Late antique studies rarely encompass the barbarians who lived beyond the Rhine and Hadrian’s Wall; even the applicability of the term ‘Late Antiquity’ to some regions under discussion has been debated.[1] Much of the ‘late antique problematic’ has concerned specific issues of long-term Roman socio-cultural continuity; it has tended to focus on the Mediterranean, especially the eastern Mediterranean; and it has frequently dwelt upon especially Christian religious topics best confronted via written sources, largely absent beyond the limites before the seventh century. Yet there are powerful reasons to include northern barbaricum within a late antique archaeology. Exclusion perpetuates the idea that Romans and barbarians lived in separate, opposed worlds. This notion must be questioned if we are not to render rather hollow a key founding principle of late antique studies, that the ‘decline and fall’ of the Western Empire should not be seen as marking a chasm within European history. An examination of the peoples north of the imperial frontiers allows us to explore the extent to which the West’s political demise was (or was not) a dramatic shift, and why. Even histories written within the traditional ‘decline and fall’ paradigm have tended to ignore these northerners.[2] Discussion usually ends once the barbarians have migrated, foreclosing any understanding of migration’s causes and mechanisms. Relations between Rome and the regions beyond Hadrian’s Wall and the Rhine could be intimate or sporadic but the Empire remained an important presence. The Western Empire’s fifth-century disintegration therefore affected different areas in varying ways. Nonetheless, in this survey of these ‘barbarians’ between the start of the fourth century and the middle of the seventh, we will see that change was dynamic throughout Late Antiquity, not just in the dramatic fifth century.
Political Geography
The principal late antique political development among the northern peoples was the third-century appearance of larger confederacies along the imperial limites. On the lower Rhine frontier the Franks emerged, their name probably meaning something like ‘Fierce People’. This grouping presumably incorporated the Sicambri, Bructeri, Chattuarii (recorded well into the early middle ages, who may be related to the Chatti) and Chamavi reported by earlier historians and ethnographers. Whether there were two Frankish confederacies, on the Lower and Middle Rhine, seems debatable at this date; the Ripuarian (river-dwelling) Franks are not attested as such until the seventh century, though the Romans certainly referred to the Franks closest to the sea as Salii (‘Salty’ Franks). In the angle left by Roman withdrawal from the Upper Rhine and Danube agri decimates, the Alamanni – ‘All-Men’ – emerged. The relationships between this group and the earlier Suebi and any link between its formation and migrations from northern Germania are again debatable.[3] More recent interpretations see an active Roman role in this confederacy’s formation within the abandoned provinces. The third new group in Germania Magna was that of the Saxons, the first certain reference to whom comes, again, in the third century. The Saxons incorporated the Anglii (Angles), Eucii (Jutes) and Frisii attested by the early imperial writers.[4]
Behind these new confederacies were other Germanic-speaking barbari: the Burgundians, the Vandals and the Lombards, whose initial location is unclear. By the later fourth century the Burgundians bordered the Alamanni to the east; the Lombards spread their power along the Elbe; and, by the early fifth century, there were Vandals close enough to the Rhine for them to be driven across that river by the fall-out from the establishment of Hunnic hegemony in the Carpathian basin. A group of Sueves accompanied the Vandals in that migration. Who they were is problematic. Like the Vandals, Burgundians and Lombards, they bear a name recorded by early Roman geographers, which may be significant. They possibly represent an Alamannic faction (with whom the name Suevi is sometimes associated). Between these groups and the Danube frontier, where Sarmatians, Goths and later Huns were to be found, in the fourth century we catch the last glimpses of the Marcomanni (‘Border Men’), once a great and threatening confederacy, and the Quadi. By the late fifth century, these had been replaced by equally shadowy groups: Skiri, Rugi, Juthungi, Gepids and Heruls.[5] The Heruls are particularly interesting, appearing in and around the Bay of Biscay as well as on the Danube in the fifth century, serving the Byzantine Empire in the sixth and apparently still retaining contact with groups based in Jutland.[6] The distribution of these peoples is noteworthy. The new confederacies are located on the imperial frontiers (counting the North Sea as a frontier) whereas the peoples in the band behind have names attested since early imperial times (though one should not necessarily assume a direct a linkage).
