Modern knowledge of the grammarian Consentius’ De barbarismis et metaplasmis, a work valuable for the study of the Latin language,Footnote ¹ dates back to a relatively recent past: it was only in 1817 that its editio princeps was published by Ph.C. Buttmann,Footnote ² just a few years after the legal scholar A.W. Cramer came across a mention of the then unknown treatise in a ninth-century MS in the Bayerische Staatsbibliothek of Munich, numbered Clm 14666.Footnote ³ Based on this solitary manuscript, H. Keil published the short treatise in the fifth volume of his Grammatici Latini.Footnote ⁴ With no little enthusiasm did W.M. Lindsay announce his unearthing of what, in his own words, had ‘long been a “desideratum”, a second authority’ for this text, in the MS F 15 III d at the Universitätsbibliothek Basel;Footnote ⁵ this was followed by E.O. Winstedt's complete collationFootnote ⁶ and M. Niedermann's critical edition.Footnote ⁷ After about a century now there comes to light a third authority, surprisingly enough in a codex which has enjoyed such fame in the past decades that one might wonder how Consentius could have gone unnoticed in it for so long: this is the eleventh-century MS of Venice, Biblioteca Nazionale Marciana Lat. Z. 497 (= 1811),Footnote ⁸ in which the De barbarismis et metaplasmis is contained on fols. 84^vb 39 - 90^va 39;Footnote ⁹ moreover, a so-far-unnoticed quotation from it (32.9-14), together with one from Consentius’ De nomine et uerbo (Consent. gramm. V 353.6-7), is found on fol. 41^vb 16–21 of the same manuscript in a famous grammatical florilegium. The codex, written in Romanesque minuscule and probably originating in Rome, is regarded as a handbook of liberal arts designed by Lawrence Archbishop of Amalfi, formerly a monk at Montecassino, thereafter a teacher in Florence and Rome, where he died in about 1049. Based on palaeographical evidence, F.L. Newton rightfully assumed as an exemplar for this codex a MS in Beneventan script, as some features can be detected that betray the scribal imitation of that typical South Italian script, namely the use of the distinctive abbreviation for eius as ‘ei in ligature with stroke through the descender of the i’,Footnote ¹⁰ the Beneventan ti ligature for the assibilated sound, and the 2-shaped Beneventan interrogation sign,Footnote ¹¹ to which I would add the typical abbreviation for in as a long i cut by a horizontal strokeFootnote ¹² and the confusion of a and t.Footnote ¹³ Interestingly enough, none of these features is found on fols. 66–95, those containing the new Consentius: from a codicological point of view, this is an autonomous section, written by a different scribe from the rest of the MS and preserving some grammatical texts generally attributed to insular authors, such as Smaragdus’ Liber in partibus Donati (fols. 66–81^vb) and part of the compilation entitled Pauca de barbarismo Footnote ¹⁴ (fols. 81^vb - 84^vb), which precedes the De barbarismis et metaplasmis; not surprisingly, the new text of Consentius displays numerous features of the Insular script, such as the symbols for enim, autem, eius, est, nihil and et. On this basis it is most likely that this whole section was never included in the Beneventan exemplar, but was added at the time and place of copying of our MS in order to enrich the grammatical content.Footnote ¹⁵

One cannot say that the scribes did a very accurate job in copying this text over time: there are many mistakes, a fair number of them apparently owing to the misunderstanding of abbreviations, such as post instead of per (Consent. 7.17 and 8.19) or the various outcomes of the Insular symbol for inter, which was at times omittedFootnote ¹⁶ (dum for interdum at 13.17, est for interest at 29.22), at times wrongly interpreted (et for inter at 9.14, 10.22, 27.21, 31.11, 31.17, 31.21; uel for inter at 19.10). Spelling too betrays an Insular influence, yet to a lesser extent than in the other two MSS of the De barbarismis et metaplasmis:Footnote ¹⁷ confusions of i and e happen (i for e in difinitionem at 2.17, in diffinitionibus at 22.20, 29.6, etc.; e for i in incedat at 13.21, in operet at 4.2, etc.) as well as the replacement of u by o (motatio at 8.22, motatus at 4.17, motet at 8.25, etc.), nor is the doubling and shortening of s exceptional (dissyllabum at 7.1, trissyllabum at 13.8, etc.; expresioni at 15.17, moloso at 13.23, pasus at 8.3, etc.). Quite an interesting point is the occasional transliteration into the Greek alphabet of some technical words of Greek origin (like tmesin on p. 9.4, epenthesis on p. 20.17, syncope on p. 20.22, etc.): the same is not found in the other MSS and could possibly go back to a very early stage of the tradition.

