José M. González, The Epic Rhapsode and His Craft: Homeric Performance in a Diachronic Perspective
Key to the Books of the Iliad and the Odyssey
Part I. The ‘Homeric Question’
1. Dictation Theories and Pre-Hellenic Literacy 2. Dictation Theories and Archaic Art 3. The Technology of Writing 4. The Euboian Connection 5. Archaic Inscriptions before 650 BC 6. Early Homeric Scholarship and Editions Part Ⅱ. Rhapsodic Performance in Pre-Classical Greece
7. Homer the Rhapsode 8. Hesiod the Rhapsode Part Ⅲ. Rhapsodic Performance in High-Classical Athens
9. The Rhapsode in Classical Athens 10. The Rhapsode in Performance Part Ⅳ. Rhapsodic Performance in the Late Classical and Post-Classical Periods
11. The Performance of Drama and Epic in Late-Classical Athens 12. The Performance of Homer after Ⅳ BC Part V. Aristotle on Performance
13. Rhapsodic hypokrisis and Aristotelian lexis 14. The Aristotelian tekhnē of hypokrisis Conclusion Appendix. The Origin of the Term hypokritēs Bibliography
5. Archaic Inscriptions before 650 BC
A collection of the earliest inscriptions down to 650 BC is conveniently found in Powell 1991:119–186.  In this chapter I review the linguistic evidence they contribute and confirm that it agrees with the chronology established above. I will only comment on features that are relevant to my inquiry and will largely pass in silence over the ones that are unproblematic.
Of interest, for example, are the Euboian fragments inscribed with the owner’s name Tεισον[ (125).  This is doubtless Tείσον[ος, the genitive of the broadly attested Tείσων.  Tεισ- is derived from τεῖσαι, the aorist of τίνω.  Johnston does not date these fragments, but they are surely much earlier than the Eretrian law IG Ⅻ.9 1273/4 from ca. 525 BC with the reading τίν[υ]σθα(ι) on which Cassio (1991–1993:190) so heavily relies to argue against the view that τει- should be restored to archaic Greek texts that have rather uniformly transmitted τῑ́νυμαι/τίννυμαι and ἔτῑσα for τείνυμαι and ἔτεισα.  These Euboian fragments disprove Cassio’s statement that at least in the West Ionic area during the eighth and seventh centuries the form τείνυμι was not in use: quite the contrary, τεῖσαι presupposes it. 
At 127–128 we meet with a late geometric kratēr fragment that Johnston (1983:64) has dated to ca. 700 (LG Ⅱ). The dipinto on it reads ]ινος μ’ εποιεσε̣[, and it does not print the -ϝ- in *εποιϝεσε. If this were proof that all inner -ϝ- had been lost by the end of the eighth century, it might place some strain on my chronology. But this conclusion is hardly necessary. The dating of Euboian LG pottery presents special challenges.  Derivative from Athenian and Corinthian styles in its beginning, scholars remain uncertain how the Euboian artists kept pace with their models. The dearth of stratified deposits and other chronological controls contributes to diverging expert judgments,  and the boundary of late-geometric and subgeometric cannot be specified narrowly.  Buchner (1970–1971:67), the excavator of Mazzola, the site by the Mezzavia hill where the dipinto was found, did not give the fragment an absolute date. He merely noted of its site that “[it could] be assigned to the period between the middle of the eighth and the beginning of the seventh century,” and added that it was deserted during the first quarter of the seventh century (64).  By about 680 BC the island exhibits a general pattern of depopulation and the oriental elements in evidence during the LG Ⅱ disappear from the record.  Given that archaeological contexts are often dated by their pottery, in the absence of Buchner’s detailed excavation report one cannot know what else he based his chronology upon.  Klein (1972:38) thought that the site’s pottery “might well suggest the output of a small group of artists in perhaps a single workshop during the second half of the eighth century BC,” although he too acknowledged that the construction of the excavated structures from which the pottery was retrieved started ca. 750 and lasted 50–75 years, with the last phase of structure Ⅳ beginning by about 700 BC.  Presumably these are the reasons why both Jeffery (1976:64) and Johnston (1983:64) chose about 700 BC as the date for the fragment.  Thus, downdating the dipinto to the first decades of the seventh century is perhaps not impossible,  and this is precisely what Johnston (in Jeffery 1990:453.1a) allows for, where his mature judgment is a date of ca. 700–675.  There is another reason why even a date of ca. 700 BC for the dipinto hardly means that all inner -ϝ- had been lost by the end of the eighth century. Scholars have disagreed whether intervocalic and postconsonantal -ϝ- were lost contemporaneously. On the basis of different reconstructions of Attic reversion, Hoffmann (1891–1898:3.344) thought that -ϝ- had been lost earlier between vowels than after rho. Brugmann seemed to disagree, and only asserted the unproblematic contrast between initial ϝ-, which held out longer, and inner -ϝ-, which was lost at an earlier time regardless of context. But he had to accept that the loss of ϝ in Arcadian between vowels and after ν and its preservation after ρ supported Hoffmann’s view.  Therefore, even assuming that the dipinto accurately reflects the potter’s pronunciation, there is no need to assume that the (occasional?) loss of ϝ after a diphthong ending in iota must imply the wholesale loss of intervocalic and postconsonantal ϝ.  Indeed, it is surely significant that the dipinto’s phonic context, from which -ϝ- is missing, is no ordinary postvocalic or postconsonantal one, and this for two reasons. First, because the initial π- would have encouraged the loss of ϝ by a dissimilation of labials. Szemerényi (1974:29) makes this phonetic development responsible for the early loss of ϝ in παῖς < *παϝις, even in “so conservative a dialect as Cypriote, which tenaciously preserves intervocalic digamma”; and he adds that “by the same type of dissimilation … the original πολέϝα, πολέϝων, etc., could very early, and that means before the general loss of intervocalic digamma, develop into πολέα, πολέων, etc.” (my emphasis). This is, by the way, the reason why the form παίζει rather than *παϝιζει on the Dipylon oinokhoē does not pose a problem to my chronology. The second reason to consider the environment exceptional is that, since the iota in the diphthong -oy- would prevent vowel contact upon the loss of -ϝ-, the sequence -V i ywV j - might well have facilitated an occasional, early neglect of -ϝ-.  Thus in Eretria we find both ποιϝ[εσας (IG Ⅻ.9 257, ca. 550)  and the roughly contemporaneous ποιεσ̣α̣[ι and ποιει (IG Ⅻ.9 1273/4, 3.2–3). Note also the inconsistent use of -ϝ- in ποι]ϝέοι and ποιέοι in the same line of the Elean inscription IvO 16.18. And the co-occurrence of πρόξενϝος and ἐποίει in early sixth-century Corcyra (IG Ⅸ.1 867).  Therefore, the attested μ’ εποιεσε̣ does not suggest, much less commend, the view that the sequences -εϝᾱ̆-, like -ε(h)α-, had already devolved into -εᾱ̆-, with the consequent possibility of contraction.
This view is confirmed by Δηϊδαμαν on an inscription from Amorgos (144).  The etymology of δηι- is contested, but it seems fairly certain that *daːw- underlies it (cf. Frisk 1973–1979 s.vv. δαΐ, δαίω, and δήϊος; and Chantraine 1999 s.v. δήιος). Like πόλ-ις, Δηι- could be an i-stem root extension of *δᾱϝ-: *δᾱϝ-ις, gen. *δᾱϝιoς. According to Lejeune (1972:172), the outcome of the intervocalic cluster -wy- in *δᾱ́ϝιος can be explained either as a metathesis -wy- > -yw- or as assimilation -wy- > -yy-. I suggest here again *dæːwy- > *dæːyw- with early loss of -w- after -y- and homorganic glide development /dæːyy-/; or simply (and indistinguishably) *dæːwy- > *dæːyy- by an assimilation that is to be dated rather early.
Further confirmation comes from the famous Naxian Nikandre inscription found on Delos and dated ca. 650.  Besides the synizesis of Δεινομένε͜ος and the loss of -w- in the cluster Δϝεινο-,  of interest is the ῑ̓ο- of ἰοχεαίρηι. Its etymology is *ῐ̓σϝ-ος. Lejeune (1972:136 §130) classifies it with the “groupe ancien *-sw- entre voyelles” that experienced voicing *-zw- followed by a softening of the -z- to *-hw- and, in Ionic, its loss with CL. Thus, we have in ῑ̓ϝο- the same sequence that obtained for ποιϝ-. Correspondingly, we can expect the same early tendency toward neglect and loss of the -ϝ-.
One final item might be added. It is a Protocorinthian lēkythos now in the Boston Museum of Fine Arts, dated to no earlier than ca. 700 and no later than ca. 650.  Jeffery (1990:83–84) tentatively ascribed it to Eretria (most scholars think it is Chalcidian). All agree that the alphabet is Euboian (Buck 1998:192 no. 9). Jeffery observes that the date of a copy need not be as early as the date of its original, and she adds that the lettering does not suggest so early a date as ca. 700. The inscription includes the (gen.) name Ἀγασιλε̄́ϝο̄  and the verb ἐποίε̄σεν. Here we see, again, an intervocalic -ϝ- in -λεϝο that coexists with an ἐποίε̄σεν that has neglected its own. Buck (1902:47) called the -ϝ- in Ἀγασιλεϝο “the first incontestable example of the preservation of an original ϝ in the Attic-Ionic dialects.” Hoffman had opined that the ϝ in Naxian ἀϝυτοῦ and Attic ἀϝυτ̣[άρ (both secondary glide sounds) proved nothing more than an acquaintance with the character and its sound as it was still being used in Doric.  Thumb countered that these words pointed to a not too remote time in which the etymological value of ϝ was still preserved in Attic-Ionic.  Buck (1902:47) went further: “As Thumb remarks, there is nothing against the assumption that in the Ionic of the Cyclades and the West the ϝ was still preserved in [the] 8th century at least; and this statement might be made to include the early 7th century without fear of controversion” (his emphasis). 
5.1 The Inscription from Cumae
The previous section reflects an exhaustive review of all the inscriptions included in Powell (1991:119–186) and mentions all that might seem to call my argument into question. I do not believe that I have met there with any insuperable objection. I have only left out one inscription that deserves a more detailed treatment. This is what follows below. It accompanies two partial abecedaria (Corinthian and Euboian) on an early Protocorinthian conical lēkythos discovered at a grave in Cumae and dated to ca. 700 BC.  Until recently, the prevailing opinion was that the inscription, hisamenetinnuna, in careful and elegant Euboian letters, was a later Etruscan addition, possibly the name of the vase’s owner. This judgment was not uncontested, however, since a persuasive Etruscan reading had not yet been offered.  Cassio is not the first scholar who has sought to construe it as Greek,  but his is certainly the most formidable attempt to date.  But, as I will show below, I do not think that his impressive tour de force ultimately succeeds. Perhaps more importantly, it is also unnecessary, for the convincing analysis of the inscription as unexceptional Etruscan by Colonna (1995:332–341) makes this scholar’s view preferable to the problematic Greek of Cassio.  Nothing needs to be added to Colonna’s able interpretation. He translates, “Hisa Tinnuna ‘fa (como dono)’,” an extraordinary gift inscription that reflects the ordinary practice of Etrustcan elite gift exchange.  Here I will only comment on those aspects of Cassio’s analysis whose combined weight makes his argument unsupportable. My criticism is twofold. Not only do I consider his linguistic reasoning wanting or unconvincing. Even if I should accept it for the sake of argument, I find implausible the meaning he ascribes to the inscription. I will take each of these points in turn.
Cassio (1991–1993:190) wants to read hisa as ἵσα < ϝίσϝα. But as Lejeune (1972:177) explains, the syllabification ϝι.σϝος leads to ἴσος whereas ϝισ.ϝος leads to the East-Ionic ἶσος and to the Hellenistic koinē ἵσος.  Therefore, Euboian syllabification would call for ἴσα, with the spiritus lenis. Adducing the Heraklean hίσον as a parallel bears no weight, since the expected form ἴσον appears twice in the same inscription (Del 3 62.149 and 170 = Schwyzer 1987:19–25), compared to a single hίσον at 175, and this in a clause that also contains hομολόγως. Its abnormal aspiration is doubtless analogical (cf. hοκτώ at 34 and hενενήκοντα at 36).  This is only the first of a number of linguistic objections that can be raised against Cassio’s argument.
Moreover, Del Barrio Vega, the foremost authority on the Euboian dialect, believes that word-initial ϝ- was in use in Euboia and its colonies during the seventh and the sixth centuries.  Thumb and Scherer (1959:261 §14) claimed that West Ionic had lost ϝ- at an early date and that the attested Euboian use of ϝ- was limited to foreign names like Ϝᾶχυς and Ϝιο̄́ (on Chalcidic vases)  and to diphthongal or secondary glides like the ones in Ὀϝατίε̄ς (SGDI 5295) and δύϝο (SEG 4.64.28–29). But Del Barrio Vega (1987:215) notes: “[W]hy not think that the -w- was still pronounced in the seventh and sixth centuries, and even in the fifth, just as the inscriptions show?”  As Kretschmer (1894:71) long ago noted in connection with the Chalcidic vases, at the time of their foundation Cumae and Pithekoussai must have had ϝ in current use, since the Etruscans borrowed it from them for /w/, usually transcribed (e.g. lavtni); and, in the south (including Volsinii and Vulci) before the second half of the sixth century, they combined it with -h- in or to represent their labial voiceless fricative /f/ (e.g. vhelmus).  Del Barrio Vega (1987:216) cites two examples from Eretria (the former is disputed): ποιϝ[εσας (IG Ⅻ.9 257, ca. 550) and δυϝε (IG Ⅻ.9 1273/4, 2.2, ca. 525).  To these, add Bartoněk and Buchner 1995:174 no. 36, a single ϝ on an LG Ⅱ amphora (ca. 725–680).  No examples from other Euboian cities or from Oropos survive, but this is not surprising considering that their oldest inscriptions date from the first half of the fifth century.  Thus, we should expect ϝίσα (if not ϝίσϝα), rather than hίσα.
5.1.2 Missing iotas?
Before I deal with the alleged form τίν(ν)υμαι, I must raise two other objections against Cassio’s argument. Faced with the wish to construe tinnuna as an infinitive, he declares that the failure to write the final iota has parallels in Attica (for which he cites Threatte 1980:269) and in the Eretrian inscription IG Ⅻ.9 1273/4, 1.2.  But since Threatte (1980:269) pronounces this a “very rare” phenomenon “doubtless due to careless omission,” it is doubtful whether we can legitimately extend the suspicion to the inscription from Cumae, whose beautiful letter forms indicate careful execution. The same charge of carelessness must be levied against the Eretrian inscription (see below, §5.1.3). The second objection regards his construal of mene as μένει. He writes: “In un primo tempo avevo pensato a μένε imperativo … e dubitavo dell’interpretazione μένε̄ = μένει: sono grato a M. Peters di avermi incoraggiato ad adottare quest’ultima soluzione, che ora mi sembra l’unica giusta” (Cassio 1991–1993:192n25). Lejeune 1972:229–230 §240 does not bear out this license, pace Cassio. Its sole support is in the Berezan lead letter dated to ca. 550–500 BC (for which see SEG 26.845).  But this must be judged indicative of monophthongization that is perhaps characteristic of archaic Milesian, as Dubois (1996:184) suggests, for the writer renders with ε all instances of ει, whether infinitives, primary third sg. endings -ει, diphthongs (true and spurious), the monosyllable εἰ, etc.  This monophthongization was only a tendency, for it coexists with preserved diphthongs—a tendency that was highly localized to the area of Olbia and might well date to the late archaic period (see previous footnote). This cannot be considered proper motivation for an alleged similar treatment 150–200 years earlier in the carefully inscribed Protocorinthian lēkythos. Cassio (1991–1993:192) may be taking a slightly different view when he hints at an error, the imperative μένε for the intended (impersonal) indicative μένει. I do not think this alternative makes the reading any more plausible. Note in contrast the regular ending -ει of hαιρέσει on Nestor’s cup (CEG 1.252–253 no. 454)  or the κλέφσει of the Tataie lēkythos (IG XⅣ 865),  both roughly contemporaneous Euboian inscriptions (the former from Pithekoussai, the latter from Cumae). It is true that Nestor’s cup shows exceptional attention to arrangement and lettering; but the consistency of its letter sizes, shapes, and other dimensions are comparable to those in hisamenetinnuna. Both resemble what Immerwahr (1990:19) calls “band script.” I do not believe that Cassio can successfully maintain that Nestor’s cup displays the closing -ει of hαιρέσει because its writing was “più conservatrice” than the one that presently concerns us, as accords with the impression of great accuracy in its inscribed writing (Cassio 1991–1993:192). Not only because the care and accuracy of their writing are comparable, but also because care is not naturally correlated with ‘conservative.’ What would correspond to carelessness? Innovation?
