^{page 129 note 1} In 1. 2 the vertical stroke in this letter seems to have been forgotten by the engraver.

^{page 130 note 1} Rut in the barxe-fragment it seems that -r is syllabic in one word; see § 21 below.

^{page 131 note 1} See below § 22.

^{page 131 note 2} Mr. Bosanquet tells me also of an insc. (of the 1st cent. B.c.), from half-way between Candia and Fraesos, containing what is explained by Hiller von Gärtringen (Hermes 1901, p. 452) as an epithet of Hermes, spelt κυφαρισσιφα We know too little of this form, or indeed of the Gr. κυπαρισσος itself, to be greatly helped, but it is certainly to be observed that both in this form and in Πραῖσος the p is followed by r, which may have caused in this dialect, as Prof. Ridgeway suggests to me, some kind of aspiration in the pronunciation of the preceding explosive. One or two of the more isolated forms discussed by Gustav Meyer, Gr. Gram.² § 207, may be due to parallel dialectal changes.

^{page 131 note 3} I have to thank the acting editor, Mr. Cecil Smith, for very kind help in securing a correct engraving.

^{page 133 note 1} ‘Its centre lies about ¼ in, to the left of the hasta of ’

^{page 134 note 1} It was suggested by Mr. A. E. Cowley, in the Athenaeum, March 16, 1902, that the hasta was used to mark off the proper names [like the ‘cartouches’ in Egyptian hieroglyphic inscc] and he proposed therefore, to regard Barxe and Agset (which he read Apset) in that way. But there seems some difficulty in supposing that two names in two separate lines, which can hardly both refer to one person, and which come too far from the beginning (see § 20) to be naturally referred to the person, if there was one, whom the insc. chiefly concerned, should be decorated in this way, especially if Anait is thrown in without any such token of respect. But I quite agree that, if they were proper names, they would not suggest an I-Eu. language.

^{page 134 note 2} See below § 30.

^{page 136 note 1} In a letter of October 2, 1902.

^{page 136 note 2} Tzetzes in Lycophr. 1307; Schol. Ven. ad Il. 1. 39; cf. Strabo 13. 1. 48., p. 604.

^{page 136 note 3} Most commonly σμ became μ e.g. in μία for ^*σμ-ια, cf. Lat. sem-el. The precise conditions of this change have not been determined for any of the dialects.

^{page 136 note 4} Kretschmer does not give the sources from which he takes these names, and I have not been able to ascertain in what part of the island the places lie. Σύρινθος Πύρανθος are from Steph. Byz.

^{page 136 note 5} I owe this reference to Mr. Bosanquet.

^{page 137 note 1} Iliad, 2, 647.

^{page 137 note 2} I have to thank my friends Professor Flamstead Walters for his kindness in verifying these glosses and some other references in this section, and Mr. H. T. Francis of Caius for similar help.

^{page 137 note 3} I take the accents in this section from Dittenberger, but I imagine they are mainly conjectural.

^{page 137 note 4} The antiquity of their friendship, ‘to all the Cretans (Κρητάεις)’ is described rather wordily in II. 22–25: it was ‘attested by oracles and all men's knowledge.’

^{page 138 note 1} Of names west of this line (excluding pure Greek formations like Κρίου μέτωποι and adding in italics names taken from Pliny 4. 20. 59) we may note, proceeding from E. to W., on the North, Eleuthernae, ᾿Αμφίμαλλα Rhithymna, Κίσαμος, ᾿Απτέρα, Κυδωνία, Τίτυρος ὔρος Pergamum, Δικτυν ναῖον ὔρος, Κίμαρος (prom.) Φαλάσαρνα and on the south ἡ Γόρτυν, ῾Ρύτιον, φαῖστος, Λίσσην, φοίνιξ ὁ Λαμπέων the island Gaudus, and in the centre the mountain ridge of Ida (᾿´Ιδα) continued in the West by that called τὰ Λευκά I add these names here because it seems likely that the Eteocretans once occupied a larger area of the island than that to which we find them confined in historical times.

