Roussou ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 2017 Stephanie Roussou The Reception of Herodian in the Byzantine Period: The Case of Theognostus Stephanie Roussou HEOGNOSTUS was a Byzantine grammarian of the ninth century, from whom a work on correct Greek spelling has come down to us. This work is known under the Greek titles Περὶ ὀρθογραφίας (“On Orthography”) or Κανόνες (“Rules”) and under the Latin title Canones. Over the centuries various Greek vowel sounds had fallen together in pronunciation, while the traditional spelling remained un- changed. Theognostus’ work aimed to help people of his time to spell correctly, a task which had become very difficult. The work is important for understanding Byzantine scholarship both because of the rare insight it provides into ninth-century linguistic thought and because of its influence on the etymo- logical lexica composed by later Byzantine scholars. At the same time the work is vital for our understanding of ancient scholarship because it preserves much ancient scholarly ma- terial that has not survived independently. In the preface to the Canones Theognostus (p.69.4–5)1 men- tions that he has gathered the rules from Herodian’s Περὶ καθολικῆς προσῳδίας (“On Prosody in General”), the most in- fluential but now lost ancient treatise on Greek accentuation.2 1 Ed. K. Alpers, Theognostos Περὶ ὀρθογραφίας: Überlieferung, Quellen und Text der Kanones 1–84 (Hamburg 1964). Subsequent citations will be to the edition of J. A. Cramer, Anecd.Ox. II (Oxford 1835: cited first by rule number and then by page and line of Cramer). 2 οὓς (sc. κανόνας) δὲ ἐκ τῆς πολυΰλου βίβλου τῆς καθόλου Ἡρωδιανοῦ ἀναλεξάµενος (“having gathered the rules from Herodian’s copious work T STEPHANIE ROUSSOU 483 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 Theognostus is writing a work on orthography and yet he recognizes the Περὶ καθολικῆς προσῳδίας as a fundamental source. That he derived material from Herodian is well known. What has been less clear is quite how Theognostus reuses a work on accentuation for his own work on orthography, and what prompted him to use the Περὶ καθολικῆς προσῳδίας as a basis for his own work in the first place. This article will examine the transformation of Herodian’s accentuation rules into orthographic ones. We begin with an introduction to Herodian’s treatise. Section 2 will then discuss Pseudo- Arcadius’ and John Philoponus’ perception of the Περὶ καθολι- κῆς προσῳδίας, and section 3 will discuss the relationship between Herodian and Theognostus, and consider whether Theognostus consulted Herodian’s work in its original form or in an epitome. Section 4 will compare the arrangement of Herodian’s treatise with that of Theognostus’ Canones, and finally section 5 will deal with Theognostus’ transformation of Herodian’s accentuation rules into orthographic ones. 1. Herodian’s Περὶ καθολικῆς προσῳδίας Περὶ καθολικῆς προσῳδίας was Herodian’s most important work and at the same time the most important ancient work on Greek accentuation. It does not survive in its original form, but in epitomes and fragments.3 The two fully-preserved epitomes are one attributed to Theodosius or Arcadius (conventionally called Pseudo-Arcadius’ epitome) and another by John Philo- ponus of Alexandria,4 while the two fragmentary epitomes are ___ Περὶ καθολικῆς προσῳδίας”). On the value of Theognostus’ Περὶ ὀρθογρα- φίας as a source of evidence for the Περὶ καθολικῆς προσῳδίας see A. Lentz, Herodiani technici reliquiae I (Leipzig 1867) clxxx–clxxxiv; C. Galland, De Arcadii qui fertur libro de accentibus (Strasbourg 1882) 33–35; K. Krumba- cher, Geschichte der byzantinischen Litteratur2 (Munich 1897) 585–586; C. Wen- del, “Theognostos,” RE 5A (1934) 1985–1987; Alpers, Theognostos 27–28. 3 See E. Dickey, Ancient Greek Scholarship (Oxford/New York 2007) 75–77. 4 See A. Wouters, “P. Ant. 2.67: A Compendium of Herodian’s Περὶ Κα- θολικῆς Προσῳδίας, Book V,” in P. Naster et al. (eds.), Miscellanea in honorem Josephi Vergote (Leuven 1975–1976) 601–613, esp. 602; Dickey, Ancient Greek 484 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 a fourth-century parchment codex from Antinoöpolis and a palimpsest dated to the first half of the tenth century.5 The two fully-preserved epitomes differ significantly from each other, which makes the reconstruction of Herodian’s work difficult. Philoponus prescribes rules for the accentuation of the oblique cases once the accent of the base or dictionary form of a word is known, whereas Pseudo-Arcadius’ epitome deals mainly with which syllable of a base form will have the accent. Pseudo- Arcadius devotes fourteen books to the accentuation of nouns and adjectives in the nominative, organising his rules according to the terminations of words, but only one book to the accentu- ation of the oblique cases. From the surviving epitomes and fragments of Herodian’s work we learn that in its original form this work contained not only long lists of words exemplifying accentuation rules, but also word meanings (especially for rare and unusual words) and passages illustrating the uses of listed words. 2. Pseudo-Arcadius’ and John Philoponus’ perception of the Περὶ καθολικῆς προσῳδίας Herodian’s treatise, though dealing mainly with accentua- tion, was perceived and used by late antique and Byzantine scholars with various aims and from various perspectives. Pseudo-Arcadius in the prologue to his epitome comments that Herodian’s original work could be used not only to learn cor- rect word accents, but also to learn the meanings of words and to see passages illustrating their uses (2.15–17):6 κἀκεῖθεν αὐτὰ λήψεται ὁ µὴ µόνον τὴν τῶν προσῳδιῶν ὀρθότη- τα ζητῶν, ἀλλὰ καὶ περὶ τὰς δυνάµεις καὶ χρήσεις φιλοµαθῶν. ___ Scholarship 75–77. 5 A. Wouters, The Grammatical Papyri from Graeco-Roman Egypt (Brussels 1979) 216–224; H. Hunger “Palimpsest-Fragmente aus Herodians Καθολι- κὴ Προσῳδία, Buch 5–7,” JÖBG 16 (1967) 1–33. 