¹ With ten different words for the waters, Ammianus surpasses in elaborate variation the nine synonyms of mare in the opening of Catullus 64. On Ammianus' descriptive technique, see M. Roberts on the aduentus of Constantius (‘The treatment of narrative in late antique literature: Ammianus Marcellinus (16.10), Rutilius Namatianus, and Paulinus of Pella’, Philologus 132 (1988), 181–95,Google Scholar at 183): ‘the effect is of a series of brilliantly eye-catching but discrete visual impressions, which in part by their very brilliance deter the reader from trying to piece together the individual scenes into a coherently ordered whole.’

² See Section III below.

³ See for example E. A. Thompson, The Historical Work of Ammianus Marcellinus (1947), 13, and J. F. Matthews, The Roman Empire of Ammianus (1989), 14–17. Matthews also (192) has a brief appreciation of the passage in the context of the structure of Book 26. Lepelley, C., ‘Le présage du nouveau désastre de Cannes: la signification du raz de marée du 21 juillet 365 dans l'imaginaire d'Ammien Marcellin’, Kôkalos 36–37 (1990–1991) [1994] 359–74,Google Scholar offers a literary interpretation of the tsunami as presaging Adrianople. Jacques, F. and Bousquet, B., ‘Le raz de marée du 21 juillet 365’, MEFRA 96 (1984), 423–61,CrossRefGoogle Scholar have some pertinent observations at 450–1. The only other discussions of its literary purpose and merits I have found are G. Sabbah, La méthode d'Ammien Marcellin (1978), 555–8, an enigmatic comment from Steele, R. B., ‘Ammianus Marcellinus’, Classical World 16 (1922), 18–24, 27–8,Google Scholar at 20, and before that the unflattering remark of Edward Gibbon, The History of the Decline and Fall of the Roman Empire (1776–88, ed. D. Womersley, 1995), ch. 26, 1.1023 n. I: ‘Such is the bad taste of Ammianus, that it is not easy to distinguish his facts from his metaphors.’

⁴ Bibliography concerned with the realities of the event can be found in E. Guidoboni (with A. Comastri and G. Traina, trans. B. Phillips), Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century (1994) (a supplemented and corrected translation of the catalogue in E. Guidoboni (ed.), I terremoti prima del mille in Italia e nell' area mediterranea (1989), which also has several relevant articles). See also N. N. Ambraseys, C. P. Melville and R. D. Adams, The Seismicity of Egypt, Arabia and the Red Sea: a Historical Review (1994), 22–4; Waldherr, G., ‘Die Geburt der “kosmischen Katastrophe”. Das seismische Großereignis am 21. Juli 365 n. Chr.’, Orbis Terrarum 3 (1997), 169–201;Google ScholarKelletat, D., ‘Geologische Belege katastrophaler Erdkrusten-bewegungen 365 AD im Raum von Kreta’, in Olhausen, E. and Sonnabend, H. (eds), Naturkatastrophen in der antiken Welt: Stuttgarter Kolloquium zur historischen Geographie des Altertums 6, 1996 (1998), 156–61;Google ScholarStiros, S., ‘The AD 365 Crete earthquake and possible seismic clustering during the fourth to sixth centuries AD in the eastern Mediterranean: a review of historical and archaeological data’, Journal of Structural Geology 23 (2001), 545–62;CrossRefGoogle ScholarPrice, S., Higham, T., Nixon, L., and Moody, J., ‘Relative sea-level changes in Crete: reassessment of radiocarbon dates from Sphakia and West Crete’, BSA 97 (2002), 171–200Google Scholar. See these last two items for more comprehensive bibliography.

⁵ For articles discussing the sources and their providentialist tendencies, see especially Jacques and Bousquet, op. cit. (n. 3) and Henry, M., ‘Le témoignage de Libanius et les phénomènes sismiques de IVe siècle de notre ère. Essai d'interpretation’, Phoenix 39 (1985), 36–61CrossRefGoogle Scholar. See also Baudy, G. J., ‘Die Wiederkehr des Typhon. Katastrophen-Topoi in nachjulianischer Rhetorik und Annalistik: zu literarischen Reflexen des 21 Juli 365 n.C.’, JAC 35 (1992), 47–82;Google ScholarMazza, M., ‘Cataclismi e calamità naturali: la documentazione letteraria’, Kôkalos 36–37 (1990–1991) [1994], 307–30;Google Scholar Lepelley, op. cit. (n. 3); Waldherr, op. cit. (n. 4). Providentialism is narrowly defined by Paschoud, F., ‘Justice et providence chez Ammien Marcellin’, Hestiasis. Studi di tarda antichità offerti a Salvatore Calderone vol. I (1986), 139–61,Google Scholar at 139: ‘the conception according to which the fulfilment or non-fulfilment of religious commands has as a rapid, necessary and plainly visible consequence either punishment or reward.’

