NOTES ON THE TEXT OF VARRO'S DE LINGVA LATINA

Abstract

This article discusses the text of seven passages in the etymological books 5–7 of Varro's De lingua Latina, and proposes new conjectures for all of them. The discussions are of direct relevance to the interpretation of fragments and testimonies of lost Latin authors quoted by Varro: the scenic poets Naevius, Pacuvius, Caecilius Statius, Juventius and Atilius, and the grammarian Aurelius Opillus. The starting point for the discussions is the new Oxford edition of Varro's De lingua Latina by Wolfgang de Melo.

INTRODUCTION

Varro's treatise On the Latin language (De lingua Latina, henceforth Ling.), written in the mid 40s of the first century b.c., is certainly not a mainstream text but rather a curiosity for enthusiasts interested in arcane Latin words, ancient linguistics, fragmentary poetry and other such topics. Only portions of this work survive: three books on Latin etymology (5–7), in which Varro quotes extensively from early Latin poets, and three on Latin morphology (8–10). Many passages are difficult to understand, not least owing to the textual tradition, which is based on the single witness F: Florence, Biblioteca Medicea Laurenziana plut. 51.10, a manuscript of the late eleventh century written in the Beneventan script of Montecassino, which preserves the text in a heavily distorted form. In addition, it is at times hard to decide whether an expression is obscure because of an error in the manuscript or owing to Varro himself, who worked quickly and, especially in the books on etymology, recycled material from his earlier antiquarian writings. This is probably the reason why the abrupt style of the etymological books (5–7) differs from the style of the later and more carefully prepared books on morphology (8–10).

In the course of preparing a new edition of the Fragments of Roman Comedy,¹ we examined the De lingua Latina, particularly the books devoted to etymology. In doing so, we were fortunate to be able to use de Melo's recent Oxford edition, published in 2019.² De Melo provides the first continuous commentary on the De lingua Latina truly worthy of the name, in which, inter alia, he identifies and discusses a number of quotations from Roman poetry which have been overlooked by their modern collectors so far.³ For one of them, cicur ingenium optineo (Ling. 7.91), which de Melo most plausibly identified as an incomplete line consisting of three anapaests, we consider the possibility that it stems from a play of Pacuvius.⁴

As for the edition, de Melo has re-examined the text of the Laurentianus and then, often following the Loeb edition of Kent,⁵ intervened in it much more frequently than the editors of what has so far been the standard text—namely, the 1910 Teubner edition by Goetz and Schoell.⁶ While with these editions we too often felt left with a barely comprehensible Latin text, de Melo has provided us with a text that is readable throughout.⁷

Moreover, de Melo's new edition gives fresh impetus to the critical examination of the text and animates fellow researchers to engage with its problems. In the hope of making a small contribution to this task, we present some conjectures which are the fruit of our examination of Varro's De lingua Latina. We concentrate entirely on the constitution of its text and not on the fragments of poetry that have come down to us from him. However, most of our conjectures are of direct relevance to the interpretation of fragments and testimonies of the lost Latin authors, both poets and grammarians, whom Varro quotes. The person responsible for a conjecture is indicated in the heading to the respective section, of which he or she is the sole author. The starting point for our textual discussions is the edition of de Melo, unless indicated otherwise.

1. LING. 5.62 (VINCENT GRAF)

In explaining the names of the gods, Varro derives Venus from uincire ‘to bind’ and cites a verse by an unknown comic poet to support his claim:

hinc comicus (comicos F, corr. Laetus): ‘huic uictrix Venus; uidesne haec?’

Instead of retaining the transmitted comicos, de Melo (like all recent editors) follows Pomponio Leto, who corrected the ungrammatical form to comicus.⁸ Although the emendation is most likely correct, it fails to heal the passage entirely. In general, De lingua Latina contains only a small number of anonymous fragments of Roman literature.⁹ In the fifth book there are only two other instances where a (part of a) verse is quoted anonymously, and in one of them it is quite possible that Varro is reproducing a proverb, not poetry, as Ribbeck had already suspected.¹⁰ All other quotations in the fifth book are assigned to individual poets. In the few cases where Varro quotes poetry without naming an author, he usually uses the passive voice or a word such as poeta, never the nominalized adjective comicus or a comparable generic term (such as tragicus, lyricus, etc.). In fact, nominalized adjectives of this kind are so rare that only one parallel can be found in the De lingua Latina. In this case, however, the term is not used in isolation but is linked to a name: Iuuentius comicus dicebat (Ling. 7.65).¹¹ Therefore, in view of this parallel and with regard to Varro's usual diction, one should assume that the name of a poet has been omitted:

hinc <…> comicus: ‘huic uictrix Venus; uidesne haec?’

