UNCOVERING THE DOUAI FRAGMENT: COMPOSING POLYPHONY AND ENCODING A COMPOSER IN THE LATE FOURTEENTH CENTURY

Abstract

Douai, Bibliothèque Marceline Desbordes-Valmore, MS 1105/3 fragment 74.4/1 is a musical fragment from the second half of the fourteenth century, composed of two bifolios copied in black mensural notation. It not only contains a concordance for Multipliciter amando, a motet that, until now, has only survived in the famous Chantilly manuscript, but also four unica (another complete motet, two incomplete ones and a three-voice Gloria). After presenting a codicological overview of the fragment, I analyse distinctive features of each of the three complete pieces, based on the first modern transcription of the source. Special attention is given to the motet Ferre solet, which, by way of literary encoding mechanisms, reveals its composer’s name (Frater Johannes Vavassoris) and its precise date of composition (1373), making these four folios a crucial witness for understanding fourteenth-century musical culture.

Douai, Bibliothèque Marceline Desbordes-Valmore, MS 1105/3 fragment 74.4/1 (hereafter the ‘Douai fragment’) is composed of two parchment bifolios and contains five polyphonic pieces copied in black full mensural notation. ¹ Among these pieces, four are unica: two voice parts that are most likely the motetus (O martir beatissime, fol. 1^r) and the triplum (Salus honor, fol. 4^v) of two incomplete motets; a three-voice Gloria (fols. 2^v–3^r) with a long melismatic Amen; ² and a complete motet Ferre solet/Ana theos de gracia/Ave Maria (fols. 3^v–4^r). In addition, the Douai fragment contains the three-voice motet Multipliciter amando/Favore habundare/Letificat (fols. 1^v–2^r), which has a concordance in the Chantilly manuscript. ³ Among these pieces, Ferre solet stands out in particular because its text contains the name of a hitherto unknown composer and a date of composition. These invaluable pieces of information, hidden in the motet by way of an acrostic, transform the Douai fragment into a crucial witness for understanding fourteenth-century musical culture.

This article provides a preliminary analysis of the Douai fragment in order to reassess its significance for fourteenth-century music history. After a codicological overview of the source, I shall introduce the texts and music of the two incomplete motets. The second part of this article then focuses on the distinctive aspects of the three complete pieces. Transcriptions of all five pieces are presented in Appendix 1, and an edition and translation of the texts of the motets by Leofranc Holford-Strevens can be found in Appendix 2.

CODICOLOGICAL OVERVIEW

The fragment reproduced in Figure 1, available in colour online on DIAMM, ⁴ was discovered during the nineteenth century. The four folios were used in manuscript bindings, but it is not clear whether they were reused as binding materials for one or multiple books. In the middle of the nineteenth century, Edmond de Coussemaker mentions ‘two incomplete musical fragments from the fourteenth century being used as flyleaves in a manuscript from the Douai library [which] are written in black and red notation’. ⁵ He then describes red notes that were previously black void notes and mentions a canonic instruction dealing with the contratenor voice part. ⁶ It seems that Coussemaker refers to fol. 4^r of the current Douai fragment, which would indicate that in 1852 at least fol. 4^r was still in its host volume and was visible. ⁷ However, Coussemaker does not say what that host volume was. The fragment is mentioned again in the nineteenth century in a catalogue published in 1878 by Chrétien Deshaines: at that time, the four folios now constituting the Douai fragment had already been taken out of their former host manuscript(s). Here again, Deshaines does not make any reference to the host manuscript(s), explaining only briefly that they may still be preserved at the Douai library. ⁸ The host manuscript(s) nonetheless left traces: fols. 2 and 4 each have five sewing holes on their left-hand margins. In addition, four green-edged holes from copper nails are on the right-hand margin of fol. 2. Two similar holes are to be found on fol. 4 as well. Therefore, it seems that these two folios were reused in the same book bound on five bands and held by two clasps, though fol. 4 may have been trimmed too extensively to have been perforated by the second nail of each clasp. Furthermore fols. 1 and 3 have only four sewing holes and no apparent traces of clasps, which seems to indicate that they were bound in a different host manuscript. The result of leather offsets onto fols. 1^r, 2^r, 3^v and 4^v seems to indicate that these were the sides that were pasted down on the wooden board. More generally, however, there are no legible relevant marginalia and no signs of old library sigla that could aid in the identification of the host manuscripts(s). ⁹ Consequently, very little is known about the provenance of the fragment. Barbara Haggh has nonetheless suggested that it was either bought as part of a nineteenth-century collection of Douai sources or that it originates from a monastery or abbey in the Douai region, such as the Marchiennes Abbey. ¹⁰ Several clues in the source itself discussed below confirm a northern French or southern Low Countries origin.