Cassio (1991–1993:194–200) argues that τίν(ν)υμαι, not τείνυμαι, must have been the West Ionic form current in the eighth and the seventh centuries. His principal piece of evidence is the law of Eretria IG Ⅻ.9 1273/4 cited above.  But as I noted, the name Tείσων and the subjunctive τείσε⟨ι⟩ in the law (1.3–4) suggest otherwise. The e-grade -νυμι verbs (like δείκνυμι) were a Greek innovation. Indo-European used a zero-grade instead.  The synchronic paradigmatic relation between τίνω and τεῖσαι reflected a diachronic fact.  The long iota of Homeric τῑ́νυμαι can only be explained in one of four ways: it may reflect τείνυμαι, a Greek innovative e-grade -νυ- verb built on the original IE e-grade sigma aorist;  the length may be analogical, after the East Ionic τῑ́νω;  its form may have been τῐ́ννυμαι, corrected to τῑ́νυμαι upon simplification of -νν-;  or it may be a metrically lengthened *τῐ́νυμαι. The first and second alternatives are possible and acceptable; West (1998a:xxxv) rightly disposes of the fourth. The third is unpersuasive, unless one accepts the notion I am questioning here: that the paradosis was primarily in written form. If the text was shaped by an oral paradosis, such a simplification could not happen by scribal error. An oral mechanism (like sound change or analogy) that plausibly accounts for the data must then be offered.
What, then, about the Eretrian law? The first thing to note is that the upsilon in τίν[υ]σθα(ι) has been supplied. Editors have near unanimously agreed on the supplement.  Cassio (1991–1993:196n35) makes much of this, but one cannot dismiss Attic τίνεσθαι out of hand, especially in the face of forms that hint at a measure of Attic dialectal influence.  If Cairns (1984:152–153) is right, for lexical and syntactic reasons one cannot accept φ]υγία in line 1.3. Instead, he supplies the Attic neut. pl. ὑγιᾶ. An objection to this supplement on the grounds of its Attic form is lessened by the Attic shape of the otherwise unattested φυγία and by hẸṚẠI, the neatly chiseled out word that follows τείσει and has been read as the Attic dative of Hera.  These arguably Atticizing forms suggest that τίνεσθαι must be seriously considered.  If so, Cassio is left without his main support for an ancient τῐ́ν(ν)υμαι. To those who accept the upsilon I would like to make an alternative proposal: that the mason chiseled τίνυσθα in error, intending τείνυσθαι instead. My proposal is conjectural, of course, but no more than the ultimately unprovable existence of the upsilon. The four lines that concern us are all by one hand and make an independent section. Its text is marred by various errors: the iota that should end τίν[ ]σθα is missing; instead of TEIΣEI, the mason inscribed TEIΣET;  and if Cairns is correct that hẸṚẠI is ἧραι for ἆραι, the η for the correct Ionic ᾱ might yet be another error, and so would be the lack of interpunction after ὑγιᾶ. It is apparent that the letter cutter was not a careful worker, a fact that lends some credibility to my conjecture that τι- was supposed to be τει-.
There are only two other alleged examples of τίνυμαι, both from Crete. The first, [τ]ινυμε̣[νο, is of greater significance because of its sixth-century date.  But it may only be a mirage: “[L]a forme, souvent citée, [τ]ινυμε̣[νο n’est, à vrai dire, pas identifiable … . Après ατελειας, le découpage des mots n’est pas satisfaisant sémantiquement. Il est préférable de rester circonspect sur la forme en question.”  The other is a late fourth-century ἀποτινυ[.  Whether the -ι- is long or short cannot be determined. Cassio (1991–1993:194n31) looks for support in the Cretan προδίκνυτι,  which scans ⏑ ⏑ – ⏑.  But as Schwyzer (GG I.696n8) remarks, even though it is often regarded as the old zero-grade form of δείκνυμι, various aspects of this second-century BC epigram undermine the notion that it preserves a hoary archaism. Of particular interest is the abbreviation (twice) of the long ᾱ in νᾱός (νᾰόν 10, νᾰῶ 11), which must be metri gratia.  Since προδεικ- could not be similarly shortened, perhaps the author decided to resort to an artificial zero-grade instead.  There is not much, then, to be said for the existence of a zero-grade τῐ́νυμαι other than the alleged case of the Cumae inscription.
If Cassio were correct to suppose the Homeric τῑ́νυμαι comes from τῐ́ννυμαι, it is hard to understand why the latter is only represented in some manuscripts. We would have expected it to predominate, since it was in agreement with later Hellenistic use and hence not an obvious target of late scribal tampering.  That there are no instances of τείνυμαι, however, need not arouse suspicion, for the analogical pressure of τῑ́νω together with the Hellenistic exchange of -ῑ- for -ει- and predilection for -ννυμι sufficiently accounts for its absence. In the end, therefore, as West (1998a:xxxvi) observes, the most plausible interpretation of the Homeric data is to assume a form τείνυμαι that was expelled from the manuscripts in the Hellenistic period,  when τιννυ- was common.
Ultimately unproven and, in my view, unconvincing, is Cassio’s theory that -ννυμαι in his alleged τῐ́ννυμαι was analogical to σβέννυμι and ἕννυμι,  even if we believe that he succeeds in showing that these two verbs are authentic Ionic forms. And one cannot forget that the active form τίνυμι is not attested before the end of the fourth century BC, if in fact the Cretan ἀποτινύ[ in IC 2, Tituli locorum incertorum no. 1.8, is correctly supplemented as ἀποτινύ[μεν.  Watkins (1995a:44–45) seeks to improve upon Cassio by assuming a long iota in hῖσα, which is at odds with the absence of CL in Euboian but more easily accounts for the spiritus asper; and by suggesting that the inscriber was quoting known poetry but (unintentionally?) distorted his model with the geminate tinnuna. In Watkins’s reconstruction, the line would have been a perfect pentameter hemistich. But even discounting the arguable opaqueness of the sentence and the implausibility of the vase’s alleged function (on which see next), if we are willing to jettison the Euboian character of the inscription, to assume an unknown original that has been unaccountably distorted, and to suppose a literary context that helps to clarify its cryptic conciseness, we have largely lost interpretive control and almost any attractive case can be made at will.
5.1.4 The alleged meaning
To turn away from linguistic matters and address the message: I believe that the meaning Cassio (1991–1993:191–192) gives to his reading is implausible. I have no objection to his brief comments regarding the ideology of strict retribution (and the use of τίνειν with ἶσα, ὁμοῖα, vel sim.).  But to succeed, Cassio needs to convince us that the following facts are credible: i) the owner of the lēkythos thought it necessary or advisable to add on the bottom of the vase a gnomic statement on the certainty of repayment for wrong committed whose target was a hypothetical tomb raider; ⅱ) he did so because he anticipated that the potential tomb raider would look at the bottom of the lēkythos, read the inscription, and apply it to himself; ⅲ) his expectation was that somehow this imagined act of reading would protect the tomb or its contents, even though the tomb raider (perforce, if incredibly, literate) could only read the sentence after he had broken into the tomb;  ⅳ) the meaning suggested, “it is fated (lit. ‘it remains’) to repay the same,”  follows intelligibly from the inscription as it stands.
I need not say much about i–ⅲ to underline their implausibility. One thing is to inscribe prominently a curse on Tataie’s lēkythos ‘he who steals me will go blind’. The statement conspicuously spirals around the body of the vase.  Drawing attention to itself, it might actually dissuade a potential thief. Its meaning is also clear: the vase owner is named (‘I am Tataie’s lēkythos’); the threat is issued (‘he will go blind’); the condition for its fulfillment, spelled out (‘whoever steals me’).  Compare this with an inscription that would only be read if, of all the objects in the tomb, the intruder should focus on a small vase (the lēkythos); he should then pick it up and, having discovered the writing on the bottom, try to understand it.  But, even if he were to do so, it would be too late: he would have broken into the tomb already. Where then is the deterrence? The controversy regarding CEG 1.256 no. 459,  whether it is a funerary monument or not, is largely irrelevant.  Cassio wishes to prove the existence of funerary curses before 400 BC in order to establish a parallel, even if a hundred years later, perhaps because he feels that his argument labors under the weight of excessive exceptionality. To me, of greater significance is that the meaning of what he alleges as a curse is clear: the threat is stated (‘may Zeus utterly destroy him’); the condition for its fulfillment is spelled out (‘whoever harms this’). All the examples adduced are of similarly lucid logic: if A, then B; or whoever A, he/to him B.
In our inscription, this structure is missing. Unless we assume, with Watkins (see above) a missing literary context that would supplement it with otherwise missing material, according to Cassio we only have the ‘B’ part: ‘it remains to repay the same’. Not only is the ‘A’ part missing that would specify the identity of the target (e.g. ‘whoever robs this tomb’); the ‘B’ part is impersonal! Not even ‘there remains for you to repay the same,’ resorting for concision to an indefinite ‘you.’ The indeterminate nature of the hisa exacerbates the opaqueness. Laboring already under the weight of mystifying compression, why not state at least the crime? ‘It remains [for you] to repay the same for your trespass [into this tomb]’; or ‘it is fated [for you] to repay the same for your robbery.’ Not so. Add to the lack of a specific target (‘you,’ ‘whoever robs,’ vel sim.) the lack of a specific crime and a specific punishment. The passage from Aiskhylos’ Suppliants, which Cassio (1991–1993:191) calls upon to support his argument, is actually detrimental to it. For in lines 434–436 the chorus tells Pelasgos: ‘For know [well]: whichever [course] you bring to pass is left to your children and your household, to repay […] a like justice (penalty?) [for it]’.  Regardless of how one construes the grammar, the target of the implicit threat is clear (‘to your children and household’) and the content of the threat, well defined (‘to repay whichever thing you bring to pass’). Even if μένει is taken impersonally, the statement does not suffer from the opacity of Cassio’s reading. Aiskhylos Agamemnōn 1563–1564 leads to the same result. With Zeus on the throne standing for the permanence of his cosmic order, we read: μίμνει δὲ μίμνοντος ἐν θρόνῳ Διὸς | παθεῖν τὸν ἔρξαντα· θέσμιον γάρ. Here the infinitive clause (‘that the doer [must] suffer’) is the subject of μίμνει. The parallel between Zeus remaining on the throne and the principle of justice remaining (as law) on the throne (‘for it is settled’, θέσμιον γάρ) makes the construction with μίμνειν readily intelligible. Not so in the case of the lēkythos: ‘to repay the same remains’. Can the inconspicuous writing on the bottom of a lēkythos deposited in a tomb be plausibly credited with such an oracular pronouncement?
5.2 Nestor’s Cup (CEG no. 454)
The purpose of this section is to probe further the notion of a definitive Euboian textual phase during the geometric and early archaic periods. In an excellent article,  Cassio seeks to answer the most urgent question that the famous Ischian cup poses for the Homerist: what is the relationship between the inscribed verses and the Homeric epic tradition in circulation towards the late eighth century BC? For ease of reference, I print here the text of the inscription after CEG 1.252–253 no. 454, with the substitution of ε̣[μ]ι̣ for Hansen’s ε̣[ἰμ]ι̣ (on which see footnote 84 below): 
Nέστορός : ε̣[μ]ι̣ : εὔποτ[ον] : ποτέριον. |
hὸς δ’ ἂν το̄δε πίεσι : ποτερί[ο] : αὐτίκα κε̄νον |
hίμερος hαιρέσει : καλλιστε̣[φά]ν̣ο : Ἀφροδίτες
hὸς δ’ ἂν το̄δε πίεσι : ποτερί[ο] : αὐτίκα κε̄νον |
hίμερος hαιρέσει : καλλιστε̣[φά]ν̣ο : Ἀφροδίτες
Naming a ‘Nestor’ in close connection with a cup invites us to think of Λ 632–637, the celebrated description of Gerenian Nestor’s δέπας περικαλλές. But did it elicit similar associations in the mind of the late eighth-century symposiast who drank from it? S. West (1994b:14) can help us to set the stage of this review. She notes that “ποτήριον and καλλιστέφανος are not to be found in the Iliad and Odyssey; the convenient-looking phrases αὐτίκα κεῖνον and ἵμερος αἱρήσει do not figure in the Homeric stock formulae.”  She therefore assigns their pedigree to an otherwise unattested Pylian saga, a poetic tradition less sophisticated than the Iliadic that celebrated the exploits of Nestor’s youth. Whether the language on the skyphos can be shown to reflect the attested Homeric poems is of great potential significance. If on balance it points elsewhere, this fact would make the alleged Euboian phase of Homeric poetry even less plausible than it already appears. For is it likely that, if something like our Iliad was in circulation in late eighth-century Euboia—that, if in fact the performance of this poetic tradition on Euboia was flourishing so vibrantly as to place a definitive and normative Euboian stamp on its language—the earliest attested inscription that arguably refers to a well-known Iliadic hero would fail to echo the formulaic language of the Iliad? Even if we conclude that the Nestor on the cup cannot refer to Gerenian Nestor, if on balance the diction of the inscription is shown to be independent from the Homeric tradition as we know it, we must also conclude that this tradition did not serve as the model for the cup, or else that it did under a shape rather different from the one ultimately preserved in writing. But, if it did not serve as its model, unless we posit an unlikely symposiast of unique ability who composed the verses de novo, with S. West we must assume the existence of another, similarly vigorous poetic tradition that incorporated a figure by the name Nestor and inspired the poet of the skyphos. The alternatives are stark for the proponents of the Euboian phase. If the Homeric tradition was not known in Pithekoussai or was known in a substantially different form from the one preserved by the vulgate, this would render the alleged Euboian phase untenable. In particular, one could not argue that dictated scripts of Euboian performances might be responsible for the written texts of the Homeric poems. But if the Homeric tradition was known substantially as we know it and simply failed to be the model of the skyphos poet, we would have to accept the coexistence of two vigorous, formulaically disjoint hexametric oral traditions that shared a named figure and, incredibly, a possession peculiar to this figure; and we would have to embrace the corollary of the near complete disappearance from the record of the tradition represented by the skyphos. If in fact the verses on Nestor’s cup point away from the Homeric diction in our texts, I submit that it is far more likely that a ‘Homeric’ tradition was known in Pithekoussai, but one in diction and themes different from our own, and yet, as the cup attests, one that included elements (the hero and his cup) still familiar to us from the vulgate.
