## Kiichiro Itsumi

Print publication date: 2009

Print ISBN-13: 9780199229611

Published to Oxford Scholarship Online: January 2009

DOI: 10.1093/acprof:oso/9780199229611.001.0001

# Aeolic Phrases

Chapter:
(p. 24 ) 5. Aeolic Phrases
Source:
Pindaric Metre: The ‘Other Half’
Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780199229611.003.0005

# Abstract and Keywords

This chapter describes the typology of various aeolic phrases and their characteristics. The main topics are the aeolic base, the second position of reversed dodrans, catalexis, prolongation, and resolution. Statistics are provided so as to make clear the tendency of Pindar. Difference from the general tendency found in a particular verse provides the criteria by which stanza-forms are classified later in Chapter 8.

# A. The Basic Structure

Aeolic cola are characterized by the presence of one or other of two ‘asymmetrical’ phrases:

 —⌣⌣—⌣— dodrans (dod) —⌣—⌣⌣— reversed dodrans (rdod)

Both phrases have six positions, and both have a double short and a single short flanked by longs. The difference lies in the order. They are mirror images of each other.

From the structural point of view, these two basic phrases, the dodrans and the reversed dodrans, can be preceded by ‘aeolic base’ (hereafter called ‘full base’, ∘∘), or by single anceps (‘half‐base’, ×). 40

 no base —⌣⌣—⌣— —⌣—⌣⌣— dodrans (dod) reversed dodrans (rdod) half‐base ×—⌣⌣—⌣— ×—⌣—⌣⌣— telesillean (tel) heptasyllable (hepta) full base ∘∘—⌣⌣—⌣— ∘∘—⌣—⌣⌣— glyconic (gl) wilamowitzianum (wil)

## 1.The second position of reversed dodrans

The second position of the reversed dodrans is shown above as short. In the mainstream of Greek metre it is anceps (—×—⌣⌣—). But in Pindaric usage the position is hardly ever realized as long. Of 63 examples which contain the reversed dodrans, counting together the reversed dodrans proper and those prolonged by half or full base, none has the second position realized as long (i.e. ———⌣⌣—) consistently throughout all the repetitions. In most examples (56 phrases), a long syllable never appears and a short syllable always fills the anceps position. Even in the other seven, short anceps is dominant, and long anceps is often (but not always) caused by proper nouns. Apparently in (p. 25 ) Pindar, the long anceps is an occasional licence. But it must be noted that not all the long syllables can be eliminated as exceptional. There are a few examples which undoubtedly have anceps at the position, like aeolic cola in tragedy. These defy easy emendation, and attempts to emend them metri causa should be rejected. As I shall argue later, these examples are judged to belong to stanza‐forms in rigidly aeolic style (Class I). For more details, see E below.

## 2.Full base

‘Aeolic base’ consists of two ancipitia in Lesbian poetry. Thus the notation ×× is appropriate there, and a base in this form is a distinctive characteristic of aeolic metre. But in Pindaric usage, as in Attic poetry, the base is actually occupied by the following combinations of syllables: ——, —⌣, ⌣—, and ⌣⌣⌣. 41 For a base in this form, I use the symbol ∘∘.

There are considerable differences in the frequency of each form of the base. Pindar predominantly employs —⌣. Responsion between different forms is not entirely free, being admitted only between —— and —⌣ (therefore the notation —× is applicable for the scheme for particular passages), 42 but the other two forms (⌣— and ⌣⌣⌣) correspond only with themselves, except in one or two cases. So, while the notation ×— can be used for tragedy and elsewhere, it is not appropriate for Pindar. 43 This peculiarity of Pindar is shared without exception by all the phrases starting with the base. For the statistical detail, see C below.

## (p. 26 ) 3.Half‐base

The aeolic base can be reduced to a single anceps: ‘half‐base’. It can be —, ⌣, or ⌣⌣. The last form, ⌣⌣, stands in the same relationship to — or ⌣ as ⌣⌣⌣ at full base does to —— etc. Responsion is possible between — and ⌣, but not between ⌣⌣ and — or ⌣. I use the symbol × to cover all these forms (including ⌣⌣) for convenience’s sake. For the statistics, see D below.

## 4.Catalexis

The final ⌣— of the dodrans can be changed into —; this process is called catalexis. This is an a priori definition of catalexis, and it implies that the final position of, for example, pherecratean is not anceps but triseme (equal to three morae, ⨼). 44

The following are catalectic cola:

Catalexis is naturally not applicable to the reversed dodrans. 45

Some scholars posit phrases truncated yet further; for example:

∘∘—⌣⌣—

Incidentally, this form subsumes as one of its realizations the reversed dodrans (—⌣—⌣⌣—). The doubly truncated phrase can be dispensed with, so far as Pindaric examples are concerned: —⌣—⌣⌣— and ⌣⌣⌣—⌣⌣— are the reversed dodrans, while ⌣——⌣⌣— is acephalous e (∧e)+d. 46 It is not necessary to assume (p. 27 ) the existence of still shorter phrases, since ×—⌣⌣— and —⌣⌣— are freer D/e phrases (× d and d).

