WRITERS, RASCALS AND REBELS: INFORMATION WARS IN THE RES GESTAE OF AMMIANUS MARCELLINUS

Abstract This article examines how the historian deals with ‘information’ broadly conceived, especially its acquisition, retention and loss. Ammianus details a complex interplay between those who control information and those who must work with an information deficit. Just as this dialogue plays out within the text, however, so too does it with respect to the author's methodology, which dances between the poles of incomplete and complete information depending on circumstance. Ammianus thus becomes an author as hard to pin down as many of his characters, by turns the omniscient narrator and the dumbfounded participant.


INTRODUCTION
Gibbon set the tone for scholarship on Ammianus Marcellinus when, in the late 1700s, he declared him to be an 'accurate and faithful guide'. 1 Recent work, however, has redirected discussion more towards his artfulness and elusiveness.Rees, in an intriguing article on the Res Gestae of Ammianus Marcellinus (hereafter RG), has gone so far as to argue that the author actively misdirects his readers with cross-references, that the earlier (believed lost) books never existed in the first place and that Ammianus aims at a comprehensive program of misinformation which comprises nothing less than his entire methodology. 2One ambition of this paper is to continue the spirit of Rees's investigation while softening his extreme conclusion, and it too begins with questions of methodology and misinformation.Ammianus outlines an 'information wars' dynamic which runs through his own methodology as much as it does the action described, cultivating a consistent thematic dialogue between information and disinformation, certainty and confusion.
A second goal, deriving from the first, is to restate the link between the author and his work.G. Kelly argued over a decade ago that it 'is time … that scholarship on Ammianus moved away from the figure of the author'. 3This was a call to re-site the RG as a literary product, as the main focus, rather than as of secondary importance to discovering the biography or credentials of Ammianus himself.A consequence of this move, though it has led to some interesting scholarship especially as regards the * I would like to thank Mark Humphries, Caroline Rowland and Keith Ward for their comments at various stages of the preparation of this paper.Thanks also are due to the anonymous reviewer at CQ who gave me much food for thought.The remaining errors are my own.RG's kinship to other texts, has also meant that at times the author's impact on his own work is downplayed in favour of tracing 'a dialogue with other books'. 4This contribution hopes to trace a dialogue within the book.What follows will discuss how the war over information plays out in Ammianus' methodology, before analysing a pair of rebellion passages, namely those concerning Silvanus in Cologne (15.5) and Firmus in Roman North Africa (28-9).It will show how there is a constant struggle by characters (and author) for more complete information and control over its dissemination.To be sure, that confusion would reign on the ground in these situations is perhaps to be expected.Ammianus himself, when he arrives on the scene at Cologne, expresses surprise at how quickly the situation had deteriorated, blaming abstract Rumor and Fama (15.5.24).The ambition here is to suggest, firstly, that the 'information war' exists; and, secondly, that it even moves between different planes of the history, from action to methodology and from political events to learned digressions.
We begin with the writer himself, however, because the manner in which he grapples with the problem sets the tone for his characters.At stake is how much information is available, who is brokering it, and the use to which it is put.The author's so-called prefatory statements (15.1.1,26.1.1) 5and the sphragis (31.16.9), in which Ammianus declares himself a former soldier and a Greek, but also in which he dwells on how silence or falsity literally destroys (corrumpere) truth, provide a lens through which readers may recognize and interpret the 'information war'.

1: Ammianus
The various digressions hint at Ammianus' preoccupation with acquiring and presenting information.Modern readers of the RG might regard the geographical and ethnographical digressions as awkward intrusions into the narrative of events and thus to be treated separately from the history proper.But such digressions were integral to ancient historiography, from Herodotus to Procopius, and similar methodological concerns underlay both elements. 6In Ammianus' case, his claim at the beginning of the digression on Thrace to combine in his account things seen with things read (uisa uel lecta quaedam perspicua fide monstrare, 22.8.1), corresponds fairly closely with sentiments in the first preface at 15.1.1 (see below).The digressions reveal aspects of Ammianus' historiography more widely. 4N.W. Bernstein, 'The siege of Amida and epic tradition: Ammianus Marcellinus, Res Gestae 19.1-9', Mnemosyne 72 (2019), 994-1012, at 1009.The article itself analyses the Homeric (and other) allusions in the Amida siege narrative.
5 'So-called' because although the general preface which Ammianus would likely have written at the beginning of the work does not survive, the statements at 15.1.1 and 26.1.1 signal a methodological shift and thus serve a prefatory purpose.This shift is discussed in the first section. 6For a useful survey see E. Dench, 'Ethnography and history', in J. Marincola (ed.),A Companion to Greek and Roman Historiography (Chichester, 2011), 493-503.This modern unease with the digressions exists pointedly in W. Hamilton and A. Wallace-Hadrill's brilliant Penguin edition of the RG (London, 1986; repr.2004), aimed at the general public.Though highlighting the importance of the digression to the ancient historian's art (p.25), they nevertheless excise a number of them from the English text for fear of provoking tedium.
The ancient writers are conflicted regarding the first origin of the Gauls, the account they have left of the matter being half-full.But afterwards Timagenes, a Greek in both carefulness and language, collected from a great many books these things which have remained unknown for a long time.Following his authority, having driven away obscurity, I shall instruct on the matter precisely and openly. 7is extract demonstrates Ammianus' purpose not only in the Gallic passage to which it belongs, but also his aim in the work more generally-to correct misunderstanding and to provide information, where before it was absent or incomplete.A succession of words denotes the extent of the confusion.First, Ammianus deals with what has come before.Earlier sources are undecided (ambigentes), and Ammianus acknowledges the work of Timagenes, whose care (diligentia) in sifting through earlier material has saved many lost facts.Then, Ammianus covers his personal contribution to the matter, mentioning the quality of his information.First, he shall treat his predecessor as an authority ( fides), thereby enhancing his own credentials; then he shall set forth his own points with precision and openly, literally 'removing darkness' (obscuritate dimota).8After quality, he moves on to the amount of information available-the ancient writers' account is half-full (notitiam … semiplenam) until Timagenes, who had perused a great number of books (multiplicibus libris) in order to write his own.Information's quality and quantity is a recurring theme.
Ammianus claims to clear away confusion, especially in the digressions.The following is from his account of the earthquake that struck Nicomedia in August 358 (17.7.9): adesse tempus existimo, pauca dicere quae de terrae pulsibus coniectura ueteres collegerunt.ad ipsius enim ueritatis arcana, non modo haec nostra uulgaris inscitia, sed ne sempiterna quidem lucubrationibus longis nondum exhausta, physicorum iurgia penetrarunt.I think the time is here to say a few things on the theories concerning earthquakes which the ancients have collected.Neither this vulgar ignorance of ours, nor even the never-ending debates of the natural philosophers which, indeed, are not yet finished despite long sessions deep into the night, have penetrated the secrets of the truth itself.
