¹ Michael Syncellus lived in the period 760/1–846, and composed the aforementioned book between 810 and 813: D. Donnet, Le Traité de la Construction de la Phrase de Michel le Syncelle de Jérusalem: Histoire du Texte, Édition, Traduction et Commentaire (Brussels and Rome, 1982), 3–4; R.H. Robin, The Byzantine Grammarians. Their Place in History (Berlin and New York, 1993), 149–62; N. Wilson, Scholars of Byzantium (London, 1996²), 76–8. References are to lines and paragraphs of Donnet's edition. Throughout the article I use the term ‘anastrophe’ to denote not the ‘throwing back of the accent’ but the ‘inversion of the natural order’, which occurs when a preposition stands after its case.

² References are to pages and lines of the text in G. Dindorf, Ἰωάννου Ἀλεξανδρέως Τονικὰ παραγγέλματα. Αἰλίου Ἡρωδιανοῦ Περὶ σχημάτων (Leipzig, 1825). Dindorf's edition relies on a flawed collation of a single manuscript, the Hauniensis 1965 of the late fifteenth century (A), ignoring the other important and almost complete manuscript of the work, the sixteenth-century Vindobonensis phil.gr. 240 (V): see Xenis, G., ‘Herodian and Strattis: a further link?’, RhM 156 (2013), 106–10Google Scholar, at 106 with nn. 1–3. There are three more manuscripts, which are fragmentary and worthless for the constitution of the text: Vaticanus 1766, Taurinensis 271 and Ambrosianus C 69 sup. (gr. 184). On the first two see P. Egenolff, Die orthoepischen Stücke der byzantinischen Litteratur (Leipzig, 1887), 38 and A. Dyck, ‘Aelius Herodian: recent studies and prospects for future research’, in ANRW 2.34.1 (Berlin and New York, 1993), 772–94, at 777. The third is described by A. Martini and D. Bassi, Catalogus codicum Graecorum Bibliothecae Ambrosianae, vol. 1 (Milan, 1906), 194–200, and P. Géhin, ‘Évagre le Pontique dans un recueil de mélanges grammaticaux du fonds Pinelli, l'Ambr. C 69 sup.’, in C.M. Mazzucchi and C. Pasini (edd.), Nuove ricerche sui manoscritti greci dell’ Ambrosiana: Atti del Convegno, Milan, 5–6 June 2003 (Milan, 2004), 265–312, at 290–301. For the needs of the present study I have consulted microfilms of A and V, and have provided text-critical material at the points where this seemed necessary.

³ Herodian's work will be henceforth referred to as Cath. Pr.

⁴ These two sections do not occur in the other three epitomes of the Cath. Pr.: (i) [Arcadius'] epitome; (ii) the palimpsest fragment from Books 5–7, preserved in the Vindobonensis Hist. gr. 10; (iii) the epitome of Book 5, preserved in P.Ant. 2.67. All four epitomes are described by Dyck (n. 2), 776–83. S. Roussou has recently established a new critical text of the first eight books of [Arcadius'] epitome: ‘Pseudo-Arcadius’ epitome of Herodian's Περὶ Καθολικῆς Προσῳδίας with a critical edition and notes on Books 1–8’ (Diss., University of Oxford, 2011).

⁵ See Hunger, H., ‘Palimpsest-Fragmente aus Herodians Καθολικὴ προσῳδία, Buch 5–7; Cod. Vindob. Hist. gr. 10’, JÖByz 16 (1967), 1–33 Google Scholar, at 24, 25, 25 respectively.

⁶ See also A. Ludwich, Ιndex auctorum et dialectorum, s.v. Ἀπολλώνιος ὁ Ῥόδιος in A. Lentz, Herodiani Technici Reliquiae, vol. 2.2 (Leipzig, 1870), 1209. I shall return to the Herodianic teaching about the preposition ἀνά below.

