[This is an article published in CR 55 (2005), 555-7. The published version may differ slightly.]
First, a warning. I had asked Professor Reeve’s advice while investigating Vegetius’ prose rhythm and had answered his queries on Thesaurus material. Since he also kindly sent me a copy of the book, I am ill placed to criticise it objectively; but it is clear enough that this edition of the Epitoma is greatly superior to its predecessors.
The main advance is in our knowledge of the manuscript tradition. There are more than two hundred witnesses, investigated, with a remarkable variety of approaches, in two earlier articles, ‘Editorial opportunities and obligations’ RFIC 123 (1995) 479-99 and ‘The transmission of Vegetius’s Epitoma rei militaris’ Aevum 74 (2000) 243-354. Reeve has three families, ε (ruthlessly weeded), δ (clarified by discounting contaminated witnesses and rightly dissociated from ε) and β (once represented by a single manuscript, now by two). A clearer picture of δ also shows that it ended at 4.39.1. Fortunately its place is taken by φ, the descendent of a δ manuscript, which supplied the missing ending from yet another source. For the main part of the text agreement of any two families against the third almost always gives the better reading. Each family occasionally preserves the truth when the others agree in error, ε most often, β very rarely. Doubtless there was manuscript conjecture and coincidental error; but there will also have been variants in the archetype. For instance, at 2 cap. 13 centuriis atque vexillis peditum, preconsonantal atque usually serves rhythm; is it chance that ε omits peditum to give Vegetius’ favourite clausula? But the same heading at 2.13 tit. has peditum in all families. Perhaps, then, peditum was an addition to the archetype in both places. At 3.12.1 one might even speculate about authorial variants: explora of εβ fits the patch of second person instruction in this part of book 3, rare elsewhere (see D. Schenk, Flavius Vegetius Renatus, Die Quellen der Epitoma Rei Militaris, [Leipzig 1930], 59-61); δ’s equally Vegetian convenit explorare might be an attempt to remove a potentially tactless imperative (admittedly others remain).
Reeve sees traces of authorial revision in two places where the manuscripts agree. The last sentence of the summary that prefaces the whole work is strangely isolated. Editors have deleted or transposed it to the beginning of 1.1, where it connects well with the following sentence, badly with the one before. Reeve sees the summary as a consequence of the amalgamation of the separately published first book into a larger work, with the final sentence intended as a link to the specific themes of book one. The whole summary is then out of place. Transposition to the end of 1 prol. is possible; but book one’s theme begins at 1 prol. 5. Perhaps it never found its proper place. At 2.15 Meineke suggested transposing sections 8-9 to follow section 5 and deleting from the beginning of 2.15.6 prima acies principum, secunda hastatorum armis talibus docetur instructa; Reeve follows, remarking ‘secundas … Vegetii curas hic Meineke pro prioribus’. Milner’s simple transposition is another possibility; but Reeve’s view gives an explanation for the displacement.
The introduction, in English, discusses not only textual questions but also such matters as author and addressee. There are some new arguments here; but the discussion of the tradition is more descriptive. Arguments for many claims must be sought in the earlier articles. The introduction also handles larger textual questions, such as the articulation of the text (the discussion of the chapter headings is particularly important) and the character of the archetype. Reeve thinks it need be no later than a copy put into circulation by Vegetius. This does not mean that there is no place for conjecture; but certainly where the archetype’s reading is certain, corruptions are neither many nor deep.
The text’s presentation is uncluttered and readable, with light punctuation, helpful paragraphs, few critical signs. The only misprint I noticed was 4.8.4 machinamanta. Thankfully Önnerfors’ section numbers are retained, with a few minor alterations (the index s.v. exspargere has not kept up, with 3.18.15 for 3.19.1). Tidiness has costs: conjectural additions like nota at 3.12.7 appear without brackets; the deleted first draft at 2.15.6 (see above) is removed to the apparatus rather than bracketed within the text.
