¹ Jacoby, FGrHist 225 (Magnus of Carrhae), Komm. 633f., Chalmers, W. R., CQ x (1960), 152ff.CrossRefGoogle Scholar, Dillemann, L., Syria xxxviii (1961), 115–135Google Scholar, Paschoud, F., Zosime (Paris 1971) i pp. xlii–livGoogle Scholar (Cf. ii¹ [1979], xii–xix), and Rosen, K., Ammianus Marcellino (Darmstadt 1982), 53, 66fGoogle Scholar., review the earlier work. Some scholars consider that the similarities have wider compass, e.g., Barnes, T. D., The Sources of the Historia Augusta, Latomus clv (Brussels 1978) 117ffGoogle Scholar. Elliott, T. G., Ammianus Marcellinus and fourth century history (Sarasota and Toronto 1983) 224ff.Google Scholar, Baker, A., Eunapius and Zosimus: problems of chronology and composition (Diss. Providence 1987) 20f.Google Scholar, n.5. I wish to thank an anonymous reader for this journal for his careful and thoughtful criticism.

² See below p.4.

³ Syria xxxviii (1961) 125–131.

⁴ De ratione quae intercedat inter Zosimi et Ammiani de bello a Iuliano imperatore cum Persis gesto relationes (Diss. Bonn 1870).

⁵ Sudhaus' demonstration of a literary affiliation between Zosimus and Ammianus was accepted by Mendelssohn, Ludwig, Zosimus (Leipzig 1887), xxxixGoogle Scholar, though he explained it differently (see above). Klein, W., Klio Beiheft xiii (1914) 42Google Scholar, termed it an ‘ausgezeichnete Vergleichung’; Klotz, A., RhM lxxi (1916) 461Google Scholar, took it for granted as self-evident. Thompson, E. A., The historical work of Ammianus Marcellinus (Cambridge 1947)Google Scholar, though he emphasized disagreement between Ammianus and Zosimus (30), did not dispute the presence of the similarities (31,134, 137) citing Mendelssohn. Subsequent writers mention Sudhaus or Mendelssohn singly or together, while Paschoud, who called Sudhaus' dissertation an ‘étude … pleine de mérite’ in 1971 (Zosime i, p.xlii) actually reverted to Sudhaus' own theory of the relationship eight years later in ii¹ xii–xix (see n.8 below). On Sudhaus, see further below nn.29, 39.

⁶ PLRE I 653f.

⁷ Zosimus, lii–xlvii; for Magnus, see Jacoby, FGrHist 225.

⁸ Jacoby (Komm. 634) followed Mendelssohn, Klein (with reservations) and Klotz. Brok, M. F. A., De perzische Expeditie van Keizer Julianus volgens Ammianus Marcellinus (Diss. Groningen 1959) 17Google Scholar, assumes a common source but is indefinite about its identity or nature. The major objection to Oribasius as the putative source of both Eunapius and Ammianus is the fact that Eunapius singled out Oribasius' memoir as a work specifically written for his private use (see below p.O). Paschoud (ii¹ xviii) seeks to evade the difficulty by suggesting that Ammianus and Oribasius met together (with Libanius) at Antioch in autumn 363, thus supplying an opportunity for Ammianus to consult ‘the journal’ of Oribasius, a journal which later, in ‘amplified’ form, was transmitted to Eunapius. But even if we believe that ‘the journal’ existed, it is arbitrary to assume (1) that its existence was a matter of common knowledge (so that others might wish to consult it); (2) that it had attained sufficient elaboration so as to serve Ammianus in the manner alleged (which turns on detail and alleged verbal reminiscences). Furthermore, even if we assume (against all likelihood) that Ammianus was already engaged in writing his own history of that expedition, we have no warrant further to assume that he enjoyed access to Oribasius and could have persuaded the man he never names, much less thanks, to share his memoranda with him. Above all, Eunapius' words about Oribasius imply his unique possession of a work written for him alone (see below n.18), and the assumption that Ammianus enjoyed access to another work, similar but different, written by the very man lauded by Eunapius for his special gift to himself, undercuts Eunapius' claim. But these are subsidiary matters; the main question, surely, is whether or not the evidence of the texts compels the conclusion, as Sudhaus supposed, that Ammianus and Eunapius had recourse to the same literary work, not whether some means can be devised to account for Ammianus' special access to Oribasius' alleged notebook, as if it were a certainty that Ammianus consulted it.

