Skip to main content Accessibility help
Hostname: page-component-55b6f6c457-9lvz7 Total loading time: 1.763 Render date: 2021-09-28T16:50:34.947Z Has data issue: true Feature Flags: { "shouldUseShareProductTool": true, "shouldUseHypothesis": true, "isUnsiloEnabled": true, "metricsAbstractViews": false, "figures": true, "newCiteModal": false, "newCitedByModal": true, "newEcommerce": true, "newUsageEvents": true }


Published online by Cambridge University Press:  25 August 2021

Tim Whitmarsh*
University of Cambridge, UK
Rights & Permissions[Opens in a new window]


This article considers a short text that was widely circulated in the mid- Roman Empire, in both a four-line and a six-line version, usually on gemstones. The text is a poem of sorts, but of a quite distinctive type. Part of it can be scanned according to the rules of classical (quantitative) metre, but more striking is the consistent rhythmic (stressed) pattern. Stressed poetry is not otherwise attested so early; this text may point to a substrate, now largely hidden from view, of popular verse that preceded the metrical revolutions of late antiquity and the Byzantine world. The poem is also a piece of visual artistry, designed to be looked at (particularly in its gemstone format). This hybrid status, between high art and popular culture, can also be detected in the content of the poem, which gestures towards both the poetics of intellectual elitism (using intertextual allusion, and dismissing the views of the masses) and a level of sexually aggressive assertion of embodied selfhood. It is a valuable witness to a form of middling literature (and a middling demographic), caught between aspirations to elite-style individuality and the mimetic imperative of an empire-wide consumer culture.

Research Article
Copyright © The Author(s), 2021. Published by Cambridge University Press on behalf of The Cambridge Philological Society

Scholarship on Greek poetry of the second and third centuries CE has boomed in recent years.Footnote 1 The old assumption (as old as Plutarch, at least in one form)Footnote 2 that the era was definitively prosaic is no longer tenable: that poetry retained both its centrality in civic life and its prestige within literary culture is now undeniable. By and large, however, attention has focused on elite poetics.Footnote 3 In this article I consider an anonymous popular text – a poem, I believe, but that identification presumes the discussion below – that was widely circulated across the Empire. My aim is twofold: to collate and publish it; and to reflect on what it can tell us about Greek metrics, poetics and literary value in the Roman period. This brief text, I argue, shines important new light on the emergence of stress-based (as distinct from quantitative) poetry. It has much to tell us about the ingenuity of ‘subliterary literature’, the interaction between material form and poetic content possible in such texts, and the complex psychology of popular literary production and circulation in the eastern Roman Empire.

The text

In Appendix 1 I collect twenty recorded versions of our text, some fragmentary and some abbreviated by design.Footnote 4 The majority survive in the form of inscriptions on gemstones (mostly cameos); one is a graffito from Cartagena, Spain. I begin with an edition:

Apart from the fragmentary item 20, all gemstone versions give the colometry reproduced above. The graffito version seems systematically to amalgamate two lines into one, producing a three-line text (see Appendix 2).

In those cases where the artefacts (including the graffito) on which the text is recorded have been dated by modern specialists, estimates have all ranged between the second and third centuries CE. The transmitted spelling is also compatible with that era, although the evidence is not decisive. There are two noteworthy features:

  1. (i) An imperial date is suggested by the general substitution of -ι for -ει endings in the third person singular of the present tense (μέλι, συνφέρι), and in the second-person singular imperative of the ε contract verb (φίλι). On inscriptions this substitution is attested as early as the third century BCE, but becomes commoner after 100 CE.Footnote 5 The confusion of -ι and -ει (which is in fact commoner in the inverse form) is indicative of a gradual phonetic convergence that seems to have left pronunciation of the two largely indistinguishable by the second century CE.Footnote 6

  2. (ii) Less indicative for the purposes of dating are the spellings συνφέρι and συφέρι. In inscriptions συνφ- is ‘comparatively well attested in the last three centuries before Christ’, although συμφ- is approximately twice as common.Footnote 7 συφέρι is more unusual, but the loss of nasals before stops can be paralleled in inscriptions already from the classical period.Footnote 8

Of course, there is no guarantee that the text itself is not older, and the very variability of the spelling should caution us against any prima facie assumption that what we have is anything like the ‘original’ text. But given that the material artefacts indicate a consistently Roman date-range, and no comparable Hellenistic-era inscriptions have been located to date, it seems safe to conclude that our text probably originates in the Roman period, i.e. in the first or second century CE.

The diction is unambitious. The verbs belong to the beginner's Greek lexicon; there are no nouns, adjectives or adverbs. There is no sign of Atticism: in particular, the third-person imperative -έτωσαν ending, which is regular for the koine of the era, is censured by Atticist authorities, who prefer -όντων.Footnote 9 -όντων is the more recherché, literary option in prose at any rate (-έτωσαν, however, is found high poetry).Footnote 10 The spelling συφέρι (in no. 8) also reflects a feature that is ‘not normally found in decrees and documents in which the writing is of a high standard’.Footnote 11 In terms of language, then, our text and its inscribers do not lay claim to literary elevation. This is perhaps what one would expect, given the relatively modest value of the gems themselves: agate, onyx and sardonyx, the material on which the majority of texts are inscribed, are all varieties of chalcedony, an abundant mineral in the Mediterranean region.


In spite of this apparent want of literary ambition, Carlo Gallavotti has claimed that our text is metrical. Combining the six lines into three pairs appears to yield one hemiamb (∪ — ∪ — ∪ — ∪) and two anacreontics (∪ ∪ — ∪ — ∪ — —):Footnote 12

∪ — ∪ — ∪ — ∪
λέγουσιν ᾇ θέλουσιν
∪ ∪ — ∪ — ∪ — —
λεγέτωσαν οὐ μέλι μοι
∪ ∪ — ∪ — ∪ — —
σὺ φίλι με συνφέρι σοι

To scan the first line, however, Gallavotti was forced to interpret its alpha not as the relative ἅ (a short vowel, producing a light syllable) but as (long) ᾇ, explaining this as a Doric form of the adverbial ᾗ (‘in which way’). Given the resolutely unpretentious diction in the rest of the text, however, it is implausible to imagine that readers would avoid the obvious and supply instead a recondite form in a different dialect. While the final two anacreontics are secure, the first verse is not metrical, at least in the form in which we have it.Footnote 13 There are three explanations for the anomalous metrical assemblage:

  1. (i) The anacreontics are intended, and the first line represents an attempt at metre (whether hemiambic or otherwise) that is ‘flawed’ by classical standards.

  2. (ii) The two lines of anacreontics are intended, but the first is deliberately extrametrical.Footnote 14

  3. (iii) The ‘anacreontics’ are unintended.

Before we assess these possibilities we should consider another striking metrical feature. Our text appears to make use of the stress accent to govern rhythm, in the manner of post-antique Greek poetry: the accents appear on the first of every four syllables (ἅ and σύ are treated as unstressed for these purposes, as they would be in Byzantine stress-based poetry).Footnote 15

It is, to be sure, far from self-evident that the option of adopting a stress-based rhythm would have been open to a poet of the second century CE or thereabouts. As is well known, the shift from pitch to stress accent, accompanying the loss of vowel quantity, seems to have begun in the Hellenistic period.Footnote 16 Greek metre was, however, by that stage already congealed in its classical, quantitative forms. Since Wilhelm Meyer's study of 1885 it has been dogma that the earliest unequivocal examples of poetry based around stress rather than quantity come in the form of two fourth-century texts by the experimental poet Gregory of Nazianzus, the De uirginitate Footnote 17 and the Hymnus uespertinus,Footnote 18 composed in bipartite lines of 14–16 syllables none of which has any quantitative metre and almost all of which show accented stress on the penultimate syllable. This feature, paroxytonesis, can be detected in some quantitative poetry from the High Empire: the chief examples are the choliambics of Babrius and the so-called meiouric (‘mouse-tailed’) hexameters found in Lucian and various papyri. Some have interpreted the incorporation of paroxytonesis into ‘classical’ metres as a precursor of the later shift to stress-based metre, but this is not universally accepted.Footnote 19 Either way, the stress-based rhythm found in our text is of a significantly more elaborate and thoroughgoing variety than mere paroxytonesis, and unparalleled (so far) in poetry of the High Empire.Footnote 20

The twenty-three (or, in the shorter version, fifteen) syllables can be understood as trochaic verses consisting of four syllables, with a caesura following the word-break in the third syllable (in the first three verses the caesura follows a free-standing word; in the final three it follows an enclitic attached to the previous word).Footnote 21 The final verse is catalectic, ending at the caesura (see Table 1).

Table 1 Trochaic scansion by stress rhythm

Alternatively, of course, in order to respect the word-breaks and the colometry in which the text has been transmitted one might understand the rhythm as iambic with a missing first syllable (the ‘caesura’ would then simply become the verse-end). At one level nothing rests on the trochaic/iambic distinction: our text is surely a sui generis experiment with an attractive, memorable rhythm, rather than an attempt to adhere to a formally regulated poetic scheme. But if we consider it from the wider perspective of the history of Greco-Roman metrics then the trochaic/iambic issue becomes significant. At first sight the ‘missing’ syllable is a curiosity, but two related phenomena provide some important context:

  1. (i) The Latin comedies of Plautus make use of iambic and trochaic septenarii in sung sections; this metre appears to have been in common use in oral culture too. The septenarius is in fact a misnomer:Footnote 22 it is rather a catalectic octonarius, consisting of seven full feet plus a half-foot (i.e. fifteen syllables in all). According to Roman conventions, the short version of our text would be considered a trochaic septenarius (the longer version might be considered an extension of the same scheme over twenty-three syllables). Traditional Latin verse patterns are, like Greek, quantitative rather than stress-based, but as has often been noted accentual patterns can track quantitative ones closely in Latin (unlike in Greek).Footnote 23 The uersus quadratus – a trochaic septenarius that often displayed strong correlation between word accent and heavy quantity – was commonly used in acclamations, for example the one that accompanied the return of Germanicus in 19 CE: Salua Roma, salua patria, saluus est Germanicus.Footnote 24 It is notable too that many such septenarii are (like this one, and like our Greek text) characterised by phonetic repetition.