⁷ The matter merits a few more words of explanation. How to accent a preposition falling between the components of noun phrases of such types as ποταμοῦ (ἀπο) Σελλήεντος, Ξάνθῳ (ἐπι) δινήεντι, μάχῃ (ἐνι) κυδιανείρῃ, ἡμετέρῳ (ἐνι) οἴκῳ was a question that divided the opinions of ancient grammarians. Apollonius Dyscolus (Synt. 442.1) and Herodian thought the preposition threw back its accent regardless of whether the προσηγορικώτερος or the κυριώτερος component of the noun phrase preceded the preposition, while Ptolemy of Ascalon and Nicias thought the throwing back of the accent occurred only if the προσηγορικώτερος word came first in the sequence (schol. Il. 2.839b¹). Here we need to clarify that common nouns counted as προσηγορικώτερος compared to proper nouns, and adjectives counted as προσηγορικώτερος compared to common nouns. Now if we consider the example mentioned in the main text, i.e. Il. 1.30 ἡμετέρῳ ἔνι οἴκῳ, since the προσηγορικώτερος word comes first, we conclude that Herodian and Apollonius Dyscolus will have been joined by Ptolemy of Ascalon and Nicias in making ἔνι a paroxytone preposition. For other scholars’ views on such matters, see P. Probert, A New Short Guide to the Accentuation of Ancient Greek (London, 2003), 128. As regards schol. Il. 18.92b¹, a good analysis is offered by K. Lehrs, Quaestiones Epicae (Königsberg, 1837), 81–3. For Ap. Dysc. Synt. 442.1, see R. Pfeiffer, Ausgewählte Schriften, ed. W. Bühler (Munich, 1960), 113 with n. 39. In the text of schol. Il. 18.839b¹ perhaps we should read an additional δεῖν, i.e. p. 344, line 6ff. ‘ἐμοὶ μέντοι δοκεῖ <δεῖν> τὸ τοιοῦτον ὅλον εἶδος ἀναγινώσκειν κατὰ ἀναστροφήν…’.

⁸ Another instance is 1608ff. (§ 190); for the source see Donnet (n. 1), 508.

⁹ For instance, Donnet (n. 1), 507 notes Michael's clear links to e.g. Dionysius Thrax 90–91.1 Uhlig.

¹⁰ Mich. Sync. 1606 (§ 189) ἢ φῶς ἢ σκότος, ἢ ζωὴ ἢ θάνατος, ἢ χαρὰ ἢ λύπη.

¹¹ In a disjunctive construction τὸ γὰρ ἕτερον ἀληθές, τὸ δὲ ἕτερον ψευδές (Ioh. Alex. 41.2), but in a subdisjunctive construction ἢ γὰρ τὸ ἓν ἢ τὸ ἕτερον ἢ καὶ ἀμφότερα (Ioh. Alex. 41.8). In 41.17 John further explains that the parts of a disjunctive construction are ἐναντία … μαχόμενα καὶ ἅμα εἶναι οὐ δυνάμενα.

¹² The text is corrupt, and Consbruch makes the following note in the apparatus: ‘Heliodorus haec fere scripsisse videtur ὅτε τῷ ἢ διαζευκτικῷ ἐπιφέρεται φωνῆεν’.

¹³ For the two manuscripts of the Τονικὰ παραγγέλματα see n. 2 above.

¹⁴ He refers to the note on Il. 1.365, but this note has been lost.

¹⁵ R. Schneider, Librorum Apollonii Deperditorum Fragmenta (Leipzig, 1910), 193 s.v. ἐγκλίνω; LSJ include both meanings in the same category (6), but creating two subdivisions would do more justice to the matter.

¹⁶ Dindorf (n. 2), praef. xi.

¹⁷ To be precise, many of Michael's MSS offer orthographic variants of συναληλιμμέναι: see Donnet's (n. 1) app. crit. ad loc.

¹⁸ Dindorf (n. 2) had found the true reading in a ‘MS περὶ πρ(οθέσεων)’, but relegated it to the critical apparatus (praef. xii). A. Lentz, Herodiani Technici Reliquiae, vol. 1 (Leipzig, 1867) was right to put it in his text (p. 486, line 1).

¹⁹ διαζευκτικῆς V (already Lehrs [n. 7], 52): διαζευκτικ [sic, without ending] A; Dindorf (n. 2) printed διαζευκτικοί, which has no manuscript support and is also mistaken in the context of the sentence.

²⁰ For the meaning of τάξις here, cf. e.g. Hdn. ap. schol. Il. 18.410d: … ὃ καὶ γίνεται ἐτεός. ἐγένετο δὲ καὶ τὸ πληθυντικὸν οὐδέτερον ἐν ἐπιρρηματικῇ τάξει· ‘ὡς ἐτὰ Τημενίδος χρύσεον γένος’ (Callim. fr. 780 Pf.); Hdn. Mon. 923.31 τὸ δὲ ‘Γυρῇσίν μιν πρῶτα Ποσειδάων ἐπέλασσε, | πέτρῃσι μεγάλῃσι’ (Od. 4.500-1) εἰ καὶ ἰδίως λέγεται, ἀλλ’ ὅμως ἔχει τάξιν ἐπιθετικήν.

²¹ The addition of θέλουσι was proposed by Bloch: see Dindorf (n. 2), praef. xi. Dindorf wrongly decided to leave it out of his text.

