Empires and Barbarians

by Peter Heather

  19 Eutropius, Breviarium 8.2.

  20 For first- and second-century references to the Goths, see Tacitus, Germania 43–4; Strabo, Geography 7.1.3 (‘Butones’); Ptolemy Geography 3.5.8. Kulikowski (2007), chapters 3–4; cf. Jordanes, Getica 4.25–8 (on Filimer: see p. 122).

  21 For more detail on the Tervingi, see Chapter 2 above. Jordanes’ anachronisms were first demonstrated in Heather (1991), chapters 1–2 (where I show my own scepticism of Jordanes, pace Kulikowski).

  22 On the first and second century, see Shchukin (1990); and cf. Batty (2007), 353ff. on Bastarnae, Sarmatians, and Dacians of various kinds (with full references, and noting the distorting political agendas that have sometimes been applied to these materials). For an introduction to Ulfila and his Bible, see Heather and Matthews (1991), chapters 5–7.

  23 For the first- and second-century placement of Goths, see note 20 above. Rugi: Tacitus, Germania 44. Vandals: Courtois (1955), chapter 1. (Kulikowski does not discuss the broader range of evidence.)

  24 Carpi: see note 38 below. On the 330s, see Anonymous Valesianus I.6.30.376, chapter 4. On the migration habit, see above, p. 30.

  25 In technical terms, these transformations occurred in period B2–C1a/b. For fuller discussion, see Heather (1996), 43–50, drawing on the materials mentioned in note 9, and now supplemented by Shchukin (2005). Kulikowski (2007), 60ff. dismisses the importance of the archaeological evidence in very general terms without discussing the phenomenon of Wielbark expansion.

  26 For introductions to this material, see Kazanski (1991); Shchukin (2005), with a fuller literature listed at Heather (1996), 47–50.

  27 See Kazanski (1991); Heather (1996), 47–50; Shchukin (2005).

  28 Jordanes, Getica 4.25–8.

  29 Jordanes, Getica 16–17.90–100 records the third-century triumphs of the Amal King Ostrogotha. The king is entirely mythical, however, invented to explain why the Ostrogoths were so called, and his name has been added to known historical events: see Heather (1991), 22–3, 368.

  30 For more detail, see Heather (1991), chapter 1 and 84–9.

  31 Batavi: Tacitus, Histories 4.12, Germania 29. Chatti, Bructeri and Ampsivarii: Tacitus, Annals 58, Germania 33.

  32 The element of fragmentation under Filimer is quoted on p. 122. Berig: Getica 4.25–6, with 17.94–5. Goffart (1988), 84ff. is reasonably concerned to undermine old assumptions that Gothic oral history suffuses the Getica, but is arguably a little too dismissive: see Heather (1991), 5–6, 57–8, 61–2.

  33 See e.g. Borodzej et al. (1989); Kokowski (1995); Shchukin (2005).

  34 See Drinkwater (2007), chapter 2 and 85–9 (with references).

  35 See Ionita (1976).

  36 On Heruli casualties, see George Syncellus, Chronicle, ed. Bonn, I.717. For other figures from the Aegean expedition (2,000 boats and 320,000 men), see Historia Augusta: Claudius 8.1. Cannabaudes’ defeat is said to have cost 5,000 Gothic dead: Historia Augusta: Aurelius 22.2. Much of this material derives from the contemporary account of Dexippus. If the parallel with Viking activity is to be taken to the ultimate, one would suspect that relatively small groups made the initial moves, only for their very success to encourage larger entities to participate in the action. The state of the third-century evidence, however, does not make such a chronological progression certain. For further commentary on scale, see Batty (2007), 390ff.

  37 Langobardi: Dio 72.1.9. Quadi: Dio 72.20.2 (explicitly pandemei, ‘all the people’).

  38 For the protest of the Carpi, see Peter the Patrician fr. 8. For the exodus on to Roman soil, see Aurelius Victor, Caesars 39.43; Consularia Constantinopolitana, s.a. 295. See, more generally, Bichir (1976), chapter 14. A total of six campaigns were fought against the Carpi during the reign of the Emperor Galerius (293–311).

  39 Naristi: see p. 98. Limigantes: see p. 85. On the Greek cities, the classic works of Minns (1913) and Rostovzeff (1922) remain essential. For an introduction to the archaeological evidence that has since become available, see Batty (2007), 284–9 (with references).

  40 Drinkwater (2007), 43–5 rightly rejects the recent tendency to claim that Alamanni did not exist before the 290s, but then attempts to make all the action of the third century, including the whole settlement of the Agri Decumates, into the result of warband activity. This argument fails to convince.

