'rational' minds trying to

nd their own limits: probabilistic

& rationally tuned computer music

lucie nezri
Master's Thesis —Institute of Sonology, 2022
fi

 Abstrac

This project activates a process of multiple translations and transitions between

mathematical concept(s), musical concept(s) and percept(s), highlighting some commonalities
between mathematical and musical creativities. This research is mainly focused on the eld of
computer generated music, speci cally where harmony and probabilities play a central role. The
thesis presents, in a chronological order, the experiments and compositions prompted by this
research, and woven between this, I present extensive analysis and commentary about the in uence
of works by other composers

Cover page taken from Catherine Crister Hennix, "Fragments From a Writing of the Unconscious
In Catherine Crister Hennix, Poësy Matters and Other Matters, Vol. II, New York: Blank Forms Editions, 2019, p.103.

fi
"

fi
fl
 Acknowledgements

Many (many) thanks to

Richard Barrett, Raviv Ganchrow, Gabriel Paiuk, Kees Tazelaa

Edgars Rubeni
Maciejs Skrzeczkowsk
Leslee Smucke
Hilde Wollenstei

Graham Flet
Johan van Krei
Anne La Berg
Andrea Vogri
Ji Youn Kan
Michael Winte

Niels Davids
Ranjeet Hegd
Kim H
Martin Huryc
Andrejs Poikān

Fund for Excellence of the Royal Conservatoir

Studio Loo

and the constellation of supportive, beloved ones

iii

 Table of Content

Abstrac ii
Acknowledgement iii
Table of Content iv
Figure vii
1. Introductio 1
2. Analysis and reconstruction of the algorithm of James Tenney's Arbor Vitae (2006 3
2.1. A starting point for this researc 3
2.1.1.A signi cant in uence in my musical pat 3
2.1.2.James Tenney: musical acoustics & "indeterminacy 4
2.1.3.Brief overview of the two main components of Arbor Vita 5
2.2. Arbor Vitae's harmonic structure / Tree diagra 6
2.2.1.An extended rational tuning syste 7
2.2.2.A tree echoing to physical models of string 8
2.2.3.A nite, recursive set of integer multiple 9
2.2.4.A tree implying a hierarchy and interconnectedness of note 10
2.2.4.1.Hierarch 10
2.2.4.2.Combinatoric 10
2.3. Arbor Vitae's non-deterministic algorith 12
2.3.1.General description of the algorith 13
2.3.2.Presence of "ef cient" computational technique 16
2.4. Some re ections on Arbor Vitae's musical informatio 17
3. for edgars (2021 18
3.1.Starting point: rational tunings and septimal minor quarter ton 18
3.2.Furthering my understanding of sets and probabilistic space 21
3.2.1.Sets ( nite, recursive, countable, recursion) / computable function 21
3.2.2.Algorithmic randomness and probability function 22
3.2.3.Finite and in nite probabilistic space 23
3.3.Musical implications of sets and probability spaces in the piec 25
3.3.1.Probability spaces in for edgar 25
3.3.2.Overall probabilistic scenario for the piec 27
3.4.Performing the piec 32
iv
fi
fi
s
fi
fl
t
y
n
fi
s
fi
fl
)
s
s

s
e
s
h
m
m
s
m
s
h
e
s
m
s
s
"
n
e
s
s
e
e
s
)
 3.5.Ambivalent sounding of the piece & Gestalt principle 32
3.5.1.Abrupt changes and continuous processes (continuation principle 33
3.5.2.Harmonic fusion, beating patterns, harmonicity (proximity principle 33
3.5.3.Auditory masking (similarity principle 36
3.5.4.Melodic contour & Uniform probability distribution (ergodicity 37
4. Pioneering "indeterminate" pieces examined through computability and
computationalit 39
4.1.Iannis Xenakis's Herma, Gestalts & Computable analysi 40
4.1.1.General presentation of the piec 40
4.1.2.Xenakis's compositional methodology and musical strategies in Herm 41
4.1.3.Computable and Gestalt analysis of the piec 44
4.1.4.Conclusio 47
4.2.John Cage's Number Pieces & Computational analysi 48
4.2.1.General presentation about the piece and Cage's notion of harmon 48
4.2.2.Computational analysis: Number Pieces as stochastic processe 49
5. Other parallel trajectories between musical and mathematical creativitie 54
5.1.Clarence Barlow, Approximating Pi (2007 54
5.2.Catherine Christer Hennix and Brouwerian continuu 56
5.2.1.Catherine Christer Hennix & intuitionis 56
5.2.2.Engaging with formalizations of the continuu 57
5.2.2.1.Continuum and Raga: "attunement" to musical interval 57
5.2.2.2.Continuum and Brouwerian's choice sequence 58
5.3.Michael Winter, Approximating Omega (2010 59
5.3.1.In uence of digital philosoph 59
5.3.2.Deriving a piece's structure from the structure of a computer progra 61
6. Arbor Vitae, soni catio 64
6.1.Two distinct aesthetic experience 64
6.2.A form of transcription & interpretation of the piec 65
6.3.Imaginatio 65
7. for blandine and maciej (2022 66
7.1.Brief description of the piec 66
7.1.1.Harpsichord par 67
7.1.2.Electronic par 67

v
fl
n
n
t
t
y
fi
n
e
y
)
e
s
)
m
)
e
m
)
s
e
m
s
s
s
s
s
)
)
y
m
)
a
s
 7.1.3.Combination of the two sound layer 68
7.2.Multiple re-writings/transcription of the scor 69
7.2.1.First version of the scor 69
7.2.2.Second version of the scor 72
7.2.3.Third version of the scor 74
7.3.Listening experiences of for blandine and macie 76
7.3.1.Successive recordings of the piece's part 76
7.3.2.Finalized version of the piec 77
7.4.Delving into the Unmeasured preludes notatio 78
7.4.1.Unmeasured scores as a probabilistic scenario: a personal accoun 78
7.4.2.Musicological grounds of unmeasured prelude 79
7.4.3.Computational analysis of unmeasured prelude 81
7.5.Comparing three methodologies and relations to unmeasurednes 83
7.6.Conclusio 85
8. Conclusion (and opening 86
Bibliograph 88
Appendi 92

vi
x
n
y
e
e
e
)
e
s
s
e
n
s
s
j
s
t
 Figure

Fig. 2.1: Transcription of Arbor Vitae's harmonic structure from Tenney's original note 6

Fig. 2.2: Harmonic Nodes 2 – 13 on a string & irreducible divisions of a strin 8

Fig. 2.3: My translation of Tenney’s algorithm into SuperCollider cod 9

Fig. 2.4: Score excerpt (6'20"-6'30"). The score shows three main roots i.e. "greatest common
divisors" or "temporary tonal centers": 9, 15, and 25 or 35 11

Fig. 2.5 : Pseudo-code of Arbor Vitae written by Michael Winte 13

Fig. 2.6: Excerpt of the algorithm for computing a root Arbor Vitae's on SuperCollide 15

Fig 2.7: a representation of Arbor Vitae's nested sequencing of event 16

Fig. 3.1: Tuning (a) in for edgar 19

Fig. 3.2 : for edgars initial tuning scheme: a computable, recursive, enumerable nite set of pitches
(6 open strings and their respective subsets of harmonics 19

Fig. 3.3: Tuning (b) in for edgar 20

Fig. 3.4: A simple scheme of the piece's core ide 25

Fig. 3.5 Scheme of the tuning/weighing of pitches at the beginning of for edgar 28

Fig. 3.6 Scheme of the tuning/weighing of pitches as the piece progresse 28

Fig. 3.7 Scheme of the tuning/weighing of pitches at the end of for edgar 29

Fig. 3.8: An excerpt of for edgars's simple prototype computer program on Supercollider for
generating the guitar scor 30

Fig. 3.9: An excerpt of for edgars's scor 31

Fig 4.1: A methodology common to Xenakis, Tenney and 41

Fig 5.1: The digits of the rst 1000 approximations of π, shown to the tenth digit, as amplitudes of a
10-partial spectrum in Barlow's Approximating Pi 55

Fig. 5.2: Computer program approximating Omega in its ascii representation. Michael Winter,
Approximating Omega, score (p.5 61

Fig. 5.3: Michael Winter, excerpt of instructions of Approximating Omega (p.6 62

Fig 5.4: Michael Winter, excerpt of the score of Approximating Omega (p.7 63

Fig. 7.1: for blandine, p.3, rst initial versio 71

Fig. 7.2.a: for blandine, p.3, (zoomed-in) second versio 72

Fig. 7.2.b: for blandine, p.3, second version of with slurs and color 73

vii
s

fi
e
fi
s
s
)
e
n
a
.
n
)
I
.
r
s
s
e
s
s
)
g
)
s
fi
r
s
 Fig. 7.3: for blandine, p.3, third version with additional slurs and comments by Maciej
Skrzeczkowski 75

Fig. 7.4: One page of Couperin's prélude non mesuré (André Bauyn's manuscript transcription 80

Fig 7.5: A ow-chart representing Couperin's unmeasured prelude, from composition to

interpretatio 83

Fig 7.6: A ow-chart representing Tidhar's analysis of the Couperin's unmeasured prelude 84

Fig 7.7: A ow-chart representing for blandine and maciej, from composition to performanc 84

viii
fl
fl
fl
n
.
7
e
)
 1. Introduction

What interested me most when starting this research was nding commonalities between the
mathematical activity and the act of music-making, and to witness how a composer can gradually
build a musical edi ce from mathematical concepts. This interest was prompted by Fernando
Zalamea's philosophy of mathematics, which states:

What is proper and speci c to mathematical activity is […] to transit between a

world of ideas of possible, free relations, and a world much richer in
determinations, full of precisions and delimitations.1

In response to this, I believe the same "transition" is observable in general in the

compositional activity and is particularly striking in certain pieces of music, including my own
work. This thesis only gives illustrations of this idea by focusing on computer-based music and
computer-based analysis of music. The pieces discussed here may appear diverse, and even
disparate in their aesthetic interests and philosophical and conceptual background. However, they
all share some speci cs. Primarily, the pieces emphasize the musical parameter of pitch, a care for
harmony and/or rational intonation — while also stressing how these aspects of music can be
understood as perceptual phenomena. Secondly, the pieces introduced in this thesis have a strong
relationship with computability and/or computational processes. Either these pieces have been
generated thanks to computational processes, or analyzed thanks to computational processes.
Finally, these pieces all present a degree of randomness, or some of their aspects are left unde ned
by their composer. Simply stated, a primary interest of mine is to determine whether it is possible to
conceive of a computer-generated piece of music devoid of a nite procedure, with no precise set of
instructions, which is also still able to articulate a relationship between how (in)determinism can be
embedded in (in)computability.

In my own work, these ideas play out in the way I develop music on the basis of
computational techniques on the one hand, for instance by writing computer programs for
generating conjointly scores for instruments and real-time electronics; while developing a exibility
when notating my pieces and a certain "rigorous," "systematic" openness for their performance on
the other hand. I am also interested in the parallelism between the rules of listening and those of
composing, with a particular focus on pitch and harmony. Prior to this project, I was already
inspired by and playing with certain categories in psychoacoustics: pitch perception thresholds,

1Fernando Zalamea, preface to Albert Lautman, Mathematics, Ideas and the Physical Real, Continuum International Publishing
Group, 2011, p.xxviii.
1

fi
fi
fi

fi
fi
fl
fi
perception of consonance/dissonance, critical bandwidth. With time, I became more interested in
creating in my music an intermediary "situation" for listeners, between the bare perception of pitch
and harmonic relationships, and of the music's structure. This interest led me to spend time
developing speci c musical strategies as well as listening and analyzing my own pieces of music —
a process which will become apparent throughout this thesis.

2
fi

 2. Analysis and reconstruction of the algorithm of James Tenney's Arbor Vitae

(2006)

2.1. A starting point for this research

2.1.1. A signi cant in uence in my musical pat

James Tenney's Arbor Vitae has been one of the most impactful pieces in my musical
development. For me, it illustrates a quintessential juncture between mathematical concepts,
musical concepts and percepts. When I listened to the piece for the rst time in 2018, I was struck
by its clarity and consistency, and its unpredictability. Shortly after, I found out about the general
algorithm2 from which the piece was entirely governed. Developing an understanding of the
different perceptual, musical, and mathematical implications derived from the algorithmic
procedures used in the piece provided me with a means for gradually incorporating them into my
own musical vocabulary

Speci c areas of the piece particularly draw my attention. First, I found out about the
possibility of composing "numerically driven music" through Arbor Vitae or how to explore the
abstractive capacities of computation in relation to sound. This encouraged me to start developing
"computational techniques" of composition, i.e. compositional strategies to derive music, sound and
scores exclusively from computer programs — something I had never done previously. More
speci cally, I started delving into probabilistic processes for composing music. Back in December
2018, I already intended to re-construct the algorithm of Arbor Vitae for the purpose of producing a
new electronic version of it. It is no coincidence that this project was only realized during my
Master's project, thanks to the technical assistance of Andrea Vogrig: 3 or 4 years is the time I
needed to get some basic ideas beneath Arbor Vitae's algorithm.
Besides the algorithmic aspect of the piece, another essential part of learning from Tenney's
piece had to do harmony. Arbor Vitae and its complex mathematical edi ce encapsulate the
composer's exploration of harmony as a perceptual phenomenon, i.e. a framing of musical intervals
as periodic patterns, more or less easily grasped by listeners.

2See Winter, Michael. "On James Tenney’s Arbor Vitae for String Quartet". In Contemporary Music Review, Vol. 27, No. 1,
Routledge, February 2008, pp.131 - 150.
3
fi

fi
fi
fl
.

fi
fi
 Finally, the piece inspired me to incorporate physical modelling of strings as a tool for
composing. In a metaphorical sense and similarly to the "numeric/mathematical" intuitions one gets
when contemplating a tree and getting absorbed by the intricate relation between its branches and
roots, I see Arbor Vitae as poetically representing an unfolding of the temporal / harmonic structure
of a string. In this regard, the poetic presence of a physical model of a string in Arbor Vitae
de nitely (and unconsciously) played a fundamental role in my pieces for edgars and for blandine,
as both were composed for string instruments.

The rst months spent on Arbor Vitae in autumn 2020 and spring 2021 were based on my listening
experience of the piece performed,3 and my own programming of the piece's algorithm

2.1.2. James Tenney: musical acoustics & "indeterminacy

James Tenney was an American composer and music theorist, whose numerous elds of
investigation in music and writing include musical acoustics, musical form, perception, graphic
notation, computer and algorithmic music, and intonation. Guided by the question of how sound
and music are heard, he got involved in computer and algorithmic music from the early 60s. At that
time, Tenney was already becoming interested in the cognitive effects of harmonic relationships,
developing methods to deal with the gradations of dissonance to consonance, noise to pitch, and
eventually, a more general theory of musical perception. His compositional techniques were
continuously informed by musical acoustics, algorithmic procedures and particularly stochastic
processes as a means for the composer to expand "variety" in his music4. In fact, Tenney's
compositions usually illustrate a reconciliation and multiple nuances between random and
deterministic processes at different structural levels. This echoes to the second major feature of
Tenney's work: his fascination with the notions of determinacy and indeterminacy. Inspired by the
writings and pieces of John Cage and Iannis Xenakis, these two notions and musical in uences/
philosophies are often (and equally) present in Tenney's work. Tenney considered determinacy and
indeterminacy as very relative: "whatever those polar concepts are, they really are not at opposite
ends of some spectrum, but rather points on a circle"5 that may eventually converge.

3Quatuor Bozzini, Arbor Vitæ, CQB 0806_NUM, 2008, Bandcamp Audio. https://collectionqb.bandcamp.com/album/arbor-vit . See
Appendix 1.1.

4 See James Tenney, Computer Music Experiences, 1961–1964, Electronic Music Reports 1 (1969): "If I had to name a single
attribute of music that has been more essential to my aesthetic than any other, it would be variety."

5Charles Amirkhanian, Morning Concert: Composer Jim Tenney, interview, KPFMA-FM, 1976, https://archive.org/details/
MC_1976_01_12, 63'00"-65'00."
4
fi
fi

"

fl
fi
 Tenney's very last piece Arbor Vitae ("Tree of Life") synthesizes the composer's musical
interests and such paradoxical aspects of indeterminacy. The piece also mixes opposite polarities
and or structures of seemingly opposing logics. Predominantly, Arbor Vitae consists in a structural
counterpoint between determinism/predictability and randomness/unknowability. From a technical
point of view, the piece's algorithm is stunning in its simultaneous high degrees of simplicity and
complexity

2.1.3. Brief overview of the two main components of Arbor Vita

Arbor Vitae originates from two main components. The rst is a simple deterministic
algorithm for pitch based on set theory and the harmonic series. The algorithm allows to compute a
"tree" consisting of nite branching structures made of a limited set of integer multiples. In the
piece, this tree is used as a pitch diagram: a just intonation pitch space, derived from harmonics of
B- at (± 58.27 Hertz), up to its 1331st harmonic. The second component of the piece is a complex
non-deterministic algorithm based on probability functions for generating sequences of notes that
are both random and harmonically woven. This algorithm organizes the overall temporal and
melodic progression of the piece, allowing for a random progression through the harmonic tree
consisting of non-repeating sequences of notes. The algorithm designed by Tenney enables various
kinds of computations and processes: sequential (re)calculations of probabilities, statistical
feedback, multiple tendency-masks (for pitch transposition, dynamics, durations) — resulting in a
"complex, time-variant probabilistic scheme."

In 2020-2021, I focused on Arbor Vitae's harmonic tree structure as well as a purely

technical approach to the piece's algorithm. This focus allowed me to start re ecting on my
experience listening to the piece.

5
fl
.

fi
e

fi
fl
 2.2. Arbor Vitae's harmonic structure / Tree diagram

Arbor Vitae's tree has its origin in a musical concept of tuning/harmony understood as
perceptual phenomena. In Tenney's music, this musical concept may expressed thanks to
mathematical structure and concepts based on set theory.

134 M. Winter

Figure 1 Harmonic structure. All graphs are transcribed from Tenney’s original notes.

Fig. 2.1: Transcription of Arbor Vitae's harmonic structure from Tenney's original notes 6
(2500 – 4000 ), the harmony comprises tones of the same pitch class as partials
(branches) of the currently chosen root, which is the 15th harmonic of the
fundamental (even though members of the 15th harmonic’s pitch class, approxi-
mately A 712¢, are not sounding). For example, the C-sharp 725¢5 in the 1st violin
part is a just major third (derived from the 5th partial) above A 712¢, which is a just
major seventh (derived from the 15th partial) above B-flat. In Figure 2, a cents
deviation is written directly above each note. The top number next to the arrow
extending from each note indicates the partial in relation to the root and the bottom
number indicates the partial in relation to the fundamental.
Figure 3 (10 3000 – 10 4000 ) shows a root transition from the 49th partial to the 9th
partial of the fundamental. Both roots lie on the 3rd diagonal because 49 and 9 are
6See Michael Winter, "On James Tenney’s Arbor Vitae for String Quartet". In Contemporary Music Review, Vol. 27, No. 1,
Routledge, February 2008, pp.131-150.
6

 2.2.1. An extended rational tuning system

First, Arbor Vitae's tree corresponds to an extended just intonation pitch space, forming the
harmonic structure of the piece. Whereas intonation is usually de ned as "the judicious placement
of frequency"7 or a pitch accuracy, it can also considered as the practice of playing "in tune." There
are two common approaches to intonation. The rst is predetermined, considering every written
note as representing a xed pitch part of a wider, xed tuning system. The second is more dynamic
and dependent on the context of a piece or a performance, for instance where musicians are able to
tune themselves to each other according to a tone decided on the spot. Arbor Vitae and my work in
general relate to the rst hypothesis, and speci cally the concept of rational intonation: musical
tunings calculated and related by ratios of integers or whole numbers.