⁶ The main focus of Section I below.

⁷ Section II below.

⁸ See Lepelley, op. cit. (n. 3).

⁹ Section IV below.

¹⁰ Section III below.

¹¹ See for examples and rebuttals Jacques and Bousquet, op. cit. (n. 3), 424–5, and n. 20 below.

¹² The theory of a Mediterranean earthquake of an extent not just unusual but unimaginable can only be rescued by positing a chain of earthquakes with various epicentres (see Di Vita, A., ‘Sismi, urbanistica e cronologia assoluta: terremoti e urbanistica nelle città di Tripolitania fra il I secolo a. C. e il IV d. C.’, in L'Afrique dans l'Occident romain, Collection de l'École Francaise de Rome 134 (1990), 425–94,Google Scholar at 466–74). However, for this period we have records and archaeological evidence of major and distinct earthquakes in mainland Greece, Sicily, Cyrenaica, and Palestine, as well as three Bithynian earthquakes in A.D. 358, 362, and 368 (Guidoboni, op. cit. (n. 4, 1994), 257–77, nos 147–56).

¹³ See Jacques and Bousquet, op. cit. (n. 3), 441–3; E. Guidoboni, G. Ferrari and C. Margottini, ‘Unachiave di lettura per la sismicità antica: la ricerca deigemelli del terremoto del 365 d. C.’, in Guidoboni, op. cit. (n. 4, 1989), 552–73.

¹⁴ See for example, Pirazzoli, P. A. et al. , ‘Historical environmental changes at Phalasarna Harbor, West Crete’, Geoarchaeology 7 (1992), 371–92;CrossRefGoogle Scholar Kelletat, op. cit. (n. 4); Stiros, op. cit. (n. 4).

¹⁵ Price et al., op. cit. (n. 4), 180–95.

¹⁶ e.g. Guidoboni, op. cit. (n. 4, 1994); Ambraseys et al., op. cit. (n. 4).

¹⁷ Guidoboni, op. cit. (n. 4, 1994), 271. Guidoboni ascribes to it a strength of X–XI on the EMS (European macro-seismic) scale, as great as any other earthquake in antiquity. The force of tsunamis can generally be correlated to the size of the earthquake that generates them. See E. Bryant, Tsunami: The Underrated Hazard (2001), 15.

¹⁸ See Guidoboni, op. cit. (n. 4, 1994), 272–3; Stiros, op. cit. (n. 4), 558–9.

¹⁹ See Bachielli, L., ‘A Cyrenaica earthquake post 364 A.D.: written sources and archaeological evidence’, Annali di Geofisica 38 (1995), 977–82Google Scholar. On mainland Greece, see R. M. Rothaus, Corinth, the First City of Greece: an Urban History of Late Antique Cult and Religion (2000), 18–21; Stiros, op. cit. (n. 4), 559.

²⁰ Earthquake damage in North Africa is detached from the event of A.D. 365 by Lepelley, C., ‘L'Afrique du nord et le prétendu séisme universel du 21 juillet 365’, MEFRA 96 (1984), 463–91;CrossRefGoogle Scholar that at Sounion on Cyprus by Guidoboni, Ferrari and Margottini, op. cit. (n. 13), 570–1. On Sicily, contrast Di Vita, A., ‘La villa di Piazza Armerina e l'arte musiva in Sicilia’, Kôkalos 18–19 (1972–1973), 251–63,Google Scholar at 256–7, with R. J. A. Wilson, Sicily under the Roman Empire (1990), 185–7, who argues for an earthquake in A.D. 363/4 as well as the tsunami of A.D. 365 (cf. E. Guidoboni, A. Muggia and G. Valensise, ‘Aims and methods in territorial archaeology: possible clues to a strong fourth-century A.D. earthquake in the Straits of Messina (southern Italy)’, in W. McGuire et al., The Archaeology of Geological Catastrophe (2000), 45–70).

²¹ This assessment reconciles Ammianus, who puts the earthquake just after dawn, and Theophanes, who puts it in the night. It may be over-interpretation to note that Ammianus was probably resident at Antioch at the time, where sunrise is over twenty minutes earlier than Alexandria, and where a distant earth-quake felt under cover of darkness further west might have been felt shortly after dawn.

²² As well as Ammianus, see e.g. Jerome, Chron. a.366, Comm. in c. xv Isaiae 5, Vita S. Hil. 40, Theophanes AM 5859, the biographer of Athanasius (PG 25.CCX), and George the Monk (560–1).

