¹ I am here echoing Hirshberg, J. ‘Hexachordal and Modal Structures in Machauf's Polyphonic Chansons’, Studies in Musicology in Honor of Otto E. Albrecht (Clifton, NJ, 1980)Google Scholar, 20: ‘the purpose of the present study is to offer a systematic description of the pitch-structuring process in Machauf's polyphonic chansons and to integrate existing information into a logical pattern’. Hirshberg's attempt was a strong impetus for the present study, as was that of Hughes, D. G., ‘A View of the Passing of Gothic Music: Line and Counterpoint, 1380–1430’, Ph.D. dissertation, Harvard University (1956)Google Scholar, though I often approach similar concerns very differently from these two authors. For the concept of tonal type to be developed below I am obviously indebted to Powers, H., «Tonal Types and Modal Categories’, Journal of the American Musicological Society (1981), 428–70.CrossRefGoogle Scholar An important new study that takes a preliminary form of the present article as one of its points of departure is Plumley, Y. M., The Grammar of Fourteenth-Century Melody: Tonal Categories, Pitch Functions, and Melodic Structures in the Polyphonic Chansons of Machaut and his Successors (New York, 1996).Google Scholar

² For this study, 384 chansons were considered. For the works of Machaut, see The Works of Guillaume de Machaut, 2 vols., ed. Schrade, L. (Paris and Monaco, 1956).Google Scholar For the later French secular repertoire, see French Secular Compositions, 3 vols., ed. Apel, W. (n.p., 1970–1972)Google Scholar and French Secular Music, 5 vols., ed. Greene, G. (Paris and Monaco, 1981–9; henceforth PMFC 18–22).Google Scholar In the following I shall use an algebraic shorthand for referring to songs, always citing Machaut after Schrade (e.g., Machaut's B15 is ballade 15 in that edition) and citing other chansons after Apel (e.g., FSC B136 refers to a ballade that is no. 136 in Apel's edition) unless only Greene has edited it (e.g., RPMFC22:35 is a rondeau that is number 35 in volume 22 of the PMFC edition).

³ On the emergence of the new courtly song style in the early fourteenth century, see Earp, L., ‘Lyrics for Reading and Lyrics for Singing in Late Medieval France: The Development of the Dance Lyric from Adam de la Halle to Guillaume de Machauf’, in The Union of Words and Music in Medieval Poetry, ed. Baltzer, R. A., Cable, T. and Wimsatt, J. I. (Austin, TX, 1991), 101–31.Google Scholar

⁴ Implicit in this broad generalization is my acceptance of the notion that composition of chansons as cantus-tenor duets was normally a simultaneous, not a successive process. I accept that in this process, as a composer spins out the voices, there might be some give and take between priorities that would occasionally favour the tenor. (For an attack on the overly simplistic, paradigmatic assumption of successive composition in the Middle Ages, see Leech-Wilkinson, D., ‘Machaufs Rose, Lis and the Problem of Early Music Analysis’, Music Analysis, 3 (1984), 9–11.)CrossRefGoogle Scholar

⁵ See, for instance, Hoppin, R., ‘Tonal Organization in Music Before the Renaissance’, Paul Pisk. Essays in His Honor (Austin, TX, 1966), 25–37;Google ScholarMarggraf, W., ‘Tonalität und Harmonik in der französischen Chanson zwischen Machaut und Dufay’, Archiv für Musikwissenschaft, 23 (1966), 11– 31;CrossRefGoogle Scholar and Hirshberg, ‘Hexachordal and Modal Structures’.

⁶ An argument for the ‘compositional precedence of the cantus over the tenor part’ and cautionary words about ‘our traditional understanding of the tenor as the “structural” voice in medieval polyphonic song’ have also been made recently in respect to the Italian trecento song repertoire. See Long, M., ‘Landini's Musical Patrimony’, Journal of the American Musicological Society, 40 (1987), 44–5.Google Scholar

⁷ For a categorization of the variety of cadence types in Machauf's chansons, see Hirshberg, ‘Hexachordal and Modal Structures’.

