Introduction to Greek metre

Greek metre measures syllables. It counts them, and to a large extent it also weighs them. Each syllable is given a relative duration, and this duration is called its quantity.^[1] Each verse has a fixed count of syllables or positions. Add or remove one, and the verse no longer scans.^[2] The pattern of long and short syllables that fills those positions is the rhythm.

Quantity, not stress. English verse is built from stressed and unstressed syllables. English speakers tend to space their stresses evenly, hurrying over runs of unstressed syllables.^[3] Greek verse does not work like this. The classical Greek word accent was not a stress. It was a rise and fall in the pitch of the voice.^[4] It played no part in the construction of verse.^[2]

What the verse regulates is length. A long syllable was heard as taking twice as long as a short one, in the relation of a crotchet to a quaver.^[5] An English line may look like a Greek colon when it is written in metrical signs. But English readers mark its rhythm by stress, over syllables of roughly equal length. Greek syllables had to keep their durations.^[6]

The material. A syllable is long in two cases. Its vowel may be long or a diphthong. Or two or more consonants may stand between its vowel and the next.^[7] Inside a verse the words run on as one stream of sound. They are cut into syllables across word-ends.^[8] At the end of each verse that continuity breaks, whether or not the sense pauses there.^[9] Here is the first line of the Iliad as a first taste. Each syllable stands over its sign, — long or ⏑ short, with its transliteration below:

The -εω of Πηληϊάδεω is one syllable. The two vowels are run together, as often happens within a word when the first vowel is ε.^[10] This running together is often called synizesis.^[11]

Why it is like this: the Indo-European inheritance. Greek is an Indo-European language. Its metres can be compared with those of other Indo-European peoples, above all with the hymns of the Rigveda. West dates the Rigveda roughly 1200–1000 BC, and Battezzato no later than the thirteenth century BCE.^[12] The comparison shows fairly clearly that Greek metre has Indo-European origins.^[13] The features common to Greek and Vedic verse, and presumably inherited, are these:^[2]

Three more shared features can be added to the list: a free beginning, called the “basis”; an indifferent final syllable; and a caesura near mid-verse.^[14]

Greek regulated rhythm more thoroughly than the parent system did, and it used more patterns. Most Greek verse also admits resolution and contraction. The Aeolic metres of Sappho and Alcaeus admit neither. In the light of the Indian evidence they look relatively primitive, as Antoine Meillet pointed out in 1923.^[15] They are probably the most conservative group of Greek metres, a judgement that also goes back to Meillet.^[16]

By the eighth and seventh centuries there were at least three regional traditions, each with its own forms. They were the Ionian (epic, elegy, iambus), the Aeolian (the two Lesbian poets) and the Dorian (the Peloponnese and the western colonies).^[17]

Kiparsky adds one more feature, which he regards as unique to Indo-European verse. This is the licensed equivalence of ⏑ — and — ⏑, which is called syncopation or anaclasis. In his view it weighs more than Meillet’s fixed syllable count, catalexis and fixed cadence, because those occur in many of the world’s metres.^[18]

How the hexameter arose is disputed. It has been explained as a borrowing (Meister, Meillet; rejected by West 1997). It has also been explained as a fusion of lyric cola (West in 1973, Nagy in 1974, Berg in 1978, modified by Tichy in 1981), or as a pair of syncopated iambic dimeters (Kiparsky).^[19] The Modern theory tab gives the details.

Metre and music. Most Greek lyric was composed to be sung, to the lyre, barbitos, kithara or aulos.^[20] In song a long normally took the time of two shorts.^[21] In some metres it took the time of three. In recitation the ratio was perhaps about 5:3.^[22]

Aristotle distinguished metron from rhythmos. All verse has metron. Rhythmos belongs only to verse delivered to a fixed beat: lyric and drama, but not, in his day, epic.^[23] The tab Rhythm and music says more.

Notes

1. ^ West 1987, p. 10.
2. ^ ^{a b c} West 1987, p. 2.
3. ^ West 1987, p. 7; D’Angour 2022, p. 133.
4. ^ West 1987, p. 2; Battezzato 2009, p. 143; D’Angour 2022, p. 132.
5. ^ D’Angour 2022, p. 132; Natoli et al. 2022, p. 334.
6. ^ D’Angour 2022, p. 133; see the Rhythm step of First steps, which sets the opening of Blake’s “The Tyger”, — ⏑ — ⏑ — ⏑ —, beside the first words of the Iliad.
7. ^ West 1987, p. 2; Battezzato 2009, pp. 130–131.
8. ^ West 1987, p. 2; Battezzato 2009, p. 130.
9. ^ West 1987, p. 2; Battezzato 2009, p. 131.
10. ^ West 1987, p. 15.
11. ^ Battezzato 2009, p. 139 n. 14.
12. ^ West 1987, p. 2; Battezzato 2009, p. 133.
13. ^ West 1987, p. 1.
14. ^ Battezzato 2009, p. 133.
15. ^ West 1987, pp. 2–3.
16. ^ Battezzato 2009, pp. 133–134.
17. ^ West 1987, p. 3.
18. ^ Kiparsky 2018, p. 102; Gunkel and Hackstein 2018, p. 1.
19. ^ Kiparsky 2018, pp. 104–106 with nn. 24–25; Battezzato 2009, p. 140.
20. ^ D’Angour 2022, p. 132.
21. ^ West 1987, pp. 6–7; Battezzato 2009, p. 143.
22. ^ West 1987, pp. 6–7.
23. ^ Kiparsky 2018, p. 79 n. 3.

Every syllable in a verse is long or short. Deciding which rests on one distinction that must be kept clear: vowel length is not syllable length.^[1]

Long and short vowels. ε and ο are short. η and ω are long, and so are all diphthongs. α, ι and υ vary from word to word, so their length has to be learned with each word.^[2] On this site a long one may be marked ᾱ and a short one ᾰ. A few rules of thumb help:^[3]

• A circumflexed vowel can only be long.
  Example: πᾶσι (pāsi), “to all” (Homer, Iliad 1.5): the circumflex shows that its α is long.
• An acute on the third syllable from the end (the antepenultimate), or a circumflex on the second from the end (the penultimate), means that the last vowel is short, unless it is αι or οι.
  Examples: ἄλγεα (algea), “woes”, has its acute on the third syllable from the end, so its final -ᾰ is short. μῆνιν (mēnin), “wrath”, has a circumflex on the second syllable from the end, so its -ῐ- is short.
• Vowels that arise from contraction are long.
  Example: ἐτίμᾱ (etimā), “he honoured”, has a long final α, because it comes from -αε (-ae).
• Certain endings are long. They include -ᾱς in masculine nominatives and in the genitive singular and accusative plural of the first declension, and -ᾱ in the dual.
  Examples: νεανίᾱς (neaniās), “young man”, a masculine nominative; χώρᾱς (chōrās), “of the land”, a genitive singular.
• Others are short. They include the -α of the neuter plural, final -ι and -υ in most words, and all prepositions, ἀνά, διά, σύν.
  Examples: δῶρᾰ (dōra), “gifts”, a neuter plural; ἔτῐ (eti), “still”; διᾰ́ (dia), “through”.

From vowels to syllables. A period is divided into syllables without regard to word-division. (A period is a stretch of verse within which the words run on without a metrical break, often a single line)^[4] The consonant immediately before a vowel begins its syllable. So ποδηγός (podēgos), “guide”, is cut πο-δη-γός, not by its etymology. A consonant after the vowel and not followed by another vowel stays with it and adds to its length. So two consonants together are normally divided between syllables.^[5]

A syllable that ends in its vowel is called open. A syllable that ends in a consonant is called closed. A closed syllable is long, and so is any syllable with a long vowel or a diphthong. Only an open syllable with a short vowel is short.^[6] The aspirate is not a consonant, and ζ, ξ, ψ each count as two.^[7]

So in κρύπτω (kryptō), “I hide”, the υ is short, but the closed syllable κρυπ (kryp) is long, and that is what counts. In ῥίπτω (rhīptō), “I throw”, the ι is long by nature as well. For the same reason it is better not to say that a vowel is “lengthened by position”.^[8]

In one line: a syllable is short only if it is open and its vowel is short; every other syllable is long.

Hephaestion’s definitions. The ancient handbook puts the same rules in its own terms. A short syllable has a short vowel “not at the end of a word”, with at most one single consonant before the next vowel.^[9] A long one has a long vowel or diphthong with a consonant after it, closing its own syllable or beginning the next. Without such a consonant the syllable is not outright long but “common” (1.2). Position he defines thus (1.3):

Θέσει μακραὶ γίνονται, ὅταν βραχέος ὄντος ἢ βραχυνομένου φωνήεντος σύμφωνα πίπτῃ μεταξὺ αὐτοῦ καὶ τοῦ τῆς ἑξῆς συλλαβῆς φωνήεντος πλείονα ἑνὸς ἁπλοῦ.
Syllables become long by position when the vowel is short, or treated as short, and more than one single consonant falls between it and the vowel of the next syllable.

