AM – anno mundi, year of the world

BCBook of Constitutions: ed. Von Salis (1892); trans. Drew (1972)

CAH 1. 12 – Cambridge Ancient History, 1st edn, vol. 12: The Imperial Crisis and Recovery AD 193–324, ed. S. A. Cook et al. (Cambridge, 1939)

CAH 2. 7. 2 – Cambridge Ancient History, 2nd edn, vol. 7.2: The Rise of Rome to 220 BC, ed. F. W. Walbank et al. (Cambridge, 1989)

CAH 2. 8 – Cambridge Ancient History, 2nd edn, vol. 8: Rome and the Mediterranean to 133 BC, ed. A. E. Astin et al. (Cambridge, 1989)

CAH 2. 9 – Cambridge Ancient History, 2nd edn, vol. 9: The Last Age of the Roman Republic 146–43 BC, ed. J. A. Crook et al. (Cambridge, 1994)

CAH 2. 10 – Cambridge Ancient History, 2nd edn, vol. 10: The Augustan Empire 43 BCAD 69, ed. A. K. Bowman et al. (Cambridge, 1996)

CAH 2. 11 – Cambridge Ancient History, 2nd edn, vol. 11: The High Empire 70–192 AD, ed. A. K. Bowman et al. (Cambridge, 2000)

CE – Code of Euric: in Zeumer, ed. (1902)

Chron. Gall. 452 – Gallic Chronicle of 452: in Mommsen, ed. (1892)

Chron. Gall. 511 – Gallic Chronicle of 511: in Mommsen, ed. (1892)

CIL – corpus inscriptionum latinarum

CJ – Codex Justinianus: in Kreuger, ed. (1877)

CM 1, 2 – Chronica Minora, vols 1 and 2: ed. Mommsen (1892; 1894)

CMH 1. 1 – Cambridge Medieval History, 1st edn, vol. 1: The Christian Empire, ed. J. B. Bury et al. (Cambridge, 1911)

CTh – Codex Theodosianus (Theodosian Code): ed. Mommsen and Kreuger (1905); trans. Pharr (1952)

fr., frr. – fragment(s)

HE – Ecclesiastical History (Historia Ecclesiastica)

ILT – P. Gauckler (ed.), Rapport sur des inscriptions latines découvertes en Tunisie de 1900 à 1905 (Paris, 1907)

MGH – Monumenta Germaniae Historica

Not. Dig. Occ. – Notitia Dignitatum, western Empire: in Seeck, ed. (1876)

Not. Dig. Or. – Notitia Dignitatum, eastern Empire: in Seeck, ed. (1876)

Nov. Theod. – Novels of the Emperor Theodosius II: in Mommsen and Kreuger, eds (1905); trans. Pharr (1952)

Nov. Val. – Novels of the Emperor Valentinian III: in Mommsen and Kreuger, eds (1905); trans. Pharr (1952)

Or. – Oration, Orationes

P. Columbia 123 – Papyri from the Columbia University collection no. 123: ed. W. L. Westermann and A. A. Schiller, Apokrimata: Decisions of Septimius Severus on Legal Matters (New York, 1954)

P. Ital. – J.-O. Tjader (ed.), Die nichtliterarischen lateinischen Papyri Italiens aus der Zeit 445–700, 3 vols (Lund, 1954–82)

Pan. Lat. – Latin Prose Panegyrics: ed. and trans. Nixon and Rogers (1994)

PLRE 1 – A. H. M. Jones et al. (eds), The Prosopography of the Later Roman Empire, vol. 1, AD 260–395 (Cambridge, 1971)

PLRE 2 – J. R. Martindale (ed.), The Prosopography of the Later Roman Empire, vol. 2, AD 395–527 (Cambridge, 1980)

ref., refs – reference(s)

s.a. – sub anno, under the year



1    Fibulae, as they’re known in the scholarly literature.

2    The volumes generated by the European Science Foundation project can perhaps stand as a metaphor for the general state of scholarship: they encompass a multiplicity of stimulating essays, but no general overview (although, of course, that was not their point).

3    The truth of this is immediately apparent in the chapters devoted to the fourth and fifth centuries in the last volume of the old Cambridge Ancient History and the first volume of the old Cambridge Medieval History, both published in the 1910s, which project the same orthodoxies about inevitable Roman decline and collapse. They remained essentially unchallenged until the 1960s.

4    In saying this, I make not the slightest criticism of projects like ‘The Transformation of the Roman World’. The aim there was to advance participants’ knowledge and understanding by exposing them to the specialized work of others and, in doing so, to enable them to do their own work better. It is that drive which its volumes reflect, and I can gratefully testify to having learned a huge amount during five happy years of participation.


1    Caesar Gallic War, 6. 1.

2    Gallic War 3. 37.

3    St Bernard Pass: Gallic War 3. 1–6 Alesia: Gallic War 7. 75ff. Uxellodunum: Gallic War 8. 33ff. For further reading on the Roman army and its training methods, see CAH 2. 10, Ch. 11; CAH 2. 11, Ch. 9.

4    There were some additions. Areas between the Upper Rhine and Upper Danube – the Taunus/Wetterau salient and the Neckar region – were annexed before the end of the century. A much larger extension came under Trajan. At the start of the second century, he launched a series of campaigns (101–2, 105–6) which eventually added the whole of Transylvanian Dacia to the Empire. This territory was abandoned by the emperor Aurelian (before AD 275). Good general accounts of Rome’s rise can be found in CAH 2. 7. 2, Chs 8–10.

5    Acco: Gallic War 6. 44. Avaricum: Gallic War 7. 27–8.

6    Indutiomarus: Gallic War 5. 58. 4–6. Catuvoleus: Gallic War 6. 31. Ambiorix: Gallic War 8. 25. 1.

7    Gibbon (1897), 160ff. Jones (1964), Ch. 25. Several studies have surveyed the many different explanations for the end of the Empire offered over the years: e.g. Demandt (1984); Kagan (1992).

8    On Rome, see, amongst many possibilities, Krautheimer (1980) with refs. Ostia: Meiggs (1973). Carthage is discussed in more detail in Ch. 6 below. An excellent introduction to the Empire is Cornell and Matthews (1982).

9    The Roman Republic is generally held to have lasted down to the reign of Augustus, the first Roman emperor, although he retained many republican constitutional trappings. Even before his reign, Rome had already acquired overseas territories by conquest, and must therefore be reckoned an imperial power.

10   On Symmachus in particular, see Matthews (1974); for detailed annotations, see the new French translation of his works (Callu (1972–2002)) and the ongoing volumes of the Italian commentary project. Good introductions to the senators of Rome in the late antique period are Matthews (1975); Arnheim (1972); Chastagnol (1960).

11   Letters 1. 52. 1.

12   E.g. ‘A is – always – followed by B’, or ‘A will – in the future – be followed by B’, or ‘A would – if certain conditions apply – be followed by B’.

13   Palladius: Symmachus Letters 1.15. On this education more generally, see the excellent study of Kaster (1988).

14   Letters 1.1.

15   His speeches won less favour after his death than his Letters: the seven we have survive only in one damaged manuscript, which would originally have contained many more.

16   Sometimes, what were originally a grandee’s marginalia eventually became incorporated by mistake into the text proper, giving modern editors the occasionally tricky job of separating original text from subsequent commentary. After Symmachus’ death, theSaturnalia of Macrobius recalled the literary and philosophical ideals of Symmachus and his friends in fictional dialogue form, so as to transmit a potted version of the classical heritage to his son. On the ancient roots of the ongoing scholarly tradition that saved many classical texts, see Matthews (1975), Ch. 1.

17   Homes Dudden (1935), 39. In the view of Boissier (1891), vol. 2, 183, they are ‘the dullest epistles in the Latin language’.

18   Excuses: Symmachus Letters 3. 4. Much of the prevailing etiquette is sorted out in Matthews (1974), (1986); Bruggisser (1993).

19   Caesar: Adcock (1956). The bibliography on Cicero is enormous, but see e.g. Rawson (1975) and, most recently on his oratory, Fantham (2004).

20   Food: Ammianus 27. 3. 8–9. Wine: Ammianus 27. 3. 4.

21   Symmachus Letters 5. 62.

22   Symmachus Letters 6. 33, 6. 42.

23   Symmachus Letters 4. 58–62, 5. 56.

24   Symmachus Letters 6. 43.

25   Ideology: Dvornik (1966). For an introduction to the ceremonial life of the Empire, see Matthews (1989), Chs 11–12; MacCormack (1981). The quotation is from Ammianus 16. 10. 10.

26   Development of Roman law: Robinson (1997); Honoré (1994); Millar (1992), Chs 7–8. Taxation: Millar (1992), Ch. 4; Jones (1964), Ch. 13.

27   The two most important imperial titles in the late period were Augustus and Caesar, both originally deriving from personal names (Julius Caesar and his nephew Augustus). In the fourth century, Augustus became the title adopted by senior emperors, while Caesar was reserved for junior colleagues.

28   Matthews (1989), 235 with refs.

29   Themistius Or. 6. 83 c–d.

30   General development of the imperial office: Millar (1992), esp. Chs 2 and 5; Matthews (1989), Ch. 11.

31   Pan. Lat. 6. 22. 6.

32   Introductions to the late Roman army: Jones (1964); Elton (1996b); Whitby (2002).

33   Growth of bureaucracy: e.g. Matthews (1975), Chs 2–4; Heather (1994b).

34   The Theodorus incident is recounted widely in the sources: Ammianus 29. 1, with full list at PLRE, 1, 898.

35   The deeper history of this development is well explored in CAH, 2. 11, Ch. 4.

36   A contemporary of Symmachus who figures in the letter collection, Petronius Probus, was, for instance, Praetorian Prefect (roughly the equivalent of first minister) for Italy, Africa and the western Balkans for about eight years altogether, in two separate stints.

37   For an introduction to these developments, see Jones (1964), Ch. 18; Dagron (1974); Heather (1994b).

38   General development of Trier: Wightman (1967).

39   Those for the presentation of crown gold were shorter than normal imperial speeches, presumably because there were so many of them – one from each city of the Empire – that the imperial personage might be driven out of his imperial mind if they went on too long.

40   The bibliography on towns in the Roman Empire is enormous, but for introductions to their importance – physical, administrative and political – see Jones (1964), Ch. 19.

41   On Konz and the Moselle villas, see Wightmann (1967), Ch. 4. The literature on the villa as a cultural phenomenon is as profuse as that on towns, but see e.g. Percival (1976).

42   Letters 9. 88; the letter was first identified by Roda (1981).

43   On Ausonius’ Latinity, see Green (1991). Ausonius’ maternal grandfather was a major landowner among the Aedui of central Gaul. His mother’s brother was a successful rhetorician who became court tutor to one of the emperor Constantine’s family in Constantinople. Ausonius tells us less about his paternal ancestry, thereby generating suspicions that it may not have been so respectable, but his father was a doctor who owned property in south-western Gaul and his uncle made a mercantile fortune.

44   For Aristotle, this constituted the only good life, and someone living isolated on his estates was bound to be less rational. Our word ‘idiot’ comes from the Greek (idiotes) for someone shunning participation in this kind of local community.

45   Gonzalez (1986); trans. M. H. Crawford.

46   Villas were always divided into the pars rustica (‘country part’, the working farm) and the pars urbana (‘urban part’, for civilized living). The pars urbana incorporated substantial public rooms for entertaining peers, as well as baths, so that life in the villa was anything but idiotic. There are many good studies on the ideological adjustments involved in becoming Roman. See e.g. Woolf (1998); Keay and Terrenato (2001); D. J. Mattingly (2002).

47   Symmachus Letters 1. 14.

48   The next three quotations are from Mosella ll. 161–7, 335–48, 399–404.

49   Quintilian’s contribution to the Latin tradition is examined in e.g. Leeman (1963).

50   Using the concordance to Symmachus’ works (Lomanto (1983)), I count getting on for twenty mentions of Baiae and its pleasures in his correspondence.

51   Letters 1. 14.

52   According to the expert on the subject, Jones (1964), 528, by c. 370 ‘The third class of the comitiva [countship] was still conferred, but on persons of very humble degree, decurions who had completed their obligations to their cities, and the patrons of the guilds of bakers and butchers at Rome.’

53   On Ausonius’ extraordinary rise to prominence, see Matthews (1975), Ch. 3.


1    Tacitus Annals 1. 61. 1–6.

2    Wells (2003), esp. Chs 2–3 and appendices, is a good introduction to the myth of Arminius and the recent archaeological finds. Its account of the battle, however, is very odd, envisaging a massacre that was all over in an hour while making no comment on the fact that the best source describes a drawn-out four-day struggle fought out over a considerable distance (Cassius Dio 56. 19–22 (no other source contradicts Dio)).

3    Dahn ((1861–1909), (1877).

4    An excellent survey, bringing out the strategic differences between the various frontiers, is Whittaker (1994).

5    As Kossinna put it in the 1926 version of The Origin of the Germani, ‘clearly defined, sharply distinctive, bounded archaeological provinces correspond unquestionably to the territories of particular peoples and tribes’. Kossinna’s ideas were disseminated most influentially in the Anglophone world through the works of V. Gordon Childe (1926), (1927). For an introduction to recent reinterpretations of the meaning of archaeological cultural areas, see Renfrew and Bahn (1991).

6    Ariovistus: Caesar Gallic War 1. 31–53; Veleda: Tacitus Histories 4. 61, 65; 5. 22, 24. Useful introductions to early Germanic society are Todd (1975); (1992); Hachmann (1971).

7    Bructeri: Tacitus Germania 33; Ampsivarii: Annals 13. 56.

8    Annals 1. 68.

9    Elton (1996b), 66–9.

10   Strabo 4. 5. 3.

11   On Chinese expansion, see e.g. Lattimore (1940). For an introduction to the Oppida culture: Cunliffe and Rowley (1976); Cunliffe (1997). On the Jastorf culture, Schutz (1983), Ch. 6. For the Germani adopting La Tène cultural forms, see Hachmann et al. (1962). Much has been written on the dynamics of Roman imperial expansion, but for an introduction see CAH 2. 9 esp. Ch. 8a; CAH 2. 10 esp. Chs 4, 15; Isaac (1992), esp. Ch. 9; Whittaker (1994), Chs 2–3. Modern studies have shown that the process was much more anarchic than old views of planned conquests implied.

