Skip to main content Accessibility help
Hostname: page-component-54cdcc668b-9glzh Total loading time: 0.386 Render date: 2021-03-08T22:16:17.585Z Has data issue: true Feature Flags: { "shouldUseShareProductTool": true, "shouldUseHypothesis": true, "isUnsiloEnabled": true, "metricsAbstractViews": false, "figures": false, "newCiteModal": false, "newCitedByModal": true }


Published online by Cambridge University Press:  24 April 2013

Jarrett T. Welsh
University of Toronto
E-mail address:


In a paper in an earlier issue of this journal I endeavoured to show that Nonius Marcellus’ three glossarial sources known as ‘Gloss. iii’, ‘Alph. Verb’ and ‘Alph. Adverb’ (lists 26, 27 and 28 in W.M. Lindsay's analysis of Nonius’ sources) were compiled by a lexicographer who paid attention to both metre and sense when excerpting works of Republican poetry. That compiler always excerpted quotations of poetry such that they consisted of, or began or ended with, a metrically complete verse. That method has produced quotations of high quality that are, on several counts, considerably more transparent and more reliable than those preserved by less diligent sources.

Research Article
Copyright © The Classical Association 2013

Access options

Get access to the full version of this content by using one of the access options below.


1 Welsh, J.T., ‘The methods of Nonius Marcellus’ sources 26, 27 and 28’, CQ 62 (2012), 827–45CrossRefGoogle Scholar. In citing fragments of scenic verse I use the numbering of Ribbeck, O., Scaenicae Romanorum Poesis Fragmenta (Leipzig, 1897–83)Google Scholar. The source of an individual scholar's conjectures is recorded only in the first instance, except where confusion may arise. I wish again to record my gratitude to the Journal's anonymous reader and to its editor for helpful comments on this study, work on which was supported by the Social Sciences and Humanities Research Council of Canada.

2 Lindsay, W.M., Nonius Marcellus’ Dictionary of Republican Latin (Oxford, 1901)Google Scholar.

3 I continue to treat these three works as having been written by a single author, but would note again that this identification is only probably, and not certainly, correct.

4 On this fragment and Bothe's conjecture <tam> iucundum, see Welsh (n. 1), 843 n. 61.

5 Ribbeck, O., Scaenicae Romanorum Poesis Fragmenta (Leipzig, 1871–32)Google Scholar, 2.174; Müller, L., Noni Marcelli Compendiosa Doctrina (Leipzig, 1888), 181Google Scholar.

6 Neukirch, J.H., De Fabula Togata Romanorum (Leipzig, 1833), 192Google Scholar.

7 Onions, J.H., Nonius Marcellus: De Conpendiosa Doctrina I–III (Oxford, 1895), 156Google Scholar.

8 F. Buecheler's conjectures are reported in the apparatus of Ribbeck (n. 1).

9 Lipsius, J., Epistolicarum Quaestionum libri V (Antwerp, 1585)Google Scholar, 190 (= id., Opera Omnia Quae ad Criticam Proprie Spectant [Antwerp, 1600], 338).

10 dictitare is the usual word in comedy (Ter. Haut. 22, Phorm. 4, 743; Plaut. Trin. 99, fr. 75) but any attempt to restore dictites here produces barbaric verse.

11 In this paper I use Gratwick's systems of sublinear dots and alphabetic notation to indicate scansion, for which see Gratwick, A.S., Plautus: Menaechmi (Cambridge, 1993), 51–2Google Scholar.

12 Delrius, M.A. (= Del Rio), Syntagma Tragoediae Latinae in tres partes distinctum (Antwerp, 1593), 1.176Google Scholar.

13 Bothe, F.H., ‘Emendationes Nonianae’, RhM 5 (1837), 250300, at 253Google Scholar.

14 Cf. Livius, trag. 23 obsecro te, Anciale, matri ne quid tuae aduorsus fuas, Accius 555–6, and the fragment printed as Pomponius Secundus 1–2 (on which see § 18 below); compare the elevated or earnest tones of the speakers of Plaut. Curc. 629–31, Men. 1032–3, Titinius 32–3, Ter. An. 326–7, 899, Haut. 644–6, 1028, 1048–9, etc.