Figure 1(a) Douai fragment 74.4/1, fol. 1^r, O martir beatissime

Figure 1(b) Douai fragment 74.4/1, fols. 1^v–2^r, Multipliciter amando

Figure 1(c) Douai fragment 74.4/1, fols. 2^v–3^r, Gloria

Figure 1(d) Douai fragment 74.4/1, fols. 3^v–4^r, Ferre solet

Figure 1(e) Douai fragment 74.4/1, fol. 4^v, Salus honor

The dimensions of each leaf are approximately 301 × 245 mm. These are not the original dimensions because the folios have been trimmed to fit the binding of the host manuscript(s). As a result, text and music are missing on the right-hand side of fols. 3^r and 4^r. However, rhythmic repetitions in the pieces on these folios confirm that rarely more than the length of two imperfect breves has been lost (more frequently only the length of one minim or one semibreve is missing), the implication being that the current dimensions of the written areas of the leaves may be close to their original size, though portions of the margins were doubtless lost in trimming. In addition, damage caused by rebinding (holes, cuts, brown-red marks left by leather offset) sometimes make a detailed reading of the content very difficult. However, overall, the parchment is relatively well preserved.

The folios are currently joined together with paper, which probably dates from their rediscovery in the nineteenth century. ¹¹ The disposition of the music indicates that the folios are contiguous, despite the lack of foliation. ¹² The parchment is very thin, and it is difficult to determine the hair and flesh sides. Given the fact that the binding of the folios is not original, it is currently not possible to assess with certainty whether they were separate adjacent leaves or two nested bifolios at a centre gathering.

Four ruling lines, in leadpoint, provide a frame to contain the writing block, though these have now faded at various points. Black ink, used for the text and the music, is generally very well preserved (fols. 1^r, 2^v, 3^r, 4^v), but it is seriously faded in some places (fols. 1^v, 2^r, 3^v). Red ink was used for proportional and mensural changes in the motet Ferre solet (fols. 3^v–4^r). The polyphonic Gloria features notable (though very simple) penwork initials: in all three voices, the letter ‘E’ (from ‘Et in terra pax’) is decorated with a small dragon-like creature. Similar initials (at least in terms of size) were perhaps intended for Ferre solet, since the indication of the first letter of the motetus voice is given in lower case before the musical notation begins (since the left margin of fol. 3^v has been cut, no possible initial for the triplum can be read). A larger penwork initial is drawn to indicate the ‘Tenor’ voice of the motet Multipliciter amando (fol. 2^r). The initials of the triplum and motetus of Multipliciter amando and of the voice part on fol. 4^v have minor and modestly decorated initials. Given the lack of space before the music and text of fol. 1^r were copied, it is highly unlikely that an initial was intended to be added.

Each folio has four small margins (most likely larger before the trimming), leaving a writing block of 260 × 205 mm in which all pieces are laid out in choirbook format. Folios are ruled with eleven five-line staves in black ink (now pale or dark brown), ¹³ which are approximately 16 mm in height. The consistency in the line-spacing suggests the use of a single-stave rastrum. ¹⁴ On fol. 1^v the scribe used the upper ruling as a sixth line of the stave. The tenor voice of the Gloria begins on fol. 2^v but, because of a lack of space, the music scribe added a twelfth stave in free-hand on fol. 3^r in order to accommodate the end of the voice (see Figure 2).

Figure 2. Layout of the Douai fragment. Diagonal lines indicate staves that were prepared but not filled with music. Names of voice parts not present in the source are bracketed. Grey areas and dashed lines mark possible reconstructions of the missing folios. Roman numerals in (c) refer to the bipartite musical division of the Gloria

In the two complete motets, the names of the lower voices are given (tenor and, where applicable, contratenor and solus tenor). None of the three voices of the Gloria are identified by name. With regard to the music on fols. 1^r and 4^v, Margaret Bent suggests that these voices formed part of two three-voice motets without contratenors. ¹⁵ It is interesting to note that they had different layouts: while fol. 4^v adopts a similar layout to Multipliciter amando (fols. 1^v–2^r), the potential triplum and tenor of O martir were most likely copied one after the other on the preceding (now missing) folio (see Figure 2(a)). Such inconsistency of mise-en-page is not unique to this fragment but can also be found in several other contemporaneous manuscripts, including Chantilly. ¹⁶

Text and music were closely coordinated in the process of copying. This is especially clear at line breaks: occasional vertical guide-lines improve note–syllable alignment, as in the Gloria and the incomplete motet Salus honor. On one occasion the scribe even extended the stave in free-hand to make sure that notes would align with the corresponding syllables of the text (see fol. 1^r, end of stave 5). This strongly suggests that, as usual, the text was copied before the music, as confirmed by the unnecessary blank space sometimes left after melismas (e.g. fol. 2^v, stave 6). ¹⁷ One hand copied both text and music, perhaps with the exception of the red ink on fols. 3^v–4^r. ¹⁸ Mensuration signs are used only in Multipliciter amando (fols. 1^v–2^r), but three other pieces feature mensuration changes: the three-voice Gloria, the incomplete motet O martir beatissime (fol. 1^r) and Ferre solet. As will be specified below, these changes in Ferre solet are indicated by coloration, while no indication is given for the changes in the Gloria and in O martir beatissime. Table 1 provides a summary of the contents.

Table 1. The contents of the Douai fragment

TWO INCOMPLETE MOTETS

O martir beatissime (fol. 1^r)

The text of the voice part on fol. 1^r (most likely the motetus of a three-voice motet) is in praise of St John the Baptist. It consists of three nine-line strophes followed by three five-line strophes.