⁸ Rather than adopt some modern convention, in the following I shall use a slightly expanded medieval (Odonian) pitch notation with italicized letter-names (CC-GG (Gammaut), A-G, a-g, aa-gg) to indicate specific pitches in four specific registers (subgrave, grave, acute, superacute). Capital letters in roman will be used to refer to pitch-classes when specific registers are not critical to the discussion at hand.

⁹ Two chansons with g final at the unison in the cantus–tenor pair, Machauf's ballade Je suis aussi com cilz (B20) and Galiof's ballade Le sault perilleux (FSC B29), are supported by the contratenor at the underfifth on C. One anonymous ballade, J'ay grant desespoir (FSC B 150), has an octave cadence to g-G over contratenor C. These are all certainly to be thought of as ‘in G’. Likewise, Bartolino da Padua's madrigal with French text, La douce cere (FSC 11), which ends with cantus and tenor on e over contratenor a, is ‘in E’. Compare the position of Hirshberg, who is also sensitive to the problematic status of the tenor final as significant pitch: ‘Consideration of the finalis [e.g., tenor final] by itself, as the factor that determines and represents the “tonality” of a given composition, is insufficient. For example, the fact that the piece under consideration ends on C or on F does not automatically qualify it as being “in C” or “in F” ' (Hirshberg,‘Hexachordal and Modal Structures’, 26). Hirschberg regards the Machaut ballade Se je me pleing (B15) with cantus final on G supported by tenor on C as ‘weakly in C’ (‘Hexachordal and Modal Structures’, 36); in my way of thinking it is unequivocally ‘in G’.

¹⁰ A mid-fifteenth-century theorist provides near contemporary acknowledgement of certain of these norms in his differentiation of odd and even modes by the range above the tenor final. He states that when the cantus is in the even modes (i.e., the plagal modes 2, 4, 6, 8) it can ascend to a twelfth or thirteenth over the tenor final (‘finis tenoris’) and descend three or four notes under the cantus final (‘finis illius discantus’); this is the normal situation in the FSC chansons. When the cantus is in the odd modes (i.e., the authentic modes 1, 3, 5, 7), it can ascend to the fifteenth over the tenor final and descend one step below the cantus final, a configuration much more common in the chanson after c.1420 than before. (See Anonymus XI, Tractatus de musica plana et mensurabilis, ed. Coussemaker, E. de, Scriptorum de musical medii aevi nova series, 4 vols. (Paris, 1864–76), III, 465.)Google Scholar

¹¹ As in ‘Odo’, Dialogus, for which there is a convenient translation in Strunk, O., Source Readings in Music History (New York, 1950), 106.Google Scholar

¹² Ellsworth, O., The Berkeley Manuscript (Lincoln, NE, 1984), 53–5, 67.Google Scholar

¹³ Regarding where melodies are notated in monophony, upper and lower boundaries are clearly observed on the ranges employed (a central twelfth C-g expanded a step wider on either side to the fourteenth B-aa) by notating the few plagal melodies mainly among the higher finals. For cantus lines in polyphony the notated ranges lie in a wider span (a central double octave D-dd expanded a step wider on either side to the seventeenth C-ee); the few authentic cantus melodies. are mainly in the lowest register, with finals on F and G.

¹⁴ The Works of Johannes Ciconia, ed. Bent, M. and Hallmark, A. (Paris and Monaco, 1985), no. 42.Google Scholar

¹⁵ Roughly half the low G melodies are authentic rather than plagal, and these represent all the authentic melodies outside of Machaut; clearly, the use of low G this way is self-conscious.

¹⁶ On the lowest cantus finals, see also Lefferts, P. M., ‘Subtilitas in the Tonal Language of Fumeux fume’, Early Music, 16 (1988), 181 n8.Google Scholar As regards the highest finals, superacute bbb occurs only in Rodericus's notationally and tonally extraordinary ballade Angelorum psalat (BPMFC19:77). Superacute dd occurs in two anonymous, anomalous chansons with ‘authentic’ cantus that cadence at the top of their range, Filz il te faut (FSC B146) and Ne celle amour (FSC B167). Two anonymous chansons which cadence to acute-register g, Pour che que je ne puis (FSC B169) and Il vient bien sans appeller (FSC R251), are the only others to have this unusual arrangement of ‘authentic’ cantus and a final cadence at the top of the cantus range. (See Table 1.)

