Philostratus’ Vita Apollonii Tyanei (VA; 3rd c. CE) is one of the most crucial books in the post-classical Greek canon. This pearl of the Second Sophistic appeals to all kinds of readers, from students of the novel, of historical grammar, of rhetoric and of medieval biography and hagiography, to historians and philologists. “The need of a new critical edition of the Vita Apollonii Tyanei (VA) by Philostratus has long been felt.” These words, written by Boter himself in a chapter published in 2009, were doubtlessly true. The last critical edition, by C.L. Kayser, dates back to 1870 (his editio maior dates back to 1844). Boter’s edition is therefore a much-anticipated addition to Bibliotheca Teubneriana.
Yet, more specifically, why is a new critical edition needed? Such an edition, it is generally agreed, should improve on previous editions in at least one of these aspects:[1]
- Description and collation of new witnesses (or previously imprecisely described / collated witnesses).
- Filiation of the witnesses (for example, by providing a new or a better stemma, which will allow for a more precise determination of the text of the archetype).
- Correction of anomalies in the archetype, whether by better judgement in the choosing of variant readings (selectio), by better use of the indirect tradition, by better use of known conjectures, by reporting unknown ones, by the production of conjectures of one’s own, etc.
Boter has advanced our knowledge of the tradition in all three aspects. In what follows I will quote not only from his edition but from two crucial articles of his: “Towards a new critical edition of Philostratus’ Life of Apollonius: the affiliation of the manuscripts”, in K. Demoen, D. Praet (eds.), Theios Sophistes, Leiden – Boston, 2009, p. 21-56; and “Studies in the textual tradition of Philostratus’ Life of Apollonius of Tyana”, Revue d’histoire des textes n.s., 9 (2014), 1-49.
Kayser made full collations of six out of seventeen manuscripts known to him. In 1983, E. Crisci made a partial collation of all seventeen mss.[2] Boter is aware of twenty-nine manuscripts, six of which he considers in this edition to be primary, and which he describes in detail (“Hac in praefatione tantummodo de codicibus quos vocant primarios agam”, viii). Boter’s collations, together with his description of the most important mss., are a clear addition to the study of the manuscript tradition of the VA.
In this edition, these collations are put to good use, first in order to produce a solid stemma (see below), secondly by supplying readings hitherto unknown. Boter is often able to replace the usual reading simply by a more careful collation, as on p. 99.12, where πρώτῃ, and not μιᾷ, is needed; ms. A (Parisinus gr. 1801, 14th c.) contained both readings (πρώτῃ μιᾷ), and the original reading was in all likelihood the numeral α´. The double reading of A had not been recorded before.
Boter does provide a stemma in this book, but he does not explain what errors allowed him to build it. Readers who feel the need to know by themselves whether they agree with Boter in the classification of families will have to delve into his contributions of 2009 and 2014. It would have been useful if the main findings had been repeated in this book—especially since Boter’s criteria are on occasion debatable, as can be gathered also from his previous contributions. We have seen that ms. A contained a double reading. Boter states more generally: “The archetype must have contained variant readings. This appears from cases as 46.20 [Kayser’s edition; 43.1 in this edition]: here F (which is the only primary MS deriving from γ here) has οὐ for σὺ; A and E have οὐ in the text, with σὺ added above the line; BMPU have σὺ in the text. For another instance, see 126.30 [Kayser’s edition; 111.20 in this edition], ξυγκλείσαντας Αγ: ξυγκλείσαντες Aslβ” (Boter 2009: 26). Yet it is by no means evident from these examples that the archetype contained variant readings. In the first one, οὐ is impossible; any clever copyist could have conjectured σὺ. This is perhaps even clearer in the second example, where an accusative is needed. In both cases the explanation may be that the archetype contained the faulty reading, and that the error was corrected independently by two or more scribes. Of course, it may also be the case that the archetype did contain variant readings. But Boter’s examples are of no value towards proving this. More crucial is another argument, quoted by Boter but due to another scholar. “Crisci (pp. 150–154) notes three places where double readings in A correspond to the readings of the two families: 22.8 ἤρετο Aslβ, ἔροιτο Aγ; 65.1 φέρεσθαι Aslβ, φαίνεσθαι Aγ; 284.4 ἐπὶ Aslγ, ἀπὸ Aβ” (Boter 2009: 26).
Despite the minor shortcomings underlined in the preceding paragraph, the advancement in the collation and description of the manuscripts is substantial. The advancement in the filiation of witnesses is no less clear.
Kayser divided the manuscripts into two families which are essentially identical to Boter’s β and γ (both dependent upon a lost subarchetype α in Boter’s stemma), with one crucial difference: Kayser took A, often the sole carrier of the genuine reading, to be a part of the first family; Boter argued in 2009, in the steps of Crisci, that A is a whole branch on its own right, equal in weight to α. The tradition would thus be bipartite: A on the one hand, family β and γ on the other. Family β would in turn be bipartite, with E against MPT; family γ also, with F and Q as only surviving witnesses, all other manuscripts being apographs.
In the present edition, Boter has diverged further from Kayser: E is no longer a part of the first family, as in 2009, but its only surviving member (already in 2014 he took MP to be apographs of E), while S is upgraded: it is now presented as a brother of A, whereas previously it had been placed in γ as an apograph of F. In truth, S is independent from F only between 265.17, where it switched exemplars, and 287.9, where it breaks off. Boter was aware of this fact, as were others before him, already in 2009.
