You are on page 1of 24

American Philological Association

Alternating Rhythm in Archaic Greek Poetry
Author(s): Joel B. Lidov
Source: Transactions of the American Philological Association (1974-), Vol. 119 (1989), pp. 63-
85
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/284261 .
Accessed: 07/10/2011 14:53

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Philological Association and The Johns Hopkins University Press are collaborating with JSTOR to
digitize, preserve and extend access to Transactions of the American Philological Association (1974-).

http://www.jstor.org

Transactionsof the AmericanPhilological Association 119 (1989) 63-85

ALTERNATING RHYTHM IN ARCHAIC GREEK
POETRY

JOELB. LIDOV
Queens College and The GraduateCenter,CUNY
Most early Greek verse appears to be composed of a series of long syllables
which are separatedby another metrical "element":a single short syllable, or
two short syllables (for which a long syllable appearsinstead in some lines), or
a single syllable which is short in some lines, long in others. Such verse is said
to exhibit "alternating rhythm."'It is a feature of some of the most familiar
verse forms; for instance,the iambic trimeter(Archilochus30.1):
^ouoca aotkXv voupoivrq; ircpTirro

the Sapphichendecasyllable(Sappho 16.1):
Oi .?ev unrijov o(xpO6ov, Oit8? neTOoCV

the dactylic hexameter(Archilochus1.1):
eiti. 8' Eyo) 09cpadxov
atEiv 'Ev-aookto oavaKTox;

On the other hand, it appearsto breakdown in so common a verse as the second
line of the elegiac couplet, the "pentameter"(Archilochus 1.2):
Icat Mowo.omvipcOTovSopov ?7rtlor6CdvoS

To the best of my knowledge alternationis not considereda constitutivefeature
of Greekverse in any moder handbookon Greekmeter.
But the fact remains that alternatingrhythmis an attractivelysimple unify-
ing phenomenon for a very large portion of surviving Greek poetry, and-in
one guise or another-it has kept recurringas an explanatoryelement in theo-
ries of Greek meter (partI). I will propose a new definition, one that makes al-
ternationa feature of an abstractlevel of metricaldescription,and use that as a

1 Paul Maas, Greek Metre, trans. H.
Lloyd-Jones (Oxford 1962 [orig. ed. 1923])
32. This and the following are referred to by their authors' last name: Bruno
Snell, Griechische Metrik, 3rd. ed. (Gottingen 1962); Dietmar Korzeniewski,
Griechische Metrik, Die Altertumswissenschaft (Darmstadt 1968); M. L. West,
Greek Metre (Oxford 1982); W. Sidney Allen, Accent and Rhythm: Prosodic
Features of Latin and Greek: A Study in Theory and Reconstruction, Cambridge
Studies in Linguistics, 12 (Cambridge 1973). Two works of A. M. Dale are further
distinguished by their initials: The Lyric Meters of Greek Drama, 2d ed.
(Cambridge 1968) and 'The Metrical Units of Greek Lyric Verse, I, II, III,"
Collected Papers (Cambridge 1969), pp. 41-97 (= CQ 49 (1950), 138-48; n.s. 1
(1951), 20-30, 119-29). Archilochus is cited from West's edition; Sappho and
Alcaeus from Voigt's (same numbers as LP); Pindar from Snell's. Other lyric po-
ets are quoted from Page, PMG or SLG; references to Page look to his texts or
critical notes at the fragment cited.

1. I will look at the consequences of my proposal for our un- derstandingof the history of Greek metric. I believe that Porter's method makes possible a simplified presentationof the differentmeters which also clarifies in- dividual traits. English and German scholars. 70. Dale adopts the concept. 95. Originally some of their differences-espe- . and. 4 The "New Metric" is a useful term for a style of analysis common to a group of. necessarily makes some conclusions tentative.) Ancient rhythmicaltheory concerneditself with the ratios of the actual durationof the arsis and thesis in each type of foot. Porter's method of notation as a means of describing the various forms that alternatingrhythmcan take (partIII). most recently. Much of the currentdisregardof alternationprobablyhas its origin in the response to this misreading. 145-47. so I have omitted statistics from this study (for an instructive example of such mis- leading figures.64 Joel B. West. it rejects some assumptionsof Greek metricaltheorythat Classicists generallyac- cept.4 2 I will not deal here with the complexities of fifth-century choral lyric. and concentratedon isolating the metrical segments of which a line of verse was comprised. for each of these is subdividedinto an arsis and a thesis. and a verse can be taken as a stream of alternatingarses and theses.) A theory of alternationis latent in the ancient metricians' method of de- scribing verse in termsof feet.3 (Unfortunately. I will illustratethe potential of this system with a rhythmicalanalysis of two poems by Simonides (partV).g. especially. Dale. Snell. the figures Page gives for the Sapphic stanza in Sappho and Alcaeus [below. and puts exceptional and ambiguous uses into a context (part IV). to be a factor in the analysis of Greekverses. MUGV 49-50. for the most part. N. No one suggested that study of rhythm could addressanythingexcept the problemof relative duration. 120-21 C. Wilamowitz. and. neither ancient nor modem scholarship is consistent about which portion to call by each term. A system of analysis based on alternationaccommodates descriptions of historicalchange and it is well-adaptedto the characterizationand comparisonof styles and to the close study of individualpoems. p. See scholia A to Hephaestion. and rhythmceased. especially. I also exclude Corinna. I summarize my reasons for also rejecting certain assumptions now currentin the discipline of linguistics. Blass. one should examine. and most popular song. The "musical"school of the nineteenthcenturyinferredfrom the ancientdiscus- sions that the feet were temporallyequal measureswith an initial beat. Maas. What I propose here revises the answers usually given to the question:what are the primarystructuralfeatures that distinguish Greek verse?-that is to say. Given the state of the evidence.The "New Metric"restoredthe ancientseparationof rhythmand me- ter as fields of study. according to which units that have the same arses and theses. e. but in different orders. LMGD 41 n. 110-11. Lidov basis for a review of all archaicpoetry and some early classical verse-formsin orderto show the pervasivenessof alternation(partII). and try to verify. of course. 3 One manifestation of alternation is in the doctrine of epiploke. of successive syllables within a segment. In concluding.in performance. The fragmentary nature of much of the archaic tradition. note 14] 324).2Then I will re-introduce H. it would be misleading to try to give a count of the ac- tual number of instances of features which are relatively more or less common. originate from a continuous series in which they are overlaid or "intertwined" so that the beginning of one coincides with the end of another.. (In an appendix.

He calls such continuous patterns"epiploke.Mass. II (Milan1973). Loeb Classical Monographs(Cambridge. p. see Massimo Lenchantinde Gubematis and Gianfranco Fabiano.but he views such a rhythmas cially in the case of Maas-were more importantthan their commonstyle.-.g.Taking his cue from reports of Heliodorus'stheories. so far as possible. 38).zwischen ihnen.Cole sees the poetic structurebased on epiploke as somethingthat evolved in the archaicperiod. If we look at the two most recent handbookson Greek meter..where he is adding several paragraphsof supplementalremarksto his presentationof the metricalelements.Cole arguesthat rhythmsubsists in the cyclical recurrence of fixed patternsof long and short(or anceps) syllables. "in einem gleichmaissig abwechselnden Rhythmusfolgen nicht zwei Stfibeaufeinander".or x . 43 of his handbook. Thomas Cole has now taken a somewhat different approachin Epiploke: Rhythmical Continuity and Poetic Structure in Greek Lyric. longum. or between which occur" the other elements (sect.Snell never men- tions the word. 53. pages 401-32. For Wilamowitz. Within the theories that posit segments as the primary metrical phe- nomenon. AlternatingRhythmin ArchaicGreekPoetry 65 Thus. the place of alternation was explicit but historically delimited: it had characterizedthe structure of the basic verse segments from which cola evolved. Bruno Snell ignores alternationcompletely. This introductoryremark closely resembles the definition by Maas which I paraphrasedin the first sen- tence. he is merely describingone class of metra in the statement. 1988 [I obtainedthis book too late to incorpo- rate it into the main body of the discussion]). in fact. wenn zwei. we see that Dietmar KorzeniewskiretainsWilamowitz's emphasis on "thesis-counting"but does not adhere to his historicalmodel. of the notions associated with them. Ostensibly.sozusagen. nattirlichkurzeoder (am Schlusse) fur kurz geltende. spoken tragic verse is three repetitionsof the sequenceanceps. Dale frequently considers the possible rhythmic distinctions amongmetricaltypes. particularly those which show a fixed numberof Hebungen(the regularlyrecurrentlong syl- lables) in a variety of manifestations."Problemie orientamentidi metricaGreco-Latina" in Introduzioneallo studio della CulturaClassica. For a fuller and more precise differentiationof modem ap- proaches to metrics. . as a way to account for the persistence of alternation-or the regularrecurrenceof the elementum longum-without recourse to the arsis/thesis distinction and without compro- mising the descriptionby units. They differ considerablyin the extent to which rhythmis important. vermutlich konnte auch einmal eine im innern fehlen" (Griechische Verskunst[Berlin 1921 and repr. 6-7. longum. oft auch vor and auch hinter ihnen stehen ein bis zwei Silben. His theory of internalresponsion can be seen.6It seems to me that unacknowledged alternation lurks behind Maas's observation that "most metra consist of two longa before. die wir als Hebungenbezeichnen.it also makes room for a system of describing verses as series of basic sequences (often called cola and metra) composed of elements in a fixed order:e.' but to keep our minds clear. after. 5 Pp.breve. and primarilydirects his attentionto explaining fifth-century forms. Maas declared in his "Introduction"that it is "necessarynot only to avoid using the terms 'arsis' and 'thesis.eines Verses bildet eine Anzahl von Langen. alternationhas had varying roles to play. 6 "Das Knochengeriist.] 88)."5This proscription serves to clear the air of the confusions associated with the terms. but Maas's definitiondoes not occur until Sect."Individualverses can be understoodas lengths of single or mixed epiploke which start and end at altemativepoints of demarcationwithin the pattern.

