369. The quantity of a syllable—that is to say, the time which it takes in pronunciation—may be determined either by the length of the vowel (or vowels) which it contains, or by the character of the consonants which separate it from the next vowel sound. In ancient technical language, the vowel may be long by its own nature (φύσει), or by its position (θέσει).
The assumptions that all long syllables are equal, and that a long syllable is equal in quantity to two short syllables, are not strictly true of the natural quantity in ordinary pronunciation. Since every consonant takes some time to pronounce, it is evident that the first syllables of the words ὄφις, ὀφρύς, ὀμφή, ὄμβρος are different in length, and so again are the first syllables of Ὦτος, ὤτρυνον. Again, the diphthongs ῃ, ηυ, etc., are longer than the single vowels η, ω, etc., and also longer than the diphthongs ει, ευ, οι, ου. In short, the poetical "quantities" must not be supposed to answer exactly to the natural or inherent length of the syllables. The poetical or metrical value is founded upon the natural length, but is the result of a sort of compromise, by which minor varieties of quantity are neglected, and the syllables thereby adapted to the demands of a simple rhythm.
It has been shown, however, that the general rule of position rests upon a sound physiological basis. "The insertion of a consonant may be regarded as equivalent in respect of time to the change of a short vowel into a long one." (Brücke, Die physiologischen Grundlagen der neuhοchdeutschen Verskunst, p. 70; quoted by Hartel).
370. Position. The general rule is that when a short vowel is followed by two consonants the syllable is long.
Regarding this rule it is to be observed that
- Exceptions are almost wholly confined to combinations of a mute (especially a tenuis) with a following liquid. But even with these combinations the general rule is observed in the great maiority of the instances.
- Most of the exceptions are found with words which could not otherwise be brought into the hexameter: such as Ἀφροδίτη, Ἀμφιτρύων, βροτῶν, τράπεζα, προσηύδα, etc.
- The remaining exceptions are nearly all instances in which the vowel is separated by diaeresis from the following consonants.
Il. 18.122 καί τινᾰ Τρωϊάδων
Il. 24.795 καὶ τά γε χρυσείην
The chief exceptions in Homer are as follows1
τρ: in Ἀμφιτρύαν, ἐτράφην (Il. 23.84—but see the note on § 42 in the Appendix, p. 390), τετράκυκλον (Il. 24.324), φαρέτρης (Il. 8.323), Ὀτρυντεύς (II. 20.383-4); and in ἀλλότριος (unless we scan -ι͜ος, -ι͜ου, etc.).
Before τράπεζα, τρίαινα, τρίτη (τριήκοντα, etc.), τραπείομεν (τράποντο, προ-τραπέ- σθαι, etc.), τράγους, τροποῖς, τρέφει (Od. 5.422, 13.410), τροφοῦ (Od. 19.489), τρέμον (Od. 11.527).
Before a diaeresis, καί τινα Τρωϊάδων (Il. 18.122).
πρ: in ἀλλοπρόσαλλος (Il. 5.831); before προσηύδα, πρόσωπον, προΐκτης, πρόσω, and other compounds of πρό and πρός (προκείμενα, προσαΐξας, etc.); also before πρὸς ἀλλήλους, πρὸ ἄστεος, and one or two similar phrases (cp. ll. 13.799, 17.726).
Before Πριαμίδης (Il.), πρίν (Il. 1.97 οὐδʼ ὅ γε πρίν κτλ., cp. 19.313, Od. 14.334, 17.597); πρῶτος (Od. 3.320, 17.275), προσφάσθαι (Od. 23.106).
κρ: in δακρύοισι (Od. 18.173), δακρυπλώειν (Od. 19.122), ἑνέκρυψε (Od. 5.488), κεκρυμμένα (Od. 23.110).
Before Κρονΐων, Κρόνου παῖς, κραταιός, Κραταιΐς, κράτος μέγα (Il. 20.121), κράνεια, κρυφηδόν, κραδαίνω, κρατευτάων, κρεῶν.
Add Il. 11.697 εἵλετο κρινάμενος; Od. 8.92 κατὰ κρᾶτα (κὰκ κρᾶτα?), 12.99 δέ τε κρᾱτί.
βρ: in βροτός and its derivatives, as ἀβρότη, ἀμφίβροτος; also before βραχίων.
δρ: in ἀμφι-δρυφής (Il. 2.700), and before δράκων, Δρύας, δρόμους. Also Il. 11.69 τὰ δὲ δράγματα (unless we read δάργματα, as Hartel suggests).
θρ: in ἀλλόθροος (Od. 1.183, etc.), and before θρόνων, etc., and θρασειάων. Also in ll. 5.462 ἡγήτορι Θρῃκῶν.
φρ: in Ἀφροδίτη: and Od. 15.444 ἡμῖν δʼ ἐπι-φράσσετʼ ὄλεθρον. Cp. Hes. Op. 655 προπεφραδμένα.
χρ: before χρέος or χρέως (Od. 8.353), and in ll. 23.186 ῥοδόεντι δὲ χρῖεν, Il. 24.795 καὶ τά γε χρυσείην.
τλ: in σχετλίη (Il. 3.414), which however may be scanned - -.
