The Poet Nemesianus and the Historia Augusta

Abstract Lurking in the Historia Augusta's life of the short-lived Emperor Carus is what appears to be a reference to the genuine contemporary poet Nemesianus and an extant work by him, the Cynegetica. Given the HA's predilection for ‘bogus authors’, this is rather surprising, but because some of what the HA says about Nemesianus is true, the otherwise unique details of his life and works that it provides have been generally accepted. We show first that the reference to the Cynegetica is an incorporated gloss in the text of the HA, one that reveals that the text was being read and studied in northern Francia. We then demonstrate that the name ‘Olympius’, which the HA gives to Nemesianus, is not authentic, offering an analysis of the text's onomastic habits more generally. We show that ‘Olympius Nemesianus’ is one of several invented authors in the HA, lent a superficial plausibility by borrowing the name of a real ancient writer. Finally, we reflect on the way that these conclusions might undermine two developing tendencies in the study of the Historia Augusta.

authentic to the text of the vita Cari and the poet Nemesianus did not have the name Olympius. What we have here is a combination of the HA's predilection for inventing bogus scriptores vaguely modelled after real authors, and of honest medieval scholarship oundering in the face of the work's addiction to bafement.

II TEXT
For some 500 years, every edition of the Historia Augusta has printed ἁλιευτικὰ, κυνηγετικὰ, ναυτικὰ in this passage, three Greek words in Greek characters. It might, then, come as a slight surprise to students of the text that the most important manuscript of the HA, Vat. Pal. lat. 899 (P), produced in the ninth century and of uncertain origin, does not transmit all three in Greek letters. 9 Instead it offers a strange hybrid (f. 214v): nam &cum olympionemesiano contendit quialieutica ΚΥΝHΓΕΤΙΚA &nautica scripsit This peculiar inconsistency is also found in all the manuscripts of the HA derived from P, including Bamberg Msc. Class. 54 (B) (f. 207r) and Paris lat. 5816 (L) (f. 109v). 10 On this evidence, printing these three words in Greek might seem like a logical editorial decision, since all three are Greek words and one of them is written in Greek characters. Yet the reason why every edition since the early sixteenth century does indeed print them in Greek is not sound editorial practice -P, after all, was not widely known for more than a century after the rst printed edition had been producedbut rather sheer textual inertia.
The editio princeps of the HA, which was published at Milan in 1475, did not print any of the words in Greek (or leave them blank to be lled in by hand). 11 Instead, it offered a (clumsy) Latin transliteration (sig. S8r): nam & cum Olympio Nemesiano contendit: qui in halienitica cynegetica. & nautica. scripsit 12 Now this is interesting because the sources of the editio princeps are L (or a derivative of it) and an unknown manuscript belonging to a widely dispersed fourteenth-century family known as Σ, which itself is probably ultimately derived from P with extensive contamination and tampering. 13 The earliest manuscript of that family, Florence Laur. Plut. 20 sin. 6 (D), does not transmit any of the words in Greek (f. 103v): 9 A digital facsimile can be found online at https://digi.ub.uni-heidelberg.de/diglit/bav_pal_lat_899. On the origins of P, see below. 10 Digital facsimiles can be found online at https://bavarikon.de/object/bav:SBB-KHB-00000SBB00000112 and https://gallica.bnf.fr/ark:/12148/btv1b84469323. On the Bamberg manuscript, copied at Fulda in the second quarter of the ninth century, see Bischoff,Katalog I.216. On L, which was annotated by Petrarch, see Malta 2014, especially n. 1. 11 Gesamtkatalog M44203; ISTC is00340000, prepared by Bonus Accursius and printed by Philippus de Lavagna in two parts (20 July: Suetonius; 22 December: the Historia Augusta with Eutropius and Paul the Deacon, Historia Romana). 12 The u/n confusion in halienitica suggests that the compositor could not clearly read what was in front of him. 13 On the editio princeps, see Stover 2020a: 172-3. The same study adduces evidence for Σ's ultimate descent from P, including misunderstood transposition marks and sutured intratextual lacunae. Valentini 2021 has restated the case for Σ's independence, without responding to these specic examples. The evidence she brings forward, with regard to textual arrangement and individual readings, can be equally explained as contamination from the non-P source, and does not disprove derivation from P. The fact that she brings up an entirely unexceptional case of an archetypal line (47 characters long) omitted in P and supplied in the margin by the corrector (f. 199r) with a tie-mark indicating the proper position after cepit imperium, correctly reinserted by B (f. 193r), as a 'frutto di trasmissione orizzontale derivante da una collazione con Σ' (223)  For this reason, it seems very likely that the source for the Latin transliteration of the cynegetica in the ed. pr. is its (unidentied) Σ source. The second edition of the HA was printed at Venice in 1489: as has recently been demonstrated, it is extremely valuable because it is based on the ed. pr. and a separate tradition of the text, independent of P and transmitting authentic information. 15 Unfortunately, for this passage, it merely reproduces the reading from the ed. pr., including the gross error in halienitica (sig. M6r).
The next edition, printed roughly 500 years ago, was (unlike its two predecessors) prepared by a respected humanist scholar with a critical eye for philological matters, Johannes Baptista Egnatius, and issued from the venerable house of Manutius in Venice in 1516. The basis for his text was the earlier Venetian edition of 1489, but Egnatius also had occasional recourse to yet another Σ manuscript, which was housed in the Marciana in Venice. 16 More to the point, however, Egnatius was an adventurous editor, never hesitant to make radical changes to the transmitted text, where sense, order, or style demanded it. 17 For example, he separated the lives of Caesars and usurpers from the main sequence of Augusti and printed them after the rest of the text, to make the rst part read more like a normal imperial history, treating just the emperors in sequence.
He was in the habit of using a well-known bon-mot on thieves, indeed one in Greek, which means this in Latin: 'if he who has stolen a lot gives a little to his advocates, he will be acquitted', which in Greek is as follows: 'He who has stolen a lot and given a little will be acquitted'. P did not transmit the Greek in Greek characters, but instead offered a curious Latin transliteration opolla clepsas oliga dus ecfeuxente (f. 107r). The rst two printed editions just left a blank space where the Greek should be. This was not good enough for Egnatius, but his manuscript probably offered him no aid. 18 Instead, he boldly inserted his own translation: ὅστις ἄν πολλὰ κεκλόπημεν [sic] ὀλίγα δὲ τοῖς βοηθοῦσιν ἔδωκεν σῶος ἔσται (sig. q4r, f. 124r). 19 Hence it should come as no surprise that when Egnatius came to Carus 11.2, he chose with no manuscript authority to print all three titles in Greek (sig. 2g5v): nam & cum Olympio Nemesiano contendit, qui ἁλιευτικὰ κυνηγετικὰ & ναυτικὰ scripsit characters, not including of course the sauts du même au même on ff. 23v, 79r, 91v, and 196v. See below for two further cases where derivation from P looks most plausible. 14 A digital facsimile can be found online at http://mss.bmlonline.it/s.aspx?Id=AWOMq8cMI1A4r7GxMYFT.
Every edition since, down to the Budé published by Paschoud in 2002, has followed suit. Hence, we arrive at the present paradoxical state of affairs: every edition prints κυνηγετικὰ not because P transmits ΚΥΝHΓΕΤΙΚA, but rather because of the bold liberties Egnatius took ve centuries ago with his text.
