The Meter and Metrical Style of the New Poem

POxy. 1787, including Sappho 58 V—and so the “New Poem”—has been assigned to Book Four of the Alexandrian edition of her works since its first publication (Ox. Pap. XV:26; cf. Lobel 1925:xii). Because the assignment is a conjecture, this is a good moment to review the status of the question, and at the same time to observe, and try to appreciate, an apparent peculiarity in Sappho’s use of the meter in this poem. First, however, I will attempt to place the meter into the context of what we have of the work of Sappho and Alcaeus.

The Metrical Type of the New Poem

The meter seen in the New Poem had been known from complete examples provided by four quotations: Sa. 58.26 V, 81.4–8 V, 82a V, 91 V. The last two are from Hephaestion Enchiridion 7.5 (p. 36.15–16 C = Sappho 154 V Test.) who calls it aiolikon and says that “Sappho used it often (πολλῷ αὐτῷ ἐχρήσατο).” In general, such statements should be taken with a grain of salt. The ancient metricians, who liked to identify the “first discoverer” of each metrical form, also often named the “frequent user”; it is likely that both identifications were driven by a desire to explain the often inconsistent nomenclature. In this case, Hephaestion associates the aiolikon with an Aeolic poet. [1] Sappho’s use of the form is now also confirmed in several of the lines restored by the combination of the old and new papyri, and the number of fragments in P.Oxy. 1787 makes it clear that in this case we can take Hephaestion literally.

Modern handbooks mostly describe it (in the terminology established by Bruno Snell) as an acephalous hipponactean with a double choriambic expansion, or ^hipp2c:

x — ⏑ ⏑ — — ⏑ ⏑ — — ⏑ ⏑ — ⏑ — ⏒

It belongs to a larger group of metrical forms, or cola, which in modern usage are also often called aeolic. Individual aeolic forms allow no variations that would change the syllable count (contraction or resolution), but a few examples suggest that they can tolerate internally an exchange in the order of a long and short syllable. This is consistent with the theory that they represent more conservatively than other forms the derivation of Greek meters from an Indo-European syllable-counting system. More specifically, this form belongs with a group of cola that are most easily described in terms of the relation of each to the glyconic (or glyconean, gl): x x — ⏑ ⏑ — ⏑ —. The hipponactean is one syllable longer than a glyconic, the pherecratean one syllable shorter. The extra syllable of the hipponactean can be long or short, but as the final syllable of a verse line, it is conventionally assumed to be long. Here the hipponactean has only one indeterminate syllable at the start, and so is called acephalous, on the assumption that two syllables are normal in the aeolic basis. Since in fact one syllable bases are a normal variant, there would be some advantage in using West’s term hagisichorean (1982:30) to avoid the impression that the verse is somehow felt to be foreshortened, but I will retain the traditional term in this discussion. The acephalous hipponactean and the glyconic are both eight syllables long, and there is no reason to give one formal priority over the other (although the glyconic may descend from an earlier prehistoric stage). It is probably more useful to retain the emphasis on syllable-count and think of them as eight-syllable forms of the same type that have different endings: one blunt, — ⏑ —, and one pendant, ⏑ — —. Aeolic forms of this type can be expanded by repetition(s) of the internal sequence of two short syllables: choriambic expansion when the whole sequence — ⏑ ⏑ — is repeated and dactylic expansion when — ⏑ ⏑ is repeated. Lines could also be expanded externally by a full or partial iambic metron (x — ⏑ —, — ⏑ —, ⏑ — —) before or after the specifically aeolic segment, but there are only a couple of examples in the Lesbian poets of internal and external expansion of the same verse. Given that the iambic metra and the aeolic forms can merge seamlessly, how one divides a verse line between them and identifies the exact aeolic form involved generally depends on the theoretical assumptions behind the analysis.

