Silensis and Aferesis in the Vita Mathildis: How Donizo’s Marginalia Explain the Battle of Tricontai (1091/1092)

* An early version of the paper was presented at the International Medieval Congress, University of Leeds, in 2015. My special thanks go to David Townsend and John Ward for their very useful feedback regarding my interpretation of Donizo’s poem. Any errors in what follows are entirely mine. My thanks also to Sarah Wilk.

Introduction

The year 2015 marked the 900th anniversary of the death of Matilda of Tuscany as well as the rough date of the completion of the principal source for her life and achievements, the poet Donizo of Canossa’s Vita Mathildis. Unfortunately, none of the poem’s critical editions adequately address the marginal notes that appear opposite six challenging passages in the authoritative manuscript, Vat. lat. 4922. [1] L. A. Muratori refrained from noting these marginalia in the edition he produced for the first series of the Rerum Italicarum Scriptores in 1724 (republished by Migne in the Patrologia Latina in 1853). [2] Ludwig Bethmann transcribed only three of the six in the edition he made for the Monumenta Germaniae Historica in 1856. [3] Luigi Simeoni mentioned none of the marginalia when he wrote what has since become the standard edition of the poem for the second series of the Rerum Italicarum Scriptores in 1931 (published 1940). [4] In fact, Simeoni declined to consider the marginalia even in his extensive footnotes, where he speculated on the meanings of passages to which they explicitly refer. The facsimile edition that Paolo Golinelli published in 1984 reproduces the marginalia as they appear in Vat. lat. 4922, but his corresponding Latin transcription – as well as the facing Italian and German translations – disregard them. [5] Other translation editions exhibit equal disinterest in the marginalia [6].

These six neglected notes are nevertheless crucial to understanding Donizo’s poem – not least because the hand that wrote them appears to be that of Donizo himself. The text, context, and codicological features of Vat. lat. 4922 provide ample evidence that Donizo was directly involved in producing this manuscript. [7] To begin with, it has an unimpeachable provenance: Vat. lat. 4922 is the lavishly illustrated presentation codex from which all extant copies descend. [8] The script is also informative, since it is the product of two hands: the body was written confidently in dark black ink; but it was frequently corrected by a less regular though coeval hand writing in weak brown ink. On fol. 80v, this latter hand even supplies two missing lines that the first hand omitted entirely. [9] The corrections then appear to be Donizo’s autograph. Further evidence of Donizo’s involvement is provided by the brief panegyric that the poet appended to the poem shortly after Matilda’s death in 1115. In this addendum, Donizo detailed the measures he took to prepare the final manuscript for her, noting that he was finishing the indices when he learned she had died. [10] This epilogue in fact remains bound to the rest of Vat. lat. 4922 in a final quire, which bears codicological features distinct from those of the rest of the manuscript (such as having 24 lines per page as opposed to 19). [11] Medieval scribes also acknowledged Vat. lat. 4922’s authority, and at least one noted the marginalia as well. In 1234, when the monks of Frassinoro [12] wanted to make a reliable witness of the text, they borrowed Vat. lat. 4922 from Donizo’s monastery of Sant’Apollonio at Canossa. [13] This careful copy – now the second oldest manuscript of Donizo’s poem extant– preserves all six of Vat. lat. 4922’s marginalia. [14] The monks then returned the original to Canossa, where it resided at least until the fourteenth century, when the copy now preserved in Reggio Emilia was made directly from it [15].

What could Donizo have been hoping to accomplish by intervening in the margins of his own presentation codex? He was not simply correcting errors, because he and the scribe fixed those by erasure or interlinear note; after all, Donizo had access to the text before it was completed. Nor was he making mundane, lexical glosses, because again, interlinear notes sufficed for those. The words written in the side margins are therefore something else. These six notabilia are in fact signposts calling the reader’s attention to figures of speech (i.e. «silensis», «aferesis»), and they serve as Donizo’s personal map for navigating the poem’s literary flourishes. These rhetorical tropes chart the author’s true course through some of the enigmatic verses for which the Vita Mathildis has now become infamous [Ghirardini 1987, 37-57] [16].

The significance of the marginalia can be proven by focusing on Donizo’s otherwise vexing description of the Battle of Tricontai (1091/2). The scansion of these lines, their medieval prose epitomes, and the military maneuvers and topography they describe make sense only in light of the marginalia. Collectively, these factors compel us to reject the most recent translations of these verses and to revise our understanding of the course and outcome of the battle. They enable us to appreciate the way in which Donizo deployed the marginalia to light the path for his readers through some of the most obscure passages of his poem.

The Battle of Tricontai

Donizo’s account of the Battle of Tricontai has long vexed editors and translators alike. The passage in question occurs during Donizo’s description of King Henry IV’s second military expedition to Italy (1090-1097), which marked an important moment in the Investiture Wars. After his famous humiliation before Pope Gregory VII and Matilda at the countess’s castle of Canossa in 1077, [17] Henry had returned to Italy with an army in 1081, spending the next three years attacking the reformers. By 1084, he had taken most of Rome, installed his own appointee (Archbishop Guibert of Ravenna) as Pope Clement III, and been invested with the imperial crown. Soon after, Gregory himself fled the city, only to die in exile in Salerno the following year.

