¹ It is not found in his grammatical or metrical treatises, Commentarius in Artem Donati, De Finalibus, De Centum Metris, De Metris Horatii.

² Nor is Havet substantiated by a consideration of any of the ancient grammarians edited by Keil in his Grammatici Latini. The attempts made by them to define or explain rhythmus are as follows: (1) Varro, as quoted by Diomedes (Keil, , i. 513Google Scholar): ‘Varro dicit inter rhythmum, qui Latine numeras vocatur, et metrum hoc interesse quod inter materiam et regulam.’ (2) Remmius Palaemon, as given in Victorinus' Ars Palaemonis De Metrica Institutione (Keil, , vi. 206Google Scholar, who cites it in the running title as Victorini de Metris et de Hex.): ‘Metro quid videtur esse consimile? Rhythmus. Rhythmus quid est? Verborum modulata compositio non metrica ratione sed numerosa scansione ad iudicium aurium examinata, ut puta valuti sunt cantica poetarum vulgarium. Rhythmus ergo in metro non est? Potest esse. Quid ergo distat a metro? Quod rhythmus per se sine metro esse potest, metrum sine rhythmo esse non potest. Quod liquidius ita definitur: metrum est ratio cum modulatione, rhythmus sine ratione metrica modulatio. Plerumque tamen casu quodam etiam invenies rationem metricam in rhythmo, non artificii observatione servata, sed sono et ipsa modulatione ducente.’ (3) Terentianus Maurus, De Metris (Keil, , vi. 374Google Scholar, lines 1628–33): ‘Hoc sat erit monuisse, locis quod quinque frequenteriugem videmus inveniri dactylum; sed non et sextum pes hic sibi vindicat umquam, nisi quando rhythmum non metrum componimus. Namque metrum certique pedes numerusque coercent; dimensa rhythmum continet lex temporum.’ (4) Marius Victorinus (Keil, , vi. 41 f.Google Scholar): ‘… cuius origo de arsi et thesi manare dinoscitur. Nam rhythmus est pedum temporumque iunctura velox divisa in arsin et thesin, vel tempus quo syllabas metimur. Latine numerus dicitur…. Differt autem rhythmus a metro, quod metrum in verbis, rhythmus in modulatione ac motu corporis sit, et quod metrum pedum sit quaedam compositio, rhythmus autem temporum inter se ordo quidam, et quod metrum certo numero syllabarum vel pedum finitum sit, rhythmus autem numquam numero circumscribatur; nani ut volet protrahit tempora, ita ut breve tempus plerumque longum efficiat, longum contrahat. Unde et rhythmus, id est a rhysi et fluore quodam, nuncupatur. Rhythmorum autem tres esse differentias volunt in dactylo, iambo, et paeone….’ Ibid. 44: ‘Inter pedem autem et rhythmum hoc interest, quod pes sine rhythmo esse non potest, rhythmus autem sine pede decurrit; non enim gradiuntur mele pedum mensionibus sed rhythmis fiunt.’ Ibid. p. 113: ‘… sic ne loqui quidem aut verbum ullum emittere quod non in pedes aliquos et in rhythmos incidat, qui alterna syllabarum sublatione et positione continentur’. (5) Atilius Fortunatianus (Keil, , vi. 282Google Scholar): ‘Inter metrum et rhythmum hoc interest, quod metrum circa divisionem pedum versatur, rhythmus circa sonum, quod etiam metrum sine plasmate prolatum proprietatem suam servat, rhythmus autem nunvquam sine plasmate valebit. Est etiam rhythmus et in corporali motu; cum enim histrio indecenter signum aliquod expressit, arythmos dicimus, 〈cum〉 decenter, eurythmos. Item si fuerit aequalitas corporis modice temperata, eurythmos; inaequalis vero et toris quibusdam confusa arythmos appellato….’ (6) Charisius (Keil, , i. 289Google Scholar): ‘Sequitur autem per rhythmon et melos, non quia non omnia metra rhythmoe sint. Multis enim generibus dicitur rhythmos secundum qualitatem rei subiectae, quam corporalem vocant, ut homines, columnae, velporticus dicuntur rhythmos 〈vel〉 sine rhythmo…. Nihil est enim inter rhythmon et metron nisi quod rhythmos est metrum fluens, metron autem sit rhythmos clausus….’ (7) Diomedes (Keil, , i. 468Google Scholar): ‘Rhythmi certa dimensione temporum terminantur et pro nostro arbitrio nunc brevius artari nunc longius provehi possunt.’ Ibid. 473: ‘Rhythmus est pedum temporumque iunctura cum levitate sine modo. Alii sic: Rhythmus est versus imago modulata, servans numerum syllabarum, positionem saepe sublationemque continens.’ Ibid. 474: ‘… metrum est quod certis pedum quantitatibus qualitatibusque rhythmo discriminatur. Distat enim metrum a rhythmo, quod metrum certa qualitate ac numero syllabarum temporumque finitur certisque pedibus constat ac clauditur, rhythmus autem temporum ac syllabarum pedumque congruentia infinitum multiplicatur ac profluit.’ (8) Mallius Theodorus (Keil, , vi. 588Google Scholar): ‘Siquid ergo praeter haec [ = eight metres just listed] quod non ad certam pedum legem sed ad temporum rationem modumque referatur vel scribet quispiam vel ab alio scriptum leget, id non metrum sed rhythmum esse sciai….’ (9) Beda (Keil, , vii. 258 f.Google Scholar): ‘Haec de metris eminentioribus commemorasse sufficiat…. Praeterea sunt metra alia perplura, quae in libris centimetrorum … quisque cupit reperiet…. Videtur autem rhythmus metris esse consimilis, quae est verborum modulata compositio, non metrica ratione sed numero syllabarum ad iudicium aurium examinata, ut sunt carmina vulgarium poetarum. Et quidem rhythmus per se sine metro esse potest, metrum vero sine rhythmo esse non potest. Quod liquidais ita definitur: metrum est ratio cum modulatione, rhythmus modulatio sine ratione. Plerumque tarnen casu quodam invenies etiam rationem in rhythmo, non artifici〈i〉 moderatione servata sed sono et ipsa modulatione ducente, quem vulgares poetae necesse est rustice, docti faciant docte. Quo modo et ad instar iambici metri pulcherrime factus est hymnus ille praeclarus:

