Bryn Mawr Classical Review 2007.12.30
A.S. Hollis, Fragments of Roman Poetry c. 60 BC-AD 20. Oxford: Oxford University Press, 2007. Pp. xviii, 440. ISBN 978-0-19-814698-8. £80.00.
Reviewed by David Butterfield, Christ's College, Cambridge (email@example.com)
Word count: 9408 words
The task of editing and discussing fragments of ancient literature is a formidable one and any hasty attempt to do so typically finds an equally short-lived readership within Classical scholarship. It is therefore with both relief and respect that one learns from the preface to Adrian Hollis' (henceforth H.) edition that the volume has been some sixteen years in the making, albeit with the editor's attention having often been focused upon other projects in that period. The resultant product of H.'s labour is an engaging and informative work, which contains considerable original material and will accordingly hold an important place within literary scholarship upon Latin fragments for some time to come.
The somewhat forbidding nature of fragments, severed by definition from their original context and often bereft of the details of their authorship, has kept most scholars from engaging actively with the diffuse corpus provided by Roman literature. After the successive labours of Scaliger minor, Pithoeus, Stephanus major, Burman minor and Wernsdorf (inter alios), the modern era of editing Latin fragments arrived in the latter half of the nineteenth century, when Emil Baehrens produced, as a supplement to his Poetae Latini Minores, the ambitiously titled Fragmenta Poetarum Latinorum (1886), a volume extensively reworked by Willy Morel (1927), then refreshed and furnished with a bibliography by Karl Büchner (1982), itself updated by Jürgen Blänsdorf (1995). It should occasion some surprise, however, that it was only in 1993 that a work attempting to include and discuss the full breadth of Roman fragments (omitting those texts well served by a Skutsch or Jocelyn, or respectably by a Marx and Krenkel) saw print. The achievement of Edward Courtney's The fragmentary Latin poets continues to be one of major significance for Latin scholars but, compressed into a single volume, necessarily cannot serve as the last word on the subject. In short, Roman fragments yet remain comparatively starved of critical attention, with the gigantische Lebenswerk on the full corpus still waiting to be written.
H. has, however, made a welcome and notable step forward in his Fragments of Roman Poetry (henceforth FRP). The extended title of the work announces that the book is designed to cover a period of roughly eighty years (c. 60 B.C.-A.D. 20) and the preface (v-vii) expands upon the significance of this period, which Velleius Paterculus (I.17.2, II.36.2-3) seemed to regard as the artistic zenith of Latin literature that had preceded him. Beyond the preface and a series of clear but short introductory discussions (1-8), H. treats, in rough chronological order, 28 poets, followed by a number of fragments preserved anonymously. Each fragment in the work (which H. terms, on the model of the Supplementum Hellenisticum of P. Parsons and H. Lloyd-Jones, an 'item') is followed, in this order, by English translation, the surrounding Latin testimonium(/-a), (occasionally) a brief critical apparatus and, where appropriate, succinct editorial comment. There are, in total, 262 items, numbered continuously throughout. These items are typically arranged so that purely allusive testimonia open each author's corpus, which are then followed by genuine poetic fragments duly sorted into the works (if plural) to which they most plausibly belong. After each collection of items comes an introduction to the poet (except for the most minor authors), followed by detailed linguistic, literary and, where deemed appropriate, textual discussion of each fragment.
After the chronologically arranged poets (11-388) comes a selection of 27 adespota (389-419), likewise provided with translation, occasional apparatus and commentary. Closing the volume is a most useful, though often brief, discussion of poets whose name alone survives (420-30). Finally, the work ends with a short bibliography (431-5)--oddly restricted to secondary sources--and a single index (437-40), which covers English terms, some proper names and the occasional Latin technical term (e.g. litterator, urbanitas, domina). A useful concordance of H.'s numbering against those of Courtney and Blänsdorf can be found towards the beginning of the volume (xiii-xviii).
So much for the layout, now let us turn to the purpose of the work. The preface records that H. wishes 'to attempt some integration of the poetic scene in Rome during the period' (vi). This is a major undertaking and one that has resisted most attempts previously. Yet by allowing himself a comparatively narrow temporal window for a full-scale book, H. has been able to devote much attention to this aim, and its results, as I will say in due course, are positive. As may surprise some, we learn from the preface that the projected volume was conceived before, and so independently of, Courtney's more ambitious work (which treats some seven centuries of fragmentary Latin). Nonetheless, '[i]t was decided that [H.'s and Courtney's] projects were sufficiently distinct that both could continue' (vi). This decision, made in 1989, was prudent: H.'s book has considerable independent value, and its delayed appearance has allowed plenty of time both for Courtney's ideas to diffuse into wider scholarship and for H. to profit in particular from two recent works alongside Courtney's: Blänsdorf's revision of Büchner's revision of Morel's Fragmenta Poetarum Latinorum (Teubner, Stuttgart/Leipzig, 1995) and R.A. Kaster's major edition of Suetonius' de grammaticis et rhetoribus (Oxford, 1995). In spite of the great, if not monstrous, volume of secondary literature in the field, which can so easily reduce such scholarly undertakings to a mere tapestry of references bearing little novelty or utility in itself, it is pleasing to note that H. has apparently sought to write FRP so that it can stand alone, at least for the scholar approaching the field afresh. The work can therefore be read with profit when outside a Classical library.
In the course of this review, which will be longer than most, I shall structure my comments around three distinct questions concerning FRP. Firstly, comprehensiveness: how extensively is the fragmentary literature of the period in question covered? Secondly, literary analysis: to what extent does the work contribute, whether for the first time or by developing previous ideas, to our interpretation of fragmentary Latin literature? Thirdly and finally, linguistic analysis: to what degree does the work enrich our knowledge of late Republic to Tiberian Latin, as well stylistic and metrical learning? It will become clear that my responses to these questions are largely positive, although more muted as regards the third. A list of particular points of discussion exemplifying some of what I say below will be densely appended at the end of the review.