During the fifth century, the Elbe valley fell under the sway of a newly-emergent confederacy, the Thuringians, while Jutland and neighbouring islands may have been taken over by the Danes, first mentioned in sixth-century historical sources.[7] Otherwise, little is known of Scandinavian political geography. The Bavarians, another new people, come into view in the late sixth and seventh centuries, around the upper Danube.[8]
North of Hadrian’s Wall, picti appear in the sources. Whether their name is a genuine ethnonym or a simple Latin description (‘the painted Men’) is unknown. Irish sources call the seventh-century Picts ‘Cruithne’, a Q-Celtic rendition of a word like ‘Pritani’ (Britons), but this is difficult to push back into the fourth century. Back-projection from the ‘early historic’ period bedevils popular conceptions of the Picts. Ammianus claims there were two principal groups of Picti: Verturiones and Dicalydones. Once believed to have been located around Strathearn, the bases of the Verturiones may instead have been further north, towards the Tay, where Anglo-Saxon sources later mention people with a similar name. If, as seems plausible, the Dicalydones may be linked to the earlier Caledonii, place-names suggest that their core lands probably lay north and west of the Forth estuary.[9] We are traditionally accustomed to populating the region between the Hadrianic and Antonine Walls with groups described as ‘British’: Votadini, Selgovae and so on. These peoples, attested by early Roman geographers, also show up in place-names and sometimes reappear in the post-imperial era. The Votadini are the Gododdin, for example, and the Maetae are probably the Miathi mentioned in Adomnán’s Life of Columba.[10] However, Roman sources refer to all people north of Hadrian’s Wall as picti. It seems most plausible to propose that, as in Germania, a new confederacy had emerged, incorporating earlier, smaller units.[11] The core lands of the southern Dicalydones may have lain around the Forth but their confederacy stretched as far south as the Wall. From around 600 we can perceive a Pictish kingdom (perhaps two) north of the Forth (now certainly styling themselves Picti) and an emerging Scottish kingdom in Argyle. The latter straddled the North Channel; its principal royal dynasty, the Dal Riata, also competed for the overkingship of the Ulaid (see below). Between the Forth-Clyde line and Hadrian’s Wall we can now detect some small kingdoms (such as the Gododdin) claiming a British identity. By the mid-seventh century those on the east had been conquered by Anglo-Saxon Bernicia, while a British kingdom based upon Dumbarton, sometimes referred to as Strathclyde, encompassed the western regions.[12]
Irish politics are difficult to establish without written accounts. Early Roman geographers had given detailed accounts of Ireland’s rivers and numerous peoples but it is usually thought that in the late Roman period large, if loose, political units existed, possibly those referred to in later Irish legendary material. By the end of our period, however, Irish politics had taken on the contours they were to retain throughout the remainder of the first millennium. The southern region later known as Munster (Mumu or the Kingdom of Cashel) formed a confederacy, the leadership of which was fought over by various branches of the Eóganacht and others. Eastwards, across the River Barrow, lay the Laigin (later Leinster) and to north of them, in what became Meath (Mide: the middle), the kingdom of the Southern Uí Néill, with kings usually drawn from the Clann Colmáin. West of the southern Uí Néill was the confederate kingdom of Connacht. The region to the north was divided from east to west between the confederacies of the Ulaid (whence Ulster), Airgíalla and the Northern Uí Néill. I have used the term confederacy to represent the loose realms ruled by a Rí Ruirech (a king over kings). The title was usually competed for between branches of royal houses from different kingdoms.[13]
Part 1: Preliminary Analytical Points
a: Ethnicity and ethnic ascription
Ethnic identification remains a hotly debated topic in late antique archaeology. The problem reaches back to the early twentieth century and the development of the notion of the archaeological ‘culture’, defined by pottery and other artefacts, house-types, burial styles and so on repeatedly occurring together in a particular region. Such ‘cultures’ could be categorised by material cultural features (e.g. the prehistoric ‘Beaker Folk’) or by a type-site (such as Hallstadt for the Iron age culture). These are acceptable ways of organising data; the problems arose when cultures were believed to represent actual peoples and their spread or decline to mark conquest or absorption of, or by, other peoples.[14]
By the second quarter of the twentieth century Roman- and Germanic-Iron-Age Cultures (henceforth capitalised for clarity) were normally identified with historically-attested peoples. In the 1920s, for example, a series of volumes by Nils Åberg discussed the archaeology of the Anglo-Saxons, Franks, Ostrogoths and Visigoths.[15] Early medieval ‘barbarian’ peoples were essentially defined by their metalwork, above all by their jewellery. This contrasts with Early-Roman-Iron-Age archaeology in Germania Magna, where pottery types were often used to identify different Kulturgruppen between the Rhine and Scandinavia.[16] Sometimes linking Culture and people is a reasonable working hypothesis, as with the ‘North Sea Coast’ Culture and the Saxon confederacy, or perhaps the Przeworsk Culture and the Vandals.[17] In other areas, the reverse side of the coin is encountered. The ‘problem of the Picts’ discussed in a classic 1955 edited volume was, largely, that they had no Culture.[18] Apart from the enigmatic symbol stones, the Picts seemingly had no archaeological manifestation. Finally, some culture groups were defined essentially by the unsophisticated expedient of labelling a particular area’s archaeology according to the ethnic group known from written sources to have occupied the region during the centuries in question. The best example is the Černjachov-Sintana-de-MurešCulture associated with the pre-migratory Goths. This meets none of the strictly archaeological requirements of a Culture, being (as, most notably, in its ‘mixed-rite’ cemeteries) a mélange of influences and characteristics. The claimed spread of the Culture (and thus, allegedly, of Gothic people) is even argued from absences: weapons are not (always) placed in burials.[19] This argument would be ruled out of court in any discipline that followed the rules of logic. The concept of the Culture far surpassed its analytical utility.
