[Exento de evaluación]

This is the question that was put to me by Oxford University Press when I first proposed a new edition to them. ‘We published an Oxford Classical Text of the EE only in 1991; why on earth should we want to publish another one only 30 years later?’ (I paraphrase.) And it is a good question: normally the ‘Oxford Text’ of any Greek or Latin work is supposed to be the gold standard, and if Oxford itself starts replacing its own editions after a mere three decades, that might be seen to be a threat to the brand itself.

The truth is, however, that the problems with the 1991 edition of Walzer and Mingay were clear from the very start – or, to be more precise, in the following year, when Jonathan Barnes published a review of the volume. Although Barnes conceded that ‘it will surely be the standard text of EE for decades to come’, he concluded that ‘WM’s text is, on average, no better and no worse than S[usemihl]’s old Teubner [1884]. And as for the app[aratus] crit[icus], “slipshod” is a generous description’ (‘An OCT of the EE’, p.31). Here is Pierluigi Donini, in the introduction to his Italian translation of 1999: ‘… un’edizione largamente inaffidabile, incline ad accogliere congetture superflue e spesso palesemente errate, fino agli eccessi difficilmente giustificabili raggiunti nella costituzione del testo della pagina conclusiva dell’opera’ (xxiv-xxv); Donini goes on to print Walzer/Mingay’s text opposite his translation, but that translation is of a text that is in many places quite different from the one printed, and a significant proportion of the notes in Donini’s volume are devoted to signalling the differences between the OCT text and the one Donini actually translates. That text, i.e., Donini’s, owes much to Dirlmeier’s 1962 translation and commentary, a volume that – despite Donini’s generous praise of it – contains too many problems of its own: it is rife with bizarre interpretations, authoritarian in tone and approach, and singularly lacking in philosophical insight (it is now being replaced with a new translation and commentary by Friedemann Buddensiek; a replacement for the Dirlmeier Nicomachean Ethics, by Dorothea Frede, has already appeared). It might be true that Dirlmeier’s version of the text ‘rimane di gran lunga il meglio oggi disponibile per gli studiosi dell’EE’ (Donini, 1999), though Rackham’s 1935 Loeb edition is probably at least as usable, but anyone who tries to use Dirlmeier – as I have since it first appeared – will recognise that whatever advances it may make, in many contexts it adds more darkness than new light. Dirlmeier's EE, then, overall, represented no real substitute for Walzer/Mingay, despite all the latter's faults. Some recent English translators – A.J.P.Kenny in 2011 and Brad Inwood and Raphael Woolf in 2013 – are more complimentary about Walzer/Mingay than Barnes was, but nevertheless frequently find themselves having to deviate from the 1991 Oxford text, and I have to add that I think they also use the freedoms of the translator to skate over many other problems in the Walzer/Mingay text.

Walzer and Mingay had a big advantage over Susemihl, who in 1884 had an as yet only incomplete view of the relationships between the extant manuscripts. Our understanding of those relationships was transformed in 1971 by the work of Dieter Harlfinger, in the ground-breaking study of the manuscripts he prepared for the fifth Symposium Aristotelicum in 1969, published in the proceedings of the Symposium in 1971. Harlfinger established with certainty that the manuscripts of the EE all stemmed from a single archetype, no longer extant ('ω'), and that these manuscripts divided into two distinct families, one family represented by Vaticanus 1342 (‘P’) and Cantabrigiensis Ii.5.44 (‘C’) – both copied by the same hand, in the 13^th century, from the same source – together with their descendants, the other family consisting of Laurentianus 81,15 (‘L’), from the beginning of the 15^th century, and the descendants of L. Susemihl did not make use of L itself (as of course Walzer and Mingay did in 1991, having Harlfinger’s stemma in front of them); since Susemihl did use two of the manuscripts deriving from L, this did not make as much difference as one might have supposed (the lesser manuscripts tend on the whole to be fairly faithful copies), but it is not too difficult to see that once we know which manuscripts have the greater authority the matter of deciding between alternative readings in any part of the text is at least in principle much simplified. So Walzer and Mingay had important advantages over Susemihl in 1884.