Once more there is consistent focus on the scholarly debate that has come before, the difficulties in understanding that debate, and Ammianus' implied role in clearing up the knotty problems for the benefit of his readers.Indeed, Ammianus himself takes on greater agency than in the previous extract, in which he had relied on Timagenes' reading of the ancient authorities.Here, though, the historian claims to have read much of the relevant literature and, as the digression progresses, he duly namedrops a number of eminent writers. 9s in the previous extract, both the quality and quantity of information is at issue.On the one hand, Ammianus notes 'our vulgar ignorance' (nostra uulgaris inscitia), unfavourably comparing contemporary levels of knowledge to those he imagines prevailing in earlier periods of Roman history; on the other hand, despite the charge that such debates are eternal and still not concluded (sed ne sempiterna quidem lucubrationibus longis nondum exhausta), more remains to be said.Indeed, he nods to his wider theme of information and its gathering with reference to the night-time work of earlier authors (lucubrationibus longis) as a means of underlining quite how troublesome is the topic amongst scholars.Ker has noted how there had long been a 'culture of lucubratio' in Roman writing; 10 and Ammianus references writing by candlelight to demonstrate that even then, despite the hard work of many scholars, confusion still reigned, and a fresh voice (Ammianus himself) was needed to banish the darkness.In both this and the previous passage, Ammianus takes up the position of confident interpreter expertly sifting through information and providing a more complete picture of the question for his readers-he alleviates perplexity. 11ust as frequently, however, Ammianus qualifies his own authority, confessing himself to be as confused and as partial as any of the characters in the text.An example occurs in his account of the trials instituted under Valens at Antioch beginning in 372: et quoniam addici post cruciabiles poenas uidimus multos, ut in tenebrosis rebus confusione cuncta miscente, summatim quia nos penitissima gestorum memoria fugit, quae recolere possumus, expeditius absoluemus.
The victims that I have seen hauled away after torturous punishments are so numerous that everything is confused in my recollection and I am left in the dark.Since the details of all that happened escape me, I will give a briefer account of what my memory retains. 12re Ammianus sets aside the persona of the informed, confident narrator and instead emphasizes the subjectivity of his own viewpoint.Even though he writes of events that he has personally witnessed, everything is jumbled up (confusione cuncta miscente) and darkness has descended (ut in tenebrosis rebus), leaving his memory of the events incomplete (memoria fugit) and the resultant account necessarily brief (expeditius).Once again, our author plays on the theme of deficient information to demonstrate the difficulties stemming from his methodology.These difficulties had been present even in the digressions, and they could be overcome only through rigorous research and book-learning, combined with eyewitness reports or autopsy.
Such confessions of incomplete knowledge, which may (but need not) be historiographical topoi, are a feature of the later books especially. 13There have been several explanations for this.One theory is that Ammianus is surer of himself during the early books when he had to rely more on written sources, whereas the later books rest more on the fallible nature of his own memory and notes. 14But this is not completely satisfactory.For one thing, other ancient authors, like Livy, still stress the incompleteness of memory even if they are largely using earlier written sources as the basis of their narrative. 15For another, there is a great deal of evidence that learned people in the ancient and medieval world possessed exceptionally strong memories owing to the prominence of orality in their schooling especially and societies generally. 16It is unclear how strong Ammianus' memory was, but it is likely that he knew certain authors by heart. 17In the best treatment of Ammianus' historical memory, Fornara demonstrates that it was precisely because Ammianus was still an active eyewitness of the events catalogued in Books 15-25 that he includes more detail in the earlier books than the later ones. 18But this explanation is not sufficient either, since even in the later books there are suggestions that Ammianus was an eyewitness, 13 e.g.28.1.2(records only those things worthy of memory), 29.3.1 (urges his readers to think on that which he passes over in silence since he cannot include everything that happened under Valentinian), 31.5.10 (asks his readers not to demand of him an exact tally of the slain in the battles between the Goths and Romans).
14 e.g.E.A. Thompson, The Historical Work of Ammianus Marcellinus (Cambridge, 1947), 25-39. 15Livy 8.11.1 haec, etsi omnis diuini humanique moris memoria aboleuit noua peregrinaque omnia priscis ac patriis praeferendo, haud ab re duxi uerbis quoque ipsis, ut tradita nuncupataque sunt, referre ('These particulars, even though the memory of every religious and secular usage has been wiped out by men's preference of the new and outlandish to the ancient and homebred, I have thought it not foreign to my purpose to repeat, and in the very words in which they were formulated and handed down').The translation is that of the Loeb edition by B.O. Foster (Cambridge, MA, 1926). 16The classic Roman statement on memory is made by Cicero (De or.2.351-4), who advises that the best way to remember a speech is by thinking of a basilica with niches containing the topics to be covered.The Rhetorica ad Herennium, whose author was once thought to be Cicero as well, is another useful take on memory.Writing just after Ammianus, Augustine devotes many pages to memory and its retention at the beginning of the tenth book of his Confessions.P. Brown, Augustine of Hippo.A Biography (London, 1967), 36 notes that Roman schooling 'imposed a crushing load on the memory'.The best study on memory in the past is that by M.J. Carruthers, The Book of Memory: A Study of Memory in Medieval Culture (Cambridge, 2008 2 ).Though working on medieval Europe, she argues (p.9) that 'medieval culture was fundamentally memorial, to the same profound degree that modern culture in the West is documentary'. 17M.L.W. Laistner, The Greater Roman Historians (Berkeley and Los Angeles, 1947), 152 (Virgil); H. Teitler, 'Visa vel lecta?Ammianus on Persia and the Persians', in J.W. Drijvers and D. Hunt (edd.),The Late Roman World and its Historian (London, 1999), 191-197, at 194  (Herodotus), though concluding that Ammianus may have misremembered a quotation from that author.C. Wickham, The Inheritance of Rome: A History of Europe from 400 to 1000 (London, 2010), 29 suggests that western officials were expected to know Virgil by heart; and eastern ones Homer.In the fifth century, Sidonius praises Avitus for teaching Theoderic II to learn Virgil by heart (Carm.5.495-7).