In the wake of the Emperor’s apparently triumphal withdrawal to Germany in 1084, however, Matilda and her allies were able to rally the reformers. The countess won a significant military victory over Henry’s Lombard allies at the Battle of Sorbara later in 1084 [Hay 2008, 95-117]. In the years that followed, Matilda worked tirelessly to install reformist candidates in the dioceses of central and northern Italy, and helped to arrange the elections of two successive reforming popes (Victor III and Urban II). By 1090, [18] she had also solidified the reformers’ wavering military alliance by marrying the young son and heir of Duke Welf IV of Bavaria. Welf IV had previously led the south German opposition to Henry and was the founder of the German branch of the house of Este, whose name would eventually become synonymous with the papal party itself in Italy (i.e. Guelph). The alliance between the dynasties of Welf and Canossa thus represented a threat the emperor had to confront: each possessed extensive lands astride the main paths to Rome and a pedigree of rebellion against the crown. And so, in the spring of 1090, Henry entered Italy in force for a second time. [19] His main target now was not the pope but Matilda, who had proven the reform party’s staunchest defender.

The emperor first invested Mantua, the chief city of the house of Canossa on the Po plain. After nearly a year of vigorous siege, Henry succeeded in capturing it, together with much of the surrounding region, although a few of Matilda’s other garrisons north of the river held out. [20] The winter of 1091/2 marked a seasonal lull in the fighting, as the combatants were momentarily separated by the waters of the Po. While the emperor consolidated his gains north of the river and rested his troops in preparation for a spring push into the heart of Matilda’s territory, the countess fortified her positions south of the Po and invited Henry to overextend himself [21].

This relative pause was interrupted when the countess learned that the emperor had dismissed the bulk of his forces and crossed the Adige River. At its closest point, the Adige bends to within about 40 kilometres of Mantua, to the city’s north and east. [22] Donizo describes how Matilda resolved to seize the moment:

Nam postquam magna fuit arx minervia capta,
Tempore rex hiemis athesis trans flumen abivit, [570]
Principibus vacuus tunc, militibus neque fultus.
Hoc didicit nempe prudens comitissa, repente
Mille viros fortes numero plus, iussit ad hostem
Pergere bellandum, campi certamine tantum.

For after the great fortress Minervia was captured, [23]
During the winter the king crossed the River Adige,
Free of the princes, and not supported by knights.
The wise Countess learned this with certainty;
She suddenly ordered brave men, more than a thousand in number,
To go to fight the enemy, but only in the open field.

Knowing the location of Minervia would help fix the Emperor’s position before the battle, since Donizo states that Henry crossed the Adige after capturing it. Was the emperor carelessly advancing into disputed territory such as the southern part of the county of Mantua, or was he withdrawing for the winter into the friendlier regions to the east of the Adige, in the counties of Verona, Vicenza, and Padua? [24] Unfortunately, the identification of the Minervia has been a matter of dispute at least since the eighteenth century, when Muratori observed that there are candidates on both sides of the Adige. Eads argues that Manerbio (just south of Brescia) and Manerba del Garda (on the eponymous lake) are the most likely locations [Eads 2010, 36-37]. These are both to the north of Mantua and west of the Adige. Another piece of evidence that should be considered, however, is the Epitome Canossiana, a prose summary of the Vita Mathildis that appears with Donizo’s poem in the Reggio Codex, written in the early fourteenth century. The Epitome states that Matilda sent her thousand brave men to recover Minervia specifically («ad recuperationem minervii»), and that she directed them towards the site of the fortress («et eos dirigeret ad locum illum»). [25] This might give further support to the identification of Minervia with Manerbe, four kilometers to the east of the Adige. As Eads observes, Manerbe is «suggestively close to Tricontai», being only a few kilometers from the known site of the battle [Eads 2010, 36]. The Epitome Canossiana is a questionable witness, however. Its summary of the Vita Mathildis is both careless and garbled, and, as Eads notes, there is no corroborating evidence of a castle, much less a «great» one, at Minerbe [Eads 2010, 36]. The most that can be said—but indeed the crucial point to make, as will be discussed below—is that all the potential locations for Minervia are north of the Po.

Donizo’s phrase «campi certamine tantum» is also somewhat enigmatic, but it might make more sense if his «tantum» is read as tantumodo («only» or «merely»), as most Italian translators have interpreted it. By the phrase, Donizo is indicating that Matilda cautioned her soldiers to avoid any fortified enemy positions even as she ordered the strike against the emperor. Her caution made sense because on the strategic level, Henry had a numerical advantage and was steadily gaining ground; the countess could not allow her forces to get pinned down in what was rapidly becoming enemy territory. Her troops would risk disaster if they stopped to lay siege to castles or walled towns [Hay 2008, 131-132], and similarly if they attempted to force their way across defended bridges or fords such as the main crossing of the Adige at Legnano (which was then held by the imperialist bishop of Verona). [26] Matilda therefore instructed her soldiers to attack only if they could catch the enemy in the open field – ideally by surprise, as they had done with great success at Sorbara in 1084.

