Thursday, 30 August 2018

The Last Two Volumes of the Ammianus Commentary

The following review, recently pre-published online, will appear in the 2018 volume of Journal of Roman Studies (for earlier thoughts arising, see here).

J. den Boeft, J. W. Drijvers, D. den Hengst, and H. C. Teitler, Philological and Historical Commentary on Ammianus Marcellinus XXX. Leiden: Brill, 2015. Pp. xix + 257, maps. ISBN 9789004299955 (bound), 9789004300927 (e-book). €112.

J. den Boeft, J. W. Drijvers, D. den Hengst, and H. C. Teitler, Philological and Historical Commentary on Ammianus Marcellinus XXXI. Leiden: Brill, 2018. Pp. xix + 357, maps. ISBN 9789004353817 (bound), 9789004353824 (e-book). €169.

Near the start of his 30th book Ammianus Marcellinus flashes forward to the disasters of the Gothic war under Valens and wonders if his narrative will ever get that far (si ad ea quoque uenerimus, 30.2.8). It is a rhetorical device, of course, one of many indications of the arduous nature of the historian’s task. The authors of the Dutch commentary on Ammianus must also have wondered if they would get to the end of this mighty enterprise. Its history reaches back more than 80 years. Pieter de Jonge produced the volumes on the first six surviving books, 14-19, amid his duties as a headmaster between 1935 and 1982 (English replacing German from 1948 onwards). While De Jonge’s dedication, ambition, and feel for his author merit admiration, these volumes are uneven, especially on historical topics, and will need replacing in due course.

This cannot be said of the work of his successors, known informally as the Quadriga Batavorum, who have now completed the remaining 12 volumes on Books 20 to 31. Three of them, the Latinists Jan den Boeft and Daniël den Hengst and the Roman historian Hans Teitler, have been working on the project since the 1980s; a second historian, Jan Willem Drijvers, joined for Book 22 in 1995. Since the completion of the second hexad (20-25) in 2005, the team has accelerated, publishing the remaining six books at the rate of approximately one every two years. The insight, comprehensiveness, and reliability of the commentaries have only increased over the years, and the last two volumes here reviewed are well up to the exacting and exalted standards of its predecessors (see my reviews of Books 25 to 29 in JRS 99 (2009), 294-6; 103(2013), 351-3; and 105 (2015), 475-8).

Ammianus’ last two books have contrasting structures. Book 30 is dominated by the death on 17 November 375 of the emperor Valentinian, and a lengthy tripartite obituary (chapters 6-9), but before that, as in the previous books on Valentinian and his brother Valens, the narrative switches between eastern and western events in geographically distinct blocs often covering several years: in chs 1 and 2.1-8, Roman foreign relations with Armenia and Persia from c. 374 to 377/8; in chs 2.9-12, 3, and 5, western events of 374 and 375; in ch. 4 an entertaining, overwritten, and highly allusive digression on the venality and incompetence of eastern lawyers, hanging on the emperor Valens’ withdrawal from hearing court cases. The book closes with an account of the elevation of the four-year-old Valentinian II by junta of high officials five days after his homonymous father’s death (ch. 10). Book 31, by contrast, is all but monographic (but the commentators reject Kulikowski’s proposal (JRS 102 (2012), 79-102) that it was originally a separate monograph written much earlier): the book focuses almost exclusively on how the Gothic migrations in response to the pressure of the Huns turned from a peaceful crossing of potential allies in 376 into a Roman-Gothic war, climaxing on 9 August 378 in the Battle of Adrianople and Valens’ death. Certain important events of the period – Saracen and Isaurian revolts – are left out altogether, and the only western episode, the successful campaign of the emperor Gratian against the Lentienses (ch. 10), is only included to explain why Gratian was delayed in bringing reinforcements to the east and why his uncle Valens’ jealousy led him to engage without awaiting them. The ‘Chronology’ in the front matter of each volume and the excellent chronological guidance passim is very welcome, and in Book 31 the commentators also point out Ammianus’ historical omissions.