²³ The earthquake of 8 August 1303 was felt in Venice and did damage in Alexandria and Acre; that of 12 October 1856 was felt from Italy to Lebanon, and that of 27 August 1886 from Italy to Syria and Egypt (Guidoboni, Ferrari and Margottini, op. cit. (n. 13), 556–8).

²⁴ The stories contained in Theophanes and George the Monk (see Section II below) also imply an earthquake of some force in Alexandria.

²⁵ On tsunami dynamics, see Bryant, op. cit. (n. 17), ch. 2.

²⁶ Local geography also leaves Methone prone to tsunami damage, according to Henry, op. cit. (n. 5), 42.

²⁷ For example, in the Chilean earthquake of 1868, a tsunami threw the American warship Wateree and the Peruvian warship America 3 km inland (Bryant, op. cit. (n. 17), 52).

²⁸ For behaviour of tsunamis at coasts see Bryant, op. cit. (n. 17), 32–7.

²⁹ The entry for A.D. 365 was published in about A.D. 370. See R. W. Burgess, The Chronicle of Hydatius and the Consularia Constantinopolitana (1993), 196. Significantly, this earliest explicit narrative mentions only the sea passing its boundaries, not the earthquake.

³⁰ Nova Patrum Bibliotheca VI.1, p. 15, now SC 317, p. 269.

³¹ The first of these passages was adduced by Henry, op. cit. (n. 5), 41 and n. 17. The comparison of Procopius to Typhon is also found at 90a, and at 93a he is compared to a raging torrent.

³² Libanius may make fleeting reference to the tsunami in the middle of a list of other regrettable consequences of Julian's death: (‘fear shook both earth and sea’). But the context is troubling, and it might be possible to explain this phrase metaphorically. On Or. 18.292–3 see below, p. 147.

³³ He describes having seen salt marshes in the eastern Nile Delta in A.D. 399, which had been created by a great tsunami. Two other possible reflections of this earthquake can be found in HA Gallieni duo 5 (an anachronistic fiction) and Zosimus 4.18 (a story displaced by chronological error).

³⁴ Orlandi, T. (ed.), Constantini episcopi urbis Siout Encomia in Athanasium duo (CSCO 349/Copt. 37) (1974), 1.22–5, 27Google Scholar. See Mazza, op. cit. (n. 5), 318–20.

³⁵ See Jacques and Bousquet, op. cit. (n. 3), 456–61 for a list mentioning all sources except Themistius and Bishop Constantine's Encomium (see n. 34). The only error is to see the anonymous biographer of Athanasius as exclusively following Socrates (p. 458).

³⁶ On ancient perceptions of earthquakes, see Traina, G., ‘Terremoti e societ̀a romana: problemi di mentalità e uso delle informazioni’, ASNP 15 (1985), 867–87Google Scholar.

³⁷ See n. 5 for Paschoud's definition of providentialism.

³⁸ Baudy, op. cit. (n. 5), has gone as far as to see the tsunami of A.D. 365 as unalloyed providentialism, a fiction arisen because of its aptness for mythologizing the usurpation of Procopius and the reign of Julian. Baudy's conspiracy theory requires a double forgery of Constantinopolitan and Alexandrian records, and astonishing credulity in the contemporaries Jerome and Ammianus. He is correct, however, about the susceptibility of almost all the literary sources to providentialist readings.

³⁹ See F. Kolb, Herrscherideologie in der Spätantike (2001), 98–9, for the characteristics of a legitimate proclamation.

⁴⁰ (‘And wh o wasn't struck … by that deluge, the surf and the triple wave, begun at night but rendered great in the daylight, when a man hateful to the gods, who'd always lived in the position of scribe, dared from the ink and the quill to put into his mind the domination of the Roman Empire.’) See Baudy, op. cit. (n. 5), 52–8, for a striking denial of this text's reference to the tsunami.

⁴¹ Similarly Jerome's Chronicle postpones it until after the death of Procopius in A.D. 366.

⁴² See Lenski, N., ‘Initium mali Romano imperio:contemporary reactions to the Battle of Adrianople’, TAPA 127 (1997), 129–68,Google Scholar at 150–2.

⁴³ cf. Eusebius, VC 3.3, where Constantine represents the persecutor Licinius as a dragon.

⁴⁴ Sozomen's use of Libanius is noticed by Jacques and Bousquet, op. cit. (n. 3), 449 n. 106, 452 n. 118, and Mazza, op. cit. (n. 5), 318.

⁴⁵ Trans. Norman.

