Bryn Mawr Classical Review 02.03.15

Carolyn Higbie, Measure and Music. Oxford: Clarendon Press, 1990. Pp. viii + 231. ISBN 0-19-814387-7. $49.95.

Reviewed by Harry R. Barnes, University of Texas at San Antonio.

[[Greek transliteration: Readers of e-bmcr are reminded that this is still an imperfect medium. This review has more or less the TLG Beta system; others that pass on the net will have other representations. There are difficulties: comments and advice welcome.]]

The study of enjambement in Homeric epic and later Greek poetry has received sporadic attention throughout the 20th Century, beginning with the technical studies of Bassett. Since Parry's (1929) comparative study there has been general consensus that the enjambement styles of oral and literate poetry differ in some respects from each other, and that the differences are attributable to the oral poet's greater dependence upon the hexameter line as the structural frame for formulaic composition. With only one exception, Lyding (1949), all previous enjambement studies have been based upon samples of the epic(s) rather than the whole. Most of these studies have been primarily statistical in nature and have not dealt with the important aesthetic questions of how and why the poet uses enjambement in the creation of specific lines and what compositional factors may impel the poet to divide formulaic phrases between lines. Higbie's study, based upon her analysis of enjambement types in the whole Iliad, makes a valuable advance by combining both statistical and aesthetic approaches. In this respect it will serve as a model to those engaged in technical studies. In presentation and argument, however, the book is less effective. Higbie fails to make explicit important definitions and assumptions, for example, how she defines a sentence and distinguishes it from an independent clause (Chapters 2-4); how she determines the "primary" form of formulas upon which modification occurs (Chapters 5-6). Higbie' s presentation of her argument in some cases hinders comprehension: there are no internal subdivisions within chapters to guide the reader; certain chapters (3 and 4) do not present a clear program at the outset, and occasionally, in following digressions and the abundant and extended discussions of specific examples, one may lose the thread of the encompassing argument. In sum though, while Higbie's presentation of her arguments is not entirely successful, this is an important book for Homeric studies and oral studies generally because of the terrain that it covers and the standard that it sets for meticulous analysis. In the remainder of this review I will evaluate Higbie's approach and conclusions, indicating those that are particularly helpful or problematic. Chapters 2-5 will be the focus here since they represent the core of her research, Chapter 6 having been provided principally as an illustration of her techniques applied.


In the Introduction Higbie informs the reader that her original intention was to "provide as complete a picture of enjambement in the Iliad as possible, based on a close reading of the entire poem and presented statistically, book by book" (1). Soon realizing that types of enjambement were related to sentence and clause break within the verse, she proceeded to gather statistics on the location of verse-internal sense breaks and to relate them to those on enjambement types in order to provide a fuller picture of the poet's technique of verse construction. After a review of pertinent literature on enjambement and formulaic composition (Chapter 1), this statistical study occupies Chapters 2-4. In Chapter 5 Higbie lays aside the statistical method and explores the ways in which formulas may be modified when their elements fall in different lines. The types of modifications considered include: mobility, rearrangement of word order, expansion, separation (including enjambement of the constituent words). Higbie employs Hainsworth's terminology and generally follows his approach closely, but she does not limit herself, as he does, to noun + epithet formulas. While special attention is paid to formulas in enjambement, Higbie's study is wide-ranging and covers a variety of metrical and grammatical factors that tend either to localize formulas or to promote mobility and flexibility. Among these factors, Higbie cites certain cases of preferred word-order, but primarily invokes patterns of localization of metrical word-types, making extensive use of O'Neill's statistics on word-localization. In Chapter 6 Higbie illustrates her method of analysis on a brief passage containing at least one example of each category of enjambement (Il. 15.292-331).

Chapter 2: The Relationship between Verses Defining and Classifying Types of Enjambement.

Most enjambement studies to date have classified types of enjambement either according to the three categories set out by Parry (no enjambement, unperiodic (adding), necessary) or to those of Kirk, who retains Parry's first two categories but subdivide s necessary into periodic, integral and violent. Higbie follows Kirk's model, but further subdivides the progressive and periodic types into internal and external, depending upon whether the enjambing portion of the sentence falls within the same clause (internal) or adds a new clause (external). Anyone who attempts to classify enjambement according to any of these systems soon realizes that, not infrequently, lines defy such neat categorization and, in the absence of clear guiding principles, it is virtually impossible to avoid subjectivity and inconsistency.1 The particular value, then, of Chapter 2 lies in Higbie's detailed discussion of numerous marginal cases and her clear explanations of the choices that she makes in classifying them. Such a basic discussion has been mostly lacking in previous studies and will greatly benefit those working on this topic. This does not mean that the element of subjectivity has been eliminated or that all will agree unanimously with her classifications.2