18 C.W. Fornara, 'The prefaces of Ammianus Marcellinus', in M. Griffith and D. Mastronarde (edd.),Cabinet of the Muses: Essays on Classical and Comparative Literature in Honor of Thomas G. Rosenmeyer (Atlanta, 1990), 163-72, at 166. or at the very least relying on eyewitness reports. 19And yet still here he dwells on, and even introduces, confusion.
Another explanation for these confessions of incomplete knowledge is that, as Ammianus approaches living memory, he practices a form of self-censorship in order to avoid giving offence and attracting the attention of a potentially repressive (and stridently Christian) regime. 20This is a popular view amongst those scholars who see the historian as engaging in covert polemic against the emperor and his Christian adherents, 21 and there is evidence that, for instance, the Latin panegyrists took a highly selective approach to the very recent past in order to avoid sensitive issues. 22Ammianus himself alludes to this possibility when he invokes the example of Phrynichus of Athens, who was severely fined by the state for reminding his countrymen of the sack of Miletus by Darius during the Ionian revolt in 493 B.C.E. 23Our historian may have feared a similar fate or worse.
However, a recent article by Kaldellis has questioned the prevailing view that late antique authors necessarily had to practice self-censorship in order to survive. 2419 e.g.omnes ea tempestate uelut in Cimmeriis tenebris reptabamus, 29.2.4.'Our Hypatius' (noster Hypatius, 29.2.16) has been suggested as a possible source by several scholars, not least Barnes (n.9), 60, who thinks that the possessive adjective indicates personal friendship.
22 C.E.V. Nixon, 'The use of the past by the Gallic panegyrists', in G.W. Clarke et al. (ed.), Reading the Past in Late Antiquity (Rushcutters Bay, NSW, 1990), 1-36, at 17.A good example may be found in Euseb.Vit.Const.4.68, which casts the succession of Constantine the Great's sons after their father's death in 337 as an orderly affair and makes no mention of the massacre by the soldiers of numerous relatives and nobles, not least Constantine's nephews, Hannibalianus and Dalmatius, the latter of whom was a full Caesar at the time of his murder.Julian (Ep.ad Ath.270c-d), who was deemed too young to be a threat, later blames Constantius II (his cousin) for orchestrating the massacre.For a forensic discussion of all the available evidence, see R.W. Burgess, 'The summer of blood: The "Great Massacre" of 337 and the promotion of the sons of Constantine', DOP 62 (2008), 5-51.
23 28.1.3-4.For the story of Phrynichus see Hdt. 6.21.10.R. Lizzi, 'Ammiano e l'autocensura dello storico', in F.E. Consolino (ed.), Letteratura e propaganda nell'Occidente latino da Augusto ai regni romanobarbarici (Rome, 2000), 67-105 argues that Phrynichus' fate caused Ammianus to tailor his account of the treason trials in Rome under Valentinian (28.1) so as not to cause offence to the leading Roman families, his audience. 24A. Kaldellis, 'How perilous was it to write political history in Late Antiquity?',Studies in Late Antiquity 1 (2017), 38-64, at 40 claims that Pelagius is the only late Roman historian to be put to death for his writing, contrasting this with the earlier empire where there is evidence that writers were more likely to be condemned.Tacitus (Ann.4.34) notes the fate of Cremutius Cordus, who was encouraged to starve himself to death under Tiberius; and the poet Ovid was famously banished to Tomis for 'a poem and an error' (Tr.2.207).
Kaldellis goes on to argue that although many writers did engage in self-censorship, there is a significant minority who did not, citing Procopius, Jordanes and Eunapius. 25Of these, the latter is closest in time to Ammianus, and the most anti-Christian.Eunapius likely published two versions of his history; the first, appearing around 395, contained a great deal of anti-Christian rhetoric which needed to be excised for the second version, appearing after 404. 26While it was conceivable to publish material against the government, doing so may have drawn the attention of censors and forced redactions.This possibility leads to the somewhat contradictory observation that authors may be both more and less likely to release potentially inflammatory material, since a writer was unlikely to be executed even if they wrote highly vituperative texts. 27In any case, it is not enough to state, as many have, that fear of repression automatically made Ammianus practice self-censorship, since we know that several authors categorically did not do this and lived to tell of it. 28aldellis makes a second observation which bears upon the question of Ammianus' self-selectiveness, and that is to do with the age of those authors who did in fact decide to publish sensitive material: generally, they were men in late middle or even old age.Jordanes was around seventy, or at least in retirement, when he wrote the Romana; Eunapius was around sixty in 404 for the second version of his history, while Procopius was at least fifty, and probably older, when he wrote On Wars and the Secret History. 29Given that Ammianus identifies himself as a young man (adulescens, 16.10.21)whilst serving under Ursicinus in Gaul in 357, it is likely that he had a birth date in the early to mid-330s.Most historians agree that he left the army sometime after Julian's death in 363 to travel the empire gathering material for the RG, after which he retired to Rome where he published and died, probably around 391. 30 Several scholars 25 To this list two works of Synesius of Cyrene may be added: De Regno and De Prouidentia.See P.J. Heather, 'The anti-Scythian tirade of Synesius' "De Regno"', Phoenix 42 (1988), 152-72 for the former, and T.D. Barnes, 'Synesius in Constantinople', GRBS 27 (1986), 93-112 for the latter, which especially has been read as a thinly disguised critique of two high officials in the Constantinople of 400, even if it was never actually read before the royal court.
