Musica Ficta and Harmony in Machaut's Songs

Author(s): Thomas Brothers

Reviewed work(s):
Source: The Journal of Musicology, Vol. 15, No. 4 (Autumn, 1997), pp. 501-528
Published by: University of California Press
Stable URL: http://www.jstor.org/stable/764005 .
Accessed: 14/02/2012 15:35

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

University of California Press is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of Musicology.

http://www.jstor.org
 Musica Ficta and Harmony
in Machaut's Songs*
THOMAS BROTHERS

7The concept of harmony easily finds a place in

discussion of late-medieval polyphony, so we should not be shy about
using the word. The casual and universal sense of the word, describ-
ing something that has been carefully put together, with agreeable
results, did find its way into musical writings of the period. When we
speak of "harmony" in modern music theory we fold this static con-
ception into a more complicated dynamic one, with the main interest
being systematic organization of vertical sonorities. (But here we may
also detect the more universal definition, the formation of an agree- 501
able whole.) The medievals, too, developed a way to relate vertical
sonorities to one another. It would surely surprise them to discover
how much later centuries made of the practice they invented, though
it should not surprise us that the modern practice traces its origins
to this period, the source of so much that happened later. The late-
medieval practice provides both an antecedent to common practice
harmony and the most accessible point of entry into the style for the
modern ear. And within this practice there is an important role for
musica ficta, that is, pitches not included in the traditional Guidonian
gamut.
The phrase causa necessitatiswould have come most readily to an
informed medieval musician's mind, one imagines, if he were asked
about the role of musica ficta in harmony. Musica ficta is necessary,
theorists say again and again, because it makes possible the adjust-
ment of vertical sonorities so that they form perfect rather than di-
minished or augmented intervals. When they lay out the rudimentary
rules of discant, music theorists often say that false or invented music
Volume XV * Number 4 * Fall 1997
The Journal of Musicology C 1997 by the Regents of the University of California
* Parts of this article appear in my ChromaticBeauty in the
Late-MedievalChanson:An Interpretationof ManuscriptAcciden-
tals (Cambridge University Press, 1997); used by permission
of Cambridge University Press.
 THE JOURNAL OF MUSICOLOGY

is useful for making good consonances--which is another way of

saying good harmony.' The Anonymous of St. Emmeram states the
traditional explanation with a fresh metaphor:

Here he wishes to express the fact that not only true music suffices
for discant, but rather never could discant be perfectly known or
put together unless false music, that is, contrived music, held out its
hand as to a stepchildfor the putting together of it. And note that its
invention was a necessityin this art, and this is because without it we
were not able in any way to have a perfect knowledgeof the concords
in disparate melodies, according to a correct and true condition of
proportion in them. Let us choose this often, therefore, if it is nec-
essary,since without it nobody can reach the perfection of disparate
melodies in a praiseworthyfashion.2

In this paradigm we are dealing with a static phenomenon, the

formation of a concordant whole; in Hucbald's phrase, "Harmony is
the proper union of two notes. "3 Marchetto may be read as recogniz-
ing the transformation of this traditional, static conception into the
forward-looking dynamic one that will occupy us in this study:
502
Boethius writes, 'A dissonance occurs when two pitches struck to-
gether do not wish to be mixed. Each, rather, strives to flee,' that is,
each pitch seeks to go to the locationwhere it will produce a pleasant,
amicable,sweet mixture, that is, a consonance.4

Marchetto of Padua observes that armonia, euphonia, simphonia, and consonantia

are simply different words for describing the same thing. See The Lucidarium in arte
musicaeplanae, ed. and trans. by Jan W. Herlinger (Chicago, 1985), 206-07.
2 "Hic vult
exprimere quod non solum vera musica sufficit ad discantum, immo
nonquam discantus sciri perfecte poterit vel componi, nisi falsa musica, id est ficta, ad
compositionem ipsius manum porrexerit ad vitricem. Et nota, quod eius inventio fuit
necessitas in hac arte, et hoc est, quia sine ipsa in diversis cantibus perfectam concor-
dantiarum notitiam habere nullatenus poteramus, secundum rectam et veram in eis
habitudinem proportionis. Hanc igitur, si necesse sit, saepius eligamus, sine qua nullus
ad perfectionem cantuum diversorum poterit laudabiliter pervenire." De Musica Men-
surata: The Anonymous of St. Emmeram, ed., trans. and with commentary by Jeremy
Yudkin (Bloomington, IN, 1990), 274-75.
In this essay I shall concentrate on "propinquity" applications of musica ficta to the
exclusion of "necessity" applications. The latter is in much need of study, however. For
it is clear that simultaneously sounding diminished and augmented intervals are idi-
omatic in some repertories. Peter Urquhart has recently argued against the hegemony
of causa necessitatisand in favor of greater tolerance for cross relations in music of the
early sixteenth century; see "Cross-Relations by Franc-Flemish Composers after Jos-
quin" Tijdschrift van de Vereniging voor Nederlandse MuziekgeschiedenisXLIII (1993),
3-41. I take up the topic in connection with music ca. 1400 in "Sharps in Med&efu:
Questions of Style and Analysis," Studi Musicali XXIV (1995), 3-25.
3 See Herlinger, ed., The Lucidarium,204-05.
4 "Boetius: Dissonantia fit cum duo soni simul pulsi sibimet permisceri nolunt, sed
sibi quisque gliscit ire, hoc est quod uterque sonus ad locum ire cupit ubi est permixtio
 BROTHERS

But this is not quite what Boethius says:

Consonance is a mixture of high and low sound falling pleasantly

and uniformlyon the ears. Dissonance,on the other hand, is a harsh
and unpleasant percussion of two sounds coming to the ear inter-
mingled with each other. For as long as they are unwilling to blend
together and each somehow strives to be heard unimpaired, and
since one interferes with the other, each is transmittedto the sense
unpleasantly.5

Boethius views consonance and dissonance as static qualities, while

Marchetto is interested in the dynamic relationship between the two.
For Boethius, consonance is amicable, dissonance hateful; the two will
never meet. For Marchetto and his age, the imperfect consonance-a
special kind of dissonance, in the language used by this theorist-has
become very useful. From statements like "the sixth seeks out the
octave, and this rule always holds" (Ars contrapunctussecundumPhilip-
pum de Vitriaco),we get a clear sense of how the phenomenon relates
to the modern-day idea of harmonic progression. Similarly, from an
anonymous mid fourteenth-century English treatise: 503
The unison, because of its immobility,is called a perfect consonance.
... Imperfect consonances are chiefly named so by virtue of their
instability,for they move from one place to another and possess in
themselves no definite [numerical]proportion.6

The progression is characterized by tension and resolution, and it is,

in this way, analogous to our conception of the dominant-tonic ca-
dence.7 The theorists are describing a kind of syntax, the rules of

iocunda, amicabilis, et suavis, hoc est consonantia." See Herlinger, ed., The Lucidarium,
200-01.
5 Fundamentalsof Music, ed., and trans. by Calvin M. Bower (New Haven, 1989),
16.
6 "Unisonus
propter suam immobilitatem perfecta concordantia dicitur ... Pre-
sertim imperfecta concordantia ab instabilitate sua merito denominatur que de loco
movetur in locum et per se inter nullas certas invenitur proportiones." Quoted and
translated in Sarah Fuller, "On Sonority in Fourteenth-Century Polyphony: Some Pre-
liminary Reflections," Journal of Music TheoryXXX (1986), 44. Reference to Ars contra-
punctus secundumPhilippumde Vitriacogiven in Richard Crocker, "Discant, Counterpoint
and Harmony," Journal of the AmericanMusicological SocietyXV (1962), 12.
7 The historical connection between this harmonic practice and harmony in the
system of tonality is stressed by Crocker (Ibid., p. 17): "The formulas of the fifteenth
century, then, are indeed functional: they depend upon the two-note progressions of
discant. They also sound like the familiar progressions of 'functional harmony,' which
simply means that triadic functions and progressions develop in unbroken continuity
out of discant." Carl Dahlhaus also recognizes the connection: "And one of the basic
 THE JOURNAL OF MUSICOLOGY

which are easily perceived by the initiated listener and central to

fourteenth-century style.
Quickly, musica ficta came to play a role in this dynamic concep-
tion of organizing vertical sonorities, just as it continued to play a role
in the static conception. The inflections enhance the directional ten-
dency that imperfect consonances have acquired. Again, Marchetto,
the earliest source:

