W.'s new Herodotus edition fully accomplishes its aims: to correct the shortcomings of the previous Oxford edition (Hude 1906, 1920 and 1927) and to provide a text as well as an apparatus criticus enriched by precious new conjectures of recent scholars, many of which are unedited, like those by MaasFootnote 1 (1.8.3–118 ἅμα … γυνή] codd. W.: del. Maas; 6.129.1–1661 τοῦ γάμου] codd. W.: del. Maas, etc.), while a few others are advanced by the editor himself. Some of W.'s new and interesting corrections have been adopted in the text: for example 2.35.1; 8.108.3.
For the aforesaid reasons, the work stands as innovative and precious; its apparatus criticus is straightforward and, compared with previous editions (especially H. Rosén's Teubner edition [1987–1997]), it appears significantly reduced: the choice to remove certain errors and the orthographic details is remarkable. W. rightfully assumes that ‘the manuscripts are unreliable and do not enable us to restore with confidence the forms preferred by the author’ (volume 1, p. vi) and claims to prefer the Ionic forms. Thus, some Attic forms are not accounted for, as in the case of ξυν- instead of συν- (5.93.2–1375 συμμάχων] ξυμ- CP, etc.).
The editor's decision to use a double asterisk to mark those passages which can be interpreted as afterthoughts by the author – like certain sections not properly integrated into the context (e.g. 5.9.3 del. Reiske), or certain duplications (e.g. 5.69.2, cf. 5.66.2) – is remarkable and truly innovative. W. does not exclude the possibility that many of these passages could be interpolations by ancient readers, but the choice to use the asterisk underlines the specificity of Herodotus’ text: a text composed to be read aloud to a large public audience or to smaller private groups, which ‘in order to ensure that the audience can follow the train of thought’ (p. vii) might present a certain amount of inconsistencies and repetitions.
An accurate philological monograph, Herodotea, accompanies the edition and enables contact with W.'s craft: most of the notes deal with passages which have raised questions in the past and require special attention. It also contains a detailed analysis of the choices made among the transmitted readings (e.g. Herodotea p. 48 on 3.9.3) and of the numerous conjectures, which are all thoroughly examined, both when they are accepted in the text (e.g. Herodotea p. 64 3.116.3–1776 πάντα Powell: αὗται A: αὐτά d; Herodotea p. 160 8.69.1–769 ἀποκρίσι Blaydes: ἀνακρίσει A: κρίσει d: ὑποκρίσι Richards; Herodotea p. 175 9.2.3–17 ἴσχειν Corcella: ἰσχυρὰ d: om. A) and when they are merely quoted in the apparatus (e.g. Herodotea p. 25 2.2.1 ἠθέλησε codd. W.: ἐξήνυσε Rödiger; Herodotea p. 193 9.110.2–1609 μοῦνον codd.: μοῦνος W.). In some cases, the commentary offers an overview of the choices of previous editors and, furthermore, it enumerates in detail the reasons that lay behind some of the innovative decisions made: for example in the note to 2.35.1 (Herodotea p. 28), W. initially summarises the reasons that led all previous editors to consider the reading ἢ ἡ ἄλλη πᾶσα χώρη incorrect; however he later on presents Herodotean parallels, which could reinforce the authenticity of the transmitted reading, ultimately to motivate his own personal correction of χώρη into γῆ, without neglecting a hypothesis about the genesis of a probable corruption in χώρη.
Generally speaking, the philological commentary could also become exemplary for considerations on philological methodology itself.
The apparatus criticus pays due attention to the variants in the papyri, which compensates for one of the deficiencies of the previous Oxford edition, while making a huge step forward compared with the last Teubner edition, where the variants in the papyri were already accounted for. W. does not overlook any kind of papyrus variant; therefore in the apparatus he also includes the readings of the most recently published papyri, neglected by Rosén: for example P. Oxy. 3376: 2.65.2 and 66; P. Oxy. 3379: 4.168 (used by Corcella [1993]); P. Oxy. 4455 (edited in 1998): 5.52 and 54. The most relevant contribution here is the fact that the apparatus includes variants attested in unpublished papyri (see, e.g., 8.112.1 and 8.130.2). Especially important is the case of 2.79.1–1153, where an unpublished papyrus (P. Oxy. inv. 85/40a) now confirms the conjecture ἁπαξὰ formulated by Hoffman (2.79.1–1153 ἐπάξιά d: ἅπαξ ἅ A: ἁπαξὰ P. Oxy. inv. 85/40a Hoffman). W., consequently, prints ἁπαξοί with Hoffman in 7.96.2–1218 (ἐπάξιοι codd.) too; in the commentary (Herodotea p. 33), W. reveals a consideration that Maas advanced to Powell, where he believed Hoffman's assumption concerning 2.79 to be ‘völlig zwingend’, and does not overlook Vannicelli's decisive contribution, according to which the unusual adjective ἁπαξά/ἁπαξοί was intended to mean ‘uniqueness’ (P. Vannicelli, QUCC 96 [2010], 87–97). Less certain is the restitution of ἁπαξοί in 7.81–1093 (ἄλλοι codd., cf. Herodotea p. 134).