To advance my own conclusions: Nestor’s cup stands in a long line of ‘I am’ inscriptions that are implicitly performative.  They betoken a culture whose ‘literature’ lived in public and semi-public performance, not in written texts disseminated within circles of readers. On balance, the diction of the cup proves independent from the language of our Homeric poems. Because I do not think composition de novo plausible, I believe that the traditions of performance underlying the verses prove that Euboian poetry susceptible of inclusion under the label ‘Homeric’ was substantially different from the one eventually written down and preserved in our vulgate texts. It is ultimately not of much consequence whether we posit the currency of a Pylian strand of which only echoes survive in the grand Panhellenic tradition. The label ‘Homeric’ applied to late eighth-century epic traditions circulating in Euboia should be capacious enough to draw into its sphere related stories only obliquely reflected by the final canonical form of the poems. Whether we can, or should, use that label at all becomes largely a question of semantics. What matters is the understanding that we are dealing with a formative period, during which the formulaic diction and the narrative repertory must have been in significant flux. Were we today privy to its contents, we might well recognize some elements and be surprised by others. If something like Λ 632–637 stands behind the verses on the skyphos, we must read them ironically. The cup is a humble vase, far from the literary description in type and worth. My guess is that the owner was himself named Nestor. This gave place to the irony and accounts for the resumption of το̄δε: ‘I [too] am Nestor’s cup—[but] one that is good to drink from. Whoever drinks from this cup, straightway desire instilled by beautifully garlanded Aphrodite will seize him’. The reason for the το̄δε is the implicit contrast established by the irony of the first line. It is as if the cup were boasting, ‘I too have Nestor as owner!’ Adding, ‘I am not the massive four-handled cup, but one that is good to drink from. And whoever drinks from this cup will not be refreshed for war, but inflamed with desire.’ Had the contrast been drawn explicitly, e.g. with ‘too’ (‘I too am Nestor’s cup’), a personal pronoun (μου or ἐμοῦ) would have sufficed. With the contrast ironically implicit, the deictic το̄δε, which (as many have recognized) serves to mark a contrast, becomes necessary. This motivates what is otherwise an exceptional switch from first to third person.  If my reading is right, the switch is only apparent, for the cup continues to speak; only, it refers to itself by an act of verbal pointing. 
What follows here is a summary of the findings in Cassio 1994. He argues the case well, although he reaches what I think is the wrong conclusion: that there is no proof that Nestor’s cup respects an epic tradition other than the Homeric (65). I could embrace this statement if, by Homeric, Cassio meant a very fluid ‘Homeric’ tradition that could be thought of as rather different from the ultimately received. But item (b) of his conclusion points in another direction. There he notes that the adjective καλλιστέφανος bespeaks the end, not the beginning, of the epic tradition (“the world of the Odyssey and the Hymns,” 65). This suggests a fixed text of poems that are by and large recognizably the same as our own. Item (c), on the other hand, points away from his first one: the verse-end use of κεῖνος is not Homeric. Cassio realizes that items (a) and (c) are at odds: “Se ci troviamo alla fine della tradizione epica l’autore degli esametri della Coppa deve avere già conosciuto ampiamente un Omero in cui ἐκεῖνος si trovava in fine di esametro ed in enjambment” (65). But there is no evidence, only the scholar’s wish, for such a ‘Homer.’ On the contrary, as my review of Cassio’s article will presently show, the inscribed hexameters diverge markedly from the attested Homeric idiom in ways more than one. What, then, is to be made of the assertion that the skyphos respects no poetic tradition other than the Homeric? The implications of Cassio’s concluding item (c), hardly inconsequential for a proper estimation of the late eighth-century shape of the ‘Homeric’ epic tradition generally, are even starker for the theory of a definite geometric and early-archaic Euboian textual phase: “Se la redazione finale o dei poemi omerici, o della sola Odissea, si deve ad aedi euboici, perché sulla Coppa non troviamo ἐκε̄νον? Inoltre, come si è visto, l’uso di δ’ ἂν nella Coppa … si allontana in maniera vistosa dall’uso omerico e in particolare de quello dell’Odissea” (65).
5.2.1 δ’ ἄν versus δέ κε
Cassio does not focus on the non-Homeric εὔποτον or ποτήριον.  Instead, he addresses himself to three points that are better able to establish dependence on, or divergence from, the attested Homeric diction: the use of ὃς δ’ ἄν; the verse-final κε̄νον in enjambment; and the noun-epithet καλλιστεφάνο̄ Ἀφροδίτε̄ς. Let us take them up in turn. Since it is followed by a word that starts with a consonant, ὃς δ’ ἄν scans – –. As Cassio notes, relative clauses that start – δ’ ἂν V – (– ⏑ ⏑) can be replaced without consequence by – δέ κ’ V – (– ⏑ ⏑); those that start – δ’ ἂν C – (– –) can be replaced by – δέ κε C– (– ⏑ ⏑). In principle, proponents of the evolutionary model have no problem accepting any one Homeric instance of δ’ ἄν as authentic. But Cassio is right that, as a rhetorical matter, to convince the skeptic that ὃς δ’ ἄν on the skyphos conforms to late eighth-century Homeric poems substantially like our own, one would have to find in our vulgate parallel instances of this construction that cannot be trivially changed to ὃς δέ κ(εν). Otherwise, a skeptic could always question the antiquity of the adduced parallel and doubt its relevance. An examination of the poems does not support dependence. Constructions resembling the syntax of the inscription are instanced only four times: ὃν δ’ ἄν (Θ 10 Ο 348), ὅσσοι δ’ ἂν (Τ 230), and ὃς δ’ ἂν (τ 332). All except Τ 230 can be trivially rewritten with κε(ν).  We simply do not find the equivalent of ὃς δ’ ἂν C-. This contrasts with eleven instances of ὃς δέ κε C- or its equivalent.  Sharper still is the numerical contrast between δέ κε C- and δ’ ἄν C-, irrespective of syntactic construction, wherever δέ κε and δ’ ἄν stand in arsis. There are eight instances of the latter if we allow for οὐδ’ ἄν, and three if we do not.  The Odyssey presents a single instance (of οὐδ’ ἄν) against twenty-four of δέ κε. As Cassio observes, the Homeric text often offers δέ κε C- where there would be no prosodic impediment to the Ionic δ’ ἂν C-. The cup’s inscription, however, follows the Euboian dialect and conspicuously diverges from Homeric usage. This severely undercuts the theory that Euboia was the setting in which the Homeric texts attained their definitive and normative shape, and renders extremely implausible the view that the Odyssey “might well be a Euboean poem.” 
5.2.2 κεῖνος versus ἐκεῖνος
An even greater obstacle to this theory is posed by the use of verse-end κεῖνον in enjambment. Homeric usage is clear: ἐκεῖνος appears at the end of the verse; κεῖνος, elsewhere.  In the Iliad, verse-end ἐκεῖνος occurs at Ι 63 646 Λ 653 Σ 188; of these, the first and third are irreducible to κεῖνος. The irreducible cases in the Odyssey are twelve.  Reciprocally, all κεῖνος at verse-end are reducible to ἐκεῖνος except ἤματι κείνῳ (Β 37 482 Δ 543 Σ 324 Φ 517), οὐδέ τι κείνῃ (ν 111), and εἵματι κείνου (ξ 501).  Because irreducible ἐκεῖνος is often found in enjambment (twice in the Iliad and five times in the Odyssey), Cassio (1994:59–60) speculates that its preferential adoption for verse-end position and the use of enjambment were related phenomena that entered Homeric diction at a relatively recent date. Hence, if Nestor’s cup reflected Homeric diction, we would expect αὐτίκ’ ἐκεῖνον.  As Cassio (1994:60) remarks, “[il redattore del testo della Coppa] ha usato invece κε̃νον, in fine di esametro ed in enjambment, allontanandosi quindi in maniera netta dall’usus omerico.” This distance is all the more conspicuous if, as Peters (1998:596) thinks, the West Ionic deictic was ἐκεῖνος and not κεῖνος.  Peters tries to turn the cup’s use of κε̃νον, arguably a high hurdle for the Euboian phase, to this theory’s advantage. Because he believes in a monumental poet who hailed from Oropos,  he thinks that instances of ἐκεῖνος in the poems reflect Homer’s own dialect and are not traditional (not formulaic).  From this premise, he infers that the cup’s verse-end κεῖνος is precisely what we should expect if the inscriber was aiming at the high register of traditional language, and that he is therefore following ur-Homeric usage—i.e. his poetry is dependent on proto-Homeric diction. Hence, the skyphos would not militate against the alleged Euboian phase but show engagement with the proto-Homeric tradition at the place and time Peters imagines the monumental poet composing his definitive version of the tradition. Now, aside from the doubtful contention that ἐκεῖνος is the proper Euboic form, this argument depends on the unprovable claim of an ‘archaizing’ intent that assays for the high register of what passed for traditional in late eighth-century Euboia. It also makes the author of the cup’s verses at best a parallel conduit of the epic tradition that flowed through Homer into the Iliad and the Odyssey. If defamiliarizing for the sake of literary effect was the aim of the inscriber, why use the (for Peters) lower-register δ’ ἄν? Should we not expect δέ κε?  Why not, too, the more traditional ἐυστέφανος? Against Peters’s clever argument one can also lay the charge that it is ultimately unfalsifiable: the cup does not follow Homer’s dialect, but it is alleged to follow pre-Homeric diction; hence it is not un-Homeric but simply non-Homeric! I would find this reasoning more persuasive if the poems exhibited a greater number of reducible and irreducible verse-end κεῖνος. Is it plausible that Homer’s personal (and ‘untraditional’) dialect could have displaced in all but one case (υ 265) every reducible instance of verse-end κεῖνος?  Moreover, if formulas predating Homer instanced κεῖνος both at verse-end  and elsewhere, why should Homer have shown such a marked preference for injecting his personal dialect at verse-end?  Could he not just as well have innovated with ἐκεῖνος in other sedes? In sum, Peters is open to the same criticism Cassio (1994:62n48) directs against Dihle for claiming that, although καλλιστέφανος is absent from Homer, it is attested “in der archaischen und archaisierenden, an das Epos anknüpfenden Dichtung”: “Adottando questo criterio qualsiasi innovazione degli Inni rispetto a Omero può ricevere una patente di grande antichità, e possiamo senza problemi costruirci un altepischer Sprachschatz di nostro gradimento.”
5.2.3 καλλιστεφανο Aφροδιτες
Cassio (1994:60–64) also probes the expression καλλιστεφανο Aφροδιτες from the point of view of Homeric formulaic diction. From his analysis he concludes that it is a development that presupposes the existence of ἐυστέφανος Kυθέρεια (63). Hence, he believes that it points to the chronological endpoint of the development of epic diction which dates the Iliad to a time before the inscribing of the cup: “A mio parere quindi … è del tutto improbabile ipotizzare per καλλιστεφάνο̄ Ἀφροδίτες un antecedente che non sia quello omerico” (63). I cannot embrace his conclusion, which is reached through a chain of doubtful links. Cassio is probably right that -στεφαν- in the sense of ‘crown’ presupposes the orientalizing period, when the use of the crown in real life spread and became established (62). Perhaps it is also true that the application of ἐυστέφανος to goddesses hints at a growing familiarity with statues adorned with crowns or poloi (63). The alternation ἐυ- and καλλι- is in evidence in our poems.  But this does not establish that καλλιστεφανο Aφροδιτες is derived from, and therefore presupposes, the familiar diction of the Iliad and the Odyssey. The question to answer is whether καλλιστέφανος or ἐυστέφανος is older, if in fact one predates the other and this fact can be ascertained; and, if a priority in time is plausibly established, what rationale suggests that the younger form was actually derived from the older. The poems do not exhibit these adjectives in a suppletive paradigm of formulaic economy, though well they would have, had either of them been long enough in use as an epithet, given their complementary metrical shapes (– – ⏑ ⏑ –and ⏑ – ⏑ ⏑ – respectively). ἐυστέφανος is used once of Artemis (Φ 511), once of Thebes (Τ 99), once of the heroine Mykene (β 120, with v.l. ἐυπλόκαμος), and three times of Aphrodite (θ 267 288 σ 193). When this goddess is in view, the diction is ἐϋστεφάνου τ’ Ἀφροδίτης (θ 267), ἐϋστεφάνου Kυθερείης (θ 288), and ἐϋστέφανος Kυθέρεια (σ 193). The latter two are the only cases of Kυθέρεια in the Homeric poems. If we include the hymns and Hesiod, there are nine other instances of it (five in the hymns and four in the Theogony). Four of the five epithets of Kυθέρεια in the hymns are ἐυστέφανος (twice with the v.l. ἰοστέφανος).  In the Theogony it is accompanied by an epithet twice, and both times it is ἐυστέφανος. The outcome of this survey is that Kυθέρεια takes the epithet ἐυστέφανος in eight of its nine surviving instances. Whereas this appellation is rare in the poems (it appears only twice), Ἀφροδίτη is used forty-two times,  and its epithets, other than the sole occurrence of ἐυστέφανος, are δῖα, Διὸς θυγάτηρ/κούρη, φιλομμειδής, and χρυσέη/χρυσείη. All are used more than once except for the single instance of κούρη (Υ 105), an uneconomical spondaic equivalent of the more common θυγάτηρ. Hence, ἐυστέφανος as a divine epithet with the meaning ‘well-crowned’ (vel sim.) cannot have entered the diction of the poems long before the time when the Ischian skyphos was inscribed. In the case of Aphrodite, it was preferentially paired with the appellation Kυθέρεια.  And καλλιστέφανος appears only twice in archaic epic, namely, in the Hymn to Demeter. It is true that in Nestor’s cup ἐυστεφάνο̄ could not have been used instead of καλλιστέφανο̄. One might therefore argue that the latter represents the necessary formulaic modification of the former. But if the practice attested by our Homeric poems had served as a model, we would never have expected that a verse ending with Ἀφροδίτ- | should call for an epithet with the metrical shape – – ⏑ ⏑ –: all her epithets in the Iliad and the Odyssey are of the shapes ⏑ – ⏑ ⏑ – or ⏑ – – – (or subsets thereof like – – and –);  none are shaped – – ⏑ ⏑ –. Neither is the hiatus -στεφανο Aφρο- necessary,  and Homeric usage would have commended the alternative Kυθερείης, which is also a perfectly acceptable Euboic form.  To meet with these two departures from the Homeric versification attested by the extant poems at the very time when the corresponding formulaic canons were allegedly being established undermines Cassio’s conclusion that the cup is in fact dependent on the predating diction of the poems. If we had found at least one or two occurrences in the Homeric poems of καλλιστέφανός τ’ Ἀφροδίτη (vel sim.), it might be reasonable to conclude i) that it was a formulaic innovation developed to supplement the metrical shape of the more common epithets; and ⅱ) that it postdated, and probably presupposed, the more numerous ἐυστέφανος. But to infer this from the Ischian cup one must assume what Cassio set out to prove: that in fact the diction of the skyphos derives from the Homeric poems. As the case stands, the only safe verdict is that the cup decisively departs from Homeric usage, and this judgment deals another blow to the theory of a Euboian definitive phase.  And even if we should grant arguendo that the Iliad and the Odyssey, in versions substantially like our own, were in circulation in Pithekoussai and that their language inspired the verses on the vase, we are still left with an impossible obstacle to the Euboian theory. For, if καλλιστέφανος and ἐκεῖνος belong to the end stage of the epic tradition, and Euboia is the alleged home of the last and definitive formative stage of the Homeric poems, why the cup’s conspicuous departures from Homeric diction? Reciprocally, why do the poems largely or absolutely fail to instance the formulaic practice allegedly current at the time when and place where their text was definitively shaped? No wonder Cassio (1994:66) concludes: “Non mi sento di offrire una soluzione netta di queste aporie; ma c’è da chiedersi se la tesi euboica, per quanto affascinante, non debba essere in tutto o in parte riconsiderata.”
