Rome's Gothic Wars: From the Third Century to Alaric (Key Conflicts of Classical Antiquity)

by Michael Kulikowski

  As we saw in the last chapter, powerful barbarian polities tended to arise on the Roman frontiers in response to the existence of the empire, a function of the changes which complex and imperial cultures can work on neighbouring cultures that are less socially stratified and less technologically advanced. These were the social forces that created the coalitions of the Franks and Alamanni along the Rhine and the upper Danube in the third century, and we have suggested that the Goths on the lower Danube should be understood in the same way. Before we can go on to address that question in more detail, we need to think about how the Goths or any other barbarian group differed from other ones. More particularly, we need to consider the ways in which both Greek and Roman writers, and we ourselves, go about ‘telling the difference’, as Walter Pohl has put it.[43]

  Barbarian Identity: Graeco-Roman Ethnography

  How do we tell a Goth from a Frank or an Alaman from a Sarmatian? How did the Romans do so? In more abstract terms, how does anyone tell themselves and those with whom they identify from other people with whom they do not? The definition of difference was a pressing concern for Greek and Roman writers, for whom ethnography – the literary description of non-Greeks and non-Romans – was so well known a genre that Virgil could parody it in his fourth Georgic with a poetic ethnography of bees. Modern scholars, in trying to explain the ancient sources with all their myriad names of peoples, strive both to understand the criteria by which ancient writers told their subjects apart, and to establish criteria by which we can do the same thing. From these two questions there follows a third: how did the different peoples we meet in our sources tell themselves apart from their neighbours? This question is much more difficult, because none of the peoples to the north of the Graeco-Roman world left behind written sources from which we might extract such information. Archaeology, if we can use it for this purpose, might provide an answer, but as we shall see, reading ethnic or group identity in the archaeological evidence is very difficult in most circumstances. Let us, however, take the contents of the ancient literary sources first.

  The three words Greek and Roman sources most often use to describe barbarian groups are gens, natio and ethnos (gentes, nationes, and ethne in the plural). The first two words are Latin, the third Greek, and the modern English derivatives of each word are plain to see. Theoretically, in etymological terms, the word gens refers primarily to an extended family, the word natio to a community of such gentes, but in practice the words were interchangeable and their Greek equivalent is ethnos. There is no good English word which we can use to translate any of the three terms. ‘Tribe’ (equivalent to the modern German Stamm) is useful because it implies a sense of community and perhaps a blood relationship (real or fictive), but it also connotes the primitive in a way that only the Latin gentes conveys (and even then, only in the plural and when used of non-Romans). ‘People’ might work, but especially in American English it implies a sense of political purpose which is absent from the Greek and Latin. ‘Nation’ and ‘race’ are too weighted down with modern baggage to be of any use. Modern scholars have settled on the boring, but deliberately neutral, word ‘group’ as the safest way of translating gens, natio, or ethnos in the context of late antique barbarians. This is quite sensible, because it prevents us from implying political or cultural characteristics without meaning to do so. On the other hand, it is very important for us to realize that when the Greek and Roman sources on which we must rely use the words gens, natio, or ethnos, they do indeed mean to imply a coherent, interrelated group of non-Greeks and non-Romans that can be identified as different and which share a sense of belonging together because they do in fact belong together. In other words, the Greeks and Romans did not share our conceptual concerns about the existential nature of barbarian groups – they worried about how to tell such groups apart.

  A Distorting Mirror: Interpretatio romana

  All our evidence for the differentiation of barbarian groups is filtered through ancient Graeco-Roman perceptions of alterity, of the non-Greek or non-Roman. This filter is what scholars call the interpretatio romana, the ‘Roman interpretation’, or perhaps Roman distortion, of the barbarian reality it claims to report. The interpretatio romana poses real difficulties, in part because a cognitive disjunction lies at the heart of Graeco-Roman ethnographic thinking. On the one hand, at a very real level Greeks and Romans believed all barbarians were fundamentally the same. The very word barbaros may be onomatopoeia, coined in order to describe the sound that came out of barbarians’ mouths – a noise like that of animals, rather than language which was the special preserve of the Greeks.[44] Barbarians lacked language and so they were all the same. And yet they were not: ethnography, in fact, existed to tell all those others apart. It set out to abstract from the universal ‘other’ that was the barbarian a set of gentes or ethne which gave shape and order to the world beyond civilization. Although Roman generals on the frontiers had very practical experience of, and sometimes extremely detailed information about, the neighbours whom they had to fight, the ethnographic tradition was not as concerned with such practical matters as it was with abstracting reality into analytical categories. These categories might pattern the experience of reality as much as they were derived from it. For this purpose, Greek and Roman writers had a series of criteria that they could use to analyse identity and difference among their barbarian neighbours or subjects. Chief among these were habits of dress and clothing; traditional weaponry or fighting styles; sex habits and gender roles; religion; and perhaps most importantly language.

