AN INTERPOLATION FAMILY IN THE POETICS Текст научной статьи по специальности «Языкознание и литературоведение»

mmmiiuh] graecia antiqua hiuuuii

philologia classica vol. 15. fasc. 2. 2020

UDC 821.14

An Interpolation Family in the Poetics*

Michael M. Pozdnev

St. Petersburg State University,

7-9, Universitetskaya nab., St. Petersburg, 199034, Russian Federation; m.pozdnev@spbu.ru

For citation: Pozdnev M. M. An Interpolation Family in the Poetics. Philologia Classica 2020, 15 (2),

173-190. https://doi.org/10.21638/spbu20.2020.201

The ms. tradition of the Poetics is a mine, quite unexpectedly, when it comes to composition on literary matters: four independent witnesses — Parisinus 1741 (A), Riccardianus 46 (B), and mediaeval translations into Latin by William of Moerbecke and Arabic by Abu-Bishr Mat-ta made with the help of a Syriac interlinear (not to mention the recentiores which still could prove of some stemmatic value, as for instance Par. gr. 2038, Vat. gr. 1400, Berol. Philipp. and Mon. 493) — allow in most cases for a safe reconstruction of an archetype. Common errors suggest that this text differed from the autograph in some twenty passages, largely interpolations, ranging from a couple of words to a number of phrases. Several intrusions prove to be typologically close. All of them correct what was deemed to be inaccurate or loose argumentation by inserting syntactic complements or references adding cohesion. As a result, both the style and context go largely neglected. The first paragraphs of ch. 6, central to the Poetics, suffered most. This text also came down to us in a Syriac translation having a heavily glossed uncial ms. as its source. Insertions in ch. 6 cause 'harmonising' additions to the following text of the treatise. The 'family of interpolations' under discussion is tentatively attributed to a professor of Aristotelianism of late antiquity (the most suitable candidate seems to be Themistius): a school-room copy diffused by his pupils became the common ancestor of both the extant Greek mss. of the Poetics and the reconstructed Greek sources of the mediaeval translations. A fresh collation of the Syriac text together with the evidence of variae lectiones in the oldest independent Greek mss. offer a glimpse into the workings of his mind. Keywords: Aristotle's Poetics, manuscript evidence, mediaeval translations, textual criticism.

* The work summarized in this paper is part of the project "Thesaurus criticus to Aristotle's Poetics" supported by the Russian Science Foundation (Project Nr. 18-18-0060). I am deeply indebted to Yury N. Arzhanov for all information concerning the Syriac translation, and to Michael N. Suvorov for his indispensable help in collating the Arabic one. I also want to thank Tatiana Kostyleva for valuable suggestions regarding both form and matter of what follows.

© St. Petersburg State University, 2020

It is but common knowledge that the Aristotelian Poetics reached a wide intellectual public with the 16th century translations and voluminous exegetic commentaries of which the outgoing Cinquecento already saw no less than six.1 Teachers of classics might know that this triumphal comeback had been prepared by the extensive scribal activity of late Byzantine and Italian copyists resulting in some 30 humanist mss.2 previous to the 1508 Aldine editio princeps.3 But probably only experts are aware of the fact that the Poetics was likewise popular with the mediaeval scribes and scholars, however peculiar its content may have appeared to their minds. The text is witnessed astonishingly well for a treatise on a matter in which the Middle Ages displayed little or no interest.4 Apart from some deeply contaminated recentiores that might still prove disputable as regards their provenance,5 the editors have at their disposal two Greek codices of equal stemmatic value, Parisinus gr. 1741 (A) and Riccardianus 46 (B), dating back to the 10th and the 12th centuries respectively, a Latin translation by William of Moerbecke made in 1278 from the ms. independent from either of these (O)6 and an Arabic translation by Abu-Bishr Matta b. Yunus created about the beginning of the 10th c. (Ar) which, again, derives ultimately from a sovereign Greek source (Z).7 To this comes a short fragment of the Syriac translation, cited by Jakob bar Shakko (f!241) in his Book of the dialogues (Syr). The Syriac version is

1 Listed by Schrier (1998, 21-30). The seventh could have been supplied by Gian Vincenzo Pinelli (f 1601), but he did not manage to bring it to print. His ms., Ambrosianus 647 (P 204 sup.), consists of 51 fol. in Quadro covered with notes in Italian. Casual observation has revealed that Pinelli dealt rather with exegesis in general than with particular text issues. Of the "great commentaries" (cf. Weinberg 1953, 100) Vincenzo Maggi (1550) and especially Pietro Vettori (1560) are the most successful as concerns emendation.

2 A total of 39 late ms. dating from ca. mid-15th to the beginning of 19th c. is now registered. 29 were collated by Lobel (1933); Harlfinger and Reitsch (1970, 37-42) added further two. All other collations including that made by Taran for his and Gutas' 2012 "Editio maior" are selective. The collation of all known copies has been accomplished for Thesaurus criticus (see n. 1). The apographs prove to be of no little, should it only be heuristic, value, to say nothing of their importance for the reception studies, of which an instance is given in: Pozdnev 2019.

3 Sicherl 1992, 113-116 supports Lobel 1933, 31-32 in assuming that its primary source has been Parisinus gr. 2038 corrected after Ambrosianus B 78 sup. However, in quite a few cases Ald. coincides with the corrections made in Berolinensis Phillippicus 1599 (196), but differs from the Ambrosianus.

4 Strange as it might seem, no first-hand references to the Poetics by the Byzantine literates have so far been discovered. The thinkers of the Christian West ignored aesthetics. The Arabic poetics was too closely tailored to the local literary and linguistic context to be influenced by Aristotelian critical concepts. The only three original mediaeval philosophers, who make some use of them, largely misapprehending their actual meaning, are al-Farabi, who was a student of Matta in Bagdad, Avicenna and Averroes; their attitude is summarized in: Kemal 1991 and 2003 (here esp. 24-27).

5 This mostly refers to Parisinus gr. 2038 which has recently been discussed in: Taran 2016, in reply to Janko 2013. But Vaticanus gr. 1400, Berol. Philippicus and Monacensis 493, to cite just the most pertinent examples, also display peculiar readings and corrections probably inferred from an unknown apograph of A, or, less probably, O, but not B which, contrary to Lobel (1933, 18; 30-31), was unknown to the humanist scribes. Corrections by the later hand could derive from the printed editions.

6 Moerbecke's text (edited in 1953 by Minio-Paluello, Arist. Lat. vol. 33) is transmitted by two mss. (Toletanus Capit. 47.10, ca. 1280, and Etonensis 129, ca. 1300) which agree with each other in nearly all problematic cases.

7 The ms. is the celebrated Parisinus ar. 2346 (Ancien fonds Arabe 882 a) of the 11th c., first published by Margoliouth (1887). On which see Gutas 1998, 147. The monumental posthumous edition by Tkatsch (Vol. I 1928, Vol. II 1932) is at some places supplemented and corrected by Gutas in his notes to the 'Editio maior'.

supposed to have been the basis for the Arabic and thus the earliest surviving document of the tradition.8

The main branches of the paradosis are geographically and culturally dispersed so as to create an impression of far-reaching openness. There certainly existed an archetype (w9) other than in the hand of Aristotle himself or his writing attendants: a number of mistakes and omissions common to all mentioned witnesses are of a later origin; some of these we examine below in detail, and some are listed in the last footnote. But what could its nature possibly be? Was it a Late Ancient recension or a Byzantine school copy? Given the present state of knowledge, every date between ca. 80-60 BC (the famous Tyrannion's edition of Aristotle's and Theophrast's writings reported by Strabo10) and the mid-9th c. AD (the latest possible date of the Greek ms. used by the Syriac translator11) can be considered. There remains uncertainty concerning the script of the hypothetic protograph. Was it in uncial letters or, perhaps, already in minuscule? Did it contain glosses or other comments in any form? To clarify all this is clearly a demanding task, as multiple transmission-indicative passages should be scrutinized. Yet there are some cases that seem to be especially instructive for elucidating the appearance of the common ancestor, while the data of textual criticism at our disposal are vast enough to promptly provide additional evidence. One such case is examined below. It is a series of analogous interpolations traceable in more than one passage of the Poetics, but mainly in the paragraphs following the definition of tragedy in chapter VI. This section is witnessed by all the aforementioned sources including the Syriac. It has been the object of most attention since the earliest stage of text reproduction:12 any scholar or copyist, ancient or mediaeval, could supply

8 As follows from the analysis by Gutas (2012, 108-109), partly confirmed by the collation carried out anew for Thesaurus criticus (see n. 1 above), the relationship between Ar. and Syr. is more complicated than usually assumed. That Abu-Bishr in some of his unfortunate renderings follows the Syriac source is fairly apparent. But it is likewise clear that other erroneous readings of Syr. are avoided in Ar.; for more details see below, n. 12, 15, 31 and 46.

9 H in Taran's sigla, though the majuscule, perhaps, is more appropriate for the text written or dictated by Aristotle himself.

10 13.1.54. The editions produced by pipXionwXai tivsc; ypa^suai cpauXoic; xpwpsvoi Kai ouk avTipaAAovT«; (if we should believe this part of Strabo's story) must have become obsolete after the corpus was edited anew by Andronicus who collated a lot of antigrapha and inscribed the rolls: Plut. Sulla 26; Porph. Vit. Plot. 24; for the discussion see Hatzimichali 2013, 15-23. That a copy of the Poetics in Andron-ican recension has survived to become an ancestor of A, B and £ is barely thinkable. But in any case, it might have been one of its descendants scattered in the ancient libraries, from which a number of mistakes and omissions (as in 1456a2-3 discussed below) were transmitted to the source of w.