and treats it as a norm.7M. ("Thesis-counting" is the best English term available for giving attention to the relative numbers of what Wilamowitz called Ilebungen and Korzeniewski usually calls Stdbe or (feste) Longa. note 13 and Allen 96-99. West. She proposed that in one style of choral lyric.. with and with- out intermediateanceps. The applicationof the rules determines"the basic rhythmicalmovements"and ". This theory takes account of the two primaryprosodic contrastsin Greek metric-the contrastbetween single and double short sylla- bles and the contrast between short and long syllables-and makes them the basis of verse-structure. ).. to a much greater extent than it influences the German writers. Lidov natural. and for explaining the rule against anceps succeeding anceps or breve (6-8.But Dale achieved this simplicity at a cost: her method complicates the descriptionsof the larger segments and makes the anceps into an anomalousstructuralelement. Dale providedthe most significantdeparturefrom the kinds of analy- sis I have been discussing. Each princeps must have a short adjacent to it..a rhythmwhich often continues smoothly from one pe- riod into the next. 10 Dale. colometric. on the other hand. and original." and "2. it seems to me.Dale's theory subsumes the phenomenawhich alterna- tion describes without sacrificing a rigorousadherenceto analysis by units (and without appealing to performance). and ionic. and appear indifferent to whether that pattern reflects an aural or an abstract scheme. Heft 2 (April 1973) 129). for the analysis of anaclasis. She introduces the further disadvantage of requiring three dif- ferent types of verse construction: periodic. MUGLV. 8 The rules are: "1. This belief affects her expositions. bacchiac. His insistence on "thesis-counting" as a basis for analyzing the inter-relationships of cola can also be seen as a reflection of the unstated primacy of alternation in his theoretical presentation. but it will be obvious that I am in debt to her magisterial essay. the only effective units were the cretic (.Instead. neither adja- cent to one another nor separated by more than two other positions" form a recognizable pattern. Kannicht to point out that the theory of a fixed count is an unhelpful genetic interpretationof the fact that "das unterschiedliche Altemieren von obligaten longa und ein oder zwei brevia [ist] eine Grundtatsache des griechischen Versbaus" (Review. he attributes to her a "militantly durative-quan- titative attitude to Greek verse rhythm").-) and the choriamb (. L. both here and in Lyric Metres.spaced.10 7 Specifically."9 A. who are much more concerned with the graphic patterns formed by their notation. The differences between her approach and the one I will pursue make it impossi- ble to note particular points of influence. not as the recogni- tion of the presence or absence of prosodically relevant distinctive feature(s) (see below. and which she herself had pursued in her book on dramaticlyric. M. and stichic. 9 P. it provides him a norm for describing the character of the doch- miac. West's discussion makes clear that he understands such rhythm as a performance feature. at least.. 22. No short syllable might be adjacent to a long syllable not occupying a long posi- tion. 8 it is simply "der gleichmasssige Rhythmus"). on p.."he formulatestwo "rules of con- trast"which operate to insure that "fixed long positions.8He calls these positions principes. This leads R. It is charac- teristic of Dale's discussions (as of West's)." 18-19.in most metresgives a clear sense of rhythm. Gnomon 45. He relates it to delivery in procession or with dancing.) Korzeniewski himself actually seems to take alternation as a secondary phenomenon to thesis-counting. that she regards the perception of long and short as the discernment of actual time measures (rhythm is a question of the ratios between them).simple.. at the beginning of his discussion of "rhythm. does not explicitly acknowledge the existence of alternation. .66 Joel B.

First. 12 See his "Elements of Versification" in W. In answer to this a more thorough and subtle analysis of "podic" meters is presented by Devine and Stephens in "The Abstractness of Metrical Patterns: Generative Metrics and Explicit Traditional Metrics. Hateful Contraries [Lexington. New York 1962). however. K. Halle. K. Versification: Major Language Types (New York 1972) 1-21. Allen 104-5. Janua Linguarum.form and structure form the domain of meter (the other two are composition and performance). but as the distinguishing characteristicof the abstractscheme which all verses in Greekrealize. A." chs. Beardsley. Warren. it treats alternationnot as a featureof perceptible rhythm in performance(which it may often also be). Beardsley. and. Sebeok. 108-45) and E. K. The ap- proach taken here. in which regulated language begins and ends. defines four different levels necessary for the study of poetic language. ed. informally. See for example R. Devine and L. "Rhythm and 'Exercises in Abstraction'. 2nd. Fundamentals of Language. Zirmunskij. Theory of Literature (3rd ed. Wimsatt and M. by itself.. K. 1966) 135-48. Janua Linguarum.11In applying these methods of analysis to Greek verse.and will consider the significance of 11 See W. These studies stress the impor- tance of the normativefunctionof the audience's rhythmicalideas-meter is as much in the ear of the audience as in the voice of the poet. also differs significantly from those with a linguistic orientation. I adopt one additionalessential procedure from a theoreticalmodel for all verse thatJohn Lotz has elaborated. present a useful description of "two basic techniques" of analyzing pattems: "structural" and "transformational-generative. C. and M.. Series Minor. 1960. Jakobson. There are useful analogies to the procedure followed here in James McAuley's practical guide to the scansion of English. The argument has been influenced by linguistic studies. it makes the observationof alternationprimaryin the analysis of Greek verse. see the appendix." Poetics 16 (1975) 411-30. by the studies of the phoneme in the work of Roman Jakobson and others of the Prague Linguistic Circle. or frames. where it has been derived from Russian Formalism. that method is presented more fully in V. Wimsatt. pp. AlternatingRhythmin ArchaicGreekPoetry 67 The analysis of Greek verse as alternatingrhythmthat I will now propose differs from previous hypothesesabout the role of alternationin Greekpoetry in two ways. for a unit of composition which is. Style in Language (Cambridge. A simple distinction between abstractelements and the syllables that realize them is fa- miliar in classical studies from the work of Maas. Such a distinction in levels has been developed more extensively in certainstudies of English meter. 207-8. see pages 170-72 in R. A. Series Major. 1965]. . and "Metric Typology" in T." Halle demonstrates the potential of a generative analysis of levels for understand- ing rhythmic variation in "On Meter and Prosody. Versification: A Short Introduction (East Lansing. Ky. not necessarily metrical. On the background of this approach in Russian Formalism. Introduction to Metrics (The Hague 1966). 1966). Wellek and A. C.but I do take for grantedthe existence of the period (or. line) as a unit of organization.." in Bierwisch and Heidolph. note 13). Progress in Linguistics.I sepa- rate the discussion of the regulationof prosodic materialfrom the discussion of the units. W." PMLA 77 (1962) 668-74. in the style of the New Metric.. Schwartz." PMLA 74 (1959) 585-98 (reprinted in W. both necessary and sufficient for the creationof verse structure. 1 (The Hague 1971). 1-3. Part I. "Phonology and Phonetics. ed.. "The Concept of Meter: An Exercise in Abstraction. that teach the value of analyzing the components of a language system into a set of binary con- trasts and of distinguishing separate stages or levels in the complex process that extends from the production of sound to its comprehension-in particular. Mich.12Insteadof searching. My focus here is on the former. Mass. eds. second. ed. Stephens (below. Wimsatt and M. Wimsatt. 43 (The Hague 1970) 64-80 (he refers mainly to English but takes note of Greek).