The melody is divided into two rhythmic patterns repeated three times each. The first rhythmic pattern (talea A) uses perfect modus and imperfect tempus with minor prolation. The second rhythmic pattern (talea B) is in perfect tempus with minor prolation. While this change was empirically determined (it is not signalled by any mensuration sign or coloration and there is no notational confirmation, for instance by presence of perfect breves), it is possible that a verbal canon copied on the preceding, missing folio explained the mensuration to be used. In addition, the musical structure fits neatly with the poetic design, since each rhythmic pattern corresponds to a specific rhyme structure (the three neuvains fit the three A taleas and the three cinquains fit the three B taleas). ¹⁹ The change of mensuration between the A taleas and B taleas can also be understood as a 2:1 diminution, where the perfect long of A becomes the perfect breve of B and where the imperfect breve of A becomes the minor semibreve of B. ²⁰ Similarly, the strophic structure of the poem can be considered as being in diminution, where the last three cinquains are diminutions of the three neuvains. ²¹ Diminution as a compositional strategy is particularly relevant in a motet dedicated to John the Baptist because it reflects the Baptist’s ‘diminishing’ quality, as encapsulated in the following declaration: ‘[Christ] must increase, I must decrease.’ ²²

Salus honor (fol. 4^v)

This single voice is most likely the triplum of a three-voice motet. Its poem follows a consistent rhyme scheme: aabaab, bbcbbc, etc. The first letter of each three-line group spells out the name ‘Sanctus Lambertus’, that is, St Lambert of Maastricht, who comes to be praised in the poem. In the source, all the letters of the acrostic are capitalised. ²³ This suggests a northern French or Belgian provenance for at least this motet, as St Lambert was (and still is) the patron saint of Liège. ²⁴ This seems to confirm the provenance of the fragment as a whole, given its current preservation and the hypothesis made by Barbara Haggh that it comes from the Douai region.

This voice is in imperfect tempus with minor prolation and the rests covering three lines of the stave indicate perfect modus. The piece consists of an opening section, in which the rhythm of the first four longs is stated twice, leaving the last three longs rhythmically free. After that, the music is regularly divided into three taleas. One can suppose that the missing voices were built on the same rhythmic structure. The first strophe of the poem is sung during the opening section of music and the following strophes are evenly distributed between each talea (five strophes per talea).

GLORIA

The Gloria of the Douai fragment has three voices, all of which are texted. The spellings are strictly the same except on one occasion (see below, ‘ad dexteram’). Capitalisation is not always the same in all three voices, as can be seen in the following edition:

This mass setting has two striking features: a new interpolated trope and a pronounced rhythmic design. This section considers each in turn. ²⁵

A New Trope

The Gloria is furnished with a short trope (bb. 99–104): ²⁶

… tu solus Altissimus … you alone the Most High

salus indeficiens nostra our unfailing salvation,

tu nos pietate illustra enlighten us by your mercy,

Ihesu Christe cum sancto spiritu … Jesus Christ, with the Holy Spirit …

According to the current state of knowledge regarding Gloria tropes, this textual interpolation is unique to this polyphonic setting. ²⁷ Moreover, by the late fourteenth century, Gloria tropes are quite rare, with the most popular being the Marian trope Spiritus etalme. ²⁸ The fact that the trope of the Douai Gloria is versified and rhymed is still more unusual, though not unprecedented. ²⁹ It may be suggested that the text of this trope is evocative of the sequence for the first Sunday of Advent that opens similarly: ‘Salus aeterna, Indeficiens mundi vita’ (‘Salvation eternal, life of the world unfailing’). ³⁰ However, since Gloria tropes were not necessarily added to create a connection with the liturgy of the day, it is not possible to place it within the annual cycle of the liturgy. ³¹ The composer gave some aural emphasis to the trope: cadences surround its appearance (b. 99 and b. 104) and the second verse of the trope is set to a strict homorhythmic texture, a unique feature in the polyphonic setting of the piece.

Structuring Polyphony

The three-voice Gloria is comprised of an opening section (here designated ‘introitus’) and a further two parts. The introitus (twenty breves) and the first part (237 breves) set the text of the Gloria syllabically in imperfect tempus, while the second part is characterised by a long melismatic Amen (forty-seven breves) set in perfect tempus. ³² All parts use minor prolation. The Amen section is therefore audibly separated from the previous section by a change of mensuration; notably, it also comes to be distinguished by a clear cadence point that brings the first part of the Gloria to a close. This process of separating structurally a lengthy Amen section is a common compositional process in the fourteenth century. ³³ In addition, it is visually distinct because the three voices of the Amen are copied one after the other on fol. 3^r and not in the usual choirbook format of the fragment (see Figure 1(c) above). ³⁴

The Gloria from the Douai fragment opens with a passage which restates the first sentence of the text twice, setting it to two contrasting vocal duets (triplum/tenor and triplum/motetus respectively). ³⁵ Part I and Part II in turn are divided into four subsections each, easily identifiable by cadence points and rhythmic repetitions as shown in Table 2. Table 2 also offers an overview of textually defined musical breaks in analogous Glorias. ³⁶ It shows that the first two breaks are quite conventional (‘Laudamus te’ and ‘Domine fili unigenite’), but the last two are much rarer (‘Qui tollis … suscipe’ and ‘Tu solus altissimus’).

