Bryn Mawr Classical Review 2006.01.03

Ludwig Bernays, Ars poetica. Studien zu formalen Aspekten der antiken Dichtung. Prismata. Beiträge zur Altertumswissenschaft, IX.   Frankfurt am Main:  Peter Lang, 2000.  Pp. 237.  ISBN 3-631-34685-9.  DM79.00.  


Reviewed by Georg, The Graf v. Gries München München , München (HoiAristoi@hotmail.com)

The volume under review contains twenty-four papers by Ludwig Bernays devoted to formal aspects of ancient poetry such as numerical verse proportions, verse structure, accentual patterns within the same quantitative metre, strophic correspondence, sound effects, and formal means of thematic division within an epic book or a single poem. Problems of textual editing and interpolation as well as interpretation of some lyric or elegiac poems receive some attention, too. Although Latin poetry prevails, four papers concentrate on Greek poetry: 1. Goethes Gedicht "Nach dem Anakreon" (p. 9-17), 4. Textgliederungen in den homerischen Epen (p. 37-50), 5. Ein Gedicht des Archilochos (p. 51-54), and 6. Ein Gedicht des Tyrtaios zwischen Homer, Horaz und Ovid (p. 55-62).1 The Latin poets dealt with are Virgil, Horace, Propertius, Ovid, Prudentius, Boethius, Seneca (as the former's model), and finally two medieval poets from Reichenau: Gottschalk (born ca. 803) and Walahfrid Strabo (born ca. 808).

It is difficult to review such a volume, as individual papers cover problems of philological detail. Therefore the reviewer, especially if he disagrees with the author on some points, has to report on details in detail. Hence the length of this review.

Bernays expresses on p. 19 regret that -- at least partly -- because of the gap between sciences and humanities, numerical principles governing the structures of ancient poems usually remain outside philological consideration. Although he himself understands why philologists are prejudiced against numerical proportions and ratios in poetry, he argues that observers admiring the beauty of ancient temples and other works of architecture fail to see the manifold mathematical calculations and proportions underlying them (p. 19) and that connoisseurs of the music of Johann Sebastian Bach do not recognize the complicated counterpoint plan of each fugue on its hearing, although it is ultimately recognizable in the process of a careful study (p. 90). Following this indication Bernays is going to uncover amongst other things some mathematical principles that remain hidden to a majority of the recipients of ancient poetry. Since these principles are in the one case, especially in short poems, a matter of fact, but in the other not so, the best attitude is to follow the facts. Bernays, however, tends to disregard the facts excessively; instead, in order to satisfy his preconceptions or wishes, he tries to persuade the reader that his own numerical inventions are reality. I would like to demonstrate this in detail through a few examples.