The next sentence is the most problematic one. It explains why Henry was able to recall his troops and surprise Matilda’s forces on the battlefield. A brief survey of how editors and translators have approached this passage in the years since Bethmann produced his justifiably admired edition in 1856 reveals the significant challenges the line has posed. The full passage reads thus in Vat. Lat. 4922 and its earliest medieval copies:

Cumque padi latices, athesis necnon vada sissent, [575]
Insidiis plenus rex, ipsos octo diebus
Vitavit, tardans proprias revocando phalanges [27].

Bethmann modernized the capitalization and removed some of Donizo’s punctuation, but otherwise transcribed the line in essentially the same way:

Cumque Padi latices Athesis necnon vada sissent, [575]
Insidiis plenus rex ipsos octo diebus
Vitavit, tardans proprias revocando phalanges [28].

The first Italian translations of this passage followed Bethmann’s edition. They took the subject of line 575 to be Matilda’s troops (who were mentioned in the previous lines), and they read the verb «sissent» as «transissent»: since her soldiers «crossed» the waters of the Po and indeed the fords of the Adige, the king was forced to resort to tricks to avoid them, and he did so for eight days until he was able to recall his army. This was for example the essence of the translations by Francesco Davoli in 1888 [29] and Natale Grimaldi in 1928 [30].

A few years later, however, Simeoni produced a new Latin edition, which offered a rather different transcription and punctuation, and which was furnished with footnotes in Italian that explicitly rejected Grimaldi’s translation. Simeoni’s Latin read:

Cumque Padi latices, Athesis, non vada sissent, [575]
Insidiis plenus rex, ipsos octo diebus
Vitavit tardans, proprias revocando phalanges [31].

Simeoni’s editorializing was significant because it reversed the meaning of the passage: it rendered the rivers impassable, at least for the eight days before the battle. The change consisted of several parts. The first was new punctuation: Simeoni added a comma after Athesis to challenge any association between «fords» and «Adige». To him, the «latices» («waters») of the Po and the Adige were the verb’s subject, and the «vada» («fords») its direct object. Incidentally, he also shifted the comma that appears after «Vitavit» in line 577 to appear after «tardans». Next was a new interpretation of the verb «sissent». Whereas Davoli and Grimaldi had translated it as «[tran]sissent» (the pluperfect subjunctive of «to cross»), Simeoni read it as the pluperfect subjunctive of sino («to allow»), in the syncopated form «sissent» (for «sivissent»). [32] Simeoni indicated this by adding a somewhat ambiguously worded note to the line that reads, «Athesis…. vada sissent, avendo permesso i guadi, non: ebbero passato». [33] Finally, Simeoni made perhaps his most radical alteration by changing line 575’s intensifier into a negative: he rendered Donizo’s «necnon» as a simple «non». In Simeoni’s edition of the line, the waters of the Po and the Adige denied Matilda’s forces any crossing, and their impassibility was what provided the emperor eight days to recall his own troops.

The editions and translations published after Simeoni’s edition have generally accepted his reading, which has become the standard. Later editors and translators have nevertheless sometimes made further alterations that seem designed to fit Donizo’s text into Simeoni’s sense. In the introduction to their Latin verse and facing Italian prose edition, Bellocchi and Marzi mentioned that they had to take measures to correct some errors they found in Simeoni’s text. [34] They thus returned the comma Simeoni had moved in line 577 to its rightful place after «Vitavit». [35] Nevertheless, Bellocchi and Marzi rendered line 575 in the same way Simeoni had, retaining his editorial comma after «Athesis» and truncating Donizo’s «necnon» into a simple «non»: «Cumque Padi latices, Athesis, non vada sissent». Bellocchi and Marzi also gave an Italian translation of the passage that is similar to Simeoni’s reading. [36]

The most recent attempt to make critical sense of the line was that of Paolo Golinelli. The lavish suite of editions he introduced in 1984 included a facsimile of Vat. lat. 4922, a Latin transcription, and facing Italian and German translations (the latter by Axel Janeck). [37] While the facsimile reproduced the passage exactly as it appears in the manuscript, the Latin transcription offered still another version of the line. Golinelli changed the comma Simeoni inserted after «Athesis» into the enclitic «-que» and separated Donizo’s «nec» from his «non»:

Cumque Padi latices, Athesisque nec non vada sissent, [575]
Insidiis plenus rex, ipsos octo diebus
Vitavit, tardans proprias revocando phalanges [38].

Despite these changes, the German and Italian translations based upon Golinelli’s transcription remained very much in the mold of Simeoni [39].

These alterations testify to the ongoing difficulties modern readers have experienced in understanding the text and meaning of this passage, at least since the publication of Simeoni’s edition, and they also raise some thorny interpretive questions. There are however easier and better solutions to the problems line 575 presents—solutions that allow the line to revert to the form Donizo originally intended it to have. To understand these, one must appreciate the full range of metrical, military, and paleographical complications the recent editors’ changes to the passage entail, as well as consider the context offered by the poem’s medieval prose epitomes. But most of all, one must read Donizo’s text together with the marginalia that appear in Vat. lat. 4922, because the marginalia are the author’s guides to traversing precisely these sorts of passages.