(I) Comprehensiveness. The preface informs us that H. has not included quite everything from his approximate eighty-year window of c. 60 B.C. to c. A.D. 20: he has chosen to exclude the poetry of the two Ciceros (on which, incidentally, much work remains to be done), the uersus populares and 'the fragments of poets who have for the most part survived' (v), which removes from consideration those of Lucretius, Catullus, Ovid and Manilius, along with the two epigrams assigned to Virgil, the three words attributed by Charisius to Tibullus and the iambic fragment of Julius Caesar. Unlike Baehrens (along with his three revisers) and Courtney, however, H. has kept the door open for tragic fragments, thus admitting Gracchus and the remnants of Varius Rufus' tragedies. The collection, however, is not comprehensive, even ignoring those omissions above. Although H. nowhere declares it to be, his silence about what else has been excluded does not aid the curious reader. Indeed, the work does not include the long fragments of the anonymous de bello Actiaco (although three authors to whom it has been attributed--Rabirius, Varius Rufus and Cornelius Severus--figure in the selection), the elegiac distich of Gaius Asinius Gallus and the various poetic insults of Pompeius Lenaeus that can be extracted from Suet. gramm. 15. Furthermore, no reason is given for the inclusion of Egnatius but the exclusion of Gannius. Whether it was right to exclude the hexameter attributed to a Verrius by Prisican (I.383.12) and thereby reject Courtney's suggestion (FLP, 291) that M. Verrius Flaccus was the author, it seems impossible to say. The adespota are avowedly selecta but it is nonetheless surprising that the interesting verses of the obtrectatores Vergilii and the curious quatrain de Crassicio are excluded, the latter being relegated to a testimonium of Calvus (7(b)). Poems such as these could happily have been exchanged for less interesting items, such as 255 (which inspires a discussion of only three sentences), or those that are purely testimonia (244, 249). It is also unfortunate that the title of the book, chosen in order to provide an acronym distinct from those of the stirps Baehrensiana and Courtney, does not emphasise that its contents are solely Latin and that it therefore excludes the Greek epigrams of Tullius Laurea (anth. Pal. VII.17, 294 and XI.12) and Scaevola (ibid. IX.217). Epigraphic poetry is not included. Notwithstanding these omissions, however, H.'s selection remains varied, robust and (most would agree) adequate.
H. has evidently not employed the narrow definition of a poetic 'fragment', i.e. the verbatim citation in another work of an incomplete piece of poetry drawn from a larger work that does not survive, whether at this particular place or at all. Instead he seems to regard a fragment more loosely as a small amount of poetry, not necessarily incomplete, of an author from whom little or no other poetry survives. The fact that this wider definition increases the amount of material overlooked by the collection does not much worry H., who declares in the preface (v) that the fact 'that some of the items in this book are not 'fragments' but complete short poems...troubles me not at all'. This catholic approach does however allow the inclusion of the Qasr Ibrim 'fragment' of Gallus, which Büchner-Blänsdorf and Courtney were also naturally keen to print, and other poems that would more naturally appear in a collection akin to Riese's Anthologia Latina.
Of H.'s selection the most paper is given to Varro Atacinus (50pp.), Gallus (44pp.), Cinna and Calvus (both 38pp.), the least to Volumnius (1p.), Arbonius Silo, Dorcatius and Sextilius Ena (all 2pp.). The scale of the treatment typically reflects the volume of extant material but certain distortions are evident; for instance, Gallus receives nearly a page of discussion per surviving word of poetry, whereas the 22 and a half hexameters of Albinovanus Pedo (therefore the second longest fragment in the collection, after the 25 hexameters of Cornelius Severus) receive the equivalent of a page of commentary to every fifteen words. The shortest fragments are of only one word (8 (Cinna) and 46 (Memmius)) and the smallest poetic corpus, that of Hortensius Hortalus (99), amounts to only a single word (and it is to be doubted whether we even know in what case it was employed). As mentioned above, those poets for whom we have much testimony but no extant word of poetry are relegated from the main body of the work to an appendix.
(II) Literary analysis: what is the extent of FRP's contribution towards our understanding of the literature of the period? It is in the field of literary analysis and tentative reconstruction of fragments that H. is most at home, and there is in FRP a considerable body of material that is both original and thought-provoking. In terms of the number of suggestions aired here for the first time H.'s book is remarkable when set against most modern scholarly works. H.'s extensive commentary throughout the work, which he rightly terms the 'heart' of the book (v), is, for the most part, genuinely rich and invigorating, with much attention being given to literary sources, contemporary parallels and subsequent imitations. Although such comparisons have only a limited use in themselves, from them H. repeatedly draws instructive conclusions, opening numerous paths for future research. With regard to H.'s conjectures (and that they must remain) about the original literary contexts of the fragments, he confesses that his approach is bold. Yet the reader of FRP should be thankful that H. has both chosen not to sit on the fence and opted for the bolder side, since his suggestions, whether convincing or not, confront the reader with the very questions that should be being asked when handling material of this difficult nature.
Another strength of the commentary is H.'s close familiarity with Hellenistic poetic material: germane parallels are drawn not only for the more obvious imitations (such as Varro Atacinus' Argonautae vs Apollonius Rhodius) but also to highlight many subtle Hellenising nuances within various poets of the collection. With most of these authors Greek influence is of course of signal importance, although in some instances it is likely that H. oversteps the mark when positing Greek influence (in spite of his own warning on p.231 that Latin commentators can make this mistake). Such a wealth of Hellenistic material does eclipse, however, an area of influence which H. rarely treats, namely that of pre-Lucretian Latin poetry. Occasional mention is made of Ennius and Accius but H.'s general silence about other early Roman influences could allow misleading inferences to be drawn by the reader.
All who use FRP will be particularly thankful for the breadth of H.'s mythological knowledge, which is often of interest beyond the immediate interpretation of the given passage. Especially impressive is the introduction to Calvus' Io (20-25) where a wealth of material is brought before the reader, and a potential reconstruction of Calvus' work is carried out by means of a masterly comparison with Ovid's subsequent treatment of the tale in Met. I. Other discussions of particular merit and depth are those of the route posited by Cinna's propempticon (4-5), that introducing Varro's chorographia (109-19) and that upon the anonymous 238.
The most strikingly impressive treatment of a poet is that of Gallus, which almost deserves to be termed a full-scale edition of his small poetic, but comparatively large testimonial, corpus. It is with this section that H. most achieves, I believe, his aim of relating the poet to his better-known contemporaries, and his skilful handling of the testimonia cannot fail to instruct. It is a shame, as noted at the close of the lengthy treatment of the poet, that the important monograph of M. Capasso (Il ritorno di Cornelio Gallo (Naples, 2003)) could not be included in a work published four years later, since (as stated on p.252) H. only learnt of the book's existence in 2006, whereas the preface is dated December 2005; one likewise finds no treatment of the ideas adduced by G.E. Manzoni (Foroiuliensis poeta: vita e poesia di Cornelio Gallo (Milan, 1995)).
