Catalexis and Anceps in Pindar: A Search for Rhythmical Logic Pearson, Lionel Greek, Roman and Byzantine Studies; Summer 1974; 15, 2; ProQuest pg. 171

Catalexis and Anceps in Pindar: 
A Search for Rhythmical Logic 

Lionel Pearson 

THE PRESENT STUDY had its origin in an attempt to understand the 
rhythmical pattern of Pindar's dactylo-epitrite odes.1 Dactylic 
rhythm is in 4-time (what Aristoxenus calls the tcoc AbYOC or 

2:2 ratio) and trochaic or iambic is in 3-time (the 8£7TAacwc AbYOC or 2:1 
ratio), and when Greek lyric poetry appears to combine these rhythms 
it may be thought to be alternating between 4-time and 3-time, an 
alternation or p-~Taf1oA~2 which ancient musicians would understand 
and to which they would respond without difficulty. It might seem to 
follow that a dactylo-epitrite ode offers alternation between 4-time 
(dactylic) and 7-time (epitrite), the 4:3 ratio. 

But here difficulties arise. It should first be noticed that Aristoxenus 
specifically rejects 7-time, the E7T'TacTJp-ov P-EY€()OC, and the 4:3 ratio, 
saying it is irrational and unrhythmic,3 though he does not deny, it 
seems, that epitrite feet occasionally occur (this is what he appears to 
say in a fragment quoted by Michael Psellus).4 We may be inclined to 

1 Some of the argument of this article was presented orally in papers to the Classical 
Section of the Philological Association of the Pacific Coast at Gonzaga University in Novem­
ber 1970 and to the Institute of Classical Studies, University of London, in February 1972. 
The present version includes many modifications and, it is hoped, some improvements 
and is more strictly limited in its scope. 

II Cf. Dion.Hal. De Compo Verb. 19, (Aristot.] Prob. 19.15, and Quint. Inst.Or. 9.4.50, where 
he explains the difference between rhythmi and metra: Sunt et ilia discrimina, quod rhythmis 
libera spatia, metris finita sunt; et his certae clausulae, illi quomodo coeperant currunt usque ad 
P.ETa{JOA.qV, id est transitum in aliud genus rhythmi. By rhythmi Quintilian means not only 
<rhythms' but poems (like Pin dar's odes) that are not written in a fixed <metre' like iambic 
trimeter or trochaic tetrameter. Cf. Carlo Del Grande, La Metrica greca, in Enciclopedia 
Classica, SHZ. n, V.ii (1960) 230-32. The term metabole is evidently as old as Aristoxenus, as 
shown in the fragment quoted by pluto De Mus. 1143B. Cf. L. Laloy, Aristoxfne de Tarente 
(Paris 1904) 324, and R. Westphal, Melik und Rhythmik des class. Hellenentums I (Leipzig 1883, 
repro 1965) 160-61. 

3 Rhythm. 35 (Westphal), 302 (Mor.), p.25 (Pighi): n ilE E1TTaC7'Jp.ov p.lyE8oe O?JK ;XE( 
il(atpEcW 1Toil(~v' Tpdiw yap )..ap.{Javop.lvwv )"6ywv €V Tote E1TTa o?JilEte €CT(V ;ppv8p.oc V cLErc p.b 
€CT(V & TOU €1T(TptTOV, ilEVTEpoe ilE & nvv 1TlVTE 1TPOC Ta avo, TplTOC ilE & TOU E~a1T)..aclov. 

;I Psellus, fr.9, p.26 (Pighi): Twv 1To8(KWV )"6ywv E?J~vlcTaTot Elc(v ol TpEic- 0 TE TOU tcov Kal 
& TOU 8mAaclov Ka~ & TOU -9p.(o)..{ov. ytvETal ill 1TOTE 1Tove Kat Jv Tpm)..acltp )..6ytp, ylvETa, Ka~ €v 
€mTptTcp, 

171 



172 CATALEXIS AND ANCEPS IN PINDAR 

disregard his rejection of 7-time, since musicians today do not find 
7-time more difficult or awkward than 5-time, the 3:2 ratio, which he 
allows.1) It is possible that elsewhere in his treatise on rhythm he rec­
onciled his theoretical objections with the practice of Greek com­
posers. But he may be telling us, as modern critics do, that although 
certain odes appear to be < dactylo-epitrite', we must not be content 
with this superficial definition but seek some more satisfactory 
rhythmical explanation. 

Ancient writers insist on the distinction between metre and 
rhythm, but little or nothing is said of this distinction by the current 
school of metridans, whose doctrines are most commonly presented 
to students in this country. They maintain that lyric metre can be 
analyzed into <metrical units' linked by syllabae ancipites, and that an 
apparent epitrite consists of the metrical unit -u- (a cretic) followed 
by a syllaba anceps.6 It is a fundamental objection to this explanation 
that it is unrhythmic, that it does not stop to ask about the prevailing 
rhythm or the alternating rhythms in a strophe.? And while seeking 
to explain away the epitrite, it provides a substitute which is open to 
exactly the same objection. It is a recognized principle of ancient 
rhythmical theory, quoted by Aristotle himself, that Hone {3ac£c does 

G 7/4 or 7/8 time is formally marked in some XX-century compositions, but it is not 
common (the example best known to me is Guillaume Lekeu, Sonata for violin and piano, 
second movement, published in 1907). 5/4 time, on the other hand, has become quite 
familiar since the late romantic period, and Westphal's remark (op.cit. [supra n.2] II.clx) 
that it is "unserer modernen Musik so gut wie fremd" seems curiously outmoded today. 

• Gf. especially Paul Maas, Greek Metre, trans. H. Lloyd-Jones (Oxford 1962) §55, and A. M. 
Dale, "The Metrical Units of Greek Lyric Verse," in Golleaed Papers (Cambridge 1969) 41-
97 (for dactylo-epitrite more particularly 53-60). I do not wish to decry the useful work that 
metricians have done by their statistical analysis of lyric verse, e.g. Dale, "Metrical Analyses 
of Tragic Choruses, i, Dactylo-Epitrite," BIGS Suppl. 21.1 (1971). My complaint is that such 
statistical analysis is often regarded as an answer to a problem instead of a collection of 
evidence which needs interpretation. 

7 Gf. Del Grande, op.cit. (supra n.2) 214-15. Indeed the same criticism can be made of 
modem metricians that Emile Martin makes of the late Greek writers on metric: "Au lieu 
de se laisser guider par l'oreille et Ie sentiment inne du rythme, ces metriciens decom­
posent mecaniquement les metres sans se souder de leur valeur rythmique propre," Essai 
sur les rythmes de la chanson grecque antique (Paris 1953)50-51. Like the metricians oflate anti­
quity, modem metricians have produced a highly complex system. In [Aristot.] Prob. 19.15 
we are told that antistrophic choral odes were not rhythmically complex but fairly simple, 
litc pv8p.dc yap ~C7"' 1C1l~ lv~ JLI!TPIi£Tllt. Even if this last statement is not literally true (if the 
words are given the meaning to which we are accustomed), it should warn us that we must 
look for a prevailing rhythm. 