^{page 138 note 2} This proverb has been kindly explained to me by Dr. Jane Harrison. Pandareus, King of Lycia, stole (or received from Tantalus, who stole), the golden dog of Minos, a life-like work of Hephaestus, which had stood in the precinct of Zeus. When Hermes came to demand the dog back, Pandareus denied that he had ever had it, but the dog was found, and the thief punished. One of the versions of the story is illustrated by an early black-figured vase in the Louvre (Pottier, Cat. A., 478, pl. 17. 1). For further details I must refer the reader to c. v. of Miss Harrison's forthcoming work on Greek Ritual and Keligion, or to the Art. Pandareus in Roscher's Lexicon of Mythology.

^{page 139 note 1} This is the only meaning of the Dat. after ῥέζειν recognised in L. and S., and it is common. The victim stands always in the Acc.

^{page 139 note 2} My friend Dr. J. G. Frazer has kindly called my attention to a curious account of some pre-Hellenic customs at Phaestos recorded by Antonius Liberalis, Metamorph. 17 ad fin.: there was a festival called the ἐκδύσια in honour of φυτίη Δητώ and in memory of a miraculous change of sex. See further Ain. Journ. Archaeol. 3. 458. If any weight be attached to the statement of Staphylos quoted above, Phaestos may have been originally Eteocretan, since it was the southernmost town of any note in Crete.;

^{page 141 note 1} The one exception would be -itsφα at the end of l. 1 of the φraisoi-insc. But we have seen some reason to separate the -sφα.

^{page 141 note 2} Take almost any page of Pauli's Corpus Inscc. Etrusc.:

.

^{page 141 note 3} In view of some conjectures hazarded by Mr. A. E. Cowley in the Athenaeum (March 16, 1901), it is well to add that this negative statement applies also to Finnish, at all events so far as its vocalism is concerned. See any page of Eliot's, selections (Finnish Gram. Oxford, 1890).Google Scholar

^{page 142 note 1} e.g. 11. 8. 4 p. 512 and 11. 14. 16 p. 532. Other references will be found in Spiegel l.c.

^{page 142 note 2} Or 541, according to Busolt, , Griech. Gesch. ², ii. p. 502.Google Scholar

^{page 142 note 3} [After this paper was in type I received from Mr. Bosanquet a few brief expressions of opinion sent to him by scholars to whom he had submitted photographs of the nomos-insc. It is best to quote them as they stand, though some of the suggestions are excluded by our improved text.]

M. Salomon Reinach writes: ‘I am convinced that the language is Aryan, though by some strange chance the last word of the first line is pure Semitic.’ [M. Reinach seems to be thinking of the IIeb), mitsvah ‘commandment.’]

M. Bréal writes: Tout inconnus qu'ils soient, les mots sermonem redolent Graecum. On peut supposer quelque dialecte grec ayant son aspect particulier, et dont les règles se laisseraient voir si l'on avait un spécimen moins court et moins mutilé. Dans l'hypothèse grecque, on pourrait signaler le mot ἔστε trois fois répété (1. 3, 4 et 5). Ce mot, ainsi que la répétition du nom de Praesos, pourrait faire croire è une délimitation de territoire.

Le mot finissant en -σατο (1. 3) a l'air d'être répété è la ligne suivante.

Hactenus mihi videor satis tolerabiliter potuisse dicere. Reliqua rectius omittam.'

Prof. A. H. Sayce writes: ‘The inscription found some years ago on the site of Praesos which I will call Pr. I. falls into two clauses, each of which begins with a word, or words, with the suffix -t, and therefore presumably in the Nom. Sing., and ends with a word with the suffix -t, presumably the 3rd pers. of a verb. I read… νκαλμιτκ[ι]ος βαρσε α. . οα(?)ρκ. απσετ Μεγ[σ]αρ κρκοκλες γε . . ας ἐπγναναιτ Perhaps the second word is an early form of the name of Praesos.

‘The newly discovered inscription, Pr. II., can similarly be broken up into two initial clauses each with its Nominative and 3rd person of the Verb. We must assume that, as in Pr. I., so in Pr. II., the perpendicular line can denote a division between words as well as the letter 1. Accordingly we get: . . . οναδεσεμε τεπιμιτ σφα . . . [σαα]δυφ . . . αραλα φραισοι ιναι . . . . ρες and . . . καναμος ελος φραισον . . . [ι]τ σααδοφ τενο . . . μαγρα ιναι ρε . . ..