As a final point on H.'s literary commentary, I wish to note how refreshing it was to be spared the layers of literary-critical jargon that veil many treatments of Augustan poetry, typically kindling in the scholar nothing but impatience. Rather, H. tackles the passages head-on and accordingly makes swift progress in his discussions. What a pleasant relief it was to see the appearance of the useful short-hand 'Silver Latin' (p.291) without apology or prolix qualification!
(III) Linguistic analysis: what is the value of H.'s work as regards linguistic, stylistic, prosodic and textual matters? Although it is certain that the volume contains many interesting comments on these topics, in this field where so much more work remains to be done H. could have often developed his observations further. Since material to be investigated is so limited, clear-headed discussion of linguistic and metrical facts, amounting to a small-scale stylistic analysis, can often provide a surer footing for criticism than conjectures about fragments' wider literary contexts. Courtney, for one, made a number of important advances in these areas, thereby paving the way for more detailed discussion or, where appropriate, dissent. H. often pauses for informative discussion of rare lexical items and semantic nuances but less often treats at length grammatical or syntactical oddities.
In particular, metrical curiosities often deserve greater discussion than they receive, or at any rate explicit recording of their rarity or dubious nature. For example, the sort of lengthening defended at 31.1 (at the thesis of the fourth foot) is at once remarkable and unmotivated: it is more likely the result of transposition of adjacent words than a genuine metrical anomaly (one nonetheless misses references to Vollmer and Norden). The lengthening at the close of the former hemiepes of a pentameter at 174.6 is also dubious (a genuinely long final syllable would be a seemingly inappropriate archaism): it is insufficient to defend it by comparison with Greek practice, or with Virgil's hexameters, or with the prevocalic but choriambic scansion of Oceanus at 256.3, that being a Greek word, placed before the hexametrical caesura and of an uncertain date and stylistic register. The Augustan elegists only offer three instances of such lengthening at the pentameter (compounds of ire are not ad rem), all of which can be easily removed by emendation; accordingly Thilo's deposuit<que> ad loc. is a prudent suggestion; if the paradosis is to be defended, comparison should be made with two poems in the Anthologia Latina commonly attributed to Seneca and here referenced by Courtney. Unelided atque goes uncommented in Varius Rufus (147.2) and receives only brief treatment in Valgius Rufus (166.2); it would have been useful to expand further on both the rough chronological development of its treatment, and how its use differs according to genre (a matter I treat in a forthcoming Mnemosyne article). The heavily diaeretic nature of 33 (if hexametric) goes uncommented upon, as does the noteworthy rhythm of 75. E.J. Kenney's objection to the elision in thesi of Mueller's iam at 6.4 is not discussed (comparison with V. Aen. VI.629, XII.582 and Prop. II.16.33 would have proven its possibility but considerable rarity).
We are told in the introduction to 106-7 that 'the Sequani themselves could not be so-called in dactylic verse' without mention of the fact that a composer as refined as Virgil stomachs the elision of the cretic proper noun Pollio four times in his bucolica (shortening it to a dactyl, as Horace does, no doubt being anathema). More could have been said at 84.7 and elsewhere about the compulsory or free use of the spondaic base in the Phalaecian hendecasyllable, drawing upon O. Skutsch's important conclusions about the chronological implications for Catullus' poems. It is observed on p.213 that 'the short final 'o' in 'Cato' [in the elegiac distich preserved at schol. Pers. II.36] seems unlikely in our poet [= Varro At.]', without discussion of the widespread phenomenon of iambic correption, by no means uncommon in proper names. At 7(b) we are only asked to 'note' the prosody of a short final 'e' before the following 'Smyrna' without any discussion of Dawes' Canon (Misc. Crit. p.3ff.), subsequent scholarly debate and indeed contemporary authors' habits on the matter. I would disagree with two points made on the metre of 148: two disyllabic words closing a hexameter, preceded by a choriambic word group, is not a 'slight infelicity' but rather a rough, rare and remarkable rhythm; on the contrary, Virgil's preceding the penthimemeral caesura with et at Geo. II.506 is not likewise a 'slight infelicity' but a common stylistic feature of his versification, in which conjunctive particles and prepositions begin the third foot, bolstered by a strong caesura in the second. As regards Cornelius Severus' comparatively long hexameter extract (219), H. states that the poet has a 'dislike of elision' without observing the rarity of elided enim (only 2% are elided in Ovid, 3% in Lucretius and 6% in Virgil). In the brief discussion of the full hiatus exhibited in the anonymous 248 (possibly, as H. states, a grammarian's fabrication) no mention is made of the obvious but crucial fact that both words are of Greek origin. It is unfortunate that H. nowhere discusses Hermann's Bridge, for, although it did not stand as a firm metrical bridge for Latin poets, its being breached often has important stylistic implications (cf. the fact that Tibullus only once breaks it (and that in a special instance) in his first book of elegies). As a final comment, the rarity of the rhythm opening 43.1 (two dactylic words) deserved further comment, for although his citation of Lucr. VI.1272 (not 1270, a line numbering drawn from Brieger?) is a genuine parallel, such a rhythm occurs in that poet fewer than once every thousand hexameters.