⁴⁶ The passage was dissociated from the great tsunami by Henry, op. cit. (n. 5), and by Jacques and Bousquet, op. cit. (n. 3), 428–36. All other evidence suggests a date at the start of A.D. 365, and Libanius' failure to mention the great tsunami confirms that 21 July is a terminus ante quem (see H.-U. Wiemer, Libanios und Julian (1995), 261–8).

⁴⁷ Henry, op. cit. (n. 5), 44–60, suggests that the earthquakes of Libanius, Or. 18.292 must have taken place before Julian's death, and she is followed by Guidoboni, op. cit. (n. 4, 1994), 259–62. The assumption is that Libanius and others saw earthquakes exclusively as portents. But earthquakes can also function as nature's responses following human events, and these earthquakes come in a list of events subsequent to Julian's death. Jacques and Bousquet, op. cit. (n. 3), 424–37, therefore seem correct to suggest that the earthquakes described by Libanius belong in the years A.D. 362–4.

⁴⁸ See n. 46.

⁴⁹ The passage embarrasses those commentators who perceive him as the father of scientific history (Hornblower, S., A Commentary on Thucydides vol. I (1997)Google Scholar ad loc.). For a convincing demonstration of Ammianus' sharp practice with the eclipse in A.D. 360 (20.3) in order to create an omen, see T. D. Barnes, Ammianus Marcellinus and the Representation of Historical Reality (1998), 102–6.

⁵⁰ Libanius may later use the tsunami in such a way in his speech On Avenging Julian (see n. 32).

⁵¹ Lepelley, op. cit. (n. 20), 363 n. 3.

⁵² One of Ammianus' sources for the digression, Pliny's Natural History (2.192), reveals that this is why thunderstorms so often precede earthquakes.

⁵³ The other sources with an Alexandrian focus are, in chronological order, the Index of Festal Letters of Athanasius, Sozomen, Bishop Constantine's Encomium of Athanasius, Theophanes, and George the Monk.

⁵⁴ A. Bauer and J. Strzygowski, Eine alexandrinische Weltchronik. Text und Miniaturen eines griechischen Papyrus der Sammlung W. Golenischev (1906).

⁵⁵ C. Mango and R. Scott, The Chronicle of Theophanes Confessor (1997), lxxviii–lxxx.

⁵⁶ George's earthquake and tsunami is unquestionably that of A.D. 365, mingled with an earthquake at Germe in Bithynia (Guidoboni, op. cit. (n. 4, 1994), no. 156) which Socrates places in A.D. 368 or 369 and which could be identifiable with the Nicaea earthquake of 11 October A.D. 368. That George misdates the event to the reign of Gratian may be the typical confusion of the Byzantine chronicler when faced with multiple Augusti. By the time of the Germe earthquake, Gratian had been proclaimed Augustus by his father. An alternative source of confusion could be Gratian's consulate of A.D. 366, the year to which Theophanes (like a number of other sources) misplaces the tsunami.

⁵⁷ Bryant, op. cit. (n. 17), passim, e.g. 161–2.

⁵⁸ Theophanes, trans. Mango and Scott: ‘(a) In indiction 8 there was a great earthquake by night throughout the whole world, so that in Alexandria [(b) ships moored to the shore were lifted high up over the top of tall buildings and walls and were carried within [the city] into courtyards and houses], (c) When the water had receded, they remained on dry land, (d) The people fled from the city because of the earthquake but when they saw the ships on the dry land they went up to them to loot their cargoes, (e) But the returning water covered them all.’ George: ‘(a) In these times there occurred a great and very fearsome earthquake to the extent that at Alexandria (c) the sea disappeared for a long time and boats were found lying as if on dry land, (d) And a multitude of people ran to see the unexpected wonder, (e) and when the water turned around and came back further than its accustomed place, 50,000 people were drowned; and some of the ships moored there the waters covered …’

⁵⁹ Rebuffat, R., ‘Cuicul, le 21 juillet 365’, Antiquités Africaines 15 (1980), 309–28,CrossRefGoogle Scholar at 322.

⁶⁰ Apart from the impossibility of Ammianus being available in Byzantium as late as the ninth century, Theophanes would have needed to combine him with another detailed source: plainly any source which had the details in Theophanes would also have those which could be found in Ammianus.

⁶¹ The reading of our main manuscript, V, is cernebantur, and though cernerentur, the reading of Gelenius (G), may represent the lost Hersfeldensis (M), it is as likely to come from Accursius' edition (A). The indicative is supported by S. Blomgren, De sermone Ammiani Marcellini quaestiones variae (1937), 56–7, on the analogy of 21.6.9, where Ammianus has used the indicative in a result clause. In this case, however, a later subordinated verb (suspicerent) is in the subjunctive, and whereas one might be unsurprised at an author forgetting the subordination in a long sentence, one should not expect it to be forgotten and then remembered later in the sentence. Easier to suppose that cernerentur has been corrupted to cernebantur because of the nearby dabatur.