⁹ Thompson (n.5) 31, 137; he altered his views in Latin Historians, ed. Dorey, T. A. (New York 1966) 152ffGoogle Scholar.

¹⁰ Chalmers, CQ x (1960), 152ff.; Barnes is cited above in n.1; Bowersock, G. W., Julian the Apostate (Cambridge, Mass. 1978) 7Google Scholar, Blockley, R. C., The fragmentary classicising historians of the later Roman Empire (Liverpool 1983) ii p.viiGoogle Scholar, Matthews, J., The Roman Empire of Ammianus (London 1989) 175.Google Scholar Matthews, more than the others, insists on a comparatively limited use of Eunapius by Ammianus.

¹¹ Dependent theories which draw the elusive Nicomachus Flavianus and the Epitome de Caesaribus into the hypothetical web need not concern us here; for these see Paschoud, Zosime i, p.lv and Rosen (n.1 above).

¹² The conventional date for the publication of Ammianus' history is 391/2; until recently, Eunapius was supposed to have ended his history in 395; the new date, after 378 (see the literature cited in note 10), would therefore make it theoretically possible for Ammianus to have read and used Eunapius.

¹³ See, e.g., Chalmers, CQ x (1960) 156. See below p.3.

¹⁴ We should expect harmony as to the essential components of the story, e.g., the itinerary followed by the expedition. The serious disagreements which divide Ammianus and Zosimus in this respect, however, undermine the notion of literary dependency because they make this ‘dependency’ inescapably capricious. But a capricious theory is a contradiction in terms. See below pp.7–10.

¹⁵ Mommsen, Ges. Sehr. vii 437, convinced as he was of a bond between the two writers, kept, with better judgement, to the idea of a common source.

¹⁶ According to the calculations of R. Goulet, JHS c (1980) 60ff., Eunapius reached his fifteenth birthday between 26 Sept.363 and 25 Sept.364 (64). See Fornara, CQ xxxix (1989) 517–523 for an emendation of the critical passage VS x.8.3.

¹⁷ Cf. F 1 ad fin.: εγὼ δὲ κατὰ τὸ πιστεύειν ἐμαυτῶι γράφω, ἀνδράσιν ἑπόμενος οἱ τοῦ καθ’ ἡμᾶς βίου μακρῶι προεῖχον κατὰ παιδείαν καὶ διατεταμένως ἐνῆγον μὴ σιωπᾶν τά κοινά τῶν ἕργων … ἐγίνετο δὲ ἐκείνοις τε κἀμοὶ κοινὸν тὸ ἕργον τόδε, καὶ πάντα γε ἐς τὸν Ἰουλιανὸν ἀναφέρειν ἐδόκει….

¹⁸ The concluding words of the fragment (ὥστε οὐκ ἦν ἀναβολὴ κτλ.) also deserve attention, for they, taken with συνετέλει, inevitably suggest that Oribasius had attempted no memoir about Julian until this special occasion arose—for otherwise Eunapius could hardly have felt (as he states) the necessity for instant compliance. See also nn.8 and 20.

