Dealing with fragmentary texts is an unavoidable task for anyone working on the Greco-Roman world with the awareness that only a tiny portion of the texts produced in antiquity has survived the perilous process of transmission. Since the Renaissance, generations of scholars have painstakingly collected and sifted quotations, paraphrases and allusions in later authors and grammatical sources, laying the foundation for our knowledge of large parts of that lost world. More recently a spectacular increase was made possible by the papyrological discoveries.Footnote 1 Even so, after centuries of explorations in this field, the evidence provided by the indirect sources and, in particular, by the still partly unpublished troves of medieval lexicographical excerpts and compilations is far from having been exhaustively exploited.Footnote 2
In this paper I draw attention to one of these unpublished texts, short but extremely rich. Its interest lies not only in the new information it provides about lost words and the reconstruction and the interpretation of poetic quotations (especially regarding a fragment of the late fifth-century/early fourth-century b.c.e. poet Antimachus of Colophon), but also on what we can learn from it about the ways these texts were reused and reinterpreted within the frame of different linguistic theories and scholarly contexts over the centuries. More particularly this excerpt opens an intriguing window over some less-known aspects of the comparative study of Greek and Latin in the ancient world.
THE TEXT
In 1975 A. Diller described, and partly published, a number of lexicographical marginalia present in one of the most important Strabo manuscripts, Parisinus graecus 1397. Among several other texts, Diller also mentioned, but did not transcribe, a previously unknown quotation of Antimachus fr. 118 Wyss (now fr. 154 Matthews).Footnote 3 The new text has not been used or mentioned in Matthews's recent edition and, to the best of my knowledge, the material present in this marginal note has remained unpublished.
The annotation is part of a series of miscellaneous notes, mainly of lexicographical character, added on the margins of this very important tenth-century manuscript after its medieval restoration, probably at the beginning of the fourteenth century.Footnote 4 This Paris manuscript is the only complete copy of the first half of Strabo's Geography (Books 1–9) known to have survived to the later Middle Ages in its unabridged form. It suffered extensive damage especially in its initial and final portions (ff. 1–39 and 173–232), where mice gnawed away a substantial part of the sheets and, as a consequence of this, was restored using large strips of parchment (in some cases palimpsest), which were glued over the damaged pages. The glued portions in many cases also covered parts of the earlier writing of Strabo's text. These were completed by a scribe who made use of the preserved text (now covered by the glued parchment), of some other manuscript offering Strabo's abridged text and, in some cases, of his own conjectures. After completing this operation, according to Diller, the same scribe ‘strangely abused the codex, especially the very part he had restored, filling the lower margin of many pages with various excerpts that have nothing to do with Strabo’, including the text we are going to examine in detail here.Footnote 5
Our excerpt is found on f. 232v, on the lower margin of the last preserved page of the manuscript, the most extensively damaged one. The pieces of parchment originally glued on to the manuscript have by now been separated once again. As a consequence, the left hand part of the annotation, which was on one such piece, now corresponds to Par. gr. 1397 (2) f. 1r. I give a transcription of the first five lines and of the beginning of the sixth one:Footnote 6
Par. gr. 1397 (2) fol. 1r + Par. gr. 1397 (1) fol. 232 v, marg. inf.Footnote 7
1 π(αρὰ) ̔Ρωμαί(οιϲ) ἡ τηθὶϲ \τιτθίϲ/ (καὶ) παρ᾿ Ἕ −
2 -λ̣λ̣η̣c̣ι̣νFootnote 8̣ ἐξ ̣αμ̣α (?) (ἐϲτι)Footnote 9 γραϕόμ(ενον?)Footnote 10 ἀμίτα (καὶ) δηλ(oῖ̣)Footnote 11 τὴ̣ν τηθ\τιτθ/ίδ̣(α)
3 αἶκλ(ον) τὸ δεῖπνον κ̣α̣λ̣ο̣ῦ̣ϲι (καὶ) \(κατὰ)/ Λάκων(αϲ) ἀϕ᾿ οὗ τὸ ἀριϲτητήρι(ον) ϕέραικλ(ον): - ϕάρεϲ (sic) ὁ κεκομμ(ένοϲ) \ἠλεϲμ(ένοϲ)/ ϲῖτοϲ ὃϲ λ(έγεται) χόν(δροϲ)
4 (καὶ) ̓Αντίμαχο̣c̣Footnote 12 αἰεὶ ἐν πενταπόλ\o/ει ἀεὶ ϕάρ(··)Footnote 13 χατέουϲι: (καὶ) Καλλίμαχ(οϲ) ἢ ἄϕαρον ϕαρόωϲι: ῥάδικ(α) τ(ὴν) ῥίζα̣ν̣
5 (καὶ) Νίκανδρ(οϲ): κόψαν(αϲ) (sic: legendum κόψαντεϲ) ῥάδικ(α) πολυϲτεϕέ(οϲ) κοτίνοιο: ὁ αὐτ(ὸϲ) ῥυτὴ\ν/ τὸ πήγαν(ον) λ(έγει)
6 (ὡς) ῥυτῆ γε μὲ\η/ ν ἴκελ(ον) ὀδμ\ϲμ/ήν : -
The sequence can be divided into five sections, each discussing one (or more) rare words. The beginning of the first entry is obviously incomplete. The second section deals with a Laconian addendum lexicis. In the next three sections Antimachus, Callimachus and Nicander are quoted as sources for the discussion of the meaning of various rare and difficult words.
