seems to have had the value, not merely of a letter, but of a syllable, viz. eF (generally changed into ev), or Fe by transposition. Thus from ἶσος (Flofos) we have νηὸς ἐξίσης, from Fίκελος ἐπιδείκελος, FeFoικώς by the side of the Ionic ἐοικώς and the Attic εικώς 6.

If we compare ekoσt with viginti, we shall see that the e is a long syllable caused by the digamma with the e, i. e. Feĺkool for Fikool (Fíkaтı)". This, again, by a singular capability of reduplication, quite consistent with the genius of the Greek language, became Fe-Fe-ikool. So we have in Od. xii. 78,

οὐδ ̓ εἴ οἱ χεῖρές τε Γεξείκοσι καὶ πόδες εἶεν.

There must have been an old aorist loaTо, 'it made itself like,' i. e. 'it appeared.' As in lonu, it took the digamma (compare our wise); and thus from Fe-ioarо arose è-Fe-loaTO. Compare Od. v. 398,

ὡς Ὀδυσῆς ἀσπαστὸν ἐξείσατο γαῖα καὶ ὕλη. Where the initial F has vanished from the first Fe. Again, we have ἔειπε (ἐξειπε) by the side of εἶπε or Feῖπε, ἐέλδωρ, ἐέργει, ἀν-έελπτος (ἀνα-Γελπτος), ἐΐσκω, ἐέρση, ἐέλσαι.

In all these it is evident that ee could not have been an open dissyllable. The Homeric Fépyeɩ passed into epyeɩ of the later Attic, ëpye of the Ionic; while the aspirated elpyet perhaps represents oFe-épyw. The Homeric Fefίσκω is evidently Fe-Γείσκω. Hesiod too has εἰς ὦπα Fεξίσκειν, Opp. 62.

The above facts appear most clearly from the transition of oFos (or ds), suus, into both Feòs and Fós 8. Thus, in Theog. 467, we have

παῖδας ἐξοὺς κατέπινε, Ῥέην δ' ἔχε πένθος ἄλαστον.

6 Used also in Il. xxi. 254, тậ eikus hicev. On the other hand, Thucydides uses the form άπεοικότως.

This is a more reasonable explanation than to conclude, with Curtius (134), that "the diphthong in the first syllable seems to be a mistake." This indeed appears to be the true explanation of the o in οἶνος, οἶκος, Οϊλεὺς = Ειλεύς. The sound of the diphthong represented wi or hwi, as oloтpos and ỏïords are our word whizz. Probably then olvos was pronounced wheenos.

s Compare meus with uós. Dr. Flack (Proleg. p. 42) gives the various epic forms σFòs, σeFds, ¿Fds, Fòs, Feós. The old Romans said sis oculis for suis, &c.,

pronounced swis.

But in Opp. 328,

ὅς τε κασιγνήτοιο Γεοῦ ἀνὰ δέμνια βαίνει.

Pindar, as well as Homer, uses the simpler form Fós. What is rather remarkable, the still further curtailed form ôs seems to have been used in early times; for we find in Od. v. 407,

ὀχθήσας δ' ἄρα Γεἶπε πρὸς ἂν μεγαλήτορα θυμόν.

And here indeed it would be easy to suppose the original reading was Feῖπε Γεῷ μεγαλήτορι θυμῷ. But a little after the time of Peisistratus, if we may trust an apparently genuine epigram quoted by Thucydides (vi. 54), the word was used without any digamma :—

μνῆμα τόδ' ἧς ἀρχῆς Πεισίστρατος Ιππίου υἱός.

That the digamma often represented Fe or eF, is also shown by the words ἔαρ, ἐαρινὸς, εἰαρινός. Comparing the Latin ver, we conclude that the old word was Fap (for Feoap). Hence we obtain Fe-ap and eF-ap, respectively čap and elap (year), and εἰαρινὸς for ἐξαρινός. Vernus is evidently Faρινos, as nocturnus is VUKтepivòs, and aeternus is aeviterinus. Indeed, the words aetas (aevitas) and aevum compared with alov seem to show that the original form was either ȧ-eF-wv or aiFwv. A good illustration of the facility with which ef became Fe by transposition, is ἕκηλος by the side of εὔκηλος, i. e. Fέκηλος and ἔκηλος, both from ἑκών.