6 I quote from my edition, Pseudo-Arcadius’ Epitome of Herodian’s De prosodia catholica (Oxford forthcoming), but the page and line numbers are those of M. Schmidt, Ἐπιτοµὴ τῆς καθολικῆς προσῳδίας Ἡρωδιανοῦ (Jena 1860). All translations of passages of Greek are mine. STEPHANIE ROUSSOU 485 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 and the person who not only seeks correctness in prosody but is also fond of learning the meanings of words and examples of usage [i.e. passages cited] will find out these from there [i.e. from Herodian’s original]. Both epitomators also point out that accenting a word correctly is a complex process, requiring knowledge of many features of the word in question. Specifically, in John Philo- ponus we read (1.1–2):7 τὸ ἑκάστῃ λέξει τὴν δέουσαν προσῳδίαν τιθέναι συµπέρασµα σχεδὸν πάσης τῆς γραµµατικῆς τυγχάνει µεθόδου. to put the correct accent on each word is the end result of almost the whole of grammatical enquiry. and in Pseudo-Arcadius (1.9–12): τὸ γὰρ ἅµα γένος, εἶδος, σχῆµα, κατάληξιν, παράληξιν, ἀρχήν, χρόνον, στοιχεῖον, πάθος παρατηρεῖν, καὶ ὅσα ἄλλα, ἢ τὰ πλεῖστα τούτων, χαλεπὸν καὶ δυσέφικτον. for to pay attention simultaneously to the gender, the deriva- tional status, the compositional status, the ending, the penulti- mate syllable, the beginning, the quantities, the letters, and the transformations of form, and whatever else, or most of these, is difficult and hard to achieve. Both epitomators thus recognise that the Περὶ καθολικῆς προσῳδίας contained multiple kinds of knowledge, and Pseudo-Arcadius in particular recognises that this can be exploited for multiple purposes beyond accentuation. 3. The relationship between Herodian and Theognostus Given that Herodian’s original work has not come down to us, that we do not know until exactly when the original work was still available, and that epitomes were created as early as the fourth century (i.e. two centuries after the work was com- posed), one might reasonably wonder whether Theognostus consulted Herodian’s original work or an epitome. Pseudo- Arcadius says in his preface that he divided the large rules of 7 Ed. G. A. Xenis, Iohannes Alexandrinus Praecepta Tonica (Göttingen 2015). 486 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 Herodian into smaller rules, to make them easier to understand (1.6–12): ἐπεὶ γὰρ τὸ πολύυλον τῶν ὁρισµῶν ἐν πολλοῖς κανόσιν ἀθρόως κείµενον δύσληπτον ἦν, καταδιῃρέθη τοῦτο, ἵν’ εὔληπτα γένη- ται διαιρεθέντα ⟨τὰ⟩ ὑφ’ ἓν κείµενα τῷ Ἡρωδιανῷ since the mass of conditions that appear all together in many rules was hard to comprehend, this has been divided, so that those that are brought under one rule by Herodian would be- come easy to comprehend, being divided There are places where Theognostus has a single rule cor- responding to two or more rules in Pseudo-Arcadius, but also places where more than one rule of Theognostus corresponds to a single rule in Pseudo-Arcadius.8 There is no clear answer to the question whether Theognostus consulted Herodian’s original work or an epitome, but the question is in any case secondary to our present purposes. The focus of our discussion will not be the precise form in which Herodian’s material was available to Theognostus, but how he reuses and transforms the material in order to turn a treatise on accentuation into a treatise on orthography. 4. Comparison of the arrangement of Herodian’s Περὶ καθολικῆς προσῳδίας with Theognostus’ Canones The spellings of masculine, feminine, and neuter nominal terminations in the nominative in Theognostus are treated in the same order as in Herodian’s Περὶ καθολικῆς προσῳδίας.9 8 Examples of places where Pseudo-Arcadius has more rules than Theo- gnostus: Ps.-Arcad. 93.1–8+93.9–13 ~ Theogn. 416 = p.75.5–8; Ps.-Arcad. 13.8–12+14.3–8 ~ Theogn. 159 = p.29.19–21. Examples of places where Theognostus has more rules than Pseudo-Arcadius: Ps.-Arcad. 110.10–11 ~ Theogn. 592+593+594+595 = p.100.20–21+100.22–23+100.24–26+ 100.27–30; Ps.-Arcad. 9.3–6 ~ Theogn. 172+173 = p.31.6–8+31.9–11; Ps.- Arcad. 9.17–10.2 ~ Theogn. 175+176 = p.31.24–32+ 31.33–32.3. Some of these passages are discussed by Galland, De Arcadii qui fertur libro de accentibus 33–34. 9 The order of the books in the original Περὶ καθολικῆς προσῳδίας can be reconstructed on the basis of the order of the books in Pseudo-Arcadius’ epitome. The similarities in the arrangement of the material between that STEPHANIE ROUSSOU 487 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 Herodian dealt with the terminations of the nominals10 in Books 1–13 following the same order as in the Τέχνη γραµ- µατική attributed to Dionysius Thrax (GG I.1 15.1–16.1),11 with some minor variations. The variations from the Τέχνη γραµµατική found in the Περὶ καθολικῆς προσῳδίας are also found in Theognostus. Thus, the similarities between Herodian and Theognostus in the arrangement of nominal terminations in the nominative, as opposed to the arrangement in the Τέχνη γραµµατική, together with the fact that Theognostus indicates in his preface that he employed the Περὶ καθολικῆς προσῳδίας as a source, suggest that Theognostus followed Herodian’s ar- rangement of nominal terminations in the nominative as a model. In the Τέχνη γραµµατική the nominal terminations are arranged as follows: τελικὰ ἀρσενικῶν ὀνοµάτων †ἀνεπεκτάτων κατ’ εὐθεῖαν καὶ ἑνικὴν πτῶσιν στοιχεῖά ἐστι πέντε· ν ξ ρ ς ψ, οἷον Δίων Φοῖνιξ Νέστωρ Πάρις Πέλοψ. θηλυκῶν δὲ ὀκτώ· α η ω ν ξ ρ ς ψ, οἷον Μοῦσα Ἑλένη Κλειώ χελιδών ἕλιξ µήτηρ Θέτις λαῖλαψ. the terminations of the masculine nouns that are parisyllabic12 in the nominative singular are five: ν ξ ρ ς ψ, for example Δίων Φοῖνιξ Νέστωρ Πάρις Πέλοψ. And [the terminations] of the feminine nouns are eight: α η ω ν ξ ρ ς ψ, for example Μοῦσα Ἑλένη Κλειώ χελιδών ἕλιξ µήτηρ Θέτις λαῖλαψ. The first ten books in the Περὶ καθολικῆς προσῳδίας deal ___ epitome and the two fragmentary epitomes suggest that Pseudo-Arcadius retained the order of the books as this appeared in Herodian’s original work. For further information see section 2.2.1.4 of my Pseudo-Arcadius’ Epitome. 10 The modern term “nominal” (or “nominal form”) is a cover term for both nouns and adjectives. 11 Ed. G. Uhlig, Dionysii Thracis Ars Grammatica (Leipzig 1883). 12 The term ‘parisyllabic’ refers to forms of a given paradigm retaining the same number of syllables in all inflected forms, while ‘imparisyllabic’ refers to forms of a given paradigm acquiring an extra syllable in any of their inflected forms (oblique cases): e.g. ὁ Δρῆς, τοῦ Δρῆ is parisyllabic because it maintains the same number of syllables through the paradigm, while ὁ Κρής, τοῦ Κρητός acquires a syllable in the oblique cases. 