  2. (ii) Our text can also be analysed as seven syllables of stress-based trochaics (λέγουσιν ἁ θέλουσιν) + eight syllables of iambics (λεγέτωσαν οὐ μέλι μοι: short version) + a further eight syllables of iambics (συ φίλι με συνφέρι σοι: longer version).Footnote 25 The fifteen-syllable Byzantine politikos stikhos (‘political verse’), which gained popularity from the tenth century onwards (but the origins of which have been traced back at least to the sixth century),Footnote 26 was similarly based around the combination of octasyllabic and heptasyllabic iambic or trochaic cola.Footnote 27 It has been clearly shown that the late-antique roots of the politikos lie in various different, fluid combinations of eight- and seven-syllable cola,Footnote 28 and that ‘oral accentual verse based on paired cola of 8 and 7 syllables with mixed trochaic/iambic rhythms was already in use from at least late antiquity’.Footnote 29

These two parallels offer some evidence, then, that (a) in both Latin and Greek iambics and trochaics were considered congenial to stressed rhythm, and that (b) such rhythms might be delivered in stikhoi of fifteen syllables, which could in Greek (according to a process that began at least as early as late antiquity) be split into hemistichs of eight and seven syllables. If, as Michael Jeffreys once argued, the politikos stikhos was the end result of a slow evolution of the acclamatory Latin trochaic septenarius, then our text might even be seen as evidence for a ‘missing link’ between Latin and Greek stressed poetry.Footnote 30 Alternatively, if Eduard Fraenkel was right to see the classical Greek (quantitative) trochaic tetrameter catalectic as originally underlying the Latin septenarius,Footnote 31 our text might better be understood as the result of an independent conversion of the Greek quantitative form into a stress-based equivalent. Such questions, however, fall well outside our present remit.

Where does this leave our text, metrically? It seems unlikely that the rhythms might be accidental: can we really believe that a six-fold repetition of a rhythm is the result of chance – particularly when the metrical form is so historically credible? What is more, both the short and the long versions display the same rhythmic pattern (i.e. a combination of seven-syllable and eight-syllable cola), suggesting that the scheme is intrinsic. At the same time, however, the anacreontics seem equally unlikely to be accidental (the non- or sub-metrical first line notwithstanding). What to conclude? One possibility is that an original poem in quantitative verse has been rewritten as a rhythmic poem. Given the frequency of metrical errors in inscribed poetry elsewhere, however, it seems more likely that what we have is the original poem, representing a genuine attempt at (at least capturing some flavours of) quantitative verse, perhaps intended to appeal simultaneously to classically educated readers alongside appreciators of popular verse.Footnote 32 What seems clear, however, is that in the poem's present state the rhythmic metre is more systematic, and likely to have been experienced as dominant.

Other ‘poetic’ features

Other striking features suggest a design that we might call ‘poetic’. One is the use of half-rhymes. As we have seen, the longer version is best understood aurally as six lines of (stress-based) trochaic tetrasyllables, the last of which is catalectic. These six lines divide naturally into 2 × 3-verse phonetic groups. The first is dominated by the initial sequence λέγουσιν/θέλουσιν/-γέτωσαν, the assonance of which is disguised in visual form, and revealed only once the poem is verbalised with the stress rhythm. In the second group of three verses the half-rhymes are even more pronounced: οὐ μέλι μοι/σὺ φίλι με/συνφέρι σοι. In the first group the half-rhymes occupy the first three positions in the trochaic tetrasyllable; in the second half, they run from caesura to caesura. This transition helps to emphasise the closural nature of the (catalectic) final verse's caesura.

Our text is visually as well as aurally poetic. The creators of the gemstone versions were clearly aiming for symmetry and elegant ‘diagrammatic’ patterns.Footnote 33 The Budapest version (no. 5) offers the most stunning example (Fig. 1).

Figure 1. The Budapest version (no. 5). Photograph: Aquincum Museum, reproduced with permission

The six lines are made of nine or eight letters. The first line, which has eight, is slightly distended; the fifth, which also has eight letters, has been elongated by the addition of what is apparently a leaf symbol (the same feature appears in catalogue entries nos. 3 and 7; one of the Paris cameos, no. 9, has μαι (phonetic) instead of με, presumably to fill out the line). The effect is to produce a square, with columnar diagrammatic patterns down either side. On the left, the ΛΕ of lines 1 and 3 and the ΑΘ of line 2 have been deliberately assimilated to create the impression of patterning; similarly in the last three lines the Ϲ has been almost closed, so as to assimilate it to Ο (note how much rounder it is than other sigmas), and so that the repetition of the second Υ is brought out. At the end of lines 1–3 we have ϹΙΝ/ϹΙΝ/ϹΑΝ, and then ΛΙΜΟΙ/ΛΙΜΕ and ΟΙ/ΟΙ. Other patterns include the diagonal run of epsilons going down from the first line. The letters start breaking apart from their positions in the words, and taking on independent lives as visual indices. The cameo takes on a mesmerising quality, as if the letters had a significance that went beyond their immediate function. We have not yet arrived in the fourth-century world of Optatian and the carmina cancellata, where the visual patterning of letters is so intense that the poems ‘vacillate between words and pictures’;Footnote 34 but we are, it seems, well en route.Footnote 35


Our text is, therefore, certainly poem-like, then, even if it differs markedly from a classical poem in its use of stress rhythm, its colloquialism, its inconsistent use of quantitative metre, its deployment of rhyme and its play with visual layout. I turn now from form to content. The text appears in two primary forms (leaving aside minor variations in spelling). The commoner version (nos. 2–12 in the catalogue) has six lines; no doubt no. 19 originally was of this kind too. Versions 13–18, however, lack the last two lines. In Appendix 2 I argue that no. 1, a graffito from Cartagena, originally consisted of the short version, and that the longer version was subsequently superimposed by a second hand.

This short version of the text means ‘they say what they like; let them say it; I don't care’. This reads as a popular-philosophical proverb: the speaker's view is set in defiant opposition to that of an unspecified majority. As a sentiment, this is almost infinitely adaptable, to suit practically any countercultural context. In a Greek context, however, the claim will have resonated as a claim to philosophical independence. The validation of an individual perspective in contrast to popular belief (δόξα) is of course a definitively philosophical stance, from at least Heraclitus and Parmenides onwards.Footnote 36 In the Roman era the opposition between δόξα and various modalities of reality (φύσις, τὰ πράγματα, ἡ ἀλήθɛια etc.) is a cornerstone of practical ethics.Footnote 37 But the shorter version does not specify exactly how or in what field this independent-mindedness manifests itself. Presumably that is the point: what is expressed here is primarily an attitude, adaptable to numerous different real-life scenarios.Footnote 38

The culminating οὐ μέλι μοι adds a more forceful note of self-assertion. It recalls the programmatic declarations of independence that characteristically mark philosophers as critically independent individuals. τί ἡμῖν … τῆς τῶν πολλῶν δόξης μέλει; asks Socrates in Plato's Crito.Footnote 39 Such strident assertions of ‘not caring’ are almost a badge of philosophical identity, particularly when it comes to death, about which Socrates and the Epicureans did not care.Footnote 40 Another popular inscription of the Roman Empire carries a prominent assertion of uncaring with philosophical overtones: the epitaph οὔκ ἤμην, ἐγενόμην, oὐκ εἴμι, οὐ μέλει μοι.Footnote 41 Both οὐ μέλει μοι inscriptions elevate the individual above the masses by borrowing from a philosophical trope (thus exemplifying the traffic between high philosophy and popular morality that has been well discussed by Teresa Morgan).Footnote 42

οὐ μέλει μοι, indeed, may even carry a countercultural charge. μέλει μοι means ‘I acknowledge my responsibility’: it is a marker of submission to social expectations. Homer's Hector famously says πόλεμος δ᾽ ἄνδρεσσι μελήσει | πᾶσι, μάλιστα δ᾽ ἐμοί (Il. 6.492–3), a form of words that also reappears with several different subjects in the Odyssey. For all the multiple ironies compressed into this sentence, for all the subtle prompting for the listener/reader to imagine Andromache's unspoken response, πόλεμος … μελήσει … ἐμοί remains a powerful articulation of the demands of the male citizen-warrior superego. To reject ‘care’ is therefore a defiant rejection of responsibility. We might think of Herodotus’ equally famous story of Hippoclides dancing away his marriage, with its epigrammatic conclusion: ‘that's of no concern to Hippoclides!’ (οὐ φροντὶς Ἱπποκλείδῃ, Hdt. 6.129). To reject ‘care’ in this way is to assert individualism within a social context that demands submission and recognition of obligation.Footnote 43

Most transmitted versions of the text, however, carry an extra two lines, which change its meaning. We shift suddenly from speaking abstractly about what ‘they’ say to a more dramatic relationship between ‘you’ (σύ, σοι) and the ‘me’ (μοι) introduced at the end of line 4. The aggressive imperative φίλι με and the presumptuous συνφέρι σοι create a new urgency. Gone is the vague proverbialising: readers are now summoned, qua addressees, into a metaleptic ‘drama of position’.Footnote 44 We are now in an apparently erotic scenario, where the words that ‘they say’ ask to be reread retrospectively as an expression of wider society's disapproval of an unconventional relationship. Catullus 5 provides an obvious parallel:

Viuamus mea Lesbia, atque amemus,
rumoresque senum seueriorum
omnes unius aestimemus assis!
Let's live, darling Lesbia, and let's love;
And let's price all the grouching
Of old grumps at one as!