The apparatus is readable and intelligible, but some omissions cause it to be less useful for a comprehensive understanding of Herodotus’ manuscript tradition than that of the previous Oxford edition. The archetype of Herodotus’ tradition had two lines of descent, a fact that has been widely acknowledged and accepted, even by W. (Herodotea pp. xiii–xiv): the Florentine family (A-Family W., a Hude) and the Roman family (d W. and Hude).
W. states that the stemmatic relations are not entirely clear (p. ix); he does not produce a stemma codicum of the extant manuscripts and the criteria he adopts to choose the manuscripts to be reported in the apparatus remain unclear.
The tenth-century Laur. Plut. 70.3 (A) and another two manuscripts which are close to A, though independent from it, namely Ang. gr. 83 (B) and Laur. Conv. Soppr. 207 (C), belong to the first family. W. favours only one manuscript, Laur. Plut. 70.3 (A), considering it ‘probably the best MS’ (Herodotea p. xv). Although he states ‘I cite some minor variants simply in order to confirm stemmatic relationships’ (Herodotea p. xiii), in the apparatus he reports only a few readings of B and C; thus, the readings common to A, B and C are mentioned as exclusive readings of A. Therefore, when facing the variants of A, there are doubts as to whether the mentioned A-variants must be accounted for exclusively as a readings of A, or as readings of the entire A-family (i.e. of A, B and C); hence, the relationships between the surviving witnesses and the division into two families remain unclear. Furthermore, in the commentary W. usually refers to A, not to the A-family: for example in Herodotea p. 47 he describes a misinterpretation of the uncial script of the A-family as an error of A.
The issue is particularly relevant because the first part of A (ff. 1–138) might not be original, but a facsimile copy.Footnote 2 In any case (leaving aside the complicated codicological configuration of the first part of A), the readings of the Roman family also transmitted in C are repeatedly printed in the text without specifying the concordance with C; in these cases, it might have been important to report the agreement between d and C in order to support W.'s choices, also through the stemmatic method (e.g. 2.19.2–285 πέλας AΒ: πέλασας dC W.; 2.96.3–1380 κατάρροον A: κατὰ ῥόον dC W., etc.). An indication in the apparatus of the cases when C agrees with d would have been important, also in the case where the reading of A is printed in the text (2.10.1–128 ἱρέες A W.: ἱερέες dC, etc.). Despite the explicit statement that C presents some excellent or plausible variants, due to some marginalia of the common source of A, B and C (overlooked by the copyist of A and B), or as a consequence of a contact with an unknown branch of the tradition (Herodotea pp. xv–xvi), the readings of C are only mentioned when they are considered to be of special interest (e.g. some linguistic particularities of C: 4.26.1–335 προτίθενται Ad: προτίθεαται C, etc.), apart – of course – from the cases where these are printed in the text (e.g. 9.105 -1520 Εὐθοίνου C: Εὐθόνου A: Εὐθύνου d, etc.).
Concerning the Roman family, as opposed to the A-family, W. does not only favour one manuscript, but reports in the apparatus the hyparchetype d variants, reconstructed through the tenth-century Vat. gr. 2369 (D) and through the lost ancestor of several later copies (the so-called β; which W. renames as r); the choice also to use the Pal. gr. 176 (Y), basically considering it a member of the Roman family (see Herodotea p. xviii and R. Cantore, Prometheus 31 [2005], 97–117), is innovative in relation to previous editions.
It might have been more appropriate to refer constantly to Y and, as exhaustively explained on other occasions (I refer to research I conducted, which could have not been taken into account by W.: R. Cantore, Per la storia del testo di Erodoto. Studi sulla famiglia Romana [2013], specifically pp. 177–82), not exclusively where it transmits a correct reading. More specifically, the concordances between Y and r are useful in order to exclude the singular errors of D: sometimes W. does not report the errors of D in the apparatus (e.g. 3.4.1–52 τοιόνδε A: τοιοῦτον d; τοιοῦτο D) and this seems to be correct; at other times, however, he reports the error of D, forgetting to indicate the important agreement between r and Y (e.g. in 3.23.2–364 W. prints κρήνην τινά and in the apparatus he writes τινά r: τινάς D: om. A; but Y also has τινά with r).
In order to reconstruct the text of the sub-group r, in the 1–2.120 section, it might have been more useful to mention consistently Laur. Plut. 70.6 (T) in the apparatus; for the whole text, on the other side, the use of Pal. gr. 215, instead of S (Cantab. Coll. Emman. 30), in addition to V (Vindob. Hist. gr. 85) would have been advisable. Pal. gr. 215 is a manuscript crafted together with V, in the same cultural environment and from a common lost source; S, on the contrary, is nothing but a copy of V and could be occasionally mentioned in the apparatus for only some conjectures (see ibid. pp. 40–52).
