¹ The false identification is made explicit at iii. 6. I: Σάξōνεσ…ϰōνάδōνσ μōῑραν σΦῶν ὄντασ … έκπἐμπōνι.

² For the presentation of this view, see the note by Paschoud to Zosimus iii. 6 (Paschoud, F. (ed.), Zosime: Histoire nouvelle, torn. 2, i (Paris, 1979), 76–7).Google Scholar

³ See Ireland's, R. chapter, ‘Transmission of the text’ in Hassall, M. W. C. and Ireland, R. (eds.), De rebus bellicis (Oxford, 1979), especially 54–5.Google Scholar

⁴ If this hypothesis is granted, a reason for the non-appearance of these units in the Notitia Dignitatum may be put forward. St Ambrose (Ep. 40, 23 = PL 16, coll. 1109–10) records that Saxon troops played a leading part in the emperor Theodosius’ victories over Magnus Maximus in the Balkans in 388; and Zosimus (iv. 30) describes the transfer of a part of the Egyptian army to the Balkans in 379 or early 380. On their arrival (iv. 31. 1), these men were assigned to units already there, and the Roman system of documentation seems to have broken down : μηδὲ άπōγραΦῆσ ἔτι τῶν ἐν τōῑσ στρατιωτικōῑσ ἀριθμōῑσ ἀναφερōμένων Φνλαττōμένησ. If Saxon units had formed part of the Egyptian garrison at an earlier date, they may be assumed (with the exception of the unit postulated for Not. Dig., Or., xxxi. 62) to have participated in this movement of troops and to have lost their identity in the resulting administrative confusion.

⁵ Hoffmann, D., Das spätrömische Bewegungsheer und die Notitia Dignitatum (Düsseldorf, 1969–1970), I, 140 and IIGoogle Scholar, 48, where the author cites Skeat, T. C. (ed.), Papyri from Panopolis in the Chester Beatty Library, Dublin (Dublin, 1964), 2, 11. 292 ff.Google Scholar

⁶ Skeat, , op. cit. (note 5), 151.Google Scholar

⁷ Ammianus Marcellinus, with an English translation by Rolfe, J. C. (Cambridge, Mass., 1935–1940). This translation appears on page 161 of vol. 3.Google Scholar

⁸ This, and the observations on Ammianus' style which follow, have been made possible by consultation of Archbold's, G. J. D. E.microfiche, Concordance to the History of Ammianus Marcellinus (Toronto, 1981).Google Scholar

⁹ ‘Huic pugnae Nevitta, postea consul, equestris praepositus turmae, et adfuisse et fortiter fecisse firmatur’ (xvii. 6. 3).

¹⁰ ‘Verum articulorum dolore Frigerido praepedito, vel certe (ut obtrectatores finxere malivoli), morbum causante, ne ferventibus proeliis interesset’ (xxxi. 7. 5).

¹¹ When Ammianus summarizes this campaign at xxx. 7. 8 his sarcasm has weakened: the Saxons were defeated ‘malefido quidem sed utili commento’.

¹² Clark, C. U. (ed.), Ammiani Marcellini rerum gestarum libri qui supersunt (Berolini, 1910–1915).Google Scholar

¹³ See Clark, C. U., The text tradition of Ammianus Marcellinus (New Haven, Conn., 1904), 10 and passimGoogle Scholar. Clark, however, thought that M and V were copied from a common archetype (cf. also the stemma printed on p. v of his edition of Ammianus). This was refuted in 1936: see Seyfarth, W., Klio xlviii (1967), 214.Google Scholar

¹⁴ These sheets contain four passages: xxiii. 6. 37-xxiii. 6. 45; xxviii. 4. 21-xxviii. 5. 2; xxviii. 5. u-xxviii. 6. 5; and xxx. 2. 5-xxx. 4. 2.

¹⁵ For Ghelen's, modus operandiGoogle Scholar, see Ireland, , op. cit. (note 3), especially 48.Google Scholar

¹⁶ See xxi. 5. 1; xxix. 6. 14; xxxi. 3. 6; and xxxi. 7. 6.