Aeolic forms, especially the various expanded forms, make up the great bulk of what has been recovered of Lesbian poetry. Poems of unexpanded gl are not found, but in Sappho, stanzas variously formed from unexpanded glyconics and expanded ones are found together in one papyrus (P.Berol. 9722); in Alcaeus there is evidence of several such stanzas from a variety of sources. Simple ^hipp is found in four lines attributed to Sappho (168b V), but the attribution is disputed. Expanded forms are much more common. Examples of dactylic expansion of glyconic, especially, and of pherecratean are found for both poets, and choriambic expansion of glyconic—in particular, gl2c—is well represented in both. Sappho also has choriambic expansion of pherecratean, and of course, ^hipp2c. There is no firm evidence that Alcaeus ever used ^hipp2c, although two of his fragments (48 V, 61 V) have line ends that are compatible with it. Alcaeus, in fact, has very few examples of hipponacteans at all; the ones that do occur are the product of uncertain analyses. [2]

The description I have just given follows one entirely modern theory, and I have tried to set it out in a way that is compatible with a description in terms of Indo-European derivation. [3] Other modern descriptions have put more emphasis on the choriamb as the distinctive element (making dactylic expansion secondary); some have focused on the mix of dactylic and trochaic feet (and make choriambic expansion secondary); some follow more closely ancient authorities. All of these latter are from the late Hellenistic or Roman imperial period and offer entirely different groupings. Hephaestion, for example, in the passage cited above, calls the meter of the New Poem an acatalectic major ionic tetrameter. The corresponding glyconic form is an acatalectic antispastic tetrameter (10.6, p. 34.11–12 C). Other contemporary theories would have described both in terms of internal segments that could be related to the more familiar forms of iambic or dactylic poetry. Forms with dactylic expansion were usually related to dactylic verse, rather than to other forms of aeolic. If we want to know what Sappho or Alcaeus would have thought of the relation of the various meters to each other, we can only infer it from how they combined them in stanzas.

Most of the metrical forms we know occur as single lines; that is, we know more line types than stanza types, especially in the case of Sappho’s poetry. This may be in part because ancient authorities tended to quote only single lines, even in cases in which we know that the lines are parts of stanzas. But, as will be discussed below, a series of identical lines arranged in couplets seems to have been a frequent form of composition. The stanza types that do survive—and there are more for Alcaeus than Sappho—are especially valuable for the negative evidence they give (the examples are most easily seen in Voigt’s “Conspectus” for each poet; see also West 1982:31–33). There is no stanza which combines both choriambic and dactylic expansion, although there are stanzas which combine each with the simple forms or (more rarely) with externally expanded forms. The use of the contrast between blunt and pendant endings to set off sections of verse—for example glyconic and pherecratean—familiar from the versification of Anacreon and throughout dramatic lyric, is not found in the Lesbian poets; with the exception of the Alcaic stanza, they use one or the other type of ending consistently in a stanza. [4] Variations between lines in a stanza are largely achieved by the amount of internal or external expansion. [5] To put this positively, by maintaining uniformity in construction, they were able to make a variety of distinctive song types out of a smaller variety of often similar materials. And within the limits of the state of our evidence, the New Poem, written entirely in ^hipp2c, appears to exhibit a metrical form that is distinctively Sappho’s.

I will consider below the particular similarity of the New Poem with others written in repeated lines and with the corresponding metrical type gl2c.

The Arrangement of Meters in Books

While it is generally assumed that the Alexandrian edition of Sappho arranged the poems by meter, the evidence associating different metrical patterns with different books is quite sparse (unless otherwise cited, see Sappho 226–234 V and Sappho T29–T36 Campbell with his notes). No ancient source actually says that the poems of Sappho were arranged by meter, but the descriptions that say certain meters were found in certain books leads to this conclusion. In addition, the papyri that show multiple poems in the same (or very similar) meters or stanzas (P.Berol. 5006, 9722; P.Oxy. 1231+2166(a), 1787) suggest a metrical arrangement. And in the case of the first three books the ancient sources tell us that each of them contained poems of a single metrical type. Since any hypotheses we make about the distribution of meters in the remaining books need to take account of all of them, my discussion of Book Four here is necessarily somewhat simplified. [6]

There is no evidence whatsoever, direct or indirect, for Book Four. As Page explained, the assignment of the meter used in the New Poem to Book Four was based on the observation that P.Oxy. 1787 contains fragments of “a considerable number of poems, … The evidence, as far as it goes, indicates that these manuscripts come from a book wholly or largely homogeneous in metre; if so, it appears reasonable to assign to it the fourth place in the series, since the first three books were metrically homogeneous whereas the fifth was not” (1955:114–115). In fact, there are not enough lines assigned to this papyrus (or in this meter) to guarantee that they constitute a book “wholly or largely” in one meter (nor do they guarantee enough columns), not even one half the size of the 1320 lines recorded by the colophon to Book One, but we can safely say that there were many poems in this meter which apparently belonged to one book. I emphasize this because of a potential complication in the idea of a “book.” Hunt reported in his original edition of P.Oxy. 1787 (Ox. Pap. XV:126–127) that fragments of other poetry in the same hand were found with these. Now that we possess one (albeit much earlier) papyrus combining poems in this meter with non-Lesbian poetry, the possibility that P.Oxy. 1787 does not represent a book roll from an edition of Sappho, but some other kind of compilation, seems less far-fetched. [7] I shall proceed, however, on the assumption that there are a sufficient number of poems in the same meter to represent a selection from a single book, even if the roll whose fragments we possess is not actually from a complete edition.