That having been said, the stemmatic analysis puts the Venice MS in a good position that is of great help for the constitutio textus, for it demonstrably depends upon neither of the extant MSS.

First of all we are allowed to set at the top of the tradition an archetype, which I shall call , for there are errors common to all three MSS, of which I select some more remarkable ones:Footnote ¹⁸ at 1.11-12 the MSS readFootnote ¹⁹ haec definitio separat metaplasmum eo quod firmum dixit in communi sermone, where firmum clearly is a corruption of fieri barbarismum (as the editors conjecture); at 2.3-5 the manuscript tradition has barbarismus enim secundum definitiones superiores uel accentibus uel litteris uel syllabis uel temporibus uel adspirationibus peccat, where uel accentibus is clearly misplaced, as it never takes the first position in such lists (indeed it is placed after syllabis by Niedermann); at 5.9 the MSS read apocope est litterae ut Achilli pro Achillis, where a definition is wanted before the example, so that the editors rightly state a lacuna after apocope est; similarly at 25.21–26.1, where the MSS have ipsi iam scripturam corruptam relinquunt, ut est ‘relliquias Danaum’ et ‘tanton me crimine dignum duxisti’. addidit enim unam litteram l per metaplasmum, item contra Lucilius ‘ore corupto’, after contra there lies a lacuna in which the grammarian must have remarked on the removal of the e in tantone.

Next there are plenty of errors common to B and V, but separating them from M, of which I only list some noteworthy omissions: at 7.12-14, item si dicat aliquis ‘neutrum’ disyllabum, quod trisyllabum fere enuntiamus, barbarismum faciet, BV omit barbarismum faciet; at 8.13-15, antithesis est cum pro littera alia littera ponitur ut mutatio fiat, numerus autem syllabarum non desit, BV omit autem syllabarum; at 16.22, haec sunt exempla de his tribus litteris, BV omit exempla; at 24.18-20, item defenditur ‘ponitur spes sibi quisque’, cur ‘te’ breuis sit, cum ‘te’ longam duae consonae efficere debuerint, ut ‘unde spissa coma’, they omit ut ‘unde spissa coma’; at 24.1-3, sicut utique breuis est ‘o mihi praeteritos referat si Iupiter annos’, they omit o mihi praeteritos referat si, which seems relevant to the discussion; at 25.11, qui uero ultra hoc rationes defensionum quaerit, BV omit ultra hoc; at 28.11-12 BV omit euenit dubitatio in ut ecce de una uocali, quae syllaba sit, euenit dubitatio. B being older than V, it obviously cannot be derived from it, nor is V derived from B, as it avoids some of its errors: at 2.5 B omits uel temporibus, which MV have; at 12.19-21, ut barbarismi similiter fieri posse uideantur, ueluti est diaeresis et episynaliphe, the correct episynaliphe is replaced by sistole in B alone, which happens also at 13.5, nonne uidetur per episynaliphen barbarismum facere, where B alone has sistolen; at 22.5-6, peccant hi non in regula, non in prolatione, sed in sensu, MV have sed in sensu, which is omitted by B. Therefore, B and V are dependent upon a hyparchetype.

That this hyparchetype is not M is self-evident, as this MS is more recent than B, besides having errors avoided by BV, such as the omission of insitum at 16.7 (et re uera insitum alterutrum uitium quibusdam gentibus est), the omission of a nobis at 17.17-18 (haec in his litteris a nobis adnotata sunt), or the omission by saut du même au même of barbarismum quidem per adiectionem litterae, soloecismum at 19.24-6 (dicitur et barbarismum facere et soloecismum. barbarismum quidem per adiectionem litterae, soloecismum per artis regulas).