In Catullus, as has often been noted, the idiom is mercantile: rejecting the rumores … senum seueriorum is couched in terms of revaluing, as if to subvert conventional society's preoccupation with finance and quantification.Footnote 45 In the Greek text, by contrast, we are given no indication of the subjects of the first three lines’ third-person verbs. The emphasis is less upon the social distinction between ‘we’ and ‘they’, and more on that between words (λέγουσιν, λεγέτωσαν) and the ἔργον of ‘love’ (if we take φίλι euphemistically).Footnote 46

The Hellenistic Anacreontea provide further parallels. Gallavotti noted that poem 7 W2 begins λέγουσιν αἱ γυναῖκες; the women, we are told, mock him for enjoying sensory pleasure while old, but the poet's contrary view is that the closer we get to death the more appropriate it is τὸ τερπνὰ παίζειν. Gallavotti concluded that in our text too the subject of λέγουσιν must be ‘the women’, and that the speaker is therefore an Anacreontic old man asserting that his virility is unimpaired. The Anacreontea certainly provide rich parallels for the rejection of δόξα in favour of sympotic delights (including sex) – indeed, richer still than Gallavotti saw. In particular, ‘not caring’ is a repeated theme. In poem 8 W2 the speaker asserts that the wealth of Gyges οὔ μοι μέλει; what, rather, ἐμοὶ μέλει (the phrase appears three times) is anointing his moustache with perfumes, wreathing his head with roses and living for today. Poem 45 asks τί μοι πόνων, τί μοι γόων, | τί μοι μέλει μεριμνῶν; (4–5). Fr. 4.3 asks τί Πλειάδων μέλει μοι, | τί γὰρ καλοῦ Βοώτου; (10–11).

‘Not caring’ for mainstream tastes is, therefore, not just a philosophical but also a distinctively Anacreontic twist on the lyric persona's idiosyncratic self-definition in defiance of popular tastes (most famously instantiated in Sappho fr. 16 V).Footnote 47 But Gallavotti goes too far in identifying the speaker and the scenario precisely.Footnote 48 As we shall see in the next section, the ‘users’ of this inscription were diverse and widely distributed. The text in fact avoids determining any specific scenario, allowing individuals instead to superimpose a scenario of their choice. Certainly the last lines strongly suggest erotic aggression, and even recall (albeit distantly) magical agōgai that seek to induce passion in the victim, and to lead them forcibly to the spellweaver.Footnote 49 One magical gemstone (now in the British Museum) carries the imperative inscription φίλει με in a context that suggests that the female owner was seeking to secure the ongoing affection of a partner.Footnote 50 But the range of relationships potentially covered by φιλεῖν is broad; this couplet need mean nothing more than ‘show me affection and you'll benefit from it’. Since we all spend much of our time wishing we were loved by one person or another, the sentiment is malleable enough to suit practically any wearer.


The few scholars who have considered our text have focused on trying to establish the demography of its wearers and the contexts of its use.Footnote 51 Gallavotti, as we have seen, believed it to be a poem in the Anacreontic tradition. Angelos Chaniotis, relatedly, has claimed that such inscriptions may have been read out at night (perhaps at symposia).Footnote 52 Gabriella Bevilacqua, by contrast, places it in a category of intimate formulae that appealed primarily to ‘un pubblico femminile’.Footnote 53 Given the plasticity of meaning established in the previous section, however, trying to identify one particular class of wearer or user seems the wrong approach. We can track the diversity of the inscription's uses along the axes of both geography and gender. In terms of geography, it was distributed across an extraordinarily wide geographical range: from Spain (no. 1) to Mesopotamia (no. 8). In terms of gender, item no. 5 was discovered around the neck of a young woman buried in what is now Hungary; while the shorter version of item no. 1 was scribbled on a wall in Spain by a man apparently called Eurypylus (see Appendix 2). The inscription's appeal was therefore not limited to one place or to one sex. Perhaps the Anacreontic parallels prompted men to read it as a piece of ironic sexual braggadocio, while parallels with female magical spells encouraged women to see it as an expression of desire to bind someone's affection. But it seems plausible that the text's crucial evolutionary adaptation in the literary contest of survival of the fittest was its very malleability, its openness to multiple renarrativisation in multiple different contexts.

Classicists are trained to locate cultural production with precise authors, individuals, frames; and to consider the reverse-engineering of those elements to be the scholar's primary duty. Who wrote it? In what polis? For what audience? To play what role? With what agenda? These are legitimate questions to ask of classical poetry, but our text is a different kind of animal. Whereas classical poetry achieves fixity and canonicity via its author function,Footnote 54 our text is anonymous;Footnote 55 it is the property not of its creator but of the network that sustains its circulation, and which authorises expansion, compression and variation at the level of detail.Footnote 56 Its real ‘author’ is its consumer, wearer and reader. In a mobile, internationalised economy, accessories that give cultural prestige are those that tie the wearer not to a particular point of origin but to a larger, pan-imperial, elite.

Our text, therefore, appealed not because it identified its wearer as a certain kind of person, but for precisely the opposite reason: because it allowed individuals to escape local pigeon-holing, and claim participation in an indeterminate network of translocal sophisticates who ‘get’ this kind of playful, elliptical discourse freighted with covert sexual aggression. Indeed, the text's content might be held to express precisely this rejection of the epichoric. ‘I don't care what they say’ articulates, as we have seen, a defiant individualism that differentiates those who imagine themselves in the subject position from the trivialities of gossip, and assimilates them instead to archetypes drawn from the literary tradition, whether morally obdurate philosophers or charismatically indulgent lyric personae. The text tells its readers that the logos of the unidentified ‘they’ is to be discounted; what matters is instead the reality of the intimacy shared between ‘you’ and ‘me’. At the same time, of course, the individuality proclaimed by the text is undermined by that text's broad dissemination. The individual who ventriloquises this text asserts autonomy, embodiment and intimacy, but – paradoxically – finds this in a pre-fabricated text. Let us consider briefly the one case where we can see our text ‘in action’, namely the Spanish graffito (no. 1 in the catalogue in Appendix 1):

Ευρ]ύπυλος λέγι ὃς κὲ Ἀ[…
λέγουσιν ἃ θέλο[υσιν
λεγέτωσαν⋅ οὐ †μελ{ισ}ο{υ̣α̣}
(σὺ) φίλι μ]ε, συνφέρι σοι⋅ ταῦ[τα
(προσ)έγ]ρ̣αψα ΣΓ            5

According to my hypothesis (see Appendix 2), the two horizontal lines enclose the original text (the nonsensical ending of which has been ‘emended’), while a different hand has added the final two lines. What is significant for our purposes is the emphasis upon individual identification. Eurypylus’ autobiographical claim to ‘say’ or ‘speak’ (λέγι) these words drags him into the narrative drama, which centres precisely upon the correct use of words (λέγουσιν, λεγέτωσαν). This shared vocabulary, together with the insertion of his own name within the horizontal lines, folds his own assertion of selfhood into the textual scenario (albeit unmetrically). The individual, Eurypylus, has found his own place within the text's capacious narrative space. But such a widely disseminated text cannot be decisively claimed by one individual. If my hypothesis is right, a second author, identifying him- or herself as ΣΓ, has added the eight syllables of the longer version, thus ‘capping’ Eurypylus not only by correcting his text but also by undermining his individualising claim to be the ‘speaker’ of a poem that is in fact common coin.

We are, perhaps, not far (in essence, if admittedly not in terms of scale) from the paradox of ‘mass individualism’ that has been identified as a characteristic of late-capitalist consumer cultureFootnote 57 (and which Monty Python's Life of Brian famously satirised).Footnote 58 The ‘I’ performed by our text is always pre-scripted, and whatever claims to anti-conformist individuality it allows one to perform are pre-rehearsed. This individualist paradox becomes prima facie even more marked if we consider the material form of the inscriptions. Some of these gems may have been worked in the same workshops (nos. 13 and 15 in the catalogue in Appendix 1). The glass-paste medallion (no. 5), for all its elegance, is probably a cheap replica created from a mould. No. 14 is an intaglio, which may have been used to seal documents. As Verity Platt notes, seals themselves create secondary images: they ‘combine the beauty and expense of precious stones and metals with a specific practical function, for the seal matrix – the carved image – can be replicated ad infinitum in a variety of pliable materials which are not precious at all’.Footnote 59 Even as the text insists on the embodied, erotic/affective individuality of the bearer, setting the singular μοι against the amorphous third-person plurality of λέγουσιν, the intaglio's material form exists as a material reminder of the text's multiple reproducibility.Footnote 60


Given the two lines of anacreontics identified by Gallavotti, and indeed the Anacreontic flavour of the ‘careless’ rejection of the words of others, it is possible that our text started out life as a quantitatively metical poem; in the canonical form in which it circulated, however, across the Roman Empire in the second and third centuries CE, its metre was tied to stressed rhythm. This makes it the earliest example of a Greek stress-based poem identified to date. But there is no great surprise in this: it is highly likely that stressed poetry was circulating in oral form long before it manifested itself in high literature. Indeed, its adoption for late-antique Christian hymns, designed as they were to appeal to a broad audience, is strong evidence that stressed poetry was at that stage deeply rooted in oral contexts. If we knew more about the oral culture of the High Empire we would no doubt have many more parallels.