LIONEL PEARSON 173 

not make apv()p.oc,"8 evidently a proverbial saying like «one swallow 
does not make a summer," "one step does not make a waltz." Single 
dactylic or trochaic feet do not occur in Pindar; there is always at least 
a dipody to establish 4-time or 3-time; but the epitrite quite often 
occurs (apparently) in isolation, as at the beginning of Olympian III: 

TvvoapLoaLC T€ cP,AOg€LVOLC ao€,v KaMmAoKap.cp ()' ·EMv~ 
vvl - v vi - -I - v - - I - v vi - v vl-

Can it be correct to analyze this as dactylic rhythm interrupted by 
a single epitrite basis, with a quick return to the original dactylic 
rhythm, or should we say that the epitrite is really a trochaic dipody, 
with its last syllable commonly lengthened, -v-v, comparable to 
the iambic dipody or metron, which establishes the rhythm of the 
iambic trimeter and has its first syllable apparently anceps, U-u-? 
And how is the isolated cretie (the 'metrical unit' substituted for the 
epitrite) preferable, when it occurs in isolation and, as a single basis, 
cannot establish rhythm? 

The whole notion of syllabae ancipites is also alien to rhythmical 
theory. Should we be content to record what we observe, ignoring the 
theory, or seek some way of reconciling our observation with it? 
Modern metricians pay little attention to ancient rhythmic theory, on 
the ground that direct evidence for it is inadequate (since only a frag­
ment of Aristoxenus' Elementa Rhythmica survives and Aristoxenus, in 
any case, may not represent orthodox opinion) and the indirect evi­
dence, from later writers, is hopelessly confused. Some metridans are 
prepared to regard the anceps as having some indeterminate quantity 
or an intermediate length between long and short (this seems to me 
a very curious attitude, as though in a restaurant which offered the 
alternative of tea or coffee they would be content if offered something 
in between tea and coffee to drink). They ignore the objection of 
Aristoxenus, who will not accept any ratio except 2:2, 2:1, and 3:2.9 

8 Arist. Pol. 2.1263b, bJC7T£P KtiV £i TLC -ri}v cvl4wvtav 7TOL7]W£V ofWrPwvtav 7j TOV pv8p.ov f3rJ.CLV 

p.lav (as examples of paradox). 
9 Paul Maas in his Greek Metre (which orthodox metricians regard as an authoritative 

handbook) suggests that the time value of the anceps "lay somewhere between that of the 
longum and breve" (§51). He appears not to be interested in seeking the support or con­
sulting the opinion of ancient writers on such a matter. Others. who use the term 'irra­
tional' to describe the final long syllable of the apparent epitrite. claim that they have the 
authority of Aristoxenus for this term (cf e.g. W. J. Koster. Traite de metrique grecqueZ [Ley­
den 1953] 27. A. Kolar. De re metrica poetarum Graecorum et Romanorum [prague 1957] 44). 

Aristoxenus in fact says that 'irrational' feet may occur in which there is a ratio between 



174 CATALEXIS AND ANCEPS IN PINDAR 

But apart altogether from ancient theory, the syllaba anceps seems to 
pose an impossible problem for the performer: how can a perform­
ance be anything but chaotic, if musicians and dancers never know 
when a long note may be shortened? The problem is not serious, per­
haps, in spoken dramatic dialogue, but in choral lyric it cannot be 
passed over as of no importance. It is all very well to maintain that the 
rhythmical instinct of the ancient Greeks was different from ours, but 
persons who make such statements must recognize that the burden 
of proof rests on them; it is for them to show how and why ancient 
rhythmic instinct and ancient hands and feet and voices were different 
from ours.l0 Unless they can provide proof, it may be easier to ex­
plain away the anceps as well as the epitrite. 

But before attempting or even admitting the need for such a task 
it may be best to dear away some preliminary difficulties. If one starts 
out with the traditional interpretation of dactylo-epitrite rhythm, 
accepting provisionally that it is an alternation of dactylic, 4-time, 
with epitrite, 7-time, one soon discovers that Pindar's words will not 
fit easily into such a regular pattern of alternation. The final syllable 
of the epitrite often appears as short, and so also, though much less 
frequently, does its third syllable, -v~- (unless the evidence of the 
manuscripts is neglected).l1 The final syllable of the dactylic hemiepes 
is also often short, -vv-vv-~, and in their catalectic form both 
epitrite and hemiepes often have their final syllable short, -v~, 
-vv-vv~. These variations would trouble no one if they occurred 
only at the end of a strophe or in some other place where a final close 
was clearly intended. But the variation between long and short 
syllables, long in one strophe and short in another, is not restricted to 

arsis and thesis other than 2:2, 2:1, or 3:2. But he means that they occur when someone 
sings or dances out of time, just as 'irregular' intervals occur (between a fourth and a fifth 
or fractions of a quarter tone) when someone sings out of tune; such irrational feet, he says, 
have no place in rhythmopoiia and are not KaTa rqv TOV pv8JLOv cpuc,,", just as the irregular in­
tervals are impractical for purposes of performance and have no place in melody. His 
statement is dear and lucid in Rhythm. 20 (W.) , 292-96 (M.), pp. 22-23 (Pighi). He could not 
express himself in this way if he found 'irrational' feet in every strophe of Pindar' s odes. Cf. 
Laloy, op.cit. (supra n.2) 299-300. 

10 Cf. e.g. F. Greif, "Etudes sur la musique antique," REG 26 (1913) 289: "11 ne sect de rien 
de se refugier dans la nuie des temps recuIes pour affirmer que la conception rythmique 
des anciens etait differente de la natre." 

11 Cf. C. M. Bowra, "An Alleged Anomaly in Pindar's Metric," CQ 24 (1930) 174-82, who 
argues the case for alternate readings or emendations. The problem of the tertia brevis 
must be reserved for another occasion. 



LIONEL PEARSON 175 

places where a final close seems to be indicated. It occurs constantly 
in the middle of a strophe, for example in Pythian III, which begins: 

"H8eAov Xipwva KE (/hAvpioav, 
, , ~O'C' "'\1 

EL XpEWV 'TOU ap.,E'TEpac a7TO Yl\wccac 
, "I'. (}" y , \, , KOtVOV EV£"aC at E7TOC, sWEtV TOV aTrO£XO/LEVOV, 

0 , '<;' , 
vpavwa yovov ••• 

In the first line the catalectic hemiepes, following the epitrite, has 
its final syllable long, as in other triads, but in the third line the cata· 
lectic hemiepes, after an epitrite dimeter, has its final syllable short, 
though it is long in other triads. If long syllables are regularly given 
twice the value of short, how is it possible to have this variation be· 
tween one triad and another without altering the rhythm and can· 
fusing the dancers and musicians? It must also be asked if the singers 
continue without rest or break after the catalectic hemiepes. The two 
questions can be answered together: if the singer has a rest after a 
catalectic hemiepes, it does not affect the rhythm whether the final 
syllable is long or short. Thus clarity and rhythm are preserved if the 
words are intended to be sung like this: 

7 4 
4d.lddl4d.l.ll d.l.ll d---I 

7 4 
4d.lddl4d.l.ll d.l.ll ddl 

7 4 I .I ~ ___ (str. a) 
4 d .I d cl 1 cl .I d cl 1 4 cl .I .I 1 d .I .I 1 d ___ (str. 13) 

The sixth line of the first strophe provides an even clearer illustra· 
tion: 

l' , \ 0 '. t. " <;, , o tOe EWV pE'I'EV 7TO'TE 'TEK'TOVa vWOvvtac. 