Table 2. Musical division of the Gloria with corresponding text openings

Furthermore, these analogous Glorias often contain several additional breaks throughout their musical setting, for example at ‘Gratias agimus’ (Glorias 39, 41, 43, 47), ‘Domine Deus Rex’ (Glorias 30, 33, 39), ‘Qui tollis … miserere’ (Glorias 30, 39, 40, 43, 47), ‘Qui sedes’ (Glorias 33, 34, 39, 40, 43, 48) and ‘Cum sancto’ (Glorias 30, 39, 43, 48). Interestingly, some of these, such as before the text ‘Domine Deus Rex’, ‘Qui tollis … miserere’ and ‘Cum sancto’, are also given aural importance in the Douai Gloria (see bb. 27, 55 and 111). Specifically, the musical settings at these points are all based on the same rhythmic pattern: hockets, preceded and followed by short melismatic passages characterised by rhythmic exchanges between the three voices (see Examples 1 and 2 below). As such, they mark out the end of each section identifed in Table 2. Additionally, Table 2 shows that the strategy of concluding this mass setting in a major mensuration is not unusual. However, it also illustrates that in spite of similarities in the musical division of the text of the Gloria, no systematic approach to dividing up the text in musical settings was applied by composers. These initial observations in turn invite broader comparison with fourteenth-century Glorias, though this lies beyond the scope of this article.

Example 1 Hockets in the Gloria (b. 28)

Example 2 Rhythmic exchanges in the Gloria (bb. 24–7)

Within the subsections of the Douai Gloria, five recurring units, or ‘ideas’, with discrete musical material can be discerned (see Examples 1–5). ³⁷ These musical ideas sometimes overlap with each other: for instance, Example 3 (conjunct melodic motion) also introduces a hocket idea (Example 1). These five ideas are featured interchangeably throughout the introitus and the four subsections of Part I. By contrast, in Part II each idea is used as a means to set the subsections apart from each other. For instance, in Part I, Example 1 (hockets) and Example 2 (rhythmic exchanges) are consistently used one after another at the end of each section, thus articulating the structure of the piece through musical repetition. In Part II, however, these ideas serve as unique motivic units that underpin the rhythmic construction of Section IIb and Section IIc respectively. The structural function of these musical ideas might have helped the composer to visualise and therefore to organise his composition in his mind, and the performer to remember it. ³⁸ The elaboration of rhythmic and melodic units to create a pronounced alternation of texture points towards a clear compositional strategy: repetitions within carefully conceived structures were a source of delight among listeners, performers and, most likely, composers as well. ³⁹ Therefore, in all likelihood, the rhythmic and structural repetitions of this Gloria must also have been shaped by aesthetic concerns.

Example 3 Conjunct melodic motion in the Gloria (tenor, bb. 112–19)

Example 4 Melodic sequences in the Gloria (triplum, bb. 85–9)

Example 5 Goups of minims in the Gloria repeating intervals of a second (bb. 8–9)

MULTIPLICITER AMANDO: CONCORDANCE AND TRANSMISSION

The isorhythmic motet Multipliciter amando (fols. 1^v–2^r) is the only piece in the Douai fragment that has a concordance: it also survives in Chantilly, in which it was a unicum until the Douai fragment was discovered. Based on its rhythmic complexity, especially in the upper voices, Ursula Günther has suggested that this motet was composed after the 1380s or in the 1390s. ⁴⁰ Despite its presence in the well-studied Chantilly manuscript, the motet in itself has never received much scholarly attention. ⁴¹ This might be due to the difficulty of understanding its textual content. Just like the motet Portio nature/Ida capillorum or the triplum of the motet Apta caro, the texts of Multipliciter amando seem to belong to a category of motets which cannot easily be translated. ⁴² Both the triplum and the motetus appear to deal with the power of love (both physical and emotional) to give comfort and pleasure. ⁴³ The tenor was drawn from the last three words of the chant Introibo ad altare Dei, originally for the Communion of Septuagesima: ‘letificat iuventutem meam’ (‘makes my youth joyful’). ⁴⁴ These words can be understood from both a secular and a sacred point of view and might therefore evoke tension between earthly and divine love, echoing the texts of the upper voices. ⁴⁵ While the version transmitted in the Douai fragment does not offer any further clarification of the meaning of the text, this witness nevertheless presents an opportunity for a more precise musical understanding of the motet by comparing it with Chantilly.

The most apparent variant between the Chantilly witness and Douai can be found in the use of mensuration signs in Douai to indicate shifts from imperfect tempus with major prolation to perfect tempus with minor prolation, while the scribe of Chantilly used coloration. Because of the use of two different systems, the shift from minor prolation to major prolation encountered at the end of each talea is indicated by black void notation in Douai and by red void notation in Chantilly. The use of mensuration signs instead of red notation can point to practical aspects of the copying process: for example, when discussing the use of red notation, the author of the Tractatus figurarum allows for the use of black void notation instead of red if red ink is not available. ⁴⁶ Here the use of mensuration signs is another hint that red ink was indeed not available at the moment of copying the Douai fragment. This hypothesis is also confirmed by the lack of red staves observed above and by the use of black void notation later filled in red in the lower voices of the motet Ferre solet. In addition, the first four mensuration signs in Douai are above or below the stave, whereas all the others are placed appropriately within the stave. This could be an additional hint that the music scribe was copying the motet from a version with red notation and realised while copying that mensuration signs were needed if the motet was being notated without coloration.

