The text of the Αοὐρελιους graffito

As will be argued, Aurelius was most likely a merchant trader and thereby possessed at least a basic literacy and numeracy suitable for commercial endeavours, including the ability to sign – or, in this instance, carve – his name. Thus, both the inscriber and the referent Aurelius were the same individual. The assertion cannot be definitively proven, for a literate slave may have done the deed at his master’s behest, but it is the most probable interpretation. Previous engagement with the text likewise accepted Aurelius as the one who inscribed the letters without an intermediary.

The graffito is approximately 41 cm long, including the ‘7’ sign that follows the lunate sigma. Aurelius displayed good stonemasonry and a steady hand with his carving. The characters are clearly formed and occupy the same relative horizontal plane with similar heights to one another and with generally even spacing. The regularity displayed at Zinkekra indicates that Aurelius took his time, slowly inscribing while ‘perched precariously with [his] back to the steep drop’ of a height of approximately 100 metres (Hawthorne et al. Reference Hawthorne, Mattingly, Daniels and Mattingly2010, 19 and 39). In addition to a display of personal courage, the steadiness suggests a familiarity with the shapes of the Greek letters. The palaeographic evidence makes it highly improbable that Aurelius carved the ‘7’ marking, which Daniels (Reference Daniels1968, 182–83) hypothesised was a ‘centurial sign’. Subsequent photographic analyses have shown that the sign was made by a different hand, one of the many individuals who left their mark on the rock over the centuries (Desanges and Lancel Reference Desanges and Lancel1970, 12–13; Rebuffat Reference Rebuffat1982, 502 n. 76; Ruprechtsberger Reference Ruprechtsberger1997, 73). Furthermore, the editio princeps did not associate the two carvings with each other (Sattin Reference Sattin1965, 77).

The graffito probably postdates the 212 promulgation of the Antonine Constitution. The decree’s expansion of citizenship led millions to adopt the nomen Aurelius, which then became the most popular in the Greek East and a close second (after Iulius) in the Latin West (Salway Reference Salway1994, 134). The third century also coincides with the Severan expansion of the Roman frontier zone southwards in Tripolitania, establishing garrisons at the oases of Cidamus (modern Ghadames), Myd[…] or Mad[…]) at modern Gheriat el-Garbia, and Gholaia (Bu Njem). The army followed an already established sedentary Romano-Libyan population (Guédon Reference Guédon2018, 100–23). A terminus ante quem is much more uncertain and cannot be definitely stated. The text most probably belongs to the third century and plausibly, but less so, to succeeding ones.

Aurelius had room on the rock to add a praenomen and cognomen, a patronymic, or all three elements if he so chose. While formal or public inscriptions typically dictated an expanded nomenclature to situate the person within the relational nexus of the community in question, e.g., Aurelius Epagri f(ilius) (Aurelius the son of Epagrus) as named in a dedication to Apollo from Oea (IRT 229), conventions were looser for ancient graffiti. There, informality was a defining characteristic (Banes Reference Banes2022, 17–18). Identification in graffiti often consists of a single name, whether painted on the walls of Pompeii (where they form the most numerous type of informal writing; Langner Reference Langner2001, 22–24), etched by pilgrims seeking intercession throughout the late antique East (Nowakowski Reference Nowakowski, Nowak, Łajtar and Urbanik2017, 239–40), or carved into rocks in the thousands by bored shepherds on and beyond the Arabian frontier (MacDonald Reference MacDonald, Škrabal, Mascia, Osthof and Ratzke2023). Common though a name it may have been across the Roman world, ‘Aurelius’ sufficed in Aurelius’ self-memorialisation in the innermost Sahara.