  Unfortunately, each of these classificatory criteria posed interpretative problems for ancient ethnographers, because none of them was infallibly diagnostic of ethnic difference. In the case of language, for instance, there were considerably more gentes than there were languages. There were, equally, many fewer fighting styles than there were people who deployed them. And couldn’t a set of stereotyped ‘barbarian’ clothing be used to signify any barbarian in artwork? Public victory monuments have a series of iconographic codes which shout out ‘barbarian’ to the viewer, be they peaked ‘Phrygian’ caps, trousers for Germans and Persians, or hair worn in a ‘Suevic’ topknot. Our ancient writers were fully aware that their classificatory categories were problematical. For that reason, even though they believed that such categories could indeed be used to separate gens from gens, Greek and Roman authors also deployed ethnographic categories as broad existential sets, into which new or newly encountered barbarians could be slotted as necessary, according to whichever classificatory criteria seemed most empirically appropriate at a given time. Categories like German, Celt or Scythian were very broad, their definitions open to discussion. It was, for instance, a very long time before the distinction between Germans and Celts came to be generally accepted among Graeco-Roman authors, and for many years Germans and Celts were regularly taken to be the same. Or again, while in the third century Dexippus could classify Alamanni as Scythians, no fourth-century author could do the same because the Alamanni were clearly fixed as Germani by then.[45] What all this means for us is a constant confrontation with the limitations of our Greek and Roman sources for the barbarians. Their belief in an eternal barbarian type explains the constant identification of the Goths with Herodotus’ Scythians, and also explains why fourth-century authors can freely combine ancient or poetic barbarians like the Cimmerians or Gelonians with the very real Iuthungi and Franci of their own day.

  These conceptual contradictions, or cognitive disjunctures, are pervasive and because they are all we have, they interpose a real barrier between us and the barbarians. We lack nearly any sense of whether or not such Graeco-Roman categories meant anything to the people who were fixed within them. In the case of such meta-categories as German or Scythian the answer, from all we can tell, is no. Nothing in our sources, even filtered through an interpretatio romana as they are, suggests that the later empire’s Germani felt any kinship amongst themselves, or that Goths and Sarmatians, both Scythians in our source
s, were aware of any similarities between themselves. We are on much less certain grounds with more specific ethnonyms – Iuthungi, Iazyges, or Tervingi, for example – which seem to designate groups that shared a sense of kinship and engaged in common actions for that reason. Unlike German or Scythian, these names for smaller groups may have been generated by their users themselves, rather than imposed from outside by Greeks and Romans.

  Even if that is true, however, it tells us very little about how a sense of identity was constituted within or between barbarian groups. How did the Tervingi tell themselves apart from the Greuthungi, who both appear in our fourth-century sources as political divisions of the Goths? In other words, can we get at barbarians’ own criteria of identity and alterity? Language must surely have been important in creating a sense of alterity. Yet despite the deep rooted nineteenth-century conviction that belonging to the same language family produces some sort of shared identity, too many different gentes spoke mutually intelligible languages for a common tongue to contribute much to a sense of identity. Religion may have been more significant – some of our ethnonyms, for instance that of the Suevi, may originally have referred not to political or kinship units, but rather to a variety of groups who shared sacred cult sites. Unfortunately, we have virtually no access to authentic traces of barbarian religion, certainly not enough to chart what function, if any, it had in defining the boundaries of identity and alterity. What of dress? Clothing does have, and has always had, a very important function in expressing identity and alterity. Precisely because it is instantly visible, clothing can serve an emblematic function for those in a position to decode what any particular item of dress, or any combination of such items, means. Greeks and Romans were fully aware of the importance of clothing as a signifier of identity: imperial laws from the fourth century restrict the wearing of ‘barbarian’ costume in certain places, exemplified by a ban on trousers in the city of Rome from A.D. 397.[46] Yet if we try to move from the recognition that clothing could be used to tell the difference to an analysis of how it did so, we run up against one of the most vexed and vexing questions of late antique and early medieval studies: what can archaeological evidence tell us about identity, and ethnic identity more specifically?