I begin with No. 2. Zahlenverhältnisse in der Dichtung Vergils (p. 19-25). Bernays observes that Theocritus' Idyll 4 consists like Verg. Ecl. 4 of 63 lines, but he at once adds in the parentheses "9 mal 7". And he does the same almost everywhere, e. g. p. 21: "264 (8 mal 33)" and "198 (6 mal 33)", and passim. This is a cabbalistic approach, ridiculed once by R. G. M. Nisbet in JRS 69 (1979) p. 231.2 In mathematical abstraction 63 = 9 x 7 or 3 x 3 x 7, but in an eclogue composed of 63 hexameters such a count is illicit unless the eclogue's inner units consist organically of 7 verses each, i. e. v. 1-7 is one thought, subject, or period, then v. 8-14, then 15-21 and so on. As soon as this is not so, a philologist has to show the inner structure of the poem as it is. On Verg. Ecl. 4 one can agree with K. Büchner, P. Vergilius Maro, der Dichter der Römer, RE VIII A. 1, col. 1202 (= Separatum, Stuttgart 1966, col. 182). For the relationship between verse numbers and thematic units as an artistic principle of ancient poetry one can still consult K. Witte, Der Bukoliker Vergil, Stuttgart 1922. He is not infallible, but he shows that it is numerical assymetry which was sought by Theocritus and his imitator Virgil, and certainly not only by them. Also Homer followed this principle, which Bernays attempts somehow to conceal. An example of the phenomenon occurs in Verg. Ecl. 4: it opens with three verses (1-3) and ends with four (60-63). If it is legitimate to take this as a frame of 7 verses, so it is legitimate only a posteriori, as the eclogue indeed shows units multiplying the number 7, namely 3 + 14 + 28 + 14 + 4, but in no way a priori.3 An illustration of what I mean and criticize is the following sentence from Bernays p. 20-21: "So lässt sich das Längenverhältnis 5 : 4 zwischen Weissagungs- und Rahmenteil (5 mal 7 Weissagungs- , 4 mal 7 Rahmenverse) auch umkehren, indem man die noch nicht an den neuen Gott, sondern an Asinius Pollio gerichteten ersten sieben Weissagungsverse (Vv. 11-17) dem Rahmenteil zurechnet. Auf diese Weise ist das Gedicht nahezu exakt symmetrisch gegliedert in 17 Anfangs-, 18 End- und 28 Kernverse. Durch Texteinschnitte nach den Versen 14 und 49 (wie in Theokrits Gedicht) ist die 4. Ekloge aber auch exakt symmetrisch im Verhältnis 2 : 5 : 2 gegliedert, wobei den 28 Kernversen ebenso viele Rahmenverse sowie die sieben "Scharnierverse" 15-17 und 46-49 gegenüberstehen" and so forth. This is obviously still fairly close to what Virgil's Eclogue 4 allows one to extract. But what shall we do with single books of epic poems? Are they, or most of them, structured with arithmetic exactitude as rigorously as Bernays tries to persuade the reader? I do not deny that epic poets cared about composition and structure of whole epic poems as well as of single books, but Bernays undertakes to show that they follow the realm of absolute figures and divisible ratios. This is often unconvincing. Here are a few examples.

On p. 21-25 a treatment of Aeneid 1-5 and Odyssey 5-12 follows that of Eclogue 4. I can refer to only a few points here, as this kind of argument necessarily requires much space. Bernays shows that the proem to the Aeneid consists of 33 lines divided into 11 and 22 and that these figures (here the apriorism begins) are "Grundmass für weitere Texte der Aeneis" (p. 21). Between the proem and Aeneas' report (1. 34 - 2. 3) he counts 726 i. e. exactly 22 x 33 lines (one is tempted to say 2 x 11 x 3 x 11) which yield the ratio 4 : 3 : 4 (264 + 198 + 264 or 8 x 33 + 6 x 33 + 8 x 33 lines). This is not exact, for the narration ends already at 2. 2, unless we assume that the boundary points are the lines 1. 496 and 2. 3 in which the word regina occurs. Nevertheless, the structures do not depend solely on single words detracted from their narrative units. The section on Carthage ends in 1. 493, not, as Bernays says, in 1. 495. He also wishes the retrospective narration to end in 3. 714, because the closing line 715, he maintains, does not belong to the retrospective part, nor does 2. 3. Therefore, I ask, to which part does 3. 715 belong? And why has 2. 3, which on p. 22 does not belong to a certain section, been attributed to the same section on p. 21? The answer may be: to yield the figure 726 instead of 725, which cannot be divided by 33. Bernays needs 726. Therefore he includes Aen. 2. 3. In the second step it is also not true that there are 725 lines between 3. 714 and 5. 16, because 4. 528 must be deleted (see the apparatus in the edition of R. A. B. Mynors). This fact of course spoils Bernays' arithmetic exactitude (see further p. 23-25 where it has been taken into account). It confirms, however, the view expressed by Kurt Witte mentioned above.