The unwary reader may read this as dactylic sequence, 

-vv - - -vv -vv -vv-. 

But comparison with the second strophe, 

• '0' , 0 ' , H , 7Tap EV V'TaTtp yvwfloav 7T£ WV, 7TaV'Ta tCaV'Tt votp, 

shows that the apparent dactyl, -¢;EV 7TO'TI., is really epitrite catalectic, 
-v~. How is the singer to mark the rhythm, keeping it the same in 
each strophe, unless he sings: 



176 CATALEXIS AND ANCEPS IN PINDAR 

4 7 4 
4 d .I .I 1 d d 1 4 d .1.11 ---I 4 d .I .I 1 ... (str. ex), 

474 4 d .I .I 1 d d 1 4 d .I d 1 ---I 4 d .I .I 1 •.• (str. f3)? 

The question may be put in different form by returning to the 
opening lines of the strophe. If there is no rest after the catalectic 
hemiepes, what happens to the rhythm? Is the dactylic rhythm suc­
ceeded by anapaestic, or is there a single note, a 'non-foot', between 
two feetl2 before the dactylic rhythm resumes? Far simpler than such 
rhythmic complexity is the solution that the term' catalectic' suggests 
of itself. Although authors of metrical handbooks generally avoid 
giving any intelligible definition of the term, the catalectic form of a 
metrical sequence is so called not simply because it leaves the final up­
beat of the normal sequence silent, but because it is an appropriate 
way of reaching a close-of 'ceasing'.IS To modern taste also it seems 
normal to reach a close on the downbeat and to take a rest on the up­
beat before resuming the rhythmical pattern; the upbeat, the 
remaining megethos of the foot, can be observed as well in silence as in 
sound.14 

11 Aristoxenus will not allow a shorter foot than a -rplCTJJI.Ov pky€8oe (Rhythm. 31 [W.], 302 
[M.], p.24 [pighiD. Far less can a short syllable be a 'foot', "Or, p.Ev oW ~e EIIoe 'XP0vov 1TOVe 
o?J/~ av €r7] q,av€pov, 18 (W.), 290 (M.), p.21 (pighi). 

11 D. S. Raven, Greek Metre (London 1962) 37-38, will hardly help anyone to understand 
catalexis when he calls it "suppression of a syllable" or "a special form of syncopation." 
Bruno Snell, Griechische Metrik3 (Gottingen 1962) 5 n.2, is more correct but also quite enig­
matic when he speaks of catalexis as "eine Silbe in zwei Blementen" and adds "denn auch 
hier gleicht natiirlich die Zeit der Pause die fehlende Silbenzeit aus." Likewise Paul Maas, 
op.cit. (supra n.6) §58, is content to say that, when a pause is reached, "the group that pre­
cedes the pause, the clausula, usually differs from the rest in form by being abbreviated, or, 
as we say, catalectic." In §SI, however, he takes the true explanation for granted-an admir­
able paragraph, but baffling to anyone who has not thought out the solution for himself and 
reached the same conclusion. 

U The statement will seem obvious enough to any persons who are accustomed to sing 
"the home of the brave" (not "the fearless"), "blow, blow, thou winter wind" (not "thou 
wind of winter"), or "ein' feste Burg ist unser Gott" (not "unser Herr Gott"). Metricians of 
the school now in fashion may regard such a statement as totally irrelevant to ancient 
poetry. because they refuse to use the terms 'upbeat' and 'downbeat' or arsis and thesis in 
any discussion of metre, as though it were irrelevant how the dancers kept step or how the 
musicians kept the time with their feet. And yet one supposes they must have read De 
Sublim. 41 and the remarks of Aristoxenus about "marking the rhythm and making it per­
ceptible to the senses" in Rhythm. 16 (W.), 289 (M.), p.21 (pighi). A 'foot' after all is not called 
a foot because it is a mark on a piece of paper. 

With regard to 'rests' in choral lyric we have the clear statement of Quintilian (a writer 
whose authOrity should never be despised): lnania quoque rempora rhythmi fanUus accipient, 



LIONEL PEARSON 177 

Metricians seem unwilling to concern themselves with such practi­
cal details as measured rests, and indeed show little interest in prob­
lems of performance. It seems they are content to apply the adjective 
anceps without asking what the poet-composer expects his perform­
ers to do with syllabae ancipites. Their fundamental mistake is in treat­
ing these syllables in isolation; but the proof and explanation of this 
accusation must wait until later in this article. 

A rest, whether of one, two, or three chronoi (the equivalent in 4/4 
time of a quarter note, a half note, or a dotted half note), at a cata­
lens is certainly the easiest way of reconciling the variations between 
long and short considered so far. And does anyone propose to read a 
dactylic pentameter otherwise than as: 

dJ.l1 dJ.l1 d .... 1 dJJI dJJI d .... 1 

It should also be asked if singers are expected to sing an entire 
strophe or epode without any rests in which to take breath. And does 
it follow that the orchestra is silent or the dancers stationary during 
such rests 1 It seems to me most unlikely that the dancers came to a 
standstill until the end of a strophe or even the end of a triad; dancers 
do not interrupt their movement in order to breathe, and it may in­
volve a greater effort to stop than to keep moving. Unfortunately for 
modern scholarship no musical score or choreographic design for any 
of Pindar's odes has survived from antiquity, and the actual musical 
texts that have been discovered provide only very little help. We 
have to do the best we can with the words alone. It is almost as 
though Schubert's music were lost, and we had only some poems of 
Goethe and Heine preserved as "Schubert's compositions." But it is 
not quite so bad as that, because Greek verse recognizes quantitative 
differences and cannot be set to music with the same rhythmic free­
dom as English or German or French.15 Despite our handicap, we 
must not fall into the error of treating Pindar's poems as though they 
were meant for private reading or declamation. They were meant for 
musical performance; Pindar was a composer as well as a poet. 

It is equally important to remember the differences between sung 
and spoken verse. Spoken Greek had a tonic accent, and it is customary 

quamquam haec in m£tris accidunt. Maior tattren illic licentia est, ubi tempora etiam ef m£tiuntur et 
pedum ft digitorum ictu, ft intervalla signant quibusdam notis atque aestimant quot breves illud 
spatium habeat (Inst. 9.4.51). For his distinction of rhythmi and m£tra see n.2 above. 

15 Cf the good remarks of Emile Martin, op.at. (supra n.7) 82-85. 



178 CATALEXIS AND ANCEPS IN PINDAR 

to say that it had no stress accent. But music, especially music that 
accompanies dancing or marching, will introduce a stress accent, 
sometimes stronger, sometimes weaker, with a downbeat and an up­
beat that dancers and marchers will recognize, a thesis or basis when 
they will put their foot on the ground and an arsis when they will 
raise it.16 Modern metricians speak of the rhythm of Greek verse as 
«purely quantitative." This may to a great extent be accurate of 
spoken verse, but it cannot be correct of choral lyric. Without Pindar's 
music we cannot hope to discover all the dynamic effects and subtle­
ties that he intended, but the combination of rhythmic pattern and 
words should tell us much of what we want to know. For example, it 
will sometimes suggest that a strong emphasis or fortissimo is de­
manded elsewhere than on a downbeat and elsewhere than on a 
syllable that is accented tonically.I7 

There is also much to be learnt from the conventions of spoken 
verse towards solving the mystery of the anceps. In iambic trimeter, as 
in dactylic hexameter, it causes no difficulty that the last syllable of 
the verse appears to be anceps. When the end of a verse in iambic trim­
eter coincides with the end of a sentence or clause, we can quite 
properly say that the final iambus may take the form .I .I ~, since a 
break or rest before starting the new verse will be appropriate. But 
what happens when a break at the end of a verse would be inappro­
priate? 