He lists five ways it happens, from Τίρυνς and Ἕ-κτωρ to ἕξ and ἔξω (1.3). His “common” syllable (κοινή) arises in three ways. The first is a long vowel or diphthong before a vowel (ἄνδρα μοι ἔννεπε μοῦσα). The second is a short vowel before stop + liquid (ὅ-πλον, ἄ-κρον). The third is a word-final short syllable followed by at most one consonant (1.4, 1.7, 1.10). In modern terms these are correption, “mute and liquid” and lengthening.^[10] Such syllables are called anceps syllables. They have nothing to do with anceps positions in a scheme.^[11]

Muta cum liquida. A stop followed by a liquid or nasal may begin the next syllable. So τέκνα in tragedy is tek-na (— ⏑) or te-kna (⏑ ⏑).^[12] Where the stop closes one syllable and the liquid opens the next, the syllable is outright long. Hephaestion cites Alcaeus’ ἔκ μ’ ἐλάσας.^[13] The κ of ἐκ is never carried over. So ἐκ-λῦε “release from” must begin long, while ἔκλυε “he heard” may be — ⏑ ⏑ or ⏑ ⏑ ⏑.^[14]

Final syllables. The last position of a period may always be long or short. A short syllable where the pattern wants a long one is called a brevis in longo. The missing time is made up by the pause.^[15] Hephaestion says the same: Παντὸς μέτρου ἀδιάφορός ἐστιν ἡ τελευταία συλλαβή, “in every metre the final syllable is indifferent”. He cites Iliad 2.1–2, which end in long -ταί and short -νος.^[16]

Worked examples. Syllables are cut as the metre cuts them, across word-ends.

Translations: “the wrath … goddess”; “laid woes”; “darkness, my light”; “Erebus, brightest for me”; “I hide”; “the darkness”.

The two Ajax lines are separate verses, so the ς of φάος closes its syllable at the pause. If the two lines ran on without a break (in synapheia), it would carry over: pha-o-se-re-bo- ⏑ ⏑ ⏑ ⏑ ⏑.^[18]

Notes

1. ^ West 1987, pp. 10, 13.
2. ^ West 1987, p. 10; see Phonology › First steps, where each vowel can be heard long and short, and the diphthong and the δίχρονα, the vowels “of two lengths”, are explained.
3. ^ West 1987, pp. 10–11.
4. ^ West 1987, p. 3.
5. ^ West 1987, p. 12; Battezzato 2009, p. 130.
6. ^ West 1987, p. 12; Battezzato 2009, pp. 130–131.
7. ^ West 1987, p. 12 n. 1.
8. ^ ^{a b c} West 1987, p. 13.
9. ^ Hephaestion, Ench. 1.1.
10. ^ See Prosody.
11. ^ West 1987, pp. 12–13.
12. ^ West 1987, p. 16.
13. ^ Hephaestion, Ench. 1.8.
14. ^ West 1987, p. 17.
15. ^ West 1987, p. 4; Battezzato 2009, p. 131.
16. ^ Hephaestion, Ench. 4.5.
17. ^ West 1987, p. 14.
18. ^ ^{a b} West 1987, p. 12.
19. ^ West 1987, p. 12 with n. 1.

A new name may describe a familiar part of a verse. It may also belong to a different analysis. West’s position and Maas’s element are equivalent terms. Their princeps and longum are another equivalent pair.^[1] The divisions of a lyric line need more care: two analyses can agree on every quantity while grouping the syllables differently.

The table sets four approaches beside one another. The examples below explain each column.

The position terms are documented in West’s glossary. The numerical use of element is illustrated in Maas’s hexameter rules. The two lyric analyses come from the passages examined below.^[2]

Positions and syllables

A position belongs to the pattern. A syllable belongs to the spoken words that fill it.^[3] In a hexameter foot, the first part is long. The second part can hold two shorts or one long. West calls these parts princeps and biceps.^[4]

In the first foot below, μῆ (mē) fills the princeps. The text span νιν ἄ (nin a) fills the biceps with two shorts. In speech the n joins the following vowel: the grid displays νι (ni) and ν ἄ (na). The third foot has a contracted biceps: Πη (pē) is one long syllable. Contraction changes the syllables occupying that part of the pattern. It does not add or remove a foot.^[5]

Maas’s “eighth element” is therefore the biceps of the fourth foot, not the eighth spoken syllable. His “monosyllabic” element is one filled by a single long. That distinction matters when reading a rule about where a word may end.^[6]

Units in choral song

A letter can stand for a whole sequence of quantities. In West’s dactylo-epitrite notation, e means — ⏑ — and D means — ⏑ ⏑ — ⏑ ⏑ —.^[7] The following example contains both:

Take the second line, Medea 411. Its three groups are easy to locate:

For this line, e — D and s — dd describe the same quantities. The difference is the notation, not an alternative pronunciation of the Greek.^[8] This comparison applies to the complete groups shown here. It is not a rule for replacing every d or D in every book.

The Sapphic line

A difference in grouping is more substantial than a change of letter. Hephaestion divides the Sapphic eleven-syllable line into a trochaic pair, a choriamb and an iambic close. The example shows that division:^[9]

West writes the line as ∧ia hag. His first group has three positions, — ⏑ —. The remaining eight form the hagesichorean, × — ⏑ ⏑ — ⏑ — —.^[10] Both patterns below are shown with a long final position. A short syllable can fill that position at period-end, as -τα (ta) does here.^[11] The divisions can be aligned:

The fourth position belongs to Hephaestion’s first group but to West’s second. A shared scansion therefore does not imply a shared account of how the verse is built. The Published schemes guide shows how to keep quantities, groupings and line endings distinct.

Notes

1. ^ West 1987, pp. 86, 88.
2. ^ West 1987, pp. 86, 88; Maas 1962, p. 62; Dale 1971, p. 50; Hephaestion, Ench. 14.1.
3. ^ West 1987, p. 5.
4. ^ West 1987, p. 6.
5. ^ West 1987, pp. 5–6.
6. ^ Maas 1962, p. 62.
7. ^ West 1987, pp. ix–x.
8. ^ West 1987, p. 65; Dale 1971, p. 50.
9. ^ Hephaestion, Ench. 14.1.
10. ^ West 1987, pp. 34–35.
11. ^ West 1987, p. 4.

A published scheme is a compact account of a poem’s pattern. The symbols describe quantities and boundaries. Letters such as D stand for longer groups. The words printed beneath the scheme show how a particular passage fills that pattern.^[1]

A responding pair

The sign ~ means “in responsion with”. Thus Antigone 781 ~ 791 tells us to compare the first line of the strophe with the first line of the antistrophe. A strophe is a stanza whose metrical form recurs. Its answering stanza is the antistrophe.^[2]

Here both lines contain eight syllables. Read down each column: the two syllables at that position have the same quantity. Then read across to hear the complete lines:

The scheme is shared; the words are not. The short Ἔ- (e-) of 781 answers short σύ (sy) in 791. Long -ρως (-rōs) answers long καί (kai). Responsion compares corresponding positions, not corresponding words.^[3]

Cola and periods

A colon is a metrical phrase, usually no more than about twelve syllables. A period is a complete span of rhythm, and it may contain more than one colon.^[4] The layout on the printed page need not give every colon a separate line. Conversely, a period may occupy more than one printed line.

In West’s notation, a single bar | marks word-end. A double bar ‖ marks period-end. Three bars ||| mark strophe-end.^[5] These boundaries answer different questions: where a word stops, where the rhythmic span closes, and where the stanza ends.

The opening of the first stasimon of Medea supplies a worked example. D expands to seven quantity signs, — ⏑ ⏑ — ⏑ ⏑ —. The sign e expands to three, — ⏑ —. Single syllables link these groups:^[6]

In line 411, καὶ δίκα (kai dikā) fills e. The next καί (kai) fills the linking long. The rest, πάντα πάλιν στρέφεται (panta palin strephetai), fills D. Expanding e — D therefore gives the eleven syllables shown in that row. The letters name groups, not individual syllables.

Lines 412–414 show why line division must be read separately from period division. Lourenço gives two successive groups, D — e and D —. Dale presents the whole passage from ἀνδράσι (andrasi) to ἄραρε (arare) as one period. The quantities agree; the way the analysis sets out the span differs.^[7]

The two rows together are one continuous period in Dale’s analysis. A break on the printed page does not by itself establish period-end.

Editorial keys

An editor’s key establishes the local meaning of a sign. West’s ⏓ means “usually long”, and ⏒ means “usually short”. Voigt uses those signs with a stated threshold: at least eight of twelve occurrences have the indicated usual quantity.^[8] These are summaries of a pattern across examples. They are not instructions to lengthen or shorten every syllable beneath the sign.

The first syllable of Medea 410 above is short ἄ- (a-). Dale’s responding scheme allows that opening even though its sign is ⏓, usually long. The short syllable remains short in this particular line.^[9] This is the distinction between a general scheme and a particular scansion.

A work may also distinguish an author’s analysis from an editor’s addition. Dale’s posthumous Metrical Analyses prints certain descriptions in parentheses because Webster supplied them where her intended analysis was uncertain. Those parentheses are part of that book’s editorial practice.^[10] They should not be assigned a metrical meaning borrowed from another book.

