3 Early polyphony to circa 1200

Sarah Fuller

Some surface aspects of musical art for the adornment of ecclesiastical song s have been outlined here. 1

Here begin mellifluous songs of organum upon the sweetest praises of heaven. 2

But in whatever way it is done…in producing diaphony the precentor must harmoniously sing in praise of the creator. 3


As the above quotations suggest, early Western polyphony was broadly viewed as a way of elaborating and adorning monophonic chant , was heard as beautiful and sweet in sound, and was considered an appropriate means of religious praise. The close connection with religious worship is not surprising, given that the extant written record – both theoretical descriptions and musical notations – of polyphony stems from literate ecclesiastical and monastic spheres. But that record is decidedly incomplete and radically discontinuous. This is because early polyphony was ‘produced’ or ‘made’, not ‘composed’ in the present-day sense of ‘composition’: it was in the first instance sounded, not notated. Polyphony arose and continued as a performance practice, a way of elaborating a known monophonic melody with a second line that was produced according to accepted conventions. 4 Polyphonic singing enhanced worship through amplification of monophonic song. The earliest treatise to describe the phenomenon, the Musica Enchiriadis from circa 850, treats it as a familiar extempore practice, one known under two names: diaphony and organum. These alternative terms (the latter characterized as ‘customary’), refer to different core facets of polyphony: diaphonia to the dual sounds produced between two voices, the sounding-apart of the pair; organum to the perfect consonances or symphoniae (fourth, fifth, octave) that controlled the relationship between voices. 5 Guido of Arezzo's formulation in Chapter 18 of Micrologus (ca1030) was widely read and often quoted or paraphrased by later theorists: ‘Diaphony sounds as a disjunction of pitches [or lines] which we call organum since these differentiated pitches/lines both sound separately in concord and concord in their separate sounding.’ 6 Diaphony remained a two-part phenomenon through the mid twelfth century, although in the most basic types attested in the earliest treatises (and still practised into the fourteenth century) either voice or both might be doubled at the octave.

From a present-day historical perspective, ‘early ’ medieval polyphony is generally understood to span a time period of roughly three centuries, from the mid ninth century, when the first written descriptions of the phenomenon appear, to around the mid twelfth century, when polyphonic repertories preceding the advent of Parisian polyphony were notated. Parisian polyphony (sometimes designated ‘Notre Dame’ polyphony, after the famous cathedral) is considered a prominent articulation point in the history of polyphony because of many factors: the magnitude of its notated repertory, transmission of that music in multiple sources, an innovative technology of rhythmic notation, its status as a foundation for later developments in polyphony (pieces for 3–4 voices, the thirteenth-century motet), the extensive theoretical literature that accrued around it. This chapter deals principally with what can be discerned about diaphony or organum up to the advent of the Parisian repertory. Although ca1160 is a rough cut-off point for this survey, basic techniques for generating polyphony engendered in this ‘early’ period persisted for decades beyond the 1100s.

Broadly speaking, our knowledge of early polyphony stems from two sorts of material traces: instructions in theoretical treatises and notations of organum in musical neumes. The two are decidedly different in nature: the former essentially didactic and verbal in expression, the latter ‘practical’ representations of pieces in musical notation that directly evoke sound. Because they are generally cast as basic training for novices or summary digests of simple principles, the theoretical dicta must not be taken as assured guides to the actual practices of skilled organum singers. Indeed comparison between theorists' precepts and collateral notated repertory invariably shows discrepancies (a situation hardly unique to this material and time period). 7 But because the notations transmitting the repertory are all indeterminate to varying degrees, and because the active interpretative traditions that informed those notations evaporated centuries ago, any modern reconstructions of the music (in editions or in performances) require historical grounding in theoretical precepts contemporary with the notations. 8

