Words with Initial ϝ-

Book Nav

390. The former existence of the ϝ in a given Homeric word may be inferred either from its appearance in some other dialect of Greek, or (where this kind of evidence fails) from the corresponding forms in the cognate languages. Thus an original ϝείκοσι is supported by the forms ϝίκατι and ϝείκατι on Doric and Boeotian inscriptions, by the Laconian βείκατι (given by Hesychius), and again by Latin viginti, Sanskrit νiṁçati, etc.; an original ϝέσπερος by the form ϝεσπαρίων on a Locrian inscription, as well as by Latin νesper; original ϝιδεῖν, ϝοῖδα, etc., by ϝίστορες on inscriptions, γοῖδα and γοίδημι in Hesychius (erroneously so written, as Ahrens showed, for ϝοῖδα and ϝοίδημι), and also by Latin νideο, Sanskrit νedmi, νeda, English wit, etc. We do not, however, propose to discuss the external evidence, as it may be called, by which the loss of an initial ϝ is proved, but only to consider the degree and manner in which the former existence of such a letter can be shown to have affected the versification of Homer. For this purpose it will be enough to give a list of the chief words in question, and in a few cases a statement, by way of specimen, of some of the attempts made to restore the ϝ to the text.1


The initial ϝ is to be traced by the hiatus in

Il. 5.161 ἐξ αὐχένα ἄξῃ

Il. 8.403 κατά θʼ ἅρματα ἄξω

(with similar phrases in 8.417, 23.341, 467) less decisively by the lengthening of the final -ιν of the preceding word in

Il. 4.214 πάλιν ἄγεν ὀξέες ὄγκοι

The evidence against an initial consonant is very slight. In

Od. 19.539 πᾶσι κατʼ αὐχένας ἦξε

we should read αὐχένʼ ἔαξε (Bekk.), understanding the singular distributively (§ 170). In

Il. 23.392 for ἵππειον δέ οἱ ἦξε

may be read ἵππειόν οἱ ἔαξε.

ἄναξ (ἄνασσα, ἀνάσσειν)

The words of this group occur in Homer about 300 times, and in about 80 instances they are preceded by a final short vowel which would ordinarily be elided. This calculation does not include the phrase ἶφι ἀνάσσειν, or the numerous examples of hiatus after the dative singular in -ι and the genitives in -οιο, -ειο,-ᾱο.2

The cases in which a slight correction of the text is needed to make room for the ϝ are as follows.

Il. 1.288 πάντεσσι δʼ ἀνάσσειν
              (read πᾶσιν δέ)

Il. 9.73 πολέεσσι δʼ ἀνάσσεις
            (read πολέσιν δέ, § 389)

Il. 2.672 Χαρόποιό τʼ ἄνακτος
              (read Χαρόπου τε)

Il. 7.162 (= 23.288) πρῶτος μὲν ἄναξ
                             (read perhaps πρώτιστα)

Il. 15.453 κροτέοντες· ἄναξ
                (read κpοτέοvτε, the dual)

Il. 16.371 (= 507) λίπον ἅρματʼ ἀνάκταων
                           (read ἅρμα, § 170)

Il. 16.523 σύ πέρ μοι, ἄναξ, τόδε καρτερὸν ἕλκος ἄκεσσαι
                (read με)

Il. 23.49 ὄτρυνον, ἄναξ
              (read ὄτρυvε, the present imperative)

Il. 23.517 ὅς ῥά τʼ ἄνακτα
                (read ὅς τε or ὅς ῥα)

Od. 9.452 ἦ σύ γʼ ἄνακτος
                 (omit γʼ)

Od. 17.189 χαλεπαὶ δέ τʼ ἀνάκταων
                   (omit τʼ)

Od. 21.56 (= 83) τόξον ἄνακτος
                          (read τόξα)

The imperfect ἤνασσε, which occurs five times, can always be changed into ἐάνασσε. The remaining passages are

Il. 19.124 σὸν γένος· οὔ οἱ ἀεικὲς ἀνασσέμεν Ἀργείοισιν
                (a verse which is possibly interpolated)

Il. 20.67 ἔναντα Ποσειδάωνος ἄνακτος
              (in the probably spurious θεομαχία)

Il. 24.449 & 452 ποίησαν ἄνακτι

Od. 14.40 ἀντιθέου γὰρ ἄνακτος κτλ.