The Notation tables give the signs used here. The Terms and analyses comparison explains changes of terminology. For a variant Greek reading beneath the text, rather than a metrical scheme.^[11]

Notes

1. ^ West 1987, pp. 4–6, ix–x.
2. ^ West 1987, pp. ix, 4.
3. ^ West 1987, p. 4.
4. ^ West 1987, pp. 3–4.
5. ^ West 1987, p. ix.
6. ^ West 1987, pp. ix–x, 65.
7. ^ Lourenço 2011, p. 149; Dale 1971, pp. 50–51.
8. ^ West 1987, p. ix; Voigt 1971, p. 15.
9. ^ Dale 1971, p. 50.
10. ^ Dale 1971, p. vii.
11. ^ See the Apparatus example.

The basic rule is that a syllable is long if it has a long vowel or if it is closed. This rule has two sets of exceptions. The first concerns what happens when vowels meet. The second concerns how consonants are treated.^[1] The vowels come first. Each rule has one example.

Elision. A short final vowel is mostly dropped before a word beginning with a vowel, unless period-end intervenes. This dropping is called elision. Some final vowels cannot be elided. They are ἅ, τά; τί, τι; ὅτι (except perhaps in epic); ἄχρι, μέχρι; περί (except in Pindar); the -ι of the vocative and, outside epic, of the dative; ὅ, τό, πρό; -ᾱο and -οιο (except in Doric lyric); and -υ generally (but Aeolic ἀπύ). -αι and -οι are elided only within restricted limits. Of a word ending in two vowels, only the last is elided.^[2] The elided vowel is replaced in writing by an apostrophe. The second line of the Iliad has two elisions, μυρί’ (myri’) for μυρία (myria) and ἄλγε’ (alge’) for ἄλγεα (algea):

Elision is rarer in the stricter styles of verse. In the iambic trimeter about one word-end in nine is elided in Semonides, and about one in seven in Sophocles.^[3] In the surviving music an elided vowel is normally not written, and it gets no note of its own.^[4]

The accent of an elided syllable is not lost. It moves back to the syllable before: πολλὰ ἔπαθον, “I suffered much”, becomes πόλλ᾽ ἔπαθον. Prepositions and conjunctions are the exception. They keep the accent where it was, or have none: παρ᾽ ἐμοῦ, “from me”, ἀλλ᾽ ἐγώ, “but I”.^[5]

Epic correption. In epic, elegy and some lyric, a long vowel or diphthong at the end of a word normally counts as short when the next word starts with a vowel. This shortening is called correption. It happens almost only in double-short rhythm.^[6] Hephaestion calls such syllables “common” and cites this very line:^[7]

Correption also occurs inside a word. In Attic poets it is frequent with αι and οι (ποιεῖν ⏑ —). It is invariable with the vocalic endings of οὗτος, ἐκεῖνος and the like before the emphatic -ί (αὑτηΐ, τουτουΐ).^[6]

Synecphonesis. Adjacent vowels may be run together as one syllable. This is called synecphonesis. The names synizesis, synaloepha and crasis are used for the same thing, crasis especially when the merger is written.^[8] Teaching books often keep synizesis for a merger inside a word and synecphonesis for one across word-end (so the Cambridge lecture handout, a teaching source). West does not distinguish them.

Hephaestion defines it like this: συνεκφώνησίς ἐστιν, ὁπόταν δύο συλλαβαὶ σύμφωνον μὴ ἔχουσαι μεταξὺ ἀλλήλων ἀντὶ μιᾶς παραληφθῶσι, “when two syllables with no consonant between them are taken as one”.^[9] It is commonest after a monosyllable (especially the article or relative, καί, ἤ, δή, μή, ὦ) or after ἐγώ, ἐπεί. Between longer words it is tolerated mainly in Sappho, Hipponax and Old Comedy:^[10]

εἰλαπίνη ἠὲ γάμος (Od. 1.226) — ⏑ ⏑ — ⏑ ⏑ — a feast or a wedding? (-η ἠ- one syllable)

In Sappho’s hymn the two vowels of ὠράνω αἴθερος (ōranō aitheros), “from heaven through the air”, are run together across the word-end. The line runs on into the next, so the second line is shown too:^[10]

When the merger is written, it is called crasis. In the fifth line of the same hymn καὶ ἀτέρωτα (kai aterōta) is written κἀτέρωτα (katerōta):

Within a word it is frequent after ε (θεός as one syllable).^[10] Other examples are ἐρέω and the -εω of Πηληϊάδεω.^[11]

Crasis is limited to the most closely linked pairs of words. In Attic the article and the relative pronoun merge with a following vowel, but other small words elide instead. καί merges, as in κἀναθήματα, “and the offerings”, while ἀλλά elides, as in ἀλλ᾽ ἀνδρός, “but of a man”.^[12] Crasis between two full words is rare outside set phrases such as τύχἀγαθῇ, “with good luck”.^[13]

Prodelision (aphaeresis). Sometimes a long final vowel swallows a following short initial vowel. This is usually the ἐ- of an augment or of ἐπι-.^[10] The augment is the ἐ- that marks a past tense, as in ἐκαλεῖτο (ekaleito), “he was called”. In this line of Sophocles it disappears after the long -η of Ἰόλη (Iolē), and is replaced in writing by an apostrophe:

The opening ⏑ ⏑ — of this line is the first-foot anapaest that tragedy allows.^[14]

Prodelision grows commoner as the style grows freer. There are about 4 cases in every 1,000 lines of Aeschylus, 13 in Sophocles and 23 in Aristophanes.^[15]

Hiatus. Vowels that meet at word-juncture and keep their full value are said to be in hiatus. Epic and elegy allow hiatus, above all at certain points of the line, though many cases involve a lost initial ϝ. Doric lyric virtually restricts it to such cases. Ionic lyric and iambus restrict it to the pronoun ἕ, οἱ, ἑό and to the place after ἤ. Lesbian lyric avoids it altogether. Tragedy admits it only around exclamations, urgent imperatives and a few set phrases.^[16] In this line of Homer, μοι (moi) stands in hiatus before ἔριδος (eridos) and keeps its length in the princeps, while καί (kai) before ἀρωγῆς is correpted:^[17]

Tragedy also avoids hiatus from one line to the next when the sentence runs on. In Euripides’ Medea about 3 pairs of lines in 100 have hiatus across the line-end in the middle of a phrase, against 17 in 100 where a new phrase begins. Comedy is much freer.^[18]

Notes

1. ^ West 1987, p. 13.
2. ^ West 1987, pp. 13–14.
3. ^ Devine and Stephens 1994, p. 261.
4. ^ Devine and Stephens 1994, p. 260.
5. ^ Devine and Stephens 1994, pp. 265, 360.
6. ^ ^{a b} West 1987, p. 14.
7. ^ Hephaestion, Ench. 1.4.
8. ^ West 1987, pp. 14–15, 86.
9. ^ Hephaestion, Ench. 2.1.
10. ^ ^{a b c d} West 1987, p. 15.
11. ^ Battezzato 2009, p. 139 n. 14.
12. ^ Devine and Stephens 1994, p. 268.
13. ^ Devine and Stephens 1994, p. 269.
14. ^ West 1987, p. 26.
15. ^ Devine and Stephens 1994, p. 270.
16. ^ West 1987, pp. 15–16.
17. ^ West 1987, p. 22.
18. ^ Devine and Stephens 1994, p. 426.

How consonants decide the length of a syllable, beyond the basic rule that a closed syllable is long.^[1]

Two kinds of consonant. The rules below depend on two kinds of consonant. A stop (also called a mute or plosive) is made by closing the mouth for a moment, so that the breath is checked: κ γ χ, τ δ θ, π β φ. A continuant is made with the mouth only partly closed, or through the nose, so that the breath keeps flowing. The continuants are the liquids λ ρ, the nasals μ ν, the fricative σ, and the semivowel ϝ. Because the breath keeps flowing, a continuant can be prolonged more readily than a stop.^[2]

Position across word-end. A word-final consonant before a word-initial consonant closes its syllable and makes it long, as ὃς before μάλα in Odyssey 1.1. Before a vowel it joins the next syllable.^[3]

Mute and liquid (muta cum liquida). A stop followed by a liquid or nasal is often taken as beginning the next syllable. A preceding short vowel is then left in an open, short syllable. So τέκνα (tekna), “children”, may be tek-na (— ⏑) or te-kna (⏑ ⏑).^[4] Voiced stops and nasals resist this more.

