The Song of Seikilos: a Musically Notated Ancient Greek Poem

Armand D’Angour

So-called “Lyric” poetry, as composed by Ancient Greek poets of the 6th century BC such as Sappho, Alcaeus, and Anacreon, properly means “poetry sung to the lyre”. The words of their poems were originally sung, but what survives of the songs are solely the rhythmically composed “lyrics” without their melodies; forms of melodic notation were devised in the 5th century BC at the earliest.[1]

There is, however, an example of such a song that does survive complete with its melody and, though much later in date than the bulk of the lyric corpus, throws an intriguing light on the principles of composition of an Ancient Greek song. This is the Song of Seikilos, a four-verse poem which survives inscribed in six rows of elegant Greek uppercase letters on a cylindrical marble column now in the National Museum of Copenhagen. It is dated by its letter forms to the 2nd century AD.[2]

The column, and a detail of the words inscribed around it (National Museum of Denmark, Copenhagen).

Words and music

The Seikilos Song is prefaced by an unmelodised verse couplet that reads:

I am an image and a stone; Seikilos sets me up here

      As a long-lasting marker of undying memory.

There follows the song itself, a short text of 17 words, above which sit small alphabetic signs, C Z K I etc., representing the pitches of the melody as anciently notated for voice. On top of a few of those signs are found some linear symbols and dots. The linear symbols are used to indicate the intended duration of some of the words’ syllables (– for a double time unit, ⏗ for a triple time unit) and the dot indicates the movement of the sung line (a rhythmical or melodic rise is heard on the dotted elements of the line):

Transcribed into modern stave notation, the piece can be sung by a male or female voice or voices, at any suitable pitch and tempo, and accompanied by strings or woodwind as desired. It should be clear that the idea of a standardised performance is entirely foreign to music before the age of recording. The shape of the melody has unmistake­able aural links to European music of much later centuries, such as 10th-century Gregorian plainsong:

A rendering of the song into English words with the same number of syllables (i.e. an isosyllabic translation) allows the melody to be sung as follows. The notes of a modern scale, here using G as the tonic, are placed above the words’ vowels, and a transliteration of the Greek is added underneath:

Interpreting the Stone

The pillar on which the Song is written was discovered in 1883 near the town of Aydın, Turkey, once the Greco-Roman city of Tralles in Asia Minor, during the construction of a railroad under the direction of a Mr Edward Purser. The alarming and droll story of the find tells that Mrs Purser decided that it would make a good flowerpot stand, and had the base sawn straight, thus destroying what might have been a vital part of the text. Fortunately, before she did so a copy had been taken (i.e. a ‘squeeze’ – paper pressed onto the stone and rubbed to leave an impression) preserving at least some of the surviving bottom line. This shows that the word ZĒI, “is alive”, had been inscribed at the end underneath the author’s name, and that the name SEIKILOS was itself followed by another name of which only the first part survives as EUTER-.

ZĒI – “s/he is alive” – was a conventional formula on grave markers indicating that the dedicator had survived the dedicatee to create a living memorial for him or her. This has led to the song/stone being given the title of the “Seikilos Epitaph”, and the fragmentary “EUTER…” being taken either as the name of a deceased wife (ΕΥΤΕΡΠΗΙ, “to Euterpe”) or Seikilos’s patronymic (ΕΥΤΕΡΠΟΥ, Seikilos “son of Euterpes”).

These suppositions, along with a number of other features of the traditional interpretation, are open to challenge. In particular, I think that the song and inscription was not in fact intended as an epitaph, although it deliberately and almost cheekily adopts the common epitaphic formula that has misled interpreters. I will argue instead that the inscription was created by its author Seikilos as a monument to preserve for posterity evidence of his musical and poetic inventiveness – as indeed it has.

Preliminary evidence for this view is provided by the poetic couplet that precedes the song. It comprises in Greek an elegiac distich (dactylic hexameter followed by pentameter). As with the song, its words are wrapped around the column as they happen to fall, in five rows (the layout is indicated by the slashes added to the Greek text below):

I am an image and a stone; Seikilos sets me up here

   as a long-lasting marker of undying memory.