12   Trans. Dodgeon and Lieu (1991), 43–6, 50, 57.

13   For an introduction to Near Eastern history in the early Roman imperial period, see Millar (1993).

14   Chronicon Paschale, 510.

15   On the Sasanian revolution, see Christiansen (1944); Howard Johnson (1995); McAdams (1965); Dodgeon and Lieu (1991).

16   Global figures: Agathias History 5. 13; John Lydus On the Months (de Mensibus) 1. 27. General discussions: Jones (1964), 679–86 (inclined to accept an increase up to 600,000 after Diocletian); Hoffmann (1969); Elton (1996a); Whitby (2002). The widely disseminated arguments of MacMullen (1963) on the military ineffectiveness of limitanei have been overturned.

17   On the confiscations of city funds, see Crawford (1975), which remains controversial. Constantius returned one-quarter to the cities of Africa, Valentinian and Valens one-third to all cities; cf. Jones (1964), Ch. 19.

18   On these measures, see Jones (1964), esp. Ch. 13 and 623–30.

19   The one exception, a heavy Roman defeat in 363, was caused by the overambition of the emperor Julian, to which we will return in a moment. For relations between Rome and Persia in the fourth century see e.g. Dodgeon and Lieu (1991); Matthews (1989), Chs 4 and 7.

20   For an introduction to these third-century events, see Jones (1964), Ch. 1; Drinkwater (1987).

21   This quotation is from Ammianus 28. 5. 4, and the next two from 28. 5. 7.

22   It was mentioned by two contemporary spin-doctors (Pan. Lat. 7. 10ff.; 10. 16. 5–6), and made it into one of the main fourth-century annalistic Histories: Eutropius 10. 3. 2.

23   Relatio 47.

24   Themistius Or. 10. 131b–c. On Themistius and his career, see Heather and Moncur (2001), esp. Ch. 1.

25   Roman ideological constructions of the barbarian were directly descended from those of the Greeks. See e.g. Dauge (1981); Ferris (2002).

26   Calo Levi (1952) and McCormick (1986) amongst many others underline the importance of victory.

27   Themistius Or. 5. 66a–c with Matthews (1989), Ch. 7, and Smith (1999) on the campaign.

28   Themistius Or. 6. 73c–75a.

29   The next two quotations are from Themistius Or. 10. 205a–b and 10. 202d–203a.

30   For full analysis of the different circumstances of Constantine’s peace with the Goths in 332 and Valens’ in 369, see Heather (1991), Ch. 3, with full refs. Themistius Or. 8. 116, delivered in March 368, refers to the arrival of the Iberian prince Bacurius at Valens’ headquarters and dates the start of Chosroes’ manoeuvres to the middle of the Gothic war. The story of the other waterborne fourth-century summit is told at Ammianus 30.3.

31   The 369 treaty is reported as a reasonable outcome to the war at Ammianus 27. 5. 9; Zosimus 4. 11. 4. Even after defeating the Alamanni in the 350s, the emperor Julian still made annual gifts part of the peace treaties he negotiated with their various kings: Heather (2001); cf. Klöse (1934) for many earlier examples. A Gothic force of perhaps 3,000 (no small number when expeditionary armies probably numbered about 20–30,000) had been provided on four occasions since Constantine’s victory over the Goths in 332: in 348 (Libanius Or. 59. 89), 360 (Ammianus 20. 8. 1), 363 (Ammianus 23. 2. 7) and 365 (Ammianus 26. 10. 3).

32   Statue: Themistius Or. 15. 191a. Oath: Ammianus 27. 5. 9. On the use of hostages: Braund (1984). Cultural influence did not always have the desired effect. Arminius, three and a half centuries before, had served as a Roman officer in the auxiliaries before plotting Varus’ destruction.

33   On these two MSS, see respectively Tjäder (1972); Gryson (1980).

34   The two main sources for Ulfilas – the letter of Auxentius and fr. 2. 5 of the Church history of Philostorgius – pose a problem over the dates of his ordination and his time in Gothia. See Heather and Matthews (1991), Ch. 5, with refs and translations for the reasoning behind my preferred solution. For a similar community of Roman prisoners among the seventh-century Avars, see Miracles of St Demetrius 285–6.

35   The Gospel text of the Codex Argenteus preserves Ulfilas’ work more or less untouched, whereas others worked after his death on the Epistle text: Friedrichsen (1926), (1939). Before Ulfilas, the Goths used runes for divinatory and other limited purposes, but, as mentioned elsewhere, Gothic did not exist as a written language. Ulfilas had first to devise an alphabet for it, which he did working largely from Greek with a few additions for particular Gothic sounds, and then render the Bible into the language he had created.

36   For an introduction to these theological debates and the manoeuvres, see Hanson (1988); Kopecek (1979).

37   Ammianus 16. 12. 26 and 63 for the numbers; 16. 12 generally for the battle.

38   For the theory and practice of these treaties, see Heather (2001).

39   Chnodomarius: Ammianus 16. 12. Macrianus: Ammianus 29. 4. 2. Alamannic, Burgundian and Frankish wars: Ammianus 28. 5. 9–10, 30. 3. 7.

40   It is traditional to envisage only two, the Visigoths and Ostrogoths, but this is an anachronism. As we shall see in Ch. 5, the traditional equation between the Tervingi and the Visigoths cannot stand; the latter were created on Roman soil in the reign of Alaric in the 390s.

41   The evidence is an amalgam of written and archaeological materials (Heather (1996), Ch. 3).

42   There is a little archaeological evidence, but linguistic evidence is very much stronger. The fifth- and early sixth-century Burgundian language is distinctly east rather than west Germanic, despite the fact that they were then living in the west (Haubrichs (forthcoming)).

43   Frankish subgroups: Gregory of Tours Histories 2. 9. Strasbourg: Ammianus 16. 12. Chnodomarius and Macrianus: n. 39 above. Vadomarius: Ammianus 21. 3–4. One source refers to Athanaric as the ‘Judge of Kings’ (Ambrose On the Holy Spirit prologue 17), and the confederation of the Tervingi contained a number of subservient ‘kings’ (the Greek and Latin terms probably translate the Germanic reiks, which may have meant ‘noble’ rather than ‘monarch’).

44   On Wijster and Feddersen Wierde, see Van Es (1967); Haarnagel (1979). For a broader run of evidence with refs: Heather (1996), Ch. 3.

45   Poland: Urbancyzk (1997), 40. Gothic pottery and combs: Heather and Matthews (1991), Ch.3 with refs. Glass: Rau (1972); cf. Hedeager (1987) and Heather (1996), Ch. 3, for broader discussion.

46   Heather (1996), 65–75 with refs.

47   Tervingi: Wolfram (1988), 62ff.

48   Heather (1996), 66, 70–2 with full refs.

49   Ørsnes (1968); cf. Hedeager (1987) with refs.

50   Ammianus 16. 12. 60.

51   Passion of St Saba 4. 4, 7. 1–5.

52   Annals 13. 57.

53   Continental evidence: Heather (2000); Wickham (1992). Anglo-Saxons: Harke (1990); a number of those buried with weapons couldn’t possibly have fought, including one individual with spina bifida who couldn’t even have walked.

54   Militarized freedmen among the Germani appear in sixth- and seventh-century Visigothic and Frankish law codes and also in fifth- and sixth-century literary evidence: Heather (1996), App. 2; Heather (2000).

55   Wormald (1999), Ch. 2.

56   Passion of St Saba passim.

57   Tacitus Annals 12. 25.

58   Ammianus 17. 12. 9.


1    Ammianus 28. 6. 26.

2    Probably those due from Valentinian’s accession in 364. Offerings of crown gold, such as that taken by Symmachus to Valentinian in 369 (Ch. 1), were used to finance these payments.

3    Army returns: Jones (1964), Ch. 19. Lost cash: Symmachus Relationes 23. Mac- Mullen (1988) lovingly catalogues many of the documented scams.

4    Libanius Letters. 66.2 trans. in Norman (1992) as Letter 52. Themistius on an emperor’s friends: e.g. Or. 1. 10c ff. Connection and position: Matthews (1975), esp. Chs 1–2.

5    The creation of the Roman Empire had been driven by the link between revenues from overseas and political influence (Ch. 2). The same was true of the British Empire: Ferguson (2001). Valentinian and corruption: Ammianus 30. 9.

6    The Theophanes archive is edited and discussed in Roberts and Turner (1952), 104–56.

7    Including Chanel No. 5, I am reliably informed.

8    Many letter collections – including that of Symmachus – betray signs of having been put together from originals stored in such a fashion, and the archives of the early papacy (usually taken to reflect late Roman governmental practice) certainly operated in this way (Noble (1990); Markus (1997), App.). To find anything, therefore, you’d have to know which year it belonged to; there is no sign of any attempt to cross-reference by subject or place. Kelly (1994) explores what is known about the imperial archives in Constantinople.

9    To decide a city’s tax bill, the landed resources of each civitas were divided into units called iugera (sing., iugum). The iugum was a unit of value, not size, so a iugum of better-quality land would be smaller than a iugum of poorer quality. Each iugum was ascribed the same value of annual output, and had the same amount of tax levied on it. Deciding the total number of iugera for each city was the work of central government and required a thorough survey of agricultural assets (much complained about in our sources). Even this was not carried out uniformly throughout the Empire. In Syria, the assessors distinguished between three qualities of arable land and two types of olive grove. Elsewhere, a much simpler distinction between arable and pastoral was applied; while in Egypt and North Africa, existing land measures were approximated to the new tax units and no reassessment took place, presumably in part because it was too difficult and in part because it might have aroused resistance. The poll tax, similarly, was not applied equally throughout the Empire, sometimes applying to urban and rural plebeians, sometimes just the latter. On the details of Roman taxation, see Jones (1964), Ch. 13.

10   Local communities are known to have got into debt, for instance, in their determination to build the kind of buildings that would secure the necessary constitutional grant. In this context, the case of Irni in Spain is very interesting. Archaeologists have so far failed to identify its site, so unimpressive are the remains in the vicinity, and despite its wonderful constitution you begin to wonder whether it was ever much more than a notional or legal town.

11   P. Columbia 123; cf. Millar (1992), 245; Honoré (1994).

12   CTh 1. 2.

13   Jones (1964), Ch. 25, is the most considered and coherent statement of this kind of analysis. The orthodoxy is expressed in more strident terms in some of the older literature: Rostovtzeff (1957); CAH 1. 12, esp. Ch. 7; CMH 1. 1. esp. Ch. 19.

14   For the standard vision of ‘curial flight’, see Jones (1964), 737–63, commenting above all on the massive run of legislation collected at CTh 12. 1.

15   Agri deserti: Jones (1964), 812–23. Tying of peasantry to their land: Jones (1964), 795–812.

16   His findings were published most fully in Tchalenko (1953–8). More recent work has suggested that some of his conclusions as to the source of these villages’ prosperity need revision, but not the basic fact of it (e.g. Tate (1989)).

17   I have seen figures between 5 and 8 million bandied about.

18   Recent summaries and discussions of the survey evidence are: Lewitt (1991); Whittaker and Garnsey (1998); Ward Perkins (2000); Duncan Jones (2003).

19   In medieval England (up to c. 1300), bonds of serfdom tightened as the population grew and the peasants’ need for land increased, but loosened after the Black Death, when landowners needed labour, rather than land. Revised view of agri deserti: Whittaker (1976). Tax and subsistence agriculture: Hopkins (1980).

20   In imperial capital cities like Trier, Antioch and Constantinople at the highest level, and lesser regional ones such as Aphrodisias in south-west Asia MinOr. On this development, see e.g. Jones (1964), Ch. 19; Roueché (1989).

21   Some ancient historians use the Greek-derived term ‘euergetism’, ‘good works’, to describe the local competitive display of the early Roman period commemorated in the thousands upon thousands of inscriptions surviving from the first two and a half centuries AD. This is at least in part euphemistic.

22   Libanius Or. 42. 24–5.

23   Waiting lists: CTh 6. 30. 16; cf. Libanius Letters 358–9, 365–6, 362, 875–6. General reassessment of the problem: Heather (1994b) with full refs.

24   Among the main losers were builders and inscription-cutters in the small towns of the Empire.

25   Efficiency: Elton (1996a); Whitby (2002). Amongst the outside groups providing contingents for campaigns were the Gothic Tervingi (see Ch. 2). The really damaging sea-change came, after 382, when large contingents were recruited from barbarian groups who had established themselves on Roman soil and began to act as centrifugal political forces (see Chs 4 and 5). Much of the traditional argument has focused on the behaviour and loyalty of generals-cumpoliticians of non-Roman origin who rose to positions of great influence. Although called ‘barbarians’, many of these, like Stilicho, were second-generation immigrants and hence Roman; and anyway, the behaviour of such men betrays no sign of disloyalty. On Stilicho in particular, see in more detail p. 216ff.

26   Gibbon (1897), vol. 4, 162–3 (from General Observations on the Fall of the Roman Empire in the West).

27   Liebeschuetz (1972), 37–8, 104–5.

28   Coinage: Calo Levi (1952). Ursulus: Ammianus 20. 11. 5 and 22. 3. 7–8. Tax minimization: Jones (1964), 462ff., who, in my view, is inclined to make too much of what is highly normal human behaviour.

29   Grammarians: Augustine Confessions, esp. books 2–4 passim. This cultural revolution, and the cases of Melania and Paulinus of Nola in particular, have been much explored in recent years. For an introduction, see the many works of P. R. L. Brown, esp. (1981), (1995); Markus (1990); Trout (1999).

30   CTh 16. 5. 42 of the year 408 debarred pagans from imperial service. On the ideologies of the Christian Empire, see Dvornik (1966).

31   The story of this remarkable MS is told in Matthews (2000), Ch. 3.

32   There were currently two emperors, Valentinian III in the west and Theodosius II in the east.

33   The Egyptian papyri preserve some nice acclamation evidence: Jones (1964), 722ff.

34   Scholarship has concentrated on the Christianization of the Empire, so that the full story of the other side of the coin has yet to be told. For an introduction, see Jones (1964), Ch. 22; Markus (1990).