rex aeterne domine,

rerum creator omnium,

qui eras ante saecula

semper cum patre filius,

et alii Ambrosiani non pauci. Item ad formara metri trochaici canunt hymnum de die iudicii per alphabetum:

apparebit repentina

dies magna domini,

fur obscura velut nocte

improvisos occupans.’

(10) Audax (Keil, , vii. 331Google Scholar): ‘Metro quid videtur esse consimile? Rhythmus. Quid est rhythmus? Verborum modulata compositio, non metrica ratione sed numero ad iudicium aurium examinata, ut puta veluti sunt cantica vulgarium poetarum. Rhythmus ergo in metro non potest inesse? Distat quidem a metro; verum tamen rhythmus per se sine metro esse potest, metrum sine rhythmo esse non potest. Quod liquidius ita definitur: metrum est ratio cum modulatione, rhythmus modulatio sine ratione. 〈Plerumque tamen casu quodam invenies etiam rationem〉 in rhythmo, non artificii observatione servata sed sono et ipsa modulatione ducente.’ (11) Anon, fragm. Parisinum (Keil, , vi. 631Google Scholar): ‘Dicimus rhythmum esse ubi tantum legitimi pedes sunt et nullo modo certo fine; metrum esse ubi pedes legitimi certo fine coercentur. Rursus quoniam eorum metrorum quae certo fine clauduntur alia sunt in quibus non habetur ratio divisionis circa medium, alia sunt in quibus habetur, est haec differentia notanda vocabulis. Quapropter illud ubi non habetur haec ratio rhythmi 〈genus〉 metrum vocatum est; hoc autem ubi habetur versus nominatur.’

¹ The words scandere, scansio occur in his notes on Georg, ii. 256Google Scholar; Aen. ii. 778Google Scholar 〈ter〉; x. 129; while metrum, metra, propter metrum, contra metrum, metri ratio, metrica ratio, metri causa (and causa metri), metri necessitas are found over three-score times scattered throughout the commentary. Cf. also De Pedibus in his Commentarius in Arterius Donati (Keil, , iv. 425 f.Google Scholar), De Centum Metris (ibid. 456–67), and De Metris Horatii (ibid. 468–72), passim.