PARALLELA GRAECO-LATINA
A. Greek ἀμμία/ἀμμίη (?) vs. Latin amita.
The first section lacks its beginning and the preserved portion is very difficult to read. It is clear however, that the text presupposes some sort of parallel involving, on the one hand, the Latin word amita (‘paternal aunt’), transcribed in Greek and, on the other, the meanings of the Greek words τηθίϲ and τιτθίϲ. These two Greek terms occur together in several ancient lexicographical texts. The starting point of this lexicographical tradition seems to be the work of Aristophanes of Byzantium περὶ ϲυγγενῶν. Most of the evidence on this section of his work, including the terms mentioned in our excerpt, is collected in Slater's apparatus to Aristophanes' frr. 220–3 and 224–7.Footnote 14 There, among other matters, the overlaps and differences among the words τήθη, μάμμη and τηθίϲ are discussed. The reconstruction of the archetype of the various sources is a complex matter that does not need to detain us here. One of the fullest versions is provided by Eustathius in his Commentary on the Iliad, 971.34 ff. van der Valk:
περὶ δέ γε τήθηϲ καὶ μάμμηϲ καὶ οὕτω ϕαϲί τινεϲ. Τήθην οἱ Ἕλληνεϲ τὴν πατρὸϲ ἢ μητρὸϲ μητέρα, οἱ δὲ παλαιοὶ (legendum πολλοί, Nauck) ἀκύρωϲ μάμμην καὶ μαῖαν, μάμμην γὰρ Ἀττικοὶ καὶ μαμμαίαν (legendum μαμμίαν, Nauck),Footnote 15 τὴν μητέρα καλοῦϲιν, ἣν δὲ οἱ πολλοὶ τήθην, τηθίδα, μαῖαν δέ, ὡϲ ἡμεῖϲ, τὴν ὑπὸ Ἰώνων ὀμϕαλητόμον, ὡϲ καὶ ἀλλαχοῦ δηλοῦται.
The sense of the passage (restored by Nauck's necessary conjectures) is rendered by Slater as follows: ‘(1) the Greeks call τήθη ‘grandmother’ (vulgarly μάμμη and μαῖα), (2) in Attic authors μάμμη and μαμμία mean ‘mother’, (3) and they call τηθίϲ ‘aunt’ what is vulgarly τήθη, (4) and μαῖα means for them ‘midwife’ <sc. only, not grandmother > '.Footnote 16 The word τηθίϲ is well attested with the meaning of ‘paternal aunt’ already in classical Attic literature. Its alternative, later sense of ‘wet nurse’, on the other hand, was easily evoked by its confusion with the word τιτθίϲ, frequently attested in our sources (and exemplified also by the addition supra lineam in our manuscript).
In our excerpt, if we are to judge from the structure of the following sections, the starting point must have been a Greek word whose sound and meaning was evocative of a Latin one (which in this case is preserved, amita), and τηθίϲ does not fit the bill: it must have been the Greek explanation of the word originally discussed in the now lost initial part of the excerpt. There are a few possible candidates for the lost word that occasioned our entry, including ἄμμα, ἀμμάϲ.Footnote 17 In my opinion, however, the most likely starting point must have been the rare word ἀμμία, which is explained as μήτηρ, τροϕόϲ in Hsch. s.v.Footnote 18 This would provide a good semantic overlap with τηθίϲ and τιτθίϲ, while being phonetically close to its supposed Latin equivalent. And the fact that τηθίϲ is well attested with a meaning overlapping with that of Latin amita (‘aunt’) would have facilitated the equation.
The only literary attestation of the word is in Herodas 1.7, where the transmitted ἀμμία (no accent and breathing in the papyrus) has been corrected into the Ionic form ἀμμίη by Herwerden, followed by the vast majority of the editors.Footnote 19 Given the apparent fondness of our lexicographical excerpt for Hellenistic poetic sources, it is even conceivable that in fact this passage may originally have provided the entry's starting point.