How difficulties arise in determining enjambement types can be seen in Higbie's statement: "The primary factor in determining enjambement is the degree of expectation of or grammatical need for what follows the verse end" (29). Degree of expectation and grammatical need are not the same, in fact, they are occasionally incompatible factors and the necessity of choosing between them can lead to inconsistency in interpretation. In most cases Higbie uses mere grammatical completeness as the determinant of enjambement type. For example, when a subject is omitted or held over in enjambement Higbie counts it as unnecessary grammatically since "the subject is implicit in the verb form and need not be overtly stated; semantically, it may be necessary to avoid confusion, but the structural requirements of the Greek sentence are complete without it" (38). On these grounds she classes both 15.397 and 12.162 as type 1a (adding internal). In the former Patroclus is the subject of the preceding narrative, hence the poet has no need to repeat his name, in fact, to do so would be ungainly: 1a is clearly the correct choice. In the latter, however, there is no noun or pronoun in the immediately preceding narrative that could stand as the subject of the verbs in 162, hence there is no possible way of predicting the subject of the sentence at the verse end. Certainly both poet and listeners realize that the sentence is incomplete and must continue in 163. (Cf. 7.277-8, where two heralds have been introduced and the listener cannot tell which one EI)=PE/ TE MU=QON until he hears the name in 278 [166].) In this case Higbie simply overrules the factor of expectation in favor of that of grammatical completeness. Similarly, lines 4.460 and 4.502 are classed as type 1a, even though the enjambed subject AI)XMH/ is necessary, if not for the grammatical construction, at least for the sense of the verb PE/RHSE(N) in the preceding lines and certainly for that of the definite article H( in 4.502 (36).3 (Cf. 13.521-2, classed as type 1a (101), and 20.201-2, classed as type 3 (96), for inconsistent treatment of verbs requiring some continuation, either infinitive or direct object, for the sentence to make reasonable sense.) More puzzling is the classification of 6.319 as type 1a, where the substantive (AI)XMH/), upon which the genitive (DOURO/S) depends, falls in 320 (114). Although these sentences are all complete grammatically at the verse end, on the basis of contextual requirements (expectation) one would count all of the enjambements discussed so far as necessary, but this may not be the best solution.

A similar problem arises later in the discussion of the verb XRH/ divided by enjambement from a complementary infinitive (51). Higbie argues that, because XRH/ can be used without an infinitive (or genitive noun 1x) complement, when the complement falls in the next line the enjambement type is adding (type 1). The expression OU)DE/ TI/ SE XRH/ by itself means "You shouldn't" or "You have no need (for)." Used by itself in 19.420 it makes perfect sense and there is no expectation that anything else need follow; similarly in 7.109-10 the genitive complement may be considered unnecessary because the sense "You shouldn't" is prepared satisfactorily by A)FRAI/NEIS. The listeners would understand the meaning of the clause at the line-end and it is reasonable to consider TAU/THS A)FROSU/NHS, from the listener's perspective, as continuing the sentence in adding enjambement (type 1a). Lines 9.496-7, on the other hand, are different in that OU)DE/ TI/ SE XRH/ alone stands as a virtual non sequitur after the imperative: A)LL' A)XILEU=, DA/MASON QUMO\N ME/GAN: "Achilles, stifle your anger. You shouldn't." The elliptical prohibition neither clarifies nor qualifies Phoenix' command and the listeners could not help but wonder what Achilles shouldn't. While the clause is complete grammatically, the sense of the sentence is not, and here the effect of continuing with the infinitive is not at all similar to that of adding enjambement. Since elsewhere in the Iliad XRH/ takes an infinitive complement 24 out of 28 times, one assumes that, when the sense of XRH/ alone was ambiguous or incomplete, the audience would have expected the infinitive and, conversely, found its omission jarring. Interestingly, Higbie notes here that the inflexion of the singer's voice on XRH/ would help alert the audience to the structure, i.e., that there was more to follow. (I note that Cantilena also considers the enjambement in 9.496 to be type 3 [47]. Cf. similar examples in 5.633-4 (51) and 22.108 (97), which Higbie also classifies as type 1a.)

If Higbie followed the principle of "grammar before expectation" consistently then we might, despite reservations, chose to establish this as the rule to follow in future studies. That she does not is seen in her determination of clausal enjambement (type 2) in 15.187-92 (45). I do not think that I will be alone in maintaining that, if the mention of Poseidon's portion in 190 is sufficient to create the expectation that the other two portions must also be related (and I do not think that it is sufficient), and hence the enjambement is clausal rather than adding, surely a verb whose subject cannot possibly be deduced from the preceding context creates an expectation of an entirely greater order. Higbie's choice results in confusion over which factor, grammar or expectation, is to have precedence and in what situations.

The problem is not simply one of inconsistency but rather is inherent in this system of classification. Higbie appears to recognize this in her comment on 4.502-3: "None the less, 503 is also considered to be in adding enjambement with 502 because the definite article can be used as a substantive; it would hardly seem appropriate to characterize the enjambement as violent, the only other option" (37, emphasis mine). The criterion that Higbie applies here, that of minimal grammatical completeness, was first adopted by Parry and has been followed in subsequent studies. While it does provide a reasonably consistent standard for classification, ironically, it is a thoroughly text-based approach and bears little relation to the way the audience apprehended the poem and even less to the way in which the oral poet composed it. The point is not academic. The enjambement in the verses discussed above that Higbie classifies as type 1a is actually a composite type: by the text-based, grammatical criterion the enjambement is type 1a (adding), but in the poet's mind as he composed the verse it is type 3 (necessary), in some cases even type 4 (violent) e.g. 4.502-3 (36). All previous comparative enjambement studies have demonstrated that literary hexameter is differentiated from its oral predecessors by a significantly higher degree of necessary and/or violent enjambement. It is, therefore, reasonable to speculate that this composite type, conceptually "necessary," will also be more prevalent in literary hexameter poetry. If this is the case, then classifying these enjambements as type 1a cannot help but blur the distinction between oral and literate composition. Quite possibly this problem would have become apparent if Higbie had undertaken a comparative study of early and late hexameter. Even if this suspicion does not hold up, such enjambements represent a clear problem for classification and it would be worthwhile to determine exactly how frequent they are. The simplest solution would be to establish a separate category for them. On the other hand, I suspect that a more useful approach would be to adopt a simplified version of Lyding's categories, which describe enjambements by grammatical types, upon which a modified version of Higbie's classifications could be overlaid.4 Higbie dismisses Lyding's method because the categories are too numerous and in some instances redundant, hence not "functional," by which I presume she means not easily amenable to statistical presentation (10-11). Yet such an approach would produce the same statistical data as Higbie's, assuming that a relatively simple system of enjambement categories was maintained, but these categories would be supplemented by the grammatical classification. This would reduce the elements of inconsistency and subjectivity and would have the additional benefit of providing data for a thorough analysis of the effects of word-order preferences upon enjambement, a topic that is of some interest to Higbie in Chapter 5 (see also 48ff.).