The diesis is a 5th of a whole tone, occurringwhen, for instance,any

whole tone is divided in two in order to color some dissonance such
as a third, a sixth, or a tenth striving toward some consonance.8

Marchetto's idiosyncrasies (a new division of the whole tone, and a

new sign, mentioned elsewhere, to notate it) do not obscure the pe-
riod's standard view. Fourteenth-century theorists recommend the
use of musica ficta to adjust an imperfect sonority moving to a perfect
one so that while one voice moves by semitone the other moves by
whole tone. French theorists describe the practice, but they say little
more.9 Prosdocimo de' Beldomandi (borrowing, in part, from Mar-
504 chetto) explains how it works:

Last, for understanding the placement of these two signs, round b

and square b ... these signs are to be applied to imperfectly conso-
nant intervals-the third, the sixth, the tenth, and the like-as is
necessaryto enlarge or diminish them to give them majoror minor
inflectionsas appropriate,becausesuch intervalsought sometimesto
be major and sometimes minor in counterpoint;and if you wish to
know the difference-when they should be majorand when minor-
you should consider the location to which you must move immedi-
ately after leaving the imperfect consonance; then you must see

ideas of contrapuntal theory from the 14th through the 17th century is that the vari-
ation of intervallic quality-the tendency of dissonance toward consonance, or of
imperfect consonance toward perfect consonance--forms the driving force behind
music's forward motion. A chain of sixths striving toward the perfection of an octave
differs of course in degree, but not in principle, from Rameau's progression of seventh
chords whose goal is a triad--an accordparfait." Studieson the Origin of Harmonic Tonaity,
trans by Robert O. Gjerdingen (Princeton, 1990), 22-23 and 29.
8
"Dyesis quinta pars est toni, puta cum aliquis tonus bipartitur propter aliquam
dissonantiam colorandam supple terciam, sextam, sive decimam tendendo ad aliquam
consonantiam." The Lucidarium, 140-41.
9 Ars contrapunctisecundumJohannemde Muris, ed. Edmond de Coussemaker, Scrip-
torum de musica medii aevi nova series a Gerbertinaaltera (Paris, 1864-76), vol. 3, esp.
59-60; Petrus frater dictus Palma ociosa, Compendiumde discantu mensurabili,ed. Jo-
hannes Wolf, "Ein Beitrag zur Diskantlehre des 14. Jahrhunderts," Sammelbdndeder
InternationalenMusikgesellschaftXV (1913-14), esp. 507-10o and 512-16; Parisian anon-
ymous of 1375 (Goscalcus?), ed. by Oliver B. Ellsworth as The BerkeleyManuscript:
Universityof California Music Library,MS. 744 (Lincoln, NB, 1984), 1 10-13-
 BROTHERS

whether the location you leave is more distant from that location you
intend immediatelyto reach, making the imperfect consonance ma-
jor or making it minor: for you should always choose that form,
whether majoror minor, that is less distant from that location which
you intend to reach, and you should, by means of the signs posited
above, make a major interval minor or, contrariwise,a minor one
major as appropriate. There is no reason for this other than a
sweeter-soundingharmony.Why the sweeter-soundingharmonyre-
sults from this can be ascribed to the sufficientlypersuasivereason
that the propertyof the imperfect thing is to seek the perfect, which
it cannot do except through approximatingitself to the perfect. This
is because the closer the imperfect consonance approachesthe per-
fect one it intends to reach, the more perfect it becomes, and the
sweeter the resulting harmony.'o

Following the lead of the Italian theorists, we could refer to this as

the "propinquity" application; through analogy with causa necessitatis
would come causa propinquitatis.
Today there is considerable momentum behind the position that
the propinquity application was conventionally applied as a perfor-
mance practice--applied, that is, even in the absence of a sign to 505
indicate it, and with such universality that the composer did not need
to indicate it. This position has deep roots in modern musicology
(to use an appropriate, nineteenth-century metaphor). It received a
strong boost from Hugo Riemann, who viewed the matter as one of
recovering a practice that was self-evident during the period, though
lost to us." Yet it is clear that for Riemann the matter was much larger

10 "Item ultimo
pro noticia collocationis istorum duorum signorum, scilicet b
rotundi et b quadri ... in vocum combinationibus imperfecte consonantibus, sicut sunt
tercia, sexta, decima, et huiusmodi, ponenda sunt etiam hec signa secundum quod
oportet addere vel diminuere in ipsas reducendo ad maioritatem vel minoritatem op-
portunas, eo quod tales combinationes in contrapuncto esse debent aliquando maiores
et aliquando minores; et si hanc diversitatem scire cupis, quando, scilicet, ipse debent
esse maiores et quando minores, considerare debes locum ad quem immediate accedere
debes post tuum recessum a tali consonantia imperfecta, et tunc videre debes an locus
a quo recedis magis distet a loco ad quem immediate accedere intendis, faciendo talem
consonantiam imperfectam maiorem an in faciendo ipsam minorem, quoniam illam
semper sumere debes que minus distat a loco ad quem immediate accedere intendis,
sive illa sit maior sive minor, et debes tunc facere ipsam per signa superius posita, de
maiori minorem vel e contra, scilicet de minori maiorem, secundum quod oportet,
cuius ratio non est alia quam dulcior armonia. Sed quare hec dulcior armonia ex hoc
proveniat potest talis assignari ratio satis persuasiva, quoniam si de ratione imperfecti
sit sui appetere perfectionem, quod aliter esse non potest quam per approximationem
sui ad rem perfectam. Hinc est quod quanto consonantia imperfect magis appropin-
quat perfecte ad quam accedere intendit, tanto perfectior efficitur, et inde dulcior
armonia causatur. Contrapunctus,ed. and trans. by Jan Herlinger, (Lincoln, 1984),
79-85.
"' Verlorengegangene Selbstversta'ndlichkeitenin der Musik des 15.-16. Jahrhunderts:
Die Musica Ficta; eine Ehrenrettung,Musikalisches Magazin 17 (Langensalza, 1907).
 THE JOURNAL OF MUSICOLOGY

than simply trying to recover details of a lost performance practice. As

Raymond Haggh observes, Riemann must be "readin the light of his
desire as an editor to make earlier music conform as much as possible
to a major or a minor tonality."12 There could be no better demon-
stration than Riemann's project of my point that late-medieval har-
monic practice provides the main point of entry to the style for the
modern ear. And one need not look very far to see how potent the
mix of "lost self-evidence" with conceptions of rudimentary tonality
has remained throughout the twentieth century.'3
But in spite of sustained twentieth-century interest in the phe-
nomenon, no theorist from the fourteenth century ever proclaims
or even alludes to the existence of an unnotated convention for the
propinquity application. There is no statement even remotely similar
to the one Tinctoris famously makes, that it is asinine to notate the
soft b when the context makes clear the need for it (the context, in this
case, being indirect tritones).14 Theorists typically discuss the appli-
cations of musica ficta with reference to the signs for soft b and hard
b,just as Prosdocimus does in the quotation given above. In fact, there

506
12
-Raymond H. Haggh, commentary to his translated edition of Hugo Riemann,
History of Music Theory,BooksI and II: Polyphonic Theoryto the SixteenthCentury(Lincoln,
1962), 396-97.
13 One need
only consider Riemann's special emphasis on the "Dufay epoch,"
which is "hardest hit by these incorrect judgments" of modern editors who are reluctant
to add accidentals to the notated songs, to recognize the lasting impact of this point
of view. Without too much effort, I think, one could trace reflections of this thought
through the work of Heinrich Besseler, Edward Lowinsky, and Howard Mayer Brown,
to mention only one spectacular and highly influential line of scholarship. In various
ways, each of these scholars takes the analogy between "common-practice" harmony
and the late-medieval practice too far in analysis of Du Fay's music. Quotation from
Riemann from History of Music Theory,p. 330. For Besseler, see Bourdon und Fauxbour-
don, revised edition by Peter Giilke (Leipzig, 1974); for Lowinsky, "Canon Technique
and Simultaneous Conception in Fifteenth-Century Music: A Comparison of North
and South," in Robert Weaver, ed., Essays on the Music of J. S. Bach and other Divers
Subjects:A Tributeto GerhardHerz (Louisville, 1981), 181-222; for Brown, Music in the
Renaissance (Englewood Cliffs, 1976), 27-59 (section on Du Fay).
14 Tinctoris, Liber de natura et proprietatetonorumChapter 8, Corpus Scriptorumde
Musica 22 (N.p.: American Institute of Musicology, 1975) vol. 1, 73-74; translated
edition by Albert Seay, Johannes Tinctoris: Concerningthe Nature and Proprietyof Tones
(Colorado Springs, 1976), 10-14. Pitch designations in this article are italicized, and
they follow late-medieval letter notation of the Guidonian gamut, with c indicating our
middle C, C the octave below, and cc the octave above. Capital letters are used when the
discussion does not involve a specific octave placement.
In his survey of fourteenth- and fifteenth-century theorists, Klaus-Jtirgen Sachs
concludes that the propinquity application was neither universally followed in practice
nor universally prescribed in theory. In Sachs's view, the propinquity application was
optional, in contrast to the necessity application. See "Die Contrapunctus-Lehre im 14.
und 15. Jahrhundert," in Frieder Zaminer, ed., Die mittelalterlicheLehrevon der Mehrstim-
migkeit, Geschichte der Musiktheorie V (Darmstadt), 199-20o8.
 BROTHERS

