The lost first book of Ammianus Marcellinus

Ammianus Marcellinus’s history survives as eighteen books numbered from 14 to 31, describing the years 353 to 378. His last paragraph (31.16.9) declares that his work had begun with Nerva’s accession in 96; the extant books cross-refer about thirty-five times (without book numbers) to the lost books. This paper offers a new conjecture about the nature of these lost books, which have troubled many scholars who have given them thought.

Michael 1880 (cf. Rowell 1964 and 1966) argued that Ammianus wrote two works, the lost thirteen books only representing the beginning of the second work. Rees 2014 suggests that the lost thirteen books never existed, and that the cross-references and transmitted book numbers are a feint. Barnes 1998: 20-31 points out that the extant eighteen books are clearly gathered into three hexads or groups of six, and that thirteen lost books is a striking anomaly in this context. He proposes that at some point in the transmission book XVIIII was misread as book XIIII: the lost material fell not into thirteen books but eighteen, divided into three hexads. Kulikowski 2012 accepts Barnes’s aesthetic argument but argues that book 31 is an anomaly, an earlier monograph tacked uneasily onto thirty finished books.

This paper accepts and reinforces Barnes’s point on structure and aesthetics. In a work written for the codex, Ammianus shows traditional artistry in beginnings and endings, and hexadic composition. I shall demonstrate that 26-31, 20-25 and 14-19 are all coherent groupings. Barnes’s claim that thirteen books is a tight fit for Ammianus’s coverage of 250 years is less compelling. However, there is another way in which hexadic structure could have been combined with thirteen lost books: two lost hexads were prefaced by an additional book, which offered an index or summary of the other thirty books. The structure, then, was 1 + 6 + 6 + 6 + 6 + 6.

An index book is surprisingly well-attested in multi-book prose works from antiquity. The first book of Pliny’s Natural History lists the contents of the 36 following books; Aulus Gellius’s chapter list was originally printed with the preface as a preliminary book before the other 20 (Rocchi and Holford Strevens 2018); Valerius Maximus may have had a similar first book. Such readers’ aids are well-attested in Greek historiography; Polybius’s lost book 40 contained a chronological summary and list of contents (Walbank 1972: 16-17, 1957-79: 743-44). Twenty-three books of Cassius Dio’s history have a surviving index that contains a table of contents and relevant consular fasti – perhaps Dio’s own work (Mallan 2016). Many Christian works like Eusebius’s Life of Constantine have capitula (Winkelmann 1991). The idea that the lost books began with an index solves an aesthetic problem, and coheres with Ammianus’ place in both Latin and Greek traditions (Barnes 1998, Kelly 2008). This paper offers an unprovable conjecture, but one that solves real problems in a neat, economical, and plausible way. It will interest not only Ammianus aficionados, but anybody interested in classical historiography and book history.