One might well be puzzled by Cassio’s conviction that despite the many departures he acknowledges in his paper the verses of Nestor’s cup must presuppose Homeric epic—a presupposition that for him does not involve a fluid tradition of multiforms, but largely fixed texts near identical with our own.  Although he owns that a priori it is “anything but unlikely” that epic traditions other than the Homeric might have existed,  he avers that in the case of Nestor’s cup such proof is not forthcoming. All departures can be explained by the influence of the local dialect. I do not dispute that what Hackstein (2010:421) calls “the practice of translating hexametric epic poetry from one Greek dialect into another” did exist, and that it may legitimately explain language that plausibly postdates the textual fixation of the poems and for which there is good reason to suspect dependence upon them.  But this cannot be safely assumed for Nestor’s cup, which Cassio regards as only one or two generations younger than the alleged composition of the Iliad by Homer.  What under different circumstances might have been read as signs of dialectal ‘translation’ must, in this case, be interpreted as evidence of independent epic diction. Ultimately, we cannot assert that a well defined, independent epic tradition lies behind Nestor’s cup. But the independent diction does prevent us from embracing the suggestion that the Homeric poems achieved their normative and definitive form in Euboia during the first few hundred years of the first millennium BC. 
5.2.4 An eighth-century bookhand?
In closing, I now turn to the execution of the inscription. The careful letter shapes, the arrangement in three lines (two of them hexameters), and the use of interpunction for the cup’s inscription have fostered the fanciful view, startlingly well received, that this inscription proves the existence of an eighth-century book script. The best known exponent of this theory is Immerwahr: “Jeffery remarks that elsewhere in inscriptions poetry is never written in individual lines; but surely in manuscripts such separation must have been customary, especially in epic texts. I would suggest that the graffiti on this vase are influenced by eighth-century book script. … [W]e must recognize the existence of a more regular script for literary and documentary purposes from the later eighth century on” (1990:19). His judgment was embraced by Cassio 1999: “[È] stato sostenuto già da Heubeck … e più di recente da H. R. Immerwahr, W. Burkert e M. L. West, che il tipo di scrittura e l’intera mise en page del testo graffito … sarebbero influenzati da scrittura libraria dell’Ⅷ secolo a.C.” (67).  This has now been raised to the next level by Bartoněk and Buchner (1995:183), who propose the existence of a Pithekoussan scribal school. To put these views in perspective it is important to mark that, as noticed above, the writing of the Cumae lēkythos is of comparable regularity and shows similar care. ‘Calligraphy’ (if the anachronism is at all adequate) is surely characteristic of the writer and varies greatly from vase to vase and location to location. It is fair to say that an inscription’s careful writing proves that it is not the inscriber’s first attempt at writing. It is quite another to claim that only the inscriber’s familiarity with a different, pre-existing writing practice (in this case, the writing of poetry books on perishable material) can explain the traits (regularity of script, punctuation, etc.) that make this inscription exceptional. Nothing prevents the view that Pithekoussan inscriptions on the whole may even display greater regularity of arrangement and writing than inscriptions from other locations. It is well conceivable that a skilled and careful artisan may set trends and standards in his local community. If scholars can detect artisans and workshops elsewhere on the basis of their workmanship, why should this not be the case in Pithekoussai? 
None of this, however, requires a belief in an eighth-century book market unless one starts with the prejudice that a vase inscription must be synchronically derivative of other kinds of writing current in its local culture. This prejudice of synchronic secondary status must be distinguished from the hypothesis, diachronically sound, that, insofar as the Greeks may have derived alphabetic writing from Semitic sources (Phoenician or Aramaic or both), it is reasonable to look for Semitic antecedents for the use of punctuation in Nestor’s cup. It makes particular sense to posit such lines of influence at a trading post like Pithekoussai,  where a Semitic minority lived side by side with the Greeks. Ridgway (1992:111–118) reviews the evidence for oriental residents at Pithekoussai. An imported amphora with several markings used in an enkhytrismos burial is of special importance.  Two of these markings can be read as the Aramaic for ‘200’ and ‘double’.  Since the capacity of the amphora is two-hundred times the average capacity of Attic-Ionic kotylai, Ridgway speculates that the vessel contained unguent to be repackaged in Pithekoussai and that it was dispatched with an Aramaic description of its contents in the expectation that the recipient would understand it. He further suggests that Phoenician Ialysos on Rhodes (where the Nestor skyphos was made) was the source whence it was dispatched before it started to be bottled in aryballoi at this location for distribution (in the LG Ⅱ period). The overlap of Aramaic and Phoenician elements is justified by the triangular west Phoenician and Punic religious symbol that points to death and afterlife.  This sign relates to the funerary use of the amphora and it suggests that a member of the family (Ridgway proposes the father) was of Levantine origin and observed this non-Greek usage at the interment of his child. Further evidence of Levantine residents in Pithekoussai comes from the Lyre-Player Group seals, scarabs, and Levantine aryballoi (probably imported from Rhodes).  The number of tombs that contain at least one of these items is about one third of the excavated cemetery population of 750–725 BC. This proportion demonstrates the degree to which the settlement was permeated by Levantine influences.
In contrast to Pithekoussai, ordinarily one cannot assume that Euboian metropoleis had sufficient exposure to foreign Semitic cultural practices—e.g. the alleged recording of poetry on perishable materials—to adopt them in turn. Direct contact with the Levant cannot be assumed and must be demonstrated. Lefkandi’s archaeological record indeed proves that it had commercial ties with Egypt and the Levant towards the end of the tenth century.  Popham (1994:28–30) argues for direct communication between Euboia and the Syro-Palestinian coast, without the intermediacy of Cyprus as a redistribution center, and for Euboian ships as the vehicles of this commercial exchange. But Sherratt (2003:229–230) has vigorously countered on the basis of the chronological and spatial distribution of Greek early Iron-Age pottery in the east Mediterranean that trade was “almost entirely in the hands of eastern (especially Tyrian) carriers,” and that “there is no evidence that the inhabitants of Euboea, or for that matter other Greek-speaking inhabitants of the Aegean, frequently if ever found themselves in the position between the eleventh and eighth centuries of seeing full-blown Cypriot literacy in operation in its own context. Since in this period at least it does not seem to have travelled outside the island, it seems very likely that they did not even know of its existence” (230).  Clearly, this judgment would apply a fortiori to Aramaic and Phoenician literacy. Moreover, in and of itself trade is not a robust enough context for the cultural exchange that the adoption of foreign writing practices presupposes. Only trade in written artifacts (inscriptions, manuscripts, etc.) could conceivably have brought to the attention of the aristocrats in the Euboian metropoleis the technologies of written literacy that, according to some, the Greeks allegedly adopted from the more advanced Semitic cultures in the eighth century BC. Where Greeks and Arameans (or Phoenicians) lived side by side, however, the stimulus for the adoption of foreign customs can be credibly attributed to the stable and culturally embedded context in which the various cultural practices intersected. The excavations at the sanctuary of Apollo Daphnephoros in Eretria have only turned up one sherd from ca. 800–750 BC inscribed with Semitic letters. Pfyffer et al. (2005:76–77) think that the author might have been a Luwian merchant passing through Eretria, a plausible conjecture given the city’s relationship with northern Syria at the time.  But mark the contrast in distribution and number between the eighth-century inscriptions found on Pithekoussai and those from Eretria: “[L]e geste d’inscrire, et plus encore celui d’inscrire des lettres, est encore rare dans le sanctuaire au 8e siècle. Il est plus rarement attesté encore ailleurs à Erétrie.”  It seems that the use of writing was generally limited to the sanctuary, although only one inscription (Pfyffer et al. 2005:61 no. 5) can be classified as ‘religious.’
To return to Pithekoussai and Nestor’s cup: the exceptional cremation of a minor and the goods deposited in the tomb where the vase was found (a few silver items, unusual in children’s graves, and several oriental arryballoi) led Ridgway (1992:116) to make the fascinating proposal that the skyphos belonged to a family that was partly non-Greek, probably Levantine. If true, this further commends attributing the inscription’s interpunction to Semitic practice, not to the fantasy of an eighth-century book hand. Heubeck (1979:115) is likely right that the punctuation must have been intended to aid the oral performance of the verses.  This establishes the unsurprising claim that from time immemorial those acquainted with epic poetry apprehended the individual line of verse as a unit of performance. Their intuition of phrasing extended to sublinear units like metrical cola. Marking these subdivisions visually (whether they were uttered as pauses or reflected by the intonation) cannot be assumed to reflect ordinary practice, as the neglect of word division during the historical period proves.  It is possible that the non-Greek owners of the cup needed aids to performance that natives could dispense with. Be that as it may, if the cup’s use of interpunction to separate words and clauses really derived from practices observed in the writing of poetry books in circulation at the time, it is inexplicable why later attested papyri fail to follow them. Much more plausible is the provision of aids to oral performance that might help non-native speakers of Greek to read aloud the verses—aids, moreover, familiar to them from their original cultural milieu.  Indeed, as Lipiński (2001:97 §9.11) notes, the use of punctuation by alphabetic scripts (vertical strokes and interpunction) is attested in the Aramaic Tell Fekherye inscription of the ninth century BC.  Ugaritic had already resorted to a small vertical wedge. The practice continued in the west-Semitic inscriptions of the eleventh and tenth centuries.  The ninth-century Moabite Mesha inscription  uses a dot to separate words and small strokes to mark out sentences and contextual units. Three dots occur in the Lachish ewer from the thirteenth or twelfth century (Cross 1954:20) and in two lines of the Tell Fekherye inscription. Pairs of dots and single dots are even more common. 
As to the arrangement of the verses, their placement between the handles will have allowed the owner to view and use them as a prompt to performance while holding up the cup. The alignment seeks to keep the text fully between the handles and in view of the drinker. This mise en page seems to have followed a false start with a spelling error (NH- for NE-).  The natural distinction in kind between the first line, which identifies the owner and the object, and the following two, which describe its properties, helps to explain the line division. Even without positing that the inscriber understood or felt the hexameter as a unit of composition and performance, the second line was bound to be divided from the third if it was to avoid running against the now isolated NH. It is obvious that the space left between the end of κενον and NH was not large enough for hιμερος.
[ back ] 1. All page numbers refer to this work unless otherwise noted. See also Powell 1989. Note that some of Powell’s depictions are scarcely accurate and mislead the reader (cf. Johnston 1992). For an earlier catalog, see Heubeck 1979:109–126. I have also consulted the recent editions of Pithekoussan inscriptions by Arena 1994:15–23, Dubois 1995, and Bartoněk and Buchner 1995; and the edition of Eretrian inscriptions, primarily from the sanctuary of Apollo Daphnephoros, by Pfyffer et al. 2005.
[ back ] 2. Cf. also Johnston 1983:67–68.
[ back ] 3. See, for example, LGPN I (×7), ⅢA (×7), ⅢB (×9). Cf. Dubois 1995:31 no. 6 and Bartoněk and Buchner 1995:157–158 nos. 6–7.
[ back ] 4. Cf. Ἰάσων < ἰάσασθαι.
[ back ] 5. See Brugmann-Thumb GG 337. Cf. Hackstein 1997–1998:31n16 and Hackstein 2002:88. Inscription 1273/4 appears in pp. ⅷ–ⅸ of the addenda ultima to IG Ⅻ.9.
[ back ] 6. Cassio (1991–1993:196) had noticed τείσε⟨ι⟩ a couple of lines later in the same Eretrian law and refers to “il contrasto” of τίν[υ]σθαι with it. Presumably, he accepts an otherwise unattested paradigmatic relationship between these two forms, the zero- and the e-grade (see further below, §5.1.3). That in the Homeric text τῑ́νω and τῑ́νυμαι/τίννυμαι are paradigmatically complementary is no object. Note that τεῖσαι cannot be derived synchronically from *τίνϝω: its aorist would be τῖσαι.
[ back ] 7. For a general up-to-date study, see Verdan et al. 2008.
[ back ] 8. Cf. Coldstream 1995:260.
[ back ] 9. Coldstream 2003:438. Verdan et al. 2008:110 write: “Il y a une véritable continuité, dans la production eubéene, entre la fin du Géométrique et le début de l’époque archaïque. On aurait aimé pouvoir conclure en abordant plus en détail cette transition entre les deux périodes, mais les ensembles étudiés ne nous le permettent pas. Il est ailleurs difficile de fixer une limite inférieure pour les plus tardifs d’entre eux.” On the beginning of LG I and the transition between LG I and LG Ⅱ in Pithekoussai, see Verdan et al. 2008:109 and n. 741 (with chronological tables at 135).
[ back ] 10. For reports on the site see Buchner 1970–1971, Buchner 1972, and Klein 1972 (Buchner’s articles depend on Klein’s work and tentative conclusions). For a nice derivative summary see Ridgway 1992:91–96, with a bibliographic note at 152.
[ back ] 11. See Nizzo 2007:84. Descoeudres and Kearsley (1983:28) extend the manufacture of Euboian LG Ⅱ, type Ⅵ chevron skyphoi into the early seventh century. This is of relevance if the fragments of Ischian skyphoi that they list in 23 as nos. 17–18 are correctly identified as type Ⅵ. Nizzo (2007:153) includes under B390(AL)B1 skyphoi that could be as late as Level 28, a stratum that he dates to ca. 680 (cf. 2007:85, fig. 39). For Nizzo’s adjustment to Coldstream’s date of 690 (Coldstream 2003:326–327), see Nizzo 2007:83. Nizzo follows Neeft 1987, with whom Buchner disagrees (Bartoněk and Buchner 1995:201–202, with reference to the discussion in Neeft 1987:372–380).
[ back ] 12. Buchner (1972:367) seems to corroborate that pottery is the primary (if not the sole) basis: “La ceramica più recente contenuta in questo scarico appartiene al primo quarto del Ⅶ sec.” In this same work, at 369, he mentions middle Protocorinthian kotylai. Similarly, Ridgway 1992: “The investigation yielded remains … that can be dated by its ceramic finds to the period between the middle of the eighth and the beginning of the seventh centuries” (92). Cf. Descoeudres and Kearsley 1983:10.
[ back ] 13. The absolute dates for the various structures seem to me no more than educated guesses based on the assumed chronological boundaries for the occupation of the site and the relative order of the structures and their building phases. The kratēr fragment was retrieved from under one of the foundation stones of structure Ⅱ (Buchner 1970–1971:67). If Strabo’s report of earthquakes (5.247) is credible, tremors and falling boulders from the Mezzavia ridge must have forced repairs and reconstruction on more than one occasion. That all structures but I, which was suddenly abandoned, show signs of reconstruction seems to confirm this view. Klein (1972:36) observes that “[f]allen-down walls elsewhere and constant rebuilding are the sad reminders of harsh natural conditions which … drove away many of the original colonists.” One should assume that this “constant rebuilding,” which must have happened sporadically right up to the time when the settlement was abandoned, may occasionally have included the foundation of collapsed walls. Hence, the location of the kratēr’s find does not require a terminus ante quem for it of 700.
[ back ] 14. The year 700 marks the end of Athens’ geometric style and the beginning of proto-Attic.
[ back ] 15. Note the caution of Turfa 1994: “The First Western Greeks can’t solve all our problems, after all: I still approach absolute chronology for the 8th–7th centuries with great trepidation. The dates of Greek pottery styles are keyed to colony foundations and the like, yet these were often dated by the circular route. Thus, P[ithekoussai] is dated by Coldstream’s Geometric scheme, which incorporated material from P[ithekoussai]—probably they are perfectly correct, but there are still no guarantees.”
[ back ] 16. In Bartoněk and Buchner 1995:177 no. 43, Buchner writes: “Johnston … datiert es ‘c. 700’. Es its anzunehmen, daß es etwas älter ist.” Without new evidence or analysis I do not believe that we must follow Buchner.
[ back ] 17. Brugmann-Thumb GG 45. From the sixth-century ϝιόλεως, which he thought a genuine Attic orthographic archaism, Thumb (1898:334) argued that even for Attica there was no need to push the loss of initial ϝ- to a “vorhistorische, d. h. nebelhafte Epoche.” But cf. Threatte 1980:23.