¹⁷ Ghelen may have been influenced at this point by the unsensational description of the Saxons at xxx. 7. 8 as ‘semper quolibet inexplorato ruentes’, or by the maritime connotations of the description of the Maratocupreni as breaking out ‘quocumque ventus duxerat’, which is intended purely metaphorically.

¹⁸ Classical Quarterly, n.s. xxix (1979), 470–8.Google Scholar

¹⁹ See vol. 3, p. 52 of his edition (op. cit., note 7).

²⁰ It is regrettable that, like Rolfe, A. R. Birley has failed to see the significance of xxvii. 8. 4. In his entry for Theodosius in The Fasti of Roman Britain (Oxford, 1981), 333–4, he quotes extensively from this chapter of Ammianus; but section 4 is represented by three dots. The omission of this section makes a proper understanding of this passage impossible.Google Scholar

²¹ On this, see Mommsen, T., Hermes xvi (1881), 602–36.Google Scholar

²² It is wrong to assume that this digression would have proved to be a mine of information about the state of Britain in the fourth century. His description of the ‘situs Brittaniae’ would probably have been derived from Ptolemy; the behaviour of the ocean mentioned in section 4, as well as the peculiarities of the Straits of Dover alluded to in section 6, would have come from Timagenes of Alexandria, whose Periplus, written towards the end of the first century B.C., would probably have extended as far as the mouth of the Rhine. See Mommsen, , op. cit. (note 21), 612–27.Google Scholar

²³ The former expression is discussed by Tomlin, , op. cit. (note 18). But he thinks (p. 470, n. 4) that the ‘eo tempore’ of xxvii. 8. 5 refers to a precise moment in time.Google Scholar

²⁴ Clark's textual apparatus to this passage makes this clear.

²⁵ xxviii. 5. 9, the Burgundians; xxix. 5. 26, the Mazices; xxxi. 16. 3, the Huns and the Alans, together.

²⁶ xiv. 10. 4; xvi. 12. 23; xvii. 8. 1; xxii. 13. 1; xxiii. 2. 3; xxvi. 5. 7. The main exception occurs at xxviii. 1. 53. The word is used alone at xxix. 1. 38 and xxiii. 1. 6; but in the latter instance the correct reading should probably be ‘scaevum’.

²⁷ Zosimus iii. 3. I:τōὐσ δὲ βαρβάρōϒσ … μέχρι σχεδὸτῶν πρὸσ θαλάττŋ πόλεων διελθντασ.

²⁸ Zosimu s iii. 6. 2: φράγκων ὸμὸρων αὐτōῑΣ ὄντων. Cf. also Julian, Oration i, 34D, where the Franks and Saxons are stated to co-operate κατὰ τὸ ξϒγγενὲχ.

²⁹ Zosimus iii. I. 1: ϰωνστάντιōχ … θεώμενōχ…φράγκōϒχ μὲν και Ἀλαμαννōὐχ και σάξōναχ άδη τεσσαράκōντα πάλειχ … κατειληφἈταχ.

³⁰ Tomlin, , op. cit. (note 18), draws attention to the unsatisfactory state of the manuscript tradition (p. 474, n. 28). But he does not develop this line of thought.Google Scholar

³¹ The details may be found in Clark. The poor quality of Castellus' edition is emphasized in his ‘Text tradition of Ammianu s Marcellinus’ (cf. note 13), pp. 10 and 67. In his preface to his edition he dubs it ‘editio mendosissimd’.

³² In fact, ‘Scotti’ was inserted above ‘secuti’ in V in another hand. But the status of this correction cannot be authenticated.