On p. 24 we have to read in the second paragraph (line 3 and 5) 758 and 1517 instead of 759 and 1518. Bernays counted 1518 lines in one section and 1515 in the other, but in order to diminish the gap of three lines between them he now adds immediately "1515 that means 1517 if we include 2. 3 and 3. 715". Why does he now include 3. 715 which he previously excluded? The answer is obvious: because 1517 comes closer to 1518 (without the deleted 4. 528 it is 1517) than 1515 does. Thus Bernays does whatever he can in order to arrive at figures he needs. Also Odyssey 9-12 contain a flash-back. Having done his counts, Bernays says that Virgil seems to have taken certain numerical proportions and even absolute figures from Homer. As a proof of the latter he purports that both Aen. 1. 173 - 2. 2 and 3. 175 - 4. 582 consist of 486 lines (we now know that in the second case there are 485) which is exactly the length of the Odyssey 8. This is, however, a serious blunder, as Od. 8 consists of 586 lines (five hundred and eighty six).4 This fact again spoils the author's argument (p. 25 lines 1-3). He corrects this unhappy mistake as he counts 1478 lines from Od. 5. 279 to 8. 586 (i. e. Book 8 has 586 lines), but creates a new one: there are 1479 lines (I follow van Thiel's edition; using Von der Mühll's leads to the same result).

It is true that Virgil writing the Carthage episode had Od. 5 in mind. Bernays divides this book as follows: 225 + 43 + 225 lines, but also 43 + 182 + 43 + 182 + 43 lines. In Od. 11 he observes that the verses 335-377 form a unit of 43 lines. So far so good. But now he says that with an inclusion of the section 11. 335-377 there are 1517 lines between 9. 1 and 11. 377, exactly as many as in the flash-back section of the Aeneid (with an inclusion of 2. 3 and 3. 715). Now his problem is the number of lines remaining till the end of Od. 12 (there are 717 lines). He should start counting from 11. 378, but he does -- miraculously -- from 11. 335: "Der Anfangsteil der homerischen Rückblende einschliesslich des Zwischentexts ist aber auch fast haargenau doppelt so lang (1517 : 759 Verse) wie der Endteil der Rückblende einschliesslich des Zwischentexts" (p. 25, my italics). Why does Bernays count these 43 lines (11. 335 - 377) twice? Because he needs the ratio 2 : 1. Therefore he counts from 9. 1 to 11. 377 (1517 lines) and again from 11. 335 to 12. 453. Here he counts 759 lines, 5 but there are, unfortunately, 760 (760 + 760 = 1520) or even 761 lines with 11. 638a (van Thiel). I follow van Thiel. If we double 761 we get 1522. Although I am criticizing Bernays' method in view of the facts, I may wish to support him. Therefore I will count every line of Od. 9-12 only once following Von der Mühll's edition (i. e. without 11. 638a). The total of lines of Od. 9-10 is 566 + 574 = 1140, of Od. 11-12 is 640 + 453 = 1093, altogether 2233 lines. 2233 = 2200 + 33. 33 is the exact length of the proem to the Aeneid. Its first part consists of 11 lines. The remaining 2200 = 200 × 11 or 11 + 66 × 33 + 11 or 33 × 33 + 22 + 33 × 33. If, however, van Thiel is right, the result will be 1140 + 1094 = 2234. I am afraid that Bernays, unlike Kurt Witte, would not be happy with 2200 + 34, but he would be with 2200 + 33. Therefore, I am sure that he would challenge van Thiel's decision to insert 11. 638a only because this insertion destroys the perfect proportion and affects the proof that Virgil imitated the number of lines in Od. 9-12 exactly.6 Since Virgil has left the Aeneid without his final touch, one missing or one supernumerary line is not so bad. Scholars who reckon that Virgil began the Aeneid with Ille ego qui7 would anyway add four lines to the proem and count 14 + 22 = 37 lines (instead of 11 + 22 = 33).

Similar ways of arriving at exact figures not covered by the facts can be observed in No. 4. Textgliederungen in den homerischen Epen (p. 37-50). Here too verses introducing or finishing a speech are sometimes counted together with the speech, sometimes not (cf. p. 40).8 In A. Ludwich's edition of the Iliad 1. 265 is deleted. Taking this fact into account would have allowed Bernays to arrive at the ratio 488 : 122 he wished to obtain without any further artifice.9 His artifice, I have to say, is annoying, especially if compared with the playfulness with which Umberto Eco amuses the readers of Foucault's Pendulum. Furthermore, Bernays comments in this article on the structure of Il. 3 and 4 and Od. 13 and 15, and it is certainly worth reading for its other merits.10 At the end (p. 49-50) he comes to the conclusion that the division of the Odyssey in 24 books must be Homer's own, i. e. cannot be a later act. He thus supports the unitarians with a good argument.