⁶² TLL s.v. do, 1689. 48–81.

⁶³ Ammianus uses ut opinari dabatur and related forms (e.g. ut intellegi datur) in many places: 16.10.n, 19.2.31, 21.14.5, 22.9.11, 24.8.5, 26.1.7, 28.2.8, 30.5.18, 31.13.12. The conventional translation is a weak one: for 26.10.16, ‘as one might well believe’ (Rolfe), ‘wie man glauben darf’ (Seyfarth), ‘come spoteva ritenere’ (Selem), ‘pouvait-on penser’ (Marié), ‘one must suppose’ (Hamilton). But dabatur should imply a donor. At 16.10.11 Constantius' statuesque behaviour and restraint were ‘signs of no mean endurance, one was given to think, which had been bestowed uniquely on him’ (‘patientiae non mediocris indicia, ut existimari dabatur, uni illi concessae’). This refers obliquely but unmistakably to individuals (‘spin doctors’) who interpreted Constantius’ actions. Similarly (30.5.18) the night before his death, Valentinian in a dream ‘saw his absent wife sitting with disordered hair and dressed in mourning clothes: one was given to think (aestimari dabatur) that this was his fortuna about to leave’. One might interpret this as the narrator's hesitant deduction, but a reference to a source is likelier. What the emperor saw on the last night of his life needs a witness. Constantius had a dream which he told his proximi (21.14.2), which was thought to be his genius. Julian on the eve of his death (25.2.3) also saw his fortuna, the genius publicus, leaving him with veiled head, and Ammianus' source for this information is openly if inexactly expressed: ‘ut confessus est proximis’.

⁶⁴ Arida humus occurs in Sallust, Jugurtha twice, in Statius, Thebaid and Apuleius, De mundo. In these places it refers to dry (e.g. desert) earth. Nowhere else in Latin is it used in this sense of dry land as opposed to sea. For the noun arida as a caique of ξηρά see TLL s.v. aridus, 569, 38–56.

⁶⁵ The reference to bodies floating face-up and face-down recalls the true information recorded by Pliny (HN 7.77), that men's corpses tend to float supine, and women's prone.

⁶⁶ As they are by Thompson, op. cit. (n. 3), 29–31, and Fornara, C. W., ‘Julian's Persian expedition in Ammianus and Zosimus’, JHS 111 (1991), 1–15CrossRefGoogle Scholar.

⁶⁷ Traditional Quellenforschung can be exemplified by the proposal of Seeck, Otto (‘Zur Chronologie und Quellenkritik des Ammianus Marcellinus’, Hermes 41 (1906), 481–539)Google Scholar that Ammianus showed the conflation of a ‘Thucydidean’ Roman source and an eastern ‘Annalistic’ source: a proposal admirably debunked by Thompson, op. cit. (n. 3). The firmest current supporter of the application of traditional source criticism to Ammianus can only argue his position with allowance for the historian's independent spirit (F. Paschoud, ‘Valentinien travesti, ou: De la malignité d'Ammien’, in J. den Boeft, D. den Hengst and H. C. Teitler (eds), Cognitio Gestorum (1992), 67–84).

⁶⁸ Looking at Tacitus' adaptation of such texts as the Res gestae diui Augusti, the Senatus consultum de Pisone, and the speech of Claudius, Woodman similarly argues that source criticism should be treated as intertextuality (C. S. Kraus and A. J. Woodman, Latin Historians (1997), 97–102). Whilst fewer inscriptional source texts survive for Ammianus, there are many more panegyrics, letters, invectives, and laws extant that may have influenced him —indeed Ammianus could offer insights into how historians in earlier periods used documents.

⁶⁹ See Kelly, G. A. J., ‘The old Rome and the new:Ammianus Marcellinus' silences on Constantinople’, CQ 53 (2003), 588–607,CrossRefGoogle Scholar at 603–6. See also Sabbah, op. cit. (n. 3), 340–50.

⁷⁰ Ammianus is a prime example for the examination of autopsy in Classical historiography by J. Marincola, Authority and Tradition in Ancient Historiography (1997). Some have found comparison with later Greek authors fruitful: N. J. E. Austin, ‘Autobiography and history: some later Roman historians and their veracity’, in B. Croke and A. M. Emmett (eds), History and Historians in Late Antiquity (1983), 54–65, compares Ammianus' use of autobiography with that of Procopius, and to a lesser extent that of the fragmentary historians Olympiodorus of Thebes and Priscus of Panium.