A rst speci city of rational intonation is that it inherently relates to the harmonic series.8
As "rationally tuned" intervals produce periodically repeating patterns of sound, rational intonation
allows for making a combined periodicity of pitched sounds, and, in brief, to compose harmonic
relationships as composite periodic sounds. Thus composing with rational intonation entails a care
for the speci c perceptual effects of harmonic relationships. For instance, a composer may take into
consideration how a listener may adjust to degrees of precision in tuning or mistuning in time, i.e.
tuning tolerance. In Tenney's, this attention to harmony is expressed by representing rationally
tuned intervals thanks to what the composer called a harmonic distances, laid out in a metric space
— such as Arbor Vitae's tree. In this sense, Arbor Vitae may be partially seen as an experiment with
auditory perception and processing of more or less complex tonal/harmonic structures,9 and/or with
the listener's tuning tolerance in time. A listener may need more or less time to get "attuned" to the
piece's tonal information or harmonic distances from B- at, especially in the midst of the various
degrees of timbral consonance and dissonance produced by the instrumentation

7 Clarence Barlow, "On Rami cations of Intonation". In KunstMUSIK, No. 16, Cologne, Germany, 2014.

8For an extended introduction to Just Intonation, see Marc Sabat and Wolfgang von Schweinitz. "Intonation — An Experimental
Application of Extended Rational Tuning." In Johann Sebastian Bach RICERCAR Musikalisches Opfer 1, 2001.

9 Usually, such tree-like diagrams are typically in music cognition research and experiments on auditory perception and human
processing of music. See for instance: Jamshed J. Bharucha and Peter M. Todd, Modeling the Perception of Tonal Structure with
Neural Nets, Computer Music Journal, Vol.13, No.4 (1989) pp.44-53.
7
fi
fi
fi
fi
fi
fi

fi
fi
fi
fl
fi

 2.2.2. A tree echoing to physical models of string

Being composed as a string quartet, Arbor Vitae implies that Tenney paid a particular
amount of attention to the phenomenon of string resonance. The effects of rational intonation are
especially pronounced on string instruments, producing difference tones and reinforcing common
partials. In this regard, I believe Arbor Vitae's harmonic tree and instrumentation work together
toward the making of a mass, composite sound of "a string." Subjectively and intuitively, I equated
Arbor Vitae's tree structure with a string model drawn from physics. This intuition of mine was due
to my previous, biased understanding of "strings," imbued with representations such as the ones
bellow, dividing a string into sections at their potential nodes of resonance. Each note of Arbor
Vitae could be seen a speci c node of resonance of a string — if only a nger as thin as a needle
that could reach such a level of precision.

Fig. 2.2: Harmonic Nodes 2 – 13 on a string10 & irreducible divisions of a string

To me, these string models representations parallel how rational intonation systems can
appeal to an in nite amount of possible con gurations and result in various degrees of precision and
extension. Thus many tools and approaches have been developed to circumscribe, map and organize
rational intonation systems and intervals.11 In Arbor Vitae, this can be done mainly thanks to basic
concepts of set theory.

10 The image on the left is taken from Ellen Fallow eld, "Multiphonics: Basics", Cellomap, https://cellomap.com/multiphonics-
basics/, and the right one from Caspar Johannes Walter, "Meantone Circles, Texts, https://www.casparjohanneswalter.de/texts/
meantone_circles

11 See for instance the great works of Catherine Lamb, Marc Sabat, Chiyoko Szlavnics, Larry Polansky…
8
fi
fi
fi
s

fi
fi

 2.2.3. A nite, recursive set of integer multiples

Arbor Vitae's harmonic lattice-structure is a nite set of integer multiples. If these integer
multiples seem to be arranged in a dense and intricate network, the way they are initially computed
is rather simple.

Fig. 2.3: My translation of Tenney’s algorithm into SuperCollider code12

for computing Arbor Vitae's tree

They are grouped into sets and subsets referred to as roots and branches. More precisely,
roots refer to harmonics or integer multiple of B- at; while branches correspond to roots multiplied
(once or several times) by 1, 3, 5, 7 or 11. Roots and branches belong to so-called diagonals, i.e.
"sets of harmonics (in relation to the fundamental) that have the same number of prime factors (not
necessarily distinct)."13 For example, roots having been multiplied 3 times (by 1, 3, 5, or 7) all
belong to the 4th diagonal; while the rst diagonal consists of the fundamental.14 This implies a
simultaneously arithmetical and harmonic hierarchy between these notes.

12 See Appendix 1.2.

13 Michael Winter, "On James Tenney’s Arbor Vitae for String Quartet," ibid. p.133.

14 ibid. p.133: "the 2nd, 3rd and 4th diagonals comprise harmonics with 1, 2 and 3 prime factors, respectively."
9
fi

fi

fl
fi

 2.2.4. A tree implying a hierarchy and interconnectedness of notes

2.2.4.1.Hierarch

From a musical point of view, Arbor Vitae's harmonic structure implies a hierarchy of notes
expressed in the tree diagram (see Fig 2.1) through the presence of what Tenney called diagonals,
roots and branches. First, all the roots and branches are derived from the rst diagonal, meaning
that any note in the piece is conceived in its harmonic relation to B- at; this de nes the main,
unvarying and "pervading" tonal center in the piece. Hence any interval may be represented in
relation to the fundamental. For instance, "if roots are prime numbers (harmonics), branches are
compound numbers made of the roots (harmonics of harmonics) [of B- at]."15 Thus the branch 33
is necessarily a byproduct of the root 11 (3x11), which is a byproduct of 1 (1x11). The further a
chosen branch is from the fundamental root on the diagram, the more complex the interval between
this frequency value and B- at. For example, branches on the 4th diagonals correspond to the most
complex harmonics of the fundamental root in the piece (605, 847, 1331). Conversely, roots and
branches belonging to the 2nd diagonal correspond to simple intervals or harmonic distances that
relate to the fundamental such as thirds, or fths and minor sevenths. Moreover, diagonals and roots
may be seen as "secondary" tonal centers (this is explained in the subsection 2.3.4.2). For instance,
the branch 33 is a byproduct of the root 11, just like the branches 55, 77 and 121. The four branches,
harmonically speaking, belong directly to the root 11

2.2.4.2. Combinatorics

Arbor Vitae's tree diagram evokes combinatorics: it represents all the possible harmonic
relationships to B- at and sequences of harmonic events, i.e. how one element of the tree may lead
to another. One of the tree's main features is the strong connectivity of its elements, and this
explains why this diagram can quickly entail complex mathematical/harmonic relationships which
use very simple computations. For instance, the harmonic 385 connects to 3 harmonics on the 3rd
diagonal (77, 55, 35), and these are, respectively, connected to 3 harmonics on the 2nd diagonal (11,
7, 5). Thus, in total, 385 may relate to potentially 6 different "tonal contexts."
Practically speaking, when the elements of the tree are turned into notes, each note may be
numerically labeled in according to different harmonic relationships, like shown in the score bellow
(Fig. 2.4). A rst number may indicate the harmonic in reference to B- at of the same pitch class as

15 Michael Winter, "On James Tenney’s Arbor Vitae for String Quartet," ibid, p.133
10
y

fi
fl
fl
fi
.

fl
fl
fl
fi
fi
the written note (see upper number next to each note circled in purple on Fig.2.4). This number is
the upper one on the score. On the cello stave, these numbers are 63, 27 and 175. Another number
may correspond to the harmonic of B- at that is "greatest common divisor" of successive notes
during a portion of the piece (see bottom left of each note circled in blue on Fig. 2.4). On the cello
stave, these numbers are 9 for the rst two notes and 25 or 35 for the third notes. I call this "greater
common divisor" a "temporary tonal center" around which the successive notes of the sequence are
derived. Finally, a last number represents the partial in relation to the current "temporary tonal
center" (bottom right of each note circled in orange on Fig.2.4). In the cello stave, this number is
successively 7, 3, and 7 or 5 for the third notes.

Fig. 2.4: Score excerpt (6'20"-6'30")16. The score shows three main roots i.e. "greatest common
divisors" or "temporary tonal centers": 9, 15, and 25 or 35.

Following this combinatoric feature, "pitches may simultaneously imply more than one
tonality since a branch may be a compound integer in relation to the root" and "polytonal
harmonies" can emerge since "sets of branches may share a common divisor that is not the [chosen]
root."17 In fact, this combinatoric feature contributes to create a continuity in the piece, illustrated in
Fig.2.4 by the lines between the notes that share the same "temporary tonal center." As a listener

16 Michael Winter, "On James Tenney’s Arbor Vitae for String Quartet," ibid., p.137
17 Michael Winter, "On James Tenney’s Arbor Vitae for String Quartet," ibid., p.136
11
fi
fl

and discoverer of Tenney's music, this may be the element of the piece that I nd most fascinating
but also understood the least at rst. The piece begins with a succession of tonalities, one after the
other. Gradually — and rather imperceptibly, it weaves together simultaneous, polytonal harmonies.
The piece has no xed tonality, rather creating a multiplicity of temporary tonalities and "micro-
scales" and yet, it grounds the listener in an overall tonal consistency and co-existence of
harmonically related pitches. Initially, I assumed this was due to the harmonic tree governing the
construction of the work, however, I now realize it is also related to the temporal/melodic
organization of the piece.

2.3. Arbor Vitae's non-deterministic algorithm

The second main component of the piece is its non-deterministic/probabilistic algorithm

determining the overall melodic and temporal organization of the piece. The algorithm uses the
harmonic tree as a sample space for probabilities, attaching evolving probability weights to each
element of the tree. On the broad picture, Arbor Vitae's probabilistic scenario is very clear: going
through the entirety of the tree diagram in 13 minutes using the most diverse sequence of pitches.
However, the algorithm designed for fulling this probabilistic scenario is very complex. Arbor
Vitae's computer program consists of hundreds of lines of codes, intermingling sophisticated
algorithmic/probabilistic processes and an overarching, temporal sequencing common to these
processes. From a strictly programming point of view, the complexity of the algorithm relates to
one of the most challenging aspects of probabilistic algorithms: nding the balance between
probability weights and events, i.e. the making of the probabilistic bounds needed for an event to
happen.

12
fi

fi
fi
fi
 2.3.1. General description of the algorithm

Fig. 2.5 : Pseudo-code of Arbor Vitae written by Michael Winter18

In the core algorithm of the piece, Arbor Vitae's harmonic tree is considered as nite
probabilistic space. Some portions or the totality of the tree are sequentially explored by the
algorithm according to a precise and rather simple timeline. Determined, durational bounds (starting
and ending points) delineate different "stages" in the pitch selection process, similar to tendency
masks.19 These stages consist in randomly selecting pitches from the 4th diagonal and gradually
including the entirety of the tree towards the end of the piece

What interest me most is what happens within each of these "stages" and how Tenney's
algorithm generates random sequences of pitches thanks to both random variables and evolving
probability weights temporarily attached to each pitch of the harmonic tree. Simplifying Arbor
Vitae's overall algorithmic procedure to its outmost, the probability of generating a given root
depends on two main computations. First, at the initialization stage, a random number is generated
and its value is compared to each of the root probabilities belonging to one or several diagonals.
The chosen root will be the one whose probability value is the closest to this random number. Then,
the probability of selecting the next roots depends on the roots that were previously chosen on a

18 Michael, Winter "On James Tenney!s Arbor Vitae for String Quartet," ibid. p 146.

19 See Fig.2.6 and the ~rdiag function on my algorithm.

13

fi
given (or several) diagonals, for a given amount of time. A similar process also takes place for the
selection of branches calculated once a root has been chosen. Throughout the piece, one walks
down into the tree by recursively evaluating its branches appropriately: each root/branch probability
is therefore constantly, sorted, collected, recalculated, and readjusted, as well as the sum of the
probabilities of all the branches, roots and diagonals. This aspect of the algorithm is said to be
conditional on the previous states of the system. In Tenney's work, this procedure of recalculation
and comparison of probabilities is often referred to as dissonant statistical feedback20. Dissonant
statistical feedback was meant for Tenney to avoid tautologies of pitch classes and, for instance, to
ensure that one pitch will not be repeated twice in a row and that a set of pitches are quasi-
uniformly distributed throughout a piece.21 As explained above, dissonant statistical feedback is
present both for the selection of roots and branches in the piece.
However, a speci city in Arbor Vitae is that, at certain times during the piece, a new random
number is generated and this reinitializes all the probabilities of the roots and/or branches to 0,
triggering new dissonant statistical feedback processes at different levels of the harmonic tree. This
aspect of the algorithm is more similar to conditional probability but using a Markov chain, i.e.
stochastic processes where the probability distribution over future states depends only on the
present state.22

20This procedure was named in this way by Tenney's successors (and I am not sure "dissonant" was the most appropriate terminology
as Tenney's algorithm can be abstracted from timbral considerations). See Polansky, Larry, Alex Barnett and Michael Winter. "A Few
More Words About James Tenney: Dissonant Counterpoint and Statistical Feedback." In Journal of Mathematics and Music, Volume
5:2 (2011): 63-82. This procedure was used by Tenney in previous compositions, like Changes. See James Tenney, About "Changes!:
Sixty-four studies for six harps, Perspectives of New Music 25 (1987), no. 1–2, 64–87, (p. 82): "Just after a pitch is chosen for an
element, [the probability of] that pitch is reduced to a very small value, and then increased step by step, with the generation of each
succeeding element (at any other pitch), until it is again equal to 1.0. The result of this procedure is that the immediate recurrence of
a given pitch is made highly unlikely (although not impossible)."

21 See Winter, Michael. "On James Tenney!s Arbor Vitae for String Quartet", ibid. p.139.

22See Wikipedia, "Markov Chain": "it is a process for which predictions can be made regarding future outcomes based solely on its
present state and such predictions are just as good as the ones that could be made knowing the process's full history. In other
words, conditional on the present state of the system, its future and past states are independent." https://en.wikipedia.org/wiki/
Markov_chain
14
fi
 Fig. 2.6: Excerpt of the algorithm for computing a root Arbor Vitae's on SuperCollider23

I will not get into the details of how the nal notes and durations were derived for the piece,
but solely mention this was done thanks to additional semi-determined, semi-probabilistic schemes
for pitch-range and durations. The accumulation of these probabilistic processes shows that
complexity and random variation were essential features of the piece.

23 See Appendix 1.2.

15
fi

 2.3.2. Presence of "ef cient" computational techniques

To me, Arbor Vitae illustrates how a composer may develop musical processes on the basis
of ef cient computational techniques and an economy of means — two aspects which are
particularly inspiring to me. First, as mentioned above and very classically, the algorithm allows
random variation. Hence a multiplicity of versions of the pieces may be generated, without
changing the structure of the piece. It is also possible to change any of the algorithm's variables,
entailing a scalability of Arbor Vitae's musical processes, still respecting the piece's core structure.
But the most striking features of Arbor Vitae's algorithm are its recursivity and nestedness, making
it particularly elegant, concise and ef cient while allowing a complexity in its outputs. The
algorithm is based on recursive structures in the form of chains and nested recursion. The presence
of recursion in Arbor Vitae implies a form of interdependence and nestedness between its different
routines. First, these recursive structures stem from interdependent modules combined together,
such as the generation of random variables, the recalculation of probabilities, the recalculation of
the sum of these probabilities. The same interdependent modules are then re-used at different levels,
or in different routines of the algorithm. These routines are nally nested inside each other, resulting
in the formation of sub-subsequences within a subsequence within a sequence (and so on). For
instance, the routine for the selection of branches is nested in the routine for the selection of roots,
which is itself nested in the routine for the selection of diagonals.

diagonals routine

roots routine

branches routine

Fig 2.7: a representation of Arbor Vitae's nested sequencing of event

16
fi
fi
fi

fi

 2.4. Some re ections on Arbor Vitae's musical information

Swells and tensions between the different levels of musical information

The overarching sequencing of tendency masks in the piece's algorithm results in the design
of a macroscopic swell. In Tenney's music, a swell is usually articulated as "an expanding then
contracting pitch range, a crescendo/plateau/decrescendo dynamic swell and an increasing then
decreasing temporal density"24. This "single gestalt" is present on the macro-level of Arbor Vitae
and applied simultaneously for the piece's different parameters — creating a dramatic tension in the
piece, contrasting with the subtleties of its inner developments in terms of intonation. The piece
starts from the highest branches of the trees softly played as arti cial harmonics. It then gradually
includes the entirety of the branches, expanding the piece’s pitch range, along with a decrease of
durations and a continual crescendo. From the second half of the piece, the pitch range starts to
contract (its upper limit going down), while probabilistically travelling more towards the
fundamental frequency. Meanwhile, the piece gets louder and its durations shorter. During the last
fourth of Arbor Vitae, the different parameters return more or less to their initial states at the
beginning of the piece: narrow pitch range, high notes, soft dynamics, relatively long durations

From a listener perspective and on this macroscopic level, Arbor Vitae's sonic information
can roughly be perceived as very much predictable and reducible to this simple, distinctive gesture
of a swell. On the microscopic level though, and as a listener, I have continuously felt challenged in
the grasping of the piece's melodic-harmonic information — which very much resonates with the
complexity of the microscopic programming level of the piece. To me, all the precision of the
piece's nested, recursive programming is conveyed in the complexity of the sonic information of the
piece on a microscopic level: the multiple tonalities in the piece are a result of the way each looping
sequence of roots are constituted from one branch after the other, and the way each note follows the
other with no repetition, etc. As much as I started understanding the piece's elaborate algorithm, my
listening experiences remained limited. Arbor Vitae's micro-melodic/pitch sequences were so
complex and intermingled with the timbral richness of the string instruments that they tended to
recede into the background of my listening experience. Instead, I was only able to hear more drastic
changes of tonalities occurring at turning points in the piece, as well as the overall swell of the
piece. This made me wonder if I would get to a different understanding of the piece by hearing a
synthesized version of it with sine tones. This curiosity led me to make a soni cation of the
algorithm that I realized in September 2022.

24 Winter, Michael. "On James Tenney!s Arbor Vitae for String Quartet", ibid, p.131.
17

fl

fi
fi
.