¹⁷ I have removed from consideration here three anomalous pieces with possible finals on A or E. One is a textless later addition to the Ivrea codex edited as a rondeau by Greene (RPMFC22:79) with cantus final on acute e harmonized at the underfifth by a. Given the proportions of the prima and secunda pars (thirteen bars to six bars) it may be a rondeau with three-line refrain or perhaps a virelai; if the latter, then the medial cadence on octave aa over a is in fact the final cadence. In any event the piece is odd tonally: the voices have narrow ranges with no overlap and the tenor is plagal. Another song edited by Greene as ending on E, the rondeau Faus semblaunt (RPMFC22:35), I take instead to be an F piece, prefering to solve a source conflict by amending cantus final e to f rather than tenor final F to E. Again the piece is tonally odd: there is little overlap of voices, the tenor is plagal and there are unusual cadence types. Finally, Greene edits the textless rondeau J'ay mis ce rondelet (FSC R252 = RPMFC22:42) in three voices with a final cadence on cantus aa over the tenor a twelfth below on D. I incline to Apel's edition in four voices and read the final of the second cantus on d as tonally defining.

¹⁸ In monophonic chansons by these composers the finals A, C, D are relatively rare, and there is just one with final on E. In the wider trouvére repertoire, melodies are found with finals on all the diatonic tones, including A, E, B and Bb (T. Karp, ‘Troubadours, trouvères, III, 2: Music, modality’, New Grove Dictionary of Music (1980), vol. 19, 200–1).

¹⁹ In the sixteenth century, seven finals are used. Arranged as a series of fifths, they are Bb F C G D A E. The first six (all but E) are used in conjunction with the scala b mollis (the cantus mollis system), and the latter six (all but Bb) are used in the scala h duralis (the cantus durus system). (See Ex. 2d.) The scala ficta (cantus fictus system) with signature Eb (and finals on Bb F C G D) becomes more and more rare after the generation of Josquin. See Powers, H., ‘Mode’, New Grove Dictionary of Music and Musicians (1980), vol. 12, 395–6;Google Scholar Powers, ‘Tonal Types’, 438; and Bent, M., ‘Diatonic Ficta’, Early Music History, 4 (1984), 11–12.CrossRefGoogle Scholar Many interesting questions can be framed about the evolution of choices of system and final across the time-span 1350–1550; the disappearance and re-emergence of signature sharps present a classic example. In the fourteenth and very early fifteenth centuries, composers of chansons were more adventurous about the use of systems and types, while less adventurous about the choice of finals, than in the later fifteenth century.

²⁰ In contrast to Hirshberg (Hirshberg, ‘Hexachordal and Modal Structures, 21, 39–40), I do not agree that the gamuts secundum artem and secundum usum of the Berkeley theory treatise (starting on FF and GG, respectively) are sufficient to encompass the tonal phenomena in Machaut's chansons both conceptually and practically. However, the compiler/author of that treatise must certainly have been attempting to come to grips with the use of lower registers and flatward tonal types through his mention of the gamut secundem artem (Ellsworth, The Berkeley Manuscript, 35–7). I would further note here that, contrary to Andrew Hughes's statement that ‘there exists no positive evidence for sharp signatures’ (A. Hughes, Manuscript Accidentals: Ficta in Focus 1350–1450 (n.p., 1972), 47), a transposition of the systems to the sharp side employing signature sharps, though rare, is not unknown. In the fourteenth-century chanson repertoire there are the anonymous rondeau Ay las quant je pans (FSC R238) in the two-sharp system with grave-register G final and the Anthonellus de Caserta ballade Notes pour moi (FSC B7) in the one-sharp system with acute-register g final, along with the Solage ballade S'aincy estoit (FSC B100), which begins in the one-sharp system but cadences in the two-flat system. Examples outside the French chanson repertoire include fourteenth-century English pieces such as the motets Suffragiose virgini and Herodes in atrio, a Gloria: Spiritus et alme setting and the Kyrie Cuthberte. One later continental example is the four-voice setting of Dufa's Se la face ay pale in Trent, Museo Provinciate d'arte, Castello del Buon Consiglio, MS 87, fols. 424v–425r, in which a b-mi clef is employed in the discantus secundus to imply the hexachordal system on C, G, D, with recta choice between F and F#.