## 5.Prolongation

Just as phrases may be abbreviated (catalexis), so they may be prolonged. Prolongation is possible in both dodrans and reversed dodrans; for example,

∘∘—⌣⌣—⌣—— hipponactean (hipp)

is a glyconic prolonged by —. Presumably the value of this — is not triseme (nor true long) but anceps (or perhaps the long similar to the final long in the dactylic hexameter), comparable with that of link anceps in D/e, in Pindaric metre. Admittedly, the hipponactean could be seen as a catalectic form of ∘∘—⌣⌣—⌣—⌣—, but in practice this phrase is extremely rare, and is not found with a hipponactean as clausula. The same applies to the hagesichorean. Some phrases prolonged by — have no name. This type of phrase is designated by the notation +1, e.g. wil+1 (see the chart below). I analyse the aristophanean

—⌣⌣—⌣——

as dodrans prolonged by — (dod+1). The aristophanean is certainly the catalectic form of —⌣⌣—⌣—⌣— in Attic poetry, but it is not found with that function in Pindar. 47 Rather, one of the two examples functions as prolonged dodrans (dod+1):

O1e7 ⌣— —⌣— —⌣⌣—⌣— —⌣⌣—⌣——ǁ ∧e e dod ar

(p. 28 ) Beside prolongation by — (+1) there are longer ones like

 ×—⌣—⌣⌣—⌣— hepta+2 48 ∘∘—⌣⌣—⌣—⌣—— gl+3 (phalaecian)

Outside Pindaric metre, the ending ⌣—— can be taken as bacchiac. The name ‘bacchiac’ is not neutral since it inevitably implies ‘catalectic iambic’. ‘Phalaecian’ may therefore be less tendentious than gl+ba. But whether the last postion is created by catalexis or not is uncertain. It may be anceps like — (+1). I use the notation +3 without any implication. For the same reason, the ending ⌣— (+2) must be differentiated from the spondee following an aeolic colon.

In fact, we cannot always distinguish between catalexis and prolongation, although such a distinction may have been made in performance, i.e. audibly by music and visibly by dance. For example, some examples of —⌣—⌣⌣—— could in theory be rdod+1 (prolongation), but in practice we have no way to distinguish it from pherecratean with base in the form —⌣ (catalexis). The most delicate case is

N3e2 —⌣—⌣⌣— —⌣—⌣⌣——ǁ

where it is very tempting to suppose that reversed dodrans is repeated, with the second dodrans followed by anceps. Also, the pherecratean in the apparent priapean dicolon at O1s1 may not be pherecratean, but in fact rdod+1 (for the analysis of O1s1, see further 7. 5). On the other hand, so delicate a classification risks producing too many borderline cases. So hereafter I shall treat every example as pherecratean. The same is the case with rdod+2 (=gl) and rdod+3 (=hipp). 49

Below I set out all the theoretically possible forms. The forms in square brackets are the ones which can or should be analysed in different ways, and * is attached to forms which do not occur in the eighteen majors. (p. 29 )

 dodrans reversed dodrans prolonged by — —⌣⌣—⌣—— [—⌣—⌣⌣—— ] aristophanean (ar) [rdod+1 = ph] ×—⌣⌣—⌣—— ×—⌣—⌣⌣—— hagesichorean (hag) *hepta+1 ∘∘—⌣⌣—⌣—— ∘∘—⌣—⌣⌣—— hipponactean (hipp) wil+1 ⌣— [—⌣⌣—⌣—⌣— ] [—⌣—⌣⌣—⌣— ] [dod+2 = d ⌣ e] 50 [rdod+2 = gl] [×—⌣⌣—⌣—⌣— ] ×—⌣—⌣⌣—⌣— [tel+2 = × d ⌣ e] hepta+2 [ —⌣—⌣⌣—⌣—⌣— ] ∘∘—⌣—⌣⌣—⌣— *[gl+2 = rdod ⌣ e] 51 wil+2 ⌣—— [—⌣⌣—⌣—⌣—— ] [—⌣—⌣⌣—⌣—— ] *[dod+3 = d ⌣ e —] [rdod+3 = hipp] ×—⌣⌣—⌣—⌣—— ×—⌣—⌣⌣—⌣—— tel+3 hepta+3 ∘∘—⌣⌣—⌣—⌣—— ∘∘—⌣—⌣⌣—⌣—— gl+3 (phalaecian) wil+3

Although examples are very scarce—indeed only one, P8e7, and that a forced analysis—the following phrase (a type of dodecasyllable) is taken as a single aeolic phrase:

hepta+2+3 ×—⌣—⌣⌣—⌣—⌣——

The enneasyllabic phrase, hepta+2, is prolonged in the same manner as the glyconic in gl+3 by ⌣——. Its theoretical correlative

wil+2+3 ∘∘—⌣—⌣⌣—⌣—⌣——

does not in fact occur in the eighteen majors.

# (p. 30 ) B. Classification by Ending and Frequency

All the aeolic phrases are arranged below in a different synoptic chart according to the number of positions preceded by the choriamb. In the following chart the number of occurrences is given in the rightmost column:

 zero ending reversed dodrans —⌣—⌣⌣— 23 heptasyllable ×—⌣—⌣⌣— 12 wilamowitzianum ∘∘—⌣—⌣⌣— 15 +1 ending adonean —⌣⌣—— 2 reizianum ×—⌣⌣—— 9 pherecratean ∘∘—⌣⌣—— 11 *hepta+1 ×—⌣—⌣⌣—— 0 52 wil+1 ∘∘—⌣—⌣⌣—— 1 +2 ending dodrans —⌣⌣—⌣— 10 telesillean ×—⌣⌣—⌣— 29 glyconic ∘∘—⌣⌣—⌣— 49 hepta+2 ×—⌣—⌣⌣—⌣— 3 wil+2 ∘∘—⌣—⌣⌣—⌣— 4 +3 ending aristophanean —⌣⌣—⌣—— 2 hagesichorean ×—⌣⌣—⌣—— 1 hipponactean ∘∘—⌣⌣—⌣—— 2 hepta+3 ×—⌣—⌣⌣—⌣—— 2 wil+3 ∘∘—⌣—⌣⌣—⌣—— 2 [+4 ending] 53 +5 ending tel+3 ×—⌣⌣—⌣—⌣—— 4 gl+3 ∘∘—⌣⌣—⌣—⌣—— 3 hepta+2+3 ×—⌣—⌣⌣—⌣—⌣—— 1 *wil+2+3 ∘∘—⌣—⌣⌣—⌣—⌣—— 0