Music historians generally parse pre-Parisian polyphony in three phases (see Figure 3.1 ). The earliest is organum based on the concept of parallelism with a pre-existent chant . This is attested by the oldest treatises of the ninth into the eleventh century and is manifested in the Winchester organa of ca1000. The middle phase, described by theorists towards the end of the eleventh century, is a more freely associative organum in which the added line is melodically independent from the principal melody. The final phase is a florid organum characterized by decorative figures in the organal voice and frequent contrary motion between the lines. The twelfth-century repertories of Aquitanian and Compostelan polyphony exemplify this individual and elaborate organum. But such ‘phases’ are not to be viewed as stages in which one practice supplants the previous one; rather, each practice simply adds another strand to accrued possibilities. Earlier practices continue actively alongside later ones, as we know from fourteenth-century theorists who describe how to produce ‘simple’ parallel polyphony, and from late medieval notations of polyphony in retrospective styles. 9 This continued vitality of early traditions for producing diaphony may be traced both to regional habits and to the level of musical competency available in local situations. 10


Figure 3.1 Cumulative phases of early organum production (ov = organal voice; pv = principal voice; u = unison)

The changing course of new protocols for realizing polyphony is more apparent in the pedagogical treatises – of which there is a fairly steady (if chronologically blurred) record – than in the substantial notated repertories where there is a considerable gap between the Winchester organa of ca1000 and the emergence of Aquitanian polyphony around 1100. But the treatises, limited as they are to giving compact directions for improvising music in two parts, offer only partial guidance to actual practices and would all have required considerable amplification by a seasoned singing master. The Musica Enchiriadis author admits to providing only a surface view of polyphonic adornment, while a later author, coping with a situation in which the organal voice exercises considerable choice relative to the cantus , simply acknowledges the inadequacy of written precepts: ‘All this is better revealed through the practice of organum singers than through rules.’ 11 The repertories, along with sporadic notations of individual polyphonic pieces, not only manifest creative musical facets that extend well beyond didactic formulations, but also provide essential (if partial) insight into ritual contexts for polyphony. Polyphonic examples in the earliest treatises, such as the Musica Enchiriadis and Guido's Micrologus , are largely isolated phrases drawn from recitation tones and office antiphon s – brief, elementary illustrations accessible to novice singers. In contrast, the Winchester repertory presents a strikingly rich image of liturgical contexts for early polyphony, including as it does organum for core chant s of the mass and offices (alleluias, proses, Kyrie s, responsories, etc.) collected in liturgically ordered groups and preserved as complete pieces. This record of intense polyphonic activity around 1000 at Winchester Cathedral constitutes our best glimpse of the possible scope of organum singing in major European churches and monasteries as the first millennium drew to a close. 12 Treatises within the second phase characteristically offer only capsule illustrations, some on an alleluia incipit, but some abstract exempla with no liturgical association. The twelfth-century Aquitanian polyphonic repertory, however, manifests a new forum for polyphonic production, one removed from the formal liturgy, centred instead on rhymed, often strophic religious poetry or versus. 13 The mixture of two-voice liturgical settings and devotional versus in the coeval Compostelan repertory indicates that polyphonic adornment spanned a broad spectrum of worship and celebration (formal and informal) at the Compostela shrine. Although limited in its focus on veneration of a single saint, the Compostelan repertory evidences a breadth of polyphonic activity congruent with that registered in the far more extensive collections of Parisian organa and conductus.

Partial and incomplete though they are, the written traces of early polyphonic practice constitute the foundation for any account of Western polyphonic production from the ninth into the mid twelfth century that tracks the striking changes in idiom that transpired over that time period. The following account of key way stations, which follows a loosely chronological path, should be read with due awareness of the condition of polyphony as an extempore (and characteristically ephemeral) means of sonic elaboration practised in ecclesiastical spheres and of the persistence of the earliest protocols for generating organum alongside more complex procedures.

The earliest layer of organum teaching extends from the Musica Enchiriadis (ca850) to Guido's Micrologus (ca1030), and includes various affiliates and adaptations (of indeterminate date) of these core texts. Although differing in specific formulations and explanations, the treatises largely concur on general procedures. 14 The singers start with a pre-existent chant (designated the principal voice) and supply to it a complementary organal voice. This voice lies below the principal chant, relates to it through the perfect consonances or symphoniae of fifth and fourth, and mainly matches it note for note in rhythmic synchronization (see Examples 3.1a and 3.1b ). Two procedures are taught. In one, a strict organum, the second singer parallels the principal voice consistently at the fifth below. In the other, a flexible organum, he follows it loosely at the fourth below. The Musica Enchiriadis theorist differentiates these two types of organum according to the primary interval of coupling, fifth or fourth respectively, while Guido, in Micrologus , distinguishes them as durus , or strict, and mollis , or flexible. 15