Od. 14.395 εἰ μέν κεν νοστήσῃ ἄναξ

Od. 14.438 κύδαινε δὲ θυμὸν ἄνακτος

Οδ. 24.30 ἧς περ ἄνασσες

ἄρνα (ἄρνες, etc.)

The ϝ is supported by three instances of hiatus

ll. 4.158 αἷμά τε ἀρνῶν

Il. 4.435 ὄπα ἀρνῶν

Il. 8.131 ἠΰτε ἄρνες

and by the metrical length given to the preceding syllable in

ll. 3.103 ἐς δίφρον ἄρνας

Il. 16.352 λύκοι ἄρνεσσι

The passages which need correction are

ll. 3.103 οἴσετε δʼ ἄρνʼ (the δὲ is better omitted)

ll. 3.119 ἠδʼ ἄρνʼ ἐκέλενεν (read ἰδὲ ἄρνʼ)

Il. 22.263 οὐδὲ λύκοι τε καὶ ἄρνες (omit τε)

Od. 4.86 ἵνα τʼ ἄρνες ἄφαρ κεραοὶ τελέθουσι (omit τʼ)

Οδ. 9.226 ἐρίφους τε καὶ ἄρνας

Note, however, that the evidence for ϝ is confined to the Iliad, and that the derivative ἀρvειόs shows no trace of it.


The presence of an initial consonant is shown by hiatus in nearly 80 places. In two places the text is uncertain

Il. 24.320 ὑπὲρ ἄστεος 

but διὰ ἄστεος in the Bankes papyrus, and several MSS. and

Od. 3.260 ἑκὰς ἄστεος

ἑκὰς Ἄργεος in most MSS.

Two passages admit of the easiest correction.

Il. 3.140 ἀνδρός τε προτέροιο καὶ ἄστεος (read πpοτέρου)

Il. 15.455 τοὺς μὲν ὅ γʼ Ἀστυνόμῳ (omit γε or μέν)

Two remain, viz.

Il. 11.733 ἀμφίσταντο δὴ ἄστυ (ἀμφέσταν Bekk.)

Il. 18.274 νύκτα μὲν εἰν ἀγορῇ σθένος ἕξομεν ἄστυ δὲ πύργοι (ἕξετε Bekk.)

The changes made by Bekker in these places are not improbable, but are hardly so obvious as to exclude other hypotheses.

ἔαρ, εἰαρινός

Hiatus is found in

Il. 8.307 νοτίῃσί τε εἰαρινῇσι

and a short final syllable is lengthened in

Od. 19.519 ἀείδῃσιν ἔαρος

In the phrase ὥρῃ ἐν εἰαρινῇ we should doubtless omit the ἐν, as in Od. 5.485 ὥρῃ χειμερίῃ (Bentl.).


The ϝ appears in ἀνὰ εἴκοσι (Od. 9.209), and the combination καὶ εἴκοσι (which occurs 9 times, including the compounds with δυαωκαιεικοσι-).

In Il. 11.25 χρυσοῖο καὶ εἴκοσι read χρυσοῦ, and in the combination τε καὶ εἴκοσι (in three places) omit τε. In the recurring ἤλυθον εἰκοστῷ ἔτεϊ κτλ. Bekker reads ἦλθον ἐεικοστῷ (Cοbet well compares Od. 23.102 ἔλθοι ἐεικοστῷ κτλ.), On Od. 5.34 ἤματί κʼ εἰκοστῷ κτλ. see § 389.


Twο instances of hiatus indicate ϝ, in Il. 24.100, 718, besides many places in which the word is preceded by a dative singular, as οὐδένι εἴκων, κάρτεϊ εἴκων.