There are, however, other crucial factors involved in editing a text. First and foremost, the editor needs as exact a knowledge as possible of the texts contained in the manuscripts. Susemihl’s reporting of the manuscripts he did use (Oxoniensis Corpus Christi College 112, or ‘Z’, Marcianus 213, a.k.a. ‘M.’, P, C, and Palatinus 165, a direct copy of P) is far from complete; Walzer/Mingay do better, for a much wider range of manuscripts, but I have discovered that they too misreport, on a considerable scale; they misrepresent P and therefore its relationship to C, its twin, particularly often. Walzer had evidently collated ‘many manuscripts’ in his youth, and inherited all the notes made by the great W.D. Ross (Sir David Ross), who had worked on a new EE text and actually submitted it to the Oxford Press, only for it to be rejected as needing further work. Eventually the project was entrusted to Walzer (or rather given back to him: he had actually given notice in Gnomon in 1951 of an OCT EE that was to appear under his name, though later in the same year it had apparently been passed on to H. Langerbeck). However Walzer – whose seminars on the EE in Oxford I attended as a graduate student – himself died before finalising his edition, and his student Jean Mingay was left with a good deal of work to do before it could finally be submitted. (I have here somewhat shortened the history of the 1991 OCT: the number of scholars who worked on the EE in the second half of the twentieth century in Oxford but died before its completion is somewhat worrying, putting one in mind of the lethal effects of the missing book of Aristotle's Poetics in Umberto Eco's Il nome della rosa. At least I have not worked in Oxford.)

But I digress: Ross's notes as inherited by Walzer apparently included the results of collations of ‘all’ the manuscripts (Walzer/Mingay 1991: 11); a student of Langerbeck’s contributed notes of a collation of Marcianus 213. But these remembered or second-hand reports on the manuscripts, even when added to Walzer's own early forays among them, evidently left much to be desired. We should not be too surprised: now we have the singular advantage of having some of the MSS online (L, for example), and the rest can be accessed through the microfilm collection in the Aristoteles Zentrum in Berlin. As a result, I have had instant access at least to imperfect reproductions of all the manuscripts, and can go back and check wherever I am in doubt; I may still make many mistakes, but mistakes are for the most part now avoidable, and readily rectified. Contrast that with the situation that obtained before the internet, and before the setting up of Aristoteles Zentrum in Berlin: any collator had to go and consult the manuscripts, scattered across Europe, in person – and occasionally one can find a log of all the people who have consulted a particular manuscript, and for how long. My own experience, having now been able to consult P, C and L in situ, is that accurate collation takes a very long time; that I could do what I needed in only two weeks with each of P, C and L was only because I had spent months looking at them beforehand in (photographs of) microfilm. In any case the inaccuracy and/or incompleteness of the collations behind the 1991 OCT already constitutes one reason why we need a replacement. As I wrote the first version of this paper, I had returned from a visit to Berlin to check on those of the ‘inferior’ manuscripts that are not in Florence and Rome; problems to do with the coronavirus pandemic forced the abandonment of a visit to Munich to see another manuscript that I shall describe in a moment, together with two copies of the Aldine edition that belonged to the scholar Petrus Victorius, i.e., Piero Vettori, at least one of which includes valuable marginalia in his hand.

A second reason for planning a new text of the Eudemian Ethics has already been mentioned: that the text printed by Walzer and Mingay is, as Donini put it, ‘inclined to welcome conjectures that are not only superfluous but often obviously mistaken’. The text of EE is, as everyone knows, in many parts highly corrupted: Harlfinger goes so far as to say (‘Überlieferungsgeschichte’, p.29) that the restoration of the text ‘obliegt letzten Endes doch der divinatio’. This is in my view something of an overstatement, but it is certainly true that the manuscripts, even taken together, sometimes offer what is, simply stated, no better than nonsense. Generations of editors, commentators and translators have then suggested their own solutions, to add to conjectures already made in the margins, and occasionally in the text itself, by medieval and Renaissance scholars, and choosing between the bewildering range of alternative readings on offer becomes extraordinarily difficult. Walzer/Mingay's choices, to put it bluntly, frequently do not inspire confidence.