11 At present there is no certainty as to the terminus ante quem of Syr. and, respectively, of £. The only thing established so far is that, contrary to the earlier belief, the Syriac translation cannot be attributed to Ishaq b. Hunain, who flourished in the second half of the 9th c. In a tentative 'scenario' Gutas considers the late 8th c. as the approximate date of the Syriac translation. By that time the copies of Greek ms. containing the Poetics were kept in "the libraries of the Nestorian Christian monasteries in northern Iraq" (Gutas 2012, 108). Yury N. Arzhanov, who is about to give a new edition offragmentum Syriacum, believes that by 780s the translation was already completed. The dating of Syr. largely relies on the interpretation of Patriarch Timothy's letters (Ibid. 80, test. 3-4) and cannot be discussed here at length. Anyway, there is all ground to conclude that the Eastern branch of transmission begins to ramify around 700, and thus the time span for an archetype in question should be narrowed to ca. 1st-7th c. which is an extended period indeed.

12 The interest in enigmatic catharsis-formula is manifest already in neoplatonic scholarship (Iam-bl. Myst. 1, 11; Procl. In Plat. Rem publ. 1, 42, 10-16; Olymp. In Alcib. 54-55). It is then reflected in both Syriac and Arabic interpretative renderings of 1449b28-29, but also e. g. in a gloss on ¿\ssivwv (1452b32-33 tautr|v ^opspwv Kai ¿\ssivwv dvai |ii|ir|TiKr|v) included in the Arabic text (Tkatsch 1928, 246).

what he considered an improvement on the Aristotelian text and thought. In the following an attempt is made to detect traces of that work.

The first passage in need of revision is close to the beginning of the chapter. Having defined tragedy in a rather peculiar way, Aristotle then takes some care to clarify what is meant by ^Suo^vw Xoyw xwpi; ¿Kaaxw xwv eiSwv ¿v xot; |opioit; (the second colon of the definition, 1449b25-26). This surely needs clarification, since ^Sua^evw is metaphoric and the subordinate modifiers are arranged in an almost perplexing manner: it is quite clear, why both A and B read eraaxou, and the Eastern translations perceive X^piq as a preposition, the implication being that Z could have had eraaxou as well. The mistake13 caused by irregularity of expression is little informative for the filiation, especially as O seems to have escaped it: Moerbecke translates by seorsum, not by sine.14 Equally confusing is the final syntagma. While |opia (viz. x^; xpaywSiac;) might be readily understood as 'smaller parts', or sections, of tragedy like prologue or stasimon, ei'S^ is vague enough to encompass any meaning, either technical or not. Aristotle explains that 'sweetened' signifies metrical and musical forms of speech: Se ^Suo^evov |ev Xoyov xov exovxa puG|ov Kai ap|oviav Kai ^¿Xoc; (1449b28-29). "Harmony" normally goes together with "rhythm" to signify two main musical means. Illustrative parallels in the Poetics are 1447a23-24 (ap|ovia |ev Kai puG|W xpw^evat |ovov ^ xe auX^xiK^ Kai ^ KiGaptaxtK^) and 1448b20-21 (raxa ^uaiv Se ovxo; %tv xou |i|etaGai Kai x^; apiovia; Kai xou puG|ou); cf. Pol. 1340b17; 1341b19; Rhet. 1403b17. But ^¿Xo; cannot be attached to puG^o; Kai ap|ovia in a way it stands in the above cited text. Aoyo; o exei ^¿Xoc; is an impossible expression. "Rhythm, harmony and song' looks absurd, because song virtually is speech embellished by rhythm and harmony. Kassel was right in following Tyrwhitt who athetized Kai ^¿Xoc;, Gallavotti and Taran wrongly let it stay.15 The interpolation common to all witnesses could not have been made independently by A, B, O and Z. Supposedly, Kai ^¿Xoc; goes back to the maker of «, whose reasons are shortly to be determined.

The rhythm, or metre (cf. 1448b21: xa yap (lexpa oxi |opia xwv puG|wv ¿axi ^avepov), can effortlessly make its work without the aid of harmony (cf. 1447a26-27: auxw Se xw

13 Emended by Reiz (1786, 12) and then by Tyrwhitt in the significant 1794 edition (pp. 137-138); the correct meaning is supposed already by Alessandro dei Pazzi's translation (a 1536 Aldine ed.) "separatim singulis generibus in partibus agentibus", though the parallel Greek text edited by his son Guillelmo reproduces the vulgate reading.

14 Moerbecke's translation is very different from that of Matta. The latter interprets almost every pattern he pretends to have understood, expands the original and glosses on it in trying to convey the meaning of things he had never heard of. What he creates is a chaotic combination of literalism and abstraction. In both he could have been misguided by the Syriac source (which he sometimes misreads). The Arabic of his translation is so bewilderingly confused that in many places the text is hardly readable at all. It must have been typical of all Matta's translations (of which only the Poetics survives), and he was justly blamed for that by his contemporaries: Margoliouth 1905, 86. The text might have been an outline intended for those who were trained in Matta's school. On the contrary, Moerbecke is strict and cold. He uses no intermediaries and obviously does not need any. He strongly adheres to the style of mirror-translating, mastering it well enough to make an impression that he understands most of the patterns he writes about. In this he is greatly helped by the structure of Latin which is far closer to Greek than the Semitic languages. His wording is thoughtful but rather poorly chosen (cf. the indices in: Minio-Paluello 1953) and could serve the reconstruction of small words with far more certainty than that of Matta.

15 Gallavotti is almost obsessively conservative, but Kai ^¿Xoc; is retained also by Dupont-Roc/Lallo (1980) and Guastini (2010), to name the editors of the last decades. Vettori was the first to suspect ^¿Xoc;. He replaced it with |i£tpov. Cf. Tyrwhitt 1794, 140: "Victorius pro ^¿Xo; legit |i£tpov, quia ^¿Xo; et ap|iovia idem valent. Sed etiam |^rpov et puG|o; idem valent. .. .Voces igitur Kai ^¿Xo; potius delendas credo, ut ex glossemate ortas." After Tyrwhitt the athetesis was argued for by Spengel 1841, 1263.

puBpœ pipeirai xwpl; âppoviat; ^ xœv opx^awv). All segments of tragedy are metrical but not all of them musical. This should help us to grasp the sense of a second Aristotelian comment immediately following the first one: to ôè xwpl; Tot; eïôeai to ôtà péTpœv evia povov nepaiveaBai Kal ndAiv §repa ôtà péAout; (1449b29-31, here the instrumental dative is reproduced by A, B and O thus justifying ÉKdaTœ in the preceding colon, whereas Syr. and Ar. render the text, again, as if it had xwpl; ™v eiôœv). 'Metres' are substituted for 'rhythm', and thus one could infer (as Bywater actually did16) that 'song' is synonymous with 'harmony'. Yet, péAot; comprises both harmony and rhythm. Choral parts and monodies are, of course, metrical. Aware of this difference the interpolator pedantically added Kal péAot; to supposedly insufficient puBpov Kal âppoviav.17 The question is, whether he inferred the words himself or was it a transfer, more or less mechanical, from an annotated text, the motifs of an annotator being then purely didactic. The lengthier interpolations examined below might shed some light on this point.

In the lines under focus popia (let out elliptically after evia and §repa) exhibits no cladistic meaning, even less than in 1448b21, where it has the same etymological sense of a 'smaller part': metre is not a species,18 but one of the constituents, or properties, of rhythm (others being, for instance, tempo and intensity). Neither do eïôn bear any specific connotation. Aristotle avoids saying anything like "there are two kinds of speech embellishments"; he does not itemize eïôn Tœv ^ôuapdTœv in a way he enumerates and specifies Tpayœôia; eïôn in chapter XVIII, 1455b32-56a3. Still, eïôn denoting 'species', as in ch. XVIII, is so characteristic of his vocabulary that the word, especially accompanied by popia, is not so lightly passed by: it lingers in the reader's mind ready to be remembered on an appropriate occasion.

Next a reader is offered several appendices to the core definition. The description of the six 'parts, or 'elements', of tragedy is shaped in similar syntactic structures and remarkably rich in repetitions (cf. 1449b37: ou; avdyK^ noioùç Tiva; eïvai, 49b38-50a1: Tà; npd^ei; eïvai ^apev noid; Tiva;, 50a6-7: KaG' o noioùç Tiva; eïvai ^apev toû; npdrrovTac;, and then again, 50a8-9: KaG' o noid Ti; èaxtv ^ Tpayœôia), some of them doubtlessly aucto-rial, others perhaps not. Initially, Aristotle employs the same term he has just used for the formal subdivisions of tragic play: ènel ôè npdrrovTec; noioùvTai T^v pipnaiv, npœTov pèv è£ avdyK^; av eï'n ti popiov Tpayœôia; ô t^; ô^eœç Koapo; (1449b32-33). 19 The iteration

16 Bywater 1909, 161: "The Kai is explanatory, péAoc; being added as a synonym to explain âppovia". In retaining Kal péAoc; Bywater is misled by the doubtful parallels he had found for péAoc; Kal péTpa (Probl. 920a12: noAAanAdaia... Tà péAr| èv Tat; Tpaywôiai; twv péTpwv; Plat. Symp. 187d; Lysis 205a; Isocr. 191b). He seems to assume that Aristotle wanted to explain a more specific term with a common one. However, it is hardly possible to explain âppovia with péAoc;: song is rhythm enriched by harmony. Âppovia is used by Aristotle in numerous difficult contexts without any explanation; âppovia Kal puBpo; ("les élément mélodique et rythmique", as Dupont-Roc and Lallot elegantly put it) is utterly self-sufficient.

17 In the similar way Kal péTpov is inferred in ch. I, 1447b25: oîov puBpw Kal péAei Kal pÉTpw. Citing the tricolon Brescia (1984, 58-59) asserts that Kal péAoc; should be retained. But Kal péTpw in 47b25, contrary to Kal péAo; in 49b29, is not out of place, if we consider the examples that follow: ^ Te twv ÔiBupappiKÛv noi^ai; Kal ^ twv vopwv Kal ^ Te Tpaywôia Kal rç Kwpwôia. After nomos which can be pure music Aristotle mentions the dramatic genres, and thus needs a specific element of rhythm appliable to them.