35   Gibbon’s case is actually much easier to make in relation to the Arab takeover of the eastern half of the Empire in the seventh century, where hostility between Greek and Syrian Orthodox (the latter often wrongly called monophysites) played a role in the process. On the contrary, religious disaffection played no substantial part in the Germanic takeover of the west in the fifth century.

36   On this softly-softly approach of fourth-century emperors to Christianizing their upper classes, see (most recently) Brown (1995); Bradbury (1994); Barnes (1995); Heather and Moncur (2001), esp. Ch. 1.

37   ‘Constitution’ is the official term for a formal imperial edict.

38   Rather than in shorthand. Late Roman bureaucratese had many ways of speeding up the laborious process of writing out, but shorthand involved some risk that a word might then be read wrongly.

39   Dvornik (1966); Barnish (1992), introduction; Heather (1993).

40   The full story is told in Matthews (2000), esp. Chs 4–5.

41   Matthews (1986) is a fascinating exploration of how Symmachus found ways to express criticism and conflict within the massive constraints imposed by the etiquette of Roman public life.

42   For Augustine’s education, see Brown (1967), esp. Chs 3–4.

43   Cohortales of Aphrodisias and Egypt: Roueché (1989), 73–5. Legal profession: Jones (1964), Ch. 14. McLynn (1994) and Van Dam (2003) are recent studies of the first generation of upper-class bishops; cf. Brown (1967), esp. Chs 17–19, on the stir created by the highly educated Augustine among the backwoods bishops of North Africa.

44   In large cities, chariot-racing was conducted through four teams, or factions: greens, blues, reds and whites. These were highly organized and could occasionally be mobilized – not least through rioting – to exert political pressure or to fulfil civic functions.

45   Maratacupreni: Ammianus 28. 2.

46   Letters 1. 1–5 (quotation above from 1. 5. 2). Further estate management letters: 2. 30–1; 5. 81, 6. 66, 6. 81, 7. 126. Another good example of late Roman ‘improvement’ is Paulinus of Pella Eucharisticon 187–97.

47   E.g. Letters 2. 87, 6. 11.

48   Letters 2. 13 (refers to CTh 4. 4. 2 of 389); cf. more generally 6. 2, 11, 27; 7. 12.

49   Letters 1. 6.

50   Letters 6. 3; cf. other references to marriage at 4. 14, 4. 55, 9. 83, 106, 107.

51   On the ‘Orfitus affair’, see Symmachus Relationes 34.

52   ‘Normal’ letters: Letters 1. 74; 3. 4; 4. 68; 5. 18; 6. 9 and the pair 5. 54 and 66. ‘Official’ letters: Relationes 16, 19, 28, 33, 38, 39, 41.

53   Letter 1. 12 (to his father). Sicilian baths: 1. 10, 2. 26, 2. 60, 5. 93; 6. 70, 7. 7, 7. 18, 8. 42. Builders: 6. 70. Cf., more generally, 1. 10, 2. 2, 6. 11, 6. 49, 7. 32.

54   Baiae in autumn: Letter 1. 7 (cf. 1. 3). The 396 ‘trip’: 5. 21, 93; 7. 24, 31, 69; 8. 2, 23, 27, 61; 9. 111 and 125. Other estate refs: 1. 5, 2. 59, 3. 23, 3. 50, 7. 35, 7. 59, 9. 83.

55   Though Symmachus could be dismissive of hunting as an immature pastime: Letter 4. 18.

56   Too busy: Letter 1. 35. Cf. Letters 1. 24 (included with it, a gift of Pliny’s Natural History), 3. 11 (on a Latin translation of Aristotle’s Constitutions), 4. 20 (on learning Greek with his son).

57   Letters 2. 47, 48; 3. 4; 3. 23.

58   Daily health letters: Letters 6. 32, with 6. 4 and 6. 29 on diet. More generally: 1. 48, 2. 22, 2. 55, 5. 25, 6.20.

59   P. Ital. 10–11 is a superb illustration of the complexities of Roman property transfer, showing that legal exchange was finalized only when the new owner was inscribed as such in the relevant town’s property register.

60   Priscus fr. 11. 2, pp. 267–73.

61   The headings of the Theodosian Code are deeply revealing – e.g. Purchase Contracts, Dowries, Inheritance – as are the curricula laid out for teaching Roman law: see Honoré (1978), Ch. 6, on both the new curriculum introduction in the sixth century by Justinian and the one it superseded. Eighteenth-century England: Linebaugh (1991), esp. Ch. 3.

62   The emperor Theodosius’ propaganda, for instance, made a great deal of the fact that he’d restored the position of a senatorial family that had been ruined by his predecessor Valens (Themistius Or. 16. 212d; 34. 18).

63   Themistius Or. 8. 114d.


1    Except where otherwise indicated, the quotations in this chapter are from Ammianus book 31, this one from 31. 4. 3.

2    Two hundred thousand: Eunapius fr. 42; cf. Lenski (2002), 354–5. In the 470s a force of something like 10,000 Gothic warriors dragged their families and possessions around with them in a train of at least 2,000 wagons (Malchus fr. 20). My own thoughts on numbers are based on the fact that Valens attacked the Goths at Hadrianople thinking that he was currently facing only about 10,000 Goths (Ammianus 31. 12. 3), at a point when he seems to have thought that he was facing only the Tervingi. Combatants to noncombatants has usually been estimated at 1:4 or 1:5, suggesting total numbers for the Tervingi of perhaps 50,000. The evidence suggests that the Greuthungi were similar in numbers to the Tervingi.

3    Ammianus 31. 2.

4    See Maenchen-Helfen (1973), Chs 8–9.

5    Atitle rather than – as per Walt Disney – a personal name.

6    Many studies have tackled this thorny question, but for an introduction see Maenchen-Helfen, (1945) and Twitchett and Loewe (1986), esp. 383–405.

7    Jordanes Getica 24. 123–6; cf. Vasiliev (1936), on the story. Twentieth-century commentary: Bury (1928).

8    For an introduction to the Chionitae and the Guptas, see Encyclopaedia Iranica: Yarshater (1985–2004, ongoing).

9    Ammianus 31. 3. 2: ‘he found release from his fears by taking his own life’.

10   Ammianus 31. 3. 7 describes these walls as extending ‘from the river Gerasus [the modern Pruth] to the Danube and skirting the lands of the Taifali [Oltenia]’. I argue my view in Heather (1996), 100, with refs to alternatives.

11   Ammianus 31. 4. 12.

12   This is the traditional chronology. Wanke (1990) has argued instead for the spring of 376 on no very good grounds, and Lenski (2002), 182ff., 325f., for the early summer on the grounds that it was the Goths’ arrival that encouraged Valens to make aggressive noises towards the Persians in the summer of 376. I find it it inconceivable, however, that an emperor who had wrapped up the Balkan front before confronting Persia in the late 360s (see Ch. 2) would have deliberately stirred up conflict in Armenia after hearing that the Danube was again in turmoil; so for me this confirms that the Goths arrived on the Danube only after Valens had made his moves in the east, hence in late summer 376 at the earliest.

13   Ammianus 31. 3. 8.

14   Ammianus 31. 3. 3.

15   As emerges from a huge Hunnic attack of that year, which went through the Caucasus rather than over the Danube (see further p. 202ff.).

16   Ammianus 31. 2. 8–9. Zosimus 4. 20. 4–5.

17   The archaeological evidence for the Huns’ bow is collected and discussed in Harmatta (1951); Laszlo (1951); Bona (1991), 167–74. The history of the recurve bow and information on best recorded shots are from Klopsteg (1927). Klopsteg’s attempts to model firepower have been superseded by the mathematically based work of Kooi; for an introduction, see Kooi (1991), (1994). These and his other studies can be read online at his website Key variables affecting the performance of a bow include length of limbs, shape of cross-section, elastic properties of the material used, draw length, arrow weight, weight and elastic properties of the string.

18   Olympiodorus fr. 19.

19   Jordanes Getica 49. 254: attributed to Priscus.

20   Laszlo (1951); Harmatta (1951).

21   Sources: Ammianus 31. 4. 4; Eunapius fr. 42; Socrates HE 4. 34; Sozomen HE 6. 37. Most tell a broadly similar story, but Ammianus is much more detailed, and Eunapius places more emphasis on Gothic treachery.

22   Ammianus records a justification for the admission of the Limigantes that strongly recalls that given in relation to the Goths of 376: ‘[By accepting the Limigantes] Constantius would gain more child-producing subjects and be able to muster a strong force of recruits’ (19. 11. 7).

23   St Croix (1981), App. III, gives an exhaustive list of known moments of immigration. The story of the Sciri is in Sozomen HE 9.3 and CTh 5. 6. 3. For full discussion of Roman policy and literature, see Heather (1991), 123–30.

24   The quotations are from Ammianus 10. 11. 10–15.

25   Ammianus 31. 5. 9.

26   Valens’ aggressive moves against the Persians in summer 376 are well constructed in Lenski (2002), 180–5. Cf. n. 12 above: in my opinion (not Lenski’s) all this must place the arrival of the Goths on the Danube after Valens’ aggression – i.e., late summer at the earliest.

27   If Sozomen HE 6. 37. 6 can be given any credence, Ulfilas may have taken part in the diplomatic process; cf. Heather and Matthews (1991), 104–6.

28   Ammianus 31. 4. 12.

29   For detailed discussion of the treaty’s terms, see Heather (1991), 122–8. There are two points of controversy. On the basis of Socrates HE 4. 33, some date the conversion of the Tervingi before 376; cf. Lenski (1995). This, however, contradicts Ammianus’ detailed contemporary account of events. Socrates, later and much less well informed, is very unlikely to be correct. I hold, therefore, to the conclusion first reached in Heather (1986). The other controversy, from the same source, centres on the report of Eunapius fr. 42 that the Goths were meant to surrender their weapons on crossing into the Empire, but didn’t. This source also claims that the Goths swore a secret oath never to stop until they destroyed the Empire. But neither the secret oath nor the illegal smuggling of weapons are reported anywhere else, particularly not in Ammianus, and were clearly used by Eunapius as devices to explain the Goths’ later victory at Hadrianople. Neither is convincing, in my view, since Valens was hoping to use Gothic auxiliaries alongside his regular troops, and the Goths, as we shall see in a moment, were much too wary of the Empire to have entered it unarmed. Most scholars who accept the disarming story reject the secret oath (e.g. Lenski (2002), 343ff.), a solution which seems to me arbitrary.

30   Themistius Or. 13. 163c after Socrates Symposium 203d.

31   Ammianus 31. 5. 5–8.

32   Other Roman kidnaps: Ammianus 21. 4. 1–5; 27. 10. 3; 29. 4. 2ff.; 29. 6. 5; 30. 1. 18–21. Many other scholars see malice aforethought in Lupicinus’ banquet (e.g. Lenski (2002), 328) without drawing conclusions about what this suggests about the orders he had received from Valens.

33   Ammianus 31. 4.

34   Diuque deliberans: Ammianus 31. 3. 8.

35   Ammianus 31. 5.

36   It is not fanciful to suppose that the Goths were aware of the Persian situation. Barbarians watched troop movements on the other side of the frontier closely (Ammianus 31. 10. 3–5), and contacts were close enough for information to pass. The continued contact between the Tervingi and Greuthungi is powerful evidence of the Goths’ suspicions.

37   Valens could still contemplate hiring Gothic auxiliaries for his Persian war sometime in winter 376/7 (Ammianus 30. 2. 6), which must have been before the outbreak of war.

38   There is an immense bibliography on the development of different parts of the Roman Balkans, but no overview. The preceding paragraphs are a synthesis based on such major studies as Mocsy (1974); Lengyel and Radan (1980); Wilkes (1960); and Hoddinott (1975), with monographs such as Poulter (1995), and Mango (1985) on Constantinople.

39   In Germanophone historiography, it has been traditional since Várady (1969) to argue that the Greuthungi of Alatheus and Saphrax were composed of three equal ethnic contingents: Goths, Alans and Huns (the so-called Drei Völker group). Part of this scholarly fantasy is the argument that Ammianus’ report of Huns and Alans joining the revolt in autumn 377 (see below) depicts the moment when Alatheus and Saphrax joined the Tervingi in revolt. This is nonsense. The Huns and Alans of autumn 377 (Ammianus mentions no Goths) were entirely separate from Alatheus and Saphrax, who were already south of the Danube and probably joined in the revolt immediately after the revolt of Lupicinus. For full refs and further discussion, see Heather (1991), App. B; cf. Lenski (2002), 330–1.

40   Ammianus 31. 6. 4.

41   On the forts, see Scorpan (1980); Petrovic (1996); their garrisons are listed at Not. Dig. Or. 39.

42   Ammianus 31. 6. 5–8.

43   According to Roman itineraries, a road station called Ad Salices (‘By the Willows’) was located in the far north of the Dobrudja, but Ammianus says this confrontation took place near Marcianople, 150 kilometres further south. At the start of the revolt proper, the Gothic wagon train had already wound its way to within 15 kilometres of Marcianople, and it’s hard to see why the Goths would have retreated again so far north. So perhaps oppidum Salices should not be identified with Ad Salices.

44   Summer to autumn: Ammianus 31. 8. 2. The 377 campaign is recounted at Ammianus 31. 7.

45   Ammianus 31. 10. 1: so early November, perhaps.

46   New alliance and collapse of blockade: Ammianus 31. 8.4. Cf. n. 39 above: some argue that this was the moment when Alatheus and Saphrax joined Fritigern in revolt. Note, however, that Ammianus makes no mention of Greuthungi in this context.

47   On the Arabs, see Lenski (2002), 335f. with refs. Destruction of villas: Poulter (1999).

48   Ammianus 31. 11.

49   The story of these months is told in Ammianus 31. 10–11.

50   Ammianus’ account of the battle can be found at 31. 12. Larger estimates of Gothic losses: Hoffmann (1969), 444 n. 138, 450–8; cf. e.g. Lenski (2002), 339 with refs. But if Valens had really had 30,000-plus troops to deploy against the Goths, I doubt that he would have needed to wait for Gratian, nor had to worry about whether the Goths were all there or not, because I don’t believe that even the combined Goths could put many more than 20,000 men in the field (see n. 2 above).