It is capricious to speculate which name has dropped out. However, since Juventius, as a rather unknown poet, is more in need of the attribute comicus than a poet such as Plautus, and the omission of his name is palaeographically plausible, he himself is a good candidate.

2. LING. 7.54 (SILVIA OTTAVIANO)

Speaking about the verb carere (‘to card [wool]’), which is attested only in Plautus’ Menaechmi, Varro mentions an otherwise unknown play of Naevius: idem hoc est uerbum in Cemetria N<a>euii.

For the title of this play the Florentine manuscript transmits the word cemetria, a reading that is likely corrupt. De Melo, though not ignoring the difficulty, accepts the text of F without mentioning in the apparatus criticus any of the many conjectures made by editors of Varro. Aldus had Cosmetria and Turnebus suggested the interesting Commotria (that is, Κομμώτρια, ‘Dresser’, ‘Tirewoman’). Further conjectures are Cementria (Scioppius) and Demetria (Goetz and Schoell). I propose Cimetria, meaning ‘the girl from Cimetra’. Cimetra was a Samnite town mentioned by Livy 10.15.6 (297 b.c.) Fabius etiam urbem Cimetram cepit (cf. TLL Onom. 2.442.74). Parallels for such a title¹² can be found in the togata ¹³ (for example the Brundisinae [attested as Brundisina in some sources] of Afranius and the Veliterna of Titinius), but we might be dealing with a praetexta as well (cf. the Sabinae of Ennius), especially if we change the text to Cimetra: a drama about the conquest of a city, like Naevius’ Clastidium or Ennius’ Ambracia.

3. LING. 7.65 (VINCENT GRAF)

In a difficult passage, Varro explains the meaning of some words he found in a verse of Plautus. The words are extremely rare, and this has led to major problems in the transmission of the verse and of the accompanying gloss:

ibidem [Plaut. fr. 99 Monda]: ‘scrattae, scrupipedae, strittabillae, tantulae’.¹⁴ ab excreando scrattae ‘siccas’ significat. scrupipedam [scruppidam F, corr. A. Spengel praeeunte Müller] Aurelius scribit ab scauripeda [auscauripeda F, corr. Bothe]; Iuventius comicus dicebat a uermiculo piloso, qui solet esse in fronde cum multis pedibus; Valerius a pede ac scrupea. … strittabillas a strittilando; strittare ab eo qui sistit aegre.

In the same place: ‘Scrattae “wretched”, scrupipeda “bony”, strittabillae “tottering”, so useless’. Scrattae ‘wretched women’, from excreare ‘to cough out’, means siccae ‘dry’ ones. Aurelius writes that scrupipeda ‘bony’ is from scauripeda ‘having deformed ankles’; the comic poet Juventius said that it was from a hairy caterpillar which with its many feet is to be found on foliage. Valerius believes that it is from pes ‘foot’ and scrupea ‘difficulty’. … Strittabillae ‘tottering women’ is from strittilare ‘to totter’; strittare ‘to totter’ is said of a person who sistit aegre ‘barely keeps on his or her feet’. [Transl. de Melo]

While the etymologies for scrattae and for strittabillae are unambiguous, Varro seems to give three explanations for scrupipeda:¹⁵ 1) Aurelius derives the word from scauripeda, a hapax legomenon the meaning of which we do not know; 2) the comic poet Juventius¹⁶ says that it stems from a hairy caterpillar cum multis pedibus ‘with many feet’ that lives among leaves; 3) Valerius claims that scrupipeda is composed of the elements scrupea ‘obstacle’ and pes ‘foot’. De Melo believes that Varro preferred Valerius’ explanation, although this is not explicitly stated.¹⁷