¹⁹ Fornara, , The nature of history (Berkeley 1983) 181.Google Scholar

²⁰ For the recent rehabilitation of Oribasius see Chalmers, , CQ x (1960) 155.Google Scholar Whether Oribasius had kept a diary which he used to refresh his memory of the circumstantial and technical details is unknown. As Seeck inferred, Hermes xli (1906), 531, the memorandum presented to Eunapius was evidently composed long after the expedition. Chalmers, , CQ x (1960) 156Google Scholar, doubts Seeck's dating of the memorandum (which Seeck presented without argument) because ‘there is … no real evidence to substantiate it.’ The evidence is real enough. Eunapius did not attain his thirtieth birthday until the year of Adrianople, 378. This was no Praxagoras (Jacoby, FGrHist 219 T 1.9), exulting in his youthful prowess, but a man who, as the fragment quoted above proves, had acquired sufficient reputation to be importuned by distinguished people to write his history; the date cannot be close in time to the expedition.

²¹ For an assessment of the fragments see Thompson (n.5) 136, Chalmers, , CQ x (1960) 155f.Google Scholar, Blockley (n.8) i 7ff.

²² He is placed by Paschoud, Zosime i p.xvii, between 498 and 518. Identification of the historian with various homonyms (see Paschoud's discussion, pp.xvii–xx) proves too speculative to be useful. See further Paschoud's bibliographical appendix in iii² (1989) 80f.

²³ See p.9. The removal of Magnus from consideration as Zosimus' source follows from Mendelssohn's inability to show that Zosimus abandoned his main source, Eunapius. The objection to Eunapius' (systematic) use of Magnus flows from the assertion of Eunapius quoted above from F 8 Muell., 15 Bl. For Thompson's rejection of Magnus as the same person as the tribune who distinguished himself at Maiazamalcha, see the Appendix.

²⁴ CQ vii (1957) 129ff.

²⁵ See pp.130–133. Cameron, A. D. E., CQ xiii (1963) 235Google Scholar, suggests other candidates. Norman concludes (132f.) that Zosimus may have gotten the version we also find in Ammianus from some other source unknown to us. ‘If, on the other hand, Ammianus is to be relied upon [as accurate] and Zosimus had access to the information found in him, Zosimus' conduct is almost inexplicable, since he but half corrects a point of detail and leaves so much undone.’

²⁶ It is conceivable that the Suda quoted an allusion by Eunapius to a variant making Magnus the first of the three, and it is also conceivable that the author of this variant was Magnus himself. Norman (129) is doubtless right that Magnus would not have described himself as ‘brave and extraordinarily daring,’ but he might well have written of his exploit in such a grandiloquent way as to enable Eunapius to accuse him of braggadocio by the use of this expression.

²⁷ Zosimus, p.xlv.

²⁸ CQ x (1960) 154. Paschoud explains, Zosime ii¹ n.6, that the new material consists primarily of ‘anecdotes arétalogiques’. But Zosimus' words (in the final sentence quoted above) imply that what is new is substantive, for the qualifying clause (καὶ μάλιστα ὅσα …) connects with the preceding words, εἰρήσεται καὶ ἡμὶν συντόμως ἕκαστα, while these apparently refer to the historical data.

²⁹ Sudhaus (n.5 above) simply assumed (3 with 89) that similarities (of any and every sort) proved literary affiliation, and was therefore primarily concerned to harmonize the two texts, sometimes urging emendation, e.g., Zos. iii 12.2 (Edessa), 13.2 (a fleet number), the names Loukianos and Konstantios (iii 13.3 with Amm. xxiii 3.9), sometimes merely stating that disagreement is parvi momenti (a much used phrase), especially when numbers are in question. Occasionally he contended that apparent disagreement must be illusory since Ammianus and Zosimus are so often in agreement (e.g. 33,40). Sometimes Sudhaus submerged disagreement by discussion of the ‘historical question’ (e.g. 73,74). Substantive differences are often ignored (e.g. 41,43), as well as sequential aberrations (e.g. 50). On one occasion, Sudhaus explains a disagreement (Amm. xxiv 4.4; Zos. iii 20.2) by postulating ‘confusion’ engendered in the various participants by the rush of events, as if this naturally explained the textual disagreement. Klein (n.5 above) used Sudhaus's material for his attempt to isolate the ‘fragments of Magnus’; Klotz differs chiefly to the extent that he took some of the more illusory ‘parallels’ as evidence for Ammianus' use of a second source. See RhM lxxi (1916) 488, 505. (Borries' paper, Hermes xxvii (1892) 170–209, contributes little. A specimen (174): ‘In Ammians Darstellung der Feldzüge Julians finden sich also Widersprüche und Incohärenzen. Daraus ergibt sich, dass Ammian für die Schilderung der Thätigkeit Julians zwei Quellen—nichts berechtigt uns, mehr zu anzunehmen—in einander verflochten hat, und zwar in recht wenig geschickter Weise.’)