488 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 with masculine and feminine nominals ending in -ν, -ξ, -ρ, -ς, -ψ in the nominative, Book 11 with feminine nominals ending in -α, and Book 12 with feminine nominals ending in -η and -ω. The only difference in these books is that in the Περὶ καθολικῆς προσῳδίας feminines ending in a consonant have already been dealt with in Books 1–10 together with mascu- lines, and so in Books 11 and 12 only feminines ending in vowels are treated, while in the Τέχνη feminines with vowel terminations precede those with consonant terminations. Theognostus arranges the nominative terminations in the same order as in Herodian, even in the case where Herodian’s ar- rangement differs from that of the Τέχνη. Specifically, rules 143–429 = pp.26.24–77.2 deal with the masculine and fem- inine nominals ending in -ν, -ξ, -ρ, and -ς in the nominative (but words ending in -ψ are not treated immediately after the nominals in -ς but later on, in rules 573–576 = pp.97.29–98.9) and rules 580–718 = pp.98.24–118.22 deal with feminine nominals ending in -α, -η, and -ω. The treatment of the neuter nouns in the nominative in Book 13 again follows the same arrangement as we find in the Τέχνη γραµµατική (GG I.1 16.2–3 οὐδετέρων δὲ ἕξ· α ι ν ρ ς υ, οἷον ἅρµα µέλι δένδρον ὕδωρ δέπας δόρυ, “and [the terminations] of neuter nouns are six: α ι ν ρ ς υ, for example ἅρµα µέλι δένδρον ὕδωρ δέπας δόρυ”). There is only a minor variation in the case of the termination -υ: this is treated not at the end but third, probably so that Herodian can deal with all the vowel terminations together at the beginning.13 Theognostus here again uses the same arrangement as Herodian, not the slightly different arrangement found in the Τέχνη. Specifically, in two cases where we find spelling rules for neuters one after another, the terminations are treated in the order -α, -ι, -υ, -ν, -ρ, -ς,14 13 Cf. P. Egenolff, Die orthoepischen Stücke der byzantinischen Litteratur (Leipzig 1887) 8. 14 In the first case rules 431–434 = pp.77.12–78.20 and 436 = p.78.28– 31 deal with -α, rule 437 = pp.78.32–79.9 with -ι, rule 438 = p.79.10–15 with -υ, rule 440 = p.79.23–25 with -ν, rules 441–447 = pp.79.26–80.13 STEPHANIE ROUSSOU 489 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 although sometimes the treatment of these terminations is in- terrupted by other spelling rules. 5. Theognostus’ transformation of Herodian’s accentuation rules into orthographic ones Lentz pointed out that Theognostus inverted Herodian’s material in order to create orthographic rules from prosodic ones, but did not expand on this point.15 Theognostus derives material from Herodian’s work, but since his goal is to prescribe rules on orthography, facts about accentuation are presented as conditions for orthographic rules, whereas for Herodian orthography provides conditions for accentuation rules. In what follows we discuss pairs of passages from Pseudo- Arcadius’ epitome and Theognostus, in order to shed light on Thegnostus’ method. Sub-sections (a)–(c) point out differences that we would expect to follow from the fact that Pseudo- Arcadius is concerned with accentuation and Theognostus with spelling. Sub-sections (d) and (e) show how Theognostus retains some material which is actually superfluous for his purposes. This material is particularly helpful to us in revealing that The- ognostus worked from a treatise on accentuation. (a) Treatment of terminations together where they sounded the same in Theognostus’ day In Pseudo-Arcadius’ epitome (58.1–4) we find an accentua- tion rule about proparoxytone words ending in -αικος and -οικος that are not adjectives: (1a) τὰ εἰς κος παραληγόµενα αι ἢ οι µὴ ἐπιθετικὰ προπαροξύ- νεται· Πάταικος, Φάλαικος, Μίθαικος, σόλοικος, µόνοικος (τὸ κατὰ συναλοιφήν). τὸ Εὐβοϊκός ὀξύνεται ἐπιθετικὸν ὄν. [Words] ending in κος which have αι or οι in their penultimate syllable and are not adjectives, are proparoxytone, e.g. Πάται- ___ and 449 = p.80.19–21 with -ρ, and rule 448 = p.80.14–18 with -ς; in the second case rule 729 = p.120.13–26 deals with -α, 730 = p.120.27–30 with -ι, 731 = p.120.31–34 with -υ, rules 732–792 = pp.121.1–131.24 with -ν, and rule 793 = p.131.25–30 with -ς. 15 Lentz, Herodiani technici reliquiae clxxx. 490 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 κος, Φάλαικος, Μίθαικος, σόλοικος, µόνοικος (with contrac- tion). Εὐβοϊκός is oxytone because it is an adjective. Theognostus is interested in prescribing orthographic rules for terminations that sound identical in his time.16 Thus, he treats the termination -αικος together with the identical-sounding -εκος (326 = p.60.24–27). His rule contains no information about the accentuation of the words cited as examples, because there is no way of making use of the accent here to help with the spelling: (1b) τὰ διὰ τοῦ αικος ὑπὲρ δύο συλλαβάς, σπάνια· τὰ δὲ ὄντα, διὰ τῆς αι διφθόγγου γράφονται· οἷον, Φάλαικος, Πάταικος, Μίθαικος (ὄνοµα κύριον)· µόνον τὸ ἀλώπεκος σεσηµείωται διὰ τοῦ ε ψιλοῦ.17 [Words] ending in αικος which have more than two syllables are rare; those which exist are spelled with the diphthong αι, e.g. Φάλαικος, Πάταικος, Μίθαικος (a proper name); only ἀλώπεκος which is spelled with ε is an exception. Further examples of this type of process can be seen in pas- sages (2b) and (3b). In (2a) Pseudo-Arcadius (59.5–10) gives a rule on words which have only one gender (µονογενῆ), end in -λος preceded by either ι or υ, do not begin with χ, and are recessively accented. At the end of this accentuation rule, the phrase τὸ δὲ ψιλός τριγενές explains why ψιλός does not display the recessive accentuation prescribed for the µονογενῆ: (2a) τὰ εἰς λος µονογενῆ [ὑπὲρ δύο συλλαβὰς] µὴ ἀπὸ τοῦ χ ἀρχόµενα ἔχοντα δίχρονον ἐκτεταµένον, εἰ µή τις εἴη διαστολή, βαρύνεται· Ἶλος (τὸ κύριον, ἰλλός δὲ ὁ διεστραµµένος), πῖλος, µύλος, στῦλος, γρῦλος. τὸ δὲ χυλός καὶ χιλός (ἡ τροφή) ὀξύνε- ται ἀπὸ τοῦ χ ἀρχόµενα. τὸ δὲ ψιλός τριγενές. 16 On the principle of ἀντίστοιχα (i.e. vowels which sounded identically: αι ⁓ ε, ει ⁓ ι ⁓ η, οι ⁓ υ, ο ⁓ ω) in Theognostus and in the Byzantine period in general see Krumbacher, Geschichte 585. On the ἀντίστοιχα in the Byzantine period in general see further E. Follieri, “ΑΝΤΙΣΤΟΙΧΑ,” Δίπτυχα 4 (1986/7) 217–228, esp. 219–220. 