In addition to the book-by-book evidence for metrical patterns, we can also consider the possibility that stanza structure played a role in the arrangement of poems into books. In the Greek sources, [8] poems are distinguished on the basis of whether they are written in lines or in systems (i.e., stanzas). The latter can be composed of similar or dissimilar metra. Sappho Book Two is cited in the Peri Poematon as an example of poems composed in stanzas in similar metra (1.1, p. 63.7–8 C). This is then further explained under the rubric of “common” systems, and the example given this time is poems such as those in Sappho Books Two and Three: these are in stanzas because they are marked off every two lines and the total number of lines is always even (that is, they are written in couplets, or distichs), but since the lines are identical, they could also be said to be written in lines (1.2, p. 63.15–24 C). Paragraphoi that represent such markings are in the papyri for Book Two, in which all the poems are in the meter gl2d. And both the Oxyrhynchus and Cologne papyri have paragraphoi that mark off distichs in the poems we are now attributing to Book Four, even though that book is not cited as part of the example. However, since the author of Peri Poematon was only using Sappho as an example—not describing her edition—he did not need to include Book Four when he cited Two and Three. But there may have been another reason to omit it. P.Oxy. 2290 = Sa. 88 V contains lines that are compatible with ^hipp2c. The papyrus fragment contains the right edge of a column and every third line appears to be distinctly shorter—although not very much shorter—suggesting that it contained three-line stanzas (see Voigt’s apparatus and the note ad loc. added in Lobel-Page 1963:67). Some poems from P.Oxy. 1787 appear to follow the same pattern; in that case, Book Four would have contained at least that one other type of stanza form. [9]

I suggest that a series of considerations affected the distribution of meters. Books One to Three contained poems in the only three meters in which there was enough material to fill a book. That would be consistent with the fact that no source associates other books with a single meter. Beginning with Book Four, then, different meters were combined in each book. [10] One source, the single piece of evidence for the meter of Book Seven, indirectly suggests that within each book poems were still grouped by meter. [11] Unfortunately, this passage is corrupt: Hephaestion Enchiridion 10.5, p. 34.7–8 C reads: …ᾧ μέτρῳ ἔγραψεν ᾄσματα καὶ Σαπφὼ ἐπὶ † τῆς τοῦ ἑβδόμου· (the mss. vary in regard to the extra article), “a meter in which Sappho also wrote songs at † the of the seventh Book.” It appears to indicate that the use of this meter characterized one part of the book. [12] If we accept that as a general rule of arrangement, the next question is, what governed the selection of meters to be combined into a single book. The possible combination of two- and three-line stanzas in P.Oxy. 1787, and the collection of similar stanzas in P.Berol. 9722 (whatever book they belonged to) suggest that perceived metrical similarity determined the groupings. At the same time, the order of the books themselves might reflect a movement from those with uniform lines to stanzas with similar lines of varying length to more mixed stanzas. Although Book One may owe its position to the special prominence of the Sapphic stanza, it may also have been seen to use continuous identical lines in a system, since the “epodos” that concludes the Sapphic stanza may not have been considered a line in the same sense as the previous ones. [13]

Book Three included lines with double choriambic expansion, but only of the form gly2c; since there were in addition poems in lines with double expansion in the form ^hipp2c, it is reasonable to think that they would have been assigned to Book Four. It is reasonable because the general similarity of the two lines seems obvious—they have the same number of syllables and the same relationship to each other as the hendecasyllables of the Sapphic and Alcaic strophe—and the pattern of couplets continues. [14] A variant three-line stanza, using the same long line twice, could have been part of the same book if metrical similarity was a determining criterion. Perhaps this book could have also contained the fifteen-syllable pher2c (which shares the further similarity of a pendant ending with ^hipp2c); that verse pattern is illustrated by Sappho 140 V, the mourning for Adonis (it may share thematic affinities with the New Poem; the original editors [Gronewald and Daniel 2004] made an explicit connection between the New Poem and Phaon/Adonis, and, at the least, the irreparability of death is common to both sets of lines).

In the absence of direct evidence of any sort, anything we say about Book Four is subject to doubt. There is also no evidence for Book Six, and in fact we have no explicit reason to exclude any meter from any book after Book Three. But until more evidence is discovered, the criteria I have described at least confirm the reasonableness of assuming that the New Poem belongs to Book Four and give us some sense of its place in the whole. If this is the case, the reader of the ancient edition of Sappho encountered poems in each of the first four books—half the edition—that were in repetitions of single, long verse forms. First, the externally expanded eleven-syllable line of the Sapphic stanza filled Book One, and this was followed by even longer lines with internal expansion written mostly in couplets: the fourteen-syllable lines of glyconics with double-dactylic expansion in Book Two, and the sixteen-syllable lines of glyconics and of acephalous hipponacteans, both with double choriambic expansion, in Book Three and much of Book Four.

Metrical Style in the New Poem

I turn now to how Sappho used the verse form in the New Poem. In her Conspectus Metrorum Voigt indicates that the lines with double choriambic expansion tend to have a word-end between the choriambs. This is true for gl2c (also called the major asclepiad) in both Sappho’s and Alcaeus’s practice, [15]

x x — ⏑ ⏑ — ¦ — ⏑ ⏑ — ¦ — ⏑ ⏑ — ⏑ ⏒

and Sappho’s treatment of ^hipp2c is similar:

x — ⏑ ⏑ — ¦ — ⏑ ⏑ — ¦ — ⏑ ⏑ — ¦ ⏑ — ⏒

Voigt marks a word-end preference when word-end occurs in at least half of the instances. Such counting depends on many assumptions, not least on how a “word” is to be defined and how to treat partial or uncertain lines. I have not tried to double-check her counts, but they can certainly be accepted as broadly indicative, and in themselves they are not surprising. (Horace turned the division between choriambs into a rule in his asclepiads.) Thus, the ine which Hephaestion first quotes (Sappho 83a V) to illustrate the meter that also occurs in the New Poem can be taken as a model: [16]

εὐμορφότερα | Μνασιδίκα | τὰς ἀπάλας | Γυρίννως

As a matter of style, however, we expect that Sappho will not follow this pattern rigidly; one of the characteristics of her composition is variety. [17] And when variety is absent, it is typically absent for a reason. This can be seen most clearly in Book One, where she normally avoids long sequences with word-ends (or with no word-ends) in any one given position but does have continuous word-ends in two passages. Both of these have a strongly unified and marked character: in every line but the last of fr. 31 V there are word-ends after the fourth position—where they are, in addition, unexpected, since there tends to be a bridge in that position in other poems—and, less remarkably, in Poem 1.18–25 V, the epiphany of Aphrodite, word-ends fall unvaryingly in the position after the fifth syllable (see Lidov 1993:510–513). We can use two four-line fragments, Sappho 55 V in major asclepiads and Sappho 81.4–7 V in this meter, as samples of her style of composition in a verse with extended choriambic expansion. In the following schemes, broken lines indicate word-ends formed by appositives, and light type textually uncertain passages. [18]

Sappho 55:


κατθάνοισα δὲ κείσηι͜οὐδέ ποτα μναμοσύνα σέθεν
ἔσσετ’ οὐδὲ †ποκ’ -†ὔστερον· οὐ γὰρ πεδέχηις βρόδων
τὼν ἐκ Πιερίας, ἀλλ’ ἀφάνης κἀν Ἀίδα δόμωι
φοιτάσηις πεδ’ ἀμαύρων νεκύων ἐκπεποταμένα.

In these lines disparaging the unmusical lady, the second halves of the lines are marked both by the consistent word-end at the tenth and fourteenth position and by the sense. The content is repetitive, but, notwithstanding the enjambments, every line rises to a point of emphasis at the end. (There is reinforcement in the sound of the responding alphas of μναμοσύνα and κἀν Ἀίδα, picked up in the close of the six-syllable finale.) But the word-end pattern that is stressed is the normal one, with a cut between the choriambs. Before the tenth position there is more variety, with no or almost no word-ends repeated two lines in a row, save, as expected, between the first two choriambs.