One might have expected that the matter would have been investi­
gated before now, but Maas is content to say: "The last element of the 
line ... is always anceps in so far as any last syllable of a line may be 
prosodically long or short (Aristeides Quintilianus 1.21 p.44 uses the 
term aSt&'cpopoc). But since internal responsion very often requires a 
longum, and hardly ever a breve, at this place, and since we have to 
reckon with the possibility that even a short final syllable may have 

18 There has been confusion about the terms arsis and thesis, because later antiquity re­
versed their meanings, applying them to the raising and lowering of the voice instead of 
the feet. Metricians (cf especially Maas, Greek Metre §8) have taken advantage of this confu­
sion, claiming that it absolves them from attaching any importance to the terms. Maas, 
indeed, regards them as "superfluous concepts." He says he is trying to describe "the most 
important phenomena with as few preconceptions as possible," but seems not to recognize 
that it is just as much a preconception to deny as to admit that one perceives something. It 
must be remembered constantly, of course, that musical downbeat is not the same as 
verbal stress accent, though it will often correspond with it in languages that have a strong 
verbal accent. 

17 This is a matter which I hope to discuss in a subsequent article. 



LIONEL PEARSON 179 

been made prosodically long by the presence of a pause after it (brevis 
in longo), every final element is noted as a longum. [But it would be 
better to note it as a <finale', and to denote it by the symbol r.".."18] 

A study of some passages from Attic tragedy soon convinced me 
that there was no need for such equivocallanguage about iambic trim­
eter, and that it was not a matter of indifference whether a particu­
lar verse ended in a long or a short syllable. Wherever I looked I 
found that in nearly every case in which the final syllable was short, a 
break or rest before the first word of the next verse was appropriate, 
even though modern convention might not think it correct to insert a 
punctuation mark. On the few occasions when, at first glance, it 
seemed better to continue without a break, brief reflection convinced 
me that a break was in fact acceptable, if not actually necessary. On 
the other hand, when it was essential (or clearly desirable) that the 
speaker, in order to make his meaning clear, should proceed without 
break from one verse to the next, the final syllable was regularly long, 
either by nature or by position, when the first word of the next verse 
began with a consonant (or a double consonant, if the short final 
syllable ended in a vowel).19 

It may seem rash to express oneself so dogmatically without offer­
ing a complete statistical analysis of all surviving Attic drama, but an 
examination of representative passages from the three tragedians pro­
duced remarkably consistent results. The passages which I chose to 

examine were the messenger scene from Aeschylus, Persae (294-531), 
the prologue and first epeisodion from Sophocles, Oedipus Tyrannus 
(1-150, 216-460), and Euripides, Hecuba (1-58, 218--443), and the 
messenger scene from Iphigenia in Tauris (1284-1489). The selection 
was made quite at random, except that messenger scenes and pro­
logues were included as containing fewer changes of speaker. 

18 Greek Metre §34. I prefer to avoid the term 'pause' and the symbol 17\, which suggests 
an indeterminate interval, such as occurs at the end of a period or if some special effect is 
intended. I use the term 'rest' in its musical sense of a measured interval, which can be de­
noted by precise musical symbols, and 'break' to denote a 'stop', as indicated in music by " 
where a singer may take breath by robbing the preceding note of some of its value. 

19 J. Descroix, Le trimetre iambique (Macon 1931) 288-95, examines the most striking 
examples in the tragedians of continUity between one verse and another-where "a ce 
repos du metre ne correspond aucun arret du sens." He does not point out that the final 
syllable is almost invariably long in the passages that he cites. In some of them the verse 
ends with 7Tap&., vmS, or E7Tt, and when the following word does not begin with a double con­
sonant, as in Soph. Trach. 539 p.dic v7T(3 xAalVTJc. it is followed by a nu, as in Trach. 557 and 
Phil. 626, aM' EYW p.tv €lp.' l7Tt vaw. An actor would certainly be tempted to say something 
like E7TtVVaW, so as to avoid the break. 



180 CATALEXIS AND ANCEPS IN PINDAR 

The Sophocles passages will be considered first, since they supply 
some of the most remarkable examples. It will be noted that modern 
editors add a punctuation mark only slightly more frequently after 
short final syllables than after long. But as one looks at verses where 
there is no comma after a short syllable, one can see immediately that 
a break is always permissible, often highly appropriate, sometimes 
indeed demanded by hiatus: 

2 Tlvac 'TTo(J' ;8pac TeXc8€ /Lo, (JOeX~€T€ 
• , \.Lt' '/: ' tKrYJptO'C K/UWOtC£V €S€CT€/L/L€VOt; 

15 opCf.c /L~V ~/Lac ~AlKOt 'TTpOc~/LE(Ja 
[3w/LotCt Totc cotc, 

228 ,'\ "\ \ '\ 
'TT€tcETat yap alV\o /L€V 

, \, t' , A t"" '[3\ [3' aCT€P'Y€C OVOEV, ')'1}c 0 a'TTEtCW a I\a 7Jc. 

It may surprise some readers that a break is permitted after on (59) 
or «Xv (12), when it precedes its verb,20 but they will remember that 
phrases like 87JAov6n and TaX' «Xv can be used parenthetically and that 
a break after other uses of relative pronouns and adverbs seems nor­
mal enough. Sometimes a break at the end of the verse is effective in 
laying emphasis on the word or phrase that follows: 

246 EtT€ TLC 

ErC Q)v MA7J(JEV E'TE 'TTAEt6vwv /L'Ta, 

249 "'1:. I ° tKOLCW Et SVV€CTLOC 
, .... ,.... I ,,_ ~, 

EV TOtC E/LOtC Y€VOtT E/LOV CVVEtoOTOC ••• 

More remarkable is the exploitation of hiatus: 

" \ - I EVVE'TTW C€ Tep K7JPVY/LaTt 350 
l' A" ep'TTEP 'TTPOE t'TTac E/L/LEVEW, 

401 \' t' A \ \ • (J' 't' Kl\a,WV OOKE tC /LOt Kat CV XW CVV EtC TaoE 
, \ , 

aY7Jl\arYJCEW. 

The jolt at the end of the verse with hiatus after the short final 
syllable is unmistakable and startling, and it puts a special emphasis 
on the word that follows.21 The interval occupied by the hiatus, the 

10 cf. Soph. Ant. 61, 98, 649, 1043; Phil. 46, 64. 
11 Cf. Soph. Phil. 268, Ajax 323, Eur. Ion 3. 