Turning back to the different versions of Douai and Chantilly, a note-to-note comparison reveals that the tenors of the two sources are rhythmically and melodically identical apart from four instances, ⁴⁷ and even the ligation of the melody is strictly the same. In contrast, a comparison of the triplum voices of Douai and Chantilly reveals forty-two variants (see Example 6 for a critical comparison of the two sources): ⁴⁸

• eighteen variants in pitch

• fourteen variants in rhythm

• two variants in pitch and rhythm at the same time

• four variants in written accidentals

• three missing notes in one of the two sources

• one extra note in one of the two sources

Example 6 Triplum of Multipliciter amando with variants in the versions of Chantilly and Douai

The comparison presented in Example 6 demonstrates that ten out of the fourteen rhythmic variants are errors in the Chantilly version, as is the extra note and two of the three missing notes. ⁴⁹ Günther and Harrison also identified these rhythmic errors when transcribing the triplum from Chantilly: ⁵⁰ Douai can now clarify their rhythmic emendations. Regarding pitch variants – the most common variant type – it has been demonstrated that Chantilly contains a number of errors and, given the results of the rhythmic comparison, it seems reasonable to assume that Douai offers a reading of the motet more musically consistent than Chantilly regarding the pitches as well. ⁵¹ Finally, a comparison of the texts also reveals significant differences: the texts themselves present some discrepancies between the two versions, though neither helps in clarifying the intended meaning (see Appendix 2). The discrepancies are either due to the difficulty of reading Chantilly (the ink has faded in some places) or to the mistakes in Chantilly. For instance, in her 1965 edition, Günther observed that the word ‘secam’ (line 7) is ‘unintelligible’. ⁵² The Douai variant helps to clarify the reading by clearly indicating ‘presertim’. Other difficult readings include the first word of line 18: Günther first transcribed it as ‘virus’, ⁵³ but changed it to ‘vibex’ or ‘nibex’ in her second edition. ⁵⁴ Harrison chose the reading ‘vibrus’. ⁵⁵ Douai quite clearly reads ‘nil’, which fits neatly with the poetic structure of the poem. ⁵⁶ The text underlay in Chantilly is highly problematic, as the very contrasting solutions provided by Günther’s and Harrison’s transcriptions illustrate. Nonetheless, the verses of the triplum are evenly distributed among the taleas (one strophe per talea). The scribe of Douai is very systematic in his alignment: each verse starts at the same location in the three taleas. ⁵⁷ He is also relatively precise in the alignment during melismatic passages – a point that resonates with my discussion above, which highlighted the extra attention he paid in the alignment of text and music. Such consistency has not been found in Chantilly. Therefore, it can be argued that musically and textually, Douai presents a more consistent reading of Multipliciter amando than Chantilly.

FERRE SOLET: COMPOSER AND COMPLEXITY

The Marian motet Ferre solet is by far the most complex piece of the Douai fragment, both in terms of its musical setting as well as its text. After presenting the overall structure of the piece, I shall discuss three of its most notable features: the solus tenor, the two Latin canonic instructions and the literary device which encodes a Marian prayer, the name of a hitherto unknown composer and a precise date of composition.

The tenor of the motet is drawn from the Ave Maria antiphon CAO 1539. It presents a few variants from the most standard versions of the antiphon, especially at the beginning: the antiphon usually starts on f, whereas this tenor begins on g. However, a comparison of resources on the Cantus Index database reveals some chant sources that also start on g. ⁵⁸ The melisma on ‘tecum’ at bars 51–9 of Ferre solet (g f g f ) is also very similar to the chant copied twice in Fribourg, Bibliothèque des Cordeliers, 2, which reads g f g. ⁵⁹ However, the list of chant melodies connected to this antiphon displayed in the Cantus Index presents many variants and it is therefore difficult to assess whether the tenor melody of Ferre solet is drawn from a specific context or if the composer intentionally changed the antiphon for compositional reasons. In any case, the Marian focus is very clear and plays a central role in the piece.

The motet is comprised of an introitus followed by two larger musical sections. The introitus is sung by the triplum and the motetus while the lower voices are silent. Each of the two larger sections encompasses one statement of the tenor color and the contratenor color. These sections are divided into three periods each, which are isorhythmic in all voices. Within each of these periods, the lower voices are composed of two segments: moving from one to the other, the tenor switches from black to red notation and the contratenor switches from red to black notation. The second section (over the second color) introduces new rhythmic ideas in the upper voices, again divided into three periods each. Furthermore, the tenor and contratenor, reading off their respective line of notation, take on a new form by halving the durations of the first segment of each period and doubling the durations of the second segment. As a result, the tenor and the contratenor periods have the same durational value in each of the two larger sections of the motet and switch between black and red notation at the same points. This manipulation by halving durational values also creates a symmetry in the duration of each segment across the first and second sections. ⁶⁰ The motetalso includes a fifth voice, a solus tenor, that works the same way as the contratenor (see Figure 3). ⁶¹