Sappho 81:


σὺ δὲ στεφάνοις, ὦ Δίκα, πέρθεσθ’ ἐράτοις φόβαισιν
ὄρπακας ἀνήτω συν<α>έρραισ’ ἀπάλαισι χέρσιν·
εὐάνθεα † γὰρ – πέλεται? καὶ Χάριτες μάκαιρα<ι>
μᾶλλον †προτερην?, ἀστεφανώτοισι δ’ ἀπυστρέφονται.

Here the cut between the choriambs is not so strongly marked, though it occurs at least once in the four lines at each position. What pattern of cuts there is, after the tenth syllable, is muted by elision (and a prepositive). These lines tend to have longer words (individual words of three or more syllables account for 50 of the 64 syllables in Sappho 81, 39 in Sappho 55), and the general effect is of variety and continuous flow, without any prominent interior structure. The sense, too, develops continuously across the two couplets.

Even if one deems such judgments excessively subjective, the patterning in these lines makes a clear contrast with what happens in the New Poem (West’s text, without supplements):


x — ⏑ ⏑ — — ἰ]οκ[ό]λπων κάλα δῶρα, παῖδες,
x — ⏑ ⏑ — —]ν̣ φιλάοιδον λιγύραν χελύνναν·
x — ⏑ ⏑ — —] π̣οτ’ [ἔ]ο̣ντα χρόα γῆρας ἤδη
x — ⏑ ⏑ — — ἐγ]ένοντο τρίχες ἐκ μελαίναν·
βάρυς δὲ μ’ ὀ [θ]ῦμος πεπόηται, γόνα δ’ [ο]ὐ φέροισι,       5
τὰ δή ποτα λαίψηρ’ ἔον ὄρχησθ’ ἴσα νεβρίοισι.

τὰ <μὲν> στεναχίσδω θαμέως· ἀλλὰ τί κεν ποείην;
ἀγήραον ἄνθρωπον ἔοντ’ οὐ δύνατον γένεσθαι.

καὶ γάρ π̣[ο]τ̣α̣ Τίθωνον ἔφαντο βροδόπαχυν Αὔων
ἔρωι ⏑ ⏑ — ανβάμεν’ ἐς ἔσχατα γᾶς φέροισα[ν,      10

ἔοντα̣ [κ]ά̣λ̣ο̣ν καὶ νέον, ἀλλ’ αὖτον ὔμως ἔμαρψε
χρόνωι π̣ό̣λ̣ι̣ο̣ν̣ γῆρας, ἔχ[ο]ντ’ ἀθανάταν ἄκοιτιν.

The expected cut between the second and third choriamb is entirely missing in the second half of the verse for the first six lines, whereas a pyrrhic-shaped word—in every case an unelided, lexical word—is marked off in five of these lines. Such words are neither rare nor common. They occur on average about once in every ten lines of the hendecasyllable of Book One, but never twice in a row. They were entirely absent from the eight lines I analyzed above. [19] The series of words here—κάλα, χρόα, τρίχες, γόνα, ἴσα—represents a notable rhythmic variation. This is combined with the persistent deviation from the normal line structure in the preceding four syllables of the first five lines, in which the preserved or highly probable word occupies a minor ionic shape that bridges the expected cut: ἰ]οκ[ό]λπων, φιλάοιδον, π̣οτ’ [ἔ]ο̣ντα, ἐγ]ένοντο, πεπόηται. Together these oddities must both unify the six lines that describe the singer’s woes and contribute to a peculiar effect. There is an obvious parallel in the persistence of cuts after the fourth syllable and the absence of them after the fifth in the “pathography” of fr. 31. In this poem the deviations come to a sudden halt with punctuation between the choriambs in line 7, as the singer dismisses the complaints and turns to the consolation or recompense suggested by the myth. After that there are no more isolated pyrrhic words in the poem (only the word group καὶ νέον in line 11), and we find varied expressions of the normal structure for the last half of the verse. On the other hand, the trisyllabic final cadence—weakly (after a prepositive) and inconsistently marked in the first six lines—is enunciated with increasing clarity, so that only a god’s name violates it in the last six lines.

The absence of the opening of the first four lines makes it impossible to know whether a similar rigidity characterized the beginnings of the verses, but the persistent absence of a cut between the choriambs from lines 5–10 and the word-end following the spondee in the fifth and sixth syllables in lines 5–7 suggest that it did. The supplements Μοίσαν and λεῦκον for lines 1 and 4, which seem certain for other reasons, would also fulfill the pattern, so it seems all the more likely. This could have consequences for how we supplement the rest of the text. Given the pattern, the article in line 2, καὶ τὰ]ν, printed by West (following Gronewald-Daniel’s πάλιν τὰμ̣), which would be unexpected simply as an article, seems all the more improbable; I have suggested φώνα]ν or αὔδα]ν as the object of a verb (allowing the last two words of the line to be read as genitive plurals). If we are to keep ἄπαλον in line 3, then Gronewald’s and Daniel’s original suggestion of ἄπαλόν μοι would cause less variation than Di Benedetto’s ἄπαλον πρὶν, printed by West.