51   Ammianus 31. 16. 7.

52   Ammianus 31. 12.

53   Eunapius frr. 47–8, survive from what was originally seemingly a lengthy account of these cities’ tribulations.

54   This reconstruction results from the realization that Zosimus 4. 24–33, based on Eunapius’ history, is actually a coherent account of events between 378 and 382, which has been ruined only by Zosimus importing into his text a second account of the war at 4. 34. Most scholars think that the Greuthungi made a separate peace with Gratian and were settled in Pannonia in 380, but I remain unconvinced; see Heather (1991), 147ff. and App. B, for further argument and full refs.

55   Themistius Or. 16. 210b–c.

56   There is literally a hole at the crucial point in the manuscript at Themistius Or. 34. 24, which makes it unclear whether he’s referring to Goths being settled in Macedonia post-382, or only to their attack on the province before the peace agreement.

57   These leaders are presumably those who would have held semi-autonomous reiks status beneath Athanaric the iudex (Judge) in the old days of the Tervingi confederation north of the Danube (Ch. 2), and from among whom Fritigern and Alatheus and Saphrax had emerged in 376.

58   The significance of this treaty has long been recognized: e.g. Mommsen (1910), Stallknecht (1969). For more detailed discussion and full refs, see Heather (1991), 158ff.

59   To the end of the chapter, unless otherwise indicated, the quotations are from Themistius Or. 16.

60   Themistius Or. 14. 181b–c.

61   Zosimus 4. 32–3, with Heather (1991), 152–5; Heather and Moncur (2001), Ch. 4; cf. n. 54 above.

62   These Goths were more Greuthungi led by a certain Odotheus: Zosimus 4. 35. 1, 38–9; Claudian On the Fourth Consulate of the Emperor Honorius, 626ff.

63   Poulter (1995), (1999).

64   ‘These men crossed over into Asia under the law of war, and, having depopulated [vast tracts] . . . settled in this territory which they now inhabit. And neither Pompey nor Lucullus destroyed them, although this was perfectly possible, nor Augustus, nor the emperors after him; rather, they remitted their sins and assimilated them into the empire. And now no one would ever refer to the Galatians as barbarian but as thoroughly Roman. For while their ancestral name has endured, their way of life is now the same as our own. They pay the same taxes as we do, they enlist in the same ranks as we do, they accept governors on the same terms as the rest and abide by the same laws. So will we see the Scythians [Goths] do likewise within a short time’ (Or. 16. 211c–d).

65   Ambrose Commentary on the Gospel According to Luke 10: 10: trans. Maenchen- Helfen (1973), 20.


1    Jerome Commentary on Ezekiel, Preface to book 1. Pagan comment: Augustine Sermon 296; for a full survey of the reactions, see Courcelle (1964), 58ff.

2    On Olympiodorus and his parrot, and subsequent uses of the text, see Matthews (1970); Zosimus made the join between Eunapius and Olympiodorus at 5. 26. 1.

3    Zosimus 5. 26. 3–5.

4    Radagaisus as Goth: sources as PLRE 2, 934.

5    These refugees included inhabitants of Scarbantia who took with them the body of St Qurinus. CTh 10. 10. 25 and 5. 7. 2 also refer to them: Alföldy (1974), 213ff.

6    Claudian Gothic War ll. 363ff. (cf. 414–15); cf. Courtois (1955), 38ff.

7    Alamanni and Quadi are mentioned by Jerome as participating in the Rhine crossing (Letter 123. 15). Wolfram (1988), 387 n. 55, takes a similar view, but Thompson (1982b), 152–3, disagrees. Fifth-century Suevi: Pohl (1980), 274–6.

8    Jerome Letter 123. 15. Fourth-century Sarmatians: Ammianus 17. 12–13 (Sarmatians who did not participate in 406 continued to live on the Danube (Pohl (1980), 276–7)).

9    March to join Valens: Ammianus 31. 11. 6. Western Roman army: Zosimus 4. 35. 2.

10   Sozomen HE 9. 25. 1–7; cf. CTh 5. 6. 2. Thompson (1996), 63–4, is clear that Uldin was a relatively minor figure. For an alternative view, to my mind mistaken, see Maenchen-Helfen (1973), 59–72, esp. 71.

11   Sidonius Poems 12.

12   Fourth-century Burgundians: e.g. Matthews (1989), 306ff.; on their subsequent movements: e.g. Demougeot (1979), 432, 491–3.

13   Zosimus 5. 35. 5–6.

14   Paulinus of Pella Eucharisticon 377–98.

15   CTh 5. 6. 2.

16   Perhaps more, since we are not entirely sure of late Roman unit sizes.

17   Radagaisus’ warriors: Olympiodorus fr. 9: Photius’ summary says ‘12,000 nobles’, but this is usually taken as a confused total figure: e.g. Wolfram (1988), 169–70; Heather (1991), 213–14. According to Augustine, Radagaisus had ‘more than’ 100,000 followers (City of God 5. 23); according to Orosius (7. 33. 4), 200,000; and according to Zosimus, 400,000 (5. 26). None of these command much authority.

18   Procopius Wars 3. 5. 18–19 says that 80,000 was the number of warriors, but more contemporary and better-informed Victor of Vita (History of the Persecution 1. 2) gives this as a total figure, reporting that the king divided his followers into 80 groups of notionally 1,000 apiece, their real size being somewhat smaller. Eighty thousand warriors is unbelievable, making the new group twice as powerful in military terms as the largest possible estimate for Alaric’s Goths. Victor lived among the Vandals and there is a good chance that he generally knew what he was talking about, even though his work is highly polemical. Goffart (1980), 231–4, takes a dismissive view even of Victor’s evidence (Victor notes, incidentally, that others had even in his day confused the total figure for the number of warriors), but on p. 33 is happy enough on a priori grounds to suppose that Vandal–Alan forces must have numbered tens of thousands.

19   Jerome Chronicle 2389. Orosius 7. 32. 11.

20   Some would place it much closer to, or even before, the year 400 (Heather (1996), 117ff., for a summary with full refs).

21   Radagaisus: Olympiodorus fr. 9 distinguishes optimatoi (‘the best’) from the rest; cf. Heather (1996), App. 1, more generally.

22   For: e.g. Lot (1939), 78–9; Courtois (1955), 39–40; Musset (1965), 103–4; Demougeot (1979), 415. Against: Goffart (1980), 2ff., esp. 16–17; cf. Maenchen- Helfen (1973), 60–1, 71–2.

23   Some Hunnic raiding parties got as far as the Danube in 376 (Ch. 4), but they were operating from a power-base much further to the east.

24   Claudian Against Rufinus 2. 26ff. records two threats to the eastern Empire in 395, one through the Caucasus, one on the Danube; 36ff. makes clear that Alaric’s Goths are the Danubian threat – not, as has sometimes been supposed, a second group of Huns settled much further west.

25   Marcellinus Comes s.a. 427; cf. Jordanes Getica 32. 166.

26   The general was Aetius, of whom we will have much to say in the next two chapters; the Hunnic help he had recruited seven years earlier presumably came from the same region (refs as PLRE 2, 22–4).

27   Royal tombs: Priscus fr. 6. 1. On Attila’s camp: Browning (1953), 143–5.

28   On Olympiodorus’ affiliations, Matthews (1970). The sea journey: Olympiodorus frr. 19 and 28, both surely referring to the same crossing; cf. Croke (1977), 353. Others have envisaged Olympiodorus visiting the Pontus – e.g. Demougeot (1970), 391–2 – but we also know that in 409 Honorius was expecting the imminent arrival of ten thousand Huns (Zosimus 5. 50. 1). This was after the defeat of Uldin, implying that contact must have been made in the meantime with other Huns. It is tempting to associate this new alliance with general Aetius’ period as hostage among the Huns in c. 410 (refs as PLRE 2, 22).

29   CTh 7. 17. 1, 28 January 412: ‘We decree that there shall be assigned to the Moesian border ninety patrol craft of recent construction and that ten more shall be added to these by the repair of old craft; and on the Scythian border, which is rather widespread and extensive there shall be assigned one hundred and ten such new craft, with fifteen added by the restoration of antiquated ones . . . These shall be equipped with all their weapons and supplies at the instance of the duke and shall be constructed on the responsibility of his office staff.’

30   Mango (1985), 46ff.

31   It doesn’t matter that much to the argument whether the mass of the Huns arrived west of the Carpathians during the crisis of 405–8 or in the course of the next decade or so. After 376, when the Tervingi and Greuthungi abandoned their homes north of the Black Sea, it took some years for the Huns to reach the Lower Danube frontier. Given that their arrival north of the Black Sea had been preceded by a huge demographic convulsion east of the Carpathians, the Huns’ intrusion on to the Hungarian Plain in 410–20 would certainly have been preceded by similar upheavals to the west.

32   In dealing with an attack on Italy in 401/2, Stilicho had drawn off forces from Gaul and Britain, and there’s every reason to suppose he did the same this time.

33   The Alans were perhaps recruited from north of the province of Raetia (Claudian Gothic War, 400–3); the Huns were sent by the – at this point – still biddable Uldin.

34   Orosius 7. 37. 37ff.

35   Commonitorium 2. 184.

36   The last two quotations are from Prosper of Aquitaine Epigramma 17–22; 25–6. A good introduction to these poems is Roberts (1992); see also Courcelle (1964), 79ff.

37   Hydatius Chronicle 49.

38   Wars 3. 33.

39   Orosius 7. 43. 14.

40   The Goths of Fritigern did much the same thing in Macedonia in c. 380 after three years of looting (see Ch. 4).

41   Zosimus 6. 1. 2 slightly conflated with Olympiodorus fr. 13, on the original of which Zosimus directly drew. The Olympiodorus fragment makes clear that the British usurpations began before 1 January 407 (the seventh consulship of Honorius), whereas Zosimus misunderstood his source to say that they began in the seventh consulship.

42   This emerges, despite some garbling, from Zosimus 6. 3.

43   Orosius 7. 40. 4.

44   This, the consensus view, has been argued against by Liebeschuetz (1990), 48–85, but I remain content with the counter-arguments offered in Heather (1991), 193–9.

45   Orosius 7. 35. 19.

46   This was later the case with Attila, who was made a notional imperial general in order to facilitate the pretence that the cash subsidy paid to him was salary for his troops, rather than a subsidy. I suspect that the dodge was first hatched in the case of Alaric, fifty or so years before. On the background to Alaric’s revolt, see in more detail, Heather (1991), 181–92, 199ff.

47   In my view, this had happened as early as 395, when the revolt broke out. Others argue that it didn’t happen until 408/9 when Alaric’s brother-in-law Athaulf joined him in Italy with a force of Huns and Goths from Pannonia. The date, however, is really a matter of detail. The fundamental point is that old distinctions had been put aside. This was presumably made much easier by the suppression of the two groups’ former leaders (Fritigern, Alatheus and Saphrax, under the treaty of 382), who might have had an interest in preserving the difference. For full details and refs to alternative views, see Heather (1991), 213–14, App. B.

48   Cameron (1970), 159ff.; cf. 474ff. for a convincing unravelling of the two main sources’ confusion regarding the two campaigns.

49   On the fall of Eutropius: Cameron (1970), Ch. 6; Heather (1988).

50   On Gainas and Constantinopolitan politics: Cameron and Long (1993), Chs 5–6 and 8. The Silvanus episode: Ammianus 15.5. The literature on Alaric’s first Italian adventure is enormous, but for an introduction, see Heather (1991), 207ff.

51   CTh 15. 14. 11.

52   He had last held office in 383, and had made a number of bad decisions in between, including backing the usurper Maximus (Matthews (1974)).

53   Letters 5. 51; 6. 2 and 27.

54   In Rufinem (Against Rufinus) 2. 4–6.

55   On Stilicho in general, see Cameron (1970), esp. Chs 2–3, and Matthews (1975), Ch. 10. This view of Stilicho’s manoeuvres with Alaric is argued in more detail with full refs in Heather (1991), 211–13.

56   The next two quotations are from Zosimus 5. 32. 1; 5. 33. 1–2.

57   CTh 9. 42. 20–2.

58   The fall of Stilicho and Olympiodorus: Matthews (1970); (1975), 270–83.

59   According to Zosimus, 30,000 fighters joined Alaric after the pogrom, but that is impossibly high. I suspect that he again misunderstood Olympiodorus, who was certainly his source here, thinking that 30,000 was the number of recruits when it was actually the total for Alaric’s force including the recruits. Thirty thousand would be my guess for the approximate size of Alaric’s force after the reinforcements arrived; i.e. 10,000 each from the Greuthungi and Tervingi of 376, combined with the 10,000-plus former followers of Radagaisus that Stilicho had drafted into the Roman army.

60   The next two quotations are from Zosimus 5. 48. 3 and 5. 50. 3–51. 1.

61   Sarus and Alaric had quarrelled at some point before the latter came to Italy (Zosimus 6. 13.2). Sarus met his death at the hands of Athaulf (Olympiodorus fr. 18), and Sergeric benefited from the assassination of Athaulf and his family (Olympiodorus fr. 26. 1).

62   For more detailed accounts of the events leading to the sack, see e.g., with full refs, Matthews (1975), Ch. 11; Heather (1991), 213–18.

63   Livy 5. 41. 8–9. The sources for Alaric’s sack are gathered in Courcelle (1964), 45–55.

64   The quotations in the rest of this passage, unless otherwise stated, are from St Augustine City of God, here 3. 17.

65   City of God, 2. 19–20.

66   The ‘lust for domination’ follows Sallust Catiline War 2. 2; the quotation is from City of God 3. 14.

67   The quotation is from City of God 2. 29. Augustine continues: ‘Among these very enemies are hidden [the Heavenly City’s] future citizens; and when confronted with them she must not think it a fruitless task to bear with their hostility until she finds them confessing the faith. In the same way, while the City of God is on pilgrimage in this world, she has in her midst some who are united with her in participation of the sacraments, but who will not join with her in the eternal glory of the saints’ (City of God 2. 34–5).

68   The quotation is from City of God 2.18. For introductions to The City of God and the full range of response to the sack, see Courcelle (1964), 67–77; Brown (1967), Chs 25–7.

69   The next five quotations are from Rutilius De Reditu Suo. The poem is translated in Keene and Savage-Armstrong (1907). On the journey and its circumstances, see Cameron (1967); Matthews (1975), 325–8.