²² The closest parallel is 28.24 τὰ μὲν οὖν ἀπὸ τῶν πτωτικῶν σύνταξιν ἐπιρρηματικὴν μετελθόντων σύνταξιν ὀλίγην παραθήσομαι, which is surely corrupt. Might the original text have been τὰ μὲν οὖν ἀπὸ τῶν πτωτικῶν σύνταξιν ἐπιρρηματικὴν μετελθόντα [σύνταξιν] ὀλίγα <ἃ> παραθήσομαι?

²³ See next note.

²⁴ The text as transmitted by the MSS and printed by Dindorf (n. 2) (τῶν δὲ περισπωμένων ὑποδείγματα ταῦτα, ὑγεῖα περισπωμένως ὑγειᾶθεν) is corrupt. Lentz (n. 18), 500.20-1 offers the correct version: τῶν δὲ <προ>περισπωμένων ὑποδείγματα ταῦτα· Ὑσιαί, <προ>περισπωμένως Ὑσιᾶθεν.

²⁵ The text as trasmitted by the MSS and printed by Dindorf (n. 2) (τὸ δὲ ἀλίας παρ‘ Ἱππώνακτι παροξύνεται, ἀπὸ τοῦ ἄλις, πλεονάσαν τὸ α) is corrupt. The correct version is: τὸ δὲ ἅλιας παρ' Ἱππώνακτι (fr. 133 West) <προ>παροξύνεται ἀπὸ τοῦ ἅλις πλεονάσαν τῷ -α-. Some critical information on the passage is necessary: ἅλιας Dindorf (n. 2), praef. xvii (cf. Et. Gen. α 486, Et. Sym. α 590, Etym. Magn. α 837 L.-L.): ἀλίας AV; Ἱππώνακτι Dindorf: ἱππόνακτι ΑV; <προ>παροξύνεται Lentz (n. 18), 511.9; ἅλις V (iam Dindorf praef. xvii): ἄλις Α; τῷ scripsi: τὸ AV.

²⁶ <παρα>- addidi.

²⁷ G. Xenis, ‘Ἰωάννου Ἀλεξανδρέως Τονικὰ Παραγγέλματα: μεθοδολογικὰ καὶ διορθωτικά’, in G. Xenis (ed.), Literature, Scholarship, Philosophy, and History: Classical Studies in Memory of Ioannis Taifacos (Stuttgart, 2015), 288 with n. 23.

²⁸ At Ioh. Alex. 28.8 we must read αἱ συναληλιμμέναι οὐκ ἀναστρέφονται instead of Dindorf's mistaken αἱ συνηλλαγμέναι οὐκ ἀναστρέφονται; see p. 5, section (2) ‘Variant readings’, with n. 18.

²⁹ For more on the nature of item (v) see below.

³⁰ Lentz, though unaware of Michael's piece, still in his edition of the Cath. Pr. felt uneasy about John's structure, and changed it by placing (ii) + (i) (i.e. keeping John's order at this point) before (iii): (ii) = 480.10–18; (i) = 480.18–22; (iii) = 481.32–484.14; on 480.10–22 Lentz notes in his apparatus: ‘ex Io. Al. 27.27 huc retraxi’.

³¹ E. Hiller, Quaestiones Herodianeae (Bonn, 1866), 31.

³² This evidence is provided by [Arcadius] and Theognostus, two authors who follow the Cath. Pr. very closely. The relevant material is this: [Arcadius'] Table of Contents 4.9 (ed. M. Schmidt, Ἐπιτομὴ τῆς καθολικῆς προσῳδίας Ἡρωδιανοῦ [Jena, 1860]): τὸ τεσσαρεσκαιδέκατον (sc. βιβλίον) περιέχει πᾶν μονοσύλλαβον ὄνομα; Book 14 appears in [Arcadius] 143.7–146.8. Theognostus treats the orthography of monosyllabic nouns in his Canones 794–816, in Anec. Oxon. vol. 2.131.31–136.4 Cramer, after he had treated the orthography of non-monosyllabic nouns. On Theognostus' dependence on the Cath. Pr. and his method of transforming Herodian's accentuation rules into orthographical ones, see recently Roussou (n. 4), 136–41 and the bibliography cited therein. Roussou (pp. 136–7, citing also a similar argument made by A. Wouters, ‘P. Ant. 2.67: a compendium of Herodian's Περὶ Καθολικῆς Προσῳδίας, Book V’, in P. Naster and H. De Meulenaere and J. Quaegebeur (edd.), Miscellanea in honorem Josephi Vergote [Leuven, 1975–6], 601–13, at 602) expresses doubts about the truth of Theognostus' claim that he employed original Herodian. Her sole basis is the idea that ‘the creation of epitomes as early as the fourth century a.d. is likely to have discontinued the transmission of the original work’. This is certainly too weak an argument to discredit Theognostus' statement. To test the truth of such claims one needs to examine the internal evidence that is available; on this methodological point see also n. 35 below.