The remainder of Chapter 2 describes necessary (type 3) and violent (type 4) enjambement with numerous illustrations. Most frequent among the violent enjambements are those in which a new clause begins at the bucolic diaeresis and clause introductory material (conjunctions, particle chains and adverbs) is divided from the body of the clause in the next verse.5 Here Higbie restricts the cases of leading adverbs causing violent enjambement to "sentence adverbs, that colour the meaning of an entire clause rather than modify only the verb, an adjective, or another adverb" (52-3); other leading adverbs are classified as creating necessary enjambement (type 3). The reason for such restriction is not made clear and appears arbitrary in view of the fact that cases of an adjective preceding the modified noun in the following verse are classed as violent enjambements. Do isolated adverbs create less grammatical need or expectation than adjectives? One could also make a case for classifying cases of dependent infinitives preceding their governing verb in the next line as violent rather than necessary enjambements. Higbie does not do so, but in Chapter 5 we learn that the word-order, finite verb precedes infinitive is standard in the Iliad, and that there are only eleven instances in which both the order is reversed and the words fall in separate lines (158-9). "Naked" infinitives surely involve grammatical need and expectation equal to those created by unattached adjectives and sentence adverbs.

Chapter 3: Interpreting the Statistics

In Chapter 3 Higbie comments upon the results of her complete statistical analysis of enjambement in the Iliad. Her findings generally corroborate those of earlier sample studies when adjustments are made for the different classification schemes. For example, comparing Higbie's statistics with those for Parry's 600 line sample (Tables 3.1, p. 82; 3.3, p.86)

Parry; 0: 48.5; 1: 24.8; 2: 26.6
Higbie; 0: 38.8; 1a: 21.3; 2-4: 24.5
1b: 15.4
we find an alteration in the frequency of lines with no enjambement (type 0) and those with adding enjambement (type 1). This reflects Parry's broader definition of the sentence: "I define the sentence as any independent clause or group of clauses introduced by a coordinate conjunction or by asyndeton" (253). Higbie, on the other hand, considers most clauses introduced by a coordinate conjunction as continuations of the sentence in adding enjambement. Therefore, most of Higbie's type 1b would be counted by Parry as having no enjambement (type 0). Following Parry's definition would result in the 15.4% of Higbie's type 1b being divided between her types 0 and 1 (most given to the former) and I suspect that the correspondence between her results and Parry's would then be close. Parry's type 2 (necessary) includes all of Higbie's types 2-4 and here their results correspond. It seems to me that, since Higbie does not intend to pursue a book by book analysis of enjambement but is interested in the relationship between genre and enjambement types, it would have been more interesting and useful to present the statistical results, not by book (Tables 3.2 and 3.3), but by genre: narrative, speech, catalog, battle, simile, etc.

A troublesome problem that comes up in Chapter 3 and again in Chapter 4, one which Higbie should have addressed clearly in Chapter 2, is how she distinguishes between a clause and a sentence. What, if any, attention does she pay to the punctuation of the OCT? Since the determination of enjambement types 0 and 1 (Chapter 3) and the statistics on internal sentence boundaries (ISB) and internal clause boundaries (ICB) (Chapter 4) both depend upon a distinction between sentence- and clause-break, lack of clear definitions is an egregious omission. In Chapter 1 Higbie comments approvingly upon Parry' decision to ignore punctuation marks in defining the sentence (8) and later iterates this view: "Part of the difficulty lies in determining where sentence and clause breaks occur. Modern punctuation of the Greek text is a dangerous guide" (90). Yet in Chapter 3 it becomes clear that her own practice in this regard is inconsistent. In some instances she follows the OCT punctuation: "The punctuation of different editors affects the frequency of such unenjambed verses, so numbers and patterns will vary, though only slightly... Sometimes clauses with parallel construction are punctuated as separate sentences for no apparent reason, as in 1.436-9 , but the punctuation has been accepted in this analysis" (68). On the other hand, KAI/ = "and" is always considered connective, despite the OCT punctuation, which sometimes counts it as a sentence starter (68). In Chapter 4, OU)DE/ ME PEI/SEIS (11.648) preceded by a comma is treated as a clause, elsewhere (6.360, etc.) preceded by a colon it is a sentence (94). Similarly she accepts the erratic OCT punctuation in determining sentences or clauses after W(/S (E)/)FAT(O), etc. in speech conclusions (97). From the limited number os examples cited it is impossible to determine the extent to which this inconsistency affects the statistics reported in Chapters 3 and 4, but the absence of a clear distinction between sentence and clause and inconsistency in the treatment of punctuation call into question the accuracy and usefulness of these statistics.