The simple, alluring beat, coupled with its half-rhymes, must have been one reason for the text's popularity. Another was its adaptability into an elegant, patterned colometry that appealed aesthetically to the eye. But form is not the only explanation for its success. Our text allowed its bearer to stake a claim to individuality by rejecting social orthodoxy (what ‘they say’), and asserting instead a strong bond between ‘you’ and ‘me’. Such claims to individuality were, however, pre-scripted, in a double sense. First, the ‘careless’ rhetoric is borrowed from high literature and philosophy. The owner who says ‘I do not care what they say’ does in fact care what the classical litterati say. Second, the materiality of the gemstones themselves – the fact that they are churned out by workshops and exported over the Empire, and that intaglios at least might have been used to create new copies in wax or clay – expose the iterability of these claims to individuality. I have stressed that we need not see this as a failure or an undermining of the text's individualist message; rather, it is a reminder that the identity of the anti-social individualist is itself necessarily a social one. A gem, intimately embraced by the body, can be imagined as a prosthetic extension of one's own truest self; but it is also, and simultaneously, an alien object superimposed on the body, a reminder of our subjection to society's irresistible demands to purchase goods, to display them, to perform our identities through fashion.

The agency that shaped these texts and the artefacts that bore them is thus multiply distributed.Footnote 61 The individuals who created the gems were many: the poet, the miners, importers and crafters of the gem and its setting, the new owner who commissioned (or simply bought) it. At a more abstract level, the need for a text like this was created by new social pressures that were the result of the emergence of a pan-imperial, translocal Hellenism.Footnote 62 The paradox of large-scale societies is that while they homogenise cultural expression in terms of language, dress-codes and accessorisation, they also increase the pressure on individuals to seek out new ways of seceding from such normative pressures. These two Newtonian cultural forces are equipollent, leaving the individual suspended in an immobile state, neither fully individualised nor wholly part of the imagined community of the like-minded.

Let us turn in conclusion to our text's relationship to literate culture. The role of paideia (civilised education) as a marker of social distinction in the High Empire is now well established.Footnote 63 This cultural ‘superego’ can be detected in our text in the traces of quantitative verse and the distant allusions to philosophical and lyric personae; and, indeed, more generally in the lyric mise en scène that is presumed. Wearers of our text will be staking a claim, however indirect, to membership of the educated elite. At the same time, the text displays a number of markers of independence from the strict demands of classicism: subliterary diction and morphology, and most prominently the superimposed stress rhythm (borrowed from the popular uersus quadratus – whether in the well-attested Latin variety or in a Greek form that is otherwise unknown to us). Our text shows that it is aware of the poetic rules established by the canon; it simply does not care to abide by them (let the high theorists of quantitative metre say what they will). This text exemplifies both the pull of and the push against the normative classicism enshrined by the Second Sophistic. This tension was the stimulus for the creation of an experimental poem – let us finally give it that title – that was, apparently, unprecedented in the Greek world, and in formal terms at any rate astonishingly sophisticated.

Appendix 1: Catalogue

Our text is found widely, and almost exclusively, on gemstones (see however no. 1 for an important exception). Given the popularity of the inscription, and the difficulty of tracing gemstones, the following catalogue cannot claim to be complete. It does offer, however, a representative sample of the various possible variants (and it is unlikely that new publications would change the overall picture). All of the items are, to the very best of my knowledge, owned legitimately according to the 1972 UNESCO Antiquities and Art Treasures Act.

  1. 1. Graffito, Cartagena (Spain), 2nd–3rd cent. CE. From an upper-storey room. Lines 1–3 are written between two incised horizontal strokes.


    Ευρ]ύπυλος λέγι ὃς κὲ Ἀ[…

    λέγουσιν ἃ θέλο[υσιν

    λεγέτωσαν⋅ οὐ †μελ{ισ}ο{υ̣α̣}


    (σὺ) φίλι μ]ε, συνφέρι σοι⋅ ταῦ[τα

    (προσ)έγ]ρ̣αψα ΣΓ     5

    SEG 62.768 (A. Chaniotis, T. Corsten, N. Papazarkadas and R. A. Tybout) = Hispania Epigraphica 18 (2009) 133–4 no. 246 (J. Curbera) = IGEP 292 (M. P. de Hoz) = Stylow (Reference Stylow, Noguera and Madrid2009). On the text see Appendix 2 below.

  2. 2. Inscription on sardonyx gem. From the collection of Fulvio Orsini.

    λέγουσιν ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι ⋅ | σὺ φίλι με, | συνφέρι σοι

    SEG 44.1704 = Pannuti (Reference Pannuti1994) 337–9 n. 299.

  3. 3. Inscription on agate gem. Possibly from the Orsini Collection. Provenance unknown.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι ⋅ | σὺ φίλι με ❧, | συνφέρι σοι

    SEG 44.1704 = Pannuti (Reference Pannuti1994) 340 n. 301.

  4. 4. Inscription on sardonyx gem. Acquired in Egypt by Wilhelm Froehner (Bakhoum and Hellmann (Reference Bakhoum and Hellmann1992) 172).

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλει μοι⋅ ‖ σὺ φίλι με, | συνφέρι σοι

    Bibliothèque Nationale de France. SEG 42.1620.

  5. 5. Cameo on a medallion of glass paste. Found in a sarcophagus around the neck of a deceased young woman, Aquincum (Hungary). 2nd–3rd cent. CE.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μἐλι μοι⋅ | σὺ φίλι με, | συνφέρι σοι

    Budapest History Museum. SEG 29.1047 n. 15 = IGPannonia 96.

  6. 6. Medallion inscription.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι⋅ | σὺ φίλει με, | συνφέρει σοι

    Athens, Νομισματικό Μουσείο, Συλλογή Καραπάνου, αρ. ευρ. 734 (reported at Cabanis (Reference Cabanis2012) 312 n. 136).

  7. 7. Inscription on agate cameo.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι⋅ ‖ σὺ φίλι με ❧, | συμφέρι σοι

    British Museum. SEG 42.933.1a = Walters Reference Walters1926 no. 3707.

  8. 8. Inscribed cameo. ‘Aus der Sammlung von M. Peretie in Beirut: “La pierre a ete trouvée aux environs de Bagdad, sans doute dans les ruines de l'antique Seleucie”’ (Merkelbach and Stauber).

    λέγ[ουσιν] | ἅ θέλ[ουσιν]⋅ | λεγέτ[ωσαν], | οὐ μέλι μοι⋅ | σὺ φίλι με, συφέρι σοι

    IK Estremo oriente (Babylonia) 87 = Merkelbach and Stauber (Reference Merkelbach and Stauber2005) 508.

  9. 9. Inscribed sardonyx cameo. ‘Trouvé à Lutz, près d'Oroza (Hongrie)’ (Babelon).

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι⋅ | σὺ φίλι μαι, | συμφέρι σοι

    Bibliothèque Nationale de France. Babelon (Reference Babelon1897) 347 = Le Blant (Reference Le Blant1896) no. 150.

  10. 10. Inscribed sardonyx cameo.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι⋅ | σὺ φίλι με, | συμφέρι σοι

    Bibliothèque Nationale de France. Babelon (Reference Babelon1897) 348.

  11. 11. Agate. ‘Ex collect. Calveti medici’ (Boeckh ap. CIG).

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι ⋅ | σὺ φίλει με, | συνφέρι σοι

    CIG 7293.

  12. 12. ‘Fragment de camée copié à Rome, en 1885, au Musée de la Propagande’ (Le Blant)

    λέγ[…] | ἃ θέλ[…] | λεγέτ[…] | οὐ μέλε[…] | σὺ φίλει μ[.] | […]νφέρι[…]

    Le Blant (Reference Le Blant1896) no. 147.

  13. 13. Inscribed sardonyx cameo.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλει μοι

    Bibliothèque Nationale de France. Babelon (Reference Babelon1897) 349. Probably = CIG 7295, ‘gemma collectionis de Portici … ex Herculan.’ Same workshop as no. 15?

  14. 14. Gold pendant with black and white onyx intaglio with inscription, ca 3rd cent. CE. Unknown provenance.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι

    SEG 37.1750 = ‘Objects with Greek inscriptions on record’, in Sales Catalogue Sotheby's Monaco, 5th December 1987: Antiquités et Objets d'Art. Collection de Martine, Comtesse de Béhague provenant de la Succession du Marquis de Ganay (non vidi).

  15. 15. Inscription on onyx/sardonyx gem. Possibly from the Orsini Collection. Same workshop as no. 13?

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι

    SEG 44.1704 = Pannuti (Reference Pannuti1994) 339–40 no. 300.

  16. 16. Inscription on agate cameo.

    λέγουσιν | ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι

    British Museum. Walters (Reference Walters1926) 3706.

  17. 17. Onyx inscription, dated to the Roman imperial period. In the possession of ‘Dr Piperidis’ (Paribeni and Romanelli).

    λέγουσιν ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλι μοι

    Paribeni and Romanelli (Reference Paribeni and Romanelli1914) col. 25 no. 14.

  18. 18. ‘Camée vu en 1884 dans la collection de M. Auguste Castellani’ (Le Blant).

    λέγουσιν ἃ θέλουσιν⋅ | λεγέτωσαν, | οὐ μέλη μι

    Le Blant (Reference Le Blant1896) no. 148.

  19. 19. Agate/onyx. In the Pourtalès-Gorgier collection in Boeckh's time (the collection was sold in 1865).

    – – – | φίλι με | συμφέρι σοι

    CIG 7294 = Le Blant (Reference Le Blant1896) no. 146.

  20. 20. ‘Fragment communiqué par M. Auguste Castellani’ (Le Blant).

    λέγουσιν ἃ θέλου [ | ν […..] εγ […]| [vac. 3 lines]

    Le Blant (Reference Le Blant1896) no. 149.