The decision to carve only a single name into the rock is unsurprising and conforms to epigraphic practices found among graffiti. Without doubt the most perplexing element of this Saharan text is the use of Greek characters to present a non-standard rendition of the Latin name Aurelius, Αοὐρελιους, instead of the orthographic Αὐρήλιος. Sustained analysis reveals a surprising amount about its author, not least his linguistic background and geographic origin. Aurelius carved his name as it sounded to him, writing in Greek a phonetic rendering borrowed from his native Latin. That the name is transliterated is not a novel finding; a recent work on the languages of Late Antique North Africa recognised it as such (Wilson Reference Wilson2024, 79). Here, we will go further and argue that the idiosyncratic spelling of Αοὐρελιους – it is a hapax legomenon for variants of Aurelius in Greek – derives from Aurelius’ origins in coastal Tripolitania.

Use of the Greek Αου to replicate the sound made by the Latin Au is one of the earliest and most frequently occurring epigraphic transliterations of the syllable (Eckinger Reference Eckinger1892, 14–15). Aurelius’ writing ους at the end of the word is not only a transliteration, but it is also indicative that he was either uncertain or unaware of first-declension nominative endings in Greek, -ος. Similar instances of ους replacing -ος can be found throughout the Roman world as examples of low-level biliteracy, but one in particular from North Africa warrants mention. Excavations at a Jewish cemetery in Carthage uncovered several funerary epitaphs in Greek, of which one was actually Latin written with the Greek alphabet. The inscription commemorates Ἀπώνιους πουερ ‘Ρούφους, which is Aponius puer Rufus (Aponius, son of Rufus) (Ferron Reference Ferron1956, 114).

The e in Aurelius is long. Both the conventional Αὐρήλιος as well as a phonetic transliteration should produce η and not the ε witnessed in Αοὐρελιους. While Koine Greek was collapsing the audible distinction between η and ε (Adams Reference Adams2003, 484), the great majority of Egyptian papyri do not demonstrate vowel variation following ρ, where standard orthography requires η, useful as a proxy for colloquial pronunciations. Variation does not increase beyond minor levels until the rise of Coptic from the fifth century onwards (Henriksson et al. Reference Henriksson, Dahlgren, Vierros and Reggiani2025, 197). Thus, Αοὐρελιους with an ε is not a borrowed Graecism. Other Latin speakers writing in Greek characters distinguished long e and short e in their transliterations. When Cornelius etched his name in Greek characters onto a rock above the temple of Hathor at Deir el-Medina, in Egypt’s western Thebes region, he wrote Κορνηλειους, where the standard form is usually Κορνήλιος (Bingen Reference Bingen1981, 342). Cornelius used ει for the long i and ους for the nominative ending -ος, but knew to represent the long e in his name with an η. Aurelius’ Αοὐρελιους would, however, be a phonetic transliteration according to a particular regional variant of Latin: African. The Latin language in North Africa was losing distinctive vowel lengths, among other linguistic developments (Wilson Reference Wilson2024, 75–76), and the process includes a conflation of short e with long e (Adams Reference Adams2007, 648–49).

Aurelius then seems to be a native Latin speaker from North Africa. Proconsularis is preferable to one of the Mauretanian provinces, given the former’s proximity to Zinkekra and the central Sahara. Furthermore, Tripolitania seems more likely than the portions of Africa Proconsularis comprising modern Tunisia for the same reason. Greek inscriptions in Tripolitania before the East Roman reconquest under Belisarius are primarily found in coastal cities, many of which were dedicated by the various foreign merchants resident in the region. Lepcis Magna has by far the most (189 inscriptions according to the Die Epigraphische Datenbank Clauss-Slaby (EDCS), with Sabratha (35, including unpublished material from the Sarapeion; Wilson Reference Wilson2024, 78) and Oea (8) a distant second and third. A background in those coastal cities would have afforded Aurelius the means and opportunity to learn even rudimentary Greek.