LIONEL PEARSON 181 

'noteworthy silence' as ancient critics called it, is distinctive.22 The 
dramatists permit hiatus only at the end of a verse, where it is quite 
co~mon, and not all examples of it are equally startling.23 

No long investigation is needed to show that the final syllable is 
long when it is important that no break be made at the end of a verse, 
for example: 

16 t "~I , 
OL f.1.€V OVO€1TW f.1.aKpav 

'(J (J' t~'" fl ... 1TT€C aL C €VOVT€C J OL O€ cvv YTJ~ ap€LCJ 

r .... " 'Z ' .,~ "f) , L€PTJCJ €yw f.1.€V TJVOCJ OLOe T '[J €WV 

A.€KTOl 

No break can be allowed after fJ-aKpaV or iJ(JEWV, though it may be 
desirable after flap€'ic. Another clear example is: 

236 ,M '\:I " ~ --- ,.. ~ ,,,.. 
TOV avop a7TavOW TOVTOVJ OCTtc ECTL J YTJC 
~ ~, .3: .,' '(J , , TTJCO J 'IC €yw KpaTTJ TE KaL povovc V€fJ-W ••• 

And there are some striking examples in the speeches of Tiresias :24 

316 CPEV CP€VJ CPPOV€'iv cfJc 8EWOV Ev(Ja fJ-..q TEA.TJ 

A.v'l} cPPOVOVv-rL J 

324 t "" \ ,~ , , , , ,/..,' '" 
opw yap ovoe COL TO COY "t'WVTJfJ- LOV , , 

1TpOC Ka,pov' 

337 amv EfJ-EfJ-tJiW rYjv EfJ-~VJ rYjv c..qv 8' OfJ-ov 

valovcav 013 KaTE'i8€c ••• 

Parallel examples in these different categories are easily found in 
the other plays. In Persae (364f): 

l' , " .1..\' ,... ", (J , EVT av "t'I\EyWV aKTLCLV TJI\LOC X ova 

A.~~?7J 
22 Cf Dion.Hal. Dem. 38: ofdKJIV'Tat yap tJ1T6 T€ p,oVCtKcJJV Ka~ P.€TptKWV 0 Ota /L'COV TWV 

</>wVTJ'V'TWV xp6voc €Tlpwv 7Tap"p.{JoAfj ypa/Lp.aTWV .q/Lt,pcfJvwv ava7TAr/povc(}at Bvva/L€Voc. TOV-rO 

8' OQK &v JylYV€TO p.~ ctW7TfjC TWOC agw'\6yov Ot€tPYOVC7JC Ta q,wvq6'7'a a7T' ill1jAWV. He has 
just remarked that the use of hiatus helps to give the effect of TD up.v6v which is 
characteristic of the 'austere' style. Modern critics are inclined not to think of hiatus as 
serving any purpose and generally notice only how much or how little care is taken to 
avoid it. Maas says that in Greek poetry it is "generally carefully avoided except at a 
pause" (Greek Metre §141). 'Pause' is not a good term to describe the break at the end of a 
verse; the xp6voc or CtW7T.J at hiatus is hardly a pause (certainly not in the passages cited), 
and it should be remembered that one cannot take breath at a hiatus unless there is a 
longer rest or pause, such as there might be at the end of a period. 

23 Other interesting examples in the OT are 7, 70, 97, 108,300,352,421. Hiatus after a long 
vowel or diphthong in the final syllable occurs at 31, 60, 124,222,258,317,372,404,413. It 
seems not to produce the same kind of jerk, since it is easier to insert a semivowel, as 
described by Dionysius of Halicarnassus (see preceding note). 

24 Cf also Soph. Phil. 263-{)7, 320-23, 327-28, 337-38. 



182 CATALEXIS AND ANCEPS IN PINDAR 

the break after the short final syllable puts emphasis on the unex­
pected word that follows,25 and in 358-60: 

<lE'\ \ , " , \ \ \ 1\ 
I\/\TjVEC OV J.LEVO LEV, UI\/\U CEI\J.LUCLV 
" , (j I "'\\ "\\ 

VUWV E7TEV OpOV'TEC UI\/\OC UI\/\OCE 

~ " ..I.. I R' , I 
OpUCJ.LCfJ KpV~UUp ~W'TOV EKCWCOLU'TO, 

the contrast between the continuity of dAJ.LUCW vuwv26 (with long 
final syllable) and the break after aAAOCE are noteworthy.27 

The practice of Euripides seems to be the same. The ghost of Poly­
dorus in Hecuba begins (If): 

<lH "(j" \, '\ KW VEKPWV KEV J.Lwvu KUL CKO'TOV 7TVI\UC 

Amwv, 

with a long final syllable at the end of the verse;28 and a little later 
when he says (sf) 

c\ , ~, " X '\ I oc 'TTjV<O > UpKTTJV EpCOV1Jnuv 7TI\UKU , 
C7TnpEL, 

no break is wanted, since C7TElpEL contains no surprise, and the final 
syllable of 7TAaKct is evidently lengthened by the double consonant 
that follows.29 As Hecuba approaches he makes good use of hiatus in 
two consecutive verses (52£) :30 

In the scene that follows, Polyxena produces an effective break at 
the end of verses with final short syllables: 

Z6 cf. Aesch. PV 226, Supp. 326, Cho. 267, Bum. 470, and with hiatus 190. 
lI8 Modern editions, evidendy with manuscript authority but without comment, include 

the final nu which makes the syllable long by position. 
17 Other good passages from the Persae which show a break after a short final syllable are 

309-10,335-36. In 399-405, after a contrast between two short final syllables and one long, 
there is an excellent example of hiatus; then in ~AEV8EPOVrE III -nai8ac, the break after a 
short final syllable puts tremendous emphasis on the next word. 

18 Cf. 10,13,30,35,45, and Soph. Phil. 1,9, 13,35,58,77 . 
• 1 Cf. Soph. Phil. 243. 
30 Cf. in Hecuba's speeches 276, 284, 289,340,387. There are also some good examples in 

her appeals to Odysseus, 286, 385. 



LIONEL PEARSON 183 

351 €7TE'T' JOpi1>07]v JA7TtoWV KaAwv V7TO 

f3aCLA€VCL vt5JLcP7]. 

377 e \ ~." " ~ \ \ , , avwv u av € '7] fLalV\oV €v'Tvx€e'T€poe 

~ ~wv. 

On each occasion the break is the Inore noteworthy because a series of 
long final syllables has preceded, 346-50, 369-76. Odysseus has long 
final syllables, except at the end of a sentence or clause, and only at 
the end of the scene does he make a break at the end of a verse (394f): 

'" , ~ ()' , , al\(C KOp7]C C7]C avaTOC. ov 7TPOCOtC'T€OC 
"\ , \"\ , 
alV\oc 7TpOC alV\CfJ. 

Is this slight hesitancy of speech the first hint that he gives of human 
feeling?31 

The messenger in Iphigenia in Tauris is equally sparing in his use of 
short final syllables, and when he has one the effect is noticeable,32 as 
in: 

1331 )A' A' , , .1..' , yaJL€fLVOVOC 7TatC. we a7Topp7]'TOV <yl\oya 

et50vca ••• 

1342 cP6f3CfJ S' & JL1] xpijv €icopa.v Kae~JL€()a 
CLyfj. 