To sum up this section, I repeat the view that the extremely careful composition of ancient epic poems can also be approached with the help of figures and ratios drawing on the amount of lines marking turning points in the epic plot. It goes, however, too far if one requires from the epic poets that they strictly follow an arithmetical symmetry. Devices such as 'if you wish the verse X to occur exactly between A and B and you do not find X in the middle, add or subtract some verses between AX or between XB, count some verses twice on whatever reasoning, include or exclude lines introducing or finishing a speech, and you will somehow get what you expect' are to be dispensed with. As Bernays is a very careful and sensitive observer of philologically relevant detail, he would have done more if he had not insisted on exact figures, but worked with numerical approximations observable in the texts he considers. Everyone would agree with him that Homer and Virgil had a good feeling for structural symmetry11 and considered quantitative proportions among units of certain length more carefully than it is commonly known, and his papers I have referred to so far would be more readable. On the other hand, shorter poems especially of later periods were often structured with exactitude in terms of length or even position of key-words. No. 5 and 6 in Bernays' collection12 prove it to be true even in Greek archaic poets. These two pieces are worth studying. No. 1. Goethes Gedicht "Nach dem Anakreon" (p. 9-17) deals with Anacreont. 43 imitated by Goethe. Also here Bernays demonstrates how carefully he observes formal, lexical, metrical and acoustic phenomena. Nevertheless, even here he sometimes tends to exaggerate, for instance on p. 13 line 1-6 from the bottom13 or on p. 15 n. 8; FANEI= : GU/NAI are not rhyme-resembling, as -EI= and -AI do not rhyme in Greek. Even if they did, the position of the accent excludes rhyme anyway at least as far as it is understood in English, German, Italian, Polish or Russian poetry.14 Also quantities and a position in a verse have to agree. On p. 218 " (tr)anquillo - Aquilo" in Boeth. 2. 3. 9 and 11 are qualified as perhaps a jingle. As far as I see the poet does what he can to lessen such an impression: not only is tranquillo a molossic word while the vowels 'a' and 'i' in Aquilo are short, but also the distance of both words from each other, their position in the verse and their accentuation seem rather to favour my view.

In No. 3. Textprobleme in Eklogen Vergils (p. 27-35) Bernays deals with Verg. Ecl. 8. 76 and 28a, the attribution of Ecl. 9. 46-50 to one of the shepherds, Ecl. 10. 17 (cf. 2. 34 and 5. 45), and with structural aspects of Ecl. 2, 9, and 10. No. 7. Versstrukturen und Textgliederungen in der 16. Epode des Horaz (p. 63-71)15 introduces a series of nine papers devoted to Horace; here Bernays defends the transmitted order of lines 61-62 (like Otto Keller, Epilegomena zu Horaz, 2. Theil, Leipzig 1880, p. 404 f., not named in the paper), but he is certainly wrong in favouring sacrauit over secreuit in line 63 (see p. 65 n. 32) and does not say where the first verb is -- as he maintains -- well attested in the sense he requires; secernere means 'to allot by detraction'.16 He also argues -- rightly, as I am convinced -- for the priority of Virgil's Eclogue 4 before Horace's Epode 16 (cf. note 26). The following papers are all, as we may expect, kept in the same flow of meticulous observation of structure, metre and accentual patterns. Therefore I shall not report on obvious virtues of Bernays', but concentrate on other, mostly textual matters, on which a classicist has to disagree.

