Volume 3 of the Budé edition covers Book 3 of Proclus' Parmenides commentary. Part 1 comprises two chapters, followed by a list of abbreviations. The first chapter includes a critical overview of previous editions, by V. Cousin (1821–1827 and 1864) and the OCT by C. Steel et al. (2007–2009). Two appendices follow recording discrepancies with the Oxford edition, and responding to Steel's review of Volume 1 of the Budé (Mnemosyne 63 [2010], 120–42); three further appendices list the non-significant variant readings from the main manuscripts. A second and considerably shorter chapter analyses Proclus' argument. Part 2 comprises the text with facing French translation and notes, a list of scholia, ‘complementary’ notes to the text and translation, and further ‘supplementary’ notes. It ends with indexes of names, sources and themes, a list of errata and a table of contents.

This complex structure illustrates the strengths and weaknesses of this edition. The text, translation and notes are a monument of impeccable and rigorous scholarship, based on a meticulous analysis of a complex manuscript tradition and a solid knowledge of Proclus' philosophy. Through a detailed study of Proclus' doctrines and style, careful cross-referencing to Proclus' other works and irrefutable use of palaeographical evidence, the editors have advanced a number of brilliant conjectures, which considerably improve the text and clarify a number of hitherto obscure passages. It provides for the first time an accurate translation of the text, since previous translations were principally based on Cousin's unreliable edition.

However, the exceptional quality of the work is overshadowed by a mass of extraneous material, endless subsections and appendices, which, in the present reviewer's opinion, should not have been published. The adversarial tone used to present this material indicates that the editors wished to mount a virulent, point-by-point attack against the Oxford edition, which was published after S. and his collaborator C. Steel split over the interpretation of the manuscript tradition. As such, most of Part 1 will be of little use to the general reader, and make for very unpleasant reading.

Much has been said about the differences between the Oxford and the Budé editions. Suffice it to say that the complexity of the manuscript tradition makes it very difficult to record in a systematic and consistent way all significant variant readings in the apparatus criticus. Indeed, the two most important testimonies (Parisinus gr. 1810, = A, which is the oldest extant manuscript, and Moerbeke's Latin translation, which preserves another, better and more complete text than the extant Greek manuscripts) are, to various degrees, revisions of the original text. Manuscript A is the result of a substantial revision by the Byzantine scholar George Pachymeres. However, it is not always possible to distinguish with absolute certainty between A's original model and Pachymeres' revision. In this context, the Oxford and Budé editions differ on the way they record the variant readings in the apparatus: L. and S. view all variant readings of A as Pachymeres' conjectures, thus to be recorded only if they are adopted in the text. In contrast, Steel et al. adopt a more prudent approach and record all readings of A in the apparatus, allowing for the fact that some variants in A might well be authentic readings from Pachymeres' model.

Similarly, it is sometimes difficult to determine whether Moerbeke's Latin translation reflects a better text, or is the result of a revision (due to e.g. the constraints of the Latin syntax, the non-mechanical character of the translation, misinterpretations or corrections): in these cases, it is not always clear whether or not the editors should include Moerbeke's readings in the apparatus.

A good example of the way in which the two editions differ is at 784.9, where both texts adopt Bessarion's (= M^b) conjecture ὅπερ αὐτὸν against the manuscripts' (= Σ or s) ὃ παρ' αὐτὸν. The Oxford apparatus reads ὅπερ M^b: ὃ παρ' Σ ὃ πρὸς A; the Budé edition reads ὅπερ αὐτὸν M^b: ὃ παρ' αὐτὸν s quod … ab ipso g. Moerbeke's (= g) reading is probably a bad conjecture (see p. 93 n. 5: ‘La traduction latine quod ab ipso est probablement une conjecture de Moerbeke sur ὃ παρ’ αὐτὸν'). It is therefore not mentioned in the Oxford edition, yet it is recorded in the Budé apparatus; in contrast, the Oxford edition mentions Pachymeres's (= A) own conjecture, whilst the Budé edition does not. This example illustrates how difficult it is to accommodate the textual evidence to the editorial conventions of the apparatus criticus. It also shows that the Budé edition is not always as consistent and systematic as it purports to be. Had it respected its editorial criteria à la lettre, it would not have recorded Moerbeke's conjecture in the apparatus: for instance, at 784.17 πρώτου A: πρῶτον s, the apparatus does not record Moerbeke's reading (primo) because primo could render either πρώτου or πρῶτον (p. 97 n. 8); the same criteria are applied at 804.18 and 809.21, where Moerbeke's readings are too ambiguous to be of any textual value and are therefore not included in the apparatus (p. 177 n. 3; p. 196 n. 4). Another inconsistency is at 792.15 (ἢ scripsimus: ἡ Σ om. g): the editors rightly state that Moerbeke probably read the article ἡ in his exemplar but did not translate it (p. 125 n. 8), which implies that the absence of the article in g should not have been recorded in the apparatus.

These minor inconsistencies make all the more surprising the numerous stings at the Oxford editors, who are systematically accused of philological incompetence when they record variant readings that might not (but could) be significant. More generally, the editors' bias against the Oxford edition leads them to make a number of unfounded assumptions. For instance, at 788.21–5, L. and S. castigate Steel et al. for stating that they have corrected the text on the basis of Moerbeke's translation and Bessarion's good conjecture (the Oxford apparatus reads πολλῷ … ἀγαθόν huc transposuimus ex g et M^b), on the grounds that the Latin translation was sufficient for establishing the correct text. Instead the Budé editors devote a page and a half to an explication of Bessarion's conjecture (pp. 116–17). In the reviewer's opinion, it seems more likely that the Oxford editors chose to record Bessarion's conjecture in their apparatus because this was the most economical solution. Another instance is at 819.19, where the Budé edition clearly misrepresents the Oxford editors' intention: as the Oxford apparatus makes clear (τῇ ὕλῃ scripsimus (cf. secundum materiam g): ὕλῃ Westerink ὕλη A Cous), and contrary to what L. and S. imply (p. 233 n. 2), Steel et al. were well aware that Moerbeke's version could not be used for conjecturing the presence of the article.

It is highly regrettable that these rather mean-spirited attacks have not been expunged from the published version of the book, because they risk discrediting the otherwise excellent quality of the work. There are indeed numerous instances where L. and S. considerably improve the text, by adopting variant readings that the Oxford editors have missed (see e.g. pp. 106–7, 125, 143) and by offering several excellent conjectures: see, for instance, 794.21 (λογοῦντος scripsimus: ἄγοντος Σ ducente g); at 803.12, ἐκεῖνα γὰρ αἰώνια πρῶτον τὰ … ἐπορέγει τὴν μετάδοσιν is corrected into ἐκεῖνα γὰρ αἰώνια πρώτως ὄντα τούτων … ἐπορέγει τὴν μετάδοσιν on the basis of the Latin translation (primarum harum) and a reconstruction of the progressive corruption of the text (see p. 169 n. 8); at 819.21–6, a meticulous analysis of the palaeographical evidence not only restores the verb ⌊προ⌋ϋπάρχειν, as in the Oxford edition, but also {προ}ὑπάρχει; at 803.17, the manuscripts' nonsensical καὶ τῶν πολλῶν is admirably corrected into καὶ πολλά.

The Budé edition is a considerable scholarly achievement, which contributes greatly to our understanding of a text that played a fundamental role in the history of philosophy. Sadly this achievement is weakened by the editors' questionable bias against the Oxford edition – all the more so in that, as a quick glance at the apparatus shows, the Budé edition remains greatly indebted to the work of Steel and his team.