(p. 31 ) There are 145 phrases with zero or +2 ending against 40 with +1, +3, or +5. At the moment we should not jump to the conclusion that blunt aeolic endings are preferred to pendent, because phrases with zero ending and phrases with +2 can stand at the middle of the verse as well as at verse‐end and, consequently, have more chance of being employed. The frequency of blunt ending will be discussed at greater length later (8. A. 4).

# C. The Full‐Base Group

The following table consists of the same material as the table above, but in a different arrangement.

 pherecratean ∘∘—⌣⌣—— 11 glyconic ∘∘—⌣⌣—⌣— 49 hipponactean ∘∘—⌣⌣—⌣—— 2 gl+3 ∘∘—⌣⌣—⌣—⌣—— 3 wilamowitzianum ∘∘—⌣—⌣⌣— 15 wil+1 ∘∘—⌣—⌣⌣—— 1 wil+2 ∘∘—⌣—⌣⌣—⌣— 4 wil+3 ∘∘—⌣—⌣⌣—⌣—— 2

The full‐base group can be further divided according to the nature of the base.

 —⌣ ⌣⌣⌣ ⌣— —— —× ×— ⩊̲× (⩊̲—) 54 total gl 12 12 8 3 11 1 1 (1) 49 ph 6 1 1 3 11 hipp 2 2 gl+3 1 1 1 3 wil 9 1 2 3 15 wil+1 1 1 wil+2 1 2 1 4 wil+3 2 2 32 15 12 6 19 1 1 (1) 87

The four left‐hand columns after the stub indicate the number of the phrases in which exact responsion is rigidly kept throughout all the (p. 32 ) repetitions. There are in total 65 instances (32+15+12+6). 55 The proportion of these is remarkably high (65/87 = 74.7%). In other words, the anceps does not work as its name suggests. Another characteristic is the rarity of examples of —— (6 in total), compared with 31 of —⌣ and 15 of ⌣⌣⌣. It should be remembered that —— is the most usual form in tragedy. 56 In contrast, Pindar seems usually to have avoided the ‘heavy’ full base (——).

## 1.The base —×

Pindar’s preference for repeating the same form of full base is still more evident when one examines each repetition of the 19 examples of —× (for the terms ‘example’ and ‘repetition’, see Key). 57 Two peculiar groups emerge. First, in six cases —— occurs only in a limited number of repetitions, in most cases one repetition only, while the others are of the shape —⌣, and, moreover, all the —— without exception involve proper nouns. For example, at six out of seven repetitions of glyconic in I8s5c and I8s6, aeolic base is consistently occupied by —⌣, the exceptions being in s5c at v. 55c (= v. 56 Sn.) and in s6 v. 16 (= v. 16a Sn.). At both v. 55c and v. 16, Αἴγινα, a proper noun and a key word of I8, occupies the base ——, and consequently the scheme is given as —×—⌣⌣—⌣—. The same is true in the other four examples (P2e5, N4s5, N4s6, I7e5). Secondly, there are also some examples where a strong preference is observable, but where, nevertheless, there is a single deviation from the norm which (p. 33 ) does not involve a proper noun. Thus, at O9s2, P10s1, and P10e1, —⌣ is found in every repetition of the verse except one, but no proper noun is involved. This tendency is a manifestation of the ‘All‐but‐One’ rule (see Part III, C). At O9s5 and P8e3/4, —⌣ being the norm, —— is used more than once, in all instances but one with a proper noun. Conversely, at O9s8, —— is the norm, but —⌣ occurs once, without proper noun.

So in 65 examples the aeolic base is of identical form throughout and in another 12 examples it deviates from the chosen norm only very rarely. Yet Pindar does not always show the same attachment to a single form. In seven verses for which the notation —× is appropriate, no such strong preference is observable: O9s3 (3 longs/8 repetitions), P2s2 (4/8), P8s5 (4/10), P8e2 (2/5), N2s3 (3/5), N2s4 (2/5), N4s3 (8/12). In these cases the ‘anceps’ is really employed as anceps.

The situation is intriguing. A simplistic metrical rule is not appropriate. An explanation is to be sought not in a general metrical theory but in the examination of each metrical context; in other words, of the style of each stanza‐form as a whole. Interestingly, the occurrence of —× is concentrated in a limited number of stanza‐forms or odes: O9s (4 examples), P2e (1), P2s (1), P8s (1), P8e (2), P10s (1), P10e (1), N2s (2), N4s (3), I7e (1), I8s (2). Adding to them the examples of —— (6 in total), it is reasonable to suppose that Pindar employed ——, whether in responsion with —⌣ or not, only in some limited stanza‐forms. As will be argued later, the metrical context of these stanza‐forms is aeolic, not freer D/e. Thus the presence of —— is to be accepted as one of the criteria of Class I (aeolic) stanza‐forms.