Linking a Culture to ethnicity is problematic. The principal difficulty comes from assuming that ethnicity leaves a passive archaeological manifestation. Ethnicity is a question of belief – in one’s own group or groups’ reality and/or in the difference of others. Objects, naturally, can have no ethnic identity; no ethnic ascription can emerge from a purely archaeological reading of the material.[20] Material culture can be manipulated to convey signs about ethnicity or other group identities. Pottery, for instance, can reveal attitudes to food production and beliefs about cooking, often part and parcel of the ideas making up a belief in a particular group-identity. However, the relationship between material culture and ethnicity – even where it exists – is complex. Moerman’s classic study revealed that members of the South-East Asian Lue could list a series of characteristics which they felt defined their group. When scrutinised, however, it emerged that many of these features were very rarely observed any longer or were used by neighbouring groups too.[21] Pohl has revealed a similar state of affairs in post-imperial Europe.[22] Larrick recorded the African Loikop Samburu belief that different spearheads had associations with neighbouring peoples, and to convey the ethnic characteristics (bravery, cunning, etc.) thought to typify those peoples. They were, however, used within Loikop society to signify different age-grades.[23] This would play havoc with any attempt to map ethnic groups from the archaeological distribution of spearheads. Material culture associated with one group could clearly be used to emphasise other forms of identity among its neighbours. The Late Roman army’s adoption of features such as trousers to express the ferocity supposedly characteristic of barbarians is one example. At a different level of society, slaves’ shaggy hooded cloaks were said to be barbarian (like many slaves themselves).[24] Among the barbarians, Roman imports were employed to denote social and political status and power.[25] It is perfectly possible that other material cultural forms crossed ethnic boundaries in analogous fashion. For a secure ethnic identification of material cultural traits we need reliable written sources; throughout late antiquity, barbaricum produced none of these. The only documentary accounts are external, Graceo-Roman ones. The nigh-insurmountable problems involved in using these sources to discuss barbarian society are well-known, if not always acknowledged.[26]
It has been insufficiently acknowledged that ethnicity is multi-layered.[27] A fourth-century barbarian warrior from North Sea Coastal Germania might have been treated as a Saxon by the Romans and adopted this confederate identity in particular circumstances. Among Saxons, however, he might have stressed a lower-level ethnic identity, perhaps ‘Angle’, ‘Jute’ or ‘Frisian’. These groups are attested before the appearance of the Saxon confederacy, and again after the fifth-century crisis. They were probably subsumed within the confederacy but this does not imply their ‘disappearance’ or ‘reappearance’, still less testify to ‘incomplete assimilation’.[28] It rather suggests such identities’ reordering according to their perceived political value. Similar processes may have occurred among the Pictish confederacy, where several early Roman peoples reappear in the post-imperial era, and among the migratory and later Goths.[29] Crises among larger groups are often marked by the reappearance of other ethnicities, presumably existing previously beneath the confederate identity. Thus, if material culture related to ethnicity it is now impossible to detect which level of identity that was.
