BMCR 2003.08.16

Virgilius Maro grammaticus: Opera omnia. Bibliotheca scriptorum Graecorum et Romanorum Teubneriana

, , Opera omnia. Bibliotheca scriptorum Graecorum et Romanorum Teubneriana. Munich and Leipzig: K.G. Saur, 2003. xviii, 266 pages ; 21 cm.. ISBN 3598712332. EUR 128.00.

The seventh-century grammarian Virgilius Maro compiled a set of fifteen epitomae, of which twelve are preserved without their dedication, and eight epistolae on the eight parts of speech, which all survive together with a preface. These works are written in a Latin sometimes straightforward, sometimes wilfully perverse, and convey doctrine that ranges from the commonplace to the absurd, including the existence of twelve Latinities; the two lists differ, but one includes a Latinitas philosophica of which Virgilius has much to say elsewhere. Another highlight is the division of words into discontinuous syllables, and of phrases into groups of identical letters ( scinderatio fonorum). All this is illustrated with quotations from bogus authors and bogus quotations from known names; both the author’s origins and his intentions have been disputed, but early diffusion points to Ireland,1 and the content indicates satire on grammarians, perhaps even on the Church.2 His works, last edited by Giovanni Polara (Naples, 1979) with an Italian translation, are now presented here by Bengt Löfstedt (hereinafter L.), who has devoted many years to the study and editing of post-antique, particularly Irish, grammarians. This edition is said to represent Bernhard Bischoff’s testament, being based on materials collected by him over a lifetime but handed over to L. as death approached (pp. ix-x).

At the outset, readers may be disappointed to find nothing in L.’s preface on the controversial topics of Virgilius’ origin and intentions; they will be more so not to find a clear statement of his editorial principles. On the one hand, his edition is based on expert knowledge that few of his readers share and on judgements that few are able to challenge; on the other, we are required to take more on trust than we might wish to take from any editor of any text.

Although L. justifies his new edition not on any defects of Polara’s text, which he declares to need little correction, but on those of his apparatus, asserting that unless we know what is in Virgilius’ manuscripts we shall never understand him (p. ix) — not indeed that we necessarily shall once we do — his account of the tradition hardly extends beyond descriptions of the individual manuscripts. Nor, having asserted that stemmatic methods will not avail the editor, who must evaluate every reading of each manuscript by itself (p. xiv), does he set out the criteria for such evaluation, though no-one seems better qualified to have gained insights into the habits both of the author and of his copyists. When he admits that since he has no conception of ‘philosophical Latin’ the text of passages dealing with it is highly insecure,3 we should like to be told how he has edited them, for where interpretatio and emendatio fail, there the principles of recensio should be at their clearest. All in all L.’s preface is extremely disappointing, as in its painfully plain and paratactic Latinity, so in its chariness of argument and explanation.

That chariness extends to the apparatus, which is rarely used to explain his choice of reading, even when it differs from Polara’s; we are evidently expected to rely on our general knowledge of late Latin and to avail ourselves of the Index Verborum et Formarum, with whose help, for example, when puzzled at the outset by the mention of a prophet ‘eon Persas’ ( Epist. pr. 3), for which previous editors had read ‘con’ for ‘eon’, we discover that eon is philosophical Latin for apud ( Epist. 7. 11), that con is an adverb meaning illuc ( Epit. 9. 13), and that nevertheless Mai and his successors have wantonly substituted the latter for the former.4 Was a cross-reference too much to add at the first occurrence?

At this point, exposition will become clearer if I treat of the tradition before returning to the text. The letters (a fragment in Karlsruhe apart) and the last epitome are preserved only in a Neapolitan manuscript of the ninth century, Bibl. Naz. IV. A. 34 = N, whose scribe was not only careless but undertook to abridge his text; the other epitomes are preserved in three other ninth-century manuscripts, Amiens, BM 426 = A, Paris, BNF lat. 13026 = P, supposedly the best (p. xi), Vienna, ÖNB ser. nov. 85 + ser. nov. 3762) = V. The first eleven are also in Oxford, Bodleian Library, D’Orville 147 = X (dated 1465; the worst), and part of the eighth is on the single leaf Munich, BSB Clm 29478 (1, olim 29014, = Mo (s. ix).5 Polara, who knew neither X nor Mo, distinguished ANPV as descendants of a single, variant-bearing, hyparchetype from five other manuscripts of the direct tradition, all containing only excerpts; the view that X and Mo also derive from the same source as ANPV seems to emerge, not only from L.’s silence, but from his suspicion that X’s occasional good readings are the fruit of conjecture (p. xii).