[ back ] 18. For chronological distinctions in the loss of -w- see Peters 1980:300, who considers (but does not ultimately adopt) a possible loss of -w- before iota (a near-converse of what I am proposing here) prior to Attic reversion after rho, while allowing for other intervocalic losses after the contraction of /e/ + /a/ → /æː/. It is interesting to observe that the context of this particular proposal is *paræːwiyæː → *paræːyyæː.
[ back ] 19. A homorganic glide -oy. y e- would develop for earlier -oy.we-. In time, intervocalic -yy- would weaken and we meet with outcomes like the Euboian μ’ εποεσεν (IG Ⅻ.9 43, ca. 450). Cf. the Eretrian homorganic glide δυϝε in IG Ⅻ.9 1273/4, 2.2.
[ back ] 20. Del Barrio Vega 1987:216, with Jeffery 1990:85, 87 no. 10 and Pfyffer et al. 2005:79C (“daté du début du 6e siècle”). Others read ποιε̣[ (SGDI Ⅳ.851–852).
[ back ] 21. Also SGDI 3188 and Buck 1998:294 no. 93.
[ back ] 22. See Jeffery 1990:293 and 304 no. 15. She tentatively dates it to ca. 700–650. The reading is disputed: Δηιδαμανι̣ (dat.) according to Jeffery (received by Powell and by Ruijgh 1997:583); Δηϊδαμαν (voc.) according to SGDI 5351 and IG Ⅻ.7 442.
[ back ] 23. CEG 1.221–222 no. 403, Del 3 758, Jeffery 1990:303 no. 2. Some date it as late as 625 BC. The word ϙορη instances the loss of -w- after rho (not surprising if its date is no earlier than 650 and perhaps as late as 625), although, strictly speaking, the date that matters is not the Naxian but the Attic for the loss of -w- in -rw- clusters (as a terminus ante quem for the second CL). The earliest attested examples of κόρη in Attica seem to be: the Phrasikleia inscription IG I3 1261 from ca. 540 (κόρε); the now lost IG I3 509bis, tentatively dated in DAA 358 to 558 BC (ϙόρει); and IG I3 618d from 520–510 BC (κόρει).
[ back ] 24. Unsurprising at this date, especially since it is rarely preserved by any dialect: Δϝεινία at Corinth (IG Ⅳ 358) and in Boiotia (IG Ⅶ 2742); Xελιδϝόν in Aitolia (IG Ⅸ.12 1.86); and δϝίς in Corinth (CEG 1.189–190 no. 355, with addenda in CEG 2.303; cf. SEG 29.337).
[ back ] 25. MFA 98.900. It is no. 22 in Jeffery 1990:88, where it is placed under the heading “Inscriptions attributed to Eretria.” For Tarbell (1902:45), the vase’s first publisher, “the chances appear to be strongly in favor of the view that the vase and inscription were made by a native of Chalcis.” Friis Johansen (1923:171) accepted Furtwängler’s opinion that it was imitation Protocorinthian and Bechtel’s proposal in SGDI 5292 of a Boiotian author under Chalcidian influence (he recorded no judgment about its date). Hoppin (1924:3) thought that it belonged “to the middle period of the Proto-Corinthian style before the Oriental influence had made itself felt.” It was therefore “not later than the beginning of the seventh century.” This is the dating accepted by the Museum’s curators (“[e]arly Protocorinthian Period, about 700 B.C.”), although Jeffery tentatively prefers ca. 650?, the lower bound in Tarbell 1902:42 (“the latest date that is to be thought of”).
[ back ] 26. Bechtel disputes this reading by Buck on the grounds that it cannot be pure Ionic since ἐποίε̄σεν lacks intervocalic -ϝ- (SGDI 5292). But this assumes what I am disputing here, namely, that -ϝ- after iota was lost more or less contemporaneously with all other intervocalic contexts. (To explain the form, Bechtel proposed a Doric masc. ᾱ-stem genitive ending -ο̆ with a stem Ἀγασιλη- reminiscent of Boiotian hypocorisms Mνασίλλει etc.) Thumb and Scherer (1959:261–262) read Ἀγασίλεϝο̄. But QM before the loss of intervocalic -ϝ- is improbable (cf. Peters 1980:300 and Méndez Dosuna 1993:97), while positing a secondary (i.e. non-etymological) -ϝ- glide where there was once an etymological -ϝ- strikes me as special pleading.
[ back ] 27. For the Attic ]ν ἀϝυτ̣[άρ, which some read and supplement as ναϝυπ̣[εγός or ]ν ἀϝυτ̣[ός, see Jeffery 1990:76 no. 7; Threatte 1980:23; and IG I3 589, whose editors accept it as Attic and date it to ca. 650–600 BC.
[ back ] 28. Both apud Buck 1902:47.
[ back ] 29. Lejeune 1945:103n1 agrees: “Sur le maintien du digamma et la forme du génitif, il faut donner raison à Buck … .” Cf. Dubois 1995:59–61 no. 22.
[ back ] 30. See, with further bibliography: Jeffery 1990:116–117, 130 no. 2; Powell 1991:156; Dubois 1995:36–40 no. 11; and Bartoněk and Buchner 1995:201–204 no. C.2. For the date, see Cassio 1991–1993:187n5 and Colonna 1995:333n35. SEG 41.848 contains a summary of the arguments. If Neeft 1987 is right to fix 680/675 as the end date for the EPC globular aryballos—late examples of which were found in the tomb—the lēkythos would have a terminus ante quem of 675 BC. Buchner disagrees with Neeft, but he admits that if we downdate the lēkythos, currently assigned to 700–690, by 10–15 years, “[das macht] im übrigen keinen großen Unterschied” (Bartoněk and Buchner 1995:202).
[ back ] 31. Frederiksen 1984:119 notes that “no recent Etruscologist would accept that the language is Etruscan, and it is hard to detect in its letters or morphology any links with the known languages of Italy.” And he adds: “The possibilities seem to be that we have either a pre-Italic native tongue of the Cumae area, or else a yet unidentified tongue spoken by a foreign visitor to Pithecusae and Cumae” (120).
[ back ] 32. See Cassio 1991–1993:189.
[ back ] 33. His article was deemed “brilliant” by Watkins 1995a:42. See 42–45 for Watkins’s own analysis.
[ back ] 34. Colonna 1995: “[L]’ipotesi etrusca [è] oggi sostenibile senza forzature di sorta” (336). His reading is approved by Ridgway 1997: “[I]t has been authoritatively argued that the well-known Tataie and the notoriously enigmatic hisamenetinnuna (c. 700–690) inscriptions at Cumae represent two more early Etruscans there, of which the latter—Hisa Tinnuna—was, pace Frederiksen 1984, 119 and Cassio 1993, using the Euboean version of the Greek alphabet to write, or to have something written, in his own language” (336, his emphasis; cf. 337–338). See also Colonna 2002:198.
[ back ] 35. See Cristofani 1975, esp. 145–152 and Stoddart and Whitley 1988:768. Bartoněk and Buchner 1995:202n15 reject Colonna’s interpretation with two arguments. First, they observe that the lēkythos was designed for an exclusively funerary use. This follows from its decoration—a serpent that winds around the base of the missing neck—and the failure to turn up similar vases in the settlement of Pithekoussai, although such lēkythoi are common as corredi both at Pithekoussai and Cumae. This observation may well be true, but I do not see why it should preclude an Etruscan gift to a Greek corredo (at least, Ridgway 1997:337 does not think it does). We will never know for sure how the alphabetic inscription ended up on the same vase. In view of the Corinthian style of the lēkythos, the switch of script from Euboic to Corinthian commends as its source a Corinthian potter working in Cumae. If so, the presence of the scribbling was not deemed an impediment to the solemn gesture of giving. Buchner and Bartoněk’s conjecture that a child of the man who wrote what they think is Greek, on seeing his father write, said, “‘ich kann auch schon schreiben’ und kritzelte die Buchstaben” (Greek and Corinthian!), is as moving as it is implausible.
[ back ] 36. On the spiritus asper as a reflex of initial ϝ-, see e.g. Solmsen 1901:186–220, Sommer 1905:83–136, Schwyzer GG I.227 §C.d.β4–5, and Sihler 1995:183 §188. There is still much that we do not understand, but, as Sihler notes, that nearly all cases of PIE *w- appearing as spiritus asper in Greek are followed by -σ- is unlikely to be a coincidence, even if no broadly accepted phonetic mechanism has been advanced (Sihler 1995:184; cf. Méndez Dosuna 1985:112n57). Schwyzer GG I.306 believes that the koinē form ἵσος is a hyper-Atticism. The view that its spiritus asper is somehow analogical is generally shared (e.g. Sommer 1905:105 and Méndez Dosuna 1985:113). If so, an early form hisa for Cassio’s conjectured *ϝίσϝα would be impossible. But even granting ex hypothesi that such a phonetic development could happen, I note that it is at Ephesos (e.g. Del 3 708a3.3, ante 321 BC; IEph 1435, 322 BC) and Samos (e.g. IG Ⅻ.6.1 18, ca. 320 BC) where we find the earliest evidence of non-psilotic ἵσος (which becomes rather common during the Hellenistic period, especially in the expression ἐφ’ ἵσῃ καὶ ὁμοίῃ; cf. Sommer 1905:106n1). This suggests that we only consider the East-Ionic syllabification ϝισ.ϝος as possibly developing into ἷσος (perhaps from the devoicing of the initial ϝ-). Cassio’s reading is thwarted by this hypothesis too, which accords with Lejeune 1972:176–177 §183. (On the meaning ad loc. of Lejeune’s “s (appuyant)” and “s appuyé,” see Clédat 1917:71 §60.) I consider immediately below the anomalous case of the Heraklean hίσον in IG XⅣ 645.175 (=Del 3 62, ca. Ⅳ/Ⅲ BC). That ἶσος is the regular East-Ionic outcome is disputed by Horrocks (1987:277) on the basis of ἴ̆σως in Herodas 2.79. He thinks that Homeric ἶσος reflects artificial lengthening after the loss of postconsonantal -w-, which was needed because the meter had treated syllables before the cluster -σϝ- as heavy. In other words, even though ϝι.σϝος would have been the syllabification in East Ionic before simplification, nevertheless the first syllable ϝι- would have scanned long when the corresponding Homeric verses were composed. (The alternative, that there was no CL although -C.w- was the syllabification before simplification, is precluded by linguistic studies of CL. Horrocks 1987:277 seems to invoke this impossible alternative when he writes that “syllable division prior to simplification should not be assumed to be a reliable guide to syllable quantity after.”) “[A]ll cases of short vowel before group with /w/ counted as heavy syllables before the loss of /w/, but only short vowels before liquids and nasals were lengthened in the everyday spoken language of the East Ionic region” (Horrocks 1987:277). I find Horrocks’s theory hard to accept. It requires differing syllabification schemes for -Cw- clusters within East Ionic that are not supported by the evidence. One may not with any plausibility allow ἴ̆σως in Herodas 2.79 to trump the ample evidence of Homeric poetry for East-Ionic everyday spoken language except under the questionable assumption that the Iliad and the Odyssey were fossilized early enough to prevent any linguistic updating in performance. Should we not expect at least one of the seventy-three occurrences of isos in Homer to scan ἴσος, just as it does occasionally in Hesiod? (E.g. Hesiod Works and Days 752; cf. Edwards 1971:107 and LfgE s.v.) In support of his implausible theory, Horrocks adduces the Boiotian inscription no. 37.1 in Buck 1998:227, which features a καλϝόν that scans ⏑ - (=CEG 1.178–179 no. 334, with addenda in CEG 2.302–303, which Hansen dates to “ca. 550–25?”; cf. Buck 1909:80 and SEG 29.449). Yet Boiotian is generally held to have simplified -ρϝ- and -νϝ- (and, presumably, -λϝ-) without CL (Bechtel 1963:1.230). If so, the syllabification of καλϝόν before simplification must have been κα.λϝόν. Assuming that this was indeed the syllabification uttered by a hypothetical reader of this inscription, the meter could only be observed by the artificial lengthening to κᾱ- of the first vowel. For what could be the meaning of the statement that “the syllabification was κα.λϝόν,” if not that the speaker correlated κα- with the long thesis and -λϝό- with first breve? (The closing -ν would syllabify with the following word, ἄγαλμα.) If so, the uses Horrocks cites at 278 must all be instances of metrical license. (Alternatively, they would call metri gratia for the syllabification καλ.ϝόν.) Of course, Horrocks does not assume for the inscription the syllabification κα.λϝόν. He adduces it precisely to make the opposite point: that the heaviness of the first syllable (which calls for καλ.ϝόν) does not correlate with the simplified Boiotian κᾰλόν attested later—implying that καλ.ϝόν devolves directly into κᾰλόν. But, once again, this impossible postulate is contradicted by linguistic studies of CL, which actually do tie syllabification before simplification to the presence or absence of CL in the outcome. If not metrical license, the truth must be otherwise: there was a change of syllabification from καλ.ϝόν to κα.λϝόν, and it is the latter that results in κᾰλόν. The dates of the inscriptions adduced by Horrocks are: ca. 550–525 BC for CEG 1.178–179 no. 334; ca. 625–600 BC for IG Ⅸ.1 867 (=CEG 1.78–79 no. 143); and ca. 575–550 BC for IG Ⅸ.1 869 (=CEG 1.80–81 no. 146). IG Ⅶ 2533 (=CEG 2.195–196 no. 786.ⅳ), dated to ca. 335 (see CEG 2.104 ad no. 630, with bibliography), should be set aside since the correct reading is not Kορϝείδας but Kορβ̣είδας (so Hansen ad loc., accepted by the LGPN Ⅲ.B.243). Even if it were etymologically derived from κορϝ-, the β for ϝ bespeaks in practice a rather different phonetic value that cannot serve to support Horrocks’s argument. (Cf. Blümel 1982:85n70.) The chronological range of the adduced evidence is therefore ca. 625–525. We may extend the lower boundary into the early fifth century if, with Nachmanson (1909:144), we accept that Korinna used the form κόρϝα and we reckon her the contemporary of Pindar. Given that her own name appears in PMG 657 as Kόριννα (⏑ - x), we may date the change to the innovative syllabification to the beginning of the fifth century BC. Regarding the simplification of -rw-, Bechtel 1963:1.230 writes: “Die jüngren Inschriften bieten nur Formen mit der Kürze” (i.e. without CL). He is followed by Thumb and Scherer 1959:30 and Blümel 1982:85. But, since the Boiotian script regularly used the same grapheme for ο and ω until the end of the fifth century BC (cf. Jeffery 1990:94 and Blümel 1982:32–33), without further information one cannot readily infer the quantity of O in an inscribed KOΡΑ, as the hesitation in SGDI 908 shows: “Kωρα- (oder Kορα-).” IG Ⅶ 587 prints Kόρα after Meister’s later judgment in the “Nachträge” to SGDI I (p. 404): “Die Inschrift ist Kόρα zu lesen.” He bases his judgment on later inscriptions (with Ionic script) like IG Ⅶ 710 which Koumanoudēs (1872–1881:298) dated to the Hellenistic period, when Attic influence cannot be precluded. (Cf. Bottin 2000:90 §39, who lists among Boiotian traits “κώρα < κορϝα.”) At any rate, I can readily embrace the consensus that -nw- and -rw- were simplified without CL, so long as the intervening adoption of the innovative syllabification -.nw- and -.rw- is granted. Be that as it may, even if Horrocks is right, his theory does not require us to depart from the expected Euboian syllabification ϝι.σϝος, which leads not to hisa but isa.