 3. for edgars (2021

The rst musical outcome of this Master's project is my piece for edgars,25 written for
electric guitar and electronics. The composition originated from my basic understanding of some
mathematical concepts encountered in Arbor Vitae and illustrates ways of applying them to my own
work. The piece for edgars is one of the rst works where I fully integrated rational tuning systems
as a part of my compositional interests. More speci cally, it is inspired from the formalized
structures found in set theory for considering rational intonation. The composition also looks into
the use different probabilistic sample spaces for generating pitch material and how their
combination may result into mixing different musical notions of pitch intervals.
Many paradoxes confronted me while "imagining" and composing for edgars — and most of
them express my speci c way of engaging with harmony and intervals of pitch. These paradoxes
were rendered in contrasting and ambiguous listening possibilities that were inherently proposed by
this piece

3.1. Starting point: rational tunings and septimal minor quarter tone

Rather than following the western tradition of tonal harmony, I have mainly learned about
harmony from the perspective proposed by Tenney and other researchers, and based on simple
mathematical principles. Such a perspective allows tuning to be central in the composition of a
piece. For instance, I am inspired by the standpoint of the American composer Larry Polansky
presents a compositional focus that seeks "to allow tuning possibilities to generate and become the
form and structure of the work"26. I tried to adopt a similar standpoint when composing my piece
for edgars, which structure originates from the combination of two rational tuning systems.
To ground my rst encounter with rational intonation, I developed some mathematical and
graphical representations of the two tuning system. These methods helped clarifying the harmonic
patterns and sonorities I would eventually decide to use. Doing some back and forth between the
graphs and simulations of tuning systems with sine tones were decisive steps for preparing the
piece.27

25 See Appendix 2.1.

26Larry Polansky, "A Few Words About Tuning," in The Just Intonation Issue, Ed. Nate Wooley, Sound American https://
www.soundamerican.org/issues/just-intonation/few-words-about-tuning

27 See Appendix 2.2 and 2.3.

18
.

fi
)

fi
fi
fi
fi

 In for edgars, a nite set of frequency values was computed from one "basic" frequency
( f0 ) and each value was then derived from a recursive chain of multiplication. This is shown for
instance in the following tuning system used in the piece, which I called "tuning (a)"

- f0 (1/1)
- f1= f0* (3/2) — Perfect fth in relation to f0
- f2= f1*(4/3) — Perfect fourth in relation to f1 (<=> an octave from f0 (2/1))
- f3= f2*(5/4) — Major third in relation to f2 (and from f0 (5/2))
- f4= f3*(6/5) — Minor third in relation to f3 (<=> a perfect fth from f0 (6/2))
- f5= f4*(7/6) — Septimal minor third in relation to f4

Fig. 3.1: Tuning (a) in for edgars

Moving on from this, each frequency value was used to tune each open guitar string.
Respective subsets of frequencies were then computed from these six initial frequency values,
eventually corresponding to the harmonics of each of the open strings (up to their 7th partials).

Fig. 3.2 : for edgars initial tuning scheme: a computable, recursive, enumerable nite set of pitches
(6 open strings and their respective subsets of harmonics)

The piece for edgars used two rational tuning system with intervals that produce repeating
patterns of sounds of two sorts. When these patterns of sounds are very slow, they give rise to what
the composer Marc Sabat calls harmonic fusion28 i.e. a distinctive periodic resonance. Conversely,
when these patterns are faster, they may be perceived as regular beating frequencies. In for edgars,

28Marc Sabat and Wolfgang von Schweinitz. "Intonation — An Experimental Application of Extended Rational Tuning." In Johann
Sebastian Bach RICERCAR Musikalisches Opfer 1, 2001, p.2.
19

fi
fi
fi

fi
:

most intervals were very simple, based on fths, octaves and thirds (see Fig.3.1) and created
harmonic fusion. But my particular interest in the piece was to enhance the interval between f4 (6/5)
and f5 (7/6), corresponding to a septimal minor quarter tone (SMQT or the ratio (35/36)).29 This
interval carries a few speci cities. First, it creates a periodic patterns between harmonic fusion and
a sensation of beating. Indeed, this interval may fall into the range of a critical bandwidth30 and the
grasping of this interval depends on the ability for listeners to distinguish timbres in a mass of
sound. Then, a SMQT tends to introduce an overall ambiguity in a tuning/scale, disrupting the
traditional major/minor polarities. This SQMT entailed important harmonic consequences in the
other tuning system used in for edgars, tuning (b) :

- f0 (1/1)
- f1= f0* (3/2) — Perfect fth in relation to f0
- f2= f1*(4/3) — Perfect fourth in relation to f1 (+ an octave (2/1) from f0)
- f3= f2*(5/4) — Major third in relation to f2, (+ an octave and a major third (5/4) from f0)
- f4= f3*(7/6) — Septimal minor third in relation to f3 (+a minor sixth (35/12) from f0) +
septimal "color" spread to f4 subset of overtones
- f5= f4*(6/5) — Major Third in relation to f4 (and septimal minor seventh (7/2) from f0) +
septimal "color" spread to f5 subset of overtones

Fig. 3.3: Tuning (b) in for edgars

Although it shares many frequencies in common with tuning (a), tuning (b) resulted in a
"septimal colorization" of its pitch set. This tuning also implies a more harmonic structure in
relation to ( f0 ) and more beating sensations.

Clarence Barlow's until… for guitar and electronics was inspiring in that regard: the piece introduces a difference of a septimal
29

minor quarter-tone in the tune of 3 pairs of strings, subtly enhanced by the electronics.

30This obviously depends on which frequencies this interval is being played. If two frequencies separated from a SMQT would be on
a very high frequency range, it is more dif cult to distinguish them as separate tones.
20
fi
fi
fi
fi

 3.2. Furthering my understanding of sets and probabilistic spaces

Before working on the overall structure of for edgars, I wanted to de ne a simple and clear
(and, ideally, effective), general frame of algorithmic, musical structure, a piece "pro le" that could
potentially be revisited and used for different solo instruments, durations, and tunings. Looking at
the sounds, in particular those from the guitar, would only happen at the last compositional stage.
Overall, my wish was to compose a gradual transition between the two tuning systems presented
above. If I was expecting this transition to result in ambiguous perceptual effects for listeners, I also
knew there were many ways I could conceive (compose and think of) this transition. Following
from the mathematical concepts I encountered in Arbor Vitae, I wished to delve into the ones which
I found most inspiring and accessible, i.e. sets and probability. I thus decided to get back to the very
basics of these two foundational branches of mathematics and see how they could prompt new
musical ideas — or reveal preexisting, intuitive ones.

3.2.1. Sets ( nite, recursive, countable, recursion) / computable function

While working on the for edgars, it was useful to take a fresh look at some basic de nitions
pertaining to sets and computable functions. This helped me developing ef cient approaches for
working with rational tuning systems and imagining new possibilities for composing the piece.

A set is a collection of single values. An enumerable or countable set is one whose members
can be enumerated. A set can be nite or in nite: nite sets have a nite number of elements,
whereas in nite sets have an in nite number of elements. In nite sets are the essence of set theory,
while countable sets are foundational to computability theory. For the purpose of this discussion, I
will focus brie y on explaining the latter.
Initially founded by Alan Turing, computability theory allows for the reduction of complex
phenomenon to minimal logical constraints. To be more precise, computability theory is concerned
with the study of computable functions either to enumerate a countable set or to approximate
uncountable sets. Computable functions suppose the existence of a nite, effective procedure, i.e.
an algorithm with a precise set of instructions and a nite number of discrete steps, which explicitly
state how to compute a function. Given an input of the function domain, the algorithm can return
the corresponding output. A fundamental aspect of computability theory lies in the notion of
recursion, i.e. the process of de ning a mathematical object in terms of smaller versions of itself.
This is done thanks to recursive function, i.e. an algorithm which calls itself during its execution;

21
fi
fi
fl
fi
fi
fi

fi
fi
fi
fi
s

fi
fi
fi
fi
fi
fi
 hence its outputs depend on previously computed values. Lastly and logically, a recursive set entails
that there is a computable recursive function to list all its elements.31

Already in the last chapter, I have shown how Arbor Vitae is fundamentally based on
computable and recursive functions. The most simple example of recursion was found in the piece's
harmonic structure: one single function listed all its elements, simply by calling itself 3 times.
Indeed, the set of branches were derived from the previously computed set of roots, which were
also derived from the previously computed set of branches.32 Likewise, recursion particularly
informs my approach to rational tuning systems and my interest in tuning system that are
completely or partially, recursive sets, as shown in the tuning systems used in for edgars (See
Fig.3.2 and 3.3). A much more complex example of recursion was also illustrated with Arbor Vitae's
statistical feedback probabilistic algorithm that de nes the temporal and harmonic unfolding of the
piece. When composing for edgars, I was not interested in using such complex algorithms based on
recursion. However, I knew that the most crucial aspect was nding the right balance of complexity
and variety between the harmonic and melodic musical structures throughout the piece. To address
this, I returned to the basics of algorithmic randomness and probabilities.

3.2.2. Algorithmic randomness and probability function

First, I got back to more foundational de nitions explaining algorithmic randomness and
probabilities and illustrate how the two concepts are intimately connected.
With the use of computers being ubiquitous in society, the word ‘random’ has become a
much more common cultural trope. However, randomness remains an elusive concept, re ecting
different types of conceptual (and mathematical) settings.33 Randomness is mostly referred to when
dealing with sequences of datas that have no discernible patterns or regularity. Such random
sequences can be nite or in nite. In this section, I will solely focus on algorithmic randomness,
that being "individual random in nite sequences which can be modelled thanks to the tools of

31 The elements of the list may be out of order or with repeats. Rebecca Weber, Computability Theory, Student Mathematical Library,
Volume 62, Providence: American Mathematical Society, 2012, p.97.

32My SuperCollider program was a great illustration of this recursion: ~diag4 is computed from ~diag3, ~diag3 from ~diag2 and
~diag2 from ~diag1. See Fig.2.3.

33Giuseppe Longo, Catuscia Palamidessi and Thierry Paul "Some Bridging Results and Challenges in Classical, Quantum and
Computational Randomness". In Hector Zenil, Randomness Through Computation: Some Answers, More Questions, World
Scienti c, 2011, p.77.
22
fi
fi
fi
fi
s

fi
fi
fi

fl
computability theory"34, i.e. computable functions. To start with, algorithmic random sequences
follow probability functions. A probability function expresses a probability distribution or the
statistical properties of the random sequence. To each probability function is attached a probability
sample space which corresponds to a set of the function's possible outputs or events. In the case of
algorithmic randomness, the sample space is a discrete one. Additionally, the probability of the
entire sample space is equal to 1, and for every event, referred to as a subset of the sample space,
the probability must be comprised between 0 and, at most, 1.
It is my belief that these initial de nitions already show how composing music with
algorithmic random sequences requires one to "compose a probability," i.e. : to establish the types
of probabilistic spaces one wants to work with, and to determine a range of probabilities for each
possible outputs of a probability function. Such modelling a musical/algorithmic random process
goes against a common, simplistic understanding of "randomness" and "indeterminacy" as "non-
intentionality" in the composition of a piece. Similarly, composing with random sequences does not
automatically equate with a sequence that will sound super cially "chaotic" or "non-intentional" or
"indeterminate" in the perceptual domain — except if this is intended by the composer.35 In my
pieces, starting with for edgars, composing with random sequences is intentional: my probability
functions are usually very simply and precisely de ned, perhaps to the point where one can wonder
to which extend probabilities and randomness are relevant tools to generate the material in the
piece. Nevertheless, I nd the generative aspect and potential of this compositional, and
probabilistic process, to be crucial. As a composer, I only specify the big "lines" of a particular
scenario, thereby avoiding the need to completely decide every note in the piece. More
fundamentally, the generative and random aspect of the work allows me to maintain a level of
playfulness, discovery and distance with my material as it is inherently malleable and provisional.

3.2.3. Finite and in nite probabilistic spaces

When I started working on for edgars, I was drawn to two types of randomized algorithms,
depending on their probability sample spaces. The rst randomized algorithm I worked into this
piece dealt with a nite probability sample space. The latter is the most commonly encountered in
stochastic music composition and consists of a nite and countable set of values. An algorithm
randomly picks each individual value belonging to a set and then generates a sequence of data
following an overall weighing of the selection process, certain values of the set being potentially

34 Giuseppe Longo, Catuscia Palamidessi and Thierry Paul, Randomness Through Computation: Some Answers, More Questions,
ibid. p.78.

35 See Section 4.1 on Iannis Xenakis's Herma.

23
fi
fi
fi
fi
fi
fi
fi
fi

 more prone to be picked than others. Arbor Vitae's harmonic tree is typical example of a nite
probability sample space whose elements are weighted differently throughout the piece.

The second type of probability, also used in for edgars, concerns so-called in nite
probability sample spaces. The latter have sets of values that delineate the extremes of some
metrics. In nite sample spaces are countable when they are made of discrete elements that can be
counted and there is an in nite number of them (like in the geometric series). What intrigued me
most were in nite sample spaces that are uncountable. Similar to an in nity of distinct points on a
continuous line segment, the elements of such sample space cannot be counted. That said, whether
they are countable or uncountable, in nite probabilities are computable: any point or element
belonging to in nite sample spaces can potentially be selected by a randomized algorithm. In for
edgars, I recognized that there was a strong potential to consider rationally tuned musical intervals
as possible in nite uncountable sample spaces, or "continuous line segments" whose elements
could be randomly picked by an algorithm. My interest in this idea also extended from the fact that
I had rarely encountered this orientation to musical intervals in other compositions in rational
intonation.36 Additionally, this pointed to a speci c relation to musical intervals I was trying to
decipher in my own work, which is associated a notion of a frequency continuum and how a listener
has a "tendency to treat a range of [pitch] values along a physical continuum as if they were the
same until one reaches a point at which the percept abruptly changes."37 To be exact, I was curious
to experiment with in nite sample space corresponding to very narrow musical intervals, such as
the septimal minor quarter tone, and by so doing, to experiment with listeners's tuning tolerances. In
truth, I was also fascinated with the idea of an "in nite" slicing of such narrow intervals because it
required a great precision which would hardly be replicable in any acoustic environment and
'soni ed' with synthesized sounds.

36 Except in the work of Chiyoko Szlavnics and in the writings of Catherine Christer Hennix (See Section 5.2)

37 Siu-Lan Tan, Peter Pfordresher and Rom Harré, Psychology of Music, Psychology Press, 2010, p. 98.

24
fi
fi
fi
fi
fi
fi
fi

fi
fi
fi
fi
fi
fi
 3.3. Musical implications of sets and probability spaces in the piec

Revisiting these basic mathematical concepts on probabilities led me to imagine a

paradoxical situation in for edgars, where nite and in nite sample spaces could be derived from
the same sets of numbers or frequency values. Yet the cohabitation between two kinds of
probabilistic spaces entailed antagonist objectives in the treatment of sets. In a nite sample space,
these numbers are considered as potential sole, discrete "outputs" of the probability functions. In an
in nite sample space, these numbers constitute the boundaries of the function's uncountable
outputs. Inevitably, this made me curious to experiment with the musical implications of this idea in
for edgars.

3.3.1. Probability spaces in for edgars

3.3.1.1. Finite and in nite probabilities for the parameter of pitc

3.3.1.1.1. A gradual transition of tunings (combined probabilistic spaces

As previously mentioned, my plan was to compose a gradual and probabilistic transition
existing between two tuning systems, one that was a result of the combination of the two types of
nite or in nite probabilistic spaces. In the way I conceived it, the transition between these tuning
systems would in fact correspond to a transition illustrated in the following gure :

tuning (b) tuning (b) + tuning (a) + SMQT (a)–(b) tuning (a)

= nite probability space (1) = nite probability spaces (1 + 2) = nite probability space (2
harmonic relevance + in nite probability space (1–2 harmonic relevance
harmonic relevance and/or harmonic irrelevance

beginning of the piece middle part end of the piece

Fig. 3.4: A simple scheme of the piece's core idea

The left of the above gure shows the rst sample space used in the piece (called " nite
sample space (1)") which is made of a nite collection of pitches belonging to tuning (b). This
means that at the beginning of the piece, the guitar and electronics were exclusively tuned to this
speci c tuning. Consequently, an overall harmonic relevance was highlighted as the two sound

25
fi
fi
fi
fi
fi
fi
fi
fi
fi
)

 
)

fi

fi
fi
fi

fi
h

fi
fi
fi

 layers reinforced the frequencies (and overtones) belonging to tuning (b). Similarly, the end of the
piece culminates into tuning (a) exclusively.
However, as shown in Fig. 3.4, for edgars's middle part combines a nite sample space
(" nite sample spaces (1+ 2)") consisting of tuning (a) and tuning (b) with an in nite sample space
which introduces a form of continuum of frequencies between the two tunings ("in nite sample
space (1–2)"). The introduction of an in nite probability space entails to "subvert" the rational
tunings (a) and (b) and consider how these can be made more dynamic, exible but also less
precise. Hence in the middle of the piece, a harmonic irrelevance was introduced since the guitar
and/or the sine tones evolved in "out-of-tune" realms, neither fully belonging to tuning (a) nor
tuning (b)

To sum up, the last paragraphs touch on the main paradox of for edgars which lays in the
blending within one single piece of harmonic relevance (i.e. a system of rational intonation that is
treated as a nite sample space) and harmonic irrelevance (partial "blurring" of this rational
intonation system when this system is treated as an in nite sample space)

3.3.1.1.2. Practical realization

First, the two pitch sets of tuning (a) and (b) were used as nite probabilistic spaces. The
individual values of the two tunings were then randomly selected by an algorithm, according to
evolving weighing functions (more in Section 3.3.3). This rst approach was mainly employed for
the guitar part and its xed score. Then, to include the idea of in nite probabilistic space in the
narrow range of a semi-minor quarter tone (SMQT), I designed my algorithm so that both the
respective subsets of overtones and the frequency values of f0 (1/1), f1 (f0* (3/2)) and f3 (f2* (5/4))
would randomly modulate in a range of an SMQT38. Each of these frequencies f would oscillate
between an upper boundary equal to f, and a lower boundary equal to f transposed by an SMQT
down. In other words, f could be multiplied by a random number comprised between (1/1) and
(35/36). Weighing functions were used in the part of the algorithm as well.

3.3.1.2. Finite probabilities for the parameter of rhythm / duration

Regarding meter and durations, the guitar and the electronics followed one unique stochastic model,
applied independently to each of the two layers. The same nite collection of durations (1", 3", 5",
7", 11") were used throughout the entirety of the piece, corresponding to the probability nite
sample space for durations. A very simple stochastic algorithm selected individual durations among

38Retrospectively, this decision to solely stick to modulations within a SMQT may have impoverished for edgars's potential. It also
shows some hesitations in my way of using just intonation.
26
fi
.

fi
fi
fi
fi
s

fi
fi
fi
fi
.

fl
fi
fi
fi
fi
this set, used for deciding when the guitar should play and when the electronics change in frequency
and amplitude. In theory, I thought this would result in an overall metrical/rhythmical consistency
between the two layers. In reality, I ended investigating the indirect interactions or loose temporal
ties between the guitar part and the electronics. For the electronics, the computer program followed
these durations with an exact precision while the performer was considering them as "averages,"
thus with more exibility. By doing so, another aleatoric dimension of the piece emerged: variable,
microscopic, random temporal offsets between the electronics and the guitar.