Redefining the direct tradition as those manuscripts which he presumes (whether ‘praesumo’ bears its strict philosophical sense, or simply means ‘credo’, is not clear) to contain the whole text or its greater part (p. x), L. throws together the excerpts with the quotations in Munich, BSB Clm 6415, which he calls Clm, as belonging to the indirect tradition. That is, at least in principle, to confuse the free-standing but faithful excerpt with the extract adapted to a new context and a new purpose; it is competent for L. to demonstrate that there is no difference in practice, but he does not attempt to do so. However, by refusing to note Clm’s ‘lectiones falsas’ (p. xiii), which phrase conflates the accidental and the deliberate, he implies that they preserve no vestiges of an independent tradition.6 It would follow that the source of ANPV, or rather AMoNPVX, is not a hyparchetype (p. xiv), but the archetype, however corrupt, of the entire tradition. That may be so, but one would expect to see the case stated; and if it be so, is it really true that no stemmatic relations can be established amongst the individual manuscripts? Even if detailed discussion were reserved for a specialist journal,7 the user of a textual edition is entitled to the broad lines of argument, with a few illustrative examples.

Since I live in Oxford, and have recently holidayed in Munich, I can confirm from full collation of Mo and partial collation of X and Clm, that L. is generally accurate in his positive reporting; but at Epit. 7. 91 X reads ‘mintantium’ not ‘mindantium’, and at Epit. 9. 102 ‘sed sic estentur’, where L. writes ‘caestentur Clm‘, the manuscript reads, without word-division, ‘sedsicaestentur’. Negative reporting is less reliable, often citing or suppressing the same facts at haphazard. In Mo, L. does not indicate where the text is illegible save at Epit. 8. 162, where the ink is lost, but the indentations of the pen seem as least as compatible with Winterbottom’s ‘illa’ as with the ‘ita’ of ανπχ;8 in the same line he fails to report that ‘a’ is omitted, and having noted the corruption ‘Latinis’ for ‘Latinus’ (120) and the normalization ‘nisi’ for ‘nissi’ (144), he omits the exact parallels ‘priis’ (130) and ‘conpositiones’ (150-1). Likewise in X, which at Epit. 185-6 ‘ne, sicut sepe notauimus, duae uoces esse potentur [= putentur ]’ presents ‘putauimus’ and ‘notentur’, L. cites the latter but not the former. If at Epist. 2. 237 it was worth noting that Clm reads ‘monosyllabum’ for ‘monosyllabe’, why not at 234?

At Epit. 9. 104 ‘coniungebant uel duo praeterita’ Clm puts the verb last, at 109 it reads ‘ceteris’ for ‘reliquis’; had either been noted, the reader would have asked whether it was a deliberate intervention by the compiler or a variant within the tradition and then craved guidance on Virgilius’ usage. At Epist. 2. 231, N’s ‘Ista sunt’ can hardly be the full authentic text, since ‘ista’ ought then to be the other pronouns, excluding hic, not, as it turns out to be, the six articles (always neuter in Virgilius) including it; although Clm’s ‘Quot sunt articuli ascultemus. Isti sunt articuli’ has been adjusted at least as to the normalized gender, we may wonder whether Virgilius wrote ‘Ista sunt articula’.9

By passing over the errors and omissions of X (p. x), L. leaves the reader mistakenly supposing it present when it is not: to take but one example, in Epit. 7, which in places is heavily abridged, he does not distinguish those passages that are present from those that are not, even when ll. 149-91 are represented only by ‘dictito .i. saepe dico uiso uisas uisat primae coniugationis’, retaining ‘ (X)‘ in the register of witnesses even when X preserves not a word of text. On the other hand, he indicates in the margin the pages of A, N, and P.