A longer and more detailed analysis might reveal further subtle­
ties, but enough evidence has been offered to suggest some definite 
conclusions; if a word ends in a short syllable, where a long syllable 
would be more normal and Inore strictly in keeping with the de­
mands of the metre, this seems to be a way of telling the actor that he 
should make a break and perhaps put some special emphasis on the 

31 Cf the hesitancy of Odysseus in Soph. Phil. 79-80, where an actor might give the im­
pression that he was not quite sure how to deal with Neoptolemus: 

;go,8a, 1I'a'i, ,pvc£, c£ p..ry 1I'£rPvKoTa 
TOlaVra ,pwv£'iv p.TJ8€ T£)(IIacOa, KaKa. 

Similarly in 83. Neoptolemus shows no such hesitancy in his speech. 
3S Cf also 1337, 1369 (hiatus), 1385, 1397 and 1399 (both with hiatus). Examples from 

comedy are easily found. Many good instances occur in Ar. Ran. 830-94, as well as in the 
trochaic tetrameters of 905-91-with Aeschylus and Euripides supplying some of the best 
examples. See also Men. Epitrep. 94-176. The comic effect intended by Eupolis (fr.73 [K.]) in 

ou 'Yap cLUa 1I'PO­
povAwp.a pacTa'ovc, rijc 1I'OA£WC p.Eya 

is surely the greater because 1I'PO- is a short syllable and the actor cannot help stopping in 
the middle of the word. 



184 CATALEXIS AND ANCEPS IN PINDAR 

word that follows. In spoken verse the words themselves should be 
sufficient guide to the actor. In choral lyric the melody would help to 
show how the words should be phrased, but if there is to be variation 
in phrasing or dynamics between one strophe and another, the in­
structions should be contained in the words. 

Music is more than a pattern of longs and shorts combined with 
highs and lows in a system of harmony. I venture to say that Pindar, 
like ourselves, felt that there must also be rhythmic movement and 
dynamic variation before there could be music, and that some in­
structions to his performers lie hidden in his verse if we are bold 
enough to search for them; and that if we are content with the current 
system of metrical analysis, we are making too little effort to under­
stand and appreciate his art. An effort of our musical imagination is 
necessary if we are to discover his purposes, and it will be best to take 
the easier questions before the difficult ones. 

A passage from Pythian III was chosen33 to show that a metrical 
pattern like 

-v---v---vv -vv - -vv -vv 

seems to mean 

or that -vv - - -v ~ -vv -vv means 

d ~ ~ I d did ~ I :1::: I d ~ ~ I d ~ ~ I 
If we look at the third epode of Olympian III: 

, QA , ,J. \' '\ , 
EVCEfJEt. YVWJLC£- .,.,VI\<XCCOVTEC JL<XIC<XpWV TEI\ET<XC. 
,~, . , ,"... ,~ , 

Et. 0 <XPLCTEVEt. JLEV VOwp. ICTE<XVWV OE 
, .... , 

XPVCOC <XLOOLECTaTOC • 
.... ~, " , 

VVV OE 7TpOC Ecxanav • • • 

we might think that after JLalC&pwv TEAET&C (-vv -) and al8odcTaTOC 

(- - -u-) there could be a full stop, a real pause of indefinite 
length. But this cannot be right, because in the other epodes, though 
there is word division at the same point, there is no suggestion of a 
full close. This does not mean that there can be no break or rest; in­
deed, if there is no rest, there is a difficult change of rhythm, with 
dactylic changing to anapaestic movement: 

aa See pp. 175-76 above. 



LIONEL PEARSON 185 

. d ~ I d did .II d d Id ~ ~ I d ~.l1 d d I .I did d I .I .I d 

On the principle that Pindar's odes were not meant for professional 
singers,S4 it is likely that he intended the easier alternative: 

el..J1 elell el..J1 elell el..J..J1 el..J..J1 el ... 1 el..J1 elell el..J.I 

It was not for no reason, surely, that he made the word division at the 
same point in each epode, and in epodes a and f3 as well as yarest of 
two chronoi (a half note in 4/4 time) wo~ld be appropriate for the Sing­
ers; the orchestra could sound a note during the rest, the dancers 
could take a step, and there would be no break in the rhythm. Like­
wise at Xpvc6c alSodC'Ta'TOC, vvv S~ 'lTpOC we may suppose that 

-u---v--vv 

means el ..J I el el I el ..J I el --- I el .I .I I· 
The same solution is surely right in Olympian VI, when dactylic 

movement ends catalectically: 

v ,., , ~ (J' (J" ApVC€aC V'lTOC'Taca:vT€C €V'T€'X€£ 'lTPO vpcp al\aJLov 
, 'fI(J \, Kwvac we O'T€ a:7J'TOV JL€yapov 

'lT~oJL€V' ••• 

i.e. el I el .I I el el I el .I I el el I el .I ..I I el ..I ..I I el ---I 
el..l..ll el..l.ll elell d..l..ll d---I 
el ..I ..I I .•. 

The break (not a pause!) before the triumphant word '1T(X~OJL€V is 
appropriate and effective, as beforeEwc'Tpchov vt6e in the antistrophe. 
Other similar examples, where a rest is effective as emphasizing the 
word that follows, can be found in the epode of Olympian VI, in the 
strophe of Olympian XI, Pythian IV and Nemean V, where, in the first 
strophe, the syllable before the break is short, so that the rest is 
longer: 

, ,t).. ", ,... f3 (J!t:! • " ayvw't"a'T €'lT aV'Tac a JLWOC € C'Tao'T 

d..J..I1 elell d..JI..I~---1 el..J.l1 
I propose to use the term catalexis to describe this common feature 

3' In [Arist.] Prob. 19.15 we are told that earlier dithyrambs, unlike those of later times, 
were antistrophic: ai·T/.oJl ~ OTt '1"6 'lTa.\at6J1 at l>'f!tJ{Jf!pot lxopf!vOJl aVro{' 'lToMovc oW ay(rJJltCTUCWC 
~SnJl xa.\ar6J1 ~JI, WCTf! lJlapp.OJlta pl>.T) bfjSOJl. pna/3d.>.>..f!tJl yap 'lToMac Pf!'I"a/3o>.ac 'I"CP €v~ plfov ~ 
'ToLe 'lToMoLe, Ka~ 'TCP ayWJltc'Tfj ~ 'I"OLC '1"6 .q8oc tPvM.'TTovctJI. Std ti'IT>'ovc'Tf!pa l'ITolovv aVroic 'Ta PI.>'11. 
7} ~ uvrlcrpo,poc a'IT'\ow' f!rC pv8",oc yap lcrt Ka~ €V, ",f!TPf!tTru. 