As a result, no one has ever thought to question why P reads ΚΥΝHΓΕΤΙΚA. It is, in every respect, anomalous. The word stands out on the page: it is written in much larger letters than the surrounding Latin characters and the letter-forms are awkward and laboured. And no wonder, for they are, in fact, the only Greek characters written in the whole manuscript. In every other instance where the HA offers Greek, P transmits a Latin transliteration, as in the passage of Alexander Severus discussed above, or at Pertinax 13.5, where it offers the awkward christologum (f. 49v). 20 By chance, this passage survives in one of the few fragments from the other ninth-century manuscript of the HA written at Murbach (M), preserved in a list of collations included in the 1518 Froben edition of the text published at Basel: it has instead the (misspelled) χρησόλογος. 21 In addition, according to Froben, the Greek of the saying in the life of Alexander discussed above was found in M. 22 P's treatment of Greek is also generally anomalous. It is not uncommon for individual Greek words to be written in Latin characters by medieval scribes and indeed, in many cases, such transliteration may go back to the original author. 23 Writing out a whole line in Latin transliteration is, however, much less common and usually restricted to texts which were meant to be publicly recited in a liturgical, monastic, or pedagogical context. Extant examples include prayers like the Pater noster and Greek creeds. 24 These transliterations almost invariably reect contemporary pronunciation of Greek. 25  Murbach manuscript containing his commentary on Isaiah. 30 Paris lat. 10910 (Fredegar), of the early eighth century, features a striking drawing of Eusebius and Jerome (with attendant goose) on f. 23v. There is a Latin caption written in Greek letters underneath them, which a ninth-century hand has transcribed. 31 An alternation between writing in Greek characters and transliterating to Latin is also found in a glossary (which shows an interest in late ancient historical texts) which is plausibly associated with Saint-Denis and its abbot Hilduin . 32 In all these cases too, there are hints of roughly contemporary pronunciation, Greek and Latin. 33 This may tell us something about the Historia Augusta. Three of our ninth-century witnesses -P, Π, and Mare derived from the same archetype. The natural explanation of their divergence is that that archetype offered Greek text for the (surprisingly few) points where the HA transmits Greek, with a Latin transliteration above the line. The conscientious scribe of M copied out the Greek, where the slightly hastier copyist of P generally offered the Latin.
What we have here, then, are three independent anomalies. One is the fact that the HA, in an utterly uncharacteristic fashion for one of the later lives, appears to transmit genuine literary lore in saying that Nemesianus wrote a Cynegetica. The second is that P puts this work's title into Greek characters, while transmitting the titles of the other two (invented) works in Latin letters. The third is that everywhere else P uses Latin transliteration for Greek, even for extended quotation, in a manner that is essentially unparalleled for the genre. Any one of these individually could be explained away through special pleading, but their conjunction exceeds the limits of credibility. One hypothesis, however, could explain all three: what if ΚΥΝHΓΕΤΙΚA is an incorporated gloss?
Nemesianus' Cynegetica is a rare text, but we have hints that it was not quite so rare in the early ninth century. 34 In the course of a lengthy denunciation of his homonymous nephew, the bishop of Laon, Hincmar, archbishop of Reims from 845 to 882, tells us that he had read the work as a schoolboy: … aliter respondere non potui, nisi, ut venatores ferae lustra sequentes agere auditu et lectione puer scolarius in libro, qui inscribitur Kynegeticon, Cartaginensis Aurelii didici, hac illaque discurrendo, retrograda etiam vestigia repetendo anfractus tuos vestigando explicare studerem. 35 … I could not otherwise answer, except that I should hasten to unravel your circumlocutions by running about, also seeking out your backwards-hastening footsteps, and investigating them, as hunters do when seeking the lairs of wild beasts, which I learnt from listening and reading when a schoolboy in the book which is entitled Cynegeticon, written by Aurelius of Carthage.
A few lines later, he goes on to quote the poem's opening: On the connection to Hilduin, see Lapidge 2017: 770-1; Cinato ( per litteras) has noted that there are also palaeographic features in the manuscript that would plausibly connect it to the region of Paris. 33 Note for example the intervocalic 'g' in pogeian; this is the phenomenon that gave rise to such widely distributed forms as Apulegius in the Middle Ages (for the similar process by which intervocalic 'g' was omitted, see Stotz 1996Stotz -2004 … and since I am even compelled, approaching your new subscription, or I should say your new presumption, to undergo with a pure heart personal quarrels: 'their thousand ways and mournful toil; their swift running hither and thither, and the battles of the untroubled countryside'. Hincmar of Reims is one of the Carolingian authors about whose life we know enough to understand exactly what he is saying in this reminiscence: he had read the Cynegetica during his education at Saint-Denis in Paris in the 810s, or perhaps even the early 820s, studying under Hilduin. 37 Somewhat disappointingly, the extant ninth-century manuscript of the Cynegetica (written in northern France, probably at Saint-Denis itself), Paris lat. 7561, does not transmit the name of the work in Greek characters (p. 18): The rest of the slim manuscript tradition follows suit. 39 If, however, we take a closer look at Hincmar, we nd him referring to the text in a garbled combination of Greek and Latin letters, at least according to the manuscript of his Against Hincmar of Laon produced at Reims during his lifetime (Paris lat. 2865, f. 115r), KYNEGETICON. 40 Evidently, the text of Nemesianus that Hincmar had read did have a title in Greek, or at the least, in his school-days there was a fashion for referring to it in Greek. This was rather a concession to Hellenism on his part, since elsewhere in the same work he fulminated against his namesake: Although there are sufcient and adequate Latin words, which you could have put in those places, where instead you have put Greek and obscure ones, now and then even Irish and other barbarous, as you fancied, bastardised and corrupt things. 41 It is a striking fact that the only two ninth-century mentions of Nemesianus' work outside the actual manuscript that transmits it spell the title KYN-. Chronologically speaking, Hincmar's halcyon schooldays take us awfully close to the period in which P itself was copied, probably sometime between 825 and 830. 42 In this context, it seems probable that in P's archetype somebody inserted the Greek word ΚΥΝHΓΕΤΙΚA in the margin next to Nemesianus' name, because they recognised the name of the poet. Greek writing was popular in medieval scholia: a Venn diagram 36 Cf. Nemes., Cyn. 1-2, which has hilaresque rather than tristesque. As de Gianni 2011 has argued, this is a deliberate alteration of the text by Hincmar to suit his polemical purpose. 37 On Hincmar's life in general, see Stone 2015; and on his early life, see Devisse 1975Devisse -1976Devisse , ii.1089 There is no title in this manuscript. On its provenance, see Bischoff 42 Pecere 1995: 337 dates it 'all'inizio del secondo quarto del secolo IX'. This is partially based on Bischoff's palaeographic dating to the second quarter of the ninth century, with the external constraints of the date of B, which Pecere wants to make before 842, in the abbacy of Hrabanus at Fulda, and the De rectoribus Christianis of Sedulius Scottus, which reects his reading of P (see Dorfbauer 2020). of pedants who enjoyed putting their rudimentary Greek skills on display and people who wrote glosses would have a very substantial overlap. 43 This would have produced a mise-en-page something like this 44 : fuisse praedicatur utomnes po&as suitemporis vicerit. nam &cum olympionemesiano contendit quialieutica &nautica scripsit quinque omnibus coloniis inlustratus ΚΥΝHΓΕΤΙΚA P's copyist blithely incorporated the gloss in the textas indeed he was wont to do on any number of other occasions. 45 Indeed, in one case on the culinary predilections of Aelius ( pernam at Ael. 5.4), we have good evidence that P and M incorporated an interlinear gloss at two different places in the line. 46 In another, discussing the literary output of the Emperor Trajan, P transmits nam et de suis dilectis multa versibus composuit amatoria carmina scripsit. (Hadr. 15.9). Casaubon judiciously deleted amatoria carmina scripsit as an incorporated gloss. In the same passage, Π transmits amatoria carmina versibus composuit (f. 142v, shamefully not recorded in any edition since Gruter). The most reasonable interpretation of this passage is that amatoria carmina was an interlinear gloss on de suis dilectis, which displaced the original in Π and was incorporated into P with the addition of scripsit for sense. If ΚΥΝHΓΕΤΙΚA were a gloss like these, it would not have been equipped with the usual transliteration above the line, and the scribe would have had actually to copy the characters as he saw them, hence why this is the sole word in Greek characters found in P. 47 From a conventional standpoint, this is a somewhat perverse suggestion: the normal process of editing seeks to purge errors that have crept into the text over the centuries of its transmission, not remove truths. The nature of the HA, however, forces a reversal: because the text, particularly in its later lives, is so determinedly ctive, we ought to be very suspicious of that in it which seems to be true, especially with regard to literary matters. The 'truth' of the HA is generally falsehood and invention and what is true in the usual sense of the term is normally error in the HA. What, indeed, are the chances that, of the literally dozens of invented authors and facts about literary history in the later lives, just one happens to be true? Given that we know Nemesianus was being read in one school, at least, and indeed recalled fondly by a former schoolboy, around the time that P was being copied, it seems far more likely that a Carolingian reader was telling the truth, and that the HA was engaged in its customary obfuscation. At any rate, the fact that the title is written in Greek characters in P, against its otherwise universal practice, demands some explanation, and one that respects what we know of ninth-century scribal habit.