⁷¹ cf. Eutropius' parenthetical mention of his participation in the same campaign (‘cui expeditioni ego quoque interfui’, 10.16).

⁷² e.g. 14.4.6, 17.4.6, 22.8.1, 23.6.36.

⁷³ See n. 3 above.

⁷⁴ Matthews, op. cit. (n. 3), 17, and Henry, op. cit. (n. 5), 39. Methone was on the sea route from Asia Minor to the West (Henry, 43 n. 22, cites Procopius, Wars 5.7.28, 32).

⁷⁵ Momigliano, A. D., ‘The lonely historian Ammianus Marcellinus’, ASNP n.s. 3.4.4 (1974), 1393–407Google Scholar (reprinted in Sesto contribute alia storiam degli studi classici e del mondo antico (1980), 142–57).

⁷⁶ ‘Addici post cruciabiles poenas uidimus multos’ (29.1.24); ‘namque ut pressius loquar, omnes ea tempestate uelut in Cimmeriis tenebris reptabamus, paria conuiuis Siculi Dionysii pauitantes’ (29.2.4).

⁷⁷ Austin, op. cit. (n. 70), 64.

⁷⁸ I do not speak only of literary sources, though herein lies an explanation of the difference in involvement between (for example) Ammianus' distant autopsy of the Persian campaign of A.D. 363, well covered in many other works (Libanius, Eunapius, Magnus of Carrhae), and the greater vividness of the Silvanus affair (on the frontier and restricted to Cologne).

⁷⁹ 19.1.2,19.1.9,19.43,19.6.7.

⁸⁰ Modern scholarly doubts begin with M. Büdinger, Ammianus Marcellinus und die Eigenart seines Geschichtswerkes (1895), 28, who called the passage a ‘Fantasiestück’. See Paschoud, F., “Se non è vero, è ben trovato”: tradition littéraire et vérité historique chez Ammien Marcellin’, Chiron 19 (1989), 37–54,Google Scholar at 41–53, and works cited there.

⁸¹ ‘Nisi oculorum deficeret acies, ad quinquagesimum usque lapidem quoduis etiam minutissimum apparebat’ (cf. 16.10.14). Contrast Rolfe's translation: ‘unless one's eyesight was impaired, even the smallest object was visible at a distance of fifty miles.’

⁸² ‘Quo usque nobis Doriscum Thraciae oppidum et agminatim intra consaepta exercitus recensitos Graecia fabulosa narrabis? cum nos cauti uel, utuerius dixerim, timidi nihil exaggeremus praeter eaquae fidei testimonia neque dubia neque incerta monstrarunt.’

⁸³ Ammianus tends to associate the fabulous with Homer (22.16.10. 23.6.53, 27.4.3) and Herodotus (18.6.23, 23.6.7, and perhaps also 31.13.19, though more obviously a reference to Homer).

⁸⁴ Paschoud, op. cit. (n. 80), 39, claims that narratives of personal experiences are more prone to be distortions of the truth. F. Paschoud, ‘Utrum in Amidae obsidione narranda Ammianus veritati magis an argutis commends studuerit’, in Loquela vivida. Donum Natalicium Nicolao Sallmann (1999), 81–6, pursues the argument.

⁸⁵ On arrangement in hexads, see Barnes, op. cit. (n. 49), 23–6.

⁸⁶ On these annalistic notices, see Barnes, op. cit. (n. 49), 92–3.

⁸⁷ For a discussion see Sabbah, op. cit. (n. 3), 557–8.

⁸⁸ The first person is admittedly used for an organizational purpose at the end of Book 28 (6.30), to refer forward. At the end of some books, avoidance of the first person may have been purposeful (e.g. Book 18).

⁸⁹ Matthews, op. cit. (n. 3), 192.

⁹⁰ Hdt. 8.129; Thuc. 3.89. See Smid, T. C., ‘Tsunamis in Greek Literature’, G & R 17 (1970), 100–4Google Scholar. For doubts as to Ammianus' knowledge of Herodotus and Thucydides, see Fornara, C. W., ‘Studies in Ammianus Marcellinus II: Ammianus' use and knowledge of Greek and Latin literature’, Historia 41 (1992), 420–38,Google Scholar and, on Herodotus, H. C. Teitler, ‘Visavel lecta? Ammianus on Persia and the Persians’, in J. W. Drijvers and E. D. Hunt (eds), The Late Roman World and its Historian: Interpreting Ammianus Marcellinus (1999), 216–23.