Example 3.1a Musica Enchiriadis , Chapter 15, Sit gloria Domini , diaphony at the fifth with octave doublings

Example 3.1b Guido of Arezzo, Micrologus , Chapter 19, Ipsi soli , flexible diaphony at the fourth

Diaphony at the fifth is strictly parallel, as is evident in Example 3.1a where the organal voice persistently shadows the principal chant at the fifth below. 16 Either voice (or both) may be doubled at the octave (above or below) to amplify the aural effect.

Flexible organum at the fourth is more complex and entails more decisions by the organum singer ( Example 3.1b ). Depending on factors of mode, linear path, and register of the parent chant, the organal voice may parallel the chant at the fourth, remain stationary on a lower boundary tone, or join it in unison at a phrase ending, an action Guido calls occursus. Articulating moments of occursus are determined by textual units in conjunction with the chant's phrasing. Guido's presentation of the D-mode antiphon Ipsi soli shows the organal voice observing a mandatory tritus boundary within each phrase and varying its phrase endings according to context. 17 For this melody, unison occursus is strategically located only at the ends of the first and final phrases, with the last the sole close on the modal final. The non-consonant seconds and thirds introduced by the boundary tone convention, along with unison occursus, impart to flexible organum a sound palette considerably more varied than that of the consistently parallel strict organum.

Beyond technical aspects of how to produce appropriately consonant organum, the early treatises provide some informative comments on general performance aspects of early diaphony. Several mention doubling either voice or both at the octave or double octave, and associate this with boys and men singing together. That organum should be sung at a moderate tempo and with due attentiveness is another recurring theme. One teacher specifically remarks that this kind of music is so weighty and deliberate that the usual rhythmic ratios of chant cannot be observed. 18 The importance of textual articulations, of musically observing units of text syntax (phrase, clause, sentence), is often mentioned, especially in conjunction with flexible organum and decisions about occursus. In addition to extolling diaphony as an appropriate means of praising God, the theorists call attention to its beauty of sound: ‘Truly, delivered with restrained care, which is most proper to it, and attentive management of concords, the sweetness of the song will be most beautiful.’ 19 This appreciation of beautiful sound is also manifested by the scribe of the Winchester organa, who variously characterizes the organa he copies as ‘melliflua’, ‘pulcherrima’, ‘iocunda’. 20

The remarkable collection of Winchester organa, preserved uniquely in the manuscript Cambridge , Corpus Christi College 473, manifests practices much richer and more diverse than those promulgated in the elementary teaching manuals. Although the organal voices are notated separately from their companion chants, and in non-diastematic neumes, the nature of this polyphony has been well established. 21 In general, the organa manifest traits characteristic of flexible diaphony at the fourth. The organal voice mainly operates below the principal one, matches it rhythmically note for note, and responds to it with a mix of parallel motion, sustained boundary pitches, and unison convergence. But compared with the theoretical expositions, the Winchester pieces show more open choice of reiterated boundary tones, more frequent occursus and provision of an extra tone or two in the organal voice, freedom occasionally for the added line to rise above the principal voice, and overall sensitivity to the text syntax and structure of a complete chant. 22 Marginal annotations in the manuscript not only point toward actual performance of the repertory, but also indicate a willingness to alter or vary passages in the organal voices originally notated. The Alleluia. Ascendens Christus , one in a major cycle of 53 polyphonic alleluias, clearly ranges beyond the usual short theoretical examples in complexity and variety of organal voice movement ( Example 3.2 ). 23 The principal voice, a second-mode melody, is narrow in range, circling within the fourth d–g, with c as a frequent lower neighbour to the final, d. Only twice in this excerpt does it rise to the fifth above the final, once at the beginning of the jubilus and once at the suggestive words ‘in altum’ (on high). The organal voice accordingly remains in a low register, often reciting on the sub-final, c, or the final, d (notes 12–19, 38–44, 66–73). It joins the principal voice in unison at the end of basic text units and of the jubilus (notes 20, 52, 69, 73, 90) and also converges at numerous points within melismas (e.g. notes 11, 28). Although the organum sometimes sounds at the fourth (notably when the chant rises, notes 75–82), it liberally intermixes seconds, thirds and unisons. Seconds are quite frequent, not only at occursus (notes 10–11, 31–35) but also elsewhere (notes 42–3, 54, 70). The overall sound quality is hardly that of theoretically mandated symphonia of the fourth, although more descents to the tetrachord zone below the final would have increased the consonant sound quality. 24 Sometimes the notation indicates a momentary position above the chant (notes 29–30, 36), sometimes the two voices sing in unison (notes 55–9, 86–90), and sometimes extra notes in the organum enhance the effect of a convergence (notes 32–5, 44, 70–3). In its liberal approach to flexible organum, this alleluia setting is representative of the Winchester repertory as a whole, and hints at abundant creativity in the actual practice of organum around 1000. The few extant notations of early Continental polyphony manifest a similarly inventive approach to producing flexible organum at the fourth. 25