two short syllables. 61. or. its prosodic equivalent.(The first distinction 13 The impossibility of a "middle quantity" is demonstrated by A.the long syllable expressing the elementumanceps is assumed not to differ from an unresolved elementum longum. At the other extreme. the distinctionbetween iambic and trochaicor between tetrameterand hexameter is not relevant to the problem of alternation. "Anceps" GRBS 16 (1975) 197-215. I assume that the opposition of M and S as a normative rhythmical idea grows out of the opposition of long and short. because the analysis of continuous alternationis without regardfor beginnings and ends. At one extremeare dramaticanapaestictetrameters. but the contrasting posi- tions are recognizable by broadinduction.13 Second.in which thereis extensive variation between two elements at both positions. and the one where there is a double-breveis the M. if at all. in Rome. and the M follows from our perceptionof other. this description assumes two and only two prosodically significant quantities. in Greece. I will call one the positio stabilior = S and the other the posi- tio mutabilior = M. In tragic trimeterand heroic hexameterone can easily see which is the more regularposition.) . Devine and Laurence D. Stephens. This valuable ar- ticle addresses the confusions in theories that base verse rhythm on the actual du- ration of syllables. one position is expressed regularlyby only one prosodic element.and "meter"for the variousrhythmicframesthatoccur in Greek (cf. thatgroup of elements representsthe S positions.in particular. because if we varied the element at the M we would still have recognizably alternating verse (thoughof a differenttype).68 Joel B. but the one in which there is a longum is the S. it would be unhelpful to describe the most abstract level with terms that originate in per- formance features: the raising of the foot. That is. we cannot identify alternationwithoutthat syllable. Thus we can readily distinguish different types of verse in alternating rhythmby the extent to which the element in the S position is varied. but if we replacedthe longa we would not. and the appendix. (It would be better to use the terms "heavy" and "light" for syllable quantity as recommended by Allen 55. although it would have been possible to redefine "arsis" and "thesis" as names for the two normative positions. In many metrical patterns there occurs an alternationbetween a position which is filled with a more regularexpression and one filled by a more varied one.one set will include some elements thatare regularlylong in all the responding lines. or of the voice. I think it is useful to keep the term "rhythm"for the phe- nomena of regulation. note 11. See above. long and short. Within some types of verse the S and the M are almost equally regular or varied.In practicethe S is induced from the persistentrecurrenceof a prosodicallylong syllable. on the other hand. the use of "dactylic" and "dactylic hexameter" in the traditional nomenclature). perceived as the elementum longum. interveningelements (including one expressed by anotherlong syllable). but I have kept the more fa- miliar terminology. the otherby variouselements: the two positions can be identifiedin comparison with each other. Alternating rhythm is protean. Lidov beginnings and ends. Because. it is the per- sistence of an elementum longum formed by a long syllable that marks the S. M. and by the way in which the M is expressed. lyric dactyls one could call either position the regular one. as an occasional variant. But when the elements are perceived accordingto a patternof alternation. First. ultimately. within a series of pure. Two warnings are in order.

The variationpresentedby catalectic trochaics(and iambics and anapests. For by whatever name we call a segment such as: EpaooviSrM 'Epa(spovis6 XapXa Xopitka x x S M S M S M it is alternating. quoted at the beginning. the fragments of the epic cycle. at a "pause")a break in the alternationshould not be construed as an abnormalityof special interest. the Hymns.it must be able to identify nor- mal variations. or representthe scheme in its every detail. But a theory which postu- lates the existence of an abstractscheme. the early iambus (including cho- liambus).4ov xiv xstvv X. -X .the examples with which I be- gan. especially. ?pao. and. since an S follows an S directly across the line end. The example of elegiac verse.smaller class of verses. the trochaictetrametercatalectic. A third.in particular. in the fifth century. analyses by cola. Hesiod.I have already mentioned the line of the Sapphic stanza (there is also no breakbefore or after the adonic). In lyric verse.points up an importantconsideration (Archilochus 118): iI %J.) Dactylic hexameters and iambic trimeters. In these two types alternationis directly expressed in the prosody and continues without interrup- tion from the end of one line to the beginning of the next. and anapaestic series whetheror not they end in a catalecticdimeter. by mass. make up. It is neithernecessary nor expected thata specific verse design which realizes an abstract scheme express every element or characteristic of the scheme. the trimetersof drama. metra or sequences distract us from the commonness of alternation. so common in comedy) can readily be describedby a general rule: the regularityof alternationin Greek meter is valid only within a period. p t% v N. _- . Between periods (that is. not only a pause. prompts us to broadenthis principle.tYevo0to X?Ipa Noo _ _ _ _ _ _ 9ty?iv Successive verses would not express the alternatingpatternbetween themselves.We can also add the epodic constructions of Archilochus (which completes his corpus). So alternatingrhythmincludes trochaic and elegiac verse.but between the halves of the pen- tameter: SM S M S /S M S M S/ The more general rule is that a patternwhich associates an expected word-end. . and describes verse in termsof it. AlternatingRhythmin ArchaicGreekPoetry 69 can be drawn simply by stating whether the verse begins in the M or the S position. can be of use only if it can also categorize and describe those forms in which the scheme is only partly represented. most surviving early Greekpoetry:Homer. Alternationis brokennot only at the period-endbetween the pentameterand a following hexameter. s SMSM SM SM S SMSM S Ei yap x. Alternationis continuouswithin the Alcaic stanza(Alcaeus 208): d(rovv&rCpJt ? aL &TvE. This discontinuity illustratesan essential featureof the kind of theory I am adopting. with a break in the alternation can be counted among the normal varieties of expression of alternatingrhythm.

A 4-7.V word-end allows ..though the exceptions are not unimpor- tant. pherecratic. k. 6.. TO y&apaozpov xcepteerzat (S) S M S S M S MS In Horace's imitations of these forms word-end is the rule between the contiguous S positions. 1971) 15-26 (with frequent word-ends marked and using conservative texts).But afterthe base the typical movementsof glyconic.. Voigt shows that the major asclepiad (gl2C)in both poets. C 3 c-d). Lidov To uev yYap?vOev Kiia KUivEat M S MSM S M S M S X0 6' i?VOcV.1-2 = 9-10 Sn. for example.. the relation of such lines to the asclepiadic type is unclear. -.x (but not between stanzas). X vai (poprO.Ea n)iv ?Xaiva 1 S_ M __ S SM _ _M_ -SM . At the startof the line the two indifferentsyllables of the aeolic base may break the rhythm. usually occur in Lesbian verse in the form of an anaclasticdimeter.--x/. to be interpreted as MSM/M. The most importantexception is asclepiadic move- ment. C 3 h. now often called "choriambic"expansion of the aeolic rhythms. theanacreontic (.. Ionics aside. i. Aeolic verse generally alternates. though they are not nearly so common as verses with unbroken alternation.e. or together with it.70 Joel B. which preservesit. Denys Page.. -.. . no examples in Alcaeus).And the anacreontic itself is an alternating form.x. Sappho and Alcaeus: An Introduction to the Study of Ancient Lesbian Poetry (Oxford 1959). -. .and forms like the asclepiad constitute an exceptional class.and hipponacteanare alternating: SM SMS xx- S M SM X X .Unlike "dactylic"expansion.14Ionics. When they are combined... the monodies of Anacreon belong among the types of alternatingrhythmdescribedso far. is muchpreferred..-). but the practiceof the Lesbianpoets is less regular. 102 LP. In Sa...and that forms with any other kind of breakare quite rare. and the minor asclepiad (glc) in Alcaeus (barely exampled in Sappho) usually have word-end at the break (her "Conspectus" numbers Sa. though here-as Mass notes in his paragraphon alterna- tion-the form . Ale.. The text of 132 LP.- SM SMSM (the two longa forming the "catalectic"or "pendant"ending of the second two will be discussed below). . . Sappho et Alcaeus: Fragmenta (Amsterdam 1963. in fact. 318-26 (with more analytical discussion and more regularized texts). but it may be an instance of the kind of break in iambo-trochaic discussed below. this involves a breakin the alternation(Alcaeus 352): nIlvol?v. where the same line also occurs. -.. 14 See E. a preferable alternative to the analysis MSMM will be discussed below.-M.. its hipponactean congener in Sappho (x . -. -.V is uncertain. Voigt (Hamm). A glance through a synopsis of Sappho's and Alcaeus' meters will show that these forms are not uncommon.&ip?tL 6 OV X0 r?oOV M SM SM - SM X X-.as in Bacch..