No. 8. Zur architektonischen Struktur der horazischen Satiren (p. 73-81) and No. 9. Zur Textgliederung in der Ars poetica des Horaz (p. 83-90) deal with hexameter poetry. While No. 9 is worth studying, No. 8 contains one serious point calling for contradiction.17 Bernays wishes to defend the pivotal structure of Satire 1. 10. This is only possible if he retains the opening verses *1-*8 commonly regarded as spurious (p. 79-80). On these verses it is enough to read O. Keller, op. cit., p. 504-506.18 The interpolator probably thought in the same way as Bernays does. Operations with figures were not unusual in Late Antiquity and Middle Ages. A similar problem of an interpolation is dealt with in No. 12. Zur Frage der Interpolationen im Text der Horaz-Ode 4. 8 (p. 111-120). Bernays wishes Carm. 4. 8. 16 and 17 to be genuine at any price. He writes (p. 115): "Für die Authentizität der Verse c. 4. 8. 16 und 17 gerade mit ihren formalen Eigentümlichkeiten dürfte nicht zuletzt auch die Tatsache sprechen, dass diese Verse -- sofern vom ganzen überlieferten Text ausgegangen wird -- genau am Ende der ersten Texthälfte stehen," and then -- and this is illegitimate -- he quotes parallels where this applies thinking that if this applies elsewhere, where the text is sound, it must also apply in 4. 8 where it may be unsound. To this I answer as follows: if I composed my own strophe of four lines and inserted it in Carm. 1. 5 after verse 8, according to Bernays this strophe would have to be genuine, because it would stand exactly in the middle between the strophes 1-2 and 4-5, as do lines 9-12 in Carm. 1. 4, 1. 6, 1. 10 e. a. I know one philologist who defends the whole of 4. 8 as Horace's, namely Otto Keller, op. cit., p. 324-331, but he, unlike Bernays, at least does not lure his readers.19 Whoever wishes to take an opposite view in this difficult matter, may either read items given by Bernays in the footnotes on p. 111 or the sound account in H. P. Syndikus, op. cit., p. 346-356, or both. As far as an absolute word-end after the sixth syllable in the Asclepiadean verse is concerned, Bernays excuses its lack in line 17 purporting that Horace knew that Karthago is a compound noun; thus at least the compound boundary occurs at the position in question like that of de|torquere in 2. 12. 25.20 If, however, Horace knew linguistic details about the Punic Karthago, he would also have known that the ending -alis and the like in oblique cases of the names Hannibal, Hasdrubal, Hamilcar is to be scanned as a trochee. He, however, scans it as a pyrrhic as do all dactylic poets who have introduced the licence metri causa.21

In No. 10. Sprachliche Knacknüsse in Texten des Horaz (p. 91-97) Bernays appears as a 'nut-cracker'. He cracks a few of Horace's nuts to our satisfaction, others are, however, too hard for him, e. g. Carm. 3. 9. 10 citharae sciens, where Bernays maintains that sciens with a genitive is unusual in Latin and quotes 1. 15. 24-26 Sthenelus sciens pugnae ... with a statement that it is not sure whether pugnae ought to be taken as a genitive (p. 95 f.). Therefore, I must say, sciens: locorum (Sall. Iug. 85. 45; 97. 3), regionum (Iug. 100. 3), belli (Iug. 97. 5), earum rerum (Varro Rust. 3. 3. 1), agendi (Vell. 2. 88. 2), or scientior ... artificii (Rhet. Her. 2. 50),22 to recall only early examples, would have to be condemned as unusual Latin. But Bernays does not give up (do read p. 96!). He readily takes sciens pugnae for a dative, but if it has to be a genitive he would describe it as a Homericism. I ask why Bernays does not emend sciens to scia in 3. 9. 10 and to scius in 1. 15. 24, if he only regards sciens with a genitive as 'unusual Latin', but not scius viz. inscius with a genitive (p. 95)? In 1. 15. 24-25 he also does not take opus est into account and joins sciens with imperitare equis. 23 One is speechless. Over a page has been wasted to disqualify good Latin grammar, in order to say that Bernays would prefer to punctuate dulcis, docta, modos et citharae sciens (scil. Chloe), thus having an accusative object modos going with sciens. The only good thing on p. 96 f. concerns reading of Latin verses. Bernays justly attacks the old-fashioned school practice of stressing each longum with an artificial 'verse accent' wrongly called 'ictus' (cf. p. 100 n. 58). In Odes 3. 9. 9 he favours Peerlkamp's transposition regit Chloe (p. 95 n. 50), and in No. 11 devoted to this poem (p. 99-109) he develops further ideas from the preceding piece. I do not know why he sticks to the reading Thraessa instead of Thressa. In reading reiectae in verse 20 of this Ode I follow the MSS (see H. P. Syndikus, op. cit., p. 104, p. 109 n. 34). No. 13 is an interpretation of Horace, Odes 4. 11 (p. 121-127).