70   The sun god, pulled in a chariot, by horses, across the sky.

71   The Carmen de Providentia Dei is partly translated and fully analysed in Roberts (1992).

72   Usually known by his full name – although the Flavius would normally be dropped – to prevent possible confusions with the usurper Constantine III. Full refs to the sources for his career can be found at PLRE 2, 321–5. The best account of his career is Matthews (1975), Chs 12–14.

73   Olympiodorus fr. 23.

74   This usurper is not the Magnus Maximus defeated by Theodosius I in 387, but a much more obscure claimant of the same name.

75   Matthews (1975), 313–15 (there is a slight question mark over the geography).

76   On the pay rise, see Sivan (1985) who dates it to 416; but it was much more likely to have been an earlier measure to stimulate loyalty among troops who had followed the usurpers and were now to fight barbarians.

77   Orosius 7. 43. 2–3.

78   Fr. 24.

79   Sergeric: see n. 61 above. For more detailed discussion of Constantius and the Goths, with full refs, see Heather (1991), 219–24. One modius is approximately one-quarter of a bushel.

80   Chronicle 24.

81   If by an indirect route: via Photius’ ninth-century summary of the church historian Philostorgius, who used Olympiodorus’ work. The key passage can be found at Philostorgius HE 12. 4–5, or Olympiodorus fr. 26. 2.

82   North Africa in the 440s, see p. 292–3. On the larger historiographical issue which has surrounded the economic form of barbarian settlement inside the Empire, see p. 423–4ff. with refs.

83   Two distinguished historians of a previous generation, Professors Thompson (1956) (revolting peasants) and Wallace-Hadrill (1961) (possibly Saxon pirates), virtually came to blows over this. I would particularly recommend to readers the appendix to Thompson (1982c), whose substantive discussion begins ‘Unfortunately, in 1961 discussion of the matter was thrown into confusion by some ill-judged pages of J. M. Wallace-Hadrill . . .’

84   The authorities in Constantinople have sometimes been criticized for not doing more, but that’s not very realistic: Demougeot (1951), esp. pt III, places a complete split between east and west far too early. More balanced are Kaegi (1968), Ch. 1, and Thompson (1950). For fuller discussion of Constantinople and the west, see Ch. 9.

85   Maenchen-Helfen (1973), 69, doubted that these Huns ever arrived, although most scholars have thought the auxiliaries did eventually turn up. As part of this agreement, a young man by the name of Aetius, of whom we shall see a great deal more in subsequent chapters, spent time as a hostage among the Huns, which does suggest that the negotiated relationship had a real point.

86   The agricultural surplus of the Empire can be thought of as being divided between the landowning classes (who received their share in the form of rent) and the imperial authorities (who received theirs in the form of tax). In the interests of satisfying the Goths within a fairly restricted area, Constantius was perhaps willing to forgo the Empire’s share of the Garonne surplus, so that the area’s tax revenues – as well as rents on public lands – could be allocated to the Goths’ support. Again, something similar happened in North Africa.

87   This quotation and the next one are from Zosimus 6. 5. 2–3; 6. 10. 2.

88   On the Saxon shore defences, see Pearson (2002). For introductions to the range of opinion generated by these (and a few other references), see e.g. Campbell (1982), Ch. 1; Higham (1992), esp. Chs 3–4; Salway (1981), Ch. 15.

89   De Reditu Suo 1. 208–13.

90   Taking a slightly more optimistic view than Matthews (1975), 336–8, on the basis of much the same evidence. Cologne and Trier did not fall into Frankish hands until 457 (Book of the History of the Franks 8).

91   CTh 11. 28. 7 and 12.

92   Not. Dig. Occ. 5 and 6; 7.

93   You can tell the date because, within each category of unit, regiments are arranged in the chronological order of their creation, and there are just a handful named after Valentinian, the son born to Constantius and Placidia in July 419. On all this and what follows, see above all, Jones (1964), App. II.

94   Uncertainty over late Roman unit sizes (p. 62) prevents greater precision, but 500 men per regiment is the minimum figure.

95   As with the losses after Hadrianople, we don’t know what percentage of a unit had to die before the decision was taken not to re-form it (cf. p. 181).

96   For the Notitia and the western army, see Jones (1964), App. III, and the various papers in Bartholomew (1976).

97   Eucharisticon 302–10.


1    Ammianus 22. 7. 3–4.

2    Fr. 33.

3    Ammianus 28. 1. 24–5.

4    On imperial ceremonial in general and Julian’s impropriety, see Matthews (1989), Chs 11–12; MacCormack (1981). On patterns of appointment, a good introduction is Matthews (1975), passim, with many of the essays collected in Matthews (1985), esp. Matthews (1971) and (1974).

5    Fr. 11.1.

6    For an introduction to regime change Roman-style, see Matthews (1975), e.g. 64ff. on the aftermath of the death of Valentinian I.

7    Ammianus 27. 22. 2–6.

8    Many of Libanius’ letters, for instance, are surprisingly aggressive to potential patrons, demanding that they show what they are made of: see e.g. Bradbury (2004), Letters 2, 5, 8, 9 etc.

9    The two commanders were first sentenced to exile but killed on the journey.

10   The Letters of Ravenna date the assassination to 7 March 413, but this seems too early in the year. Ships did not normally pass between Carthage and Italy between November and March, and we need time for Heraclianus first to land, be defeated and return to Africa. Perhaps he landed in Italy on 7 March. See generally Orosius 7. 42. 12–14; other sources as PLRE 2, 540.

11   Olympiodorus fr. 33.1.

12   On Galla Placidia, see Oost (1968).

13   Fr. 38.

14   PLRE 2, 1024.

15   Olympiodorus fr. 43.2.

16   See Matthews (1970) on Olympiodorus, with Matthews (1975), Ch. 15, on the history.

17   Prosper Tiro s.a. 425; Chron. Gall. 452 no. 102; the figure of 60,000 is impossibly large.

18   For Aetius, Felix and Boniface, refs as PLRE 2, 23–4, 238–40, 463–4. The main secondary accounts I have drawn upon in this and subsequent discussions of Aetius’ activities are: Mommsen (1901); Stein (1959), Ch. 9; Zecchini (1983); Stickler (2002).

19   Refs. as PLRE 2, 22–3.

20   Ammianus 31. 2. 25.

21   No law codes survive from the Vandal kingdom, but Procopius’ narrative of the Byzantine conquest notes in passing the pattern observed among other Germanic groups of this period of having two distinct military castes, which I take to be those of the ‘free’ and the ‘freed’ (see Ch. 2). The remains of the Przeworsk culture, from which the Vandals derived, also suggest no obvious difference in social structure compared with other better-documented Germani. For an introduction to the fragile and relatively egalitarian social structures of nomads, see Cribb (1991) with full refs.

22   Gregory of Tours Histories 2. 9.

23   Chronicle 42, 49, 67–8, esp. 68: Alani qui Vandalis et Sueuis potentabantur: ‘the Alans who were ruling over the Vandals and Sueves’.

24   Hydatius Chronicle 68.

25   Depending on the extent of the exaggeration reported by Victor of Vita, see p. 198–9. On the official titulature of the kings, see Wolfram (1967).

26   All narrative accounts of the Vandals and Alans in Spain are largely based on the Chronicle of Hydatius. Dating is affected by the controversy over how extant versions of the text should be edited: see e.g. Burgess (1993), 27ff., for an introduction to the dispute. My notes here follow the referencing system used in Mommsen (1894); for our purposes, these arguments fortunately affect details rather than overview.

27   Hydatius was writing when Visigoths were pillaging widely in Spain, and he is always critical of them. Given how readily they had helped defeat Vandals and Alans in the later 410s, it is unclear why they should have had such a change of heart in 422.

28   Jordanes Getica 33: 168.

29   Not. Dig. Occ. 25.

30   Wars 3. 3. 22f.

31   The peace was negotiated by the senator Darius, with whom Augustine exchanged pleasantries (Augustine Letters 229–31).

32   The historical reconstruction offered in the rest of this section depends upon Courtois (1955), 155ff. (with refs).

33   Not. Dig. Occ. 26.

34   The quotation above is from Victor of Vita 1. 3. Amongst the bishops tortured Victor names Pampinianus of Vita and Mansuetus of Urusi.

35   Letter 220. Olympiodorus fr. 42.

36   The illustrator of one of the later copies of this manuscript mistook these for upside-down ducks, presumably with rigor mortis having already set in, since Africa is holding them by the necks and the bodies are sticking up in the air.

37   As a politically incorrect and chronologically befuddled sergeant of the British Eighth Army put it in late 1941, after a lecture from the enthusiastic Director of Army Education in Tripoli on the wonders of Roman North Africa: ‘We now know all about this place; it is full of the ruins of buildings put up by the Eyeties before the war, but as far as I can see, it is now just full of camels and Western Oriental Gentlemen’ (quoted in Manton (1988), 139).

38   Even these were tending to dwarf: a sure sign that they had been cut off from the mainstream for some time.

39   Tripolitania was also administered from Carthage in the Roman period.

40   Historians of Roman Africa writing in the nineteenth and early twentieth centuries supposed that large-scale immigration from Italy was an important contributory factOr. There certainly was some immigration. When Carthage was refounded as a Roman colony in 29 BC, for instance, 30,000 Italian settlers crossed the sea. Also, the third legion was stationed in North Africa from 23 BC, generating a steady trickle of veteran settlers who bought farms and established townships such as Diana Veteranorum, Timgad, Thurburbo Maius and Djemila. But by the third century AD the vast majority of the 600 Roman towns in North Africa were inhabited by indigenous Romano-Africans, so that the role of immigrants must not be overstated.

41   A famous inscription from Ain Zraia (ancient Zarai) in Algeria records that whereas most goods were taxed at 2–2.5 per cent, animals and their products (textiles, hides etc.) were rated at only one-fifth to one-third of 1 per cent.

42   The ‘colonial’ view of North Africa can be found in Baradez (1949). See also the more recent revisions of Whittaker (1994), 145–51, and the collected essays of Shaw (1995a), nos 1, 3, 5, 6, and (1995b), no. 7. Readily accessible general accounts of Roman North Africa are: Raven (1993); Manton (1988).

43   Carthage: Ennabli (1992), 76–86. Utica: Procopius Wars 3. 11. 13–15.

44   CTh 13.5 and 6 contain a long run of relevant imperial pronouncements.

45   Lepelley (1979–81). Even elsewhere, as we saw in Ch. 3, this change had more to do with the restructuring of city finance in the third-century crisis and its political repercussions than with economic decline.

46   ILT 243.

47   CIL 8, 18587. Again, the French colonial view held that it was European settlers who brought with them new water technologies, such as the 50-kilometre aqueduct which serviced Carthage in the late antique period, required to make North Africa bloom. But that massive construction brought water to the city for luxury, not agricultural, purposes, particularly to service its massive Antonine baths.

48   Mactar inscription, olives and rural expansion: Raven (1993), 84–6, 92–6. A good general account of the rural survey findings is Mattingly and Hitchner (1995).

49   Expositio Totius Mundi 61.

50   The amusements of the amphitheatre included everything from wild beast hunts, in which two hunters might be pitted against nine bears in a battle to the death, to risqué representations of the loves of Jupiter, to novelty acts such as acrobats acting out dramas on tightropes. At the circus, you could find the everpopular chariot-racing, but the young Augustine particularly loved the theatre. At the age of sixteen he had come to Carthage, where ‘a cauldron of illicit loves leapt and boiled about’ him; and in response to the stage: ‘In my wretchedness, I loved to be made sad and sought for things to be sad about: and in the misery of others – though fictitious and only on the stage – the more my tears were set to flowing, the more pleasure did I get from the drama and the more powerfully did it hold me’ (Confessions 3. 2).

51   For an introduction to the findings of the UNESCO excavations, see Ennabli (1992).

52   Gregory of Tours 2. 8, after Renatus Frigeridus. On the background of Aetius, see most recently Stickler (2002), 20–5, with refs.

53   Zosimus 6. 2. 4–5; Hydatius Chronicle, 125, 128.

54   For revisionary views of the Bagaudae, see Drinkwater (1993); Halsall (1993); arguing against an older, more Marxist strand of interpretation, e.g. Thompson (1956).

55   They vary in form from Latin prose, to hexameters, to elegiac couplets, to Phalacean metre.

56   The base and commemorative inscription still survive (CIL 6. 1724).

57   On Merobaudes, see generally Clover (1971), who discusses the manuscript as well as editing and translating the Poems; with further discussion in PLRE 2, 756–8.

58   The quotations in the next few pages, unless otherwise indicated, are from Merobaudes’ Panegyrics 1 and 2.

59   The insurgents are called ‘Nori’, but this pre-Roman conquest tribe hadn’t existed for centuries, so they may well have been another quasi-Bagaudic group.

60   Sources as PLRE 2, 166; cf. Courtois (1955), 155–71; Stickler (2002), 232–47.

61   Priscus fr. 11. 1, p. 243; the date and extent of the concession has occasioned much debate: for an introduction, see Maenchen-Helfen (1973), 87ff.

62   Merobaudes Panegyric 1 fr. IIB.

63   Refs as PLRE 2, pp. 24–5; most recent commentary, with full refs: Stickler (2002), 48ff. Procopius Wars 3. 3. 15 refers to Aetius (and Boniface) as the last of the Romans.

64   Quodvultdeus In Barbarian Times 2. 5.

65   Merobaudes Carmen IV.

66   Nov. Val. 5. 1, 6. 1.

67   Nov. Val. 9.

68   Force from the east: Theophanes AM 5941, with CJ 12. 8. 2 on Pentadius. Secondary commentary: Courtois (1955), 171–5.

69   Panegyric 2. 51–3. The next two quotations are from Panegyric 2. 61–7, 2. 98–104.

70   Nov. Val. 34: 13 July 451.

71   Basic terms of the 442 treaty: Procopius Wars 3. 14. 13, with Clover (1971).

72   Panegyric 2. 25–33.

73   Carmen 1. 5–10.

74   ‘Allotment’ may conjure up visions of Vandals and Alans on the outskirts of Carthage tending their vegetable patches, putting up sheds and comparing the sizes of their marrows. And, in fact, that connotation is not entirely inappropriate: see next note.

75   Victor of Vita 2. 39. I owe this point to Moderan (forthcoming), whose discussion renders redundant the suggestion of Goffart (1980), 67–8 n., that the Vandal settlement after 439 merely reallocated provincial tax revenues. Goffart’s suggestion was based on an argument by analogy rather than detailed consideration of the North African evidence. On this historiographical issue, see further pp. 423–4ff.