Figure 3. Overall structure of Ferre solet

Solus Tenor

The solus tenor of Ferre solet is labelled as such by the scribe at the bottom of fol. 4^r of the fragment. A solus tenor is a conflation of two lower parts in a motet, whose function is still the subject of scholarly debate. First considered by Heinrich Besseler as a reduction made after composition for performance in a reduced scoring, this hypothesis was then challenged by Margaret Bent, who considers the solus tenor primarily as an aid in the compositional process at a time when music was not visually aligned in score. ⁶² According to Bent, only motets with a grammatically essential contratenor would have needed a solus tenor, that is, a contratenor which shares an essential harmonic foundation with the tenor. In Ferre solet, while the contratenor appears mostly inessential in this sense, it nevertheless does provide a foundation for several unsupported fourths in the upper voices, thus qualifying it on a harmonic level as essential (see, for instance, bb. 133, 144, 148, 160 and 171). ⁶³ From these two essential lower voices, the composer of Ferre solet would have been able to produce a conflation of tenor and contratenor voices – a solus tenor – which must certainly have supported the composition of the upper parts.

On the other hand, the copying of a solus tenor greatly assists the performance of a four-voice piece as a three-voice piece, as it saves the performers the worry of devising a compositional solution of their own. In a complex piece like Ferre solet, having a written version of a solus tenor already prepared to replace the tenor–contratenor pair would have been of considerable help. Therefore, while Ferre solet does not necessarily allow for a more precise understanding of the function of solus tenors, it nonetheless adds to the repertoire of pieces that survived with this musical feature and may well promote new discussion in the future.

Two Latin Canonic Instructions

The lower voices of the motet are governed by different proportions and mensurations, as mentioned above (see Figure 3). Their structure can be summed up as follows:

• Tenor

  • Color I, in each of the three taleas:

  • black (perfect modus) at notated values (18 breves)

  • red (perfect modus) at half notated values (9 breves)

  • Color II, in each of the three taleas:

  • black (perfect modus) at half notated values (9 breves)

  • red (perfect modus) at notated values (18 breves)

• Contratenor

  • Color I, in each of the three taleas:

  • red (perfect modus) at half notated values (18 breves)

  • black (imperfect modus) at half notated values (9 breves)

  • Color II, in each of the three taleas:

  • red (perfect modus) at half their previous values (i.e. at a quarter of the notated values) (9 breves)

  • black (imperfect modus) at notated values (18 breves)

Two separate Latin canonic instructions offer explanations for the reading (and performance) of the two tenor and contratenor colores: the first verbal canon, copied at the end of the motetus (Figure 4), provides a solution for two iterations of the tenor color, whose melody is copied on fol. 3^v. The second canon, copied under the first line of the contratenor (Figure 5), presents rules for two iterations of the contratenor and solus tenor color (whose melodies are copied on fol. 4^r). The texts lack clarity due to the inconsistent state of preservation of fol. 4. ⁶⁴ In what follows, I provide my interpretation of how the instructions set out a solution for reading the lower voices of the motet.

Figure 4. First canonic instruction in Ferre solet (Douai fragment, fol. 4^r), end of stave 8

Figure 5. Second canonic instruction in Ferre solet (Douai fragment, fol. 4^r), below stave 9

The first canon (Figure 4) contains two different instructions. ⁶⁵ The first two lines deal with the poetic structure of the motetus, which is analysed below. The following four lines explain how to perform the tenor, whose melody is stated twice. Line 3 seems to belong with the beginning of line 4 (‘the colour and figure remaining/with the whole *** in both places’). They indicate that in both statements of the tenor melody, a reading must take into account the different ink colours (they must ‘remain’); that is to say, these colours have an effect on the proportions of this voice. It is likely that the trimmed end of line 4 (‘with the whole ***’) contained information about the modus, which must stay perfect in both statements of the tenor melody. The end of line 4 clearly explains that during the first statement the black notes must be read as they are written (‘the black notes are sung the first time as [written]’), that is, in integer valor. By contrast, ‘the others’ (i.e. the red notes) must be read at double the notated value (line 5: ‘the others are uttered at half value’), that is in duple proportion to the integer valor. It is the reverse during the second statement (line 6: ‘and learn that it is sung the other way round’): the black notes must be read at double the notated values and the red notes must be read as written.

The contratenor instruction is written below the contratenor on fol. 4^r (Figure 5). The first line clarifies the modus to apply to red and black notation and presents the solution for the first statement of the contratenor melody. The meaning of the text is somewhat unclear owing to the trimming of the folio on the right side and the faded ink. The first portion of the text seems to indicate that during the first statement of the melody, the contratenor must maintain a perfect modus when reading the red notes (i.e. they are ‘still under perfection’). ⁶⁶ However, exactly how the imperfect modus of the black notes and the duple proportion are to be applied to both the red and black notes of the contratenor is something that is not clearly expressed in the canon’s current state. That noted, the mention of ‘totaliter’ (‘completely’) may be a hint that both ink colours have the same proportional meaning for the entire first statement of the melody. ⁶⁷

The second line of the verbal canon deals with the second statement of the contratenor melody (‘But the second time’) and is much clearer: the beginning indicates that the red notes of the second statement must be read at half their previous values (‘at half from the first time’), i.e. a quarter of the notated value. The end of the second line indicates that in the second statement of the contratenor melody the black notes must be sung ‘as notated and coloured’, i.e. in integer valor.