18 Cf. Bywater 1909, 11: "Imitation, then, being natural to us — as also the sense of harmony and rhythm, the metres being obviously species of rhythms." In contrast to other constituents of rhythm metres are 'fractional' (Tpr|Td) and thus should not be used in oration (Rhet. 1408b29, Bywater's conjecture Tpr|paTa is misleading; cf. the commentary by Lucas 1972, 74).

19 Aristotle makes little effort to fully elaborate his thought: the reader is left to conjecture, why the non-narrative mode of delivery (npdrrovTec;) presupposes visualization (ô^ew; Koapo; must be peri-

seems to be deliberate, albeit not intentional, as Aristotle adds xi, presumably to discern |opiov, here meaning 'constituent', from |opia in the former paragraph. He then decides in favour of variety choosing ^¿pn to define basic components of the tragic genre. These he catalogues briefly, building on the classification of mimesis through media, modes and objects discussed in ch. I (1447a16-17).

'Music and word' are perfectly comprehensible as mimetic means; eixa |eXonoiia Kai X¿^l;, ¿v xouxoit; yap noiouvxai x^v |i|naiv (1449b33-34) needs no comment. The next following explanation of X¿^lc; and |eXonoiia is not only superfluous, but bizarre in thought and weird in expression: X¿Y« Se X¿^lv |ev aux^v x^v xwv ^¿xpwv auvGeaiv, |eXonoiiav Se o x^v Suva|iv ^avepav exei naaav (1449b34-36). Clearly, X¿^lc; can signify both poetic diction (as in 1449a23) and spoken dialect (1459a12, cf. Rhet. 1408b33). But, contrary to Bywater and others, it cannot possibly mean the "composition of verses", should it be used "in reference to the poet"20 or in any other reference. A few paragraphs later in the same chapter Aristotle gives an acute general definition of X¿^lc; preceded by a back-reference: X¿Y« S¿, wonep npoxepov ei'pnxai, X¿^lv eivai x^v Sia x^; ovo|aaia; epmveiav, o Kai ¿ni xwv ¿^¿xpwv Kai ¿ni xwv Xoywv exei x^v aux^v Suva|iv (1450b14-15). "If the reference is to 6, 1449b34", Bywater comments, "Aristotle must have forgotten [!] the exact terms of his previous formula." Still, to quote Bywater's next utterance, "the point is the same in both instances".21 This sounds even weirder than the above cited 'Aristotelian' passage. The "speech" that "interprets by giving names", and thus "has the same function in verse and prose" is evidently quite another thing than ^ xwv ^¿xpwv ouvGeoi;, that is to say, the versification of speech. wonep npoxepov ei'pnxai could refer to language as one of mimetic media. Much more credible, however, is that it is put forth by an interpolator with no other aim than to recall his own self-made definition.

The logic of X¿^l; defined as ^ xwv ^¿xpwv ouvGeoi; is easily discernible, indeed if we discredit its genuineness. Music encompasses rhythm, resp. metre. The reader easily infers it, because he was previously informed on what the speech embellishments are. But metre is formally absent, tragic play, we are told, has no media except |eXonoiia Kai X¿^lc;. In an attempt to improve on this seeming shortage, the interpolator supplied a notion of X¿£l; which included metres. He modelled it on the neighbouring phrase, X¿Y« yap |uGov xouxov x^v auvGeaiv xwv npay|dx«v (1450a4-5; xwv ^¿xpwv ouvGeoi; mirrors ouvGeoi; xwv npay|dxwv), but worded in such a way as to stress his point: aux^v empha-

phrastic), especially as he knows that a tragedy goes perfectly well "without movement", its quality being clear "from mere reading" (1462a10-12). Lucas (1972, 99) might be right in interpreting mpdrrovre; as "the visible actors". A piece can be judged by reading but it is always (at least, to Aristotle's mind) written for staging. It is this uncertainty of expression that induces insertions.

20 Bywater 1909, 162. To support this Bywater has to conjecture X¿í;lv |iev raurnv. Lucas (1972, 99) follows Bywater without accepting тaírrr|v: "The whole is short for x^v xwv ovoidTOv ¿v auvGeaiv". However, letting out the main term is a strange way to shorten, and the accent made on verse would, by all means, contradict the definition given in 1450b14-15. Gottfried Hermann boldly emends aux^v x^v xwv ovo^dx^ auvGeaiv (1802, 16; cf. 116: "Quid metra hic sibi volunt?" etc.). This matches well with ch. XXII, 1458a28-29: Kaxa |ev ouv x^v xwv <aXX«v> ovo^dx^ auvGeaiv oux oiov xe xouxo noirjaai, Kaxa Se x^v ^exa^opwv ¿v5¿xeral — no matter whether we read aXXwv with Z, as Kassel and Taran prefer to, or side with Bywater (1909, 294) in taking ovo^dx^ for Kupiwv ovo^dx^. But pace Hermann there is no palae-ographical support to his suggestion: even if contracted ovo^dx^ could not be taken for |^xp«v. Besides, ^ xwv ovo^dx^ auvGeai; would be a simplistic definition both for 'speech' and 'literary composition' (as in Dionysius of Halicarnassus; the £p|r|veia of Demetrius is evidently a more 'Aristotelian' term for that, cf. Poet. 1450b14 cited above).

21 Bywater 1909, 175.

sizes t^v twv ^éxpwv aúvGeatv ("the mere metrical arrangement of the words" [Butcher], "l'agencement même des mètres" [Dupont-Roc/Lallot]).22 Having afterwards encountered the original definition he supplied the reminiscence cliché, perfectly careless about how to reconcile the two concepts. Moreover, he risked a further enhancement to the text he had before his eyes. In 1450b12-13, immediately before the back-reference, all ms. evidence gives TÉTaptov ôè twv ^èv Àôywv ^ Àé^iç. Desperate to grasp the point of twv ^èv Àôywv Kassel prints it sub cruce. The "speaking of words" is indeed pointless, but "speech as related to words", with defining genitive (as implied by ^év23), makes sense if opposed to the previously suggested idea of speech as dealing with metres. Tarán's decision to exclude it appears correct: the interpolator's obstinate endeavour to coordinate individual sections of the treatise as regards terminology is fairly manifest in wanep npótspov eïpr|Tai and will be observed again later.

The unassuming account of ^sÀonoiia as something "of which every function is manifest" is due to the same scholastic punctiliousness. With no new information provided (cf. 1450b15-16: twv ôè Àoinwv ^ ^sÀonoiia ^éyiaTov twv ^ôua^ôtwv) it is a mere pendant to the definition of Àé^iç. Besides, it is oddly formulated, nâaav being at times misread as nâaiv (and despondently replaced by it in the 1550 Maggi-Lombardi edition; cf. "a tutti puo essere chiaro" [Guastini], "whose sense every one understands" [Butcher]) or misinterpreted as nàvTwç ("parfaitement claire" [Dupont-Roc/Lallot]). nâaa ôùva^iç is not altogether alien to Aristotle (cf. SE 172a36: npoç anaaav téxv^v Kai ôùva^iv), but nâç does not go well with ôùva^iç, if the latter means 'function', or 'significance', as seems to be the case here.24 Yet, it becomes usable, if one perceives music as consisting of "rhythm, harmony and melos" and accordingly disintegrates its functions to evasively declare that none of these actually need explanation.

The syntax of the next lines (1449b36-1450a10, where the Syriac fragment ends), is confused to the point of incoherence; attempts to isolate interpolations have proved to be unsuccessful, though it is very unlikely that the text is authentic in every word. The paragraph starts with a protasis marker, but opinions widely differ as to where the protasis ends. Bywater punctuates the text as follows:

snsi ôè npd^swç sari ^raiç, npáttstai ôè ûnô tivwv npattóvtwv, ouç áváyKr noioùç Tivaç slvai Katá ts to r|9oç Kai t^v ôiàvoiav, ôià yàp toútwv Kai tàç npà^aç slvai fa^sv noiàç tivaç, néfUKsv a'itia ôùo tâ>v npá^swv slvai, ôiàvoia Kai r|9oç, Kai Katà taûtaç Kai tuYxávouai Kai ánotuyxávouai nàvtsç. sativ ôè t^ç ^èv npá^sw; ó ^ù9oç ^ ^r|aiç- Àéyw yàp ^ù9ov toùtov t^v aúv9saiv tâ>v npay^átwv, tà ôè ^9n, Ka9' o noioùç tivaç slvai fa^sv toùç пpáttovtac, ô^voiav ôé, sv oaoiç ÀÉYovtsç ànoôsiKvùaaiv ti ^ Kai ànofaivovtai

22 Else (1957, 236) tries to save aür^v r^v rwv psrpwv aüvGsaiv by reinterpreting afrrr|v: "'just', i.e. that very composition of the verses that was implied by 5ia psrpwv nspaivsaGai. As^ic;, then, is the composition of the spoken verses, the dialogue." This was objected to by Verdenius (1960, 257): "(1) 'just' is a very awkward way of referring to b30 and (2) it is very improbable that Xs^i; should be confined to the dialogue. The fact that in b30 povov has been added to 5ia psrpwv shows that pstpov in itself does not denote a dialogue."

23 In a late humanist Cod. Ferrarensis Cl. II. 348 psv is left out: its maker must have been embarrassed by the emphasis laid on rwv Xoywv. Normally, the learned copyist follows Vettori, but in this case he took the decision upon himself.

24 Cf. Dupont-Roc/Lallot 1980, 53. Schmitt (2008, 9) fancifully translates: "dessen wirkende Kraft ganz im Äußeren präsent ist", but this is too complicated for a clarification, the pregnant meaning of cpavspav requiring a further comment.

Yvw||v. avdYKr| ouv t^; tpaY^Sia; ^¿pn elvai KaG' o noid xi; ¿axiv ^ xpaY^Sta-

xauxa S' ¿axi |uGo; Kai ^Gn Kai X¿^l; Kai Sidvoia Kai Oyi; Kai |£Xonoiia.