In the last couplet of the poem, the expected cuts are expressed in all instances, softened only by elision. The return to the normative structure brings the myth and the emotional journey to a close (one can again compare Sappho 31 V, where, in the final hendecasyllable, the normative cut after the fifth syllable returns, and there is no cut after the fourth). This pattern is maintained by as much as can be recovered from the quotation by Athenaeus that would occupy lines 15 and 16, if these are part of the same poem.

The observation of these tendencies can have, at best, a suggestive value when it comes to answering the questions of what to print, and a cautionary value when we try to fill in verses with our own composition; more importantly, they can also aid us in appreciating the artistry of the texts that we do have or can reconstruct.


Campbell, D. A., ed. 1982. Greek Lyric I. Cambridge, MA. Corrected second edition 1990.

Consbruch, M., ed. 1906. Hephaestionis Enchiridion cum commentariis veteribus. Leipzig. Reprinted Stuttgart 1971.

Gallavotti, C. 1957. Saffo e Alceo: Testimonianze e frammenti. Vol. 1: Saffo. Revised 3rd edition. Naples.

Gronewald, M., and R. W. Daniel. 2004. “Ein neuer Sappho-Papyrus.” Zeitschrift für Papyrologie und Epigraphik 147:1–8.

Hunt, A. S. 1922. “New Classical Fragments. 1787. Sappho, Book IV.” Oxyrhynchus Papyri Part XV (ed. B. P. Grenfell and A. S. Hunt) 26–46. London.

Korzeniewski, D. 1968. Griechische Metrik. Die Altertumswissenschaft. Darmstadt.

Liberman, G., ed. 2002. Alcée: Fragments. Paris.

———. 2007. “L’édition alexandrine de Sappho.” Atti del convegno internazionale di studi ‘I papiri di Saffo e di Alceo’: Firenze 8–9 giugno 2006 (ed. G. Bastianini and A. Casanova) 41–65. Studi e Testi di Papirologia n.s. 9. Florence.

Lidov, J. B. 1993. “The Second Stanza of Sappho 31: Another Look.” American Journal of Philology 114:503–535.

Lobel, E., ed. 1925. ΣΑΠΦΟΥΣ ΜΕΛΗ: The Fragments of the Lyrical Poems of Sappho. Oxford.

Lobel, E., and D. L. Page, eds. 1963. Poetarum Lesbiorum Fragmenta. Oxford. First edition 1955. (= L-P)

L-P. See Lobel and Page 1963.

Nagy, G. 1996. “Metrical Convergences and Divergences in Early Greek Poetry and Song.” Struttura e storia dell’esametro greco, Vol. 2 (ed. M. Fantuzzi and R. Pretagostini) 63–110. Studi di metrica classica 10. Rome. Revised version of Appendix to Pindar’s Homer (Baltimore 1990).

Page, D. L. 1955. Sappho and Alcaeus: An Introduction to the Study of Ancient Lesbian Poetry. Oxford. Corrected reprint 1959.

Parker, L. P. E. 2001. “Consilium et ratio? Papyrus A of Bacchylides and Alexandrian Metrical Scholarship.” Classical Quarterly 51:23–52.

Powell, J., trans. 1993. Sappho, a Garland: The Poems and Fragments of Sappho. New York.

V. See Voigt 1971.

van Ophuijsen, J. M. 1993. “On Poems: Two Hephaestionic Texts and One Chapter from Aristides Quintilianus on the Composition of Verse.” Aufstieg und Niedergang der römischen Welt II.34.1 (ed. W. Haase) 796–869.

Voigt, E.-M., ed. 1971. Sappho et Alcaeus: Fragmenta. Amsterdam. (= V)

West, M. L. 1982. Greek Metre. Oxford.

Yatromanolakis, D. 1999. “Alexandrian Sappho Revisited.” Harvard Studies in Classical Philology 99:179–195.