³³ He treats the accent of monosyllables on pages 7.16–8.8, and of non-monosyllables on pages 8.8–9.18.

³⁴ We have already seen that the other three surviving epitomes do not contain any material or at least any comparable material on the accentuation of prepositions in anastrophe.

³⁵ Donnet (n. 1), 491 advanced the following argument: ‘La ressemblance entre Philopon et M. le S. est vraiment frappante, et si l'on en conclut que M. le S. se réclame d'Hérodien, c'est qu'il le cite parmi ses sources, mais non Jean Philopon (cf. §1)’. In his argument Donnet seems too ready to take the truth of Michael's claim for granted, and perhaps because of this, he did not see the need to argue for his position on the basis of internal evidence. A by-product of his line of thought is that he missed the opportunity to identify the Herodianic material preserved exclusively by Michael. It is clear that I have followed a different methodology: I have not used Michael's claim as a reason for my conclusion, but as an item that confirms my conclusion, which I have drawn on the basis of other (internal) evidence and after evaluating the relative worth of competing hypotheses.

³⁶ For Choeroboscus see e.g. the Index auctorum et dialectorum s.v. ἡ Καθόλου in A. Hilgard (ed.), Theodosii Alexandrini Canones. Georgii Choerobosci Scholia. Sophronii Patriarchae Alexandrini Excerpta, vol. II (Leipzig, 1894), 439. Roussou's ([n. 4], e.g. 86) suspicion that Choeroboscus had not read the Cath. Pr. in the original is based on the same weak argument as the one which led her to doubt that Theognostus had employed original Herodian: see above n. 32. Theognostus names Herodian's Cath. Pr. as a source for his Περὶ ὀρθογραφίας in the following passage of the dedicatory epistle that precedes his orthographical treatise: … οὓς δὲ ([sc.] τοὺς κανόνας) ἐκ τῆς πολυύλου βίβλου τῆς Καθόλου Ἡρωδιανοῦ ἀναλεξάμενος (Theognost. 69.4, ed. K. Alpers, Theognostos Περὶ ὀρθογραφίας: Überlieferung, Quellen und Text der Kanones 1–84 [Hamburg, 1964]). For Theognostus see also the bibliography cited above in n. 32, and Dyck, A.R., ‘On Etymologicum Genuinum β 120 (65.7 Berger)’, CPh 75.3 (1980), 253–4Google Scholar, at 253.

³⁷ Lentz, A., ‘Das 15. und 21. Buch des Arcadius’, Philologus 19 (1863), 111–8CrossRefGoogle Scholar, at 113. On the theory that resulted from his misjudgement concerning the cohesion of the chapter see below n. 42.

³⁸ ἀναστραφήσονται here bears the second meaning of those listed in n. 1 above: ‘will take a recessive accent’.

³⁹ Hiller (n. 31), 31.

⁴⁰ The Appendix of the Cath. Pr. followed the last book of the work, i.e. Book 20, and is also referred to by some scholars as Book 21.

⁴¹ Hiller did not phrase his hypothesis in such detail, but this is what he seems to mean.

⁴² Lentz (n. 37), 113 (also Lentz [n. 18], lviii), who (wrongly) took the coherence of John's chapter on prepositions to be unproblematic, put forward the theory that the accentuation of anastrophized prepositions was treated twice in the Cath. Pr., once in Book 18 and once in the Appendix, and implies that John reflects the treatment of Book 18. So in his reconstruction of the Cath. Pr., the section on the anastrophized prepositions figures among the contents of Book 18. Roussou (n. 4), 104 (and 76 n. 127), who does not discuss Hiller's position, follows Lentz in thinking that Herodian had dealt with the subject both in Book 18 and the Appendix.

⁴³ Items at 480.16, 480.18, 484.1, 484.4, 484.5, 484.8 belong to the section that will have to be transferred from Book 18 to the Appendix of the Cath. Pr.

⁴⁴ Xenis (n. 27). The six items mentioned in this paragraph have been included in the recently published critical edition G.A. Xenis, Iohannes Alexandrinus. Praecepta Tonica (Berlin/Munich/Boston, 2015).

⁴⁵ Dyck (n. 2), 793.