Published online by Cambridge University Press: 22 April 2026
[The graphs 〈etuvanoinu〉 and 〈etokoinu〉 on the bronze tablet of Idalium do not contain optatives duwánoi and
, but aorist indicatives éduwan and édōk'. Éduwan comes from *édoan for édon in the same way that éthijan comes from *éthean for éthen. The infinitive dowénai is scarcely dow-énai: evidence outside Italic for a root form dow- is not convincing. Either it is do-wénai, with a suffix otherwise unknown in Greek, or a development, partly analogic, from *do-énai. 〈oinu〉 is perhaps oîn nu, containing a particle cognate with Attic oûn; PGk. *óen could account for both, since a sound law PGk. *-en > Cypr. -in can be set up. The Cypriote thematic active infinitive ending is -ēn, not -en.]
1 Eduard Schwyzer, Dialectorum graecarum exempla epigraphica potiora (1923); referred to hereafter as Schwyzer.
2 Carl D. Buck, The Greek dialects (1955); referred to hereafter as Buck.
3 Olivier Masson, Les inscriptions chypriotes syllabiques (1961); referred to hereafter as Masson.
4 The lack of special signs ha he hi ho hu is no evidence that Cypriote lacked h, any more than the lack of signs to distinguish voiced, voiceless, and aspirated stops indicates that the three orders had fallen together. Cf. Thumb (revised by Scherer), Handbuch der griechischen Dialekte2 2.164 (1959); referred to hereafter as Thumb-Scherer.
5 Die Inschrift von Idalion und das kyprische Syllabar 88 (1874). On pages 93 and 95 Schmidt left the undeciphered signs in their original shape.
6 ‘Die wichtigsten kyprischen Inschriften’, Curtius Studien 7.219–64 (1875).
7 BB 6.153.
8 Cf. the examples in Liddell-Scott-Jones, A Greek-English lexicon, s.v. nûn.
9 Sammlung der griechischen Dialekt-Inschriften, vol. 1, ed. Hermann Collitz, 29–30 (1884); referred to hereafter as Collitz-Bechtel.
10 Die griechischen Dialekte 2.154–5 (1889); referred to hereafter as Meister.
11 Die griechischen Dialekte 1.69 (1891); referred to hereafter as Hoffmann.
12 Inscriptiones graecae ad inlustrandas dialectos3 no. 4 (1910).
13 Cf. note 1.
14 Cf. note 2.
15 Cf. note 3.
16 Except that Solmsen and Schwyzer accent the first word ê. In Eduard Schwyzer and Albert Debrunner, Griechische Grammatik, vol. 2 (referred to hereafter as Schwyzer-Debrunner 2) the accent is è on page 322 and ê on page 335.
17 Meister 275, Hoffmann 264, Buck 126, Masson 239.
18 Die griechischen Dialekte 1.431 (1921); referred to hereafter as Bechtel.
19 1.700, 702 (1939); this volume is referred to hereafter as Schwyzer, Gram.
20 The last two mention only duwánoi.
21 The former of these seems more likely to me, despite Leumann's arguments in favor of the latter, Morphologische Neuerungen im altindischen Verbalsystem 27, 45–6. But for the present discussion it is immaterial whether it was Greek and Italic or Indo-Iranian that innovated in the reduplication.
22 Still preserved, thematicized, in Mod. Gk. dhidho (Thumb, Handbuch der neugriech. Volkssprache'2 125) beside dhíno (ibid. 128), which is a late remodeling of dhidho.
23 Contrasting remarkably with their conservatism in preserving words like iatêr, kasígnētos, oîwos.
24 The root shape duw- has been explained by Bartholomae, as will be shown below.
25 Hesychius. This is the only example where -ánō is preceded by a nonobstruent consonant.
26 For a list see Schwyzer, Gram. 1.700.
27 List ibid. 699. Note that in luggánō the internal nasal belongs to the root, at least descriptively, and that pimplánetai, Il. 9.679, is, strictly, a ne/o-extension of pím-plē-mi, pím-pla-men.
28 The Baltic causative-factitive formant -in- (e.g. Lith. bùdinti Vecken', leñgvinti ‘lighten‘) has a quite different function, is not limited to present stems, and phonologically can continue *-in- just as well as *-n-.