² Cf. mulcend Nemes. Ecl. i. 53Google Scholar; laudand ibid, ii. 80; experiund Auson. 11. vi. 4 (ed. Peiper); fand V. xxi. 18; manend XIII. v. 2; faciend XIV. v. 4; patiend ibid.; profand XVIII. xxii. 54; percontand XIX. xxxiii. 15. This shortening had appeared at least as early as Juvenal (vigiland iii. 232) and probably as early as Seneca, who uses vincend and lugend at the beginning of iambic lines (Troad. 264Google Scholar; Herc. Oet. 1862Google Scholar). I suspect that the loss of the ancient long quantities of vowels lies at the bottom of Servius' complaint about the line … cristaque tegit galea aurea rubra (Aen. ix. 50Google Scholar): ‘Duo ablativi sunt et duo nominativi, quos metrica ratione discernimus; nam rubra crista longae sunt ultimae, quia ablativi sunt casus. Sane huius modi versus pessimi sunt.’ Cf. [Sergii] Explanationes in Artem Donati (Keil, , iv. 522Google Scholar): ‘Syllabas natura longas difficile est scire; sed hanc ambiguitatem sola probant auctoritatis exempla, cum versus poetae scandere coeperis. Syllabas quae positione longae fiunt facile est dinoscere.’

³ So in Marius Victorinus' parallel list.

¹ Diomedes ascribes the invention of saturnians to Naevius.

¹ The ratio is 36:7.

² Cf. p. 136.

³ Cf. p. 135.

⁴ Atilius Fortunatianus is exceedingly mild in stating ‘hic versus obscuras quibusdam videtur quia passim et sine cura eo homines utebantur’. The schoolmen must have felt as confused and helpless in this matter as Cicero did about the settarii of the Roman comic writers, which, he says (Orat. 184Google Scholar), ‘sic saepe sunt abiecti ut non numquam vix in eis numeras et versus intellegi possit’. Like him (ibid. 198) they felt that in verse ‘certa quaedam et definita lex est, quam sequi sit necesse’. But their trouble was that the law which they had framed on a basis of quantity proved to be a dead letter.

¹ Their modern followers, e.g. Leo (Der saturnische Vers, 1905Google Scholar), use this line as a bed of Procrustes (as Lindsay terms it) on which they lay the scores of variant lines and proceed to hack and wrack them. The length to which Leo is forced to go may be illustrated by his insistence (p. 56) on putting the ictus even on an enclitic, in the hemistich ibidemque vir summus, and thus equating this syllable metrically with the second syllable of dabunt. To mate quantitatively with malum dabunt Metelli Naevio poetae such lines as quamde mare saevom vires quoi sunt magnae or nam diva monetas filia docuit or duonorum optumo fuise viro would have struck a Greek metrician aghast. One could understand a poet's occasional lapse, but not a system built on lapses.

² The closest present two variants. ‘Is hie situs quei nunquam victus est virtutei’ (v. Buecheler, Carm. Lat. Epigr. 9. 4)Google Scholar, has two irrational feet, but neither one is placed where Terentianus Maurus obviously thought this permissible, i.e. at the beginning of the line. So ‘Parens timens heic vovit voto hoc solut[o’ (ibid. 4. 2), if one accepts the hiatus voto hoc, presents two variants. The following present three each: ‘consol censor aidilis hic fuet a[pud vos’ (ibid. 6. 4), with the old quantities censōr and fuēt; ‘consol censor aidilis quei fuit apud vos’ (ibid. 7. 4), with similar quantities; ‘utei sesed lubent[es be]ne iouent optantis’ (ibid. 2. 6); ‘de]cuma facta poloucta leibereis lubetes’ (ibid. 4. 3). ‘Quei apice insigne Dial[is fl]aminis gesistei’ (ibid. 8. 1) would also present three variants from the ‘typical’ line if one accepts the two hiatus and understands insigne=insignem; Buecheler, however, insists that it is the neuter. One might also include here the line ‘honc oino ploirume cosentiont R[omane’ (ibid. 6. 1) if it did not violate what seems to be a fundamental, viz. a clean word-break between the hemistichs. Lines showing four variants also occur: ‘bene rem geras et valeas, dormias sine qura’ (ibid. 11. 3); ‘hospes gratum est quom apud meas restitistei seedes’ (ibid. 11. 2), with elision of quom and synizesis in meas; ‘subigit omne Loucanam opsidesque abdoucit’ (ibid. 7. 6), if one can stomach subigit; ‘facile facteis superases gloriam maiorum’ (ibid. 8. 5), with the same difficulty about facile. Finally, a line of five variants occurs once (if the text is sound): ‘annos gnatus (viginti) is [div]eis m[an]datus’ (ibid. 9. 5). The rest of the lines, amounting to some 42½, in these pre-Bassian inscriptional Saturnians of Buecheler, Carm. Lat. Epigr. 1–13Google Scholar, simply do not conform; they present still more serious anomalies in one or other or both hemistichs. (I include the Carmen Arvale as amounting to 4 lines and 2 hemistichs. It was cut in a.d. 218, but, although partly unintelligible, doubtless represents a form that had been in use for centuries, and so may fairly be included among pre-Bassian documents. This view is supported by the case of the Carmen Saliare; cf. Varro, , L.L. vii. 2Google Scholar: ‘Aelii … interpretationem carminum Saliorum videbis et exili littera expedita〈m〉, et praeterita obscura multa;’ Hor. Epist. 11. i. 86 f.Google Scholar: ‘iam Saliare Numae carmen qui laudat et illud quod mecum ignorat solus volt scire videri;’ Quint. Inst. Orat. i. vi. 40 f.Google Scholar: ‘… ab ultimis et iam oblitteratis repetita temporibus, qualia sunt … Saliorum carmina vix sacerdotibus suis satis intellecta. Sed illa mutari vetat religio, et consecratis utendum est.’)