  41 See above Chapter 9.

  42 For female burial costume, see note 26 above. For an introduction to the Gothic Bible, see Heather and Matthews (1991), chapters 5–7. The contrast with the originally Norse Rurikid dynasty, who quickly took Slavic names (see above Chapter 10), is extremely striking. See also Chapter 6 below for discussion of the linguistic evidence from the Anglo-Saxon conquest of lowland Britain.

  43 Quadi: Dio 72.20.2. Hasding Vandals: Dio 72. 12.1.

  44 ‘Commonsense’: Drinkwater (2007), 48. For Burgundian linguistic evidence, see Haubrichs (2003), (forthcoming).

  45 For a qualitative definition of ‘mass’ migration, see pp. 31–2 above. If ‘mass’ sounds too redolent of the invasion hypothesis, then alternative terms might be found (perhaps ‘significant’?), but there is surely virtue in bringing first-millennium usage into line with the norms prevailing in specialist migration studies.

  46 Panegyrici Latini 3 [11].16–18.

  47 See above Chapter 2.

  48 On military inscriptions, see Speidel (1977), 716–18; cf. Batty (2007), 384–7. On shipping, see Zosimus 1.32.2–3.

  49 I would therefore strongly argue that the ‘interaction’ theme that has been so marked a feature of frontier studies in recent years – e.g. Whittaker (1994); Elton (1996) – must be balanced with a proper appreciation of the frontier’s equally real military function.

  50 See pp. 43ff. Oddly, Drinkwater (2007), 48–50, while accepting the evidence for increased competition within the Germanic world, refuses to recognize that this would naturally lead to increased pressure on the Roman frontier, amongst other areas, as groups sought to escape the heightened dangers of their existence. Wells (1999), chapter 9 is similarly – and equally oddly – ‘internalist’ in interpretation, seeking to locate the causes of third-century disturbances within the frontier zone, and particular the Roman side of it.

  51 Ammianus 26.5, 27.1; cf. Drinkwater (2007), chapter 8.

  52 Tacitus, Annals 12.25.

  53 See e.g. Anokhin (1980); Frolova (1983); Raev (1986).

  54 Peter the Patrician fr. 8.

  55 The rhythms of Roman frontier management perhaps aided the process. Thinning out the frontier zone periodically, as the Romans did, to reduce overcrowding and the potential for violence (see p. 85), can only have made it easier for more peripheral groups eventually to build up a sufficient manpower advantage to overthrow established Roman clients.

  56 Cf. Chapter 2 above, p. 101. I would in any case strongly argue that freeman and retinue society were unlikely to be completely separate.

  57 Jordanes, Getica 55.282 (‘ascitis certis ex satellitibus patris et ex populo amatores sibi clientesque consocians’).

  58 For references, see note 10 above, with Kmiecinski (1968) on Odry. Descriptive terms like ‘semi-nomad’ are sometimes used, but to my mind misleadingly. What we’re talking about here are mixed farming populations, who kept many animals, perhaps measured their wealth in cattle, but also engaged in extensive arable agriculture, despite lacking the techniques to maintain the fertility of individual fields over the long term.

  4. MIGRATION AND FRONTIER COLLAPSE

  1 Before these tumultuous events of the late fourth century, the western border of Alanic territory lay on the River Don. This just about made them outer clients on Rome’s Lower Danube frontier, especially since the Empire retained strong contacts with the southern Crimea. But they can only be classed as complete outsiders when it comes to the convulsion of 405–8, which affected the Middle Danubian frontier region.

  2 The same basic vision of the crisis can be found, amongst other sources, in Ammianus 31; Eunapius frr. (and Zosimu
s 4.20.3 ff., which is largely but not completely dependent on Eunapius); Socrates, Historia Ecclesiastica 4.34; Sozomen, Historia Ecclesiastica 6.37. The total figure of 200,000 is provided by Eunapius fr. 42, whose account is generally vague and rhetorical, and therefore unconvincing by itself: see Paschoud (1971–89), vol. 2, 376 n. 143. The figure has, however, been accepted by some: see e.g. Lenski (2002), 354–5 (with references). On the 10,000 warriors, see Ammianus 31.12.3; these may have represented only the Tervingi: see Heather (1991), 139. On the wagon trains, see Ammianus 31.7, 31.11.4–5, 31.12.1ff. On social dependants, see e.g. Ammianus 31.4.1.ff.; Zosimus 4.20.6.

  3 Matthews (1989) stresses Ammianus’ literary artistry, where Barnes (1998) stresses his lack of candour. These two most recent studies disagree on many things but both stress that Ammianus is not a straightforward read. For further comment, see Drijvers & Hunt (1999); G. Kelly (2008).

  4 For the ‘more secret’ archive, see Ammianus 14.9.1. For career documents, see Ammianus 28.1.30. For military dispatches, see Sabbah (1978).

  5 On the migration topos, see Kulikowski (2002). On causation, see Halsall (2007), chapter 6.

  6 For examples of the migration topos in action, see pp. 122 and 251 above. Ammianus on warbands: e.g. 14.4; 17.2; 27.2; 28.5. Ammianus on Strasbourg: 16.12.7; 31.8.3.