The opening words are standardly translated “I am a stone image”, taking the second word (H) as the definite article (ἡ) used as a semi-demonstrative: “I, this stone, am an image.” But this is awkward in Greek, and H is better interpret­ed as ἤ, “or”. The column is speaking (“I am”), and presenting itself as an image of text OR a block of stone; it is either or both (i.e. the “or” is conjunctive, as in “for better or for worse”, not disjunctive as in “either here or there”). The opening statement immediately draws attention to the stone itself and the inscription thereon, rather than to any supposed dedicatee.

A map of the broader Greek world, highlighting Sicily and Aydin (ancient Tralleis).

The dedicator is then named as Seikilos, a person otherwise unknown. By the period in which the stone was carved, the pronunciation of Greek had long changed to be more similar to that of Greek today than to that of Classical times; so the way the name was pronounced would have been Sí-ki-los, with syllables of equal length (rather than “Say-ki-los” with a long first syllable). This makes it evident that it was a Greek form of the Latin Siculus or Sicelus, “Sicilian”; it seems likely that the author was a Sicilian in origin – perhaps one who had travelled to Asia Minor to pursue a musical career (a 3rd-century inscription records “a convention of musicians from Ionia and the Hellespont”).

The statement “Seikilos sets me up here” continues what is implied by EIMI, “I am”: that the stone/inscription is to be imagined as speaking in person, in the present tense, to the reader. The object of “sets up” (ΤΙΘΗΣΙ) can equally be the stone or the text of the song inscribed on it. The stated purpose of Seikilos in setting up the stone/song is given by the second verse: it is to be a “long-lasting marker”. The Greek word for “long-lasting”, poluchronion, incorporates the word chronos (“time”), a technical term for the duration of a musical note; and sēma (“marker”) can connote an alphabetic letter as well as a grave marker.

These double meanings suggest that Seikilos not only wants to claim that the inscribed song will last “through time”, but also that the text of his song puts on view “the marks of many time-elements” (chronoi) or musical notes. For immediately following this couplet, which the stone presents in a manner akin to epitaphs that “speak” to the passer-by (e.g. Simonides’ epitaph at Thermopylae: “Go tell the Spartans, stranger, that here / I lie obedient to their commands”) the stone stops speaking – and starts singing.

A modern reproduction (1955) of Simonides’ epigram on Kolonos Hill near Thermopylae, Greece.

Metre and Rhyme

One further aspect of poetic interest is Seikilos’ use of metre and rhyme, and his adaptation of a standard Classical metrical form (the iambic dimeter) to a more modern style of composition. The words of the song indicate a scansion as follows:

At first sight this looks metrically erratic, but it is in fact very regular. Each verse essentially consists of the rhythm, with standard variations, di dum di dum | di dum di dum (the rhythm of Hilaire Belloc’s “The nicest child I ever knew | was Charles Augustus Fortescue”). But it is complicated by the way the iambic metron (⋃  − ⋃  − )  is subject to resolution (two short beats for one long, e.g. ⋃ ⋃⋃ ⋃   −), syncopation (syncope means a missing or prolonged beat, standardly indicated by as in ⋃  −  −), and anaclasis (swapping round the first two elements of ⋃ − ⋃ −   to get  − ⋃ ⋃ −  as in verse 2). A more accurate represent­ation, incorporating the missing beats, shows how all the verses have similar durations:

While this pattern is read as indicating the rhythm of “The nicest child I ever knew”, once the additional melodic duration signs on the stone are taken into account a different 6/8 rhythm emerges, as in the transcription. Moreover, pronounced correctly for the 2nd-century context, the words ZĒN (ζῆν, phonetically [zi:n]) and APAITEI (ἀπαιτεῖ, [ape:ti:]) at the end of the penultimate and final verses should be recognised as creating an assonance (rhyming vowel or diphthong), just like PHAINOU and LUPOU in the opening verses. The verse endings thus rhyme AABB, a rhyme scheme of the kind that is never found in Classical lyric. These features remind us that the song should not simply be assessed by Classical harmonic theory or metrical rules.