  Archaeology, Identity, Ethnicity

  The material remains of the frontier regions are an extremely valuable source for barbarian social history, as will become clear in the next chapter, but they are much less useful as evidence for ancient ethnic divisions. Although that fact has been demonstrated by a great deal of recent work – both practical and theoretical – it flies in the face of more than a century of scholarship. The correlation of particular types of material evidence with particular barbarian groups named in the literary sources has long been, and remains, normal practice, as does the tracing of migrations on the basis of artefacts. The origins of these approaches lie in the early twentieth century and are particularly associated with the archaeologist Gustav Kossinna, though they underpin the work of other great archaeologists of the European barbaricum like Hans Zeiß and Joachim Werner. Kossinna’s Siedlungsarchäologie (‘settlement archaeology’) postulated that materially homogeneous archaeological cultures could be matched up with the ethnic groups attested in our literary sources, and also with the language groups defined by philologists. The shifting extensions of material cultures should therefore be interpreted as the movements of peoples. The rigidity of Kossinna’s approach has long been repudiated, but its legacy is pervasive. One widespread belief ultimately rooted in that legacy is that artefacts themselves carry ethnicity: that one particular form of brooch is Gothic, another Vandalic, and that wherever we find such brooches we can locate Goths and Vandals. This ‘ethnic ascription’ – the attaching of ethnic identity to particular material artefacts – is still ubiquitous in archaeological study of the barbarians, as is the designation of complexes of material evidence with ethnic names drawn from our literary evidence. Ethnic ascription is what allows some scholars to maintain that the Gothic migration recorded in Jordanes is also visible in the archaeological evidence.

  Unfortunately, it has now been definitively shown that artefacts do not carry ethnicity in such a fashion.[48] Whether in the cemeteries from which most of our artefacts come or in the remains of barbarian settlements, material evidence tells us a great deal about vertical social relationships – those between different status levels within a society – but much less about horizontal relationships between ethnic or linguistic groups with separate identities. Thus while it is comparatively easy to characterize vertical distinctions within a single archaeological assemblage – such as bigger houses, better grave goods – defining assemblages by contrast to others is much more difficult. For one thing, it is a wholly artificial process that involves selecting out several characteristics – for instance the positioning of weapons in burials, or particular brooch forms or building techniques – and holding them to be diagnostic, either singly or in combination, of a particular archaeological culture. The selection of defining characteristics can itself be a problem, as there is always a danger of taking as diagnostic characteristics that are actually very widely diffused. But even if we avoid that danger, we are still making another problematical assumption: that the characteristics we have selected as definitive are the same ones that contemporaries would have recognized as defining their sense of identity or alterity. That assumption can never be possible in purely archaeological terms. Although we can be sure that some items of dress were used as emblems of identity and alterity – of belonging and exclusion – we need the human voice of the past to tell us which items communicated that sense of identity and how they did so. As we have seen, the only human voice that exists for our late antique barbarians is that of an interpretatio romana which is as alien to the barbarian perspective as we are ourselves. For that reason, we cannot be confident that our archaeological cultures really do represent something other than our own selection of dead remains – that they do in fact identify some sense of cultural identity that living contemporaries might have recognized. In consequence, we risk investing the material evidence with a historical significance it does not intrinsically possess. That is to say, we risk turning an abstract set of material markers, which we have ourselves selected, into a historically real group of humans to which we then attribute a collective identity or ascribe collective actions. These intrinsic risks are only exacerbated when we draw a connection between an archaeological culture and a historical group named in our sources.

  The Goths and the Sântana-de-Mureş/Černjachov Culture

  It is, of course, sometimes possible to draw a legitimate connection between the material evidence and the barbarians named in our Greek and Roman authorities. If a well-dated material culture is widely present in a region in which our sources locate a named ethnic group over a substantial period of time, then we can say with some certainty that the named ethnic group used that material culture. The correspondence is never absolute, however. All of the archaeological culture zones that we know extend over regions in which the literary sources describe more than one ethnic or political grouping. In other words, a material culture is never identical with a particular ethnic grouping we find in the written sources. The single best illustration of this theoretical position is, as it happens, the Goths themselves. We know that Goths first appear in contemporary literary sources in the early decades of the third century and that, in the company of various other named groups, they posed a threat to the peace of the empire from bases in the region to the north and west of the Black Sea. As we shall see in the next chapter, by the earlier fourth century the Goths had unquestionably become the most powerful group in that region. In that same region – roughly between Volhynia in the north, the Carpathians in the west, the Danube and Black Sea to the south and the Donets to the east – a single archaeological culture is visible from the late third until the early fifth century. This archaeological culture is known as the Sântana-de-Mureş/Černjachov culture and is reasonab
ly well dated on archaeological grounds. That is to say, the region in which the Goths were dominant fell within the Sântana-de-Mureş/Černjachov cultural zone. This means that we can use the socio-historical evidence of that material culture to help describe fourth-century Gothic social structures and economic relations – as we will in the next chapter.