Books 21–25 contain some of the most memorable parts of Livy’s Ab Urbe Condita : the beginning of the Second Punic War, Hannibal’s crossing the Alps and the Apennines, and the devastating Roman defeats at the Trebia, Lake Trasimene and Cannae. In 2016 John Briscoe published a magisterial new edition (henceforth: ed.) of this important text, which combines a deep knowledge of Livy and Livian scholarship with critical skill and independence of judgment. While the accuracy of this edition leaves something to be desired (see below), it constitutes a major step forward in the textual criticism of Livy. In 2018 it was followed by a valuable companion volume entitled Liviana. Studies on Livy, which makes some additions to Briscoe’s past work on Books 31–45, but whose main focus is Books 21–25: it contains an extensive list of corrections to the edition as well as a discussion of many of the problems affecting the text (see the table of contents below). Given that the edition and the companion volume constitute a meaningful whole, it seemed best to review the two of them together, even though this has involved a significant delay.1
Books 21–25 of the Ab Urbe Condita have reached us through the venerable Codex Puteanus or P (Paris, Bibliothèque Nationale de France, lat. 5730), an uncial manuscript transcribed probably in Italy during the fourth century CE. At the start of this codex, a number of leaves containing about two-thirds of Book 21 have been lost; but here we can rely on M and C, two copies made of P during the early Middle Ages, when it was still complete; and on the descendants of R, another early medieval copy of P that only survives in part. Studies by Michael D. Reeve and others have left us with a good understanding of the stemma codicum, so the editor’s task is now to document the readings of the authoritative sources (that is to say, those of P where it survives and those of its earliest descendants where it does not) and reconstruct the original text. The challenge lies in this final step, as the age of the Puteanus is not matched by its quality: almost every page is marred by serious corruption. Those who set out to edit these five books of the Ab Urbe Condita have their work cut out for them.
Briscoe’s achievements are best understood when set against three other modern critical editions of this pentad: the Oxford Classical Text prepared by C. F. Walters and published after his death by R. S. Conway (1929); T. A. Dorey’s editions of Books 21–22 (1971) and 23–25 (1976) published by Teubner of Leipzig; and the editions of four of these five books published in the Collection des Universités de France by Paul Jal (Book 21: 1988; 23: 2001; 24: 2005) and Fabienne Nicolet-Croizat (Book 25: 1992). In his preface to the 1929 Oxford Classical Text (at p. vii), Conway highlights the superiority of P and states that CRM are all copies of it; but the critical apparatus reports in full the readings of seven manuscripts PCRMBDA. Jal too acknowledges that P is the archetype,2 but he quotes the readings of sixteen manuscripts, while Nicolet-Croizat uses as many as eighteen, producing an impressive amount of irrelevant evidence. Dorey already put an end to this needless exuberance and based his edition on P only, quoting the other manuscripts only as sources of conjectures, except for those passages where the Puteanus is not preserved. His apparatus, however, is so austere that it conceals some of the corruptions in P.
Briscoe states that his ‘aim is to give all the information that is needed and not to give information that is not needed’ (ed., p. xx). He wisely follows Dorey in reporting in full the readings of P alone where it is preserved, which provides the correct basis for the reconstruction of the text. However, his apparatus is the opposite of austere: it is richer than that of any recent Oxford Classical Text I can think of, occupying over one-third of some pages. It is the result of a study of all the scholarship on the text of this pentad from the editio princeps of 1469 until today; the bibliography alone covers ten pages (ed., pp. xxx–xxxix). This very rich apparatus is backed up by fifteen pages of ‘Apparatui critico addenda’ that follow the text (ed., pp. 365–74) and, in Liviana, by a list of further readings of P (pp. 167–73). This approach has yielded a true editio maior of this pentad.
How has this rich material influenced Briscoe’s reconstruction of the text? Here too a comparison with his predecessors may be helpful. Walters and Conway produced a radically new text that incorporated some of their own supplements and conjectures. They were followed closely by Jal and especially by Nicolet-Croizat, although before them Dorey had already set out in another direction, adopting earlier conjectures and supplements that had been rejected by Walters and Conway, and proposing ones of his own. Briscoe’s edition constitutes a fresh point of departure. He has thought carefully about most problems affecting the text and even where Walters and Conway, Dorey and Jal or Nicolet-Croizat print the same conjecture or supplement, he has often adopted a different solution. It is especially useful that he discusses a couple of hundred problems in the central section of Liviana (at pp. 23–151), which constitutes in effect a textual commentary on this pentad and helps the reader to understand the reasoning behind his editorial decisions. (Briscoe only changed his mind about a handful of problems in the two-year interval between his edition and Liviana.)
