A passage of Clearchus' Περὶ γρίϕων (‘On riddles’),Footnote ¹ quoted by Athenaeus (10.454f–455a), preserves five lines from a Hymn to Pan by the early Hellenistic poet Castorion of Soloi, at the same time providing the reader with a clue to the puzzle that this fragment – according to the poet's own words – conveys:

As Clearchus/Athenaeus says, the peculiarity of Castorion's lines is threefold: (a) each iambic metronFootnote ³ is made of eleven letters; (b) metra are divided by word end; (c) metra thus are – or should be – interchangeable within each line.Footnote ⁴ Point (a) is self-evident,Footnote ⁵ and need not detain us here. Point (c) was elucidated some three decades ago in a brilliant paper by Peter Bing, who persuasively demonstrated that, in fact, interchangeability is only apparent, and that a competent reader was expected to realize that Castorion's arrangement turns out to be the best one.Footnote ⁶ Yet something remains to say on point (b), and it is on this very point that my brief note will focus.

Lines divided into three equal parts are not unattested in Greek poetry of the archaic and classical periods. A notorious dactylic hexameter of such shape is Il. 7.238: οἶδ᾿ ἐπὶ δεξιά, οἶδ᾿ ἐπ᾿ ἀριστερὰ νωμῆσαι βῶν;Footnote ⁷ see also Archil. fr. 2.1 W.²: ἐν δορὶ μέν μοι μᾶζα μεμαγμένη, ἐν δορὶ δ᾿ οἶνος.Footnote ⁸ Trimeters such as Soph. Phil. 1300 – ἆ, μηδαμῶς, μή, πρὸς θεῶν, μὴ ᾿ϕῇς βέλος – are rare in tragedy, but become far less so in Aristophanes (Ach. 163, Nub. 661, Av. 175, Ran. 755, etc.);Footnote ⁹ in the first half of the fifth century bc we find one in a verse inscription from Aegina (CEG 451.1), where each iambic metron is marked both metrically and graphically, μὲ ἐκ τᾶς hοδȏ | λhαβὸν λίθον | στάσες σκοπόνFootnote ¹⁰ (in the second and third metron, even single feet are divided by caesura). Anapaestic systems usually feature a word end after each metron except the last before the catalexis,Footnote ¹¹ and in the anonymous fragment PMG 1027c (Terpander fr. °9 Gostoli), each trisyllabic foot – a molossus – consists of a single metrical unit: ὦ Zηνὸς καὶ Λήδας κάλλιστοι σωτῆρες.Footnote ¹² Some ancient critics did not approve of such devices,Footnote ¹³ but Castorion's practice was far from unprecedented. What seems to have gone unnoticed is that the fourth and fifth lines, at first sight, do not appear consistent with the overall picture. Both μή and ὅς usually count among appositives, that is words – not only enclitics and proclitics, but also articles, prepositions, conjunctions, particles, relative pronouns, and so forth – closely cohering with either what precedes or, as in this case, what follows, excluding the possibility of a metrical break in between.Footnote ¹⁴ Accordingly, μὴ-σοϕῷ and ὃς-μείλιγμ(α) form a metrical unit – a Wortbild in German, a parola metrica in Italian. If this were to apply to Castorion's text, the word break after the second metron at lines 4–5 should be regarded as apparent. What are we to do with that?

An easy – in all likelihood, too easy – answer might be that Castorion's poem was written (or even inscribed?) στοιχηδόν, i.e. in single monometers:

and so on. This would presumably enhance the perception of both δύσγνωστα μή and κηρόχυτον ὅς as single blocks, no matter which kind of syntactical role the monosyllables may have. The suggestion can be traced back to Wilamowitz, who nonetheless expressed it in a brief footnote, with no mention of metrical problems at all.Footnote ¹⁵ In fact, this is quite unlikely: as Bing rightly objected, ‘if this were so, neither the eleven letters nor the interchangeability of the metra would be very puzzling as they would be visually obvious. Such a reading is antithetical to the challenge of v. 4.’Footnote ¹⁶

I think that a suitable explanation may rather be found in the interaction of metre and syntax. At line five, κηρόχυτον ὃς μείλιγμ᾿ ἱεῖς involves anastrophe (the normal word order being, of course, ὃς κηρόχυτον κτλ.); and in such cases it is certainly possible that at least some poets treated the postponed element, albeit usually prepositive, as postpositive.Footnote ¹⁷ Accordingly, in Call. Vict. Sos. fr. 384.55 Pf. παύριστον τό κεν ἀνδρὶ παρ᾿ ἀϕνειῷ τις ἴδοιτο we cannot rule out the possibility of a metrical unit παύριστον-τό-κεν (though this would in some way offend against Giseke's Law), and in Castorion we may reasonably think of κηρόχυτον-ὅς. It is true that a metrical unit of this kind would apparently weaken the hephthemimeral caesura, but this is not unparalleled: scholars know only too well that iambic trimeters often allow postpositives to stand after the main caesura, and prepositives to stand before it.Footnote ¹⁸ What about line four? I find it hard to resist the temptation of interpreting it in the very same way. Castorion's ‘ideal’ reader – one who had been able to identify the metrical refinement of these lines – could realize that line four too was to be read not as δύσγνωστα μὴ-σοϕῷ κλύειν, ‘difficult for dull listeners’, but δύσγνωστα-μὴ σοϕῷ κλύειν, ‘not difficult for clever listeners’: a learned audience will receive the glass half full, rather than half empty. Rhythmical patterns would thus be absolutely consistent in all five lines. Some years ago, Kathryn Gutzwiller further developed Bing's point, acutely arguing that ‘Castorion is self-consciously poeticizing the practice of metathesis, which critics used to demonstrate an excellent arrangement producing beautiful sound’: in our text, ‘the critical tool of metathesis is incorporated into poetic practice’.Footnote ¹⁹ I wonder whether Castorion was also playing with anastrophe, turning it from a figure of speech into the litmus test to sort the σοϕός from the ἀπαίδευτος.

From this point of view, let me add a last remark on κλύειν (line four). Does it merely refer to the poet's audience, or does it convey a more pointed allusion to the very act of listening, implying that Castorion's refined poetry is ‘not difficult for the clever when they listen to it’? The presence of exactly eleven letters in each metron ‘could only be experienced visually through reading’,Footnote ²⁰ but the right syntax and rhythm of lines four and five could be better appreciated when those lines were read aloud in the proper way: that is, enhancing the word ends after μή and ὅς and not before them. I am well aware that this last point is very speculative: maybe I am just over-interpreting. Yet such a learned exploitation of the complementarity of reading and listening, of vision and voice, would be quite at home in the cultural atmosphere of the early Hellenistic age.