29 Paradigmatic example Arm. Ik'anem, Gk. limpánō ‘I leave’ beside Lat. linquō, Skt. rinákti, Lith. dial, palinkt. Note that the Armenian present of ‘give’ is not *t(g)anem or the like, but tarn.
30 It is altogether unlikely that the Greek root pneu- ‘breathe’ should have arisen from a nasal present *p-ne-w-, as supposed by Pedersen, IF 2.314, and tentatively accepted by Schwyzer, Gram. 1.696.
31 New Testament adaptations of hístēmi are rather histáō, histánō, stánō; cf. Blass-Debrunner (tr. Funk), A Greek grammar of the New Testament 9–10 §93.
32 Cf. Blass-Debrunner §73 and Schwyzer-Debrunner 2.286. From the Classical Greek tense opposition hístēmi : héstēka comes the Modern Greek lexical opposition stíno, sténo ‘I set up’ : stéko ‘I stand’.
33 Van Herwerden, Lexicon graecum suppletorium et dialecticum2 (1910), s.v. dídōmi, in fact takes
as aorist, but does not explain its formation.
34 The only unquestionable example of such a pair I have been able to find is khéō : khúto. In the type phúō : éphu the present probably has a ye/o-suffix, and I suspect that Schulze was right, QE 63 n. 1, in seeing the same suffix in seúomai beside aor. súto; here -eu- would be analogic for *-ei-, as in the type basileúō for basiteíō. In the type dékhomai : dékto, the aorist is not quite above suspicion of being an archaic kind of s-aorist; cf. Schwyzer, Gram. 1.751, with Nachtrag. Thematic and athematic forms of the root *klew- are distributed without regard for aspect: kléomai ‘I am famed’ is present, but so must be the participle underlying the proper names Klúmenos and Kluménē, so that the athematic aorist relics klûthi, klûte beside the usual ékluon ‘I heard’ cannot be paired with kléomai as a second example where the thematic vowel is an aspect marker.
35 Cf. Chantraine, Grammaire homérique 2.195, and Schwyzer-Debrunner 2.256–60.
36 Figures from Gehring, Index homericus 212–3.
37 But one may suspect that our text has otrúnoite rather than otrúnaite mainly because of the parallelism with didoîte. In 9.518 otrúnō is no doubt aorist (cf. theiõ 517), but its formal ambiguity may have contributed to the aspect of otrúnoite in 11.357.
38 Which leads to what I think is the truth underlying Vendryes' doctrine : Indo-European roots of lexically telic aspect formed uncharacterized aorists and characterized presents. This I learned from Paul Tedesco in 1953, and in 1960 I heard it again from Werner Winter; cf. also Meillet, Introduction à l'étude comparative des langues indo-européennes8 198–9.
39 The aspect difficulty would of course not apply to
if it were an aorist optative created in the way fancifully outlined above.
40 Anósija ‘unholy things’, suggested by Liddell-Scott-Jones 148, fits the sense better than the anosija ‘Friedlosigkeit’ of Deecke and Siegismund 256. Although aitía, axia (Schwyzer, Gram. 1.469) attest the possibility of an abstract noun anosia to the adjective anósios, the meaning would surely be ‘impiety’, as given by Buck 352. But ‘may impiety befall him’ is a strange sort of curse; impiety would be a necessary and automatic consequence of breaking the agreements, not something to be wished on the violator.
41 Philologus 35.40. Ahrens actually gives the verbs as duwánoije and
, accepting the then current reading of nu as je and imagining an optative formation *-oijon *-oijes -oije corresponding to Skt. -eyam *-eyas *-eyat.
42 Schwyzer-Debrunner 2.322 accept the common view of a prescriptive optative, as does also Schwyzer in his notes to his edition of the inscription, where he refers to Slotty's treatment in Der Gebrauch des Konjunktivs und Optativs in den griechischen Dialekten; but on page 335 of Schwyzer-Debrunner, Sommer's view of the construction is quoted, with no reference to a possible alternative. Apparently then one or both of the two Swiss scholars found Sommer's view convincing, at least for a time.
43 It was probably a mistake of the 19th-century scholars to accept Deecke and Siegismund's value za for the first sign of this word rather than Schmidt's ga, for which Lejeune, BSL 50.70–1 (1954), has given persuasive arguments.