The spread of material culture might imply an expansion of ethnic or political identity, if we can plausibly link the former with the latter, but the Lue and the Loikop examples give us pause for thought. A growth in the area covered by a Culture could imply population movement but it need not. Political ascendancy could lead to the adoption of the dominant group’s cultural forms by subordinated leaders, demonstrating the new roots of their power, but with few if any people changing location. Goods also move with trade and exchange; the frequent discovery of Roman artefacts’ in barbaricum – which naturally does not reveal Roman migration and settlement – constantly reminds us not to assume that the movement of material necessarily implies a migration of people. A repeated problem in the historical use of archaeology is the reading of material that spreads in the ‘right’ direction – that suggested by the traditional ‘barbarian migrations’ narrative – as revealing population movement, whereas the diffusion of cultural artefacts in the ‘wrong’ direction – in a direction running counter to the historical narrative – is read as representing trade or exchange.[30]
The archaeological record is not produced by chance but as the result of deliberate choice by historical actors.[31] Thus migration will only be archaeologically detectable where the migrants chose to make themselves visible in the material cultural record. This is by no means commonplace. Most late antique migrations are archaeologically invisible. Between the first and fourth centuries many thousands of people moved from barbaricum into the Roman Empire, leaving no almost material trace at all. The few archaeological signs of late antique population movements relate primarily to the Anglo-Saxon migration and some short-distance drift across the Rhine.
The extent to which even a rigorously-defined archaeological ‘culture’ matches an identifiable ethnic or political group or the movement of the latter is strictly limited. As a default position, straightforward one-to-one correlations should be treated as inherently implausible. This critically-aware stance remains uncommon in the Late Antique archaeology of the northern peoples, where crude ethnic ascriptions persist.
b: Banded barbaricum
The interpretation presented in this chapter draws heavily upon the idea of a roughly ‘banded’ barbaricum.[32] This is not presented as a hard and fast ‘fact’ but as a means of structuring and interpreting the data. As such it provides a framework, which can be refined, corrected or rejected with new research. We might initially imagine barbaricum as three bands. Closest to the limes, contacts with the Empire were dense and frequent. This band has been called a ‘buffer zone’ but this term is predicated upon the notions of a natural hostility between the Empire and its neighbours and of an automatic barbarian desire to invade imperial territory (hence the need for a ‘buffer zone’).[33] I will instead call this the ‘frontier band’. Behind this was territory where contacts with Rome were less frequent, possibly more structured and more politically important. The proportion of higher-value Roman artefacts to other imports is often higher. I will term this the ‘intermediate band’. Furthest away was the ‘outer band’, where contact with Rome was scarcer.
This schema requires caveats. It is immediately complicated by seaborne trade, which could bring Roman goods in some quantity and with steady frequency to areas far from the limes, like Denmark. It is also more analytically descriptive than geographically predictive; there is no algorithmic relationship between geographical distance from the frontier and the ‘band’ within which a settlement is found. Some sites within the fourth-century ‘Saxon homelands’ manifest relationships that would enable their consideration almost as a frontier region. Limited contacts with the Empire mean, conversely, that Ireland never shows ‘frontier band’ archaeological characteristics, even though no territory intervenes between the island and Rome’s British provinces. At a micro level, relationships can change over much shorter distances. Coastal ‘terps’ (see below) show steady relationships with the Empire, whereas those only slightly further inland reveal few or no Roman connections.[34]
The interpretation of Roman imports is also important.[35] There were many means by which objects could enter barbaricum: straightforward commerce, more or less as we would understand it today; diplomatic payment or gift; loot; the property of barbarian soldiers returning home after serving the Empire; and so on. We must also examine potential differences between the means by which objects were brought to barbaricum and the ways in which they were distributed thereafter. A barbarian leader might control a port of trade, to which Roman goods were brought in commercial transactions, but then distribute these to his followers as gifts. Before making confident statements close analyses are required of distribution patterns, the dating of material and the contexts within which objects are found; fine decorated Roman tablewares found in rubbish pits in settlements suggest very different contacts, and social use, from the same ceramics found in exceptional, lavishly-furnished burials.