B. Greek ϕέραικλον vs. Latin fer(i)culum (?).
The second section starts as a discussion of the well-known ‘Laconian’ word αἶκλον (also spelt as ἄϊκλον). Our excerpt explains it with the term δεῖπνον. Its exact definition had kept ancient lexicographers busy, and had been the object of a polemical attack of Didymus against Polemon the Periegetes (fr. 86 Preller). Polemon (as transmitted by Athenaeus 4.139b–c) seems to have stated that the meaning of the word among the Dorians was δεῖπνον (quoting, to this effect, Epicharmus, frr. 34 and 109 KA), but that at Sparta it indicated the breads and meat that were brought in a wicker basket after the meal in the ϕειδίτια. According to Didymus (ap. Ath. 4.140c–d) Polemon's explanation was wrong:Footnote 20 the Laconians did not give this name to ‘the portion after the meal’, nor to ‘the food given to the ϕειδίται after the meal’, as the food brought in after the meal was rather called ἔπαικλα (also spelt ἐπάϊκλα). Athenaeus again, thanks to Didymus, also offers plentiful information, from various sources, about the ways ἔπαικλα were prepared.Footnote 21
The word αἶκλον in our excerpt, however, is adduced only to introduce a second one, the otherwise (to the best of my knowledge) unattested ϕέραικλον. This is explained as ἀριcτητήριον, in its turn a fairly rare word (relatively frequent only in the Byzantine period) which, in its other attested occurrences, indicates a dining hall or a building.Footnote 22 The formation of the word, however, implies that ϕέραικλον should rather mean an object used to carry an αἶκλον. Several of the sources dealing with the ἔπαικλα say that they were ‘brought in’ after the ϕειδίτια, using the verb ϕέρω, or its compounds such as εἰϲϕέρω (in Polemon, who talks not about the ἔπαικλα but about the αἶκλον itself) and περιϕέρω. We can conjecture that ϕέραικλον might have designated the wicker basket (ἀρριχίϲ), used to bring in the food, that Polemon mentioned in this context.
No Latin parallel is quoted in this section, but it is not difficult to guess that in this case too the issue must originally have been here, as in the other passages, exactly the similarity between a rare Greek word and a common Latin one. The missing Latin parallel in this case is easily provided by fer(i)culum, a word one of whose meanings (‘a dish on which food is served’) broadly overlaps with that we should expect for ϕέραικλον. If this was the case, the explanation with ἀριcτητήριον would either be based on an unattested meaning of this latter wordFootnote 23 or, perhaps, on a later confusion in the transmission of the lexicographical material, at a stage when the meaning of the Latin equivalent was not familiar any more. No source is mentioned in our excerpt for ϕέραικλον. If in this case too the source was poetic, Alcman, who uses not only αἶκλον but also singular compounds such as ϲυναικλία, would be the likeliest candidate, since our excerpt explicitly mentions the Laconians.Footnote 24 Other poets who use αἶκλον are Epicharmus (frs. 34 and 109 KA, quoted by Polemon, and Kaibel's conjecture ad 122.4 KA), and Euphorion fr. 24a.12 van Groningen (= 24.12 Lightfoot, Suppl. Hell. 413.12). If we consider the apparent preference for Hellenistic poets detectable in other portions of this excerpt, Euphorion would be another good candidate as its possible source here.
C. Greek ϕάροc/ϕᾶροc vs. Latin far.
I shall come back to the discussion of this passage, arguably the one richest in new information, at the end of this section.
D and E. Greek ῥάδιξ vs. Latin radix; Greek ῥυτή vs. Latin ruta.
The last two sections are both based on quotations from Nicander's Theriaca. In the first one, line 378 of the poem is quoted. Here too, as in sections B and C, no Latin word is explicitly mentioned. It is obvious, though, that the Greco-Roman parallelism must have been, once again, the starting point for our lexicographer. Only here, in the whole corpus of the Greek texts included in the TLG database, the Greek word ῥάδιξ means or is interpreted as meaning ῥίζα, ‘root’.Footnote 25 The sense of ῥάδιξ, a very rare word in Greek, in all its other occurrences, and in most ancient and all modern interpretations of the Nicander passage quoted in this excerpt, is that of ‘branch’ of a tree (or, more specifically, a ‘palm-tree branch’).Footnote 26 The interpretation offered in our manuscript was obviously prompted by the analogy with the Latin word radix. In this the source of our lexicographical excerpt may have been preceded (or followed) by Plin. HN 30.85 who, as argued by W. Morel in 1928, seems to have (mis-)understood this same passage of Nicander attributing to the Greek word the meaning of Latin radix.Footnote 27
The last section too is based on a quotation of Nicander, Ther. 523 (legendum ῥυτῆι γε μὲν εἴκελον ὀδμήν).Footnote 28 Nicander is the earliest among the very few attested Greek authors who used the word ῥυτή with the same meaning as the much more usual πήγανον (Latin Ruta). According to the scholia on this passage, this meaning of the word had been discussed by a certain Iolaus, in a work dealing with (the language of) the Peloponnesian cities.Footnote 29 Once again the Latin link is not explicitly mentioned in the preserved text, but is very clearly in the background: Latin ruta accounts for the equivalence between Greek ῥυτή (equivalent in form to the Latin word) and Greek πήγανον (its semantic equivalent). This time the same meaning is found also in later medical authors writing at an age when the influence of Latin was arguably stronger.Footnote 30
The equivalence is mentioned also by Varro, De lingua Latina 5.103: ruta quam nunc πήγανον appellant (scilicet Graeci), within a series of Latin horticultural words that Varro thought were of foreign origin. Since the series includes also radix (sic enim antiqui Graeci, quam nunc ῥάϕανον: an otherwise unattested meaning of the word, though the equivalence between Greek ῥάϕανον and Latin radix is attested by Dioscorides, Materia medica: RV 2.112.1 and 4.175.1), it is not unlikely that both Varro and our excerpt may derive from the same source, or, at least, that they may represent the same linguistic lore.