The standard of the text and its apparatus is perhaps the only real shortcoming of the work. Since, however, I have considerably more to say on this matter than would interest all readers of this review, I defer much of the discussion of these points until later. I shall limit myself here to discussing the more important textual points. It is admirable that a number of conjectures appear in the work for the first time, many being from the hand of the work's dedicatee, and H.'s long-term colleague, R.G.M. Nisbet and some 23 from H. (of which twelve are recorded only in the commentary and nine are supplements rather than corrections). All concern the texts edited in the work, save for H.'s suggestion (p.77) of altering deo to deum at Ou. Her. IX.43. It is evident that H. enjoys verse composition--a skill whose value is much underrated by modern Latinists--and many of his versions are pleasing to read and perhaps rather close to the truth. Nonetheless, such a talent could perhaps have been usefully applied towards finding plausible Latin for lacunas within the texts, not least for the two lines of Valgius Rufus (166.5-6), rather than to adding context to fragments whose sense is otherwise clear. Two conjectures are palmary, namely Nisbet's abisse (78.1) and Cu<ma>na (89); H.'s Phoebo at 111.4 and fratri (accepting Unger's concumbere) at 178.4, along with Nisbet's hauriat (after La Bigne's uis) at 112.4 and suggestion of nominative 'flauus...galerus' at 213, also deserve serious attention. H.'s conjecture pertusam for perfusam in a two word fragment of Valgius Rufus (170) is certainly attractive, although Pliny's mention of a peruncta peluis (h. n. XXXI.46) perhaps provides a potential parallel for the paradosis. On occasion H. reassigns the location of a fragment within an author's corpus, the most convincing example being his transferring a fragment (131) of Varro Atacinus' Argonautae from the second to the fourth book (a suggestion H. announced in 1995). A number, however, of H.'s more attractive textual conjectures have been preceded: te at 10.2, conjectured by H. as a graduate student, had already struck J.C. Collins (Illustrations of Tennyson (London, 1891), 27) and, independently, A. d'Elia (Latinarum litterarum historia (Naples, 1952), 102), although neither explicitly records his slight change; Thynia at 185.4 appears in the notes of Valpy's Delphin edition of Catullus ad XXV.7 and perhaps earlier; in a similar vein, Lachmann appears to have anticipated H.'s suggestion that a fragment of Calvus (37) is iambic tetrameter catalectic spanning two lines (witness the colometry in item 20 on p.87 of his third (posthumous) edition (Berlin, 1874)); regardless of this fact, D.A. Russell's suggestion (ap. Courtney) of iambic dimeters deserved explicit mention and discussion; finally, it is unclear from H.'s wording whether, in suggesting the supplement narrantur at 217.2 (comm.), he thought it his own: it is Morel's.
A number of other conjectures in the apparatus can be pushed back earlier than H. and other modern editors realise: 31.1, the transposition docuit sanctas occurred to T. Clayton and C.S. Jerram in their edition of Aeneid I-VI (London, 1865) ad IV.58 before Castorina; tellus for terra at 57.2 can be antedated by some three centuries, since it appears in Ulisse Aldrovandri's citation of the fragment in his posthumous Serpentum et draconum historia (Bologna, 1640); qua at 112.4 is Scaliger's; Thynica at 185.4 is not Avallone's own creation but occurred not only to H.A. Lion (Maecenatiana (Goettingen, 1824)) but long ago to Pithoeus; at 186.3, however, ninnio should presumably not be Pithoeus' correction but Scaliger's; afuerant at 219.24, attributed to Kiessling, also goes back at least to Scaliger; Courtney's doubling of d in redduxi at 236.4 is an admirable change but can be traced back to Orelli's Eclogae (Zurich, 1833) and probably beyond.
Before turning to further discussing of the text of FRP I should say something about utility. Notwithstanding the importance of the work's contents, the book is not particularly easy to use for continual reading (as opposed to reference), since it demands a great deal of flicking between pages. As mentioned earlier, the 'text' section is separated from the general introduction and commentary that follow, with the result that the main introduction to each author is buried in the midst of that poet's chapter, inserted, with only a minute gap, after the final text item. The placing of the introduction to Calvus, positioned at the very bottom of p.57, sufficiently illustrates this point. Thus layout, which may not be H.'s own, is not particularly lucid. More natural would be to place the general introduction before the appearance of the fragments to provide due historical and literary context. This would allow a clear display of the possible 'floruit' of each poet, information currently difficult to find in haste.
It can occasionally be difficult to distinguish (at a glance) Latin from English words, since the former are typically not italicised but in roman within quotation marks. Further, translations of fragments appear in the same size and font as the Latin text, directly below which they appear, which also has its incoveniences. Some may query the decision to print the testimonium separately after the fragment and translation, typically recording only the first and last word of the fragment within the testimonium. On more than one occasion (e.g. 43, 192) these words differ from the printed text, almost implying that the textual error in the fragment is not due to scribal error but rather the mistake of the citing author (an implication intensified when, as occasionally, the name of the author, rather than the manuscript sigla, is recorded in the apparatus). At times only one word is omitted in the testimonium, marked by a hyphen (e.g. 91, 115, 201 et, 205), and with one- or two-word fragments they are simply repeated (e.g. 92, 192). It would have been better to follow the more standard practice of printing the fragment in the midst of its testimonium (but well spaced and in larger print), and then following this with an apparatus and lastly, if appropriate, a translation of the whole. In the instances where there is more than one testimonium the later or less informative one(s) could be printed below without repeating the fragment; as it stands, multiple testimonia are cramped together, not beginning on separate lines.
The ordering of fragments within chapters is not always clear although it may well be that a given methodology could not always be employed. With an instance such as Furius Bibaculus, however, even if Suetonius did not know the author of 86, there seems no good reason to transpose it to after 84-5.
It is within the scope of utility to say a few words about the translation. The benefit that H.'s translation brings is limited and of use primarily for demonstrating how H. reads each fragment. Perhaps he would defend its presence because it allows readers with little or no Latin to engage with the book (the dust jacket advising that the book is 'accessible to readers without Latin'). And yet, if that is so, one would expect a translation of the surrounding testimonia, but this only occurs most rarely, with items 8, 46, 245 and 246 (and for items 194-6, 202 and 238 a separate translation is provided for both the fragment and the testimonium, although I know not why; at 99 only the shorter of the two testimonia is rendered). Furthermore, large tracts of Latin, as well as Greek (a language readers cannot be expected necessarily to have), appear in the commentary without translation: of Velleius Paterculus (147 words), of Propertius (117 words), of Tacitus (107 words) among others. Sometimes such passages in the commentary do have a translation, as with Apollonius Rhodius in the commentary to 126, Euphorion in that to 139, and Nicander in that to the anonymous 237, although it is not clear why these merited special treatment (the often difficult Oppian and Nonnius, for instance, are left untranslated in the introduction to 7-10). The variation in format of H.'s translations from Jerome's chronicon evident at 47(d), 71(a), 93, 138(a) and 146(a) will be noticed by few and inconvenience no one.