Early poetry admits the short scansion only at the beginning of a word (δειλοῖσι βροτοῖσι — — ⏑ ⏑ — ⏑) or in otherwise intractable words (Ἀφροδίτη ⏑ ⏑ — —). The iambi of Archilochus and Semonides and normal Lesbian lyric avoid even that. In Stesichorus it is the rule. Theognis, Solon and Pindar are freer. In Attic speech short syllables stayed short before these groups, and this is the normal scansion in comic dialogue. It is called Attic correption. Tragedy allows the long scansion within the word.^[5] Hephaestion adds that a semivowel before the liquid (ἀμνός, ἑσμός) makes the syllable fully long.^[6]

Lengthening by an initial continuant. In epic an initial λ, μ, ν, ρ, σ (and originally ϝ) could be held. It then closed the preceding syllable. This happens nearly always in the princeps. With ρ the effect stays common in iambus and lyric, and it is almost invariable in Attic dialogue. With the other consonants only isolated examples occur, in elegy and lyric:^[7]

Digamma. Many Homeric words once began with ϝ (w) and, especially in formulae, are scanned as if it were still there. Hiatus before them is not hiatus, and a preceding syllable may be lengthened. Among them are ἄναξ, ἄστυ, ἔπος, ἔργον, ἰδεῖν and οἶδα, ἴσος, and the pronoun ἕ, οἱ. In the pronoun the ϝ is almost always effective.^[8]

Lengthening by a final continuant. In epic, elegy, Pindar and Bacchylides a final ν, ρ or ς is occasionally treated as closing its syllable although a vowel follows. In Attic drama this happens only with the abrupt interjection εἶεν, which in isolation is two long syllables.^[9]

Three regional practices. The rules of this tab and the last, side by side:

Maas states the rule more generally. For him it is a ban on word-end after a long anceps. Tragedy and the early iambographers keep this ban, and comic dialogue breaks it on average about once in five lines.^[10]

Notes

1. ^ West 1987, pp. 12, 16–18.
2. ^ West 1987, p. 11.
3. ^ West 1987, p. 12.
4. ^ West 1987, p. 16; Natoli et al. 2022, pp. 335–336.
5. ^ West 1987, pp. 16–17.
6. ^ Hephaestion, Ench. 1.7.
7. ^ West 1987, pp. 17, 22.
8. ^ West 1987, pp. 22–23.
9. ^ West 1987, p. 18.
10. ^ Maas 1962, pp. 34–35.

Greek verse is built in layers. From the smallest unit up:

One line, layer by layer. The first line of Sophocles’ Oedipus Tyrannus is an iambic trimeter. It has twelve positions. They make six feet of two positions each, or three metra of four, × — ⏑ —.^[1] A regular word-end after the fifth position, the caesura, divides it into two cola.^[2] The whole line is one period, which ends at ‖:

Why “period” rather than “line”. A verse is defined by synapheia, not by the page. Sometimes the continuity runs over several written lines, as in the last two lines of a Sapphic stanza. There the neutral term period is preferred.^[3] In the same way, a verse can be defined as a sequence between two metrical pauses. It is then different from the line written on the page. The division of a poem into lines is called colometry.^[4]

Within a period the scansion of a word may be changed by the word after it. At period-end it cannot, and the last position is free.^[5] Hephaestion says that every metre ends in a complete word, and that its last syllable is indifferent.^[6] Some lines have halves that behave like separate periods. The ancients called such lines ἀσυνάρτητα, “disconnected”. They did not see that this is ordinary period-end.^[7]

The first stanza of Sappho’s hymn to Aphrodite shows the difference between a line and a period. It is written in four lines, but it has only three periods:

Lines 1 and 2 are periods. Line 3 runs into line 4, so the stanza is three periods, not four.^[8]

Caesura and diaeresis. A caesura is a place where word-end regularly falls inside a period. In the hexameter it normally falls at one of two places in the third foot. One is after the third princeps, — | ⏑ ⏑, and is called “masculine”. The other is after the following short, — ⏑ | ⏑, and is called “feminine”. The feminine caesura is commoner, by about 4:3. Occasionally a long word runs through the third foot to a caesura after the fourth princeps. The caesurae after the third and fourth principes are called penthemimeral and hephthemimeral.^[9] The iambic trimeter has the same two options. Its caesura comes after the first position of the third foot or, failing that, of the fourth:^[2]

Line 1 has the penthemimeral caesura, and line 3 the hephthemimeral. Line 2 ends in a brevis in longo.^[2]

Diaeresis is a common name for word-end that coincides with the end of a metron or colon.^[10] Usage differs. West calls the fixed break between the pentameter’s two hemiepes a caesura.^[11] Battezzato keeps “caesura” for word-end inside a metron and “diaeresis” for word-end between metra.^[12]

The regular caesura shows that the hexameter is made of two cola. These cola, more than any six feet, are what it is built from, and many epic formulae fill one or the other.^[9] The two commonest types of hexameter can likewise be built from two cola joined at the caesura: a hemiepes with a paroemiac, or a feminine hemiepes with an enoplion.^[13] Cola may be set off by caesura and syntax. Or they may run together seamlessly, which is called synartesis.^[14]

Bridges. A bridge (zeugma) is the opposite of a caesura. It is a point where poets avoid ending a word.^[15]

Hermann’s Bridge is a bridge in the hexameter. Word-end between the two shorts of the fourth biceps is avoided. It is breached about once in 550 lines of Homer.^[16] Maas places the same bridge after the “fourth trochee” and gives a different rate: a word ends there only about once in 390 lines.^[17] Each figure is given here as its author states it.

Porson’s Law applies in early iambus and tragedy. The rhythm — — | — ⏑ — ‖ is avoided at line-end. That is, a long syllable in the third anceps of the trimeter must share a word with the syllable after it, unless either of the two is a monosyllable. Comedy ignores the law.^[18] Havet’s Law is the corresponding rule for the opening of the trochaic tetrameter.^[19]

Maas treats both as one general rule for verse containing the sequence × — ⏑ — ×. In such verse, apart from the central caesura, a word may not end after an anceps that is filled by a long syllable. In the trochaic tetrameter of the early iambographers and of tragedy he finds no exception. In their trimeter he finds only one, a Persian proper name at Aeschylus, Persians 321. The dialogue of comedy, by contrast, breaks the rule on average about once in five lines. This is one of the chief marks that set comic dialogue apart from tragic.^[20] Prepositives count with the following word, and postpositives with the preceding one.^[21]

Phrasing. Syntax need not match metre, but poets avoid serious clashes. As they approach a caesura or period-end, they let the sentence run to it or a little past it, rather than start a new one just before it. Prepositives are not placed immediately before the boundary, and postpositives are not placed immediately after it.^[22] Tragedians occasionally break this at line-end. Sophocles and the comedians even elide across the line-break.^[23]

The stricter the style, the more often sentence-end and line-end coincide. In Aeschylus and Sophocles about 34 full stops in 100 fall inside a line rather than at its end. In Euripides’ later plays the figure rises to about 40, and in his satyr play Cyclops it is 39.^[24] The critic Dionysius of Halicarnassus already observed that phrasing which cuts across the verse makes the metre harder to hear.^[25]

A postpositive is a small word that cannot begin a sentence. It leans on what comes before it. The particles ἄρα, αὖ, γάρ, μέν, δέ, δή, νυν, οὖν, τε, τοι and τοίνυν are postpositives. They usually stand after the first or second word of their sentence.^[26] Greek can also split a phrase that belongs together and put another word, usually the verb, inside it, as in καινὸν ἀγγέλλεις ἔπος, “new is the word you bring” (Troades 55). This is called hyperbaton. It is common in verse but not confined to it, and the word put first is usually the one with the emphasis.^[27]

How period-end is detected. In choral lyric, where the text is not laid out in obvious lines, period-end must be inferred from these signs:^[28]

Period-end established at a point in one strophe holds at the corresponding place in all the others. In the same way, evidence from the others holds for it. Metre can also exclude period-end. An acatalectic dactylic tetrameter cannot be followed by period-end, since its last foot would then be — ⏑ × and not — ⏑ ⏑.^[29] In tragedy no strophe occurs more than twice. There these signs show period-end only here and there, and corruption may leave the scheme uncertain.^[30]

Brevis in longo suggests that a short syllable before a pause was drawn out in delivery. Speakers of many languages lengthen the last syllable before a pause. Some of the musical papyri mark the last syllable of a line with a sign for extra length.^[31]

Notes

1. ^ West 1987, p. 5.
2. ^ ^{a b c} West 1987, p. 25.
3. ^ West 1987, p. 3; D’Angour 2022, p. 136.
4. ^ Battezzato 2009, pp. 131–132.
5. ^ West 1987, pp. 3–4.
6. ^ Hephaestion, Ench. 4.5–6.
7. ^ West 1987, p. 32.
8. ^ West 1987, p. 34.
9. ^ ^{a b} West 1987, p. 19.
10. ^ West 1987, p. 86.
11. ^ West 1987, p. 24.
12. ^ Battezzato 2009, p. 132.
13. ^ Battezzato 2009, p. 138.
14. ^ West 1987, pp. 4, 90.
15. ^ West 1987, pp. 21, 90; Battezzato 2009, p. 132.
16. ^ West 1987, p. 21.
17. ^ Maas 1962, p. 60.
18. ^ West 1987, p. 25; Battezzato 2009, p. 137.
19. ^ West 1987, p. 28.
20. ^ Maas 1962, pp. 34–35.
21. ^ West 1987, pp. 9, 20–21.
22. ^ West 1987, pp. 8–9.
23. ^ West 1987, p. 27.
24. ^ Devine and Stephens 1994, pp. 424–425.
25. ^ Devine and Stephens 1994, p. 425.
26. ^ Clyde 1857, p. 171.
27. ^ Clyde 1857, p. 171; Devine and Stephens 1994, pp. 483–484.
28. ^ West 1987, pp. 4–5, 36–37.
29. ^ West 1987, pp. 36–37.
30. ^ West 1987, pp. 51–52.
31. ^ Devine and Stephens 1994, pp. 426–427.