GREEK AND LATIN COMPARATIVE LEXICOGRAPHY
Even before dealing with our last section, the third one, some considerations on the general character of this excerpt may be in order. It is clear that the text of our manuscript represents a strongly abridged and altered version of a curious and learned lexicographical sequence. Its beginning looks incomplete, but also sections B, D and E in the preserved form do not mention at all the Latin words which must have inspired the selection of the Greek terms explained. All the words discussed in the excerpt are very rare, and at least one of them is not attested elsewhere. The lexicographical sequence quotes three poets, Antimachus, Callimachus and Nicander. The text of Nicander was easily available in the Middle Ages, though the interpretation provided by our source for the first passage (on ῥάδιξ) is not paralleled in the extant exegetical corpus. The Callimachean quotation derives from his lost Hecale (fr. 111 Hollis). This particular line is also known thanks to other lexicographical sources but, once again, our text offers a different interpretation of the text. The situation is similar, but even more complex, as we shall see, for the Antimachus quotation, where our only other source offers a shorter text (though this may be due to an accident of transmission) and an altogether different explanation. The quotation of ϕέραικλον is unique.
Our author's aim obviously was to explore the affinity between Greek and Latin, by showing that some rare Greek words have a very close Latin cognate. In some of the cases the aim is achieved by a more or less considerable distortion of the word's actual etymology (nothing unusual for an ancient grammarian, whose etymological criteria were very different from current ones) and, at least in one case (Nic. Ther. 378), of its generally accepted meaning in the quoted text. Even in this case, though, the misleading interpretation finds an interesting parallel in a Latin author, Pliny (HN 30.85, cf. above, p. 638).
The comparison between Latin and Greek had been an object of interest for Greek (and Latin) grammarians at least from the late Hellenistic period. It was a fairly widespread opinion, indeed, that Latin was in fact a Greek dialect and, more precisely, an Aeolic one. The series of scholars who dealt with the issue seems to start in the second century if, as has been argued, Cato the Elder had already expressed a similar view, and this opinion was certainly well established by Varro's time. Its exponents include one Hypsicrates, quoted by Varro in De lingua Latina (5.88, fr. 1 Funaioli) and by Cloatius Verus in the early Augustan age (ap. Gell. 16.12.5, fr. 2 Funaioli), perhaps the same person as the historian Hypsicrates of Amisus (FGrH 190), one Tyrannio (of Amisus too, if this was Tyrannio the elder; it it is possible, though, that this was his pupil, Diocles, also known as the younger Tyrannio),Footnote 31 Varro himself, Claudius Didymus and Seleucus.Footnote 32 The best-known grammarian by far who showed interest in this theory, however, is without doubt Philoxenus of Alexandria who, in the first century b.c.e., wrote a work Περὶ τῆϲ τῶν ̔Ρωμαίων διαλέκτου which has left considerable traces in the later lexicographical tradition.Footnote 33 In some of the fragments attributed to this work, Philoxenus employs the characteristic method for explaining word derivations that he had used in his famous work ̔Ρηματικόν. According to this, the various words are brought back to a monosyllabic root from which, thanks to additions, movements of letters and other changes, they are derived. This, however, does not happen for all the preserved fragments of his work On the Language of the Romans, at least in the form in which they have been transmitted to us.Footnote 34 In some of the fragments no Latin parallel form is actually quoted, but this is certainly due (as in our excerpt) to the selection of later transmission.Footnote 35 Philoxenus, who recurs extensively to poetic authorities,Footnote 36 also makes use of Greek dialectal peculiarities in his writings. As far as I can judge, his main sources seem to have been literary texts rather than collections of local idioms.Footnote 37 Our lexicographer is working in the same tradition and it might be argued that some of his material may go back to Philoxenus himself. Many other grammarians, however, are known to have dealt with similar exercises in comparative lexicography including, among others, Juba II, King of Mauretania, author of a work by the telling title Ὁμοιότητεc (Resemblances), in at least fifteen books, which exerted a certain influence on Plutarch.Footnote 38 Even Varro's writings remained available to Greek readers till as late as the sixth century, when Joannes Lydus quotes him twice, perhaps from his lost work De origine linguae Latinae, or from a lost book of De lingua Latina, exactly regarding the Aeolic origin of Latin (frr. 295–6 Funaioli= frr. 6 and 8 Traglia). The interest in the interaction and the comparison between the two languages, often with a noticeably more practical slant, Footnote 39 of course continues in later periods,Footnote 40 and occasionally emerges also well into Byzantine times.Footnote 41 The fact that our text condenses so much recondite material in so few lines, however, suggests that its ultimate source is not likely to be later than the early Imperial period.Footnote 42 If Philoxenus remains a possible candidate, it is mainly for the reason that his work continued to exert a much stronger influence on later lexicographical sources than any other similar work of which we know. The lack of clues unambiguously pointing to his linguistic theories and the presence of some features arguably at odds with his principles must, however, leave the question open.Footnote 43
THE THIRD EXCERPT AND ANTIMACHUS, FR. 154 MATTHEWS
1. ϕαροc and far. Antimachus, fr. 154: the text.