If this hypothesis is true, it gives us a tantalising hint about the much-discussed question of the Schriftheimat of P, and where the archetype of the HA was preserved. The old idea that P was written in Italy has recently been discredited: indeed, the fact that its two siblings, M and Π, are both northern (the former written at Murbach, the latter at Lorsch) makes it virtually impossible that P was copied from the same archetype close to the same time in Italy. 48 Perhaps instead we ought to look for the sort of centre which had the tradition of Greek study that might have inspired transliteration, and one in which Nemesianus' Cynegetica was used in schools. In other words, Saint-Denis itself, or a place closely associated with it.

III NAMES
The one obviously authentic detail about Nemesianus provided by the HA may very well be an interpolation. That has considerable implications for its other claims for him. Despite occasional attempts to identify the Halieutica and Nautica with extant Latin poems, it should now be even clearer that they are pure inventions, designed to pad the résumé of the poet. 49 The name 'Olympius' has, however, achieved broader acceptance. It was enshrined in the second edition of the Prosopographia Imperii Romani and in The Prosopography of the Later Roman Empire. 50 Similarly, both Syme and Paschoud were inclined to accept it. 51 From these authoritative sources, 'Olympius' has bled into other more general works of reference and is now comfortably accommodated in the broader scholarly literaturenot a bad career for a datum from one of antiquity's least reliable informants. 52 Clearly, it requires more systematic treatment.
In the late third and fourth centuries A.D., the onomastic landscape of the Roman world had changed radically from that of the early empire. 53 Individuals were generally referred to by a single nameinvariably their last if they had more than onewhich scholarship has come to call the 'diacritic'. Trawling through the letters of Symmachus, for example, illustrates quite how ubiquitous this system of reference had become in even the stufest of circles. 54 In ofcial usage, e.g. consular dates in papyri, the diacritic was often supplemented by one more name, always placed before it. 55 Depending on the bearer's background and the context, this additional name might indicate descentusually a gentilicium, but sometimes also a cognomen that had come to play a similar roleor status: Flavius, the nomen of the emperor Constantine, is the obvious example. 56 Much more rarely, in elevated literary or epigraphic contexts, we nd individuals with four, ve, or even more names, generally gentilicia but occasionally even praenomina as well, 48 See Dorfbauer 2020; and Stover 2020b. 49 For an overview of earlier efforts to identify extant poems as (fragments of) these, see Volpilhac 1975, 11-12. Verdière 1974 made a quixotic attempt to revive the identication of Anthologia Latina 718 (Riese 2 ), the Ad Oceanum and AL 720 (Riese 2 ), the Ponticon, as part of the Nautica. This was briskly dismissed by Smolak 1993: 358, who correctly concluded: 'il est douteux qu'ils aient jamais existé'. For a more realistic estimate of the Ad Oceanum, see Canal 2013. 50 PIR 2 A.1562 (Stein); PLRE I 'Nemesianus 2'. Somewhat surprisingly, they both also gingerly admitted the ἁλιευτικά and ναυτικά to historical reality with a parenthetical question mark. 51 Syme 1971: 279 and n. 2, 'there is no reason to doubt the "Olympius"'. Paschoud 2002: 366: 'Olympius n'est donné que par le présent passage de l'HA, mais il n'y a pas de raison sérieuse de mettre en doute l'authenticité de cette partie de son nom'. 52 Smolak 1993;Scoureld 2012;Uden 2018. In the literature, see e.g. Chastagnol 1976;Küppers 1987;Altmayer 2014: 27-8 it is noteworthy that Jakobi 2014 omits it. 53 On late antique names, see Cameron 1985 and(especially) Salway 1994. 54 Most easily done through the helpful index nominum in Seeck 1883: 342-52, where the names actually used by Symmachus are in small caps (a very useful practice which ought to be more widely adopted by editors). 55 This is most easily grasped by looking at the entries in 'Appendix D' of Bagnall and Worp 2004. Procedure varied for emperors, but for private individuals the dual name is the normal pattern. Consular dating became regular in Egypt only from the Tetrarchic period onwards (Bagnall and Worp 1979: 282;1982). See in general also the important article by Salway 2008: 280-5. 56 See Cameron 1988. something that hints that the 'diacritic system' conceals the full richness of late ancient onomastics.