3.3.2.Overall probabilistic scenario for the piece

Weighing functions were another essential tool for me to create a continuous, smooth
transition or trajectory between tuning (b) and (a) and give speci c directions to the piece's different
random processes.
The rst probabilistic trend relates to evolving sets of pitches available for the random
selection of notes on the guitar part. For making the guitar score, I used a computer program that
weighted and selected which string and which note (open string or harmonics) would be played. I
gave more weight to speci c strings and harmonics throughout the piece (the details of this decision
are explained in my description of the code, found in section 3.8). At a later stage, and when
possible, I combined these notes into pairs or triads so they could be played as chords by the
performer.39
The second probabilistic direction given to the piece concerns the gradual transposition of a
SMQT down throughout the piece, from (1/1) to (35/36). This transposition mostly affected the
electronics but also the guitar part, which was retuned throughout the piece. At rst there was
greater probability for f0 (1/1), f1 (f0* (3/2)) and f3 (f2* (5/4)) to be multiplied by (1/1), then
gradually by a random number comprised between (1/1) and (35/36) and nally by (35/36). The
output of this algorithm was used to change the frequencies of the sine tones in real-time during the
performance of the piece. Noticeably, the frequency (f0 *(35/36)) acts as a "bridge" in the piece: it
belongs to the two sets of pitches (a) and (b); it is also sustained in the electronics and becomes
enhanced in the guitar part throughout.
Finally, amplitude changes are applied to the sine tones throughout the piece. Thanks to
other weighing functions, the durations between each guitar note or between each frequency or
amplitude change to the electronics are expected to be longer at the beginning and end of the piece
and shorter at the middle of the piece

39 This process was an embryonic compositional strategy elaborated a lot further in for blandine & maciej.
27

fi
fl
fi
.

fi
fi

fi
 Further (optional) explanation of each stages of the piece are given bellow.

Fig. 3.5 Scheme of the tuning/weighing of pitches at the beginning of for edgars

At the beginning of the piece, the guitar's string VI is tuned lower to D. String IV and string
VI are the most prominent. More complex melodic and harmonic lines in relation to D occur since
there were more chances for higher harmonics to be selected by the computer program. The dotted
"strings" on the graph above are thus more of the "background" of the piece. The 16 sine tones are
sustained and their frequency values remain constant.

Fig. 3.6 Scheme of the tuning/weighing of pitches as the piece progresses

As the piece progresses, some of the frequencies of the sine tones randomly change between
their initial value and this value is then multiplied by (35/36) via a rst stochastic routine and a
second stochastic routine affects their amplitude. Meanwhile, strings VI, V and IV are retuned one
by one a quarter tone down by the performer at three different places in the piece. The guitar part
28

fi

 moves towards simpler harmonic relationships in relation with A-47cents. The probability
distribution for selecting the strings is a quasi-uniform distribution: all strings of the guitar have a
more or less equal chance to be played, just as higher and lower harmonics.40 For the random
selection of durations, shorter durations for the electronics and longer ones on the guitar are
preferred when arriving towards the middle of the piece

Fig. 3.7 Scheme of the tuning/weighing of pitches at the end of for edgars

At the end of the piece, some of the sine tones have been transposed a quarter tone down,
string II and string V are the most prominent on the guitar part (both corresponding to an A-47c).
Lower harmonics have more chance to be stochastically selected, meaning that the melodic line on
the guitar is more harmonically simple

40At rst, the harmonically simpler intervals with D are slightly more prone to being picked, and gradually, the harmonically simpler
intervals with A-47c are more prone to be picked. But I do believe this idea too timid in for edgars but was expanded in for blandine
and maciej (See Chapter 7).
29
fi
.

 Fig. 3.8: An excerpt of for edgars's simple prototype computer program on Supercollider for
generating the guitar scor

Here are a few explanations about the simple computer program used to generate the score.
- stringSelected corresponds to the set of available pitches, to which is attached an array of
probability weights (not present on the screenshot).
- Once a pitch has been selected, it can be multiplied by one of the values of harmLimit1,
corresponding to their harmonic (up until the 7th harmonic). harmLimit2 allows to transpose this
harmonic up till the 4th octave. chosenHmc1 are the probability weights attached to harmLimit1
and harmLimit2, which are the probability weights attached to harmLimit2.
- Throughout the piece, the ve variables stringSelected, harmLimit1, harmLimit2, chosenHmc1
and chosenHmc2 evolve.
- A similar operation is applied for the electronics of the piece
There is no recursion in the program but every function is connected to the other: for
example, the function for choosing the octave transposition of an harmonic depends on the function
for choosing the harmonic, which depends on the function for choosing an open string.

30
e

fi

 Finally, the score bellow shows how sticking to numbers only was the easiest notation for
both me and the performer.

Fig. 3.9: An excerpt of for edgars's score41

The performer has a certain amount of seconds to play one or several notes. The orange
boxes indicate to the performer when to play several pitches together as a cluster or group of notes.
They were grouped together as non-repeating clusters of notes — an idea expanded further in my
next piece.

41 See Appendix 2.4.

31

 3.4. Performing the piec

Overall, collaborating with Edgars Rubenis has given me an opportunity to clarify my role
in relation to performers, and the agency I expect them to have in my pieces. When asking Edgars
about his approach when playing the piece, he answered that his strategy was "ignorance." Never
having played in a "classical ensemble" nor "knowing" its rules, Edgars usually "ignores" the other
players in improvisation settings. Thus "ignorance" of the rules of classical ensemble playing and
ignoring other instrumentalists became Edgars's approach to musicianship in general — and was
realized in my piece as well. This unique way of playing meant that Edgars intentionally ignored
what was going on with the electronics, avoiding interacting with them as much as possible. In
addition, Edgars had never worked with harmonics or just intonation before, and this provided a
freshness in the way he related to his instrument
In my view, Edgars's "ignorance" when performing and his freshness with harmonics
contributed to a successful approach for playing the piece. At least, it deeply echoed with the way
the piece was composed using random processes, but also the unknown dimension of my
compositional process and the recent "leap" I had made with working with just intonation. In brief,
writing this piece has shown me how the role knowledge, or lack thereof (ignorance), can have in
assisting a performance of this type of music. This led me to begin discovering an approach to my
work that touched on certain of limits of "knowledge," speci cally when composing, but also while
a performer plays — and while a listener listens.

3.5. Ambivalent sounding of the piece & Gestalt principle

From a listener’s perspective, for edgars and its hesitant sounding raises many questions
related to the perception and cognition of auditory/musical information in the piece. On the one
hand, the subtleties of certain amplitude changes or frequency modulations are barely perceivable.
On the other hand, the piece's musical information is very intricate: frequency modulations of the
sine tones occur within an extremely narrow range; frequency modulations between the electronics
and the guitar also occur; and different pitch sets unfold and evolve from more simpler harmonic
relationships to more complex ones. Re ecting back on for edgars, the piece alternates between
either very low or very high perceptual and cognitive thresholds for listeners, especially for pitch.
The piece met some mixed reactions: some listeners were able to hear most of the piece's features
while others were only able to discern a few of them. Even more interestingly, some listeners had an
ambiguous response to the piece, being able to grasp and follow the piece's musical information at
times, and for some, not at all. These remarks from listeners sometimes made me question whether
32
e

fl
.

fi
certain compositional decisions I made were successful. For instance, I wished to understand "why"
the piece prompted such drastically different listening experiences among listeners, and, more
importantly, to draw some connexions between these reactions and my compositional approach to
rational tuning and/or probabilities

This led me to consider analyzing the piece through the lens of Gestalt theory. This theory
seemed applicable as it posits a way to predict thresholds of perception and allows to make better
sense of the workings of human cognition. The following Gestalt analysis of for edgars helped me
gain clarity and announced a different compositional approach for my next piece. In general, three
Gestalt principles could be found, albeit ambiguously, in the piece. These were continuity, similarity
and proximity.

3.5.1. Abrupt changes and continuous processes (continuation principle

Originally applied to the visual realm, the continuity principle states that elements arranged
as "lines" or "curves" tend to be perceived together as a grouped unit, rather than elements that are
not placed on these lines or curves. From this perspective, the sustained, justly tuned tones in for
edgars could be understood as temporal "lines," evoking the continuity of periodic sounds found in
drone music for instance. "Lines" may also be metaphorically found in the piece's linear and gradual
processes of transition. Yet, a speci city of my work concerns the simultaneity of continuity in the
form of sustained tones and continuous processes (transition), and (abrupt, and often dense) changes
(of frequencies and dynamics). For instance, in the middle part of for edgars’s, the linearity of sine
tones tends to be broken apart because of the presence of abrupt changes that are applied to them, or
the juxtaposition of the guitar's sharp, short, disruptive sounds that occur on top of them.
Nevertheless, as the density of changes diminishes during the piece, these "lines," along with their
just intonation signature are found again. One unique regular beating (that corresponds to the
interval of a SMQT) is even featured and allowed to grow in amplitude at the end of the piece.

3.5.2. Harmonic fusion, beating patterns, harmonicity (proximity principle)

According to the proximity principle, elements that are close together seem more related
than elements spaced farther apart. This principle is particularly helpful when trying to assess the
perceptual effects of harmonic fusion and beating patterns in for edgars. First, the beginning and the
end of the piece enhance a sensation of harmonic fusion between the tones of the guitar and the
electronics, a sensation that goes even beyond a proximity of tones. Secondly, an ambiguous
33

fi
)

situation occurs in the middle of the piece, combining harmonic fusion and beating patterns. These
beating frequencies unfold within a SMQT, an interval made of very proximate tones and already
very close to a critical bandwidth.42 Overall, the simultaneity of harmonic fusion and beating
patterns conveys a certain ambiguity or indecisive focus of listening. The latter is either directed to
the complete fusion of tones or regularity of the periodic patterns derived from the rationally tuned
intervals; or gravitates towards the separation of proximate tones, i.e. a variety of beating
frequencies contained in a SMQT.

From a compositional standpoint, for edgars raises a question about the balance between
minute details and variations in pitch, especially with respect to the size of a small interval and the
time needed for a listener to grasp these details. In that regard, it is interesting to link the precision
of just intonation systems with how our perception and processing of pitch intervals occurs. For
instance, we know that it is dif cult for any listener to discriminate intervals that do not cross the
perceptual boundary between simpler intervallic ratios, as such intervals are too small for our
auditory system to distinguish these very detailed variations.43 This is a reason why most just
intonation system set a prime limit, i.e. a largest common denominator for all the ratios of integers
present in the system. The most common just intonation systems are 2-, 3-, 5-limit; some composers
are adventurous to go up to the 19-limit but such complex limits are more rare. Overall, for edgars
goes up to 7-limit.44 However, its electronic part may extend far beyond when producing intervals
smaller than an SMQT, as these intervals correspond to ratios of very large whole numbers (a
random example could be: 1344/1345).
The composer Clarence Barlow worked with these limits extensively and regards that "it is
implausible to expect convincing music with an arbitrarily large prime limit without an
appropriately developed, well-considered musical grammar for the system that results from it".45
His comment raises the question about the minimum amount of harmonicity46 needed in a piece of

42 A critical bandwidth refers to the narrow range which the ear can perceive two different pitches or intervals as equivalent.

43Siu-Lan Tan, Peter Pfordresher and Rom Harré, Psychology of Music, Psychology Press, 2010, p. 98: "Frequency is a continuous
variable, with an in nite number of possible values, and so it would be inef cient for our auditory system to respond selectively to
every possible value."

44 To be totally accurate, one sine tone tuned is tuned to( f(0)* 11/2), meaning that the piece has a -11 limit dimension.

45 Clarence Barlow, On Musiquantics, Musikinformatik & Medientechnik, Musikwissenschaftliches Institute der Johannes
Gutenberg-Universität Mainz, Report No.51, Mainz, 2012, p.39. Barlow considers that it is mostly relevant to set "the minimum
harmonicity to 0.02, allowing to encounter no fewer than 256 intervals in the semitone range of 550 to 650 cents."

46Barlow, On Musiquantics, ibid. p.61:" harmonicity ('intervallic clarity') stems from the numerical simplicity of the ratio between
the two frequencies of an interval, whereby timbre is of hardly any signi cance; string music, for instance, can be re-instrumentated
for winds and/or transposed to extreme registers without losing its harmonic meaning." The harmonicity-algorithm allows to order
chords according to their degrees of harmonicity, or 'scaled-harmonicity' and to implement similarities of interval chord structures
and proximity in chord progression by adjacent intervals and/or common tones.
34
fi
fi

fi
fi
music. Harmonicity, as de ned by Barlow, is a psychological phenomena of "intervallic clarity" for
listeners, and more broadly the harmonic sense given to an interval. For Barlow, harmonicity is
independent from timbre and distinguished from the physiological and timbre-dependent
phenomena of consonance or dissonance.47 Rather, harmonicity relies on "contextually adjusted
listening"48 and the uniqueness of a listener's attention. Harmonicity is also dependent on the
context of a given composition, and particularly the time span given in the music for listeners to
grasp the composition's harmonic complexity. In the speci c context of rationally tuned music, time
may allow listeners to develop a certain tuning tolerance when confronted with complex harmonic
relationships which can then allow them to proceed to a more or less conscious "rationalisation" of
the notes and pitch intervals. Interestingly, Barlow considers that in "harmonically more complex
music, the minimum harmonicity should be set lower than in more harmonically simple music;
lower sensibility results in a higher tuning tolerance." In other words, in harmonically simple music,
a very complex tuning may still be understood and clear for listeners, but in harmonically complex
music, such tuning would probably not be grasped.

By composing for edgars I learned much about this topic, especially when it comes to
understanding music through the lens of harmonicity and how harmonicity plays a signi cant role
in the listening experience of my piece. My compositional standpoint was precisely not to x any
minimum "harmonicity" for the electronics in the middle of the piece, and to instead play around
with qualities of transparency or a certain blurriness of harmonic relationships. That said, the
question remains whether the beginning and end of the piece were long and harmonically simple
enough to ground the listeners with intervals that could be easily discerned—namely ones that allow
for musical sense to emerge from the microscopic and variable harmonic intervals found at the
middle of the piece.

47 Barlow, On Musiquantics, ibid. p.61. "consonance refers to the ‘smoothness’ (dissonance to the ‘roughness’) of a sound. In the
lowest piano octave, a perfect fourth – due to the larger interval size of the critical bandwidth there – can be shown to be more
dissonant than a tritone, a whole-tone more dissonant than a semitone. The psychological phenomenon harmonicity originates in the
brain’s time-related perception of neuron ring: the physiological phenomenon consonance originates in distances on the ear’s basilar
membrane"

48 Barlow, On Musiquantics, ibid. p.61.

35
fi
fi

fi
fi
fi
 3.5.3. Auditory masking (similarity principle

The similarity principle states that we tend to group element together when they appear to be
similar. This principle nds a speci c parallel in for edgars with its intermittent effect of auditory
masking. Masking occurs when the perception of one sound is affected by the presence of another
sound.49 Masking acts on our perception of timbre, in this case the combination of sine tones and
the guitar, one masking or being absorbed by the other. Partial masking by one sound on another
results in a reduced perceived loudness of the weaker partial. It can be "a very complex process,
where amplitude and frequency of both masking and masked sounds interact in ways which are far
from linear, and are also strongly affected by phenomena such as beats and difference tones."50
Initially, considerations about auditory masking in uenced the composition of for edgars's
spatiality. This is most apparent at the beginning and end of the piece, as I intended to emphasize
the timbral separation between the guitar (having its own ampli er on the right side of the room)
and the electronics (two loudspeakers, left and center). Furthermore, as the piece unfolds, my plan
was to increase the volume of the central loudspeaker playing the sine tones, which have the lowest
frequencies, to create a strong auditory masking effect with the guitar sounds. Though the reality
was that the perceptual distinction between the two sound sources in the piece was, at times, very
strong, and far blurrier at other times. Masking is thus highly unpredictable in for edgars and is like
this for several reasons: the harmonics of the guitar are sometimes weaker than the sine tones, and,
conversely, the sine tones can sometimes disappear in the spectral richness of the guitar’s open
strings. Moreover, the sharp attacks of the guitar sounds either get absorbed into the sine tones or
come suddenly to the foreground. As a result of these circumstances the timbral merging between
the two layers appears and disappears in unforeseeable ways.

Last but not least, I realized in for edgars how timbral masking in uences the perception of
the just intonation's periodic signature of the piece, or its beating effects. In theory, the
psychological phenomena of harmonicity or "intervallic clarity" is independent from timbre. But it
is also true that a clean timbre serves (or suggests) a clean intonation. In other words, the cleaner the
timbre and the simpler the rational harmonic relationships are the more graspable they will be for
listeners. This is illustrated at the beginning and the end of for edgars, where the periodic patterns
of tunings (a) or (b) come out well and are coherent with the timbral simplicity of sine tones and the
guitar harmonics. The two sound layers reinforce an overall harmonic simplicity or each other in

49 Wikipedia, Auditory Masking, https://en.wikipedia.org/wiki/Auditory_masking

50Pedro Manuel Branco dos Santos Bento, The Harpsichord: Its Timbre, Its Tuning Process, and Their Interrelations, University of
Edinburgh, 2013, p.58.
36
fi
fi
)

fl

fi
fl
their "in-tune-ness," largely because the sine tones are sustained in a "clean" intonation thus acting
particularly like an anchor for listeners, providing a stable, sonic context to hear the tuning of the
piece. However, when the intonation is less clean in the middle of the piece, it is as if the guitar and
sine tones start working against one another, "unmasking" each other's presence and their
differences of tunings. The sine tones also start misleading the listener’s ear as the sonic context
they initially provided breaks apart
These last few remarks reveals another paradox present in the piece. Besides the clashing of
what I called earlier harmonic relevance and harmonic irrelevance in the structure piece, there is a
clash between "clean" timbre or transparent sound sources like sine tones and the harmonics of the
guitar, with "unclean" intonation. I found these unfamiliar, strange musical territories quite
interesting and explored them further in the next piece I will discuss, for blandine and maciej.

3.5.4. Melodic contour & Uniform probability distribution (ergodicity)

One aspect of the piece which may be the easiest to process for the listener concerns
grasping the work’s different macro-sequences. Speci cally these are so-called moments of
"plateaux" where the guitar is retuned and the electronics are also kept steady, primarily for the
purpose of announcing new stages in the piece, where new sets of pitches and harmonics are
introduced on the guitar. Due to the nature of the available pitch sets, and how they evolve
throughout the piece, it is hypothetically possible to hear more complex harmonic and melodic
relationships on the guitar at the beginning of the piece, and a more simple harmonic cloud in the
middle. But within each of these macro-sequences, the piece’s melodic contour is dif cult to grasp,
probably because of the use of random processes for pitches solely using a quasi-uniform
probability distribution throughout the entirety of the piece. For each section or each available pitch
set, the pitches had a more or less equal chance to be selected by the algorithm. As a result, when
considering for edgars as a whole, one may mostly retain an overall ergodicity in pitch: everything
is "different" yet everything sounds the same. I came up with the conclusion that composing/
shaping the probability distribution for weighing the random selection of notes in the piece seemed
crucial from a listener’s perspective.

37
.

fi
fi

 Conclusio

In retrospect, there was room for progress in the piece. For instance, I could have gone much
further in the variety of harmonic material in the piece, or incorporated 'harmonicity' as a parameter
in the piece. I could also have counterbalanced better the degrees of complexity of harmonic
relationships with degrees of "timbral clarity" — especially in my way of using the sine waves in
the piece. Finally, playing with auditory masking could have been bene cial, for instance by
drastically separating the guitar part from the electronics at some moments in the piece. These
intuitions were eventually explored in my next piece for blandine & maciej. In the meantime, I was
curious to see how other composers had dealt with rational tuning systems and probabilities in their
music; and composers I found in uential in this regard will be examined in the next chapter

38
n

fl
fi
.