[ back ] 1. Liberman (2002:I.cviii) aptly prefaces his collection of metrical testimonia for Alcaeus with a general notice from Aelius Theon (Progymnasmata 73 p. 32 Patillon-Bolognesi): ὥσπερ Ἀριστοφάνειόν τι μέτρον καὶ Σαπφικὸν καὶ Ἀλκαϊκὸν καὶ ἄλλο ἀπ’ ἄλλου λέγεται, οὐχ ὡς τούτων τῶν ποιητῶν μόνων ἢ πρώτων ἐξευρηκότων τὰ μέτρα, ἀλλ’ ὅτι αὐτοῖς ἐπὶ τὸ πλεῖστον ἐχρήσαντο (“so a certain meter [is called] Aristophanean or Sapphic or Alcaic, and another is called after another poet, not because these poets alone [used] or first found the meters, but because they made most use of them”).

[ back ] 2. Other types of meter—dactylic, ionic, and iambo-trochaic—also occur in both poets. Since many of these occur in citations from ancient metricians, who were more interested in cataloguing types than in describing a body of work, it is difficult to know just how large a role non-aeolic forms play.

[ back ] 3. For a history of the aeolic forms, see Nagy 1996.

[ back ] 4. One analysis of Alcaeus 130b V yields pendant hipp together with blunt gl and glc, as well as an exceptional combination of the latter with ^glc. But the alternate analysis, combining only expanded and unexpanded gl, conforms much more on all counts to Alcaeus’s practice elsewhere.

[ back ] 5. There is some very slight evidence for asynartete verse in the Archilochean style, with contrasting lines (see section D in each Conspectus in Voigt); such verse forms may also account for some of the isolated iambo-trochaic lines cited in antiquity.

[ back ] 6. I hope to present a fuller discussion elsewhere. Most of the difficulties concern Book Five, and are not immediately relevant to the discussion of the meter of the New Poem. Page offers a summary discussion of the evidence for distribution of meters (1959:112–116). Liberman 2007 provides a fuller presentation, with a spare critique and his own hypotheses; I am in agreement with him on many points. Yatromanolakis 1999 surveys the evidence for the number of books and, against Page, reaffirms Wilamowitz’s conclusion that there were only eight books and that the last was referred to as “Epithalamia“; Liberman argues for nine books in eight rolls. I shall assume that there were eight books, and not review the question further here.

[ back ] 7. This is also the case with P.Oxy. 1788, from the same find but in a different hand; the aeolic fragments were assigned to Alcaeus by the first editors (= Alcaeus 115–128a V) on the basis of their style and of their variety in meter and content, on the assumption that they all came from one book of one poet. Subsequent discussions have all been based on the same assumption (see Liberman 2007:55–56n59, with reference to previous literature, where he withdraws his earlier ascription to Sappho). But there exists the possibility that they are from a collection, and are by more than one author.

[ back ] 8. Stanzas are discussed in the Peri Poematos and Peri Poematon that are found with the mss. of Hephaestion Enchiridion (they can be assumed to epitomize the same source; see van Ophuijsen 1993); the latter contains the fuller treatment of this topic. There is nothing of additional significance in the Latin sources.

[ back ] 9. What survives for the first two lines of the supposed stanza of Sa. 88 V is, at most, the first seven of the last nine syllables, …]⏑——⏑⏑—⏑[—x. For the third line, the end survives, …]⏑⏑—⏑—x. As Voigt indicates, it is likely that the third line is one choriamb shorter. In P.Oxy. 1787, the same pattern is possible for Sappho 73 V, and (more doubtfully) 64a V and 86 V (note especially line 4); in Sappho 65 V, there is a paragraphos after line 9, and confusing traces of what would be another both two and three lines earlier. Gallavotti 1957 gathers these fragments as a set of poems separate from the ones in distichs (nos. 72–75 in his edition).

[ back ] 10. As Liberman 2007:49 observes, the inclusion of more than one meter in the book would also be consistent with the difference between Hephaestion’s language introducing ^hipp2c, Σαπφὼ πολλῷ αὐτῷ ἐχρήσατο (Enchiridion 11.5 [p. 36.15–16 C] = Sa. 154 V Test.), and the formula that he uses for the meters of Books Two and Three, ᾧ τὸ δεύτερον (τρίτον) ὅλον Σαπφοῦς γέγραπται, Enchiridion 7.7 [23.15 C], 10.6 [34.12 C] = Sappho 227, 229 V).