17 In the quoted passages from Theognostus’ Canones I have made some minor changes to the punctuation, compared with Cramer’s edition. STEPHANIE ROUSSOU 491 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 [Words] [with more than two syllables] having only one gender and ending in λος, if they do not begin with χ and they have a long anceps vowel, are recessive if there is no distinction [be- tween different meanings], e.g. Ἶλος (the proper name, but ἰλλός “distorted”), πῖλος, µύλος, στῦλος, γρῦλος. But χυλός and χιλός (“food”) are oxytone because they begin with χ. And ψιλός has separate forms for each of the three genders. In (2b) Theognostus composes one rule on disyllabic words in -ιλος, -ειλος, and -ηλος (328 = pp.60.33–61.9), which sounded the same in his day, but -υλος is kept distinct, probably because it was just about still distinct in educated speech of Theo- gnostus’ day.18 Theognostus announces at the beginning of his rule that the words treated are the µονογενῆ, “words of one gender,” both proper names and common nouns. There are also adjectives ending in -ιλος, and indeed Theognostus goes on to include the word ψιλός as an example, without any ac- knowledgement of the fact that this is an adjective rather than a “word of one gender.” Since words in -ιλος can be proper names, common nouns, or adjectives, the type of nominal makes little sense as a condition for the orthographic rule. We may conclude that the mention of various nominal types at the beginning of the rule (µονογενῆ, εἴτε κύρια, εἴτε προσηγορικά) is left over from an accentuation rule such as the one in (2a).19 (2b) τὰ διὰ τοῦ ιλος δισύλλαβα µονογενῆ, εἴτε κύρια, εἴτε προσηγορικὰ βαρύτονα, διὰ µακροῦ τοῦ ι γράφεται· οἷον, Ἶλος (τὸ κύριον· ἐπὶ γὰρ τῆς καταδύσεως τοῦ θηρίου ὀξύνεται), πῖλος, κτῖλος, Βρῖλος (ὄνοµα κύριον), Μῖλος (ὄνοµα νήσου), φίλος· τοῦτο συστέλλει τὸ ι· κτῖλος (ὁ προηγούµενος τῶν προ- βάτων κριός)· οἷς ὅµοιον τὸ ψιλός, χιλός, εἰ καὶ περὶ τὸν τόνον διήλλαξεν· τὸ δειλός ἀπὸ τοῦ δέος διὰ τῆς ει διφθόγγου γράφε- ται· σεσηµείωται τὸ Τῆλος (ὄνοµα κύριον λιµένος καὶ πόλεως), Μῆλος, Βῆλος (ὁ Ζεύς), δῆλος (ὁ φανερός), Σφῆλος (ὄνοµα κύ- ριον), ἧλος (τὸ δασυνόµενον)· Ἦλος τὸ κύριον, ὃ καὶ ψιλοῦται. 18 G. Horrocks, Greek: A History of the Language and its Speakers2 (Chichester 2010) 168–169. 19 See also the discussion of passages (10a) and (10b). 492 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 Disyllabic [words] which end in ιλος have only one gender and are barytone, whether proper names or common nouns, are spelled with ι, e.g. Ἶλος (the proper name; for the hole of an animal is oxytone), πῖλος, κτῖλος, Βρῖλος (a proper name), Μῖλος (an island name), φίλος; this has its ι short; κτῖλος (the ram that leads the sheep); and similar to these are ψιλός, χιλός, even if they differ in accentuation; δειλός from δέος is spelled with the diphthong ει. Exceptions are Τῆλος (a proper name of a harbour or city), Μῆλος, Βῆλος (“Zeus”), δῆλος (“clear”), Σφῆ- λος (proper name), ἧλος (the aspirated one), and Ἦλος the proper name, which is unaspirated. In passage (3a) Pseudo-Arcadius (69.8–10) deals with words with more than two syllables ending in -εµος: (3a) τὰ εἰς µος ὑπερδισύλλαβα παραληγόµενα τῷ ε προπαρο- ξύνεται· Τήλεµος, Ἔχεµος, πόλεµος, ἄνεµος. τὸ δὲ θελεµός ἀπὸ τοῦ θεληµός ὀξύνεται. Words ending in µος that have more than two syllables and have ε in their penultimate syllable are proparoxytone: e.g. Τήλεµος, Ἔχεµος, πόλεµος, ἄνεµος. But θελεµός from θεληµός is oxy- tone. In (3b) Theognostus deals with the termination -εµος side by side with the identical-sounding -αιµος (344 = p.64.13–18). His comment καὶ τὸ θελεµός ὀξύτονον looks left over from an accentuation rule like the one in (3a), since it does not matter for Theognostus’ rule how a word is accented.20 (3b) τὰ διὰ τοῦ εµος ἁπλᾶ ὑπὲρ δύο συλλαβὰς διὰ τοῦ ε ψιλοῦ γράφονται· οἷον, ἄνεµος, πόλεµος, ἤρεµος, Ἔχεµος (ὄνοµα κύ- ριον), Τήλεµος, τρίλεµος, Ἰάλεµος, κάλλεµος, φάλεµος, ἄγρε- µος, ἄνθεµος (ἐξ οὗ τὸ ἀνθεµόεις)· καὶ τὸ θελεµός ὀξύτονον· πρόσκειται ἁπλᾶ διὰ τὸ ἄναιµος σύνθετον παρὰ τὸ αἷµα, καὶ εἴτι παρ’ αὐτὸ ἕτερον, σύναιµος, πολύαιµος. Uncompounded [words] ending in εµος which have more than two syllables are spelled with ε, e.g. ἄνεµος, πόλεµος, ἤρεµος, Ἔχεµος (a proper name), Τήλεµος, τρίλεµος, Ἰάλεµος, κάλλε- µος, φάλεµος, ἄγρεµος, ἄνθεµος (from which comes ἀνθεµόεις); 20 Cf. the discussion of passage (6b). STEPHANIE ROUSSOU 493 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 and θελεµός is oxytone; the word “uncompounded” is included because of ἄναιµος which is a compound from αἷµα, and any- thing else based on this word, e.g. σύναιµος, πολύαιµος. (b) Omission of information about accentuation The transformation of the accentuation rules into or- thographic ones sometimes involves omitting information on accentuation, where this does not help us to decide between two possible spellings of a sound. In passage (4a) Pseudo-Arcadius (8.10–12) deals with the accentuation of two-termination words ending in -ιν, which are said to be oxytone: (4a) τὰ εἰς ιν δικατάληκτα ἁπλᾶ µὲν ὀξύνεται· δελφίν καὶ δελ- φίς, Τελχίν καὶ Τελχίς, Σαλαµίν καὶ Σαλαµίς, ἀκτίν καὶ ἀκτίς. Uncompounded two-termination [words] ending in ιν are oxytone: δελφίν and δελφίς, Τελχίν and Τελχίς, Σαλαµίν and Σαλαµίς, ἀκτίν and ἀκτίς. Theognostus in (4b) gives instead a spelling rule on two-termi- nation words ending in -ιν (146 = p.27.14–15). The examples cited are all oxytone, but he does not explicitly mention their accentuation: (4b) τὰ εἰς ιν δικατάληκτα διὰ τοῦ ι γράφεται· δελφίν, ῥίν, Ἐλευσίν, θίν. Two-termination [words ending] in ιν are spelled with ι, e.g. δελφίν, ῥίν, Ἐλευσίν, θίν. Passage (5a) is Pseudo-Arcadius’ accentuation rule on mascu- line and feminine monosyllabic words ending in -ρ which are oxytone (144.8–10): (5a) τὰ εἰς ρ ἀρσενικὰ καὶ θηλυκὰ µονοσύλλαβα ὀξύνεται· ψάρ, κάρ, σήρ, θήρ, κήρ (ἡ θανατηφόρος µοῖρα), χείρ, φθείρ, φώρ (ὁ κλέπτης). Monosyllabic masculines and feminines ending in ρ are oxytone: ψάρ, κάρ, σήρ, θήρ, κήρ (“death-bringing fate”), χείρ, φθείρ, φώρ (“thief”). Theognostus’ orthographic rule in (5b) deals with the spelling of words ending in -ρ (223 = p.41.13–20). Following this initial statement of the principle, we get a list of words ending in -ηρ. 494 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 These words are all oxytone, but Theognostus does not men- tion their accentuation: (5b) πᾶν ὄνοµα εἰς ρ λῆγον, πᾶσαν δίφθογγον ἀποστρέφεται πλὴν τῆς ει ποσότητος συλλαβῆς ἐν τῷ κανόνι µὴ τηρουµένης· οἷον, κήρ, θήρ, ἀνήρ, αἰθήρ· σεσηµείωται τὸ χείρ καὶ φθείρ ἀπὸ τῆς σαρκός21 διὰ τῆς ει διφθόγγου γραφόµενα· τὸ γὰρ κύριον ὄνοµα οἱ πλείους διὰ τοῦ ι γράφουσι· τὸ αὐτόχειρ· ἑκατόγχειρ· πολύχειρ, διὰ τῆς ει διφθόγγου γραφόµενα σύνθετά ἐστι παρὰ τὸ χείρ, καὶ τοῦ ἁπλοῦ τὴν γραφὴν ἐφύλαξαν. Every nominal ending in ρ rejects every diphthong apart from ει when the quantity of the syllable is not maintained in the rule, e.g. κήρ, θήρ, ἀνήρ, αἰθήρ; χείρ and φθείρ [generated] from the flesh, spelled with the diphthong ει; for most people spell the proper name with the ι; αὐτόχειρ, ἑκατόγχειρ, πολύχειρ spelled with the diphthong ει are compounded from χείρ, and have maintained the spelling of the uncompounded form. In passage (6a) Pseudo-Arcadius (115.6–10) prescribes reces- sive accentuation for trisyllabic nouns ending in -ερᾱ: (6a) τὰ εἰς ρα µακροκατάληκτα τρισύλλαβα παραληγόµενα τῷ ε µονογενῆ βαρύνεται, εἰ µὴ ἐπιθετικὰ εἴη· ἡµέρα, ἀσκέρα (εἶδος ὑποδήµατος), διφθέρα, λακέρα. τὸ δὲ Γλυκερά ὀξύνεται, εἴτε κύριον, εἴτε ἐπιθετικὸν εἴη. Trisyllabic [words] ending in ρα, ending in a long syllable, with an ε in their penultimate syllable, and with only one gender, are recessive if they are not adjectives, e.g. ἡµέρα, ἀσκέρα (a kind of sandal), διφθέρα, λακέρα. But Γλυκερά is oxytone, whether it is the proper name or the adjective. In his spelling rule on trisyllabic nouns ending in -ερᾱ Theo- gnostus (638.1–8 = pp.106.28–107.3) again does not specify the accent, and this time he includes a comment showing that he considers the accent irrelevant to the spelling rule. After a series of recessive examples he moves on to γλυκερά, φοβερά, and πενθερά. He acknowledges that these differ in accent from the µονογενῆ “words of one gender,” but points out that the differ- 21 For the phrase φθείρ ἀπὸ τῆς σαρκός cf. Arist. Hist.An. 556b28 οἱ δὲ φθεῖρες ἐκ τῶν σαρκῶν. STEPHANIE ROUSSOU 495 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 ence extends only to accent, not spelling (περὶ τόνον οὐ περὶ τὴν γραφὴν διαφορά): (6b) τὰ διὰ τοῦ ερα τρισύλλαβα µονογενῆ µακροκατάληκτα, πρὸ τέλους ἔχοντα τὸν τόνον, διὰ τοῦ ε ψιλοῦ γράφονται· οἷον, ἡµέρα, ἑσπέρα, διφθέρα, χολέρα, ὑπέρα, τροµέρα, ἀλέρα, κυδέρα, Θεσκέρα (ἡ τιθηνὸς Διονύσου)· τὸ γλυκερά, φοβερά, πενθερά οὐ µονογενῆ, ὅµως καὶ οὕτως περὶ τόνον οὐ περὶ τὴν γραφὴν διαφορά· πρόσκειται πρὸ τέλους ἔχοντα τὸν τόνον, διὰ τὸ χίµαιρα, µάχαιρα· ταῦτα γὰρ προπαροξύνεται, καὶ διὰ τῆς αι διφθόγγου γράφεται. Trisyllabic [words ending] in ερα which have only one gender and have a long final syllable are spelled with ε, if they have their accent on the penultimate syllable, e.g. ἡµέρα, ἑσπέρα, διφθέρα, χολέρα, ὑπέρα, τροµέρα, ἀλέρα, κυδέρα, Θεσκέρα (the nurse of Dionysus). γλυκερά, φοβερά, and πενθερά are not words of only one gender, but even so the difference is not in the accent but in the spelling. The phrase “having their accent in the penultimate syllable” is included because of χίµαιρα and µάχαιρα; for these are proparoxytone, and they are spelled with the diphthong αι. (c) Addition of orthographic principles In some instances Theognostus not only omits information about the accent but adds orthographic principles that will help the reader choose between different ways of spelling the same sound. In passage (7a) Pseudo-Arcadius (8.14–16) has an accentua- tion rule on Aeolic words ending in -υν, which are recessive: (7a) τὰ εἰς υν λήγοντα βαρύνεται ἐξαιρέτως παρὰ τοῖς Αἰο- λεῦσι· Φόρκυν, Πόλτυν, Γόρτυν, µόσ⟨σ⟩υν, τέκτυν (ὁ τέκτων). [Words] ending in υν are recessive specifically in Aeolic, e.g. Φόρκυν, Πόλτυν, Γόρτυν, µόσ⟨σ⟩υν, τέκτυν (“carpenter”). In (7b) Theognostus (147 = p.27.16–17) creates an ortho- graphic rule on words ending in -υν and comments that a nom- inative singular cannot have a termination -οιν (οὐκ οἶδε τὴν οι δίφθογγον). All the examples cited are recessive, but this is not explicitly mentioned. The point that no nominative singular 496 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 ends in -οιν helps readers to choose the spelling -υν rather than the identical-sounding -οιν:22 (7b) τὰ εἰς υν ἐπ’ εὐθείας ἑνικῆς οὐκ οἶδε τὴν οι δίφθογγον· Φόρκυν, Πόλτυν, Κόλυν, Κάπυν, Γόρτυν (ὄνοµα ἥρωος) The [nominals] ending in [the sound] υν in the nominative singular do not know the diphthong οι [i.e. they are not spelled with οιν], e.g. Φόρκυν, Πόλτυν, Κόλυν, Κάπυν, Γόρτυν (the name of a hero) (d) Conditions for accentuation rules which are not conditions for the spelling of words are presented side by side with their opposite (or side by side with an alternative), as alternatives in the orthographic rule Another feature of Theognostus’ method of transforming Herodian’s accentuation rules into orthographic ones involves words and phrases which were originally conditions for the accentuation rules but which are irrelevant to the orthographic rules, and which appear as alternatives next to each other in the orthographic rule, usually in the pattern εἴτε … εἴτε. When Theognostus says εἴτε x … εἴτε y this is often a sign that he has combined more than one accentuation rule. The alternatives found in the orthographic rules regard: (i) the type of accent (e.g. one finds the phrase “whether oxytone or barytone”), (ii) the number of syllables (e.g. one finds the phrase “whether di- syllabic or with more than two syllables”), (iii) the type of nom- inal (e.g. one finds “whether a nominal, or common noun, or an adjective”), (iv) the type of declension (e.g. one finds “wheth- er they are parisyllabically or imparisyllabically declined”), (v) the vowel length of the penultimate syllable (e.g. one finds “whether they have the ι long or short”). The number of syllables is often significant for accentuation rules such as we find in Pseudo-Arcadius. He has two separate accentuation rules on words ending in -βων (8.17–19, 9.1–2): one rule for the disyllabic words (passage 8a1), which are re- cessive, and one for those with more than two syllables (8a2), 22 On Theognostus’ arrangement of the treatment of the vowels accord- ing to how they sounded in his time see Krumbacher, Geschichte 585. STEPHANIE ROUSSOU 497 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 which are oxytone: (8a1) τὰ εἰς βων δισύλλαβα ἀρσενικὰ βαρύνεται, εἰ µὴ µέρος σώµατος δηλοῖ, οἷον Στίβων, τρίβων, Στράβων, ἄµβων. Masculine disyllabic [words] ending in βων are recessive, unless they indicate a part of the body, e.g. Στίβων, τρίβων, Στράβων, ἄµβων. (8a2) τὰ εἰς βων ὑπερδισύλλαβα ὀξύνεται· Ἀλαβών, ἀρραβών. τὸ δὲ Χαρναβῶν περισπᾶται. Words ending in βων that have more than two syllables are oxy- tone: Ἀλαβών, ἀρραβών. But Χαρναβῶν is perispomenon. However, the number of syllables is not normally crucial for the spelling of a word’s termination, which is Theognostus’ concern. In (8b) Theognostus therefore has one rule for words ending in -βων with two or more syllables (169 = p.30.29–34). Some of these are recessive and some are oxytone, but this division into two accentual types does not interfere with their being similarly declined and spelled. Although Theognostus combines these into one rule, his method of working from accentuation rules is visible in the superfluous phrase εἴτε βαρύτονα εἴτε ὀξύτονα.23 Furthermore, the phrase εἴτε δι- σύλλαβα, εἴτε ὑπὲρ δύο συλλαβάς also looks left over from the combination of two Herodianic rules into one. Theognostus could have said, more succinctly, “of more than one syllable,” but he retains from his source the concept of disyllabic words and the concept of words with more than two syllables: (8b) τὰ εἰς βων λήγοντα καθαρόν, εἴτε δισύλλαβα, εἴτε ὑπὲρ δύο συλλαβάς, εἴτε βαρύτονα, εἴτε ὀξύτονα, διὰ τοῦ ω κλί- νεται· οἷον, Σκρίβων Σκρίβωνος, Στράβων Στράβωνος, Κίκων24 Κίκωνος, στίβων στίβωνος, Ἀλαβών Ἀλαβῶνος, ἀρραβών ἀρρα- βῶνος· σεσηµείωται τὸ Χαρναβῶν περισπώµενον, διὰ τοῦ ντ κλινόµενον καὶ φυλάττον τὸ ω. 23 Similar examples include rules 177 = p.32.4–10, 236 = p.43.19–22, 268 = p.49.21–9, and 342 = p.63.31–64.6. 