Book 29 of the Dutch commentary on Ammianus Marcellinus has appeared at the now regular interval of two years after its predecessor. Three of the four authors have been on the team since Book 20 in 1987 (Drijvers joined for Book 22) and they look set to reach the 31st and last book in 2017. The applause merited by previous volumes (see my reviews of 25 and 26 in JRS 99 (2009), 294–6, and of 27 and 28 in JRS 103 (2013), 351–3) is equally due here. Reviews of commentaries will tend to pick up on points of detailed disagreement, but any quibbles below should be read bearing in mind the consistent thoroughness, good judgement and originality of the authors across linguistic, literary and historical scholarship.

In Book 29, as in 27, 28 and 30, sections tend to cover particular regions for periods of several years; as in earlier volumes, the authors follow the introduction with a useful chronological discussion, though the problems are less vexed in this book. Book 29 begins with treason trials under the emperor Valens in Antioch and elsewhere in the eastern provinces (chs 1–2); these should perhaps be seen as starting in a.d. 372 rather than in winter 371/2. The account forms a pair with the Roman magic and adultery trials at the start of the previous book (various significant intratexts are pointed out). The commentators also demonstrate the interesting likelihood that Ammianus used a handbook of magical practices for the famous scene where the conspirators divine the first letters of the next emperor's name, ΘΕΟΔ-. Meanwhile, in the West, under the baleful influence of the prefect Maximinus, the emperor Valentinian also permits cruel injustices (3), but remains an exemplary Commander-in-Chief (4). The long fifth chapter describes the Mauretanian campaigns of Count Theodosius, Valentinian's best general, against the rebel Firmus, between 373 and 375. Sallust's Jugurtha is an obvious influence. The fact that Theodosius' homonymous son later became emperor (fulfilling the conspirators' prophecy) has led to the confusing juxtaposition in Ammianus' narrative of panegyrical celebration with frank description of the hero's old-fashioned discipline. In a previous article (in the commentators' edited book Ammianus after Julian (2007)), Drijvers had sympathized with the view of Robin Seager, who argued beguilingly in Histos 1999 that Ammianus subtly and deliberately undermines Count Theodosius; however, the detailed examination of relevant passages here leads, regrettably but rightly, to a rejection of this argument. The sixth and last chapter tells how the treacherous murder of King Gabinius of the Quadi led to barbarian attacks across the Danube and how the younger Theodosius as dux Moesiae successfully resisted, before closing with the peaceful urban prefecture of Claudius (attested in 374), including a brilliantly impressionistic description of Rome transformed by the Tiber's floods into an archipelago: the commentators let their appreciation of the writing shine through at such moments. But in commenting on the fact that, after describing every prefecture of Rome between 353 and 372, Ammianus omits at least two (xii, xviii, 246, cf. ix), they should at least have mentioned the theory of Otto Seeck (Hermes 18 (1883), 291), recently revived by Timothy Barnes (1998, App. 8), that an account of these prefectures has been lost in the lacuna of 29.5.1.

For the text of Book 29 is not good. The principal ms, Vat. Lat. 1473 (V), is beset by a series of lacunae marked as around three lines long, between 29.3.4 and 29.5.1 and again between 29.5.22 and 36 (probably not coincidentally, these fall within a quaternion misplaced after 29.1.17 earlier in the transmission). For the most part Den Boeft et al. ably reconstruct the contents of lacunae, and are also in commanding form on textual problems elsewhere: they suggest or consider over forty changes from the text of Seyfarth's conservative Teubner. With a few exceptions where the text of V is restored (rightly with aliqua at 29.2.13 and procincti at 29.4.5, dubiously with consonans against Gelenius' consonos at 29.1.31), these are conjectural emendations, including about half a dozen of their own. The best is at 29.2.17, where they modestly credit comparison to Suetonius, Tib. 61.5 for poenarum maturitate (‘an early end to their tortures’); at 29.2.19 they use comparison to Ammianus' source Gellius in repairing the exemplary tale of Dolabella and the woman of Smyrna. There are a few places where problems in the text have not been spotted. At 29.2.6–8, they regrettably stick to an extraordinarily forced interpretation of the text, admittedly found in all translations that I know of, that makes the conspirator Heliodorus a court chamberlain — who then has secret discussions with the court! The passage is correctly interpreted by Josef Češka in SPFB 39 (1994), 139–45. At the start of ch. 6, the transmitted text tells us that the Quadorum natio mota est inexcita repentino: they point out reasonably that inexcita, unaroused, needs to be qualified by something like diu (transposable, following Heraeus, from a few lines below); but the problem with repentino is not whether it can be an adverb, which it can, but its position. The simplest solution is the conjecture of ms E: motu est excita repentino. As I have commented in previous reviews, Den Boeft et al. often adduce Ammianus' exceptionally regular accentual clausulation as a factor in textual decisions, but sometimes do not mention it when it supports a case made on other grounds: 29.2.24 nóta <ac> pervulgáta (Gelenius), 29.1.32 lítterae pósterae (Heraeus, for postrémae); they sometimes regard linguistic rarities as defensible even though against the cursus. In 29.4.5 there is no intrinsic problem with the participial form animati replacing a main verb, or with suspecti (E's emendation of V's suspencti) having an active meaning (though nowhere else in Ammianus), but prose rhythm requires animati <sunt >  and suspicati.

For all these minor disagreements, this is a model work of collaborative and interdisciplinary scholarship. Fans of Ammianus look forward eagerly to the quadriga Batavorum on 30 and 31.