The manuscripts of the Cynegetica call the poet M(arcus) Aurelius Nemesianus. 57 That was clearly true also of the text Hincmar had read, since he refers to him as Aurelii. 58 The archetype of those manuscripts of the Eclogues which attribute the poems correctly seem to have called their author Aurelianus Nemesianus, Aurelianus an easy slip from Aurelius by dittography. 59 There is nothing in the rich tradition of Nemesianus' workdirect or indirectthat suggests he had any other names but these. 60 Moreover, the names transmitted by the manuscripts of Nemesianus are entirely compatible with what we know otherwise of late antique onomastics and obviously authentic. Nemesianusa cognomen derived from the Greek names Nemesis or Nemesiosis a rare but attested diacritic in the Latin-speaking regions of the Roman world. 61 Given the poet's association with Carthage, it is perhaps particularly interesting that the bishop of Thubunae in Numidia in the time of St Cyprian was one Nemesianus. 62 He was still being commemorated as a martyr in the middle of the fourth century and is mentioned (indeed quoted) by St Augustine. 63 The gentilicium 'Aurelius', sometimes with its attendant praenomen 'M(arcus)', was extremely common in the later Roman empire, including in Africa. 64 That was because it had been widely taken by those whom the constitutio Antoniniana of A.D. 212 had enfranchised, to commemorate their benefactor, 57 For Paris lat. 7561, see above, Section II. Paris lat. 4839 (f. 20r) has MAURELII MENESINI KATAGINENSIS / CYNEGETICON. Vienna 3261 (f. 48r) has M. AURELII NEMESIANI / CARTHAGINENSIS / CYNEGETICON (and a similar notice at the end of the poem). 58 This, a gentilicium, would not on its own have been the correct way to refer to someone in Late Antiquity, but it does usefully conrm the testimony of the manuscripts. 59 The transmission of Nemesianus' Eclogues, separate from the Cynegetica, is complex and bound up with that of Calpurnius Siculus, to whom alone the second family of MSS attributes them (see Reeve 1983 -Paris lat. 17903 f. 74r attributes the extracts of the poems it contains to one Scalpurius, presumably a mangling of Calpurniusark:/12148/btv1b52500967c). The two extant manuscripts of the rst family, however, Florence, BML plut. 90 inf.12 and Naples V.A.8, call the poet Aurelianus Nemesianus (see e.g. Williams 1986: 9-10). The lost manuscript, which Niccolo Angeli saw, belonging to Taddeo Ugoleto seems to have done the same (Volpilhac 1975: 34). The twelfth-century catalogue from Prüfening mentions bucolica Aureliani (Manitius 1935: 120 65 Generally, though of course not in every individual instance, the Aurelii were the ordinary inhabitants of the Roman Empire, the broad mass of the population who had not achieved Roman citizenship before the early third century. 66 It is likely that (say) Nemesianus' grandfather was one of those suddenly elevated in status by Caracalla. This would make the poet, with his aristocratic interest in the chase and his transparently sophisticated literary culture, an interesting example of the rise of these 'new Romans' and their descendants to positions of power and privilege.
There is no particular reason to sully this coherent picture on the testimony of the HA alone. Given that it is a text famous for its made-up names, the economical solution is that this too has simply been invented. To instead accept that Nemesianus had the name 'Olympius', there are two routes that we could take, neither of which has much to recommend it. One option, canvassed occasionally in the early modern period, is that it was a genuine additional family name of Nemesianus. 67 There are two major reasons to reject this. First, it is difcult to understand how it would have been omitted from an archetype that was formal enough to include the poet's praenomen (a rare item indeed by Late Antiquity). 68 Second, in Late Antiquity Olympius was a common diacritic, but only very rarely used as an additional (familial) name, perhaps precisely because it was so common and thus not usefully distinctive. 69 It seems unwise to accept something that is inherently unlikely because the HA claims it. The alternative explanation of Olympius, supercially more plausible and adopted by both PIR 2 and PLRE I, is to treat it as a signum. 70 Signa were nicknames (a category sometimes referred to as supernomina), ending in -ius, used fairly widely in the third and fourth centuries, generally in high-status and literary contexts. 71 They could replace the diacritic: hence  Cagnat et al. 1923) published in the same volume (it seems unlikely given the very low number of individuals they nd overall). ILT, however, offers only a selective re-edition of and supplement to CIL VIII and its various supplements (ILAfr publishes texts not included in those supplements): it is not a comprehensive or representative sample of North African inscriptions (hence why Kracker and Scholz 2012, table 2, found only 258 individuals epigraphically attested in the region, under half the number they nd for Britain!). In Salway 1994: 134 and n. 59, the statement that Aurelii account for 23 per cent of the nomina in the Christian inscriptions of Rome and Carthage appears to be a misreading of Kajanto 1963: 16, where the gures relate only to Rome. The nomen Aurelius was certainly very common in North Africa and its distribution would repay further study, though that would require that the Christian inscriptions be taken properly into account. 65 See the brief but perspicacious summary of Salway 1994: 133-6 and the more substantial overviews by Buraselis 2007: 94-120;Rizakis 2011;Besson 2020: 75-104. The important paper by Blanco-Pérez 2016 is of particular signicance for M. Aurelii as a third-century phenomenon. 66 The social status of the Aurelii was demonstrated for Egypt, the region where we can most condently be more than impressionistic, in two classic studies by Keenan 1973; In North Africa, their relatively humble position is perhaps best illustrated by the fact that in the Timgad Album, which lists members of the city's council in the 360s, there are a mere four Aurelii amongst the 204 individuals who have a gentilicium and one of those is a cleric (Chastagnol 1978: 49, 53, 94). There were evidently not many local landowners with the name. For the view that the overwhelming majority of the Empire's inhabitants were not enfranchised before 212, see the compelling paper by Lavan 2016 (cf. Lavan 2019, showing the limited numerical impact of enfranchisement through military service). 67 See Volpilhac 1975: 7-8, who was rightly sceptical. 68 On the decline of praenomina, see Salomies 1987: 390-413. 69 PLRE I registers 18 bearers of it, with only two individuals using it before their diacritic. Of these, Tamesius Olympius 'Augentius 1' is a slightly doubtful case, as the name Olympii is actually detached on the inscription (ILS 4269) that attests it. OPEL III.112 does not register it as a nomen. 70 Though it has not often been noted, Olympius is an attested signum: Kajanto 1967: 86. In contrast to PIR 2 , PLRE I does not explicitly register Olympius as Nemesianus' signum, but its placement of it after his diacritic indicates as much. 71 On signa, see Woudhuysen 2019 and the literature cited therein. The terminology is in some senses unsatisfactory. Later Romans used signum to indicate some additional names (mostly but by no means always the writer called Firmianus, generally known to us by his signum Lactantius. They could also be combined with it as a dual name, and the order signum + diacritic is attested. 72 While sometimes ubiquitous, they were often used only occasionally: L. Aurelius Avianius Symmachus, father of the famous orator, had the signum 'Phosphorius', but though he is unusually well attested, the nickname is known to us only from a single inscription. 73 The rather slippery nature of signa is perhaps why the idea of 'Olympius' Nemesianus has seemed plausible. To see why it should be rejected, we must briey consider the way that the HA used names. The HA is justly famous for its fraudulent inventiveness when it comes to onomastics, but focus on this has perhaps led to a neglect of the more prosaic but important subject of the way that it uses names. 74 This is not the place for a comprehensive investigation: this would be a major work in itself, and one complicated by the fact that so many of the individuals named are invented. Nevertheless, a few important onomastic habits of the HA can usefully be picked out, seen most clearly when set against what was otherwise standard practice. Most Latin historians of the fourth century referred to the vast majority of individuals by a single name, their diacritic. They particularly avoided referring to gures of the third and fourth centuries by more than one name, though they were slightly looser with those of earlier periods, perhaps because polyonomy had an antique avour. They were especially parsimonious with the (to them) traditional combination of gentilicium and cognomen, usual in the works of Tacitus or the letters of Pliny (for example). 