In Chapter 3 Higbie also raises the question of the relevance of enjambement to arguments for the oral composition of a text. While hers is not a comparative study, Higbie asserts that her statistics on enjambement will make it possible "to comment upon the relative orality of the Iliad" (66). It turns out, however, that she agrees with Janko (quoting him): "enjambement is a negative test only, [because] the 'oral style' of enjambement can be used by literate poets." Therefore (quoting Higbie), "a low frequency of necessary enjambement is consistent with oral composition but does not require it" (81). I would suggest that in this case Higbie is overly cautious: the indisputable fact that literate poets can use "the oral style of enjambement," hardly means that this is the habitual practice of any one of them. As Higbie shows in subsequent chapters, oral poets can also manipulate their diction in complex ways, some of which may be indistinguishable from those of literate poets. The difference is seen in aggregate samples and, until someone produces a large sample of literate hexameter that eschews complex enjambement, there is no evidence to dispute the conclusions of earlier studies that there is a direct correlation between composition by the line (no enjambement) and adding enjambement, and oral / formulaic composition.6

Chapter 4: Verse Internal Boundaries

In Chapter 4 Higbie examines sentences that begin and/or end in verse-internal positions. In such cases the verse may contain either two complete sentences or (more frequently) one complete sentence beginning or ending at a verse boundary and one partial sentence; or the sentence may be "skewed," the term Higbie applies to sentences that both begin and end in interior positions, almost always involving enjambement of the sentence but, rarely, contained within one verse. Having established these distinctions, Higbie discusses the poet's use of the shortest sentences, those filling two feet or less. Many examples are cited. Higbie's interesting discovery here is the connection between short sentences and genre: short sentences are frequently used in speeches to convey excitement and strong emotion (92-99). Higbie goes on to demonstrate the correlation between the location of internal sentence boundaries (ISB) and enjambement types (see Table 4.4 & 4.5 p.133). The figures show that when a sentence begins earlier in the verse than the bucolic diaeresis the poet most frequently ends the sentence with the verse-end (type 0) or continues in adding style (type 1). On the other hand, when a new sentence begins at the bucolic diaeresis, necessary and violent enjambement are far more likely (roughly 5x that of types 0 and 1). From this she concludes that: "there appears to be metrical pressure to avoid sentences shorter than half a verse" (99) and later: "There is metrical pressure for a sentence to be at least half a verse in length" (104). The evidence presented, however, does not support the claim that "metrical pressure" (which is not defined) has any effect upon sentence length. There are, after all, only five syllables available to the poet between the bucolic diaeresis and the verse end and complete sentences, even clauses, of this length are bound to be rare in any poetic form, oral or literate, regardless of meter.7

Higbie continues with a discussion of the four possible metrical shapes of sentence beginnings in interior positions (100-4). While this would seem relevant to the topic of verse-internal boundaries, having read this passage several times I am still at a loss to find a clear connection. Various examples of clause-beginning words and phrases are discussed from the point of view of their mobility within the verse. These examples, the reader is informed, will lead to a consideration of the "hierarchy of forces which contribute to the shape of the lines" (104), referring, presumably to the observation that certain name + epithet combinations are more highly localized at the verse-end (the example cited is QEOI=S E)PIEI/KEL' A)XILLEU= -- but where else could it appear -- straddling the mid-line caesura?!) than certain sentence beginning phrases like OU)=D' A)/RA PW/ TI. The discussion is then tied up ring-composition style with a second reference to the supposed "metrical pressure for a sentence to be at least half a verse in length" (104). Here the detailed scrutiny of textual examples within an argument that is only tenuously connected to its surroundings is apt to leave the reader disoriented. There follows an apparent digression on various theories of the origin of the hexameter (104-5) that might better have been incorporated in the review of the literature (Chapter 1).

Higbie goes on to discuss verses with two internal sentence boundaries and "skewed sentences." The two types are similar in that each involve sentences that neither begin nor end at verse boundaries; considering their infrequency, the former might be considered an unenjambed subclass of skewed sentences. Verses with two ISB's are rare (only eight examples) and all fall in speeches. Adducing numerous formulaic parallels, Higbie finds nothing particularly unusual about these eight verses except for their rarity and their confinement to speeches. She suggests that this type of verse reflects the high emotion of the speaker and I find this to be the case in about half of the examples (112).

Skewed sentences (those involving enjambement) occur in a variety of contexts throughout the Iliad, but are, perhaps, associated more with battle scenes than with other genres (118). Among sentences of this type the ratio of necessary and violent enjambement to adding enjambement (320 : 389 = 1 : 1.2, Table 4.9, p. 137) is higher than that in the Iliad as a whole (3034 : 5760 = 1 : 1.8, Table 3.2, p. 84), which is not surprising since almost 3/4 of these sentences begin in the second half of the verse (Table 4.8 p.136). Higbie's treatment of skewed sentences raises several related fundamental questions. First, are sentences (whether "skewed" or not) in which necessary enjambement occurs less formulaic than those that are either unenjambed or continued in adding style? Internal and external factors suggest that they might be: the disposition of words within a necessarily enjambed sentence may conflict with more frequent patterns of localization of formulas. Within the development of hexameter poetry, the correlation between higher frequencies of necessary enjambement and literate composition could suggest that sentences with necessary enjambement are either less formulaic or depend more upon formulaic modifications than their unenjambed counterparts. Having established some objective correlation between enjambement type and degree of formulaic content within the sentence, one would be in a better position to investigate whether skewed sentences as a type, or perhaps only those involving necessary enjambement, are less formulaic than "non-skewed" sentences. Higbie claims that skewed sentences are "constructed of normal formulaic material," but two of her examples, including the longest, are described as lacking formulas (10.513-4, p. 115) or containing few formulas (21.241-8, 118f.).