Appendix 2

The graffito from Cartagena (Fig. 2) invites separate discussion. I am grateful to Professor Noguera Celdrán for supplying a high-resolution photograph, which has helped to clarify a number of issues. I begin with a conventional transcription:










]ύπυλοϲ λέγι ὃϲ κὲ Ἀ[…

]λέγουσιν ἃ θέλο[

]γέτωϲαν, οὐ μέλι̣ ϲ̣οι̣[


]ε ϲυνφέρι ϲοι⋅ ταυ̣[

]αψαϲγ 5

The obvious explanation for the horizontal lines is that they enclose the text. This would imply that the original inscription consisted of the first three lines.

  • Line 1. Clearly someone speaks (λέγι) the following words. As editors have seen, the most likely name to end in -ypylos is Eurypylos, but other options are possible.Footnote 64 Αt the end of the line, κὲ (thus accented) should be taken as a variant spelling of καὶ (so de Hoz):Footnote 65 the alternative, epic/dialectic κε (= ἄν),Footnote 66 is surely implausible. We are presumably to imagine ellipsis of a verb such as καλεῖται or προσαγορεύεται (cf. D. L. 1.79: Πιττακὸς … ὃς καὶ Μικρὸς προσηγορεύθη). The crossbar of the Α seems secure, but it is just about conceivable that it is a scratch disguising a Λ, in which case a phrase such as ὁ ϲκελλόϲ (‘the bandy-legged’) could be supplemented.

  • Line 2. Two horizontal strokes appear to be visible at the left-hand edge (see the detail in Fig. 3). Previous editors have identified a Ϲ, but this seems unlikely, given the unevenness of the marks (elsewhere sigmata are written in neat, compact loops). Expansion of the digital photograph reveals that these marks are continuation of the cracking of the plaster, and thus may be the result of damage caused by the building's collapse rather than design. Certainly the lower of the two looks unlikely to be intentional. The upper could be deliberate, but may not be part of a letter. Could it be a diacritical mark indicating the start of the quoted poem? It is worth noting that if we disregard this ‘letter’ then it would appear from the substantial gap before its start that line 2 is indented relative to line 1 (perhaps indicating the start of the quotation of the poem).

  • Line 3. The problems lie at the end, and they are severe (see the detail in Fig. 4). Should we read μέλι ϲοι, with other editors? (The gemstone versions of our text have μέλι μοι.) If the letter following the Λ is indeed an Ι then it cuts across the Λ; and if the next letter is indeed a Ϲ, it is unusually flat and narrow. Both letters are eccentric compared to the rest of the text. The Ο, however, is secure: the author characteristically writes this letter with two semi-circular strokes, as here. I believe that the original inscription read ΜΕΛΟ, and that the two strokes between Λ and Ο have been inserted later (note that the gap between Λ and Ο is consistent with spacing elsewhere). The letter following the Ο has been read as I, but this seems wrong: it shows a slanted stem and a second stroke, firm and deliberate, issuing diagonally up to the right from the centre of the stem; what letter might have been intended here is uncertain (an erratic Υ?). This is another oddity: whereas the rest of the writing is relatively neat, this stroke is longer than usual, slants and breaks down through the horizontal line below. We can also perhaps detect another diagonal stroke where the plaster is broken to the right: an Α or a Λ? (But there is no trace of the second diagonal that one would expect in both cases.) The text thus seems chaotic here. It seems that an original text containing the meaningless μελο has been ‘emended’ by at least one scrawling hand. μελο might be explained as a failed attempt at an first-person present-tense verb meaning ‘I care’ (perhaps an error for μέλω, or an unfinished μέλομαι; or the author may have been distracted by the example of θέλουσιν in the line above).Footnote 67

  • Line 4. We are now below the second horizontal line. The writing appears to be that of a different hand: the letters are larger, cruder and more uneven; we can also detect interverbal spacing. This hand may or may not be the one that made the additions to line 3. If Ε is to be read at the start of the line, we should no doubt fill in ϲὺ φίλι με. Expansion of the digital image (see Figs. 5 and 6) shows that ε could indeed be read: although neither of the two horizontal strokes is connected to the rest of a letter on the left, line 1 contains a close parallel for an ε written in this way. The letter seems to have been written by creating an upward loop beginning at the bottom, with the pressure released at the top (causing some jittering). A crossbar was then added, not necessarily connected to the loop. The gap following this letter is compatible with the interverbal spacing employed in the final two lines. Since this and the following line appear to be the work of a second, more haphazard, hand, perhaps the entirety of ϲὺ φίλι με was written on this line, even though this would have jutted out further to the left than previous lines. Alternatively, part of the phrase may have been added to the end of line 3 (although this would have encroached into Eurypylus’ inscription, which was contained between the horizontal lines). The left-hand margin of this line will have been further to the left than the line above, so perhaps just φίλι με (without ϲύ) was written: this would align with the start of Ευρ]ύπυλοϲ. But since the other left-hand margins seem to have been unequal, there is no need to presume the alignment of this one. ταῦτα seems inevitable at the end of the line.

  • Line 5 (including the end of line 4). ]ΑΨΑϹΓ is all that is clearly legible. Stylow and Curbera read ]ΛΑΨΑϹΓ, which leads them to the supplement ταῦ[τα β]λάψαϲ. (Curbera also wonders about κολάψαϲ, from κολάπτειν = ‘chisel’). There are indeed further marks to the left of ΑΨΑϹΓ (see the detail in Fig. 7): a vertical stem, and a connected thin, diagonal mark rising from the middle of the stem towards the right. This, however, is incompatible with Λ. I believe it could be a Ρ. The Ρ in line four is comparable (see the detail in Fig. 8): it has a strong vertical stem, and a sail-shaped loop consisting of a strong lower part and a weaker upper part (which would have to be completely invisible in an Ρ at the start of line five). This would be compatible with the suggestion of De Hoz and Chaniotis et al., viz. ταῦτα γράψαϲ γ, a plausible option in terms of sense. Suggestions for the final gamma include the numeral 3 (de Hoz: ‘habiendo escrito esto 3 veces’), an abbreviation for γράμματα (also de Hoz, translating as ‘habiendo escrito estas letras’), and an abbreviation for a name (Chaniotis et al.). The last is the most plausible. I propose in fact ταῦτα ἔγραψα ϹΓ, the last two referring to the name of the second author (no doubt a Roman-style double name or a name + patronymic). The left-hand margin will have been indented considerably relative to the one above, perhaps in imitation of the marginal variation of the text enclosed within the lines. Alternatively, προσέγρ]αψα ϹΓ would begin at roughly the same point as σὺ φίλι με in line 4.

My proposal, therefore, is that the original inscription, prior to correction, read:


Ευρ]ύπυλοϲ λέγι ὃϲ κὲ Ἀ[…

λέγουσιν ἃ θέλο[υϲιν

λεγέτωϲαν⋅ οὐ †μέλο [


Here is a complete edition of the original and the overlain inscriptions, together with an apparatus criticus:


Ευρ]ύπυλος λέγι ὃς κὲ Ἀ[…

λέγουσιν ἃ θέλο[υσιν

λεγέτωσαν⋅ οὐ †μελ{ισ}ο{υ̣α̣}


(σὺ) φίλι μ]ε, συνφέρι σοι⋅ ταῦ[τα

(προσ)έγ]ρ̣αψα ΣΓ       5

  1. 1 Ευρ]ύπυλος uel Ἀνθρούπυλος, Δρουπύλος         ὃς κὲ Ἀ[ Chaniotis et al. : ὅς κε Ἀ de Hoz : ΟΣΚΕΑ Stylow, Curbera

  2. 3 †μέλο manus prima : {ισ} et {υ̣α̣} add. alius uel alii

  3. 4 –5 (σὺ?) φίλι … ΣΓ add. alius

  4. 5 προσέγρ]αψα (uel ἔγρ]αψα) ΣΓ scripsi : γράψας Γ (Chaniotis et al.) : γράψας γ’ (uel γ[ράμματα]) de Hoz : βλάψας Γ Stylow, Curbera (uel κολάψαϲ Γ)

The first inscription, between the lines, is therefore the ‘short’ version of the text, apparently inscribed on the wall by Eurypylus. The second hand, identified as belonging to ΣΓ, ‘completed’ the text.

Figure 2. Graffito, Cartagena (Spain), 2nd–3rd cent. CE. From an upper-storey room. Photograph: José Miguel Noguera Celdrán, reproduced with permission

Figure 3. Detail of start of line 2: Ϲ or cracked/gouged plaster?

Figure 4. Detail of end of line 3: -Λ{ΙϹ}Ο{ΥΑ}?

Figure 5. Detail of the start of line 4: Ε?

Figure 6. A parallel from line 1? Ε, showing jittering at the top and floating crossbar

Figure 7. Detail of line 5: ΡΑ?

Figure 8. Detail of line 4: Ρ


I am grateful to a number of individuals for their generosity and insights, including Pavlos Avlamis, Marc Bonaventura, Ben Cartlidge, Renaud Gagné, Geoff Horrocks, Anna Lefteratou, José Miguel Noguera Celdrán, Verity Platt, Enrico Prodi, Zóltan Quittner, Katerina Kourtoglou, Jona Lendering and Leah Wild. CCJ's anonymous referee offered invaluable advice on phonetics and related issues. Versions of these arguments have been aired at the A Caucus seminar in the Faculty of Classics, Cambridge in October 2020, and at a meeting of the Classical Association of Aberdeen in February 2021.