24 This example seems to be out of place since it does not end in -βων. 498 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 [Words] ending in βων preceded by a vowel, whether they are disyllabic or have more than two syllables, and whether they are barytone or oxytone, are declined with ω: e.g. Σκρίβων Σκρί- βωνος, Στράβων Στράβωνος, Κίκων Κίκωνος, στίβων στίβωνος, Ἀλαβών Ἀλαβῶνος, ἀρραβών ἀρραβῶνος; the perispomenon Χαρναβῶν, which is declined with ντ and maintains the ω, is an exception. Another example can be gained by comparing Pseudo- Arcadius’ treatment of the termination -πη at 130.2–17 with Theognostus’ orthographic rule on words ending in -υπη. Pseudo-Arcadius has three separate rules, two on disyllabic words and one on those with more than two syllables. In pas- sage (9) Theognostus (704 = p.116.22–27) has one large spell- ing rule on words ending in -υπη, both the disyllabic ones and those with more than two syllables, and both the oxytone ones and the recessive ones. Here again, the phrase εἴτε δισύλλαβα, εἴτε ὑπὲρ δύο συλλαβάς betrays that this phrase results from the combination of more than one Herodianic rule. Moreover, accentuation does not provide any condition for the ortho- graphic rule, but the occurrence of the phrase εἴτε ὀξύτονα εἴτε βαρύτονα suggests that the source for this rule was an ac- centuation rule:25 (9) τὰ διὰ τοῦ υπη, εἴτε δισύλλαβα, εἴτε ὑπὲρ δύο συλλαβάς, εἴτε ὀξύτονα, εἴτε βαρύτονα, µονογενῆ διὰ τοῦ υ ψιλοῦ γρά- φεται· οἷον, λύπη, γύπη, κύπη (εἶδος πλοίου), τυπή ὀξυτόνως, Ἀρυπή (ἡ πόλις), λατυπή ὀξυτόνως, χαµαιτυπή· τὸ λύπη βαρυ- τόνως· πρόσκειται µονογενῆ, διὰ τὸ λοιπή ὀξυνόµενον, καὶ διὰ τῆς οι διφθόγγου γραφόµενον, τριγενὲς γάρ. [Words] ending in υπη, whether they are disyllabic or have more than two syllables, and whether they are oxytone or bary- tone, are spelled with υ if they have only one gender, e.g. λύπη, γύπη, κύπη (a kind of ship); τυπή is oxytone; Ἀρυπή (a city), λατυπή oxytone, χαµαιτυπή; but λύπη is barytone; “which only have one gender” is included because of the oxytone λοιπή, 25 A further example of the phrase εἴτε βαρύτονα εἴτε ὀξύτονα comes from the comparison of Pseudo-Arcadius 6.10–8.9 with Theognostus 145 = p.27.7–13. STEPHANIE ROUSSOU 499 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 which is spelled with the diphthong οι: for it has a separate form for each of the three genders [i.e. it is a three-termination adjec- tive]. Sometimes Pseudo-Arcadius differentiates between proper names, common nouns, and adjectives with the same termina- tion, while Theognostus has one rule on words of the same termination, regardless of what type of nominal they belong to. The type of nominal is irrelevant for Theognostus’ purpose, but he betrays his method of working by including a phrase such as “whether they are proper nouns, common nouns, or adjectives.” Thus, in (10a) Pseudo-Arcadius (58.15–19) pre- scribes a recessive accent for words ending in -υκος which have more than two syllables and are either proper names or com- mon nouns. This accentuation rule excludes adjectives because they do not have a recessive accent but an oxytone one: (10a) τὰ εἰς υκος ὑπερδισύλλαβα κύρια ἢ προσηγορικὰ µὴ ἔχοντα κατ’ ἰδίαν θηλυκὰ κτητικῆς ἐννοίας ἐχόµενα βαρύνεται· Ἴβυκος, Ἄµυκος, κώρυκος, Ἴνυκος. τὸ δὲ Λιβυκός καὶ θηλυκός καὶ ἁλυκός ὀξύνεται θηλυκὰ ἔχοντα. Proper names or other nominals ending in υκος which have more than two syllables, and which do not have a separate form for the feminine [i.e. common nouns and two-termination adjec- tives], are recessive: Ἴβυκος, Ἄµυκος, κώρυκος, Ἴνυκος. But Λι- βυκός and θηλυκός and ἁλυκός, which have feminines [i.e. a separate form for the feminine], are oxytone. In (10b) Theognostus (324 = p.60.10–18) gives an orthographic rule for proper names, common nouns, and adjectives that end in the sound ‘-υκος’ and have more than two syllables. Al- though these three types of nominals are not all accented in the same way, they have their termination spelled in the same way. Any kind of nominal ending in the sound ‘-υκος’ is generally spelled with the letters υκος, not with οικος (which sounded the same in Theognostus’ day), and so the rule could have simply said τὰ εἰς κος λήγοντα (διὰ τοῦ υ) or τὰ εἰς υκος λήγοντα without specifying the different types of nominals which can be spelled in this way. Thus, the phrase εἴτε κύρια, εἴτε προσηγορικά, εἴτε ἐπίθετα reveals that Theognostus based his 500 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 rule on an accentuation rule for which it was relevant to distinguish between these different types of nominals: (10b) τὰ εἰς κος λήγοντα καθαρὸν ὑπὲρ δύο συλλαβάς, εἴτε κύρια, εἴτε προσηγορικά, εἴτε ἐπίθετα µὴ ἔχοντα κατ’ ἰδίαν θηλυκά, οὐκ οἶδεν ἐν τῇ πρὸ τέλους τὴν οι δίφθογγον· οἷον, Ἴβυκος, Ἄµυκος, Κώρυκος, Σίµυκος, Δίνυκος (ὄνοµα πόλεως), Σίτυκος, Ἁρπάλυκος (ὄνοµα κύριον), Εὔτυκος, Ἐλίσυκος (ὄνοµα ἔθνους), Κήρυκος (ὄνοµα κύριον), πλὴν τοῦ σόλοικος· τὸ θη- λυκός, Λιβυκός, ἁλυκός, τὴν γραφὴν φυλάξαντα τὸν τόνον καὶ τὴν σηµασίαν ἐνήλλαξαν, κτητικῆς ὕλης τυγχάνοντα. [Words] ending in κος preceded by a vowel, and having more than two syllables, whether they are proper names, common nouns, or adjectives which do not have a separate feminine form [i.e. two-termination adjectives], do not know the diphthong οι in the penultimate [i.e. the diphthong οι is not usual], e.g. Ἴβυ- κος, Ἄµυκος, Κώρυκος, Σίµυκος, Δίνυκος (a city name), Σίτυκος, Ἁρπάλυκος (a proper name), Εὔτυκος, Ἐλίσυκος (a nation name), Κήρυκος (a proper name), apart from σόλοικος; but θηλυκός, Λιβυκός, ἁλυκός which have maintained their spell- ing, have changed their accent and meaning in obtaining a possessive matter. The accentuation is not really relevant for Theognostus’ spell- ing rule in (10b): the oxytone words listed are all spelled with υ in the penultimate, just like the proparoxytone words except for σόλοικος. The accentuation does not help to distinguish be- tween two spellings of the same sound here, and yet having stated his orthographic rule Theognostus first gives examples that all happen to be proparoxytone, then mentions an excep- tion (σόλοικος) and then makes a special remark for the oxy- tone words θηλυκός, Λιβυκός, and ἁλυκός. In addition, the status of the condition “which do not have a separate feminine form” is unclear in Theognostus’ rule because the words cited as examples do have a separate feminine form: this too looks left over from an accent rule such as (10a). Sometimes the parisyllabic or imparisyllabic declension of certain words is relevant for their accentuation. In (11a), for example, Pseudo-Arcadius (145.11–14) distinguishes between monosyllabic words ending in -ης which are imparisyllabically STEPHANIE ROUSSOU 501 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 declined and are oxytone, and those which are parisyllabically declined and are perispomenon: (11a) τὰ εἰς ης µονοσύλλαβα τυχόντα φυσικῆς καταλήξεως ἢ ὀξύνεται ἢ περισπᾶται. καὶ ὀξύνεται µὲν τὰ περιττοσυλλάβως κλινόµενα: Κρής, σής, Γνής (ὁ Ῥόδιος)· περισπᾶται δὲ τὰ ἰσο- σύλλαβα· Δρῆς, Τρῆς κύρια. Monosyllabic [words] ending in ης are either oxytone or peri- spomenon if they have their natural termination. And those which are declined imparisyllabically are oxytone: Κρής, σής, Γνής (“Rhodian”). The parisyllabic ones are perispomenon: Δρῆς, Τρῆς, [which are] proper names. Theognostus (808 = p.134.21–8), by contrast, has a single rule on monosyllabic words ending in -ης. No distinction between parisyllabic and imparisyllabic declensions is relevant for his orthographic rule, but the presence of this distinction in his source shows through in his phrase ἰσοσυλλάβως καὶ περιττο- συλλάβως κλινόµενα, “whether they are parisyllabically or im- parisyllabically declined”: (11b) τὰ (addendum) εἰς ης µονοσύλλαβα, ὀξύτονα δὴ λέγω καὶ περισπώµενα, ἰσοσυλλάβως καὶ περιττοσυλλάβως κλινόµενα, σπάνιά ἐστιν· τὸ βλής καὶ θής (ὁ διὰ τροφὴν δουλεύων ἐλεύ- θερος· παρὰ δὲ Ἀττικοῖς ὁ ἄτιµος, καὶ µηδὲν ἔχων), Κλής (ἐπὶ τοῦ ποταµοῦ), Κρής (ὃ ἐπὶ τοῦ ἔθνους)· τὸ γὰρ οὐδέτερον ἐκ συναιρέσεως ὂν περισπᾶται· Γνής (ὄνοµα ἔθνους), γλής, φρής, σής (ὁ σκώληξ, καὶ τὸ ποταµοῦ ὄνοµα)· περισπᾶται δὲ τὸ δρῆς καὶ τρῆς, ὡς δὲ τινὲς βούλονται καὶ τὸ ζῆς. Monosyllabic words ending in ης, I mean oxytone and peri- spomenon ones, whether they are parisyllabically or imparisyl- labically declined, are rare: βλής and θής (“free man working for someone else for food,” but in Attic writers “one who is of low status and has nothing”), Κλής (used of the river), Κρής (which [is used] of the ethnic group); for the neuter, which is formed by contraction, is perispomenon; Γνής (the name of a nation), γλής, φρής, σής (“worm” and a river name); δρῆς and τρῆς are peri- spomena, and according to some also ζῆς. In passage (12a) Pseudo-Arcadius (15.21–16.2) has a rule on words in -ων that are perispomenon when they are declined with ντ: 502 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 (12a) τὰ εἰς ων ἐν τέλει ἔχοντα τὸν τόνον καὶ κλινόµενα διὰ τοῦ ντ περισπᾶται· Ξενοφῶν, Ἱπποκῶν. σεσηµείωται τὸ Ποσειδῶν καὶ ταῶν καὶ τυφῶν περισπώµενα µέν, οὐ µέντοι διὰ τοῦ ντ κλινόµενα. The [words] ending in ων which have their accent on the final syllable and are declined with ντ are perispomenon, e.g. Ξενο- φῶν, Ἱπποκῶν. [The words] Ποσειδῶν and ταῶν and τυφῶν, which are perispomenon but are not declined with ντ, are ex- ceptions. However, the declension with ντ or ν is irrelevant to the spelling of the words ending in -ων, which is Theognostus’ concern (161 = p.29.29–31). The genitive is to be spelled with ω regardless of whether the word declines with ντ or with ν. The comment κἄντε διὰ τοῦ ντ κλίνοιτο, κἄν τε µή in (12b) is strictly speaking redundant for Theognostus, and looks left over from an accentuation rule such as (12a): (12b) τὰ εἰς ων περισπώµενα, κἄντε διὰ τοῦ ντ κλίνοιτο, κἄν τε µή, φυλάττει τὸ ω κατὰ τὴν γενικήν· Χαρναβῶν Χαρναβῶντος, Σολοµῶν Σολοµῶντος, Τυφῶν Τυφῶνος. The perispomenon words ending in ων, whether they are de- clined with ντ or not, maintain the ω in the genitive, e.g. Χαρνα- βῶν Χαρναβῶντος, Σολοµῶν Σολοµῶντος, Τυφῶν Τυφῶνος. The length of a vowel in the penultimate syllable often plays a significant role in determining the accent of a word. In the accentuation rule on disyllabic verbs ending in -νω, Pseudo- Arcadius (182.17–183.1) says that if the penultimate syllable contains a short anceps vowel then these verbs will be peri- spomenon: (13a1) τὰ εἰς νω δισύλλαβα ἔχοντα τὴν πρὸ τέλους συλλαβὴν εἰς δίχρονον συνεσταλµένον καταλήγουσαν, ⟨ἀρχόµενα⟩ ἢ ἀπὸ συµφώνου ἢ συµφώνων µὴ ἀντιστοίχων, περισπᾶται, χωρὶς εἰ µὴ κατ’ ἐπένθεσιν εἴη τοῦ ν γινόµενα· κυνῶ καὶ ἐν συνθέσει προσκυνῶ, πλανῶ, σινῶ (τὸ βλάπτω). Disyllabic [verbs] ending in νω that have their penultimate syl- lable ending in a short anceps vowel, or which begin with a con- sonant or consonants that do not correspond, are perispomenon unless they are formed with the insertion of ν: κυνῶ, and in com- position προσκυνῶ, πλανῶ, σινῶ (“damage”). STEPHANIE ROUSSOU 503 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 In another rule on the disyllabic verbs ending in -νω, Pseudo- Arcadius (183.22–24) says that if the penultimate syllable has a long ι the word will be recessive, unless the verb was formed after its corresponding noun. Thus, κρίνω, πίνω, κλίνω, and σίνω are said to be recessive, while ῥινῶ which derives from the noun ῥίνη is perispomenon: (13a2) τὰ εἰς ινω δισύλλαβα ἔχοντα τὸ ι ἐκτεταµένον βαρύνε- ται, εἰ µὴ προκατάρχοιτο ὄνοµα· κρίνω, πίνω, κλίνω, σίνω. τὸ δὲ ῥινῶ παρὰ τὴν ῥίνην. Disyllabic [verbs] ending in ινω that have a long ι are recessive, unless a nominal was formed before [the verb]: κρίνω, πίνω, κλίνω, σίνω. ῥινῶ [is derived] from ῥίνη. Theognostus (870 = p.144.3–9), so long as the word termina- tion is spelled in the same way, does not need to distinguish between verbs whose penultimate has a long ι or a short ι. His orthographic rule therefore deals with all disyllabic words end- ing in the sound ‘-ινω’, but he nevertheless includes a remark saying that the iota in the penultimate syllable can be either long or short (εἴτε µακρὸν ἔχει τὸ ι, εἴτε βραχύ): (13b) τὰ διὰ τοῦ ινω δισύλλαβα, εἴτε µακρὸν ἔχει τὸ ι, εἴτε βραχύ, µὴ ἔχοντα ἐν τῷ µέλλοντι ἀναφαινόµενον τὸ ε, διὰ µόνον τοῦ ι γράφεται· οἷον, σίνω, πίνω, φθίνω, κλίνω, τίνω (τὸ ἀπο- δίδωµι), οὗ παράγωγον τὸ τιννύω· γίνω (τὸ κατασκευάζω), ἐπὶ γὰρ τοῦ γεννῶ διὰ τῆς ει διφθόγγου· κτείνω δὲ καὶ τείνω τὸ ἁπλῶ, διὰ τῆς ει διφθόγγου, τενῶ γὰρ καὶ κτενῶ οἱ µέλλοντες. Disyllabic [verbs] ending in ινω, whether they have the ι long or short, are spelled only with ι if they do not have an ε appearing in the future, e.g. σίνω, πίνω, φθίνω, κλίνω, τίνω (“give