Although some aspects of the text remain unclear, these types of verbal canon are not uncommon. In her dissertation ‘Verbal Canons and Notational Complexity in Fifteenth-Century Music’, Emily Zazulia proposes three main ways in which composers used verbal canons in the fourteenth and fifteenth centuries. Two of them concern notational issues, in particular the use of coloration and mensuration or proportion signs. The third type of canon was intended ‘to instruct singers to repeat a given line, often changing the way they read the music on subsequent repetitions’. ⁶⁸ The canons of Ferre solet belong to this third category: they indicate which proportions and mensurations the singer must use for the first and the second melodic statements of the tenor, contratenor and solus tenor. ⁶⁹ As such, the canons have a clear structural value and therefore ‘assume a dynamic, essential role, initiating musical processes rather than merely explaining otherwise complete notation’. ⁷⁰ The canon of Ferre solet might thus exemplify the ‘seminal role the visual form of music [could] have played in the act of composition’. ⁷¹

Before turning to the analysis of the texts of the motet, an explanation of the tenor and contratenor (and solus tenor) syncopations is necessary, since they are notably unusual. ⁷² All perfect modus segments are syncopated, i.e. the red and the black segments of the tenor and the red segments of the contratenor. The rhythmic result, however, remains in multiples of three.

As an example, Figures 6 and 7 show the first and the second segments of the first talea with the original notation and the transcription. In the contratenor, the single breve annotated Ia is made up to three by the two-breve rest at the end of the segment, annotated Ib (Figure 6). In the tenor, the two-breve rest annotated IIa is made up to three by the one-breve rest IIb (Figure 6). Similarly, the one-breve rest annotated IIIa on Figure 7 is made up to three by the two-breve rest IIIb. The original notation makes this very clear by the use of dots: dots of division clearly indicate which longs are perfect and which are to be imperfected by the following breve. ⁷³ Moreover, the separation dot between the final two-breve and one-breve rests indicates these two rests are not meant to form a perfection. The editorial barlines of the transcription provided in Appendix 1 show the syncopations by isolating the syncopating note and grouping the bars according to the modus. These syncopations, in addition to the proportional and mensural changes, add significant rhythmic complexities to the motet. ⁷⁴

Figure 6. Tenor and contratenor syncopations in Ferre solet (talea 1, segment 1)

Figure 7. Tenor syncopations in Ferre solet (talea 1, segment 2)

Literary Encoding Mechanisms

In addition to a complex compositional design, the motet Ferre solet is built upon three fascinating literary encoding mechanisms: ⁷⁵

Motetus

The first poetic construction is found in the motetus text: the last syllables of each line, grouped in threes, form the first four words of the antiphon used in the tenor: ‘ave maria gracia plena’. Such poetic constructions are very common in devotional poetry and not unusual in motet settings. For example, these same words are also the basis of the acrostics in the triplum of Ave regina celorum/Mater innocencie/Tenor of Marchetto of Padua and the triplum and motetus of Aurora vultu pulcrior/Ave virginum flos et vita in the Turin codex. ⁷⁶ However, the fact that Ferre solet ciphers the prayer through a telestich already points to the exceptional character of this piece.

The second poetic construction lies in the text of the triplum. A fragmented acrostic ciphers the clerical status and the name of a composer hitherto unknown: Frater Johannes Vavassoris. ⁷⁷ Interestingly, to each of the three words corresponds a specific way of forming an acrostic (‘Frater’ is formed by the first letter of each line, ‘Johannes’ by the first letter of each first and third line, ‘Vavassoris’ by the first letter of every third line).

The third poetic construction which sets this motet apart most especially appears again in the motetus. In fact, the acrostic interwoven into the triplum can be read as the beginning of a much larger sentence continued in the motetus: here, another acrostic, this time composed of the first syllables of motetus lines, reads: ‘Anno domini millesimo trecentesimo septuagesimo tercio fecit istum motetum’ (‘[Frater Johannes Vavassoris] made this motet in the year of our Lord thirteen seventy-three [Ave Maria gracia plena]’). ⁷⁸

Acrostic composition constitutes part of a very long poetic tradition and one that was highly developed by the Middle Ages. ⁷⁹ Indeed, acrostics are far from uncommon in fourteenth-century motets. As in Ferre solet, they can be formed of either the first letters or first syllables of the text and sometimes only a few of these letters (or syllables) are to be taken into account. Such acrostics often take on religious significance, through quotation from the Bible (‘Adina’ in Machaut’s Amours qui a le pouoir/Faus samblant m’a deceu/Vidi Dominum ⁸⁰ ), other liturgical phrases (‘Deo gratias’ in an anonymous motet of the Turin codex ⁸¹ ) or the Marian prayer. ⁸² The dedicatee is also frequently named in an acrostic, ⁸³ and this poetic device can also be used as a means of reinforcing the meaning of the whole motet. ⁸⁴ However, authorial acrostics are very rare in fourteenth-century motets: ⁸⁵ only Marchetto of Padua’s motet Ave regina celorum/Mater innocentie contains one in the motetus voice. ⁸⁶ However, even if both Ferre solet and Ave regina celorum are Marian motets, their literary encoding mechanisms seem to have slightly different functions. In Marchetto’s motet, the Marian prayer is inserted in the triplum. His name is presented in the motetus and in the accusative case, which, according to Anne Walters Robertson, ‘signals his plea to be the recipient of grace’. ⁸⁷ By contrast, in Ferre solet, the Marian prayer is presented as a telestich in the motetus, while ‘Vavassoris’ is inserted in the triplum and in the nominative case, with the date inserted in the motetus: he clearly presents himself as the narrator and the maker (‘fecit’) of the prayer. ⁸⁸