is no strong evidence from the theoretical discourse that any conven-

tions for applying unnotated inflections in polyphonic music existed
during the fourteenth century. At the least, this means that we should
make every effort to make sense of musical texts as they have been
transmitted to us. We should take seriously the possibility that the
texts are literally and completely--though perhaps incorrectly-
notated. It would be foolish to insist that there were no self-evident
performance practices, and that we should abandon any hope of re-
covering them. But it is important to realize how extensively inter-
pretations of polyphonic music have been conditioned by the heavy
emphasis on musica ficta as an aspect of performance practice. We
shall see how favored notions of conventions of performance practice
obscure subtleties of compositional practice that may be very impor-
tant indeed in our attempt to make sense of the impressive poly-
phonic achievement from the late middle ages.
The treatises we rely upon so heavily usually combine theory and
pedagogy. Often they lean more toward the latter than the former,
but on both counts the texts deal in basic matters. They cover the
rudimentary and universal fundamentals of music. For this reason,
they are of only limited use in higher-level interpretation, and this is 507
especially true, I think, for interpreting nuances in use of musica ficta.
For example, the treatises make clear that the propinquity adjustment
may be made with either sharps or flats. They do not tell, however,
how to choose between the two. The following progression, then, may
be adjusted in one of two ways (see Example 1). Margaret Bent and
Andrew Hughes have tried to find a way through this dilemma by
suggesting that codes of performance practice favored musica recta
(that is, the pitches of the traditional Guidonian gamut) over musica
ficta; following them, one would choose b-flat over g-sharp in this
example. But on the basis of a survey and critique of the theoretical
discourse, Karol Berger has argued against this line of reasoning.
Moreover, Berger gives evidence suggesting that, to the contrary,
theorists seem to favor mi inflections over fa inflections. Applied to
our example, his argument would yield g-sharp.15
Confirmation--together with substantial Ber-
from a of qualification--of In our
ger's insight emerges survey manuscript evidence.16
15 Berger discusses the matter, critiquing the position of Bent and Hughes, in
Musica Ficta: Theoriesof AccidentalInflectionsin Vocal Polyphonyfrom Marchettoda Padova
to GioseffoZarlino (Cambridge, 1987), 83-84, citing Bent and Hughes in note 64, p. 216.
Berger's review of evidence suggesting mi preference over fa preference is on pages
139-50.
16
Primarily that of the Machaut sources, but also evidence drawn from a survey
of the chanson during the period ca. 1275-1475; see my ChromaticBeauty in the Late-
Medieval Chanson, Chapter 2.
 THE JOURNAL OF MUSICOLOGY

EXAMPLE1. Options for inflecting the progression minor sixth to

octave.
la lb Ic

example, b-flat produces, in the language of solmization, a mi clausula,

a distinctive arrival marked by half step above the lowest sounding
pitch (solmizing a as "mi"). In modal terms, the progression suggests
phrygian or deuterus mode, and in modern-day terms, the formula
features the "lowered leading tone" or "lowered second." Patterns of
manuscript evidence make clear not only that lowered leading tones
hold a minority status but also that the formula is used in a special
way. From a survey of "nodal points" that define the bare outlines of
the formes fixes-ouvert, clos and final cadences in ballades and mid-
point and final cadences in rondeaux-one can see that, when he uses
them, Machaut places lowered leading tones exclusively in ouvert po-
508 sition.'7 From this we must conclude that the lowered leading tone
cadence brings a more tentative resolution relative to the raised lead-
ing tone. It is suited to a cadence in ouvert position in the same way
that arrival on an imperfect consonance is so suited.'8 The composer
of the anonymously transmitted Pour tant se j'ay le barbegrise'9 takes

'7 Lowered-second cadences never occur in final position in Machaut's poly-

phonic ballades, rondeaux and virelais, and they occupy ouvert position in twenty songs
(achieved either by signed accidentals or by locating the arrival on E). From the bal-
lades, see nos. 1, 5, 8, 10, 13, 14, 18, 23, 34, 36, 39, 41, and 42; from the rondeaux, nos.
3, 5, 6, 10, 17; and from the virelais, nos. 23 and 29. Numberings follow the edition of
Leo Schrade, The Worksof Guillaume de Machaut Polyphonic Music of the Fourteenth
Century, nos. 2-3 (Monaco, 1956). Another formula that Machaut uses to destabilize
an arrival on a perfect consonance in ouvert position is to approach a perfect fifth
sounded by the cantus via the lowered upper neighbor-for example, arrival on D/a via
b-flat, or a/levia f For this, see ballades nos. 4, 6, 40, and rondeaux 2, 7, 9, 11 and 12.
This formula, too, is completely avoided at final cadences. These observations are based
on a survey of the Machaut sources. Of course, it is hardly a straightforward matter to
identify these events, given the two central textual problems (common to all study of
accidentals throughout the fourteenth and fifteenth centuries) of variants and duration
of the sign's effect. In a case like this, one may rely on the strength of the pattern to
work through these problems.
18 Ouvertcadences feature cantus and tenor sounding a held imperfect sonority in
the following songs by Machaut: Ballades 3, 7, 15, 16, 20, 21, 22, 25, 26, 28, 32;
Rondeau 18; and Virelais 26, 28, 30, 31.
19 Oxford 213 f. 47. Modern edition in Gilbert Reaney, ed., Early Fifteenth-Century
Music Corpus Mensurabilis Musicae no. XI, vol. 4 (American Institute of Musicology,
1977), 31-32.
 BROTHERS

advantage of this distinction in an unusual way. Typically, the mid-

point arrival in a rondeau will be located a step (or perhaps two steps)
above the final cadence. In Pour tant sej'ay le barbegrise, midpoint and
final cadence both feature arrival on D/ald. The two cadences are
distinguished from one another by the respective propinquity appli-
cations that lead up to them. For the midpoint cadence, e-flat is signed
and for the final, c-sharp. Instead of marking an ouvert-closstructure
through different points of arrival, such as would impart forward
direction and, as a result, strength to the final cadence, the composer
manipulates the propinquity application, taking advantage of the dis-
tinction between lowered second and raised seventh, to achieve the
same result.
The general point to draw from these observations is that acci-
dentals may be used to define a hierarchy of harmonic gestures. The
raised leading tone is stronger than the lowered one, and this leads to
association of an ouvert quality with the latter. (One could analyze the
emergence of a "phrygian affect" toward the end of the fifteenth
century in this way, I think; in any event, it is clear that harmonic
instability is an important part of that affect.) So when Berger states
(p. 143) that "the fourteenth-century evidence, and in particular that 509
of Marchetto and Petrus, argues . . that where a choice of the leading
tone had to be made, the sharp, and not the flat, was the preferred
solution," my qualification would be that the sharp is the preferred
"solution" (backing off from this word because in Berger's statement
it points to choices made as part of a performance practice) only in the
sense that it is stronger and more commonly used. The flat is second-
ary because it is held for special effect. Patterns of use in the nodal
points of Machaut's ballades and rondeaux are consistent enough to
support the idea that the hierarchical distinction is always valid, even
for interior cadences, where formal conventions do not direct the
analysis; in any event, this is a premise that we can work with.
Sarah Fuller's work on harmony in ars nova polyphony provides a
good base for developing further our understanding of how acciden-
tals may be used to define hierarchies of harmonic gestures.20 Build-
ing upon the basic precept that progressions from imperfect to per-
fect consonances are associated with tension and resolution, Fuller
seeks to coordinate a range of sonority types with relative degree of
stability and instability. This allows her to rank progressions along a
scale ranging from "neutral or non-committed to definitely directed"
(p. 51). At one end of the spectrum stands the "doubly imperfect"