75 In the Res Gestae, more than 90 per cent of the men named by Ammianus are given only a single name: of the remainder, a considerable proportion are gures of Roman antiquity, whose (usually) dual names were hallowed by long usage. 76 Amongst actors in the narrative, the majority of those referred to by two names receive a diacritic and a nickname. 77 Only a handful are identied by gentilicium and cognomen, and Ammianus had a partiality (perhaps rhythmical) for reversing their usual order. 78 Ammianus was an idiosyncratic author, but in his use of names he was remarkably ordinary. Festus shows a very similar pattern in his account of the Roman Empire's dealings with the Parthians and Persians (Breviarium 19-29). He names 33 individuals: only six of these (18 per cent), none later than the early third century, have more than one name and only two of those are referred to by a ending -ius), but most of what we now think of as signa are not introduced by any formula at all, nor is there much explicit evidence for what they were called. Yet supernomina terminating in -ius are a clearly visible category in our evidence (see Kajanto 1967, 52-4). 72 Woudhuysen 2019: 851-2. 73 PLRE I 'Symmachus 3'. The inscription is ILS 1275. In contrast, the praetorian prefect of the 360s Saturninius Secundus signo Salutius is more often called Salutius than Secundus by the sources (PLRE I 'Secundus 3'). 74 For 'bogus names', see, canonically, Syme 1966. As Burgersdijk 2016 has already noted, an onomasticon of the bogus names in the HA is a major desideratumthe closest work that currently exists is Domaszewski 1918, which deserves to be taken more seriously than it perhaps has, but is now seriously dated. 75 1.7). See also Rusticus Iulianus (27.6.1). The reversal may also speak to the inuence of Tacitus, who occasionally did the same (Goodyear 1972: 148). In general, on the phenomenon of cognomen + nomen, see Shackleton Bailey 1965: 402-3. gentilicium and cognomen. 79 There is a similar pattern in the tenth book of Eutropius' breviarium, which covers 305-364. There are twenty-seven named individuals in the book. Of these, only three (11 per cent) have more than one name, only one of those is contemporary with the events described, and only one (much earlier) individual is given a gentilicium. 80 These were not habits that had suddenly emerged in the middle of the fourth century and thus likely to be avoided by any conscientious forger seeking to pretend their work was written in the era of Diocletian and Constantine. Our largest corpora of secular Latin prose from that period are to be found in the Mathesis of Firmicus Maternus and the nine contemporary speeches of the Panegyrici Latini. 81 Firmicus mentions some forty-ve historical individuals in the course of his lengthy astrological treatise. 82 Of these, only two are referred to by more than one name: Germanicus appears as Julius Caesar and Cicero is called Marcus Tullius. 83 Firmicus always refers to contemporaries by a single name, even when he varies which of their names he chooses: his patron, for example, is either Mavortius (his signum) or Lollianus (his diacritic), but never both together. 84 The Panegyrici name some forty individuals from Roman history. 85 Of these, a mere ve (12.5 per cent) receive more than one name: none of those ve is later than the rst century B.C. and only two of them are given a gentilicium. 86 We are not exactly over-endowed with Latin historiography from this period, but we do have Lactantius' On the Deaths of the Persecutors (c. 315). Lactantius was interested in names, sensitive (for instance) to the fact that Diocletian had changed his (9.11). Yet, while he names some thirty-nine individuals in his narrative, only two (5 per cent) can really be said to receive more than one name: Tiberius Caesar (2.1, the praenomen is written out in full in the only manuscript) and Maximianus qui est dictus Herculius (8.1). Lactantius even curtly refers to Tarquinius Superbus as 'Superbus' (28.4). Clearly, the diacritic system was already well established for Latin authors of the late third and early fourth centuries.
In contrast to all this, the HA delighted in multiple names and especially in fathering them on (putatively) contemporary gures. In the Quadriga Tyrannorum, for example, it names some fty-seven individuals. This is one of the shorter lives (c. 2,300 words) and gives some avour of the sheer number of named individuals that the text ings at its readers (contrast Lactantius' sparseness in the c. 11,000 words of the DMP  1, 6, 5.praef.1, 1.1, 7.1, 6.1.1, 1.10 ,22.1,28.2,31.26,37,32.1,33.1,40.1,7.1.2,26.12,8.1.1,6,4.14. Lollianus: 1.proem.8,3.3,10.1,15,2.29.20,3.proem.2,4.proem.3,5.1.38. It is rather striking that (outside the rst book, where both are found), Lollianus predominates in the earlier books and Mavortius in the later. On the relationship between Firmicus and Mavortius, see Woudhuysen 2018. 85 They include the names of many mythical gures and divinities as well, invariably referred to by a single name. Our count of thirty-nine includes (e.g.) Perseus, the king of Macedonia, but not (for instance) Romulus or Remus. 86 Julius Caesar is named as C. Caesar at Pan. Lat. 12.6.1. Fulvius Nobilior (the victor over the Aetolians in 189 B.C.: see Nixon and Saylor Rodgers 1994: 159 and n. 1) appears as Fulvius ille nobilior at 9.7.3, Q. Maximus (Cunctator) at 12.15.5. Pompey the great appears as Pompeiumque Magnum at 7.5.2 and Cn. Pompeius at 12.8.1. Scipio Africanus the elder is P. Scipio at 7.13.5 and 10.8.1. these, twelve (21 per cent) have more than one name. 87 The majority of those (eight) are made out to be contemporary with the author or with the events narrated: they are mostly putative sources or informants. 88 Most (nine) of those referred to by more than one name have a gentilicium. In fact, the HA's author was something of a connoisseur of gentilicia, deploying some very rare ones indeed: Aurunculeius, Larcius, Masticius, Verconius. 89 All this suggests that when confronted by the need to invent gures, his usual practice was to give them a gentilicium and a cognomen. 90 This is obvious, for example, from the names of the six scriptores to whom the lives are attributed: Aelius Spartianus, Iulius Capitolinus, Vulcacius Gallicanus, Aelius Lampridius, Trebellius Pollio and Flavius Vopiscus. 91 So, confronted by the 'Olympius Nemesianus', our default assumption ought to be that the name is supposed to mirror the usual pattern of nomen + cognomen, just as the invented Aurelius Apollinaris in the same passage does, than which it is no more authentic. Nothing licenses us to make the rationalising assumption that this just happens to be a disguised signum.
In fact, we can go somewhat further than this. Rather unsurprisingly, the HA had a fondness for nicknames. Yet its author did not generally introduce them to the reader without some ourish (where was the fun in that?). Instead, he tended explicitly to signal that they were nicknames, by using some formula: cognomine, appellatus est, cognominatus, cognomento, etc. 92 He also delighted in offering (sometimes elaborate) explanations for them. So, we learn that Gordian I was nicknamed 'Africanus' not because he was proclaimed emperor in Africa, but because he was descended from the Scipios. The HA's author further informs us that 'in very many books' he has found that Gordian and his son were also both nicknamed Antoninus or Antonius. 93 We get similarly elaborate explanations for why Antoninus was 'Pius', why Septimius Severus was nicknamed Pertinax, and why Aurelian mockingly suggested the Senate call him carpisculum, amongst many other instances. 94 Signa were, of course, a kind of nickname and crucially the HA's author showed exactly the same desire to highlight and explain them as he did for other sorts of supernomina. 95 In fact, the HA provides one of our very few explicit ancient discussions of how someone gained their signum, in its account of the reign of Commodus: Menses quoque in honorem eius pro Augusto Commodum, pro Septembri Herculem, pro Octobri Invictum, pro Novembri Exsuperatorium, pro Decembri Amazonium ex signo ipsius adulatores vocabant. Amazonius autem vocatus est ex amore concubinae suae Marciae, quam pictam in Amazone diligebat, propter quam et ipse Amazonico habitu in harenam Romanam procedere voluit. 96 His atterers even renamed the months in his honour: 'Commodus' for August, 'Hercules' for September, 'Invictus' for October, 'Exsuperatorius' for November, 'Amazonius' for December, from his signum. He was called Amazonius, moreover, due to his love of his concubine Marcia, whom he delighted to see painted as an Amazon and on whose behalf he himself wanted to enter the arena at Rome in the dress of an Amazon.