15 Plaut. Curc. 137, Merc. 321, Persa 656; Ter. Haut. 1052, Eun. 95. Cf. Adams, J.N., ‘Female speech in Latin comedy’, Antichthon 18 (1984), 4377CrossRefGoogle Scholar, at 56–8 and 60.

16 Jocelyn, H.D., The Tragedies of Ennius (Cambridge, 1969)Google Scholar, 181 n. 2 lists instances where the manuscripts of Nonius present an obvious mismatch between author and title, which should be removed by positing similar lacunae. Such errors are common in lexicographical texts; cf. Jocelyn, H.D., ‘Ancient scholarship and Virgil's use of Republican Latin poetry’, CQ 15 (1965), 126–44CrossRefGoogle Scholar, at 132.

17 For the content of that scene, see especially Fantham, E., ‘Towards a dramatic reconstruction of the fourth act of Plautus’ Amphitruo’, Philologus 117 (1973), 197214CrossRefGoogle Scholar, at 208–9.

18 Bothe (n. 13), 283; the conjecture is very weak.

19 Neither ‘pocas cosas en poca cantidad’ (López, A. López, Fabularum Togatarum Fragmenta: Edición crítica [Salamanca, 1983], 259Google Scholar) nor ‘un peu de petites choses’ (Daviault, A., Comoedia Togata: Fragments [Paris, 1981], 226Google Scholar) passes muster. Plaut. Curc. 123 de paulo paululum does not rescue the obvious corruption.

20 Cf. Ter. Eun. 281 tum tu igitur paullulum da mi operae.

21 Bothe, F.H., Poetarum Latii Scenicorum Fragmenta Vol. 2: Fragmenta Comicorum (Halberstadt, 1824), 86Google Scholar.

22 See Ribbeck (n. 5), 2.xxxvi; Rychlewska, L., Sexti Turpilii Comici Fragmenta (Leipzig, 1971), 33Google Scholar.

23 Cf. Plaut. Persa 1 and Pseud. 1113, and see Questa, C., La Metrica di Plauto e di Terenzio (Urbino, 2007)Google Scholar, 331. A comment in his apparatus shows that Müller (n. 5), 103, had entertained a similar scansion but his preferred solution was rather different.

24 Junius, H., Nonius Marcellus De Proprietate Sermonum (Antwerp, 1565), 103Google Scholar.

25 The text of this fragment is insecure, and the version given above represents the best text available; for further discussion, see Jocelyn, ad loc. My argument about Anon.’s methods casts further doubt on Lindsay's arrangement of this fragment as the end of a trochaic octonarius followed by a complete trochaic septenarius.

26 Cf. the similar but not precisely parallel treatment of adverbs in –iter (see § 6 below) or of infinitives in –ier, which occur overwhelmingly at line end and only rarely at the mid-line break of long verses, but are set with considerably more freedom in mixed-metre cantica (cf. Plaut. Cas. 220, 723, Men. 1005a; Ter. Ad. 535).

27 In Plautus, the first anceps is resolved 13 times; the second, 4 times; the third, 11 times; the fourth, twice.

28 Cf. Jocelyn on Enn. trag. 110. His comment could be strengthened, for in tragic verse, apart from the particular case of duriter in Enn. trag. 258 (see below, n. 36), only Accius 120 has such an adverb in an unexpected position, which is itself not to be trusted; see § 9 below.

29 Lipsius’ justification, that Purgamentum is never otherwise quoted but that the Priuignus often is, is not compelling, but suspicion may nevertheless attach to the title since the quotation does not provide a complete iambo-trochaic verse.

30 Cf. Plaut. Curc. 128 (anapaests), Poen. 235 (bacchiacs), Pseud. 1290 (cretics), Rud. 265a (catalectic iambic dimeters), Afranius 12 (anapaests).