there are sixteen instances of MS S M within the lines (using Snell's edition) against 81 between lines (even without including the instancesbetween the last line of a stanza and the first of the one which follows it.g. supplemented by "The Versification of the New Stesichorus (P. In Haslam's discus- sion. The dactylo- epitrite has an alternating movement (Pindar. SMSM S M S M S MSMS Within a stanza most interruptionsin the alternationfall between lines." 56.. 0. But we can add to the list the odes of Pindarand Bacchylides in this meter. when the lattermay be an antistropheor an epode).Maas. Haslam. the instance of rhythmic break between periods is much higher in the epodes of this sample (following 83% of the non-final periods) than it is in the strophes (following 52%)." QUCC 17 (1974) 7-57. exdxe in Maas's notation): M&aTp & %pVooTc?pdvtov e. for the most part. and derives the dactylo-epitrite from "the two types of stabilised Greek rhythm.. assigns the work of generalizationto 15 Note how in 16 P what appears to be a change from iambic to trochaic movement between lines 3 and 4 (MS/SM) is obscured by a tribrach as well as marked by word-end. long/double-short alternation (dactylo- anapaestic) and long/single-short alternation (iambo-trochaic)" ("Versification. is similarly consistent and enters into our account without need for detailed comment. ". Taking the Olympianand Pythiandactylo-epitriticodes as a sample.. 1/282a P is notable. my emphasis). Despite some patent exceptions.1. The stanza of the Geryoneis (8/185 P = S17) shows further examples of the type that involves apparent successive M posi- tions. see Michael W.Metre" 17. Alc.I refer to not just the lyric dactyls and iambs. the poetry of Alcman that we have is overwhelmingly alternating.. . AlternatingRhythmin ArchaicGreekPoetry 71 Among choral lyrics one large class can easily be deal with. l1 & 6[v S M SM SM SM q Xploov ifpvoq q a&7ia6[v gi]ov M S M S M S MM SS Stesichorus' practice.16The same is true for Ibycus./ 2 tro Agl ia / ." GRBS 19 (1978) 29-57.." for str. "Stesichorean Metre. for example.. It is worth noting in this regardthat where period- end is definitively established (by hiatus or brevis-in-longo). There a variety are of ways to accomplish the task of describingindividual There are a variety of ways to accomplish the task of describingindividual lines of Greek poetry in terms of a verse design thatcan be seen as a realization of a general pattern.Lille 76abc). Intralinearbreaks in rhythmwill be discussed below. in accordance with the general principle that they be accompaniedby obligatory word-end. though the asclepiadic movement of the last line of the epode of Ibyc. 8.0kwv. 4-5 of 3 P should not obscure the un- brokenalternationin and between these lines: oXpavc 5xlov?M pvo. He re- marks on the occurrences of "break[s] in metrical sequence" which occur at period junctures (e. 16 For a survey.. 1 P and 3 P do not break alternationeven over line end... 26). as does one line-end in the epode of the Lille poem. alternation appears to be a fundamental feature of these verse forms.15An analysis such as Snell's "..

breve).and totally to rely on them here would be additionallycon- fusing.e. it is markedC (cf.. is dependent on the pattern that the B and D elements establish.There are only four possibilities..72 Joel B. But neitherthe traditionalvocabulary of scansion nor Maas's elements (which do not imply units of length) are par- ticularlywell adaptedto referringboth to ever more generalabstractschemes and to the verse instance. note 11. and if it is not present at all. and I will employ it here in orderto examine more closely the types of alternatingverse and the con- texts in which breaksin alternationoccur....). common)..Its use can be called a kind of normal variation.S).. the position is markedA (i. Lidov the six elements and the basic sequences. 24 (Leiden 1972). it would eventually defeat the expectation of alternation.. of D alone.... Porter's system has previously been published-although explained in terms quite different from the ones used here-by Julia Loomis.DDD. anceps.I designate it C. = . 13). Allen (110) introducesthe term"tension"to describethe relationof actualstruc- turesto an ideal one (whichmay rarelyor neverbe actuallycomposed). N. if by two short syllables. The advan- tage of this system is that it allows us to describe the prosody in terms of a normative rhythmic pattern at the most elementary level: it shows how the membersof the audience can assimilate what they hear (the longs and shorts of scansion) to the patternsthat their knowledge of verse-form (as a part of their linguistic culture)leads them to expect. I assume that it is an apparentlongum createdby the prosodicphenomenonof contractionand retainthe D designation... Porter's de- scriptionof the elements of rhythmcomplementshis study of normativeframes in "TheEarlyGreekHexameter. It will be useful to refer to these two types as isosyllabic and isochronic alternation. absent).. on the other hand.Use of B by itself produces an alternationbetween elements which have one syllable (. Porter simply assumes. To distinguishverses or otherunits. Studies in Catullan Verse: An Analysis of WordTypes and Patterns in the Polymetra.see above.not between the units in a verse. When it appearsin responsion with a single-short."YCS12 (1951) 3-63... and at times able to overcomethe slight reluctanceof an occasionalphrase to take the requiredshape-so long as the reluctanceis not too great"(above..BBB .- _.. open (M. The long syllable which often appearsas the expression of an element in the M position. without notation. the recurrentelementa longa in the S position. A method of notationdeveloped by H. I also use C whenever the natureof a prosodically long syllable expressing the element in 17 Cf. These are writtendown in order of occurrence. we can take from Porter's system the terms rising (M.. Porterworks more efficiently. If the M position is representedby a single short syllable.S). because they reuse for a different purpose the terminology and (especially) symbols thatdescribe the longs and shorts of prosody. When a long syllable appearsin situationswhere responsiondefinitely indicates a patternthat allows two short syllables. .. it is designated B (cf...as long as we keep in mind that the terms refer to the relationbetween the elements of alternation.). Mnemosyne Supp.. note 13).able to resolve ambiguoussyllables in its favour. it goes on in our mind as a patternof expectation. = . it is designatedD (cf.._. double-breve).17 Since the B and D elements. Without them.. on the basis of whetherthey begin or end with the M or S position. alone or in combination.If it can be repre- sented by either a long or short syllable. falling (S... 2-4 (Loomis retains "arsis"and "thesis". between elements which each have two morae (.M).. and closed (S.. -.they claim a certainpriority. he identifies only the elements of the M position. can create a perceptiblealternation.M). McAuley:"Oncethe metricalpatternis clear.

19C is frequently found before. and Ibycus 36/317a. _ _ _ pa 8icxTuaKxpipiSo oB? D D D? 5 6 D? D D D tpog&o V ~ gLav iTpotc vtv pXO. The rhythm of the first line of the Sapphic stanza is BCDBC. note 16. But contraction and the C ele- ment are not. see the example in the previous note.. for example.BCBCBC. 35-39.8 P (and 3 responding lines). p.. It is thus not a fea- ture of the verse system.. I prefer West's analysis: he includes this word-final syllable at the end of line 3.2 P where I suspect. for example. where there are long syllables in the M position. yet here it looks like an aeolic basis that prepares the change of rhythmical context before the asclepiadic ending.. Stesichorus S148 i. This last group includes those cases in which there is no instance of a true short syllable among the responsions (if any). BCD. but in which the verse design also appearsto exclude a double-short. of course. making it brevis-in-longo. but in practice it is not to be distinguished from CD (if it is to be brought into the system at all. immediately before a pause.8 also show BD. When the aeolic base is accommodated to continuous alternation. after.. But we could obtain some partial re- sponsion of line 4 with other lines by calling this element and the initial ones of lines 3 and 5. to be found in the same context. these represent strophe 1 and 3 in his reconstruction of the metrical scheme of the Eriphyle). 53. dactylo-epitriteshows.. Haslam dis- cusses the problem this presents in light of Stesichorus's practice generally (above. consistent use of . page 76). often make it unclear whether one is looking at BD or CD..6 and 7 (supplements) and ii. but can hardly prove. (But see below.. BD seems certain in Ibycus 1/282.DDDD.. as a general rule.. 19 Uncertainties of responsion. but what is especially distinctive in aeolic and sets it apartfrom other types is its use of the sequence DB = . Some generalizationsabout early Greek verse now emerge quite readily. ?... -. .. but a feature of the language itself-whether a phenomenon of the phonological determinants of prosodic features (perhaps a consequence of length as a redundant feature of syllable quantity?) or a generalization of vowel contrac- tion in the evolution of poetic language (see Allen 255-59). aJixet. the design appears to be a consistent D rhythm. For example.?vov All other M positions in the poem have either a double-short or a long.18 In this notation iambo-trochiac rhythm is the series . is uncommon in all archaic verse. see below. The sequence B D. dactylo-anapaesticis . of the Alcaic CBCDB. . the exceptions 18 Maas's biceps is not a true parallel to anceps because contraction appears to be not a metrical element (that is. this changes the verse design. on C and contraction.. in Page's edition of Ibycus 6/287 the initial position of line 4 has a short syllable = B: D? D D D 3 oavt rXiLaaGoi rooaoio. DDCBC BCDD and DDCDD. AlternatingRhythmin ArchaicGreekPoetry 73 the M position is obscure. on the other hand. or when there is no evidence from external responsion to tell us whether the apparent longum is a contractionor an anceps.. even though the verse system regulates its appearance. A certain arbitrariness seems inevitable in the use of C. C.. an interpretationof prosodic material for met- rical purposes). or between two (especially) or more D's. The series BCD appears within many aeolic forms.could produce the impression of BD...) "Pendant"endings will be discussed below.