³³ This argument, when taken in conjunction with the preceding one, completes the elimination of the Scots from Ammianus' narrative. Pan. Lot. xii. 5. 2 will be discussed below; but it should be pointed out here that the references in Claudian (De Hi cons. Honorii, 55; De iv cons. Honorii, 33) which purport to show a ‘Scottish’ involvement in fourth-century British history are too deeply embedded in passages of poetical exaggeration to be allowed to count as historical evidence. It follows that there is no literary justification for a search, by archaeologists, for evidence of Irish settlement in Wales and the south-west as early as the mid-fourth century.

³⁴ A barbarian invasion is, however, mentioned in the chapter summaries to xxvii. 8 and xxviii. 3. But these were not written by Ammianus; they first appear in the seventeenth-century printed editions, and show only that, as a result of a hasty reading of the text, misunderstandings had arisen. Rolfe's practice, in the Loeb edition, of printing these summaries at the beginning of each chapter (instead of at the beginning of each book, as earlier editors had done) creates a false impression. They are not evidence for fourth-century Britain.

³⁵ This view is elaborated above, p. 171.

³⁶ Even descriptions of what appear to be genuine actions cannot be relied upon. At xxviii. 3. 1–2, Theodosius is described as ‘nihil gregariis imperans, cuius non ipse primitias, alacri capesseret mente. Hocque genere cum strenui militis munia et praeclari ducis curas expleret…. This is suspiciously reminiscent of the description of Julian at xvii. 1. 2 as ‘omnis operae conturmalem, auctoritate magnificum ditcem, plus laboris indicere sibi quam militi, sicut perspicue contigit, adsuetutri’; and, as Ammianus admits (xvi. 1. 3), he found it difficult, in speaking of Julian, to avoid straying into ‘laudativa materia’. Similarly, the reference at xxviii. 3. 9 to the escort which he received ‘ad usque freturn’ must be regarded as pure fiction.

³⁷ op. cit. (note 20), 338, n. 12. A garbled reference by Zosimus at iv. 35. 3 shows that his source, Eunapius, contained a fuller account of Theodosius' British expedition. Zosimus chose only the Valentinus incident for inclusion in his own work.

³⁸ Ireland, , op. cit. (note 3), 103–4, 106, 107–8.Google Scholar

³⁹ This supposition is supported by the numerous occasions on which the early mediaeval copyist failed to understand the text he had in front of him, and produced a garbled version in the captions.

⁴⁰ See his contribution, ‘What was the Notitia Dignitatum for?’, in Goodburn, R. and Bartholomew, P. (eds.), Aspects of the Notitia Dignitatum (Oxford, 1976), 1–9.Google Scholar

⁴¹ This view obviously needs to be argued in detail elsewhere.

⁴² It must be pointed out that no reliance at all can be placed on the numeral ‘V at iii. 32.

⁴³ At this point, V is hopelessly corrupt. ‘Arcanos’ is an emendation of ‘areanos’; but since no other ancient writer mentions the ‘Arcani’, their restoration here is quite insecure. It is unwise to attempt to make deductions from the apparent form of their name to their function. It is also unsatisfactory to base speculations about the duties of these troops upon the ‘areanos’ of the manuscripts, since this form occurs nowhere else.

⁴⁴ Breeze, D. J. and Dobson, B., Hadrian's Wall (London, 1976), 221–6.Google Scholar

⁴⁵ The text of Ammianus xxviii. 5. 1 provides a n illustration of the dubious authenticity of ‘limes’. According to Clark and the other modern editors, the object of the Saxon attack on north-east Gaul in 370 was ‘Romanum limitem’. But the reading of M (which survives at this point) is ‘Romanum militem’; and M is followed by V. ‘Limitem’ appears only as a correction in an inferior fifteenth-century manuscript, and in Ghelen. This indicates the readiness with which Renaissance scholars thought of ‘limites’ in the context of late Roman military operations; it also shows how unscrupulous Ghelen could be in setting aside the testimony of M (when he felt so inclined) in favour of a conjecture of his own. The decision of the modern editors to accept ‘limitem’ instead of the better attested ‘militem’ must appear distinctly questionable.