[ back ] 37. Cf. Meister 1871:397–403 §8, Bechtel 1963:2.384–385, and Uguzzoni 1968:38 §8. Schwyzer GG I.305 §2: “ἵσος nach ὅμοιος.” On this matter Méndez Dosuna 1985:111–116 should be consulted. Note his statement at 113: “[E]s muy probable que la analogía haya desempeñado un papel de importancia en casos como los que acabamos de mencionar” (these cases include ἵσος). See also Sihler 1995:184 §188.
[ back ] 38. Del Barrio Vega 1987:215–219 §3.11. Also Del Barrio Vega 1991:21 §9.
[ back ] 39. Dubois 1995:114–125 no. 43 conveniently collects the inscriptions on these vases.
[ back ] 40. “Pero, ¿por qué no pensar que la -w- era pronunciada todavía en los siglos Ⅶ y Ⅵ, e incluso en el s. V, tal como indican las inscripciones?” Similarly in Del Barrio Vega 1991: “[P]ensamos que no hay ningún argumento válido en contra de la existencia de la digamma en euboico en los s. Ⅶ y Ⅵ” (21).
[ back ] 41. Scholars are not agreed whether Etruscan /f/ was labiodental (Rix 1984:209) or bilabial (Bonfante and Bonfante 2002:77–79 and 125n46; also Lejeune 1966:149). For the origin of the Etruscan alphabet, see Cristofani 1972 and Rix 1984:202.
[ back ] 42. Both cited above. For the latter, see immediately below.
[ back ] 43. Bartoněk and Buchner (1995:173) note that “ein ebenso unspezifisches ϝοῖνος (Wein) erscheint unwahrscheinlich.” Why this should be so is not explained and they have no better suggestion in its place.
[ back ] 44. Thumb and Scherer 1959:262 §311.14a attribute ϝοικέο̄ν and ϝοι, the two examples from Rhegion (SGDI 5276), to the Doric element of its population (cf. Del Barrio Vega 1987:217–218). Two inscriptions on bronze helmets dedicated by the people of Rhegion to Zeus at Olympia do not preserve intervocalic ϝ: το̄]ι Διὶ Ῥεγῖνοι Γελεαίον (SEG 24.303; cf. SEG 45.407) and Διὶ Ῥεγῖνοι Λοκ̣[ρο̄ν (SEG 24.305), both dated to Ⅵ/V BC. Kunze (1967:101–103) suggests the early fifth century for both, in particular, the late 490s on historical grounds for the former, which Landi (1979:326 no. 225) unaccountably dates to the sixth century. (Cf. Dubois 1995:96–97 nos. 33–34.) If ϝοικέο̄ν and ϝοι are not ascribed to Doric influence and one needs an explanation for the neglect of intervocalic ϝ in Διί, I need only point out that it follows iota and that it stands between two identical vowels.
[ back ] 45. On which see below, §5.1.3.
[ back ] 46. Dubois 1996:50–55 no. 23; Effenterre and Ruzé 1994–1995:2.261–263 no. 72; Jeffery 1990:478 no. 60c. Most scholars date it to ca. 500 (Jeffery apud Chadwick 1973:35: “not far from 500 BC”).
[ back ] 47. Dubois (1996:184) cites other examples of this “propensity” towards monophthongization: “Le phénomène qui ne subsiste quasiment plus à Milet que dans le verbe ἐ̃πε = εἶπε … était donc vraisemblablement un trait phonologique du milésien du Ⅶe siècle.” The Milesian ἐ̃πεν (Bechtel 1963:3.34) occurs on an inscription from the Delphinion published in Kawerau and Rehm 1914:276–277 no. 132a. Rehm writes that the script is “spätarchaisch, schwerlich lange vor 500.” Note the dotted theta ʘ, the H-shaped eta, the A with horizontal crossbar, and the non-sloping epsilon with horizontal bars (add to these the isosceles lambda Λ). For Jeffery 1990:343 no. 39, it is “c. 500–480?” and it “bears what is apparently our latest example of a boustrophedon text from Miletos” (335). There is a second instance of ἐ̃πεν in Rehm 1958:8 no. 11, an inscription now lost, edited from a squeeze by Haussoullier. Only while surveying the development of the archaic script of Didyma and Miletos does Rehm offer an approximate date for it (1958:1–3). His table on page 2 collects letter forms from no. 11, the inscription from the Delphinion cited above, and others. Not only does Rehm state that no. 11 features the “erreichte Endform” of epsilon (2); it also exhibits the final H-shape of eta and the dotted theta that characterize the third and final phase of the script (1) (Concerning the dotted theta, Rehm 1958:8 observes: “[D]er senkrechte Strich im ʘ dürfte Steinverletzung sein. Ich habe ʘ notiert.”) Hence, the letter shapes of no. 11 are late-archaic. Rehm’s vague dating follows by inference from a conjectural date of 540 BC for his key comparandum, an inscription by Aiakes, the father of the Samian tyrant Polykrates: “[Die Aeakesinschrift] kann man mit unserer n. 11 wie mit Mil. I 3 n. 31 wohl vergleichen. Für alles, was vor n. 11 liegt, haben wir die Möglichkeit, in die erste Hälfte des 6. Jhs. hinaufzugehen” (3). This probably led Bravo (1980:858) and, after him, SEG 30.1291 to date no. 11 to the sixth century, although it is clear that only the second half of the sixth century would do justice to Rehm’s words. Rehm himself makes clear, however, that 540 is too early and no. 11 must actually be ca. 500. For “Mil. I 3 n. 31” is dated to “nicht lange vor 500” (Kawerau and Rehm 1914:162), confirming Jeffery’s assignment to “c. 500?” and her judgment that “[a]ccording to the squeeze the lettering should not be earlier than the late archaic period” (Jeffery 1990:335 and 343 no. 36). Jeffery adds that this lettering “may perhaps be compared with that of 39” (335), i.e. with the inscription from the Delphinion that is our only other source of ἐ̃πεν. Unwarranted higher dates for no. 11 are common in the literature. Robinson (1981:350 D-2) gives the “6th cent[ury]” without discussion. Fontenrose (1988:180), similarly, the “early sixth century.” As further support for a date of ca. 500, I might cite Newton (1862–1863:2.783 no. 70), who observed that its letters are “rather less archaic” than his no. 66, an inscription that he had assigned to the first half of the sixth century (for Newton’s drawing of no. 11, see Roehl 1882:132 no. 489d). Newton’s no. 66 is Jeffery 1990:342 no. 22, “c. 600–575?” (Fontenrose’s and Robinson’s dates for the Delphinion inscription are also too early.) I submit that these two contemporaneous, late-archaic Didymaean oracular responses have the same source: an official at Didyma or Miletos who instructed the inscribers, whose tendency to monophthongize ει- was exceptional and did not reflect a Milesian linguistic habit (not even a marginal one). I cannot speculate further about his identity, since we know nothing certain about the archaic administration of the Oracle of Apollo at Didyma. Cf. Fontenrose 1988:45 and Greaves 2002:123–124. I cannot agree with Greaves or Jackson 1995 that no. 11 concerns guidance to private individuals—why inscribe the answer? An anxious question about the propriety of piracy does not seem an endorsement of individual piety suitable for display (cf. Robinson 1981:67). Bravo’s view (1980:858) is preferable. Note also Parke (1985:28), who allows for “the community of Miletus” as the questioners. Jackson (1995:95n1) mischaracterizes Jeffery’s prudent dating as “non-committal.” The few remarks in Günther (1971:18, with nn61–62) add nothing of interest. Dubois apparently assumes that ἐ̃πεν exhibits a genuine Milesian linguistic trait; and that, inasmuch as it is common to Miletos and its colonies but exceptional in the metropolis, it must predate the foundation of Berezan. But all his examples from the area of Olbia are also ca. 500 or later. I wonder if the two Milesian ἐ̃πεν might not reflect a colonial linguistic development that traveled back to the metropolis in the person of the official at Didyma charged with overseeing the inscribing. If so, this monophthongization is late-archaic and not a valid parallel to the alleged Cumaean μενε̄. Another oddity in the Berezan lead letter, perhaps not unrelated to its monophthongal ει, is the dative Mατασυ for Mατασυι (also printed as Mατασιν!). Cf. Miller 1975 and Merkelbach 1975.
[ back ] 48. With addenda in CEG 2.304.
[ back ] 49. Schwyzer 1987:373 no. 786 and Jeffery 1990:240 no. 3.
[ back ] 50. For its dating to 525 BC see Cairns 1984:148.
[ back ] 51. Schwyzer GG I.697 §γ.
[ back ] 52. τίνω comes from the zero-grade of the root by -νϝω thematic affixation (Schwyzer GG I.698 §η). τεῖσαι is formed directly upon the root’s e-grade. Cf. García-Ramón 2007.
[ back ] 53. So Chantraine GH I.303, after Wackernagel 1970:77–81.
[ back ] 54. Or by confusion with τῑ́ω < *k w i-ye-, derived from the root aorist of *k w ey/k w i. Cf. LIV 2 378. τίω and τίνω are differently derived from the same root (LIV 2 380).
[ back ] 55. Accounting for an original derivation with -νν- and motivating the subsequent simplification is problematic. See below, §5.1.3.
[ back ] 56. Even Wackernagel 1970:255, in his addendum to page 80, without observing that this undermines his argument.
[ back ] 57. Cassio (1991–1993:198n41) seeks to do away with this (to him) inconvenient alternative by asserting that the τίνομαι attested to by Attic drama in the sense of “mi vendico,” like the Homeric τίνυμαι, was never current in ordinary speech: “τίνομαι non abbia mai fatto veramente parte del dialetto attico parlato” (202, my emphasis). His argument ails on several counts. First, because it is not clear that the relevant sentence of the inscription, which is notoriously difficult to translate, bears out the meaning “esiga il pagamento della penalità (δίκεν)” (198n41)—that is, it is not clear that the alleged τίν[υ]σθα(ι) is a middle and is to be taken with δίκην in the sense of “vendicarsi” (‘to inflict punishment’ or ‘to take vengeance’, with χρέματα δόκιμα in apposition). Furthermore, construing with Cassio the opening ΔIKEN as the object of τίν[υ]σθα(ι) comes at a high cost, for we must then accept the strained hyperbaton of the intervening ἐπεὰν κατομόσει. Effenterre and Ruzé (1994–1995:1.330), who with some hesitation read ΔIKEN as δίκην, reject this unnatural word order and instead suspect that the beginning of the sentence is missing. A second weakness in Cassio’s argument is strictly semantic. It concerns the marked division he makes in the application of τίν[υ]σθα(ι) to two separate spheres, the moral (‘to exact punishment’) and the material (‘to exact payment of money’). This division underlies his assertion that for the meaning “vindicarsi” Attic substituted τίνυμαι by τιμωροῦμαι, and for “riscuotere una somma di denaro” by πράττω; and it underlies, further, the pivotal claim that Attic also eliminated τίνομαι “nel senso ‘mi vendico’” (198 §11). However, that the distinction adduced by Cassio led to differential outcomes in the ‘elimination’ of τίνομαι depending on its sphere of use is not only questionable but of dubious application to our case: Attic often uses ἀποτίνω for ‘to repay’ (or ‘to compensate for harm’; cf. DGE s.v. I.b), and in its passive voice (specifically, ἀποτίνεται, ‘is repaid’) it appears three times in Athēnaiōn Politeia 54.2. This is all that the verb need mean in the inscription: ‘let good money be repaid by the third day’ (cf. Lysias 1.29). Even if Cassio were right that Attic instances of τίνομαι are all of a high register (1991–1993:202–204)—in essence, a thematic literary equivalent of the athematic epic form—why should this register not be appropriate for the solemn language of legal disposition? At any rate, this assertion is tenuous, for it depends: on the notion that tragedy is invariably distant from everyday speech (presumably, Cassio would similarly dispose of Solon fr. 4.16); on dismissing instances in comedy as paratragic (Aristophanes Birds 370) or themselves of a lyric high register (Aristophanes Women at the Thesmophoria 686); on making the Herodotean instance at 9.120.10 a superficial thematic adaptation of epic diction (cf. 1.73.4, 2.108.1, 3.75.15, 4.205.2, 6.136.11, etc.); and on dismissing Xenophon’s language (Anabasis 18.104.22.168; Kyrou paideia 22.214.171.124 and 126.96.36.199–6) as poetical and having very little to do with actual Attic usage. Should any other instances of the offending verb exist in Attic, is there any doubt that Cassio would similarly indict them? Even on Cassio’s own terms, with so many opportunities for exposure to a τίνομαι allegedly not true to Attic idiom—through Attic tragedy, comedy, historiography, and elegy—one might reasonably question the propriety of classifying this form as non-Attic in the first place.
[ back ] 58. The sentence is difficult. With the objectionable φυγία it has been translated, “and [there is] exile unless he pays [to Hera?].” But why chisel out ‘Hera’? This only makes sense if hẸṚẠI is not ‘to Hera’ and instead hides a penalty that was later abolished. (For another view, see Vanderpool and Wallace 1964:386.) If so, Cairns’s pairing δόκιμα with κα[ ]υγια and his reading hẸṚẠI as an apodotic imperatival infinitive makes better sense. The interpunction, however, creates difficulties: “[T]he stone-cutter should have punctuated after ὑγιᾶ but did not” (Cairns 1984:153). That there is an error of neglect is not implausible (see immediately below). I do not know whether to accept the alternative supplement in Cairns (1991:304–305), where he argues that hẸṚẠI is the aorist ἆραι (from ἄρνυμαι/ἀείρω) with eta for long alpha and its spiritus asper borrowed from αἱρέω. The case for the aberrant asper is defended in Cairns 1983:22. But the eta for alpha is neither generally Ionic (Smyth 1894:272–273 §305.Ib; Thumb and Scherer 1959:257 §311.10b) nor specifically Euboian (Del Barrio Vega 1987:165 §3.7.1), and Cairns offers no justification for it. Whatever the case, this does not lessen the strength of Cairns’s argument for ὑγιᾶ and against φυγία.
[ back ] 59. Already suggested by Ziebarth in IG Ⅻ.9.
[ back ] 60. Vanderpool and Wallace 1964: “[The letter] was clearly tau, presumably by mistake for iota … . It seems possible that the stonecutter himself recognized his error and tried to erase the crossbar, for the stone above and around it seems to have been rubbed away” (386). No such solution could fix an omitted epsilon between the tau and the iota of τίν[υ]σθαι. The espilon is too large and adding the horizontal bars to the iota (for τέν-) would not do.
[ back ] 61. IC 2, Axos no. 1.3.
[ back ] 62. Bile 1988:238.
[ back ] 63. IC 2, Tituli locorum incertorum no. 1.8.
[ back ] 64. IC 1.272–273 Phaistos no. 3.1 (=SGDI 5112).
[ back ] 65. Guarducci writes: “In προδίκνυτι (⏑ ⏑ ⏑ ⏑) leges metricas laesas esse apparet” (273).
[ back ] 66. Also attested in IC 1.138 Lato no. 24.2 (=SGDI 5083), IC 1.170–171 Lebena no. 21.2 (=SGDI 5088), and IC 1.254–255 Olus no. 9.1 (=SGDI 5105). Cf. Bechtel 1963:2.676.
[ back ] 67. Even the syllabification is peculiar, προ.δι.κνυ.τι.
[ back ] 68. West 1998a: “τιννυ- [traditur] in paucis quibusdam [libris] neque illis optimis” (xxxⅵ).
[ back ] 69. But there is no need to reconstruct an original *δίκνυμι.
[ back ] 70. Cassio 1991–1993: “in maniere ‘impressionistica’, non proporzionale” (196).
[ back ] 71. Cassio 1991–1993:194. Cf. Bile 1988:238.
[ back ] 72. Cf. Loney 2010, esp. 169–218.
