure lands available to them for food production, and unable to break into fortified storehouses, the Goths were forced to pillage in order to eat, and, because so much food was required, it was extremely difficult for them to stay in the one place. Already in autumn 377, there was nothing left north of the Haemus Mountains, and the pattern of the subsequent war years, in so far as we can reconstruct it, saw them moving from one part of the Balkans to another. Sometimes it was the Roman army that forced them on, but this restlessness was largely attributable to their lack of secure food supplies.

Victory at Hadrianople allowed the Goths to range as they wished in Thrace during the rest of 378. The next year, however, even though the Empire had no more than light skirmishing forces available in the eastern Balkans, they shifted the centre of their operations further west into Illyricum, the combined Gothic force advancing north-west over the Succi Pass into Dacia and Upper Moesia (map 6). In 380, Tervingi and Greuthungi then divided, perhaps because of the difficulty of supplying their combined numbers. Alatheus and Saphrax moved further north into Pannonia, where they were defeated, it seems, by the forces of the western emperor Gratian. The Tervingi under Fritigern moved south and east along the Morava–Vardar trunk road to Thessalonica and the provinces of Macedonia and Thessaly. They seem to have learned from previous experience, contenting themselves with exacting only a moderate tribute from the cities – repeatedly taking protection money – rather than trashing the place and moving on. Whether this would have continued we cannot know, because in 381 forces of the western Empire drove the Goths back into Thrace, perhaps this time along the Via Egnatia rather than through the heart of the Balkans. It was in Thrace again, finally, in 382 that peace was made.54

The Roman Empire, however, could not in the end, after six years of war, claim total victory, although the formal ceremony that inaugurated the peace treaty on 3 October 382 certainly took the form of a Gothic surrender. Themistius was again an eyewitness, and he leaves us in no doubt:

We have seen their leaders and chiefs, not making a show of surrendering a tattered standard, but giving up the weapons and swords with which up to that day they had held power, and clinging to the king’s [the emperor Theodosius’] knees more tightly than Thetis, according to Homer, clung to the knees of Zeus when she besought him on her son’s behalf, until they won a kindly nod and a voice which did not rouse war but was full of kindness, full of peace, full of benevolence and the forgiveness of sins.55

But Themistius’ vocabulary immediately signals that this was not the kind of peace deal that normally followed Roman victories over hostile would-be immigrants. The language of ‘kindness’, ‘benevolence’ and ‘forgiveness’ strikes a new note, and the difference is not merely rhetorical. For the surrender generated no theatrical bloodbaths, no mass selling of Goths into slavery, no large-scale distributions of Gothic captives as unfree farm labourers. When, in 383, an emperor wanted to reassure the population of Rome that the Empire was once more secure, it was Sarmatians who were slaughtered in the Colosseum, not Goths. But the Goths had killed a Roman emperor, destroyed a Roman army, and laid waste with fire and rapine large tracts of the Roman Balkans. In a world where a Roman emperor considered himself well within his rights to throw a fit if ‘barbarian’ ambassadors didn’t grovel with sufficient conviction, the absence of revenge, punishment and example-setting in the peace settlement of 382 is extraordinary.

Once again, we don’t know everything we’d like to know about the terms agreed. They clearly broke new ground in some important ways, but although they were strikingly generous to them, the Goths did not get everything they may have wanted. Before Hadrianople, Gothic peace offers tended towards the possibility of Thrace becoming an independent Gothic kingdom. Fritigern, as we’ve seen, was also angling to have Valens recognize him as new overall leader of all the Gothic immigrants. Neither of these things happened. Neither Fritigern nor Alatheus nor Saphrax survived to participate in the peace deal. They may have died in battle somewhere, but, if not, I have no problem in seeing their overthrow as part of the price the Goths had to pay for peace. The Empire needed tokens of victory to show off to its taxpayers, and the survival – indeed prosperity – of the victors of Hadrianople would have been completely unacceptable. Indeed, for the next decade or so, in a replay within the frontier of the policy commonly pursued towards the Alamanni beyond the Rhine (see Chapter 3), the Romans refused to recognize any overall Gothic leader, no doubt hoping to keep them politically divided. Nor did the Goths as a whole get Thrace as an independent fiefdom. The integrity of the diocese of Thrace as a centrally run unit of the Roman Empire was reasserted with vigour. Frontier fortifications were rebuilt and remanned where necessary; Roman law and tax-gathering resumed. In this sense, Gothic ambitions had been pruned right back.