⁴⁶ Antiquity xxxv (1961), 318–9.Google Scholar

⁴⁷ The existence of such a proletariat is one outcome of the development which Reece has sketched for Romano-British towns in the fourth century. He suggested (World Archaeology xii (1980), 77–92, especially 86–7) that, in the later Roman period, the successful and cultured members of the upper classes moved out of the towns into the countryside; conversely, the less well-off moved away from the countryside into the towns. The latter would have been the impoverished residue of the British Iron Age tribes, or Ammianus' ‘gentes’. This move can be detected archaeologically; for in the later fourth century the quality of occupation in the towns changes dramatically for the worse. They were not pleasant places in which to live, and the occurrence of urban disorders would seem to be only too probable.Google Scholar

⁴⁸ Goffart, W., Barbarians and Romans, a.d. 418–584 (Princeton, 1980), 42–3.Google Scholar

⁴⁹ Seeck, O., (ed.), Notitia Dignitatum (Berolini, 1876), 180–1. The fact that ‘Rutupis’ (at line 19) is printed in italics indicates that it has been supplied by the editor. The text of this chapter of the Notitia DignitatumGoogle Scholar, as it appears in the Oxford manuscript, is illustrated in Mothersole, J., The Saxon Shore (London, 1924), plate facing p. 18; ‘Rutupis’ is plainly absent. The plate facing p. 19 shows the picture from the head of the chapter, with ‘Rutupis’ added among the captions.Google Scholar

⁵⁰ Boon, G. C., Isca, 3rd ed. (Cardiff, 1972), 62–4.Google Scholar

⁵¹ These occasions are: xiv. 7. 2; xiv. 10. 4; xv. 5. 29; xvi. 4. 4; xxii. 14. 1; xxiv. 3. 14; xxiv. 8. 2; and xxxi. 8. 4. ‘Inopia’ is qualified by a noun meaning food at xiv. 6. 19; xviii. 7. 8; xix. 10. I; xxvi. 3. 6; xxvi. 8. 3; xxvii. 5. 7; xxviii. 1. 17; xxx. 6. 2; and xxxi. 4. 11; and by a noun meaning drink at xiv. 6.1; xv. 7. 3; and xxii. 13. 4. At xxv. 7. 14, food and drink occur together (‘potus inopia et cibi’). Only five passages show ‘inopia’ qualified by a noun with a different meaning: xvii. 7. 5 (auxiliorum); xviii. 1. 4 (documentorum); xxvii. 6. 2 (dignorum); xxx. I. 8 (navium); and xxx. 4. 16 (allegationum).

⁵² Berchem, D. van, Memoires de la Société nationale des antiquaires de France lxxx (1937), 193–4.Google Scholar

⁵³ Cf. Zosimus ii. 33–5: τό πλέōν τῶν σπήσ;εων μέρōχ ɛἰσ κέρρσ;τōῦ στρατηγōῦ καὶ τῶν ύπηρετōνμένων χτōύτῳ ωρετν. Codex Theodosianus vii. 4. 3, issued in 357, records that the Count of Africa had wrongfully appropriated the food supply; in the same year, Barbatio, magister peditum, burnt part of the food supply of the army fighting against the Alamanni, possibly to cause embarrassment to Julian (Ammianus xvi. 11. 12–13).

⁵⁴ In 360, the praetorian prefect Florentius tried to cause difficulties for Julian, who was at Paris, by failing to send the annona from Vienne (Ammianus xx. 4. 6 and xx. 4. 9). Earlier, in 354, food supplies were held back from the troops at Cabyllona, in order that the life of the praetorian prefect Rufinus might be put at risk (Ammianus xiv. 10. 3–5).

⁵⁵ One interesting possibility may be mentioned here. Codex Theodosianus xi. 1. 15, dated to 19th May, 366, is addressed to Probus, praetorian prefect of the Gauls. If he can be identified with the Petronius Probus whose avarice is contemptuously described by Ammianus at xxvii. 11. 3 and xxx. 5. 5–6, the disappearance of the corn-supply for the British troops, arranged jointly with his deputy, the Vicar of the Britains, will offer no cause for surprise. The misdeeds of Petronius Probus were very familiar to Ammianus' readers. His references to them are ironical and very ambiguous. If his readers knew that Probus was indirectly involved in the British events of 367, the lack of clarity with which Ammianus begins his description of them may be readily understood.