Twο places may be easily corrected

Il. 4.529 μηδʼ εἴκετε

read μὴ εἴκετε with asyndeton as Od. 24.54 ἴσχεσθʼ Ἀργεῖοι, μὴ φεύγετε

Il. 12.48 τῇ τʼ εἴκουσι

(omit τε). In Od. 12.117 for θεοῖσιν ὑπείξεαι read θεοῖς ὑποείξεαι (Bekk.) There remains Il. 1.294 εἰ δὴ σοὶ πᾶν ἔργον ὑπείξομαι.

ἕοικα, ἐίσκω, εἶκελος

The ϝ of ἔοικα appears from hiatus in 46 instances (not counting the numerous places in which it follows a dative in -ι). The adverse instances are 11 in number, besides the form ἐπ-έοικε (which occurs 11 times). The corresponding present εἴκω is generally recognized in Il. 18.520

ὅθι σφίσιν εἶκε λοχῆσαι
where it suited them to be in ambush

The form ἐΐσκω has hiatus before it in 3 places, but twice rejects ϝ (Od. 9.321, 11.363). The adjective εἴκελος or ἴκελος usually needs an initial consonant (except ll. 19.282, Od. 11.207)

It seems probable that this is the same word as εἶκω to yield. The notion of giving way easily passes into that of suiting or fitting, hence conforming to, resembling.

ἑκών, ἕκητι, ἕκηλος

Hiatus indicating ϝ is found in 22 places (not reckoning οὔ τι ἑκών Il. 8.81, etc.).

In Od. 4.649 for αὐτὸς ἑκών we may read αὐτὸς ἐγών (cp. Od. 2.133, where both these forms are found in good MSS). In Od. 17.478 ἔσθιʼ ἕκηλος two MSS. have ἔσθʼ (i. e. ἔσθε), The remaining exceptions are

with ἑκώv, Il. 23.434, 585; Od. 5.100 (where we may read τίς κε, or perhaps τίς δὲ ἑκὼν . . . διαδράμοι the optative without ἄν being used as in negative clauses, § 299.f)

with ἕκηλος, Il. 8.512; Od. 2.311 (ἐϋφραίνεσθ’ ἐΰκηλον Bekk.)

ἑκάς, ἕκατος, etc.

Traces of ϝ are to be seen in the hiatus

νῦν δὲ ἑκάς
(Il. 5.791, 13.107)

ἀλλὰ ἑκάς
(Od. 15.33)

οὐδὲ ἑκηβολίαι
(Il. 5.54)

and in the lengthening in Ἀπόλλωνος ἑκάτοιο (Il. 7.83, 20.295), ἐϋπλόκαμος Ἑκαμήδη, etc.

The exceptions are Il. 1.21 & 438, 17.333, 20.422, 22.15 & 302; Od. 7.321—mostly admitting of easy correction.


The original ϝ of this word (recently found on a Locrian inscription, see Curt. Stud. ii. 441 ff.) is traced by means of hiatus in 115 places. The adverse instances, however, are about 50 in number, and the proportion that can be removed by emendation is not so large as in most cases (see L. Meyer, K. Z. viii. 166. About a fourth of the exceptions appear in the recurring phrase μένος καὶ θυμὸν ἑκάστου.

The form ἑκάτερθε shows slight traces of initial ϝ in

Od. 6.19 σταθμοῖϊν ἑκάτερθε

Od. 11.578 γῦπε δέ μιν ἑκάτερθε

Od. 22.181 τὼ δʼ ἔσταν ἑκάτερθε

It is preceded by elisiοn in ll. 20.153 (omit ῥʼ), and in Il. 24.273, Od. 7.91 (omit δʼ).

εἴλω (ἔλσαι, ἐάλην), ἁλῶναι, ἅλις

The ϝ is shown by hiatus in

Il. 1.409 ἀμφʼ ἅλα ἔλσαι

Il. 16.403 ἥστο ἀλείς

(and five other examples of this tense, viz. Il. 5.823, 21.571 & 607, 22.308; Od. 24. 538).

Il. 18.287 κεκόρησθε ἐελμένοι

Il. 12.172 ἠὲ ἁλῶναι
               (so 14.81)

Il. 21.281 εἵμαρτο ἁλῶναι
               (so Od. 5.312, 24.34)

Il. 81.495 τῇ γε ἁλώμεναι

Before ἅλις hiatus occurs in about 12 places; cp. also Il. 23.420 εἰνάτερες ἅλις ἦσαν.