³ Cf. Varro, , R.R. 1. ii. 27Google Scholar: ‘Ego tui memini, medere meis pedibus, terra pestem teneto, salus hic maneto in meis pedibus.’ Here, as in the Carmen Anale (partly), to some extent in the elogium of Mummius (Buecheler, , Carm-Lat. Epigr. 3)Google Scholar, and perhaps the epitaph of Eurysaces (ibid. 13), the hemistich appears as the unit.

¹ Cf., e.g., Sturtevant, , The Pronunciation of Greek and Latin, pp. 192–205.Google Scholar

² Ibid., pp. 206–18.

³ Aristonicus observed that the line (Il. xxiii. 644)Google Scholar: ἔργων τοιούτων ἑμὲ δὲ χρὴ γήραι λυγρ could be read as an iambic trimeter. For the parallel situation in Latin cf. Rufinus, , Commentarium in Metra TerentianaGoogle Scholar (Keil, , vi. 555)Google Scholar: ‘Bassus ad Neronem de iambico sic dicit: “lambicus autem, cum pedes etiam dactylici generis adsumat, desinit iambicus videri nisi percussione ita moderaveris ut cum pedem supplodes iambum ferias; ideoque illa loca percussionis non recipiunt alium quam iambum et ei parem tribrachyn, aut alterius exhibuerint metri speciem. Quod dico exemple faciam illustrius. Est in Eunucho Terentii statini in prima pagina hic versus trimetras:

Exclusit, revocat. Redeam? Non si me obsecret.

Hunc incipe ferire; videberis heroum habere inter manus. Ad summam paucis syllabis in postremo mutatis totus erit herous:

Exclusit, revocat. Redeam? Non si mea fiat. Ponam dubium secundo loco pedem, quo propius accedam:

Heros Atrides, caelitum tester fidem.”’

There are plenty of lines in Plautus and Terence whose opening can be distinguished from that of a dactylic hexameter only by the ictus.

⁴ In speech heard at a distance, frequently only the strongly accented syllables are audible.

¹ Mer. of Ven. 11. viiGoogle Scholar, sub init.

² Even in the iambics and trochaics of the early dramatists the accent frequently overrode the quantities in the same way as in English verse. Cf. Lindsay, , The Captivi of Plautus (1900), introd. pp. 30 ff.Google Scholar on the ‘Law of Breves Breviantes’. It is not, however, a case of lengthening under the ictus, as postulated by Leo.

³ Even Lindsay, who in Am. Jour, of Philol. xiv (1893), pp. 139–70, 305–34Google Scholar, gives in my judgement the best account of this old metre, is sometimes inclined to over-stress certain ‘laws’, which he then frequently discovers to be disobeyed.

⁴ Hardie, W. R., Res Metrica (1920), p. 197Google Scholar, properly asks: ‘Which theory will best account for the greatest number of the extant lines?’ He then enters the remarkable complaint (p. 198) that ‘the accentual view explained too many: i.e. it did not account for them convincingly, it was too complaisant or flexible in its scheme’, and abandoning his criterion proceeds to travel the quantitative road, but, like Andromache, ἐντροπαλιζόμενος (cf. p. 203: ‘The quantitative scheme, it must be admitted, requires many licences to make it workable; so many that its claim to be a real scheme becomes rather doubtful’; ibid., p. 205: ‘There are numerous other variations, some slight, some perplexing in the extreme’; ibid., p. 206: ‘Against them must be set the most perplexing variations of all, when the second part of the Une assumes so attenuated a form

parisuma fuit

fuisse virum

that one is tempted to resume the accentual theory and suppose that the principle of the second part was simply that there must be accents in it’). I should say that for ‘complaisance’ and ‘flexibility’ the accentual theory is no match for the quantitative; according to Leo's account the Saturnian shows between 80 and 90 quantitative forms of the first hemistich alone.