As stated in the treatment of my third question, I have found the quality of the critical apparatuses often to be dissatisfactory. The short preface to the work does not state the importance of the contribution made in the textual field by the Baehrens-Morel-Büchner-Blänsdorf Teubner dynasty and, to a major degree, Courtney's FLP, with the result that H.'s evident debt to their apparatuses is nowhere recorded, beyond his saying that he has 'profited' from their work (vi). This is unfortunate. Although it is now customary, albeit for no obvious reason, for literary works to draw freely upon another's edited text without acknowledgement or even declaration, in a close and often technical study of fragmentary Latin poetry, the importance of the text and its history is paramount. Since H. includes no discussion or evaluation of previous editions of Latin fragments, the charitable reader would presume that his silence indicates that much of the textual work is H.'s own. This is by no means the case, beyond the introduction of his own conjectures and those of Nisbet. H. has provided an apparatus for 88 of the 262 items in the collection, presumably limiting himself only to those passages where he deems it useful to record manuscript dissent. This is understandable for a work attempting to cover so much ground within two boards, but the reader is left uneasy at apparent inconsistencies, finding an orthographic trifle recorded such as (187.3) 'benest B1 (i.e.a.c.) : bene est (Bc (i.e.p.c.)φ ψ, i.e. all the rest of the ms tradition) but silence about some words of poetry being conjectural supplements (e.g. lucens in 17 or regis in 250) or highly tentative readings (e.g. -que superbi at 123.4).
In my own copy I have found it useful to add an apparatus to 22 items which H. leaves bare, and very often to make further additions to his critical notes elsewhere. To cite only a few examples of where an apparatus would have been useful (I leave other instances for the appendix of this review): at 84.3 it should be recorded that selibra also appears in M, and selibia in B (both with a rogue et preceding), which suggests that the correct selibra could be pushed back not only to β and Γ (the α family being contaminated by the X branch?) but also to the hyparchetype Y; within the testimony to 123 it is misleading that over twenty words provided as conjectural supplements to Schol. Veron. ad Aen. II.82 are printed as though definitive; at 195.2 the reader should be told that for Hector is found ore sectora or ore sed ora in our witnesses (ABV), which Bursian tried to retain more faithfully by suggesting Orcus / Hectora <habet>; it must be recorded at 233 that serum is a supplement for in of ML (om. V), even if the lemma and Haupt's critical acumen make it an assured suggestion.
It is of course an important rule that the textual apparatus be visible at the same time as the relevant portion of text, yet on occasion this is not the case in FRP (e.g. 6, 118, 219). Although this only means a turn of the page, the tenuous nature of the thread on which our knowledge of fragmentary Latin poetry hangs is left yet more obscure to the uninitiated reader. It is also unfortunate, although partly understandable, that H. is not able (except uniquely with item 33) to record what manuscripts his sigla denote--sigla which are drawn, like the various means employed of distinguishing hands within a manuscript, from others' editions--and therefore which witnesses are typically more faithful than others. A summary similar to that at the opening of Blänsdorf's Teubner could have cleared this up to some degree.
Although most scholars would naturally head for the welcoming shelter of the aedes Teubnerianae where the apparatus is concerned, the fact that H. has gone to the trouble of adding an apparatus to some third of his items implies that their construction and content are sufficiently rigorous. Unfortunately, on a number of occasions, H.'s apparatus is unclear or misleading.
H. wavers between employing a positive and a negative apparatus: he typically uses the former (sometimes writing codd., sometimes a closing square bracket, which can cause confusion when used alongside the same symbol for supplements to a papyrus fragment (as at 145), but instances of the latter are not hard to come by (e.g. 22, 127, 166.3, 217). Clarity is better than dogmatism on this point but H.'s choice between the two does not seem based upon a case-by-case assessment. On one occasion (187) both negative and positive are used in the two-line fragment. On another (218.1), the two distinct mascots of a positive apparatus--the square bracket and colon--are combined (likewise on p.263), with the result that the reader is not told that grauantur, the reading H. prints, is in fact found only in printed editions of Diomedes. The vagaries of H.'s approach to constructing an apparatus can be clearly witnessed with those to fragments 199-201: all three involve a simple correction of the paradosis, but although those to 199 and 200 are in the form 'X codd., corr. A', 201 instead appears in the form 'Y A : X codd.'. Likewise, there seems no obvious reason behind his writing, in the apparatus to 219.24 'afuerunt Heinsius: -at codd.: afuerant Kiessling'. Such uariatio, it is true, only occasionally confuses the reader but is nonetheless curious. More alarming and certainly misleading is H.'s carefree use of the colon, which often oversteps its strict limits by relating conjectures and readings that in reality treat different amounts of the text. Such misuse can be witnessed at, e.g., 102, 120.1, 124, 128, 222.2, 235.2.
The unnecessary practice of recording the name of a scholar who has been preceded in his conjecture alongside that of the preceding scholar is occasionally in evidence (e.g. with Courtney at 29.1, Alfonsi at 43a.2, H. himself at 166.8, Heyne in a discussion on p.237, alii at 174.5, Scaliger at 181.1 and Vinetus at 181.2 being given a place). A genuinely bizarre feature is H.'s occasional but deliberate inclusion in his apparatus of Courtney's typographical slips in FLP, namely at 57.1 (on which see below), 76 and 194.5: this is neither necessary nor a reflection of the purpose of an apparatus, viz to record as clearly as possible manuscript evidence and plausible textual conjectures. It would surely have been enough for H. to note these errors, as he does (vii), in his preface. The presence in the apparatus of 'edd.' or 'editores' alongside a conjecture is likewise of little use.
Substantial dependence upon the Teubner text is particularly clear in a number of places, such as in the apparatus to 262, where a sequence of 33 words is transcribed verbatim (or cf. the retention in the critical note of the uncommon word order 'rec. man.' at 113). Similarly, I imagine that H. implies that antiqui is evidenced in the mss of Priscian (GLK II.189) before the first fragment of Ticidas (102) because the curt expression of Büchner-Blänsdorf, unlike the clarity of Keil and Morel, implies that the nonsense atticidas veiled only Ticida (a correction, incidentally, not of Mueller but of Hertz). Why state at 30.1 'quatiens om. RHLDGK' (mss that should be italicised) if the reader has no direct means of knowing which manuscripts of Priscian do contain the word?The phrase has been drawn directly from the Teubner, in which work the statement can be readily interpreted to mean that only codices AB retain the word. Nonetheless, some evidence can be found of H.'s avoiding mistakes of the Teubner text: at 124.1 the assertion of Büchner-Blänsdorf that the manuscript of Charisius offers Tiphyn as its correction not Typhin is silently corrected; at 192 the strange error of creating a manuscript 'R' from the first initial of Unger is not inherited from Blänsdorf; at 214 Baehrens' conjecture is rightly given as gemmata, not geminata, est.