26 D. Rohrbacher, The Historians of Late Antiquity (London, 2002), 67.For possible Eunapian influence on Ammianus (and vice versa) see J.H.W.G.Liebeschuetz, 'Pagan historiography and the decline of the empire', in G. Marasco (ed.), Greek and Roman Historiography in Late Antiquity: Fourth to Sixth Century A.D. (Leiden, 2003), 177-218. 27We have no way of knowing, of course, if Eunapius suffered more subtle punishments, like property confiscation or blackmail; and the threat of such things, though not as extreme as death, could have convinced other authors to avoid criticizing the regime.In a different context, P. Brown, Through the Eye of a Needle.Wealth, the Fall of Rome, and the Making of Christianity in the West, 350-550 AD (Princeton, 2012), 45 discusses the notion of 'gentle violence', in which simply the presence of a Christian emperor must have exerted invisible pressures on pagans at court. 28It should be noted that Kaldellis (n. 24, 54-5) believes that Ammianus was extremely wary of offending emperor Theodosius. 29These works are not all universally hostile to the regime.Kaldellis highlights a few sensitive passages in his article.The Secret History (Historia Arcana) was not published until after Procopius' death which, given its scathing picture of Justinian, Theodora and Belisarius, is unsurprising.Authors may have been free to score certain rhetorical points, but unvarnished polemic, in which you deemed the empress a prostitute known to all mankind (Arc.9.29) and the emperor a demon (12.26), was a step or three too far. 30T.D. Barnes (n.9), 1-2 is a useful summary of Ammianus' possible movements.As for dating, the Serapeum in Alexandria, infamously destroyed by a Christian mob in 392, is extolled in the present tense as rivalling the Capitol in Rome at 22.16.12.Strictly, this reference confirms only that Book 22 was published before 392, but most scholars think that if Ammianus were able (i.e.still alive), he would have glossed its destruction, especially since he glosses, during the preamble to Procopius' uprising, the notarius Neoterius' future consulship in 390 (26.5.14), a comparatively have argued that as Ammianus published he was embittered, perhaps because of his relatively undistinguished military career. 31And yet, if this were true, it would make it less, not more likely that Ammianus would practice self-censorship-his career was over, he had no reason to court the regime, and therefore had a greater degree of freedom to say what he wanted. 32 argue that Ammianus intentionally creates confusion in his narrative, not to avoid the censors, but as a means of illustrating the fragility of information and the difficulty inherent in attaining its completeness.To be sure, C. Kelly has suggested that in the later books Ammianus purposely creates a disjointed narrative which is hard to knit together. 33But while Kelly sees Ammianus as creating confusion only in the later books, as a means of critiquing a dystopian empire post-Julian, it is possible to extend the analysis to the earlier books too, and it is all in service to the author's wider theme concerning the interaction between complete and incomplete information.
The two prefaces at 15.1.1 and 26.1.1 give a clue as to the tightrope Ammianus walks between complete and incomplete information in the text more widely.There are significant differences between the first preface and the second.The one at 15.1.1 focusses on the detail that will be offered in the succeeding books.As was noted in the passages discussed above, Ammianus here stresses his privileged position-he includes only that which he witnessed himself (ea quae uidere licuit per aetatem), or that which he acquired through the questioning of those involved (uel perplexe interrogando uersatos in medio scire). 34Everything deemed fit for inclusion will be strictly accurate (limatius), and Ammianus pre-empts the critics who might complain that it is too detailed and long-winded (nihil obtrectatores longi, ut putant, operis formidantes) with the observation that nothing can be allowed to hinder the knowledge of events (cognitioni gestorum).Ammianus' voice is confident, open, and he states explicitly that he will indulge in detail and shun brevity.
The second preface at 26.1.1 reads differently.It is true that Ammianus does mention the dangers associated with writing about the immediate past, but most of the passage serves as a defence of his methodology.Whereas before he luxuriates in the detail he can offer either as a witness himself or as a questioner of those who were, and in fact minor event.D. Rohmann, 'The destruction of the Serapeum of Alexandria, its library, and the immediate reactions', Klio 104 (2022), 334-62, at 357 has argued that Ammianus was actually aware of the Serapeum's destruction, which Rohmann places in 391, but decided not to mention it for fear of provoking Christians 'in the heated atmosphere of the immediate aftermath of the destruction of the Serapeum'.As I argue in the text, however, even if we allow that simply glossing the event would have attracted negative attention, Ammianus is less wary of the censors (real or imagined) than might be thought.
31 T.D. Barnes (n.9), 63 theorizes that Ammianus' involvement in enforcing Julian's anti-Christian policies hindered his career after Julian.M. Kulikowski, 'Regional Dynasties and Imperial Court', in J. Wienand (ed.), Contested Monarchy: Integrating the Roman Empire in the Fourth Century AD (Oxford, 2015), 135-48, at 146 contrasts Ammianus with his fellow protector Agilo-the latter, argues Kulikowski, enjoyed better contacts and thus received many promotions.A. Cameron, Poetry and Propaganda at the Court of Honorius (Oxford, 1970), 30, reading the diatribes at 14.6 and 28.4 as influenced by personal experience, thinks that Ammianus was 'embittered' because late fourth-century Rome was largely unwelcoming to foreigners.

WRITERS, RASCALS AND REBELS
takes the attack to the critics (obtrectatores), now he is somewhat on the back foot, claiming that he had intended not to cover the recent past in order to avoid such people who might condemn him for missing out minor details (humilium minutias causarum), 35 when the principles of history demand only a survey of the highlights (negotiorum celsitudines).
It is significant that while Ammianus does mention contemporary danger as a reason to avoid recent history, most of his attention covers how much weight to give the highlights (celsitudines) and the trivialities (minutiae). 36He expresses a similar concern at 28.1.15where he points out that 'not everything which occurs amongst the unwashed masses is worthy of narrative' alongside a complaint regarding the unreliability of the public records in this time of 'new madness' (nouo furore). 37In other words, at issue in the two prefaces is not so much fear of a repressive regime inducing self-censorship as much as it is how much detail to include in both parts of the history-Ammianus' preoccupation is chiefly literary, not political, and as literary statements the prefaces function as a key for the surviving text.The first preface at 15.1.1 signals Ammianus' desire to aim for complete information; the second at 26.1.1 signals his desire to jettison completeness and instead record only the most important matters.Both methods are always in play, however. 38t is difficult to say why Ammianus exhibits such an interest in this information dynamic.As already noted, many scholars suspect him to be aware of the issue of censorship, and for them this awareness explains some of his selectiveness.But the contemporary political situation need not be the cause, at least not directly.Instead, it is better to look to the author himself (or rather, the author as presented by the text), and the peculiar traits he picks out for comment on his characters, for an explanation.The contribution of the two prefaces (15.1.1,26.1.1)to this theme has already been seen.But there are several passages in the wider text on which it is worth reflecting.Intelligence and knowledge, or the capacity for knowledge, are central.The emperor Julian excels in this regard, and his intelligence is greatly admired by Ammianus, who neatly combines it with an aptitude for military matters upon his first entry in the text (16.1.5).During the narrative of the campaign in Persia, Julian's unusual capacity for knowledge merits occasional mention, culminating in a death-bed scene which has him debating the finer points of philosophy with specialists (25.3.23). 39mmianus covers the intelligence of all the emperors either in the narrative or in their respective necrologies: Constantius has a narrow spirit (ut erat angusti pectoris, 19.12.5) and mixes up Christian dogma with his own old woman's superstition (anili superstitione confundens, 21.16.18);Valens' mind is uncultured, and he is unlearned in both art and war (subagrestis ingenii, nec bellicis nec liberalibus studiis eruditus, 31.14.5).Valentinian, though, is commended for his handwriting, painting, crafting, inventing and the strength of his memory and speech (30.9.4).Even Jovian, whose reign comprises just a few pages, is noted as only moderately educated (mediocriter eruditus, 25.10.15),albeit a keen student of Christianity.Zosimus writes on all these emperors too, but does not share Ammianus' fascination with knowledge and information gathering.Julian, whose intelligence is a centrepiece of his portrayal in the RG, appears chiefly as a general and an administrator (and the quintessential pagan) in Zosimus. 40mmianus is keenly aware of the power of information: he explicitly praises the 'noble system of instruction' ( praeclara informatio doctrinarum, 29.2.18) which cultivates even the harshest natures. 41As with Ammianus' methodology in the digressions, information acquisition is clearly an active process, and several times he notes its utility; for instance, how a knowledge of history brings with it great perspective and the possibility of recovery after the disastrous battle of Adrianople in 378 (31.5.11).But the task of interpretation can be a struggle.In Persia, Ammianus himself helps to decode a message through his knowledge of history, despite its great obscurity (18.6.19); and elsewhere investigators appointed by Constantius to examine the fall of Amida to the Persians in 359 become bogged down in trivial detail far removed from the wider facts of the case (20.2.3).These examples derive from different circumstances, but the underlying theme remains how much information is available, or able to be gleaned, and the use to which that information is put.