1 Among book-length studies of the last twenty years see Höschele (Reference Höschele2006); Baumbach et al. (Reference Baumbach, Bär and Schmid-Dümmler2007); Bär (Reference Bär2009); Maciver (Reference Maciver2012); Miguélez Cavero (Reference Miguélez Cavero2013); Whitmarsh (Reference Whitmarsh2013) 137–210; Lightfoot (Reference Lightfoot2014); Benaissa (Reference Benaissa2018); Rosenmeyer (Reference Rosenmeyer2018); Greensmith (Reference Greensmith2020); Kneebone (Reference Kneebone2020); Perale (Reference Perale2020). I am grateful to a number of individuals for their generosity and insights, including Pavlos Avlamis, Marc Bonaventura, Ben Cartlidge, Renaud Gagné, Geoff Horrocks, Anna Lefteratou, José Miguel Noguera Celdrán, Verity Platt, Enrico Prodi, Zóltan Quittner, Katerina Kourtoglou, Jona Lendering and Leah Wild. CCJ's anonymous referee offered invaluable advice on phonetics and related issues. Versions of these arguments have been aired at the A Caucus seminar in the Faculty of Classics, Cambridge in October 2020, and at a meeting of the Classical Association of Aberdeen in February 2021.

2 Plut. Mor. 403a–9c (see Whitmarsh (Reference Whitmarsh2013) 197–9).

3 The epigraphic collection provided by Merkelbach and Stauber (Reference Merkelbach and Stauber1998–2004) offers a rich resource that has not yet been fully exploited.

4 Uncertainty arises because a number of nineteenth-century notices (among them Boeckh's minimalist entries in CIG) record inscriptions on items that were at the time in private collections; it is not always clear what has happened to these items since, and whether they are identical with others recorded more recently.

5 Threatte (Reference Threatte1980) 207, 199.

6 Vessella (Reference Vessella2018) 40–4.

7 Threatte (Reference Threatte1980) 610.

8 Threatte (Reference Threatte1980) 485–8, with 486 on omission of μ before φ (three instances, from the fifth and fourth centuries BCE); see also 485 on νύφη.

9 Apollonius Dyscolus Pron. p. 85 Schneider; Moeris α 27; Lesbonax, De figuris 13b. This advice is followed by practising Atticists such as Aelius Aristides and Athenaeus. Lucian's preference, however, is for -έτωσαν, a form that he even ascribes to his insanely obscurantist hyperatticist Lexiphanes (θελγέτωσαν, Lex. 23). ‘The shorter Attic -των is used eight times … These forms, however, all occur in the mock laws of the Saturnalia, where they are interspersed to add a legal flavour to the laws’ (Deferrari (Reference Deferrari1916) 20).

10 E.g. Nonnus Dion. 2.270, 43.159, 48.895.

11 Threatte (Reference Threatte1980) 485.

12 Gallavotti (Reference Gallavotti1988) 23–8.

13 Another anacreontic could be generated from e.g. ἃ θέλουσιν αἰτιῶνται (but the abyss of conjecture is deep).

14 Prosimetry is found elsewhere in the imperial period. Stramaglia (Reference Stramaglia1992) 136–9 notes its embrace by the novel in particular (cf. not only the Iolaus fragment = Stephens and Winkler (Reference Stephens and Winkler1995) 358–66 and Tinouphis = Stephens and Winkler (Reference Stephens and Winkler1995) 400–8, but also e.g. Petronius, Chariton, P.Turner 8, and episodes in The Alexander Romance and Apollonius King of Tyre). The phenomenon can however also be detected in a broader range of texts: see the (elliptically brief) survey of West (Reference West1982) 164–5. One might also consider texts such as Lucian's Charon (Deriu Reference Deriu2015) or even Athenaeus’ Deipnosophists, where verse is ‘quoted’ (as it is in Chariton) within prose.

15 See Maas (Reference Maas1963) 512 on the non-accentuation in the kontakia of Romanos of prepositives (a class in which he includes both relatives and pronouns (see additionally Maas (Reference Maas and Lloyd-Jones1962) 84)). To my knowledge the only acknowledgement of the stress accent in our text is at Politis (Reference Politis1911–12 [1920–31]) 191 (where, tellingly, it is misdated to the Byzantine period; Politis also appears to misunderstand the metre). I am grateful to Katerina Kourtoglou for this reference.

16 Much remains uncertain regarding the chronology, and the order in which different vowels and diphthongs collapsed, but it is clear that the phenomenon begins in the Hellenistic period (e.g. Horrocks (Reference Horrocks2010) 167). There is much debate over the role of accent in classical poetics: for a recent, radical proposal see David (Reference David2006.

17 No. 3 in the Carmina moralia (Migne, PG 37.521–968).

18 Meyer (Reference Meyer1885) 49–51, 313–15, 400–9. Stress-based prose clausulae are also attested from around 400 CE (Norden (Reference Norden1923) 922–3).

19 Allen (Reference Allen1973) 267–8 discusses Babrius, raising the possibility that ‘phonetic prominence had come to be associated with the falling melodic pattern’ (268); see contra Devine and Stephens (Reference Devine and Stephens1985) 136–7; Luzzatto (Reference Luzzatto1985). On paroxytonesis in meiouric hexameters see Dihle (Reference Dihle1954) 184–5; West (Reference West1982) 173–4. There are some signs of regularisation of the position of the stress accent in Nonnus (Miguélez Cavero Reference Miguélez Cavero2008) 109).

20 The earliest surviving examples of elaborate metres based around stress are usually held to come in the kontakion form, apparently originated by Romanos in the sixth century CE. On these metres see Maas (Reference Maas1963) 511–38; Maas and Trypanis (Reference Maas and Trypanis1970) 210–17; Koder (Reference Koder1983); Lauxtermann (Reference Lauxtermann1999) 55–68.

21 The third syllable of the trochaic verse is not accentuated; secondary stress, however, is generated by the momentary pause at the caesura.

22 Luque Moreno (Reference Luque Moreno2017).

23 See esp. Zinn (Reference Zinn1997) [1940] on Horace. Controversy surrounds the role of ictus, a ‘beat’ used (according to late antique grammarians) to teach quantitative metre: see e.g. Zeleny (Reference Zeleny2008) (forcefully contra), and Becker (Reference Becker2010), arguing for coincidence of ictus and word accent in the Sapphics of Horace's Carmen saeculare.

24 Suet. Cal. 6.1; see further Fraenkel (Reference Fraenkel1927) 360–5; Jeffreys (Reference Jeffreys1974) 183–5.

25 As above, I follow the Byzantine accentuation for these purposes. Politis (Reference Politis1911–12 [1920–31]) 191 n. 14 inexplicably claims to identify two trochaic tetrameters + one iambic tetrameter.

26 Jeffreys (Reference Jeffreys1974) 171; Koder (Reference Koder1983); Lauxtermann (Reference Lauxtermann1999) 35–6 (and note also that the ‘pairing of colons, especially hepta- and octosyllabic ones, can be traced back to the late fourth century’, 59–60).

27 Horrocks (Reference Horrocks2010) 328. Byzantine critics, indeed, often confuse the two (Jeffreys (Reference Jeffreys1974) 183).

28 Lauxtermann (Reference Lauxtermann1999).

29 Horrocks (Reference Horrocks2010) 328. In sixth-century accentual octosyllables the proportion of iambics to trochaics is approximately 2/3 to 1/3 (Lauxtermann (Reference Lauxtermann1999) 52).

30 Jeffreys (Reference Jeffreys1974) 184–95. Jeffreys points out an early seventh-century court acclamation in accentual trochaic tetrameters, incorporating a catalectic fourth line (187–8; text from Maas (Reference Maas1912) 34): και οὐδείς τολμᾴ λαλήσαι, ἀλλ᾽ ὅλους ἐφίμωσεν. This offers a clear metrical parallel for our text (albeit with the caesura after the eighth rather than the seventh syllable). In the tenth century Symeon the New Theologian wrote in a variety of stress metres including the trochaic octasyllable: since he ‘merely repeats what he heard being sung by the common folk, the trochaic octasyllable is likely to have existed for quite a long time in popular songs. But for how long a period, I cannot say with absolute certainty’ (Lauxtermann (Reference Lauxtermann1999) 53). From the time of the High Roman Empire?

31 Fraenkel (Reference Fraenkel1927), citing such lines as Ar. Eq. 247 (παῖε παῖε τὸν πανοῦργον καὶ ταραξιππόστρατον). On this metre see West (Reference West1982) 40–2, 91–2.

32 Byzantine poetry offers some parallels for the combination of quantitative and stress-based schemes: see Lauxtermann (Reference Lauxtermann1999) 44.

33 On the ‘diagrammatic’ quality of some imperial Greek poetry see Whitmarsh (Reference Whitmarsh, Galinsky and Lapatin2016). Renaud Gagné points out to me that the first five lines of the gemstone format supply the acrostic λάλος (‘eloquent’). An anonymous referee for CCJ speculates that Lalos may even be a proper name, noting its particular prevalence in southern Italy.

34 Squire (Reference Squire2016) 184.

35 Marc Bonaventura has pointed out to me that the layout places the iotacised ΜΕΛΙ in the centre of the visual field; this may be a visual pun, given that μέλι = ‘honey’ is frequently used as an endearment (Bevilacqua (Reference Bevilacqua1991) 230–1).

36 Arguably all ‘rationalistic’ stances are similarly anti-populist: cf. Hecataeus’ τάδε γράφω, ὥς μοι δοκεῖ ἀληθέα εἶναι⋅ οἱ γὰρ Ἑλλήνων λόγοι πολλοί τε καὶ γελοῖοι, ὡς ἐμοὶ φαίνονται, εἰσίν (1a FGrH F1).