⁹¹ P. de Jonge, Philological and Historical Commentary on Ammianus Marcellinus Book XVII (1977), ad loc.

⁹² See OLD s.v. insido, ib. Ammianus elsewhere shows a fondness for the irregular perfect form insidi rather than insedi, though the choice of the form at this point has the advantage of making perfectly clear that the verb is insido not insideo.

⁹³ Trans. Fairclough/ Goold.

⁹⁴ For other uses of this commonplace see e.g. Nisbet, R. G. M. and Hubbard, M., A Commentary on Horace Odes Book I (1970)Google Scholar, ad Carm. 1.2.9. In that poem Horace, like Ammianus here, ‘describes his not as a quaint possibility but as something that has actually happened’.

⁹⁵ See the excellent but brief discussion of the tsunami as symbol in Sabbah, op. cit. (n. 3), 555–8, which has strongly influenced this present analysis.

⁹⁶ Sabbah, op. cit. (n. 3), 556.

⁹⁷ Ammianus also alludes to Ovid, Tristia 5.12.27–8, ‘correcting’ an allusion to this same passage of Vergil: ‘uertitur in teneram cariem rimisque dehiscit / siqua diu solitis cumba uacauit aquis’ (‘if any barque has long been out of its accustomed waters, it turns to soft decay and gapes with cracks’). Ovid's allusion to Vergil recalls the etymology of fatisco from fatim hisco (see Servius ad Aen. 1.123).

⁹⁸ That is to say, beyond its literal meaning, that the boat had been rotting for a long time, diuturnus also implies the antiquity of the image of the yawning boat. Meta-textual markers of this sort are more commonplace in discussions of Latin poetry (see the influential S. Hinds, Allusion and Intertext (1998), ch.1).

⁹⁹ To be distinguished from other rumours which Ammianus does not explicitly deny, that Julian had given Procopius a purple cloak, or advised him before the Persian expedition to claim the Empire if the situation arose (23.3.2, 26.6.2).

¹⁰⁰ Linguistic similarities suggest that both passages, though undeniably cliché, allude to Cicero's Pro Sestio: cf. 20 ‘clauum tanti imperi tenere et gubernacula rei publicae tractare’ (‘to hold the helm of so great an empire and guide the rudder of the republic’), an allusion noted by Lindenbrog ad 26.1.5. Cf. also 46 ‘hanc rei publicae nauem, ereptis senatui gubernaculis fluitantem in alto’ (‘this ship of the republic, tossing about on the deep with the helm torn from the senate’), a parallel identified by H. Michael, De Ammiani Marcellini studiis Ciceronianis (1874), 27.

¹⁰¹ ‘Et cum sciamus adeo experimenta quosdam referre improuidos, ut bella interdum uicti et naufragi repetant maria, et ad difficultates redeant quibus succubuere saepissime, sunt qui reprehendant paria repetisse principem ubique uictorem.’

¹⁰² On Ammianus' denigration of Jovian see P. Heather, ‘Ammianus on Jovian: history and literature’, in Drijvers and Hunt, op. cit. (n. 90), 105–16 and Lenski, N., ‘The election of Jovian and the role of the late imperial guards’, Klio 82 (2000), 492–515CrossRefGoogle Scholar. For identification and interpretation of the allusion to Plato, see Barnes, op. cit. (n. 49), 141.

¹⁰³ For contemporary use of the image of the helmsman see e.g. Claudian, IV Cos. 59–62 (quoted in n. 112) and 419–27.

¹⁰⁴ Sabbah, op. cit. (n. 3), 556. He describes Julian as ‘fidèle au dieu solaire’, presumably implying a (some what fanciful) connection to the sun's rays touching the seabed.

¹⁰⁵ See n. 32.

¹⁰⁶ Lepelley, op. cit. (n. 3), 364–6.

¹⁰⁷ cf. Amm. 31.5.12, where inundare is used of the Teutones and Cimbri, a republican comparandum for Adrianople. Pacatus (3.3) describes the situation at the start of Theodosius' reign: ‘barbaris nationibus Romano nomini quasi quodam diluuio superfusis’, and Claudian speaks of the same period in IV Cos 49–51: ‘cum barbaries … mixto turbine gentes iamdeserta suos in nos transfunderet Arctos’. Nautical metaphors can also be found in Themistius (see below nn. 111–13).

¹⁰⁸ Lepelley, op. cit. (n. 3), 366–7, lists signs warning of Julian's demise.

¹⁰⁹ ‘… quorum asperitate post multiplices pugnarum aerumnas saepe res Romana uexata postremoomnem amisit exercitum cum rectore.’