31 As in Questa, C., Titi Macci Plauti Cantica (Urbino, 1995)Google Scholar. Questa identifies considerably fewer glyconic verses or cola than did Lindsay, which is in no small part due to his classification of many of Lindsay's glyconics as uersus wilamowitziani or as catalectic trochaic dimeters. In any treatment of comic lyric there are bound to be disagreements.

32 Bergk, T., Kleine philologische Schriften I (Halle, 1884), 394 n. 18Google Scholar.

33 The evidence is sparse but significant; cf. Pomponius 17, 18, 75, 179. At Pomponius 54, ampliter is set as BcD before the final metron of a trochaic septenarius; that position would be aberrant for Plautus (ampliter six times, all in verse-final position) but apparently was not felt to be so by Pomponius, who has primiter in the same position of a septenarius (Pomponius 70), or by Laberius, who has miseriter similarly (149 = fr. 85 Panayotakis, C., Decimus Laberius: The Fragments [Cambridge, 2010]CrossRefGoogle Scholar).

34 Anon. is at least as likely a candidate. In the sequence firmiter, fidele, aeqviter, pvblicitvs, proterviter and ignaviter (p. 512.22–513.14), three non-sequential entries are illustrated with primary quotations taken from the ‘Lucilius i’ (9) list, and illustrated with added quotations generally taken from Anon. The other three lemmata in this sequence could have been taken from Anon. along with the added quotations to firmiter and pvblicitvs.

35 Only Accius, praet. 28 radiatum solis liquier cursu nouo (for the text, see Manuwald, G., Fabulae Praetextae: Spuren einer literarischen Gattung der Römer [Munich, 2001], 227 n. 250)Google Scholar seems to offer a marked form in a comparable metrical position, but given the slight but definite differences observed in the treatment of fuas, of infinitives in -ier and of adverbs in -iter, it would be unwise to make liquier support the irregular position of aequiter against Anon.’s usual methods.

36 Cf. Plaut. Persa 255 (amiciter), in an iambic octonarius in a polymetric canticum. It is attractive to assign this fragment similarly to a lyric context in order to explain the position of aequiter at the rarer mid-line break. In other instances where such adverbs are placed in the interior of a verse our evidence is too equivocal to be relied upon. Since dure never occurs in Republican scenic verse, the position of duriter cannot be pressed very far (contrast Enn. trag. 265 with Afranius 251; the other five instances of duriter occur at the expected positions at line end or at the opening of a septenarius). Given Pomponius’ somewhat greater metrical liberties, the placement of toruiter (Pomponius 18) is as unhelpful as was ampliter (Pomponius 54) in the preceding section (see n. 33).

37 This solution raises the question of the presence of iambic septenarii in Roman tragedy. Readers not inclined to accept their presence in that genre may prefer a different way of correcting this fragment. See Jocelyn (n. 16 [1969]), 35.

38 For this interpretation of nam ut, cf. the exchange at Plaut. Amph. 602–3 {SOS.} nam ut dudum ante lucem a portu me praemisisti domum – | AM. quid igitur? SOS. prius multo ante aedis stabam quam illo adueneram. I am suggesting that Anon. would have read the transmitted quotation of Livius’ Aegisthus (where nam is isolated and its thought incomplete) with the same irritation that Amphitruo feels while waiting for Sosia to get to the point.

39 Contrast the isolated subordinate clauses given at e.g. Nonius p. 121.13 (Pacuvius 345–6) nisi coherceo | proteruitatem atque hostio ferociam, p. 341.20 (Accius 125–6) qui aut illorum copias | fundam in campo aut nauis uram aut castra mactabo in mare. In quotations longer than one verse Anon. quotes what is necessary to establish the sense boundary but does not quote extraneous words that precede it; cf. Welsh (n. 1), 832 n. 25.

40 Cf. also the language and triumphant posturing at Plaut. Persa 753–7.

41 Cf. the discussion of pariter (Afranius 2) in §11 below. A similar problem has probably affected the text of Atta 3, where it is necessary to transpose lupantur from its transmitted position at verse end, in order to restore a trochaic septenarius.