Here emerges an important admonitory remark. It is dangerous to emend the text by introducing a word which has a long syllable at the second position of a glyconic (or equivalents), without consideration of the nature of the stanza‐form as a whole. Thus, for example, I am dubious about the transposition of πλαγχθǷντϵς at N7s8 (v. 37) proposed by Boeckh and accepted by most editors. 58

## (p. 34 ) 2.The base ⌣⌣⌣

Pindar uses ⌣⌣⌣ frequently as full base. He is perhaps the first Greek poet to introduce resolution in the aeolic base; the Lesbian poets and Anacreon did not use it, nor did Aeschylus except once. 59 It becomes common in later tragedy, especially in Euripides. Theoretically, ⌣⌣⌣ can be taken as a resolved form of —⌣ or ⌣—; but as a rule ⌣⌣⌣ does not correspond with the other forms, even in tragedy. The sole exception in Pindar is found at the beginning of P5e9:

P5e9 ⩊̲⌣—⌣⌣—⌣— —⌣⌣— ⌣ ⩊̲⌣— —⌣—ǁ gl de e

There, correspondence occurs between ⌣⌣⌣ ὕδατι (31) and —⌣ ὅϕ´ρα (62) and ϵὔχομαι (124) and —— σκυρωτάν (93). Perhaps the peculiarity of the style of P5 as a whole may be related to this irregularity (see Part II, ad loc.).

## 3.The base ⌣—

The base ⌣— is common neither in Pindar nor in tragedy. Of 12 examples in total, four are found in P2e, and two in I8s. This scarcity is partly explained by the fact that Pindar does not use glyconic or other phrases with full base κατὰ στίχον (P2e1a–4 are a rare exception). Anacreon and the dramatic poets are fond of the regularity produced by repetition, but with slight variation of the base. That is not Pindar’s manner. Thus O1s1 (but see below, 7. 6) and P8e6 are exceptional:

 O1s1 ⌣——⌣⌣—⌣— —⌣—⌣⌣——ǁ gl ph P8e6 ⌣——⌣⌣—⌣— —⌣—⌣⌣—⌣—ǁ gl gl

In other words, since the phrase ⌣——⌣⌣—⌣— and the like are not invariably employed next to an unambiguous aeolic phrase starting with full base, there are cases in which it is not easy to tell aeolic base ⌣— from acephalous e (⌣—). For example, the first phrase of

P5e2 ⌣——⌣⌣—⌣— ⌣ —⌣— —⌣—⌣⌣—ǀ

may be not a glyconic, but ∧e+dodrans followed by short anceps+e+rdod, since a number of verses in P5 start with ∧e (Part II, ad loc.). Similarly, (p. 35 )

N7s1 ⌣——⌣⌣—⌣— —⌣—⌣—ǁ gl e 2

may be ∧e+dod+e 2, taking into consideration that the resemblance between the phrases —⌣⌣—⌣— and —⌣—⌣— is fully exploited in this stanza‐form (see Part II, ad loc.). As for the wilamowitzianum, I analyse

I8s5a ⌣——⌣—⌣⌣— —⌣—⌣⌣— —⌣—⌣⌣—ǀ

as wil+rdod+rdod. But it may be ⌣— (∧e)+threefold —⌣—⌣⌣—. But at the same time it must not be forgotten that there are some examples of ⌣——⌣⌣—⌣— which are, judging from the context, most certainly glyconic: for example, N2s1, and three consecutive verses of P2e (e1b–3). It is impracticable to take all the examples of ⌣— as ∧e. For the ambiguity, see further 7. 4.

# D. The Half‐Base Group

 reizianum ×—⌣⌣—— 9 telesillean ×—⌣⌣—⌣— 29 hagesichorean ×—⌣⌣—⌣—— 1 tel+3 ×—⌣⌣—⌣—⌣—— 4 heptasyllable ×—⌣—⌣⌣— 12 hepta+2 ×—⌣—⌣⌣—⌣— 4 hepta+3 ×—⌣—⌣⌣—⌣—— 3 hepta+2+3 ×—⌣—⌣⌣—⌣—⌣—— 1

 — ⌣ ⌣⌣ × total tel 14 3 6 6 29 reiz 4 3 2 9 hag 1 1 tel+3 2 2 4 hepta 3 1 8 12 hepta+2 1 1 1 3 hepta+3 2 2 hepta+2+3 1 1 24 9 8 20 61

(p. 36 ) Generally speaking, long is preferred to short in half‐base: 24 examples against 9. 60 Especially when an aeolic phrase makes up a verse on its own, long is dominant. P10s2 starts with short anceps, and is the sole exception, while long anceps occurs at every repetition in nine phrases (= verses). Besides, at explicit anceps (×, 20 examples) where both long and short are freely employed, long is dominant. Every repetition being counted, long occupies the anceps position 74 times, short 60 times. In some verses the anceps strongly tends to be either long or short: e.g. N4s3 (long is used at 11 out of 12 repetitions; ‘All‐but‐One’ is a manifestation of extreme cases of this tendency) or N4s1 (2 out of 10). These verses, and others of N4, indicate that it is neither a metrical rule nor the metrical context that regulates the realization of the anceps. N4 is a very simple aeolic stanza (monostrophic), and its first three verses start with hepta or hepta+2. But the ratio of long to short in each verse varies considerably: 2/10 (s1), 9/3 (s2), 11/1 (s3). 61 This seems to be a matter of aesthetic preference in the individual context.