²¹ For Apel and others, the concept encompasses two phenomena I differentiate, namely ‘conflicting signatures’ and this further variability. The term ‘bitonalityx’ is not particularly satisfactory in this context; I use it here because it is a central term and concept in the well-known debate over partial signatures. It might better be reserved for a phenomenon such as underfifth harmonizations. Apel also uses ‘polytonality’ synonymously with ‘bitonality’; Lowinsky talks of ‘vacillating’ (as opposed to ‘unified’) tonality; Hoppin regards the phenomenon as one of modes and pitch levels a fifth apart, or a kind of bimodality’. See Apel, W., ‘The Partial Signatures in the Sources Up to 1450’, Acta musicologica, 10 (1938), 5, 9–10CrossRefGoogle Scholar, and ‘A Postscript', Acta muskologica, 11 (1939), 40; Lowinsky, E., The Function of Conflicting Signatures in Early Polyphonic Music, Musical Quarterly, 31 (1945), 228, 238–9, 259–60;Google ScholarHoppin, R., ‘Partial Signatures and Musica Ficta in Some Early Fifteenth-Century Sources’, journal of the American Musicological Society, 6 (1953), 200, 202–4;CrossRefGoogle ScholarLowinsky, , ‘Conflicting Views on Conflicting Signatures, Journal of the American Musicological Society, 7 (1954), 182–3, 193–4, 198–9;CrossRefGoogle ScholarHoppin, , ‘Conflicting Signatures Reviewed’, Journal of the American Musicological Society, 9 (1956), 108, 114–5.CrossRefGoogle Scholar

²² See especially Hughes, Manuscript Accidentals, 47–52; Bent, M., ‘Musica Recta and Musica Ficta’, Musica disciplina, 26 (1972), 97–100; and Bent, ‘Diatonic Ficta’.Google Scholar

²³ Powers, ‘Tonal Types and Modal Categories’, and Judd, C. C., ‘Modal Types and Ut, Re, Mi Tonalities: Tonal Coherence in Sacred Vocal Polyphony From About 1500’, Journal of the American Musicological Society, 45 (1992), 428–67.CrossRefGoogle Scholar

²⁴ It should be acknowledged that there are, inevitably, disagreements among sources (and among editors) over signatures, but these are relatively few in number. To resolve them, I have not examined the sources of every chanson, relying instead on the published incipits, editions and critical reports in the Apel and Greene volumes along with the form of the incipits reported in RISM to make my decisions, which I will not individually review here. (To give a single example, Greene edits the virelai Le grant biaute (FSC V205 = VPMFC21: 41) on two-flat c while I follow previous editors Willi Apel and Nigel Wilkins in interpreting it as in the one-flat system.) In my opinion, some pieces (I am thinking of Calextone, Le mont Aon and Fumeux fume) lack a signature because of an unusual proliferation of accidentals, which are then separately flagged, and here I do normalize (for instance, taking Fumeux fume as in the one-flat system). Signatures in Machauf's chansons need closer examination than I could give them without full recourse to all the sources, so I have mainly followed the Schrade edition, which is conservative about interpreting source flats as signatures; hence, some songs (on F, bb and c) might be placed a step too sharp in my categorization.