In numbering the epitomes L. follows A for his first eleven; since his eleventh epitome is numbered ‘IIII’ in MS N, and followed by an epitome numbered ‘V’ found nowhere else,10 he makes the latter, which is clearly the last, no. 12, without even noting that three are missing. He records that no. 8, on six accidentia uerbi, is divided in PX into 8 and 9 after the first (69-71), but not that at 128, following the fourth, Mo rubricates ‘Nunc de significatione’. Nevertheless, he is right to present a continuous text, for making a new epitome for each would raise the total from twelve to seventeen, or two too many; but by the same token we may question whether the explicit and incipit formulae printed at lines 69-71 have any business in his text. Contrariwise, since the concluding paragraphs of no. 11 have nothing to do with the main subject of etymology, this would appear to be the locus of abridgement; L.’s twelfth epitome is Stangl’s and Polara’s fifteenth. L. explains neither his numeration of the epitomes nor his placing of the letters before them, following N. against Virgilius’ explicit chronology, which makes it impossible to retain the useful sigla ‘A’ and B’ for the easily confusible Epit. and Epist. Readers with distrustful minds, remembering van den Hout’s Fronto, will suspect an attempt to enforce citation by Teubner page and line.

If now we turn to the text, it is difficult to judge L’s choice of readings without a commentary; at Epit. 8. 128 one might be inclined towards Mo’s ‘edicturus’ by Epist. 1. 3, Epit. 5. 2), especially if at 162 the scribe really wrote ‘illa’. At Epit. 8. 193, 242-3, where P, joined in the former place by N, exhibits ‘plusquam’ as the name of a verbal tense, L., having previously rebuked Polara for retaining it, gives the full form ‘plusquamperfectum’; it is possible that plusquam entered the text from school slang only after Virgilius, but we require reasoned argument, not dogmatic assertion.11

Since Virgilius does not discuss the genitive plural of pes, we are not surprised to see him use the normal form pedum (see Epist. 1. 60, 2. 273, Epit. 3. 23, 4. 21-2, 131); but at Epit. 2. 17, 3. 22 L. adopts the late by-form peduum, analogous to manuum. In the former place a fault of redaction has reduced his note to ‘pedum cett.‘; in the latter we read ‘peduum V : pedum cett.‘. Hence, even if only one manuscript presents the anomalous form, L. is ready to accept it, but he will not even conjecture it in the next line, where no manuscript happens to carry it. If there be a defensible ratio edendi here, it ought to have been stated.

Such a ratio cannot have been mere kowtowing to manuscripts, for L. is not afraid to emend: the names of Stangl, Winterbottom, and Polara occur frequently in his apparatus, as does scripsi. Perhaps he might have gone further, particularly when the only witness is N: even if the modal variation at Epist. 1. 184-5 ‘”Ego” autem dipthongo careat et sic declinatur’ were bearable, at Epist. 2. 267-8 ‘eligant utrum omnia pronomina articula uocent … an uero nullum ex articulis excepto hic in articulum recipiatur’, sense requires not articulis but pronominibus.

So far as spelling is concerned, although L. regards P as the most faithful witness (p. xi), he does not attempt to impose an orthography thence derived. At Epit. 1. 59, 2. 14, where P spells the Greek-derived verb meaning ‘to trace characters, write’ with initial cra-, L. adopts that form; but in texts preserved only by N he consistently spells caraxare even though at Epist. 1. 177, Epit. 12. 29 the first hand wrote cra- and though at Epit. 2. 14 both N and its corrector have cara-. Yet it seems unlikely that Virgilius, relying on prophetic powers borrowed from his namesake, varied his practice in accordance with the future state of his transmission. Even harder is it to understand why at Epist. 2. 239 L. accepts Mai’s (and, as he does not remark, Clm’s) ‘compositum’ for N’s ‘cumpositum’, but reads ‘cummixtum’ at Epit. 8. 158-9 on the authority of P’s ‘cummixtuim’. Matters are made no easie by Epist. 3. 595-6, where according to N prefixed cum adds nothing to concessent ( sic); one would like to know whether author, scribe, or editor has a system.

One manifest fault of editorial technique is the use of angle brackets at Epist. 2. 107 with ‘tamen’, omitted in N but preserved in the Ars Bernensis, in angle brackets, and likewise 233 ‘doctores’, 239 ‘tunc pronomen est’, both preserved in Clm, as if they had been supplied by conjecture; that would be correct only had the texts been taken specifically from N, so that Clm’s ‘adserit’ (235), which L. adopts in preference to N’s ‘adsentitur’, was a happy conjecture or an even happier interpolation. The implication is sufficiently refuted by Epit. 7. 173-5, where Clm presents an adaptation of the fuller text abbreviated in N. At Epist. 1. 200 ‘tibi’ should be in quotation-marks and wear an acute on the second vowel. Flaws in the apparatus, apart from failure to go into or come out of italic, are Epit. 2. 17 noted above Epit. 8. 158-9, where L.’s text is ‘cummixtim’: does ‘commixtim N (corr. N 2)‘ indicate correction to ‘commixtum’ or ‘cummixtum’?