If we accept the text given by de Melo, only the explanation of Valerius Soranus is understandable, while those of Aurelius and Juventius are puzzling: first, it is odd that a Latin grammarian such as Aurelius¹⁸ does not derive the first element of the word from the obvious scrupeus, a word common in mid Republican poetry,¹⁹ as Valerius does, but from the more obscure scaurus, which is unattested in surviving archaic literature.²⁰ Second, unlike Valerius, Aurelius does not explain the two elements of the word separately, as would be practical, but cites a hapax legomenon which is not appropriate for an etymology and rather requires an explanation itself. Third, as de Melo himself points out,²¹ the etymology of Juventius is unsatisfactory since it only explains the banal element -peda and not the much more distinctive first element of the word. In addition, the gloss as a whole is rather conspicuous, since Varro usually quotes no more than two different opinions from other scholars.²²

To solve these problems, I suggest a small addition to the text:

scrupipedam Aurelius scribit ab scauripeda, <id quod>²³ Iuuentius comicus dicebat, a²⁴ uermiculo piloso qui solet esse in fronde cum multis pedibus. Valerius a pede ac scrupea.

Aurelius writes that scrupipeda is from scauripeda, which is the word that the comic poet Juventius used, that is, from a hairy caterpillar which with its many feet is to be found on foliage. Valerius writes that it is from pes ‘foot’ and scrupea ‘difficulty’.

The correction combines the two rather puzzling explanations of Aurelius and Juventius into one that is readily understandable: Aurelius says that Plautus’ word scrupipeda stems from scauripeda, a rare zoological term he found in a comedy of the comic poet Juventius, where it denotes a certain caterpillar. In explaining the term, Aurelius may well be following Juventius himself, as is the case in a lemma in Paulus’ epitome of Festus’ De uerborum significatu, where the word inuoluus is explained as uermiculi genus qui se inuoluit pampino (Paul. Fest. 100 Lindsay); this definition, as Lindsay points out in his apparatus criticus,²⁵ is certainly based on Plaut. Cist. 729 inuoluolum, quae in pampini folio intorta implicat se.²⁶

As a whole, Varro's discussion of scrupipeda is similar to the discussion of praefica at Ling. 7.70.²⁷ There again, two different explanations are offered for a difficult word that is found in Plautus. As in the case of scrupipeda, Varro cites first the view of Aurelius (ut Aurelius scribit),²⁸ which is supported by reference to a treatise of Aristotle and a quotation from a comedy of Naevius; then Varro cites the explanation given by Servius Clodius (Claudius scribit), which is shorter than that of Aurelius. Since in the case of praefica the grammarians give two competing etymological explanations, Varro closes the discussion by offering a compromise (utrumque ostendit eqs.). In our passage, on the other hand, the two explanations are not mutually exclusive, so Varro can place both views side by side without further comment.

4. LING. 7.70: PART 1 (MARCUS DEUFERT)

Varro attaches his etymology of praefica to a quotation from Plautus’ Truculentus:

in Truculento: ‘sine uirtute argutum ciuem mihi habeam pro praefica’. <praefica> dicta, ut Aurelius scribit, …

De Melo follows (as do all recent editors) the lost humanistic MS B²⁹ and the editio Aldina in indicating a lacuna after praefica (as the last word of the quotation of Plautus) and then repeating praefica as the first word of Varro's own gloss. This is an economical conjecture (the omission being due to haplography), but it can be further improved without much effort. The complete sentence in Plautus reads as follows (Truc. 495–6): sine uirtute argutum ciuem mihi habeam pro praefica, | quae alios conlaudat, eapse sese uero non potest. The relative clause defines the praefica as a person who praises others, a fact that is of great relevance in the following gloss of Varro: Varro shares the view of the antiquarian Aurelius Opillus that the praefica is a mulier … quae ante domum mortui laudis eius caneret,³⁰ which he supports with a quotation of Naevius (com. 129): haec quidem hercle, opinor, praefica est: nam mortuum collaudat. It is therefore probable that Varro included what Plautus says about the activity of the praefica, namely her praising of other people, and that his quotation did not stop with praefica but with collaudat. His original text reads thus:

in Truculento: ‘sine uirtute argutum ciuem mihi habeam pro praefica, <quae alios collaudat’. praefica> dicta, ut Aurelius scribit, …