Il. 21.236 κατʼ αὐτὸν ἅλις ἔσαν

some MSS. read ἔσαν ἅλιs, and at 1.344 the same transposition may be made. The only other instance against ϝ is Il. 17.54 ὅθʼ ἅλις ἀναβέβρυχεν (ἀναβέβροχεν Zenod.), where Bentley read ὃ ἅλις ἀναβέβροχεν.

ἑλίσσω, εἰλύω

Before ἑλίσσω hiatus is found in four places, and the recurring phrases καὶ ἕλικας βοῦς and εἰλίποδας ἕλικας βοῦς point in the same direction. The only exceptions are

Od. 12.355 βοσκέσκονθʼ ἕλικες κτλ.


Il. 18.401 γναμπτάς θʼ ἕλικας

It is probable that in many places the forms ἐλέλικτο, ἐλελίχθη, etc., are old errors for ἐϝέλικτο, ἐϝελίχθη, etc. see Dawes, Misc. Crit. 177, also Heyne on Il. 1.530

Traces of ϝ in εἰλύω should perhaps be recognized in Od. 5.403 (ἐρευγόμενον, εἴλυτο) and 15.479 σάκεσιν εἰλυμένοι; cp. Il. 20.492 φλόγα εἰλυφάζει. In Il. 18.522 ἵζοντ’ εἰλυμένοι it is easy to read ἷζοv (as Bekker). The aorist participle ἐλυσθείς has no ϝ, but it may be from a different verb stem (see Buttm. Lexil. s. v. εἰλύω).

ἔλπω (ἔολπα)

The initial ϝ of this wοrd is proved by 10 instances of hiatus (including καὶ ἐλπίδος, Od. 16.101, 19.84). The perfect ἔολπα also shows traces of ϝ in the reduplicated syllable, viz. in Od. 2.275, 3.375, 5.379.


Il. 8.526 εὔχομαι ἐλπόμενος

should be εὔχομ’ ἐελπόμενος (Hoffm.) or perhaps (as Zenodotus read) ἔλπομαι εὐχόμενοs. In four places ϝέλπω can be restored by very slight corrections.

Il. 15.701 Τρωσὶν δʼ ἔλπετο
                (Τρωσὶ δέ Heyne)

Il. 18.194 ἀλλὰ καὶ αὐτὸς ὅδʼ, ἔλπομʼ
               (αὐτὸς ἑέλπομʼ Heyne)

Od. 2.91 (= 13.380) πάντας μέν ῥ’ ἔλπει (omit ῥ’).

Two others are less easy

Il. 15.539 πολέμιζε μένων, ἔτι δʼ ἔλπετο
                (μένον δʼ ἔτι ἔλπετο Bentl.)

Il. 24.491 ἔπί τʼ ἔλπεται
               (καὶ ἔλπεται Bentl.)

The passages which tell against ϝέϝολπα are

Il. 20.186 χαλεπῶς δέ σʼ ἔολπα τὸ ῥέξειν
               (read σὲ ἔολπα)

Il. 21.583 μάλʼ ἔολπας
               (μάλα ἔλπεʼ Hοffm.)

Il. 22.216 νῶί γʼ ἔολπα
               (omit γʼ)

Also, Od. 8.315, 24.313.

ἔπος, εἰπεῖν

The ϝ of ἔπος is supported by about 26 instances of hiatus, and a much larger number in which preceding syllables are lengthened (as in the common line καί μιν ἀμειβόμενος ἔπεα κτλ.).