Meter

The first issue that must be considered is scansion. A significant problem the recent editorial alterations create is that they disrupt the poem’s meter. [40] The disruption is all the more problematic because Donizo seems to have given line 575 his personal imprimatur: in Vat. lat. 4922, the corrector’s hand has amended the scribe’s initial «Cunq;» to the proper «Cumq;» (for «Cumque»). [41] The fact that the corrector checked this line quite carefully obliges modern editors to do everything possible to make good sense of it before they alter it.

Yet, the only version of the line that scans properly into heroic hexameter is the one found in Vat. lat. 4922. It actually scans quite easily:

Cūm qŭe Pă/dī lătĭ/cēs, ║ Ăthĕ/sīs nēc/nōn vădă / sīssēnt, [42]

The changes Simeoni and Golinelli introduce, by contrast, spoil the line’s scansion. Simeoni’s removal of Donizo’s «nec» creates an erroneous fifth foot with two short syllables:

Cūmqŭe Pă/dī lătĭ/cēs, ║ Ăthĕ/sīs, nōn / vădă / sīssēnt,

Golinelli’s addition of a «que» after «Athesis» merely shifts the problem to the fourth foot, which becomes an unacceptable cretic (long-short-long):

Cūmqŭe Pă/dī lătĭ/cēs, ║ Ăthĕ/sīsqŭe nēc / nōn vădă / sīssēnt.

The argument from scansion alone is not conclusive, however, for Donizo is known to have made metrical errors. Bellocchi and Marzi observe that l. 900 of the first book has only five feet, while l. 1128 of the second book has seven. [43] Errors of meter occur even in the lines that Donizo appears to have corrected personally: the same hand that changed «Cunq;» to «Cumq;» in II.575 has also corrected the word «sumptuque» in I.900, apparently without realizing that the corrected line 900 still lacks an entire foot. [44] While much of the modern criticism of Donizo’s poem is well deserved, then, his metrical errors are nevertheless rare, at least when considered in absolute terms; the vast majority of his lines scan correctly. Readers should therefore amend his text only with the greatest caution.

The Location of Tricontai

There are corroborating reasons to prefer the original manuscript’s reading as well. One of these is the location of the Battle of Tricontai. As noted above, recent treatments of line 575 render both the Po and the Adige impassable for at least eight days. Yet we know that Matilda’s troops did in the end cross both rivers, for in the verses that follow line 575 [45], Donizo picks up the story: after avoiding Matilda’s forces for eight days, the emperor was finally able to reassemble an army. Aided by a traitor within the Canossan host, who assured Matilda’s soldiers that the king was still unprepared to fight them, Henry was able to launch a surprise attack of his own at a place Donizo calls «Tres Comitatus» [l. 585]. Caught unaware in the open field, Matilda’s army broke, with the scattered survivors retreating in dejection to Canossa.

While the location of Minervia might be contestable, that of «Tres Comitatus» is not. Since at least the time of Muratori’s edition of the Vita Mathildis in 1724, it has been identified as Tricontai in the county of Padua. [46] A recent article by Valerie Eads elaborates that Tricontai was not a village or castle but a more vaguely defined area bordering some woods just to the east of the Adige. Its name came from the fact that it was the place where the three counties of Verona, Vicenza, and Padua met [Eads 2010, 34 and n. 47]. Eads points out that Donizo himself calls «Tres Comitatus» a «pagus» [II.6, l. 585], while contemporary charters of the Este dynasty (which ruled the region) refer to it as a «locus» or «terra» [Eads 2010, 34 and 38]. Thus a charter written in nearby Montagnana in the year 1100 records a donation of lands held «in comitatu Pataviensi et Veronensi et inter Tres Comitatus et silvam Caracedi». [47] The location of Tricontai is confirmed by the participation of prominent members of the Este line in the conflict. Donizo states that the traitor within the Canossan ranks was Hugh of Maine. [48] This Hugh can be identified as younger son of Adalberto Azzo II (Azzo d’Este, d. 1097), the Margrave of Milan, Count of Padua, and founder of the House of Este. Not coincidentally, Hugh was also (through Azzo) the half-uncle of Matilda’s new husband Welf (the future Duke Welf V of Bavaria). Moreover, Matilda’s diplomas record that she had just endowed Welf with lands in the nearby county of Mantua (in 1089 or 1090). [49] Matilda’s young husband himself might even have been present in the Canossan host at Tricontai, together with his treacherous half-uncle Hugh. [50] Hugh, for his part, is recorded swearing fealty to his older brother Fulk (yet another of Azzo’s sons) in the same area in 1095 [51].

The location of Donizo’s «Tres Comitatus» is thus secure: it is immediately east of the Adige, and well north of the Po. This presents an obstacle to reading line 575 as Simeoni and the more recent Italian translators do, however. For if both rivers remained impassable, why did the emperor need to resort to «tricks» [«Insidiis»], and indeed, how did a battle occur at all? Whether Minervia was to the east or the west of the Adige, and whether the Emperor had exposed himself by advancing carelessly into the county of Mantua or by withdrawing carelessly into the county of Padua, Matilda’s forces in the end had to cross not one but both rivers to get to him at Tricontai. The line does not seem to be saying that the rivers were insurmountable; it is saying something else.