76   Celestiacus: Theodoret of Cyrrhus Letters 29–36; Maria: 70.

77   This is where the allotment image breaks down, and, in any case, any residual agricultural expertise among the north-central European Vandals would have been inappropriate for the Mediterranean littoral.

78   This wasn’t as good as outright ownership, but the new tenants were given emphyteutic leases with full inheritance rights, so that they had some security (Nov. Val. 34).

79   Nov. Val. 13. This degree of remission is comparable with that granted to areas around Rome affected by Alaric’s Visigoths between 408 and 410 (see p. 245–6).

80   These were the kinds of things that emperors often gave as favours, and when the tax-base was large they could be easily absorbed. Now, however, as the law put it: ‘the weight of the tribute which is withdrawn from certain persons individually falls back on the others’ (Nov. Val. 4).

81   Nov. Val. 7. 1; slightly modified in 7.2 of 27 September 442.

82   Nov. Val. 10.

83   Four thousand two hundred solidi (one-eighth of the old total) were now to be paid by Numidia under the general tax account, together with 1,200 military subsistence allowances and 200 units of animal fodder. Five thousand solidi (again, one-eighth of the previous figure) and 50 units of animal fodder were demanded from Mauretania Sitifensis. Each military subsistence allowance and fodder unit had also been reduced by one solidus (Nov. Val. 13).

84   Following the calculations of Elton (1996a), 120–5, which must be approximately correct, even if detailed points are arguable.

85   The next two quotations are from Panegyric 2. 55–8, 75–6.


1    Opinions on Attila: Thompson (1996), 226–31; Maenchen-Helfen (1973), 94ff. Thompson and Maenchen-Helfen are the two main accounts available in English. The quotation is from Marcellinus Comes, Chronicle s.a. 447.2. A few years ago, I was asked to revise the old article on Attila in The Oxford Dictionary of the Christian Church. They said change whatever you like, but keep ‘scourge of God’.

2    One source dates Rua’s death to 434, but is demonstrably wrong (Maenchen- Helfen (1973), 91–4).

3    The next three quotations are from Priscus frr. 2, 6.1, 6.2.

4    Compare the approaches of Thompson (1945) and Blockley (1972).

5    Julian, Letter to the Athenians 279A–B; Ammianus 16. 2ff.; cf. Matthews (1989), Ch. 6.

6    For an introduction to the development of the Silk Road cities, see Boulnois (1966).

7    Ammianus 31. 10. 3–5 gives an excellent example of cross-border intelligence from winter 377/8. The Alamanni learned about Roman troop movements from a retired guardsman, but also observed them for themselves.

8    On the context of his life, see Toynbee (1973); Runciman (1929).

9    That we have part of volume 50 suggests that most of them did once exist; the quotation is from the Preface to Constantine’s Excerpts concerning Virtues and Vices.

10   For an introduction to Constantine’s project, see Lemerle (1971), 280–8.

11   I here broadly follow Maenchen-Helfen (1973), 116–17.

12   Working with his book is a bit like working with Priscus. As its editor’s introductory note explains, the author brought a ‘beautifully typed’ manuscript into the offices of California University Press in early January 1969, but then died just a few days later. The manuscript turned out not to be a completed book, but only some chapters, and despite much editing the book as finally published remains highly episodic, missing a great number of connecting passages. But none of this detracts, again like the fragments of Priscus, from the quality of what is there.

13   Maenchen-Helfen (1973), 86–103, discussing Theophanes AM 5942.

14   Given that 442–7 is six years, 8,400 pounds should have been paid in this period, but that the arrears were 6,000 suggests that only 2,400 were ever handed over.

15   Nov. Theod. 24.

16   On the recruitment of Isaurians, see Thompson (1946).

17   Quoted in Maenchen-Helfen (1973), 121.

18   Life of Hypatius 104.

19   Poulter (1995), (1999).

20   Priscus fr. 9. 3, p. 238; cf. its survival of Avar attacks, too, 150 years later: Theophylact Simocatta History 7.3. The next two quotations are from Priscus fr. 9. 3.

21   Road: fr. 11. 2, p. 263. Tents: fr. 11. 2, p. 251.

22   Priscus fr. 11. 2, p. 275.

23   We have virtually the whole of Priscus’ account transmitted in a number of different fragments. They are arranged in chronological order and translated in Gordon (1960), Ch. 3 (with commentary), and Blockley (1983), frr. ll. 1–15. 2, whose translation is quoted in what follows. Except where otherwise indicated, the quotations in the rest of this chapter are from Priscus.

24   Fr. 11. 2, pp. 247–9.

25   Circuit walls: Priscus fr. 11. 2, p. 265; buildings: fr. 11. 2, p. 275; seating arrangements: fr. 13. 1, p. 285; furnishings: fr. 11. 2, p. 275; greeting: fr. 11. 2, p. 265.

26   Rather as in the old game of Kremlin-watching, this ceremonial life made it abundantly clear when someone had been promoted or demoted.

27   The next two quotations are from Getica 34: 182; 35: 183 (= Priscus fr. 12. 2).

28   Fr. 2, p. 227.

29   Fr. 14, p. 293.

30   Priscus fr. 11. 2, p. 267. The sharing of booty is, as we shall see in a moment, fully reflected in the archaeological evidence, but also referred to in passing in the literary evidence: e.g. Priscus fr. 11. 2 p. 263f.

31   Priscus fr. 15.2, p. 297.

32   Olympiodorus fr. 19.

33   Some commentators have taken Matthews (1970) to task for simply assuming this to have been the case (see Matthews (1985), additional note), but the text is ambiguous and his original thought could easily be correct.

34   ‘Er, yes – I’m afraid I gave the best gifts to the wrong king and the Akatziri are now our mortal enemies. Sorry, Sir.’ Nearly as good a diplomatic foul-up as the occasion when the Queen referred to the battle of Waterloo as an excellent example of Anglo-German cooperation, much to the chagrin of the French.

35   It also puts a further nail in the coffin of any idea that Uldin, prominent before 411, possessed anything like the power of Attila. There are other reasons, as we have seen, for rejecting this idea (p. 198), but, just as important, it is clear that no Hun king was so dominant at such an early date.

36   Heather (1996), 113–17, on the rise of Valamer among the Pannonian Goths: he killed Vinitharius, the head of one line (but also married his granddaughter), and forced Beremud, from another, to flee; but Gensemund, Beremud’s uncle, accepted his overlordship. See further Ch. 8.

37   Part of the manoeuvre to catch Vigilas involved banning the Romans from making further purchases of Hunnic horses, and the first trouble with Attila and Bleda involved a surprise attack on market day (Priscus fr. 6. 1).

38   Itself worth the far from negligible sum of nearly seven pounds of gold.

39   The figures are those of Lindner (1981) who, in a famous article, concluded that, since Hunnic armies in the time of Attila certainly numbered tens of thousands, the Huns could no longer have been nomads, since there wasn’t enough space in Hungary for so many horses. He missed the important point that most of the manpower for Attila’s armies was provided by his Germanic subjects, not by the Huns themselves, so we don’t need to postulate so many horses anyway. The military manpower point is fully explored later in this chapter.

40   Languages: Priscus fr. 11. 2, p. 267: ‘Being a mixture of peoples, in addition to their own languages they cultivate Hunnic or Gothic or, in the case of those who have dealings with the Romans, Latin.’ Names: Maenchen-Helfen (1973), 386ff.; cf. JordanesGetica 9:58 on the adoption of names between different language groups.

41   Sozomen HE 9.5; CTh 5. 6. 3.

42   Theophanes AM 5931; cf. Procopius Wars 3. 2. 39–40, with Croke (1977). The date could be either 421 or 427.

43   Gepids, Rugi, Suevi, Sciri and Herules are all named in the post-Attilan narrative of events on the Great Hungarian Plain (see Ch. 8), while Attila intervened in the internal politics of the Franks (Priscus fr. 20. 3), which makes it very likely that he exercised some kind of hegemony over Lombards and Thuringians, and possibly Alamanni too, all of whom lived closer to the Hunnic heartland.

44   I find Priscus perfectly comprehensible, unlike Baldwin (1980), who argues that these terms were all used in a confused and confusing fashion.

45   You can tell how people were dressed by where they wore their safety-pins – all that tends to survive of clothing in most graves.

46   Reasons for archaeological invisibility can range from the dramatic, where bodies are left exposed to the elements and wild animals, to the prosaic: cremation followed by the scattering of ashes, or bodies being buried without any chronologically indentifiable grave goods – which often makes medieval northern European cemeteries undatable once populations convert to Christianity.

47   Although written sources occasionally provide information, which can be used in conjunction with archaeological evidence to identify particular groups.

48   The archaeological ‘horizons’ are differentiated from one another by discernible variations in the manner in which broadly similar sets of grave goods were decorated. In chronological order – and there are overlaps – the sequence starts with the Villafontana horizon, followed by those of Untersiebenbrunn and Domolospuszta/Bacsordas (names not for the faint of heart!).

49   Many of the Germanic groups of central Europe in the first to third centuries had practised cremation, but inhumation was already spreading more widely before the arrival of the Huns.

50   For introductions to these finds, see Bierbrauer (1980), (1989); Kazanski (1991); Tejral (1999). Wolfram (1985) has some excellent illustrations.

51   The point was first emphasized by Bury (1928).

52   Priscus fr. 15. 4, p. 299.

53   Priscus fr. 11. 2, p. 277.

54   Sources as PLRE 2, 568–9. Besides, this kind of thing is typical of the rich and underemployed – especially those of energy and determination – caught up in the empty ceremonies of a highly orchestrated court life. In Nepal in 2001 a drunken prince went berserk and shot ten of his relatives, including the reigning monarch, before killing himself; readers with slightly longer memories will recall the much covered-up death of a Saudi princess who strayed from the expected paths of inaction.

55   Plate: Priscus fr. 11. 2, pp. 263, 265, 277. Frankish succession: Priscus fr. 20. 3. On contacts with Geiseric and the general diplomatic context: Clover (1972).

56   Getica 33: 182.

57   Poem 7. 319ff.

58   My account of Attila’s two western campaigns draws heavily on Thompson (1996), Ch. 6; Maenchen-Helfen (1973), 129ff. The latter deals only with the Italian campaign: an account of the attack on Gaul was not found among the extant fragments (see n. 12 above).

59   Jordanes Getica 195.

60   The battle narrative is from Jordanes Getica 38:197–41:218.

61   Fr. 22. 2, p. 313 = Procopius Wars 3. 4. 33–4.

62   Hydatius Chronicle 154.

63   The fact that Aetius was either unable or unwilling to confront Attila head-on with another confederation, as in Gaul, has often aroused comment. Prosper Tiro s.a. 451 says that Aetius was caught unprepared, and some have believed this to be so. Maenchen-Helfen (1973), 135ff., convincingly reconstructs Aetius’ countermeasures and sets them in the context of other contemporary Roman defences of the Po valley. I also follow Maenchen-Helfen in understanding Hydatius Chronicle 154 to mean that Aetius received eastern military help in Italy, as well as benefiting from a campaign by eastern troops on the Danube.

64   Even in more modern eras, armies campaigning over these kinds of distances have come unstuck. In the summer of 1914, German armies reached the gates of Paris, before rolling back again (there’s no record that St Genevieve had anything to do with this). They were brought to a halt by a bold tactical manoeuvre on the part of the French army on the River Marne, but also by exhaustion, which gave the French their opportunity. A British trooper, falling back before the German advance, reported: ‘The greatest strain . . . was . . . fatigue . . . I fell off my horse more than once, and watched others do the same, slowly slumping forward, grabbing for their horse’s neck, in a dazed, barely conscious way. At any halt men fell asleep instantaneously’ (quoted in Keegan (1988), 107). There were, of course, a number of differences between 451 and 1914. In 1914, the distance from Belgium to Paris was covered at speed in about two weeks, the men marching 40 kilometres a day, day after day. The Hunnic advance was a much more leisurely affair. In 1914 the Germans had travelled to the German-Belgian border by train, leaving only about 500 kilometres to cover on foot and on horseback, and they did have supplies and wagon trains.

65   Getica 49: 256–8.

66   Although they did attempt to broker a peace between the Suevi and the Gallaecian provincials.

67   Any account of Spain in the 430s and 440s has to be built up from Hydatius 91–142.

68   The Pelagian heresy is named after the Romano-British theologian Pelagius, who argued, not least against Augustine, that salvation required not just Divine Grace, upon which others put the emphasis, but also great individual effort to live a virtuous life.

69   Though not any more in cash, as the non-subsistence elements of the local economy, such as its pottery industries, seem to have collapsed by c. 420.

70   The three quotations that follow are from Gildas On the Ruin of Britain 23. 5, 24. 3 and 20. 1.

71   On the crisis of the 440s and the end of Roman Britain, see e.g. Campbell (1982), Ch. 1; Higham (1992), Chs 5–8; Salway (1981), Ch. 16; Esmonde Cleary (2000).

72   The rest of North Africa, southern Gaul as far east as Arles, north-western Gaul affected by the Bagaudae, central Gaul and northern Italy affected by Attila’s campaigns.


1    Jordanes Getica 48: 246–55: 282.

2    Momigliano (1955) and Goffart (1988) come to opposite conclusions about the Cassiodorus–Jordanes relationship from more or less the same set of observations. Suggestions as to why Jordanes should have lied mostly turn on the fact that he was writing on the eve of an east Roman campaign which destroyed the Ostrogothic Italian kingdom. It has been argued that the Getica contains an important political message (from Cassiodorus, or pretending to be from Cassiodorus), urging people not to resist the east Roman forces. These hypotheses conveniently ignore the problem of how the Getica’s supposed political message was to be disseminated. The only way of turning a literary history into political propaganda is to suppose that landowners were assembled and exposed to the Getica in the way that people had been exposed to the speeches of a Themistius, a Merobaudes or a Sidonius Apollinaris. This is highly unlikely. For more technical argument, see Heather (1991), Ch. 2.

3    Cassiodorus concentrates on the royal dynasty from which Theoderic (Cassiodorus’ master) came – the Amal family – and organized Gothic history by geography, dividing it up according to the different abodes inhabited by Goths at different times. Both of these points are evident in Jordanes’ history. See further Heather (1993).