On more than one occasion H. shows a tacit preference for a reading of apparently less weight, and it is unclear what reasoning lies behind this: 59, scrabrone for crabrone, thereby demanding lengthening of the preceding open short syllable rather than following the more natural crabrone and Courtney's easy transposition of est: it is over-simplistic to say that scrabrone 'is well attested in the manuscripts quoting this fragment' (p.111), when the Servian tradition points confidently to crabrone, that of Isidore (a keen harvester of Servius) to scrabrone; 92, no reason is giving for preferring et of the paradosis over Haupt's ut in the testimonium; 140 (d) (= Ou. rem. am. 765), why potuit over poterit?; 166.1, there seems no clear reason for H.'s printing solebas for canebas of the Veronese Palimpsest (as also at 1(b)): we are not told whence the reading came nor that it is silently rejected by most subsequent scholars (I have not, however, been able to consult A. Herrmann's Die veroneser Vergilscholien (progr. Donaueschingen, 1868-9/1869-70)).
In a work of this nature the obelus is too rarely wielded, appearing only thrice in the poetic fragments (43a.2, 112.4 and 171); with such brief texts, daggers display admirable caution and are to be preferred to a makeshift decision. A similar desire to settle for a firm text is evident at 123, where it would have been better to challenge Keil's supplements (which are very unlikely to be Varro Atacinus' original words) than to translate and comment on them at length, as if the poet's ipsissima uerba. Some may object on occasion to H.'s dismissing N. Heinsius' admirable urgenti at 22 on the grounds of its being 'unnecessary' and rejecting iurisque of the same at 219.14 because the paradosis 'seems unobjectionable': the probability of any textual conjecture depends upon how more natural it would have been for the author to employ that word or phrase weighed against the chance of its coming about by scribal error in the given tradition. It should be made clear why asterisks appear in the testimonium to 237.
Since I shall record a number of my objections in the specific notes appended to this review, I limit myself at this point to recording what I deem the most notable errors in H.'s apparatus: 112.4, if Scaliger's rota, though not greatly appealing, is cited, his emendation of the following word (ferueat) should be also; 118.3, the manuscripts of Isidore offer condere alongside concedere, the correct reading contendere, which H. wrongly ascribes to the mss of Isidore, arising solely from ann. in Luc. ad III.237 (I add in passing that it is strange to record indiscriminately the variants magnum and magnam for magna but not V's magno); 126, here we read of Probus' 'cod. E', which is in reality the editio princeps (Venice, 1507), edited by Egnatius; 174, the apparatus is confused, with the paradosis being recorded under '5-6' but subsequent emendations being recorded separately under '5' and '6', yet with the repetition of the paradosis at the end of the entry to 6 (twice does H. here have to say 'v. supra' to refer to another section of the apparatus); 188.2, I fear that H. (or his typesetter) has misread the stigma (= recentiores) of other editors as an 'S' and therefore as a single manuscript (similarly the use of sigma four times for stigma in the apparatus to 157 could mislead); 227.2, it is confusing for notis to be recorded without attribution when it is not found in ABV (also at ibid.19 D, if cited elsewhere in the apparatus, should be recorded as a further witness to libris); 256.1, it is a mistake drawn from the Teubner text that the manuscripts of Augustine present te (which H. prints) for tu at de doct. Chr. III.7.11 (therefore rather than a 'clash of saints' (comm. ad loc.) between Augustine and Isidore, it is really a consensus bonorum for the rejected reading); it is also an odd editorial decision in this testimonium to replace with a lacuna Augustine's important aside before citing these verses, si bene recolo.
Typically no apparatus is provided for items which are purely testimonia, a decision which is on the one hand understandable for reasons of space but on the other potentially misleading. Very occasionally a quasi-apparatus is provided in square brackets within the testimonia (as with 7(a), 71(d), 126, 141, 195) but often without the recording of manuscripts, or without recording what the paradosis is, simply the corrector. A reader would surely be interested to know, for example, that at 36 the manuscripts of Porphyry have Licinius Gaius for Licinius Caluus; or that at 125 Anchiales is Fabricius' correction for the disarray of the mss.
It is perhaps the drawn-out composition of the work that explains many unresolved inconsistencies of layout. To take an example, H.s critical note about the attribution of a verse typically occurs below the apparatus and in larger type (cf. e.g. 107, 111, 124) but not always: with 62 it appears in small italics as if the apparatus; at 41 and 66 it is part of the apparatus; at 69, 70 and 88 these notes appear in italics as if testimonia; contrariwise, the testimoniaat 17 and 126-7 should be in italic (and the apparatus to 128, like the latter half of that at 97, should be in smaller type); 193, which is purely a testimonia, exhibits a curious fusion of an apparatus and a critical note of attribution, and is not in keeping with the style elsewhere (nor is that to 37). Some other variations of layout I cannot explain: in 109 one finds a translation that only renders part of the testimonium and is placed (as is 119) after H.'s critical note about authorship and original context; items 48(c), 110 and 172(b) are simply observations (in Latin) by H. that the poet in question is named as a source for some books of Pliny's historia naturalis, and these notes are provided with translations, once in square brackets, once in rounded, once without. Such information could happily be provided in the introduction to the poet. Occasionally the Latin of the critical apparatus is a little odd, such as the phraseology of those to 242 and 243, or the change of tense in 244.
Enough then of the apparatus. Finally, a few words may be said on orthography. H.'s approach is typically consistent, although no clear decision is reached regarding prodelision (only printed at 187.3) or sigmatic ecthlipsis (marked in Barwick's conjecture at 68.1 but not in the discussion of 30(a) or in 43a.1). Similarly, if quom is to be granted to Cinna (2), then it naturally should be to all poets in the collection before, say, Valgius Rufus (note that H. seems happy (p.363) to accept Housman's quoi at Ou. ex P. II.3.76); it is no good to print cum at Cinna 22, only to state in the discussion that 'Calvus himself would probably have written 'quom'', for whose text is being edited? It is admirable that H. prints the acc. pl. of a third declension adjectival form in -eis at Cinna 6.1, however much that jars with his constant use of 'v' for consonantal 'u', but we only find him employing such spellings here and in the Gallus papyrus (145). H. tends to shun consonantal assimilation, but not with complete consistency.