20
"On Sonority in Fourteenth-Century Polyphony: Some Preliminary Reflec-
tions," Journal of Music TheoryXXX (1986), 35-70.
 THE JOURNAL OF MUSICOLOGY

sonority (for example, DIF-sharp/b, "doubly imperfect" because the

third and sixth are combined), at the other the "doubly perfect" (for
example, D/A/d, doubly perfect because of the combined fifth and
octave). In between fall mixed sonorities (for example, DIF/a). Inflec-
tions are important: for example DIF-sharp/bis more firmly directed
than DIFIb. Not only that, the very existence of an inflection may
signal an important event-this is an insight that the present study will
try to take full advantage of. It is not Fuller's intention to identify a
rigid syntactical system. Rather, she seeks to show how the composer
manipulates a few fundamental principles involving harmony in or-
der to shape a piece. The events take their meaning from context as
much as they do from the syntactical precepts just sketched.21
That context involves voice-leading, duration, mensural position,
and, above all, organization of pitch in the piece as a whole. As Fuller
demonstrates in analysis of motets by Machaut, these factors deter-
mine how one should parse a sonority and a progression. Machaut's
Nes que on porroit,22 a ballade in which inflections do not play a terribly
important role, shows how this approach can help elucidate subtle
aspects of harmony, manipulated by the composer in the service of
510 overall design.
A good place to begin analysis of the musical side of any ballade
is with the ouvert, clos, and final cadences (see Example 2). The voice
leadings are unusual: cantus and tenor proceed in similar rather than
contrary motion, and they move by leap rather than by step. In terms
of where the arrival pitches are located, on the other hand, these
cadences follow the standard pattern in which clos and final match,
with ouvertlocated one step above (D- C- C). Also standard is the fact
that this pattern is forthrightly coordinated with the organization of
pitch throughout the piece. C is clearly emphasized, both melodically,
through the shape of the elegant cantus, and harmonically, through
emphasis on C/c/c and C/IG/c.Hence, there is no ambiguity about how
to parse the open and closed status of the three main cadences. Within
the body of the song there are a few clearly directed progressions that
highlight both C and D. Notably, several occur in weak mensural

21
Fuller: "But in actual practice, the association of objective types with specific
qualities can only be considered loosely normative, for multiple factors can act in a
compositional context to qualify the effect and function of a sonority." (p. 44) and:
"The characterizations advanced here are to be considered as useful guides to norms,
not fixed stereotypes to be forced upon musical events." (p. 45)
22
Modern edition of the song by Leo Schrade, The Worksof Guillaumede Machaut,
V.II, 122.
 BROTHERS

EXAMPLE 2. Main cadences in Machaut's Nes que on porroit, after Leo

Schrade, The Worksof Guillaumede Machaut, v. 2 p. 22.
18 Iovert cadence rToscadence( = finalcadence)
CANTUS
•I
F~ IV !

TENOR N_0 F

position (for example, the arrivals on D/ald in measure 4 and CIG/cin

measure 8). I would acknowledge Fuller's insistence that context must
direct the analysis of harmonic events by suggesting that, in these
cases, mensural position undermines harmonic progressions that are
otherwise clear and strong. And, indeed, a certain softening quality
surfaces throughout the piece. In gestures like these given as Example
3, the voices are slightly out of sync with one another, as if to decorate 511
the harmonic progression.
We should think of the unusual voice leadings at the three main
cadences in light of these events. Strong progressions in the classic
paradigm feature contrary motion by step between the two main
voices. The downward fifth leap in the clos and final cadences of Nes
que on porroit causes a superficial resemblance to the dominant-tonic
progression of common-practice tonality, and for that reason one may
be inclined to think of such a leap as bringing harmonic strength; that
would be incorrect: rather, these progressions are weaker than the
typical strong cadence of the fourteenth century. The musical atmo-
sphere is conditioned first and foremost by the leisurely, I flow of the
cantus. In light of the poem, one might characterize the musical mood
as nonchalant or even wistful. But what is most striking about the
piece is not musical expression of the poem, which is, after all, of a
standard type. Rather, we should highlight the persuasive coordina-
tion of musical events. Isolated components reinforce one another in
the service of a single musical mood.

From this analytical base, we may consider two bal-

lades by Machaut, Biaute qui toutesautrespere and De toutesflours, both
of which show the composer using accidentals to manipulate harmony
and shape musical design.
 THE JOURNAL OF MUSICOLOGY

EXAMPLE3. Measures 2 and 6 of Nes que on porroit

2 6
Cantus

Contratenor
-

Tenor

Biaute qui toutes autres pere (Example 4)23 is built around an ex-
traordinary cadential pattern in which the three nodal points each
emphasize the same pitch. The composer keeps D as the lowest pitch,
and he distinguishes ouvert and clos from one another only by the
subtle means of contrast between perfect fifth (D/a, for ouvert) and
512 perfect octave (D/d, for clos and final). (In 1336, Petrus Frater dictus
Palma Ociosa codifies a hierarchy of intervals in which octaves are
more stable than fifths.24) Furthermore, all three cadences are weak,
since, in each case, the tenor moves by leap and there is no leading
tone in the cantus (recall Nes que on porroit). It is a provocative com-
bination: the composer locates the three main cadences on the same
pitch, as if to negate large-scale forward direction, and he also makes
them weak, as if to say that the arrivals do not really conclude what
has come before.
Or at least that is one way to view the situation, and I shall try to
suggest how these cadences, together with signed accidentals, play a
role in the organization of pitch through the whole piece. Biaute qui
toutesautres pere makes little sense if we follow the precepts of modal

23 Modern edition by Leo Schrade, The Worksof Guillaume de Machaut, V.II, 74.
The song is transmitted as a duet in four of the "principal"-i.e., those sources devoted
exclusively to this composer's work-Machaut sources (Machaut C, Machaut A,
Machaut F-G, Machaut Vg), then with an added contratenor in Machaut E (an inferior
and later principal source) and Utrecht 37 (Utrecht, Universiteitsbibliotheek, 6 E 37;
a late secondary source). On the inferior quality of added contratenors in Machaut E,
see Wolfgang D6mling, Die mehrstimmigenBalladen, Rondeaux und Virelaisvon Guillaume
de Machaut: Untersuchungenzum musikalischenSatz, Mtinchner Veriffentlichungen zur
Musikgeschichte (Tutzing, 1970), 74-75. I argue for the legitimacy of concentrating on
the cantus-tenor pair to the exclusion of contratenors when dealing with problems
surrounding accidentals in "Sharps in Medie fu: Questions of Style and Analysis."
24 Compendiumde Discantu Mensurabili, Wolf, ed., 512.
 BROTHERS

EXAMPLE 4. Machaut's Biaute qui toutes autres pere, after Machaut

MS Vg.
Staff breaks and clef changes in MS Vg are indicated in this edition by
regular "bar lines" through the staff, and signatures, if any, for the
new staff of the source are given after the bar; the signature is then
repeated on each staff of the edition, following the original manu-
script. For example, there is a clef change with accompanying b-flat
signature in the cantus at measure 36, and this is followed by a staff
break, also with b-flat signature, at measure 38; both are represented
by bar lines through the staff, with the b-flat signatures indicated. If,
however, a signature flat is canceled by a contradictory sign on the
staff of MS Vg, then the signature is subsequently dropped in the
edition (for example, b-natural in the tenor at measure 25 causes b-flat
to be dropped from the remaining staffs of the edition.) Mensurstriche
mark a regular flow of perfect breves; the song famously uses perfect
tempus and imperfect tempus simultaneously in the concluding mel-
ismas of each section. MS Vg (which I am viewing through a poor
microfilm) shows two b-flats notated at the beginning of the cantus,
apparently in two different hands. Yet, there appears to have been
plenty of room allowed for both signs. The principal Machaut sources
513
agree on most of the accidentals presented in this edition. The main
question about the musical text would seem to be the presence or
absence of b-flats in the cantus, measures 5-9 (I would argue that
b-flat still holds).
5

[Cantus] Biau-te - qui tou - tes au - tres pe-re

Dou-ceur fi - ne a mon goust a-me-re

Tenor
10

En - vers moy di-verse et es-tran

Corpsdi - gne de tou - te lo-an

15

T1
ge,
.
 THE JOURNAL OF MUSICOLOGY

4. (continued)
EXAMPLE
F2. 2
0.