The use of the manuscripts U (Urb. gr. 88) and X (Vat. gr. 122) for the reconstruction of r throughout the text might be misleading. W. (pp. ix–x and Herodotea p. xviii) points out that, compared with the others manuscripts of the r group, these two manuscripts require special attention: they do not present some of the corruptions common to D and r and they do not share all the abridgements and gaps of r in Book 1, and, moreover, from time to time they agree (either individually or together) with A in the text's passages where D and r preserve a mistaken reading; consequently, especially in the initial books, U and X are quoted in the apparatus (e.g. 1.108.5–1684 παρεῖδες AUX: παρίδον DRSV, etc.).
In such cases, it might be possible to draw the erroneous conclusion that U and X are not liable to the corruptions common to the other manuscripts of the Roman family and, therefore, that they constitute fundamental witnesses in the reconstruction of the text. This might be a mistake considering the ‘physical aspects’ of the witnesses U and X: as clarified in several previous studies mentioned by W., the text of U and X in Book 1 does not derive from r, since for the initial part of the text both manuscripts (in dissimilar ways) clearly depend on a different antigraph than the one used to copy the remaining books of the Historiae.
With regard to X, the codicological and paleographical evidence account for the absence of the gaps of r in Book 1: ff. 9r-22v of Vat. gr. 122, containing Hdt. 1.42.1/Hdt. 1.112.2 (second and third quire of the manuscript), were written by a different hand from the scribe who compiled the entire manuscript. Therefore, these quires were incorporated into the manuscript by George Chrysococces, who corrected the entire codex through several marginalia and interlinear corrections. More specifically Chrysococces was able to finalise it with the aid of a manuscript different from the antigraph of X and which, as previously proven, presented a text very close to Laur. Plut. 70.6 (T) (G. De Gregorio, BollClass 23 [2002], 31–131, see 40–53).
T, only sporadically mentioned by W. in his apparatus, was written in 1318 by Nikolaos Triklines and, therefore, is one century older than X and U; it constitutes the witness that substantially explains the concordances of X and U with A. As proven by the numerous common errors found only in these two manuscripts, U is an indirect copy (if not a direct one) of T, up to 3.26 (1.27.2–349 ἀπικόμενον ἐς Πέρσας TU: ἀπικόμενον ἐς Σάρδις rell.; 2.57.1–883 διότι … ϕθέγγεσθαι] om. TU; hab. rell.; 2.91.1–1265 ϕεύγουσι … νομαίοισι] om. TU: hab. rell.; 2.110.3–1572 ἱστάναι om. TU: hab. rell.; 2.148.4–2299 ἐπ’ἐκείνοισι … ἑκάτερα] om. TU: hab. rell., etc.).Footnote 3
These errors, like many others attested only in U and T which W. does not mention in his apparatus, clearly illustrate the descent of U from T. The latter is hardly ever mentioned in the apparatus and, therefore, the few passages in which the concordance between U and T is indicated are misleading, as if these were the only cases in which the two manuscripts agree (e.g. 2.120.4–1738 παραλάμψεσθαι AX: παραλάμψασθαι DSV: παραλάψεσθαι UT).
In fact, in all the cases where W. mentions U in the apparatus (both for the individual readings and for the concordances of U with A), the readings of U coincide with T's, because U is an apograph of T. It should be noted that, in the second book, starting from chapter 130, one can find an ever-increasing number of quotations of U next to A in the apparatus (actually, the large number of cases of concordance is not systematically mentioned); this is due to the fact that T presents a ‘Roman’ text only up to 2.130 (approximately), after which T was copied from a ‘Florentine’ codex and, therefore, U, from 2.130 to 3.126, also preserves a text similar to A (cf. 2.137.1–2110 Αἰθίοπάς UTDa: Αἰθίοπά r–W.'s appar.: Αἰθίοπάς ADUXpc: Αἰθίοπά r; 2.139.1–2142 ἐς Ἑρμέω] ἐς om. aTU: hab rell.–W.'s appar.: ἐς d: om. A; 2.142.3–2203 πρὸς τροηκοσίοισί] πρὸς om. aTU: hab. d–W.'s appar.: πρὸς d: om. A; etc.).
Hence, in all cases where the concordances between A and U appear in W.'s apparatus, one should not assume that U's antigraph descends from r; the concordances between A and U up to 3.26 do not prove that U is a better witness of sub-group r than R S V. This is reinforced by the fact that, from 3.26 onwards, there are very few cases of concordance between A and U against R S and V; the rare concordances between A and U mentioned in the apparatus (e.g. 8.111.2–1355 Θεμιστοκλέος ARU: Θεμιστοκλέους DSVX) are also supported by other witnesses of r and can, therefore, be easily explained by the fact that the hyparchetype r derives from more than one source: it was contaminated by the A-family (Cantore, Per la storia op. cit., pp. 55–60).