Briscoe’s editorial decisions are of a very high quality; he combines linguistic sensitivity with erudition, critical skill and a laudable independence of judgment. As an editor, he is more cautious than most: he prefers to leave a corrupt passage in the text rather than applying a doubtful remedy. Unfilled lacunae (i.e., gaps in the transmitted text that have not been filled with a supplement) and especially cruces (which indicate corrupt passages that have not been corrected) are frequent in his text, to a degree that becomes striking if we compare his text with that of Walters and Conway. There, Book 21 contains only one unfilled lacuna and one obelized passage; in Briscoe’s edition the book contains one unfilled lacuna and as many as eleven obelized passages, one of which he declares to be sound in Liviana (pp. 38–39, on 21.37.7). In Book 22 the difference is even greater: there are no unfilled lacunae and two obelized passages in Walters and Conway, while in Briscoe we find five of the former and twelve of the latter. Dorey, Jal and Nicolet-Croizat follow Walters and Conway in this regard. If one reads Liviana carefully, one will realize what makes Briscoe’s edition different: where he feels unable to reconstruct the original text with a reasonable degree of certainty, even if only because several different reconstructions seem possible, he prefers to print the transmitted text between cruces or indicating a lacuna as appropriate. This editorial approach has the advantage of presenting the reader with what is certain and limiting inauthentic material in the text to a bare minimum. Its drawback is that it can produce a rough and disjointed text, even where this could be avoided by adopting a fairly inoffensive conjecture or supplement. In fact it is common editorial practice to print a conjecture or supplement that is not entirely certain, provided that the reader remains free to explore the alternatives in the apparatus; and a text that runs smoothly may be closer to the spirit, if not the letter, of the original than one that is disrupted by cruces and lacunae.3
What Briscoe’s editorial principles mean in practice may be elucidated with the help of 25.14.4, where he prints proxima forte †hostium† erat cohors Paeligna. Here hostium does not make sense, since the cohort in question is not Carthaginian but Roman. It follows from the logic of the narrative that it stands the closest to the Carthaginians, or to their camp. This meaning is yielded by all of the reconstructions that have been put forward. Briscoe feels unable to choose between them, so he prints hostium between cruces. To an inadvertent reader this might suggest that the passage is seriously corrupt, even though its meaning is clear and several reconstructions appear plausible to the editor.
In fact, there may well be an optimal solution to this problem. Briscoe notes that the ‘genitive hostium … can easily be emended to hosti (Madvig) or hostibus (Weissenborn); alternatively, Madvig proposed deletion, and it could indeed be a corrupted gloss by someone who thought that proxima required explanation’ ( Liviana, pp. 137–38). But it would be odd for a word meaning ‘nearest’ to lack a point of reference, and a glossator would hardly have construed proxima with the genitive. The word usually takes the dative; but one would be hard pressed to explain how hostibus or hosti could have yielded the genitive plural hostium, since nothing in the context calls for a genitive. It is easier to assume that the genitive is authentic and that it depended on a dative that has been lost. It is natural to think of hostium, which was already conjectured by Luchs and is supported in this context by 25.14.2 castris hostium iungi.
Briscoe rarely prints a reading that is downright impossible. There may be one such example at 21.10.2 Hanno unus aduersus senatum causam foederis magno silentio propter auctoritatem suam, cum adsensu audientium egit, which he prints thus. Here the transmitted text must be corrupt, since magno silentio propter auctoritatem suam cum adsensu audientium ‘is in total contradiction to 21.11.1 adeo prope omnis senatus Hannibalis erat ’ ( Liviana, p. 28). While the pars destruens of Briscoe’s reasoning is convincing, the pars construens is less conclusive: ‘[l]ike most modern editors, I have followed Eichhoff’s combination of Lipsius’ non with the transmitted cum assensu, the omission of non after suam being much easier than any of the other supplements proposed’ ( ibid.). But this supplement may be unidiomatic. The use of non cum with a noun in an ablative, without a supporting word such as sed, seems to be unparalleled. What we need is not only a negation of adsensu, but also an expression of contrast. The easiest supplement would be cum adsensu, but that might be too quick and flighty for this context of senatorial gravitas, and one would expect at sine adsensu. I believe that Livy may have written something along the lines of at haudquaquam cum adsensu. This should show two strengths of Briscoe’s editorial practice: his reporting of a broad range of supplements and conjectures and the fact that he explains many of his editorial decisions in Liviana, which is helpful even where one disagrees with him.