At the same time, the Goths were given grants of land for themselves, not to farm for others as unfree tenant farmers. We don’t know exactly where these were located. Some were north of the Haemus Mountains in Lower Moesia and Scythia close to the Danube, where the Carpi had lived around the turn of the fourth century, but there may also have been some settlements in Macedonia.56 Much more important, wherever they were, they were clearly in sufficiently large clusters to allow the political and cultural life of the Goths to continue. This is explicitly acknowledged in Roman sources of the late 390s, and shows up implicitly in the narrative of intervening events. One of the things that the Empire got from the peace deal was a military alliance. Not only did it take the normal draft of Gothic recruits for its regular army, but the Goths also agreed to provide much larger forces, serving under their own leaders, for specific campaigns. These times of special service required the emperor to negotiate with leading Goths as a group. On the one occasion for which we have details, we learn that the emperor Theodosius threw a great feast for them.57 If, in 382, the three leaders of the revolt were sacrificed as part of the peace deal, a large number of their peers clearly survived to sustain some sense of Gothic community. Under the peace, despite losing the right to operate independently under the leader of their choice, the Goths continued to enjoy the freedom to negotiate and act as one, with or against the Roman state, as we shall see in the next chapter.58 The break with established ways of dealing with immigrants could not be clearer.

According to Themistius, speaking to the Senate of Constantinople in January 383, this transformation in imperial policy was the result of some divinely inspired decision-making on the part of Valens’ successor Theodosius.59

He was the first who dared entertain the notion that the power of the Romans did not now lie in weapons, nor in breastplates, spears and unnumbered manpower, but that there was need of some other power and provision, which, to those who rule in accordance with the will of God, comes silently from that source, which subdues all nations, turns all savagery to mildness and to which alone arms, bows, cavalry, the intransigence of the Scythians, the boldness of the Alans, the madness of the Massagetai yield.

Taking his inspiration from God – and it was really to Him that he owed his appointment as eastern emperor – Theodosius understood that a better and more total victory could be won through forgiveness than by arms. Consequently, his chief negotiator ‘led the Goths [to the emperor] docile and amenable, all but twisting their hands behind their backs, so that it was a matter of doubt whether he had beaten the men in war or won their friendship’. And the overall outcome, for Romans and Goths, was better for both:

If the Goths have not been utterly wiped out, no complaint should be raised . . . Was it then better to fill Thrace with corpses or with farmers? To make it full of tombs or living men? . . . I hear from those who have returned from there that they are now turning the metal of their swords and breastplates into hoes and pruning hooks, and that while paying distant respect to Ares [god of war], they offer prayers to Demeter [goddess of corn and fruitfulness] and Dionysus [god of wine].

The Goths, Themistius told the Senate, have given up fighting for farming, and everyone has gained. Theodosius, Themistius’ new employer, had come up with a brilliant solution – forgiveness for the Goths and a compromise peace that would subdue them more thoroughly than war ever could, while considerably benefiting the Empire. Once again, it’s important to remember the tyranny of imperial ideology and the fact that Themistius was a remarkably adept propagandist (over a thirty-year period, he managed to create a niche for himself with no fewer than four imperial employers). As usual, he was being economical with the truth – before coming up with his peace deal, Theodosius had had a pretty good shot at winning the Gothic war by more conventional means.

The death of Valens had left a power vacuum which lasted until Gratian appointed Theodosius as his counterpart in the east in January 379. The new emperor had clearly been appointed to avenge Hadrianople. He came from a distinguished military family – his father was a five-star general under the emperor Valentinian I – and he had a good military record of his own. Immediately he was given temporary control of part of the prefecture of Illyricum – the dioceses of Dacia and Macedonia – which belonged to the western Empire, in order to exercise a unified control over the entire area vulnerable to the rampaging Goths. He spent his first year in office rebuilding the eastern field army: calling up veterans, recruiting new units, and drafting in more troops from Egypt and other parts of the east. Themistius’ first speech for the new emperor, in spring 379, confirms the thrust of all of this activity: the emperor’s initial self-presentation was as ‘the man to win the Gothic war’ –

It is because of . . . you [Theodosius] that we have taken a stand . . . and believe that you shall now check the impetus of success for the Scythians [the Goths] and quench the conflagration that devours all things . . . Fighting spirit returns to the cavalry and returns to the infantry. Already you make even farmers a terror to the barbarian . . . If you, though not yet in the field against the guilty ones [the Goths], have checked their wilfulness merely by pitching camp nearby and lying in blockade, what do we suppose those damned villains will suffer, when they see you readying your spear and brandishing your shield, the lightning flash from your helm gleaming close at hand?60