42 Lindsay, W.M., ‘A study of the Leiden MS of Nonius Marcellus’, AJPh 22 (1901), 2938Google Scholar, at 37–8, with further examples of disturbed quotations.

43 Second thoughts about this proposal, prompted by the referee's shared doubts about <illud>, lead me to try seṛo est; uĕṇibo ṣexta, <ṣătĭs> si ṭĭbĭ plă et, in which the displacement of uenibo would be explained in the same way, but a rather different error (omission of satis after sexta, with L 1’s sib merely anticipating tibi) would have produced the transmitted text. I am grateful to the referee for encouraging me to rethink this problem.

44 Bothe (n. 13), 280.

45 The examples are those of Questa (n. 23), 76.

46 Spengel, L., Caii Caecilii Statii comici poetae deperditarum fabularum fragmenta (Munich, 1829), 53Google Scholar.

47 p. 196.5 (Caecilius 226) and 270.5 (Caecilius 227), either from Anon. or from ‘Gloss. v’; on the latter, see Welsh (n. 1), 836 n. 37.

48 Works attributed to Licinius Macer are quoted thrice elsewhere in Nonius. Two quotations are given with a full title indication (‘Annalibus lib. II’ at p. 52.5, probably from a marginal note on Gell. NA 13.11.5; ‘Annali lib. I’ at p. 63.11, from ‘Gloss. v’). The third (p. 260.2, given without a title) was marked by Lindsay as deriving from ‘Gloss. i’. I am not confident in that attribution, because of the irregularities in Book 4 that Lindsay never fully addressed. Given Nonius’ apparent tendency in Book 4 to work, in the first instance, from the sources numbered 22–31, a work of Anon. seems rather more plausible but there is insufficient evidence for judgement. The origin and attribution of the quotation recorded against the name ‘Licinius’ without title at p. 196.21 are obscure.

49 A similar case occurs at p. 224.29, where homo, given in the manuscripts at the end of the quotation of Titinius 34/35, has wandered from its proper position at the end of the subsequent quotation of Naevius, com. 60. A possible but uncertain parallel is also found in the position of prius in Caecilius 84 together with its absence from the quotation of Plaut. Aul. 336 at Festus p. 340.17 Lindsay.

50 The reason for attributing these lemmata to ‘Gloss. i’ is not stated, but the logic is, in essence, that the beginning of a book should commence from the first of Nonius’ forty-one sources (a principle that Nonius himself does not always follow) and that the emergence of Plautine primary quotations beginning at the entries for auariter and amiciter (from source 2, the ‘Plautus i’ list) imply that, on the usual order of sources, Nonius has been using ‘Gloss. i’.

51 It is difficult to attribute to ‘Gloss. i’ the quotations of the following works: Pomponius’ Auctoratus, which is quoted only once elsewhere (p. 516.11, from list 28), both times in a complete metrical unit; Varro's Rerum diuinarum antiquitates (the Ant. are often quoted in Anon.’s work); Pacuvius’ Paulus (quoted only twice elsewhere in Nonius, never from ‘Gloss. i’); and Titinius’ Prilia (a script demonstrably cited in ‘Gloss. i’ and by Anon.).

52 The methods of ‘Gloss. i’ are considerably more difficult to determine than those of Anon. and ‘Gloss. v’. Stretches of quotations assigned to that source generally suggest that, as was seen in ‘Gloss. v’, ‘Gloss. i’ quotes complete units of sense without regard for metre. At p. 81.6–84.12, of 14 verse quotations, 3 are corrupt or otherwise uncertain; of the remaining 11, all of which are complete as far as sense, 6 are metrically complete and 5 are incomplete. For other sequences of quotations, securely attributed to ‘Gloss. i’, that present a comparable picture, see p. 94.27–95.6, 182.23–183.16 (with Lindsay [n. 2], 58 n. o) and 548.10–16.

53 For such errors, see White, D.C., ‘The method of composition and sources of Nonius Marcellus’, Studi Noniani 8 (1980), 111211, at 119–21, 129–30 and 138–9Google Scholar. For a clear explanation of ‘primary’ and of ‘added’ or ‘secondary’ quotations, see White, 116.