²⁵ Not acknowledging the appearance of the two-flat c type in Machaut chansons at all, Apel regards ‘c-minor tonality’ as a post-Machaut trait (Apel, W., ‘The Development of French Secular Music During the Fourteenth Century’, Musica disciplina, 27 (1973), 49).Google Scholar

²⁶ Other repertoires can supply more examples of some of the unusual types. Taking F, for example, Matteo da Perugia's ballata Gia rete d'amor is in three-flat F (La Cappella musicale del Duomo de Milano: Le origini e il primo maestro di capella, Matteo da Perugia, ed. Fano, F. (Milan, 1956), no. 13)Google Scholar, and Thomas Fabri's rondeau Die mey so lieflic is in natural-system F (Strohm, R., Music in Late Medieval Bruges (Oxford, 1985), ex. 2)Google Scholar. Among the fifteenth-century rondeau in the Reina manuscript are three on two-flat f: Souffice toy, Mon cuer pleure, Resioisons et se prenons (A 15th-Century Repertory From the Codex Reina, ed. Wilkins, N. (n.p., 1966), nos. 13, 33, 34). The list could be vastly extended. See also note 33 below.Google Scholar

²⁷ See Reaney, G., ‘Modes in the Fourteenth Century, in Particular in the Music of Guillaume de Machauf’, Organices voces: Festschrift Joseph Smits van Waesberghe (Amsterdam, 1963), 142–3.Google Scholar

²⁸ Conventionally behaved, authentic low-lying F and G melodies (I am thinking here also of the monophonic Machaut virelais, for instance) reach out for the closest medials above, including 5 above. Five out of the eight low G authentic melodies have a medial cadence at the fifth above, on d.

²⁹ Egidius de Murino, Tractatus de musica (Coussemaker, Scriptorum de musica, HI, 128b): ‘Rondellus habet apertum ante, quando finitur in ut, et debet esse decima; et quando finitur in la, debet esse [Couss.: este] quinta, et retro dausum’.

³⁰ The use of in alpha-major shows up earlier in monophonic trouvère chansons at ouvert cadences and continues to be prevalent in the fifteenth-century rondeau at the medial cadence; these and other repertoires (e.g., later medieval Latin songs) require further investigation for the light they may shed on and as secondary cadence tones in ‘major’, especially for plagal melodies.

³¹ Of all those chansons in the one-flat system, just one rondeau on c, the anonymous (Galiot?) Se vos me voles (FSC R277), has medial e; this piece has a cantus signature flat in the Chantilly manuscript but acute bb should (or could) be raised to natural on every appearance. Of those in the two-flat system, just three ballades on c have e, all at the pre-refrain cadence (FSC B21, B123, B171).

³² Besseler, H., Bourdon und Fauxbourdon (Leipzig, 1950)Google Scholar, 43, an observation recently cited in Fallows, D., Dufay (London, 1982), 100;Google Scholar see also Berger, K., Musica Ficta: Theories of Accidental Inflections in Vocal Polyphony from Marchetto da Padova to Giosefo Zarlino (Cambridge, 1987), 244–5 n38.Google Scholar For examples, see Dufay's chansons Navré je suis (no. 34), Donnes I'assault (no. 70) and Ne je ne dors (no. 77), and his antiphon-motet Ave regina celorum (III).

³³ For a sampling that does not pretend to be exhaustive, see Machaut's ballade De desconfort (B8, a Bb piece); Solage's rondeau Fumeux fume (FSC 103, an F piece); the Johannes Cesaris ballade Bonté bialté (an F piece); and the Dufay chansons C'est bien raison (no. 16) and Ma belle dame (no. 31), both F pieces. These shifts also happen in G pieces such as the anonymous ballade Le mont Aon (FSC 159) and in Dufay songs such as Quel fronte signorille (no. 7), Je prens congie (no. 57) and Estrines may (no. 58). From other repertoires, see the fourteenth-century English Kyrie Cuthberte (a C piece) and cantilena Stella man's (an F piece), Dufay's song-motet Flos florum (an F piece) and Josquin's motet Absalon fili mi (a Bb piece). Finally, the Pycard Credo (The Old Hall Manuscript),

³⁴ D. Hughes, ‘A View of the Passing of Gothic Music’, 93, 94. Hughes further perceived the functional equivalence of the pairs of fourths on scale degrees and , and and , in relation to the final. In a table presenting his statistics, the ‘cadences made on or below the final, with supertonic or mediant in the tenor, have been transferred to’ or respectively (Ibid, 93).