Higbie's discussion of the stylistic effects of skewed sentences would also benefit from further refinement. She asserts that in battle narrative "they underpin the picture of violence, motion, and confusion which the poet seeks to paint" (118; for examples in other contexts see p. 120). Presumably, in deviating from the more frequent pattern of close sentence to verse correspondence, the poet disorients the audience and thus, by means of a stylistic device, enlivens his depiction of tumultuous slaughter. The problem is that the effect she describes will more likely result from the use of necessary enjambement within the skewed sentences than from continuation in adding style.8 For instance, in the example cited as typical of a skewed sentence in battle narrative (16.412-13, p. 118), the clause beginning at the masculine caesura of 412 is complete at the verse end but the sentence continues in adding style with a prepositional phrase filling the next line up to the masculine caesura. This is the commonest form of enjambement in the Iliad and hardly seems to create a dissonance between the patterns of sentence and verse, or otherwise to obscure the metrical pattern of the verse. On the other hand, the disorienting effect that Higbie describes may well be created by the necessary enjambements in her next example, taken from Achilles' fight with the river (21.241-8, 118f.). Therefore, these sentences should not be treated as one phenomenon but rather separately according to their distinct enjambement types.

Chapter 4 closes with a consideration of internal clause boundary (ICB). As the table on p. 122 demonstrates, single ICB occurs in the generally the same positions as single ISB and in roughly the same proportions. In the absence of adequate distinction between sentence and clause there would seem to be little point in attempting to draw further conclusions from the relatively minor deviations in frequency between the two types.9

Chapter 5: Formulae and Enjambement:
Questions of Word Order and the Definition of "Formula"

In chapter 5 Higbie leaves behind the statistical approach of the previous chapters in order to concentrate upon "the relation between enjambement and formulae," in an attempt "to try understand whether it is unusual for a formula to be enjambed, and to explain, wherever possible, the reasons for the enjambement." In fact, the focus of this chapter is on various types of formulaic modifications, enjambement being one accident that may occur to the formula in the course of modification. Her presentation is based upon a study of all instances of necessary and violent enjambement in the Iliad, in which she collected examples of enjambed expressions that also occur within a single line. The types of formulaic expressions collected in this way are varied but include primarily verbal expressions.10 Her assumption is that the unenjambed versions are the "primary" formulas from which the enjambed versions are derived by secondary modifications. She cites numerous examples of modifications that may lead to enjambement between elements of formulas. Rather than attempt to discuss all of the examples that Higbie adduces I will first state what I see as the three conclusions that emerge from Chapter 5 and then comment on two important questions that are raised along the way, first the effects of word order on the positions of words in the verse and, secondly, problems related to terminology and the definition of "formula."

Three conclusions can be drawn from Chapter 5: (1) Proper noun + epithet formulas, especially the longer types filling the second half of the verse from the mid-line caesuras, are relatively immobile compared to verb formulas. (2) Because individual verbs have a great variety of metrical forms, verbal formulas are less subject to localization than are noun + epithet formulas; i.e., they are relatively mobile. (3) As a direct result of 1 and 2, when two formulas "compete" for the same metrical position, the mobile types either move or are modified to suit the occasion. Occasionally such adaptation results in the enjambement of parts of formulaic phrases that otherwise fall within the same verse.

A. Patterns of word order

Some types of formulas resist modification, others change their forms freely. The former class is inflexible and immobile, or "localized." Higbie comments on various factors that tend to localize formulas, including, circumstances of use, metrics (bridged positions in the verse and the length of the formula), the grammatical structure of the formula, and word-order (155). Thus, various factors influence the positioning of words in the hexameter and it is rarely possible to isolate one alone as responsible for any particular word position. For example, Higbie suggests two patterns of word-order that may affect the positioning of words in the verse: (1) finite verb precedes (proper noun + epithet) and (2) finite verb precedes infinitive. In the case of the first pattern, one wonders initially why it should be limited to proper nouns. But remembering that most of the proper noun + epithet formulas documented by Parry are designed to fill the metrical space between various caesural points in the second half of the verse and the verse end, and that the positions before the mid-line caesuras and between the mid-line caesuras and the bucolic diaeresis are convenient for many verb forms, it is hardly surprising that proper noun + epithet expressions are most frequently preceded by their verb.11 As Higbie's examples show, this pattern holds true regardless of the case of the noun (assuming it does not change its metrical shape), hence of its grammatical relationship to the verb (157). One suspects, therefore, that metrical and stylistic factors are principally responsible for localizing proper noun + epithet formulas at the verse-end. The metrical shape of the components together with the poet's demonstrated aversion for necessary enjambement will cause the verb to precede the proper noun + epithet expression in most cases. Here meter is an important variable that obscures the role (if any) of word order. The most effective way to demonstrate this hypothetical pattern of word order would be to test freely mobile nouns (perhaps those without epithets) and verbs for the frequency of the pattern: verb precedes noun (either subject or object).