As in many words the initial F has left only an aspirate breathing, so it has passed into a vowel when employed, as it constantly was, in the middle of words, or even at the end of root-syllables. Thus we have Boûs for BoFs, exeva for ëxeFa οι ἔχει-σα, χεύσω for χερ-σω, χυτὸς for χερ-τὸς (χευτός), κλυτὸς

• An example of this is edva for Fédva, which is also written čedva, i. e. ĕFedva (our wed). It is a question if àvéedvov, not åváedvov, should be read in Il. ix. 146, and ἀνέελπτα for ἀναέλπτα in Theog. 660. If a privative is a clipped form of ἀνὰ, 'the backward way,' i. e. the converse (analogous to our like and un-like, &c.), we can thus explain such forms as νήνεμος, νώνυμος, for ἀνάνεμος, ἀνόνυμος, and the unmutilated compounds åvá-FeλπTTOS, &c. Otherwise, we must assume a primitive ve (as in vérodes, 'footless') lost in Greek, but retained in the Latin ne (nefas, &c.). See Curtius, Gr. Et. 317.

Pindar, αὐάτα for ἄτα, and fans is from the root Fear,

for κλε-τός. We have, even in ὑπο-φαντιες (φάτις). Το αλαξ for which becomes a sibilant in sulcus. In other words the F became 4, as in λείος for λέFoς (levis), καίω for κάτω, νειὸς for νέος, εἶαρ for ἔξαρ, κλαίω for κλάξω (fut. κλαύσω), φατειὸς for pareFos, Scut. Herc. 161,-a form which is seen in the Latin verbal adjective sativus, &c. So perhaps ouotios for ouóFios, ὀλώϊος and ὀλοιὸς for ὀλόξιος (ὀλοφώνος).

There is some difficulty in accounting for the forms oida and oika, in which there is the double influence of the digamma in the root, and the lengthened syllable of the perfect, as in πέποιθα. The participle however is not οἰδὼς, but εἰδώς, and there are metrical reasons for thinking Fidos, Fidvia, was an epic usage, though whether a genuine or merely an imitative one, seems open to doubt.

It may be conjectured, that the true power of the F was first dropped in monosyllables, where it was not metrically necessary to avoid a hiatus. A comparison with the Latin shows that there were in the early Greek many digammated monosyllabic roots and crude forms, which became dissyllables in the Latin inflexion or vocalisation. Thus, vaFs, BoFs, oFs, KλaFS (roots vaF, BoF, ȧF), were changed in Greek into vaûs, Boûs, ois, Kλels, and in Latin into navis, bos, bovis, ovis, clavis1. Other monosyllables might be cited, as Fap (ĥp) ver, kλefs for kλǹs (whence λéa, 'lays'), λeFs, levis (Xeîos), and probably Spufs for Spûs. The Greek termination of adjectives in -s or -eùs may originally have been -Fs. Thus, nus or evs (whence cv, bene) was perhaps eFs, "Apns or 'Apeùs was ȧpefs, ýdùs was oFadF-s (as shown by suavis). There is a diversity of opinion. among scholars, whether T is a letter of the primitive alphabet (and it occurs in the earliest inscriptions), or was at first represented by the vowel-sound of F, as Franz and Donaldson maintain. Thus it is somewhat uncertain whether pv (péw),

1 That κλείειν, to shut, was originally κλέξειν or κλάξειν, is proved by the Latin claudo and clavis. Compare καίω, καύσω, κλαίω, κλαύσω.

που (πνέω), πλυ (πλέω), or ῥef, πνεί, πλεί or πλοF, are the true forms of the roots. (See Curtius, Gr. Et. 564.)