κλ: in Πάτροκλε (Il. 19.287), ἐκλίθη (Od. 19. 470—should perhaps be read ἑτέρωσε κλίθη), προσέκλινε (Od. 21.138, 165—read perhaps πρόσκλινε or ἔκλινε) and before Κλυταιμνήστρη, Κλεωναί, κλύδων, κλεηδών, κλιθῆναι (Od. 1.366). Also, in Od. 12.215 τύπτετε κληΐδεσσιν, 20.92 τῆς δʼ ἄρα κλαιούσης.
πλ: in the compounds τειχεσι-πλῆτα (Il. 5.31, 455), πρωτό-πλοος, προσέπλαζε (Od. 11.583—read perhaps πρόσ-πλαζε); before Πλάταια, πλέων sailing, πλέων more (Il. 10.252), πλέον full (Od. 20.355). Add ll. 9.382 (= Od. 4.127) Αἰγυπτίας, ὅθι πλεῖστα (with v. l. ᾖ πλεῖστα, cp. Od. 4.229), and ll. 4.329 αὐτὰρ ὁ πλησίον.
χλ: in Od. 10.234 καὶ μέλι χλαωρόν, 14.429 ἀμφὶ δὲ χλαῖναν.
To these have to be added the very few examples of a vowel remaining short before σκ and ζ.
σκ : before Σκάμανδρος, σκέπαρνον (Od. 5.237, 9.391), σκίη (Hes. Op. 589).
ζ: before Ζάκυνθος (Il. 2.634, Od. 1.246, etc.), Ζέλεια (Il. 2.824, etc.).
στ: before στέᾱτος in Od. 21.178, 183—unless it is a case of synizesis.
A comparison of these exceptions will show that in a sense we are right in attributing them to metrical necessity. There are comparatively few instances in which the two consonants do not come at the beginning of a word of the form ˘ ˉ , so that the last syllable of the preceding word must be a short one. On the other hand, the extent to which neglect of position is allowed for metrical convenience is limited, and depends on the natural quantity of the consonants in question, i. e. the actual time occupied by their pronunciation. Sonant mutes (mediae) are longer than surd mutes (tenues); gutturals are longer than dentals or labials; and of the two liquids λ is longer than ρ. Thus shortening is tolerably frequent before πρ and τρ, less so before κρ, πλ, κλ, θρ, χρ. With other combinations of mute and liquid, as φρ, βρ, δρ, and with σκ and ζ, it seems to be only admitted for the sake of words which the poet was absolutely compelled to bring in, such as Ἀφροδίτη, Σκάμανδρος, Ζάκυνθος, βροτός, with its compounds, etc. No exceptions are found before γρ, γλ, φλ, κν, κμ, or any combination other than those mentioned. In short, the harshness tolerated in a violation of the rule usually bears a direct relation to its necessity. It was impossible to have an Iliad without the names Aphrodite and Scamander, but these are felt and treated as exceptions.
The wοrd ἀνδρότης, which appears in the fixed ending λιποῦσʼ ἀνδρότητα καὶ ἥβην, should probably be written ἀδρότηs. As the original μρ οf βροτός becomes either μβρ (as ἄ-μβροτος, φθισί-μβροτος), or βρ (as νὺξ ἀ-βρότη, ἀμφῐ́-βροτος), so νρ might become vδρ (as ἀνδρός), or δp. So perhaps Ἐνυαλίῳ ἀνδρεϊφόντῃ should be Ἐνυαλίῳ ἀδριφόντῃ (˘ ˘ ˉ ˉ): cp. ἀνδρε-φόνος (Hdn. ap. Eustath. 183, 6).
The plea on which a short vowel is allowed before Σκάμανδρος and σκέπαρνον may be extended, as Fick points out (Bezz. Beitr. xiv. 316), to some forms of σκίδνημι now written without the σ, viz., κέδασθεν (Il. 15.657), κεδασθέντες, etc. Metrical necessity, however, would not justify the same license vwith σκίδναται (ἐπικίδναται Il. 2.850, etc.), ἑ-σκίδνατο, ἐ-σκέδασσε (for which ἐσκέδασε is available).
Neglect of position is perceptibly commoner in the Odyssey than in the Iliad. Apart from cases in which the necessities of meter can be pleaded, viz. proper names and words beginning with ˘ ˉ, it will be found that the proportion of examples is about 3 : 1. It will be seen, too, that some marked instances occur in Books 23 and 24 of the Iliad. In Hesiod and the Homeric Hymns the rule is still more lax. Thus in Hesiod a vowel is allowed to be short before κν (Op. 567, Fr. 95), and πν (Theοg. 319). In the scanty fragments of the Cyclic poets we find πε̆́πρωται (Cypria), πᾰτρί (Little Iliad), Ἀγχίσαο κλυτὸν κτλ. (ibid.), ᾰκριβέα (Iliupersis).
Note— To the instances of shortening before -βρ- should be added ἀβροτάξομεν (Il. 10.65), which is a derivative verb from the stem which we have in the two forms ἁμαρτ- and ἀ(μ)βροτ- (cp. ἤμβροτον). The appearance of ρο instead of ρα (for r̥) is Aeolic.
- 1. They are enumerated by La Roche, Hοmerische Untersuchungen, pp. 1-41, with his usual care and completeness.