The excerpt's third section offers much interesting material. The starting point once again was the parallel explanation of a rare Greek word thanks to a common Latin one: in this case an equation between Greek ϕαροc (variously accented) and Latin far.Footnote 44 This interpretation of the Greek word, to the best of my knowledge, is otherwise unattested in ancient sources. Several modern scholars, however, have postulated it, either when interpreting other lexicographical sources on related terms,Footnote 45 or when discussing the occurrence of this word in Alcman's Louvre partheneion.Footnote 46 Its occurrence in our excerpt, of course, is no proof that the equation is correct.Footnote 47 It only demonstrates that it was already in circulation in ancient times.
The new quotation does not add anything to the text of Callimachus, Hecale fr. 111 Hollis, which was already known thanks to other lexicographical sources. It does, however, offer a new implicit explanation for the adjective ἄϕαρον (‘without spelt’) as derived from the word ϕάροc, here interpreted as referring to a cereal, while in the other preserved sources it is explained as related to verbs meaning ‘to plough’ (ϕαρῶ τὸ ἀροτριῶ, ἀπὸ τοῦ ἀρῶ) or ‘to produce’ (ϕέρω).Footnote 48 The latter etymology leads to the explanation of the adjective ἄϕαροϲ as τὴν μηδὲν ϕέρουϲαν, substantially equivalent to the meaning suggested in the new text.Footnote 49 The excerpt has interesting new readings for the text of Antimachus, fr. 154 Matthews (119 Wyss). Part of this line has already been known for almost two centuries thanks to its quotation in Herodian's περὶ μονήρουc λέξεωc, a work of the famous second-century grammarian first published in 1823.Footnote 50 Here Herodian examines the different meanings of the word ϕαροc (variously accented), and distinguishes the cases when it is used with a long alpha from those with the short vowel. In this second group, he distinguishes its occurrences as a feminine noun from those where the word is masculine. The instances of ϕάροc with the short vowel, according to Herodian, include the proper name of the hero Pharos (masculine), who gave his name to the Egyptian island of Pharos (feminine), where the famous lighthouse was built. Herodian knows that this short-vowel form was used also as a neuter noun, with the meanings of ‘cloak’ or ‘plough’, ‘as it happens in Alcman’.Footnote 51 The mention of Alcman is followed by the Antimachus quotation, which appears in the last edition of Antimachus by V.J. Matthews in the following form: ἀλλὰ καὶ παρ᾿ Ἀντιμάχωι ἀεὶ ϕάρεοc χατέουcιν ἔχων. οὕτωc ἐν τοῖϲ ἀντιγράϕοιc εὕρηται.Footnote 52
The new excerpt offers a different text, longer at its beginning and lacking the last word of Herodian's quotation. If we accept, as we must, the slight and obviously necessary correction of χατεύουcιν to χατέουcιν (now confirmed by the new source) the first three words as quoted by Herodian easily fit into part of a hexameter. If they are immediately followed by the last word transmitted in the Herodian quotation, ἔχων they may theoretically still belong in a sequence beginning in the first foot and continuing as far as the princeps element of the fifth one ( ˘̲ -2 | ̮ ̮ -3 | ̮ ̮ -4 ̮| ̮ -5< ̮ ̮ - ̮->). This would however entail a break of Hermann's bridge, extremely rare in most competent hexameter poets at all periods, and never attested elsewhere in Antimachus, and should be safely ruled out.Footnote 53 If the words do belong to a hexameter, and we want to avoid infringement of Hermann's bridge, the only viable alternative would be to locate the whole sequence at the end of the line ( ˘̲ -3 | ̮ ̮ -4 | ̮ ̮ -5 ̮| ̮ -6 < ̮->): in this case ἔχων should be awkwardly followed by a lost monosyllabic word. From a syntactic point of view, moreover, this entails understanding the word χατέουcιν as a dative plural form of the participle of the verb (‘to those who lack’), with the consequence that Herodian would be quoting an incomplete and unintelligible sequence of words. In fact, almost all editors variously correct ἔχων to a trisyllabic word, or consider it corrupt.Footnote 54 However, none of the suggested solutions is, to say the least, particularly compelling.Footnote 55
The new text, as quoted in our excerpt, suggests instead that the sequence ἀεὶ ϕάρεοc χατέουcιν corresponds to the second hemistich of a hexameter. It should be clear now that the last word in Herodian, ἔχων, did not actually belong to Antimachus, and that Herodian's text should rather be punctuated (and perhaps also slightly corrected) as follows: ἔχον οὕτωc ἐν ἀντιγράϕοιc εὕρηται, ‘the text has been found in this form in the exemplars’.Footnote 56 For ἔχον οὕτωc referring to the textual shape of a quotation we can compare Herodian's wording in this same work (3.2.936, 4–6 Lentz): καὶ ἐπίγραμμα ἔχον οὕτωc. It is not difficult to point out various textual annotations formulated almost with the same words. A passage in a medieval commentary on Aristotle's Sophistici elenchi, for example, offers a very similar wording: εὕρηται δέ μοι καὶ ϲχόλιον οὕτωc ἔχον.Footnote 57 The note in Herodian's text implies some degree of doubt regarding the textual shape of the preceding quotation,Footnote 58 and confirms that it was actually transmitted in this form in the available manuscripts. It is quite likely that the note may belong to a later scribe or reader of Herodian, as P. Maas, quoted in the apparatus of Wyss's edition, had suggested (still sticking, of course, to the previous punctuation).Footnote 59 To be fair, it must be noted that references to the ἀντίγραϕα do occur elsewhere in Herodian's text (at least as reconstructed in Lentz's edition),Footnote 60 but if this note goes back to the grammarian himself we should revert to the assumption that he had access only to a defective copy of Antimachus, while the source of our anonymous excerpt had access to a more correct copy. This is not, of course, impossible but is certainly unlikely, given the enormous quantity of literary texts with which Herodian was familiar, and it becomes even less plausible if we think that deducing the meaning of the word he was dealing with from such a defective text would have made Antimachus' evidence almost irrelevant in the context. It is certainly more probable that the corruption is due to the transmission of Herodian's text itself, and that the note is a later (medieval) interpolation mentioning the ἀντίγραϕα of the grammarian, not those of the poet.