Stichic. A stichic metre uses one verse-pattern over and over, line by line, without grouping lines into stanzas. Most Greek poetry is of this kind, above all the hexameter and the iambic trimeter.^[1] The Iliad, for example, is hexameter after hexameter from beginning to end; the spoken scenes of a tragedy are trimeter after trimeter. Such use is also described as κατὰ στίχον (kata stichon), “line by line”.^[2] The elegiac couplet is strictly strophic, but it behaves like a stichic metre.^[3] The recited parts of drama are stichic, and the sung parts are normally strophic.^[4]

Strophic. A strophic poem is built from stanzas. A strophe is a stanza of one or more periods which recurs in the same form. Its single recurrence is called the antistrophe. The metrical agreement between the two is called responsion.^[5]

Responsion means that each position of the strophe is matched by a syllable of the same kind at the same place in the antistrophe. Here are the first lines of the strophe and antistrophe of the chorus’s song to Eros in Sophocles’ Antigone. The words are quite different, but the pattern of longs and shorts is the same, position by position:

The simplest strophes are the Ionian “epodes” of Archilochus and Hipponax. They have two or three periods. The first is usually a hexameter or trimeter, and the rest are shorter cola. In antiquity the shorter line was called the ἐπῳδός (epōidos).^[6] The Lesbians composed nearly everything in strophes of two, three or four verses.^[7] In the longer strophes one verse differs. This gives an aaA shape, in which the last verse amplifies the others, as in Sappho fr. 94.6–8:^[8]

τὰν δ’ ἔγω τάδ’ ἀμειβόμαν· ‖χαίροισ’ ἔρχεο κἄμεθεν ‖μέμναισ’· οἶσθα γὰρ ὤς σε πεδήπομεν. |||— ⏑ — ⏑ ⏑ — ⏑ — ‖ · — — — ⏑ ⏑ — ⏑ — ‖ · — — — ⏑ ⏑ — ⏑ ⏑ — ⏑ — ||| (gl ‖ gl ‖ glᵈ |||)And I answered her thus: go, and fare well, and remember me; for you know how we cherished you.

synthetic ∿

The Sapphic and Alcaic stanzas have the same shape (D’Angour’s AAAB and AABC). Each served many poems and presumably went with a particular tune. Antiquity wrote their third period over two lines, which is why our editions print four.^[9] Occasionally a whole strophe is one compound period. An example is Sappho fr. 96.12–14, ∧ia gl / gl / gl ia∧ |||, whose words run over the line-ends (scanned in Metres › Lyric stanzas and epodes, “Lesbian stanzas”).^[10]

Dorian strophes and the triad. Each Dorian strophe, so far as we know, was a new creation for its song. It was built from familiar cola but was open to innovation.^[11] The same holds for the strophes of the choral poets and of Pindar’s odes.^[12]

Stesichorus and Ibycus used the triad. A triad is a strophe and antistrophe followed by a third stanza of different form. This third stanza is called the epode (ἡ ἐπῳδός, hē epōidos, sc. στροφή, strophē). The sequence was repeated as often as the poet wished.^[11] Simonides, Bacchylides and Pindar mostly used it.^[13] A poem may be one triad. Pindar goes up to five, except for the thirteen of Pythian 4, and Bacchylides to at least eight.^[14]

Responsion as evidence. What is established at one point in a strophe holds at the corresponding point in all. West analyses the opening of the monostrophic Pythian 12 as — D — D ‖ D ⁝ — D |.^[15] Strophe 2 answers strophe 1 exactly:

αἰτέω σε, φιλάγλαε, καλλίστα βροτεᾶν πολίων, ‖ (1)Φερσεφόνας ἕδος, ἅ τ’ ὄχθαις ἔπι μηλοβότου (2)τὸν παρθενίοις ὑπό τ’ ἀπλάτοις ὀφίων κεφαλαῖς ‖ (9)ἄϊε λειβόμενον δυσπενθέϊ σὺν καμάτῳ, (10)— — ⏑ ⏑ — ⏑ ⏑ — — — ⏑ ⏑ — ⏑ ⏑ — ‖ · — ⏑ ⏑ — ⏑ ⏑ — ⁝ — — ⏑ ⏑ — ⏑ ⏑ — |I beg you, lover of splendour, fairest of mortal cities, seat of Persephone, who on the banks of sheep-grazing [Acragas] ... (the dirge) which she heard pouring, with grievous toil, from the maidens’ dread snake-heads.

synthetic ∿

(αἰτέω is two syllables by synecphonesis, as West marks it.) There may be period-end after μηλοβότου, but nothing in any of the four strophes proves it.^[15] Conversely, a single hiatus or brevis in longo settles the matter for every strophe. In Alcman’s Partheneion a brevis in longo in line 36 and hiatus in lines 38, 39 and 45 establish period-end at those places in all the strophes.^[16]

Responsion allows for anceps and final positions, and for resolution and contraction. Some sequences respond to a cognate rather than an identical form.^[2] Beyond that, responsion is usually exact, but not always. Alcman’s Partheneion ends some strophes ⏑ ⏑ — where the others end ⏑ — —, which is “without parallel”. Bacchylides 5 three times lets × — ⏑ — answer — ⏑ —. In Pindar’s dactylo-epitrite a resolution is more often a licence taken in one strophe than a feature recurring in all. In Bacchylides 17 there are eight places where syncopated iambics answer unsyncopated ones, and they prove the two equal in time.^[17] Anaclastic aeolic cola may respond with the standard form.^[18]

Drama. Sung parts run AA BB CC... . Sometimes there is a single closing epode. Occasionally a non-responding strophe stands before the pairs (proode) or among them (mesode).^[19] A strophe divided between singers is almost always divided at the same points in the antistrophe. In the comic parabasis, strophe and antistrophe are each followed by sixteen or twenty trochaic tetrameters. This arrangement is called the epirrhematic syzygy.^[20]

No tragic strophe recurs more than twice. So responsion helps the editor less, and corruption may leave no way of telling which of the pair keeps the true scheme.^[21] Late Euripides and Timotheus gave up responsion for metrically free, monostrophic song.^[22] Such poems are also called astrophic. An example is Timotheus’ Persians.^[23]

Two ancient terms. Hephaestion’s ἀσυνάρτητα (asynartēta) are “two cola which cannot be fitted together nor form a unity, taken in place of a single line”.^[24] For West the disconnection is simply period-end. On his view these are two short verses, written on one line by ancient convention.^[25] Polyschematist (πολυσχημάτιστα, polyschēmatista, Ench. 16.1) is still used of aeolic cola that respond in several forms.^[26]

Notes

1. ^ West 1987, p. 19.
2. ^ ^{a b} Battezzato 2009, p. 132.
3. ^ West 1987, pp. 23–24.
4. ^ West 1987, p. 48.
5. ^ West 1987, p. 4; Battezzato 2009, p. 132.
6. ^ West 1987, p. 31.
7. ^ Battezzato 2009, p. 135.
8. ^ West 1987, p. 34.
9. ^ West 1987, pp. 34–35 with n. 1; D’Angour 2022, p. 136.
10. ^ West 1987, p. 35.
11. ^ ^{a b} West 1987, p. 36.
12. ^ Battezzato 2009, p. 132; D’Angour 2022, p. 138.
13. ^ Battezzato 2009, pp. 132–133.
14. ^ West 1987, p. 42.
15. ^ ^{a b} West 1987, p. 43.
16. ^ West 1987, pp. 36–37.
17. ^ West 1987, pp. 37, 43–44, 47.
18. ^ West 1987, p. 33.
19. ^ West 1987, p. 50.
20. ^ West 1987, pp. 50–51.
21. ^ West 1987, pp. 51–52.
22. ^ D’Angour 2022, p. 138.
23. ^ Battezzato 2009, p. 133.
24. ^ Hephaestion, Ench. 15.1.
25. ^ West 1987, p. 32.
26. ^ West 1987, p. 89.

Here are the terms a learner meets, in alphabetical order. Each comes with the page of West 1987 on which it is defined or explained. His own glossary-index^[1] and list of symbols^[2] are the model, and the sigla in brackets are his. Where Battezzato, D’Angour, the ancient metricians or a teaching book use a word differently, the entry says so.

Schemes use these signs: — long, ⏑ short, × anceps, ⏓ usually long, ⏒ usually short, ⏔ biceps, contracted biceps, ⏕ resolvable long, ⏖ resolved long, ○ ○ two positions not both short, ⌊ and ⏗ trisemes, ⏘ tetraseme, ‖ period-end, ||| strophe-end, ∧ a suppressed position (catalexis, or syncopation in iambics).

Three groups of terms call for a warning. First, the names of feet (dactyl, iamb, spondee, cretic and the rest) are ancient. West keeps them for pointing to a place in a verse, but he analyses rhythm by metra and cola.^[3] D’Angour goes further. He regards units smaller than the colon as largely artificial, devised for analysis rather than used by poets or singers.^[4]

Second, several terms have an ancient sense and a modern one that do not match. They are asynarteton, system, prosodiac, enoplian, logaoedic and epode.