[ back ] 73. Or else the owner of the vase did not expect anyone to read the text but merely consoled himself by incising the sentiment; or the bare fact of inscribing it and depositing it in the tomb (as a magic spell of sorts) was thought to effect what it stated.
[ back ] 74. “È fatale pagare le stesse cose” (Cassio 1991–1993:190).
[ back ] 75. A photograph can be found in Friis Johansen 1923, pl. XV no. 5, or at the British Museum website (search its database for ‘Tataie’).
[ back ] 76. There is also a clever correspondence between the punishment by blindness and the act of reading, to which sight is instrumental, through which a potential offender is imagined to learn his fate.
[ back ] 77. Realizing the utter implausibility of this scenario, Bartoněk and Buchner (1995:202–203) refer the inscription to the sorrow of the father of the deceased child probably interred with the lēkythos: “‘[D]as Schicksal hat es gewollt, daß du [der Schreiber] das Gleiche erleiden mußt’ (wie etwa ein Freund von ihm, oder sein Vater, auf den Tod eines Bruders des Schreibers bezogen?)—oder völlig und ganz allgemein: ‘man muß dasselbe als Buße leisten” (203). The proposal stands indicted by these elaborate supplements: the alleged meaning must be read into words that, on this hypothesis, are downright cryptic. Why the child’s death should be the father’s penance is also baffling. Is this a consolatory message? That a child was probably interred with the lēkythos follows from the regular practice of cremation for adults (Buchner, apud Cassio 1991–1993:187n2).
[ back ] 78. With addenda in CEG 2.304.
[ back ] 79. Cassio 1991–1993:193.
[ back ] 80. ἴσθι γάρ· παισὶ τάδε καὶ δόμοις, | ὁπότερ’ ἂν κτίσηις, μένει †δρεικ????????| ?????? ?????τίνειν | ὁμοίαν θέμιν (434–436). Scholars are divided whether to read τάδε as the subject of μένει and antecedent of ὁπότερα, with τίνειν ὁμοίαν θέμιν epexegetical to τάδε (and an implicit τῶνδε that depends on θέμιν). So Haupt 1829: “scito, filios tuos domumque tuam, utrumcunque decreveris, manet hoc, ut Marti luant similem poenam” (144; cf. Sandin 2005:199 and Friis Johansen and Whittle 1980:2.338–339). Others, with Wecklein (1902:65), take τάδε as the object of τίνειν (in the sense of ‘to pay for’). I prefer (and have translated) the former, since τίνειν θέμιν seems to stand for τίνειν δίκην and would call for the gen. of what one repays (‘whichever you do, it awaits your children and household to pay the penalty for it’). If ὁμοίαν θέμιν were not present, τάδε could indeed be the direct object of τίνειν (τίνειν τάδε = ‘to pay for this’). A third alternative, by Paley (1851:39), is to take ὁπότερα adverbially and include τάδε within its clause (the μένει would be impersonal): “whether you do this [or not], it awaits your children and household to repay a like justice.”
[ back ] 81. Cassio 1994. The bibliography on the cup is voluminous. For items up to 1991, see Powell 1991:163n110 and O. Vox’s bibliography in Buchner and Ridgway 1993:751–758. To them, add S. West 1994b and the important contribution by Pavese 1996.
[ back ] 82. For a translation, see the discussion below.
[ back ] 83. Cf. Powell 1991:166n120.
[ back ] 84. Cf. Nagy 1996b:35–36, who cites Svenbro 1993:26–43. See, further, Pavese 1996:4. From close inspection of the letter sizes on a photograph I became convinced that the correct supplement for the first lacuna is simply ε̄[μ]ι. There was no need to assume the dialectally inappropriate ειμι (Pavese 1996:7–8) or the unidiomatic εστι. I was gratified to learn that Pavese (1996:7) had confirmed by autopsy that the spacing suits a lone μ. The 3rd sg. εστι departs from the sole form attested early, the 1st sg. (of which there is, comparatively speaking, what Hansen 1976:30 justifiably calls a “massive array” of instances). Although it fits the lacuna, it should be rejected, pace Watkins 1976:38–39. (Cf. Bartoněk and Buchner 1995:150–151.) Once it is epigraphically unnecessary, the only argument in its favor is the supposed difficulty posed by το̄δε as a resumption of ε̄μι. But this difficulty is largely specious. More on this below. The phonetic figure noted by Watkins 1976:39n21 adds support for εστι but cannot outweigh the array of first-singular parallels.
[ back ] 85. For the few parallels, all much later and none especially close, see Pavese 1996:16.
[ back ] 86. Pavese’s opposition to inferring from the verses a contrast between two cups is aimed at the view that the first line refers to Nestor’s Iliadic δέπας, whereas the το̄δε of verse 2 points to the Ischian skyphos. He believes that this contrast, otherwise improbable, is only tenable if one supplements εστι (Pavese 1996:16). While I share his rejection of this view (and of εστι), I believe that the contrast presented above is textually justified and makes good interpretive sense.
[ back ] 87. The closest parallel to εὔποτος is the epithet for wine ἡδύποτος (e.g. β 340; cf. Dubois 1996:71–72 no. 29). If we assume that an Iliad substantially like our own circulated at the time in Euboia, by choosing ποτήριον over δέπας the inscriber neglected a ready means of making clear at the level of diction his reference to the poem.
[ back ] 88. They are of the form – δ’ ἂν V – (– ⏑ ⏑), which can be rewritten – δέ κ’ V – (– ⏑ ⏑).
[ back ] 89. ὁππότερος δέ κε (Γ 71 92 σ 46), τῷ δέ κε (Γ 138 255), ὃς δέ κε (Ρ 229 Ψ 322 857 τ 577 φ 75), ᾧ δέ κε (Ω 531).
[ back ] 90. These are enumerated in Cassio 1994:57n14, which I have independently checked for accuracy.
[ back ] 91. West 1988:172. Contrast Pavese 1996:15.
[ back ] 92. La Roche 1866:247–250. For exceptions and statistics see below.
[ back ] 93. β 183 γ 103 113 δ 819 ξ 163 352 ο 330 σ 147 τ 322 ω 288 312 437. Add to these the reducible but probative ο 368; and the verse-final ὄφρ’ ἂν ἐκεῖθι at ρ 10, reducible to ὄφρα κε κεῖθι (cf. β 124). In a ‘probative’ instance one form cannot be turned into the other without changing the transmitted text, i.e. merely by moving the word boundary (e.g. ἐκτήσατο κεῖνος is probative because the alternative word division ἐκτήσατ’ ×ὀκεῖνος is not possible). In a ‘reducible’ instance one form can be turned into the other either by moving the word boundary or by combining a new word boundary with an alternative word end (e.g. τεύχε’ ἐκεῖνοι is reducible to τεύχεα κεῖνοι).
[ back ] 94. Reducible verse-end κεῖνος occurs at α 212 ρ 112 (non-probative) υ 265 (probative). The latter two read ἐκεῖνος in the Laurentianus 32.24.
[ back ] 95. For the elision of αὐτίκα in this sedes, see e.g. κ 237.
[ back ] 96. “[D]as Pronomen mit der ‘jener’-Deixis hat im Westion. m.E. gegen den ersten unserer Inschrift verdankten Anschein doch geradeso wie im Att. und in Homers persönlichem Dialekt ἐκεῖνος und nicht κεῖνος gelautet” (596). In 596n35 he cites with approval Bechtel 1963:3.168: “Soweit die Steine beweiskräftig sind, sagen sie aus, daß in den chalkidischen Städten ἐκεῖνος, in Kleinasien κεῖνος die Form der täglichen Rede war.” But Bechtel had no evidence for ἐκεῖνος other than a defixio from Cumae on a lead plaque, for which see Dubois 1995:54–57 no. 20 and the photograph in Landi 1979 pl. X (also SGDI 5270 with the addendum in vol. 4.889–890). Dubois dates it to the beginning of the fifth century; Landi 1979:231 no. 22 to the fifth century without further specification; both Jeffery 1990:240 no. 16 (tentatively) and Arena 1989:18 no. 21 to 450–425, before the Samnite conquest; and Schwyzer 1987:792a to the fourth century. Thus, by common consent this instance is rather late (certainly, much later than the Ischian skyphos). It only proves that ἐκεῖνος was in use at Cumae during the classical period (how broadly, one cannot say without further evidence). Del Barrio Vega (1987:407 §3.34b) supposed that the verses on Nestor’s cup were colored by Homeric diction and that κεῖνος was probably not Euboic but Ionic. This cannot be sustained in view of the cup’s glaring departures from Homeric diction. (Alternatively, she suggested the utterly improbable aphaeresis αυτικα ’κε̄νον.) After noting that the defixio from Cumae might be proof that Euboic shared ἐκει- forms with Attic, she prudently declined to assert this (“la falta de más datos nos impide afirmarlo”). Given that Nestor’s cup conspicuously departs from the diction of our Homeric poems, one should rather infer that κεῖνος was the form current in early Euboic, just as it was in early Ionic. ἐκει- forms must have been innovated in Euboic, Ionic, and Attic at a later time, perhaps independently. Not only is independent innovation conceivable because the beginning ἐ- was an old inherited deictic particle (Frisk 1973–1979 and Beekes 2010 s.v. ἐκεῖ) and the expressive accumulation of the two particles ἐ-κε- was a natural development; it is also commended by the apparently unrelated appearance of ἐκει- forms in such distant locations as Cumae and Olbia (under which I include Berezan). The instances from Olbia are: ἐκεῖ, in Dubois 1996:146–153 no. 93, dated to 550–525 (SEG 36.694, to 525–500); and ἐκεῖνοι, in Dana 2004:6, line 7 (SEG 54.694, “late 6th century” after Dana 2004:4; Jeffery 1990:479 no. N, “c. 500?”). [Santiago Álvarez and Gardeñes Santiago (2006:60) suggest ἀποδ]ώσε̄· κε̄νοι; but however attractive its meaning, this assumes that the first O in ]OΣEKENOI stood for ω, which seems unlikely given that in that same line ἀποδώ | [σε̄ν] is spelled correctly with Ω.] For Attica, the first attested uses of ἐκει- are: κἀκένοι in CEG 1.43 no. 70, “ca. vel post 500?” (=IG I3 1231; cf. CEG 1.142–143 no. 268 and 1.168–169 no. 313, both with addenda in CEG 2.302); and ἐκεῖ, IG I3 6B.33–34, “ante a. 460.” (Cf. Threatte 1996:328 §62.01b.) There is no reason why ἐκει- forms could not have been in use long before. For Thumb and Scherer 1959:276 §312.13c, “die [ionischen] Belege für ἐκεῖνος sind kaum beweiskräftig” (citing specifically the ἐκε̄νο̄ν from Cumae). With help from Dettori 1996:306–308, Peters 1998:596n35 attributes the use of ἐκει- in Olbia to Attic influence. But Dettori’s prooftexts (Dubois 1996:154–155 no. 94 and 160–164 no. 99), pace Vinogradov (1997:34), are of doubtful significance for the Ionic dialect of Olbia (Dettori 1996:308 and nn. 62, 65; cf. Bravo 2007:79), for all that they prove the growing presence of Attic speakers in the region; whereas the Olbian texts with ἐκει- exhibit many good Ionic forms and no trace of Attic influence other than the offending ἐκει-. (For the Ionic pedigree of forms like ὀπό̣[σα see Cassio 1998:16 and Ruijgh 1995:15–16.) For a general survey of ἐκει- versus κει- in Ionic, cf. Smyth 1894:207 §224.15 and 447–449 §564. I must emphasize that even if ἐκεῖνος was a genuine Ionic form (though late and secondary to κεῖνος), Homeric diction must have adopted it after the seventh century, during the definitive period of Athenian transmission (cf. Nagy 1996b:42 and Nagy 2001). For that reason, ceteris paribus, Homeric ἐκει- forms, which by then were dominant in Athens, must be considered Atticisms.
[ back ] 97. Peters 1986:319n49 and Peters 1987. Trained as an oral poet in East Ionic epic and familiar with the independent tradition of Boiotian epic, he would have introduced into his compositions forms culled from his own Euboic dialect and the diction of Boiotian epic. According to Peters, the poems were drafted in writing.
[ back ] 98. “[E]ine erst von Homer selbst in die ep. Diktion eingeführte Variante” (596n34).
[ back ] 99. Cf. Peters 1998:597n38.
[ back ] 100. All probative verse-final ἐκεῖνος are irreducible save: τεύχε’ ἐκεῖνοι (Σ 188), where no manuscript supports τεύχεα κεῖνοι; εἵματ’ ἐκείνη (ο 368), where only the long-vanished manuscript of Vespasiano Gonzaga, of unknown date (cf. Allen 1910:5–6), supports εἵματα κείνη; and ἐκτήσατο κεῖνος (υ 265), where the paradosis favors κεῖνος but the tenth-century Laurent. 32.24 prints ἐκεῖνος. Instances like οπποτεκεινων (Ι 646) are of no probative value, since the aural shape of the verse is identical whatever the assignment of the boundary epsilon. Hence the lack of documentary support for verse-final κεῖνος when the preceding word ends in -α (τεύχεα and εἵματα), since elision actually makes an aural difference. Whereas, whenever it does not, word boundary is a largely indifferent graphic choice as far as the paradosis is concerned. We therefore expect the witness of manuscripts that actually mark word boundaries to diverge in indifferent cases, since the choice, if motivated, must follow criteria that performance alone cannot establish. Such are: ὁππότ’ ἐκείνων (Ι 646), where there is manuscript support for ὁππότε κείνων (West and Allen print κείνων, van Thiel ἐκείνων); and ἐμὲ κεῖνος (α 212 ρ 112). In sum, υ 265 emerges as the only line in the two poems in which the manuscripts support a reducible κεῖνος at verse-end that actually makes an aural difference in performance.
[ back ] 101. At least the irreducible occurrences ἤματι κείνῳ are assumed by Cassio (1994:59) to be traditional.