The blanks and dense spots on distribution maps necessitate other caveats when using archaeological data on a regional scale. These can, in various times and places, show the different levels of archaeological interest in types and periods of settlement. They can also reflect geographical and ecological differences; blanks might represent ancient forest or mountains, or zones of heavy industrialisation and urbanism destroying evidence without record. Intensive agriculture might eradicate archaeological evidence without it being recognised. Conversely, especially in more recent times, ploughing can distort the map producing large numbers of chance finds. Sometimes a bias exists towards recognising some sites rather than others; furnished inhumation cemeteries, lavish bog-deposits or hoards have long been noted because valuable objects would be reported and sold when revealed by ploughing or building. Private and museum collections were often built on such finds[36]. Other sites would not be as observable. That leads to the key variable: the ephemeral nature of many types of site in barbaricum. Most settlements were built of wood, thus requiring a certain level of archaeological competence for their recognition. In some regions, the dead were disposed of in unfurnished and un-urned cremations, making cemeteries all but invisible. Some forms of house-building leave few or no traces, even where excavation techniques are quite advanced. Deep pits for rubbish-disposal and a general ‘cleanliness’ of sites will render them extremely difficult to find, even today, via field-walking or chance. One also needs to take account of the precise context within which finds were discovered – whether as rubbish in settlement sites or deliberately placed in burials for example – and this contextual information is often lacking. Everything we say must be taken as provisional and liable to modification.
c: The interlinking of imperium and barbaricum
This chapter contends that we should stop seeing the Roman Empire and the barbarian territories as opposed and confronting worlds – implicit in the term ‘buffer zone’ for example. By the time of the Tetrarchy the inhabitants of Ireland, northern Britain and Germania Magna[37] had been neighbours of a militarily- and culturally-dominant world power for 250 to 300 years. Military confrontations certainly occurred between the Romans and their neighbours[38]but we should remember that hostile encounters were never the only – and rarely the normal – form of interaction along the limes. They were only one aspect of a complex network of diplomatic relations.[39]
The expression of high status in fourth-century barbaricum is invariably coloured by Roman ideas. For barbarians the emperor was the fons et origo of real power. Historians and archaeologists have been wont to transfer comments on Germanic kingship from Caesar’s and Tacitus’ works (problematic though these texts are even for their own period) to the late Roman era but there is little evidence to suggest that these concepts persisted so long. We do not know how Roman idioms and objects were translated in their use by non-Romans but we cannot now identify any distinctively barbarian ideas of rulership in the late imperial era.[40]
The Alamans, for example, produced brooches that clearly imitated those issued to imperial officials. Possibly manufactured on high-status Höhensiedlungen (see below), they these may have been issued to royal followers as signs of status. In north-western Germania, Roman military metalwork was buried quite frequently in the cremation cemeteries of the ‘Saxon homelands’. Clearly, the families of Saxons who returned home after serving in the Roman army could think of no better way of symbolising the deceased’s prestige than to cremate them in their old uniform or publicly inter their ashes alongside badges of their imperial service. In the same part of the world, around 400, a wooden chair was included in a prestigious burial. This was carved with designs based upon the ‘chip-carved’ ornament that decorated official imperial metalwork. Copies of Roman coins were made in Scandinavia.[41]
The frontier band of barbaricum was saturated with Roman material cultural influence. At the site at the Oespeler Bach near Düsseldorf (D) imports of all sorts were found.[42] Even in areas quite far from the frontier, trans-Rhenan craftsmen had imitated Roman products since the early Roman period. By the fourth century, the ‘frontier zone’ of Germanic-speaking barbaricum might be seen as northern Gaul’s economic hinterland. Exports of Argonne ware are frequent there and bronze vessels evidently made on the Meuse were exported as far as Norway. Study of northern British artefacts shows similar patterns, especially but not only immediately beyond Hadrian’s Wall. In the late Roman period, influences from the Empire seem to have begun to penetrate Irish politics and society and perhaps became a significant engine of social change (see below).[43] This influence did not only take artefactual form. The appearance in Germania of inhumation, the standard fourth-century provincial Roman burial rite, has long been recognised as indicating imperial influence. Inhumation appeared in northern Britain too, probably in a similar cultural imitatio imperii.[44] Although approaches driven by the traditional ‘barbarian migration’ narrative have frequently argued that the Roman-barbarian frontier was deepening into the Empire, when read on their own terms the archaeological data are unambiguous in showing the opposite: the increasing ‘Romanisation’ of barbaricum. This should in no way surprise us.