We should consider whether Aurelius could have been a native Garamantian, who acquired both Greek and a new name while serving in the eastern army. The cohors Gaetulorum, stationed in Arabia, has been proposed in light of the geographic proximity between the Gaetuli astride the frontier zone of Roman Africa and the Garamantes further south (Daniels Reference Daniels, Bynon and Bynon1975, 251; Mattingly Reference Mattingly2023, 168–74, on the Gaetuli; Ruprechtsberger Reference Ruprechtsberger1997, 72–73 and Reference Ruprechtsberger2023, 46, for the cohors Gaetulorum). Legionaries and auxiliaries from the second century onward were usually recruited and stationed close to home, and so indigenous Arabs made up the ranks of the cohort by the 130s (Nehmé et al. Reference Nehmé, Bouchaud, Delhopital, Durand, Égal, Fiema, Gatier, Gazagne, Gerber and Al-Huwaytî2020, 108), not Gaetuli and not any itinerant Garamantes. If there were Garamantian auxiliaries, they most probably would have served in or near Tripolitania, not the hellenophone East, learning to speak Latin, yes, but to write with Latin characters and not Greek ones. Too many implausible events are needed for Aurelius to have been a Garamantian veteran of the Roman army.

Could Aurelius then have been a Roman soldier, even discounting the relevance of any centurial sign? The army in North Africa tended to recruit from a relative handful of coloniae. Exceptions exist: a unit of Palmyrene archers, the numerus Palmyrenorum sagittariorum, is found in Numidia with an active Syrian complement, but on the whole, a Roman soldier in North Africa from the second century onwards would have been from the region (Mattingly Reference Mattingly2023, 291–94). Using the Gholaia ostraca as a wider proxy, most of those stationed on the limes Tripolitanus had imperfect Latin literacy. The majority came from a Punic background or a Latinate socioeconomic stratum that did not permit a full educational curriculum with a grammaticus, unsurprisingly, although their spoken Latin may well have been quite good (Adams Reference Adams1994; Felice Reference Felice2019, 231–32). Even attempts by centurions at the fort to compose in Latin verse betrayed a substandard education, with one (Iasucthan) who was probably a native Punic speaker (Adams Reference Adams1999, 110). Mattingly (Reference Mattingly1995, 122) raised the prospect of soldiers sent to Garamantian territory as overseers of a sort, referencing Aurelius at Zinkekra in support. The possibility is real – though we then wonder what supervisory tasks a legionary or auxiliary would have performed – but there do not appear to be Greek inscriptions or other evidence from or in relation to soldiers on the Tripolitanian frontier. If there were Hellenophone legionaries or auxiliaries dispatched to the Fazzan, they presumably would have written their name using standard orthography.

The question still remains why Aurelius chose Greek characters. The decision did not facilitate comprehension. In fact, that may have been the purpose. Aurelius wrote to communicate more than he did to be read. An epigraphical comparandum from the other end of Africa, at Musawwarat es-Sufra in Meroe (present-day Sudan), helps explain such a paradoxical statement. It is the southernmost Latin text from the ancient world, an informal graffito carved by a passing visitor to a Meroitic temple. The author was probably a merchant or diplomat visiting Meroe around the turn of the fourth century. His text reads as follows:

Though written in Latin, the phraseology suggests the carver was a Greek speaker. Bona fortuna is a direct translation of the common expression ἀγαθῇ τυχῃ, where Latin would more naturally use feliciter, as also seen in the graffito. Domina regina corresponds with κυρία βασίλισσα, a titulature found at the nearby temple at Dakka. The visitor had been to Rome – or perhaps Alexandria – and wanted to exult in the fact rather more than he wanted others to know it. The fact that a text in Latin would have been much less intelligible than one in Greek seems to have served the visitor’s designs. The text was a boast that ‘would have been lost on virtually everyone in such a remote region. The display consists at least as much in the ability to use the language of the Romans.’ The author was a Greek speaker who used Greek phraseology to compose in Latin because the language and the script were themselves markers of status (Adams Reference Adams2003, 80–81).