[ back ] 102. The following other-than-verse-final ἐκεῖνος are not probative because the preceding word ends in epsilon: Η 77 Ξ 250 Ο 148 Ω 90 β 124 δ 731 κ 397 414 λ 390 418 615 ο 346 ρ 110 521 ψ 76 ω 90 313. In these loci ἐκεῖνος, if attested at all, is utterly marginal save at Ξ 250, where it is weakly supported, and at δ 731 ρ 110, where it has strong support and van Thiel prints it. Thus, there are only two arguable exceptions among them to the rule of verse-end ἐκεῖνος (about these, see the end of this footnote). Probative exceptions divide into two groups. The first consists of eleven instances for which the manuscripts decisively support κει- forms. They are: Ε 604 δ 152 739 ξ 42 70 122 153 283 π 376 ρ 243 φ 201. The second consists of eight instances for which the manuscripts decisively support ἐκει- forms. They may be subdivided into four patterns: | καὶ γὰρ ἐκείνῳ (β 171; but cf. ξ 70); μετ’ ἐκεῖνον (π 151); Ἶρος ἐκεῖνος (σ 239); and | οἷος ἐκειν- (Ο 94 Σ 262 β 272 ξ 491 ο 212). The latter five instances deserve special mention. Three correspond to the repeated unit | οἷος ἐκείνου θυμὸς (Ο 94 Σ 262 ο 212); two, to | οἷος ἐκεῖνος ἔην (β 272 ξ 491). I believe that two factors encouraged their emergence: first, the probably preexisting verse-final οἷος ἐκεῖνος | (Λ 653); second, the pattern | οἷος Ὀδυσσεὺς … (β 59 δ 689 ρ 538 τ 315 φ 94), which will have fostered (and have been fostered by) the substitution ἐκει- ~ Ὀδυσ- in β 272 ξ 491. This expression will have been modified and extended to the other cases. Peters’s hypothesis of a Homer whose composition at verse-end was more inattentive than at verse-opening (1998:598n39) clashes with the well-known fact that the inherited prosody of IE isosyllabic meters—whence the hexameter descends by expansion and substitution—featured greater metrical freedom at the opening of the verse than at its closing (Nagy 1974:34–36; cf. Miller 1990:172 no. 4 and 174 no. 7). Thus, extreme metrical regularity is a sign of relatively late composition. Viewed synchronically, to enhance compositional flexibility in the face of metrical preferences and requirements, performers innovated within carefully calibrated parameters of traditionality. In the present case, a marked preference for a dactylic fifth foot was best accommodated by the innovative ἐκεῖνος. Whereas verse-end κεῖνος allows for a dactyl only when the preceding syllable is short + open, ἐκεῖνος not only accommodates most short + open syllables (unless they feature an iota, which cannot be elided) but also short + closed syllables (long + open, permitted under epic correption, do not occur). Hence the strong tendency to place ἐκει- forms at verse-end vis-à-vis their marginal appearance elsewhere. The numbers are as follows. Of 27 verse-final instances of κει-/ἐκει-, 7 feature irreducible κεῖνος after iota; 17 ἐκεῖνος (1 non-probative, 14 probative and irreducible, 2 probative and reducible); and 3 κεῖνος (2 non-probative, 1 probative and reducible). Of 169 other-than-verse-final occurrences, 10 feature ἐκεῖνος, two non-probative, the rest probative and reducible. Excluding the older, irreducible verse-end κεῖνος after iota, innovative ἐκεῖνος occurs 85% of the time at verse-end (94%, if we only count probative cases). Reciprocally, it occurs 6% of the time when other-than-verse-end. The well-known exchange in phraseology between the Adonic sequence and the metrical space between the end of the first foot and the trochaic caesura (Householder and Nagy 1972:50)—and, secondarily, between the opening of the verse and the second trochee when Meyer’s law is not breached—accounts for the ἐκεῖνος that are not verse-final: -δεμεκεινος (ρ 110) ~ -σεμεκεινος | (ρ 112), cf. α 212; δ 731 ~ Ι 646; and Ο 94 etc. ~ Λ 653. Even if performer and audience notionally felt the breaks as the vulgate prints them—there is no telling aurally whether ὁππότε was elided or not—δ 731 does not breach Meyer’s law because of the heavy counterbalancing effect of the strong third- and fourth-foot caesuras. The same is true of β 272. As to ξ 491, with a masculine caesura and a bucolic diaeresis, it too avoids the offense of a subsequent trochaic caesura. (Cf. Kirk 1985:23–24.)
[ back ] 103. See Risch 1974:183–184 §68b and Risch 1987:9.
[ back ] 104. The fifth is Kυπρογενῆ (Homeric Hymn 10.1).
[ back ] 105. Every single time but two at verse-end.
[ back ] 106. The preference of -στέφανος epithets for Kυθέρεια over Ἀφροδίτης, only marginally noticeable in the Odyssey, is clear when we take Hesiod and the Homeric hymns into account: of eleven occurrences, eight correspond to ἐϋστέφανος + Kυθέρεια; the remainder comprise one each of Ἀφροδίτης with ἐϋστέφανος, φιλοστέφανος, and χρυσοστέφανος.
[ back ] 107. χρυσείη does not appear with verse-end Ἀφροδίτης or Kυθέρεια (cf. Ι 389).
[ back ] 108. Note that there is no epic correption and that the genitive ending -ου, by contraction of -οο < -οιο, was a monophthong (cf. Sihler 1995:259 §259.8). In the Iliad and Odyssey the only hiatus that I can find is χρυσῆ Ἀφροδίτη | at Χ 470 (cf. Homeric Hymn 5.93). Of course, χρυσῇ Ἀφροδίτῃ, of which there are seven examples, all but one at verse end, do not involve hiatus (cf. Schwyzer GG I.399 §2.a.α). All other examples of a comparable hiatus are in Hesiod or the Homeric hymns: φιλοστεφάνου Ἀφροδίτης in Homeric Hymn 2.102; χρυσῆ Ἀφροδίτη in Homeric Hymn 5.93; πολυχρύσου Ἀφροδίτης in Homeric Hymn 5.1, 9 and Hesiod’s Theogony 980, Works and Days 521, Shield 8, 47, and frr. 185.17 MW, 195.8, 47 MW, 253.3 MW; and χρυσο[σ]τεφά̣νο̣υ̣ Ἀ̣φ̣ρ̣οδίτης in Hesiod fr. 26.13 MW.
[ back ] 109. There was no reversion of η > ᾱ in Euboic. I am puzzled by the motive Cassio (1994:63) gives for the hiatus: “Variazione sicuramente dovuta al desiderio di usare il nome proprio tradizionale della dea e non un suo epiteto, e probabilmente anche favorita dal fatto che l’epiteto di Citerea, più tardi diventato semplicemente esornativo, era ancora fortemente sentito come legato all’isola di Citera.” I for one cannot divine whether the author of the cup’s verses preferred ‘Aphrodite’ to ‘Kυθέρεια’ because he thought the latter a mere epithet that fell short in the use of the satisfaction provided by the traditional name. Such psychological musings seem unsafe ground on which to build one’s argument. As to the second reason put forward by Cassio, are we to suppose that Kυθέρεια was rejected because it was not sufficiently Panhellenic? And yet the Odyssean tradition had no problem embracing it. Once again, how are we to explain these diverging sensibilities if the cup reflects the poetic milieu that gave the Homeric poems their definitive shape? (Cassio’s comment ibidem apropos θ 362 only has force if: i) one accepts the view articulated earlier in the same page that the formulaic starting point for the poet of this passage was “senz’altro ἐυστέφανος Kυθέρεια”; and ⅱ) from this view one concludes that any departure from said formula calls for an explanation narrowly motivated by the context. To me it seems safer and more economical to assume that, where the traditional name was possible, ceteris paribus it was likely to be used. At θ 267 it could, and so it was; at θ 288 it could not because of the hiatus, and so Kυθερείης was; neither was Aphrodite allowed at σ 193 because of the meter.)
[ back ] 110. Hackstein (2010:419) expresses a similar prejudice. Speaking of the Dipylon oinokhoē and Nestor’s cup, he observes that “[b]oth inscriptions are neither exclusively dependent on nor solely repetitive of Homeric diction. Instead of copying, they are innovative in coining epithets and phrases not found in the Iliad and Odyssey. Such non-Homeric words and phrases, however, can be shown to be generated by the same generative mechanisms that lie behind the diction of the large scale epics … .” To speak without proof of innovation against the background of the poems’ diction is to take as one’s starting point the assumption that these inscriptions are derivative of, and presuppose acquaintance with, our Homeric texts (or a near final stage thereof). Even if this were true it cannot be simply asserted, and Hackstein soon hedges by adding that the inscriptions are “not strictly dependent on Homer and may reflect para-Homeric traditions” (420). Indeed, nothing can be inferred from the bare fact that epic composition that might well be parallel to and independent of—or at least have the same standing as—that of the extant poems might have at its disposal “mechanisms” that also lie behind the diction of the Iliad and the Odyssey. The degree to which independent epic traditions might have shared diction (epithets, prosody, grammatical and lexical productive processes, etc.) can only be settled by particular textual and historical study. A priori judgments are not possible.
[ back ] 111. Cassio (1999:81) attempts a via media between Janko’s late eighth-century dictated text and Nagy’s evolutionary model. But it is clear that he believes in an early textual fixation: “[O] gran parte di quello che sarebbe poi diventato l’Iliade e l’Odissea circolava già, ovvero, come minimo, … la dizione utilizzata per Omero era già totalmente cristallizzata a quell’epoca” (70). And, discussing Theagenes of Rhegion and his alleged statement about Apollo’s reason to help Khryses, Cassio writes: “Teagene si riferiva a un certo verso dell’inizio di un certo poema in cui si parlava di Apollo amico di Crise; registrava cioè una variante rapsodica di un verso già strutturato in una narrazione immutabile” (81, his emphasis). More on Theagenes below, §6.2.
[ back ] 112. Cassio 1999: “In se stessa la possibilità dell’esistenza a epoca antica di tradizione epiche diverse da quella omerica a noi nota è tutt’altro che improbabile” (70).
[ back ] 113. Hackstein’s examples ad loc. are puzzling. Does he really believe that Corinthian ϝοι in CEG 1.251 no. 452.ⅱ depends on a parallel oral transmission of Ionic epic? (Not to mention the assumption, broadly shared but unprovable except by circular recourse to our Homeric texts, that Attic-Ionic—he writes “epic Ionic”—had lost word-initial ϝ by the beginning of the eighth century.) But if he only means that ϝοι is an archaism that goes back to a stage that is common to, and predates, both this inscription and Ionic epic, in what sense does he speak of the translation of hexametric epic poetry? One must always bear in mind S. West’s warning (1994b:14n27).
[ back ] 114. Cassio 1994:64 and n. 66.
[ back ] 115. Cf. the conclusion in Pavese 1996:20.
[ back ] 116. Burkert and West lean on Immerwahr’s judgment.
[ back ] 117. Note the pertinent observations by Klein 1972:38.
[ back ] 118. I do not intend to join the debate on the nature of Pithekoussai, whether an emporion or an apoikia. See the insightful and reasonable observations in Ridgway 1992:107–109 and Crielaard 1996:244.
[ back ] 119. Ridgway 1992:111–114. For the inscriptions, see Amadasi Guzzo 1987:23–25 and Bartoněk and Buchner 1995:187–188 no. A.1.
[ back ] 120. For the choice between Aramaic and Phoenician, see Ridgway 1992:153.
[ back ] 121. Bartoněk and Buchner 1995:188 no. A.1.E.
[ back ] 122. Ridgway 1992:115.
[ back ] 123. Popham et al. 1988–1989:118–119; Ridgway 1992:24; Walbaum 1994:54; Popham 1994; Bonnet 1995:656–657; and Crielaard 1996:244.
[ back ] 124. I cannot accept Sherratt’s theory of a fundamental link between Homeric epic and the derivation of the alphabet (2003:232), although she carefully distances herself from the untenable views of Powell 1991. Her objection against a predominantly commercial stimulus for its origin—why this should have happened in the late eighth century and not before (230)—applies to her proposal just as well. (The assertion that the surviving inscriptions are largely non-commercial holds true only under a narrow view of ‘commerce’—one that excludes, for example, marks of ownership [cf. Csapo et al. 2000:104]. And one may reasonably counter that, by their very nature, ephemeral commercial records must have been kept on cheap, perishable materials.) As the formulaic system demonstrates, poetry that eventually joined the Panhellenic stream of Homeric epic had been performed long before, even if we postulate a late eighth-century resurgence of itinerancy that transcended local concerns. So also there must have been an increase and invigoration of trade links, with the Greeks actually doing the traveling. Besides, if ethnogenesis was the aim of literacy, how does one explain the plethora of epichoric alphabets? Did all Greeks, everywhere and independently, suddenly seek to define themselves through the adoption of literacy? Sherratt’s argument at 233 assumes that all epichoric scripts descend by adaptation from a single invention. This is extremely unlikely (cf. Csapo et al. 2000:105–107, with reference to Cook and Woodhead 1959). Sherratt (2003:232n13) views Homeric diction and the Homeric poems only synchronically, as late eighth-century cultural artifacts. But, for the evolutionary model, no historical singularity privileges that particular point in time: diachronically, the dialectal mixture of Homeric diction extends from that point into the past and into the future. Time marks a distinction of degree, not kind. One may indeed speak of a formative Panhellenic period for the Homeric tradition that starts roughly at that time, with an increase in communication, a resurgence of travel and trade, the rise of supra-regional sanctuaries as points of convergence, and the formation of the political communities of metropolis and colony (Nagy 1996b:42). The phenomenon thus envisioned is the gradual reorientation of local praise poetry towards an increasingly Panhellenic audience (Nagy 1986). No such gradualism, much less the reorientation of an existing practice, is involved in the derivation of the alphabet. I do agree, however, that the Homeric poems articulate a Panhellenic linguistic consciousness and I do not doubt that the dialectal mixture of their diction contributed to their Panhellenic performance scope. We know that in the historical period dialects (and, to a lesser extent, epichoric alphabets) were preeminent instruments of intra-Hellenic cultural self-definition, and there is no reason to question that specific instances of writing (not the alphabet in the abstract) could have served to define Greekness against foreignness.
[ back ] 125. For its potential significance for the origin of the Greek alphabet, see Theurillat 2007.
[ back ] 126. Pfyffer et al. (2005:58), where they also list the six geometric graffiti from the city, five of them from the same area (E, F, G also in Bartoněk and Buchner 1995:190–193). Cf. Theurillat 2007:336.
[ back ] 127. The same must be true of the doubling of the lambda in καλλιστεφανο̄, a visual reminder of the need to draw out the consonant and of the corresponding prosodic length of the syllable καλ-.
[ back ] 128. For the use of scriptio continua, see Wingo 1972:14–15; Turner 1987:8–9; Nagy 2000b; Signes Codoñer 2004:124–128; and Nagy 2009b. If interpunction had been in regular use at such an early time, it is hard to explain why it is absent from so many of the earliest inscriptions; or why punctuation was not in regular use during the classical era (cf. Cassio 2002:127–128 and Aristotle Sophistical Refutations 177b). Neither the Derveni papyrus nor Timotheos’ Persians mark word separation. The writing habits of Mycenaean scribes (cf. Hooker 1980:44–45 and Ferrara 2010:21–22) bespeak a concern to make the texts maximally readable. Considering the challenges posed by the often ambiguous syllabic script, one must not wonder at the regular use of vertical strokes to mark word division (cf. Millard 1970:13). Readers of alphabetic scripts do not face such difficulties, and one can hardly assume the transference of the writing habits of the Mycenaean scribal class to the late eighth century BC. (For the restriction of Linear B to a narrow social class of Mycenaean palace administrators, see Sherratt 2003:228.)
[ back ] 129. There is no need, as Heubeck thinks, for hypothetical epic texts on perishable material to serve as models. All that is required is acquaintance with the Semitic use of punctuation and with the proper performance phrasing of epic hexameters (pauses, intonation, etc.).
[ back ] 130. Millard and Bordreuil 1982 and Lipiński 1975:2.19–82, esp. 50–51.
[ back ] 131. Cf. Millard 1970. See Sass 2005 (with Misgav 2009) for the downdating of some of these inscriptions.
[ back ] 132. Cooke 1903:1–14 and Donner and Röllig 1962–1964 no. 181 (text: 1.33; commentary: 2.168–179).
[ back ] 133. For later examples of punctuation on vases, see Pavese 1996:20. For Greek generally, see Jeffery 1990:50. For Attica, cf. Jeffery 1990:67, Threatte 1980:73–84, and Immerwahr 1990:168.
[ back ] 134. Cf. Pavese 1996:19. If Ridgway’s proposal of a Rhodian connection is right, perhaps the erroneous H was prompted by the Rhodian use of this sign for /eː/ (see, for example, Jeffery 1990:356 nos. 1–2). Before abandoning what would have been a much less adequate arrangement of the text, the inscriber began to correct H to E. Bartoněk and Buchner (1995:152) write: “[E]in solcher unüberlegter Anfang unter dem Henkel dünkt uns bei einer kalligraphisch und räumlich so gut ausgeführten Inschrift etwas unwahrscheinlich.” But the logic is circular: a carefully executed final layout cannot preclude a priori a former less careful attempt. Elsewhere the layout and execution is far from infallible: note the ποτοριον for ποτεριον; the tight and uneven spacing of hοσδα (the sigma was perhaps added later); and the nu of ἂν, first omitted, added interlineally in a reduced size under the tau of τοδε. It is not surprising that Bartoněk and Buchner have no alternative explanation for the two isolated letters.