Historical contextualisation is uniformly thorough. In the initial episode of Book 30, where various generals of Valens attempt unsuccessfully to detain, and then successfully to assassinate, his ally king Papa of Armenia, the commentators use Armenian sources to point out the religious subtext ignored or suppressed by Ammianus: they are perhaps slightly more willing than in previous books to see oblique jibes at Christian piety (see on 30.1.2). Equally good is the clarification of Valentinian’s movements in his last campaign on the Danube in 375, and they rightly point out the clear signs of a whitewash in the narrative of Valentinian II’s promotion (and suggest, correctly in my view, that Ammianus was therefore writing before Valentinian II’s death in May 392). If I may be permitted a personal quibble, however, their argument that this promotion took place in Brigetio (Szöny), where the young emperor’s father had died, is false. Neither Ammianus nor any other ancient source explicitly states this, though Hadrien de Valois evidently inferred it from Ammianus’ text, as indicated by his chapter heading for 30.10. But the chronicle known as Descriptio consulum or Consularia Constantinopolitana, which is exceptionally reliable for the 370s and 380s, places the event in Aquincum (i.e. the right bank of modern Budapest), and this is surely correct. It is irrelevant (p. 204) that Socrates’ church history, based on a version of this same chronicle, garbles its source and puts Aquincum in Italy. Also irrelevant is the fact that Valentinian I had not thought Aquincum suitable winter-quarters for himself, since emperors on the northern frontier would not necessarily winter in the same place as most of their troops, who were scattered over the whole region: e.g. 17.10.10 (Julian in Paris; compare also the situation at the time of his acclamation in 360), 27.10.16 (Valentinian and Gratian in Trier). Indeed the point is implied by the difficulty in finding a doctor to tend the dying emperor, as they were tending to an epidemic among the soldiers and scattered per varia, 30.6.4). Aquincum was the provincial capital of Valeria with a massive military camp and a second amphitheatre, the largest north of the Alps, for the army: it was the obvious place for an acclamation that would be approved by as many troops as possible. In Book 31, a particularly significant contribution is the late dating of Gratian’s campaign against the Lentienses in 378, in summer rather than spring, ensuring that he really was very late indeed. The massacres of Gothic teenagers in the eastern provinces after Adrianople (31.16.8) are dated to late 378, against Zuckermann’s proposal of early 379.  

While Book 30 has been transmitted in a slightly better condition than 28 and 29 (some passages of which are marred by repeated lacunae), Book 31 has particular problems of its own. The better of the two ninth-century manuscripts of Ammianus was the Hersfeldensis, which was haphazardly used for Gelenius’ 1533 edition before being dismantled in the late sixteenth century, to the extent that only a few recycled pages survive, a bifolium from book 30 among them. But by Gelenius’ time, the Hersfeldensis went no further than the penultimate chapter of book 30, where Gelenius accordingly stopped. For Book 31 the Vaticanus (V, Vat. Lat. 1873) is therefore the sole authoritative source for the text. An additional problem arises between 31.8.5 and 31.10.18, where at some point after the first copies were taken the Vaticanus lost two pages (the central bifolium of a gathering, not a single page, as on p. 146 and in most Ammianus scholarship). The apparatus of Seyfarth’s Teubner, from which the lemmata are taken, is inadequate here, as was sharply pointed out in the review by Rita Cappelletto (RFIC 109 (1981), 80-85), and unfortunately the commentators do not really make up for this inadequacy. For these pages Seyfarth printed the readings of only one of the three renaissance manuscripts that are direct copies of the Vaticanus, namely E (Vat. Lat. 2969), as well as those of Accursius’ editio princeps, which used both V and E. Since E was copied by an intelligent humanist who was prone to emendation, its readings cannot be trusted as representing the transmitted text. At a couple of points, the commentators follow Seyfarth in not giving relevant information. So at 31.8.5 a reader without access to Clark or Sabbah’s edition would infer that E’s acutius observantes was the transmitted text, whereas F and N both have, and V presumably had, the nonsense word adiutius (the commentators support Petschenig’s cautius; given the reconstructed reading of V, diutius should be considered). Likewise at 31.10.7, Mallobaudes alta pugnandi cupiditate raptatus, alta (where E and Accursius have autem) is not a conjecture of Valesius but the reading of F, N, and presumably V.