-.20 IV I discussed earlierthe verse types in which thereis a breakin rhythmiccon- tinuity...as they occur. Lines in which an element is missing can present some ambiguity in their analysis. Within the line.For analyticalpurposes. as in Alcaeus 352 (quotedabove.. --.If the sequence of two long syllables is clausular(as it so often is: "pendantclose") and the next period startswith the M.. however. as either BA and DA or as BC and DC. they are more unusual..and .e..x -. West 48-49 on the occurrenceof "resolvableanceps.. (in Maas's notation:x e e or die). p. . and described these breaksas a form of normal variationin their com- monest position: between periods and at other points at which word-end is expected. 24]. then the final long syllable will certainly be assimilated as a C element in the M position in continuous alternation. especially. Lidov tend to be at the beginning or end of well-established frames (e. 3)... This featureand the occurrenceof BD (see pre- vious note) together suggest to me that within dactylo-anapaesticsequences Stesichorusused bits of isochronicalternationmore as a kind of punctuationor dividing element than as a balancingelement in the mannerof dactylo-epitrites. there is no need ever to assume successive M positions.This variantstructurecan be analyzed as the absence of an expression of the M position in the verse design.the dactylic hexameter.. In the dactylo-epitrite. Within the line.x.. ---and .74 Joel B. the audience will expect the anceps which occurs normallyafterM = B or M = D in this rhythm. . They are not infrequentin later choral lyric..g.70). Alcaeus' as- clepiadwould be describedas (C)D ADB. that forbids anceps next to other elements in the M position. then.-- could be heardas . the exact natureof it need not be decided. but note especially str.)It may also be the case that Stesichorusexperimented with allowing the least possible rhythmicdefinitionat points where variationis allowed.where the final element is understoodas C because D is excluded).. The principlethat associates expected word-endswith interruptions in the rhythm will be considered applicable both when the two coincide and 20 Contractionand the C element appearto occur in very nearly the same con- text in the scheme for the strophe of Stesichorus'sIliou Persis (more so in Haslam's version [above. that is.. unprovablebut induced from wide observation. can be interpretedas MS S or MS M. If the next period begins with the S. they only can "correct"their first hearing as the patternproceeds.since a terminalsyllable can always be taken as long and as an expression of the S position. (Haslam's presentation implies an evolutionary development from dactylo- anapaest to dactylo-epitrite for which there is no evidence other than his arrangement of the poems.Thesequences .but the ambiguity here is quite real... In Porter's notation. or at the end of the catalectic iambic trimeter(or ithyphallic.although the transitionbetween periods will remaina momentof decreasedrhythmicaldefinition. for example. even over stanza- end (as in the case of the Alcaic stanza)." 21 See Maas 29-30.... if we allowed the possibility that . cf. ). in dactylo- - epitrites:. than in the one Page prefixes to S88- 132. Therefore. note 16. .in fact. But this sequence violates the rule. B A B or DA B) will always be preferredto M SM M (even though thatmay seem to over-definesome periodendings or some rhythmicallyobscure sequences).21 So the analysis BA (or D A) is preferable. -. then we would have to accept .the analysis M S S M (i.-) and of the pherecratic(.. the ambiguitycan be assimilated to the discontinuity of rhythmthat is associated with line-end..

not word-group ends (cf. though West 22.g. they produce a momentaryconfusion between the A and C elements as the form of the rhythmic variation at the juncture..Given the restricteduse of the A element in isosyllabic alternation. 6.. no contexts associate it regularly with dactylo-anapaesticrhythms. P.). . 2. LMGD 97. In some cases. feature of Pindar's dactylo-epitriticstanzas.. 7) word-end is preferred but not ob- ligatory. P. of A for C-produces the rhythmof the "cretic. Supp.22But before maturearchaic lyric the internalA element is uncommon. viz.as it were.The DAD se- quence-what I am calling the asclepiadic type-in aeolic verse is the only context which makes regularuse of it. . The "pendant"close . etc. (or . The characterizationis either subjective or circular. P.."The 22 Bruno Snell lists the places where the "anceps interpositum" is missing. for example. 23 Griechische Metrik 29. note the frequency of caesurae in Aesch. 2 str. if this peculiar effect is accepted. cited by Dale. West 25-26). 3/4 ep.)--/."bacchiac"addresses its distinguishingfeaturebetterthan"cretic. and for ar- chaic verse I see no reason why they should be so separated. I will risk a suggestion of a very different sort: that in some of these comic pas- sages we are being treated to a rhythmic tour de force in which three shorts func- tion as the element in the M position. etc. cf. Maas 84ff. Pindari Carmina cum Fragmentis. isolated instance of this variation.. by Alcman (p. LMGD 97-103.. 3 for a possible.24 If we need a name for syncopated isosyllabic rhythm. Cf. 42 C). It tends to occur toward the end of a stanza. assume that it did. 98. this move- ment elsewhere plays an importantrole only in comic lyric and usually with the furthervariationof resolution (producingthe "paeon"). His lists omit the points at which word-end also occcurs. In her examples. 1 ep.. 2.--.. AlternatingRhythmin ArchaicGreekPoetry 75 when.. vol. The cretic and paeonic-some- times classed as "irrational feet" or "meters in five-time" (Dale.. except in the case of the most common type. both hyporchemes to which Snell attributes "den Charaktervon leichten Tanzweisen"23). 4. but not a stanza. and Dale. fr.. Throughoutthe archaicperiod and later. It can cer- tainly be identified in contexts where there are monosyllabic elements express- ing the M. there are a certain number of lines in which there is a virtual bridge at this point. 3 str.and its tendency to distinctive con- texts. I. Note that Aris." In all but four instances of this there is a word-end after d1 (the single double-breve is itself anomalous in this rhythm.A completely regularomission of the element in the M posi- tion in a context of isosyllabic alternation-a substitution. but a caesura one syllable later. LMGD 3.appearsfre- quently as the "catalectic"variationof . 781ff. 1 ep. I. The absence of an element for the M position within the period is a regular. 418-22 = 423-27 (persistent use of such regular syn- copation obviously produces an effect quite different from the irregular use of it). for showing syncopated cretics in "a cretic-paeonic passage adulterated with trochaics. Their appearances in drama are discussed by Dale. word-endoccurs at this point.) In a few cases of the other types (e. But these manifestationsof it occur irregularly. in I."Hephaestioncites this as the rhythmof a line. but not frequent..So far as our evidence goes. Lys..-. "e-dle. for a stanza he cites Bacchylides (v. 16)-are notoriously hard to separate from syncopated iambo-trochaic. 24 Syncopation need not imply prolongation in performance. 1 str. II3 165-67. I would retain the designation "syncopated"for all the passages in which it occurs in archaic verse. by being placed a syllable apart. I am counting actual word-ends." could be read as alternating without a break up to its last word. 15 and 16 Sn. 1 str. 4. Pindar P. Furthermore. ().:-. no definition of the latter is sufficiently free of exceptions and inconsistencies to be useful. The end of the stanza in the dactylo-epitriteis a context in which iambo-trochaicrhythmsare expected.

37 C). e.3-4: T' ? 0 'D B A D Xwx(pro36Xov aypoTepav "Apu?y' "ApTep.suggest a wholly different explanation: that the alternatingelements of ionic are .-. 27 A corollary of West's rules of contrast leads him to state that "the fourth po- sition in -.. But the regularassociation of ionics with isosyllabic alternationsof M and S.with resolution. but it does yield. 886.25 Anothercontext in which the sequence. Such a description accords with the frequent use of isosyllabic alternation before the distinctive D element of aeolic (as in Sapphics and Alcaics). a yx)vaIKo:v v. should be anceps" (59)-something which does happen in tragedy (124).. 28 I suggest that this capacity of aeolic verse to keep place in the line at any moment by syllable-counting instead of alternation is what allows the rare re- .appearsin earlyarchaiclyric is the startof aeolic verse. .. and the peculiarcultic or foreign ethos that clings to so much of their use in the later tradition. The ionic has achieved greatest fame in the verse of Anacreon. the distinctionbetween 25 To look. bacchiac movement appears to have produced in tragedy the distinctive dochmiac. is similar. This would be a phenomenonparallelto that of bacchiac syncopation.is rarerthan . The anacreontichas the apparentmovement of aeolic (--.-. [-XEt X1 Kcpd'to.-) appearto be an alternationof D and A as the M element (i. for a moment. is the second OlympianOde.and . a colon frequently found in iambo-trochaic contexts. The ionic is the most peculiar of the early forms. Bacch.in the fourth line. and looks aheadto early fifth-centurylyric. in conjunction with its presumedana- clastic form. .. Ionics (. every other M element is omitted). 893-896. The aeolic base can be describedas a momentof rhythmic uncertaintyat the startof the line in which a perceptiblebacchiac movement is occasionally substitutedfor syllables thatcan be perceived as alternationwith a C element. in contrast.g. p. a peculiarity which ac- counts for the strangesound.28Indeed.-.--. and for their easy association with normalalterna- tion. 907) we find the sequence DBAD in both the third and fourthline. p.g. As aeolic base.. ---. 17 Sn. e.27 And the tendency for poems in aeolic rhythmsto be writtenin lines with a fixed numberof syllables. .iv. 70).26 In the scolia written to the metrical pattern of the "Harmodios"stanza (PMG 884-890. -. two shorts in the base create an altogether different effect for one line.in laterchoral lyric.76 Joel B. This is a more developed use of syncopated movement as partof the aeolic rhythm. -- DBBA). Lidov great monumentto irregularbacchiac movement. the regularly alternating anacreontic (above.word-end is never found.e. 26 Very rarely. which restricts the variationbetween mono- syllabic and disyllabic expressionof the same element. This illustrates the inadequacy of treating the rhythmical phenomena as the by-product of a restriction on metron or colon formation: it overly restricts rhythmical typology and history. beyond this survey. establishesa fixed frame thataccommodatessequences in ionic rhythm. ... Pure ionics are recorded for Alcman (46 P) and Alcaeus (see Heph. p D B A D B In seven of the eleven differentthirdlines thereis some kind of word-endat the rhythmicalbreak.