¹¹⁰ Marié, M.-A., Ammien Marcellin Histoire Tome V (1984), 28–9Google Scholar and ad loc.

¹¹¹ (trans. Heather and Moncur).

¹¹² Nulla relicta foret Romani nominis umbra ni pater ille tuus iamiam ruitura subisset pondera turbatamque ratem certaque leuasset naufragium commune manu. This passage follows swiftly on from that quoted in n. 107 above, which implied that the barbarians were a flood. For another ship of state metaphor see In Ruf. 1.275–7.

¹¹³ K. Rosen notes an interesting parallel between Themistius, Or. 15.197a and Amm. 31.5.14 (see ‘Wege und Irrwege der römischen Gothenpolitik in Ammians 31. Buch’, in den Boeft et al., op. cit. (n. 67), 85–90, at 86).

¹¹⁴ See e.g. 26.10.8: ‘nec similes eius nee suppares.’

¹¹⁵ See Blockley, R. C., ‘Ammianus Marcellinus on the Battle of Strasburg: art and analysis in the History’, Phoenix 31 (1977), 218–31,CrossRefGoogle Scholar and ch. 8 of my forthcoming book, Autopsy and the Art of Allusion in Ammianus Marcellinus. Strasbourg ended Book 16, which would have been exactly in the middle of the original thirty-one books of the Res Gestae.

¹¹⁶ Tomlin, R. S. O., ‘Ammianus Marcellinus, 26.4.5–6’, CQ 29 (1979), 470–8CrossRefGoogle Scholar.

¹¹⁷ One may compare Libanius (and contemplate the possibility of allusion). He juxtaposes invasions or threatened invasions by Scythians (Goths), Sarmatians, and Celts,‘and every barbarian tribe that rejoiced to live under treaty’ (Or. 18.290) with the earthquakes that portended and responded to Julian's death (292–3). The Alamann (Libanius' ‘Celtic’) attacks, from early in A.D. 365, have already been referred to at 26.5.7. The fact that the book begins at this point can therefore be seen as carefully engineered, and typical of Ammianus' altered use of chronology in the last hexad.

¹¹⁸ ‘Et postquam ex occidua plaga digressus est, etquoad fuit in terris, quieuere nationes omnes immobiles ac si quodam caduceo leniente mundano.’

¹¹⁹ Classen, C. J., ‘Greek and Roman in Ammianus Marcellinus' History’, Museum Africum 1 (1972), 39–47Google Scholar.

¹²⁰ Mommsen, T., ‘Ammians Geographica’, Hermes 16 (1881), 602–36Google Scholar.

¹²¹ See Sabbah, op. cit. (n. 3), 556.

¹²² See n. 89.

¹²³ It has been said of mise en abyme that ‘there is a two-way movement …, from the work as a whole to the section within it and back from that section to the reading of the work as a whole, in a hermeneutic circle that eventually more-or-less settles into equilibrium or the denial of equilibrium’. D. Fowler, ‘Epic in the middle of the wood: mise en abyme in the Nisus and Euryalus episode’, in A. Sharrock and H. Morales (eds), Intratextuality: Greek and Roman Textual Relations (2000), 89–113, at 109.

¹²⁴ The traditional view that Books 26–31 were written later is found for example in R. Syme, Ammianus and the Historia Augusta (1968), 9–16, and most recently restated by Sabbah, G., ‘Ammien Marcellin, Libanius, Antioche et la date des derniers livres des Res gestae’, Cassiodorus 3 (1997), 89–116Google Scholar. My owndiffident suggestion would carry more conviction if the representation of a unified work culminating in Adrianople occurred in the ‘earlier’ part of the work (14–25). More forceful arguments have been and will be made for the point of view that Ammianus conceived and wrote as a unity a history from Nerva to Valens. See for example Matthews, op. cit, (n. 3), 17–27; C. W. Fornara, ‘The prefaces of Ammianus Marcellinus’, in M. Griffith and D. Mastronarde (eds), Cabinet of the Muses: Studies in Honor of T. G. Rosenmeyer (1990), 163–72.

¹²⁵ Paschoud, op. cit. (n. 5), 158.

¹²⁶ Barnes, op. cit. (n. 49), 87. Barnes' argument against the likes of Matthews, op. cit. (n. 3), that Ammianus was a militant pagan, who consistently but subtly attacked Christianity and its adherents, seems to me capable of correction on minor details, but fundamentally right. Barnes' reception has been mixed — which testifies to Ammianus' success.

¹²⁷ R. Syme, review of Demandt, A., Zeitkritik und Geschichtsbild im Werk Ammians (1965), JRS 58 (1968), 215–18,Google Scholar at 215.