My own impression is that the text of Book 31 is in a slightly worse condition than that of 30 because of the absence of Gelenius, though it is noteworthy that the commentators propose changes from Seyfarth’s text, the lemma text, with almost exactly the same regularity in the two books (about 30 in book 30, and about 50 in book 31, which is just over one and a half times as long). Given Seyfarth’s conservatism, changes proposed to the text tend to be emendations, but there are a few vindications of the text transmitted by V: The restoration of quam at 30.8.6 with the sense potius quam is clearly right; so is the removal of Valesius’ –que at 31.16.7; etiámtum impraepedítus in the description of king Papa’s murder (30.1.20) is probably also correct, although the argument from prose rhythm against Gelenius’ reading praepeditum is false, since Ammianus also accents etiam túm in clausulae (e.g. 30.3.9). (I would also restore V’s readings at 30.1.10 iactique (for prose rhythm) and 31.16.9 aetate, doctrinis). Of the places where they advocate conjectures to improve on Seyfarth’s text, about a dozen are their own. The palm goes to 30.9.3, where sense is brilliantly rescued from ut solent occupationis spe uel impuniae quaedam sceleste committi with occultationis and impunitatis ([at the start of reigns] ‘when some criminality tends to be committed in the hope that it will be unnoticed or unpunished’: cf. 27.7.2). If other conjectures that they propose are incremental improvements, they are mostly either definitely or probably right: e.g. 30.10.1 <excohortibus, 31.6.2 uiaticum [cibos] et bidui dilationem (where V actually has the lemma form cibus, surely even likelier to be a gloss], 31.14.2 in palati<n>is (‘among courtiers’). If I have a reservation in the area of textual criticism it would be that although they take Ammianus’ immensely regular prose rhythm seriously, both as part of his style and as a tool for establishing his text, they could be bolder in accepting the consequences. For example at 31.3.1 they propose an entirely cogent solution to an irregular clausula (a transposition of the first two words of Tanaitas consuetudo nominauit), before stepping back from it in favour of accepting ‘incidental irregularities’. For all that, there is no doubt that they have contributed more to understanding the text and language of Ammianus than anybody since Charles Clark and his collaborators over a century ago.

This review has focused on matters of historical detail and textual criticism; much more could have been said about points of language, geography, intertextuality, and indeed the general mood of Ammianus (which they stand closer to Matthews’ optimism than Barnes’s pessimism) and his attitude to the history of his own times. Produced without fanfare or the support of large grants, and occupying the three original contributors long into retirement, this commentary is a model of learning and insight, and of selfless, collaborative scholarship, which will help Ammianus’ readers for centuries to come.

Wednesday, 14 March 2018

An intrusive gloss in Ammianus’ obituary of Valens

I have been reading the final volume of the magnificent commentary on Ammianus Marcellinus by Den Boeft, Den Hengst, Drijvers and Teitler (to be duly celebrated in Leiden next month). They do full justice to the magnificent climax of Ammianus’ work.

Valens’ obituary opens with his age and length of reign (31.14.1; I indicate clausulae with a hasta):

Perit autem hoc exitu Valens/ quinquagesimo anno contiguus,/ cum per annos quattuor imperasset et decem/ parvo minus. Cuius bona multis cognita dicemus et vitia./ 
And with such a death perished Valens, close to his fiftieth year, when he had held supreme command for fourteen years, a little less. We shall tell of his good qualities and his faults, which are widely known.

As the commentators note, the phrase parvo minus, ‘although easily understood, is without parallel.’ As they also note subsequently, it is incorrect. When he died on 9 August 378 Valens had been emperor for fourteen years and four months, having been appointed on 30 March 364. Thirdly, we can add that it is also fails to produce one of the regular clausulae that can be found with nearly 100% regularity in Ammianus’ text – even though had indulged in a hyperbaton to produce one immediately beforehand, cum per annos quattuor imperasset et decem, ‘when he had reigned for four years and ten’. (And if anybody suggests that pár-u-o minus might work, there is no good reason to think that v / consonantal u was ever vocalised in Ammianus’ usage, except after q- or in words like suadeo). 

All of this would suggest that the two words do not belong, and a reason for them to have made their way into the text is not hard to find: they are a gloss explaining the word contiguus, ‘close to, touching on’, in the line above, which is only very rarely used temporally. The gloss may have been either marginal or interlinear in an ancestor of the Vatican manuscript, our only authoritative source for this book, but the distance is almost exactly one line, which averaged 42 characters in the immediate exemplar of the Vaticanus.


Addendum: It would perhaps only be fair to add the case in favour of retaining the words parvo minus or something like them. The positive argument would be that on the death of Valens’ brother, before his obituary, a similar formula is found, saying correctly that he died aetatis quinquagesimo anno et quinto, imperii minus centum dies secundo et decimo (30.6. 6).