«Sissent»

A further complication with Simeoni’s edition is that it fails to explain why previous translators had read «sissent» as «[tran]sissent». This omission in a way sets its own trap for those seeking the truth about Tricontai. Eads noted that Simeoni had removed line 575’s «nec», which had appeared in all previous editions of the poem, although she found the line remained «obscure, even when corrected» [Eads 2010, 35 n. 49]. Prompted by a marginal note in Bethmann’s edition, Eads independently tried to make sense of the verb by «construing sissent as a compound of eo with the prefix truncated (s-issent)» [Eads 2010, 35 n. 49]. Eads’ article did not specify which prefix she had in mind, but in a personal communication to the present author she explained she was thinking of trans- or pertrans-. She based this astute conjecture on the Epitome Polironese, a prose summary of the Vita Mathildis from the fifteenth century, which supplies the missing prefix: «Qui cum Padum et Athicem pertransissent, eos diebus octo imperator sua astutia evitavit». [52] Eads therefore translated line 575 into English in a manner similar to the way Davoli and Grimaldi had rendered it into Italian: «So, they crossed the fords, the waters of the Po and of the Adige; the crafty king evaded them for eight days, delaying and recalling his own troops» [Eads 2010, 35].

Such a conjecture might seem tenuous until one examines the manuscript of Vat. lat. 4922 directly. In the margin of fol. 63r, in the same less assured hand and weak brown ink that historians have long suspected to be Donizo’s, appears the word «aferesis». This literary term was well known to late antique and medieval grammarians (including the authoritative Donatus): from the Greek aphaeresis («to take away»), it means the removal of the beginning of a word. [53] In the Vatican codex, [54] two additional, identical marks («..») in the same ink – one to the left of the marginal «aferesis» and the other above «sissent» – confirm the pairing of the two words. Although editors and translators from Simeoni onwards have ignored this marginal note, Bethmann had transcribed it in his 1856 edition; indeed, it was the presence of Bethmann’s note that first prompted Eads to posit the removal of the prefix. [55] Eads was on the right track then; the only mistake she made was in assuming the marginal note was Bethmann’s. In fact it was Donizo’s.

A translation of the passage that takes the marginalia into due account might therefore read:

And since [they] crossed the waters of the Po, and indeed the fords of the Adige,
The emperor, full of tricks, avoided them for eight days,
Delaying while recalling his battalions.

The other marginalia in Vat. lat. 4922 confirm this reading. The vast majority of the second hand’s numerous interventions in the manuscript are simple corrections by erasure or interlinear glosses. Thus it writes «ut» above «quo» [39v and 46v], «scutum» over «parmam» [34r], and «padus» over «Eridanum» [51r]. On 44v the same hand clarifies the antecedent for a «quos», on 34v it corrects an entire line, while on 80v it supplies two additional lines that the scribe had omitted entirely. The marginalia of the sort one finds on 63r, however, are much fewer, and they serve a different purpose. In fact there are only six such marginalia in the manuscript (including the one opposite line 575), and each of them is instructive.

Two of the five remaining marginalia (viz. foll. 63v and 74v) contain another word: «silensis». One might be tempted to read this as «silentis» (the genitive of silens, «silent»), but the context and usage, as well as Donizo’s documented infatuation with Greek words, clarify that what he meant is «syllepsis». [56] From the Greek for «taking together», syllepsis was variously defined by ancient grammarians, but generally it included figures of speech that united different clauses by a single word that pertained, strictly speaking, only to one of those clauses. Donatus provided an illustration from the Aeneid in which Vergil allowed a singular and a plural subject to take the same (singular) verb: «hic illius arma, / Hic currus fuit». [57] Medieval grammarians added further cases of syllepsis occurring in sense as well as in the written word, and sometimes defined it more generally as the confusion of singular and plural. Thus Julian of Toledo (642-690) explained that it was syllepsis when Vergil described the Trojan horse being filled «with armed soldier» («armato milite»), since of course Vergil meant many soldiers rather than one. [58] Bede added a Christian example:

Syllepsis also occurs in sense, that is, when one is given for many or many for one.
One for many, as in: “He sent the dog-fly against them and it consumed them; the frog,
and it destroyed them,” even though he did not send one fly or frog to destroy the
Egyptians, but innumerable ones [59].

The first marginal «silensis» in Vat. lat. 4922 appears on fol. 63v, next to a passage describing the outcome of the Battle of Tricontai. The text of the poem depicts the Canossan soldiers being ambushed by the emperor’s reassembled army. [60] The corrector’s hand has added a marginal «silensis» beside one of these lines, with paired marks that point to the verb «capitur»:

A turba regis miles capitur sed inhermis. [583]
Plures evadunt, fugiunt per devia saltus.

But [only] the unarmed knight is captured by the throng of the king.
More escape, they flee through the lonely places of the forest [61].

The marginal note explains that the author is using a figure of speech: that the one «miles» here stands for many [62].