back”), whose derivative is τιννύω; γίνω (“construct”), for when the meaning is γεννῶ the spelling is with the diphthong ει; κτείνω and τείνω (“stretch out”) with the diphthong ει, for τενῶ and κτενῶ are the futures. We have thus seen that instead of simply omitting irrelevant information (about the accent, the type of nominal, the number of syllables, the declension, and the vowel length in the penulti- mate) Theognostus sometimes betrays his way of working by including a phrase of the type εἴτε x … εἴτε y. 504 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 (e) Addition of a qualification in cases where words spelled in the same way but accented differently are dealt with in the same rule Pseudo-Arcadius (94.6–10) prescribes a recessive accent for masculine words ending in -ιτος that have more than two syl- lables, with the exception of adjectives, and then gives some examples of feminines which are oxytone: (14a) τὰ εἰς τος ὑπὲρ δύο συλλαβὰς ἀρσενικὰ παραληγόµενα τῷ ι µὴ ὄντα ἐπιθετικὰ βαρύνονται· βάρβιτος, Ἀδήριτος, Λήϊτος, Ἄγχιτος. τὰ δὲ θηλυκὰ ὀξύνεται· ἁµαξιτός, ἀτραπιτός, Λυχνιτός (ἡ πόλις) [δηριτός]. Masculines ending in τος that have more than two syllables and have ι in their penultimate syllable are recessive if they are not adjectives: βάρβιτος, Ἀδήριτος, Λήϊτος, Ἄγχιτος. And the feminines are oxytone: ἁµαξιτός, ἀτραπιτός, Λυχνιτός (a city) [δηριτός]. In (14b) Theognostus (419 = p.75.21–6) deals with the spelling of adjectives and proper names ending in -ιτος that have more than two syllables and are proparoxytone. He then moves on to cite some words whose termination is spelled in the same way as that of the words mentioned already, but which are accented differently. Theognostus’ decision to mention the accent or gender at all betrays the way he has worked from an accentua- tion rule for which these features were relevant: (14b) τὰ διὰ τοῦ ιτος ὑπὲρ δύο συλλαβὰς ἁπλᾶ προπαροξύτονα, ἐπιθετικά τε καὶ κύρια, διὰ τοῦ ι γραφόµενα σπάνιά ἐστιν· οἷον, Ἄγχιτος (ὄνοµα κύριον), Ἀφρόδιτος, ἀδήριτος, Ὀΐκριτος (ὄνοµα κύριον)· τὸ ἁµαξιτός, λυχνιτός, ἀτραπιτός ὀξύτονα, κατὰ τόνον καὶ κατὰ γένος, οὐ κατὰ τὴν γραφὴν ἀνακόλουθα. Uncompounded [words] which end in ιτος, have more than two syllables, and are proparoxytone, both adjectives and proper names, and are spelled with ι are rare, e.g. Ἄγχιτος (a proper name), Ἀφρόδιτος, ἀδήριτος, Ὀΐκριτος (a proper name). ἁµαξι- τός, λυχνιτός, and ἀτραπιτός, which are oxytone, are irregular as regards their accent and gender but not with regard to their spelling. In passage (15a) Pseudo-Arcadius (81.21–82.2) prescribes a recessive accent for words with more than two syllables ending in -ρος that have ει or ι in their penultimate syllable: STEPHANIE ROUSSOU 505 ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 (15a) τὰ εἰς ρος ὑπὲρ δύο συλλαβὰς παραληγόµενα τῇ ει διφθόγγῳ ἢ µόνῳ τῷ ι ἐκτεταµένῳ βαρύνεται· Κάµειρος, ὄνει- ρος, πέπειρος, µάγειρος, Στάλιρος, Σύσιρος. τὸ καυστειρός οὐ µάχεται· ἀπὸ γὰρ τοῦ η ἐτράπη εἰς ει. [Words] ending in ρος that have more than two syllables and have the diphthong ει or a long ι alone are recessive: Κάµειρος, ὄνειρος, πέπειρος, µάγειρος, Στάλιρος, Σύσιρος. And καυστει- ρός is not an exception for starting from η it underwent a change to ει. Theognostus (392.1–6 = p.71.19–24) deals with the spelling of words ending in -ειρος in (15b). After a list of proparoxytone words, we get the example καυστειρός, with a comment about its different accent. The accent of all these words is irrelevant to their spelling, and Theognostus could have simply not men- tioned the accent anywhere in this rule, but his mention of the accent betrays once again that his source was an accentuation rule: (15b) τὰ διὰ τοῦ ειρος προπαροξύτονα ἀπὸ ῥηµάτων γινόµενα διὰ τῆς ει διφθόγγου γράφονται· οἷον, µάσσω µάγειρος· Αἰολεῖς δὲ διὰ τοῦ ι· πέπτω πέπειρος, ὀνῶ (τὸ ὀφελῶ) ὄνειρος, ἀΐσσω αἴγειρος, ἥδω Ἄνδειρος (ὄνοµα ποταµοῦ)· κονῶ Κόνειρος (ὄνο- µα ἔθνους), καίω καυστειρός· τοῦτο τὴν γραφὴν ἐφύλαξεν οὐ τὸν τόνον, ὀξύνεται γάρ Proparoxytone [words] ending in ειρος which are formed from verbs are spelled with the diphthong ει, e.g. µάσσω µάγειρος; but the Aeolians spell this with ι; πέπτω πέπειρος, ὀνῶ (“benefit”) ὄνειρος, ἀΐσσω αἴγειρος, ἥδω Ἄνδειρος (a river name); κονῶ Κόνειρος (the name of a nation), καίω καυστειρός; this has maintained the [same] spelling but not the accent, for it is oxy- tone 6. Conclusions We have considered several features of Theognostus’ trans- formation of Herodian’s accentuation rules into orthographic ones, and these may now be summed up as follows: (i) Theognostus deals with terminations which sounded the same (because of changes in the pronunciation of vowels) in the context of the same rule. (ii) He sometimes omits from his rules material which concerns 506 THE RECEPTION OF HERODIAN ————— Greek, Roman, and Byzantine Studies 57 (2017) 482–506 only the accentuation of the words dealt with. (iii) He adds material pertaining to the orthography of the words (particularly information pertaining to alternative spell- ings for the same vowel sounds). (iv) His use of phrases of the type εἴτε x ... εἴτε y with regard to the accent, the type of nominal, the number of syllables, the declension, and the vowel length in the penultimate syllable, in cases where he could have simply omitted the information, re- veals his way of working from accentuation rules, and his habit of combining two or more accentuation rules into one ortho- graphic rule. (v) He adds a qualification of the kind “words x and y differ in their accentuation but not in their spelling,” in cases where words spelled in the same way but accented differently are treated in the same orthographic rule. Herodian’s systematic treatment and arrangement of ma- terial according to the terminations of words was practical and could be easily re-used for other grammatical purposes apart from accentuation. Herodian’s work apparently became attrac- tive to Theognostus, who, in aiming at a systematic treatment of the spelling of words of the same shape, managed to re-use Herodian’s systematic arrangement of material in a construc- tive way for his own purposes. Theognostus thus provides us with a window onto the reception of Herodian’s work, in some form, in the Byzantine period.26 December, 2016 Faculty of Classics University of Oxford 66 St Giles’ St., Oxford, U.K. OX1 3LU stephanie.roussou@gmail.com 26 This article was written while I held a Humboldt Post-doctoral Re- search Fellowship at the University of Cologne. I am deeply grateful to Philomen Probert for her feedback on my article.