Along similar lines, the Roman digressions (14.6, 28.4) are instructive.In these Ammianus speaks as a direct observer of the capital's society, just as he claims to have written in the first preface (utcumque potui ueritatem scrutari, ea quae uidere licuit per aetatem, 15.1.1).The city Romans, unlike their 'ethnographer', are perversely uninformed about.The nobles do not meditate on the majesty of where they were born (ubi nati sunt non reputantium, 14.6.7),they are unaware (ignorantes, 14.6.10)that their ancestors gained fame through hard work and valour, and some are so oblivious that they fail to recognize the very client they had seen the previous day (14.6.13).Others, meanwhile, hate learning as a poison when really their reading material ought to have been deep and varied, as befitted their ancestry (28.4.14-15).Knowledge, information in general, is meant to be the preserve of Romans, but many of them are unwilling or unable to acquire it and thus become all the more contemptible.Contemptible, perhaps, like barbarians, who frequently 'forget' what had already been done. 4240 Zos.9.4-12.1. 41informatio, a rare word used only here by Ammianus, may derive ultimately from Cicero, who defines the Greek πρόληψις with it at Nat. D. 1.43.By the late fourth century, it had acquired currency in Christian circles (e.g.August.Ep. 12, Rufin.Hist.6.13, Ambr.Iac.1.7.31).
42 e.g.17.6.1 (the Juthungi forget the peace and the treaties [obliti pacis et foederum]), 17.13.1 (the Limigantes ally with their former masters as if forgetful of what had happened before [nam uelut obliti priorum]), 30.6.3 (Valentinian suffers a stroke while railing against emissaries of the Quadi, castigating them as forgetful of, and ungrateful for, past acts of kindness done to them by the Romans [beneficiorum immemorem et ingratam]).
Ammianus, though, portrays himself as a lover of knowledge, and if it is the preserve of Romans, then it is also the business of Greeks. 43When it comes to presenting scientific or geographical wisdom, often in the digressions, he often self-identifies as a Greek: the doomed Caesar Gallus experienced night terrors, 'which we call fantasies' (quas φαντασίας nos appellamus, 14.11.18);earthquakes are caused by underground passages 'which we call suringai in Greek' (quas Graece σύριγγας appellamus, 17.7.11);and when he describes various types of eclipse he uses terminology 'which we say often in the Greek language' (Graeco dictitamus sermone, 20.3.4).Meanwhile Timagenes, as we saw above, is a Greek in carefulness and language (15.9.2).
To be Greek is, for Ammianus, to be interested in information.Nowhere is this clearer than in his excursus on the Egyptian obelisk (17.4.17-23),where he transmits lengthy excerpts from the Greek translation of the original hieroglyphs. 44This belief in the worth of his calling as an historian and a Greek is the likely reason for his deep interest in the information dynamic, and it is peculiarly idiosyncratic, not unlike his hero Julian, whose eccentric brand of philosopher kingship has been well documented. 45here is no need to posit Ammianus engaging either in a dystopian critique of the post-Julianic empire (since these processes occur throughout the RG), 46 or in self-censorship.

2: Ammianus' characters
It is time now to move beyond the author and examine how his theme plays out in the narrative.Just as Ammianus himself grapples with the interaction between complete and incomplete information in a methodological sense, 47 so too do his characters, but in a very real sense.This interaction between the intellectual and the practical is hinted at in the sphragis (miles quondam et Graecus, 31.16.9).Ammianus confronts not just the intellectual problem of accessing and presenting information (alluded to by Graecus), but also the activities in the real world of certain shadowy individuals (alluded to by miles quondam).In other words, the sphragis hints at a tension in the text between the scholarly researcher on the one hand, whose 'war' for information is waged in books, and the practical man of action on the other, whose 'war' is not just 43 Cf.G. Sabbah, La méthode d'Ammien Marcellin: Recherches sur la construction du discours historique dans les Res Gestae (Paris, 1978), 27 for 'l'idéal d'une connaissance complète' which governs Ammianus' approach.
44 Ammianus ascribes the translation to 'Hermapion', an individual not otherwise attested.A. Benaissa, 'Ammianus Marcellinus Res Gestae 17.4.17and the translator of the obelisk in Rome's Circus Maximus', ZPE 186 (2013), 114-18, at 115 has suggested emending in Circo Hermapionis to in Circo Maximo Apionis.Apion, a scholar of Homer and a native of Alexandria, wrote in the first century.For the intelligent manner in which Ammianus interprets this passage see F. Foster, 'Bees and vultures: Egyptian hieroglyphs in Ammianus Marcellinus', CQ 70 (2021), 884-90. 45My own view, purely speculative, is that as a Greek and a soldier in the civilian, Latin-speaking Rome of the 380s-90s, Ammianus was keenly aware of his own 'outsider' status.Consequently, he sought refuge in strenuously acquired knowledge as a means of proving his credentials to a potentially ambivalent aristocratic audience. 46This is not to deny the dark tenor of the RG post-Julian; only to argue that the dynamic appended to support this theory is by no means unique to Books 26-31. 47Ammianus is of course not unique here.Cassius Dio (53.19) notes the relative ease of source work in the Republic compared to the empire, when information channels were more tightly controlled by the emperor and his court.For discussion of this passage see B.J. Gibson, 'Rumours as causes of events in Tacitus', MD 40 (1998), 111-29, at 124-6. on the battlefield, but also in the political arena, where he must somehow sift through competing viewpoints, downright lies and other techniques of misdirection designed to hide the truth. 48This section analyses two rebellion narratives as examples of the latter. 49n the first instance, and perhaps unsurprisingly, the same processes are involved in each.What perhaps is surprising is that this is so despite them ostensibly belonging to two different prefatory spheres (after 15.1.1 and after 26.1.1).In the second instance, just as Ammianus moves between the poles of complete and incomplete information in his historiography, so too do his characters, many of whom control the flow of information to the detriment of their targets (and the emperor).As ever, it is an active process.