44 The father of Onasilus.
45 This is the regular Cypriote treatment, contrasting with the Koine-influenced édōken, onéthēken, katéthēken mentioned below in the discussion of (oinu).
46 Probably continuing PIE forms of the type *e-dhE-ent, *e-dO-ent, *e-gwA-ent, *e-bhuH-ent.
47 Itself probably analogic to the s-aorist, according to the model *édeix : édeixan = ês : X. Only after êsan had been created was the old plural êen, ên free to be reevaluated as a singular, ousting ês.
48 -an apparently also from the s-aorist, this time on the model of the first singular: édeixa : édeixan = êa : X. Otherwise Schwyzer, Gram. 1.665.
49 Attested once also in East Locrian, Schwyzer 359.
50 Likewise Schwyzer, Gram. ibid. Since éthe-an and édo-an agreed in ending with the s-aorist édeix-an, the old participles thé-nt- and dó-nt- could be remodeled to thé-ant-* and dó-ant- after the aorist participle deix-ant-. In Attic-Ionic the ending -san was not identical with the -an of the aorist, and ‘be’ had no participle *ēsant-; hence dónt-, thént-, etc. remained.
The curious absence of éthen and édon in Homer suggests that êan, éthean, and édoan are quite early in Achaean, and were the regular forms already in the pre-Aeolic stage of Greek epic, a chronology that agrees well with the presence of -an in both Arcadian and Cypriote, which prima facie suggests a date before the interruption of communication at the end of the Mycenaean empire. (Homeric unaugmented hésan on the other hand can be overlying an augmented *héen.)
51 The first of these changes is found in various other Greek dialects, e.g. anéthian for anéthean in parts of Boeotia.
52 Similarly R. E. Carter, Classical philology 48.24 (1953).
53 Martinet's treatment of the problem, Word 9.259 n. 25, is conscientious but does not carry conviction.
54 Another possibility, which avoids positing a stage [ou] between [o:] and uo, is to suppose that in */dooee/ the intervocalic /o/ became nonsyllabic and fell in with reflexes of PIE *w. Schmalstieg's derivation of -av- from *-°Hw-, Word 12.258 (1956), involves the still dubious development of v from laryngeal; a nonautomatic ablaut between preterit and infinitive that is otherwise unknown in Lithuanian derived verbs, and all but unheard of in root verbs (exceptions only
: im̃ti ‘take’ and
beside dial, ẽjo : eĩti ‘go‘); and schwa secundum, an entity of very questionable status.
55 This view of -davati I learned from Paul Tedesco in 1952–3. It is also that of André Vaillant, Manuel du vieux slave 1.317 (1948). Dav-forms do not occur at all in the oldest Church Slavic manuscripts, Zographensis and Marianus.
56 Endingless locative plus particle *-ay; cf. Benveniste, Origines 129–31.
57 Figures from Gehring, Index homericus.
58 But it is not quite certain that émmen is a genuine Homeric form; all five of its occurrences can be read émmen'.
59 An original ending -nai, assumed by some in this word (e.g. Bechtel 1.333) is unnecessary beside the unavoidable -énai of doûnai, theînai, eidénai, etc., and does not agree with the full grade characteristic of the Proto-Indo-European locative of n-stems. For the genesis within Greek of -nai in forms like dûnai and histánai cf. Schwyzer, Gram. 1.808.
60 Cf. Chantraine, Grammaire homérique 1.35; Chantraine however adopts Benveniste's *do-wnai, which, as Schwyzer observes, Gram. 1.809, is impossible.
61 Cf. Chantraine 1.42.
62 Beattie expresses well-justified doubts in his recent treatment, Rheinisches Museum 101.138–46 (1958).
63 Myc. 〈porenaqe〉, PY Tn 316 4x, is pretty clearly a noun in the accusative case; cf. e.g. Lejeune in Mycenaean studies Wingspread 1961 92–3 (1964). Ventris and Chadwick's early guess that it might be an infinitive (Documents 285) should be forgotten. In any case, a dialect with uncontracted 〈ekee〉 should offer uncontracted (poreena) as infinitive to *phórēmi.