³⁰ Even Barnes (n.1 above) 117, treats the matter as if it were unproblematical and routine: ‘Parts of Ammianus' narrative show a strong similarity to Zosimus' account of the period, in the selection of facts, in their arrangement, sometimes in apparent error, and even sometimes in verbal expression, especially when narrating Julian's Persian expedition (Ammianus, xxiii 2.6 ff; Zosimus, iii 12 ff.). A literary relationship must be inferred.’

³¹ See below, pp.11–12.

³² Among others, Klotz, , RhM lxxi (1916) 467Google Scholar, 483, 486, 489, attributes these differences to the librarii.

³³ RhM lxxi (1916) 467; CQ x (1960) 159.

³⁴ The presence of the aforesaid details in Ammianus' narrative should be enough to meet the possible objection that Ammianus might casually have ignored comparable material because he considered it neither historically important nor worthwhile artistically, for in that case these details too should have been omitted. To suppose that he might have mentioned one detail while intentionally omitting another of the same type is tantamount to the belief that he proceeded without any clear rationale for the inclusion or exclusion of material. But in that case the purpose motivating his alleged use of Eunapius is undercut.

³⁵ Syria xxxviii (1961), 156. Paschoud, Zosime ii¹ n.50 (142) objects on the grounds that the two places are not absolutely identical; he supposes, instead, that Ammianus left Phissenia out of his narrative. That argument, needless to state, presupposes a settled view of the source-relationship.

³⁶ Apart from the notice about Edessa, a significant discrepancy appears at the very beginning of the journey (Amm. xxiii 2.2–3, Zos. iii 12.1). Klotz, , Rhm lxxi (1916) 469Google Scholar, judged it an ‘arbitrary change’ made by Zosimus. Above all, the events framed by Circesium and Dura fundamentally differ, as Mommsen, Ges. Schr. vii 427, rightly insisted, though subsequent writers continue in their efforts to harmonize the two accounts. I discuss this episode in a forthcoming issue of AJAH.

³⁷ Substantial disagreement commences with the location of Gordian's tomb. Of the road between Thilutha and Diacira (Amm. xxiv 2.2–3; Zos. iii 15.2) Zosimus states generally that a number of forts followed the example of Thilutha, opting for neutrality; Ammianus knows of only one such city, Achaiachala, and adds that Julian burned a set of deserted forts; he also brings the army to Diacira by way of the place Baraxmalcha. Sitha and Megia apparently were not on Ammianus' itinerary-list; Amm. xxiv 2.7–8 gives a radically different picture from Zos. iii 16 (see below p. 10). We have already noticed the divergency with regard to the Roman-style palace and royal park (Amm. xxiv.5.1–2; Zos. iii.23.1). The arrival at the ‘Naarmalcha’ in Ammianus follows on the destruction of a munimentum (xxiv.5.7–12) ignored by Zosimus, who links it to the the first disappointing ἐλάττωμα experienced by the troops (3.24.1). From this point it becomes increasingly difficult to align the two texts, which describe the same route with a different system of names, e.g., Abouzatha and Noorda in Zosimus iii.26.1,3. Klotz, , RhM lxxi (1916), 505Google Scholar, supposed that Ammianus changed sources after xxv.7.3, chiefly because of the different description of the provinces surrendered to Persia in xxv.7.9 (cf. Zos. iii 31.1).