37 Van Hoof (Reference Van Hoof2010) 106, 126, 131.

38 One curious example of this plasticity is to be found in early-modern Scotland: a Scots–English version adapted from the first two lines (‘They haif said. | Quhat say they? | Lat thame say’) was adopted as a motto by the Earls Marischal. The phrasing, though not identical, is close enough to the Greek to suggest kinship. Did one of the Earls perhaps own such a gem? In 1593 the motto was inscribed on Marischal College in Aberdeen, a Protestant rival to the Catholic King's College, the ancestor of the modern University of Aberdeen. No doubt in this context this was a bold statement of the truth of Protestant belief, as distinct from Catholic ritual and dogmatism. It subsequently took on a life of its own: the same saying was mimetically reproduced over many Scottish lintels, as Middleton (Reference Middleton1891) 95 notes (identifying this as a ‘loose translation’ of our text). In the nineteenth century this saying gained a new life throughout the British Isles in the Latinised form Aiunt. Quid Aiunt? Aiant. The derivation of this motto from our Greek text was the subject of discussion among British amateur scholars of the 1870s: see the letter of Mr R. Hill of Bournemouth to The Guardian published on 27 November 1878, and the entry of E. T. M. Walker in Notes and Queries of 11 January 1879.

39 Pl. Cr. 44c; for comparable declarations of not caring about public opinion see Menex. 99e, Euthyd. 303c, Hi. Mi. 369d etc.

40 In Plato's Apology Socrates claims that ἐμοὶ θανάτου … μέλει … οὐδ’ ὁτιοῦν (Pl. Ap. 32d), a phrase adapted into the Epicurean slogan ὁ θάνατος οὐδὲν πρὸς ἡμᾶς (Kuria doxa 2, 11; cf. Ep. Menoec. 124–5).

41 E.g. OMS vi.109–11, GVI 1135, IGUR iii.1283, 1397, 1398. The Latin version (non fui, fui, non sum, non curo) is also widely attested. This inscription can be plausibly linked to the Epicurean ‘symmetry argument’ (Lucr. 3.832–42, 972–5), to the effect that death should not matter to us because pre-natal non-existence did (or does: see Warren (Reference Warren2004) 57–100) not matter to us. See Lattimore (Reference Lattimore1962 [1942]) 83–6 on the ‘Lucretian’ quality of such epitaphs.

42 Morgan (Reference Morgan2007) 274–99, 333–40. See esp. 299: ‘The best we can do to characterize the relationship is probably to say that in high philosophy and popular ethics we find two streams of culture, ultimately rising from many of the same sources, which sometimes mingle, each influencing the other, and sometimes run separately, along roughly parallel terrain.’

43 ‘We're pretty vacant’, sang The Sex Pistols, ‘and we don't care’.

44 Fitzgerald (Reference Fitzgerald1995).

45 ‘By applying monetary standards to human worth, Catullus implicitly negates the world to which the old men belong, which makes money – numerical quantification – the primary means of human exchange and validation. In exposing the absurdity of rendering human worth accountable, he subverts the mechanism of accounting altogether’ (Greene (Reference Greene and Skinner2007) 135). The gemstone inscription ἐγὼ δέ γε οὐ δίδωμί σοι οὐδὲ κόλλυβον (SEG 63.174) may exploit a similar kind of metaphor (‘I don't give tuppence for you’, i.e. ‘stuff you’), but other interpretations are possible (Zellmann-Rohrer (Reference Zellmann-Rohrer2018) 292–5).

46 It could even be a euphemism for ‘have sex with’, as at Asclepiades 25 (= AP 5.181) 11, with Sens (Reference Sens2011) 171. Other possible interpretations are canvassed below.

47 Zellner (Reference Zellner2007), with further literature.

48 Leah Wild has ingeniously suggested to me that the speaker might be the gem itself, inviting the bearer to kiss it so as to activate its power to ward off the malign talk of others. This would create a parallel with the talking stones, books etc. of Hellenistic epigram (Tueller Reference Tueller2008). This seems to me an eminently plausible way of reading the text, especially given (as Wild additionally notes) that a pendant would be worn intimately about the person. Nevertheless, it is unlikely to be the only, or even the dominant, reading (it would not work e.g. for the graffito).

49 Faraone (Reference Faraone1999) 55–69, emphasising the language of violent coercion that pervades such spells.

50 Faraone (Reference Faraone1999) 101. Faraone also notes that AP 5.158 (= Asclepiades iv Sens) speaks of a female owner of a belt inscribed with this phrase.

51 I shall continue to speak in general terms of ‘wearers’, but it is worth emphasising that we do not know in every case how the gem was borne. No. 5, clearly, was worn as a pendant, and others (e.g. 9, 13) apparently have loops through which a necklace could be threaded. But the casings may not be antique in every case, and without systematic expert analysis it is risky to make assumptions.

53 Bevilacqua (Reference Bevilacqua1991) 226.

54 Insightful discussion at Netz (Reference Netz2020) 96–136.

55 I am unconvinced that [Εὐρ]ύπυλος, who on the graffito claims to ‘speak’ (λέγι) the text, is the original author (as claimed by Stylow (Reference Stylow, Noguera and Madrid2009) 264 and Chaniotis et al. (Reference Chaniotis, Corsten, Papazarkadas and Tybout2012), the editors of SEG 62.768).

56 See Selden (Reference Selden2010) on such anonymous ‘text networks’, ‘autopoietic bodies of related compositions whose origins largely escape us and whose evolution, in the second and third centuries C.E., remained far from complete’ (8). Selden focuses on larger-scale prose texts, but the point is the same.

57 See e.g. Mackinney-Valentin (Reference Mackinney-Valentin2014).

58 Brian: ‘You're all individuals!’ Crowd: ‘We're all individuals!’ Brian: ‘You're all different!’ Crowd: ‘Yes, we're all different!’ Solitary voice: ‘I'm not.’ Crowd: ‘Sssh.’

59 Platt (Reference Platt2006) 238.

60 ‘Seals thus combine an intimate relationship between owner and object with a more widely circulated replicated image which acts as a public marker of the physical presence of the private self’ (Platt (Reference Platt2006) 241).

61 I borrow here the language of Gell (Reference Gell1998).

64 LGPN also gives Ἀνθρούπυλος and Δρουπύλος (and Εὀρύπυλος, a variant of Εὐρύπυλος). For references to previous editions see Appendix 1, no. 1.

65 De Hoz however prints ὅϲ κε.

66 Thus apparently Stylow, who translates as ‘quienquiera’.

67 μέλομαι = ‘I care for’ is exclusively poetic (LSJ s.v. ii).