This analysis gives us the following stemma:

V will thus prove of great help in doubtful textual choices when it agrees with M, allowing us to tell what probably was in the archetype and thereby helping us have more a balanced approach than Niedermann's sometimes overenthusiastic preference for the then recently unearthed B. In fact, this often makes us revert to Keil's reading:Footnote ²⁰ at 3.1-2 MV's nunc oportunum et tempestiuum est ut, quae sit inter barbarismum et metaplasmum distantia, praeloquamur (so Keil) is better than B's inter barbarismum et metaplasmum nunc oportunum et tempestiuum est ut, quae sit distantia, praeloquamur (so Niedermann); at 4.2 MV's faciet barbarismum suggests reverting to Keil's reading against barbarismum faciet, which was chosen by Niedermann on account of B; at 4.20-1, syncope est cum mediae parti orationis littera syllabaue subtrahitur, the agreement of M and V favours Keil's orationis against B's and Niedermann's dictionis; at 7.19-20, which Niedermann printed as hoc, sicut dixi, aut ipse poeta ita iam scriptum reliquit aut nobis subtrahendum tradit, M's and Keil's tradidit is now confirmed by V; at 8.12, quoniam quorundam longiorum indigent praeceptorum (so Niedermann following B), the agreement of M and V on the word order longiorum quorundam (printed by Keil) makes it more likely; at 8.13-14, antithesis est cum pro littera alia littera ponitur (so Keil), pro littera alia littera is backed by the agreement of M and V against littera pro littera which Niedermann chose on the account of B; at 10.14-15, simile est, cum dicit aliquis ‘lapidicinas’ (so Niedermann), B's aliquis is now less likely in view of MV's quis (so Keil); at 12.18-19, where Keil had et uicissim aliquos modos de metaplasmis, Niedermann followed B in reading alios instead of aliquos, but now V supports M and Keil; at 15.21, in a discussion about the sound of initial i, M's and Keil's ite is now confirmed by V against B's ire (chosen by Niedermann); at 19.6–8, itaque dicunt, quia non secutura casuum dispectione statim repudietur … barbarismum eum fecisse (as printed by Keil), Niedermann accepted B's quod instead of M's quia, but the latter is now read in V as well; at 27.11, uelut expressas strangulatasque displodit (so Keil), MV's displodit is superior to B's diplodit (printed by Niedermann); at 28.9-10 the word order of MV, deinde euenit ea dubitatio, ut nescias … (so Keil), seems better than that chosen by Niedermann on the account of B, deinde ea dubitatio euenit, ut nescias …; at 28.21, nam potes scandere ‘femine arden-tem’ et est synaliphe (thus Keil), B has the future erit (so Niedermann), but M's est is confirmed by V. In other cases the reading of V helps us understand the direction of the corruption: at 8.13-14, antithesis est cum pro littera alia littera ponitur, ut mutatio fiat, numerus autem syllabarum non desit, the reading of M is fiat, that of V is fit and B has sit, whence we can argue that the true reading fiat was corrupted into fit in the hyparchetype of BV and that B has gone further down the path of error by altering fit into sit so as to have the subjunctive needed.

Some lectiones singulares of V against BM seem genuine: these may be inherited, in case B and M have coincidental corruption, or may have come about by conjecture. At times they confirm, or anticipate, the emendations of modern scholars: at 6.1, antiqua e cedro Italusque paterque Sabinus (a quotation from Verg. Aen. 7.178), V alone has Italusque, against the ametrical Italus of BM; at 8.5, pessime, ubi ei, quae tollitur, accedit et consonans, the correct reading accedit is found in V alone among the MSS, against accidit of BM; at 14.5-6 putat se nescio quis facere structuram ex trochaeo et ione apelassonos, qualis structura <est> ‘esse uideatur’, Niedermann, following Heraeus, has to supply <est> which V already features although before structura; at 30.4-5, once again in a quotation (Verg. Aen. 12.23), quale est ‘nec non aurumque animusque Latinost’, V has the correct aurumque against the ametrical aurum of BM; at 31.6-9, ergo his quattuor modis sic euenit, ut quicquid illud perit, uel inter duas uocales positum sit uel inter uocalem et consonantem aut inter consonantem <et uocalem> uel inter duas consonantes, the editors’ integration is confirmed by the reading of V. In many other cases V shows better orthography, as at 5.18, dicuntur autem ectasis et systole, where V alone has ectasis against iectasis (B) and iactasis (M), or at 7.5, ‘Orphei Calliopea’ (from Verg. Ecl. 4.57), where Calliopea is in V against B's Calliopia and M's Caleopea.

In short, the unearthing of a third authority for Consentius’ De barbarismis et metaplasmis may not be considered today as much of a desideratum as Lindsay considered the discovery of the second a century ago. However, our new witness must be regarded as a significant help in understanding the ways in which an influential grammarian has come down to us as well as in establishing his text.