Example 3.2 Winchester organum, Alleluia. Ascendens Christus through first phrase of verse (Principal voice, GB-Ob Bodley 775, fol. 79v; organal voice, GB-Ccc 473, fol. 167r)

Guido's elementary teaching about strict and flexible organum in the Micrologus dates from around 1030. By the end of the eleventh century, theorists are describing a very different sort of organum practice, such that the Ad organum faciendum theorist prefaces his exposition of the new way of generating organum with the remark that Guido's teachings are ‘worthless’ and ‘scarcely to be esteemed’. 26 The symphoniae remain central to the concept of organum, but now fourths and fifths may be intermingled and are joined by octaves and unisons not just as intervals of doubling and occursus respectively, but as fully participating vertical sonorities. The older notion of boundary tones has disappeared. Remarkably, the organal voice has shifted its standard position from below the principal chant to above it. In contrast to the first-phase teaching, which stipulates but one or two appropriate organal responses to a given cantus , this second-phase pedagogy provides for, even encourages, a multiplicity of possible organa against a cantus. Ad organum faciendum organizes its teaching neatly in terms of five modes or types of producing organum, and illustrates each with a uniform cantus, the opening twelve-note phrase of the Alleluia. Justus ut palma ( Example 3.3a ). 27 Each type entails a different disposition of perfect consonances and unisons within the phrase. For example, the first two types feature different opening gambits, the first beginning conjunctly with the first cantus note (here at the octave), the second beginning disjunctly (here at the fourth) ( Example 3.3a , I, II). The third type intermingles disjunct intervals and unisons in its middle passages ( Example 3.3a , III). In a fifth category – one that bears important implications for the emergence of florid organum – the organal voice sings multiple notes against a cantus tone, a tactic here strategically located at phrase end ( Example 3.3a , V). A longer example, the beginning of this Alleluia's verse, shows how the procedures of the preceding exercises can be freely combined in a series of phrases ( Example 3.3b ). 28 The absence of boundary tones coupled with acceptance of octave intervals above the cantus give this manner of producing organum a more consistently consonant quality than that of earlier flexible organum (compare Examples 3.3a , 3.3b with Example 3.2 ). In distinguishing among inceptions, medial passages, and terminations (arrivals on unison or octave termed copula by some teachers), the Ad organum faciendum pedagogue and his associates foster in the novice singer functional awareness of constituent phrase units within the music. The open choices for intervallic successions and for ranging either above or below the given line afford the late-eleventh-century organal singer considerable scope for individuality in responding to a cantus.


Example 3.3a Ad organum faciendum , modes of organum I, II, III and V on incipit of Alleluia . Justus ut palma

Example 3.3b Ad organum faciendum , beginning of verse, Alleluia . Justus ut palma