 4. Pioneering "indeterminate" pieces examined through computability and

computationality

Before discussing more recent music, it is interesting to go back to John Cage's Number
Pieces (1988) and Xenakis's Herma (1965).51 Both of these works are so-called "indeterminate"
pieces, applying aleatoric procedures for generating pitch material. I will present the two pieces in
this section for the sake of illustrating the process of translations and transitions between
mathematical concepts, musical concepts and percepts — from an analytical perspective. Indeed,
what I nd most interesting is that these two pieces have both recently been analyzed through the
lens of computability and/or algorithmic information theories and Gestalt theory. Thanks to the
speci c methods used, the analysis of the two pieces shed new light on the paradoxical aspects
inherent to both of them. This includes Herma's effects on listening in relation to Xenakis's
compositional intentions and methods; and, with respect to the Number Pieces, more recent analysis
adds nuances to Cage's compositional standpoint and personal approach to understanding harmony.
Further, by making Xenakis and Cage's original compositional intentions "revisable," these analysis
result in a reversal of a commonly held outlook towards their pieces and how their aleatoric
dimension for generating pitch material may be understood.

To sum up, it is my hope to underscore throughout this section the existence of dynamics of
opposite logics and polarities, such as determinism and indeterminism, simplicity versus
complexity, and composing and analyzing. All of these are important, and in my opinion are present
in Herma, the Number Pieces, their analysis, as well as my own work.
.

51 See Appendix 3.1 and 3.2.

39
fi
fi

 4.1. Iannis Xenakis's Herma, Gestalts & Computable analysis

For me, Herma has much to do with the workings of a "speci c" creativity consisting of an
interplay between knowledge and ignorance. Xenakis's adventurous and brave compositional
process in Herma encapsulates the imperfect emergence of knowledge and composition/music — or
how one has to work constantly in the unknown in order to actually learn from and under these
conditions. Herma is inspiring for me for this reason. It provided me with the courage to develop
and to compose in a similar intermediary, dynamic, and creative state, oscillating between
knowledge and ignorance. I also consider Herma's experimental spirit as a template; one that
forgives any erroneous ideas about one's work.

4.1.1. General presentation of the piece

Written in 1961, the piano piece Herma is a turning point in Iannis Xenakis's musical path.
Besides being his rst piece for solo instrument, the piece departs from the solely stochastic means
of composition previously he used. Herma instead raises compositional questions having to do with
the potential to simply "deduce" music from mathematics and mathematics from music. Xenakis
started from concepts drawn from set theory to derive the structure and score of Herma, applying
logical operations to its musical parameters — mainly pitch. Herma illustrates the composer's
notion of Symbolic Music: a music that would allow a listener to "reason by pinning down our
thoughts by means of sound"52 —in other words, a music that conveys a complete transparency
between mathematical concept(s) and percept.53 As explained in his writings about the piece,
Xenakis was aiming at a perfect consistency and correspondence between the three poles of
mathematical concepts/model (sets and stochastic processes), musical concept (mainly related to
pitch), and how music is heard by listeners. Hence Herma is an interesting case study to examine if
such a complete transparency is actually valid/tenable

In reality, the piece revolves on a coexistence of both transparency and ambiguity. On the
one hand, Herma is undeniably numerically driven: its compositional methods originate from
rigorous mathematical concepts and the complexity of the music when heard could be
spontaneously interpreted as resulting from the mathematical dimension of the piece. Hence on a

52Iannis Xenakis, Formalized Music, Thought and Mathematics in Music. Harmonologio Series No.6, Revised edition, Pendragon
Press, Stuyvesant NY, 1992, p.172

53This reminds of Fred Lehrdal and Ray Jackendoff's music cognition model where the existence of an "ideal listener" is assumed to
justify the af nity between the rules of listening and those of composing, through an objective structure which can be successfully
communicated through the music. See Fred Lerdahl and Ray Jackendoff, A Generative Theory of Tonal Music, MIT Press, 1982.
40

fi
fi
.

fi
 super cial level, the piece conveys a transparency between mathematical concept(s) and percept.
However, when examined closer, one realizes that the compositional methods used by Xenakis
entailed some aesthetic and technical contradictions in relation to his initial intentions, touching
upon the way the piece was made and the listening experiences it was supposed to provide. Thus
Herma illustrates an intricate relation between mathematical concepts, musical concepts and
percept: it is transparent and it is not and this, in my opinion, makes it a fascinating work to analyze.

4.1.2. Xenakis's compositional methodology and musical strategies in Herma

4.1.2.1. Compositional methodology: a set theoretic," imaginary experiment in

outside time

First, I wish to highlight the existence of a speci c compositional methodology in Herma —

already encountered in Arbor Vitae and that I am also investigating in my work. Although Herma,
Arbor Vitae and for edgars have (very) different musical grounds and motivations, they share a
methodology that implies the following steps. The rst step takes place prior to the musical time
proper. It consists in de ning the set of all potential pitches in the piece and then all the subsets of
pitches and the (mainly combinatoric) operations that can be applied to them. The following step
consists in presenting the results of the set operations dynamically through probabilistic processes,
i.e. sequentially organize pitches into random sequences. Finally, these random sequences are
eventually transformed into sonic events.

pitch combinations
pitch sets/subsets results of sets operations into
sequenced in time thanks
new pitch combinations
set operations to stochastic processes

prior to the musical time musical tim

Fig 4.1: A methodology common to Xenakis, Tenney and

Indeed, in Formalized Music, Xenakis explains having conducted an "imaginary

experiment" to compose Herma, abstracted from any musical temporality, unfolding in what the
composer calls the outside-time.54 The experiment was supposed to embody basic concepts of set
and computability theories. The composer rst de ned, stated and sketched a map of "state classes

54 Iannis Xenakis, Formalized Music, Thought and Mathematics in Music, ibid. p.166 "an algebration of sonic events that is
independent of time (algebra outside-time)" VS "an algebration of sonic events as a function of time (algebra in-time)."
41
fi
"

fi
fi
fi
fi
fi
I

 of sonic events"55, corresponding to the set theoretic features of Herma. Typically, Xenakis adopted
as a universal pitch set R, the set of all 88 keys on the standard piano keyboard. He then derived
three subsets of pitches A, B, and C from R, to which he applied speci c elementary set operations:
intersection, union, complementation. The results of these operations between the three pitch sets
de ne the macro-structure of the piece, each section corresponding to a speci c mathematical
relationship. Then, the individual pitches in each section proceeds from stochastic processes: the
pitches are picked at random from the given sets and/or their combinations, without registral
preferences.

Herma's speci city has, on the one hand, much to do with the roles Xenakis gave to musical
time ("in-time") and listening, and on the other hand, with the musical strategies he used.

4.1.2.2. Role of mathematics in Herm

4.1.2.2.1. Questioning the possibility of an "immediate comprehension" of

mathematical concepts through sonic events

When reading Xenakis's original writings about Herma, one can discern speci c functions
attributed to sonic events and perhaps listening by the composer. It seems that listening to sonic
events is considered by Xenakis as a direct, immediate way of knowing or having access to the
mathematics beneath the piece. At least, this is my take on Xenakis's idea when he writes that
Herma's pitch sets and their mathematical operations applied to them were meant to be "deduced
mentally by the observer"56. As Xenakis invites the listeners to undertake an "intellectual task" of
looking for such relations while simultaneously listening to Herma, it seems that listening should be
equivalent to an "immediate comprehension"57 of a conceptual, "deduced" knowledge for the

55 Iannis Xenakis, Formalized Music, ibid., p.171.

56 ibid.

57 Iannis Xenakis, Formalized Music, ibid pp.171-172: "an observer must undertake an intellectual task in order to deduce from this
both classes and operations. On our plane of immediate comprehension, we replaced graphic signs by sonic events. We consider
these sonic events as symbols of abstract entities furnished with abstract logical relations on which we may effect at least the
fundamental operations of the logic of classes. We have not allowed special symbols for the statement of the classes; only the sonic
enumeration of the generic elements was allowed (though in certain cases, if the classes are already known and if there is no
ambiguity, shortcuts may be taken in the statement to admit a sort of mnemotechnical or even psychophysiological
stenosymbolization). We have not allowed special sonic symbols for the three operations which are expressed graphically by " +,
- ;only the classes resulting from these operations are expressed, and the operations are consequently deduced mentally by the
observer. In the same way the observer must deduce the relation of equality of the two classes, and the relation of implication based
on the concept of inclusion. The empty class, however, may be symbolized by a duly presented silence." My emphasis.
42
fi

fi
a

fi
fi
fi
 composer. Evidence of this is vaguely implied by Xenakis when he writes that "we can reason by
pinning down our thoughts by means of sound."58
However, this is also where I begin to nd Xenakis's writings (and overall concept of
"Symbolic Logic") puzzling. What I nd problematic is not the idea of deducing some knowledge
from music eventually but the potential confusion between direct knowledge (perception) and
deduced, indirect (intellectual) knowledge. In fact, Xenakis seems to suggest the possibility of an
immediate and deduced knowledge through the listening to sounds all at once when he mentions the
equivocal idea of "a sort of mnemotechnical or even psychophysiological stenosymbolization"59—
which is disputable. Listening is a mode of perception or sensory awareness which may lead to
immediate comprehension. In that regard, I nd the Buddhist epistemology of the Mind helpful.
The latter refers to perception as direct knowledge, and thus acknowledges that the perception of
sounds is direct knowledge. Yet for it to be immediate, such perception or direct knowledge cannot
be at the same time produced by inference, deduction or reasoning, understood in the Buddhist
epistemology as indirect knowledge or mental impressions. In other words, perception cannot lead
to mentally inferred knowledge. This idea, which seems to go against Xenakis's premises in Herma,
is summarized in the Sanskrit Algebras of Catherine Christer Hennix (another important composer
and gure presented in Chapter 5.2), when she writes: "Pure sensations stand out alone in
containing no constructive coordination. It is uncognizable by logical methods."60

4.1.2.3. Mathematics as tools for developing compositional strategies

Putting aside Xenakis's premises of "Symbolic Logic," I would say that the role of
mathematics in Herma is the same as in any other numerically driven music: mathematics just acts
as compositional "tools," used to prompt some perceptual effects, convey speci c mental
impressions to listeners, and nd new ways of making music. For instance, the random ordering of
the individual pitch sequences he used were designed to avoid any melodic or harmonic patterns
that could interfere with the listener's perception of the pitch sets and the logical operations applied
to them. Besides the ordering of the individual pitches, the "intensities and densities, as well as the
silences, are meant to help clarify the levels of the composition."61 All these musical strategies
imply an overall technical dif culty in the piece, resulting into new forms of gesturalities on the

58 ibid.

59 Iannis Xenakis, Formalized Music, p.171.

60 Catherine Christer Hennix, Poësy Matters and Other Matters, Vol. II, New York: Blank Forms Editions, 2019, pp.80-82.

61 Iannis Xenakis, Formalized Music, ibid., p.175.

43
fi
fi
fi
fi
fi
fi

fi
 side of the performer. These gestures are technically very challenging and original, like the way a
performer must play in all the registers of the piano at an extremely high speed or very precise
rhythms, constantly alternating between very big dynamic ranges or timbres (staccato/pedal).

4.1.3. Computable and Gestalt analysis of the piec

When listening to Herma, the piece has a perceptual clarity of its own, characteristic to
Xenakis's style (and perhaps other serialists of that time). The piece sounds at once hectic, dense,
but more importantly: clear. At rst glance, Herma even conveys a clear stochastic "form"62 in
listening, indicating that some aspects of the relation between some of mathematical concepts
beneath the piece and its percept are preserved. Yet, despite its super cial, external characterization
of a random process, the piece presents many aspects of mathematical inconsistency and aesthetic
contradictions. When giving the work a closer look, one may get curious to know whether Xenakis's
musical strategies were:
- mathematically rigorous (translation from a mathematical to musical concepts)
- "musically effective" in the sense of guiding the listeners to grasp the main musical concepts
beneath the piece (translation from musical concepts to the music as heard). This particularly
applies to the grasping of pitch sets and their sequencing in the piece

The above questions have been thoroughly exposed by the composer and the music theorist
Robert Wannamaker, speci cally in his meticulous study of Herma,63 which certainly helped to
complement some previous re ections I made about the composition for edgars.

4.1.3.1. Computable analysis (translation between mathematical and musical

concepts

Wannamaker's computable analysis reveals more broadly the tension between using
mathematics as tools for music-making, and/or the consistency with which these tools are used. To
me and from a general perspective, such tensions invite a composer to shape their own view on the
role of mathematics in their music.

62Expression taken from Sam Goree, Structure and Randomness in Iannis Xenakis’ Analogique A, Musical Capstone Thesis, 2017,
p.34: about Analogique A/B where "the micro-level of the piece was [most likely] composed intuitively, without the use of stochastic
techniques, to t a stochastic form."

63Robert A. Wannamaker, "Structure and Perception in Herma by Iannis Xenakis". In Music Theory Online, Society for Music
Theory, Vol.7, No. 3, May 2001, [1.3].
44
)

fi

fi
fl
fi

fi
;

 With this in mind, in theory and ideally, Herma's musical structure should stand for a
mathematical demonstration related to sets, set operations and stochastic processes. All these
operations and processes are perfectly computable and easily translatable into musical material.
However, and contrary to his writings, Xenakis did not rigorously follow the mathematical model
from which the piece was supposedly built upon.
To begin with, the validity of some of Herma's set-theoretic and stochastic features is open
to doubt: both the set-algebraic computations and stochastic processes exposed in Formalized Music
are not always in line with the actual score of Herma. Besides, what seems to be an arbitrary choice
made by Xenakis concerning the three sets of pitches A, B and C, reveals inconsistencies that,
according to Wannamaker: "appear to persist throughout the score based on the set-algebraic
computations…in which signi cant disagreement between computed sets [of pitches] and their
representations in the score is repeatedly observed."64 By this statement, it appears that certain
pitches are, for instance, present in numerous sections when they should be excluded from them if
one would strictly follow set-algebraic operations described in Formalized Music.
In addition, the "stochastic processes" mentioned by Xenakis to generate melodic sequences
from the pitch sets raise other problems, and these issues exemplify the juncture between
computability and the craft of composition. As compositional "tools," stochastic processes come
into contradiction with Xenakis's musical idea of demonstrating "completed" sets of pitches. Indeed,
from a statistical point of view, it would be extremely rare, if not impossible, that uniformly
distributed stochastic processes would be able to display the complete set of 88 values in such short
sequence of datas (see exposition of the piece: 88 different pitches out of 205 notes).65 This shows
that the resulting sequences of pitches in the piece are in fact highly improbable, suggesting
Xenakis partially composed, or wrote, the notes himself. But whether or not Xenakis did write these
notes himself, what I nd even more interesting is that, in any case, certain pitches are never
sounded in the piece. To me, this hints that Xenakis did not consistently follow the mathematical
model he was after, nor his initial musical idea of covering the entirety of the 88 pitch set.66 He just
did something slightly different.

64 Robert A. Wannamaker, "Structure and Perception in Herma by Iannis Xenakis", [4.12].

65Wannamaker, "Structure and Perception in Herma by Iannis Xenakis", [4.1]: "a computer simulation employing ten million trials
of the pitch lottery has shown that an average of 446 draws are required to obtain all 88 possible pitches (with a standard deviation of
110 draws), and that the probability of obtaining all 88 pitches in 205 or fewer draws is extremely small (about 3 in 100000). "

66ibid, [4.2]: "if the object of the composition is to demonstrate certain set-theoretic facts, then a complete representation of the sets
of interest would seem to be a necessary prerequisite".
45
fi
fi

 4.1.3.2. Gestalt analysis (translation between musical concept to percept

If focusing on Xenakis's musical concept of presenting speci c sets of pitches to the

listeners, Wannamaker believes that "the macro-structure of the pitch sets [is] not the primary
determining factor in low-level temporal gestalt formation."67
Certainly, perceptual units at the sectional level are easily demarcated thanks to radical
changes in dynamics (silences), temporal densities (varying paces) and timbre (pedal). However,
these drastic changes delineating Herma's sections do not necessarily help discriminate the different
pitch sets within each section. Similar to my experience in Arbor Vitae, Herma's density of musical
information on the microscopic level causes a discrepancy between the set-theoretical structure of
such a piece and the structures a listener can perceive and retain in their memory
Then, and this echoes an aspect of my piece for edgars, the application of stochastic
processes in the piece but also the distribution of notes among all the registers of the piano result in
an ergodicity in pitch materials: they sound statistically homogeneous when considering Herma as a
whole. This means that every sample of a given sequence of Herma would be equally representative
of the entirety of the piece in regard to pitch. Thus, for Wannamaker and from an aesthetic point of
view, this overall perceived ergodicity in pitch materials contradicts Xenakis's initial compositional
intentions

Just as I discussed in for edgars, Herma gives an example of challenges encountered when
composing with this speci c "methodology" I presented at the beginning of Chapter 4: sometimes,
the methodology simply does not work according to what a composer had planned.

67 Wannamaker, "Structure and Perception in Herma by Iannis Xenakis", ibid, [3.10].

46
.

fi
)

fi
.

 4.1.4. Conclusion

To me the conceptual and perceptual paradoxes or con icting aesthetics in the piece are not
aws. On the contrary, they participate in sculpting an overarching heterogeneous compositional
idea and musical effect. Xenakis's style is not found in one consistent application of a single
theoretical idea, but in a multiplicity of ideas merged together that create a tension both in listening
and with respect to performing the piece. Additionally, this analysis of Herma reminds me of Henry
Flynt's notion of structure art as opposed to concept art — notions both developed the same year as
Herma, in 1961. The composer, artist and philosopher criticized the idea of music as

sounds made to operate as vehicles for maintaining and communicating an

abstract "structure," irreducible to the sensible presence of the sounds. [As a
result, structure art would imply] the underdevelopment of both the structure and
of the sensible content of the art. Insofar as music seeks to satisfy both poles
(structural and sensible/musical), at the same time, both are diluted: structure
becomes impoverished as it needs to be incarnated in sounds, and music becomes
impoverished to the extent that it obeys a logic external to sounds themselves.68

Like Flynt, and as I mentioned earlier, I do believe that the insights gained from listening to
music cannot be equated with mathematical or inference-based knowledge. The radicalness of
Flynt's perspective is very refreshing in that regard. Yet I disagree with structure art being
necessarily "impoverished." Moreover, Flynt’s critique of structure art focuses the problem on the
music or the art and what it seeks. However, it seems more relevant to me to say that the intentions
of a composer are problematic when they seek to satisfy both the polarities of ‘the structural’ and
‘the sensible’. Also, Flynt’s view overlooks that a composer who seeks to satisfy one of these two
poles can end up incidentally satisfying the other. Pieces of music af liated with concept art, and
dealing primarily with the sensible presence of sounds, may be analyzed as mathematical structures
(and vice versa). Examples like John Cage's Number Pieces, which is another indeterminate piece,
is just one such telling illustration.