4    Group 1: first known from Jordanes’ account of the collapse of the Hunnic Empire, then from many sources subsequently (see further Heather (1996), 111–17). Group 2: see p. 193–4. Groups 3/6: best evidence from Malchus of Philadelphia in the 470s, origins perhaps documented in Theophanes AM 5931 (see further Heather (1996), 152ff.); cf. above Ch. 7 n. 42. Group 4: Jordanes Romana 336. Group 5: Priscus fr. 49. Group 7: Procopius Wars 8. 4. 9ff. (‘not numerous’), Buildings 3. 7. 13 (3,000 warriors strong).

5    Valamer and his nephew Theoderic united at least groups 1 and 6, but possibly also 3 and 4; see Heather (1991), Ch. 1, with full refs.

6    This reference and the next quotation are from Getica 50: 261–2, 50: 260.

7    Theoderic the Amal, the Ostrogothic king of Italy, fought against the Gepids, for instance, in 488/9 and again in the early 500s.

8    Getica 50: 262–4.

9    Getica 268–9, 272–3.

10   Priscus fr. 49.

11   Getica 272–3.

12   Getica 248–52.

13   The crucial passage is Getica 248–52 plus the discussion of Heather (1989), with full refs to previous attempts to resolve its obvious difficulties.

14   Refs as PLRE 2, 385f. Maenchen-Helfen (1973), 388 and n. 104, denies the identity of the two Edecos, but it is generally accepted; cf. (and more generally on the emergence of the kingdoms that would succeed the Huns’) Pohl (1980).

15   Except possibly the Amal-led Goths. Jordanes says that they came west of the Carpathians after the Huns fled east after following the battle of the Nedao (Getica 50: 263–4). This seems unlikely, and I suspect that they were settled in Pannonia by the Huns and not on their own initiative, but there is no way to be certain.

16   Getica 53: 272–55: 282.

17   Getica 54: 279.

18   Fr. 45.

19   Getica 52: 268 vs. 53: 273.

20   Thompson (1996), esp. Ch. 7; cf. Maenchen-Helfen (1973), 95ff.

21   Priscus fr. 11. 2, p. 259.

22   Tervingi and Greuthungi: p. 145. Burgundians: p. 198. Of various Goths, group 3 above, perhaps identical with the later Thracian Goths (n. 4 above), were detached from Hunnic overlordship by Roman military intervention, and the Amal-led Pannonian Goths (group 1) were very clear that they had been made part of the Hunnic Empire by force, even if the ‘Balamber stories’ of the Getica are confused (n. 13 above).

23   Merchant: Priscus fr. 11. 2, p. 269 l.419–p. 272 l.510. Gibbeting: Priscus fr. 14 p. 293 ll. 60–5.

24   Fr. 49.

25   Theophanes AM 5931 (group 3): the point obviously holds, whether or not group 3 can be identified with group 6 (see n. 4 above).

26   The next three quotations are from Priscus fr. 2, p. 225; p. 227; p. 227.

27   The Romans provided Attila with a succession of secretaries, including the prisoner Rusticius who wrote the odd letter (Priscus fr. 14, p. 289). This governmental machine made lists of renegade princes who had fled to the Romans and possibly kept track of the supplies required from subject groups.

28   Most dominated: the Goths who appear in Priscus fr. 49, part of which is quoted above. Least dominated: the Gepids who led the revolt against Attila’s sons (Jordanes Getica 50: 260–2). In between, the Pannonian Goths of Valamer (Jordanes Getica 48: 246–53, 52: 268ff.), with commentary by Heather (1996), 113–17, 125–6.

29   Each of the Hunnic-period ‘horizons’ is named after one of these rich burials.

30   An excellent introduction to Apahida and the other rich burials of the period is provided by the catalogue volume Menghin et al. (1987).

31   Gold certainly existed in fourth-century Germania and was being worked into plate. The famous fifth-century treasure from Romania, the Pietroasa horde (Harhoiu (1977), contains one or two items which were clearly antiques at the time of deposition and must have been made in the mid-fourth century. Roman gold coins were also far from unusual.

32   After Bierbrauer (1980).

33   E.g. Ammianus 17. 12–13, 19. 11, on Constantius’ settlement in the Middle Danube in the late 350s, with further commentary by Heather (2001).

34   Odovacar in Gaul: Gregory of Tours Histories 2.18 (sometime between 463 and 469); cf. PLRE 2, 791–3. Herules, Alans and Torcilingi: Procopius Wars 3. 1. 6: Ennodius Life of St Epiphanius 95–100.

35   Romana 336.

36   Sidonius Poems 2. 239ff.

37   The other possibly complicating factor was the arrival of a new nomad power north of the Black Sea. By the early 480s, for instance, Bulgars had established themselves close to the Empire’s Danube frontier (John of Antioch fr. 211. 4).

38   Fr. 37.

39   Fr. 45.

40   Jordanes Getica 49: 255, probably originating in Priscus and translated therefore by Blockley as Priscus fr. 24. 1.

41   Best general accounts of Aetius’ fall: Stein (1959), 347ff.; Stickler (2002), 150ff.

42   For Petronius’ career, with full refs, see PLRE 2, 749–51.

43   The next few quotations are from Priscus fr. 30.

44   On Avitus’ early career, with full refs, see PLRE 2, 196–8.

45   For a recent study of Sidonius and his life and times, see Harries (1994); Stevens (1933) remains valuable.

46   Dill (1899), 324; for a collection of similar judgements, see Harries (1994), 1–2.

47   Letter 4. 10. 2, quoted in Harries (1994), 3.

48   Many have contributed to this revolution in appreciation, and an excellent introduction is Roberts (1989) with references to other relevant studies.

49   The next few quotations are from Sidonius Poem 7.

50   Priscus fr. 30. 2.

51   Full reconstruction: Courtois (1955), 185–6.

52   Priscus fr. 32 = John of Antioch fr. 202.

53   On this Gallic literary context, see e.g. Harries (1994), Chs 1–2.

54   Sidonius Letters 1. 2.

55   The story of the campaign is told in Hydatius Chronicle 173–86. On the previous joint Romano-Gothic campaigns in Spain, see Chs 5 and 6 above.

56   It is impossible to reconstruct Burgundian history in detail, but for fuller discussion and refs, see Favrod (1997).

57   History of the Persecution 1. 13.

58   This quotation and the next are from Sidonius Poem 7. 233–6, 286–94.

59   Sidonius Poem 361ff. Next quotation: 510–18.

60   Refs as PLRE 2, 198.


1    Depending on the balance between 500- and 1,000-man units: Jones (1964), vol. 3, 364 and 379; see further p. 63. The eastern section of the Notitia dates only to the mid-390s, but eastern armies suffered no massive losses after this. In 395, the east also controlled the entire field army of Illyricum (another 26 regiments), but thereafter west Illyricum and its forces were returned to western control.

2    The argument of Goffart (1981). Constantine defeated a series of rivals between 306 and 324 to unite the Empire, having started by controlling just Britain and Gaul. Julian had been appointed Caesar in the west by his cousin Constantius in 355, but revolted in 360, uniting the whole Empire under his control on Constantius’ sudden death in 361.

3    Theodosius accepted that Persia should exercise hegemony over two-thirds of Armenia, taking just one-third for himself.

4    A fair summary of Roman–Persian relations can be found in Blockley (1992). Rubin (1986) brings out the extremely peaceful nature (in relative terms) of Roman–Persian relations in the fifth century as opposed to the sixth or fourth.

5    Not. Dig. Or. 5, 6, 8.

6    CTh 7. 17. 1 of 412.

7    421: Theophanes AM 5931 (cf. Ch. 8). Ruga: Maenchen-Helfen (1973), 81–94.

8    Zosimus 6. 8. 2–3.

9    Thus Hydatius Chronicle 154 reports: ‘The Huns . . . were slaughtered by auxiliaries sent by the Emperor Marcian and led by Aetius, and at the same time they were crushed in their settlements both by heaven-sent disasters and the army of Marcian.’ See further Ch. 7 above.

10   Cameron (1970), 176ff. disposed of the old argument that Alaric was encouraged to move into Italy by Constantinople. Edward Thompson’s judgement can be found at Thompson (1996), 161ff.; he was particularly exercised by the revenge Attila would have wreaked on the east in 453 had his own death not brought a halt to campaigning. I therefore find nothing in the detailed narrative to support the argument of Goffart (1979) (see n. 2 above).

11   Scholarly opinion on Ricimer’s policies has ebbed and flowed, with some regarding his barbarian origins as having undermined his loyalty to Rome. But the Visigothic succession was actually in the hands of another line, descended from Vallia’s successor Theoderic I, so that Ricimer would probably not have been very popular had he turned up in Visigothic Aquitaine; and Ricimer’s policies, while certainly self-interested, show no general pro-barbarian bias. O’Flynn (1983), Ch. 8, provides a general survey. On Majorian, see PLRE 2, 702–3, with refs.

12   Poems, 2, 317–18. Best narrative account: Stein (1959); 380ff.; cf. O’Flynn (1983), 111–17. For the Gallic perspective on these manoeuvres, see Harries (1994), Chs 6–7.

13   Hydatius Chronicle 217.

14   As Sidonius put it: ‘Receiving a count’s authority, he traversed the Danube bank and the whole length of the great frontier lines, exhorting, arranging, examining, equipping’ (Poems 2, 199–201). As comes rei militaris (second-grade field army general), he sorted out the mess on the Danube in 453/4, as the civil war between Attila’s sons progressed and the successor kingdoms to the Hunnic Empire emerged.

15   He continued to deal with the longer-term fall-out from the Hunnic collapse, confronting Valamer when he invaded Illyricum in search of a subsidy in c. 460, and fighting off the Hunnic fragment led by Hormidac which invaded the Empire in the 460s.

16   In the 390s, as recorded at Not. Dig. Or. 19, the field army of Illyricum comprised 26 units, something over 10,000 men. Illyricum was then divided between the east and west on Stilicho’s accession in 395, and its western army was already down to 22 units by 420 (Not. Dig. Occ. 7. 40–62). Subsequently, the region took a heavy pounding, including the loss of Pannonia to the Huns, so that by the 460s the command was largely confined to Dalmatia and its military establishment probably much smaller, although Marcellinus supplemented his regular troops with barbarian auxiliaries (Priscus frr. 29, 30). On Marcellinus in general, see MacGeorge (2002), pt 1.

17   Visigoths and Burgundians: Harries (1994), Ch. 6. Roman army of the Rhine: MacGeorge (2002), pt 2. Brittany: Galliou and Jones (1991), Chs 1–2. Franks: James (1988), Chs 2–3; Wood (1994), Ch. 3.

18   Poems, 13. 35–6.

19   Letters 1. 11.

20   Letters 1. 9.

21   Majorian had forced the Burgundians to give up some of the cities (civitates) of the Rhône valley and their revenues; most significantly Lyon, which they had seized during Avitus’ reign; and he had browbeaten the Visigoths into acknowledging his power, as well as attracting Gallic landowners. On Anthemius and the Gallo-Romans, see Harries (1994), Ch. 7.

22   The sources variously report that the marriage occurred both before the Vandal sack and upon Placidia’s arrival in Constantinople. They were perhaps betrothed in 454/5, therefore, and actually married in 462 (PLRE 2, 796–8, with Clover (1978) generally on Olybrius). Vandal sack of Rome: p. 378 above.

23   Justinian’s general Belisarius managed to conquer North Africa in 532/3.

24   The next three quotations are from Sidonius Poems 5. 53–60, 338–41, 349–50.

25   I would assert this principle with the greatest insistence, although some don’t seem to realize how much the public life of the late Empire resembled that of a one-party state: see further Heather and Moncur (2001), esp. Ch. 1.

26   Sidonius Poems 349–69, 441–69.

27   This raises the same issue faced with Geiseric in Ch. 5, as to whether Majorian’s forces were to be transported in one lift or several. Belisarius required 500 ships to move 16,000 troops (see p. 400), so Majorian’s 300 could have moved about 9,600 in one go, and I doubt that he was planning to go into battle with so few. So I suspect he was planning for at least two movements of troops, and hence would not have wanted to place his advance guard too close to Carthage. On Majorian’s campaign, see further Courtois (1955), 199–200.

28   Candidus fr. 2 = Suda X 245.

29   Respectively John Lydus On the Magistrates 3. 43; Procopius Wars 3. 6. 1; cf. Courtois (1955), 201; Stein (1959), 389–91.

30   Eleven hundred ships: Priscus fr. 53 = Theophanes AM 5961; the MS reads ‘100,000 ships’, 1,100 being an emendation based on the figure of 1,113 supplied by Cedrenus, p. 613. The amended figure would make the 468 armada the same size as that assembled for the putative 441 expedition which never sailed: see p. 290. In 532, when the emperor Justinian mounted a more exploratory expedition towards Vandal Africa, the eastern Empire assembled 500 ordinary ships, together with 92 specialist warships (dromons), which again makes 1,100 seem proportionate for an all-out effort.

31   On the fleet of 532, see Casson (1982) with refs to ancient shipping in general. To put the east Roman effort into broader perspective, the ‘invincible Armada’ which set sail from Spain in the late spring of 1588 comprised 90 great ships of 300 tons’ displacement or more, and another 40 auxiliary craft. This, however, was merely a covering force for the army of the Duke of Parma, who was meant to provide additional barges to get his men across the Channel.

32   Manpower: Procopius Wars 3. 6. 1. Marcellinus: sources as PLRE 2, 710. Heraclius: Theophanes AM 5963.

33   In contradiction of Courtois (1955), 201, who is generally an excellent guide, but wished to downplay the scale of the effort made in 468.

34   The next three quotations are from Poems 2. 14–17, 537–43 and 315–16.

35   Belisarius’ later fleet sailed from Italy for Africa on 21 June 532 and eventually put in to the Bay of Utica, close to Carthage, which was large enough to accommodate its 600 ships.

36   Poems 5. 332–5.

37   CTh 9. 40. 24; Zosimus 1. 31–3, with commentary by Heather (1996), 38–43, on the third century. Zosimus explicitly notes that ships and sailors were provided by settlements north of the Black Sea.