## 1.Position of half‐base in verse

Half‐base is used either at the beginning of a verse (in cases where the verse starts with the phrase in question) or in the middle (in cases where the phrase is preceded by others). The dominance of long is clearer at the beginning of the verse. (p. 37 )

 — ⌣ ⌣⌣ × total beginning 18 3 6 10 37 middle 6 6 2 10 24

Long anceps is exclusively used in half the verses (18 out of 37). However, there are 3 verses which invariably start with short anceps in all the repetitions: O9s10 (tel), P8e1 (hepta+2), P10s2a (hepta). At the 10 explicit ancipitia at the beginning of the verse, no strong preference is evident: long occurs in a total of 46 repetitions, short in 31. Just one verse has a high occurrence of short syllables: N4s1 (hepta+2, 10 short out of 12).

There are 6 examples of half‐base in mid‐verse where anceps is invariably short. All these phrases are preceded by…—⌣—⌣—, …—⌣⌣—⌣— or …—⌣⌣—, as the final part of the preceding phrase, so that the whole verse makes a sequence in which single short or double short alternates with single long without intervening long or anceps:

…—⌣—⌣—⌣—⌣⌣—…

…—⌣⌣—⌣—⌣—⌣⌣—…

…—⌣⌣—⌣—⌣⌣—…

These verses make a group of particularly Pindaric colour. The examples will be collected and discussed in I. 8.

## 2.Half‐base ⌣⌣

In 8 cola, ⌣⌣ occupies the base. All the cola are telesillean or its prolonged form (tel+3). I do not classify

O9e3, O13s1 ⌣⌣—⌣⌣——ǁ

as reizianum, nor the first phrase of

O1e5 ⌣⌣—⌣—⌣⌣— —⌣—⌣—ǁ

as ‘heptasyllable’ starting with ⌣⌣: O9e3, O13s1 is acephalous D and anceps, and O1e5 is ∧dd. 62

(p. 38 ) ⌣⌣ is never in responsion with —, let alone ⌣. Even outside Pindar, I do not know any certain example of ⩊̲—⌣⌣—⌣—. Whether the two double shorts in the phrase

⌣⌣—⌣⌣—⌣—

are equal, and if so, whether they are distinguishable from those of ∧D

⌣⌣—⌣⌣—

is an interesting question.

# E. The First Two Positions of Reversed Dodrans

As already demonstrated above (A. 1), there is no example which has ———⌣⌣— at all the repetitions. Thus reversed dodrans itself and phrases which include it are classified into three groups: —⌣—⌣⌣—, —×—⌣⌣—, ⌣⌣⌣—⌣⌣—.

 —⌣ —× ⌣⌣⌣ total dod 13 1 9 23 hepta 10 1 11 wil 13 2 15 wil+1 1 1 hepta+2 2 2 4 wil+2 4 4 hepta+3 2 2 wil+3 1 1 2 hepta+2+3 1 1 46 7 10 63

## 1.Reversed dodrans starting with —×

The form —×—⌣⌣—, of course, permits initial ——. Of the seven examples of —×, four (O9s6/7, N4s1, N4s2, N4s6) occur in those stanza‐forms where aeolic characteristics are most evident, which I shall call Class I. This tendency is obviously related to the fact that —— tends to be used at the full base in the Class I stanza‐forms (see C. 1 above). But even here, the repetitions with long anceps (p. 39 ) remain very much in the minority. 63 Besides these four examples, there remain a further three. In

P10s2a ⌣—×—⌣⌣—ǀ hepta

long anceps is found only at v. 8 (ἀμϕκτιόνων). If this word can be treated as a proper noun, then the irregularity would be mitigated. The sixth example,

P11s4 —×⩊̲⌣⌣— — —⌣⩊ —⌣⌣—ǁ rdod × e d

is most irregular in that not only is long anceps used in the second position, but there is resolution to the ‘left’ of aeolic nucleus in half the repetitions, and long medial anceps occurs after rdod. 64 However, even here, there are some limitations: in repetitions where resolution occurs, the preceding anceps is always short. Thus there is no example of ——⩊⌣⌣—, its avoidance agrees with the general rule that long anceps does not precede resolved long; cf. 6. C (iii). The last (P8e7) is a curious verse as a whole, and, whatever analysis is chosen, its colometry will inevitably be a forced one. I offer the following provisionally:

P8e7 —— ×—×—⌣⌣—⌣—⌣——ǁ sp hepta+2+3

The anceps position of the (assumed) reversed dodrans (hepta = × rdod) is filled with long at 3 repetitions out of 5. See Part II, ad loc.

## 2.Reversed dodrans starting with ⌣⌣⌣

Ten verses have resolution of the initial position of the reversed dodrans (⌣⌣⌣—⌣⌣—). 65 There is no phrase in which resolved position is in responsion with unresolved. Of the nine examples of resolved reversed dodrans, five stand at the beginning of the verse. One is tempted to ask whether reversed dodrans in this form sounded identical in performance with the first part of a glyconic with base in the form ⌣⌣⌣. (p. 40 )

⌣⌣⌣—⌣⌣—

⌣⌣⌣—⌣⌣—⌣—

We cannot answer this question since we do not know the time‐value of anceps. It is, however, a reasonable hypothesis, and one which I propose to adopt, that the two phrases did sound identical.

Compare the following verses, which make a spectrum:

 P2e4 ⌣⌣⌣—⌣⌣—⌣——⌣—ǁ gl e P6s3 ⌣⌣⌣—⌣⌣—⌣⩊̅—⌣—ǁ gl e N7e4 ⌣⌣⌣—⌣⌣—⌣⩊—⌣—ǁ gl e P2s7 ⌣⌣⌣—⌣⌣—⩊⌣——⌣—ǁ rdod e e O1e6b ⌣⌣⌣—⌣⌣——⌣—ǁ rdod e

The first two (P2e4, P6s3) are unambiguously gl+e. Following these two verses, I analyse N7e4 as gl+e, not as rdod+e 2. However, the next, P2s7, is not glyconic followed by long anceps+e, because there is no certain case at all of gle. 66 On the other hand, P2s7 is similar to O1e6b, where the first three shorts unambiguously belong to rdod.