Bywater, notably followed by Taran, renders Sia yap xouxwv... noid; xiva; parenthetical, and begins the apodosis with ^¿^uKev. Else extends the parenthesis to anoxuyxdvouai ndvxe;, athetizes n^uKev... nGo; (1450a1-2),25 and stretches the ¿nei clause out up to ano^aivovxai yv«mv after which he puts a dash mark, apparently perceiving avdyKn ouv kxX. as the main clause. In fact, those are the only two options we have.26 But neither is eligible. As to the suggestion of Else accepted by Kassel, it is hardly feasible that from the speculations on characters and plot of tragedy there should follow that their constituents are six in number. avdyKn ouv announces a new thought sequence which briefly summarizes the whole section on ^¿pn: the number of 'parts' is given and then expanded in a list (put in brackets by Else but, inconsequently enough, not by Kassel). Bywater's solution seems more attractive, as it shortens the protasis and avoids elimination. Yet the reasons Aristotle gives for introducing Sidvoia and nGo; into his analysis of tragedy, namely that the acting persons will necessarily exhibit certain moral qualities and make certain judgements, would provide no premise for asserting that there happen to be just two causes of action, and that it is by their actions that everyone's successes and failures are conditioned.

Accordingly, we have either to consider the text incurably corrupt, or to venture an emendation. Aristotle's primary concern is to introduce two new evaluation criteria. Hence ou; can be smoothly replaced with xouxou;, the corruption probably resulting from haplography with subsequent correlative correction (xouxou; > xou; > ou;). Cf. Poet. 1448a2: ¿nei Se |i|ouvxai ol |i|ou|evoi npaxxovxa;, avayKn Se xouxou; n anouSaiou; n ^auXou; elvai; Rhet. 1365b36: avayKn Se xouxou; ^aiveaGai apiaxou;; De gen. et corr. 332a5: ei yap ¿axi xwv ^uaiKwv a«|ax«v uX|..., avayKn ^xoi ev n Suo elvai xauxa n nXeiw. The conditional clause, then, ends with Kai x^v Sidvoiav, Sia yap xouxwv delivering an additional argument, quite naturally, in a separate colon. The next segment is meant to be parenthetical, since the connective particle is lacking. n^uKev eqs. cannot be straightforwardly athetized, least of all because it "was intended as a paraphrase of Sia yap xouxwv Kai xa; npd^ei; elvai ^a|ev noid; xiva;".27 What the parenthesis actually aims at, is to supply a passing comment on why actions are commonly qualified by characters and thoughts of the agents.28 Its first section points out at the causes of actions (the accent lies on xwv npd^ewv), while the second (with slightly adversative Kai) deals with their results. Having removed the n^uKev clause one should concomitantly eliminate Kai Kaxa xauxa; Kai xuyxdvouai Kai anoxuyxdvouai ndvxe; which is authentic enough: xuyxdvw should be considered synonymous with ¿mxuyxdvw which in the Aristotelian corpus is used elsewhere in the same combination (EE 1247a36). The text seems not sound, but its content is adequate, the reason for bracketing it away being merely formal.

The main obstacle to rendering the afore-cited section coherent (provided we read xouxou;) appears to be the appositive, or rather parenthetical, nominative Sidvoia Kai

25 See Else 1957, 240, with n. 68. Else gives a sound criticism of the attempts of Vahlen and Gomperz to transpose the segment.

26 For the overlong protasis before dvaYKr| ouv cf. Phys. 234b15. A full stop before dvaYKr| (Du-pont-Roc/Lallot) would not work: the conclusion would be lacking. Placing a colon after dvai cpa|i£v noia; Tivac; (Gallavotti 1974, 20) is an absurdity.

27 Else 1957, 240.

28 The detailed analysis is given in Schmitt 2008, 354-357.

nGo;. It fits not with impersonal n^uKev, comma before Sidvoia does not rule out the problem, but rather creates a new one, since it becomes less clear what the fem. xauxa; refers to. Kaxa xauxa; (sc. npd^ei;) Kai xuyxdvouai Kai anoxuYxdvouai ndvxe; is closely paralleled by Kaxa Se xa; npd^ei; euSai|ove; n xouvavxiov (1450a19-20),29 but B has Kaxa xauxa and the Syrian translates likewise30 possibly following the mistake of his source. We should therefore accept Sidvoiav which is the reading of A. However, it is an easier reading and looks very much like a copyist's correction. Suppose Aristotle has omitted the subject, for which a'ixia Suo affords a predicate. Having affirmed that actions are qualified by character and thought, he goes on to say that these are by their nature the two causes of action. The interpolation technique, into which some insight has been gained above, consists in filling up the seeming gaps with gloss-like additions. Thus, an interpolator would consider that explanatory Sidvoia Kai nGo; is to be added,31 regardless of how irregular such a parenthesis might be. This neglect of syntactic framework will be illustrated below by further examples.

The insertions identified so far are, of course, purely conjectural. If the uncorrupted text was previous to the archetype in question, it can hardly be witnessed anywhere except a secondary source. Yet we are almost completely ignorant about the number of intermediaries between Aristotle and w32 or between the latter and our earliest mss. Some interpolations might have been originally added over the line to be inserted in the main text later. With Kai ^¿Xo; this could easily be the case, and same is true about Sidvoia Kai nGo;. Hope remains, therefore, to detach supplements that left traces in the ms. tradition, all the more so because the tradition is, as seen above, remarkably ramified.

The Aristotelian account of "the parts of tragedy" closes with the enumeration linked to the initial divisions of mimesis, 1450a9-12: ol; |ev yap |i|ouvxai, Suo ^¿pn ¿axiv, Se |i|ouvxai, ev, a Se |i|ouvxai, xpia, Kai napa xauxa ouS¿v. The subsequent text (1450a12-15) in Kassel's edition resembles a graveyard:

toutoic; |ev ouv touK 6XiYoi auxwvt elnetv ^xpnvxai xot; dSsaiv- Kai Yap toyi; sxe

navt Kai nGo; Kai |uGov Kai X¿^lv Kai ^¿Xoc Kai Sidvoiav «oatix«;.

29 Cf. Valgimigli 1937, 7.

30 The same is to be found in Matta's version, while Moerbecke translates secundum has. In fact, it is just one of many cases where Z displays common mistakes with B against A, O (n). The following sampling from cap. I-XII illustrates that the current consensus which makes all extant Greek mss. descendants of a common ancestor (Taran's H) independent of Z must not necessarily be followed: 1448b26: xa; xwv TOiou-xwv n : xwv om. B, Z; 1449a7: ¿niaKonetv na^xei n : ¿niaKonetv apa exei B, Z; 1449b9-10: |^xpou |eYdXou n : |^xpou ^exa Xoyou B, Z; 1449b26: anaYYeXia; n : ¿naYYeXia; B, Z; 1450a4: o |uGo; ^ |i|nai; n : ^ om. B, Z; 1450a35: an|etov 6x1 Kai n : Kai om. B, Z; 1452a33: nepiroxeiai Yivovxai n : nepiroxeia Y¿vntal B, Z; 1452b17: axia^o; B, Z : axiai^^ n. See also below, on Poet. XIX, 1456a33-34.

31 In much the same manner eXeo; |ev nepi xov avd^iov, cpopo; Se nepi xov 6|oiov (1453a5-6) is added after o |ev Yap nepi xov ava^iov ¿axiv Suatuxouvta, o Se nepi xov 6|oiov.

32 For the pre-Andronican (Apellicon's?) editions see Hatzimichali 2013. To what extent the non-Aristotelian material was incorporated in the Andronicus edition, is not our task to discuss. F. Grayeff (1956) assumes that the text of the Poetics then published was to a great extent a mixture of Aristotle's thought and late peripatetic additions, but his analysis of ch. IX 1451a37-b32, a well-known passage on noinai; / laxopia and on 'giving names', condemned as incongruous and thus partly an interpolation, is not very convincing: see esp. 1956, 112-113 where he reinterprets 1451b19-21, ou aXXa Kai ¿v xat; xpaYwSiau; ¿v ¿viai; |ev ev n Suo xwv Yvwpi|wv ¿axiv 6vo|dtwv eqs. The 'harmonising' bias Grayeff points out at is in fact palpable throughout the treatise. But to discern the genuine links resulting from the author's wish to string together parts of a larger whole from what is "incongruous and merely 'harmonized'" one should base on the actual incongruities in the transmitted text and not on the faulty notions he pretends to read into it.

There is enough reason to side with Kassel in considering the passage largely inexplicable, even if the segments he obelizes might actually be explained away or emended by conjecture. oük óAíyoi aüxwv is unique in Aristotle and does not look Aristotelian, no more than the nonchalant omission of the substantive (Spa^dxwv rather than noirfrwv33) scarcely smoothed by 6c; eineiv.34 Apparently, óyi; e^ei nav needs to be cured: both ó^ei; and óyiv can be found already in the late mss., and the editions differ respectively. Yet the problem recognized by Kassel lies not here, but in nav, evidently meaning nav Spa^a. Even if Aristotle would think of Spa^a as a plain synonym for tragedy in this context, he would hardly have omitted it. Apart from being useless for the analysis, the whole phrase is painfully trivial and consists mostly of repetition. Together with stylistic discrepancies this more than suffices to consider it an insertion, though question remains as to its purpose, since it does not at all look like a gloss nor, at least prima facie, like a correction of an alleged incongruity.