64 It is not obvious whether -iienai here is formed with the same secondary -nai as Att. hié-nai (for which cf. note 59), or is to be read as -iiênai, directly aping the aorist hênai* from *he-énai (cf. Homeric didoûnai vs. Att. didónai). Or the form might even be an aorist infinitive, with *ee becoming ije in Pamphylian rather than ē. Lurje's alternative restoration to a[n]iiénai, Klio 37.9 (1959), does not affect the problems here under consideration.
65 Whether the preform of the suffix was *-en- or *-sen- is not clear to me. Perfect infinitives like eidénai are no evidence for the former, since they need not have been created before the loss of intervocalic *s; but eînai favors *-en-, since *es-senai would have given Attic-Ionic and Arcadian *esénai and it is scarcely possible that eînai goes back unremodeled to an *esen- formed when the rule ‘s plus s yields s‘ still operated. On the other hand Vedic infinitives in -ane and -áni are rare and dubious, while -sáani is well established—cf. Wackernagel-Debrunner 2.2.179, 924–5; and it seems unlikely that pre-Greek would have created hiatic forms (*segheen, *stāaenai) after the contraction of pre-IE (or PIE?) vowel sequences (*eest > *ēst > Gk. ês) and before the rise of new vowel sequences upon the loss of intervocalic *s and *y. The bulk of the evidence thus favors *-sen-, and all that is lacking is an explanation for eînai.
66 Lesbian forms like kérnan reflect spread of the thematic ending to athematic stems, the converse of the Boeotian and Thessalian type pherémen.
67 ‘Man wird gut tun, … einen Hiatuskonsonanten darin zu sehen, der nach Willkür oder Bedarf eingeschoben wird.‘
68 Thus the 〈toena〉 doéna[i] of Masson 308 (first published by Sittig in Symbolae Danielsson) is no evidence that Idalian dowénai is from a form without w, since w does not occur at all in this inscription and 〈oikonaone〉 (4) and 〈koikisi〉 (8), if rightly interpreted as containing woik- ‘house’, show loss even of initial w.
69 23 examples: katéworgon, Ketiêwes, wétei, Edaliêwes, e(u)wrētása(n)tu (2), woíkōi, basilêwos (3), álwō(n) (3), oíwōi, rhówon, hierēwijan, Diweíthemis, wépija,
(2), woi, aiweí, Edaliêwi. 〈uvaise〉 (3×) cannot count as an example of either preserved or lost w until its etymology is better known than at present. Paîdes, paídōn, paisí, paîdas (2× each) belong to a word whose original shape is not clear, and which even in Homer has only a monosyllabic stem outside the nominative and vocative singular (cf. Chantraine 1.29). If 〈panonione〉 (10) and 〈panoniose〉 (22) are rightly interpreted as ‘with all salable products’ (Buck 212), they contain a word whose traditionally accepted cognates have *w- (cf. Boisacq 1082), but which lacks all trace of it in Greek (Schwyzer, Gram. 1.225, Chantraine 1.125). But Lejeune, Mycenaean studies Wingspread 1961 108–9, gives excellent reasons for reviving Meister's connection (225–6) of the Cypriote words with the root onā- ‘use, enjoy’, which never had a w. The loss of w in orúxē (3×) < *orwúxē < *orwíxē is only to be expected, given the change of i to u; the Cypriote syllabary did not even have a sign 〈vu〉. At most one might object to the lack of a glide following the u of khrauómenon (9); but the word is written khrauzómenon in line 18, and I suspect that 〈o〉 in line 9 is a mistake for 〈zo〉.
70 Thumb's explanation is much superior to that of Wackernagel, Vermischte Beiträge 54.
71 Poekhómenon (19, 21) does not prove that old oe was kept unchanged in Idalium, since this compound exhibits analogic replacement of pos- by po-, which may well have been later than the postulated change of *oe to uwe. The lack of strict parallelism between ija, uwa, and the postulated uwe, with glide, and io, without it, is only apparent: the Cypriote syllabary lacked a sign 〈jo〉, and perforce
had to be written io.
72 On the infinitive ending see below in the discussion of 〈oinu〉. The development of ee to ë is paralleled by the development of -oo to -ō in the genitive singular of o-stems (e.g. tô passim, argúrē 6, 13, 15, 17), which was earlier than the raising of final -o to -u.