2.1: The Silvanus affair (15.5)
The rebellion of Silvanus (15.5) occupies a significant position in the RG because it is the first time that Ammianus appears in the extant text as a participant.There is a great deal of scholarship on the episode, though a contribution by Weisweiler has taken the initial step towards the analysis offered here.Weisweiler argues that Ammianus' narrative repeatedly fails to answer its own questions and that it presents readers with 'competing versions of historical truth from which they have to construct their own histories'. 50I suggest, however, that in fact there are no 'competing versions of historical truth'.Instead, Ammianus presents the truth, which nevertheless must be filtered out from the lies and misdirection which usually attends it. 51Thus readers do not 'construct their own histories' as much as learn about and perpetuate the history propounded by Ammianus.The author is a broker of information, but so too are the characters he describes.
Just as Ammianus himself is at times the omniscient narrator and at others the fallible searcher for truth, so too are Silvanus, Ursicinus and Constantius by turns fully informed on one hand and scrambling for information on the other.Ammianus executes a series of bewildering shifts in perspective as he offers competing viewpoints, with the result that the reader becomes just as confused as the participants.There are fourteen individuals packed into the Silvanus episode who all have some impact on the formation and suppression of the revolt, 52 and one of these is Ammianus himself, who is sent along with the commander Ursicinus to put a stop to the usurpation as expeditiously as possible.Each person, to a greater or lesser extent, has a stake in the information war.
Ammianus begins with an authorial comment on the seriousness of the uprising, commending the role of Fortune in ending the affair before it could cause lasting 48 Ursicinus, Ammianus' commanding officer, exemplifies this struggle at 14.9.1.Although a skilled soldier, he is hopelessly outmatched in politicking. 49The account of the fall and execution of Caesar Gallus (14.9-15.1)deals with similar themes. 50J.Weisweiler, 'Unreliable witness: failings of the narrative in Ammianus Marcellinus', in L. van Hoof and P. van Nuffelen (edd.),Literature and Society in the Fourth Century AD (Leiden, 2015), 103-33, at 107. 51 Cf. 15.2.9 ideoque et ueritas mendaciis uelabatur, et ualuere pro ueris aliquotiens falsa 'thus truth was veiled by lies and sometimes false had power over true'.Ammianus' nuanced approach to truth, especially his preferring fides to ueritas (perhaps deriving from his military oath to maintain fides), is noted by G. Sabbah (n.43), 19-21. 52This count disregards individuals like the Frank Mallobaudes (15.5.6), the 'unexpected messenger' (insperabilis nuntius, 15.5.17) who brought news to Milan of Silvanus' revolt, and Verenianus (15.5.22), the colleague of Ammianus.Though named, these individuals are silent and have no direct role in the episode.
damage to the provinces (1). 53Then he praises Silvanus for fighting on behalf of the state in Gaul, valiantly pushing back barbarian raiders.And yet straightaway Silvanus has received misleading information, for he had been given this assignment by his rival Arbitio, in the hope that he would perish in his duty (2). 54The next section (3) introduces a certain Dynamius (Dynamius quidam), who asks the commander for letters of recommendation, thus appearing to others as if he were close to Silvanus.The latter, suspecting nothing (nihil suspicans), duly provides the material, which is kept by Dynamius for some scheme in the future.
Dynamius enlists the help of co-conspirators, including the praetorian prefect Lampadius, the comes rei priuatae Eusebius and the magister officiorum Aedesius, before erasing Silvanus' words, though keeping his signature, and writing a new, incriminating version.This is brought to Constantius, who orders that those named in the letter should be summoned to explain themselves (4, 5).But Malarich (6, 7), a general at court and a Frank, complains that those devoted to the empire (dicatos imperio) should not be slandered by factions and tricks ( factiones et dolos).
These initial stages establish the shape of the chapter as it develops-a straightforward and unsuspecting general fighting on behalf of Rome in the provinces dutifully responds to a favour.But this favour is misused, and it becomes a means of undermining him.Ammianus sets up two camps: one side comprises the innocent and guileless Silvanus, and his advocates with the emperor; the other the underhanded accomplices of Dynamius and Arbitio.Silvanus is oblivious to the machinations occurring behind the scenes.Indeed, he had been sent to Gaul in the first place by a conniving colleague who hoped that he would perish; and the letter of recommendation he had provided in good faith is immediately repurposed for nefarious ends.Ammianus writes how the presumably laudatory words are rubbed away with a sponge ( peniculo … abstersa, 4), demonstrating how the little information over which Silvanus did have command is literally destroyed without his knowledge, casting forwards to Ammianus' warning in the sphragis that his own words would never ruin truth (31.16.9).At this stage, then, Silvanus is losing the information war.In fact, he is not even aware of its existence.
The emperor Constantius is little better off: he orders Silvanus brought back from Gaul to defend himself (8).However, the man selected to retrieve the commander, Apodemius, is hand-picked by the hostile Arbitio.Apodemius neglects to follow the instructions he had been given.He refuses not just to hand over the summons to Silvanus, but even to meet him, treating the man's dependents as if they were slaves and as if their master were already dead.Constantius, then, begins this episode believing himself fully informed, and he dispatches Apodemius to retrieve Silvanus.But straightaway efforts are made to further confuse the situation.In the meantime, Dynamius, the author of the whole plot and the most informed of everyone, aims to reinforce his skulduggery, this time writing a letter in his own hand (though using the name of Silvanus and Malarich), to the tribune at the Cremona armoury, urging the man to hasten unspecified preparations (9).But the tribune is confused (since he had scarcely spoken to the Frankish pair), and returns the letter to Malarich, asking him to explain clearly what he wants, since the tribune confesses himself a rustic and simple man (subagrestem et simplicem) and unused to interpreting obscure meanings (10).This tribune has little information at his disposal and, like Silvanus and Constantius, is somewhat at the mercy of those who do have complete information.