68 This is a "politically correct" summary of Flynt's provocative "Essay: Concept Art” found on: J-.P. Caron, "On Constitutive
Dissociations as a Means of World-Unmaking: Henry Flynt and Generative Aesthetics Rede ned," e- ux Journal, Issue #115
February 2021. See also Henry Flynt, "Essay: Concept Art," 1961, HenryFlynt, http://www.henry ynt.org/aesthetics/conart.html
"Contemporary structure artists… tend to claim the kind of cognitive value for their art that conventional contemporary
mathematicians claim for mathematics. Modern examples of structure art are the fugue and total serial music […] These examples
illustrate the important division of structure art into two kinds according to how the structure is appreciated. In the case of a fugue,
one is aware of its structure in listening to it; one imposes "relationships," a categorization (hopefully that intended by the composer)
on the sounds while listening to them, that is, has an "(associated) artistic structure experience" In the case of total serial music, the
structure is such that this cannot be done; one just has to read an "analysis" of the music, de nition of the relationships. Now there
are two things wrong with structure art. First, its cognitive pretensions are utterly wrong. Secondly, by trying to be music or whatever
(which has nothing to do with knowledge), and knowledge represented by structure, structure art both fails, is completely boring, as
music, and doesn't begin to explore the aesthetic possibilities structure can have when freed from trying to be music or whatever. "
My emphasis in italics.
47
fl

fl
fi
fi
fi
fl
fl
:

 4.2. John Cage's Number Pieces & Computational analysi

In the previous section, I focused on the listening experience of Herma in relation to its musical
concepts and compositional methods. I will now focus on Alexandre Popoff's recent computational
methods employed to analyze Cage's Number Pieces,69 mainly drawn from stochastics and
statistics. I will examine how these methods reframe Cage's notion of "indeterminacy" and propose
that Popoff's analysis may be seen as an act of interpretation or even performance of Cage's piece.
In the Number Pieces, Cage expanded his idea of "indeterminacy" by using chance
operations as compositional tools speci cally for building a notion of harmony and resulting into a
certain unknowability for the performer and the listener. Yet, the recent computational methods of
analysis of Cage's Number Pieces tend to make this unknowability relative, since they delineate the
predictable lines in the pieces. By doing so, the analysis also highlights in my opinion the crucial
role of Cage in composing the piece.

4.2.1. General presentation about the piece and Cage's notion of harmon

John Cage's Number Pieces are the latest cycle of pieces written by the composer between
1987 and 1992. The main feature of the Number Pieces' is their "time-bracket" structure. Within
these time-brackets, speci ed pitch intervals and silence can resound without their exact beginning
and end times being precisely de ned in the score (See Fig 4.2).

Fig.4.2: Example of time-bracket notation in Cage's Number Piece

69Alexandre Popoff, "John Cage's Number Pieces as Stochastic Processes: a Large-Scale Analysis." In arXiv: Physics and
Society (2013). See also Appendix 3.2.
48
fi
fi

fi
s

 This technique echoes Cage's notion of indeterminacy in relation to the performance of a piece.70 In
the Number Pieces, the composer used this chance procedure to obtain a myriad of pitch
combinations through the numerous possible realizations of one single score. This arguably
illustrates a reconciliation between Cage and the parameter of harmony. It is an approach that
combines Cage's life-long wish for harmony to be liberated from its classical, functional Western
rules,71 while also reinforcing his interest in composing simultaneous pitched tones. The latter was a
preoccupation he had towards the end of his career, and at that time, Cage considered harmony very
broadly, regarding it as "several sounds...being noticed at the same time, hmm? It’s quite impossible
not to have harmony, hmm?"72 In Cage's music, harmony is thus understood as both the production
and perception of simultaneous sounds — a performative act from the side of both musicians and
listeners.

This speci c notion of harmony is present in the Number Pieces. The chance operations
applied to harmony in the piece prompt introspective states for listeners: the perception of
simultaneous sounds, without no any real grasping sequences of higher-level unity. In that sense, the
Number Pieces do not explicitly offer formal or temporal grounds or a temporal structure which
might guide the listener through the unfolding of the piece. Rather, the treatment of harmony in the
Number Pieces renders the listener oblivious to comprehending any kind of structure or musical
form— and the latter speci cally becomes dissolved into the perception of simultaneous sounds.

4.2.2. Computational analysis: Number Pieces as stochastic processes

Computing numerous simulations of possible performances of such indeterminate pieces

allows for more clarity on the musical structures of these pieces which seem unidenti able at rst.
This is exempli ed in the work of the musicologist Alexandre Popoff, who thanks to computational
methods, studied the structure of Cage's Number Pieces extensively. He described and analyzed
Four and Five (1988) as stochastic processes in his 2013 paper titled 'John Cage's Number Pieces
as Stochastic Processes: a Large-Scale Analysis.' Popoff's research consists in transforming the
Number Pieces into a mathematical structure using combinatorics and computable stochastic

70See John Cage, Silence, Lectures and Writings by John Cage, 50th Anniversary Edition, Wesleyan University Press, Middletown,
2012, p.35 and the concept of an "indeterminate composition in relation to its performance."

71 See John Cage, 45' For a Speaker: "Harmony, so-called, is a forced abstract vertical relation which blots out the spontaneous
transmitting nature of each of the sounds forced into it. It is arti cial and unrealistic" in Silence, Lectures and Writings by John
Cage, 50th Anniversary Edition, Wesleyan University Press, Middletown, 2012, p.152.

72John Cage and Joan Retallack, Musicage: Cage Muses on Words, Art, Music. John Cage in conversation with Joan Retallack, ed.
Joan Retallack, Hanover, NH: University Press of New England, Wesleyan University Press, 1996, p. 108.
49

fi
fi
fi
fi

fi
fi

processes. He states that, “[b]y averaging over a large number of realizations (which is achieved
through a computer program running the determination of the parts repeatedly) we can access the
probability distributions of each pitch-class set over time, thus turning the Number Pieces into
stochastic processes."73 Hence, thanks to computational methods, the musicologist can now turn
possibilities (multiple realizations of a musical concept) into probability and probability into
mathematical concepts and a logical construct. While this chain of events is nothing new, the
methods for analyzing the music are

4.2.2.1. Computational methods of analysi

The way Popoff's reduced the Number Pieces's structure is important to review. He worked
with the following main axioms to analyze Cage's Five and Four
- First, he divided each piece into different sections, and treated each section independently
- Popoff then de ned a set of variables speci c to each section. These variables are the pitch-class
sets74 (including silence) and their starting/ending times within their corresponding time-
brackets. This step is crucial in turning a Number Piece score into a computable function.
- The musicologist used Gaussian distributions for selecting the starting and ending times in the
corresponding intervals of each time-bracket — which is, in passing, a very subjective and
disputable choice.
- The next step of the analysis corresponds to the simulation of numerous realizations of the piece
(between 104 and 105). It consisted in the generation of collections of random variables indexed
over time, with values in the possible pitch-class sets within each time-bracket. This means that
the outer limits of the time-brackets are chosen through a random selection within the given
intervals and that the selection of multiple pitches inside a time-bracket is made through a
succession of random time-mark choices.
- Finally, Popoff went through the computation and quanti cation of the probabilities of
occurrence of the chords in each time bracket, and eventually of a model for pitch-class set
evolution, with the most probable transitions between pitch-class sets for an entire section of a
piece and the whole piece

73Popoff, Alexandre, "John Cage's Number Pieces as Stochastic Processes: a Large-Scale Analysis." In arXiv: Physics and
Society (2013).

74 Popoff uses Allen Forte's notation found in The structure of atonal music, Yale University Press, 1977.
50
fi

fi
fi
:

 4.2.2.2. Uncovering of Number Pieces's harmonic biases

Popoff's study allows for visualizing Four and Five of the Number Pieces as structures, and
this enhances what could be considered as their harmonic/time biases. Instead of a purely uniform
distribution of pitch-class sets, the latter were probabilistically weighted in Four and Five, some
being more prevalent than others.75 Moreover, the probability distributions of occurrence of the
pitch-class sets are not xed but vary over time. These distributions can be either uniform within a
given duration among different pitch classes; or introduce different weights, some of them
signi cantly high for speci c pitch-class sets and low for others within another time-bracket.76
In brief, Popoff's rigorous and detailed analysis allows to rede ne Cage's musical concept of
harmony in the Number Pieces through the lens as mathematical concepts. This was an important
manner of analysis for me and one that further blurred Flynt's limited distinction between structure
art and concept art. Once articulated into mathematical concepts, Cage's way of thinking about
harmony can be understood through a more subtle perspective — to the point where Cage's
statement on harmony may seem simplistic in relation to the Number Pieces. Harmony is not
equivalent to performing or listening to 'any simultaneous sounds.' Rather, harmony is restated as a
limited set of pitches woven with rather speci c intervallic relationships, and also as a controlled
density of possible simultaneous sounds, carefully carved throughout a piece. As Popoff gives a
clear image of the probabilities of occurrence of pitch-class sets and their evolutions throughout the
unfolding of the Number Pieces, his study shows the predictable aspects of Four and Five —
nuancing the idea of the pieces's endlessly "varied" realizations, and fundamentally displaces the
usual point of focus that most of us have concerning Cage's Number Pieces. It also reveals the
tactics of attention that Cage employed in the piece: how he carried the listeners without them
noticing that their attention is directed in a rather speci c way. Although Popoff explicitly mentions
his study as not being perceptually motivated, imagining/studying the potential perceptual effects of
Four and Five on listeners, thanks to their harmonic pro les, could be another interesting avenue
for further understanding the piece

75For example, in Five Popoff studies con rm the prevalence of pitch-class sets 3-7 ([0, 2, 5]), 3-2 ([0, 1, 4]), 3-10 ([0, 3, 6]) and
4-27 ([0,2,5,8]) in Forte notation, corresponding respectively to the presence of [(M2 or m7 + P4 or P5)], [(m2 or M7 + m3 or M6)],
[(m3 or M6 + A4 or d5)] and [(M2 or m7 + P4 or P5 + m7 or M2)]. In brief, the intervals that have the most chance to occur are
major seconds/minor sevenths, minor thirds/major sixth, major sevenths/minor seconds, perfect fourths/perfect fths.

76 See Popoff's speci c analysis of the rst time-bracket in Five, ibid., pp.13-19.
51
fi
fi
fi
fi
fi
fi
.

fi

fi
fi
fi
fi
 Fig 4.3: Popoff's heatmap of the probabilities over the 87 possible pitch class sets (in
ordinate) at each time t in Five.

4.2.2.3. Computational analysis as interpretation/performanc

Popoff's analysis puts aside the (essential) reality faced by musicians when playing the piece
and listening to one another, especially the assumption that the entrance and exit times should
follow a Gaussian Curve. This view on the performance of the piece is actually restrictive and
biased. But because of it, the paper raises interesting questions about the notions of interpretation
and performance, namely, I view Popoff becoming a kind of interpreter/performer of Cage's score. I
would also claim that his approach in many ways resembles my own analysis of Arbor Vitae. To
me, this illustrates how the boundaries between musical analysis and performance can become
blurred through the use of computers. And in fact, I found out much later that Popoff started a "365-
days projects" in 2020 where he posted each day an algorithmically generated version of John
Cage’s Five on Youtube, based on his algorithmic reconstruction of the piece.77 Despite stopping his
posting after 47 days, the format indicates Popoff's position on the piece was not solely analytical
anymore but also very performative and artistic.

77Alexandre Popoff, John Cage, Five - A 365 days project, Alpof, September 2020, https://alpof.wordpress.com/2020/09/05/john-
cages- ve-a-365-days-project/. See also Alexandre Popoff, The Number Pieces of John Cage (7), Alpof, January 2017, https://
alpof.wordpress.com/2017/01/13/the-number-pieces-of-john-cage-7/.
52
fi

 Finally, Popoff's studies indirectly enhance how musically and computationally ef cient and
elegant his compositional model is — even if that was not Cage's primary intention. The Number
Pieces provide complex output based on simple instructions, one being an ideal situation for
computability theorists and computer scientists. I discovered only much later that some of the
Number Pieces were partially computed thanks to Andrew Culver's program musicfor, which
generates time brackets pitches, dynamics, and "specials".78 As much as I do not have enough
information about how this program was used by Cage, this anecdote con rms that the general
consensus about Cage's music —including my own— still needs to be refreshed
Last but not least, rather than prompting me to create my own algorithmic reconstruction of the
piece,79 Popoff's paper has certainly inspired some new musical avenues that I could explore in the
future. For example, I consider engaging with compositional strategies similar to the Number Pieces
but starting from programming and thus from an awareness of the structure of the piece — one that
I could share with the instrumentalists. This composition could combine:
- a Cagean indeterminate, aleatoric dimension in relation to its performance, allowing for
performers to work freely and independently from each other
- and a rigorous structure, fundamentally be based on algorithmic, probabilistic processes. If this
structure would be known by the performers, I could imagine it may not only inform their way
of performing but also create a sense of collective cohesion and understanding.
Overall, I believe the above points are compositional interests that I am already exploring,
although in an embryonic form (as solo pieces). Presently, a future aim of mine would be to expand
on these ideas further with several instrumentalists.

78 See Anarchic Harmony, Cage Programs https://www.anarchicharmony.org/People/Culver/CagePrograms.html.

79This has been done by Popoff and also Benny Sluchin, Mikhail Malt, A computer aided interpretation interface for John Cage’s
Number Piece Two Journées d’Informatique Musicale (JIM 2012)„ 2012, Mons, Belgium.https://hal.archives-ouvertes.fr/
hal-01580153/document.
53

fi
.

fi
 5. Other parallel trajectories between musical and mathematical creativities

The pieces I will present in this chapter were all in uential in my work as they literally
"blew my mind" in different ways: through their music, their structure, their approach to harmony,
their philosophical grounds, their distinct connexions to mathematics. The aim then in this section
will be to illustrate further the singularities with which one composer may undertake these
processes of transition between mathematical concepts, musical concepts and percepts.

5.1. Clarence Barlow, Approximating Pi (2007)80

It seems to me that a particular branch of computer music consists of presenting the

computation of mathematical objects as a musical process. This shows another, speci c take on how
mathematical objects may be used for music derivation. I nd two particularly illustrative examples
of this are the two works Approximating Pi (2007) and Approximating Omega (2010), respectively
written by the composers Clarence Barlow and Michael Winter. I will go back to Approximating
Omega in the last section of this chapter, but as indicated in their titles, these two pieces deal with
approximation. Computability theorists often work with approximation, i.e. partial computations to
represent mathematical objects whose exact form or exact numerical number is unknown or dif cult
to obtain,81 in this case the trans nite number Pi or the highly complex and truly random number,
Omega. Barlow and Winter's pieces thus work in the realm of the 'non-computable' via partial
computations, tying together musical and approximation processes

The computer-based electronic music installation, Approximating Pi, illustrates a point of

junction between just intonation, mathematics, computability theory, programming, and harmony as
a perceptual phenomena. In contrast to Xenakis’s Herma for instance, which is somewhat
controversial in that regard, the structure of Approximating Pi is fully determined by computations,
i.e. those for approximating Pi. As a reminder, Pi may be represented as the following series: π=
4(1- 1/3 +1/4 - 1/7 + 1/9…), forming a convergent sequence gradually approaching a limit point.
The limit of this sequence is the value each term of the sequence tends towards, indicating when an
irrational number such as Pi nds a form of stability where it can be approximated. In that case, Pi
converges towards 3.1415926535. In Approximating Pi, Barlow proceeds to the computation of this
series as a sample-based synthesis procedure where the "amplitudes of ten partials of a complex

80 Clarence Barlow, Approximating Pi, Compositions, http://clarlow.org/compositions-by-year/approximating-pi-8ch15/ and

Akshay1992, "ApproximatingPi", Github, https://github.com/akshay1992/ApproximatingPi. See Appendix 3.3.

81 Wikipedia, "Approximation" https://en.wikipedia.org/wiki/Approximation.

54
fi
fi
fi
fl
.

fi
fi
 tone would re ect the rst ten digits of the progressive stages of convergence [of Pi] – with each
added component of the series – toward the nal value."82 This mathematical and musical process
gradually reconstructs a stable saw tooth spectrum containing a set of ten partials of an overtone
series.83

Fig 5.1: The digits of the rst 1000 approximations of π, shown to the tenth digit, as amplitudes of a
10-partial spectrum in Barlow's Approximating Pi.84

As a listener, a process of "convergence" towards a tonal stability is made very transparent.

The listener is rst invited to follow the timbral complexity of the wave forms, their construction by
means of a gradual timbral stabilization, and this allows them to also gradually attune to the
harmonic relationships present in the piece

A very different illustration of a composer's relation to mathematics from the ones evoked
until now is found in Catherine Christer Hennix's work

82Clarence Barlow, "Algorithmic Composition, illustrated by my own work: A review of the period 1971-2008" in Proceedings of
Korean Electro-Acoustic Music Society's 2011 Annual Conference (KEAMSAC2011), Seoul, Korea, 22-23 October 2011, p.8.

83 ibid.: The piece was realized wholly in Linux GNU Pascal. Barlow "decided on a time frame of 5040 samples, i.e. 83⁄4 frames per
second. All time frames contain a set of ten partials of an overtone series, multiples of the frequency 83⁄4 Hz, which automatically
results from the width of the frame, 5040 samples."

84 ibid.
55

fl
fi
fi
fi
.

fi

 5.2. Catherine Christer Hennix and Brouwerian continuum

One could (and should) dedicate an entire research project exclusively to studying the
Norwegian composer, musician, poet, artist, mathematician, philosopher, Catherine Christer
Hennix. Her work results from an assemblage of various artistic practices (jazz, raga, maqam, blues,
poetry, noh theater), and teachings (mathematics, logic, buddhism, muskim su sm) — and this is
probably only the top of a much broader iceberg. For me, her body of works represents the epitome
of "transdisciplinarism," a highly personal artistic practice and one that is also involved with
political and social engagements of the times. If there is little, proper documentation and research
about her work, it is still worth mentioning.85 In general, Hennix's art may be described as attempts
to wed mathematical formalism and introspective states of mind.86 Logic and numbers are not only
tools to model music in Hennix's works; rather, they point at a much broader "mathematization" of
art-making and way of being. In brief, Hennix's relation to mathematics is existential, and this is
especially shown in her music.

5.2.1. Catherine Christer Hennix & intuitionis

Hennix's work is permeated by so-called intuitionism, a radically novel —and controversial

— conception of mathematics and mathematical creativity consisting of a mathematical theory of
introspection, founded by the mathematician Luitzen Egbertus Jan Brouwer around the early 1900s.
As a philosophical position, intuitionism puts the "creative subject" and its psyche at the centre of
the mathematical activity, where it can be understood as the fruit of internal acts of thinking.
Because reasoning is always attached to a speci c mathematical context and the creative subject,
intuitionists acknowledge mathematical notions have only a provisional, relative existence. In
Hennix's music, the role of the composer and performer are inspired from the Brouwerian "creative
subject," and this emphasises putting one’s intuitions at the center of a rigorous creative process.
Thus similar to the Brouwerian mathematical activity, the musical activity, is interpreted as the
result of internal or introspective states of mind; it is highly subjective, revisable, and offers
transitory solutions to a musical inquiry

85 In general, her work is not very well known or documented and this is partially due to Hennix's indifference to releasing her music
in conventional recording formats.