‘In the bilingual (Mysian and Greek) inscription discovered on one of the columns of the temple of Athena at Pergamon we find similar grammatical forms in the Mysian text. The Greek has Παρταρας Αθηναιηι the Mysian U-ÿ-^*-t-a-i…. B-a-r-t-a-r-a-s p-a-u-i-t. I would compare the divine name μητρι ᾿Ιπται found by Prof. Ramsay in an inscription at Goelde in Lydia, and translate: “to Vipta… Bartaras has dedicated.”’

[The fulness with which the epigraphic data have been treated in the opening sections of this paper precludes me from entering into any discussion of these suggestions, as it would involve fruitless repetition. This is a convenient place to mention that in the Times of April 1st, 1901 (or some closely adjacent date) Col. C. R. Conder proposed to recognise the barxe- insc. as Indo-Kuropean, adding an adventurous ‘translation.’—R.S.C.]

^{page 144 note 1} Kieper's map gives Chrysea; his authority I cannot find though I have searched the Indices to many Greek authors. If (as often in Kiepert) it be one of the later geographers, Pliny is clearly a better source.

^{page 144 note 2} The temple of Athena Salmonia near Itanus (Head, Hist. Num. p. 398), a sea-goddess, shows that initial s- in this region was preserved as in Lat. sal, not changed to h- as in Gr. ἅλς.

^{page 144 note 3} See Sillig's critical note ad loc.

^{page 144 note 4} Etym. Forsch, i. 141.

^{page 144 note 5} Bury, Gr. Hist. p. 7 ff.

^{page 145 note 1} Pauli himself was under a misconception on this point, as Kretschmer (Einleit. in d. Gesch. a. Griech. Spr. p. 275) remarks. The examples of -s- are abundant.

^{page 145 note 2} Some other possibilities as to σιαμων may perhaps be set down here, though no one of them can become in the least probable without further evidence: (1) Mr. Cecil Smith tells me of an epithet or surname of the sun-god or moon-god Men, which appears in Lydia and is spelt τιαμου (Waddington, Inscc. d' Asie Min. p. 215; Ramsay, , Cities, etc. i. p. 341Google Scholar). It is conceivable that τια- and σια- should be different attempts to represent the same (palatal?) sound in Greek alphabet. (2) In the ‘Sequanian’Celtic Calendar of Coligny, (Comptes Rendus de l'Ac. d'Inscc. 1897, 703)Google Scholar one of the winter months is called Giamon, and (3) another month, sixth earlier in the list in the same calendar, is called Samon. Mr. E. B. Nicholson has printed a most interesting, if somewhat adventurous discussion of this Calendar in his pamphlet called ‘Sequanian’ (Nutt, 1898). Thurneysen has an article on the insc. in the Zeitschr. f. Kelt. Phil. 1899, p. 523).

^{page 146 note 1} On the vexed question of the date at which θ, φ, χ became spirantic see e.g. Brugmann, Grundriss L², § 739, who inclines to a more conservative view. But such scansions as φιλόσ-φον in Aristoph. Eccl. 571, point to a partly spirantic pronunciation (e.g. something like πφh) in vulgar Attic, although spellings like ἕχω in Attic inscc. (exx. in e.g. Meisterhans Gramm, d. Att. Inschrr. ². p. 66) show equally that there was a real -h- sound in the aspirate, or it could not be imported into another syllable. But it is to be observed that this “affricate” stage had arisen in some other dialects at a very early date, and that rather numerous examples seem to come from the sources roughly grouped together as “Aeolic,” Pindar and IIesiod; cf. the Northern Βάκχος (see e.g. Gust. Meyer, Gr. Gram. ², § 210). Latin words like bracchium (from Gr. βραχίων) carry this pronunciation back to a fairly early date in Magna Graecia.

^{page 146 note 2} Professor Burrows points out to me that the form Λύττος occurs in an insc. quoted by Frazer, , Pausan 3, 313Google Scholar, the date of which can hardly be earlier, and may be much later, than 300 B.C. (See Hermann, K. F., Philologus, ix. p. 694Google Scholar).

^{page 146 note 3} On the use -φσ- for -ψ- if it stood alone no stress could be laid as it appears in many parts of Hellas (see e.g. Kirchhoff, Studien z. Gesch. d. Gr. Atph. ⁴ pp. 95, 116, 121, or the tables in Roberts' Gr. Epigraphy).