54 Several omissions support the hypothesis that Nonius, once he realized his mistake, ignored most of his ‘Plautus i’ material and resumed searching for relevant words from Poen. onwards. It is otherwise difficult to explain the position of the following entries within Book 11, the omission of the indicated Plautine verses (which Nonius would have encountered if he had followed his usual methods), or both: in blanditer (p. 510.5), the omission of Plaut. Asin. 222; in ampliter (p. 511.18), the position of the entry and the omission of Cas. 501, Cist. 598, Merc. 99 or Mil. 758; in firmiter (p. 512.22), the position of the entry and the omission of Cas. 132 or Epid. 83. The omission of these examples suggests that Nonius skipped ahead and began searching more carefully from the Poenulus, rather than from the beginning of the list.

55 Some data from three randomly selected sequences of primary quotations from the ‘Plautus i’ list: of 32 quotations in Book 1 (p. 3.26–12.21), all give complete units of sense, 22 are metrically complete, 8 are metrically incomplete and 2 are too corrupt for judgement. Of 8 quotations in the passive-voice section of Book 7 (p. 476.15–476.34), all give complete units of sense, 2 are metrically complete and 6 are metrically incomplete. Of 8 quotations in Book 8 (p. 483.8–486.22), all give complete units of sense, 2 are metrically complete, 3 are metrically incomplete and 3 are slightly corrupted. This evidence suggests that Nonius is giving quotations in units of sense without regard for metre. White (n. 53), 194, called attention to the tendency, observable throughout the work, to ‘begin rather grandiosely and then peter out’, and we should view data from the first book with that important point in mind. Cf. also the more general point made at Welsh (n. 1), 843–4.

56 This claim is not meant to rule out the possibility that any particular quotation derives from marginal annotations in one of those lists.

57 The quotation of Accius 371–2 in particular seems like the work of Anon. One expects that, had Nonius himself excerpted it from his copy of the Eurysaces, he would have given no more than the end of 372, tute ipse te confirma et conpara, which amply illustrates his definition (p. 256.10, conparare ueteres confirmare et constituere dixerunt), or just possibly the text of the fragment apart from nihil est, since the text si autem … subsistat in some fashion corroborates, by opposition, the meaning of conpara. This is necessarily a subjective judgement, but cf. the data on Nonius’ own methods of excerption, and in particular the decline in his diligence, in n. 55.

58 Daviault (n. 19), 104; López López (n. 19), 73; Guardì, T., Titinio e Atta: fabula togata, i frammenti (Milan, 1984), 48Google Scholar.

59 Cf. n. 36 above.

60 G. Hermann, ‘Adnotationes ad Io. Henr. Neukirchii librum de Fabula Togata Romanorum editum Lipsiae a. 1833’, in id., Opuscula V (Leipzig, 1834), 254–88, at 269, to produce trochaic septenarii (ergo ut sua seruet bona | nobis faciundum est, inuita ut cum ea primum blanditer | comparemus colloqui); cf. Afranius 4. Lindsay reported that Onions resuscitated the conjecture.

61 Cf. above, n. 50.

62 Cf. White (n. 53), 130 with n. 15, on material suggesting that ‘Nonius sometimes knew in advance which lists were likely to prove useful and went straight to them’.

63 For a full account of such attempts, see Ribbeck.

64 Brugman, O., Quemadmodum in iambico senario Romani ueteres uerborum accentus cum numeris consociarint (Bonn, 1874), 19Google Scholar.

65 Ribbeck suggested that the thought was incomplete and that mulcent was wanted to pair with beniuolentia, which depends to no small degree on his own reshuffling of the fragment.