The situation was similar for Aurelius at Zinkekra. He, like the visitor to Meroe, commemorated not merely his appearance in a land distant from the Roman world. Just as much, Aurelius highlighted his elevated status with a script chosen for the purpose that was as demonstrative towards that end as the writing in text itself, more so in this instance. Latin in Egypt and elsewhere in the East was a ‘super-high’ language of power and politics (Adams Reference Adams2003, 545–76). Adams wrote that the graffito at Musawwarat es-Sufra was a ‘display’, and the word is apt. Both that text and that of Aurelius possessed a visual component where the very shape of the letters separated the author from others. The Garamantes had their own language, dialects of Proto-East-West Amazigh probably shared with the Gaetuli to an extent, and their own script, undeciphered Libyan characters that are found at Zinkekra and other Garamantian sites in the thousands (Mattingly Reference Mattingly2023, 384–85). Latin was suitable for a visual spectacle had Aurelius wished it amidst the foreign linguistic and alphabetic context, though both the language and its writing system were probably known to a select few Garamantes from contacts with the Roman world via Tripolitania. Greek, though, was more unusual to Tripolitanians and Garamantians alike.

It is an axiomatic proposition that epigraphy is ‘read’ not only as a text but also as an object grounded in physical space. We should bear in mind that the Zinkekra graffito is located at a long-abandoned settlement carved into a rock at the very edge of the promontory above the area where a possible Garamantian temple was situated. Most subsequent engagement with the graffito would have been by people looking at the letters upside down, as Ruprechtsberger (Reference Ruprechtsberger2023, 46) notes. This short piece of onomastic writing was as much a boast as what was found at Musawwarat es-Sufra. He, Aurelius, was a Roman citizen who had ventured far from the known world and wished to proclaim that accomplishment with a piece of onomastic permanence.

The Αοὐρελιους graffito in its Saharan and Garamantian contexts

Attention to the physical nature of the graffito – it is a material object as much as it is written text – invites a change from the Mediterranean perspective that has dominated thus far. The focus now shifts to the Garamantes, who have been lurking in the background despite Zinkekra being in the heart of their territory. Since the publication of the text in 1965, the Garamantes have received extensive archaeological fieldwork and analysis that have transformed scholarly reconstructions of the Fazzan and its peoples (Duckworth et al. Reference Duckworth and Mattingly2020; Gatto et al. Reference Gatto, Mattingly, Ray and Sterry2019; Mattingly Reference Mattingly2003b, Mattingly et al. Reference Mattingly, Leitch, Duckworth, Cuénod, Sterry and Cole2017; Sterry and Mattingly Reference Sterry and Mattingly2020, and references therein). The transformation opens new avenues of inquiry into this short piece of writing. The Garamantes (a recent overview is found in Mattingly et al. Reference Mattingly, Merlo, Mori, Sterry, Sterry and Mattingly2020) were a pastoralist Saharan people who, starting in the early first millennium BC, sedentarised around the oases of the Wadi al-Ajal, where Zinkekra is located. They developed a complex society with social hierarchies, possessed at least one urban centre at Garama (Mattingly et al. Reference Mattingly, M., D.C. and Mattingly2013) and possibly also had a state (see the divergent works of Scheele Reference Scheele, Sterry and Mattingly2020 versus Mattingly and Sterry Reference Mattingly, Sterry, Sterry and Mattingly2020).

Zinkekra was an early Garamantian settlement, the ‘type site’ for Garamantian culture in the first millennium BC (Mattingly Reference Mattingly and Mattingly2003a, 136). Habitation began on the promontory where natural defences were aided by the construction of walls where it narrowed, and dwellings spread downwards along the escarpment as the population grew. The site was most active ca. 900–500 BC before it was abandoned in the second half of the millennium as the nearby site of Garama eclipsed it, located just a few kilometres distant (Daniels Reference Daniels1968; Hawthorne et al. Reference Hawthorne, Mattingly, Daniels and Mattingly2010; Mattingly Reference Mattingly and Mattingly2003a, 136–42, 160–66). Zinkekra is distinctive for two reasons. First, the site is home to one of the densest concentrations of rock art in Garamantian territory and is the only Garamantian settlement containing rock art (Barnett and Guagnin Reference Barnett and Guagnin2014, 176). Saharan rock art tended to cluster around special topographic spaces and featured subjects primed for divine intercession – war, livestock and wild animals, fertility, etc. (Mattingly Reference Mattingly2023, 505). Zinkekra was then one such place. The rock that Aurelius used for his carving was home to an abundance of other inscribed material. Visible in Figure 1 is a swastika-shaped emblem speculated to have a solar connection (Sattin Reference Sattin1965, 77). Also present is a later Arabic carving of the shahada, the Islamic profession of faith, four sandal outlines – possibly a ‘private’ carving given their difficulty in being perceived – and geometric ‘V-shaped motifs and lines of different lengths’ (Barnett Reference Barnett2019a, 207–33, and Reference Barnett2019b, 262). Aurelius entered into a pre-existing epigraphic community reaching backwards and forwards in time. The fact is unsurprising as graffiti tends to cluster together in both ancient times and modern.