But there is another problem, which is that once editors and critics have it in their head that EE is a corrupt text, they begin emending where there is no need for it. One absolutely certain truth about the EE is that it is written in what is typically a highly elliptical style. This is in itself part of the reason why the text is so corrupt (copyists who fail to understand a text are somewhat less likely to copy it correctly, though admittedly copying is demonstrably often a purely mechanical business: otherwise egregious errors would not be so often reproduced). But the elliptical, shorthand style of the work also encourages textual critics to step in where they should keep well away (cfr. Donini’s reference to ‘superfluous’ conjectures). My own experience is that finding the way to a solution to a problem involves first of all looking to see whether there actually is a problem at all, and then scraping off the layers of successive editorial interventions, quite often themselves developments or refinements of previous editors’ work, but generally useful only as a guide to how Aristotle might have written a particular sentence if he had been writing it out in full and in the best Greek style. In the vast majority of cases, my own restoration of the text involves either reinstating what one or more of the primary manuscripts gives us, or working out – from all the evidence, which includes corrections/glosses in those manuscripts – what they would allow us to reconstruct, with the fewest possible changes, in order to produce a good and appropriate sense. In short, the manuscripts are not in such a desperate state as has generally been assumed (‘vile’, says Jonathan Barnes, ‘hideous corruptions on every page’). It helps considerably that we are dealing with ´philosophical arguments`: we can get a long, long way with a proper application of philosophical charity (i.e., reconstructing the apparently best argument, out of respect for the author: and in any case how many really bad arguments are there in Aristotle? Very few, I would say). It has not helped that many of the editors who have worked on EE demonstrate a relatively weak grasp of the argument. (Dirlmeier (1962) is one example; von Fragstein (1974) makes some even more extraordinary proposals that could have been avoided with a clearer sense of the context and of the overall argument Aristotle can be seen to be making in that context.) Another regular failing among scholars, including Walzer and Mingay, is an over-readiness to rely on parallels with the Nicomachean Ethics – understandable, when as in Susemihl’s case there is not even a presumption that Aristotle is the author of EE (‘Aristotelis Ethica Eudemia’, it says on the outside cover of the 1967 reproduction of the 1884 Teubner volume, ‘[Aristotelis Ethica Eudemia] Eudemi Rhodii Ethica’ on the title page), but hardly acceptable in the 20^th and 21^st centuries, when we know both that EE is Aristotle’s and that in many respects it differs from NE.

A further reason why we need a new edition is that even the 1991 OCT occasionally makes no sense. In this respect Walzer/Mingay inadvertently reproduce a feature of both Susemihl and his starting-point, the 1831 Prussian edition of Bekker, a feature that in their case resulted from their being less interventionist, more concerned to follow the manuscripts (so Bekker, for example, goes so far as to print γνώσ [sic] at the end of EE VIII.1, because that is what he found in all the manuscripts he knew about; he does not fill in the obvious ending, εως, and neither, astonishingly, did any of the copyists of our Greek MSS – with one interesting exception that I shall come to). Usually in Walzer/Mingay it is the punctuation that goes wrong, much of it actually borrowed from Susemihl, because Walzer/Mingay used the Teubner text as their starting-point; but their paragraphing is also sometimes strange, breaking up the argument in unhelpful ways.