The words quae … praefica were omitted by a saut du même au même, a jump of the eye from the first praefica to the second praefica. The number of missing letters is twenty-six. This deserves to be mentioned, since Vetter once argued that the gaps in De lingua Latina often comprise thirteen letters or multiples thereof. He concluded that our manuscript tradition, with the Laurentianus being the extant archetype, derives from an uncial manuscript of De lingua Latina in two columns.³¹ If he is right, the words quae alios collaudat. Praefica filled two lines in this manuscript.

The error I postulate in Ling. 7.70 occurs almost certainly at 7.57 in Trinummo: ‘nam illum tibi | ferentarium <esse amicum inuentum intellego’. ferentarium> a ferendo … and at 7.105 in Colace: ‘nexum < …’. nexum> Manilius scribit omne quod per libram et aes geritur. In both cases the lacuna was recognized by L. Spengel. The same type of error has also been postulated at 7.87 apud Pacuuium: ‘flexanima tamquam lymphata <aut Bacchi sacris | commota’. lymphata> dicta a lympha, where all recent editors follow Turnebus, who supplied aut … commota on the basis of Cic. Diu. 1.80, and L. Spengel, who repeated Lymphata after the quotation quae causa cur illa exciderint.³² Here, however, the smaller supplement … tamquam lymphata’. <lymphata> dicta a lympha is sufficient and may well be correct, because the words aut … commota are irrelevant for Varro's explanation of lymphata. ³³

5. LING. 7.70: PART 2 (VINCENT GRAF)

The passage in which Varro explains the meaning of the word praefica poses further difficulties. In order to understand the problems, we need to start from the text given by Goetz and Schoell:

<praefica> dicta, ut Aurelius scribit, mulier ab luco quae conduceretur quae ante domum mortui laudis eius caneret. hoc factitatum Aristoteles scribit in libro qui <in>scribitur Νόμιμα βαρβαρικά, quibus testimonium est quod † fretum est N<a>euii: ‘haec quidem hercle, opinor, praefica est; nam mortuum collaudat.’ Claudius scribit: ‘quae praeficeretur ancillis, quemadmodum lamentarentur, praefica est dicta.’ utrumque ostendit a praefectione praeficam dictam.

According to this text, Varro first paraphrases the view of Aurelius Opillus, according to whom a praefica is a woman who is found in a sacred grove³⁴ and is hired to sing the praises of a deceased person in front of his house. As evidence that this custom existed in the past, Varro refers to Aristotle's Νόμιμα βαρβαρικά (fr. 604 Rose³ = 469/696 Gigon) and quotes a verse from a comedy by Naevius. He then cites an explanation by the grammarian Servius Clodius, who explains that a praefica is a woman who presides over a group of servants and tells them how to mourn. Varro concludes that ‘both’ explanations (utrumque ostendit) are proof of the fact that the word praefica is derived from praefectio.

De Melo correctly points out in his commentary that Varro's conclusion precludes such a treatment of the problem: the words utrumque ostendit can only refer to Aurelius and Clodius, not to Aristotle or Naevius, who are cited only as witnesses to the existence of the custom. However, it is in no way clear from Aurelius’ explanation that praefica is derived from praeficere. De Melo therefore follows a conjecture made by A. Spengel, who changes the passive verb conduceretur to an active conduceret as well as the singular caneret to the plural canerent:

<praefica> dicta, ut Aurelius scribit, mulier ab luco quae conduceret[ur], quae ante domum mortui laudis eius canere<n>t.

According to this text, Aurelius had explained that the praefica is a woman who is not hired to sing the praises herself, but who commissions other women to do so. Thus, since the praefica, just as in Clodius’ explanation, presides over a group of mourners, Varro can conclude that both Aurelius and Clodius derive praefica from praefectio.