•.. d'a-y
vis acuer
Sim-ple mant
ge,

30

514
poura-mermor- ray.

theory, in which analysis begins with the final.25 For aside from the
three main cadences, which emphasize D, G is emphasized as the most
important pitch. As a result, the piece is built in a contradictory way,
in which nodal points and the design of the interior flow stand at
odds with one another. Perhaps this explains the relative weakness of
5 As noted by Jehoash Hirshberg, "Hexachordal and Modal Structure in
Machaut's Polyphonic Chansons," Studiesin Musicologyin Honor of OttoE. Albrecht(Kas-
sel, 1980), 26. Hirshberg discusses the limitations of modal analysis for Machaut's
chansons, and his analysis of Biaute qui toutesautrespere (26-27, referred to as "Ballade
4") agrees in many details with that presented here. I would not feel the need, as he
does, however, to think of the song as "essentially a composition in G Dorian, although
its two sections end on D." Also absent from Hirshberg's analysis is the notion that
"lowered-second" cadences are weaker relative to "raised-seventh" ones. And finally, to
mark the difference in our analyses, I would cite Hirshberg's recognition of sharps
This is a revealingwayto sumup biases
"neededto improveintervalprogressions."
about harmonic conventions that are mixed together with assumed conventions of
practice;unconventional
performance areseenas needingimprovement.
progressions
 BROTHERS

the main cadences: in this context of contradiction, they seem not to

conclude what has gone before. And the triple repetition supports the
idea that the cadences stand outside of the main flow, for the repe-
tition negates the sense of large-scale forward direction that one ex-
pects from an ouvert-clos-finalpattern. D is associated with closure by
virtue of being repeated at the ends of sections, rather than by actu-
ally satisfying a goal implied through the polyphonic flow.
G is emphasized as the main pitch through a series of melodic and
harmonic events in Part 1 (speaking in terms of a two-part form, with
the first part repeated). The opening sonority is G/d. And the shape of
the cantus helps to make it clear that the G of measure 7 is a point of
arrival, the goal of the slightly circuitous descent from the opening d.
Even though this arrival in measure 7 splits line 2 of the poem, it
marks the end of the first main musical phrase, in which a at the
downbeat of measure 4 represents the ouvert mid-point of the phrase,
G of measure 7 its clos ending. Musica ficta helps define this structure:
the tenor's E-flat (m. 3) provides the lowered-second leading tone that
defines the ouvert quality of the mid-point arrival on D, while, at the
same time, F-sharp in the cantus points to a stronger arrival on G.
(The problem of how long accidentals remain in force is always trou- 515
bling. Here I assume that F-sharp in the cantus holds through mea-
sure 6; F-sharp in the tenor would seem to confirm this assumption.)
The sense that measure 7 marks a strong arrival is confirmed by
what happens next, for the cantus ascends back to d (mm. 7-9), where
the two voices are both poised precisely as they began, on G/d; this is
a fitting way to launch the splendid melisma of measures 9-16. In this
melisma the tenor's syncopations provide steady tension, rhythmic
and harmonic.26 G is again emphasized by the fact that this disso-
nant, polymensural passage clears out on F-sharp (m. 15), which es-
tablishes a clear sense of direction; g is also emphasized by virtue
of its position as the high boundary, obtained in m. to. And when it
perches on a for the ouvert cadence, the cantus seems to predict the
traditional step-wise relationship between ouvert and clos endings,
a-G. Part 2 opens with a digression through several highlighted
pitches (C, B-flat, C), a typical procedure for ballades. Eventually, the
importance of G is reaffirmed at another important structural mo-
ment: the sonority that serves to announce the refrain is F-sharp/a (m.
28). "Rounding" (repetition of music from the end of part 1 at the end

26 Richard Crocker has

pointed out to me how the hidden structure of this pas-
sage becomes apparent if the tenor is shifted one semibreve backward-that is, if the
rest in m. 9 is dropped, and the tenor's G is placed under the D of the superius, etc. This
produces a series of consonances between cantus and tenor in proper discant fashion.
 THE JOURNAL OF MUSICOLOGY

of part 2) brings about a slight revision of the polymensural melisma

and restatement of tle clo,, ending.
Biaute qui toutes aiutrespere cannot be reduced to a single, domi-
nating pitch. In an ambiguous design, the directional and hierarchical
signals provided by accidentals have an important role.27 The song
appears to fall relatively early in Machaut's spectacular period of in-
terest in the ballade; perhaps the restless, experimental flair signals a
breakthrough in which the composer flexes his command of the basic
rhythmic and harmonic components of the ars nova. Some seventy-
five years later, Du Fay works with a similar conception. In the ballade
Je me complains, d is emphasized, and c-sharps strengthen the sense
that it is the main pitch.28 Yet the three main cadences of the ballade
end on a/aa (the ballade lacks alternative first and second endings for
part 1). c-sharps sound above A three times to form "half cadences"
(mm. 7, lo, and 28), confirming the association of A with instability.
The traditional cantus/tenor framework is abandoned in favor of
equal-range voices marked primus, secundus and tertius. Traditional
discant cadences are irregular and scattered; and, as if to compensate,
Du Fay works out a plan of clearly directed half-cadences, regularly
516 placed strong sonorities built on D, and weak arrivals at the main
nodal points on a/aa. As in Biaute qui toutesautrespere, a few carefully
placed accidentals play an important role in the contradictory design.

De toutesflours (Example 5) draws out the central

problem of unnotated conventions for the propinquity application,
mainly because the musical text, which is stable through the principal
sources, contradicts modern assumptions about those conventions.

'7 Compare Biaute qui toutesautrespere withJe ne cuit pas (Schrade, ed., p. 85). This
latter ballade is remarkable for its quantitative emphasis on a, which, in spite of the
emphasis, is clearly established as a secondary pitch through propinquity b-flats. Aside
from the clos and final cadences, D is used sparingly, though it is prepared by the
tentative arrivals on A and by a few important c-sharps. Commentpuet (Schrade, ed., p.
154) resembles Biaute qui toutesautrespere in significant ways. The midpoint cadence of
the rondeau (D/F-sharp/a) emphasizes the same pitch as the final cadence (D/a/ld).A few
inflections lead the piece quickly through various areas: D is established immediately,
through c-sharp, but b-flat, emphasizing a, and then b-natural, emphasizing c, quickly
follow. (Schrade's c-natural in measure 3 is surely incorrect; several sources make it
clear that the sign is b-flat, which makes good sense and stands as a more characteristic
inflection, since cancellations of c-sharp are rare.) Just as quickly, D returns for the
conclusion of the first phrase. In phrase 2, inflections in the descending sequence-a
sequence not unlike that of the melismatic one from Biaute qui toutesautrespere-carry
on the steady instability. The nod toward G in this phrase provides a way to hear
D/F-sharp/a at the midpoint. The last few phrases bring the song through C and G
before it finds its way back to D. This song shows an "essential" contratenor that fully
participates in the harmonic organization of the piece.
'2 Modern edition in Heinrich Besseler, ed., Guillaume Dufay: Opera Omnia, Cor-
pus mensurabilis musicae, vol. 1, no. 6 (Rome, 1951-1966), 29.
 BROTHERS

EXAMPLE 5. Machaut's De toutes flours

The transcription follows Fuller, "Guillaume de Machaut: De toutes
flours," pp. 59-61, with accidentals adjusted to conform to Machaut
MS A. Although it is difficult to tell for sure, it appears from my
microfilm copy of this source that D-flat is signed in the tenor for the
clos ending. If so, the accidental is most likely non-authorial.
5

[Cantus] De [De]
Ga [Ga-]

Contratenor

Tenor

10

I i;i I
u 1
I ir o

tou - tes flours n'a voit et de tous 517

stes e - stoit li seur plus et de -

15 20

fruis en mon ver - gier fors u-ne seu- le ro

struis par For- tu - ne que du-re-ments'o-po

25
A

II IrI I
 THE JOURNAL OF MUSICOLOGY

EXAMPLE
5. (continued)
Si. 30 12.
12.35 35

se - se Con - tre _ ce-ste dou-ce flour pour

40
A
II

a - ma - tr sa cou - louret s'o.- dour mais se cueil - lir

45 50
518

la voy ou tre - bu - chier Autre a - pres

55_ 60

li ja- mais a - voir

A65

ne quier
,, _ 1 _ • ,I 1 i i i I
 BROTHERS

Dropping the assumptions, we may read the text as a sophisticated

display of Machaut's interest in manipulating the harmonic language
of his time. The song has been analyzed by Fuller, whose work on
harmony, as we have seen, provides one base from which we may
interpret accidentals. The perspective on accidentals offered by the
present study, however, leads to an interpretation of the song and
its transmission that differs fundamentally from that pursued by
Fuller.29
The song is transmitted by five principal and four secondary
sources. Fuller draws attention to one particularly vexing situation (m.
15 of the edition):