Higbie's second pattern, on the other hand, finite verb + infinitive can be better supported. Since verb forms take on a variety of metrical shapes and are relatively mobile as a group, one has to contend less with the metrical force of localization. At the same time the effects of localization cannot be dismissed entirely, and here they may lead to an overestimation of the force of word order. Several classes of infinitives, those ending in a long penult before the terminations -NAI, -SAI, -SQAI, -EIN, are virtually restricted to the end of the verse by their metrical shapes. 12 Higbie observes, for example, that the infinitive MA/XESQAI falls at the verse end in 86 of 91 occurrences (171). Once again it appears that meter complicates the question of word-order, since, for the same reasons adduced above in the case of verb + (proper noun + epithet) formulas, infinitives at the verse end will regularly follow finite verbs without regard to rules of word-order. Even granting, as I think we should, that finite verb precedes infinitive is a logical and natural pattern for this semantically cohesive word-group, it is unclear how great a priority the poet places upon maintaining that pattern when the two elements fall in the same line.13

B. Definition of "Formula"

Do formulas have a "primary" configuration from which all their other forms are derived by modification? If so, is that form the simplest form? These questions are fundamental to our understanding of the manner of composition. Higbie assumes an affirmative answer to both questions -- too easily I believe -- and in so doing needlessly complicates the poet's task. The reader will find a number of questionable applications of these principles in Chapter 5. Treating them all in detail could easily produce a work longer than the chapter itself, so one case in point will serve for all:


According to Higbie, in these lines the formula "O)/SSE + KALU/PTW has been split by NU/C, moved, and enjambed, further it has been conflated with three other formulae, NU/C + ME/LAINA, NU/C + KALU/PTW and TW\ DE/ OI( O)/SSE" (164).

B.1. The formulaic model for TW\ DE/ OI( O)/SSE / ... NU\C E)KA/LUYE

In its simplest configuration the formula O)/SSE + KALU/PTW is localized at the verse end, most frequently in a formula for dying, TO\N DE\ SKO/TOS O)/SSE KA/LUYE(N) (4.461, etc., 11x). The expression TW\ DE/ OI( O)/SSE, "his two eyes," and several similar expressions are also localized at the verse end (10x); in all cases a new clause begins at the bucolic diaeresis and the main verb falls in the next line. The frequency of the expression and the repetition of the metrical and syntactic structures give evidence for a formula that we may designate as: TW\ DE/ OI( O)/SSE (vel sim.) / + verb. But if these expressions are formulaic in their own right, it hardly seems meaningful to insist that TW\ DE/ OI( O)/SSE / KA/LUYE is a formulaic modification of O)/SSE KA/LUYE(N). Does the cumbersome process of splitting, moving, conflating, enjambing, supposedly necessary for the creation of this expression, correspond to the poet's actual technique? Did the poet first recall O)/SSE KA/LUYE(N), localized at the verse end and run it through the modifying permutations, or did he simply invoke the ready-made model TW\ DE/ OI( O)/SSE / + verb employing KA/LUYE for the verbal variable?

B.2. Is NU/C + KALU/PTW a formula?

The noun nuc occurs 18 times in the nominative in the Iliad and, like most nouns in the Iliad, it is frequently modified by an epithet (11x with / 7x without). NU/C is used 6 time with the verb KALU/PTW in the expression: "night (death) covered (his eyes)," i.e., "he died." In all six cases NU/C is modified by an epithet, but NU/C also occurs once sans epithet with the verb KALU/PTW (10.201). According to Higbie, "NU/C + KALU/PTW is a formula which stands alone just once in the Iliad and regularly is conflated with noun + epithet formulae, such as NU/C + E)REBENNH/" (164). Thus NU/C (alone) + KALU/PTW is taken to be the "basic" form of the expressions yet, while it is the simplest form in that it involves only the two adjacent words, it is also the unusual case and is never repeated as such -- one might even hesitate to call it a formula. It is hardly apparent that the expressions E)REBENNH\ NU\C E)KA/LUYE, etc., arose from conflating a basic formula, NU/C + KALU/PTW, with noun + epithet formulas.

There is a further complication that Higbie doesn't notice: NU/C has two meanings: "night" and, metaphorically, "death." NU/C meaning night may or may not be accompanied by an epithet, but whenever the poet intends death, NU/C is always modified by an epithet emphasizing darkness and accompanied by a form of KALU/PTW (5.310, 5.659, 11.356, 13.425, 13.580, 14.439, 22.466). From the poet's persistence in applying the epithet in these expressions, one senses that it is functional, emphasizing the darkness of death and thus signalling to the audience the metaphorical use of NU/C (cf. the alternative formula for dying: TO\N DE\ SKO/TOS O)/SSE KA/LUYE[N] 4.461, etc.). In 10.201, the example that Higbie cites as the supposed basic form from which this formula group was derived by modification, NU/C means night, not death. It is the only instance of NU/C + KALU/PTW in which NU/C means night. Note further that the poet could easily have used NU\C E)KA/LUYE ME/LAINA in 10.201 after O(/TE, but instead he altered it, intentionally omitting the epithet in order that the audience not be confused into thinking that Hector had recently died. Thus, NU\C E)KA/LUYE in 10.201 appears to be a modification of the formulas with epithet rather than the "primary" form of them.

B.3. The formulaic model for NU\C E)KA/LUYE ME/LAINA

The question raised here is similar to that in B.2. Higbie notes that NU/C + ME/LAINA "is a flexible formula that can be inverted, split, and moved," further asserting that, "in its primary form -- unsplit -- it appears only in the oblique cases." Thus NU\C E)KA/LUYE ME/LAINA results from the conflation of two basic formulas, NU/C + KALU/PTW and NU/C + ME/LAINA, the latter having been split by the verb, the former having been shifted from its accustomed position at the verse end. It would appear that the poet has performed a feat of considerable mental dexterity, and, perhaps to point toward that conclusion, Higbie notes that the location of E)KA/LUYE in 14.439 is most unusual for words of its metrical type: only 2.4% of words shaped u u - u occur in position 3.5.