There is, finally, another and particularly compelling reason for producing a new edition of EE. This is that it is now clear that there are not just three primary manuscripts, as Harlfinger proposed, but four. In 1971, when Harlfinger’s essay appeared, this fourth manuscript, which Harlfinger labels Phillipps 3085 (‘Überlieferungsgeschichte’, p.3, n.8), was at the time inaccessible, but in 1976 it was bought by the Bayerische Staatsbibliothek in Munich, and became Monacensis 635 (this is the MS that fills in the ending of γνώσ at the end of chapter 1 of EE VIII; it has no known descendants that might have copied its restored γνώσεως). When I started to work on my new edition, it was Harlfinger who suggested that I should look at this manuscript. In 1971 (Überlieferungsgeschichte᾽, p.40, n.95) he said that MS Phillips 3085 certainly did not contain the three ‘common’ books, i.e., what we typically treat as NE V-VII: these L includes as EE IV-VI, as do P and C too, though they do not write out the three books, simply saying that they are identical to NE V-VII, which in P and C are to be found earlier in the same codex (of Aristotelian works, L has EE alone, but also includes the Hieroglyphica of ‘Horapollo’/'Horus Apollo'). Otherwise, Harlfinger said (in 1971) only that the manuscript was dated to the 16^th century, giving no precise indication about its relationship with other manuscripts. But in the Sotheby’s sale catalogue for the auction in which the manuscript was bought by the library in Munich, the detailed description, prepared with Harlfinger’s help, gives the date as ‘first half of the 15^th century’, and includes the following sentence: ‘Dr Harlfinger informs us that the present manuscript belongs to the Messina family [represented by P + C] and is particularly close to the Cambridge MS’. Giuseppe Di Gregorio, in an article published in 1991 (‘Aristotele tra occidente e oriente’, pp.493-4), goes further, and asserts ‘è praticamente certo che [C] costituisca l’esemplare diretto di copia del Monacense [= Monacensis 635]; oltre allo stesso taglio del testo, con gli incipit dei libri controversi, il M. presenta tutti gli errori del [C]’, etc. In fact the ‘taglio’, if that means the ‘style’, of Mon. gr. 635, is really quite different from that of C; it does not contain the incipits of the ‘common’ books, but instead numbers EE VII as EE IV and EE VIII as EE V; and it does not 'exhibit all the errors' of C, which are in fact are mostly shared by P before P was corrected, though it (Mon. 635) certainly belongs to the same family as P and C. Further, Mon. 635 – which I call ‘B’, it being now in the Bayerische Staatsbibliothek – on quite a number of occasions offers us the right reading where either P or C go wrong, or all of P, C and L do. My conclusion is that this manuscript, B, belongs to the same recensio as P and C, but is independent of them: the differences between it and them are so numerous as to make it actually impossible (despite some circumstantial evidence adduced by di Gregorio) that it derives from either of them. Β represents a separate line of descent that branches off from the line that leads to P and C, and does so before their source, α. This is why it sometimes gets things right that they get wrong; where it does better than all three of the others, i.e., P, C and L, this is because the hyparchetype on which P, C and B together depend – i.e., the lost source I label α´ in the stemma, thus reducing the status of α, the source of P and C, from hyparchetype, as it was in Harlfinger's original stemma, to sub-hyparchetype – has preserved the reading of the archetype, ω, i.e., the common source of all four primary manuscripts, that reading then being corrupted independently in the immediate source of P and of C and in L.

The addition of B to the list of primary manuscripts may be a long way short of revelatory, but it brings with it a mass of new information large enough by itself to necessitate the wholesale revision of the Walzer/Mingay apparatus. Taken together with the other reasons I have offered, this seems to me to constitute a cast-iron case for a new edition. Such a new edition I delivered to Oxford University Press in June 2021, along with a separate volume of Notes – of about 100,000 words in length – that gives detailed explanations of my decisions about what to print, in each and every case. I hope that this separate volume, along with a very full apparatus criticus attached to the text itself, will provide scholars with the basis either for disagreeing, or, as I would prefer, agreeing, with those decisions.

Christopher Rowe, Hebden Bridge, West Yorkshire, UK, October 2020