If our lexicographic excerpt helps to settle an old problem regarding the end of the quotation, the three words that now precede the text of Antimachus raise new problems. Poetic quotations in lexicographical tradition are subject to a huge range of voluntary and involuntary manipulations. We can have no guarantee, therefore, that the new text actually corresponds to the original quotation. It is, however, only with this (ultimately unverifiable) proviso that I shall work on the assumption that no major alteration has taken place. The new text offers a quasi-hexametric sequence: αἰεὶ ἐν πενταπόλει ἀεὶ ϕάρεοc χατέουcι. A first difficulty, however, arises with the repetition of the adverb at the beginning of both hemistichs. Not only does the repetition make poor sense but, more importantly, the first αἰεὶ does not metrically fit within the sequence, as crasis with the following preposition is, of course, out of question. A possible, fairly economical solution would be to imagine that the first word was the result of a wrong anticipation of the following ἀεί or, rather, a misplaced orthographic correction, which displaced the original text of its first syllable.Footnote 61 We may toy with the idea that it was <οἱ δ᾿> or <ἀλλ᾿> but many other possibilities are available. The word πενταπόλει is quoted with an ending uncommon in poetic epic diction, where a form in –πόληϊ would normally be expected.Footnote 62 There are, however, several instances of πόλει already in archaic hexameters (from as early as the Homeric poems) and elegiac poetry. The transmitted text, moreover, entails a hiatus between the two hemistichs, as well as requiring the minimal correction of the second ἀει into α < ἰ > εὶ.Footnote 63 Hiatus at the caesura is not rare in Antimachus, but with a single possible exception (fr. 31.1 Matthews, where the presence of an elided word is a viable alternative)Footnote 64 the longum which precedes the hiatus is always followed by a biceps, realized by two short syllables.Footnote 65 This kind of hiatus, between two long syllables, however, is well attested in Hellenistic poets such as Aratus, Theocritus and Apollonius Rhodius, whose versification is certainly not less refined than that of Antimachus.Footnote 66 Given the relatively small number of complete hexameters of Antimachus actually preserved, the absence of this feature may well be due to chance. All in all, the obstacles for such a reconstruction do not seem particularly difficult to overcome. A conceivable alternative might be that, after the false start with the first αἰεί, the excerpt actually introduced the title of a work of Antimachus, but this is far less likely, as the title of none of his works is transmitted in a form even remotely similar to πεντάπολιc.
According to the source of the new excerpt, the meaning of the word ϕάροc in this Antimachus passage was ‘ground grain’, ‘spelt’ (ὁ κεκομμένοc [superscript ἠλεcμένοc] ϲῖτοc ὃϲ λέγεται χόνδροc). In Herodian, on the other hand, the quotation is introduced, after a reference to Alcman, as an example of neuter forms of the word with a short alpha, with the meanings ‘cloak’ and ‘plough’. Many occurrences of ϕάροc= ‘cloak’ are attested, and it is very unlikely, to say the least, that Herodian would have resorted to these two difficult and ambiguous passages if his intention were to offer examples of this meaning (even more so since Herodian had already adduced two Sophoclean occurrences of this meaning of the word a few lines earlier, in Lentz 3.942, 5–8).Footnote 67 It seems inescapable that Alcman and Antimachus are both quoted as examples of the meaning ‘plough’, a meaning partly paralleled elsewhere only in Hesych. s.v. βούϕαρον, where the word is actually explained as ‘the act of ploughing’ (ἡ ἄροcιc).