Some problems are especially thorny. For instance, at the end of

phrase two, on vergier(breve 15) a notated B flat in the cantus in two
principalsourcescontradictsthe normal inflection (majorsixth D-B
natural to octave C) expected at a cadence and specified at other C
endings in this song. In other copies where B flat stands as a 'signa-
ture' in the cantus, the flat is not specified. Trained performers
working from these sources would surely sing B natural to provide
the conventionalcadence progression,to fulfill the resolution of the
B naturalat breve 14, and to be consistentwith the other C cadences 519
in the piece. Does the notated B flat on vergiersignify an express
desire on someone's part to override the conventionalperformance
mode, or is it a scriballapse, a premature indication of a restored
'signature'B flat?Although a good case can be made for the second
position, based on specific disposition in the two principal sources,
patterns of voice-leadingand Machaut'snormal practiceelsewhere,
the other view cannot be completely ruled out.30

Admitting that our understanding of fourteenth-century perfor-

mance practice is not as solid as we would like it to be opens up the
possibility of accommodating contradictions and exceptions. In an

29 Sarah Fuller, "Guillaume de Machaut: De toutesflours," in Models of Musical

Analysis: Music Before 1600, ed. Mark Everist (Cambridge, Mass., 1992), 41-65. A
triplum is given in two sources; see Fuller, 62-64 for four-voiced transcription. I agree
with Fuller's view that "This triplum is doubtless a later accretion to the three-voice
setting and was probably composed by someone other than Machaut" (41). See
Friedrich Ludwig, Guillaumede Machaut: MusikalischeWerke(Leipzig, 1926-29), V.II, 48
for a concordance table, including references to two lost sources, Paris, B. N. n.a.f.
2319o, no. 27 (the "Tremodlle manuscript"), and Strasbourg, BibliothZequeMunicipale,
MS 222 C.22, no. 168. There is also a keyboard arrangement of De toutesflours in MS
Faenza, Biblioteca Comunale, Cod. 117, ff. 37v/38v. The arrangement is quite free,
yielding the source of little apparent value for our discussion of the transmission of
accidentals.
30 Fuller, "Guillaume de Machaut: De toutes
flours," 57, n. 6.
 THE JOURNAL OF MUSICOLOGY

idiosyncratic piece, especially, it is good to take seriously the possibility

that the text should be taken literally, without adding or subtracting
accidentals.31
Measure 15's b-flat is interesting for two reasons: it cancels the
b-natural in the cantus of two measures earlier, before this pitch has
a chance to reach the implied c; and it undermines the arrival on c in
measure 16 by creating a whole step instead of a half-step leading
tone. It is easy to understand how the flat could have been dropped
by a skeptical scribe (and Fuller shows herself to be one when she
drops it from her edition of a source which includes it). But the
transmission of the sign is not weak. It is firmly communicated in the
principal sources, and it is dropped only in the secondary sources
(Table 1).32 That all of the principal Machaut manuscripts are corrupt
at measure 15 while the later and secondary manuscripts give an
authoritative reading seems highly unlikely.

31 As long ago as Johannes Wolf in 1904 and as recently as Bettie Jean Harden in
1983, scholars have taken the position that accidentals in Machaut's music are com-
520 pletely notated (for Wolf, the exceptions being some cadential inflections); see Wolf,
Geschichteder Mensural-Notationvon 1250-1460 (rept. Hildesheim, 1965), 174-75; and
Harden, "Sharps, Flats and Scribes: 'Musica Ficta' in the Machaut Manuscripts," (Ph.D.
dissertation, Cornell University, 1983), passim. Harden observes (3-5) that "when re-
ferring to introducing alterations the theoretical treatises generally transmit rules for
composers, not for singers, except in their discussions of improvisation. These treatises,
while often elementary, do not treat the subject haphazardly: although they may be
incomplete in their treatment of alterations, those that get far enough to mention the
matter at all seldom bring up the possibility of unwritten accidentals. They treat acci-
dentals as alterations which by their very nature must be written. Singers are told how
to read the signs they may encounter and how to make sense of preexisting music; they
are not instructed to make alterations when the signs are absent. . . . Most of the time,
chromatic inflections additional to the written ones would be an unwarranted imposi-
tion on a musical fabric that is already complete."
32 Four of the principal Machaut manuscripts carry the sign explicitly notated at
this place. The fifth (Machaut MS F-G) has b-flat notated at the beginning of the staff
as a signature, indicating the inflection of all pitches on that line or space. There is no
evidence that signature-accidentals are weaker than internally signed accidentals, or
that they are anything other than inflections governing an entire line or space (there is
no evidence, for example, that they serve more casually, as loose guides to solmization).
The Parisian Anonymous of 1735: "whenever any of these signs for the coniuncta is
placed at the beginning on whatever line or space, all syllables on that line or space
ought to be sung according to the sign placed at the beginning--unless it is removed by
something more particular in the middle or elsewhere .. ." ("quandocumque aliquod
istorum signorum pro coniuncta ponitur in principio regule vel spacii cuiuscumque,
omnes voces illius regule vel spacii cantari debent virtute illius signi in principio positi,
nisi per aliud specialius in medio vel alibi hoc tollatur ...; trans. and ed. Oliver B.
Ellsworth, The BerkeleyManuscript, 52-53)-
For identification of the Parisian Anonymous of 1375 with the composer Goscal-
cus, see Klaus-Jiirgen Sachs, Der Contrapunctus,p. 184; Sachs's suggestion is supported
by Anna Maria Busse Berger, Mensuration and Proportion Signs: Origins and Evolution
(Oxford, 1993), 175-
 BROTHERS

Similar events occur in two songs by later composers. Solage uses

the same b-flat to weaken a cadence to c in Le Basile. The resemblance
to De toutesflours is striking (see Example 6). It is more than ironic to
discover that this flat, too, has been dropped from a modern edition,
presumably because the editor was skeptical about it. But the event
makes sense as a component of large-scale musical design: it forms
part of a gradual destabilization in which the polyphony flows away
from the firmly established C of the beginning; C returns with the clos
ending of part 1. Later, Johannes Cesaris uses a similar gesture in his
ballade Bont6 biault6 (see Example 7). Again the formal logic is clear.
Moving to cc through bb-flat weakens the arrival (as does the con-
tratenor's e, bringing imperfection), as if to undermine the strong
discant progression. The tentative arrival, implying a firmer one to
come, is well-suited to a cadence in this position, for the chord serves
to announce the ballade's refrain (recall m. 28 of Biaute qui toutesautres
pere, where an imperfect consonance was used for the same purpose).
The refrain that follows brings the piece back to its main pitch, F.
It is unlikely, I think, that all three of these texts are corrupt and
likely that they show the same interest in using accidentals to shape
the organization of pitch. Rather than imagining, as Fuller does, that 521
trained singers knew how to "correct" unconventional cadences, I
imagine them being especially skillful in negotiating the unusual
events as written. If, as is surely true, a raised leading tone makes an
arrival stronger-the theorists all agree on that-then its absence
makes an arrival weaker. If a composer wishes to make hierarchical
distinctions between various arrivals on the same pitch, then he may
accomplish this through selective use of the propinquity application.33
(Another, though not necessarily equivalent way would be to sign
some with lowered second and some with raised seventh, as we have
seen in Pour tant se j'ay le barbegrise.) Machaut makes the event even
stronger by leaving the previous b-natural dangling. We should not
expect to find this situation described in the discant treatises. But one
way to ground the event in contemporary theory would be through

33 Bettie Jean Harden independently comes to a similar position when she sug-
gests that the composer may create cadential hierarchies via the presence and absence
of accidentals. See "Sharps, Flats and Scribes," 231-34, where the ballade Gais etjolis is
discussed. Harden is led to this position not through Fuller's work on harmony (which
was not available in 1983), but by taking seriously the idea that accidentals in the
principal Machaut sources are literally and completely notated. I thank John Nadas for
drawing my attention to this dissertation.
 THE JOURNAL OF MUSICOLOGY