To answer this, we have to take a closer look at el'Seaiv. The use of elSr in lieu of ^épr| is evidently a mistake. That these are not interchangeable follows from ch. XXVI 1462b16-18: nepi ^ev oiv xpaywSia; Kai ¿nonoiíac;, Kai aurwv Kai xwv eiSwv Kai xwv ^epwv, Kai nóaa Kai tí Sia^épei <...> etp^aGw xoaaüxa. What Aristotle means by xpaywSia; e'iSr| is lucidly defined in ch. XVIII, 1455b32-1456a3, the section being crucial for the understanding of how w could have been created. In Tarán's edition the text runs as follows:

xpaywSia; Se síSr| s'iai xéaaapa [xoaauxa yap Kai xa ^épr| ¿AéxQr], H ^ev nsnAsy^évr, r|C to oAov ¿axiv nspinéxaa Kai avayvwpiaic;, H Se naGrxiK^, oíov oí xe A'íavxs; Kai oí 'I^íovs;, H Se oíov aí OGiámSs; Kai ó nrAsú; to Se xéxapxov H ánA^, oíov aí xe OopKÍSs; Kai

ó npo^nGeu; Kai oaa ¿v aSou.

At first, a note is required about H ánA^ which is not a manuscript reading but an emendation by Conrad Bursian.35 Actually, the idea is some three centuries older: ánA^ law; ^ ó^aAóv stands in the margin of the 16th century Codex Riccardianus 16.36 ó^aAóv

33 This is presupposed by the following nav. According to Else (1957, 249), "the rationale ofthe passage demands that it be the dramatic characters". Else athetizes oük oAlyoi aürwv w; sinsiv. Butcher, whom he claims to follow, prints <ndvt£;> [oük oAlyoi aürwv] w; sinsiv.

34 Placing it after nav, as Bywater tentatively suggests in the apparatus, would deprive it of its justifying force. Claas Lattmann (2015, 266-267) applies w; sinsiv to KsxpnvTai, the implication being that the poets use the elements of tragedy insufficiently: "Sie ,nutzen' sie nur, insofern sie sich ausschließlich der öyi; widmen, die ja qua Modus der Tragödie alles beinhaltet; eine intendierte und zielgerichtete Nutzung im eigentlichen Sinne erfolgt jedoch — anders, als es in Aristoteles' Augen notwendig ware — nicht. Der Einschub w; eikeiv dient damit mehr oder weniger als Ironiesignal — und spielt zum Zweck des argumentativen Effekts am Beginn des neuen Unterabschnitts zugleich mit der inhaltlichen Paradoxie, die die Aussage im Kontext impliziert und die fur die bisherige Forschung eben der Stein des Anstoßes war: Nicht wenige ,nutzen' alle qualitativen Teile, nutzen sie aber nicht." This is inventive enough, but the parallels do not convince: w; sinsiv would hardly suite to non-figurative expressions and customary words like xpdopai or give them a special connotation, while its use (in the sense of 'generally speaking') to qualify ndvTs;, otiSsv, napnoXXoi (= oük oAlyoi) and the like generalizing quantitative words is supported by dozens of passages. Hence Bursian (1859, 754) conjectured oük oAlyoi dAAa HdvT«;, "so dass in w; sinsiv eine Beschränkung des vorliegenden ndvTs; liegt, vgl. Pol. 1328b15; Plat. Alcib. 105c". But suppose w; sinsiv is to be taken prospectively. The skeptical overtone in Lattmann's reading of it is hardly supported by the harmless enumeration, with no discernible stress on o^i;, presented in the next phrase.

35 Bursian 1859, 757; Taran (2012, 280) wrongly ascribes it to Susemihl. Schrader suggested rsparw-5s;, which has gained much approval among the late 19th c. editors: Bywater 1909, 250.

36 Morel — if the notes to the 1555 Paris edition printed by him were by his hand — proposed to read dnXouv (not önÄfj, as in the reference by Pia Pattoni 2012, 159, n. 5), see [Anonymus] 1555, 83.

(sc. eíSo;) which brings forth the same concept has been inferred by several late mss., and accepted in the once famous edition of Charles Batteux.37 Originally it was meant as a correction of the inexplicable form ón; that we find in A. ánAí correlates with the previously named nenAey|évn and participates among the 'species' of epics enumerated in ch. XXIV with explicit allusion to xpaywSía; ei'Sn, 1459b8-11: xa e'íSn xaüxa Set exeiv x^v énonoiíav xfl xpaywSía, n yap ánA^v n nen\ey^evr|v n ^G^v n naGnxtK^v. Kai xa |épn e^w |eAonoiía; Kai oyew; xaüxá^ (Note that ei'Sn and |épn are juxtaposed and oyi; listed among the latter.) Unmistakably, the fourth 'kind' of tragedy is meant to be ánAí. And yet, it turns to be out of place in view of the examples that follow. One might try to argue that Prometheus Bound (if this Aeschylean tragedy is meant) and the Satyr play The Daughters of Phorcys by the same author (F 261-262 Radt) lack peripety and recognition (cf. Poet. 1452a14-17: Aéyw Se ánA^v |ev npá^iv ^c yiyvo|évn;... aveu nepinexeía; n ávayvwpta|oü ñ ^exá^aat; yíyvexat). But it is impossible to claim that all tragedies and, by implication, all Satyr plays the action of which takes place in the underworld are of the similar 'plain' composition.38

Intriguingly enough, B coincides with A in presenting the unmeaning on; written in a very peculiar, unmatched, manner — OHC — as if the scribe suddenly decided to use uncials. Bywater appears to have found the solution: on; "is but one remove from oyi; as it might be written in an uncial ms. (OHC = O+IC)".39 Kassel disbelieves this printing on; under crosses in his usual agnostic manner. Tarán's treatment of the text is, on the contrary, quite heavy-handed. on; cannot be straightforwardly replaced with ánAí, because there is one more passage in the Poetics where both A (blindly followed by a number of descendants as well as the Aldine) and B read on; for undoubtable oyi;, 1458a5: |ía yíyvexat a^oxépwv oy (Empedocl. B 88 DK); Aristotle had in mind the contracted form (cf. a4-5: aq^pn^vov Se oiov to Kpt Kai to Sw, and Strab. 8.5.3, cited by Diels: 'E|neSoKAet Sé "|ía yívexat aiqoxépwv oy" ñ óyi;-, in the next following passage by Antimachus oyi; is contracted in the same way), but a copyist unthinkingly inscribed a commonly known word. The parallel, also noted by Bywater, speaks strongly in favour of his conjecture: w most certainly had to Se xéxapxov oyi;. oyi; appears to have been written twice in a similar confusing manner — a trait of personal ductus. Coincidentally, to Se xéxapxov OHC was carefully transcribed to finally survive in B. And certainly, Prometheus, Phorcides and "all those that take place in the underworld" should more suitably be called impressive in spectacle than simple in composition.40

37 See Batteux 1771, 150, with n. 3. He refers to the same reading in Cod. Par. gr. 2117. Cf. c. XV, 1454a26: xéxapxov Se [sc. to ó|iaAóv. ó|iaAóv first occurs as a marginal note in Estensis a. T. 8. 3 (= Puntoni 100), where it might have been inserted by Giorgio Valla who owed the codex (Lobel 1933, 3). In his 1498 Latin translation Valla renders the locus quarta porro aequabilis. John Rhosus who wrote Lau-rentianus plut. 31. 14. conjectured oiKetov, and the same reading intruded into Par. Coislinanus 324 which also has ó|iaAóv in the margin. Among numerous emendations listed by Pia Pattoni (159, n. 12; 184-185) some are worse than others: [to Ss Téxapxov] <ñ Se e'neiaoSiw>Sn; Else; to Ss xéxapxov ó ño<úxio; |iü9o; Kai éneiaoSiwSn;> Post; oyKo; idem; m:ór|cu; Georgoulis; Siávoia inter alios Schmitt.

38 Cf. Pia Pattoni 2012, 160: "non risulta del tutto perspicuo perché i drammi ambientati nell'Ade (oaa év aSou) dovrebbero essere tout court ánAa."

39 Bywater 1909, 250.

40 Cf. Pia Pattoni 2012, 161-162. Like many before her (Rostagni, Valgimigli, Janko, Dupont-Roc/ Lallot et al.), Pia Pattoni stands for the authenticity of oyi;. But note that oyi; and ánAí cannot replace each other: the Iliad which is plain in action is less spectacular than the Odyssey which is complex.

This can only mean that 1456a2-3, to Se TeTapTov... aSou, is not Aristotelian, composed most likely to fill in a break in the given text. The majuscule insertion should be dated at the latest to ca. mid-8th century,41 but it could have been of much earlier origin, since the author has no difficulty in telling which ancient plays are eye-catching enough to represent o^ic. It is composed in a careless style already familiar to us. OopKiSe; is the only Satyr play mentioned in the extant text of the Poetics,42 and it is recalled to exemplify the species of tragedy. o^i; markedly differs from the other three 'kinds', not least because of its syntactical inconsistency, so we ought to ask how the interpolator came to the idea to supply it. The answer is partly provided by another striking insertion of the same origin. ToaauTa yap Kai Ta ^epn ¿A.ex9n would seem perfectly absurd,43 unless we remember that with all his negligence as to the appropriateness of the utterances, intended to be explanatory, the interpolator is painstakingly scrupulous about the terminological coherence. He does not bother himself with the actual number of ^epn, but considers it important to remind that the newly introduced concept of 'kinds' strictly corresponds to the preceding notion of 'parts'. His point, clearly indicated by yap, is to link eiSn and ^epn together into a kind of terminological unity. This allows him to further add one of the 'parts', to the insufficient list of the 'kinds'. So, he simply transfers o^i; from the catalogue in ch. VI without even trying to better adjust it to the new setting.