Forty years after Herodian's περὶ μονήρουc λέξεωc was first made available in print, a papyrus with the text of Alcman's poem from which Herodian quoted this word was actually published: it occurs at line 61 of the famous Louvre partheneion first published by É. Egger in 1863.Footnote 68 In the text itself the word is provided with a circumflex accent (ϕᾶροc). The metrical structure of the line, though, allows for both scansions. A marginal note under the column where the word occurs provides some additional information on the ancient discussions on the text and includes a repetition of the word ϕαροc (this time with no accent) followed by the name Sosiphanes and by the explanation ἄροτρον (‘plough’). We do not know whether Sosiphanes was a poet (possibly the playwright belonging to the Alexandrian Pleiad) or, as it has often been assumed, an otherwise unattested grammarian who understood the word in this way. It is clear from the preserved text of Alcman that the interpretation of the word as ‘plough’ offered by Sosiphanes (?) and Herodian was by no means unambiguously secured by the context. The same obviously must have been true for the passage in Antimachus, where the context must have allowed both the interpretation of ϕάροc as ‘plough’ (as in Herodian), and the one as ‘spelt’ (as in the new excerpt). Ancient interpreters reacted in a similar way also to the use of the iunctura ἄϕαρον ϕαρόωcι in Callimachus' Hecale (fr. 111 Hollis). In this case too, the previously known sources explained the two words as related to ‘plough’, and/or to the verb ϕέρω, while the new lexicographical excerpt apparently connects them to his equation between ϕαροc and far. All we can say is that Antimachus' line must have been compatible with a situation in which a group of people was described as chronically lacking either agricultural subsistence goods or the possibility of working the fields (which, ultimately, would produce the same result).
2. Antimachus, fragment 154: a plausible context?
If the text as reconstructed above is roughly correct, the mention of a Pentapolis seems the most promising starting point for attempting to reconstruct a possible context for this line. If the word does indeed belong to Antimachus, chronological reasons would militate against the identification with what is probably the most famous Pentapolis in the ancient world, the Libyan one. In the Hellenic world a possibly chronologically compatible Pentapolis is known in LesbosFootnote 69 but, if we look for a connection with themes treated in other remains of Antimachus' poetry, the best candidate is probably going to be the Dorian Pentapolis, centred around the sanctuary of Apollo at the Triopion, near Cnidus. It included Cnidus, Cos and the three Rhodian cities of Ialysus, Lindus and Camirus, and had ceased to be a Hexapolis at the time when Halicarnassus was expelled from the group (Hdt. 1.144). It is very likely that Antimachus told the story of the arrival of the Thessalian colonizers at Cnidus in the second book of his Lyde. Fr. 85 Matthews (= 72 Wyss), known to us thanks to a quotation in Stephanus of Byzantium among a series of poetic passages where adjectives derived from Δώτιον are used (the ultimate source is, once again, Herodian), is a complete pentameter: ϕεύγονταc γαίηc ἔκτοθι Δωτιάδοc. Already in the eighteenth century Schellenberg had connected these words to the story of Triopas' migration from Thessaly.Footnote 70 Diodorus Siculus (5.61) tells that Triopas fled to Cnidus after the sacrilegious destruction of a wood sacred to Demeter: δι᾿ ἣν αἰτίαν ὑπὸ τῶν ἐγχωρίων μιϲηθέντα ϕυγεῖν ἐκ Θετταλίαϲ, καὶ καταπλεῦϲαι μετὰ τῶν ϲυμπλευϲάντων λαῶν εἰϲ τὴν Κνιδίαν, ἐν ἧι κτίϲαι τὸ καλούμενον ἀπ' αὐτοῦ Τριόπιον. This story was narrated, with different variants, by Callimachus in his Hymn to Demeter, and by Apollonius Rhodius, in his lost Foundation of Rhodes (fr. 10 Powell). It does not seem unlikely that Antimachus, having told the story of the foundation of Triopion in Lyde Book 2, may have added some details on the subsequent story of the new colonies. Two authors of uncertain date, Dieuchidas in his Megarica (FGrH 485 F 7 = fr. 10 Piccirilli), usually dated to the fourth century,Footnote 71 and an enigmatic Aristides (FGrH 444 F 4 Jacoby) mention a story of curses, which involved the descendants of the founder hero. Dieuchidas alludes to a quarrel between Phorbas and Periergus, sons of Triopas. Periergus cursed his brother in the islands between Syme and Cnidus, later known as Araeae, the ‘Curse Islands’:
Διευχίδαϲ δ' ἐν τοῖϲ Μεγαρικοῖϲ <…> τὰϲ καλουμέναϲ, ϕηϲὶν, Ἀραιὰϲ (μεταξὺ δε τῆϲ Κνίδου καὶ τῆϲ Ϲύμηϲ εἰϲί) γενομένηϲ διαϕορᾶϲ τοῖϲ ϲυνεξορμήϲαϲι τῷ Τριόπᾳ μετὰ τὸν ἐκείνου θάνατον, καὶ τῶν μὲν εἰϲ τὸ Δώτιον ἀναχωρηϲάντων <…>, οἱ μὲν μετὰ Φόρβαντοϲ μείναντεϲ εἰϲ Ἰηλυϲὸν ἦλθον, οἱ δὲ μετὰ Περιέργου τὴν Καμειρίδα κατέϲχον. Τότε λέγεται καταράϲαϲθαι τὸν Περίεργον τῷ Φόρβαντι, καὶ διὰ τοῦτο τὰϲ νήϲουϲ Ἀραιὰϲ κληθῆναι (…).Footnote 72
Aristides' fragment interestingly, if somewhat puzzlingly, informs us that in these islands ‘the Dorians cursed the Pentapolites’: Ἀραὶ,Footnote 73 Ἰωνίαϲ νῆϲοι τρεῖϲ, οὕτω λεγόμεναι διὰ τὰϲ ἀρὰϲ, ἃϲ Δωριεῖϲ ἐποιήϲαντο πρὸϲ τοὺϲ Πενταπολίταϲ, ὡϲ Ἀριϲτείδηϲ. Ἄλλοι δ' ἄλλωϲ.Footnote 74
The content of the curses is not made clear in these sources. If the Antimachus passage refers to the same Pentapolis, we might suppose that one of their consequences was its inhabitants' chronic lack of ϕάροc, i.e. a famine. A context such as this would have allowed readers to understand the word as referring either to the cereal itself (as the source of our excerpt did), or metonymically, to the tool used for its cultivation (as Herodian seems to have done). We may well attribute this to an early case of learned amphibology, going back straight to the poet/grammarian himself, but the similar case in Alcman's partheneion should make us pause along this path.