The metrical context must be taken into consideration: reversed dodrans of the form in ⌣⌣⌣—⌣⌣— occurs in a limited number of stanza‐forms: O1e (1 example), P2s (2 examples), P2e (2 examples), P5e (2 examples), N6s (1 example), N7s (1 example), N7e (1 example). These are classified as Class II (freer D/e) or Class III (amalgamated), not as Class I (aeolic), except for P2e (an ambiguous case). On the other hand, the aeolic base in the forms —— or —× is a mark of Class I. It may not have been harmonized with resolved dodrans.

# F. Resolution

Apart from ⌣⌣⌣ at full base and ⌣⌣ at half‐base, a long position in aeolic phrase is sometimes filled by two short syllables. This can reasonably be regarded as resolution. 67 Every long position of dodrans and reversed dodrans can be resolved. For example, the following forms of glyconic are found: (p. 41 )

 Left of the choriambic nucleus N6e2 —⌣⩊⌣⌣—⌣—ǀ resolved Right of the choriambic nucleus P8s2 —⌣—⌣⌣⩊⌣—ǁ resolved resolved Final position resolved I8s5c —×—⌣⌣—⌣⩊̲ —⌣——ǁ P6s3 ⌣⌣⌣—⌣⌣—⌣⩊̅ —⌣—ǁ

Resolution must be attested by responsion with unresolved position. Examples of resolved long position corresponding with unresolved are found in four verses:

 P11s4 —⌣̄⩊̲⌣⌣— — —⌣⩊ —⌣⌣—ǁ rdod × e d N7s7 —⩊̲⌣⌣—⌣— ⩊⌣—⌣—ǁ tel e 2 P6s3 ⌣⌣⌣—⌣⌣—⌣⩊̅ —⌣—ǁ gl e I8s5c —⌣̄—⌣⌣—⌣⩊̲ —⌣— —ǁ gl e—

In the first two (P11s4, N7s7), resolution is found at the left‐hand long position of the choriambic nucleus (⩊̲⌣⌣—). Of these, the case of N7s7 is not surprising, since resolution involves a proper noun (v. 70 Εὐξένιδα), and may be taken as a special licence. But P11s4 is extraordinary. Of its eight repetitions, four have resolution, and none of the resolutions involves a proper noun (v. 9 θέμιν, v. 41 τὸ δὲ, v. 52 ἀνὰ, v. 57 καλλίονα). 68 In the other two (P6s3, I8s5c), resolution occurs at the final position of the glyconic, that is, of the dodrans (—⌣⌣—⌣⩊̲). At P6s3, the position is unresolved in only one repetition (v. 48 ἥβαν). And out of seven repetitions at I8s5c, three are resolved (v. 25c = v. 26 Sn. πινυτοί τϵ, v. 35c = v. 36 Sn. λϵχέων, v. 45c = v. 46 Sn. ἐπέων). Here the situation is similar to P11s4 in that resolved positions and unresolved are freely used in responsion.

There are 16 examples in which resolution is present at every repetition (i.e. responsion between resolved and unresolved is (p. 42 ) absent). Of these, 10 are the initial of reversed dodrans, mentioned in section E above. The others are: the final of the dodrans, N7s2, N7e3, N7e4; the left of the nucleus, N6e2 (gl); the right of the nucleus, P8s2 (gl), P11s2b (gl). In some verses it is difficult to decide which position is resolved. The sequence…⌣⌣⌣—⌣— is typically confusing. For example, see the phrases at the end of

 N7e3 —⌣—⌣⌣—⌣⌣⌣—⌣—ǁ N7e4 ⌣⌣⌣—⌣⌣—⌣⌣⌣—⌣—ǁ

These are resolved forms of either (i)…⌣——⌣— or (ii)…—⌣—⌣—. If (i) is chosen, the verses are glyconic+cretic (e). On the other hand, if (ii) is chosen, the verses are reversed dodrans+e 2. I choose (i) on the analogy of P6s3 above, but there is no certainty here; see further Part II, ad loc.

# G. Acephaly

Besides the phrases above, there is one more phrase that is probably aeolic, judging from its context:

O13s5 ⌣⌣—⌣— —⌣—⌣⌣——ǁ ∧dod ph

The first half of this verse is explicable as acephalous dodrans. Acephaly is rather common, as will be demonstrated, in freer D/e, but to what extent it occurs in aeolic cola in general is not certain. 69 At least there are no other examples in the eighteen majors. Acephalous dodrans (⌣⌣—⌣—) is familiar as a constituent of the third verse of the Attic skolion, but it is doubtful whether such an extra‐generic comparison is meaningful. Rather, the proper comparandum is the Simonidean poem 542 P. Here acephalous dodrans is most certainly attested. The poem is monostrophic, with a seven‐verse stanza: 70 (p. 43 )

 1 —⌣⌣—⌣⌣—⌣—⌣——ǀ 2/3 ⌣⌣— — —⌣—⌣⌣—⌣— —⌣—⌣⌣—⌣— × —⌣—ǀ 4/5 ⌣⌣—⌣— —⌣—⌣⌣—⌣— —⌣—⌣⌣—⌣—ǁ 6 ⌣⌣—⌣— —⌣—⌣⌣—ǀ 7 — —⌣— —⌣—⌣⌣—ǁ 8 ⌣— —⌣— —ǀ 9+10 —⌣—⌣⌣— × —⌣⌣—⌣——ǀ

Although there remain some uncertainties in the reconstruction of the strophe, it is certain that the phrase is situated at the beginning of the two verses 4/5 and 6. The basic structure of the former (4/5) is the repetition of —⌣—⌣⌣—⌣—. It is natural to extrapolate from it that ⌣⌣—⌣— is equal to —⌣—⌣⌣—⌣— minus the initial —⌣—. The latter (6) is almost identical with O13s5. These two Simonidean verses and O13s5 incorporate the palindromic movement (…⌣⌣—⌣— —⌣—⌣⌣…) found in O1s1 and elsewhere (see 8. A. 6 below).