In No. 14. Strukturen der alkäischen Elfsilbler in den Römeroden des Horaz (p. 129-138), No. 15. Versstrukturen in der Ode 3. 11 und in anderen sapphischen Oden des Horaz (p. 139-146), No. 16. Strukturtypen des Pentameters bei Properz (p. 147-157) and No. 20. Strukturen des Hemiepes in Ovids Elegien (p. 181-192) Bernays focuses on observations about word accent and sound effects.24 These are his most valuable contributions to our understanding of the rhythmic structure and sound of ancient Latin poetry and should be given due attention not only by students of Horace, Propertius or Ovid. Only remarks on textual points require the reader's utmost caution. I say e. g. a clear 'no' to almost all the textual proposals of Bernays in No. 20, as he begins to reorder words in Ovid according to his ill-omened ideas governed by numbers and assonances. On p. 190 he even transposes words in Ovid, because "dieser Wortfolge entspricht die Übersetzung von Willige" (sic!). In No. 15 he justly supports Syndikus' defence of the fifth strophe of Horace's Odes 3. 11.25

Two further items devoted to Propertius are No. 17. Zur Gliederung der Elegie 1. 3 des Properz (p. 159-164) and No. 18. Ein umstrittener Vers bei Properz (p. 165-172). In the first one Bernays wishes to delete one elegiac couplet, verses 25-26; in the second one he emends Prop. 4. 5. 64: per tenues ossa sunt numerata cutes which shows a breuis in longo in the 'Versfuge' of the pentameter to: per tenues ossa est enumerare cutes (p. 167). In No. 19. Textprobleme in Ovids Elegie am. 2. 18 (p. 173-180) he deals with verses 26-34 of this poem. In the last two cases I leave the judgement to the interested scholars. In both there are some good points.

In the last four papers of the collection the author moves towards late and medieval Latin poetry. He focuses on imitation and on formal phenomena in short lyric verses or strophes. The conclusion of the otherwise illuminating paper No. 21. Die hipponaktischen Gedichte von Horaz und Prudentius (p. 193-202) seems to me simply absurd. If Prudentius indeed were about to prepare the way for accentuated poetry, there would be no need for him to imitate the metrical variety of Horace. The tendency Bernays is wrongly attributing to Prudentius is visible in Commodianus on the one hand and in Augustine's Psalmus contra partem Donati on the other. From No. 22. Das Orpheusgedicht von Boethius und dessen Pendent bei Seneca (p. 203-214) we learn that Herc. Oet. 1031-1089 was Boethius' chief model. In No. 23. Versstrukturen und Textgliederungen im Gedicht 2. 3 des Boethius (p. 215-221) Bernays observes the position and distribution of the word accent within the verse and discusses the exchange between accentually concordant and discordant lines. Finally, No. 24. Formale Aspekte karolingischer Lyrik des Klosters Reichenau (p. 223-237) is devoted to the Praise of Reichenau by Walahfrid Strabo (born ca. 808) and the Christ Hymn by Gottschalk (born ca. 803). Bernays shows that these two poems belong to two different styles. While Strabo looks back to ancient models, Gottschalk prepares for the new stream of accentuated poetry. I rather see in Gottschalk a tendency that will soon lead to dropping the quantity as the principle of the verse-structure than, as Bernays says, "raffiniert ausgeklügelte(n) Gebrauch von Reimen" (p. 228). Anyway, Bernays contributes here to the much discussed problem of the addressee of Strabo's poem; according to him it was Gottschalk.26

Apart from the Corrigenda affixed to the volume I have found 8 mistakes in Greek words (all easy to correct for a Greek literate) and 15 other errors of which the following are disturbing: p. 52 line 13 from the bottom read "4/5", p. 95 l. 11 from the bottom divide "ci-thara", p. 111 l. 9 read "28" instead of "27", p. 111 l. 14 and in n. 74 read "Borzsák", p. 174 l. 17 from the bottom read "das 8. Distichon", p. 184 l. 11 read "Kenney nach Lenz" instead of "Kenney".