One area where this edition falls slightly short is in its reporting of the manuscripts. Out of those manuscripts that have source value, Briscoe has studied directly only one, namely O; of the rest he has either used a facsimile (PA) or a digital reproduction (RCBDEKQZMN; see ed., p. 1 n. 1, with p. xv). The readings of less important manuscripts are quoted from earlier printed editions, even when a digital reproduction is available online. It is understandable if an editor does not want to spend his time chasing up minor manuscripts, but it is surprising to find that Briscoe has not studied the Puteanus directly. His approach stands in contrast to that of ‘Conway and Walters, who spent a great deal of time examining P, [and] believed they could distinguish five different correctors’ (ed., p. viii). Briscoe comments that ‘they were far too optimistic’ and groups together all corrections to P under one heading ( ibid.), but it is hard to accept such a step backwards as long as it is not based on a study of the Puteanus itself. Could minute details of the script and the colour of the inks enable one to distinguish individual correctors after all?
It has been possible to distinguish the hands that have annotated another manuscript of Livy, namely A, better known as the Harleianus (London, British Library, Harley 2493). It was pointed out by Giuseppe Billanovich in 1951 that A had been annotated by Petrarch and by Lorenzo Valla,4 for whose annotations Briscoe uses the sigla A p and A v, reserving A c for those annotations that cannot be ascribed to either of the two (ed., pp. xvi–xvii). But in a conference paper that is about to appear in print (mentioned briefly by Briscoe in Liviana, p. 153 n. 1), Marco Petoletti argues convincingly that the fourteenth-century annotations in A should not be ascribed to Petrarch.5
Briscoe does his best to share the results of his extensive research. Bibliography is a major concern not only in the edition but also in Liviana, which starts with a chapter on the editorial history of Livy and also contains references to the exact manuscripts quoted by Briscoe in the apparatus as ‘dett.’ (chapter 9) and to the place of publication of the conjectures by Weissenborn, Madvig and H. J. Müller that are mentioned in the apparatus (chapters 10–12). The first pages of the initial chapter are more suited to the bibliographer than to the uninitiated reader; later on, there follow some entertaining anecdotes about the greatest Livian scholars of the twentieth and twenty-first centuries.
Part II of Liviana contains a short chapter (pp. 221–23) on the Paduan proto-humanist Lovato Lovati (d. 1309) and the α(a) group of manuscripts of the fourth decade of Livy. Giuseppe Billanovich had argued that Lovato had already known a text of the decade that ended at 40.12.15, as is shown by the marginal note hic deficit T. Livius next to the text of Justin 32.2.8 in a manuscript that he had copied himself, namely London, British Library, Additional MS 19906, at fol. 48v.6 In this short chapter Briscoe responds to an objection by Bernard Mineo about the exact contents of the manuscript of Livy that would have been known to Lovato, as is shown by the London manuscript. However, Petoletti has demonstrated what was suspected by Walther Ludwig and Michael Reeve, that Additional MS 19906 cannot have been written by Lovato.7 Briscoe mentions Petoletti’s article and acknowledges that ‘Billanovich … was almost certainly wrong to think that the manuscript itself was written by Lovato’ (p. 221 n. 3); but that leaves us with no reason to believe that Lovato should have known the fourth decade of Livy.