Rather than subject the song to formal metrical analysis, then, we should trust the aural impression it produces, which is one of phrasal balance reinforced by assonance. The similarity to familiar English nursery rhymes is marked, and the melody can in fact be sung to the words of ‘Humpty Dumpty’:

It is a pure coincidence (but a noteworthy one nevertheless) that the word ‘fall’ coincides with a falling cadence, that the song trots forward at the arrival of “all the King’s horses”, and that the sad end of Humpty Dumpty matches the dejected cadence of the melody. Such a comparison, conscious or otherwise, has in the past led to the song’s being considered no more than a ‘ditty’ or musical trifle. But this is an unfair assessment: the subtly imitative qualities of the melody are masterly, and never intrude on the integrity of the song as a highly effective piece of music.

Remember therefore from whence thou art fallen, and repent: illustration by W.W. Denslow, 1903.


The sentiment expressed by the Seikilos Song, that life is short and one should enjoy it while one can, is a standard viewpoint derived from Epicurean philosophy, of the kind expressed by the Roman poet Horace with carpe diem, “harvest the day” (Odes 1.11.8). It seems hardly appropriate for such a sentiment to appear on the grave marker of a deceased family member, and there are no parallels for such a dedication.

One may therefore reconsider the fragmentary signature at the end in the light of the above interpretation and analysis. The fact that EUTER- is clearly the beginning of a name led two great scholars of ancient music, Egert Pöhlmann and Martin West, to speculate that this was a patronymic (EUTERPOU, “of Euterpes”): “If his father’s name was really Euterpes,” they write, “it implies a family line of professional musicians” (they compare the patronymic of the bard Phemius “Terpiades” in Homer’s Odyssey, which would make the bard the son of “one who delights in music”). But there is a simpler explanation that does not require recourse to the family name of a mythical bard. Euterpe was, of course, one of the nine Muses, goddesses of music, song, and dance: she was in fact none other than the Muse of Lyric Poetry.

Statue of Euterpe, Friedrich Ochs, 1857 (Lustgarten, Sanssouci, Potsdam, Germany).

If Seikilos wanted to lay claim to lyric excellence, then, he could do no better than to speak of himself as the son or perhaps devotee (if there is text missing that could be so supplemented) of the Muse herself. We should thus read the signature perhaps as “SEIKILOS, child of EUTERPE” (ΣΕΙΚΙΛΟΣ ΕΥΤΕΡΠΗΣ) – a light-hearted use of the matronymic – or “SEIKILOS [dedicates this] to EUTERPE (ΣΕΙΚΙΛΟΣ ΕΥΤΕΡΠΗI).

“Shine while you live” (ZĒIS), urges Seikilos at the beginning of his song: an encouragement to unabashed self-display by a poet who claims in his sign-off that he indeed “is alive” (ZĒI). This song, then, with its poetic introductory couplet, is a self-conscious display of proud musico-poetic creativity, of the kind for which poets had long sought to be remembered by future generations. The first-century BC Roman poet Catullus asks that his poetry “might last through time for more than one generation” (1.10,  plus uno maneat perenne saeclo); and his younger contemporary Horace claims “I have raised a monument more lasting than bronze” (Odes 3.30.1, exegi monumentum aere perennius). Seikilos similarly has made a bid for his “many-noted” music to be honoured by time. It is pleasing to think that this is what his song and its inscription on stone have achieved.

Armand D’Angour is Professor of Classics at the University of Oxford. He has written on the music of Sophocles’ Ode to Man here.


1 I have speculated that the spur to the creation of a musical notation was the regular violation of natural pitch-accents by the New Musicians of the late fifth century: A.J. D’Angour, “The New Music: so what’s new,” in S. Goldhill and R. Osborne (eds.), Rethinking Revolutions (Cambridge UP, 2006) 264–83.
2 See E. Pöhlmann and M. West, Documents of Ancient Greek Music (Oxford UP, 2000) 90.