The interpolation method we attempt to observe can be called retrospective: insertions never build on what follows which suggests that they were made in the very process of copying. e'iSeaiv in ch. VI, 1450a12 turns to be very helpful for making of o^i; a 'kind'. But it could not have appeared in ch. VI in support of what is asserted in ch. XVIII. Like all other additions, it should have been founded on what precedes. That would cause us to suppose that the interpolator based on x^pic Tot; eiSeai (1449b30). As seen above, Aristoe tle describes verse and music as 'kinds', or means, of speech embellishment. Using eiSn to recapitulate the constitutive elements of tragedy would imply a shift of meaning. The effort it would require is quite needless in view of the possibility to employ the obviously appropriate ^ep^. But ei'S^ supplied as an alternative to ^epn in 1450a8, naa^c T^c TpaywSia; ^epn eivai that is in but one sentence before the puzzling toutoic... Kexpr|VTai Tot; eiSeai, appears to be witnessed in the Syriac translation. The Aristotelian text is incorporated by Bar Shakko in a leapfrog manner: he starts with the definition of tragedy (from 1449b24 onwards), interrupts the citation at nparrovTec noiouvrai t^v ^naiv (1449b31) to include alternative definition from another source,44 and takes it up again at naa^c T^c TpaywSiac. What he quotes here is literally to be rendered as follows: "so, the kinds, that is the parts, of tragedy are six in number". 'Kinds' and 'parts' are linked together by the exegetic particle usually introducing a gloss-like explanation (^o^). Letting out Ka0' o noia Tic eaTiv ^ TpaywSia which Bar Shakko's source probably has found

41 See n. 54 below.

42 It remained unnoticed by Schoder 1969, 75; hence, perhaps, the common opinion according to which the Satyrs are lacking in the Poetics.

43 Allen (1972, 81-82) proposes to refer it to another work, or "supplementary material", but such a 'blank' reference would be unparalleled; one should at least expect something like ¿v unoypacpfj. But even if Allan is right, why what was four in the scholarly remarks turned to be six in the treatise we read. Are we left to conjecture about the 'parts' that were at first considered irrelevant?

44 "Some claim that tragedy means something mournful <...>" There follows a quote from the Psalms (6:7) in Peshitta version. On the sources of Bar Shakko Arzhanov will have plenty to say in the forthcoming edition of Syr.

unessential the citation proceeds then up to |eXonoiia (1450a10) and is again interrupted by summarizing remark: "these are the six kinds of tragedy that we find according to Aristotle's view". After that the compiler jumps back to the section corresponding to ¿^ avdYKn; av el'n xi |opiov xpaY^Sia; (1449b32, the translation ignores |opiov!) and cites the text up to ndanc x^c xpaY^Sia; ^¿pn elvai e^ which is this time translated without interpretative additions but with an allusion on what precedes: "Therefore it is necessary that every tragedy should have six parts — those which we have mentioned above". Since the 'kinds' are 'mentioned above' twice in the same context, it can be claimed that eiSn, if not e'iSn n or e'iSn ^xoi was inscribed supra lineam before ^¿pn in the archetype: its maker might have recalled the word from previous paragraphs and jumped to the occasion to employ the 'Aristotelian' term. Most copyists ignored his note, not least because of its in-appropriateness, but it intruded in the text on which the Syriac translation was based.45

We cannot be sure, of course, that 'kinds' is not an invention of an intermediary Syri-ac source. But even if it was never present in any Greek ms., its reflections in Bar Shakko's text strengthen the probability that, having misapprehended xwpic xot; ei'Seai (sc. xwv nSua^dxwv), the interpolator regarded e'iSn as a near synonym of ^¿pn. He might even have thought it a better suited term, and, with typical pedantry, considered it important to supply a side remark in which it would replace ^¿pn to denote the constituents of tragedy. That provides an explanation why xouxoic |ev ouv ouk 6XiYoi auxwv 6; einetv ^xpnvxai xot; ei'Seaiv (note the emphasis on the last word) followed by the repetitive list (Kai Yap oyi; eqs.) came into being. What remains of chapter VI is a prolonged argument in favour of the priority of |UGo; (1450a15-b4), the hierarchy of remaining components appended. Hence, for an insertion aimed at stressing e'iSn no better place could have been found.

Curiously enough, the extant witnesses allow to reconstruct the same interlinear gloss in w not far after xoaauxa Yap Kai xa ^¿pn ¿X¿xGn, at the very beginning of ch. XIX, 1456a33-34. In these lines the paradosis divides between nepi |ev ouv xwv aXXwv eiSwv eipnxai, Xoinov Se nepi X¿^ewc Kai Siavoia; einetv (B, Z) and nepi |ev ouv xwv aXXwv nSn

45 In reflecting the Greek Syr. displays gaps and errors where Ar. appears faultless. The new collation of Syr. carried out by Arzhanov for Thesaurus criticus (see n. 1) has revealed that in reflecting the Greek Ar. and Syr. have common mistakes to which, however, only Syr. adds its own. These are: 1449b29: exovxa puG|iov] sxovxa ^¿YeGov Syr.; 1449b32: |opiov xpaYMSiac;] |opiov om. Syr.; 1449b36 ¿nei Se npa^ew;] Se om. Syr.; 1449b50a2-3 Kai xuYxavouai] Yap xuYx&vouai Syr.; 1450a4 X¿Yw Yap |UGov toutov] toutov om. Syr.; 1450a6-7 ¿v oaoi; X¿Yovtec anoSeiKvuaaiv xi n Kai ano^aivovxai] ¿v oaoi; anoSeiKvuaai Kai ano^aivovxai Syr.; 1450a7-8 avdYKn ouv] avdYKn Se Syr. This is quite enough for one Bekker's page. The divergences are small, but distinct. Gutas (2012, 101-102) claims that Ar. stems from the revised version of Syr. But it remains obscure who would carry out such a revision (it could hardly have been Ishaq b. Hunayn), and why would he do that kind of editorial work instead of simply translating the text anew. In trying to frame this into his revision scenario Gutas (109) is forced to admit the existence of Greek ms. other than Z: "Syriac translation itself was revised. on the basis of other, unknown, material, which conceivably could have included another Greek ms. (¥)". The latter probably "belonged to the apographs of the hyparchetype manuscript of the Syro-Arabic tradition" (103). Our conclusion would be that the Greek ms. used for the Syriac translation was copied from the ms. used for the Arabic one. Abu-Bishr had the Syriac text before his eyes. That is confirmed, apart from the common interpretations, by Ibn al Nadim's statement in Kitab al Fihrist. In stating that Matta was translating from the Syriac Nadim uses the word tafsir (lit. 'interpretation'): this can be a synonym to naql ('translation'), but also could mean specifically interpretative translation. Matta is explicitly called 'Greek' (which can, of course, mean 'Christian') and could hardly have escaped knowing some Greek, living and teaching in a Syriac monastery (Deir Qunna). We cannot be sure that all his Arabic translations listed in Fihrist were from Syriac only: these works are not attested. But his manner of translating allows to suggest that he was compiling from the Greek and the Syriac. He consulted the latter source for interpretations, probably because it was glossed.

elpnxai eqs. (A, O, rec.).46 Modern editors prefer eiSwv, but, first, eiSwv could hardly have turned into ^Sn: the words are not similar either by appearance or by spelling (even considering the itacism; besides, the Byzantine scribes were trained to be very sensitive about accents). Secondly, only some few lines above the term was given quite another meaning. However, ^Sn appears too sophisticated for a scribe's conjecture, and if we admit that a learned copyist remembered the text of ch. VI, toútoic;. .. Kéxpnvxai xoi; elSeaiv, he would have considered eiSwv acceptable like the current scholarship do.47 Still, to decide for ^Sn is clearly not a better solution. The adverb is superfluous, and ^Sn elpnxai has no parallels in the Aristotelian Corpus (wanep el'pnxai ^Sn, Met. 1053a24, is hardly a suitable one). So, we are left to suggest that eiSwv and ^Sn originate from a common source that allowed both readings. This must have been el'Sn originally inscribed over the text between xwv aAAwv and elpnxai (cf. GA 715a7-8: nepi ^ev oiv xwv aAAwv elpnxai, and Pol. 1301a19: nepi ^ev oív xwv aAAwv wv npoeiAó^eGa axeSov elpnTai), perhaps in the contracted form EIA — since A actually has ^S'. The ancestors of B and Z interpreted that as a genitive, whereas those of A and O decided for an adverb in a mistaken itacistic spelling. It is also thinkable that the maker of w implanted el'S(n) into the text before elpnTai (cf. the un-grammatical óyi; in óyi; exei nav,1450a13), or even substituted it for the original ^epwv, or ^opíwv (cf. Phys. 199b28: nepi xwv aAAwv ^epwv; GA 715a1: nepi xwv aAAwv ^opíwv el'pnxai; HA509a21: nepi ^ev oiv xwv aAAwv ^opíwv xwv ¿vxo; elpnxai, while xwv aAAwv eiSwv occurs nowhere else in the Corpus).

The above-described treatment of the source text is surely very frivolous, and even if the atheteses of toútoic;... waaúxw; (ch. VI. 1450a12-15) and to Se xéxapxov... aSou (ch. XVIII. 1456a2-3) would prove irrefutable, strong doubt persists that the insertions were meant as embellishments of a text to be widely read. The last related passage we are going to examine might bring more clarity to this issue. The following phrase marks the beginning of what in most editions is printed as ch. XII of the Poetics, 1452b14-16:

^épn Se xpaY^Sía; oí; ^sv 6; síSsai Sei xp^aQai npóxepov slno^sv, Kaxa Se to noaov Kai si;

a Siaipeixai Kexwpia^éva xáSe ¿axív'

Hereafter definitions of npóAoyo;, ¿neiaóSiov, e^oSo;, nápoSo;, axáai^ov and Ko^óc; are given, and the chapter is concluded by almost verbatim repetition of the passage (1452b25-27) in which, however, w; elSeai is skipped. It is absent in all extant branches of transmission and thus was most probably omitted in w, though without it Sei turns to be pointless: it is fairly absurd to require from tragedy the use of its constitutive elements such as plot or characters. To be sure, w; elSeai is essential: it is this syntagma that renders the phrase comprehensible — if, of course, we read it as interpolation effected by the same person whose techniques we have surveyed above. Provided our assumptions are true, his intention here is the same that he demonstrates later, in ch. XVIII: "parts that are to be used as (= identical with) kinds" is another clumsy attempt to reconcile the authentic ^épn with the interpolated elSn which to his mind was a more appropriate term to convey abstract meaning. Thus, w; elSeai is indispensable in the first transition clause. But are we to follow those who restore it in the second, as Kassel and Tarán actually do?