  7 The recruitment of this extra mercenary support has sometimes been confused with the arrival of the Greuthungi alongside the Tervingi. This is a serious mistake: see Heather (1991), 144–5, and Appendix B.

  8 On the split of the Tervingi, see Ammianus 31.3.8ff.; 31.4.13. The Greuthungi seem also to have fragmented, in that a leader called Farnobius and his followers, found alongside the main body as it crossed the Danube, then suffered an entirely different fate from the rest: see Ammianus 31.4.12; 31.9.3–4.

  9 Only Kulikowski (2002) really dares to suggest that Ammianus might be completely misleading, and even he seems to backtrack substantially in Kulikowski (2007), 123ff., which, while wanting to minimize any unity among the Goths, still accepts that they formed a mixed group of humanity ‘numbered at least in the tens of thousands, and perhaps considerably more’ (p. 130). Among the other anti-migrationists, Halsall (2007) is willing to think in terms of over 10,000 warriors and a total mixed group of 40,000 people; while Goffart (1981), (2006) has never treated the events of 376 in any detail.

  10 See Halsall (2007), 170ff., drawing particularly on the analysis of Socrates, Historia Ecclesiastica 4.33 in Lenski (1995).

  11 I therefore remain entirely happy with the analysis of the ‘Ammianus versus Socrates’ issue I offered, with references, in Heather (1986). Halsall’s desire to avoid a sequence of events that would put predatory migration at the heart of causation seems to provide the principal reason for rejecting the contemporary and more detailed Ammianus in favour of the later and less detailed Socrates, but he offers no good reasons based on the historical evidence, and in my view this line of argument allows preconception to justify unsound methodology.

  12 Ammianus 31.3.8.

  13 Zosimus 4.20.4–5.

  14 Ammianus 31.3.2–8.

  15 On the Caucasus raid, see Maenchen-Helfen (1973), 51–9 (who does think they came from the Danube). For other Goths north of the Danube in 383, see (Arimer) Achelis (1900); and (Odotheus) Zosimus 4.35.1, 4.37–9.

  16 Some Hunnic groups did operate further west before 405–8, but the numbers were very small up to about 400: just the mercenaries who joined the Goths south of the Danube in autumn 377 (see note 7 above) and another Hunnic/Alanic warband found near Raetia in the 380s (Ambrose, Epistolae 25). Uldin’s force from c.400 was clearly a bit larger, but even his command paled compared to the Hunnic forces that arrived in the Middle Danube after 405–8: see above Chapter 5. In general terms, all of this suggests to me that the action of 376 should be viewed rather along the lines of Caesar’s description of the move of the Tenctheri and Usipetes west of the Rhine in the mid-first century BC. In that case, an extended series of smaller-scale raids and attacks, rather than one outright invasion, convinced them that they could no longer live securely east of the Rhine: Caesar, Gallic War 4.1.

  17 On discussion, see Ammianus 31.3.8. On persuasion, see Heather (1991), 176–7,179–80.

  18 On the archaeology and group identity, see above Chapter 1. The particular items within the Cernjachov culture that strike me as a priori promising for distinguishing the immigrant groups are its bone combs, particular fibulae and north European, Germanic longhouse types. Unfortunately, no detailed mapping of these items has yet been made.

  19 For an introduction to Ulfila, see Heather and Matthews (1991), chapter 3. I suspect that the alternative view, of a swift social amalgamation, involves a degree of wishful thinking, largely inspired in reaction to the horrors of the Nazi era, that shrinks from accepting such unequal relationships between relatively bounded groups of human beings. The eastern expansions of the Goths and other Germanic groups in the late Roman period were enthusiastically seized upon by Hitler’s propagandists to justify the poisonous activities of the Third Reich: see Wolfram (1988), chapter 1. But a laudable determination to condemn Nazi atrocities becomes muddled thinking if we try to make the past conform to our wish rather than to the reasonable probability of its evidence.

  20 Different grades of warrior are not specifically mentioned in the Hadrianople campaign, but they do feature in the evidence for Radagaisus’ Gothic force of 405 (Olympiodorus fr 9.) and Theoderic the Amal’s Ostrogoths (see Chapter 5), as well as in later Visigothic laws. Moreover, the Historia Augusta’s vivid account of third-century mass Gothic migratory bands, complete with families and slaves, may well be based on fourth-century events (Historia Augusta, Claudius 6. 6, 8.2; cf. Chapter 3 above), and I strongly suspect it was those of lower status that the hard-pressed Tervingi were selling into Roman slavery in return for food on the banks of the Danube: Ammianus 31.4.11.