[ back ] 11. This is a separate conclusion from Lobel’s hypothesis that the poems of the same metrical form were arranged alphabetically (1925:xv).

[ back ] 12. A word such as τελευτῆς (Bergk) or ἀρχῆς (Westphal ap. Bergk) may be missing, or a confusion over an abbreviation of, e.g., τέλους (Bergk) may be behind the error.

[ back ] 13. The ancient practice of writing the poem out in three identical lines and one different line would seem to satisfy the definition of a poem in stanzas of dissimilar metra (and if it did, one could then wonder why the example was not used, given the examples from Sappho for stanzas in similar metra). And certainly some Latin sources simply describe it as four “verses” (e.g., Diomedes Thrax 1.519, 521 Keil). But Dionysius Halicarnassensis in De compositione verborum 19, discussing sources of variety in composition, remarks that Sappho and Alcaeus “composed small strophes, without much variety in the few cola, and used very few epodes” (μικρὰς ἐποιοῦντο στροφάς, ὥστ’ ἐν ὀλίγοις τοῖς κώλοις οὐ πολλὰς εἰσῆγον μεταβολάς, ἐπῳδοῖς τε πάνυ ἐχρῶντο ὀλίγοις). The masculine epodos here (in distinction from the feminine, used a few lines earlier, to indicate the third strophe of a triad) seems likely not to refer to Archilochean epodes as a genre but to what gives those their name, the epode which Peri Poematon 7.2 (71.2–3 C) defines as “something extra (περιττόν τι) attached to a long line.” That could easily be a description of the last segment of the Sapphic stanza; epodus is also used for it by Marius Victorinus (Aphthonius), 6.161–162 Keil. From that point of view, the separation of the fourth “line” serves less to mark it as a colon than to preserve the similarity of the other three—an exception that tests the rule defining stanzas of similar metra, but does not break it (although it would certainly disqualify Book One from being used as an example).

[ back ] 14. The similarity would not have been obvious according to the theories of the Enchiridion, mentioned earlier, by which the meter of Book Three is antispastic, and that of the New Poem major ionic. The treatment of the latter as a version of the same type as the former involves a category of variation not recognized in that handbook but familiar to other ancient authorities (Atilius Fortunatianus 6.296–297 Keil specifically applies it to derive the Sapphic from the Alcaic hendecasyllable by the transfer of one syllable). I am assuming that the editors of the Alexandrian edition would have been pragmatically influenced by the fifteen syllables the two forms have in common (on the pragmatism of Alexandrian colometry in general, see Parker 2001). Without that assumption the argument is weaker, and depends more on the assumption—made by Page, but which I consider unjustified—that we know that Book Five contained an even greater variety of stanza types, and for that reason should follow the book represented by P.Oxy. 1787. I propose in my interpretative essay, in this volume, that the similarity is also substantive.

[ back ] 15. The first syllable is always long in Alcaeus’s poems and usually long in Sappho’s; the second is usually long in Alcaeus’s. Alcaeus also has a bridge within the second choriamb.

[ back ] 16. Enchiridion 11.5 (36.17 C). He also quotes Sappho 91 V, ἀσαροτέρας οὐδάμα πω͜Εἴρανα, σέθεν τύχοισαν, which begins with a short syllable, and has no break (with synizesis) between the second and the third choriamb; by his ionic analysis the initial short would be a variation from the norm worth mentioning.

[ back ] 17. As Jim Powell, one of her best translators, puts it: “Sappho’s secret consists largely in keeping her caesura moving: in her sapphics the caesura (a pause in mid-verse) seldom falls in the same place in two consecutive lines” (Powell 1993:38).

[ back ] 18. In what follows I assume a rising scale of value for word-junctures, from clitic-junctures, which might fulfill the expectation of word-end but do not necessarily violate expected bridges, to elisions (including synizesis) to full word-divisions; the latter two can be reinforced by phrasing (punctuation). See Korzeniewski 1968:16–17.

[ back ] 19. I suspect that those eight are somewhat unusual in having none at all; I count another eight instances in the scraps numbered 59–88 in Voigt’s text (63.3, 7; 65.7, 9; 68.7, 8; 70.10; 71.5). Pyrrhic words may be slightly more common in lines with dactylic expansion; there are five in Sappho 44 V and one in almost every one of the remaining quotations from Book Two. The diminished opportunities to use words with internal spondees might have led to a different set of vocabulary choices (originally formulas) in the meter of Book Two. In discussing Book One I am including the third hendecasyllable.