66 Significant features include the fact that a complete sotadean is recorded in Afranius 202 although the first four words are not necessary to establish the meaning of dicta facessas; that two quotations from Pacuvius’ Teucer are adduced (which work is cited only from Anon.’s compendia, apart from one obscure case at p. 159.23 and the incidental citations of Pacuvius 314 from its appearance in a Menippean satire of Varro); and that Titinius 51–2, recorded in a metrical form characteristic of Anon., begins from a sense boundary in at. Apart from the two rare Afranian scripts (for which our evidence is insufficient), the other literary works were all demonstrably included in Anon.’s word-lists. In Enn. trag. 136, Müller's <e>dico restores the septenarius and, with it, Anon.’s methods.

67 Lindsay's attribution of it to the ‘Plautus i’ (2) list is plausible, but on my explanation it would be difficult to accept that Nonius proceeded from list 26 or 27 back to list 2 before moving onwards to list 29. The quotation of Plaut. Rud. 1061–2 violates Anon.’s methods and cannot have been set down in that form by him. In Book 4 Nonius himself was responsible for the arrangement of the entries. The fuller context of Rud. 1061–2 suggests that Nonius may have obscured the shape of Anon.’s quotation by omitting some distracting words. The full dialogue gives an exchange in which facessere is used twice, in two different senses (immo ego eloquar. :: ego, opinor, rem facesso. :: si quidem | sis pudicus, hinc facessas). Anon. therefore perhaps originally quoted that full exchange and did not distinguish between shades of meaning of facessere, but when Nonius wrote his sub-divided entry it was necessary to omit rem facesso and the preceding five words in order to make it clear that only in hinc facessas was the verb synonymous with recedere. That argument would restore Anon.’s normal methods, although it is possible that the quotation came simply from the Plautine list by an aberration in Nonius’ usual procedures.

68 See Welsh (n. 1), 842 n. 59. About this script in particular I remain sceptical of being able to differentiate between quotations from Anon. and ‘Gloss. v’.

69 Comparable confusion over marginal annotations can now be seen as reason why Nonius attributed Pomponius 26 to ‘Afranius Bucco Adoptato’ at p. 126.8; cf. Welsh (n. 1), 842 n. 59.

70 On the methods of ‘Gloss. v’ (38), see Welsh (n. 1), 836 n. 36. The quotation is given in a form more typical of list 26 than of list 38.

71 Guardì, T., Cecilio Stazio: i frammenti (Palermo, 1974), 90–1Google Scholar; cf. id., Due titoli ceciliani’, Pan 1 (1973), 1317Google Scholar, at 16–17.

72 Schmidt, B., ‘Zur römischen Tragödie’, RhM 16 (1861), 586600Google Scholar, at 596–7; Hertz, M., De Scaeuo Memore poeta (Breslau, 1869)Google Scholar, 4 n. 3; Müller (n. 5); Lindsay (n. 2), 51, tacitly accepted the attribution to Accius but was more sceptical in his edition of Nonius; Keyser, P.T., ‘Late authors in Nonius Marcellus and other evidence of his date’, HSPh 96 (1994), 369–89Google Scholar, at 372 n. 11.

73 One might imagine that Anon. recognized in Pomponius Secundus the characteristic style of early Roman tragedy and therefore did not scruple over including an excerpt from his Atreus; cf. Tarrant, R.J., ‘Greek and Roman in Seneca's tragedies’, HSPh 97 (1995), 215–30Google Scholar, at 220–2.

74 See Keyser (n. 72), 370–2, with abundant references. Such comments are particularly common in connection with ‘Gloss. v’.

75 Nonius p. 202.10, 210.37, 281.1, 415.23, 505.4.

76 Hertz (n. 72), 4 n. 3.

Full text views

Full text views reflects PDF downloads, PDFs sent to Google Drive, Dropbox and Kindle and HTML full text views.

Total number of HTML views: 18
Total number of PDF views: 58 *
View data table for this chart

* Views captured on Cambridge Core between September 2016 - 8th March 2021. This data will be updated every 24 hours.

Send article to Kindle

To send this article to your Kindle, first ensure is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

Note you can select to send to either the or variations. ‘’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Available formats

Send article to Dropbox

To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

Available formats

Send article to Google Drive

To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

Available formats

Reply to: Submit a response

Your details

Conflicting interests

Do you have any conflicting interests? *