L. provides parallel passages in a separate bank of annotations, but only closely related passages or borrowings: when Terrentius and Galbungus take a fortnight to debate the vocative of ego ( Epist. 2. 70-90), reference is made to one of the two citations in the anonymous Ad Cuimnanum, but we are not informed that this had been a matter of serious concern to grammarians both Greek and Latin.12

There are indexes of the few genuine quotations, of the parallels cited, and of words and forms, but astonishingly none of proper names; if one wishes to review the various appearances of Aeneas or Galbungus, one must find them for oneself, or use Polara.

It is likely that if L. were to explain himself at length he could give good reasons for judgements that I have failed to understand. Although if any author needs needs a full-scale commentary it is Virgilius, even within the confines of a Teubner text L. could have done more to help his reader by presenting a more expansive preface and including brief explanations in his apparatus; as it is, he has offered us the editio minor of what at full length might well be a master-work.


1. See Dáibhí Ó Cróinín, ‘The date, provenance, and earliest use of the works of Virgilius Maro Grammaticus’, Günter Bernt, Fidel Rädle, and Gabriel Silagi (eds.), Tradition und Wertung: Festschrift für Franz Brunhölzl zum 65. Geburtstag (Sigmaringen, 1989), 13-22; repr. Early Irish History and Chronology (Dublin, 2003), 191-200. Although Virgilius speaks of ‘our Gauls’ ( Epist. 2. 255, 3. 207-8, 211), and claims to have been taught by the Germanic-sounding Galbungus, he notes the primacy of the verb in Irish ( Epit. 4-6), and takes purely syllabic verse for granted; his playful attitude to language too seems thoroughly Irish.

2. Vivien Law, Authority and Grammar in the Seventh Century (Cambridge, 1995); doubted by James E. G. Zetzel, BMCR 95.10.23, but eyebrows must have been raised at ‘Domini, eloquentissimi et modestissimi uiri’ ( Epist 4. 126-7). Like the author of the Historia Augusta, he lifts the veil on his own foolery when he makes one of his teachers, called Aeneas, bid him not to trust his own inuenta more than his preceptors’ examples, and thus betray improba falsitas ( Epit. 5. 190-3).

3. At least, so I interpret ‘Confiteor me de philosophica quae dicitur Latinitate nihil comprehendere, quare ii loci, qui in ea uersantur, plus quam incerti sunt.’

4. Hence L.’s discussion in Latomus 40 (1981), 121-6 at 122-3 is superseded; other forms too cited in that article from Polara’s text are not read in L.’s.

5. L., pp. xii, xv, gives the obsolete shelfmark first, and omits the ‘(1’ in the new.

6. At Epist. 2. 237 ‘quod’ for N’s ‘quando’ is plainly an error, not an interpolation; it is therefore worthy or unworthy of record according as Clm’s exemplar was or was not independent of the direct tradition.

7. It is not clear whether L.’s view of the tradition will be set forth in the forthcoming Eranos article mentioned at p. xiv n. 1.

8. It is possible that technology could provide a definitive answer; in any case a full account of the tradition would list all such places in Mo, together with other readings that L. excluded from his critical apparatus: Epit. 8. 122 ‘dicemtce’, 123 ‘arcere erant’, 126 ‘inquid’, 127 ‘depresis’, 151 ‘scripturam’. At 156, where L. reads ‘Cummonia’, Mo normalizes the first vowel; the second is illegible, not least owing to a hole.

9. Although ‘Quot … ascultemus’ may owe its verb to the immediately preceding ‘scultemus’ in N, the compiler had no need to concoct that sentence: ‘Isti sunt articuli’ would have begun his excerpt just as well.

10. In fact N presents the epitomes in the order 1, 10 (‘secundus’), 4, 11 (‘ιιιι’ 12 (‘V’), 2, 3, 5, 6, 7, 8, 9.

11. ‘The abbreviated form … should certainly not have been kept’: Speculum 56 (1981), 205-8 at 206. I have encountered plusquam on a German academic website; cf. the abbreviated numerals in Bede, or the originally Scottish ‘penult’.

12. See Michela Rosellini, edn. of Ps. Remmii Palaemonis Regulae (Hildesheim, 2001), pp. 122-3 with my remarks, BMCR 2003.03.24 n. 10.