Unfortunately, things didn’t work out as planned. Theodosius’ new model army fell apart when it tried to take on the Goths head to head in Macedonia and Thessaly in the summer of 380. The circumstances are mysterious – the sources hint at treachery and unreliability. It was not another bloody catastrophe like Hadrianople, but there’s no doubt that Theodosius failed and that the Goths overcame a second Roman army. In the autumn, Theodosius had to hand back control of the war to Gratian’s generals, and it was they who eventually drove the Goths from Thessaly in summer 381, while he ran for cover in Constantinople to secure his political position there in the aftermath of military failure.61

Theodosius may have come up with a new plan, then, but not without trying traditional means first. He turned to diplomatic innovation in 382 only because military incapacity – the defeat of two Roman armies – required it. And this was the only time he resorted to such a deal. If he had won the war, I have not the slightest doubt that the normal terms would have been imposed upon any defeated Goths left inside the Empire. When, four years after 382, another group of Goths tried to force their way across the Danube, they were massacred in large numbers. Some of the survivors were drafted into the army, the rest distributed as unfree tenant farmers – both groups sent far afield, to Asia Minor.62

The Goths might be hounded out of rich areas like Thessaly, ground down by constant battering of their raiding parties, starved into submission. But after the summer of 380 the Romans would not risk another set-piece battle.

Given that it was impossible, as we’ve seen, to admit that a Godappointed emperor had ever been forced into a course of action by barbarians or even by circumstances beyond his control, Themistius came remarkably close, in January 383, to telling the truth, making little attempt to downplay Roman disarray at the time of Theodosius’ appointment:

. . . after the indescribable Iliad of evils on the Ister and the onset of the monstrous flame [of war], when there was not yet a king set over the affairs of the Romans, with Thrace laid waste, with Illyria laid waste, when whole armies had vanished completely like a shadow, when neither impassable mountains, unfordable rivers, nor trackless wastes stood in the way, but when finally nearly the whole of the earth and sea had united beside the barbarians.

Nor did he pretend that Theodosius could easily have chosen to press the war to a fully victorious conclusion:

. . . just suppose that this destruction was an easy matter and that we possessed the means to accomplish it without suffering any consequences, although from past experience this was neither a foregone nor likely conclusion, nevertheless just suppose, as I said, that this solution lay within our power . . .

For the man who had felt constrained to claim, in 364, that the loss of provinces, cities and fortresses to Persia was actually a Roman victory, this is not so far removed from an admission that Theodosius had had no choice but to opt for a compromise peace with the Goths.

‘This Is Not Yet the End’

THE TRADITIONAL INTEGRITY of the Roman state had been breached, but we mustn’t get carried away. We are still a long way from imperial collapse. The war on the Danube had affected only the Empire’s Balkan provinces, a relatively poor and isolated frontier zone, and even here some kind of Romanness survived. The late fourth- and early fifth-century layers of the recently excavated Roman city of Nicopolis ad Istrum are striking for the number of rich houses – 45 per cent of the urban area – that suddenly appeared inside the city walls.63 It looks as though, since their country villas were now too vulnerable, the rich were running their estates from safe inside the city walls. At the end of the war, moreover, both eastern and western emperors remained in secure occupation of their thrones, with their great revenue-producing centres such as Asia Minor, Syria, Egypt and North Africa entirely untouched. And most parts of the Empire hadn’t even seen a Goth.

In his final spin on the peace deal Themistius tried to reassure Roman taxpayers that the Goths would lose even their semi-autonomy in due course. He took, as a case in point, some Celtic-speaking barbarians who had crossed the Hellespont in 278 BC and carved out the territory of Galatia (named after themselves) in Asia Minor, but who over the next centuries became fully assimilated into Graeco- Roman culture.64 Given the huge disparity in resources between themselves and the Roman Empire, it no doubt did seem that the Goths’ present status must eventually be reversed, whether by longterm assimilation, as Themistius archly evokes, or, much more likely, by renewed conflict once the Roman army had been properly rebuilt. As events turned out, Themistius’ confidence was misplaced. The descendants of the Tervingi and Greuthungi were destined not only to survive as Goths, but would eventually carve out on Roman soil the fully independent kingdom that they had originally sought. Writing soon after Hadrianople, Bishop Ambrose of Milan summarized the prevailing crisis with admirable economy: ‘The Huns fell upon the Alans, the Alans upon the Goths and Taifali, the Goths and Taifali upon the Romans, and this is not yet the end.’65 The bishop had in mind only the ongoing war with the Goths, but his words were prescient. The Empire would never get the chance to reopen the Gothic question on its own terms. Hadrianople was indeed not yet the end, and the Empire would have many more challenges to face before the full effects of the Hunnic revolution worked themselves out.

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