46 Cf. above, n. 31.

47 For Lattmann (2015, 260) nepi pev ouv Twv aAAwv eiSwv ei'pr|Tai is the main reason for retaining ei'Seaiv in 1450a13. He does not discuss the alternative reading.

This is a more complex question than it might seem. The suspicion that the segment containing definitions of the structural components of tragedy is out of place in the Aristotelian Poetics is as old as the post-humanist age: Heinsius tried to relocate ch. XII,48 it was proscribed by several 19th century scholars49 and then by Else, who retains only the first clause arguing that eiSeai Set xp^aGai alludes to 1450a13, ^xpnvxai xot; eiSeaiv, the words he deems authentic.50 In supposing this he fails to recognise the difference between ^¿pn and eiSn, and makes no effort to explain the point of 6; eiSeai. He believes that ^¿pn Se xpaY^Sia;. eino|ev concludes ch. XI and thus has to be joined together immediately with the beginning of ch. XIII. That makes his arguments for athetesis insufficient. The syntactical linking to npoxepov eino|ev believed to be Aristotelian is hampered by ¿^e^c av ein Xe^ov xot; vtiv eipn^voi; (1452b30). Ch. XII is, of course, spurious, but, contrary to Else, not on the grounds that it breaks the connection between chapters XI and XIII. Ch. XIII introduces new subjects: wv Se Set axoxdZeaGai Kai a Set euXa^etaGai Kai noGev eaxai xo x^c xpaY^Sia; epYov (1452b28-30). The section on npoXoYo;, ¿neiaoSiov etc. is incorporated right at the watershed where the descriptive part of the Poetics ends and the prescriptive begins. This is just the proper place to include a few paragraphs on a special topic. Consequently, the dismissal of ch. XII should comprise its first colon not excluding npoxepov eino|ev, while a deliberate allusion on ^xpnvxai xot; eiSeaiv in ch. VI supports its athetesis.

What is offered in ch. XII is indeed a "purely mechanical division based on no principle and carried out in a crudest possible manner".51 But it is not primarily because of this that the section should be excised. More important is the fact that whatever the target audience of the Poetics might have been, it is for his contemporaries that Aristotle was writing. It would be ridiculous to explain them that a prologue is "that part of tragedy which precedes the entrance of the chorus", or that stasimon is a choral song "without anapests and trochees". On the contrary, a late ancient or early mediaeval professor would certainly try to explain it to his students. A modern lecturer on ancient tragedy would likewise think that without defining its formal divisions his lectures would be incomplete.

Now, the author of ch. XII writes in the first person pointing at the authorship of what precedes (cf. 1449b34-36: X¿Yw Se X¿^lv |ev aux^v x^v xwv ^¿xpwv auvGeaiv, and 1450b13: wanep npoxepov eipnxai). And yet even such a telling detail is not enough to blame him for falsification. Note that the segment is incorporated in a plainly visible manner: its end is marked by a near replica of the transition phrase. This is hardly a proper way to conceal a forgery. His interpolations reveal no desire to defraud a reader; he never disguises himself as an ancient thinker like, for instance, the author of a preface to Theo-phrastus' Characters does. Rather, he was copying the Poetics for his own use, conceivably, with an educational purpose. This would explain the casualness of ouk 6XiYoi auxwv (1450a12) and exei nav (Ibid.), the syntactical inaccuracy of oyi; [?] (1450a13), eiS(n) [?] (1456a33), Sidvoia and ^Go; (1450a2, to be compared with likewise scholarly superfluous eXeo; |ev nepi xov avd^iov, ^o^o; Se nepi xov 6|oiov, 1453a5-6) and xo Se x¿xapxov oyi;

48 Heinsius 1611, 15. He placed it between chapter VI and VII.

49 The proponents of athetesis as well as of attribution are listed in Else 1957, 360, n. 1.

50 Hence Montmollin (1951, 58-59; 125-129), who argues for the authenticity of ch. XII, proposes to change Toic; d'Ssmv in wc; d'Ssmv in 1450a13.

51 Else 1957, 362. On p. 351 Else cites a nice account of ch. XII by Gomperz ("Polizeiverordnung inmitten eines rechtsphilosophischen Werkes").

(1456a2), the salient inconsistency of Toaauta yap Kai ta pepn eAexQn (1455b32-33; in making passing remarks he has no need to remember the exact number!), as well as the 'equalizing' tendency displayed in Kai peAo; (1449b29), in the definition of poetic speech as 'the composing of verses', further on, in twv pev Aoywv ^ Ae^i; (1450b12-13), in placing o^i; among the 'kinds' of tragedy, and perhaps most of all, in switching from pepn to ei'Sn to denote one and the same thing. The didactic background becomes palpable in the forced account of music as something o t^v Suvapiv pavepav exei naaav (1449b35-36). All this creates a general impression of a school copy never intended for use outside the class-room. But if that is true, the maker of w had no need to repeat ei'Seai in the closing sentence of ch. XII: pepn Se TpaywSia; ok; pev Set xp^aGai would more than suffice to point back to his own conjecture. Thus, the decision to restore it appears preposterous.

What school it could be is not our present concern, but in contrast to the Eastern translators its master was entirely aware of what the Poetics is all about, and was too well-read in the Greek drama even for a Middle Byzantine scholar.52 He wrote in uncials,53 inserting most of his supplements directly in the core text or between the lines, the technique presupposing the use of a papyrus roll rather than a parchment codex. His enthusiasm naturally decreases, as he works through the text, and in the last chapters his presence is much less explicit, the conjectures almost lacking. Some of his notes might have been dismissed by the copyists, others reproduced in the mss. now lost. The recent research confirmed by the fresh collation of Syriac and Arabic translations has shown that the earliest reachable branch of transmission, that is the Eastern, is more ramified than previously thought, its archetype (if they stem from any) probably dating back to antiquity. The interpolations which we have attempted to trace could well have originated some half-millennium before the making of A, in the late Neoplatonic period (4th-6th c.), when Aristotle was read and commented on by many teachers in West and East. As noted by several scholars of the Poetics, the most prominent of them, and whose work on the text of the Poetics is somehow witnessed, was Themistius.54

However, to penetrate thus far is barely possible even for the eye of the keenest textual critic, and so this study has to conclude with uncertainties. We have focused on the phenomenon once called by Wilamowitz interpolationum familia.55 But, in fact, we cannot

52 One more interpolation coming from an interlinear gloss reveals literary knowledge. It is to be found in, 1454b13-14: napaSeiypa aKAnpornto; oiov rov AxiAAea AyaGwv Kai "Opnpo;. napaSeiypa aKAnpornto; was secluded by Ritter whom Taran follows. The ms. text can not be cured in a way proposed by Lobel (1929, 78): oiov tov AxiAAea aya96v (lectio facilior of B, see: Pozdnev 2015, 196-198) Kai napaSeiypa aKAnpornto; "Opnpo;. N. A. Almazova (2019, 306) supposes, not without reason, that 1455a4-6, ev Xonpopoi; oti opoio; ri; ¿Ar|Au9ev ktA, was inserted by someone who "added a famous example of discovery, which suited his own conception, but not that of the Stagirite." 1461a17-20: apa Se cpnaiv "I roi or' e; neSiov to TpwiKov aGpr|aeiev"... Kara peracpopav eipntai is much the same case.

53 Cf. Gutas 2012, 108: "Z was in uncials, representing a stage in the transmission parallel to the hypar-chetype from which the extant Greek manuscripts and exemplar of the Latin translation derive." If this is so, there remains no doubt that the common source of Z and other ms. was also in uncials.

54 He taught at Constantinople in the mid-4th c., and as a 'senator' had access to the emperor's library. His glosses in Aristotle's corpus are many. His paraphrase of the Poetics is mentioned in Ibn al-Nadim's Kitab al Fihrist. See Busch 2008, XXV, with references. And it seems that numbers were not his strong suit, cf. Or. 36 (Ynep tov Xeyeiv q nuq r& fiXoadfW Xekteov), 316d: Kai oi> npoaexopev ApiaroreAei oti to pev npwrov o xop6; eiaiwv rjSev ei; rou; Geou;, ©earn; Se npoAoyov re Kai prjaiv e^eupev, AiaxUAo; Se rpirov [!] unoKpir^v Kai OKpipavra;, ta Se nAeiw rourwv ZopoKAeou; anr|Aauaapev Kai EupimSou;

55 Wilamowitz-Moellendorff 1875, 205. The characteristics he gives concerning a group of verse interpolations starting with ^ in Euripides and Sophocles, for all their mordancy, fit well with the above examined

be sure that these interpolations have originated from one source. They might have been made by several ancient and mediaeval scholars who used the text for teaching purposes. Nor there is any means to clarify if œ was not a copy of this source, or a copy of this copy. There is a certain, relatively small, number of indicative gaps and mistakes, as well as interpolations, in the transmitted text of the Poetics that do not fall into the mentioned category and thus have found no place in the present discussion (with the only exception of ouç, 1449b37, altered in xoùxouc; just to render the passage readable).56 Any of these faults could have been committed by a person whose supplements we have studied. But they could equally well have been inherited by him from his source or added by the maker of a further copy which should then be thought of as œ. The common ancestor we look for proves to be extremely elusive.

References

Allan D. J. EIAH TPAmiAIAZ in Aristotle's Poetics. CQ 1972, 22, 81-88.

Almazova N. Recognition Based on Paralogism (Aristot. Poet. 1455a12-16). Hyperboreus 2019, 25, 302-327. [Anonymus (G. Morel?)] (ed.) Aristotelis De artepoetica liber. Paris, apud Guil. Morelium, 1555. Batteux Ch. (éd., trad., comment.) Les quatre poétiques: d'Aristote, d'Horace, de Vida, de Despréaux. Paris, Saillant et Nyon, 1771.