Spengel's solution, while palaeographically plausible,³⁵ has several problems, the first and most obvious being that there is still no verb comparable to praeficere in Aurelius’ explanation, which therefore remains rather obscure. The larger problem, however, lies in the altered role of the praefica. This contradicts not only Varro's further explanation—in Naevius, after all, it is again the praefica who laments—but also a gloss in Paul's epitome of Festus, which transmits the same fragment of Naevius and apparently follows Varro (or their common source) closely by combining the two explanations: Paul. Fest. 250 Lindsay praeficae dicuntur mulieres ad lamentandum mortuum conductae, quae dant ceteris modum plangendi, quasi in hoc ipsum praefectae.³⁶ Here again, we read that the praefica does not hire other women, but is herself hired to sing the laments. Spengel's text is thus unconvincing. Instead, I suggest a different conjecture, assuming another small lacuna owing to a jump of the eye and filling it on the basis of the evidence provided in the gloss of Paulus:

praefica dicta, ut Aurelius scribit, mulier ab luco quae conduc<eretur et praefic>eretur, quae ante domum mortui laudis eius caneret.

As in Clodius’ explanation, praeficeretur here also means ‘to be put in charge’, although in this case it is used absolutely and is constructed with a final relative clause. The TLL lists a handful of parallels for this construction, the most pertinent being Cic. Verr. 2.2.144 aliquem procuratorem praeficere qui statuis faciundis praeesset, Nep. Dion 9.1 proximo die … domum custodiis saepit, a foribus qui non discedant, certos praeficit, nauem triremem armatis ornat and Tac. Ann. 6.11 duratque simulacrum (= ‘as a simulacrum the office of a city prefect lives on’) quoties praeficitur, qui ob ferias Latinas consulare munus usurpet.³⁷

Varro's utrumque ostendit thus refers to two more or less similar explanations: according to Aurelius, a praefica is hired from the grove and ‘put in charge’ (praeficeretur) of singing a dead person's praises, while, according to Clodius, a praefica ‘is put in charge’ (praeficeretur) of the lament, presiding over a group of mourners. Varro provides two explanations, giving his argument twofold support: whichever explanation one might favour, in both cases it is established that praefica is derived from praeficere.

6. LING. 7.103 (KEVIN PROTZE)

In Book 7, chapter 103, Varro treats onomatopoetic words in Latin, particularly verbs derived from animal cries such as latrare, gannire, dibalare, rudere, eiulitare. For each of these he quotes a verse or parts of a verse. The quotations are marked explicitly with the names of the poets in their genitive forms:

… multa ab animalium uocibus tralata in homines, partim quae sunt aperta, partim obscura. perspicua, ut Ennii:

animus cum pectore latrat.

Plauti:

gannit odiosus omni totae familiae.

<Cae>cilii:

tantum rem dibalare, ut pro nilo habuerit.

Lucilii:

haec, inquam, rudet ex rostris atque eiulitabit.

Eiusdem:

quantum hinnitum atque equitatum.

If we suppose that the above pattern is consistently followed throughout the chapter, the transmitted cilii in the middle leaves most likely half a poet's name in the genitive, followed by an iambic senarius. Consequently, Pomponio Leto corrected the text to <Cae>cilii, assuming the comic poet Caecilius Statius as the author of the following verse, which indeed fits the comic genre. His conjecture has been accepted by all editors so far.

Pomponio Leto, however, did not have any proof for his conjecture, since the verse is not attested elsewhere for Caecilius. Indeed, the names of other poets that fit palaeographically may be more probable than Caecilius, who is not quoted anywhere else in the De lingua Latina. The quotations in the De lingua Latina concentrate on the major authors Accius, Ennius, Lucilius, Naevius, Pacuvius and Plautus, with more than a dozen quotations from each. Others, such as Atilius, Manilius, Matius, Terentius and Valerius Soranus, are attested two to three times each, whereas Aprissius (Ling. 6.68), Catullus (7.50),³⁸ Juventius (6.50), Pompilius (7.93) and Porcius Licinus (7.104) are quoted only once. Nevertheless, a poet already cited should be preferable to an unquoted one as the basis for a conjecture of a poet's name.

There are not many poets’ names ending in -cilii in the genitive whose works Varro quotes. <Lu>cilii can be excluded easily, since he is quoted directly after our poet's verse, followed by a quotation of eiusdem. Varro uses eiusdem consistently to indicate a further quotation of the same author. Therefore, it is unlikely to assume a quotation of Lucilius (Lucilii), followed by another one indicated with Lucilii as well.