The verse treatise that follows Ad organum faciendum in its principal source advances similar teaching but is more colloquial in expression and more informative about cultural attitudes than its technically oriented companion. Its characterizations of the relationship between the two parts announce a new, dominant status for the organal voice. The opening introduces a metaphor of friendship between the voices, figurative language that may emanate from the notion of ‘affinity’ between the note pairings in organum. 29 But by the end of the treatise, the metaphor has shifted to serve a vigorous assertion of the superiority of the organal voice. It is described as running ‘like a very strong soldier, most gratifyingly’ and ‘breaking up pitches like a prince’, a likely reference to florid elaboration in the organal line, even though the proffered examples do not show this. 30 More than Ad organum faciendum , this verse treatise encourages and illustrates contrary motion between the parts. Comments on sweetness of sound (lines 66, 85–6, 138, 140) promote genuine aesthetic appreciation for organum and mesh with similar comments from the first stage of organum theory. 31 Other brief treatises conventionally grouped with Ad organum faciendum extend explicit recognition to thirds and even sixths as acceptable organal intervals, and increasingly draw attention to contrary motion between the voices. John's treatise on Musica from around 1100 even goes so far as to advocate contrary motion between cantus and organal voice as the ‘simplest’ procedure. 32 John also encourages multi-note elaboration in the organal part. Concerning a short example in which an organal part is paired strictly note to note with a syllabic cantus, he remarks that the organal singer could well ‘double or triple’ or otherwise competently accumulate notes in the sparse line provided. 33

A subsequent state of organum pedagogy (perhaps emergent around the mid twelfth century) switches perspective radically to focus on standard binary interval progressions between organal voice and cantus. 34 The interval-progression treatise s value octaves, fifths and unisons (fourths in some strata) and feature exercises of the type, ‘If the organum is an octave above the cantus and the cantus ascends a whole-step, let the organum descend a third to be at the fifth.’ Such teaching in terms of discrete two-note segments departs from the phrase-based approach of earlier organum pedagogy, but still fits with extempore production of polyphony through the act of singing. Under this training, singers practise and internalize in ear and memory standard voice-leading moves that they can then string together to produce appropriate organum rendered through approved perfect consonances above a cantus. 35 Although contrary motion is the rule in such treatises, some include formulas with similar or parallel motion, while a very few describe a mode of production based on parallel fifths. 36 But the theoretical discussions are brief and abstract in nature; for a sense of actual production and cutting-edge trends in polyphony during the twelfth century, one must turn to notated music.

The Aquitanian repertory of polyphony manifests a full spectrum of practices that are sporadically attested in the later ‘second phase’ theoretical documents and are generally compatible with early interval-progression pedagogy. The designator ‘Aquitanian ’ comes from the type of chant notation in which this repertory is preserved, a notation practised in the area of southwestern France known as Aquitania in Carolingian times. 37 This polyphony, which seems to have flourished from around the 1090s into the second half of the 1100s, contrasts sharply with that of Winchester, not just in musical idiom but with respect to texts, religious functions, and manuscript transmission. Rather than adhering to the formal liturgy, Aquitanian polyphony comprises versus, some versus-like Benedicamus Domino tropes, and a few liturgical proses – with versus the predominant genre. Versus are settings of devotional poetry, poems that are newly created and express contemporary religious sensibilities, particularly veneration of Jesus and of Mary. Typical subjects celebrate the birth of Jesus, the motherhood of Mary, the miracle of the Virgin Birth, the union of the divine and human in the Incarnation. Such song s could be inserted into ceremonial offices, but could also have been sung on more informal occasions. 38 This polyphony is preserved in nine small to medium-sized pamphlets (libelli ) that transmit both monophonic and polyphonic songs. Each source is an individual collection with its own independent selection and ordering of pieces; from among some seventy polyphonic settings, only four pieces appear in as many as three sources, and only one appears in four sources. 39 It is likely that individual poet-singers, or their followers, copied these songs for personal use. Important beyond the history of polyphony, the Aquitanian versus repertory manifests a remarkable creative outpouring of devotional Latin song by twelfth-century poets and singers in southwestern Europe.