86Spencer. Gerhardt, "Domains of Variation: Choice Sequences, Continuously Variable Sets, Remarks on the Yellow Book." In Blank
Forms Journal, Vol. 4: Intelligent Life, Blank Forms Editions, 2019, p. 102.
56

fi

fi
 5.2.2. Engaging with formalizations of the continuu

If Barlow's Approximating Pi had to do with the mathematical formalization of irrational

numbers, Hennix's music mostly engages with mathematical formalizations of the continuum, i.e.
"theories or models that explain gradual transitions from one condition to another without abrupt
changes or discontinuities."87 This engagement was nourished by two intermingled sources

5.2.2.1. Continuum and Raga: "attunement" to musical intervals

Hennix's interest in formalizing continuous phenomena and in nity in her music stems from
her encounter with drone music and Indian raga — a melodic framework for improvisation
contained as a substructure within the tambura sound.88 In an interview, Hennix explains that "You
get your rst intuitive acquaintance with in nity through the raga and then mathematics ampli es
this concept of in nity by teaching you to formally manipulate it on paper with symbols."89
Hennix's musicianship results from a process of rst attuning to the perception of justly tuned
harmonic relationships and/or partials over a xed fundamental, sometimes intermingled in a
complex sonic event, like the tambura drone. On this basis, Hennix then formalizes her musical
compositions, ones that are endebted to mathematics. I can relate to this approach, as seen in my
attempts to get "attuned" to Arbor Vitae or the tuning systems I work with, and how my pieces are
also developed on the basis of simple mathematical concepts.
Attunement is radically put in the forefront of Hennix's work. Rather than xed notated
pieces, Hennix's music evolves through a given harmonic framework, reminding of La Monte
Young way of considering composition as

a mode of generation, always developing in time through the performer’s

attention. Intervals are not treated as relations of completed points in musical
space, but rather un nished sequences of observation, subject to further
re nement.90

This is shown in the way Hennix systematically and rigorously improvises with acoustic
instruments through their acoustical properties, or self-made justly tuned synthesizers like in her

87Wikipedia, "Continuum", https://en.wikipedia.org/wiki/Continuum. This is the most basic de nition I found to avoid getting too
deeply into philosophical/mathematical "abysses" right away!

De nition inspired from Hennix's own de nition in Marcus Boon, "Basically One to In nity: an Interview with Catherine Christer
88

Hennix." Blank Forms Journal, Vol. 2: Music From the World Tomorrow, Blank Forms Editions. New York, 2018, pp.121–142.

89 Marcus Boon, "Basically One to In nity: an Interview with Catherine Christer Hennix", ibid.

90 Gerhardt Spencer, "Minimalism and Foundations," ibid. p.236.

57
fi
fi
fi
fi
fi
fi
fi
:

fi
m

fi
fi

fi
fi
fi
fi
.

fi
 early Electronic Harpsichord.91 Moreover, attunement is what is primarily present in her work and
shared with the listeners, developing in time according to their attention. So despite Hennix’s works
owing their origin to inference and logic, they propose an epistemological perspective that is
fundamentally apperceptive — ways of knowing exclusively based on perception.

5.2.2.2. Continuum and Brouwerian's choice sequences

When getting to know Hennix's compositional methods, one can observe one main
compositional pattern, i.e. how her treatment of musical intervals is in uenced by some
foundational shifts or oscillations in the mathematics she works with. Spontaneously, I believe that
my approach to musical intervals is similarly in uenced by such an foundational oscillation in the
mathematics I am inspired by (mainly, between nite and in nite probabilistic spaces)
Frequently, Hennix's compositions express a shift from classical to intuitionistic
mathematics, or from constant structures to variable ones.92 Mainly, Hennix's music and ways of
working with musical intervals are partially inspired by set theory but mainly by the Brouwer's
notion of continuum. Based on a disapproval of the classical, set-theoretic, reduction of the world's
phenomenon to discrete terms and completed sets,93 the Brouwerian continuum is constituted
through choice sequences. The latter are blurred mathematical structures which are never being
fully nished and dynamically emerge or develop in time according to an idealized mathematician's
attention and intuition.94 More precisely, these sequences are constructed from a nite set of
elements, where each element is successively and freely chosen by a creating subject in a
potentially in nite process. Choice sequences "may follow pre-ordained rules (law-like sequences),
while others are generated quite freely by the subject (a lawless sequence),"95 meaning that they are
in fact random.
Delving into choice sequences is beyond the scope of this thesis but one can intuit how they
shaped Hennix's work. The latter evolve between rigorous modal musics (law-like sequences), her

91 The AmbientFox, Catherine Christer Hennix - The Electric Harpsichord, https://youtu.be/eXxobmct4xY . See Appendix 3.4.

92Gerhardt Spencer, "Domains of Variation: Choice Sequences, Continuously Variable Sets, Remarks on the Yellow Book", ibid
p.130.

93 Set theory provides an understanding of the world mainly in discrete or disconnected terms, and primarily tied to constant
structures. In this context, the mathematical continuum is more of an additional structure added upon a pre-existing discrete
collection of objects; and mathematical objects in general are seen as objective entities ruled by logic and deduction, independent of
the thinking mathematician,.

94Gerhardt Spencer, "Domains of Variation: Choice Sequences, Continuously Variable Sets, Remarks on the Yellow Book", ibid
p.114.

95Gerhardt Spencer, "Minimalism and Foundations" in Simplicity. Ideals of Practice in Mathematics and the Arts, New York:
Springer, 2017, p.235.
58
fi
fi
fi
fl

fi
fl
.

fi
 own compositional decisions and/or improvisations with other musicians (lawless sequences).
Often, new musical modes emerge in Hennix's music by the way she layers together just intonation
modalities from traditional, rigorous modal musics (raga, maqam, blues) in one piece, through
improvisation. Additionally, Hennix's compositions are developed thanks to computer generated
processes. In fact, Hennix treats a computer as an idealized mathematician with perfect memory and
inde nite attention, which can proceed with "in nitely long spreads of musical events, locked
together by some appropriate algorithm that recursively generates each new step on the basis of the
preceding ones."96 The latter quote highlights the importance of computable and recursive functions
in Hennix's compositional process, functions which were introduced multiple times throughout this
thesis as well.

For me, some aspects of Hennix's radical approach to mathematics resonate with the work of
Michael Winter, which will illustrate a last approach in this chapter to the process of translating
mathematical objects into musical structure.

5.3. Michael Winter, Approximating Omega (2010)97

Michael Winter's Approximating Omega synthesizes some of the ideas found in Barlow and
Hennix's work. Winter derives computer-aided and open scores in the form of 'harmonic framework'
for the performers, from the very speci cities of a mathematical object, Omega.

5.3.1. In uence of digital philosophy

A radical aspect of the work of Winter's is that it nds its source in digital philosophy, and
the hypothesis/basis where everything in life, any phenomena, any experience can be theoretically
computable. Admittedly, Winter is highly in uenced by Algorithmic Information Theory (AIT),
founded by the mathematician Gregory Chaitin, whose playfulness with mathematics is reminiscent
of Brouwerian's intuitionism and thus an approach to mathematics that foregrounds introspective
practices. For example, Chaitin equates the limits of reasoning with the limits of computability and
calls a mathematician trying to de ne randomness as "a rational mind trying to nd its own

96 Spencer Gerhardt, "Minimalism and Foundations," ibid. p.235.

97Listen to the piece here: Muirgene Leonore Gourgues, Approximating Omega, Edition Wandelweiser Records, 2019 https://
soundcloud.com/muirgeneleonoregourgues/approximating-omega and an excerpt. See Appendix 3.5.
59
fi
fl

fi
fi

fl
fi
fi

fi
limits."98 Chaitin also presents a very different mathematical paradigm whose foundations are
randomness and probabilities. For him, AIT implies:

that some mathematical truths are fundamentally probabilistic, that there is

randomness in the foundations of mathematics, that the truth content of some
statements in pure mathematics is grey, not black or white, that sometimes truth is
probabilistic, not sharp and clear […] AIT shows that randomness and
probabilities arise naturally in pure mathematics[…].99

Chaitin goes even further by saying that "the reason why mathematicians are changing their
working habits"100 is due to the computer, which has turned mathematics into a far more
"experimental"101 science, one that is gradually imitating methods employed in physics.

Winter explains that in his music he is not interested in demonstrating a mathematical

concept to the listeners. At the same time, his music is also not exclusively based on the mere
perception of sounds, but lies somewhere in-between. This is expressed in the rst part of
Approximating Omega, which is an optional part of the piece and may be accompanied by a reading
of texts by Chaitin (to give more context about the mathematical aspect of the work). A major
takeaway from this for me was that a possible didactic dimension of the work remains optional.

98 Gregory Chaitin, Exploring Randomness, Springer, 2001, p.162.

99 ibid.

100 ibid.

101 ibid.
60

fi

 5.3.2. Deriving a piece's structure from the structure of a computer program

The second, and non-optional, part of the piece is fully derived from the structure of the
computer program that approximates the truly random, highly complex number Omega — found in
the work of Chaitin. Omega is the probability for a computer program to halt. In its essence, Omega
is random, and its randomness challenges the limits of computability. Yet, as a value it is able to be
computed thanks to a very sophisticated computer program with multiple nested routines — akin to
20-piled matryoshka
In conclusion, dolls. As a comparison,
permissiveness in our LISP is Arbor
achievedVitae's
because nestedness,
functions withwhich was already
extra arguments
are evaluated but ignored and empty lists are supplied for missing arguments. There are no
fascinatingerroneous
into
expressions; only expressions that never return a value because the interpreter goes
andanchallenging for me to understand, had less than ve such nested routines (which
infinite loop.

were themselves obviously

Approximating much simpler mathematical processes).
Omega:

Section 2:

(('(&(V)(('(&(A)(('(&(R)(('(&(W)(W('(O)))))('(&(n)(*0(*.(R(Vn())n))))))))('(&(xy)(/(.x)(/(.y)()(*0(Rx(-
y))))(^(R(-x)(y))(*(+x)()))))))))('(&(xyz)(/(.x)(/(.y)(/z('(1))())(A('(0))yz))(/(.y)(Ax('(0))z)(*(=(+x)
(=(+y)z))(A(-x)(-y)(/(+x)(/(+y)1z)(/(+y)z0)))))))))))('(&(xy)(/(.x)(/(.(+(?n('(!(%)))y)))()('(1)))(A(V(
-x)(*0y))(V(-x)(*1y))0)))))

Above is the program approximating Omega given in its ascii representation. Note that the symbol O
represents a list of 1s with a length that determines how many bits of the binary expansion of Omega
are approximated.

Part 0 (time-unit, sound):

Fig.00 5.2:
0,
1,
26, Computer
27,
4
0
52,
53,
3 program
0
78,
79,
22
23
approximating
104,
105,
1
1
130,
131,
1
1
Omega
156,
157,
23
1
in its0 ascii
182,
183, 11
208, representation.
209,
6
0
234,
235,
0
7
260, Michael
261,
0
4
286, 12 Winter,
287, 0
312,
313,
1
0
2, 4 28, 14 54, 9 80, 1 106, 1 132, 1 158, 0 184, 0 210, 7 236, 23 262, 0 288, 16 314, 9
3,
4,
0
14
29,
30,
0
20
55,
56,
5
0
81,
82,
0
11
Approximating
107,
108,
1
0
133,
134,
1
1
Omega,
159, 11
160, 24
185,score
186, 23
5 (p.5
211,
212,
23
1
237,
238,
1
2
263, 14
264, 0
289,
290,
1
1
315, 3
316, 23
5, 0 31, 1 57, 19 83, 0 109, 15 135, 1 161, 0 187, 1 213, 24 239, 24 265, 22 291, 1 317, 1
6, 17 32, 0 58, 0 84, 5 110, 0 136, 1 162, 4 188, 0 214, 1 240, 1 266, 23 292, 23 318, 1
7, 1 33, 20 59, 17 85, 22 111, 19 137, 0 163, 0 189, 18 215, 1 241, 0 267, 1 293, 1 319, 0
8, 0 34, 0 60, 25 86, 1 112, 0 138, 4 164, 2 190, 22 216, 0 242, 11 268, 0 294, 1 320, 17
9,
10,
0
4
In contrast to Barlow’s Approximating Pi, the piece is not, as such, about the mathematical
35,
36,
4
0
61,
62,
0
1
87,
88,
0
11
113,
114,
8
22
139,
140,
0
14
165,
166,
1
1
191,
192,
0
4
217,
218,
18
0
243,
244,
0
7
269,
270,
11
0
295,
296,
1
0
321,
322,
0
8
11, 0 37, 21 63, 1 89, 0 115, 1 141, 0 167, 0 193, 0 219, 8 245, 23 271, 5 297, 1 323, 22
12, 14 38, 1 64, 25 90, 5 116, 0 142, 22 168, 1 194, 3 220, 22 246, 1 272, 22 298, 0 324, 1
object Omega but rather the nestedness of the routines present in the computer program
13, 0 39, 1 65, 1 91, 23 117, 8 143, 23 169, 1 195, 1 221, 1 247, 24 273, 1 299, 4 325, 0
14, 18 40, 1 66, 1 92, 1 118, 23 144, 24 170, 0 196, 1 222, 0 248, 3 274, 0 300, 0 326, 9
15, 1 41, 1 67, 1 93, 0 119, 1 145, 1 171, 18 197, 24 223, 8 249, 1 275, 11 301, 2 327, 2
approximating Omega.
16,
17,
0
0
42,
43,
1
0
68,
69,
1
1
102 In that regard, this part of the piece illustrates a very rigorous translation
94,
95,
1
0
120,
121,
1
0
146,
147,
0
11
172,
173,
0
4
198,
199,
1
0
224,
225,
23
1
250,
251,
1
1
276,
277,
0
5
302,
303,
1
1
328,
329,
23
1
18, 4 44, 4 70, 1 96, 9 122, 9 148, 0 174, 0 200, 9 226, 0 252, 1 278, 0 304, 1 330, 1
19, 0 45, 0 71, 1 97, 3 123, 0 149, 5 175, 3 201, 0 227, 11 253, 1 279, 7 305, 0 331, 3
symbol-by-symbol from the computer program's structure (see Fig. 5.2), into sonic events: distinct
20, 14 46, 14 72, 1 98, 0 124, 7 150, 22 176, 1 202, 6 228, 0 254, 1 280, 0 306, 18 332, 1
21, 0 47, 0 73, 0 99, 19 125, 22 151, 1 177, 1 203, 0 229, 7 255, 1 281, 13 307, 0 333, 1
22, 19 48, 25 74, 4 100, 22 126, 1 152, 0 178, 23 204, 7 230, 22 256, 1 282, 25 308, 17 334, 1
(or primarily short) sounds representing each symbol and the entrances and exits of various
23,
24,
1
0
49,
50,
1
0
75,
76,
0
14
101,
102,
0
8
127,
128,
0
1
153,
154,
11
0
179,
180,
24
1
205,
206,
22
1
231,
232,
1
0
257,
258,
1
1
283,
284,
0
4
309,
310,
0
8
335,
336,
1
1
25, 0 51, 9 77, 0 103, 23 129, 1 155, 5 181, 1 207, 0 233, 11 259, 1 285, 0 311, 22

continuous
!
sounds that demarcate the expressions in the program. Furthermore, Winter attempts5! to
capture the "phenomenon" of realizing this incomputable mathematical object into music.

Apart from a structure borrowed from the program itself, the transcription of a
computer program that approximates a maximally complex number such as
Omega is perhaps the most pure realization of such a mathematical object since
any computation or subset does not so fully encapsulate the phenomenon.103

102 Michael Winter, Approximating Omega, unboundedpress, 2010.

103 ibid.
61
)

fi

 Fig. 5.3: Michael Winter, excerpt of instructions of Approximating Omega (p.6)

The above instructions in Fig.5.3 consist of directions similar to a Cagean open form score,
exclusively providing pitch classes and start and end times, depending on the structure of the
computer program. It is important to note that the more nested the computer program, the denser the
sounding of the piece. The piece puts an emphasis on harmony as a perceptual phenomenon, basing
it on rational intonation and sustained tones. Thus to me, this harmonic framework is again slightly
reminiscent of Hennix work and, more hypothetically, Brouwerian's choice sequences

62

 Fig 5.4: Michael Winter, excerpt of the score of Approximating Omega (p.7)

I believe Approximating Omega is a unique sort of computer music where incomputability

plays a central role in its raison d'être. It also derives music from the nesting of routines of a
computer program — a reversed approach to nestedness than the one found in Arbor Vitae, where
nested routines resulted from imagining how the music could sound.

In summary, all these ideas are very inspiring to my work and offer a broader perspective on
how and what 'computer music' or 'mathematically driven music' can be like, that being the
subtleties and gradations between deterministic and non-deterministic compositional strategies, both
in computer-generated sonic processes or for realizing scores for performers

63
.

 6. Arbor Vitae, soni catio

My analysis immersed me in Arbor Vitae's algorithm and this was a signi cant experiment,
one that forced me to take a closer look at the "gradations and the back-and-forth of the concepts"
used in the piece. Moreover, this allowed me to see how the emergence of 'musico-mathematical'
thought can be observed.104 What I nd captivating in the piece is that Tenney goes well beyond an
initial set theoretical construction for pitch; instead he built a much broader musical edi ce from
elements and con gurations of diverse complexity and mathematical disciplines (mainly set theory,
stochastics, probabilities). In 2022, I nally soni ed my take on the piece's algorithm using sine
tones,105 which was another meaningful experience in different ways and doing this greatly
informed, for blandine, my last piece composed for this research project

6.1. Two distinct aesthetic experiences

In contrast to the string quartet version, which I see as an "embellished" version of the
piece's structure, enveloped by the timbre of the string instruments and expressive changes in
dynamics, I have always believed I could listen to another version of the piece, one reduced to its
"bare skeleton." In other words, I have always wanted to bypass the timbral complexity in the piece
and get a more direct access to the harmonic relationships and tuning, unaffected by the mass of
sound characteristic of string quartets. The synthesized version is and sounds much more precise
and this con rmed my previous experiences in for edgars, that being, an extremely "clean" timbre
allows for more precision in performing and grasping harmonic relationships.
Additionally, listening to the two versions of Arbor Vitae successively brought to light some
of my other compositional ideas. For example, I started re ecting on the possibility of emphasizing
the distinctness between an instrumental part and an electronic part in a piece, while suggesting—
through listening—how the two sound layers stem from the same structural and conceptual
grounds.

104Paraphrasing Fernando Zalamea's expression in Introduction to Lautman Albert, Mathematics, Ideas and the Physical Real,
Continuum International Publishing Group, 2011, pp.xxiv- xxvi.

105 See Appendix 1.3.

64
fi
fi
fi
n

fi
fi
fi
fl
.

fi
fi
 6.2.A form of transcription & interpretation of the piece

Similar to what I noticed with Popoff's study and 365-days act with Cage's Number Pieces, I
found out that reconstructing the piece was a form of transcription and interpretation of the piece's
original algorithm. The transcription part is easily understood: as I did not have access to the
original program of the piece, I just transcribed it to a different programming language. The process
of coding a precise, and as close a reconstruction of the piece as possible, was paved with
uncertainty and some overall ambiguity. In reality, I occasionally had to take certain "interpretative"
decisions about the algorithm of Arbor Vitae. These decisions had an impact on the output of my
reconstructed algorithm, and later, its soni cation. For instance, my soni cation ignored dynamic
changes. Certainly, I felt inspired by the generative aspect of this reconstruction, and the ease with
which one can realize and listen to many more renderings of the same piece (because it is based on
probabilistic processes). In short, I would like to expand on these ideas for future works,
speci cally, I would be curious to try running the algorithm with a different synthesis, a different
tuning system, a different length and overall pacing, etc. In other words: re-interpreting the piece's
musical concept.