⁵⁶ L'Année épigraphique (1931), 53Google Scholar. Cf. also Egger, R., Byzantion v (1929–1930), 9–32Google Scholar ( = his Römische Antike und frühes Christentum, Band 1 (Klagenfurt, 1962), 126–43).Google Scholar

⁵⁷ Reference has been made to the edition of Galletier, E., Panégyriques latins, tome 3 (Paris, 1955).Google Scholar

⁵⁸ Demandt, A., Hermes c (1972), 91–4 and 110–1. Theodosius was now magister equitum in the west, and held this post until his death.Google Scholar

⁵⁹ It is presumably this campaign to which allusion is made in the inscription from near Stobi with the phrase χάρμα μὲγα Δαρδάνων. There is no justification for the alteration of Δαρδάνων to βριττάνων, put forward by W. Ensslin (RE 5A, col. 1939), and printed by Birley, , op. cit. (note 20), 336.Google Scholar

⁶⁰ In quoting this passage, Birley, (op. cit. (note 20), 333) omits the second sentence, dealing with the Alaman and the Moor, from the third pair of contrasts. The effect of this is highly misleading, since it is implied that the Scottish victories belong with the Saxon, and that both, apparently, should be assigned to Britain.Google Scholar

⁶¹ The reading of the manuscripts is ‘Scotturri’ (not ‘Scotum’), which would make a corruption from Gothum rather more probable. At xii. 11. 4, the manuscripts read Cothus; but the occurrence of the words ‘Hunus’ and ‘Halanus’ in close connection with ‘Cothus’ make the emendation to Gothus certain. It would seem that in both cases the initial ‘G’ became corrupted to ‘C’ at an early stage in the transmission of the text. The ‘S’ at xii. 5. 2 is perhaps an erroneous repetition of the final's’ in the preceding word, ‘suas’.

⁶² This can be seen clearly in Galletier's textual apparatus. See also the introduction to tome 1 of his edition of the Panégyriques latins (Paris, 1949), where the stemma is printed on p. lv.Google Scholar

⁶³ Anglo-Saxon archaeologists, in their search for prototypes for the earliest pottery of the invasion period, have concentrated their efforts on the area between the Ems and the Weser. They have been induced to do so by the testimony of Bede. But if the interpretation of the ancient evidence offered here is correct, their investigations should be centred on an area at least 150 miles to the south-west.

⁶⁴ This bridge, which linked the right bank of the Tiber to the Isola Tiberina, has now been replaced by the Ponte Cestio. The inscription was inserted into the parapet of the new bridge. (Coarelli, F., Guida archeologica di Roma (1974), 313).Google Scholar

⁶⁵ Their geographical overlap is illustrated by the entry in Jerome's Chronicle s.a. 373: ‘Saxones caesi Deusone in regione Francorutri’. Orosius, drawing on this entry, places their defeat ‘in ipsis Francorum finibus’ (vii. 32. 10). This campaign is slightly later.

⁶⁶ Tomlin, R., Britannia v (1974), 303–9CrossRefGoogle Scholar. The attempted refutation of Tomlin's arguments by Blockley, R. C. (Britannia xi (1980), 223–5) is unconvincing.CrossRefGoogle Scholar

⁶⁷ It can hardly have been a lasting success. The invasion of the Saxons described by Ammianus at xxviii. 5. 1–7 occurred in the very next year.

⁶⁸ In fact, this is by no means certain: see Clark's textual apparatus to xxvii. 8. 1.

⁶⁹ History lxiv (1979), 259.Google Scholar

⁷⁰ It would, of course, be mistaken to argue, from the demise of the comes maritimi tractus, that the Saxons ‘must’ have played a part in the British events of 367.