73 Indeed, I see no reason not to phonemicize [e:] as /ee/ in Cypriote.
74 To avoid excessive repetition of alternatives in the following discussion, I shall arbitrarily assume that oï had contracted to oi in fifth-century Idalian.
75 Cf. Boisacq 728. The explanation as *eón ‘being’ is semantically attractive, but works phonologically only in Attic, and there only if we assume an accent shift when the word changed status from participle to particle.
76 It is especially puzzling that our Homer text agrees with Attic against the Ionic authors (other examples Chantraine 1.16). The spelling of the Amphipolitan inscription just cited suggests that West Ionic agreed with Attic on this feature, but may be an Atticism (Amphipolis was originally an Attic colony, but early became independent; the language of the inscription is otherwise mostly Ionic). As far as I know, the spelling ōn is not attested inscriptionally in any dialect that kept the contraction product of oe and oo distinct from old long ō. Bechtel 2.273 cites ôn from Epicharmus; but literary spellings can have been altered by ancient scholars who thought that all non-Attic dialects had ôn.
77 Hypotheses of Wackernagel and R. Meister are rejected by Schwyzer-Debrunner 2.586.
78 Or perhaps inalalisménan; cf. Lejeune, BSL 33.68, and Masson ad loc.
79 The initial of inalalisména must then have been imported from collocations where the morpheme en regularly became in. Likewise Homeric pinutós for *penutós, discussed recently by Eric Hamp, Gioita 38.198–9, cannot be explained by a simple pan-Achaean raising of e before nasal; attempts to explain this word should not neglect the parallel with
pointed out by Schulze, Quaestiones epicae 323 n. 3, to which can be added Libúē : Lebúa and perhaps Horn, písures : Lesb. pésures.
80 If 〈itate〉 Masson 11, 213a ‘here’ be read intháde (íntha nu, Schwyzer 683.6, is a false reading), its initial is analogic to the morpheme in ‘in’. One might also read itháde, and cf.
. The Hesychian gloss iggia: heîs, Páphioi is too obscure to contribute anything now to the elucidation of Cypriote phonology.
81 A first plural in -min would be very welcome; on a possible min ‘me’ from Chytroi see below.
82 The glosses assembled in Meister 304–14 and Hoffmann 104–26 offer no examples of either -en or -in for *-en. The statement that Homeric words like doúpesen and émarpsen come from Cypriote of course proves nothing about the shape of their endings there.
83 As was pointed out in note 45, the usual Cypriote treatment of verb forms in -e is exemplified by the édōk' which we have now identified in the Idalium Bronze.
84 Glosses like athrízein: rhigoûn, Kúprioi (Hesychius) are hardly useful as evidence for the exact shape of the Cypriote infinitive ending. More significant would be the mauíēn ‘zēteîn’ cited by Meister 2.210, if only there were something more than a chain of tenuous hypotheses to link the form with Cyprus.
85 Cf. Schwyzer Gram. 1.606 on mod. Gk. emén, eména.
86 Meister's later readings mé n', tánde ne with an enclitic ne ‘here’, e.g. Berichte der sächsischen Akademie 63.27 (1911), are rightly rejected by Masson 290.
87 E.g. Xenophon, Anabasis 1.1.2: Kûron dè metapémpetai … anabaínei oûn ho Kûros … * (Darius) sends for Cyrus … and so Cyrus goes up … '
88 Denniston 425; cf. Schwyzer-Debrunner 2.586.
89 Except perhaps in Herodotus when combined with ouk. Cf. Kühner-Gerth, Ausführliche Grammatik3 2.2.161, and Denniston 424 n. 1.
90 Schwyzer-Debrunner 2.584: ‘in Wahrheit, Wirklichkeit; in der Tat, gewiss’.
91 Cf. e.g. Sadnik and Aitzetmüller, Handwörterbuch zu den altkirchenslavischen Texten 277; Vasmer, Russisches etymologisches Wörterbuch 2.223; Pokorny, IEW 770.
92 Krause-Thomas, Tocharisches Elementarbuch 1.170, 172.
93 On the supposed nu of Schwyzer 683.6 see note 80 above.