Of the apparent exceptions, about 35 are removed by reading ἔπεσσι for ἐπέεσσι (as in Il. 5.40 χειρὸς ἑλοῦσʼ ἐπέεσσι προσηύδα, read ἑλοῦσα ἔπεσσι). This is justified by the fact that in similar words (especially βέλος) the form in -εεσσι is less frequent than that in -εσσι. A group of 11 may be corrected by scanning ἔπεα as a disyllable (˘ ˉ) in the formula φωνήσασα ἔπεα πτερόεντα προσηύδα. Another small group of exceptions is formed by phrases such as

Od. 4.706 ὀψὲ δὲ δή μιν ἔπεσσιν κτλ.

where perhaps ἑ may be put for μιν. There remain twο instances in the Iliad (5.683, 7.108), and seven in the Odyssey (11.146 7 561, 14.509, 15.375., 16.469, 17.374, 24.161).

In εἰπεῖν the ϝ is proved by about 80 instances of hiatus, besides lengthening such as we have in the forms ὥδε δέ τις εἴπεσκε, ὣς ἄρα οἱ εἰπόντι, etc. The exceptions number about 35.

Of these exceptions 10 are found in the recurring line ὄφρ’ εἴπω τά με θυμὸς ἐνὶ στήθεσσι κελεύει. It has been suggested as possible that εἴπω has here taken the place of an older ἔπω (ϝέπω), or ἔσπω (cp. ἔσπετε), This supposition would of course explain other instances of neglected ϝ, as Il. 1.64, 11.791; Od. 1.10 & 37, etc.

ἔρδω, ἔργον, etc.

The verb ἔρδω is preceded by hiatus in two clear instances, Il. 14.261, Od. 15.350. In Il. 9.540 πόλλʼ ἔρδεσκεν there is an ancient v. l. ἔρρεζεν. In Il. 10.503 ὅτι κύντατον ἔρδοι we may read κόντατα. But there are several instances on the other side in the Odyssey (viz. 1.293, 5.342 & 350, 6.258, 7.202, 8.490, 11.80).

The reduplicated form ἔοργα (for ϝέϝοργα) is preceded by hiatus in 7 places. Instances on the other side are

Il. 3.351 ὅ με πρότερος κάκʼ ἔοργε

where the aorist ἔρεξεν is more Homeric, cp. § 28

Il. 21.399 ὅσσα μʼ ἔοργας
               (ὅσσα ἔοργας Ambr.)

Il. 22.347 οἷά μʼ ἔοργας

here also με may be omitted

Od. 22.318 οὐδὲν ἐοργώς (read οὔ cp. § 356)

The noun ἔργον, with its derivative ἐργάζομαι, occurs in Homer about 250 times, and the ϝ is required to prevent hiatus in about 165 places. There are about 18 instances against ϝ.

εἴρω, ἐρέω

The ϝ of εἴρω is required by hiatus in the three places where it occurs, viz. Od. 2.162, 11.137, 13.7; that of ἐρέω by about 50 instances of lengthening (such as ἀλλʼ ἔκ τοι ἐρέω, ὥς ποτέ τις ἐρέει, and the like), against which are to be set three instances of elision (Il. 4.176, 23.787; Od. 12.156).

ἕννυμι, εἷμα, ἐσθής

The ϝ is shown by hiatus in more than 80 places, including the instances of the perfect middle (εἷμαι, ἕσσαι, etc., see § 23.5). The contrary instances are of no weight. The superfluous ῥʼ may be omitted in ἐπεί ῥʼ ἕσσαντο (three places), and τʼ similarly in Od. 14.510, 24.67. This leaves Il. 3.57; Od. 6.83, 7.259.


The ϝ (which is inferred from Latin vomo) may be restored by reading ἐϝέμεσσε for ἀπέμεσσε (Il. 14.437) and αἷμα ϝεμέων, or possibly ϝέμων (L. Meyer), for αἷμ’ ἐμέων (Il. 15.11).


Hiatus occurs in six places, after the prepositions ποτί (Od. 17.191) and ἐπί, There are no instances against ϝ.


The ϝ is supported by the lengthening of the preceding syllable in five places, such as Il. 24.765 ἐεικοστὸν ἔτος ἐστί.

In the only adverse instance, Il. 2.328 τοσσαῦτʼ ἔτεα, we may read and scan τοσαῦτα ἔτε͜α, as in the case of ἔπεα (above).