186 CATALEXIS AND ANCEPS IN PINDAR 

of dactylo-epitrite verse, calling it dactylic cat alexis in this form, 
d .I .I I d .I .I I d --- I (sometimes with final short syllable, I .I ~ --- I), 
and epitrite catalexis when it takes this form, d.l I d ___ I (or 
d .I 1.1 ~ .... ).35 

When a transition is made from dactylic movement to epitrite (or 
from epitrite to dactylic) without catalexis, there can be no such rest, 
as, for example, in the strophe of Olympian III, where the singers con­
tinue without a rest as they sing: 

TvvSap{l,aLc TE cPLAogElvOLC &SE£V KaAAL7TAOKd.fLCfJ 0' tEAtv~ KAEL­

vw 'AKpd.YGW1"a YEpalpwv EVxofLaL.38 e~pwvoc 'OAVfL7TtOvlKcx,v 

d.l.ll d.l.ll ddl d.ll ddl d.l.ll d.l.ll ddl 
d.l.ll d.l.ll ddl d.ll ddl d.l.ll d.l.ll ddl 

The final syllable of the hemiepes, however, is often short, whether 
followed by epitrite or further dactylic movement as in Olympian 
VIII in the first strophe iva fLd.VTtEC /J.VSPEC 'fL7T1)POLC, and in the first 
epode Z7]v~ YEVEOAlcp' OC CE fLEV NEfLtc! 7Tp6cPaTov. Often too the final 
syllable of the non-catalectic epitrite is short, as in Isthmian II, first 
antistrophe, 

, 0 - , 'O',J.' l:' , apyvpw EtCaL 7TpOCW7Ta fLail CXK0'jJWVOf, cxot,uCXt. 

and pythian IV, fourth epode, 

fltAOC 'APTtfLL8oc O~PEVCE Kpcxmv6v. Jg &VtKd.TOV cPaptTpac.37 

In these passages the short syllable comes before a word division, 

85 In the Olympian dactylo-epitrite odes alone the follOwing examples of catalexis may 
be noted, in some strophes and epodes 'with short final syllable' or with hiatus before the 
break (the references are to the first strophe or first epode): 
Dactylic catalexis: 01. 3.11, 14; 6.1, 2, IS, 16, 19; 7.1, 4; 8.16, 17; 11.2,3, 15; 12.1, 2,4. 
Epitrite catalexis: Ol. 3.4, 12, 13; 6.18; 7.19; 8.4,5, IS, 17,20; U.5, 16, 17; 12.3. 

81 In the Snell-Maehler edition "finis periodi (pausa')" is marked after £iJxof/4', where 
word division is constant throughout the ode, but a rest or pause at this place would break 
the rhythm and is surdy inacceptable; the singers might take a quick 'catch breath' there, 
but in the third strophe the words Bopta .froXpov warn them clearly that there cannot be a 
break. Line-division in modem printed editions is always artificial and unsatisfactory, since 
it is impossible to avoid dividing some words between two lines, and it would really be 
more proper, though it would offend conventional taste, to print the poems without line 
,division,like music. 

87 The pattern of the second example, -uu ---u-u is much more common than 
,that of the first: -u-- -u-u. It has in fact been doubted whether the first pattern was 
admitted by Pindar. At Pyth. 4.180, where the codices read: 

&p,tp1 IIayyalov 8tp.JJAa JlIU£T40V'TfiC EpaJl. 



LIONEL PEARSON 187 

and a short rest, even if not specially appropriate, would be possible. 
But there is not always a word division after these short syllables, so 
that the easy solution which has been suggested in passages considered 
so far is not always possible. Sometimes a way might be found of tak­
ing a rest before instead of after the short syllable, as in Olympian 7.78 
ic-ra-rat Ttpvv(){wv apxayl-rC[., which could be sung: 

d .I I d t.l I d .I I d did .I I d ... , 

or Olympian 8.20-21 Jglv€1T€ Kpa-rlwv mi),C[. OOl..tX~P€-rf.LOV, which could 
be sung: 

d.l.ll d.l.ll d t.ll d.l.ll d.l.ll··· 

or Isthmian 1.2 1TpaYf.La Ka~ acxol..tac lJ1Tlp-r€pov, which could be sung: 

d.l.ll d.l.ll d t.ll d.l.ll··· 

One might even go so far as to suggest that Olympian 8.17 'AI..Ktf.Ll­

oovTa OE 1Tap Kpovov >"ocp~ was sung: 

dl d.ll ddl d.l.ll d.l.ll td.ll d ... 

Variations of this expedient can be applied to other passages, some­
times with results that are rhythmically quite attractive, but the 
opening phrase of Nemean V warns us that there is sometimes no way 
of finding a place for a rest: OliK aVOptaVT01TOtOC €lf.L' ••. This looks 
and sounds like a straightforward trochaic or iambic figure, and un­
less the audience were warned by <program notes' they would not 
suspect that -v-v was a substitute for -v--. It is the same with 
the opening of Nemean I, "Af.L1TV€vf.La C€f.Lvov 'Al..cp€ov, where there is a 
word division after -v-v. But the epitrite figure, -v--, seems 
clear enough in subsequent antistophe and strophe in both odes: 

modem editors have generally followed the judgement of Boeekh in his Notae Criticae 
(p.460): "Olp.£fJAa Pindarus seribere non poterat, quum lieeret OEp.IOAOLC, quod reposui haud 
eunetanter." He keeps 1Tpocw1Ta in Isth. 2, but suggests tentatively that 1TPOCW1TOV might be 
read. In Baechylides, on the other hand, we find (5.31-32) 

p.vpla 1Tcl.IIT~ KlAW80c up.E7"lpav apETcl.V, 

and the pattern is repeated in epodes 4 and 5. Cf Maas, Greek Metre §48, with the note added 
by Lloyd-Jones. 



188 CATALEXIS AND ANCEPS IN PINDAR 

EIC S~ Kp&vov Ka~ Z7}JI&c (Nem. 5.7), €l s' o¥ov ~ X€£pWV (5.18), 
&pxa~ 8~ p'pA7}V7'a£ ()€WV (Nem. 1.8), lC'Tav S' E-TT' aoA€ta£c ()tJpa£c (1.19). 

What, then, did Pindar expect his performers to do? Postpone the 
appearance of epitrite rhythm until the antistrophe? And is there a 
real difference between the first strophe and antistrophe? But why 
present them with such a puzzle? One might even ask why he did not 
write cTJ ap.vDv ap.1TV€vp.' , AAcp€ofi ? 

Unless we are prepared to say that Pindar is deliberately avoiding 
response between strophe and antistrophe or that the anceps has a 
recognized intermediate value between long and short (with the un­
desirable rhythmic consequences that this solution involves), we are 
obliged to say that in these passages a trochee is equal in time value 
to a spondee. And we can admit at once that (in musical terms) this 
need present no difficulty-not, at least, to modern performers, if 
they are told clearly, in print, what is expected of them. We must, 
therefore, try to explain, in a musical notation that we understand, 
what demands Pindar may be making of his singers. If, for example, 
they are to sing the opening of Olympian III: 

-vv -vv - - -v- - -vv 

they can be asked to sing: 

~d~~1 d~.l1 ddl~d~ddl~d~~1 
This will preserve the same value of long and short notes (d and~) 
throughout, and the spondee will keep the ratio of 4:3 in relation to 
the trochee. But they will find it much easier to sing: 

~ 
t d ~ J I d ~ ~ I d did..} did d I c:1 ~ J I 

What they are doing now is to make the trochee equivalent in time 
value to the spondee, although in singing triplets they do not trouble 
to ask themselves what exactly is the relation of each note in the trip­
let to the notes in the preceding measure.38 

38 Cf the good remarks of Emile Martin (an experienced choirmaster), I1p.cit. (supra n.7) 
42. Metricians of an older generation spoke of 'irrational spondees' when they appeared to 
be equal in time value to a trochee, saying they were spoken 'faster' than regular spondees 
-if four chronoi had to be compressed into three, a short syllable in an 'irrational' spondee 
would take only three-quarters of a chronos. Cf P. Masqueray, TraiU de metrique grecqu 
(Paris 1899) 10. 