Hence, the poet Atilius remains the only candidate for the conjecture. Varro quotes him twice (Ling. 7.90, 7.106), and as a comic poet he is a plausible author for the quoted senarius. The middle consonant of the name t may seem to be implausibly different from the transmitted c, when compared to the case of <Cae>cilius and <Lu>cilius, but one must take into consideration the similarity of ci and ti in the script of the main witness F (eleventh century), which is written in Beneventan minuscule. Some other, more frequent errors owing to confused letters in the De lingua Latina, for example a and t, can be explained by their similarity in the Beneventan script as well.³⁹ Therefore, a Beneventan ancestor of F, younger than the uncial manuscript claimed by Vetter,⁴⁰ may be assumed.⁴¹ In particular, other instances of confusing c and t can be found in the De lingua Latina, for example in the name Accius, which is written Attius in 6.80 (corrected by Kent). Apart from Varro, Atilius is spelled Acilius in other cases, such as in MSS B and E of Cic. Fin. 1.3.⁴²

The omission of the initial A(tilius) is at least as probable as the omission of Cae (<Cae>cilius). Whilst the latter can be explained by a jump of the eye from the closing two letters of the preceding word familiae, nearly the same explanation can account for the omission of A alone. Given that familiae acilii ⁴³ consists of multiple letters that are very similar in the Beneventan script, a jump of the eye easily explains the omission of A.

Author names are crucial categories for editors of fragments. The main aim of my conjecture is to illustrate that there is a plausible alternative to Pomponio Leto's <Cae>cilii. Therefore, the fragment tantum rem dibalare, ut pro nilo habuerit should no longer be ascribed to Caecilius but should rather be included in the group fragmenta incertorum poetarum, with a note in the apparatus criticus indicating Caecilius and Atilius as possible authors. In Varro, the transmitted cilii should be printed in cruces.

7. LING. 7.107 (SILVIA OTTAVIANO)

In chapter 7.107, Varro lists several expressions from different plays of Naevius, providing an etymology for each of them (for example in Stigmatia praebia a praebendo). The last example is taken from a comedy called Technicus:

in T[h]echnico confictant a conficto conuenire dictum.

The word confictant is a brilliant conjecture of Turnebus for conficiant, the reading of the Florence manuscript. confictare appears only here and, again as a result of conjecture, in Pacuv. trag. 337 (= 255 Schierl) ubi poetae pro sua parte falsa conficta<nt>, ⁴⁴ canunt. De Melo translates confictant ‘they fabricate a story together’, probably because of the comparison with this passage.

Let us now turn to the etymology: Varro establishes a relationship between confictant and confictum, the perfect participle of confingo. While this is reasonable, the connection with conuenire seems less clear. Th. Bergk (‘Zu den lateinischen Komikern’, Neue Jahrbücher für Philologie und Pädagogik 16 [1870], 823–46, at 831) considered the text ‘völlig unverständlich’, and suggested assigning to Naevius the expression confictant conuenire. ⁴⁵ Other attempts to emend the text are far from convincing,⁴⁶ and de Melo's translation of a conficto conuenire (‘from agreeing on a fabrication’) does not make good sense. The syntax of the passage is peculiar as well. I cite de Melo again, who remarks in his commentary: ‘Varro's normal practice is to have a gerund dependent on a preposition, and to use the infinitive only if it is the ending that is relevant. Here he has chosen the infinitive conuenire despite the irrelevance of the ending because otherwise there might be unclarity due to the form conficto.’

For these reasons, I think that conuenire does not fit the context and I would rather expect confictant a conficto dictum. Therefore, conuenire should be deleted. For the derivation of a noun from a perfect participle, cf. Varro, Ling. 5.157 torus a torto;⁴⁷ 7.62 trames a transuerso.

Still, we must explain an important point: how could the word conuenire sneak into the text? I think that it was a gloss on confici, which derived from a wrong word separation of the corrupted reading conficiant. Indeed, confici can mean ‘to be united’ (TLL 4.201.14), which means the same as ‘to come together’ (conuenire).