Why would Donizo feel this needed to be explained? The figure of speech has to be identified because a few lines later, when recounting how Matilda’s soldiers eventually regrouped and retreated to Canossa, Donizo describes the captives in the plural:

Tandem collecti, remeaverunt retro, mesti
De sociis captis parvis pariter quoque magnis. [590]
Inter quos captus manfredus erat probus altus
Filius alberti, super ipsum contio regis
Maxime gaudebat, comitissae contio flebat.

Finally gathered together, they returned back, grieving
For their captured comrades great and small alike;
Amongst whom was captured the valiant Manfred,
Noble son of Albert, about whom the army of the king
Rejoiced greatly, the army of the Countess wept.

The full passage, understood together with the marginalia, thus indicates that at the Battle of Tricontai, the Canossan soldiers (plural) who were not able to arm themselves or mount up in time were captured, while the remainder fled. The more recent editions of the text, by contrast, do not explain this very well. A footnote to the line in Simeoni’s edition argues that more than one soldier had to have been captured—here again Simeoni criticizes Grimaldi’s more literal translation—but without indicating the presence of Donizo’s marginal note that explains to the reader that the captives were indeed plural [63].

The other instance of «silensis» occurs several chapters later, when Donizo is discussing a different subject: namely, the capture of St. Bernard degli Uberti during the revolt in Parma in 1104. Bernard was the abbot of Vallombrosa, papal legate, and Matilda’s candidate for the recently vacated bishopric of the city. On arrival in Parma, he began celebrating mass, but his audience considered his reformist sermon insulting to the emperor. Donizo describes how the citizens interrupted the ceremony and dragged Bernard outside the cathedral. The text on Vat. lat. 4922, fol. 74v, reads,

Iniecitque manum super ipsum civis avarus,
Et foris asportant, heu missa relinquitur orta, [1000]

And the grasping citizen laid hand upon him,
And they carry [him] outside. Alas the mass once begun is abandoned!

Here again, a marginal note guides the reader through the passage. Directly to the left of line 999, the corrector’s hand has written, «silensis id est singularis pro plurali» [64]. Unfortunately, the hand has not left any visible marks to refer the reader to a particular word in the line (though the scribe of the Lucca manuscript settles the issue by moving the «silensis» to directly above «Iniecitque»). [65] The note in Vat. lat. 4922 does at least clarify that this is a syllepsis of sense: the singular «citizen» here stands for multiple citizens (and perhaps a single «hand» for multiple hands as well), just as earlier, in Donizo’s description of the Battle of Tricontai noted above, one «knight» stood for many knights. This explains why the verb «asportant» in the following line is plural [66].

Modern editors and translators have not always fully appreciated the intent of the marginalia accompanying this line. While Bethmann transcribed the marginal note to line 999, [67] and Davoli’s translation made the avaricious citizen plural, [68] later editors and translators generally declined to reckon with Donizo’s syllepsis. Following Muratori, Simeoni looked instead to the Epitome Parmense’s prose adaptation of the line for clarification of the meaning; he suggested that a single citizen had both seized Bernard and dragged him outside. [69] Golinelli’s edition presented further challenges to both sense and meter by incorrectly rendering the punctuation mark that follows «asportant» as an «i» [70].

These two instances of syllepsis are nevertheless keys to understanding Vat. lat. 4922’s marginalia. Above all, they prove that the marginalia are not corrections. Corrections in this manuscript have been made by erasure and interlinear note; apheresis and syllepsis, by contrast, did not need to be corrected because they are not errors per se. They could be tolerated, and even considered laudable, especially in poetry. Donatus had suggested as much when he had defined syllepsis as a figure of speech and listed it amongst other rhetorical virtues; he had made the point explicit in the case of apheresis, which he described as a change in spelling (metaplasm) that could occur for the sake of either meter or ornament. [71] Commentators on Donatus’s work such as Servius explained the theory in still more detail: what distinguished virtue from vice in these cases was intent. Performed unknowingly, a departure from grammatical norms was to be avoided; performed knowingly, the same departure could be commended. [72] The marginalia in Vat. lat. 4922 thus broadcast Donizo’s intent. They signal not just that he is using a figure of speech, but that he is consciously using a figure of speech – a fact that is worth mentioning in the margins even of a presentation codex. His marginalia then are not corrections, but literary signposts that chart the true path through his poem’s rhetorical landscape. When these marginalia are neglected, that path becomes obscured.

Each of the remaining three marginalia in Vat. lat. 4922 points out a further instance of aferesis. One occurs in the first chapter of the first book of the Vita Mathildis (fol. 13r), which deals with the heroic deeds of Matilda’s ancestors. The passage describes how Queen Adelaide of Lombardy escaped from Berengar of Ivrea in the year 951. Seeking a refuge, Adelaide instructed her faithful chaplain Martin to petition the bishop of Reggio for help. Martin cautiously approached the bishop, who inquired about the status of the queen:

Presul ut agnovit, de regina rogat, o quid [186]
Est factum? quaeso michi narra, discere spero.

The bishop realized this: he asks, «O what happened
To the queen? Tell me, I beseech you, I hope to learn».