‘Career migration’ into the Empire was, as mentioned, a standard feature of barbarian life. The late Roman army may or may not have recruited more barbarians than before but, with the separation of civil and military service, the opportunity for non-Romans to rise high in the army was certainly greater.[45] Alamans and, later, Franks did very well in the fourth-century military. One Frank, Bauto married his daughter to Emperor Arcadius and another, Arbogast, briefly (388-94) became the dominant western leader.[46] The Empire continued to intervene in barbarian politics, paying large sums to barbarian groups to keep others in check and periodically launching military operations. Diplomatic payments became extremely important in politics beyond the limes. Setting up and knocking down barbarian leaders remained essential to Roman frontier policy. As had been the case since the late Republic, losing barbarian factions tended to move to imperial territory for security. That north-western barbaricum was a periphery of the Roman Empire and the dynamics involved in this relationship are hugely important in understanding fifth-century history.[47]
Part 2 is here
Notes (See Bibliography)
Part 2 is here
Notes (See Bibliography)
[4]For the early histories of these confederacies, see Capelle (2004), pp.3-50; Drinkwater (2007) pp.43-144 passim; Pohl (2000), pp.29-38, 101-15; Springer (2004), pp.11-46.
[5]For discussions, of varying critical worth, of these people, see: Gepids: Bona (1976), pp.14-19 and passim; Goffart (2006), pp.199-203; Kharalambieva (2010); Heruls: Steinacher (2010); Goffart (2006), pp.205-10; Rugi: Goffart (2006), pp.110-14; Skiri: Goffart (2006), pp.203-5.
[7]For the Thuringians, see Schmidt (1983); (1987); (1997); Fries-Knoblach, Steuer & Hines (ed.) (2014). For the Danes, see Näsman (1999).
[8]See, with various interpretations: Goffart (2006), pp.218-21; Hammer (2011); Hardt (2003); Menghin (1990);
[9]A very good overview of northern Britain and Pictish origins may be found in Fraser (2009), pp. 15-67.
[12]For the Picts, see, again, Fraser, (2009), pp. 15-67. On the Britons and Anglians in the lowlands, Lowe (1999) is valuable.
[13]Mac Niocaill (1972) pp.28-41 for a brief description of the political geography. On early Christian Irish political history, see Ó Cróinín (1995), pp.14-62; (2005).
[14]For an excellent survey of the historiography of this topic with special reference to the archaeology of the peoples that concern us, see Brather (2004).
[19]For an excellent introductory survey of the Černjachov culture in English, see. Heather & Matthews (1991), pp.51-101.
[26]See the very naïve use of Tacitus’ Germaniain Ethelberg (2011); Thurston (2002), e.g. p.46. The use of the late Anglo-Saxon epic Beowulf to study ‘early Germanic’ society in Bazelmans (2014) & Nicolay (2014) is equally naïve. On Graeco-Roman ethnography, Balsdon (1979) is a useful catalogue of descriptions. Hall (1989) and Dench (1995) are essential. More recently, see Bonfante (ed.) (2011); Skinner (2012); Almagor & Skinner (ed.) (2013).
[28]Heather (1998); see also Heather (1996), pp.174-8, 299-303. The reappearance of the Frisians may be archaeologically visible: Nieuwhof (2011).
[30]A classic example is Heather’s discussion of ‘Černjachov Culture’ glassware moving back towards Scandinavia as evidence of trade, whereas objects (such as glass) moving in the opposite direction ‘demonstrate’ Gothic or other ‘Germanic’ migration. Cp. Heather (1996), pp.18-50 (and fig. 2.4) & pp.78-79 (and fig. 3.6).
[31]The foundational texts of ‘post-processual archaeology’, which stressed this point, are Hodder (1986); Shanks & Tilley (1987a); Shanks and Tilley (1987b)
[36]This information bias has been observed from the published catalogues of the Corpus der römischen Fünde in europäischen Barbaricum.
[37]Germania Libera is a modern term, invented in the 1920s. Alföldi, M.R.-, (1997); Neumaier (1997). That a barbarian territory could be ‘free’ would be oxymoronic in Roman thought!
[39]The classic overview of the frontier is Whittaker (1994). See also Lee (1993), Elton (1996a) and, recently, Kaizer & Hekster (ed.), (2010).
[41]Material in Saxon cremations: Böhme (1974). Höhensiedlung: Hoeper & Steuer (1999); imitations of Roman coins: Winge Horsnaes (2013)
[43] Roman bronze vessels in Norway: H.-W. Böhme et al. (1980), pp. 127, 129; Argonne ware export: Elton (1996a), p.79, fig.13; imports in northern Britain: Hunter (2007).
[44]Northern German inhumation: Kleemann (1999); Bemmann (1999). Northern British inhumation: Fraser (2009), pp.36-37