Readers accustomed to Snell’s analysis may wonder why I exclude from aeolic consideration of ‘dactylic expansion’ (e.g. ∘∘—⌣⌣—⌣⌣—⌣—). There are, however, problems of ambiguity here, which I prefer to discuss following my definition of freer D/e; see below, 7. 6.

## Notes:

(40) The historical process is not taken into account here. Rather, the glyconic is the most common of all the aeolic cola in Pindaric metre as elsewhere.

(41) Some editors accept the responsion between ⌣⌣— and —— at N6e8. If the text there is correct, we must admit not only ⌣⌣— as a variation of aeolic base quite possibly unique in Greek poetry, but also the responsion between it and ——. This is outrageous. There is a notorious glyconic starting with ⌣⌣—: Aristoph. Ra. 1322 (πϵρίβαλλ᾽, ὠ̑ τέκνον, ὠλένας). Bacchylides 18 uses ⌣⌣——⌣⌣—⌣— in the environment of glyconics. I leave provisionally the transmitted text and the responsion ⩊̲—, but this does not mean that I accept it. See further, Part II, ad loc.

(42) But even this responsion is, in fact, restricted to Class I stanza‐forms. See below, 5. C.

(43) For the responsion in tragedy (with statistics), see Itsumi, ‘Glyconic’, 67–8.

(44) Cf. Parker, ‘Catalexis’, 15. Ancient metricians use the term ‘catalexis’ for mechanical amputation of the last syllable of a colon. That was a product of their conception of ‘final anceps’, which is now seen to be invalid. Cf. E. Rossi, RFIC 91 (1963), 52–71.

(45) Thus the latter phrase of the eupolidean dicolon

∘∘—×—⌣⌣— ∘∘—×—⌣—

is not the catalectic form of the former; contr. West, GM 95.

(46) Whether an actual phrase —⌣—⌣⌣— or ⌣⌣⌣—⌣⌣— should be taken (i) as the base (∘∘) + —⌣⌣—, or (ii) as reversed dodrans itself (of which the first position is resolved in the latter case) is, in a sense, a purely academic problem. A more important difference lies in the implication: if (i) is accepted, it means that ⌣——⌣⌣— is a correlative form of —⌣—⌣⌣— or ⌣⌣⌣—⌣⌣— and even that responsion between them is theoretically possible. There are 5 cases of ⌣——⌣⌣—, which do not correspond with other forms, in the eighteen majors (O1e2, P2e6, P5e5, P10s2b, P10s4). Judging from the metrical context and consistency with other related ones, it is better to take the phrase as acephalous e (∧e) + d (i.e. a freer D/e phrase); For acephale in general, see further 6, B and for the ambiguity between aeolic full base ⌣— and ∧e, see 7, (4). Thus I prefer (ii) to (i).

(47) Even in tragedy, aristophaneans are found repeated in synaphea; see Eur. Ba. 105 ff. cited in n. 28 above. Stinton supposes that link anceps occasionally occurs in aeolo‐choriambics (—⌣⌣—⌣—× in our case); at least, he admits the difficulties about this type of colon. See Stinton, ‘Pause and Period’, esp. ‘Postscript’, 64–6 (= Collected Papers, 358–61).

(48) This aeolic enneasyllable has been regarded as anceps plus glyconic by some scholars. This interpretation entails the unlikely supposition that there is a colon in Greek metre which starts with three ancipitia. And the so‐called ‘Barrett’s scheme’ (Hippolytos, Appendix I, p. 422) is refuted by Dale, LM 153 ff.

(49) Although the argument is circular, I am inclined to suppose that pherecratean in Class I stanza‐forms (aeolic) is really pherecratean (i.e. catalectic glyconic) but only seems to be so in Class III (amalgamated), being in fact rdod+1. So are other related phrases. I shall come back to this problem in the final chapter.

(50) The metrical contexts of actual examples refute the aeolic analysis of de and ⌣ de. They are collected at 8. C. 7.

(51) The analysis, rdode, is concordant with that of

—⌣—⌣⌣—⌣—⌣—⌣—

which is easily accepted as glyconic + link anceps (⌣) + e. Note that the anceps is always short in these cases. Examples are collected and discussed at 8. B. 5.

(52) There are two examples outside the eighteen majors: Pae6e10, e11.

(53) There is no phrase with +4 ending because I do not analyse phrases as aeolic that could theoretically be included in such a category. Thus —⌣⌣—⌣—⌣— in the list above I analyse as de, although it is equal to dod+2.

(54) The last form in parentheses, ⩊̲—, is based on a certainly corrupt text (N6e8); cf. n.41 above, but is included in the calculation.

(55) —⌣ (gl) P5s3, P8s2, P8e6, P11s2b, N3s4, N4s4, N6s1b, N6s2, N6e2, N7e3, I7e1, I8s4; (ph) O1s1, O1s4, O13s2, O13s5, P5e4 N3e2; (hipp) P2e8, N7s8 (wil) P5s2, P5e1, P6s1/2, P6s4, P6s5, N6s2, I8s1/2, I8s3, I8s4; (wil+2) N3s7; (wil+3) P10s5, N7s4.