^{page 146 note 4} Brugmann, , Grundriss I. § 715.Google Scholar

^{page 146 note 5} ib. § 793c.

^{page 146 note 6} ib. § 769. Possibly in Venetic, though no examples either of -ft- or -pt- occur in the record we possess; but -ht- -hs- appear in Venetic for an orig. -kt- -ks- (Rehtia-, Ahsus), Pauli Veneter p. 256, just as in Osc.-Umb. There is a small Venetic river called Tilavemptus, where the m complicates the question. The Greek equivalent would be, I imagine, τηλε-ϝεμετὸς ‘far-shooting.’

^{page 146 note 7} ib. § 724.

^{page 146 note 8} ib. §§ 774, 782.

^{page 147 note 1} The first change was probably to ofs-, then to ohs-, then os- merely. In Umbrian the change took place even in a -ps- which had arisen from contraction; but Oscan keeps this later -ps- (Osc. opsaum, ‘^*operare, aedificare.’) On these changes generally see It. Dial. ii. p. 495, and more fully Von Planta, Osk. Umbr. Gram. i. § 207 f.

^{page 147 note 2} But Pauli, Ven. p. 318, connects this and the masculine form iiuvants with Lat. i[u)uentius, a Gentile name common in Venetic areas, and this with invenis, etc. In either case the phonetic change would be the same (-ntia → -ntsa), but if Pauli's interpretation is preferred we must not call it a Participial ending. I am bound to add that Pauli derives Iiuvantsa from an imaginary ^*Iuventissa; but for this he has absolutely no analogies to offer; nor do I think it can be made at all probable that such a form would have been contracted in Venetic.

^{page 148 note 1} In Venetic it forms many derivative proper names, Ermon-, Molon- etc. Further north it is common; Teutones is of course a derivative of I.-Eu. teutā Osc. tovto etc. “ciuitas.” So Gallic Redones ‘the charioteer-folk’ (Glück, Kelt. Namen bei Caesar, p. 148). Pauli (Ven. p. 350) quotes also as Gallic the form Vennones beside Vennum, which would be a closer parallel, but he does not state where he found it, and I have not yet discovered the source.

^{page 149 note 1} This statement so far as it concerns Accent in Venetic, I cannot fully justify here. Apart from its a priori probability, it rests upon what I regard as the direct evidence of the puncts which appear in the middle of words in Venetic script. These Pauli (pp. 191–213) has failed to interpret, but it is clear to me that they denote the word-accent; e.g. in zona-s-to (= Gr. ἐ-δω-νήσετο cf. Lat. donâret) the accent is on the second syllable. Accent is denoted in a similar way, by a mark before and a mark after the accented sounds, in the Pada-text of the Rig-Veda. The commonest practice in Venetic seems to be to place a dot on either side of the last consonant of the accented syllable; but until the text of the Venetic inscc. is put upon a sounder basis (a task I hope to attempt) the theory cannot be properly tested.

^{page 149 note 2} Spelt eriimo-, where the ii- denotes the same ‘open -ῖ-’ as in the common terminations -iia -iios (cf. the Latin spelling filea) where the second -i- originally denoted merely the ‘glide’-consonant (i-) between the vowel -i- and the next vowel. But it was from this, I now believe, that the symbol ║ came to be used for e over a large part of Italy; see Ital. Dial. ii. p. 467.

^{page 149 note 3} If there were, more traces of religious formulae in the insc. one would guess that irere irereiet was part of a prayer ‘auctu augeatur’ (or ‘augeantur’) or the like; cf. ‘with blessing I will bless thee’ etc.

^{page 149 note 4} It is well to observe that Gr. ἄρχω Skt. arhati has lost no initial sound.

^{page 150 note 1} See e.g. Pauli, Veneter, p. 403. For ō→ū see ib. p. 401. For the Phrygian inscc. see Ramsay, , Journ. R. Asiatic Soc. 1883 (xv) p. 128 ff.Google Scholar; and Kuhn's Ztschr. 1887 (xxviii) p. 387 ff. Final -m does not occur and forms like κακουν mataran are umnistakeable Accusatives.

^{page 150 note 2} And almost certainly Phrygian, where (εἴτου ῾ἔστω ᾿γλουρος ‘gold’ = χλωρὸς) seems to correspond to Greek ω; cf. Kretschmer p. 224.