32In three of the fragments. I would suggest the same process is at work in these po- ems as in those in which Sapphoequatesa glyconic with a choriambicdimeter.29It is probablybest to reserve the label "ionic" for unambiguouscontexts. we can see new rhythmical variations incorporatingbreaksin alternation.5 P can be taken as BDBCB. Anacreon certainly goes furthestin experimentingwith what can be variouslycalled com- binations and substitutions of iambic and choriambic. Snell sees there his primaryexample of the re- sponsion of choriamband iamb (25).. his analysis suggests less regularresponsion than the text actually shows). We also find rhythmicalvari- ations in which alternationis not interrupted-fragments which have much of the appearanceof dactylo-epitrite.but contain clearly aeolic sequences. 1).West 58. however. AlternatingRhythmin ArchaicGreekPoetry 77 ionic movement and asclepiadic in Sappho may be too fine to draw. because of dovetailing (single-syllable overlap of colon end). sees a sequenceof anaclasticglyconics and an aristophaneum with choriambic expansion that. pre- scribes the use of the typical syncopatedform of a rhythmtowardsthe end of a stanza. but the only breaks in alternation in this fragmentare at the end of lines 2 and 6. Whereaeolics predominate. as an anaclasticglyconic by West (57). "In [11/]356 [P] Anacreonuses a straight ionic as the penultimateline of a six-line anaclastic strophe-period. or ionic anaclasis. with which I shall conclude this survey.. "has all the appearanceof ionic. 26/531.to give it definition.50/395 P has the same patternwith less textual uncertainty) he is supposingone possible structuralprinciple-one for which there are no an- tecedents. 30 When West says. which would also fall between phrases.one with a single extra choriamband an iambic metron(57-58). In 76/581 P the only break occurs in line 2. 32 In line 3 synizesis in Xp-t?a-. Page's note offers an analysis into choriambsand iambics. 1/346 P." (59. 31 There is also a break after line 7 if we retain6 in 8. In 74/579. Most of his surviving poetry is in forms alreadydiscussed. In 50/555 P the textualuncertaintiesare too great. or "choriambicexpan- sion" of glyconic..The opening sequenceof Anacreon41/386 P (72 D). we pass to the choral lyric of the late archaic and early classical periods.This responsionis groupedby Maas with other examplesof anaclasis(25-27). The initial sponsion of "choriambicdimeter"with a glyconic in Sappho.these show the way to some of the apparent rhythmic irregularity of later choral lyric. . 43/388 P (54 D) is listed by Maas among the types of anaclasis (25. Another principle of composition.a largerrole can be ascribedto asclepiadicrhythm. eliminates anotherbreak.4 P are easily removed by Wilamowitz's reading.31 59/564 P is similar.for which there are parallels." Anac.30 V In the fragmentsof Simonides. For ex- ample. Perhapsthe "straightionic" can be taken as a variantof the "straight" anacreontics.. otherwise they exemplify asclepiadic movement. Note that the sequence of elements in the M positions is the same in an anaclasticglyconic and in an anaclasticionic. For Anac.-.4 P what could be a familiar aeolic line (hipponactean) is made into something else by a syllable that breaks the rhythmbefore the D element: . Line 8 is alternating with or withoutK(XI. West takes it that the poet is using inter- changeablytwo "iambo-choriambic verses"derived from the anaclasticglyconic: one with double choriambicexpansion. my emphasis. 42/387 P (71 D) is analyzedas an iambicdimeterwith an initial choriambby Snell (33 n. between words (and separatephrases). 29 The description of these verses in terms of metrical units is especially thorny. The instances in 62/567.

exceptfor these points and the end of "lines" 1. the Danae frag- ment.. omission of "link"anceps). (with word-end preferredat the break). 3. .(The two types are more clearly distinguishedin the closing lines. in the next line we confront even more clearly an uneven transitionfrom monosyllabic M elements to an aeolic type of sequence:5 -----. Line-endis possiblebefore the double-short(so printed)or after the following long. but I suspect he was led to this remark by assuming that the poem's metri- cal structure must be found in cola within the periods. The fragmentseems to be alternating. Thus. but the text is not certain.It can also be seen as syncopationof the isosyllabic alternationthatprecedes the D element in some Lesbian verse (such as the Sapphic hendecasyllable)-a variationanalogousto syncopationin the two-fold movementof dactylo-epitrite (i...)34 Dionysius of HalicarnassuspreservedSimonides 38/543 P. or to the DBAD in the "Harmodios"stanza. The occasional C element interposedbefore a B both allows some longer stretchesof alternationand creates a clausularcadence in the last threelines./.78 Joel B.33 In Sim. His text presentsno definite line-ends..BDBABD. He is not more spe- cific. element sequences familiar in the traditionsof aeolic and of syncopated isosyllabic alternatingrhythm. This has some resemblance to the BAD infrequentin aeolic bases. 5/6 and 7/8-at 5/6 the series can be seen as either restarting or overlapping). The rest of the stanza is. but it is unlike any- thing else in the stanza. . which correspondto a pherecraticand an ithyphallic.._---. it also establishesthe me- ter reasonablywell... There is an undeniable rhythmic ambiguity here which is re- solved as Simonides continues. the second is more like what follows in that it recalls aeolic rhythm. unambiguously:CBDBABDBC B/4DBABDBABDB//6 DBABD/7C BABD //8 BABC/9 BDC/ 0 BB(C)///. the worst-case example. / The first is quite unexceptional as a rhythmic sequence.. The most remarkablefeatureis the frequentuse of broken rhythm in the recurrent series .e. it uses.. Yet it is possible to see in the rhythmof this fragmentsome regularitiesof construc- tion. 4. I will present an analysis of it as a final illustration of the theory and 33 There may be something of the same sort in the first line of Lasus 1/702 P. within and between lines and periods. the first line could be: D D C B (A) D D D B BA D . Note that there is no instance of DA or (unless in line 1) AD.for the study of metrics. but the poem is not composed of glyconics.. The first line begins with two successive D's.." of which the second and third are long enough to be printed(and numbered)as two.-)... for example.. 37/542 P we see the missing M element as a regular feature. The following sequence. more clearly. .. as a sample of metricalobscurity. and B elements with C or A. my emphasis). The sequenceBDB can be taken as a glyconic (. 34 Maas says of these stanzas that "Their internal structure reveals all the li- censes permitted in later lyric poetry" (50. or------------.. the only in- stance of paired D elements in the stanza. v allows morethanone interpretation. Responsion of stanzas establishes eight "lines...The scries of elementsmay overrunline-end(3/4.. using single D elements juxta- posed with B's. It represents. althoughhiatus does define occasional pauses (one period consists of only five syllables). .. Lidov "cretic" is not isolated by word-end.. and so not noticing the fairly simple rhythmic patterns.