EXAMPLE6. Mm. 23-26 of Solage's Le Basile

After Gordon Greene, ed., French Secular Music: Manuscript Chantilly,
Mus`e Cond6564, Polyphonic Music of the Fourteenth Century, vol.
19 (Monaco, L'Oiseau-Lyre 1982), 97.
23
Triplum

CantusI

Contrateneor r

Tenor _C

EXAMPLE 7. Mm. 25-26 of Cesaris's Bont6 biault6

After Gilbert Reaney, Early Fifteenth-CenturyMusic 1, Corpus Mensu-
522 rabilis Musicae 11 (N.p.: American Institute of Musicology, 1955),
30-32.
25 3 3

CANTUS2

TENOR

CONTRATENOR
5:

the notion of manipulating progressions so that intervals do not pro-

ceed according to their "natural" inclinations. Several fourteenth-
century theorists legitimize this technique and defend it as a means of
gaining variety (causa variationis).34
A basic principle of textual criticism, one that remains, however,
unfamiliar in study of musica ficta, may explain the instability of De
toutes flours's musical text. The dictum difficilior lectio potior follows

:4 On causa variationis, see Klaus-Jtirgen Sachs, Der Contrapunctusim 14. und 15.
Jahrhundert: Untersuchungenzum Terminus, zur Lehre und zu den Quellen Beihefte zum
Archiv fiir Musikwissenschaft vol. 8 (Wiesbaden, 1974), 66-69.
 BROTHERS

from the assumption that idiosyncratic readings tend to becomecon-

ventionalized over time.s5 Table 1 charts a process in which the mu-
sical text becomes increasingly trivialized. The table implies a concep-
tual (not necessarily a chronological) sequence, in which the unsettled
quality of the piece is gradually washed away. The principal sources
notate a highly discursive piece, with many unexpected turns. Of the
secondary sources, Florence 26, PN 568 and Modena preserve most
of the inflections notated in the principal sources; they lack mainly the
b-flats of mm. 15-22. For the scribe of PN 6771, the piece is hum-
drum: e-flat disappears from the superius entirely (not only in part 1,
as indicated in the chart, but throughout the piece), and after the first
phrase the manuscript preserves few b-flats. Obviously, if we knew
only this version of the song we would have no clue about essential
ingredients of the composer's conception-and one can only wonder
how many times that has actually happened throughout the surviving
polyphonic corpus.
The troubling b-flat easily finds a place in analysis of the song.
The cantus opens and ends the first musical phrase (mm. 1-8) on c,
and the phrase closes with what looks like a traditional cadential ges-
ture to c, with the voices moving in contrary motion from a doubly- 523
imperfect sonority to a double-perfect sonority. Yet, a b-flat signature
governs this pitch (all of the sources agree), so the arrival is relatively
weak. The sources do not contradict this signature until measure 14,
where b-natural is signed (all but two of the sources agree). And, as we
have just seen, several sources immediately contradict the b-natural,
so that the cadence in m. 16 lacks a leading tone, just as the cadence
in measure 8 does.
The composer may have good reasons for not wanting all arrivals
on c to be the same. During the first phrase, several events weaken the
sense of any clear hierarchy among the pitches. The cantus begins on
c, but with the support of F in the tenor. And in measure 6, F is again
emphasized, now as the lower and prominent boundary of the cantus;
the cantus reaches this F, significantly enough, by turning away from
c through b-flat. As often happens in late-medieval songs, pitch is
organized dynamically and unpredictably (one reason why it is mis-
leading to label this or that phrase or piece as "in" a certain modality
or tonality).
Phrase 2 brings about close juxtaposition of b-natural (mm. 9
and 14) with e-flat (mm. lo and 13-and, presumably, for the whole

35 Lee Patterson's summary of this principle -"scribes habitually trivialize their

texts" -seems especially apt for the present situation; Negotiating the Past: The Historical
Understandingof Medieval Literature(Madison, WI, 1987), 95-
 q•

TABLE 1
Distribution of accidental signs in the cantus line of De toute
Signature-accidentals are marked by parentheses, as they apply to each pitch on the given staff u
measure 5 7 o10 14 15 20 22 23
Machaut A (b-flat) (b-flat) e-flat b-nat. b-flat b-flat e-flat
Machaut F-G (b-flat) (b-flat) (e-flat) b-nat. (b-flat)2 b-flat --
(b-flat) e-flat
Machaut Vg (b-flat) (b-flat) e-flat ?3 b-flat (b-flat) (b-flat) e-flat
Machaut B (b-flat) (b-flat) e-flat b-nat. b-flat (b-flat) (b-flat) (e-flat4
Machaut E (b-flat) (b-flat) e-flat b-nat. b-flat (b-flat) (b-flat) e-flat
Fl 26 (b-flat) (b-flat) e-flat b-nat. - - - (e-flat)
PN 568 (b-flat) (b-flat) e-flat b-nat. - - - (e-flat)
Modena (b-flat) (b-flat) e-flat b-nat. - - - (e-flat)5
PN 6771 (b-flat) (b-flat) - (b-flat) (b-flat) (b-flat) (b-flat) -
'Here the question of how long a sign remains in force comes into play: the flat before the b of
b of m. 22, since the pitches follow one another so closely. In any event, measure 22 is governed by b-
as the table shows.
2The situation for this important place needs to be spelled out: the staff is governed by a b-flat sig
of the line. Hence, the flat sign could be read as either a signature or a sign referring to a specific
3The microfilm copy of this source available to me is difficult to read. Particularly, it appears to m
at measure 14, and it seems like there is some mark on the page, but it is difficult to say anything
41In what must be a copying error, the source has mm. 23-28 transposed down a third; hence, th
as a b-flat is a copyist's attempt to make sense of his error and that the correct reading is e-flat, lik
5Modena, Fl 26, and PN 568 each carry an e-flat signature here, providing e-flats for mm. 2
guidelines, entries for e-flats in mm. 24 and 25 should be made; the entries are dropped for clarity.
e-flats for these pitches, even in the absence of a signature; i.e., that the e-flats signed in mm. 9 an
immediately, in each case.
6e-flat is given as a normally placed signature, but b-flat is given before the clef, in the margin, o
occurs fairly often in this manuscript.
 BROTHERS

phrase in the cantus),36 implying a diminished fourth and increased

harmonic tension. The e-flat/b-natural combination also serves to turn
the piece more decisively away from F. b-naturals imply a strong
arrival on c, but, as we have seen, the cantus turns through b-flat and
full closure is denied. Music always offers the potential to work on two
levels and even in two contradictory ways at once. Machaut manages
to define certain moments as marked by both arrival and continuing
forward motion. The skillful touch is to create the firm sense that Part
1 of the ballade is inclined toward c while withholding closure until
the cldosending.
The composer is interested in shaping large stretches of time
through nuances in the organization of pitch, and in this sense De
toutesflours makes a good companion to Biaute qui toutes autres pere.
In De toutesflours, as in the earlier ballade, Part 1 ends with an ornate
melisma. The cantus turns twice more through b-flat (mm. 20oand 22),
the second time in another undermined progression to C/c. Harmonic
tension is not fully resolved until the clos ending, where b-natural
is signed. (Three of the principal Machaut manuscripts sign the b-
natural only at the first ending and not the second ending; the layout
on the page makes it easy to imagine the sign carrying through to the 525
second ending.) A fresh chord built on B-flat begins Part 2, repre-
senting a digression from the material of Part i; when b-natural fol-
lows (m. 35), the event seems designed to restate tension between
the two locations of B. Other references in Part 2 to the material of
Part i follow: e-flat and b-natural are juxtaposed in the cantus, and
the cadence that serves to announce the refrain (mm. 46-47) dupli-
cates the harmonic sense of Part i's ouvertending.37 The refrain itself
provides a climax in harmonic tension, a concentration of the song's
energy into a single dramatic sweep. The refrain is distinguished by a