The turn to newly created poetic texts means that the polyphonic Aquitanian versus were not necessarily based on a pre-existent melody – indeed, rarely is a monophonic ancestor documented for a polyphonic versus. Some versus music may well have been conceived as polyphony from the start, a conclusion fortified by extended melismas in which the two voices are so intertwined as to be an inseparable pair. Such concurrent conception would represent a facility of invention beyond that expressed in organum pedagogy, which regularly postulates a given cantus , even if that is reduced to a two-note element. The thirteen proses in the Aquitanian repertory are based on pre-existent chants, as are a few combinations of a Benedicamus versicle with a superimposed versus. To avoid the assumption of a given cantus, I will here designate the voices as upper and lower, a reflection of the manuscript notation and prevailing registral positions (although some voice-crossing does occur). 40

Like their monophonic siblings, polyphonic versus settings are shaped on the one hand by poetic versification and syntax, on the other by melodic norms, tonal centring, and matrices (horizontal and vertical) of perfect consonances and unisons. Other musical resources coordinated with text structure include texture and manner of text declamation. 41 As can be seen in Per partum virginis , versus poetry exhibits periodic regularities of syllable count, rhyme and line groupings in either couplets or stanzas (see Figures 3.2a and 3.2b ). The sequence or prose texts also observe regular, periodicity, the parallel couplets typical of the genre. The periodicities of poetic structure constitute a framework respected and delineated by the musicians, as can be seen in a representative polyphonic versus, Per partum virginis . 42 The poem celebrates the birth of Jesus, his mother Mary's virginal childbearing, and the positive consequences for humankind. It is cast in couplets whose clear boundaries (outlined in Figure 3.2b) are created through syllable count, rhyme, accent and syntax. Each line falls into three segments, the first two sharing a common rhyme, the last rhyming with its partner within the couplet. Odd- and even-numbered couplets are differentiated by syllable count within the lines and by verse accent, with couplets 1 and 3 dactylic, 2 and 4 and the single final line trochaic.


Figure 3.2a Text and translation of Per partum virginis

Figure 3.2b Text structure and musical phrases in Per partum virginis

The musical structure aligns with that of the poetry on many levels from the principal sectional units to phrase and motivic elements (see Examples 3.4a and 3.4b and the left-hand column of Figure 3.2b). 43 In this musical setting, the couplets are through-composed except for the fourth (4ab) where repeated music audibly underscores the verse pairing. 44 Extended terminal melismas with the voices yoked together articulate the ends of the first two lines and the final line. 45 Changes in texture demarcate sectional units. Line 1a maintains a florid upper voice up to the terminal melisma, while line 1b begins with the voices note against note, then intermixes such synchrony with florid figures in the top part. The final couplet and final line offer a sharp contrast to the opening in their very plain note-for-note texture. The rhyme words internal to poetic lines also receive their due. This is particularly clear in line 1a where a sweeping downward scale figure (distinguished in scope from the other ornamental figures) associates virginis with hominis . Such explicit association of internal rhymes does not continue in the rest of the setting, but the main rhymes are punctuated with stable arrivals on perfect consonances and on the modal final or its close associate.

Per partum virginis is tonally focused on g with the lower voice beginning and ending on that pitch, and returning often to it at the end of text segments. The first two lines are anchored on a g final that is supported in the upper voice with the upper fifth and octave, and often reinforced by an upper-voice descent to g when the lower one rises to the fifth or octave above. At the most stable phrase endings – the end of the first couplet (ethera ) and the conclusion of the piece (gratia ) – the voices unite on a unison g final. Together the two voices present a unified tonal profile that mainly projects a g–d1 –g1 matrix in coordination with the primary text units. The fourth couplet ( Example 3.4b ) offers tonal contrast in its orientation towards the fourth and fifth above the final, a shift that gives special impact to the restoration of g centring in the final line.

In keeping with their tonal compatibility, the two lines are anchored, both horizontally and vertically, on perfect consonances and unisons. A background of fifths, octaves and unisons can easily be discerned behind the elaborative figures in the upper voice, as, for instance, in the very first text line, shown in a voice-leading reduction in Example 3.5a . Such a spare background calls to mind John's invitation to the organum singer (quoted above) to proliferate notes within a foundational note-against-note context. Less consonant, or even quite dissonant, intervals sometimes enhance the motion towards a stable interval, as on the very first syllable (Per , from fourth to fifth) or in line 1b with seconds on di luitur and que .