Bernays is extremely sensitive to the form of ancient poetry. However, the approach to textual criticism displayed in these papers is not always satisfying. Manuscript variants are often not valued correctly, and poor readings are accepted as correct if they suit the views on symmetry or sound subjectively favoured by Bernays. It is a great pity, for many of his contributions are an excellent training in thorough observation, and philology needs young adepts with this skill developed. But Bernays invents far too much whenever the facts disagree with his expected calculation. The numerical discussions in particular often lead the philological discussion in the wrong direction.

To sum up: Bernays' papers are worth reading. Intelligent readers will take no harm from the unconvincing arguments, and will learn both what to do and how to do it well, and what never to do.27


Notes:


1.   The articles are given numbers by myself for the sake of convenience.
2.   Cf. G. Graf v. Gries, Sileno 28-29 (2002-2003), p. 271-289 (a review article on Helmut Seng, Vergils Eklogenbuch. Aufbau, Chronologie und Zahlenverhältnisse, Hildesheim - Zürich - New York 1999 (Spudasmata, 72)) and idem, Scripta philologica uetera, 2., veränderte Auflage, Göttingen 2005, p. 192-193 with adn. 7.
3.   On Kurt Witte see further Sileno, op. cit. (note 2), p. 278 with n. 29, p. 280-281.
4.   N. B. no book of the Odyssey consists of 486 lines.
5.   Even if it were true, 759 x 2 = 1518 against 1517. Here again Kurt Witte has to be acknowledged.
6.   Saying this I foreground textual points in which Bernays exaggerates the importance of the numbers at the expense of textual criticism.
7.   A good report of this problem is to be found in S. Koster, Ille Ego Qui. Dichter zwischen Wort und Macht, Erlangen 1988, p. 31-47.
8.   See also p. 41: Bernays counts 77 lines as an opening section of the Iliad 1 (v. 1-77) and states that the book also finishes with a section of 77 lines, but there are 76 (v. 536-611).
9.   On p. 40 we read about 122 lines, but on p. 41 Bernays says: "119 ... zu denen eventuell die Übergangsverse A 490-492 hinzuzuziehen sind."
10.   See e. g. an observation on Il. 1. 11 and 1. 356 viz. 1. 43 and 1. 357 and its bearing on book 1 and the whole epos (p. 39).
11.   See A. Weil, Sur quelques symétries dans l'Iliade, in: P. Hilton, F. Hirzebruch, R. Remmert (Edd.), Miscellanea mathematica, Berlin - Heidelberg - New York e. a., 1991, p. 305-309.
12.   For the titles of No. 5 and 6 see above p. 1. Verse 3 of Archilochus (p. 52) has a clear KATA\ TRI/TON TROXAI=ON after KU=MA as a chief caesura. In verse 27 of Tyrtaios (p. 56) GERAIOU/S can be read as an anapaest with AI shortened ante uocalem.
13.   N. B. A)NAIMO/SARKOS (verse 17) means 'with / of bloodless flesh', and not 'without blood and flesh'. Both v. Goethe's "ohne Fleisch und Blut geborne" (p. 10) and Bernays' "ohne Qual des Bluts und Fleisches" (p. 17) are semantically wrong. Bernays also forgets (p. 13 n. 7) that the so-called I subscriptum was normally pronounced as an element of diphthongs.