The confusion is cleared somewhat as the chapter develops, thanks to the intervention of a particularly informed individual.Malarich, upon receiving the letter from the bewildered tribune, realizes the severity of the plot, assembles the Franks in the palace, and laments the trickery ( fallacia) involved (11).Constantius orders the deputy magister officiorum Florentius to investigate the document (12), ostensibly written by Silvanus but in fact the work of Dynamius, and, upon scrutiny, he detects the shadow of Silvanus' original words (those he had written for the letter of recommendation) beneath the forgery.The cloud of deceit ( fallaciarum nube) is dispersed (13) and Constantius finally learns the truth of the business through trusty report (imperator doctus gesta relatione fideli).As Ammianus grapples with issues of accuracy and fidelity in his historiography, so too do his characters, for it is only through careful study that the truth is revealed.Crucially, there are not competing versions of truth; instead, there is the actual truth over which characters fight for control.
It is only upon examination of the offending document that Constantius becomes fully informed once more, and yet even this is achieved, not because the emperor is available himself, but because he has learned second-hand the truth of the matter and, presumably, the original and utterly harmless contents of the letter.It is a truism that the later Roman emperor was far removed from the daily administration of the empire, but nowhere is this clearer than in Constantius' case, and, though it is a theme which runs through the whole text, Constantius is particularly susceptible to his officials' propensity for disinformation. 55In the event, the hitherto passive Silvanus is allowed to speak indirectly as fragments of his earlier words (apicumque pristinorum reliquias) are at length retrieved.
Still, things are not yet resolved, and once more the interaction between complete and incomplete information dictates events.As regards the former, Ammianus reveals that Silvanus possessed a fuller picture than previously known-it happens that through his friends' constant messages he had been informed of Apodemius' efforts all along (15).Despite this information, however, he is still fatally uninformed; he fears Constantius' paranoid mind, completely ignorant of the fact that the emperor had already been told of and accepted Silvanus' innocence.As a result, Silvanus consults another soldier, the Frank Laniogaisus, regarding the possibility of fleeing to the tribes across the Rhine.But even this individual is surer of the situation than his commanding officer and assures him that such a course of action is impossible, for the tribes would either kill him or deliver him to Constantius.Therefore, driven by necessity, Silvanus assumes the purple (16).
Silvanus' rebellion is the product of disinformation combined with wilful misdirection on the part of the conspirators.It is also somewhat ironic (even tragic) that Silvanus decides to rebel even after he had been found innocent of aiming beyond his station.Ordinarily, it might be assumed that the 'not guilty' verdict simply could not be relayed to Silvanus fast enough, but Ammianus seems to discount this possibility with his telescoping of time.Silvanus proclaims himself at section 16 and the very next section has its reporting to the emperor in the present tense (15.5.17): dumque haec aguntur in Galliis, ad occasum inclinato iam die, perfertur Mediolanum insperabilis nuntius, aperte Siluanum demonstrans, dum ex magisterio peditum altius nititur, sollicitato exercitu ad augustum culmen euectum.
And while these things are being driven on in Gaul, as the day was already drawing to its close, an unexpected messenger reaches Milan, openly declaring that Silvanus, striving higher than the command of the infantry, had stirred up his army and risen to the highest majesty.
The distance from Cologne to Milan is just under four hundred miles as the crow flies, though the ancient route would have been considerably longer. 56It would have been impossible for a messenger to make the journey in under three to four days, 57 and yet Ammianus writes as if the announcement occurs late on the same day or very soon after.We might have expected a formulaic 'within a few days' (diebus paucis) or something similar to account for travel time. 58Instead the action in Gaul occurs simultaneously to its announcement to the emperor in Milan (dumque haec aguntur in Galliis).Ammianus compresses the events for dramatic effect.
Importantly, however, this compression also removes from the picture the question of time, which is not subject to human manipulation.Without the justification of time delay, the Silvanus episode becomes not a simple case of unfortunate circumstances exacerbated by the limitations of communication, but instead a wholly man-made construct.In this instance, in the literary world of the RG, communication is instantaneous and information is not disseminated or restricted by natural circumstance but rather by human agency.Indeed, it is interesting that Ammianus has an unexpected messenger (insperabilis nuntius) announcing the news rather than, say, abstract Rumour or the impersonal 'when it was learned'.The chapter itself reveals Ammianus' intentionality in this regard, for as soon as the shadowy operatives have done their work and Silvanus is dead, he reverts to describing the communication of information through impersonal statements.Constantius learns (we know not how) of the rebel's death with inconceivable joy (inaestimabili gaudio re cognita princeps, 15.5.35); and the people in Rome, in a prophetic moment, predict Silvanus' impending doom-'whether excited by some story or by some foreknowledge is uncertain' (incertum relatione quadam percitus an praesagio, 15.5.34).Ammianus, then, makes the revolution of Silvanus the result of clandestine manoeuvrings which keep a tight grip on the flow of information.The truth exists, but it must be actively discerned, like an author interpreting his sources. 56According to A.M. Ramsay, 'The speed of the Roman imperial post', JRS 15 (1925), 60-74, at 63, n. 1, the route between Milan and Cologne comprised several stages: from Milan over the Graian Alps to Lugdunum in Gaul; thence to Argentoratum; thence to Cologne, a distance of 822 Roman miles, which is 752 modern miles.
57 Procop.Arc.30.3-5 describes the Roman courier system in some depth, suggesting that with regular changes of mount, messengers could cover ten days' journey in just one, though he does not state how far a messenger could expect to cover in a day.Tac.Hist.1.12.1 records that in 69 it took several days for news of the mutiny of the legions at Mainz to reach Galba in Rome, via the loyalist governor in Rheims.According to Ramsay (n.56), 65, and others, the couriers must have travelled 155-250 Roman miles a day, depending on how many days one judges paucis post Kalendas Januarius diebus to entail.Even if we adopt the upper estimate of 250 Roman miles per day, though, it would still take a messenger travelling at breakneck speed around four days to reach Milan from Cologne. 58Cf. post paucissimos dies, 17.5.15.See too 15.7.3, 15.8.18, 31.11.1.