Starting with the earliest organum treatises and continuing through John's Musica and the interval-progression treatise s, voice-leading stands alongside intervallic proprieties as a primary pedagogic concern. Reflecting an accumulation of possibilities, voice-leading in polyphonic Aquitanian versus intermixes contrary, oblique and parallel motion, the latter commonly involving parallel fifths that are sometimes overt, sometimes covered by upper-voice figures. 46 The voice notated in the upper position generally sounds above the notated lower voice, but sometimes crosses below, particularly when the lower voice rises in register. 47 The most striking and characteristic voice-leading routines appear in the long terminal melismas constructed in a mosaic fashion from figures that converge and diverge in contrary motion. Verses 1a and 1b both end with melismas built from two-voice figures that characteristically pivot on a unison b and diverge out to the g–d1 fifth (see Example 3.4a ). The very last melisma of the versus has a more complex, tonally diverse structure that can be parsed in a series of contrary motion elements: octave, fifth, fourth and third converging into unisons, unisons expanding out to a fifth or fourth (see reduction, Example 3.5b ). 48 These figures decorate a dramatic scalar arch in the lower voice that rises through the g octave and then descends directly to the final. Such intertwining convergent/divergent figures, common in melismas of the polyphonic Aquitanian versus, appear to have been standard routines, easily produced extempore, adaptable to texted phrases as well.


Example 3.4a Aquitanian versus Per partum virginis , first couplet (GB-Lbl add. 36881, fol. 4r)

Example 3.4b Aquitanian versus Per partum virginis , last couplet and final verse (F-Pn fonds lat. 3549, fol. 151v)

Example 3.5a Consonance reduction, first phrase, Per partum virginis (see Example 3.4a)

Example 3.5b Consonance reduction, final melisma of Per partum virginis (see Example 3.4b ) Basic intervals are indicated below the staff. An underscore (5 ) indicates that the upper voice has crossed below the lower one. The symbol * marks a diminished fifth or augmented fourth that might or might not have been adjusted. Lower-voice note count appears above the staff.

The three notated versions of Per partum virginis – like notations of other versus inscribed in more than one source – indicate some fluidity in how singers might realize a song . Typical differences, seen in the two versions of line 1a aligned in Example 3.6 , are details of upper-voice figures and scribal indications of vocal timbre and alignment between the parts. 49 Variants in the upper-voice (the initial c1 –d1 gesture, the first and last syllables of virginis ) indicate flexibility in realizing elaborative figures. The ‘B’ scribe more often indicates special vocal production, such as liquescence or extension of a pitch. Alignment between the parts is variable (and a problematic issue). Whereas the ‘D’ scribe supplies many vertical strokes to specify syllable coordination, the ‘B’ scribe scarcely indicates syllable boundaries and is fairly cavalier about depicting vertical alignment through physical spacing. Still, ligature groupings and a few syllable strokes in ‘B’ do point to some alignments divergent from ‘D’, as on hominis . 50 Each notated version, indeed, seems to represent a somewhat individual view of ‘how the piece goes’ in keeping with a fluid, improvisational approach to producing polyphony. Still there are passages involving uncharacteristic dissonances in which one source attests a scribal mistake in another, as on et (no. 10 in Example 3.6 ), where the ‘D’ scribe apparently wrote the last note in the upper-voice figure a step too high.


Example 3.6 First line, Per partum virginis , two versions (B = F-Pn fonds lat. 3549, fol. 150v; D = GB-Lbl add. 36881, fol. 4r)

Another, less extensive, polyphonic repertory contemporary with the later Aquitanian sources is contained in the Codex Calixtinus , a substantial book dedicated to promulgation and celebration of the cult of Saint James at the pilgrimage shrine in Santiago de Compostela. 51 The collection of polyphony appears in a supplement positioned after the formal five books of the Codex Calixtinus. 52 Of the 20 pieces copied there, half occur monophonically in Book I, which transmits special divine offices for the saint's feast. As already noted, Compostelan polyphony comprises both settings of devotional poems, rubricated as ‘conductus’ or (as appropriate) ‘Benedicamus’ in Book I, and liturgical chant s: matins responsories, a responsory prosa, mass gradual and alleluia, two troped Kyrie s and three Benedicamus Domino versicles. A cultural context distinct from that of the Aquitanian sources is manifested both in the formal presentation of the polyphony within a carefully designed codex and in the combination of fixed liturgical items and ad hoc devotional poems. Insofar as the Codex Calixtinus has been associated with compilers in the region of Burgundy and with Cluniac interests, its polyphonic component would seem to reflect northern traditions. 53