ἰάxω, ἰαxὴ, ἠxὴ

The ϝ in ἰάχω and ἰαχή is chiefly indicated by 23 instances of a peculiar hiatus, viz. after a naturally short final vowel in arsis, as

ἡ δὲ μέγα ἰάχουσα

ἡμεῖς δὲ ἰάχοντες

γένετο ἰαχή

and the like. There are 3 instances of lengthening by position. The ϝ is also proved by αὐίαχος (= ἀ-ϝίϝαχος) withοut a cry. The exceptions are confined to the aorist or imperfect ἴαχον (ῑ), which never admits ϝ in Homer: see § 31.1.

The derivative ἠχήεις follows hiatus in two places (Il. 1.157; Od. 4.72); elsewhere in Homer ἠχή only occurs at the beginning of the line. The compound δυσ-ηχής (πολέμοιο δυσηχέος, Il. 2.886, etc.) is best derived from ἄχος (see Wackernagel, Dehnungsgesetz, p. 42).

ἰδεῖν, οἶδα, εἷος.

In the different forms of the 2nd aorist ἰδεῖν the ϝ is shown by upwards of 180 instances of hiatus, and about 12 instances of lengthening of a short syllable. The indicative (εἶδον in Attic) is nearly always a trisyllable (i. e. ἔϝιδον) in Homer. On the other side we have to set nearly 50 instances of neglected ϝ, about half of which are susceptible of easy emendation (such as putting ἰδεῖν for ἰδέειν, omitting superfluous δὲ, and the like).

In the perfect οἶδα there are about 125 instances of hiatus, against 24 which need emendation. Of these, however, only about seven or eight present any difficulty. The proportion is much the same with the other forms, as εἴδομαι, εἴσομαι, etc., and the nouns εἶδος (11 instances of hiatus, two adverse), ἴστωρ, ἰδρείη, εἴδωλον, etc.

ἴον (ἰόεις, ἰοδνεφές)

The ϝ is supported by hiatus in Od. 4.135, 9.426, and is nowhere inadmissible.

ἴς, ἶφι (ἴφια), ἶνες.

These words, with the derived proper names Ἰφιάνασσα, Ἴφιτος, etc., show ϝ in about 27 places, while seven or eight places need slight emendation. ἴφθιμος, which shows no trace of ϝ, is probably from a different root.


The ϝ is traced in about 30 instances of hiatus; the adverse passages being eight or nine in number. In three of these, containing the phrase ἀτεμβόμενος κίοι ἴσης (Il. 11.705, Od. 9.42 & 549) the form ἴσης should perhaps be changed to αἴσης share. Or we may recognize the Aeolic form of the word, viz. ἴσσα (Fick, Odyssee, p. 20). The other places are easily corrected.

ἴτυς, ἰτέῃ.

The ϝ is shown by hiatus (Il. 4.486, Od. 10.510). The particle τε may be left out before καὶ ἰτέαι in Il. 21.350.


The ϝ is required in 105 places by hiatus, in 14 by the lengthening of a short syllable. About 25 places are adverse.


The ϝ is required by hiatus in nearly 100 places. The adverse places are about 20 (including the names Οἰνεύς and Οἰνόμαος).

  • 1. The first systematic attempt to restore the digamma was made by Heyne in his edition of the Iliad (1802). It was based upon Bentleyʼs manuscript annotations, of which Heyne had the use. The first text with restored ϝ was published by Payne Κnight (1820). Much was done by the thorough and methodical Quaestiones Homericae of C. A. J. Hoffmann (Clausthal, 1842-48). The ϝ was again printed in the text of Bekkerʼs second edition (Bonn, 1858). The light of the comparative method was brought to bear upon it by Leskien (Rationem quam I. Bekker in restituendo digammο secutus est examinavit Dr. A. Leskien, Lipsiae, 1866). The most complete treatise on the subject is that of Κnös (Upsaliae, 1872). The most important contributions, in addition to those mentioned, have been made by Leo Meyer (K. Z. xviii. 49), and by W. Hartel (Hom. Stud. iii). Most of the cοnjectures given in this chapter come from one or other of these sources.
  • 2. For a complete analysis of the examples in the Iliad see Dawes, Miscellanea Critica, Sect. IV.