2.2: Valentinian and Firmus (28.6, 29.5) Hanaghan has argued that, unlike Julian, Constantius is an especially gullible emperor in the RG, easily led astray and fed 'alternative facts' by his courtiers. 59The Silvanus episode mentioned above is a case in point.Valentinian, however, also struggles for complete information.This is most obvious in the circumstances surrounding the misdeeds of Romanus in Africa (28.6), which leads directly to the revolt of Firmus (29.5) in 373-5. 60It is customary to read Ammianus' account of Firmus as complementary to his narrative of Procopius' rebellion (26.5-9), as yet another example of the cruelty and instability of the Valentiniani. 61There are a number of similarities with the Silvanus episode, however, especially in the circumstances surrounding its causes.As with Silvanus, the misdeeds of an imperial official on the ground complicate the situation such that the emperor is left completely ignorant of events for a long time, and this individual is quickly joined by co-conspirators.Once more Ammianus' methodology informs the action's unfolding.Similar to 15.5, Ammianus introduces a bewildering number of names into his narrative to exacerbate the confusion, even more so than in the previous episode: around twenty-three individuals are named in 28.6 as having some part or other to play in the plot or its attempted suppression.Most of these individuals appear only here.Perhaps Ammianus goes into such depth because he had access to a number of eyewitnesses and wished to do their material justice. 62At the very least, it is interesting that he includes the names of so many incidental characters, because names come in for particular attention in the second preface (26.1.1)as examples of trivial details that do not deserve a place in proper history.Just as with the Silvanus episode, Ammianus focusses on names to render the narrative all the more disorientating while also alluding to his own methodological remarks concerning the quantity of information available.
An outline of the matter shows how the information war plays out on the ground.Romanus, the new governor of Tripolis in 363, refuses to bring help to the provincials, who are under constant barbarian attack (28.6.6).Dismayed, the people of Lepcis send a pair of named individuals to Valentinian ostensibly to hand over accession gifts, but also to inform him of their plight.Romanus acts quickly and manages to get a message to the magister officiorum Remigius, his relative and a 'partner in his thievery' (rapinarum participem, 28.6.8)asking that he ensure investigation of the complaint be assigned to compliant officials.Valentinian, the recently elected emperor, receives the envoys who present a full account of the whole affair (obtulerunt decreta, textum continentia rei totius, 28.6.9).But this account is challenged by Remigius such that the emperor does not know whom to believe, and the latter orders a full investigation (disceptatio plena).Already, then, the quantity of information, presented and required, is at issue, and harks back to Ammianus' own thoughts in the methodological statements.
As the narrative advances, the quality of information becomes relevant.At length, Valentinian sends Palladius to Africa to pay the troops and conduct further investigations; specifically, to scrutinize with suitable trustworthiness what had transpired at Tripolis (et gesta per Tripolim fide congrua scrutaretur, 28.6.12).Palladius witnesses the devastation wrought on Lepcis and seems intent on getting to the truth of the matter (utque ad ueritatis perueniret indaginem, 18).These words echo claims to accuracy made by Ammianus in both prefaces and sphragis, again keeping methodology and action somewhat interlinked. 63Eventually, however, Palladius is also corrupted and (unlike Ammianus), fully committed to the 'impious art of lying' (arte mendaciorum impia, 20), returns to court, reporting that the African provincials had no cause for complaint.Only after Gratian had risen to the throne did he receive trustworthy information, and even here the language focusses on its quality and quantity (28.6.28):doctoque super nefanda fraude Gratiano imperatore fidentius … audiendi sunt missi … aperta confessione cognouit, ipsum suasisse ciuibus suis, ut grauarent mentiendo legatos.haec acta secuta est relatio gestorum pandens plenissimam fidem.
And when the emperor Gratian was given trustworthy information of the abominable deception … they were sent for a hearing … (The judges) learned (from Caecilius' [a minion of Romanus]) open confession that he himself had persuaded his citizens to make trouble for the envoys by false statements.This investigation was followed by a report, which disclosed the fullest confirmation of the acts which had been committed.
The affair is only ended when a proper investigation is completed, and the emperor and his officials win the information war against the plotters.But in order to win the war they must adopt a methodology akin to Ammianus' own.
Firmus has arguments for the defence, but he is not allowed to present them to the emperor, whose officials conceal from him the particulars of the matter with the claim that he did not need to be consulted on unimportant and unnecessary things (minima et superflua).This justification bears a striking resemblance in spirit, if not in word, to the second preface, which declares that it is best 'not to search for the trifling details of unimportant matters' (26.1.1)when writing history.The conspirators, of course, could not be more wrong as to the seriousness of the situation, but it is telling that they, like Ammianus himself, are aware of the dynamic between complete and incomplete information.The only difference is that they use the dynamic to confuse and spread disharmony and, as with Silvanus, Firmus is compelled to rebel in order to preserve his life.
As with Silvanus, too, hostile agents manage to control the flow of information to and from the court (29.5.2).A conspirator on the ground quickly enlists the help of the magister officiorum, and the situation is managed so that one of their own is sent to gather further information; only after great difficulty is the truth finally uncovered.While Silvanus' revolt and death comprise one chapter of Book 15, Ammianus splits up his account of the context surrounding Firmus' attempt and the attempt itself, which comprises the lengthy fifth chapter of Book 29.Though the presentation of the usurpations are different, similar characters are involved.In place of Dynamius and Arbitio stand Romanus and Remigius, introduced earlier (28.6.8).
Once Valentinian finally receives adequate information as to the situation in the province, he sends Theodosius the Elder, a seasoned general, to Africa.Firmus confirms the part played by shadowy conspirators, declaring to Theodosius that it was his treatment by Romanus that led to his insurrection (29.5.8).To be sure, Theodosius himself falls victim to courtly machinations in 375, but this also has obvious parallels to Constantius' sending of Ursicinus (and Ammianus himself) to deal with Silvanus.Indeed, just as Firmus blames shadowy agents for leading him to rebel, so does Silvanus, who even includes Ursicinus, the architect of his impending doom, in the group of those loyal generals so harassed by hostile operatives (15.5.28).But this commonality is not enough to save Silvanus, who is in turn deceived by his erstwhile colleague.Both rebels lose the information war to their accusers, culminating in their losing the actual war to the emperor's generals.

CONCLUSION
In 1953 Auerbach wrote that Ammianus' words 'depict the distorted, gory, and spectral reality of the age'. 64Recent scholarship has examined many facets of this 'spectral reality', suggesting that he practices self-censorship or an endlessly flexible historiography in which even the notion of truth is contingent.To go quite this far is unnecessary.Ammianus does dwell on confusion at points, while alleviating it at others, but this is a consistent dynamic (arising from his own interests) which also informs the deeds of his characters.Ammianus deploys a thoroughgoing technique of revealing quite how pervasive is the struggle between complete and incomplete information, truth and lie.The truth is often hidden behind clouds of obfuscation, and many people succumb in the information war.But (for those that survive) truth's final uncovering is all the more valuable for the struggle, and Ammianus sees himself as an intellectual warrior in this fight, as saving the truth from the silence and lies that would destroy it.

GUY WILLIAMS University of Liverpool
guy.williams@liverpool.ac.uk