even though the uncertainties of both structureand text make it fruitless to argue any individualpoint of the analysis in much detail.. although I understand them to be partof the isochronic series. In the absence of clarifying responsion. this makes the sequences of single-shortalternationquite unlike the "iambo-trochaic"that is so readily expected. in lines 3 and 27 less surprising.. 21 ___ _//22_ __ _ _ _ 23( KEXogat 6'\ ficPp. The absent element between the lines is the only certain instance of DAB.. Line 21 opens with DAD. the omission of an expression of the M position. CQ 29 (1935) 85ff. Syncopation.and . In what follows I will take advantageof whicheverreadings. my line numbers refer to the latter for convenience of reference. 17." Whether or not one goes along with West's reluctance to treat aeolic as a metrical style unrelated to geog- raphy.. AlternatingRhythmin ArchaicGreekPoetry 79 proceduresoffered in this paper. createthe fewest difficulties. if we accept CTexr-(the manuscriptsand the unlikeliness of BD leave it doubtful). 36 See West 51-52 on Ibycus's "asymmetrical cola.BCD.at the ends of lines 24 and 6.B D...37The . is very notable. especially.. where it seems to be part of the closing off of one type of rhythmicsegment. often with word-end (hiatus in 24/25.among thosePage offers. of later aeolic is found once at what is probablya major transition.. below). and so brevis-in-longo.. this makes the supposed responsion of -. which is mostly used in phrasesthat denote the infant's sleep.. and after B. In addition.." a type which she dis- cusses in a chapter on aeolic cola (157ff. lists such sequences under "prosodiac.CB... It preservesthe clarity thatcomes of not mixing contrac- tion and anceps in the same rhythm (. Dale. the syntactic separation. It varies the series and anticipatesthe clearly aeolic segment (with the god's name) that 35 Page discussesthe text morefully in JHS 71 (1951) 133ff. does occur two or three times).. and of the kind of discussion they encourage. occurs in the segment DAD..3i The fragmentinvolves . I think the kindredness of all such sequences can be assumed in Simonides.). LMGD 217. I will label all such syllables here as occurrences of the C element.. The absence of . and the probable pause: D A D A B B B B B A B D . lineation differ there from the version in PMG.. . Ec E'U S&o) 5 n6vTO. line 22 is an extended isosyllabic series ending with syncopation (cf.D DB sequences such as one finds in Ibycus36but blends them with longer segments of isochronic alternationthat include appar- ent contractions-that is. such an anomaly is not very interesting given the sharp distinction in rhythm.). Hiatus guaranteespause..thereare sequencesof alterna- tion with only single-short. the absence of clear anceps in the segments of single- short alternation makes it more likely that the C elements which appear sur- rounded by D elements are assimilated as contractions. His text and. 6/7). long syllables in M positions that are not distinc- tively initial or clausular. I will assume that lines 21-25 + 2 (= 26) begin the strophic system. EurcTo 0' a(jrixpov caKO6v Line 23 begins a new isosyllabic series with a D element.. 37 Put the other way. There are also shortersegments thatare more distinctively aeolic: ... and otherwise only at textually uncertainpoints.(C)DB. (see also Davison. I take the ultima of line 21 to be brevis-in-longo at pause.

After a break (BAD). )//7 CD D B 8( ?lt?cv T' * Q T?Ko otov EXO X6VOVv Lines 8-10 provide two examples of the sleeping infant motif in an ex- tendedisochronicsequence:40 8 D A D C9 D A 10 D D D D C 11( D o 5'& toTciC. The unique BD in 7/8. 6 D D D B6 A //7 &dapiTe n?po?i padlk? (pixav X?pa Line 7. The ending in 6 is like that of line 25 A C D C D D D AD C ?. The text Page prints in PMG would give CDCB for line 4.d T' ip?tl~ p}tiv. --..Line 5 contains the only instance of DA D that does not refer to the infant. and may have a parallel in 11/12.O)K a&iadvtOtlt 7Cop?1tai. and the echo of line 21 in 8 all affirm the suggestion thata new stanza(the epode) begins after line 7. line 25 (= 1) exhibits the aeolic isochronicrhythm: B23D D 23 __ B C 24 D B 5 //25 D D D B //26 tEapouxoixa 86?TI q(paveir. perhaps this unparalleled ._)4 Kitvr?cioa 6i 4ipLvao.. Lidov follows in 24. note 35.?-iXopa = . here C is transitional.. = ave?6Lo.. ? . Line 5 then would be BBDADC. ?K ?0o 6OTTt 8? apoaX?0ov ?tnoC. B Ac D D B D .. yaaOrvjo) v &' iO10? KvctO)C?1i. the CB is unusual but the line could be taken as a version of line 25 showing contraction. a disyllabic KV(tooCEt. the force of the exclamation. unlike the printed text. OT? dpvca A transitionalC element combines with the first D of an isochronic series that extends to line 3 to form the remarkablyshortperiod of line 2 (= 26): . is a virtual repetition of line 6.B _ )//26 C D /3 D //3 D D A( _4( ] vooqpt iKtxa.... 39 Above. an unusual type of expansion for an aeolic phrase. Ze? XidT?p.80 Joel B. .BD could be understood as BCD (cf. line 18). begun like 4. . = ?v 6atla?a O)yyvc i ot --. tE?p?LprlV?xvov38 A rhythmicallyobscure sequence at the end of the series separatesthis from a very extended isochronic series in 4-6 (the text used for 4/5 does not follow PMG39).?&T?pn?it 5oopaTt axiK?ccoy60(p 38 Reading 'T? gtv itv?ov we get CDDBB for 26/27(=2/3). but it gives a clearer A element between 3 and 4. 40 This assumes correption of the ultima of i0?i and.5?pia.

appear to close ..ci& rot 8ctvbv -r6 ye 8&tvv Ijv. cf.. His arguments. but a transition(the single D followed by B elements would be like the reading accepted for 23). CA and AC do not differ) with another instance of the DAD motif (Page's 16/17). BAD. AlternatingRhythmin ArchaicGreekPoetry 81 Page's emendation of 11 gives a doubtful transitionto a short isosyllabic phrase. B C D B A D 18 19( . If we posit period end at the end of 17 and brevis-in-longo. give three syllables. rhythmicallyambiguous se- quence (cf. This interpretationmakes line 12 not a close. which begins in 14 and contains the sleeping in- fant motif (for the transitionbetween the two series. 38. . The isosyllabic series in 12-14 is succeeded by an- other long isochronic series. CBBC. two long syllables would yield another instance of CB) and synizesis in T?aV.. Line 17. 42 I am assuming correption of the penult in the improbable aiXvav (the mss.11P.Kvgaxro." QUCC 4 (1967) 134-35. Line 18 provides the sort of short aeolic sequence alreadyseen in 23/24 but here quite clearly set off.DBAD-at least if we read the rhythmcontinuously throughto line 18. That would prepare the final phrase.41which prolongs the rhythmof 8- 10. Elsewhere BAD or BAC is found at period-ends or rhythmical transitions. and the sentence. In line 12 the sequence comes to a firm close on a clearly aeolic ending: C D 12 D B B 13. vucdi 6: auc7nci vcuav 6p v6p(p TaOei. 41 "Simon. which are not directed to metrical problems. 24/25 and 6/7): 13 B B B A / aivacv irepOe T?cv wKOpav paOciav A D C 15 D D A D 1C D xuptovTxo. . 6ga?lpov KcCKOV in line 22 (but create an atypical transitionover the period-end:B AB again). Pavese's reading. I give C.then the ending is BAB. This isochronic series continues after a brief. 17-18 would thus have sequences like those in 24-26 but without much rhythmicalclarity. I offer one reading of Page's text42but a different distributioninto lines (indicated by /) for 13 to the opening of 16.oiKc&aXeye?t o>Ub'a&vg. and then closes with the aeolic cadence DB (cf.o (90677yyov.. The opening of line 13 offers a numberof textual and prosodic possibili- ties. 6). A D D B A D 16 C _ D _ 17 KcaXOVv ?V XXavSt xpOOConov xtop9up?a K?ccg?VO... 6/7. persuade me that this is a preferable interpretation. 25.to be a pleasing contrast to the only other occurrence of BAB. np6ocoRovKcoC6v..

On the other hand. If that is the case.overlap- ping system of organization. there may have been a shift in perceptionsuch that the system of recognizable phrasescame to be regardedas the primaryone in poems in which both systems workedat once (or appearedto).complex choral lyric may be crucial only when the patternof alternationcannot provide a rhythmicalbase (and the exact determinationof their terminal points may be neither essential nor even rele- vant). by defini- tion. . to be exhibiting a metrical form based on familiar segments. repeating. A B A D AB 20 _ D B A _)19 _ _ __ KO{iKIV i?coV pp.so that it is period-endwhich is a matterof indifference to the rhythm. suitablefor a cadence. which can provide a sense of metri- cality when the patternof alternationbecomes too variable to be apprehended easily-a system which could function even when the sequence of elements ex- ceeds the limits of the most comprehensivenorm.. isosyllabic. there are stanzas in which a particularrhythmicpat- tern occurs only at the juncture of frames. Ultimately.The former mostly uses a fairly sim- ple.Perhapswhat we see here is the emergence of another. Looking back on such patterns 43 The opening is textually uncertain but syncopation seems unavoidable here.43 The amount of syncopation throughoutthese few lines is remarkable(7 of the fragment's 18 instances are in lines 16-19) and line 19. is especially marked. The poems representedby Simonides 37/542 P and 38/543 P were very dif- ferent in style.ovCa. and construction. it is possible to observe in these last two fragments an occasional use of fairly short segments of rhythmicpatternseasily recognizablefrom the earliestperiods of the tradition. Once such a system arose.82 Joel B. the perceptionof complicatedand variedpatternsof alternationmay have died out.The density of unusual features here suits the crescendo of the pathetic contrastbetween the infant's calm and his mother's and the sea's agitation (whatever may have been the use of the respondinglines in other tri- ads). finally.XT(coV XirnOVri)X?lE?. the latterexhibits an apparently free variationbetweendifferenttypes of uniformsequence. For. the earliest verse must have seemed. Line 20 is an unmistakableaeolic segment.it turnsout that rhythmicpatterningdoes not necessarily become a matterof indifferenceat the juncture of frames. It also incorporatedfree variationin the length of individual seg- ments. This traditionallowed. The traditionthus came to include styles in which there is less predictabilityas to the rhythmicsequences and the clarity of their separationinto differenttypes or segments. and mixed (aeolic) rhythms. in the course of its development in the archaic pe- riod.compoundsequence of elements. on the other.discrete cola. And.and. content. an increaseduse of syncopationand a greaterdiversity in the combination of elements. While there appears to have been some correlation between rhythmic variation(especially syncopation)and regularlyrecurringword-end.In each poem the pat- terns of elements realizing the positiones mutabiliores are a selection of the possibilities offered by a traditionof isochronic. perhapsthe next to last of a stanza. Lidov Line 19 contains the only instance of intralinearDAB. there are patternsthat continue over period-end. one the one hand. the identi- fication of familiarcola in some later.