So far, I have focused on the rich contents of the edition and Liviana, and I have not considered their presentation. Here major problems emerge. The layout of the edition is generally adequate. Each two-page opening of the text is topped by a running title that indicates the beginning and end of the passage as well as the years covered. The paragraphs by Arabic numerals placed within the text, from which they are not distinguished typographically; this is less convenient, as modern readers may well take these numerals to be part of the text. The precision of the two volumes, on the other hand, leaves much to be desired. The edition and especially Liviana are littered with misprints and other errors (many of those in the edition are corrected in Liviana at pages 209–14). There is a misprint even on the title page, where the middle word of ‘adnotatione critica instruxit’ has been omitted. There are several mistakes in the stemma codicum (ed., p. vii), where the codices M and R have exchanged places, and between manuscripts Γ and C there has appeared a line that does not correspond to anything in the account of the transmission in the preface; also, manuscript F has not been indicated as a descendant of C (compare pp. x–xi). The text of the edition too contains several major errors. At 21.43.7 (p. 49 line 1), Madvig’s convincing reconstruction transcendes. autem? Transcendisse dico duos has become the unintelligible transcendes. <“Transcendes”> dico? Duos, in the text and also in the apparatus. At 22.60.9 (p. 153 line 5), the second word of ipsis plerisque regressis has been turned into plerique. And at 25.8.8 simul ab impotentibus tyrannis liberatas (p. 344 line 5), the three words in the middle have been omitted. I shall not discuss minor misprints, nor the mistakes in the apparatus, nor those in Liviana, most of which will be evident to an attentive reader who has the edition in hand. The number and kind of these errors are would be surprising anywhere, let alone in such a renowned series of critical editions.8 Briscoe’s long list of corrections in Liviana is useful, and his honesty admirable; but it seems unlikely that they should be added manually to most copies of his text. One hopes for a second edition with corrections.
Despite these shortcomings, this edition and companion volume are priceless contributions to Livian studies. It is hardly possible to do justice to every aspect of them within a review of these dimensions. They will surely be studied by generations of scholars to come.
Table of Contents: Liviana. Studies in the Text of Livy
1. Editions of Livy
I. Books 21–5
2. Textual Discussions: Book 21
3. Textual Discussions: Book 22
4. Textual Discussions: Book 23
5. Textual Discussions: Book 24
6. Textual Discussions: Book 25
7. The Annotations of ‘A z ’ in London, BL Harley 2493
8. Readings of the Puteaneus Not Cited in the Apparatus
9. Codices deteriores
12. H. J. Müller
13. Addenda and Corrigenda
II. Books 31–45
14. The Problem of 34.4.16
15. Lovato Lovati and the α(a) Group of Fourth Decade Manuscripts
16. Commentaries on Books 38–45: Further Addenda and Corrigenda
17. Edition of Books 31–40: Addenda and Corrigenda
Appendix: bonum / malum publicum
Addenda and Corrigenda
1. This review appears as part of research projects subsidized by the Agència per la Gestió d’Ajuts Universitaris i de Recerca (Agency for the Administration of University and Research Grants) in Barcelona (2014 BP-B 00071) and the Nemzeti Kutatási, Fejlesztési és Innovációs Hivatal (National Bureau for Research, Development and Innovation) in Budapest (OTKA 2015 PD 116524).
2. In his edition of Book 21, at p. lxx.
3. On the other hand, Briscoe has removed one kind of stumbling-block from his text by relegating interpolations to the apparatus rather than leaving them in the text between square brackets, as had been done by his predecessors.
4. Giuseppe Billanovich, ‘Petrarch and the Textual Tradition of Livy’, Journal of the Warburg and Courtauld Institutes 14 (1951), 137–208.
5. Marco Petoletti, ‘Episodi della fortuna di Livio nel Trecento e il problema del Harleyano 2493’, forthcoming in Gianluigi Baldo, Luca Beltramini (eds.), A primordio Urbis. Atti del Convegno Internazionale di Studi Liviani (Padova, 21-23 ottobre 2015) (Turnhout). I thank Marco Petoletti for having shared this paper with me before it was published.
6. Giuseppe Billanovich, La tradizione del testo di Livio e le origini dell’Umanesimo. Volume I: Tradizione e fortuna di Livio tra Medioevo e Umanesimo. Parte I (Padua, 1981), 6–10 and plate I; compare already id., ‘Petrarch and the Textual Tradition of Livy’ (see n. 3 above), 208.
7. Marco Petoletti, ‘I carmina di Lovato Lovati’, Italia Medioevale e Umanistica 50 (2009), 1–50, at 7–20.
8. The errors in the text would have been caught if it had been compared automatically with the digitalized text of another edition, for example with that of Walters and Conway, which Oxford University Press surely have on file.