Prior to concluding this review, I shall offer two lists, the first recording a small number of textual suggestions that struck me in working through the volume, the second containing minor observations, textual or otherwise, that could not be conveniently touched upon in the discussion above. In each I supply only the item number of the point at issue, qualifying it where appropriate either by '(text)', '(test[imonium])', '(app[aratus])' or '(comm[entary])'.
I: 14, saecula permaneat...Dictynna is an unparalleled use of the accusative and hardly mitigated by reading Mommsen's homophonic saecula per maneat; one could retain Mommsen's postponed per (as four times in Lucretius, twice in Vergil inter alios) and read saecla per hinc (or haec) maneat, naturally suggesting Cat. I.10 plus uno maneat perenne saeclo; 43a (presumably so numbered because it was a late addition), for the highly implausible altis I conjecture ortis, thereby restoring a poetic plural for the most common luce orta; 68, for the nonsensical flumant minu I suggest flammantem in, arising by haplography of 'mm', false word division after 'nt' and the alteration of 'emin' to 'minu'; 81, if diuortia is nominative, quo for qua could be an improvement; 97, it is always dangerous to impose sigmatic ecthlipsis in an author not known to employ the licence, but it is worth stating (if only to reject the idea) that the unmetrical sequence here could scan as a dactylic hexameter if Courtney's transposition of arboreis were accepted, followed by frugibu' tegmina gignunt; 145.8-9, so many suggestions have been made to restore these lines of the Gallus fragment, that I make no apology for suggesting ac si confiteatur (iam Nisbet) idem tibi, non ego, Visce,/ scribere [or edere] dupla, Cato, iudice te, uereor, i.e. if Lycoris is prepared to praise Gallus' verses before such stern literary judges, then Gallus does not shrink from writing, or disseminating, twice as many, even with them as his literary reviewers; 150.6, since it is most unlikely that Virgil drew the whole line from Varius Rufus, a possible noun to replace perdita (for an adjective qualifying ardua is, though not impossible, less natural) would be semita (cf. Man. I.851, Sil. XV.102, Liu. IX.24.7), which could have easily been corrupted to the paradosis; 214, I suggest that the nonsense t(h)erua arose from ter uaga, with a scribal saut from the first to the second 'a'; 228.2, since it is most probable that iam quidem has entered this second line en bloc from the preceding line, I suggest, without any heed to the ductus, uana (adv.) petunt, or sic before Nisbet's tentative quaerunt; 236.2, for the impossible uelabant (repeated from the previous line) it may be worth considering a suggestion that does not emphasise the bright colour of the neck--for the hair has been dismissed as neglecti and the picture is not one of beauty--but rather its prominence, i.e. exstabant (cf. Ou. met. II.854); 261, since the elision of iam before 'a' is attested in Ovid, Propertius, Lucan and Statius (to say nothing of the 'less polished' poets), it would seem better to follow Haupt in believing that the first foot of a hexameter has been lost before iam adducta, rather than supplying with Housman madet betwixt the two; perhaps crescere would serve the purpose, with something akin to incipit beginning the next line?
II: Intro. (pp.1-2), 49 is too high a figure for the instances of sigmatic ecthlipsis in Lucretius, as I argue in a forthcoming Hermes article; 2 (comm.), it would have been useful to record that the 'one or two older scholars' who inserted the line into Catullus were most pre-Lachmannian editors (and L. Mueller after), following the lead of Muretus and Faernus; 7-10 (intro.), the spelling 'Smyrna' goes back not only to Goold's 1973 Groton edition but right to the princeps (a reference to the discussion in Ellis' orthographic excursus in his second Catullan edition (Oxford, 1878) would have been of use); 9, insert excipitur before 'haec aluus'; 21.1 (app.), Weichert's fine omina for omnia surely deserves a place here; 23, I do not believe that the paradosis uergatur has yet been solved with regard to both ductus and sense, but if Munro's superatur is to be printed in the text and Baehrens' superata est in the apparatus, so should the fact that Keil first laid these foundations with superata; 39.1, homines seems preferable to omnes not only due to its better sense and presence in two of the three traditions but because it is an unlikely corruption of omnes (whereas the inverse corruption is very easy after the widespread loss of initial 'h'); 45, it is unclear why reference is here made to Mueller's edition of Nonius in preference to Lindsay's (whose text is the basis elsewhere); 53, it strikes me as better method to regard neither of the differing versions as correct, but both as faulty remembrances of the same distich (particularly since Servius' credentials are typically superior), and therefore to fuse them in the most plausible fashion (mergitur undis, for instance, has the appearance of being genuine and closes the hexameters of Germanicus (4x), Lucan and Silius Italicus); 57.1, the situation with Courtney's printing (deliberately or no) fumantia for spumantia may be more complicated than a mere misquotation, for the conjecture goes back to Salmasius, but yet earlier we find Thomas de Cantimpre citing this verse with fumantia in his de natura rerum (c.1230-40) at VIII.12.45 (ed. Boese); 63.1, it is very difficult to believe in the veracity of p's o; 96 (comm.), the possibility is worth recording that a spondaic/anapaestic word was omitted, either by Macrobius or a subsequent scribe, after Centauros (a reconstruction like sanguine Centauros calido foedare bimembris/ incipit (cf. Ou. met. VI.238) being not impossible); 104 (comm.), no reference is made to Plutarch's mention (Brut. 48) of a philosopher Volumnius; 106 (test.), a lacuna should be marked before P. Varro, in which a fragment of Piso's history is cited; 109 (app.), the elliptical si ita is dubious Latin for 'if this (= the latter) is the case'; 111.4, the manuscripts suggest writing with aphaeresis laetitiast, as indeed was printed as long ago as 1584; 115 (test.), it should be made clear that the testimonium begins in the middle of a sentence (and the same should also be done with those to 178 and 194); 120.1, it is a strange fact, perhaps not worth recording, that nubes si ut uellera lanae, Buescu's conjecture (which H. accepts), opens section LXVIII of Bacon's Historia ventorum, written in 1622; ibid. (test.), 