Figure 1.

Photograph of the rock at Zinkekra with the Aurelius graffito in the lower foreground (image reproduced from Hawthorne et al. 2010, 39 fig. 1.34, taken by David Mattingly, 2005).

Also relevant is Zinkekra’s post-occupation history, which further imbued the inscription with a sacral connotation. Aurelius climbed through an abandoned settlement on his way to the rock found at the promontory’s eastern edge, yet the site was still active. His carving faces east, towards the rising sun, as do many Garamantian monuments and tombs (González Rodríguez Reference González Rodríguez2015, 324), including the burials that continued in cemeteries found across the slopes of the Zinkekra promontory. Such funerary sites are often the objects of religious activity across North Africa and in the Sahara, many of whose societies feature ancestor cults (Mattingly Reference Mattingly2023, 505–506). New buildings were constructed after abandonment, made from ashlar and mudbrick, larger and more complex than the one- and two-room structures found when it was inhabited, possibly temples or ritual spaces (Mattingly and Edwards Reference Mattingly, Edwards and Mattingly2003, 178–79). Epigraphic activity also seems to have continued unabated, cementing the special status of Zinkekra as a religious centre (Barnett Reference Barnett2019a, 245–47).

Graffiti in sacred spaces, and Zinkekra was one such space, provides a material monument to the impermanence of speech and movement that comprise many rituals. If ritual ‘is an activity which is recognised by a social group as having meaning and repeated by members of that social group’, then it is easy to see how graffiti, both the act of creating and the permanent creation, are themselves ritual products (Banes Reference Banes2022, 127–29, quote at p. 127), The numerous inscriptions found at Zinkekra – but not other, non-sacral Garamantian sites – should themselves be viewed in a religious light. Thus the informality that is a defining characteristic of graffiti does not necessarily mean the text is then casual or unserious. The autographic nature of Aurelius’ inscription, written with his own hand, created a more direct relationship between himself and the objects of veneration or thanksgiving. Aurelius would have noticed the carvings surrounding him as he made his ascent, seen what may have been temples and passed by the cairns and shafts marking the resting places of the dead. Details may have occluded him, the specific ritual forms or deity(ies) worshipped and such like, but Aurelius would have recognised the sacred environment into which he entered, and acknowledged its power.

Perhaps his graffito was meant as much for the divine – be they ancestor spirits or deities – as mortals, a benediction for safe passage through the desert offered to local powers upon entering their territory in a kind of religious excise duty or tariff. A diachronic feature of desert epigraphy in sacred spaces is a tendency to offer gratitude to the divine for safety amidst such a hostile environment. It can be seen at the bathhouse of the Roman fort of Gholaia with its dedication to the Roman goddess of health, Salus (Rebuffat Reference Rebuffat1987), and from successive generations of travellers to the temple at El Kanais in the Eastern Desert offering thanks to Pan – some in scratched carvings that only give their name (Adams Reference Adams2007, 215–20). Reading an offering of gratitude into what is only Aurelius’ name is, admittedly, speculative. But the Garamantian guides who would have accompanied Aurelius into the Sahara would have been more familiar with Zinkekra’s special status as a liminal place where the ancestral dead and inscribed carvings meant that barriers between the human and the divine were perhaps more permeable than in other areas of the mundane world.