Brescia G. Sul testo e la fortuna della Poetica. Note di critica aristotelica. Milazzo, SPES, 1984. Bursian C. Zu Aristoteles Poetik. Jahrb. f. cl. Philol. 1859, 5, 751-758.

Busch T. Chronologische Übersicht zur Textgeschichte [sc. der Poetik], in: Schmitt 2008, XVII-XXVI. Butcher S. H. (ed.) Aristotle. Poetics. Transl. and with crit. notes. London, Macmillan, 41911.

case: "Curabant absurdi homines, ne tragici quidam viderentur omisisse"; "alia quaevis abesse vel potius addi posse visa sunt male sedulis homuncionibus" (Ibid. 206).

56 Apart from those examined above (1450b12, xwv |iev Xoywv, 1452b25, <6; ei5eai> which is, as said, doubtful, and 1454b14: napdSeiY|a a^rpo^TO;, see n. 53), Taran (2012, 148-149) lists nine cases, most of them (*) controversial: (1) 1449b9, ^¿xpi |6vou |^xpou: the luckiest attempt to correct it is Tyrwhitt's ^¿xpi |iev xou |^xpou; (2*) 1450a29, X¿5euc Kai Siavoia;: Vahlen conjectured X¿5el Kai Siavoia, and many would like it, but in fact the ms. reading is tenable, if not that elegant; (3*) 1452a3, Kai |aXXov: the athetesis is questionable; in fact, if we let out Kai after Yivexai (as in B) and read таuта Se Yivexai ^aX^a, Kai |aXXov oxav Y¿vr|тal kxX. the transmitted text appears coherent; (4) 1454a18-19: npoaipeaiv xiva rj; here ® probably had xi; rj or xi; av rj because Moerbecke translates electionem que sit, but this could have been a scribe's conjecture; Vahlen's npoaipeaiv xiva ^xi; av rj was accepted by both Kassel and Taran, while Bywater and others simply deleted rj; in any case the corruption was present in the source text; (5*) 1456b8, cpaivoiTO ^¿a is not that dubious as it is thought of; cf. Rhet. 1369b16; 19; EN 1176a20: EE 1231a1; Probl. 928b19; 23, while Vahlen's fl 5¿ol meant as a usual expression, almost a colloquialism, is in fact a hapax; (6) 1457b33, where, as noted by Maggi, the explanation of K6a|o; is missing; the lacuna is evident, and it cannot be established at what stage of transmission it appeared; it could well have been before the time of w; (7*) 1458a28-29, Kaxa Se x^v ^exa^opdv: the ms. reading is in no way worse than ^exa^opuv conjectured by Bywater; actually, it should be x^v xwv ^exa^opuv (sc. auvGeaiv), but this is too smooth to have been altered; (8) 1460a13, avaXoYov: this is a corruption of aXoYov, as noticed by Vettori. (9*) 1461a16, aXXoi: the case is rather complicated to be explicated here with due care; to cut it short, aXXoi i.e. the correct rendering of Homeric passage should not be emended into navxe; with Grafenhan; the 'metaphor' refers to navvuxioi (see also n. 53). Hence at least three archetypal mistakes. To these six more might be added: (1) 1447b29, ¿v ai; : after Vettori ¿v oi; is accepted almost unanimously; though the corruption is trivial it is likely to go back to w; (2) 1448a21: oxe |ev dпаYY¿XXovtа n exepov xi YiYv6|evov, where we should restore oxe |ev dпaYY¿XXovta oxe Se exepov xi YiYvo|evov, since in Aristotle oxe |ev is always followed by oxe Se; it was first suggested by Sahl, rejected by Kassel, and quite reasonably accepted by Taran; (3) 1448a35, ouxoi: Spengel's emedation auxoi suggests itself; (4) 1449b25 EKdaxou: Pazzi's EKdaxw was briefly discussed above; (5) 1453a37, av o'i exGiaxoi: Bonitz was doubtlessly right in conjecturing o'i av exGiaxoi; (6) 1457a35-b1, oiov xa noXXa xwv MaaaaXiwxwv, 'Ep|oкalK65avGoc: the transmitted text is unsatisfactory: either it is lacunar, or the last word originates from a gloss.

Bywater I. (ed., transl., comment.) Aristotle on the Art of Poetry. Oxford, Clarendon,1909.

Dupont-Roc R., Lallot J. (eds). Aristote. La Poétique. Texte, traduction, notes. Paris, Éditions du Seuil, 1980.

Else G. F. (ed., comment.) Aristotle's Poetics: The Argument. Cambridge, Ma, HUP, 1957.

Gallavotti C. (ed., comment.) Aristotele. Dell'Arte Poetica. Milano — Roma, Fondazione L. Valla / Monda-dori, 1974.

Grayeff F. The Problem of the Genesis of Aristotle's Text. Phronesis 1956, 1/2, 105-122.

Guastini D. (ed., comment.) Aristotele. Poetica. Introd., trad., comment. Roma, Carocci, 2010.

Gutas D. Greek Thought, Arabic Culture: The Graeco-Arabic Translation Movement in Baghdad and Early Abbasid Society (2nd-4th/8th-10th Centuries). London — New York, Routledge, 1998.

Harlfinger D., Reitsch D. Die Aristotelica des Parisinus gr. 1741. Zur Überlieferung von Politik, Rhetorik, Physiognomonik, De signis, De ventorum situ. Philologus 1970, 114, 28-50.

Hatzimichali M. The texts of Plato and Aristotle in the first century BC, in: M. Schofield (ed.), Aristotle, Plato and Pythagoreanism in the First Century BC. Cambridge, CUP, 2013, 1-27.

Heinsius D. (ed.) Aristotelis De Poetica liber. Leiden, Elsevier, 1611.

Hermann G. (ed.) Aristotelis De Arte Poetica liber. Leipzig, G. Fleischer, 1802.

Janko R. Rec. Taran, Gutas 2012. CPh 2013, 108, 252-257

Kassel R. (ed.) Aristotelis De arte poetica liber. Oxford, Clarendon, 1965.

Kemal S. The Poetics of Alfarabi and Avicenna. Leiden, Brill, 1991.

Kemal S. The Philosophical Poetics of Alfarabi, Avicenna and Averroes: The Aristotelian Reception. London — New York, Routledge, 2003.

Lattmann C. Vom Nutzen der ôyiç Aristoteles, Poetik 1450a12-14. Philologus 2015, 159/2, 251-271.

Lobel E. The Greek Manuscripts of Aristotle's Poetics. Oxford, OUP, 1933.

Lucas D. W. (ed.) Aristotle. Poetics. Greek text, Latin app., with introd., comment., appendices. Oxford, Clarendon, 21972.

Minio-Paluello L. (ed.) Aristoteles. De Arte Poetica Guillelmo de Moerbeke interprete. Bruges — Paris, Des-clée de Brouwer, 1953.

Margoliouth D. Analecta Orientalia ad poeticam Aristoteleam. London, D. Nutt, 1887.

Margoliouth D. The Discussion between Abu Bishr Matta and Abu Sa'id al-Sirafi on the Merits of Logic and Grammar. Journal of the Royal Asiatic Society 1905, 37/1, 79-129.

de Montmollin D. La Poétique d'Aristote. Texte primitif et additions ultérieures. Neuchâtel, Messeiller, 1951.

Pia Pattoni M. Aristotele, Poetica 18, 1456a2-3 e il quarto tipo di tragedia, in: F. Malhomme, L. Miletti et al. (eds). Renaissances de la tragédie. La Poétique d'Aristote et le genre tragique, de l'Antiquité à l'époque contemporaine. Napoli, Giannini, 2013.

Pozdnev M. Honoratus Achilles — the Hero of Two Poetics. Philologia Classica, 2015, 10, 186-202. (In Russian)

Pozdnev M. A Tale of Two Manuscripts, Philologia Classica 2019, 14/1, 141-153.

Reiz F. W. (ed.) Aristotelis De poetica liber. Leipzig, Svikert, 1786.

Schmitt A. (übers., erl.) Aristoteles. Poetik. Berlin, Akademie Verlag, 2008.

Schrier O. J. The Poetics of Aristotle and the Tractatus Coislinianus. A Bibliography from about 900 till 1996. Leiden, Brill, 1998.

Schoder R. V. Literary Sources cited by Aristotle in the Poetics. CJ 1969, 65, 75.

Sicherl M. Die Aldina der Rhetores Graeci (1508-1509) und ihre handschriftlichen Vorlagen. Illinois Classical Studies 1992, 17/1, 109-134.

Spengel L. Rec. F. Ritter (ed.) Aristotelis Poetica. Coloniae, Renard, 1839. ZA 1841, 149, 1252-1275.

Taran L., Gutas D. (eds). Aristotle Poetics. Editio Maior of the Greek Text with Historical Introductions and Philological Commentaries. Leiden — Boston, Brill, 2012.

Taran L. The Text of Aristotle's Poetics in the Codex Parisinus Graecus 2038. Mnemosyne 2016, 69/5, 785-798.

Tkatsch J. (ed.) Die arabische Uebersetzung der Poetik des Aristo teles und die Grundlage der Kritik des griechischen Textes. Wien — Leipzig, Hölder-Pichler-Tempsky, 1928-1932.

Tyrwhitt T. (ed.) Aristotelis De poetica liber. Textum rec., versionem refinxit, animandvers. illustravit. Oxford, Clarendon, 1806.

Valgimigli M. Note di critica testuale alla Poetica d'Aristotele. Atti R. Istit. Veneto. 1937, 96, 1-21.

Verdenius W. J. Rec. Else 1957. Mnemosyne 1960, 13/3, 256-258.

Weinberg B. From Aristotle to Pseudo-Aristotle. Comparative Literature 1953, 5/2, 97-104.

von Wilamowitz-Moellendorff U. Analecta Euripidea. Berlin, Borntraeger, 1875.

Received: July 25, 2020 Accepted: October 27, 2020