The lines are admittedly awkward, in Donizo’s signature fashion, but the pertinent word for present purposes is «rogat». Paired marks beside it and the «aferesis» in the margin of fol. 13r confirm that a prefix has been excised from this verb as well. In the Vita Mathildis as a whole, the only prefix used before any form of the verb rogare is «inter-», and it only appears once. It is found in a parallel construction sixteen lines later in the same chapter, in which Matilda’s great-grandfather Adalberto Azzo asks Martin about the fate of Queen Adelaide in a very similar fashion:

Verbo quem largo princeps interrogat Atto [203]
Quid de regina factum sit, quo [73] sibi dicat.

Prince Atto asks him at length
What happened to the queen, that he might tell him.

The distinction between rogat and interrogat might seem inconsequential, which may be why Bethmann declined to transcribe this marginal note. Donizo however might have wanted to distinguish rhetorical questioning from interrogation, especially since the chaplain is hiding something. For Martin initially lies — first to the bishop and later to Azzo — by saying that Adelaide is dead, in order to verify his questioners’ sincerity. Only after Martin sees their tears flow does he tell them the truth and reveal the plight of the fugitive queen [Vita Mathildis II.1, ll. 186-208].

Bethmann either failed to notice or simply disregarded the remaining two marginal notes in Vat. lat. 4922, and other modern editors have done the same. They are admittedly faint and easy to overlook. But the careful copyist who produced the manuscript for the monks of Frassinoro (now Lucca 2508) in 1234 noticed them, and an examination of both affirms that they too are rhetorical signposts.

Each of the two involves another case of aferesis. The first appears in the prologue, line 53, as Donizo is extolling Matilda’s good qualities. [74] The passage begins with an enumeration of the cardinal and theological virtues, after which the scribe has written,

Ista legit totas probitatis iure coronas.

Donizo however has written «aferesis est» in the margin, indicating that the verb «legit» has been subjected to aferesis. The verb intended is obviously «collegit», because – as in Donizo’s other cases of aferesis – the full verb can easily be found in the nearby lines: it appears in the future tense («colliget») only three lines earlier (l. 50). In light of the marginal note, the line becomes much simpler and easier to understand than it sometimes appears in modern translations. Donizo is merely saying this:

She justly collected all crowns of righteousness.

The same verb is also the subject of the last of Donizo’s marginalia, which appears on fol. 16r of the Vatican manuscript. The setting of this passage is that Otto the Great has marched into Italy and defeated the rebel king of Italy, Berengar of Ivrea. The scribe writes:

Gens Alemanna capit Berengerium, superatis
Tunc Longobardis, campum legit Otto, retraxit
Illico frena retro, gaudens de rege retento.

Donizo’s very faint marginal note indicates that the line contains another case of aferesis. Once again, the verb «collegit» has been shortened to «legit» in order to fit the metre:

The German nation captures Berengar,
Then, with the Lombards overcome, Otto broke camp,
He pulled his bridles back from there, glad to have captured the king [75].

Conclusions

What light then do Donizo’s neglected notations shed on the Battle of Tricontai? Some details of the encounter must remain obscure, but others finally come into focus when one considers all the manuscript’s marginalia, together with battle’s physical and human geography, as well as the poem’s prosody and prose epitomes. Collectively, these indicate that whichever way Henry traversed the Adige River, Matilda’s forces crossed both it and the Po to get to him. This is the point Donizo is making with his crucial «necnon», and the point that is unfortunately lost in the editions of the poem that truncate this word and neglect the marginalia. Matilda’s soldiers chased the Emperor across the Po «and even» the Adige. This also explains why Matilda’s troops were so nervous that they had to be reassured by a comrade native to the region – who unfortunately for them turned out to be a traitor – that the king did not have the means to meet them there in battle. Traversing the fords of the Adige was more dangerous than crossing the wider waters of the Po because, especially after the fall of Minervia, the Adige essentially marked the natural limit of Matilda’s military power. To soldiers from Canossa, it was the Rubicon.

Vat. lat. 4922’s marginalia both merit and reward careful attention. In addition to explaining the two armies’ maneuvers leading up to the battle, they render other passages in the poem more intelligible as well, revealing for example that Matilda’s forces suffered more than a token casualty at the Battle of Tricontai, and explaining the precise circumstances of Bernard’s ejection from the cathedral of Parma. Perhaps most importantly, the marginalia prove that Donizo was more aware of his poem’s obscurities than the standard edition of his text suggests. Appreciation of this fact may not rehabilitate his reputation as a poet, but those who ignore the beacons he left behind to guide us can hardly blame him when they become lost in the depths of his rhetoric. Donizo left us these marginalia in order to light our path through that darkness.

Manuscripts

• Lucca, Biblioteca Statale (olim Governativa), MS 2508.
• Mantua, Biblioteca Comunale, MS 243 (B. IV. 17).
• Reggio Emilia, Biblioteca Comunale “Panizzi”, MS Turri E 52.
• Vatican City, Biblioteca Apostolica Vaticana, MS Vat. lat. 4922.