³⁵ ‘Modal behaviour’, as I take it here, is a shorthand description of melodies and discant structures with characteristics that accord with the precepts of modal theory of the ‘pseudo-Greek’ or Italian variety. (See Meier, B., Die Tonarten der klassischen Vokalpolyphonie (Utrecht, 1974), 32–5, and Powers, ‘Mode’, 392–6.)Google Scholar That is, with various means of musical emphasis they must articulate, horizontally and vertically, octave species consisting of pentachord-tetrachord pairs (with the pentachord always located above the final), and display a significant role for the fifth scale degree as local melodic goal and boundary tone, and as longer-range melodic and harmonic cadential goal. In this sense, ‘modal behaviour’ emerges gradually between 1340 and 1480 in the French chanson and is an especially noticeable feature of the art of Dufay, as Leo Treitler has shown (Treitler, L., ‘Tone System in the Secular Works of Dufay’, Journal of the American Musicological Society, 18 (1965), 131–69).CrossRefGoogle Scholar Mode as such plays no part in the choice of scale degrees and or and as medial cadential goals; these conventions, in fact, can probably trace their origins to the monophonic idiom of the trouvères. And in regard to melodic facture, the earliest freely composed chansons explore many possibilities for melodic and cadential articulation of pitches besides the fifth above the final.

³⁶ For a succinct account of the association of modal quality and hexachord syllable (in particular, what is called the vox rule for finals), see Powers, ‘Mode’, 391–92. Given the obvious relevance of the passage on the vox rule in the Berkeley manuscript in time and place to the chanson repertoire, something more deserves to be said about it here. The treatise has to omit three doublings of function in the natural hexachord in order to give each clavis a single function, and several mistakes are made. Despite the effort at rationalization made by Ellsworth, The Berkeley Manuscript, 85–7, the following can only be considered blatant errors: putting sol B-mollis in modes 1/2; putting fa B-quadratum in modes 7/8; and omitting sol B-mollis from modes 7/8. In addition, sol naturalis is excluded from modes 1/2; la naturalis is excluded from modes 3/4; and fa B-quadratum and ut naturalis are excluded from modes 5/6. The treatise, if amended, would predicate G re-sol, D la-re-sol and A la-re as the voces for modes 1/2, corresponding to alpha-minor and beta-minor finals; A mi-la, E mi-la and B mi for modes 3/4, corresponding to the gamma-minor class; Bb fa, F ut-fa and C ut-fa for modes 5/6, corresponding to alpha-major finals; and C sol-ut and G sol-ut for modes 7/8, corresponding to beta-major finals.

³⁷ One reasonable fit is with a well-known theory from a far-removed corner of the musicological realm, namely the system of five anhemitonic pentatonic octave species, with their expansion to hexatonic and heptatonic octave species, devised by Annie Gilchrist as a modal theory for Anglo-American folksong. Gilchrist's scheme has the virtue of being both empirically founded and consonant with the pentatonicism proposed above as a way of describing the apparent bitonality of the chanson. Many pertinent concerns, such as the necessary distinction between tonic and final, and a rich research agenda - especially into the potentially finer gradation of types based on the relative strength and function of melodic scale degrees–are opened up by her work. A convenient exposition of Gilchrist's basic theory, on which I am relying, is found in Powers, ‘Mode’, 419–20. Gilchrist's pentatonic modes 1 and 2 correspond roughly to the majorminor beta pair (‘mixolydian’ and ‘aeolian’), modes 3 and 4 to the major-minor alpha pair (‘ionian’ and ‘dorian’), and mode 5 to the gamma-minor type (‘phrygian’), respectively.

³⁸ Reaney concluded from an examination of modes in the fourteenth century that there are in effect just one ‘near major’ and one ‘near minor’ (Reaney, ‘Modes in the Fourteenth Century’, 143).