⌣⌣⌣ (gl) P2e1a, P2e4, P6s1/2, P6s3, P6s6, P8s1, P11e4, N2s4, N3s3, N4s7, N7e4, N7e5; (ph) P11e2; (gl+3) N7e5; (wil) O9s6/7.

⌣— (gl) O1s1, P2e1b, P2e2, P2e3, P5e2, P8e6, N2s1, N7s1; (wil) I8s1/2, I8s5a; (wil+2) P2e8, P5s8

—— (gl) O9s4, N2s4, I7s5a; (ph) P8e3/4; (gl+3) I7s3/4; (wil+1) P6s7/8

(56) Itsumi, ‘Glyconic’, 67–8.

(57) —× (gl) O9s3, O9s5, O9s8, P2s2, P2e5, P8e3/4, P10e1, N2s3, I7e5, I8s5c, I8s6; (ph) P8e2, P10s1, N2s4; (gl+3) O9s2; (wil) N4s3, N4s5, N4s6; (wil+2) P8s5. The examples of the other two forms (×—, ⩊̲×) are too rare to be taken into account. For the detail of ×—, see Part II, O10s6. ⩊̲× (P5e9) will be illustrated later.

(58) The paradosis of v. 37 does not make good metre and must be emended; but Boeckh’s transposition is not as easy a solution as has been supposed. See further Part II ad loc.

(59) Itsumi, ‘Glyconic’, 68.

(60) * means that half‐base is used in the middle of a verse.

— (tel) O1e6a, *O9e4, O10e6, P11e3, *P2s2, P2s8, P5s7b, N3s1, N7s6, N7s7, N7s8, *I7s5a, I7e2, I7e3; (reiz) *O9s4, *O9s5, *O9s6/7, O9s9; (tel+3) N2s2, N4s8; (hepta) P2e7, P8s3, P11s2a; (hepta+2) P11s5.

⌣ (tel) O9s10, *I8s3, *I8s7; (reiz) *O1e4, *N4s4, *N4s6; (hag) *P2s8; (hepta) P10s2a; (hepta+2) P8e1.

⌣⌣ (tel) O9s1, *P2s4, P10s6, N3s8, N3e4, *N3e4; (tel+3) I7s1, I7e4.

× (tel) O9s2 (long at 7 repetitions out of 8), *P10e6 (1/4), *N4s5 (2/12), I7s2 (2/6), *I7s3/4 (1/6), *I7e5 (2/3); (reiz) *O9s3 (7/8), *N2s3 (2/5); (hepta) *O9e8 (2/4), P8s6 (4/10), P8e2 (3/5), P10e3 (2/4), P10e4 (3/4), N4s2 (9/12), N4s3 (11/12), *I7s5a (5/6); (hepta+2) N4s1 (2/12); (hepta+3) *O9e8 (3/4), P10e5 (3/4); (hepta+2+3) *P8e7 (3/5).

(61) Or 10/2 (s2) and 12 /0 (s3), if the initial syllable of ὁ χρυσός (v. 82) is scanned long, and if Οὐλυμπίᾳ(v. 75) is read instead of Ὀλυμπίᾳ.

(62) Some of the telesilleans, ⌣⌣—⌣⌣—⌣—, may be interpreted in a different manner. See 7. 2.

(63) O9s6/7 (3 long out of 8), N4s1 (2/12), N4s2 (3/12), N4s6 (1/12). Moreover, four of these long ancipitia are in proper nouns.

(64) Resolution at aeolic nucleus is discussed below, 5. F. For the mid‐long anceps see 6. D.

(65) (rdod) O1e6b, P2s2, P2s7, P2e2, P2e3, P5e3, P5e6, N6s3, N7e2; (wil+3) N7s4.

(66) An aeolic phrase is, in general, not followed by a long anceps. There are only two exceptions (P10s6, P11s4). See below, 6. 3.

(67) Resolution of longs in aeolic verses is rare in Attic drama; cf. Itsumi, ‘Glyconic’. It is almost restricted to later Euripides. The unique example of Aeschylus is Cho. 317: τύχοιμ᾽ ἂν ἕκαθϵν οὐρίσας (⌣—⩊⌣⌣—⌣—). It worries many (e.g. Dale, BICSSuppl. 21.2 (1981), 15: ‘resolution in 317 very ugly and hardly possible in Aeschylus’), but it may be accepted by referring to N6e2 and, especially, P11s4. It is noteworthy that P11 treats the myth of Clytaemestra and Orestes. Perhaps Aeschylus is influenced by Pindar. Further, Cho. 315 = 332 (the first verse of the same strophe) may betray another influence of Pindar. —⌣⌣—⌣⌣—⌣— is, most certainly, not glyconic in Aeschylus (see Itsumi, ‘Glyconic). Rather it may be related to P11s1 (for this verse, see 7. 6 below).

(68) For the avoidance of long anceps preceding the resolution, see 6. C above.

(69) Note that O13s is the only stanza‐form where the metre shifts from non‐D/e to D/e in the middle. The alternative interpretation is: ⌣⌣ + e (⌣⌣ being substituted for anceps). See Part II, ad loc. Outside the eighteen majors, ∧dod occurs at Parth1s1; see Part II, Appendix B.

(70) I follow Page in the reconstruction of the text, but the colometry is my own. West, GM 66 gives an analysis different from both Page’s and mine.