Like their Aquitanian counterparts, conductus/versus in the Compostela repertory are songs on devotional poetic texts, but their consistent theme is veneration of Saint James. Musically, they resemble their Aquitanian cousins; indeed two of the polyphonic conductus are contrafacta of Aquitanian versus. 54 In both repertories, musical phrases and formal design conform to poetic versification, the voices are tonally coordinated and linked through perfect consonances and unisons, and texture varies from single-note coupling of the voices to florid decoration in the upper voice. One difference is a lack of extended note-against-note melismas except in the two imports from Aquitania.

The liturgical polyphonic settings, all based on a pre-existent chant, adopt florid textures with figures of three to nine notes in the upper voice decorating perfect consonances above individual cantus tones. In contrast with the conductus, there is little voice-crossing in these settings. The upper voice tends to stay above the cantus, or to join it in unison when the cantus rises or settles on the modal final. 55 Although the able scribe provides division strokes to indicate segmental coincidence between the voices, the chant neumes do not permit specification of pitch alignment within those segments. The first segment of the responsory O adjutor , for example, combines two two-note neumes in the cantus with six neumes in the upper voice (see Example 3.7a ). We do not know how much leeway twelfth-century Compostelan singers would have had to combine the voices rhythmically within a segment, but in any event the alignment (along with relative temporal delivery) is now a matter of informed conjecture. 56


Example 3.7a Codex Calixtinus , opening of matins responsory O adjutor (E-SC , fol. 217r/188r). Plainchant continuation, fol. 110v.

The polyphonic settings of the responsories and the mass gradual and alleluia involve only the soloist's portions of the chant: the opening phrase of the initial section and the major part of the verse. The remainder of the chant is sung monophonically, as occurs in Parisian settings of responsorial chants. A glance at two passages from a typical responsory, O adjutor ( Examples 3.7a , 3.7b ), shows the upper voice coupled with the D-mode chant through transparent consonances of octave, fifth and unison. Some of these intervals seem locked in place relative to pivotal cantus pitches: a primary octave always surmounts the fourth and the step below the final (A and c), a unison or (occasionally) upper fifth sounds with the fifth above the final (a). In contrast, the d final offers more intervallic choices and is variously paired with its octave, upper fifth, or unison.


Example 3.7b Codex Calixtinus, first part of verse, responsory O adjutor (E-SC , fols. 217v/188v)

The upper-voice melody proceeds conjunctly by step or third, with larger intervals of fourth or fifth generally limited to occasions when the cantus repeats a pitch. Its elaborative figures, varied in length and lineaments, display inventive ways of elaborating a vertical interval or moving through a linear space. Immediate repetition of figures is rare (one occurs in the verse on mari ), and even separated recurrences are not conspicuous. A seven-note figure in the third segment of the respond recurs a fifth higher at the beginning of the verse (see brackets, Examples 3.7a , 3.7b ). The elaborating voice moves through scalar space in varied ways. The motion from c1 down to d in the first segment is gapped. That in the third segment (on adjutor ) turns gracefully in its path, while the stepwise descent on subvenis in the third segment of the verse is elegantly protracted through reiteration of successive notes in a series of two-note neumes (clivis). This last figure notationally and aurally resembles some figures in the Vatican Organum Treatise, one of the few interval-progression manuals to transmit decorated upper voices. 57 Background voice-leading relative to the cantus is a mixture of contrary and parallel motion, with the former predominating. The second segment in the verse is essentially a series of parallel octaves (g–f–e–d), but that parallelism is masked by ornamental figures in which the upper voice approaches the octave from below as the cantus descends by step. The voices typically reach unison convergences from different directions.

Thus, although from a condensed and selective historical perspective the significance of Aquitanian and Compostelan polyphony may seem to reside in setting the stage for Parisian polyphony, the level of complexity and musical invention manifested in these repertories, as in earlier ones, represents a considerable artistic accomplishment in its own terms and within indigenous local contexts.

That the above survey has concentrated on changes over approximately three hundred years should not obscure the considerable consistencies that existed within this span of Western polyphonic practice. By way of conclusion, these basic consistencies are summarized below.