36 That the e-flat

sign in the cantus, which actually is placed just before the d of
m. 9 in all sources, carries through the phrase seems evident. This is not really a
problem in unnotated conventions; rather, it is one of not knowing how long signs
govern. In spite of theoretical testimony to the fact that an internally placed sign inflects
only the pitch it immediately precedes, it seems likely that, in many cases, the inflection
governs a longer stretch. On this problem, see Berger, Musica ficta, 19-20. Tinctoris
alone appears to recognize that the inflection may govern longer than a single note. But
the Anonymous of 1375 comes close to saying this. See Ellsworth, The BerkeleyManu-
would read Tinctoris less as trying to reform contemporary practice (as
script, 52-3. I
Berger reads him) and more as responding to the reality of contemporary practice with
a rule that is tidier and more precise than practice allows; that is, Tinctoris is once again
attempting to systematize what was probably a haphazard practice.
37 This is, I think, how this moment should be
parsed, as an ouvert cadence that
serves to announce the refrain and predict a final resolution on C. Contrast this with
Fuller's interpretation of a "shift in tonal orientation" to D and her suggestion that "the
task of the refrain is to re-establish the C-G-c sonority that has just been supplanted"
("Guillaume de Machaut: De toutesflours," 56).
 THE JOURNAL OF MUSICOLOGY

leap to high aa (m. 55, following the previous high boundaries off in
Part 1 and g in m. 39). And since the refrain also highlights e-flat (m.
49) and b-natural (m. 57), it implies, through near juxtapositions of
these pitches, an augmented fourth, a minor seventh and a dimin-
ished fourth. D/F-sharp/b-natural (m. 57) is stretched out as if to sum-
marize the cumulative tension, resolved ultimately in the final bar.
In these two ballades Machaut is interested in organizing large
sections through nuances in pitch relations, and the propinquity ap-
plication provides him with an important tool-at least, that is one
way to interpret the versions of these songs that I have analyzed.38
Accidentals provide a means to sustain tension over large stretches of
time. Sarah Fuller's perspective on this period leads her to an impor-
tant hypothesis, one that properly, in my opinion, places harmony
at the center of ars nova polyphony: "An issue-oriented history of
compositional technique might indeed claim that a primary task of
14th-century composers was-to develop control over new harmonic
resources forced to the fore by novel rhythmic practices."39 In com-
plement, one could also say that composers learned how to take ad-
vantage of the inherent flexibility of the harmonic syntax that had
526 become such an important part of their musical language. The nu-
ances that flow from this syntactic flexibility have been trampled over
as the songs moved along through the rough paths of transmission
upon which we depend. There is a conceptual problem, too: when
analysis is conditioned too heavily by the analogy with common-
practice harmony, the nuances and flexibility of fourteenth-century
harmonic syntax may disappear altogether.

38 For De toutesflours, it is possible to argue that Machaut's manipulation of the

propinquity application that I am trying to legitimize and interpret on musical grounds
serves also to highlight the syntax of the poem. The use of cadential hierarchies may be
read as a technique for providing a sort of differentiated musical punctuation. Harold
Powers cites the Musica Enchiriadis and other treatises in which concepts of cola and
comma,varying degrees of stops and articulations, are used in musical analysis; Harold
S. Powers, "Language Models and Musical Analysis," EthnomusicologyXXIV (1980),
1-6o, quotations on pp. 49 and 50-51. Accordingly, one would see the lack of a full
stop in the polyphony for each run-through of Part 1 in De toutesflours as manifesting
in musical terms the lack of a full stop in the poetry. Fuller makes the following
observation on the poem ("Guillaume de Machaut: De toutesflours," p. 43): "Despite the
rhyme-stopped endings, thought units frequently continue across line boundaries in
the poem. The continuity (technically known as enjambment)is sometimes disregarded
in the music, as in stanzas 1 and 2 where a decisive cadence at the end of line 4 (breve
3ob) denies the patent syntactic thrust from this line into the next. But Machaut does
frame his music to take some continuities of the first stanza into account." I thank
Alejandro Enrique Planchart for suggesting this textual-musical analysis to me.
39 "On Sonority in Fourteenth-Century Polyphony: Some Preliminary Reflec-
tions," 38.
 BROTHERS

The philological principle difficilior lectio potior,

standard as it is, goes against the main thrust of much research on
musica ficta, which has proceeded according to two main focal points
-as a branch of performance practice, and as the study of composi-
tional rules given by theorists. Recognizing this principle means that
it becomes more difficult to make generalizations about practice and
manuscript evidence. The only way to proceed is to isolate local ten-
dencies through analysis which simultaneously juggles as much evi-
dence as possible--evidence of sources, style (assessed chronologi-
cally, by region and by composer), theory, and especially the design of
the piece. With Machaut, we have an advantage on at least several
counts. The sources are good, even though they are still inconsistent,
to an extent; and we have a good sense of the composer's style, though
it is hard to generalize about a style that accommodates this degree of
experimentation. In the face of a chaotic situation, with so much
intractable evidence, it is tempting but ultimately unsatisfactory to
rely on the certainty of theoretical postulates.
It is likely that the questions surrounding accidentals in late-
medieval polyphony have various, complicated answers, rather than
uniform, simple ones. It is certainly possible that some codes for 527
performance practice determined some variants in some cases. It is
hard to see how research on this or any other problem related to
accidentals (I am thinking of the period of ca. 1275-1475) can pro-
ceed in any way other than placing manuscript evidence at the center
of the inquiry. The temptation to make too much out of music theory
from the Middle Ages arises from the sheer distance separating us
from the period, a distance that manifests both in conceptual alien-
ation (the thinking is so foreign) and dimness of view (the picture is so
incomplete). And there is a more subtle problem: one may detect,
here and there, and with various degrees of candor, the attitude that
this music is more rooted in theoretical abstractions than later music
is.40 But surely composers of this period, like composers of any pe-
riod, enjoyed the freedom to manipulate and even to contradict con-
ventions.
What emerges in this account is that 1) accidentals play an im-
portant role in musical syntax; and 2) this syntax is flexible rather
than rigid. Syntactic flexibility is one reason analysis is difficult. It is
hardly the kind of thing music theorists of the period write about, but
it may be just the thing that we should write about in our efforts to get

40 Edward Lowinsky: ".. .drawing a line between theory and practical music, as
Apel does, conforms with the situation obtaining in the 19th and 2oth centuries rather
than with that of the 15th and 16th." "The Function of Conflicting Signatures in Early
Polyphonic Music," The Musical QuarterlyXXXI (1945), 229.
 THE JOURNAL OF MUSICOLOGY

closer to the great musical achievements of the late Middle Ages.

There is some basis for the analogy between harmony in the late-
medieval period and harmony in common-practice tonality, but,
clearly, this analogy can mislead. Riemann thought of musica ficta
mainly as an issue of performance practice, and he viewed the phe-
nomenon as the lost component of the text that, when recovered,
brought the harmonic language in line with modern-day thought. I
view corruption as the main textual barrier, and I see the topic of
musica ficta as one that usefully draws us into a conception of orga-
nizing pitch that is distinctly different from that of common-practice
tonality. Emphasis on flexibility in syntax leads to recognition of nu-
ance and subtlety. It should come as no surprise that subtlety is an
important value for fourteenth-century music. The challenge is to
uncover and identify it, in the face of rough transmission, and then
to interpret it, in the absence of substantial guidelines from the period
to higher-level interpretation.

Duke University

528 APPENDIX

Manuscripts Cited
Faenza 117 Faenza, Biblioteca comunale, Cod. 117
Florence 26 Florence, Biblioteca nazionale centrale, Panciatichi
MS 26
Machaut MS A Paris, Bibliotheque nationale, MS f. Fr. 1584
Machaut MS B Paris, Bibliotheque nationale, MS f. fr. 1585
Machaut MS C Paris, Bibliotheque nationale, MS f. Fr. 1586
Machaut MS E Paris, Bibliotheque nationale, MS f. fr. 9221
MachautMS F-G Paris, Bibliothequenationale, MSS f. fr. 22545-
22546
Machaut MS Vg New York, Wildenstein Galleries, MS without shelf-
mark
Modena Modena, Biblioteca Estense e universitaria, MS alpha.
M.5.24
Oxford 213 Oxford, Bodleian Library, MS Canonici Misc. 213
PN 568 Paris, Bibliotheque nationale, MS it. 568
PN 6771 Paris, Bibliotheque nationale, MS n. a. f. 6771 (the
"Reina Codex")
PN 23190 Paris, Bibliothbque nationale, MS n. a. f. 23190 (the
"Tremoille manuscript")
Strasbourg 222 Strasbourg, Bibliotheque Municipale, MS 222 C.22
(destroyed by fire in 1870)
Utrecht 37 Utrecht, Bibliotheek der Rijksuniversiteit, MS 6 E
37/1, II (olim 1846)