LIONEL PEARSON 189 

Or suppose they are to sing the second strophe of Nemean I, which 
begins: 

- -v- - -v- - -v-- -vv -vv-

should they sing: 

(a) ~ did .I d did .I d did .I d d I ~ d .1.11 d .1.11 d, or 

""3"" r"'3' ~ 
(b) t did d did did d did did d did did .I .I I d .I .I I d ?39 ....... ....... ....... 
The advantage of (b) is that it gives the singer the chance to substitute 
a trochee for a spondee (a trochaic dipody instead of an epitrite, as in 
the first strophe of the ode), without disturbing the rhythmic pattern; 

~ 

he can sing d d d instead of d d (deliberately or by mistake) and 
....... 

still not be out of time. And it will not matter whether the dancers 
know in advance which version of the <epitrite' or the <anceps' the 
singers will be singing in any strophe, -v-- or -v-v, because 
they will be in time either way. And if the singers are dancing them­
selves, they can dance two even steps, a <spondee', while they sing a 
<trochee', and still maintain time, playing <two against three', feet 
against voice, just as a pianist learns to play <two against three', right 
hand against left.40 

38 One could produce the same effect by writing: 

(c) t .11.1 I .1.1 1.11 J.I 1.1 I .J.J Itd.J.I / d.J .11. 
222 

but musical editors, as well as performers, would probably prefer (b); musicians generally 
are more at home with the convention of triplets (three notes corresponding to two or 
four) than with duplets or quadruplets (two or four corresponding to three). In the lan­
guage of metrics, it is a question of whether the spondee is 'speeded up' to correspond with 
the trochee or the trochee is 'slowed down' to correspond with the spondee. When true 
trochaic rhythm (with no sign of an epitrite except at the end of a colon) alternates with 
dactylic, there is the other possibility of 'speeding up' the trochee so that two trochees 
correspond to one dactyl. Thus Alcman's a;aovc£ a' oplwv KOPvrPat T£ Kat .p4payy£c might be 
presented as: 

"3'. "3' "3' 
t d dld.l.l/ d .1.11 J oJ J.I .1.11 d d d ..... 1 ....... '-' ....... 

'0 Is this perhaps what Aristoxenus had in mind when he wrote Rhythm. 10 [W.], 292 

[M.J, p.ll [pighiJ): VOl1Tlov af xwpk Ta T£ ,..qv TOU 'l'TOaOC avvap.£v t/JvM.ccoVTa c7]p.£'ia Ka~ Tae 
tHrO rijc ;w8p.O'l'/'odae 'Y£vop.lvae 8£a£plc££c' Kai'l'Tpoc8£Tlov 8f TOtC £lPflp.IVl)£c, OTt TO: p.& lKaCTOv 
'l'TOaoe CflP.£'ia. S£ap.lv££ ica ovra Kai Tef! ap£8p.ef! Kai Tef! p.£-y18££, at a' ~o rijc ;w8p.o'l'Todac 'Y£v6p.£va£ 
a£a£ple££c 'l'ToM~v Aap.{3a.vovc£ 'l'To£KtAlav. leTa, Sf TOVTO Kai fv TOre l'l'T££Ta t/Jav£pov. Unluckily 
TO: l1T££Ta are lost. But he seems to be telling us that some performers may have to play 
triplets (trochees) while the conductor marks four beats (dactylic time), a situation with 
which modern orchestral players are quite familiar. 



190 CAT ALEXIS AND ANCEPS IN PINDAR 

What is perhaps more important for our purpose is that Aristox­
enus recognizes and describes clearly the different kind of metabole 
which takes place in version (a) and version (b). He says that one foot 
may differ from another in length (megethos) OTav Ta /.l.Eyl(J'Y} TWV 

7To8wv, &. KaTlxovcw Ol7T68€c. UVLca n. or in rhythmic type (genos), when 
the ratio is different, oTav ol A6YOL 8La~lpwCLv (XAA~AWV ol TWV 7To8wv, 
when, for example, one foot has the 2: 1 ratio, another the 2 :2. Or 
they may differ in diaeresis, oTav T6 at3'T6 /-dy€(Joc €lc uVLca J-Llp'Y} 8LaLp€(Jfj 

(Rhythm. 11 [W.], 298-99 [M.], pp. 23-24 [Pighi]). The difference, there­
fore, may be in both respects, in arithmetical ratio and length, or in 

f h " \',/..' I \' (J \ \ \ one 0 t ese respects, 'Y}TOL KaTa aJ-L'f'OT€pa. KaTa T€ TOV apt J-LOV Kat KaTa 
Ta J-L€yl(J'Y}. 7j KaTa (JuT€pa. He means that when the ratio changes from 
2:2 to 2: 1, from dactylic to trochaic, the trochee may occupy the same 
length of time (if the change is KaTa (Ju'T€pa) or a different length of 
time (if the change is Ka'Ta &J-L~6'T€pa). And it is the change <in one 
respect' which seems to be demanded by Pindar's practice, when 

~ 
trochee is substituted for spondee or dactyl, d ~ d instead of d d or -d .I .I. 

It follows, then, that the assumption which was suggested in the 
first paragraph of this article-that dactylo-epitrite is an alternation 
of 4-time and 3-time (or 4-time and 7-time)-is not correct, and it may 
be better to describe it, in modern musical terms, as written in 4-time 
with a liberal use of triplets. It also becomes unnecessary to use the 
term epitrite, which is discouraged by Aristoxenus-although, like 
Aristoxenus, we cannot fail to notice individual epitrite phrases. One 
can say now that the rhythm uses spondees, dactyls, and trochees, and 
that the trochee differs from the other two not in megethos but only in 
logos or arithmos. And at suitable places there will be catalexis, when 
the singer is silent for one or more chronoi, recovering his breath while 
his feet continue to move. 

Under this system it becomes unnecessary to make any use of the 
term syllaba anceps, though one must notice that a strophe may begin 
with an introductory note, on the upbeat, the length of which is not 
important, and that the final syllable of a period (when the dancers 
come to a standstill) may be brevis, longior, longa, or longissima (ad 
libitum). Within a period one should think not so much in terms of in­
dividual syllables but of feet and metra, and one should concern one­
self not with the absolute length of a syllable (which will certainly 



LIONEL PEARSON 191 

have varied as the omsic developed, with tempo rubato), but with the 
2: 1 ratio of long syllable to short, which in theory at least was a con­
stant ratio and was the rock on which Greek rhythmic theory was 
built. It was because of this principle, and because he could not accept 
the 4:3 ratio as rhythmic, that Aristoxenus refused to recognize the 
epitrite, despite the evidence of his ears. Rhythmic principle de­
manded that the epitrite, like the anceps, had to be explained away. 
It seems to me, at least, that Pindar's music is easier to understand 
without them, and that the demise of the anceps demands no 
lamentation. 

STANFORD UNIVERSITY 

January, 1974