There are not many books on textual criticism that are actually rooted in the practice of experienced editors—and Tarrant’s is one of the finest among them. Since it is neither voluminous nor expensive, I expect it will be widely read and equally widely reread, not only by philologists but also, as Tarrant wishes, by ‘classicists who do not specialize in the subject and [by] scholars in related disciplines’ (xi). Having said that, I am far from convinced that non-editors are the best audience for such a book, because, to my mind, it is much more an editorum in usum companion to manuals of textual criticism than an introduction to the field. It should be stressed that it is not a manual, i.e. not a work intended for untrained students: it requires at least a good knowledge of the bibliography, and ideally experience in editing.
Judging from the summary, Tarrant’s book may look indeed like a manual. The introduction deals with some general problems. Chapter 1 sketches a panorama of contemporary textual criticism as the heir to what Tarrant calls, using an expression coined by Derek Pearsall (21 n.9), a heroic age. Chapter 2 deals with the status of evidence in textual criticism. What remains is more technical: chapters 3-5 discuss respectively recension (reconstructing the oldest text available), conjecture (correcting it to go beyond the transmitted evidence), and alterations affecting the text on a larger scale (‘Interpolation, collaboration, and intertextuality’). Chapter 6, devoted to Propertius, is a practical implementation of chapter 5.1 Chapter 7 deals with the constitution and layout of the apparatus (complemented by the appendix, ‘Reading a critical apparatus’). Finally, chapter 8 is an open-ended conclusion.2
There are two leitmotifs that run throughout the book: the nature and value of evidence in textual criticism, and the subjective depreciation of the accidental (in its philosophical sense) nature of the transmission of texts.
First. Tarrant holds the view (advanced mainly in chapter 2) that evidence in textual criticism is non-existent and is only a matter of persuasion, that is, of rhetoric. A careful reader will soon catch from various explicit (41 above all) or implicit (or so I take the mundane wordplay in the title of chapter 2, ‘The rhetoric of textual criticism/textual criticism as rhetoric’) elements that Tarrant does not intend to be taken too literally. But this view is still generally irrelevant. It should be stressed first that textual criticism is a two-step process; the first part of the editorial work, that is the reconstruction of the archetype, has nothing to do with rhetorical play. It is a matter of pure logic, and as such is indeed the most ‘scientific’ part of the Humanities.3 Except in a few cases, stemmatics is a legitimate construction from the available evidence.4 Therefore, Tarrant actually criticizes not textual criticism, but only a part of it: the correction of the archetype ex ingenio (however he never says so, except 64). To be frank, classical textual criticism is full indeed of nonsensical arguments, of gratuitous textual parallels, and of frivolous claims made by people who do not know enough Latin or Greek and nothing of the history and the tradition of the texts they claim to correct. But the discipline should not be held responsible for the incompetence of those—even if they were a majority, which they are not—who wildly claim to illustrate it. And they are still probably not Tarrant’s target. In fact, he aims at the many places in classical texts where two or more possibilities cannot be definitively sorted out, but have nonetheless been the object of many assertive statements for over half a millennium. Let us hope that this criticism will be heard, for indeed it is all too relevant: there is no shame in stating that no firmly-grounded choice is possible. However, we should value carefully the alternatives. Tarrant praises Heinsius’ dealing with such cases, that is by leaving them open (46-47, 57-58, etc.). It should certainly not be inferred from this praise (and this is certainly not what Tarrant thinks) that relativism should reign over philology. Times have changed, and readers too; doing nowadays what Heinsius did in any series of Latin classics would be at best irresponsible (Housman would certainly have written ‘criminal’), because the average contemporary reader does not have the knowledge of Latin that his/her peers had in Heinsius’ times. The role of an editor will always be to make choices: the good editor is not the one who makes only good choices, but the one who justifies them and expresses their degree of authority.
Second. Tarrant makes a strong plea against neglect of, and more generally prejudice about, alterations to the classical texts throughout their history. This ranges from a condemnation of the traditional vocabulary of philology, which tends to consider texts as ailing bodies, to the defence of the intentions of ‘interpolators’. The former may be slightly overrated: many philologists use terms such as ‘corruption’ or ‘contamination’ because these have become technical terms, not because they dream of ‘crime and disease’ (31). The latter is not new from Tarrant, and has been quite unfairly criticized. The point is not to claim that any interpolation or variant is due (within the scope of classical editorship) the credit normally reserved for what is believed to be the authentic text: truth be told, this is nonsense. The vast majority of medieval scribes and scholars did exactly what we do: try to restore the text in front of them. Maybe they were wrong, maybe some even thought that they could challenge the greatest poets and prose-writers of ancient Rome, but they did not intend to deceive us and do not deserve our contempt.
In this context, I do not understand why Tarrant uses the outdated concept of error when dealing with stemmatic evidence. What is it that we name ‘error’? The fact that in some instances the text we have cannot, for whatever reason, stem back to the author. It may be an omission, an addition, or a change. Therefore it is no error, but an innovation recognizable as such. The use of ‘innovation’ instead of ‘error’ is not just a matter of language, or even of impartial objectivity towards the variants we have collected: it is a far more precise concept that lets us deal with such complicated cases as authorial errors or agreements in (non-traditional) truth at lower stages of the stemma.