Only four full manuscripts remain after the eliminatio descriptorum: A (first family) and E FQ (second family; Q begins at 128.1); S is not an apograph only for about 8% of the text of the VA, so its value is analogous to that of a fragmentary manuscript.
All families have significant separative errors (Trennfehler), which prove that none derives from another. Such an error of the first family (ms. A) is an omission of almost two lines (49.25-26); of the second family (β and γ), the presence of χωρίων instead of θηρίων (40.10). A significant error of E is ἀδικοῦντα instead of λόγος (258.13), and of FQ ἀναφύονται instead of ἀναφαίνονται (268.17).
By now it will be apparent that only three witnesses, AEF, are quoted throughout the book. Needless to say, it is a valuable addition to determine which witnesses, after a full collation and a new filiation, are inutiles. This does not mean Boter has solved all questions concerning the stemma. Already in 2009 he showed that A sides more often in error with β than with γ. He states that “[s]ome common errors may have arisen independently” (Boter 2009: 30). Surely so. But how come he is able to quote twenty times when A sides with β in error, and only four when it sides with γ? At least a postulation of contamination could have been tried. Furthermore, the explanation of common errors might be polemic. “Here A and β have πυροφόρα instead of πυρφόρα (the reading of γ)” (Boter 2009: 30). The reading is impossible; “missiles containing πυρός, ‘wheat’, instead of πῦρ, ‘fire’” (ibid.) makes no sense in the context. Boter adds: “But the corruption of πυρφόρα into πυροφόρα is easily made and may well have occurred in A and β independently” (ibid.). Indeed? It seems as likely that the archetype contained the absurd reading πυροφόρα, that the smart scribe of γ corrected it, and that A and β were faithful to their model. This possibility invalidates the example as a significant coincidence in error between A and β. In Maasian terms, the error is probably not conjunctive (contra Boter); while this is debatable, it is certainly not separative: any smart reader could have corrected it without any trouble.
An appendix of “Lectiones variantes minores” appears on pp. 297-301. The existence of this appendix leads the reader to wonder what is actually quoted in the critical apparatus. Does Boter only quote readings which could have been present in the archetype and conjectures good enough to be candidates to improve on said archetype? Quite a bit more, actually.
Boter claims that he may be open to criticism by adding a number of non-indispensable conjectures, for an apparatus “tantummodo ad textum constituendum servire debeat” (XXVIII.43). Yet variant readings of important manuscripts are regularly quoted in critical apparatuses, even though these manuscripts may neither be useful to establish the filiation of the witnesses nor have any chance to be the reading of the archetype. This is the usual practice, not a rarity. Two examples from the first page of this edition: 1.4 καὶ om. C, 1.6-7 οἱ θεοὶ μᾶλλον τὰ τοιαῦτα F (τὰ τοιαῦτα οἱ θεοὶ μᾶλλον the other mss.). In these cases, the readings of C and F are useless for the constitutio textus; the reading of the archetype is secure even without them and no correction is needed. The same happens with other critical editions. So, for instance, Mynors and Conte mention, in their critical apparatus to the Aeneid, that two later mss. have the interesting variant causa instead of causae in Aen. I.25, yet there is no possibility that this is the reading of the archetype. The editors, rightly, do not feel any need to justify the inclusion of this or similar readings in their apparatus.
My point is that an apparatus, unless extremely restricted, is not “tantummodo ad textum constituendum”.
With this we move forward to the third moment of standard modern critical editions (after collatio and recensio), often called examinatio: that of improving upon the archetype through a more careful selectio of variant readings of equal stemmatic weight, or through conjecture whenever the archetype seems to be corrupted.
Boter’s fear of composing an apparatus larger than expected is not directed to negligible readings present in important witnesses. He anticipates criticism for the inclusion of non-indispensable conjectures. He dismisses such a criticism with these words: “pietas erga maiores mihi maioris momenti est quam invidia eorum qui me hanc ob rem vituperant” (xxviii, n. 43). This humorous remark has an outdated aftertaste, not only because of the alleged pietas but also of the invidia attributed to other critics and, most importantly, because of the thought that respect, love or piety are sufficient philological reason – even jokingly – to include conjectures, or anything else, in a critical apparatus.
This said, I also believe that in case of doubt it is better to include interesting conjectures, though they may be superseded; and that reading a critical apparatus can also be a rewarding trip to the world of critics and philologists who exerted themselves to make a text more fully understandable, no matter whether in the end they failed.
All in all, Boter’s edition has made significant advances in all major features of critical editions. His collations are more systematic, his stemma more solid, and the conjectures he includes are enough to allow the reader to picture, even in problematic cases, what the genuine reading may be.
Notes
[1] For a more detailed discussion, see M.D. Reeve, “Cuius in usum?”, in M.D. Reeve, Manuscripts and Methods. Essays on Editing and Transmission, Roma 2011, 339-359, here 349. (Previously published in Journal of Roman Studies XC, 196-206.)
[2] E. Crisci, Ricerche sulla tradizione manoscritta della Vita di Apollonio di Tiana di Filostrato, bachelor’s thesis (tesi di laurea), Sapienza University of Rome, 1983.