¹ Cf. Varro, , L.L. vii. 36Google Scholar; Cic. Brut. 71, 75 f.Google Scholar; Orat. 157, 171Google Scholar; Div. i. 114Google Scholar; Quint. Inst. Oral, ix, iv. 115.Google Scholar

² Brut. 76Google Scholar: ‘Sit Ennius sane, ut est certe, perfectior … luculente quidem scripserunt [of Naevius], etiam si minus quam tu [= Ennius] polite.’

³ XXVII. xxxvii. 13.

⁴ Epist. 11. i. 156–60:Google Scholar

‘Graecia capta ferum victorem cepit et artes

intulit agresti Latio; sic horridus ille

defluxit numerus Saturnius et grave virus

munditiae pepulere; sed in longum tarnen aevom

manserunt hodieque manent vestigia ruris.’

⁵ I suggest, in passing, that if ‘malum dabunt Metelli Naevio poetae’ had been ‘typical’ Horace would never have exhibited his shudder of distaste; this line is as unblemished rhythmically as any of those found in his own lyrics.

⁶ This and the other passages cited from Livy can be read as Saturnians without the orgy of emendation which Baehrens employs (Fragm. Poet. Rom., pp. 35, 294 f.)Google Scholar in turning them into hexameters. Lindsay appears to me unnecessarily diffident about accepting some of them, committing himself only to the second half of No. 8 below, and a few lines each from Nos. i and 13. The verses are rough and irregular, one freely grants; but this is what one would expect from the testimony of Virgil, Livy, Horace, and the later grammarians, and what one is forced to acknowledge in the only reliable evidence that we have, the pre-Bassian inscriptions. Would anyone doubt that if good fortune should provide us with an additional score of Saturnian inscriptions, the variety in forms would be materially increased? I would not insist that Livy's text is an exact copy of his originals at all points, but I do not believe that it differs from them to any great extent.

¹ That these were meant to be taken as verse is established not only on intrinsic grounds but also by the use of incised horizontal lines to indicate verse endings in No. 7 and by the application of the word carmen by Cic. De Sen. 61 to the similar elogium of Calatinus.

² Established as Saturnian by the testimony of Varro and Cicero referred to above in p. 141, n. 1.

³ That these prophecies were in verse is vouched for by Liv. loc. cit. 2–5, 8, 11, and that the verse was Saturnian appears from Cic. Div. i. 114 f.Google Scholar: ‘Eodem enim modo multa a vaticinantibus saepe praedicta sunt neque solum verbis sed etiam “versibus quos ohm Fauni vatesque canebant”. Similiter Marcius et Publicius vates cecinisse dicuntur.’

¹ See p. 136.

² From which the ‘scriptor incertus’ (cf. p. 136)Google Scholar cites the first line as coming ‘ex Regilli tabula’.

³ Cic. Pro Arch. 27: ‘Decimus quidem Brutus, summus vir et imperator, Acci, amicissimi sui, carminibus templorum ac monumentorum aditus exomavit suorum.’Google Scholar

Schol. Bob. ad loc.: ‘Hic Brutus Gallaecus fuit cognomento ob res in Hispania non minus strenue quam feliciter gestas. Eius etiam nomine 〈Accii〉 poetae tragici exstat liber cuius plurimos versus quos Saturnios appellaverunt vestibulo templi Martis superscripsit Brutus.’

¹ We should not fail to remember also the prominent role of accent in the verse of the third- and second-century b.c. playwrights, or the general harmony of ictus and accent in Cicero's clausulae (cf. Zielinski, , Das Clauselgesetz in Ciceros Reden).Google Scholar

² For this custom cf. Liv. IV. xx. 2 (of b.c. 437): ‘in eum milites carmina incondita aequantes eum Romulo canere;’ ibid. liii. n (of 410 b.c.): ‘alternis inconditi versus militari licentia iactati quibus consul increpitus, Meneni celebre nomen laudibus fuit;’ Plin. N.H. xix. 144Google Scholar: ‘nec non olus quoque silvestre est triumpho divi Tuli carminibus praecipue iocisque militaribus celebratum; alternis quippe versibus exprobravere lapsana se vixisse apud Dyrrachium, praeniiorum parsimoniam cavillantes.’

³ Is it without significance that the trochaic tetrameter is not used at all by the Augustan poets?

⁴ For others, cf. Vell. Paterc, 11. lxvii. 4Google Scholar; Suet. Div. lul. 49, 51, 80Google Scholar; Calig. 6.Google Scholar

¹ See p. 133, n. 2, No. (2).

² See p. 134, n. 2, No. (9).