³⁸ I.e., those of Sudhaus-Paschoud on the one hand, and Mendelssohn and his followers on the other.

³⁹ Consider, for example, the first four correspondences with which Sudhaus began his study, namely (6), the description of Batnae: municipium Osdroenae (xxiii.2.6), πολίχνιόν τι τῆς Ὀσδροηνῆς (Zos. iii 12.2); (9) the disturbance caused by news of a Persian raid (xxiii 3.4–5; iii 12.4); (10) the commission bestowed on Procopius and Sebastianus (xxiii 3.5; iii 12.5). Careful comparison will reveal (cf. Paschoud, Zosime ii¹ n.33) that (1) the roads are defined differently, (2) the effect of the raid is not presented from the same perspective, Eunapius having apparently described a panic in the army, (3) the motivation ascribed to Julian differs, and (4) the numbers (and instructions) given Procopius and Sebastianus do not tally. An author who can ‘correct’ another (or insist on the superiority of his own recollection) does not need a ‘source’ for his knowledge of the outline of events.

⁴⁰ See n.32. Klotz here follows Sudhaus. Corruption is likely in Zos. iii 13.3, since Lukianos, named here with Antonios (Antoninus in Ammianus), is correctly written elsewhere as Loukillianos.

⁴¹ ‘Wenn schliesslich Zosim. iii 25,7 von 75 Toten auf römischer Seite spricht, Ammian hingegen nur 70 zählt, so wird das an sich kaum jemand aus einer Verschiedenheit der Quellen erklären wollen, um so weniger (my italics), als sie in den 2500 gefallenen Feinden übereinstimmen’ (Klotz, , RhM lxxi (1916) 494).Google Scholar See Thompson (n.5) 30f.

⁴² CQ x (1960) 158f.

⁴³ ibid 158.

⁴⁴ See above nn.36, 37.

⁴⁵ At Carrhae Julian formally reviews his army of 65,000 (Zos. iii 13.1); for the events attached to Circesium see n.36 above. Note the omission by Zosimus of the significant detail given by Ammianus in xxiv 1.11–15. Julian's initiative in Zos. iii 16.2 finds no counterpart in Ammianus, nor, for that matter (assuming the use of Ammianus by Eunapius) does Julian's exploit recorded by Ammianus in xxiv 2.14–15 appear in Zosimus. The speech delivered by Julian after Pirisabora takes a different turn in each writer (Amm. xxiv 3.2–9, Zos. iii 18.6) and the sequence of events varies (Amm. xxiii 3.1 with Zos. iii 19.1; xxiv 3.2 with iii 18.6). At Maiazamalcha (unnamed by Zos.) Julian's assailants in the surprise attack number two in Amm. xxiv 4.4, one in Zos. iii 20.3. Julian's near escape near Ctcsiphon was unknown to Zosimus. The pattern continues for the rest of the expedition.

⁴⁶ Eunapius must have described an attack delivered against Thilutha (προσβαλών in iii.15.1); Ammianus did not (xxiv 2.1–2). Zos.iii 16.2–17.2 stands alone; for Pirisabora see Chalmers, , CQ x (1960), 159Google Scholar, but add the difference in the numbers of the capitulating garrison (2500 in Amm. xxiv 2.22, 5000 in Zos. iii 18.4). Subtle but nonetheless real variation occurs in respect to Maiazamalcha (Amm. xxiv 4.10–31 with Zos. iii 21–22.7). Note the differing indications of time; the contrasting sketch of the emergence of the three men from the tunnel (Amm. xxiv 4.23; Zos. iii 22.5); and the competing versions of the manner in which Nabdates/Anabdates was captured (xxiv 4.26 with iii 22.6). Disagreement about the identity of the corpses found at the place named by Zosimus as Meinas-Sabatha (Amm. xxiv 5.3; Zos. iii 23.3–4) reflects a wider difference between the two writers in regard to this deserted city (Amm.) which in Zosimus is taken by storm. Just afterwards, an attack is directed against the Romans by a cuneus which either slipped out of a city (Amm. xxiv 5.5) or (Zos. iii 24.1) slipped into it. The divergencies intensify thereafter; one should compare, for example, the entire sequence culminating in the battle before Ctesiphon (Amm. xxiv 6ff. with Zos. iii 25ff.).