The HA's author did not understand signum narrowly to be a nickname in -ius (as we would generally dene it today). 97 He refers, for instance, to the emperor Aurelian as receiving in his army days the signum of manu ad ferrum to distinguish him from another tribune, also called Aurelian. 98 He also says that 'Antoninus' was the verum signum of the emperor Antoninus Pius, by which he must mean his 'real name'. 99 Yet the passage shows very clearly that what we would call a signum was not something to be introduced without some fanfare. The (very few) possible signa in the HA are all either brought up in discussions of nicknames, or explicitly marked as such. 100 So, we are told that Lucius Verus had on his staff an actor called Agrippus, cui cognomentum 93 Gordiani tres 9.3-5. 94 Hadrianus 24.3-4; Severus 17.6; Aurelianus 30.4-5. 95 Like much else in the HA's use of names, this is in contrast to other Latin historical texts: Ammianus and Eutropius (see above) both use signa without much indication that they are nicknames. 96 Commodus 11.8-9, cf. Clodius Albinus 2.4. 97 See Kajanto 1967: 52-3. 98 Aurelian 6.2. 99 Pescennius Niger 8.5, cf. Gordiani Tres 4.8 (though we might understand Antoninus to be a nickname of the Gordian in question). 100 e.g. paenularius in Diadumenus 2.8 (in a discussion of nicknames). Single names ending in -ius in the HA, which might be thought signa, are mostly attested elsewhere as names, usually Greek or gentilicia (which, as noted above, n. 89, the HA did occasionally use in isolation). A sample of the more interesting: Acholius (Alexander Severus 14.6, 48.7, 64.5; Aurelianus 12.3; a Greek name: LGPN lists ve instances); Antistius (Aurelianus 50.3, a nomen: OPEL I.60); Cecropius (Gallieni duo 14.4, 7, 9; Probus 22.3; Cecropia is registered once as a cognomen by OPEL II.46, but add a Cecropius in CIL 6.30839; a Cecropius was bishop of Nicomedia in the middle of the fourth century, Athanasius, Historia Arianorum 75.1, and another, bishop of Sebastopolis, attended the council of Chalcedon in 451); Cordius (Heliogabalus 6.3, 12.1, 15.2; a nomen, but attested as a cognomen: OPEL II.75); Encolpius (Alexander Severus 17.1, 48.7; OPEL II.117 lists only one instance but add CIL 6. 30810, 15963, 16588, 17159, 18048, 22951, 37322, 11.7249, 12.242; many for slaves or freedmen); Eugamius (Maximini duo 27.5, identied as a Greek rhetor, LGPN registers three women with the name Eugamia); Theoclius (Aurelianus 6.4; not attested but Theocles was a very common Greek name, with 191 instances in the LGPN); Toxotius (Maximini Duo 27.6, perhaps inspired by the senatorial Toxotii of the fourth century: PLRE I: 921); Tynchanius (Gordiani tres 14.7; a Tynchanius who was bishop of Apollonias featured prominently in the council of Chalcedon). Palfuerius, the Isaurian brigand (Probus 16.4) is erat Mem, informed that Clodius Albinus was nicknamed Porfyrius by his nurse because he was swaddled in purple, and introduced to Aurelius Victor, cui Pinio cognomen erat. 101 In other words, were the HA trying to lumber Nemesianus with the signum Olympius, everything about its practice elsewhere suggests that it would tell us that was what it was doing, not leave us to work it out. We can conclude, with considerable condence, that the HA's 'Olympius Nemesianus' is not actually meant to be a genuine reference to M. Aurelius Nemesianus. Perhaps the name was just a weak pun, inspired by that more famous competition at Olympia, on the idea of Numerian's poetic contests? In any case, if Olympius is not actually part of the historical Nemesianus' name, there is even more reason to suppose that ΚΥΝHΓΕΤΙΚA is an interloper in the text.
IV SCHOLIA There is one nal piece to this onomastic puzzle. Anyone who consults PLRE I or PIR 2 will be told that Nemesianus is referred to as Olympius in the ancient scholia to Statius' Thebaid. At rst sight, this seems powerful supporting evidence for the HA, one reason (perhaps) why scholars have generally taken the text at its word. In fact, however, this is a remarkable example of the durability of misguided early modern ideas. The scholia do not refer to Nemesianus by name. In a comment on Thebaid 5.388-9, they quote a poet called Olympius: AB IUNCTIS ergo ὑφ' ἕν noli accipere. OPERTA autem pinus pellibus, ut mos est. ABIUNCTIS quidam ὑφ' ἕν legunt: divisis. Sic in Olympio: 'abiungere luna iunices'. 102 FROM THE CONNECTED [AB IUNCTIS]: do not understand this as a single word. The pine COVERED also with skins, as is the custom. Some read UNYOKED [ABIUNCTIS]: meaning divided. Olympius uses it so: 'to unyoke the young cows from the moon '. 103 There is absolutely no reason to associate this half-line with Nemesianus: it does not appear in his poetry, does not relate to themes that he discussed, and is not attributed to him by the text. It is the work of some unknown poet called Olympius, which was (as a rare exception to this broad trend, but is clearly not a signum. One Cecropius is also identied as Ceronius (Gallieni duo 14.4) a name for which there is no parallel. Domaszewski 1918: 138 identied (besides Toxotius and Eugamius) four signa: Severus Archontius (Quadriga Tyrannorum 2.1), Claudius Eusthenius (Carus 18.5), Aelius Xidius (Aurelianus 12.1), and Zosimio (Claudius 14.2). Zosimio is a well-attested Greek name (LGPN lists 16 instances of Ζωσιμίων). Aelius Xidius is an editorial restoration for P's deeply corrupt aelioxi dio (Hohl's apparatus is completely misleading on this point) (cf. also the Xidius listed as a consul in the Excerpta Latina Barbari, ed. Schöne 1866-1875: 1.227). There is a Eusthenius (which does not seem to be a signum) in the Carmina XII sapientum, argued by Friedrich 2002 to be of Tetrarchic date (Gregory of Tours also had a niece with the name: PLRE III 'Eusthenia'). Isidore of Pelusium's Ep. 1247 is addressed to a priest called Archontius (cf. 1807, to the children of Archontius), one Archontius was a subdeacon at Angers in the middle of the fth century (see the letter of Lupus and Eufronius, CCSL 148 p. 140, ed. Munier), and the name occurs in two fth-century inscriptions (ICUR N.S. 8.20819, another Gaul, and AE 1975 411b). 101 Verus 8.10; Clodius Albinus 4.9; Macrinus 4.1. In the case of Victor, the nickname is Pinius, since in the formula cui cognomen erat, the name is almost always in the dative: Livy 2. 33.5, 3.12.8, 4.13.6, 23.34.16, 23.37.10, 23.39.3, 26.8.2, 26.39.15, (25.28.5 is a rare exception, but the nickname, for one Epicydes, is presumably Greek); Plin., HN 7.143;Suet.,Iul. 59.1,Claud. 26.1;Val. Max. 1.5.9,5.4.7;Verg.,Aen. 1.267,9.593. In Tyranni Triginta 8.3, Mamurius and Veturius are nicknames, but derived from Mamurius Veturius, the legendary smith. 102 The text is from Sweeney 1997, with some minor changes to orthography. 103 The precise translation of this entry is not perspicacious, but the key point seems clear enough. The personied moon is shown driving a chariot pulled by heifers on the Parabiago Plate, cf. Auson., Epist. 15.3,17.3 (iuvencae), Claud., De raptu Pros. 3.403 (iuvenci). The passage is not cited by the TLL s.v. 'luna ' (7.2.1829.40-1837.34 (Maltby andFlury)), but does appear in 'iuvenix', without much explanation of what is going on (7.2.740.38-55 (Quadlbauer)).
