379. Hiatus is a term which is used by writers on meter in more than one sense. It will be convenient here to apply it to every case in which a word ending with a vowel or diphthong is followed by a word beginning with a vowel, and the two vowel sounds are not merged together (as by elision, crasis, etc.) so as to form one syllable for the meter.
It would be more scientific, perhaps, to understand the word hiatus as implying that the two vowels are separated by a break or stoppage of vocal sound, so that the second begins with either the rough or the smooth "breathing." Thus it would be opposed to every form of diphthong (including synizesis), the characteristic of which is that the two vowels are slurred together, by shifting the position of the organs without any perceptible interruption of the current of breath. This definition, however, might exclude the case of a long vowel or diphthong shortened before an initial vowel (as τὴν δʼ ἐγὼ οὐ, where the final ω seems to be partly merged in the following ου). Again when a final ι or υ comes before a vowel without suffering elision, it is probable that the corresponding semi-vοwel (= our y or w) is developed from the vowel sound, and prevents complete hiatus.
380. Long vowels before hiatus. The general rule is that a long final vowel or diphthong coming before a vowel forms a short syllable in the meter. This shortening is very common in Homer: cp. Il. 1.299 οὔτε σοὶ οὔτε τῳ ἄλλῳ, ἐπεὶ κτλ., where it occurs in three successive feet.
But the natural quantity may be retained before hiatus when the vowel is in the arsis of the foot, as Ἀτρεΐδῃ, Ἀγαμέμνονι, ὅς κʼ εἴποι ὅτι κτλ. And in a few instances a long vowel or diphthong is allowed to remain long in thesis.
Il. 1.39 Σμινθεῦ· εἴ ποτέ τοι κτλ.
The readiness with which long syllables are allowed before hiatus varies with the several long vowels and diphthongs; partly also it depends on the pause of the sense.
The long diphthongs (as they may be called), viz. ῃ and ῳ, are the most capable of resisting the shortening influence of hiatus; next to them are ευ and ου, and the long vowels η and ω, while ει, οι, and αι are at the other end ob the scale. Α measure of this may be gained by observing how often each of these terminations is long before a vowel, and comparing the number with the total number of times that the same termination occurs. Thus it appears that out of every 100 instances οf final ῳ, it is long before hiatus about 23 times. Similarly final -ῃ is long 19 times, -ευ 6·7 times, -ου 6 times, -η 5·7 times, -ω 4 times, -ει 1·8 times, -οι 1·6 times, and -αι only 1·3 times. Thus hiatus after ῳ and ῃ is scarcely avoided, while after ει, οι and αι it is very rare.
In a large proportion of the instances in which a long vowel retains its quantity before hiatus it will be found that the hiatus coincides with a division either in the sense or the rhythm. Of the examples in the arsis of the foot, more than half occur before the penthemimeral caesura, where there is almost always a pause, while in thesis the same thing is chiefly found to occur either after the first foot, as Il. 2.209 ἠχῇ, ὡς ὅτε κτλ.; Od. 11.188 ἀγρῷ, οὐδὲ κτλ.; or after the fourth foot (in the Bucolic diaeresis).
381. Shortening of Diphthongs Before Hiatus. Regarding the nature of the process by which a diphthong before hiatus was reduced to the time or metrical value of a short syllable two probable views have been maintained.
- Curtius holds that whenever long syllables are shortened by the effect of hiatus something of the nature of elision takes place. Thus η and ω lose the second half of the vowel sound, while αι, eι, οι lose the ι. In support of this he points to the facts of crasis. Thus καὶ ἐγώ in becoming κἀγώ may be supposed to pass through the stage κα ἐγώ.
- According to an older view, which has been revived and defended with great ingenuity by Hartel,[fn]Homerische Studien, iii. pp. 7 ff.[/fn] the ι or υ in a diphthong is turned into the corresponding spirant; so that καὶ ἐγώ becomes κα-ι̯-εγώ, and ἐκ Πύλου ἐλθόν becomes ἐκ Πύλο-ϝ-ελθόν.
It is certainly in favor of this latter supposition that it does not oblige us to suppose the frequent elision of the two vowels which in general are the least liable to be elided. The explanation however is not a complete one. It does not account for the shortening of ῃ and ῳ, which on the principle assumed by Hartel would become ηι̯, ωι̯. On the whole it seems most probable that the shortening in question was effected, for diphthongs as well as for simple long vowels, by a process in which ancient grammarians would have recognized rather "synizesis"—viz. the slurring of vowels together without complete loss of any sound—than either elision or contraction. And this conclusion is supported by the general tendencies of the Ionic dialect, which was especially tolerant of hiatus, and allowed numerous combinations of vowels, such as εα, εο, εω, εοι, to have the value either of one syllable or two.[fn]The use of εο for ευ in Ionic inscriptions shows, not indeed that ευ and εο were identical in pronunciation, or that εο was a true diphthong, but certainly that εο was very like ευ, and might be monosyllabic in scansion. Probably monosyllabic εο (when it was not a mere error for ευ) stood to ευ as the σynizesis ε͜α, ε͜ω, εοι, etc. to the contracted η, ω, οἱ. See Erman in Curt. Stud. v. 292 ff[/fn]
382. Hiatuus After Short Syllables. The vowels which are not liable to elision may generally stand before hiatus. Τhus, we find ζωστῆρι ἀρηρότι (§ 376.3), πρὸ ὁδοῦ, πρὸ Ἀχαιῶν, αὐτὰρ ὁ ἐμμεμαώς, ἑτάροιο ἐνηέος, and the like. Hiatus is also tolerated occasionally in the pauses of the verse
- In the trochaic caesura of the third foot.
Il. 1.569 καί ῥʼ ἀκέουσα καθῆστο, ἐπιγνάμψασα κτλ.
Od. 3.175 τέμνειν, ὄφρα τάχιστα ὑπὲκ κτλ.
- In the Bucolic diaeresis.
Il. 8.66 ὄφρα μὲν ἠὼς ἦν καὶ ἀέξετο ἱερὸν ἦμαρ
Od. 2.57 εἰλαπινάζουσιν πίνουσί τε αἴθοπα οἶνον
The vowel of the person endings -το, -vτο seems to be especially capable of standing before hiatus in these places. It appears in more than a fourth of the whole number of instances given by Κnös (pp. 42-45).
Hiatus in the Bucolic diaeresis is commoner in the Odyssey than in the Iliad, in the proportion 2 : 1. Hiatus after the vowel ε is also comparatively rare in the Iliad; Κnös reckons 22 instances (many of them doubtful), against 40 in the Odyssey. It is worth notice that in both these points books 23 and 24 of the Iliad agree with the Odyssey, also that book 24 of the Odyssey contains an unusual number of instances of hiatus, both legitimate (Od. 11.63, 215, 328, 374, 466) and illegitimate (Od. 11.209, 351, 430).
Illegitimate hiatus, like other anomalies, may be diminished by emendation. Thus in Od. 5.135 ἠδὲ ἔφασκον we may read ἠδέ ϝʼ ἔφασκον; in 5.257 ἐπιχεύατο ὕλην we may insert ἄρʼ, on the model of Il. 5.748 ἐπεμαίετ’ ἄρʼ ἵππους. But in Il. 13.22 ἄφθιτα αἰεί must stand because ἄφθιτος αἰεί is a fixed phrase. It is unlikely, then, that hiatus was ever absolutely forbidden in epic verse.