CONCLUSION
The whole body of the marginalia added to the Paris manuscript of Strabo after its medieval restoration would deserve a publication of its own. They do not form a coherent corpus and their origin is likewise very diverse.Footnote 75 The two main grammatical sources identified so far seem to be a larger version of Herodian,Footnote 76 and one of the various versions of the work περὶ διαϕόρων λέξεων which went, among others, under the name of Ptolemy.Footnote 77 As Diller remarked, some of the notes overlap with materials found in ‘the scholia on Clement of Alexandria, Protrepticus and Paedagogus, in codex [Par.] gr. 451, where part of them are in the hand of Arethas of Caesarea’.
For other excerpts the source cannot be identified with any degree of certainty.Footnote 78 This is the case of our text. As we have seen above, it should be examined within the frame of Greco-Latin comparative lexicography, a research and teaching activity that, under different circumstances and following different paradigms, has been pursued for millennia. Our note is remarkable for its particularly concentrated focus on rare terms found in difficult authors, mainly (though perhaps not only) Hellenistic poets. This neglected excerpt contributes to a significant degree to the establishment of the text, and possibly to the interpretation, of a fragment of Antimachus. It is our only ancient source that explicitly presupposes a common explanation for Greek ϕάροc (and related words) and Latin far, and for Greek ῥάδιξ and Latin radix. And it is our only source for the Laconian word ϕέραικλον.
Even more interesting, however, is the peculiar slant of the comparative approach that this text reveals. While most of the preserved grammatical sources dealing with Greco-Latin comparisons focus on the original Greekness of Latin, the assumption underlying our excerpt seems to be different. All the selected Greek words (ἀμμίη [?], ϕέραικλον, ϕάροc, ῥάδιξ, ῥυτή) are very unusual or, at least, their meaning in the selected passages is very rare, if not unique. On the other hand, the Latin terms of comparison are perfectly usual ones: amita, fer(i)culum, far, radix, ruta. In this case, it is not the Greek that provides the key for the explanation of the Latin, but things seem to work the other way round.Footnote 79 What might the purpose of such pursuit have been? Much has been written on the possible background of ancient theories on the Greekness of Latin, looking, in some cases, for wide-ranging political and cultural implications.Footnote 80 Of course, the fact itself that the identity and even the date of our source are unknown makes it very difficult for us to explore the intellectual background of this text. Its author, however, displays thorough familiarity not only with both languages but with a syllabus of authors that is particularly advanced and recherché, and of course he expects a similar level of knowledge in his audience. This is most probably the work of a scholar active either in the late Republican period or in the early Imperial one. It provides an interesting glimpse into the possible implications of scholarly research and advanced teaching activity on Hellenistic poetry and other linguistically arduous texts during this period. The method followed by our source involves addressing them by presupposing, suggesting and even forging a common intercultural ground between the Greek ‘classical’ texts and Latin speaking audiences. This would potentially have been highly rewarding for students and scholars belonging to both speaking communities. A good vantage point, if we want to imagine one of the possible backgrounds of such a method, might be to look at the teaching activity of the likes of Statius the Elder, a grammarian born in the formerly Greek and later Lucanian town of Velia/ Οὐελία /Hyele Footnote 81 and active in mainly Greek-speaking Naples, many of whose pupils, ‘aristocrats … from all over Southern Italy’ ascended to the highest administrative ranks of the Empire.Footnote 82 His teaching syllabus, in the account provided by his son in Silvae 5.3, included a wide range of Greek authors writing in different styles and dialects,Footnote 83 such as Aeolian Sappho, Laconian Alcman, Boeotian Corinna, Syracusan Epicharmus and Sophron, and the obscure Lycophron (as well as Pindar, Stesichorus, Ibycus, Callimachus and, of course, Hesiod and Homer). As McNelis has shown, this teaching list should not be considered peculiar to the milieu of western Greek elites. Statius' activity, however, may be useful for us to envisage a possible scenario for the use of the kind of scholarship exemplified in the Strabo excerpt, where the study and interpretation of the most difficult portion of the Greek literary legacy is pursued through what looks to have been a large-scale comparison to its kindred Latin language. The ultimate origin of our text is still an unsolved problem, as is its route to the margin of a Strabo manuscript in the fourteenth century. This notwithstanding, its tenor and content offer a most interesting example of ‘intercultural’ scholarship at the high end of the teaching spectrum in the Greco-Roman world.