we have seen) a common enough name. Similarly, in a comment on Thebaid 2.58, the scholia cite a poet called Olympus: MEDIAEQUE S(ILENTIA) L(UNAE) philosophi lunam terram ‹aetheriam› esse dicunt, quae circa nostrum hoc solum circulo altiore suspensa est. Haec autem omnia corpora maiora gignit, utpote quae vicina sit caelo. Poetae denique omnes asserunt leonem de his polis ortum, quem Hercules prostravit, ut etiam Olympus ait. THE SILENCE OF THE MIDDLE MOON the philosophers say that the moon is a ‹heavenly› body, which has been hung up on a loftier orbit around this our earth. 104 This moreover gives birth to all the greater bodies, namely those which are next to the sky. All the poets allege that the lion, which Hercules defeated, sprang from these heavens, as also Olympus says.
Once again, the origin of the Nemean lion is not a subject mentioned by Nemesianus, is not obviously relevant to the themes of his poetry, and is not here ascribed to him, or indeed even to 'Olympius'. Modern editors generally insert an i into the name Olympus, but there is no particular reason so to do. While seemingly not so common as Olympius in Late Antiquity, Olympus is a very well-attested name in its own right. 105 Given that these are two references in the same set of scholia, both seemingly related to heavenly bodies, it might make sense to assume that they are references to the same poet, but there is no particular reason to favour Olympius over Olympus as the name of this presumably late antique author. 106 To understand why these two references have become part of the story of Nemesianus, we have to turn to the rst volume of Johann Christian Wernsdorf's Poetae Latini Minores, published at Altenburg in 1780. 107 This was principally taken up with the Cynegetica of Grattius and Nemesianus, extensively annotated, but also offered a selection of other works with a eld-sports theme, and some short essays on points raised by them. Among the testimonia for Nemesianus, Wernsdorf printed the scholion to Statius, Thebaid 5.389. 108 His reasoning, as he explained, was that no other poet called Olympius was known from antiquity. He also silently assimilated the Olympus of the scholia to this composite gure: the source of the emendation in modern editions. 109 The identication of Olympius as Nemesianus, while obviously weakly grounded, is not so surprising in an era before Dessau, when the HA was still mostly taken to be more or less what it claimed. 110 Wernsdorf, however, had a more daring aim than simply to ascribe another fragment to Nemesianus. In a substantial preface, he proceeded to argue on the basis of the scholia that Nemesianus was also the author of the Laus Herculis (LH), a late antique poem attributed to Claudian in the only surviving manuscript (of the eighth century). 111 The Laus Herculis, Wernsdorf suggested, was really a panegyric of the emperor Maximian, colleague of Diocletian, who was closely associated with Hercules. 112 As he explained, when he rst noted the scholion about the Nemean lion, he began to hunt for parallels. This led him to the Laus Herculis, which says that the Cretan Bull came from the moon. He acknowledged that this was not the same thing, but suggested that the scholiast wished to include Olympius as the poet who in particular had treated the labours of Hercules. 113 Hence Olympius was the author of the Laus Herculis. That Olympius and Nemesianus were one and the same he deduced from some stylistic parallels between the Cynegetica and the Laus Herculis. 114 From there he attempted to squeeze some biographical details from the LH, which he suggested were entirely compatible with what we know of Nemesianus. 115 This is an argument of considerable ingenuity and is made with gusto, but it is also plainly wrong. Leave to one side the question of the HA's reliability, and ignore the silent assimilation of Olympius and Olympus: Wernsdorf's theory is simply incompatible with the text of the scholion on Thebaid 2.58, the whole point of which is that Olympus had said that the Nemean lion came from the moon, something the LH conspicuously does not mention. 116 There is, moreover, nothing remotely like the quotation in the scholion on 5.388-9 to be found in the Laus Herculis, so one has to posit that the scholiast had access to a different lost poem by Olympius, on a similar astronomical theme. The stylistic parallels invoked by Wernsdorf are barely worthy of the name: one of them is that both poets claim inspiration from the Castalian Spring. 117 As Volpilhac also pointed out in the 1970s when there was an attempt to revive the theory, the author of the LH takes some liberties with quantities which are unthinkable in Nemesianus and difcult to imagine even in the considerably later Claudian. 118 For these reasons (and others, no doubt), Wernsdorf's theory appears to have been very largely ignored by scholars of the Laus Herculis. 119 They tend to locate the poem at some point after Claudian in the fth or perhaps sixth century, without much certainty as to the precise date. 120 Yet, in spite of this, for some reason the idea that the scholia to Statius relate to Nemesianus has stumbled on. 121 It is past time that it was put to rest.  Mustard 1916. 123 For the Eclogues, Schenkl's edition (1885) offers a very extensive apparatus of parallels. Not all of these are convincing, but they give some sense both of the allusivity of Nemesianus' verse and the richness of his reception. The same can be said of the even richer collection in Korzeniewski 1976. For the Cynegetica, the commentary by Jakobi 2014 assembles many interesting possible borrowings. 124 De Caesaribus 38.1.