  21 On Carpo-Dacai, see Zosimus 4.34.6. On Cernjachov continuity, see Kazanski (1991).

  22 For the Carpi, see Chapter 3 above. On the Sarmatian move, see Anonymous Valesianus 6.31.

  23 For Goths in the fourth century, see Chapter 2. The point about information is also applicable to the minority under Athanaric who moved into Sarmatian territory: this is what the Tervingi as a whole had tried in 332, only to be frustrated by Roman counteraction (see previous note).

  24 Noel Lenski (2002), 182ff., 325f. seeks to locate the reason for Valens’ aggression towards Persia in the Goths’ arrival, and thoughts of the extra recruits he could muster from them. I find the argument unconvincing, and remain confident that the Gothic crisis left Valens with very little room for manoeuvre: see Heather (1991), 128ff.

  25 Kulikowski (2007), 123ff. implies that the Tervingi and Greuthungi came to the Danube and requested asylum on separate occasions, so that Valens had sequential decisions to make, but this is not what Ammianus’ account suggests (31.4.12–13; 31.5.2–3).

  26 Ammianus 31.10; cf. more general frontier studies such as Whittaker (1994).

  27 Ammianus 31.5.3–4. This might possibly be Roman paranoia; I don’t think it is.

  28 For example, the Goths of Sueridas and Colias (Ammianus 31.6.1); perhaps also the Alamannic unit under Hortarius (Ammianus 29.4.7).

  29 I suspect, but am unable to prove, that this would have been particularly true of indigenous groups who merely paid some tribute to the Goths and were otherwise left substantially alone. For a similar range of relationships between the Huns and their different subjects, see Chapter 5.

  30 For full references, see PLRE 2, 934.

  31 For Vandals in Raetia, see Claudian, Gothic War 278–81, 363–5, 400–4, 414–29. On the identity of the Sueves, see most recently Goffart (2006), 82–3, who adopts the most plausible Marcomanni/Quadi approach. The Rhine crossing is generally dated 31 December 406 on the basis of Prosper, Chronicle AP 379; for the argument that the chronicler might have meant 31 December 405, see Kulikowski (2000a), 328–9. Following the counterargument of Birley (20
05), 455–60, however, Kulikowski (2007), 217 n. 37 appears less sure.

  32 For Uldin, see Sozomen, Historia Ecclesiastica 9.25.1–7; Codex Theodosianus 5.6.3. On Burgundians, see Demougeot (1979), 432; 491–3.

  33 On Olympiodorus, see, above all, Matthews (1970), with the further thoughts of Blockley (1981), (1983).

  34 For Vandal losses, see Gregory of Tours 2.29. The 1:5 ratio was customarily employed, e.g. by Schmidt (1933), 286, 293. For Vandal/Alanic numbers, see Procopius, Wars 3.5.18–19; Victor of Vita, History of the Persecution 1.2. For Burgundian numbers, see Orosius 7.32.11. On Radagaisus’ following, see note 30 above.

  35 Jerome, Chronicle 2389 (= 371 AD).

  36 On the ‘distributio’ and its significance, see Jones (1964), Appendix III. Jones’s argument is unaffected by Kulikowski (2000b) since it works from the comparison of two well-dated sections of the Notitia: the eastern field army of c.395, and its western counterpart of c.420. For the thirty ‘numeri’, see Zosimus 5.26.4. For the 12,000 followers of Radagaisus, see Olympiodorus fr. 9.

  37 Victor of Vita, History of the Persecution 1.2. I therefore take a more optimistic view of Victor than does Goffart (1980), Appendix A.

  38 Halsall (2007), 206, for example, has Radagaisus leading ‘a large force’, characterizing the Rhine crossers as a ‘huge force’ (p. 211). It is really only Drinkwater (2007), especially 323–4, who thinks that warbands will adequately explain the action.

  39 Zosimus 5.26.3 has Radagaisus engaged in widespread recruitment prior to attacking Italy (although I remain slightly worried that he is here confusing Radagaisus and the Rhine crossing). Codex Theodosianus 5.6.3 makes clear that Uldin’s following was a mixture of Huns and Sciri, and therefore a new, post-376 alliance.

  40 On Radagaisus’ followers: Zosimus 5.35.5–6. On Alans in Gaul: Paulinus of Pella, Eucharisticon 377–9. On Vandals and Alans in North Africa: Victor of Vita, History of the Persecution 1.2. For Burgundians, see notes 34 and 35 above, together, of course, with the fact that this group were able to preserve their east German dialect throughout these moves: see Chapter 3 above. No one doubts Ammianus’ report that the Goths of 376 also came with women and children in tow (31.3–4), so the basic principle that Germanic and Alanic armed forces might have moved with their dependants seems well enough established. Against this, the assertions of Drinkwater (1998), especially 273, that it is commonsense that only warriors took part in the action are underwhelming. Cf. Drinkwater (2007), 323–4.