Third, the vocabulary of positions differs by school. West’s position is Maas’s element, and West’s princeps is Maas’s longum.^[5] Battezzato writes in Maas’s terms and adds an indifferent element.^[6]

Notes

1. ^ West 1987, pp. 85–90.
2. ^ West 1987, pp. ix–x.
3. ^ West 1987, p. 5.
4. ^ D’Angour 2022, p. 134.
5. ^ West 1987, pp. 86, 88.
6. ^ Battezzato 2009, p. 131.

“Dactyl”, “iamb”, “foot” and “colon” were coined as early as the fifth century BC.^[1] Aristotle distinguished metron from rhythmos.^[2] He judged iambics the metre closest to ordinary speech.^[3]

Hephaestion’s Enchiridion. Hephaestion was a metrician of the second century AD.^[4] He wrote the Ἐγχειρίδιον περὶ μέτρων (encheiridion peri metrōn), “Handbook on Metres”, which was edited by Consbruch (1906). Chapters 1–2 define short, long and “common” syllables and synecphonesis. Chapter 3 builds the feet from syllables. Chapter 4 lists the ways a metre may close: acatalectic, catalectic (one syllable short, or two where the foot is trisyllabic), brachycatalectic and hypercatalectic.

Καλοῦνται δὲ ταῦτα πάντα ἀποθέσεις.
All these are called “closes”.^[5]

The word for these “closes” is ἀποθέσεις (apotheseis).

Chapter 4 ends with two rules. Every metre’s last syllable is indifferent, and every metre ends in a complete word (4.5–6). Chapters 5–13 treat nine basic metres: iambic, trochaic, dactylic, anapaestic, choriambic, antispastic, ionic a maiore, ionic a minore and paeonic. Each is defined by the feet it admits and the places where it admits them (e.g. 5.1, 7.1):

Τοσαῦτα περὶ τῶν ἐννέα τῶν μονοειδῶν καὶ ὁμοιοειδῶν.
So much for the nine single-form and uniform metres.^[6]

The rest is derived. Chapter 14 mixes metra “by opposition” (κατ’ ἀντιπάθειαν, kat’ antipatheian). It makes the Sapphic eleven-syllable line a trochaic pair, a choriamb and an iambic close. It also notes that Alcaeus uses the line too, so that it is unclear which of the two invented it (14.1). Chapter 15 treats the ἀσυνάρτητα (asynartēta):

Γίνεται δὲ καὶ ἀσυνάρτητα, ὁπόταν δύο κῶλα μὴ δυνάμενα ἀλλήλοις συναρτηθῆναι μηδὲ ἕνωσιν ἔχειν ἀντὶ ἑνὸς μόνου παραλαμβάνηται στίχου.
Disconnected verses also arise, when two cola that cannot be joined to one another or form a unity are used in place of a single line.^[8]

Chapter 16 treats the πολυσχημάτιστα, “multiform” verses. It ends with a promise to speak next “on the poem” (περὶ δὲ ποιήματος ἑξῆς ῥητέον). A treatise On the Poem is cited from Consbruch’s edition.^[9]

Hephaestion is no mere compiler. He reports rival names (6.2). He doubts the name “paroemiac”, since proverbs come in every metre (8.6). Of a claim of Heliodorus about prosody he says “we proved it false” (1.9). Heliodorus (c. 100 AD) had explained the unmetrical ἕως of Iliad 1.193 by a trochaic first foot. He justified it by a theory that West counts among Heliodorus’ fanciful ones. Another metrician, probably Hephaestion, rejected it.^[10]

Rhythmicians and musical theorists. The rhythmicians measured verse in time. Their smallest unit is the mora (χρόνος πρῶτος). A long has two morae and a short one.^[11] A long of three morae is called τρίσημος, and one of four τετράσημος.^[12] The modern triseme signs are theirs.^[13] They spoke of a “cyclic” foot ⏑ ⏑ — of unrationalised rhythm.^[14]

The rhythmicians called the up-beat arsis and the down-beat thesis. Ancient musical texts put a dot over the notes of the arsis.^[15] Such dots appear in the Orestes papyrus.^[16]

Aristoxenus of Tarentum was a Peripatetic of the fourth century BC.^[17] He is one of the ancient sources for an explanation of a puzzle in responsion, together with Aristides Quintilianus, Hephaestion and Dionysius of Halicarnassus. Iambic metra respond with choriambic and ionic ones, and these writers explain this by the temporal equivalence of ⏑ — and — ⏑ (Kiparsky, citing Barris 2011).^[18] Aristides also recognised metres made wholly of long syllables, in which some longs last twice as long as others.^[19]

Scholia and colometry. Before the second century BC lyric was mostly copied as prose. The division into cola is often ascribed to Aristophanes of Byzantium (c. 255–180 BC). From the second century BC, texts divided into stanzas and triads are the norm.^[20] Some third-century papyri already show colometric arrangement. Aristophanes is credited with colometrised texts of Pindar. His successor Aristarchus edited Alcaeus and probably Anacreon, and commented on Alcman, Pindar and Bacchylides. The medieval scholia descend from such works and from the responses of Didymus, Aristonicus and Theon.^[21] Hephaestion himself has scholia and a commentary by Choeroboscus.^[22]

Kept and rejected. Modern metrics keeps the syllable rules, the final-syllable and complete-word rules, catalexis, the names of feet and cola (“first paeon”),^[23] the time-units and the Alexandrian colometry. It rejects or recasts several things:

• The foot as the unit of analysis. The metron serves analysis better,^[24] and units smaller than the colon can be treated as artificial aids to analysis.^[25]
• The ἀσυνάρτητα, whose disconnection is ordinary period-end.^[26]
• The derivation of aeolic lines from feet. Hephaestion’s “Aeolic” dactyls with a free first foot^[27] and his trochee-plus-choriamb Sapphic become West’s aeolic cola: ∧ia hag for the Sapphic line, and gl²ᵈ for the Sapphic fourteen-syllable.^[28]
• Hephaestion’s λογαοιδικά (7.8). The nineteenth-century “logaoedic” has also fallen out of use.^[11]
• Prosodiac and enoplian, which moderns use in various senses.^[29]

Hephaestion files the glyconic under the antispastic metre.^[30] He does so because the Alexandrians took ⏑ — — ⏑ ⏑ — ⏑ — as its basic form.^[31] D’Angour makes a wider complaint. In his view the mass of technical terms tends to cut the study of metre off from the music and rhythm it first existed to serve.^[25]

Notes

1. ^ D’Angour 2022, p. 132.
2. ^ Kiparsky 2018, p. 79 n. 3.
3. ^ D’Angour 2022, p. 135.
4. ^ Prauscello 2021, p. 221.
5. ^ Hephaestion, Ench. 4.4.
6. ^ Hephaestion, Ench. 13.8.
7. ^ Hephaestion, Ench. 14.
8. ^ Hephaestion, Ench. 15.1.
9. ^ Battezzato 2009, p. 137 n. 12.
10. ^ West 2018, p. 373 with n. 10.
11. ^ ^{a b} West 1987, p. 88.
12. ^ Battezzato 2009, p. 143.
13. ^ West 1987, p. 7.
14. ^ West 1987, p. 38.
15. ^ West 1987, pp. 85, 90.
16. ^ D’Angour 2022, p. 139.
17. ^ Katz 2018, p. 59.
18. ^ Kiparsky 2018, p. 97 n. 19.
19. ^ West 1987, p. 40.
20. ^ Sampson 2022, p. 114.
21. ^ Phillips 2022, p. 120.
22. ^ West 2018, p. 373 n. 10.
23. ^ West 1987, p. 55.
24. ^ West 1987, p. 5.
25. ^ ^{a b} D’Angour 2022, p. 134.
26. ^ West 1987, p. 32.
27. ^ Hephaestion, Ench. 7.5–7.
28. ^ West 1987, pp. 34–35.
29. ^ West 1987, pp. 87, 89.
30. ^ Hephaestion, Ench. 10.2.
31. ^ Kiparsky 2018, p. 98.