14.   Cf. p. 40 where Bernays emphasises Thetis' "reimartige Anklänge". He splits Il. 1. 421 and 422 until the penthemimeres in three lines to demonstrate it graphically: NHUSI/ - W)KUPO/ROISI - *)AXAIOI=SI, but he conceals the fact that Thetis ends the verse with an impressive assonance which is a sought means of Greek and even more of Latin poetry. Hom. Il. 1. 421 f. run as follows: A)LLA\ SU\ ME\N NU=N NHUSI\ || PARH/MENOS W)KUPO/ROISI / MH/NI' *)AXAIOI=SIN, || POLE/MOU D' A)POPAU/EO PA/MPAN. An almost unavoidable O(MOIO/PTWTON in this kind of line may be an acoustic signal preparing the strong cadence. Bernays, however, tends to favour rhyme in ancient poetry, quite wrongly, I think. Even if he can show that in Greek and Latin more sound effects are rhyme-resembling ("reimartig") than is commonly believed, it does not mean that rhyme was sought in Greek and Roman poetry. In fact it was not. On the Semitic origin of rhyme see W. Meyer, Gesammelte Abhandlungen zur mittellateinischen Rythmik, vol. I, Berlin 1905, p. 5-7, 277-283, vol. II, Berlin 1905, p. 116-118, vol. III, Berlin 1936, see p. 395 (Register). All three volumes were reprinted together: Hildesheim - New York 1970.
15.   On problems of structure see also D. Ableitinger-Grünberger, Der junge Horaz und die Politik. Studien zur 7. und 16. Epode, Heidelberg 1971, p. 55-59 and 112. On the symmetrical structure of Ovid's proem to the Ars Amatoria 1. 1-34 see K. Zeleny, Itali Modi. Akzentrhythmen in der lateinischen Dichtung der augusteischen Zeit, Diss. Wien 2005, p. 143.
16.   On p. 67 the same accentual structure as verse 34, 46 and 50 has also verse 54: aquósus Eúrus árua rádat ímbribus. Verse 65 must not be read with the monosyllable dehinc (aerea dehinc), but as aere dehinc: Horace always measures dehinc as an iambus which has been observed already by Meineke (cf. Keller, op. cit., p. 406).
17.   and also a trifle, but an important one for Bernays: Hor. Sat. 1. 6. 45-131 is 87 lines, not 88. Sat. 1. 10. 42 and 43 have not a bucolic diaeresis, but a diaeresis after the fourth spondee.
18.   See further H. P. Syndikus, Die Lyrik des Horaz, Bd. II, 3. Aufl., Darmstadt 2001, p. 348 n. 11.
19.   Bernays does even insist on commodis against the right commodus in line 1, see p. 117 n. 82 or he changes tibias to tibiae (p. 118), because it better caps tubae.
20.   By the way, such splitting of a word for the sake of a caesura or diaeresis is illicit (cf. p. 102, p. 104 n. 64). There are verses without a caesura which are correctly constructed in terms of quantity and accentuation.
21.   Cf. H. Mysliwiec, Zur Prosodie des punischen Namens 'Hannibal' im Latein, Eos 78 (1990), p. 315-324.
22.   Cf. also Cic. De Orat. 1. 214: uir regendae rei publicae scientissimus.
23.   For sciens with an infinitive see Hor. Carm. 3. 7. 25.
24.   Two studies are worth pointing out in this context: D. West, Reading the Metre in Horace, Odes 3. 9, in: Homage to Horace. Ed. by S. J. Harrison, Oxford 1995, p. 100-107 and K. Zeleny, Itali modi. Akzentrhythmen in der lateinischen Dichtung der augusteischen Zeit, Diss. Wien 2005 -- a recent general treatment of great importance.
25.   Cf. R. G. M. Nisbet, N. Rudd, A Commentary on Horace: Odes Book III, Oxford 2004, p. 157 f. See further U. Hübner, Im Unechten Echtes. Zur Hor. Carm. 4, 8, 33, Hermes 132 (2004), p. 241-243.
26.   Two further short papers by Ludwig Bernays appeared recently in Mnemosyne 58 (2005): Zum Text der Horaz-Ode 1. 4, p. 282-284 and Das anakreontische Gedicht des Boethius, p. 443-446 [= Boeth. 3. 7], a third one is forthcoming in Museum Helveticum: Zur Priorität der 4. Ekloge Vergils gegenüber der Epode 16 des Horaz.
27.   I am indebted to S. J. Harrison for having applied the arida pumex to my English.

BMCR, Bryn Mawr College, 101 N. Merion Ave., Bryn Mawr, PA 19010