It is a singular fact, that the F when represented by v had not in itself the power of lengthening a syllable, even when it made a diphthong. Thus χυτός, κλυτὸς, ῥυτὸς, for χερτὸς &c., have the v short, as is the au in the Pindaric aŬáтa for åтa. So Fopavòs (Varuna) became ovpavòs and metrically opavós, as βόλομαι, Lat. volo, is the Aeolic form of βούλομαι. But in ἔχευα, ἀλεύασθαι, ἐπιδενὴς for ἐπιδεής, the F does make a long syllable. The inference from this is, that in ῥέει, ῥέεθρον, ἔχεα, àλéaodai, véos, and such words, the single digamma really did exist, by which the hiatus was avoided; while in the lengthened forms, exeva &c., the F was doubled, exeƑFa, àλéƑFaolai, and so

on.

ɩ,

The variation of the digamma between , v, and o, is a curious property, as showing how different from our F was the real power of the letter. We have peîlpov and πveiw by the side of ῥεύσομαι, ῥυτὸς, and πνεύσομαι, πνεῦμα, ἄμπνυτο. Hence ῥέξω, πvéƑw, may well have been the primary forms, like xéƑw = xéw, and πλέξω = πλέω, fut. πλεύσομαι. Both ἀείδειν and ἀοιδὴ come from a digammated form closely connected with αὐδᾶν, αὐλὴ, viz. ἀείδειν οἱ ἀξύδειν. The written form τραγαυδος for τραγῳδὸς is found in an inscription 2. Again, Kλeíw is another form of λéw, and if we compare κλύω and κλυτὸς, we shall arrive at the conclusion that kλéƑw was the old verb. The first verse of the "Works" might therefore be given thus ;-Moûσai Пiepín@ev ἀξυδῇσι κλέξοντες. Thus we account for the expanded form κληΐζειν, κλῄζειν.

There are some words, however, as κρειών for κρεῶν, κρείων for κρέων, εξείης for ἑξῆς (ἑξέης), εἰᾶν for ἐᾶν, λείων for λέων, χάλκειος for χάλκεος, &c., which would seem rather to depend on a different principle of arbitrary elongation, viz. the epic property of dwelling on a short syllable for metrical convenience. The large class of verbs in -ców, evidently analogous

2 See Donaldson's Greek Grammar, § 18.

to -éw, may originally have been digammated, just as eπidevns and δεύομαι appear to represent ἐπιδεξὴς and δέξομαι.

In questions of Attic orthography, such as ἀεὶ, κλάειν, κάειν, ȧeròs, for aleì, Kλalew, &c., it is evident that the rejection of the

is only a final effort to efface the lingering vestiges of the F. On the other hand, a few words in the Attic seem to have retained the F or its representative sound, for metrical reasons, as προσελεῖν, φιάλλειν, ἀρχέλειος (λεξὼς, Aesch. Pers. 299), κατέαγα (άγνυμι).

The above remarks are only intended as a popular exposition of an extremely interesting theory, and with the view of directing the attention of younger students to a subject which has not only not been taught, but is even shunned in schools and public lecture-rooms, although rather more attention is now given to phonetic laws and changes, which include the numerous substitutions for the dropped digamma. At present it is perhaps sufficient to refer the student to the important Essay on "Transformations of the F" in Book III. of G. Curtius' "Greek Etymology." I have been unwilling, however, wholly to omit, in reprinting, these remarks of my own, the result of much independent thought, especially as Dr. Flack has throughout referred to them in his edition of the Theogony3. Mr. Mahaffy indeed, who in p. 120 of vol. i. of his History of Literature "damns with faint praise" my edition of Hesiod as "overloaded with very questionable notes about the Digamma, and the etymology of old Greek words," disparages them; but he writes in the style of one who has not himself gone far into these inquiries. He evidently regards them rather as antiquarian curiosities than as having any practical bearing on the extant literature of Greece. And without doubt, investigators of the digamma must walk warily, as on slippery ground. Yet it is no real gain to scholarship to speak even of their speculations as of no importance, and wholly barren of results. Such inquiries are not by any means barren of results, when