The apparatus is clear and informative, positive when reporting the families, negative when reporting isolated readings of single manuscripts. Although simplified and selective, it seems to deliver more useful information than Önnerfors, even on spelling and morphology, in which it is complemented by an index. The level of reporting is very consistent; but one point might mislead. Some forms are regularised according to the general picture of Vegetius’ usage given by rhythm or manuscripts; but sometimes variants are reported, e.g. 2.20.4 ‘nil β: nihil εδ’, while the favoured form is elsewhere printed conjecturally without warning, e.g. 4.9.1 nil (consistency requires ‘nihil εβδ’). For accuracy I compared selected conflicts of reporting with Önnerfors for the ε manuscripts B(ern Burgerbibliothek 280) and M(unich clm 6368): Önnerfors was correct twice (3.8.14 repellet ε, 3.20.15 passos ε), Reeve eleven times (2.3.3 demittit B, 2.5.5 porro manere ε, 3.8.12 igitur M, 3.10.18 firmata ε, 3.15.4 passus om. ε, 3.20.19 defundetur B, 3.20.22 verum B, 3.20.27 strenuissimis ε, 4.9.2 ad ε, 4.14.2 de ε, 4.17.7 aspiciant ε). Reeve’s silence at 3.20.15 is probably deliberate; he warns us on p. xl-xli that he suppresses worthless variants in ε. This certainly gives a clearer apparatus, although a fuller reporting of ε’s implausible variants might have helped us in judging its plausible variants. We cannot rely on Önnerfors for this: for instance, neither editor reports the ε variants 1.1.5 prudentiamque, 2.4.5 exercitos, 2.15.1 corno. Some guidance is offered on textual questions, occasional parallels and frequent references to a third article, ‘Notes on Vegetius’ PCPS 44 (1988) 182-218. For some reason discussions in ‘Editorial Opportunities’ are rarely mentioned (e.g. p. 494 on 3.6.14, p. 496f. on 3.19 tit.).
The text differs from and generally improves on Önnerfors in nearly two hundred places (more if you count spelling and morphology). Much hardly affects the sense; but note for instance 3.15.5 volueris aciem tenuare, 4.8.5 equosque deterere, 4.9.5 cataractas tegendas, 4.41.4 nive. On several decisions there is still room for debate - a few are discussed below; but major advances are now unlikely (some points might become clearer with a satisfactory edition of the Mulomedicina). This is, then, likely to remain the standard edition for some time. All we need now is a Loeb.
Detail. 1.5.4 non tam staturae rationem convenit habere quam virium. et ipso Homero teste non fallimur (fallitur εβ). Is δ’s fallimur not a conjecture? It would be better if e. g. monemus had come before. The unsatisfactory fallitur might be authorial error, as though the previous sentence had contained dux peritus eligit (β adds hoc quisquis effecerit before et); someone (surely Vegetius?) who had forgotten the construction of the previous sentence also added literary adornment at 3.21.6. 3.18.14 Reeve rejects the transposition of the chapter division to before tamen ars belli (better before 3.19.2 cavendum), explaining the ungrammatical text that follows the title as an anacolouthon; but that leaves a chapter whose first paragraph is irrelevant to its heading. Once the chapter headings had been entered in the text, misplacement could hardly occur; but between writing them for the table of contents and creating the fair copy, it would be easy enough. A scribe (or Vegetius) might have been confused by the apparent suitability of superventibus to the chapter description. Similarly the heading to chapter 4.46, which would begin better at 4.45.4, is placed before a sentence in which alto et libero mari might have seemed appropriate to aperto marte of the title. 4.1.2 ‘tutissimo melius abesset’. Perhaps tutissima. 4.38.12 Aparcias rightly defended. Add that ἀπαρκίας is known to LSJ and that the spelling is transmitted for other Latin writers.
Nigel Holmes, Thesaurus Linguae Latinae.
[Update: I somehow failed to notice that Scriverius had already made the conjecture tutissima at 4.1.2 and that this was reported in Lang, the standard edition of the nineteenth century.]