We will never know whether the single occurrence NU\C E)KA/LUYE ME/LAINA is an ad hoc creation or an underrepresented formula. It feels like a formula, but the unusual placement of E)KA/LUYE, violating Meyer's law, suggests that this is a verse-end formula transposed to the first half. Corroborating this view are occurrences in the second half-line of the formula: NEFE/LH E)KA/LUYE ME/LAINA (17.591 = 18.22 = Od. 24.232, cf. Il. 10.210). The structure of the two expressions can be expressed as a formula-type: noun shaped (u u) - + E)KA/LUYE ME/LAINA. The adjective ME/LAS can be applied to a variety of nouns, both those dark in color (GAI=A, NHU=S, etc.) and those metaphorically dark (KH=R, FRH/N, etc.). In 14.439 the poet has taken advantage of the possibility of inserting the noun NU/C into a familiar structural prototype (i.e., composition by analogy; see Parry 319-21, Ingalls 211-26). The new expression was designed to fill the first half of the line (at least its metrical shape fits better there) and therefore the expression was transposed there from the prototype's more frequent location in the second half. That the poet was not deterred from violating Meyer's Law indicates that he thought of the expression as a unit (i.e., a formula), rather than of a group of individual words derived from "primary" formulas, split, shifted, conflated, etc. There is nothing, in fact, to show that the "primary" formulas (MELAI/NHS NUKTO/S, etc. occurring only in oblique cases, p. 164) hold primacy in the poet's mind or that the poet refers to them at all in creating this expression. By adhering rigidly to Hainsworth's categories and ignoring the other formulaic material available to the poet, Higbie arrives at an explanation of the poet's technique that is unduly complicated and bears little resemblance to actual compositional process: in this case, the poet has relied upon simple and apt use of analogy rather than cumbersome modification and enjambement of "primary" formulas.

In Chapter 5 the reader will find a number of similar problematic cases in which Higbie claims to have discovered a "basic" or "primary" formula from which enjambed versions are derived by "Hainsworthian" modifications. Among the questionable formulas that Higbie adduces are: BA/LLW + DOURI/ (167-9), NU/SSW + E)/GXEI+ (169-70), OU)TA/ZW / OU)TA/W + DOURI/ (174-6). Here I will refer the reader to a more persuasive treatment of expressions of this type by Bakker and Fabbricotti (unavailable to Higbie before publication), who demonstrate that dative expressions for "spear" are most frequently "peripheral" elements, combined with the verb ("nucleus") for metrical reasons.14 These dative expressions should not be conceived as bound formulaically to specific verbs of wounding as Higbie's formula-models suggest, rather, they are applied to variety of verbs for metrical purposes in the style of generic epithets.


This book is more important for what it attempts to do and for the questions that are raised in passing than for what it actually accomplishes. The methodology is flawed by: (1) various problems in the classification of enjambement types; (2) the failure to provide a definition of the sentence that distinguishes it functionally from an independent clause; (3) the failure to justify the concept of "primary" formulas. While these are serious problems that should be rectified in future studies, still, this book is most useful in a variety of ways. The clear discussion of enjambement classifications will facilitate the work of others in this field. Higbie is at her best in the many detailed discussions of specific textual examples, whether analyzing enjambement types or explicating formulaic content. Higbie's merging of statistical and aesthetic approaches has great potential benefit for oral-formulaic studies. Finally, Higbie has raised important questions, not only about the poet's use of enjambement, but also concerning the poet's compositional technique in the many cases that are not immediately explained by obvious formulaic manipulations and in which the distinction between oral and literate styles appears to blur.

Works Cited

  • Bakker, E., and Fabricotti, F., "Peripheral and Nuclear Semantics in Homeric Diction: the case of dative expressions for 'spear,'" Mnemosyne 44 (1991) 63-85.
  • Barnes, H. R., "Enjambement and Oral Composition," TAPA 109 (1979) 1-10.
  • Cantilena, M., Enjambement e poesia esametrica orale: una verifica (Ferrara 1980).
  • Clayman, D. L., "Sentence Length in Greek Hexameter Poetry," Quantitative Linguistics 11 (1981) 107-36.
  • Hainsworth, J.B., The Flexibility of the Homeric Formula (Oxford 1968).
  • Ingalls, W. B., "The Analogical Formula in Homer," TAPA 106 (1976) 211-26.
  • Janko, R., Homer, Hesiod and the Hymns: Diachronic Developments in Homeric Diction (Cambridge 1982).
  • Kirk, G. S., "Studies in Some Technical Aspects of Homeric Style," YCS 20 (1966) 75-152.
  • Lyding, E., Homeric Enjambement (Diss., Bryn Mawr 1949).
  • McLennan, G.R., "Enjambement in the Hymns of Callimachus," Hermes 102 (1974) 200-206.
  • O'Neill, E. G., Jr., "The Localization of Metrical Wordtypes in the Greek Hexameter," YCS 8 (1942) 105-78.
  • Parry, M., "The Distinctive Character of Enjambement in Homeric Verse," in The Making of Homeric Verse (Oxford 1971) 251-265.
  • Porter, H. N., "The Early Greek Hexameter," YCS 12 (1946) 3-63.


  • [1] Janko (1982) points out the discrepancy between various scholars' enjambement statistics for Greek hexameter texts (32).

  • [2] Cantilena's recent monograph compares enjambement patterns in various texts (Iliad 9, Odyssey 12, Homeric Hymns, Batrach, Callimachus' Hymns) and contains a similar, though less detailed, discussion of enjambement classifications. His classification scheme allows for a wider variety of periodic enjambement (type 2), interprets as necessary (type 3) certain enjambements that Higbie counts as adding (type 1), and is somewhat more restrictive in the category of violent enjambement (type 4). The two authors' statistics are not presented in such a way as to be readily comparable, but I note that in his list of enjambement types by line (43-49) Cantilena assigns a different category to seven of the forty examples from Iliad 9 listed in Higbie's Index Locorum (226): 111=4(H)/3(C); 259=4/3; 338=1a/2; 340=1a/3; 496=1/3; 625=1/3; 678=4/3.