⁴⁷ For the rejection of the hypothesis of the ‘common source’ see above p.8.

⁴⁸ The phrase is Thompson's (Thompson [n.5] 31, 136).

⁴⁹ Thompson (n.5) 28–32 concentrated on the discrepancies; Chalmers' discussion, CQ x (1960) 156ff, stands alone as a serious attempt to supply a rationale for a hypothesis of literary contact; Ridley, T. R., Historia xxii (1973) 317–330Google Scholar, is not useful.

⁵⁰ We may dispense, surely, with discussion of the historical note about Diocletian's fortification of Circesium, and the existence of ‘Trajan's tribunal’ at Ozagardana/Zaragardia. These items were part of a city's identity, the predicate of its name, the common knowledge of every visitant.

⁵¹ CQ x (1960) 156f.

⁵² CQ x (1960) 158. Cf. Matthews, The Roman Empire of Ammianus, 171f.

⁵³ As to the deer: ἔνθα καὶ πλῆθος ἐλάφων φανὲν οἱ στρατιῶται κατατοξεύσαντες ἅλις ἐχρήσατο τῆι τούτων τροφῆι (iii.14–2) [≈] in quo loco greges cervorum plures inventi sunt, quorum alii confisi missilibus, alii ponderibus illisi remorum ad satietatem omnes paverunt; pars maxima natatu assueta veloci alveo penetrato incohibili cursu evasit ad solitudines notas (xxiv. 1.5). Far less impressive are the parallel passages about the palm grove (iii.20.1 [≈] xxiv.3.12), for, among other things, the fundamental observation made by each writer is not the same: ἄλσος δὲ ἐκ φοινίκων πεποιημένον, ἐν ὦι καὶ ἄμπελοι παραπεφύκεσαν ἄχρι τῶν φοινίκων κώμης τοῖς κλήμασιν ἀνατρέχουσαι, παρέχουσαί τε ὁρᾶν τον ἐκ τῶν φοινίκων καρπὸν ἀναμεμιγμένον τοῖς βότρυσιν ≈ in his regionibus agri sunt plures consiti vineis varioque pomorum genere, ubi oriri arbores assuetae palmorum per spada ampla adusque Messenen et mare pertinent magnum instar ingentium nemorum. et quaqua incesscrit quisquam termites et spadica cernit assidua, quorum ex fructu meilis et vini conficitur abundantia. et maritari palmae ipsae dicuntur etc.

⁵⁴ It is a privilege and a pleasure to record my deep gratitude to the John Simon Guggenheim Memorial Foundation for the generosity which has enabled me to devote myself fully to scholarship during the academic year 1988–1989.

^{note page 14 1} The historical work of Ammianus Marcellinus, 29f.

^{note page 14 2} See above n.4 to main text.

^{note page 14 3} See above p.5.

^{note page 14 4} CQ vii (1957) 129.

^{note page 14 5} CQ x (1960) 153.

^{note page 14 6} CQ xiii (1963) 234f., n.3.

^{note page 14 7} RE Magnus no.27, 493.

^{note page 14 8} p. 32; cf. p.9 above.

^{note page 14 9} Pp.28–31.

^{note page 14 10} p.31.

^{note page 14 11} Ibid.

^{note page 14 12} For examples of the military involvement of tribunes see Ammianus xv.3.10. xix.92, xxv.1.9, xxv.6.3 and, especially, xviii.2.11.

^{note page 14 13} p.32.

^{note page 14 14} CQ vii (1957) 129f.

^{note page 14 15} CQ xiii (1963) 233.

^{note page 14 16} See above n.25 to main text.