From feet to rules. Modern work began from the Alexandrian colometry^[1] and from Hephaestion’s feet. As late as 1711 Joshua Barnes explained an unmetrical ἕως in Homer as an amphibrach. In his view it stood in the first foot in place of a dactyl of equal value. J. H. Voss was the first to conclude that emendation was needed. Gottfried Hermann (1816) postulated an older *εἷος. Curtius (1846), who knew more historical linguistics, introduced the ἧος, τῆος that later editors adopted.^[2]

Hermann had in 1805 observed the hexameter bridge named after him.^[3] Porson discovered his law,^[4] and also the one now called Havet’s.^[5] The publications are these. Porson stated the rule for the end of the tragic trimeter in his notes to Euripides’ Hecuba (1797, on line 347) and Phoenissae (1799, on line 1464), and in the supplement to his Hecuba (1802). Hermann found it for the end of the trochaic tetrameter in his own Hecuba (1800). Snell extended it to the other metres it governs in his Bacchylides (1961). Hermann set out his bridge in his Orphica (1805).^[6]

On the beginning of the tetrameter the two sources differ, and both are reported here. West says that Porson discovered this rule as well, although it goes by Havet’s name.^[5] Maas credits the discovery to L. Havet himself, in his Cours élémentaire de métrique (1886).^[7]

Nineteenth-century terms. “Logaoedic”, for verse in asymmetric rhythm, has dropped out of use.^[8] “Ictus” is now rarely used. It is misleading if it implies a stress that could lengthen a short.^[9] The term “syncopation” is Westphal’s. It means the omission of a short, with supposed compensatory lengthening of a neighbouring long.^[10] It is still the conventional term in Greek metrics.^[11]

Wilamowitz extended anaclasis to further metres (1886). In 1921 he made it a basic principle of metrics that iambs, trochees, ionics and the rest grew from one abstract four-syllable metron.^[12] The anaclastic glyconic ○ ○ — × — ⏑ ⏑ — is sometimes called the wilamowitzianus after him.^[13]

Maas, Dale, Snell, West. Paul Maas’s term for a position is element, and his long position is the longum. West calls them position and princeps.^[14] Maas devised the dactylo-epitrite sigla D, e, E.^[15] In these sigla an Archilochean line is written D | × E ‖.^[16]

A. M. Dale analysed all of tragedy in Metrical Analyses of Tragic Choruses (1971–83). She discussed much of it in The Lyric Metres of Greek Drama (2nd edn 1968). West sends readers to her, to Dawe’s Teubner Sophocles and to modern commentaries, and he calls Jebb’s analyses “antiquated”.^[17] Battezzato, for his part, calls Dale’s notation “idiosyncratic”.^[18]

Bruno Snell’s Teubner Pindar and Bacchylides (revised by Maehler) analyse every poem. Snell aligned related sequences vertically, with bracketed sigla for mutant forms. This method rests on the principle of motif-development, which he was the first to grasp.^[19]

West’s Greek Metre (1982) was abridged in the 1987 Introduction, whose footnotes fall from 458 to 42.^[20] D’Angour calls Greek Metre the standard full-scale treatment, though hard for beginners. He finds Raven (1962) less forbidding.^[21] Battezzato too calls West’s handbook fundamental. He also recommends Snell’s brief Griechische Metrik and Martinelli 1995.^[22] Itsumi (1982) showed that Greek has no metres answering to the loose schema × × × × — ⏑ ⏑ —.^[23]

The comparative approach. Meillet (1923) saw that the Lesbian metres look archaic beside Vedic verse. Only the Greek–Indian comparison brings the inherited principles out clearly.^[24] The landmarks are Meillet 1923, Jakobson 1952, Watkins 1963 and West’s “Indo-European Metre” (1973).^[25] Kiparsky adds syncopation to Meillet’s inherited features. He rates it higher than the others, since it seems unique to Indo-European.^[26]

How the hexameter arose is disputed. Meister and Meillet thought it borrowed from a non-Greek Aegean people, which West (1997) refuted.^[27] West (1973) derived it from a pherecratean and an expanded reizianum, or from a hemiepes and a paroemiac. Nagy (1974) derived it from a pherecratean expanded by three dactyls. Battezzato finds this view a priori likely, given an Aeolic phase in epic.^[28] Berg (1978) derived it from a dimeter and a pherecratean, and Tichy (1981) modified his account. Kiparsky criticises West, Nagy and Berg. He derives the hexameter from a pair of iambic dimeters altered by syncopation.^[29] Nagy also argues that metre is “diachronically generated by formula rather than vice versa”.^[30] Kiparsky argues against this.^[31]

Notes

1. ^ Sampson 2022, p. 114; Phillips 2022, p. 120.
2. ^ West 2018, p. 373 with nn. 11–12.
3. ^ West 1987, p. 21.
4. ^ West 1987, p. 25; Battezzato 2009, p. 137.
5. ^ ^{a b} West 1987, p. 28.
6. ^ Maas 1962, pp. 35, 60.
7. ^ Maas 1962, p. 35.
8. ^ West 1987, p. 88.
9. ^ West 1987, pp. 87–88.
10. ^ Kiparsky 2018, p. 97 n. 20.
11. ^ West 1987, p. 7.
12. ^ Kiparsky 2018, pp. 97, 103.
13. ^ West 1987, p. 90.
14. ^ West 1987, pp. 86, 88.
15. ^ Battezzato 2009, p. 141.
16. ^ Kiparsky 2018, p. 108.
17. ^ West 1987, p. 52 with n. 5.
18. ^ Battezzato 2009, p. 146.
19. ^ West 1987, pp. 43, 45–46.
20. ^ West 1987, Preface.
21. ^ D’Angour 2022, p. 141.
22. ^ Battezzato 2009, p. 145.
23. ^ Kiparsky 2018, pp. 82–83.
24. ^ West 1987, pp. 1–3.
25. ^ Gunkel and Hackstein 2018, p. 3.
26. ^ Kiparsky 2018, p. 102.
27. ^ Kiparsky 2018, p. 104 n. 24.
28. ^ Battezzato 2009, p. 140.
29. ^ Kiparsky 2018, pp. 104–108, 122.
30. ^ Nagy 1990, p. 29.
31. ^ Gunkel and Hackstein 2018, p. 1.

Handbooks such as West’s describe Greek verse in its own terms: positions, metra and a stock of named cola. Linguists have also asked why Greek metre takes the shapes it does. They answer by relating it to the sound system of the language. Two such approaches are summarised here. Neither has replaced the descriptive notation used on this site.

Generative metrics. Kiparsky treats a metre as a “grammar” rather than a “dictionary” of patterns. On his view a metre is an abstract rhythmic template, plus constraints that match it to the text’s phonology. With Jakobson, he distinguishes the verse design, the verse instance (a text’s scansion) and the delivery instance (its performance).^[1] The patterns then prove to be simple binary structures. Their variety comes from the correspondence rules. So acephaly and catalexis are licences at the verse edges, not parts of the pattern.^[2] Gunkel and Hackstein open their volume with Brogan’s verdict that metrics still has nearly all of its work ahead of it.^[3]

Two units and markedness: Golston and Riad. Golston and Riad (2005) argue that Greek lyric metre is built from only two basic units, a dactylic and a trochaic. These are the same units that run through epic and dramatic verse.^[4] Each metrical position is a “moraic trochee”. This is a heavy syllable (H) or two lights (LL), and sometimes a single light. Each verse foot is two positions.^[5] Word accent, which was pitch, plays no part.^[6] The handbooks agree on this.^[7]

Their claim about rhythm is the novel one. Every dactylic unit contains a stress “clash”, and every trochaic unit a “lapse”. So the units break constraints that keep ordinary speech rhythmic. A metre is then defined by which of these marked features it uses, and where.^[8] Their tools are dactylic (D) and trochaic (T) units, extrametricality (their term for the aeolic base, s), catalexis (L) and “quantitative metathesis” (anaclasis). With these they reduce the aeolic cola to a few combinations:^[9]

West’s schemes are from his list of symbols.^[10] On this analysis the Sapphic stanza is TDT, TDT, TDTDD. The Alcaic is s TDTL, s TDTL, s TTDDT. So the two stanzas share their units.^[11] West writes the first two Sapphic periods ∧ia hag. He allows that the third might be analysed as ∧ia hag ⁝ ad. But in his view that hides the main point: the third period is a stretched-out form of the first two.^[12]

Authorities differ. Golston and Riad reject Raven’s analysis of aeolic verse around a choriambic nucleus.^[14] They judge the colon-based systems of Korzeniewski and of West’s Greek Metre (1982) to be either provisional or ad hoc. They give two reasons. Cola such as the adonean have no existence outside metre. And West allows two analyses of the last period of the Sapphic and Alcaic stanzas.^[15] For West, by contrast, the cola are what Greek verse is actually built from, and they recur from one metre to another.^[16] Golston and Riad’s treatment of the opening anceps of the iambic trimeter as an extrametrical “base”^[17] is also their own. Their analysis is offered here as a linguist’s alternative, not as the received view.

Notes

1. ^ Kiparsky 2018, pp. 77–79.
2. ^ Kiparsky 2018, pp. 80–81.
3. ^ Gunkel and Hackstein 2018, p. 1.
4. ^ Golston and Riad 2005, pp. 77–78.
5. ^ Golston and Riad 2005, p. 79.
6. ^ Golston and Riad 2005, pp. 79–80.
7. ^ West 1987, p. 2.
8. ^ Golston and Riad 2005, pp. 77–78, 113.
9. ^ Golston and Riad 2005, pp. 94–96.
10. ^ West 1987, pp. ix–x.
11. ^ Golston and Riad 2005, pp. 83–85.
12. ^ West 1987, p. 34.
13. ^ Golston and Riad 2005, pp. 83–85, 94–96.
14. ^ Golston and Riad 2005, pp. 109–110.
15. ^ Golston and Riad 2005, pp. 111–112.
16. ^ West 1987, pp. 4, 19.
17. ^ Golston and Riad 2005, pp. 88–89.