A few remarks on the introduction. (13-14) The definition of ‘archetype’ could be narrowed by the correction Reeve made to himself (see n. 33), now in Manuscripts and Methods 94 n. 85: ‘In a set of witnesses to a text, the archetype is the latest but for which none that survives would survive’. (15) In the sentence ‘If the archetype did contain numerous variants, contamination in its descendants is almost inevitable’, the end seems to be mistaken for something like ‘contamination in its descendants is almost inevitably supposed’. (16-17) Though the stemma obviously dictates the correct text of Caes., Civ. 1.9.2, the definitive evidence lies in the parallel occurrences (e.g. 1.7.7).
Chapter 1 could be contextualized better, for its perspective does not really take into account the situation in Continental Europe, where the figure of Housman impresses much less. Something like a philological school is now almost specific to the Anglo-Saxon world. In France things are the exact opposite, and it partly explains why the Budé series is so badly received by Anglo-Saxon scholars: did it occur to them that the editions they review are, with few exceptions, made by people who were never trained in editing, even if they are very good historical or literary specialists in one particular author? Neither are, in general, their French-speaking reviewers so trained.
(50-52) The examples chosen by Tarrant are curious. The ‘errors’ of Cat. 49.7 and 101.7 are polygenetic, and so demonstrate nothing (by the way, has any advocatus diaboli tried to defend the authenticity of patronum as plural genitive in 49.7?).
(85-86) I agree with Tarrant against the authenticity of Ov. Met. 11.600-01. But it cannot be denied that Ausonius clearly has this verse in mind when writing Epist. 21.66 Green, both textually and symbolically (Juno asks Sleep to intervene because she is bored by Alcyone’s prayers for the return of her husband, who is actually dead), and Paulinus, in his answer to Ausonius, is clearly aware of it: Ult. 1[=Hartel 10].296 and Ov. Met. 11.601 are the only two occurrences of the phrase convicia humanae linguae. So the interpolation was already accepted in the 390s.
The appendix is arguably the most useful part of the book for non-specialists. However they should be extremely careful with the list of Latin adjectives (166-69), because it mixes without distinction places and former or actual collections (e.g. every Bononiensis is in Bologna but not every manuscript in the British Library is a Harleianus, and there never was any formal collection of Thuanei in the BNF). ‘nomen om. M, add. M2’ (164) though correct can be sometimes misunderstood: instead, use ‘nomen om. Mac’, or possibly ‘nomen om. M, exh. M2’. I take this opportunity to point out that the usual way of using ‘corr. <quidam>’ after a colon is erroneous: it does not mean corrigendo efficit but correxit (my thanks to Michael Reeve who pointed this to me, though it took me two years to eventually grasp it), and so it refers not to the lemma but to a rejected lesson. So e.g. in Mynors’ Catullus, the apparatus at 1.8 (habe tibi) should be ‘tibi habe V, corr. η’ (meaning ‘the archetype has tibi habe, which was corrected into habe tibi by “η”’), not ‘tibi habe V : corr. η (which should mean, but cannot, ‘the archetype has tibi habe, and the text printed is the result of a correction by “η”’).
In summary, this is an excellent book both in itself and because it puts so many things into question. It paves the way for a huge improvement in the editing of classical texts.
1. From where comes (117) the version of Prop. 4.7.4 which transposes murmur and nuper (contrary to what is written at 116)?
2. It concerns notably the use of new technologies, which convey the worst and the best. Among the latter I am surprised that the database Musisque Deoque, a wonderful achievement, is not mentioned. It is still in progress but, contrary to the Digital Latin Library (151), we can access and use it.
3. However the constitution of the stemma cannot be said to be ‘mechanical’ (14); what is mechanical is the use of it (when or where it has more than two branches) only.
4. Evidence may be scarce, but this is common to all classical, or even historical, studies; one can be blamed for using it unscientifically or not entirely, but not for having just what survived from the ancient world (on this regard, 15 n.37, what West actually demonstrates is that a well-constructed stemma can be used to reconstruct the archetype even if it is historically false). True enough, since all ‘originals’ are irremediably lost, we will never know for sure if the hypothesis (for indeed an edition is nothing but an hypothesis as a whole) is correct or not (particularly p.39??), and, worse, we do not even know if anything could really be called an ‘original’ ever. Nevertheless, it is equally true that the classical texts are far more accurate now than they were in the Middle Ages, and maybe even in Antiquity. So, it is not that vain to look for unrecoverable originals. Incidentally, the epigraph of chapter 3, from Guglielmo Gorni, is particularly unfair, since what he is speaking of is not recension but the orthographical layout of editions (a problem not really relevant for Latin texts), and since, furthermore, the first point of his essay is actually to demonstrate the original structure of Dante’s Vita nuova.