Aurelius was most likely from Tripolitania and probably not a soldier sent to reconnoitre or oversee the Garamantes. His purpose for being at Zinkekra and in the Fazzan then requires an explanation. Much of the previous discourse on Aurelius has postulated a military context for his presence in Garamantian territory, whether as a returning mercenary or visiting legionary. Most of these works were written before two paradigm shifts in the scholarly discourse. The first is the almost complete reconstruction of knowledge on the Garamantes, discussed above. Second, and almost as important, is the change in frontier studies over the past several decades. The limes was not a ‘defended border’ demarcating Roman from non-Roman territory (Isaac Reference Isaac1988), and frontiers were porous spaces with bilateral and multilateral cultural influences, negotiations of power, economic exchange and more (Whittaker Reference Whittaker1994 for an important milestone).

The military was indeed one of the primary drivers of long-range mobility in the Roman world; a second was commerce. The likeliest explanation for Aurelius’ presence in the Fazzan is one connected to trade. The last generation of fieldwork in the Sahara has firmly established the existence and magnitude of pre-Islamic networks of mobility and exchange (formerly deemed a ‘myth’ – see the title of Swanson Reference Swanson1975) between and beyond the Sahara and the Mediterranean (overviews are found in Mattingly et al. Reference Mattingly, Leitch, Duckworth, Cuénod, Sterry and Cole2017). Pottery from the Punic and Hellenistic worlds was found at Ghat in the southwestern corner of Garamantian territory (Felici Reference Felici and Liverani2005, 243). Tripolitania sent southwards wine and olive oil in many thousands of amphorae, and metals such as brass and silver for recycling. The sheer quantity of ceramics recovered from the oasis sites indicates that large swathes of Garamantian society used and consumed these products, not just the upper echelons of the elite. The Sahara in return would have sent north slaves – many probably obtained from sub-Saharan Africa – precious stones such as carnelians, dates grown in the oases and textiles (Mattingly Reference Mattingly2023, 549–55 for a recent overview). That said, we should remember that ‘the number of archaeologically identified objects that can unequivocally be shown to have come to the Roman world from the Garamantes is, currently, precisely zero’ (Wilson Reference Wilson, Mattingly, Leitch, Duckworth, Cuénod, Sterry and Cole2017, 190). Foodstuffs and other archaeologically invisible material circulated at the local, inter-oasis level within the Sahara. One of the Gholaia ostraca (O. Bu Njem 72) records the entry into the fort of a caravan of barley, three mules and four donkeys led by some Garamantes.

Only a few kilometres separated Zinkekra from the Garamantian capital at Garama, a true oasis-city with a population reaching 2,500 at its height, complete with a temple adorned with Corinthian columns, a bathhouse built in the Roman style with hypocausts and, assuredly, markets and bazaars where Aurelius could buy and sell goods (Mattingly et al. Reference Mattingly, M., D.C. and Mattingly2013). Commerce provides a more useful heuristic for Roman–Garamantian interactions than combat. Romans going south and Garamantians heading north were much more likely to do so for commercial reasons than to engage in raiding or open warfare. For the latter, the ancient sources only identify Balbus’ punitive campaign of 19 BC, the Garamantes’ involvement in Tacfarinas’ revolt (ca. 15–24 AD) and Festus’ retaliatory expedition in 70, following Oea’s hiring of Garamantian mercenaries the year before amidst the chaos of the imperial succession. Nor does the epigraphic record mention incursions by the Garamantes, although arguments from silence are admittedly not the strongest.

If Aurelius were a merchant or trader, he would have possessed the resources, incentive and circumstance to be in the Fazzan. We know that many goods were carried in desert caravans between the central Sahara and the Mediterranean. We do not know of any Roman soldiers sent to the Sahara after Festus’ incursion of 70 AD, nor even how far into the desert that force penetrated. Other reconstructions of Aurelius’ background are plausible in that they are not outright contradicted by the available evidence, which is a single word of ten letters. But these alternatives are harder to fit to both what is known and what can be justifiably argued about Aurelius and his text.