Time-values. In recitation a long was perhaps 5:3 to a short. So a long filling a biceps had to be drawn out slightly. In song the longs took a more exact relation to the shorts, normally 2:1, and in some metres 3:1.^[1] The ancients counted in morae, two to a long and one to a short.^[2]

The 3:1 long is called the triseme, written ⌊. It appears in iambic and trochaic song, where ⌊ ⏑ —, ⏑ — ⌊ or ⌊ ⌊ stands beside × — ⏑ —.^[3] It is found already in Alcman’s syncopated clausulae.^[4] In Bacchylides 17 such metra respond at eight places with unsyncopated ones, and this proves them equal in time.^[5]

The tetraseme ⊔ is worth two longs. It belongs to slow hymns. The Berlin Paean (c. 160 AD) is preserved with notation. It is an anapaestic scheme sung at half speed, with all its syllables long and some tetrasemes.^[6] Aristides Quintilianus knew several metres made wholly of long syllables, in which some longs were twice the length of others.^[7] Authorities differ. Battezzato holds that in music a long normally equalled two shorts. In his view there is only sporadic evidence for three- or four-time longs, at the end of the classical era.^[8]

Beat and ictus. Arsis is the up-beat and thesis the down-beat. Ancient notation puts a dot over the notes of the arsis.^[9] The old idea of ictus was a stress that reinforced certain positions and could lengthen a short. West rejects it as misleading.^[10] He also warns against reading the principes at equal intervals when the positions between them are unequal. English stress-habits tempt us to do this, but the asymmetry is what gives many metres their character.^[3]

The dochmiac is a case in point. English speakers read the usual mnemonic for it in four beats with silent rests, and there is no Greek evidence for this. The little evidence there is suggests two main beats, on the second and fourth elements.^[11]

The musical documents. The earliest extended notated melody on papyrus is a chorus of Euripides’ Orestes (338–45, 408 BC). It is in dochmiacs, accompanied by the aulos, and probably Euripides’ own setting. Some syllables carry stigmai, dots marking the arsis, on the first and third of the dochmiac’s five elements. This goes against the assumption that the beat always fell on the long elements.^[12] The Greeks may not have assumed that the down-beat is stressed. Lynch (2016) links arsis and thesis with pitch.^[13]

The melody shows the New Music. Its departures from traditional word-setting have been linked with the need for a notation. One of them is Euripides’ prolonged first syllable of εἱλίσσετε, mocked in Frogs 1314.^[14] In archaic lyric each syllable probably had a single note. Battezzato states this outright, and he dates to the fifth century the division of a long between two notes.^[8] Later he calls it probable, because classical writers criticised that newer practice.^[15]

The other notated document discussed here is the Song of Seikilos. It is a quatrain inscribed with notation, perhaps in the first century AD. It is the only post-classical verse with syncopation, apart from the end of the Epidaurian Hymn to Pan:^[16]

Melody and accent. Spoken Greek rose and fell in pitch. From the third century BC this was marked by the accents.^[17] The documents show moderate conformity of melody to accent. In the Orestes it cannot be complete, since one melody served strophe and antistrophe. The melody seems instead to follow a mimetic principle. It has a falling cadence on κατολοφύρομαι “I lament”, and an upward leap on ἀναβακχεύει, of frenzied leaping.^[18] Strophic song in general shows only partial correspondence. Some astrophic Hellenistic pieces with notation show close coherence.^[8]

For early lyric the most plausible hypothesis remains this. The melody followed the natural pitches of the words within one harmonia of the seven-stringed lyre.^[19] Simonides already stretched syllables mimetically, as in πῦρ for a flickering flame.^[20]

Instruments and delivery. The main accompaniments were the lyre family (lyra, barbitos, kithara) and the aulos. The aulos was a reed instrument, more like an oboe than a flute. The accompaniment mostly doubled the singer’s line. Instruments were tuned to a mode (harmonia), and each mode had its own character.^[21]

Drama used three kinds of delivery: recitation without music, recitation with music (παρακαταλογή, “recitative”) and singing.^[22] Archilochus was said to have invented parakatalogē.^[23]

Rhythms could also be given a character, an ethos, and here views differ. D’Angour holds that dochmiacs alone are tied to agitation or distress.^[11] West’s index entry “ethos of particular rhythms” points to spondaics, dochmiacs, lyric anapaests and ionics.^[24]

Notes

1. ^ West 1987, pp. 6–7.
2. ^ West 1987, pp. 7, 88.
3. ^ ^{a b} West 1987, p. 7.
4. ^ West 1987, p. 39.
5. ^ West 1987, p. 47.
6. ^ West 1987, p. 75.
7. ^ West 1987, p. 40.
8. ^ ^{a b c} Battezzato 2009, p. 143.
9. ^ West 1987, pp. 85, 90; D’Angour 2022, p. 139.
10. ^ West 1987, pp. 87–88.
11. ^ ^{a b} D’Angour 2022, p. 138.
12. ^ D’Angour 2022, p. 139.
13. ^ D’Angour 2022, p. 139 n. 6.
14. ^ D’Angour 2022, pp. 138–139 with nn. 4, 8.
15. ^ Battezzato 2021, p. 131.
16. ^ West 1987, p. 71.
17. ^ D’Angour 2022, p. 132.
18. ^ D’Angour 2022, p. 140.
19. ^ D’Angour 2022, pp. 140–141 with n. 11.
20. ^ D’Angour 2022, p. 140 n. 10.
21. ^ Battezzato 2009, pp. 143–145.
22. ^ West 1987, p. 48; see Speech and song.
23. ^ D’Angour 2022, p. 134; Battezzato 2009, p. 143.
24. ^ West 1987, pp. 40 n. 4, 56, 61–62, 87.

In drama the metre tells you how a passage was delivered. There are three modes of delivery: recitation without music, recitation with music (παρακαταλογή (parakatalogē), commonly called “recitative”), and singing. The recited parts are in stichic metres, or in long periods of uniform rhythm that behave like them. The sung parts are normally non-stichic and mostly strophic.^[1]

Seen from the audience’s side, the scheme is this. Lyric metres are used for song and dance by the chorus or a soloist. Other rhythms, above all anapaests, are used for recitative. Spoken iambics serve for most dialogue.^[2]

The borders are not sharp. Where one mode ends and the next begins is not always certain.^[1] There may be several modes between full song and plain speech.^[6] It is unknown whether parakatalogē meant a special chanted delivery.^[4] Anapaests come in two kinds, marching (recited) and lyric (sung). They are told apart by their metrical features.^[7]

How we know. The manuscripts do not mark which parts were sung. Editors decide by metre and by language. One sign is the “Doric” long α of tragic song, which the copyists often confused.^[8] The lyric metres of tragedy continue the tradition of archaic choral song, the line of Pindar and Bacchylides.^[9]

Tragic song went to the aulos, a pair of pipes with double reeds. The lyre was not the standard accompaniment, and it appears on stage only in special cases. So “tragic lyric” is, strictly, a misnomer.^[10]

A worked example: Euripides, Medea 1–266:^[11]

All Medea’s utterances before 214 are sung. The Nurse only speaks and chants. This marks her as subordinate, while the shared anapaestic metre links her to Medea.^[11] Who sings depends on dramatic prominence rather than on any fixed rule.^[12] Slaves and other low-status characters are given anapaests or hexameters but not true songs, the Phrygian slave of Orestes excepted. Male heroes in their prime do not, as a rule, sing. Sophocles is the important exception: he gives his heroes song in extreme suffering.^[13]

Strophes in drama. A tragic strophe is repeated only once, in its antistrophe. Sometimes an epode follows.^[14] No lyric passage in tragedy is longer than the lament of Libation Bearers 306–478, which has eleven strophic pairs. Sophocles and Euripides never exceed three pairs and an epode.^[15]

Change over the century. Choral song shrank as a share of the play, and actors’ solo songs grew. Some dramatists, following Agathon, used embolima. These were songs not written for one particular place in one play.^[16] The New Music of the late fifth century spread one syllable over more than one note. It also gave up strophic responsion for freer patterns. Its influence is visible in Euripides from Trojan Women (415 BC) on.^[17] The Orestes papyrus shows the first of these traits. But it dates from about 280 BC at the earliest, so whether its music is Euripides’ own remains an open question.^[18]

Metre and mood. Euripides often uses the dochmiac where a character reacts in fear or despair to news. But in his late recognition scenes (Iphigenia among the Taurians, Ion, Helen) dochmiacs also carry joy.^[19]

Notes

1. ^ ^{a b} West 1987, p. 48.
2. ^ Easterling 1997, p. 157.
3. ^ Wilson 2005, pp. 184–185.
4. ^ ^{a b} Battezzato 2005, p. 151.
5. ^ Wilson 2005, p. 185.
6. ^ Wilson 2005, p. 184.
7. ^ West 1987, p. 48; Battezzato 2005, p. 151.
8. ^ Battezzato 2005, pp. 150, 153.
9. ^ Battezzato 2005, p. 149.
10. ^ Wilson 2005, pp. 185–186; Battezzato 2005, p. 150.
11. ^ ^{a b} Easterling 1997, pp. 159–160.
12. ^ Easterling 1997, p. 159 n. 18.
13. ^ Battezzato 2005, p. 156.
14. ^ Battezzato 2005, p. 151; West 1987, p. 50.
15. ^ Battezzato 2005, p. 152.
16. ^ Easterling 1997, pp. 155–156.
17. ^ Wilson 2005, p. 190.
18. ^ Wilson 2005, p. 191; compare D’Angour’s view in Rhythm and music.
19. ^ Battezzato 2005, p. 158.