  • [3] AI)XMH\ XALKEI/H serves as the subject of PE/RHSE and as the understood direct object of PH=CE, both in 4.460 and in 4.502. Higbie classifies the enjambement in these two lines as type 1a because, "there is no hint that this subject is to be appended by the end of 460." Elsewhere, however, whenever the verb PH/GNUMI (both transitive and intransitive forms) is used in the context of piercing with a weapon, the poet is consistent in indicating the weapon, most frequently in the same clause but sometimes in an earlier clause or sentence. Similarly, with the verb PERA/W the weapon is always expressed. Thus, while the two lines in question are grammatically complete, to classify the enjambements as type 1a ignores the factor of expectation, derived from the poet's consistent practice of explicitly stating the wounding weapon whenever he uses these verbs.

    In the case of 4.502, Higbie agrees that the meaning of the sentence is unclear at the verse end because of the lack of any plausible referrent for H(: "the article, though it fulfils the grammatical wants of the clause, cannot complete the meaning, be cause the article refers to a person or object as yet unnamed and thus not predictable" (36). Both poet and audience know that the clause is incomplete and will expect the delayed subject to fall in the next line. Nonetheless, Higbie overrules the factor of expectation (in this case certainty) and classifies the enjambement as type 1a. Clearly the system of classification is inadequate if violent enjambements like this must be grouped together with the frequent adding type.

  • [4] In any case future enjambement studies should include a complete line by line list of enjambement types in order to raise the level of discussion above statistics.

  • [5] Kirk and Cantilena classify these enjambements as necessary (type 3).

  • [6] On the basis of his samples of the Iliad, Odyssey and Callimachus' Hymns Cantilena finds insufficient difference in all but violent enjambement to support this contention. There are, however, methodological problems: (1) he does not test his statistics for significance but decides arbitrarily that differences of at least 5% are significant (20); (2) his sample of only two books of the epics seems a bit skimpy to support any conclusions, especially considering that previous studies (McLennan and Barnes) have reported significant differences between the enjambement styles of Homer and Callimachus. On the other hand, Cantinela argues persuasively that there is a qualitative difference between Homer's and Callimachus' use of complex enjambements, the former being mitigated or masked by formulaic composition (31-34).

  • [7] According to Clayman's data, the average sentence in the Iliad contains 29.3 syllables and only 7.4% of its 8,407 sentences contain 10 or fewer syllables. Clayman follows the OCT (Munro and Allen, 1920) punctuation and defines a sentence as: "a string of characters and blanks terminated by a strong stop, a weak stop, or an interrogation mark, which in modern texts are indicated by a low dot, high dot and a semicolon" (111).

  • [8] Kirk notes that the fifty line section of Iliad 16 that is highest in necessary enjambement (301-350) depicts "man-took-man fighting." The pace of this section is "rapid, urgent, and heavily broken up by internal stops. Strong overrunning of the verse-end is often caused by stops at the bucolic caesura" (129). Elsewhere concerning the same passage: "Yet this passage reminds us that integral enjambement, at least, is often associated with internal stop" (115).

  • [9] A startling fact that emerges from the table, one that would seem to have implications for the study of the colometry of the hexameter (though precisely what they are is still unclear), is the relative infrequency of syntactic breaks at position 7, Fraenkel's alternate C caesura. Almost half of the verses in Porter's 1000 line sample have word-end in position 7, yet sentence and clause break here is quite infrequent compared to that in the other interior positions between 2 and 8 in which word-end is frequent. This can be seen by comparing Porters figures for the percentage of verses with word-end in these locations (from his Tables 10-19, 57ff.) with the totals of Higbie's figures for the percentage of ISB and ICB in the same locations (Table, p. 122):
    Positions% word-end (Porter)% ISB + ICB (Higbie)252.324.7361.828.05 49.952.85.560.834.7745.48.2862.842.3
    While one would not necessarily expect a direct correlation between the frequency of word-end and syntactic break, the exceptional deviation observed in position 7 needs further study.

  • [10] Higbie's understanding of "formula" is broad and inclusive, following Hainsworth's definition, "a repeated word-group" (see p. 18).

  • [11] And conversely, the relatively few proper noun + epithet formulas that are localized at the beginning of the verse are apt to be followed by their verb. Higbie cites examples of *DARDANI/DHS *PRI/AMOS (restricted metrically to positions 5 and 7) which confirm that the metrical shape of the formula was more important than word order in determining localization (158). Higbie cites Parry's work but appears not to appreciate how it complicates her argument (157).

  • [12] According to O'Neill's tables the frequencies of the relevant metrical shapes in position 12 are: u - - (92%), u u - - (96%), - u u - - (100%). The actual number of occurrences in position 12 of infinitives with the following metrical shapes are: u - - (340x), u u - - (89x), - u u - - (43x).

  • [13] That the poet feels some inclination in this direction is clear: for what it's worth, a quick test of forms of dunamai + inf. in the Iliad produces 54 cases in which the infinitive follows the verb against only 3 cases of the opposite word order.

  • [14] Peripheral elements are: "(i) as neutral as possible with respect to their context, and (ii) metrically variable." Their meaning "is subservient to the ultimate goal for which they are used by the poet, the metrical extension of their nucleus" (3).