^{page 151 note 1} See Kretschmer op. cit. p. 230 and the multitude of divergent writings he cites. He himself quotes as many exceptions to, as examples of, his own view. The solitary example alleged of Phryg. s=I.-Eu. k is the Pronoun (dat.) σεμου(ν) ῾τούτῳ᾿ which is equated with the O. C. SI. dat. semu ῾τούτῳ ᾿ and taken to come from I.-Eu. *κ῀ἰ- Both forms contain the pronominal affix -smo-, but the Phrygian σε- (fem, σα ῾ταύτῃ᾿) may, in the present state of our knowledge, just as well contain I.-Eu. κἰο-κηἰο-σἰο- or lio or even so- (like Lat. if sum and Osc. ekso-) as I.-Eu. ki-; all these pronominal roots occur in different languages (Brugmann, Grundriss II. § 409). Even in the second syllable the identity of the Phryg. and O. C. SI. forms is not certain, see Solmsen, , Kuhn's Zeitschr. 34 (1897) p. 50Google Scholar (he accepts the current view of the first syllable).

^{page 152 note 1} Kretschmer op. cit. p. 231. There is no doubt that I.-Eu. palatal gh became a spirant, but so it did in some of the centum languages.

^{page 152 note 2} Very fully collected in Pauli's Veneter; they include over 300 inscc. in the Venetic alphabet (which we may note has no q), about a score of glosses and place-names, some 200 Personal names, with another 400 or 500 from Pannonia and Illyria where a speech akin to Venetic was certainly spoken.

^{page 153 note 1} There are two apparent exceptions of the kind which prove the rule; Liquentia the river and emporium of the Latin Colony Concordia; Quasauna C.I.L. v. 3463, with a characteristic Gallic suffix, the wife of a man from Aesium, in the ager Gallicus.

^{page 153 note 2} Of course this contains I.-Eu. -κυ- not I.-Eu q. But if Venetic delabialised a full -u- it must surely have corrected the tendency of q to become qu

^{page 153 note 3} χ denotes g or something like it; e.g. Ven. Exetor gives a derivative Egtoreus in Latin alphabet. There are no signs for the Mediae in the Ven. alphabet; Boius is written φ ohiios.

^{page 154 note 1} Some possible limitations have been mentioned above.

^{page 154 note 2} §§ 22 and 31 (f.). Briefly they are: σμερ, σμίνθος,(Σμισίων) Chrysa, Σμινθεύς Χρύση Λαβύρινθος ῾Ραβίνθιος, —᾿´Ιδα Πέργαμον,—Κίσαμος

^{page 154 note 3} Words containing the root of ἄνθος Skt. anahas, must of course be put on one side.

^{page 155 note 1} On Iapygians, Bruttians and Pelasgians see Ital. Dial. 25 A. Rem. p. 15.

^{page 155 note 2} See Ital. Dial. pp. 359 ff.

^{page 155 note 3} Note here again the -σ- between vowels.

^{page 155 note 4} ‘Pistacia Terebinthus, a Mediterranean tree, yields by incision the Chio or Cyprus turpentine.’ Le Maout et Decaisne, General System of Botany [Trans. Hooker] s.v. Terebinthaceae, a reference I owe to my colleague Mr. H. Spencer Harrison, D. Sc.

^{page 155 note 5} I use this term rather than Pelasgian only because the Pelasgian question is strictly irrelevant to the matters discussed in this article.

^{page 155 note 6} If it is an I.-Eu. word, the root might be that of Gr. σμάω Eng. smear (smaḭ- or smeḭ-).

^{page 156 note 1} The name Σμινδυρίδης which must be connected with Σμινθεύς occurs at the Achaean colony Sybaris, Herodt. 6. 127.

^{page 156 note 2} The only group of sounds which could be cited as non-Indo-European is initial kb-, which is fairly frequent, e.g. in Κβονδίασσις But the combination is not merely non-Indo-European but non-human, i.e. utterly unpronounceable, unless either a vowel is understood after (or before) the k, or, which is clearly the right alternative, the β be taken to denote here, as it does in so many other places, nothing more solid than a consonantal ιι or the spirant u. I know nothing about Lycian; but so far as Kretschmer's arguments as to this language rest on the groups kb-, tb-, they are certainly misleading.