Primary Sources and Translations

• Bede, De schematibus et tropis, Kendall C. B. (ed.), Corpus Christianorum, Series Latina 123A, Turnhout: Brepols, 1975, 142-171.
• Die Urkunden Heinrichs IV., Gladiss D. von, Gawlik A. (eds.), Monumenta Historiae Germanica, Diplomata 6: 1-3, Weimar and Hannover, 1941-78.
• Die Urkunden und Briefe der Markgräfin Mathilde von Tuszien, Goez E., Goez W. (eds.), Monumenta Germaniae Historica, Laienfürsten- und Dynastenurkunden der Kaiserzeit II, Hannover: Hahnsche, 1998.
• Donatus A., Ars maior: Donat et la tradition de l’enseignement grammatical: Étude sur l’Ars Donati et sa diffusion (ive-ixe siècle) et édition critique, Holtz L. (ed.), Paris: Centre National de la Recherche Scientifique, 1981.
• Donizone, Vita Mathildis comitissae, celeberrimae principis Italiae, Rerum Italicarum Scriptores V, Muratori L. A. (ed.), Milan: Ex typographia societatis palatinae in regia curia, 1724, 335-397.
• Donizone, Vita Mathildis comitissae, celeberrimae principis Italiae, Patrologiae cursus completus, Series Latina CXLVIII, coll. 939-1036, Migne J. P. (ed.), Paris: Migne, 1853.
• Donizone, Vita Mathildis, Monumenta Germaniae Historica, Scriptores XII, 348-409, Bethmann L. (ed.), Hannover: Hahnsche, 1856.
• Donizone, Vita della grande contessa Matilde di Canossa, Davoli F. (trans.), Reggio Emilia: Stabilimento tipografico e litografico degli Artigianelli, 1888.
• Donizone, Donizone, Il Cantore di Matilde e dei Principi Canusini, Grimaldi N. (trans.), Reggio Emilia: Bizzocchi, 1928.
• Donizone, Vita Mathildis celeberrimae principis Italiae: Carmine scripta a Donizone presbytero, Simeoni L. (ed.), Rerum Italicarum Scriptores, nuova edizione 5.2, Bologna: Nicola Zanichelli, 1940.
• Donizone, Matilde e Canossa: Il Poema di Donizone, Bellocchi U., Marzi G. (eds.), Modena: Aedes Muratoriana, 1970.
• Donizone, La Vita di Matilde di Canossa, Codice Vaticano latino 4922, Paolo Golinelli (ed. and trans.), Milan: Jaca Book (Italian edition); Golinelli P. (ed.), Janeck A. (trans.), Zürich: Belser Verlag (German edition), 1984.
• Donizone, Vita di Matilde di Canossa, Golinelli P. (ed. and trans.), Milan: Jaca Book, 2008.
• Gloria A. (ed.), Codice Diplomatico Padovano: dal secolo sesto a tutto l’undecimo, Monumenti storici, Deputazione veneta di storia patria, ser. 1, Documenti, no. 2, Venice, 1877.
• Julian of Toledo, Ars Iuliani Toletani episcopi: Una gramática Latina de la España Visigoda, ed. M. Maestre Yenes, Toledo: Instituto Provincial de Investigaciones y Estudios Toledanos, 1973.
• Livy, Titi Livi ab urbe condita libri, Weissenborn W. and Müller M. (eds.), Stuttgart: Teubner, 1966.
• Orderic Vitalis, The Ecclesiastical History of Orderic Vitalis, Marjorie Chibnall (ed. and trans.), Oxford: Clarendon Press, 1969.
• Servius: Maurus Servius Honoratus, In Donati artem maiorem, in Grammatici Latini, Heinrich Keil (ed.), Leipzig: Teubner, 1855-1880, vol. 4, 421-448.
• The Fourth Grammatical Treatise, Clunies Ross M., Wellendorf J. (eds.), London: Viking Society for Northern Research, 2014.
• Vergil: Publius Vergilius Maro, Bucolics, Aeneid, and Georgics, J. B. Greenough (ed.), Boston: Ginn and Co., 1900.

Secondary Sources

• Eads V. 2010, The Last Italian Expedition of Henry IV: Re-reading the Vita Mathildis of Donizone of Canossa «Journal of Medieval Military History», VIII: 23-68
• Ghirardini L. L. 1987, La voce immortale di Canossa: Studio critico sul celebre monaco poeta Donizone, Modena: Aedes Muratoriana.
• Hay D. J. 2008, The Military Leadership of Matilda of Tuscany, 1046-1115, Manchester: Manchester University Press.
• Rippe G. 2003, Padoue et son contado (Xe-XIIIe siècle), Rome: École Française de Rome.
• Riversi E. 2013, La memoria di Canossa: Saggi di contestualizzazione della Vita Mathildis di Donizone, Pisa: Edizioni ETS.
• Schwartz G. 1993, Die Besetzung der Bistümer Reichsitaliens unter den sächsischen und salischen Kaisern mit den Listen der Bischöfe, 951-1122, Leipzig, 1913; reprint Spoleto: Fondazione Centro italiano di studi sull’alto medioevo.
• Vignodelli G. 2011, Il Filo a piombo: Il Perpendiculum di Attone di Vercelli e la storia politica del regno italico, Spoleto: Fondazione Centro italiano di studi sull’alto Medioevo.
• Vignodelli G. 2016, Politics, prophecy and satire: Atto of Vercelli’s Polipticum quod appellatur perpendiculum «Early Medieval Europe», 24 (2): 210-235.