The Poetic Dialect of Sappho and Alcaeus (New York 1981) 16-27. Bowie. In this light. But if there is any valid- ity to my suggestion that "cola.Mass 1974). Parkerraised in her notes on their original publication. and. HarvardStudiesin ComparativeLiterature33 (Cambridge. AlternatingRhythmin ArchaicGreekPoetry 83 the Alexandrianeditorsand the readersof theirtexts may well have thoughtthat they had no more internalstructurethandid a piece of prose. each allowing a single type of variation within a consistent line. rest on whatever as- sumptions we make about the way it was heard. is held to consist of a large body of verbal patternsand phrases.45If the I-E tradition. isochronicand isosyllabic are much alike.while assuming the validity of the pre-historicreconstructionon the basis of the sur- vival of the cola in the Greek tradition(all the while adducingless and less de- terminateforms as obvious cognates). Lustrum 15 [1970] 67). The methodsof analysis that we use.He finds the origin of these units in the commonest demarcationsof epiploke (demarcationswhich were also exhibitedin archaicpo- etry). E. and aeolic is distinguishedby the evolution of a peculiarcadence following a line of more indifferent structure but more rigid syllable count. understood as an abstract pattern of comparatively fixed and free positions. If equal ingenuityhad been appliedto non-I-E languagesfor as much time. note 4) also arguesthat the system of compositionby cola and metrais a late development. ComparativeStudies in Greek and Indic Meter. should be assumed to be a primary featureof Greekmetric.44 This analysis of archaic rhythms has not suggested or relied on any as- sumptions about the pre-historyof Greek verse forms. on the other. are a late development. This cadence than became a characterizingrhythmicfeatureindependentof its originalfunction. One should beware of assuming the validity of an analysis by cola for the earli- est verse on the basis of the pre-existenceof the cola in the I-E tradition. The descriptive power of this analysis suggests that rhythmical alternation. and I regretmore than he does the absenceof negative evidence. the thesis of GregoryNagy. then they cannot be used as primafacie evidence for an historicaldevelopmentof the longer verse forms out of an inherited body of common Indo-Europeancola. I find it harderthanBowie does to accept some of the deducedcommonforms that he recordsfrom earlierscholarship(see the objections L.The analysis that I have presented. We can no more prove how the originalaudienceheardGreekverse thanwe can demonstratehow the original poets pronouncedit.based on alternation. and the styles of description that we adopt. and follow out the conse- quences in practice.46then it is possible that the three basic types of rhythmic alternationarose as three different styles of regularizationwhen changes in the languagebegan to obscurean originalmetri- cal principle. would not they too have yieldedup evidenceof an enneasyllabicur-verse? 46 Cf.however. unifies and simplifies the presentationof early Greek poetry while allowing us to appreciatevariety in the use of verse forms and their development in time. P. e5 For a summaryof the historyand currentstate of theoryon the Indo-European origins of Greekverse forms see Angus M. whatever it was.on the one handemploys concepts thatare now familiarin the largerframeworkof comparativemetric. 44 Cole (above." which have been the primary object of analysis for the practitionersof the New Metric. .

I do not thereby deny that "metricaldata can hardly be explained except in terms of some property of the Greek lan- guage.53 So I have also kept in mind Suzanne Langer's 47 See Allen 27-65.36 (The Hague 1965) 18- 29. 53 On this exclusion.51 In their conclusions the performance characteristicsof style of delivery and tempo play a majorrole. 1984).48 The most rigorous application of the primaryassumption that meter must be derived from phonology is Allen's. his remarkson the relation of the disciplines of classical metrics and linguistics. see Allen 19 n. the quotationis from p. specifically. who emphasizes (p. 26. It is not clear to me what they take to be the effect of such manipulationon the arrestof long and shortvowels.The usual form of the abstractpatternas presentedby linguists is. 3 (cf. and on the theory of the syllabic pulse as a ballistic rather than controlled movement(see Allen 63-72). 48 A Theoryof Meter. I have addressed this essay to . he does not do so to the extent of positing other than phonetic qualities as cate- gories. Lidov Appendix:Alternatingpositions and linguistics The proceduresof contemporarylinguistic science.50 They have re-examinedAllen's hypothesis of stress and concluded that the metrically relevant phonetic factor was not intensity but a manipulationof actual temporaldurationin series of heavy (long) or of light (short) syllables. Ca. that linguistics without meaning is meaningless. 50Languageand Meter:Resolution. and their Prosodic Basis. 12 (Chico. it seems that the exclusion of psychological factors has cost more than it gained.let alone one directly re- lated to motor phenomena in speech. The weakness of these conclusions suggests that the methodological axioms should be open to question. or an alternationof strong and weak events (usually syllables).Janua Linguarum. on the syllable and its qualitiesas the basic event. AmericanClassical Studies."52But the conclusions reachedby the applicationof a strictperformative approachare not encouraging. Devine and Stephens. Chatmanemphasizes the psychological factor of conception. In contrast to such studies. 49 It should be rememberedthat Allen's study of metricalsystems in Greek is part of a largerstudy of prosodicphenomenagenerally. for example.note 51) 130. and 96- 101 on the characterof rhythmand its relationto motorfactors."attempt to explain the rules governing not only resolution but even the grammaticallyexpressed regularitiesof bridge on the basis of phono- logy. 51 "Stressin Greek?"TAPA115 (1985) 125-52. who call Allen's book "the first work in the field of Greek metre that can trulybe said to understandthe requirementsof scientific method and theory construction. the articlesby Halle and by Devine and Stephensreferredto in note 11. derive a metrical system out of perceived phonetic features.Porson's Bridge. I am not proposing an abstractor ideal form whose componentshave a phonologicalrepresentation.SeriesMinor. either a temporally regularrecurrenceof prominentevents (usuallya suprasegmentalfeaturesuch as stress).we are asked to suppose too much at the level of performancemerely because the theoryconstructionrequiresit. 1 and 101 n."49He has concluded that the prosody of Greek must include a non-accentual stress. Even when S.84 Joel B. as I understandthem. 16). Although I take my own procedureto be one applicationof Jakobson'sfamousdictum. or implies. 52 Devine and Stephens(above.47 See. 19) the "performativeside" because his goal is the "characterizationof the prosodic phonological systems in termsof theirphonetic exponents.

in termsof neutralization. As a consequence. I am asserting thata mentalprocess akin to those involved in syntax or semantics precedes and guides comprehensionand composition.55 Classicists. Second. AlternatingRhythmin ArchaicGreekPoetry 85 comment that "the essence of rhythmis the preparationof a new event by the ending of a previous one. and not attempted to formulate everything possible in linguistic terms (it would be possible to describe the A and C elements. 55 I would like to thankProfessorsSeth Schein and Jacob Ster for readingan earlierdraftof this essay and giving me the benefitof theiradvice and comments. . and therefore better ac- commodates the co-existence of distinctive classes of rhythmwhich can be per- formed in a variety of styles or contexts. it offers a means of unifying the rhythms of Greek verse that avoids imposing prominence or duration as the quintessential rhythmic experience."54In positing normativeconcepts-the two positions in opposition-as the constituents of rhythm (instead of the two syllable weights directly). the theory I offer has two advantages.opposingthem as a class to the B and D elements. it provides a means of ex- plaining rhythmicvariationthatalso accounts for the diachronicincrease in va- riety as a functionof the accumulatingexperienceof poets and audiencesin their linguistic culture (and for the disappearanceof variety as a consequence of cul- turaldiscontinuity). 54 Feelingand Form(New York 1953) 126. for example.which are themselvescontrastedrealizationsof alternation). This researchwas supported(in part)by grantsfrom the City Universityof New YorkPSC-CUNYResearchAwardProgram.First.