The three literary encoding mechanisms (the two acrostics and the telestich) are all highlighted visually in the fragment. First of all, the scribe reveals the connection between the antiphon used in the tenor and the telestich of the motetus in the verbal canon copied at the end of the motetus. The text states: ‘read the last syllables of this poem: “ave maria gracia plena”, to be read in these T[enors]’ (see above, Figure 4, lines 1–2). The ‘poem’ mentioned here clearly refers to the text of the motetus while the missing text after the letter ‘T’ is without doubt the word ‘Tenoribus’. ⁸⁹ The textual connection between the two voices also explains why the canon relating to the motet’s tenor statements is copied on fol. 4^r after the motetus and not on fol. 3^v near the tenor (where there is still space): it is only at this point that the scribe reveals the relation of the motetus to the tenor and, probably taking advantage of quoting this lower voice, indicates how to sing it immediately afterwards. Furthermore, all the letters that form an acrostic are highlighted by capital letters. ⁹⁰ This latter feature of the scribe’s copying process is far from coincidental: the last incomplete motet of the fragment, Salus honor, features an acrostic encoding the name of Saint Lambert (‘Sanctus Lambertus’), which is also highlighted in the source by capital letters. ⁹¹ Consequently, the scribe clearly knew and understood the acrostics. This points at the textual understanding of the scribe, which resonates with his musical understanding outlined above (assuming that the same scribe copied both texts and music): the overall very sensitive alignment of music and text, the careful adjustments from red to black notation and the very small number of mistakes that have been found (compared to, for instance, the Chantilly scribe) suggest that the scribe was quite intimately connected to the compositional milieu. ⁹² In any case, such an extreme example of textual and musical interpolation suggests that Vavassoris was responsible for all the texts and music of Ferre solet.

CONCLUSION: SIMILARITIES, CONTRASTS AND PROVENANCE

In conclusion, it can be observed that the five pieces of the Douai fragment share several structural similarities: the motets Ferre solet and Salus honor share textual and copying similarities (the acrostic, highlighted by capital letters); they both open with a rhythmically free introitus and with very similar melodic contours. Interestingly, this melodic contour is also closely related to the incipit of the Gloria, a polyphonic setting which, like these two motets, opens with a duet. Other similarities can be found between Multipliciter amando and Salus honor: for example, they both have one rhythmic pattern that comes to be stated three times in the settings. Contrastingly, Ferre solet and O martir beatissime are comprised of two rhythmic patterns stated three times in their settings, and additionally, both use diminution. However, the musical styles of these pieces seem to point in different directions: the Gloria and Salus honor are set entirely in minor prolation, contrasting with the changes of mensuration and/or proportion in the three other pieces. Similarly, the simple contrapuntal style of the Gloria appears quite distinct from the syncopations of the lower parts in Ferre solet, for instance, or the rhythmic subtleties in Multipliciter amando. In turn, the presence of these contrasting pieces next to each other invites more detailed analysis to be undertaken against the background of a broader repertorial survey.

More generally, the Douai fragment is curious: highly complex music is copied in a rather cheap source (the parchment is of low quality and contains no decoration) by a scribe who was very versed in mensural notation of that time. This opens new questions on the intended use or performance of such music, as the Douai fragment differs from more lavish sources that transmit a similar repertoire, like the aforementioned Ivrea codex or Chantilly.

Similarly, the Douai fragment raises questions about the geographical context for such a complex compositional culture: while no literary encoding mechanisms give any clue to the provenance, the close connection of the name ‘Vavasseur’ with Normandy, together with the incomplete motet dedicated to St Lambert and the current location of the fragment, strongly argue in favour of a northern French or Low Countries provenance. ⁹³ In spite of the destruction of many archives during the Hundred Years’ War and into the twentieth century, there is evidence for strong musical activity, including complex polyphony, in medieval Flanders, which was after all one of the wealthiest areas in Europe. ⁹⁴ The Douai fragment, with its most unusual motet Ferre solet, is another witness not only for the production but also for the transmission and performance of complex polyphonic music in this region and at that time. While it is necessary to proceed with caution, as dating based on stylistic grounds is particularly challenging (if not misleading) ⁹⁵ this unique case of composer identity and dating opens numerous avenues for future research, not least with regard to Frater Johannes Vavassoris and the context in which his motet was created, but also with regard to the implications of such a piece for larger narratives about music history.

University of Oslo