38   Viereck (1975), 165–6, gathers the references.

39   Mattingly (2002), 313.

40   The next three quotations are from Procopius Wars 3. 6. 18–19, 20–1 and 22–4.

41   Procopius Wars 1. 6. 10–16. It is worth comparing Basiliscus’ failure in 468 with Belisarius’ success in 532. Belisarius sailed with a smaller force, landed safely, and wrapped up the Vandal kingdom in under a year via two decisive land battles. He landed at Caputvada, south of Cape Bon, clearly much further away from Carthage than Basiliscus’ target, whatever it was. Belisarius also achieved complete strategic surprise. The North African expedition was an unexpected gamble on the part of his master the emperor Justinian, prompted by a succession dispute in the Vandal kingdom which divided the Vandal forces. Hence, when Belisarius turned up, 120 ships and 5,000 of the Vandals’ best fighters were away squashing a revolt in Sardinia, and he was able to land his men without a battle at sea. In amphibious landings, among the most difficult of military operations, the attackers should have at least a 6:1 advantage over the defenders according to modern doctrines. Basiliscus may well have been trying to land too close to Carthage, which would have brought him right into the middle of the Vandal fleet. But he may have been on a hiding to nothing anyway. News of his fleet, anticipated with so much excitement by Sidonius in January 468, could hardly be suppressed. Geiseric was bound to know of its arrival, so it is doubtful, huge as it was, that it had enough of an advantage to pull off victory in the face of forewarned and mobilized opponents. In 1588, likewise, the Spanish design was flawed: Medina Sidonia didn’t have a powerful enough fleet to hold off the English, and the Duke of Parma had nowhere near enough transports and smaller escorts to get past a Dutch inshore fleet and make it to England. Parma knew this perfectly well, and didn’t even try to get his men ready, even though he had plenty of advance warning of the Armada’s arrival.

42   His reticence has prompted a wave of inconclusive speculation on the part of excitable modern historians. The most dramatic hypothesis about Severinus is that of Lotter (1976), who argues that the Life really begins after c. 460 and not c. 453 (the death of Attila), and that Severinus was in fact the consul of 461, with a long administrative career behind him. I don’t subscribe to this (see the response in Thompson (1982)), but I do think that the events begin closer to 460 than 453, since, rather than the Huns themselves, they feature the successor states to the Hunnic Empire (such as the Rugi and the Goths of Valamer), which took some time to form.

43   On provincial development in Noricum, see Alföldi (1974), passim.

44   Not. Dig. Occ. 39: river police at Boiodurum (Passau Instadt), Asturis (Zeiselmauer) and Cannabiaca; normal cavalry units at Comagena (Tulln), Augustiana (Tralsmauer), Arelape (Pochlarn) and AD Mauros (Eferding); mounted archers at Lentia and Lacufelix.

45   For a survey of the evidence: Alföldi (1974), Ch. 12.

46   This quotation and the next are from the Life of Severinus 30. 1 and 20. 1–2.

47   Walls and militias: Comagenis Life of Severinus 2. 1; Faviana 22. 4; Lauriacum 30. 2; Batavis 22. 1; Quintanis 15. 1; barbarians 2. 1.

48   Not. Dig. Occ. 7. 40–62. The evolution of the Illyrican field army can be followed because we can compare the listing for it from AD 395, or just before, in the eastern Notitia, with the western listing from c.420 in the distributio numerorum. Theselanciariihad clearly been drafted into the field army sometime between 395 and 420.

49   Life of Severinus 4. 1–4.

50   Advance warning: Life of Severinus 30; defeating assaults: 25, 27; rescuing prisoners: 4, 31.

51   Capture by slavers: Life of Severinus 10; Tiburnia: 17; Herules’ destruction: 24, 27; Rugi and attempted transplants: 8, 31.

52   If increasingly manned by barbarians fleeing the fall-out from Hunnic collapse, the Roman army of Italy continued to exist under Ricimer; so too to some extent the army of Gaul, parts of which had gone into revolt under Aegidius in 462 (MacGeorge (2002), Ch. 6).

53   Hydatius Chronicle 238–40.

54   Getica 45: 237.

55   For further details and full refs, see Wolfram (1988), 181ff.

56   Burgundians: Favrod (1997). Franks: James (1988), 72ff.; Wood (1994), 38ff.

57   Harries (1994), 222ff.; cf. Sidonius, Letters 3. 3, on Ecdicius.

58   The next five quotations are from Sidonius Letters 1. 7. 5; 5. 5; 8. 3; 8. 9.

59   Arvandus: Sidonius Letters 1.7. Vincentius: Chron. Gall. 511, s.a. 473; cf. PLRE 2, 1168; Sidonius’ description of the trial of Arvandus was also penned for a Vincentius, but it is unknown whether it was the same man. Victorius: PLRE 2, 1162–3; Seronatus: Sidonius Letters 2.1; 4. 13; 7. 2. Unlike Arvandus, Seronatus was not Sidonius’ friend, so he was unmoved by the former’s fate. As early as the 410s, some Gallo-Roman landowners had been attracted to Athaulf’s standard as the best path to peace.

60   Solon was the legendary law-giver of Athens, who gave them their first written code. Written law had an important cultural significance for Romans.

61   But Sidonius was ever the Nimby.

62   Eucherius: Sidonius Letters 3. 8. Calminius: Letters 5. 12 (claiming that Calminius hadn’t wanted to be there). On Sidonius’ son, see PLRE 2, 114.

63   For an introduction to the Visigothic kingdom: Heather (1996), Ch. 7, with refs.

64   A common poetic designation for the Franks.

65   On Sidonius’ imprisonment and release, see most recently Harries (1994), 238ff.

66   Book of Constitutions 54. 1.

67   Goffart (1980), supported esp. by Durliat (1988); (1990). For counter-argument specifically with regard to the Burgundian kingdom: Heather (forthcoming a); Innes (forthcoming). See more generally Wickham (1993); Liebeschuetz (1997), Barnish (1986) and n. 75 to Ch. 6 above on the Vandal kingdom. The Burgundian freemen had their own dependants – freedmen and slaves – which was why they received a smaller fraction of the labour force. Subsequent Burgundian legislation dealt with matters that changed the value of one of the partners’ share in the joint estate (assarting by forest clearance, or the planting of vineyards – more valuable per hectare than arable fields), and with the pre-emption rights of the original Roman owner should his Burgundian partner decide to sell up. All of these regulations, like the original order, make much more sense in relation to actual landed property than to taxation deriving from it (BC 31, 55. 1–2, 67, 84).

68   This is an extremely unlikely omission if the collection and distribution of taxation played the key political role in the structure of the kingdom that Goffart’s argument would suggest.

69   CE frr. 276, 277; cf. Liebeschuetz (1997).

70   Best narrative account: Stein (1959), 393ff. Gundobad’s ‘sudden’ departure from Italy: Malalas 375.

71   Details: Courtois (1955), 209.

72   Count of the Domestics: Procopius Wars 5. 1. 6. Patrician: Malchus fr. 14 (not noted in PLRE 2, 791–3).

73   Life 7. 1.

74   Wars 5. 1. 8.

75   This was the distribution of forces used by Theoderic the Ostrogoth, who next ruled Italy after Odovacar (Heather (1996), Ch. 7 with refs). Goffart (1980), Ch. 3, again argued that, in both cases, rewards took the form of tax rather than land, but this doesn’t make sense. The whole point of the revolt was that Italy wasn’t generating enough tax revenue. After defeating Odovacar, Theoderic certainly allocated land, while maintaining part of the tax system (Barnish (1986)).

76   Malchus fr. 2.


1    Tax reductions: Hendy (1985), 613–69; cf. more generally on seventh-century transformation, Whittow (1996), Haldon (1990).

2    The most extravagant of recent attempts to minimize the importance of group identities is Amory (1997); cf. Amory (1993). But see, for instance, the responses of Heather (2003) or Innes (forthcoming). Romani are mentioned in law codes from the Visigothic, Burgundian and Frankish kingdoms and, in Cassiodorus’ Variae, from the Ostrogothic.

3    In that year, the Huns launched a huge attack on the Roman Empire, but to the east rather than the west of the Black Sea (p. 154).

4    Roman army of c. 420: p. 247. Tax crisis and loss of Africa: p. 296.

5    Goffart (1980), 35.

6    Carolingian collapse: Reuter (1985), (1990); cf. various essays in Gibson and Nelson (1981). For a more general overview, see Dunbabin (1985), and regional surveys in Hallam (1980). Goffart started life as a Carolingianist and I have often wondered if the processes of Carolingian collapse have not too much influenced his vision of Rome’s fall. The one exception to the ‘internalist’ rule was the Duchy of Normandy, founded by the Viking Rollo, but here the key grant of territory was not made until 911, after the main process of Carolingian collapse had already worked itself out.

7    The arrival of the Christian mission sent to Canterbury by Pope Gregory I in 597 pretty much defines the lower chronological limit of Bede’s detailed knowledge of the Anglo-Saxon past.

8    General surveys: Campbell (1982), Ch. 2; Esmonde Cleary (2002); Higham (1992). The kingdom of Kent perhaps preserved the boundaries of the old Roman civitas of the Cantii, and the same may be true of Lincoln and Anglo-Saxon Lindsey. But most early Anglo-Saxon kingdoms were much smaller than the old Roman civitates and were clearly carved out from them piecemeal: cf. the essays in Bassett (1989).

9    Historians have sometimes fought over whether the end of the Empire should be thought of as destruction or evolution. As so often, the answer to this seeming question-cum-contradiction is to be clear as to what exactly one is talking about in any particular case.

10   Typical of the traditional approach was the title of Frank Walbank’s study of 1969: The Awful Revolution. The wind of change shows up nicely in the title given to the European Science Foundation project on the same subject, The Transformation of the Roman World.

11   In the high Empire, this form of education was geared to producing expert public speakers who would excel on the town council. In the late Empire, classical Latin (and to some extent Greek) became the language of the imperial bureaucracy, the new career structure that replaced the town councils.

12   General pattern: Heather (1994). On Venantius, see George (1992). The relevant evidence from across former Roman Europe is surveyed in Riché (1976).

13   Brown (1996) explores many of the changes.

14   The eastern Church: Hussey (1990); cf. revealing particular studies such as Alexander (1958).

15   Gibbon (1897), 160ff. (quotation from p. 161).

16   Baynes (1943); Jones (1964), Ch. 25.

17   Overtaxation, in Jones’s view, was largely attributable to the need to support a large enough army to confront the barbarians and Persia; so even this, if indirectly, was due to the barbarians – although he did also identify ‘idle mouths’ in the new imperial bureaucracy (rather than, as Gibbon, in the Church) as a further source of trouble (Jones, 1964, Ch. 25).

18   Tervingi and Greuthungi: p. 145. Radagaisus: p. 198. Alaric: p. 224.

19   Rhine invaders: p. 198. Burgundians: p. 198.

20   I have not included anything for Anglo-Saxon immigrants into Britain because they did not directly cause the British provinces to drop out of the imperial system.

21   Ammianus 27. 8.

22   The cases of some later nomads, such as the sixth-century Avars, are better documented; they were on the run from the western Turks (see e.g. Pohl (1988)). For an introduction to the Eurasian Steppe nomads, see Sinor (1977); Khazanov (1984).

23   Written and archaeological evidence both suggest that the Germanic-dominated groups affected, such as the Bastarnae, were conquered or fragmented (Shchukin (1989), pt 1, Chs 7–9; pt 2, Chs 7–8).

24   The figures, as we have seen, are little better than guesses, but the Tervingi and Greuthungi may have numbered c. 10,000 warriors apiece, and Radagaisus’ force perhaps double that. The new group weighed in at more like 30,000 combatants. For detail see Heather (1991), pt 2.

25   Roman policy towards the Alamanni (p. 83–4) looks similar to the kind of preemptive action recorded against Frankish groups at Gregory of Tours Histories 2. 9. Subsequent unification under Clovis is recounted by Gregory of Tours at 2. 40–2. He dates it by implication after 507, but there is good reason to think that processes of conquest and unification had gone hand in hand between 482 and 507.

26   See, in more detail, Heather (1991), pt 3. The only new kingdom which we don’t know to have been the product of a major political realignment is that of the Burgundians. This was a second-rank power, which preserved its independence only when it could play off the Franks against the Ostrogoths, and fell to the Franks when Justinian’s conquest of Italy eliminated the latter. Two possibilities present themselves (both feasible, given the sparseness of information available). Either there was no significant fifth-century political realignment behind the creation of Burgundy, which might explain its relative lack of military power; or any alignment was not on the same scale as that which produced the other kingdoms.

27   On the freeman class, see p. 94. Many of the individuals who are known to have separated themselves from the uniting groups were defeated candidates for leadership, such as the Visigoths Modares, Fravittas and Sarus. Some Thracian Goths who stayed in the east rather than follow Theoderic the Ostrogoth to Italy were Bessas and Godisdiclus (Procopius Wars 1. 8. 3).

28   Radagaisus followers: Orosius 7. 37. 13ff. (slavery); Zosimus 5. 35. 5–6 (pogrom). Vandals and Alans: Hydatius Chronicle 67–8. Ostrogoths: Malchus frr. 15 and 18. 1–4 with Heather (1991), Ch. 8.

29   Julian’s treaties: Ammianus 17. 1. 12–13; 17. 10. 3–4, 8–9; 18. 2. 5–6, 19. The evidence for economic contact is collected and analysed by e.g. Hedeager (1978). On the Tervingi, see p. 72ff.

30   Annals 12. 25.

31   Northern Amber Route: Urbanczyk (1997).

32   E.g. Ammianus 16. 12. 17; cf. generally Klöse (1934) on gifts.

33   Ørsnes (1968).

34   Ammianus 30. 3. 7.

35   What I’m offering very briefly here and will develop further in another study (Heather (forthcoming b)) is a centre/periphery model for developments around the fringes of the Roman world. For an introduction to this kind of vision, see Rowlands et al. (1987); Chamption (1989). To my mind, it is crucial to add to such analyses a powerful element of agency: (cf. e.g. Prakash et al. (1994)). Rome’s neighbours were not passive recipients of Roman action and stimuli, but responded dynamically according to their own agendas.

*1   Italics = usurpers (emperors unrecognized in the other half of the Empire). Some minor western usurpers who never extended their power beyond one immediate locality are not included.

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