'codd. cett' is a little grandiose for two of three witnesses; 121.4 (comm.), more argument could have been useful to explain H.'s acceptance of Virgil's borrowing a whole line from Varro At. but his rejection on p.261 of his doing so from Varius Rufus; 124 (test.) does not begin intelligibly; 125 (test.) replaces with a lacuna the object in apposition with filium; 126 (test.), Courtney's (if it is his) post for propter is more economical than H.'s alteration of Actaeonis to Icarii; 127.1 (text), H.'s conjecture of tunc for nunc is somewhat unlikely because of the following guttural (see most recently Courtney (Prometheus 29 (2003), 235-40), comparing J.F. Gaertner (RhM 150 (2007), 211-24), while his suggested alternative tum is further removed from the paradosis; 129.2 (app.), Muretus seemingly first restored placida for tacita; ibid. (test. 1), de his uersibus has been omitted before dicere; ibid. (test. 2), read '56.5-6' for '55.5-6'; 132 (app.), reference could have been made to Ruhnken, who first suggested (Ep. Crit., 224) that the verse should be attributed to Book IV; 136 (comm.), to posit anapaestic scansion of Veneris before a word opening with a vowel is rash, particularly since, if dactylic, this would be in the fourth arsis; it would be better to regard the metre as iambic, perhaps the close of a trimeter if preceded by a proclitic monosyllable; 151(b) (test.), it is a shame that it goes unrecorded that epici is W. Meyer's palmary emendation for the mss' ipse; 157 (comm.), the brevity of expression with which H. says that 'Mynors quotes Varius in an unmetrical form' obscures the fact that Mynors simply retained the manuscript order of Marius Victorinus without imposing any metrical structure (hence the transpositions of Schneidewin, Buecheler or others); 166.5 (text), mihi should be punctis subnotatum; 174.7 (text), does not Baehrens' bile deserve to be recorded?; 180 (app.), the subject of tribuerunt is wanting, presumably 'edd. priores'; 181.1 (text), Watt's tantundem is attractive enough; 192 (test.), if [? tornus] is to be added after tornum, it should be recorded that L has turnus alongside MV's turnum; 194 (test.), the veracity of transposed dignum at the end of the testimonium is, to say the least, dubious; 200 (text), in order to draw out the evident pun, which may have encouraged the neuter form of humum, umidum should perhaps be printed with an initial 'h'; 205 (test.), a lacuna should be marked after duo nomina; 214, 'new' should not appear in the translation; 217.1 (app.), it is potentially misleading to use the lemma strati for stratique, even if previous editors have done the same; 219.1 (comm.), an elision such as magnanimorum hominum could hardly be described as awkward in dactylic verse; ibid.21 (comm.), acceptance of Lachmann's emendation at Lucilius 41 M. would provide a much earlier example of adjectival Emathius; ibid.22 (app.), why repeat Courtney's ed. Ven. 1490-1503 when the conjecture can be found in the edition of 1490?; 221(b) (test.), the place of Kiessling's supplement Tiberii has moved from its correct place as seen in the test. to 129 (cf. the change in the editorial supplement between 71b and 100); 228.15 (app.), the reader needs to be told in the apparatus (not only in the comm.) that the three principal manuscripts, ABV, have pectore; ibid.16, Gronovius' conjecture (of 1649) may have been preceded by the correcting hand in L; 238.1 (app.), it is worth informing the uninitiated that Micylli's Basel edition of 1535 is our only source for the text of Hyginus' fables; 262 (comm.), one would like to learn more about H.'s tentative suggestion that this opening of the/a poem is a Renaissance fabrication and about the reasons he would adduce for that surprising occurrence.
Bearing in mind the size of the work, one can say with confidence that typographical errors are relatively rare. Nonetheless, several do occur in the Latin: 5.3, Graeci for Graece (also in comm. ad loc.); 11 (test.), audio for audeo; 35 (test.) laboriosum for laboriosus; p.61, delete the second instance of in in the quotation from Prop. II.33; 150.2 (app.), read compendere for comprendere, thereby restoring a negative apparatus (the correct reading being preserved by PRATF); 158 (test.), idest; 180.1, Tibullo for Tibulle; 181.2 hominum for hominem; 185 (test.), primum should be inserted before in Bithnyia; 195 (test.), a verb (adiecit) has appeared from nowhere, an aliquid ex nilo which F. Leo also accomplished with this passage in 1878; 218 (test.), luxuriantque for luxuriatque; 219.6 (app.), hunc for tunc; 220 (comm.), surma (in headword) for syrma; 243 (comm.), Erycius for Baehrens' Erucius.
Names are occasionally misspelt, e.g. Herz (115 app.), Bahrens (179 app.), Capelli (262). H. sometimes parts from his typical procedure by Latinising the names of post-Renaissance Classical scholars: Le Clerc (184) vies with Clericus (87 test., 232) (but at least not 'Gorallus'), depending on whether H. has drawn his apparatus from Blänsdorf or Keil; without the Teubner or Courtney as his base for the tragic fragment 157, Keil's Latinisation introduces H. 'Schneidewinus' and 'Bothius', in contrast to the appearance of, say, Martinus Delrius (Del Rio) simply as 'Delrio' at 199. 'Pithou' once (39 comm.) challenges the regular 'Pithoeus', and 'Grial' appears for 'Grialius' (appropriating La Bigne's uis at the same time) in the discussion to 112. It is left either to common sense or to specialist knowledge to decide which Scaliger, Heinsius, Burman, Vossius or Baehrens is being spoken of in the apparatus, although O. Skutsch has his initial at 181.1, since the note and reference is drawn from Courtney; Pithoeus at 186 and Unger at 192 are reunited with their initials but not elsewhere; Scaliger receives 'J.J.' once and once only at 39.3.
To conclude: although I find H.'s textual work uneven in places, his broad and rich commentary--where the heart of the book genuinely does lie--is rewarding, original and instructive, and poses many important questions. Its breadth and depth guarantee it a place as a major tool on the shelves of scholars not only of Latin fragmentary poetry but also of Latin and Hellenistic poetry more widely. I believe that H. would concede that the work, though now a major presence in the field, is at its best when used alongside Courtney and Blänsdorf. Yet it is a happy certainty that all scholars now working in this area, or those bold enough to turn their minds to it in future, will find their research much aided by, and often significantly indebted to, H.'s FRP.