Latin. Almost, because the one other place it occurs is in Nemesianus (Cyn. 80-1), who fancies that he can already see the 'imperial raiment' (augustos habitus) of the brothers. The rarity of phrasing and the context strongly suggest that Victor was consciously alluding to Nemesianus here. A few decades later, Ausonius quoted Cynegetica 268 in his Gratiarum actio (14.65). 125 Further allusions to the same work can be found in the poet Avienius, in Claudian and in Dracontius. 126 The Eclogues were also popular: known to the authors of the Carmen contra Paganos and the Einsiedeln Eclogues. 127 A line from one of them was even used in a Christian epitaph from Rome, of the fourth or fth century. 128 There was a fair chance that any educated reader who came across the alleged poetic rivals of the emperor Numerian would feel a icker of recognition at the name Nemesianus. Perhaps that was the point. The HA is crowded with 'bogus' authors, but when they are put under the microscope, they often transpire to have a 'fake but accurate' feel to them. The works cited or the names given are slightly wrong, but at least some of the core details, most often the rough date, are plausible. We can even sort the bogus authors into two general categories. In the rst, we might put those who are indeed real authors, but to whom the HA attributes fake works. Examples include Gargilius Martialis, who may well have lived at the time of Alexander Severus, but actually wrote on horticulture; and Phlegon of Tralles who was a freedman of Hadrian and did write historical works, but whom the HA claims (almost certainly falsely) transmitted the letter of Hadrian that it reproduces. 129 In the second category, we might put bogus authors whose names are redolent of actual literary gures, but themselves invented: the historian Onesimus, for example, is very close to the name of the sophist Onasimus; and the Aurelius Victor, cui Pinio cognomen erat, mentioned above shares the rst two names with the actual historian Aurelius Victor (since the chronological conceit means that he cannot actually refer to Victor, who was probably born c. 310-20). A particularly striking example is Suetonius Optatianus, a supposed writer of (later) imperial biographies, whose rst name is shared with the famous biographer, and whose second (coincidentally or not) is shared with a Constantinian poet Porrius Optatianus. 130 Even sceptical scholars have been tempted to put Olympius Nemesianus in the rst categoryan actual author with fake works fathered on himbut since 'Olympius Nemesianus' cannot actually be the name of M. Aurelius Nemesianus, perhaps this instance might actually be better assigned to the second. The fact that Nemesianus was a real author with a known context was used to mint a new poet, who did not write on the terrestrial matters of husbandry and hunting, but on the maritime pastimes of sailing and shing.
These made-up references to real authors and bogus authors fabricated on the model of genuine ones were the result not simply of a delight in pure invention, but rather of the need to lend a wash of authority to what the HA claimed about the history it described. 131 That ought to make us take a closer look at some of the outwardly more respectable authors cited by the HA -Marius Maximus, for example, or Florus. 132 Perhaps the imposture has just been particularly successful in such cases. Students of the HA ought, as always, to be on their guard.

VI CONCLUSION
Recent scholarship on the Historia Augusta has tended in two opposite directions. One strand, typied by Rohrbacher's 2016 monograph The Play of Allusion in the Historia Augusta, has emphasised the literary qualities of the collection, even to the point of glossing over its historical fraudulence. Rohrbacher makes the reference to Nemesianus the centrepiece of a sophisticated engagement with the tradition of Latin pastoral in the Carus. 133 Were this true, it would provide a powerful counterpoint to the arguments we present here, showing that the scriptor denitely had the genuine Nemesianus in mind and expected the same of his readers. First, he adduces one Julius Calpurnius (mentioned earlier in the same life, 8.4) as the author of a letter about the death of Carus. 134 Following a well-worn, if unacknowledged, early modern track, Rohrbacher associates this gure with the (probably third-century) bucolic poet Calpurnius Siculus. 135 He then moves to the fact that, much later in the life, Numerian's brother Carinus is described as having a jewelled belt (17.1 balteum … gemmatum in P). Rohrbacher notes that there is also a belt with jewels in Nemesianus' Cynegetica, used to describe the garb of Diana, gemmatis balteus … nexibus, 'a girdle with jewelled fastenings' (Cyn. 92). In concluding, he points out, in very careful terms, that Calpurnius Siculus 'has the same collocation, "balteus en gemmis", in his seventh eclogue '. 136 Were this all true, then the HA would show a deep knowledge of Latin bucolic poetry in general and of Nemesianus' works in particular, which might be thought to undermine the arguments advanced here. In fact, however, none of Rohrbacher's arguments withstand scrutiny. Leaving to one side the 130 See Rohrbacher 2016: 24-5. 131 Cf. Lippold 1999: 160, who actually uses the 'reality' of Nemesianus to suggest that Aurelius Apollinaris might have been a genuine gure. 132 For some bracing scepticism on Marius Maximus, see Paschoud 1999a. 133 Rohrbacher 2016 At Carus 8.4, P actually reads iubus capurnius, which Matoci corrected to iulius calphurnius. Σ reads iubet capriniusa mistake that only makes sense if derived from an abbreviated version of P's reading (e.g. iub' capurnius in B)yet another mark of the family's ultimate derivation from P. 135 Fabricius 1697: 152. On the date of Calpurnius Siculus, see still Champlin 1978, whose conclusions stand despite the recent challenge attempted by Nauta 2021. 136 Rohrbacher 2016: 40. fact that jewels in connection with belts is a rather common idea, there is no parallel between the two belts: Carinus' is evidently itself set with gems, where the one described by Nemesianus has jewelled clasps. 137 Indeed, balteus is used in two different senses in the two passages: sword-belt (i.e. ζωστήρ) in the HA and woman's girdle (i.e. ζώνη) in Nemesianus. 138 The real parallel with Carinus' belt is found in the HA itself, where the louchest of emperors and a tting antitype of the wastrel Carinus, Gallienus, is described as 'using a jewelled belt' (HA Gallieni duo 14.6: gemmato balteo usus est). Calpurnius' balteus has even less to do with Carinus'. The reason for Rohrbacher's slight obliquity in phrasing is that Calpurnius' balteus en gemmis … radiant (7.47-8) has nothing whatsoever to do with clothing: instead, it refers to the aisle of an amphitheatre, probably the Colosseum (i.e. διάζωμα). Hence, there is no connection whatsoever between Calpurnius, Nemesianus and the HA in this phrase. The whole notion that there is some obscure engagement with the bucolic tradition in the life of Carus and his sons depends upon the idea that the bogus Olympius Nemesianus is meant actually to refer to the real bucolic poet M. Aurelius Nemesianus. Take that awayas it must be when the Cynegetica is removedand there is no reason to even bring pastoral to mind. Students of the HA inclined to nd considerable literary subtleties in the work may be themselves victims of a much more basic imposture.
The second tendency in recent HA scholarship is to attribute varying degrees of the text's fraudulence to its Carolingian transmission. Shedd, for example, has recently argued that the six fake authors are not all intrinsic to the HA as composed, but actually the result of medieval tampering, or at least misguided scholarship. 139 Even more extreme, Baker has attempted a wholesale rehabilitation of the basic reliability of the HA, suggesting that its incoherence as a collection is due to it having been assembled in the Carolingian period. 140 Neither study can actually provide evidence that the transmission of the HA is not a straightforward case of medieval copying, like that of Suetonius' De vita Caesarum, a work of the same genre, with a parallel structure, and transmitted via the same milieu. 141 By contrast, the evidence we have adduced here shows that if anything, medieval interventions in the text tried to tame its idiosyncrasies and introduce facts into the farrago of fantasy. Shedd is not wrong to note that the HA as we have it contains 'unprecedented ctions', but the genesis of those ctions is to be found in the text itself, whenever and by whomever it was composed, and not in medieval scholarship of the generation before our earliest manuscripts were copied. 142 This inquiry has brought us, by roundabout paths, to several fairly simple conclusions. (1) The text of Carus 11.2 should be printed cum Olympio Nemesiano contendit, qui Halieutica et Nautica scripsit, with ΚΥΝHΓΕΤΙΚA bracketed or banished to the apparatus and the other two works rendered in Latin characters. (2) M. Aurelius Nemesianus should no longer be saddled with the name Olympius.
(3) Olympius Nemesianus should be added to the list of 'bogus authors' in the HA. Alongside these, we have come to two more tentative ndings which merit further exploration. (1) The onomastic practices of the author of the HA deserve further study: just because the people referred to are fake does not mean that the onomastic practice by which they are named is also. (2) The archetype of the tradition of the HA might have been housed and