Allen, W. S. (1973) Accent and rhythm, Cambridge.Google Scholar
Babelon, E. (1897) Catalogue des camées antiques et modernes de la Bibliothèque Nationale, Paris.Google Scholar
Bakhoum, S. and Hellmann, M.-C. (1992) ‘Wilhelm Froehner, le commerce et les collections d'antiquités égyptiennes’, JS 1 (n.s.), 155–86.Google Scholar
Bär, S. (2009) Quintus Smyrnaeus Posthomerica 1: Die Wiedergeburt des Epos aus dem Geiste der Amazonomachie. Mit einem Kommentar zu den Versen 1–219, Göttingen.Google Scholar
Baumbach, M., Bär, S. and Schmid-Dümmler, N. N. (eds.) (2007) Quintus Smyrnaeus: transforming Homer in Second Sophistic epic, Berlin.CrossRefGoogle Scholar
Becker, A. (2010) ‘Listening to lyric: accent and ictus in the Latin Sapphic stanza’, CW 103: 159–82.Google Scholar
Benaissa, A. (2018) Dionysius: the epic fragments. Edited with introduction, translation, and commentary, Cambridge.CrossRefGoogle Scholar
Bevilacqua, G. (1991) ‘Osservazioni su alcune formule affettuose e galanti di età imperiale’, MGR 16, 225–37.Google Scholar
Cabanis, P. (2012) ‘ΤΑ ΦΥΛΑΚΤΑ ΤΗΣ ΥΣΤΕΡΗΣ ΑΡΧΑΙΟΤΗΤΑΣ: η διατήρηση και ο μετασχηματισμός τους από τη χριστιανική κοινωνία’, diss., University of Thessaloniki.Google Scholar
Chaniotis, A. (2019) ‘Epigraphy of the night’, in Noreña, C. F. and Papazarkadas, N. (eds.), From document to history: epigraphic insights into the Greco-Roman world, Leiden, 1336.Google Scholar
Chaniotis, A., Corsten, T., Papazarkadas, N. and Tybout, R. A. (2012) ‘Maxim on a wall painting (graffito)’, SEG 62, 768.Google Scholar
David, A. P. (2006) The dance of the Muses: choral theory and ancient Greek poetics, Oxford.CrossRefGoogle Scholar
Deferrari, J. (1916) Lucian's Atticism: the morphology of the verb, Princeton.Google Scholar
Deriu, M. (2015) ‘“Prosimetrum”, impresa e personaggi satirici nei Contemplantes di Luciano di Samosata’, Lexis 33, 400–16.Google Scholar
Devine, A. M. and Stephens, L. D. (1985) ‘Stress in Greek?’, TAPhA 115, 125–52.Google Scholar
Dihle, A. (1954) ‘Die Anfänge der Griechischen Akzentuierenden Verskunst’, Hermes 82: 182–99.Google Scholar
Faraone, C. (1999) Ancient Greek love magic, Cambridge, MA.Google Scholar
Fitzgerald, W. (1995) Catullan provocations: lyric poetry and the drama of position, Berkeley.CrossRefGoogle Scholar
Fraenkel, E. (1927) ‘Die Vorgeschichte des Versus quadratus’, Hermes 62, 357–70.Google Scholar
Gallavotti, C. (1988) ‘Revisione di testi epigrafici’. Part 2, Bollettino dei Classici 9, 2139.Google Scholar
Gell, A. (1998) Art and agency: an anthropological theory, Oxford.Google Scholar
Gleason, M. (1995) Making men: sophists and self-presentation in ancient Rome, Princeton.Google Scholar
Greene, E. (2007) ‘Catullus and Sappho’, in Skinner, M. (ed.), A companion to Catullus, Malden, 129–50.CrossRefGoogle Scholar
Greensmith, E. (2020) The resurrection of Homer in imperial Greek epic: Quintus Smyrnaeus’ Posthomerica and the poetics of impersonation, Cambridge.CrossRefGoogle Scholar
Horrocks, G. (2010) Greek: a history of the language and its speakers, 2nd edn, Chichester.CrossRefGoogle Scholar
Höschele, R. (2006) Verrückt nach Frauen: der epigrammatiker Rufin, Tübingen.Google Scholar
Jeffreys, M. (1974) ‘The nature and origins of the political verse’, Dumbarton Oaks Papers 28, 141–95.CrossRefGoogle Scholar
Kneebone, E. (2020) Oppian's Halieutica: charting a didactic epic, Cambridge.CrossRefGoogle Scholar
Koder, J. (1983) ‘Kontakion und politischer Vers’, JöByz 33, 4556.Google Scholar
Lattimore, R. (1962 [1940]) Themes in Greek and Latin epitaphs, Urbana.Google Scholar
Lauxtermann, M. (1999) The spring of rhythm: an essay on the political verse and other Byzantine metres, Vienna.Google Scholar
Le Blant, E. (1896) 750 inscriptions de pierres gravées inédites ou peu connues, Paris.Google Scholar
Lightfoot, J. L. (2014) Dionysius Periegetes: Description of the known world. With introduction, translation, and commentary, Oxford.Google Scholar
Luque Moreno, J. (2017) ‘Trochaicus idem septenarius et quadratus’, Estudios Clásicos 152, 111–25.Google Scholar
Luzzatto, M. H. (1985) ‘Fra poesia e retorica. La clausola del “Coliambo” di Babrio’, QUCC 19, 97127.Google Scholar
Maas, P. (1912) ‘Metrische Akklamationen der Byzantiner’, BZ 21, 2851.Google Scholar
Maas, P. (1962) Greek metre, trans. Lloyd-Jones, H., Oxford.Google Scholar
Maas, P. (1963) Sancti Romani melodi cantica: cantica genuina, Oxford.Google Scholar
Maas, P. and Trypanis, C. A. (1970) Sancti Romani melodi cantica: cantica dubia, Berlin.CrossRefGoogle Scholar
Maciver, C. (2012) Quintus Smyrnaeus’ Posthomerica: engaging Homer in late antiquity, Leiden.CrossRefGoogle Scholar
Mackinney-Valentin, M. (2014) ‘Mass-individualism: Converse All Stars and the paradox of sartorial sameness’, Clothing Cultures 1.2, 127–42.CrossRefGoogle Scholar
Merkelbach, R. and Stauber, J. (1998–2004) Steinepigramme aus dem griechischen Osten, Stuttgart.Google Scholar
Merkelbach, R. and Stauber, J. (2005) Jenseits des Euphrat: griechische Inschriften. Ein epigraphisches Lesebuch, Berlin.Google Scholar
Meyer, W. (1885) ‘Anfang und Ursprung der lateinischen und griechischen rythmischen Dichtung’, Abhandlungen der philosophisch-philologischen Classe der königlich bayerischen Akademie der Wissenschaften, 17, 265450. Repr. in W. Meyer, Gesammelte Abhandlungen zur mittellateinischen Rythmik, 3 vols. (vol. iii ed. W. Bulst), Berlin, 1905–36, ii.1–201.Google Scholar
Middleton, J. H. (1891) Engraved gems of classical times: with a catalogue of the gems in the Fitzwilliam Museum, Cambridge.Google Scholar
Miguélez Cavero, L. (2008) Poems in context: Greek poetry in the Egyptian Thebaid 200–600 AD, Berlin.CrossRefGoogle Scholar
Miguélez Cavero, L. (2013) Triphiodorus: The sack of Troy. A general study and a commentary, Berlin.CrossRefGoogle Scholar
Morgan, T. (2007) Popular morality in the early Roman Empire, Cambridge.CrossRefGoogle Scholar
Netz, R. (2020) Space, scale and canon in ancient literary culture, Cambridge.CrossRefGoogle Scholar
Norden, E. (1923) Die antike Kunstprosa vom vi. Jahrhundert v. Chr. bis in die Zeit der Renaissance. Volume ii, 4th edn, Berlin.Google Scholar
Pannuti, U. (1994) La collezione glittica del Museo archeologico nazionale di Napoli. Volume ii, Rome.Google Scholar
Paribeni, R. and Romanelli, P. (1914) ‘Studi e ricerche archeologiche nell'Anatolia meridionale’, Monumenti Antichi Pubblicati per Cura della Reale Accademia dei Lincei 23, 5274.Google Scholar
Perale, M. (2020) Adespota papyracea hexametra Graeca (APHEX): Hexameters of unknown or uncertain authorship from Graeco-Roman Egypt. Volume i: Adespota papyracea hexametra Graeca (APHex i): Hexameters of unknown or uncertain authorship from Graeco-Roman Egypt, Berlin.CrossRefGoogle Scholar
Platt, V. (2006) ‘Making an impression: replication and the ontology of the Graeco-Roman seal stone’, Art History 29, 233–57.CrossRefGoogle Scholar
Politis, N. G. (1911–12) ‘Δημώδη Βυζαντινά άσματα’, Λαογραφία 3, 622–52. Repr. in and refd from Λαογραφικά Σύμμεικτα (Athens, 1920–31), iv.186–209.Google Scholar
Rosenmeyer, P. A. (2018) The language of ruins: Greek and Latin inscriptions on the Memnon Colossus, New York.CrossRefGoogle Scholar
Schmitz, T. (1997) Bildung und Macht: zur sozialen und politischen Funktion der zweiten Sophistik in der griechischen Welt der Kaiserzeit, Munich.Google Scholar
Selden, D. (2010) ‘Text networks’, AN 8, 123.Google Scholar
Sens, A. (2011) Asclepiades of Samos: epigrams and fragments. Edited with translation and commentary, Oxford.Google Scholar
Squire, M. (2016) ‘How to read a Roman portrait? Optatian, Porfyry, Constantine and the Vultus Augusti’, PBSR 84, 179240.Google Scholar
Stephens, S. A. and Winkler, J. K. (1995) Ancient Greek novels: the fragments. Introduction, text, translation, and commentary, Princeton.CrossRefGoogle Scholar
Stramaglia, A. (1992) ‘Prosimetria narrativa e “romanzo perduto”: PTurner 8 (Con discussione e riedizione di PSI 151 [Pack2 2624] + PMil Vogliano 260)’, ZPE 92, 121–49.Google Scholar
Stylow, A. U. (2009) ‘viii. Catálogo. 10. Fragmento de panel pictórico de color rojo con titulus griego inciso’, in Noguera, J. M. and Madrid, M. J. (eds.), Arx Hasdrubalis: la ciudad reencontrada. Arqueología en el cerro del Molinete, Cartagena, Murcia, 264–5.Google Scholar
Swain, S. (1996) Hellenism and empire: language, classicism, and power in the Greek world AD 50–250, Oxford.Google Scholar
Threatte, L. (1980) The grammar of Greek inscriptions. Volume i: Phonology, Berlin.Google Scholar
Tueller, M. A. (2008) Look who's talking: innovations in voice and identity in Hellenistic epigram, Leuven.Google Scholar
Van Hoof, L. (2010) Plutarch's practical ethics: the social dynamics of philosophy, Oxford.CrossRefGoogle Scholar
Vessella, C. (2018) Sophisticated speakers: Atticistic pronunciation in the Atticist lexica, Berlin and Boston.CrossRefGoogle Scholar
Walters, H. B. (1926) Catalogue of engraved gems & cameos, Greek, Etruscan & Roman in the British Museum, London.Google Scholar
Warren, J. (2004) Facing death: Epicurus and his critics, Oxford.CrossRefGoogle Scholar
West, M. L. (1982) Greek metre, Oxford.Google Scholar
Whitmarsh, T. (2001) Greek literature and the Roman Empire: the politics of imitation, Oxford.Google Scholar
Whitmarsh, T. (2010) ‘Thinking local’, in Whitmarsh, T. (ed.), Local knowledge and microidentities in the Roman Greek world, Cambridge, 116.Google Scholar
Whitmarsh, T. (2013) Beyond the Second Sophistic: adventures in Greek postclassicism, Berkeley.Google Scholar
Whitmarsh, T. (2016) ‘The mnemology of empire and resistance: memory, oblivion and periegesis in imperial Greek culture’, in Galinsky, K. and Lapatin, K. (eds.), Cultural memories in the Roman Empire, Los Angeles, 4964.Google Scholar
Zeleny, K. (2008) Itali Modi: Akzentrhythmen in der lateinischen Dichtung der augusteischen Zeit, Vienna.CrossRefGoogle Scholar
Zellmann-Rohrer, M. (2018) ‘Notes on Greek jewelry-inscriptions’, GRBS 58, 292–8.Google Scholar
Zellner, H. (2007) ‘Sappho's alleged proof of aesthetic relativity’, GRBS 47, 257–70.Google Scholar
Zinn, E. (1997) [1940] Der Wortakzent in den lyrischen Versen des Horaz: mit einem Nachwort zur Neuauflage von Wilfried Stroh, Hildesheim.Google Scholar
Figure 0

Figure 1

Figure 2

Table 1 Trochaic scansion by stress rhythm

Figure 3

Figure 1. The Budapest version (no. 5). Photograph: Aquincum Museum, reproduced with permission

You have Access

Linked content

Please note a has been issued for this article.

Send article to Kindle

To send this article to your Kindle, first ensure is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

Note you can select to send to either the or variations. ‘’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Available formats

Send article to Dropbox

To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

Available formats

Send article to Google Drive

To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

Available formats

Reply to: Submit a response

Please enter your response.

Your details

Please enter a valid email address.

Conflicting interests

Do you have any conflicting interests? *