6.3. Imagination

Re ecting on Tenney's compositional method—starting from algorithms and without

"hearing" any sonorous renderings of them for several months—I realized how his method entailed
not only risks but also a necessity to trust the eventual sounding of the piece. The piece was the last
one Tenney ever wrote, mostly from the hospital with the assistance of Michael Winter. After
working on the algorithm quite extensively, Tenney listened to a synthesized version of the piece
with sine tones and was ecstatic. He passed away a couple of weeks after, before hearing any
performance of the piece. This is not a mere anecdote to me, but crucially illustrates the composer's
ultimate compositional method, combining highly degrees of formalization and imagination, of
algorithmic processes and inner listening — and again, trust

65
fi
fl

fi

fi
 7. for blandine and maciej (2022

The composition for blandine and maciej106 is a piece for harpsichord and computer. The
piece encapsulates many of the areas explored throughout this thesis: rational tuning systems and
harmony, probabilistic processes for generating computer-based music/scores, composing/analyzing
musical structures through the lens of mathematics and computable processes, and exploring the
perceptual effects of a piece. Overall, working on this piece reminded me of Tenney's idea of
indeterminacy as "points in a circle," which involved (in this case) revisiting some of my own
compositional and listening experiences, and to circulate between different bodies of knowledge.
Through this chapter, it is my hope that the reader may follow the constellation of ideas that
ourished from and after writing this piece. Initially, the piece started from a similar approach to
tuning and probabilities as found in for edgars. Gradually, the piece deviated towards another
approach to the harpsichord, taking into consideration some aspects of its timbre but also its
historical dimension. The latter led me to engage with a notation system inspired by L. Couperin's
Unmeasured preludes, adding more dimensions and exibility to the mathematical concepts from
which the piece originated. The mixture or tensions between computer music and early music
conventions occurs in the performance as well as in the listening of the piece. In this piece, I ended
up exploring the merging of computational creativity in music with baroque music conventions;
also largely due to the involvement of the very open minded and gifted harpsichordist Maciej
Skrzeczkowski.

7.1. Brief description of the piece

The piece initially took shape following the recent passing away of the harpsichordist
Blandine Verlet, a member of my family whose way of playing the harpsichord and interpreting
music and sound were personally very in uential. Differently from composing for edgars, which
was devoid of considerations about the historical speci cities of the electric guitar, I was curious to
examine such historical aspects of the harpsichord, its conventional role and expressivity in early
music pieces. Perhaps more than any other instrument, I am particularly fascinated by the
harpsichord because of how the creation of phrasing is intimately bound up with the instrument's
timbre and resonance

106 See Appendix 4.5.

66
fl

fl
fl
fi
 Similar to for edgars, the piece combines two probabilistic spaces ( nite/in nite) resulting
in two different musical logics (harmonic relevance and harmonic irrelevance), that are expressed
in two diametrically different acoustic and electronic sound layers.

7.1.1. Harpsichord par

I wrote for a double-manual harpsichord with the two keyboards using slightly different
extended just intonation systems for each keyboard.107 Throughout the piece, the harpsichordist
transitioned from one keyboard to the other, and from high notes to lower ones. This simple
trajectory allowed me to work with the full range and timbre of the instrument.108 Rather
unknowingly, I also developed a very unconventional and challenging technique (never seen by
Maciej before) of simultaneously playing the two keyboards all at once. As many things in my
work, this technique came from not knowing what is doable for the instrumentalist. Eventually, it
brought a certain playfulness for the performer, as well as bene cial limitations for the general
pacing of the piece.

7.1.2. Electronic par

The electronics of the piece consisted of 24 sine waves, initially tuned like the two
keyboards of the harpsichord. In contrast to the harpsichord, the sine waves also remained in the
same mid-high register. They transitioned from one type of tuning to the other thanks to real-time,
probabilistic processes. In the middle of the piece, the frequencies of the sine tones were neither
tuned to one tuning system, nor the other, but somewhere in-between. This introduced additional
microtones other than the ones found when playing the two harpsichord keyboards together. As
discussed in for edgars, the sine tones created auditory beatings, their frequencies randomly
change within small intervals corresponding to the small differences between the two types of
tunings (less than a semitone). Importantly, there were two groups of sine tones in the piece: some
were solely sustained (group A), like in for edgars; while others had short attacks (group B) and
were thus similar to the attacks made by the harpsichord. The decays of group B sine tones were
made longer throughout the piece. By doing so, the simple effect I was after was to mirror the short

107 See tuning instructions in Appendix 4.1. I will not get into the details of the tunings in this Chapter. They are different from for
edgars but follow similar principles. The difference between the two tunings is bellow a semitone. Many thanks to Richard Barrett
for having suggested the idea of using two different tunings on each of the harpsichord's manuals.

108 The 2 keyboards of the harpsichord sound differently, one being more brilliant, the other closer to a luth.
67
t

fi

fi
fi
decay of the harpsichord's highest keys, and how this decay of gets longer when moving towards
the lower range of the instrument.

7.1.3. Combination of the two sound layer

The combination of the harpsichord and electronic parts, either juxtaposed or sequentially
presented, was the most crucial aspect of the piece. Because this combination affects the function
given to the sine tones in the piece, it may also provide different points of focus for the listener.
If the two parts were played simultaneously, the sine tones would evoke the sound of the
harpsichord while potentially "clashing" with it. At the beginning of the piece, the group A of sine
tones evokes the short attacks/sounds of the harpsichord's highest keys. As the harpsichord part then
moves towards lower registers, the 24 sine tones (group B + A) start acting as "resonators" of the
harpsichord. This juxtaposition required to nd a delicate balance between the limited resonance of
the harpsichord strings and the sustained characteristics in the electronics. Like in for edgars, the
harpsichord was more or less following the general timeline of the electronics, while being loosely
synchronized with it.109 As a listener, I felt that the juxtaposition of the two layers invited me to
mainly focus on an immediate perception of harmonic relationships or the simultaneity of
interactions between the two sound layers
However, if the electronics were played after the harpsichord part, the sine tones were
reminiscent of the harpsichord part and re-informed or transformed the experience of the whole
piece — echoing some of my re ections and experiences with Arbor Vitae. In this con guration, the
durations of each note and pacing on the harpsichord part remained completely up to the performer
(incidentally, this way of working was more coherent in my piece, unmesured preludes, see Section
7.3). Then, the sine tones in for blandine and maciej endorsed two functions — related to how sine
tones are very basic sounds (with the lowest amount of timbral information). First, the sine tones
evoked a reduced representation of the harpsichord’s timbre (i.e. short attacks and short decays on
the highest registers, and slightly longer in the lower register, with its resonance being maintained
by sustaining notes). But what I found was that the sine tones also conveyed a reduced
representation of the piece's structure, which consisted mostly of pitch material that accumulated as
melodic units and was composed from transitions of tunings that did not change in register. In this
case, as a listener, I could grasp the structural similarities between the two parts while also
experiencing their distinctness. Thus this sequential presentation prompted an awareness of the

109I had written an open score where the performer spent 120 seconds per page. They had to get used to play for 9'00" and to stop
exactly at the same time as the electronics. The freedom of the performer was conditioned by external temporal boundaries.
68
fl

fi
fi

 music's structure based on previous listening experiences and a succession of "mental impressions"
— what Flynt calls (associated) artistic structure experience.
Eventually though, I opted for an in-between solution: the harpsichord part was played rst
by itself, then a sparse version of the electronics was introduced on a distant loudspeaker at a
pianissimo dynamic during the second half of the harpsichord part. Finally, the electronics were
soloed at the end of the piece, from pianissimo to mezzo-piano.

7.2. Multiple re-writings/transcription of the scor

Working on this piece raised many questions related to computer-aided composition,

notation, interpretation, and clashes of early music and algorithmic music. To illustrate some of
these questions, I will share three versions of the same page of the score110 (corresponding to the
3'45"- 5'40" section in the recording of the piece111). Each version shows a gradual discovery, both
for the performer and I, of how to work with the exible but bounded parameter of durations in the
piece — an ironic outcome as I had not originally planned to explore this aspect of composition so
intensely.

7.2.1. First version of the scor

Bare output of SuperCollider algorithm: random sequences of pitche

The algorithm I worked with is very similar to the one in for edgars. Following the idea of
transitioning from one tuning to the another, different sets of pitches are sequentially used for the
random selection of each individual pitches. I then mainly implemented weighing functions in my
algorithm. The latter gradually added or subtracted pitches throughout the piece by giving varying
weights to certain pitches. In general, more available pitches for the random selection were made
towards the middle of the piece, and fewer pitches can be found at the beginning and end. This had
an in uence on how pitches were successively repeated (or not) throughout the piece: less
successive repetitions at the middle of the piece, many more at the beginning and end. Additionally,
overall, the pitches were deployed rst on the upper register of the harpsichord and gradually
transposed down, including to the lowest notes of the instruments.
Each stave in for blandine and maciej corresponded to speci c probability distributions.
Two types of probability distribution were used: uniform distribution (allowing for each of the

110 See Appendix 4.1 for the full score.

111 See Appendix 4.5.

69
fl

fi
e

fl

fi
s

fi
available pitches of the random selection to have an equal chance of being selected) or a distribution
based on Barlow's harmonicity function. This was done to order the available pitches from simpler
to more complex harmonic relation to the root note of the piece (G); and then the pitches that were
harmonically simpler would have more chance to be picked by the algorithm. In practice, when
using harmonicity probability distribution, the resulting random sequences of pitches conveyed a
stronger presence to 'purer' intervals (octaves, fths…) and had more cadential nature than those
sequences of pitches computed with uniform distribution

No durations were generated with my probabilistic algorithm, only sequences of pitches.

Thus in the rst version of the score, see Fig.7.1, note values do not represent duration. Also, the
different ratios above the notes simply indicates the tuning systems

70
fi
fi
.

 Fig. 7.1: for blandine, p.3, rst initial version

71
fi
 7.2.2. Second version of the scor
Manual re-write: phrasing pitch sequences by grouping notes into non-repeating phrase

To rework the score, I rst coloured the different ratios to indicate explicitly on which
keyboard each note was to be played: yellow corresponded to the upper manual and green to the
lower. For practical/technical reasons, I left the rest of the notes uncoloured, so that the performer
may choose on which keyboard to play a given note. I then grouped notes into non-repeating
phrases of pitches — taking into consideration that differences in tuning would result in different
pitch material.

Fig. 7.2.a: for blandine, p.3, (zoomed-in) second version

This grouping was inspired by L. Couperin's unmeasured preludes notation (see Fig. 7.4),
consisting of slurs and curves. It became my main compositional strategy and discovery in the
piece, developing a rigid structure (determined notes in the score structured through slurs), while
offering bounded possibilities of "free phrasing" to a performer. Rather than precise articulation
instructions, I initially used this notation to suggest two main processes of differentiation of
structural/non-structural notes: one for the performer, one for the listeners. Big slurs would de ne
longer groups of notes and higher levels of perceptual units while shorter groups of notes smaller
slurs de ne lower levels of perceptual units for the listeners. As a result of my algorithm and its
random selection process, longer melodic sequences are found in the middle of the piece and shorter
melodic sequences are present at the beginning and end of the piece.

72

fi

fi

fi
Fig. 7.2.b: for blandine, p.3, second version of with slurs and colors

73
 7.2.3. Third version of the scor
Maciej Skrzeczkowski's re-reading/annotating of the score

Maciej Skrzeczkowski, like Edgars Rubenis, worked in the unknown with the piece and we
both found this process interesting. First, Maciej had never worked with just intonation before. In
addition, I initially decided not to share information about the inner workings of the piece before
inviting him to collaborate. For instance, the slurs ambiguously indicated harmonically signi cant
notes in the score, suggesting to the performer that they need to actively undergo a process of
differentiating notes into structural and non-structural categories. Speci cally, the performer needs
to go through a process of harmonic clari cation of the work's pitch material and to interpret these
slurs accordingly. I liken this idea to the analytical interpretation and transcription processes of
Couperin's unmeasured preludes. Considering Maciej's understanding of early music and
harmony,112 I was expecting him to re-shape the overall harmonic/melodic movement in the piece
— which he did with every practice. He could read the piece's slurs, indicating sustained notes (how
long to hold one key) or how to group notes together into musical, melodic units, or separate
successive groups of notes. The performer could make a combination of the latter options,
consequently deciding on the overall pacing of the piece, and add more ornaments.
In his version of the score, his phrasings denotes how he tried to nd harmonically
signi cant notes, technically doable ngerings but also a wish not to "break the sound of the
instrument" — which is what he had been mainly taught to do as a harpsichordist. Yet, when
playing simultaneously the two keyboards all at once, ‘not breaking the sound of the instrument’
can be very dif cult. As Maciej shared having felt "puzzled" by the piece, it is true that the slurs are
at times anything but harmonically and technically logical for the performer. Thus, I can sense in his
annotations to the score an internal dilemma between a wish to convey an overall, continuous
phrasing and how these internal articulations would occasionally work against my original slurs,
which tended to circumvent any sense of continuity and created recurring, intermittent disruptions

112 Using slurs served as a bridge to communicate with early music performers and creating a bridge with my music.
74
fi
fi
e

fi
fi

fi
fi

fi
.

 Fig. 7.3: for blandine, p.3, third version with additional slurs and comments by Maciej
Skrzeczkowski

75
.

 7.3. Listening experiences of for blandine and maciej

7.3.1.Successive recordings of the piece's parts

Similar to the presentation of the three successive scores, sharing three successive
recordings of the piece-in-progress expresses varying points of tension between algorithmic
processes, early music convention, notation, and interpretation that are all central to the piece.

The rst excerpt113 comes from the rst recording made of the harpsichord part, where the
instrument was tuned to Maciej's customized, semi-equal temperament with just thirds. As such, the
recording feels like a "mild" expression of its musical concept. Maciej's musicianship added a
richness to the score and multiple ways to "read" the material. This was especially noticeable in
terms of his rhythmic organization of the material; secondly, by his expressivity which added
baroque gestures (e.g. ornamentation) into his interpretation. That said, the temperament of his
instrument smoothed out the essential element of the piece, i.e. the different tunings of the two
manuals. Lastly, because Maciej approached the work as an early music performer he inevitably
played the score with a conventional harmonic interpretation, for instance, by emphasising bass
notes as roots he weighted these notes to imply some presence of traditional harmony
As we rehearsed the music we began discussing different ways to approach and interpret the
slurs in the score. This included enhancing the ones Maciej played more spontaneously and
conversely avoiding the ones that sounded more "automatic" or conventional. We discussed how to
arrive at different types of uidity when performing the piece by allocating xed durations for each
note (executing the score) and introducing long silences at unexpected times — this contrasted with
Maciej's fear of ‘breaking the sound of the instrument’. In the end, his interpretation of the piece
incorporated a variety of phrasing: in middle of the piece, he used more expressive phrasing (based
on his interpretation of the material), however, at the end of the work he played the music in a more
removed and detached way.
In parallel to this rehearsal process, I was working with simulations of the electronic part.114
The electronics were intended to act as the bare skeleton of the music. Though in contrast to a
"synthesized version" of the harpsichord part, this approach to composing the electronics owed
much to my experiences with analysing and coding an algorithm based on Arbor Vitae and

113 See Appendix 4.2.

114 See Appendix 4.3.

76
fi
fl

fi

fi
.

 experiments with a methodology of composing that dealt solely with tuning possibilities (without
knowing how the piece would sound with the instrument and relying on my imagination and some
degree of trust)

The second recording of the harpsichord part115 was made a few days before the rst
performance of the piece. It is was the rst time that Maciej nally got to try the piece with the right
tuning (with the two manuals tuned differently). For both Maciej and I, the effect of this experience
felt like a striking rediscovery of the piece, as the music spoke in a completely different way. Up to
a certain degree, this unleashed a fresh look on the instrument's vibrancy as well. Nothing was lost
from what or how Maciej had worked previously; simply, his understanding of the piece and way of
playing the instrument became less "personal." In a way, both Maciej and I felt that the piece did
not belong to either of us anymore. We also decided that we needed to continue to work separately
with our own individual imaginations: Maciej on the harpsichord, and myself on the electronics.
Maciej never heard the electronics before and the day of the rst performance of the piece and I had
not tried the electronics with Maciej either. Fortunately, this proved to be a very good decision, as it
led our full discovery of the piece during the premiere. The combination between the two sound
layers was somewhat "spacious," allowing the two parts to coexist in a rather balanced way.

7.3.2. Finalized version of the piec

By using unmeasured notation I had hoped to communicate to the listener a process of

differentiation of non-repeating sequences of notes of varying lengths. However in practice, the
delineation of sequences of notes as perceptual units was admittedly dif cult to recognise. Instead, I
now see that by mixing unmeasured notation with non-repeating groups of notes, a more
paradoxical effect is created one that simultaneously gives rise to audible baroque-like phrasing but
is contrasted by a "stochastic effect". Similar to the music heard in Herma, this "stochastic effect" in
my piece is due to its overall speed and durations, the random ordering of notes and repetitions, and
their wide range in the middle of the piece. For instance, the beginning of for blandine and maciej
features the harpsichord part as if the performer was tuning the instrument; this is due to the
repetition of the same two pitches and the simplicity of their harmonic relationship. Further on, the
piece super cially (or accidentally) conveys something akin to baroque music. However, the piece
eventually leads to a more "disorienting" experience, as the music reveals that it has structurally
nothing to do with baroque music. This is noticeable when the tuning of the second keyboard is

115 See Appendix 4.4.

77

fi
.

fi
fi
fi
fi

fi
 introduced, the piece sounds very unfamiliar, or almost like something is going wrong — this is all
the more ampli ed by the almost undetected pianissimo presence of sine tones in the background of
the piece. But gradually towards the middle of the piece, the two sound layers coherently fall more
into place. When the sine tones are nally heard on their own, they evoke a reduction of the
previous musical experience of the harpsichord. The sine tones perhaps even convey some of the
shared structural and mathematical avenues, which the two parts of the piece originated from, while
still emphasizing their distinctness from the harpsichord. For instance, as sustained tones, the
electronics are foreign to a notion of melodic phrasing, whereas the harpsichord's part consist of
short sounds whose musicality depend primarily on phrasing.

7.4. Delving into the Unmeasured preludes notation

To better understand my compositional ideas and how I sought to combine the harpsichord
and the electronics requires me to explain how I unexpectedly became interested in the history of
Couperin's unmeasured preludes and Dan Tidhar's computational analysis of them.116 I believe this
"digression" supported my own ongoing compositional process — and has opened up more
possibilities for future pieces

7.4.1. Unmeasured scores as a probabilistic scenario: a personal account

After brie y encountering Louis Couperin's scores, working with unmeasured notation
seemed like a logical step, largely because this form of notation expresses a rhythmical looseness or
a form of "indeterminacy in relation to performance". This was also something I had experimented
with in for edgars, and I noticed that it invited the performer to decipher and analyze the rhythmic-
metrical relations of the notes present in the piece. Thus, each performer can nd different ways of
interpreting and even transcribing unmeasured scores — a point that appeared connected to my
experience when reconstructing ("transcribing") Tenney's algorithm to Arbor Vitae. Additionally, I
was attracted by the idea of considering the unmeasured scores in terms of visual and musical
Gestalts. For me, slurs suggested differently sized melodic-harmonic and durational units, and,
more hypothetically, a hierarchy of musical (rhythmical/harmonic) information
Last but not least, I could see how this notation could reinforce the probabilistic dimension
of my piece. To me, an unmeasured score may be seen as a probabilistic scenario that could
potentially be analyzed through the lens of probabilities and/or as stochastic processes. I was

116Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured Preludes.
Technishe Universitaät Berlin, Berlin, December 2004.
78

fi
fl
.

fi

fi
.

 curious where my piece would stand in this regard, as I had considered myself to be composing
with both algorithm probabilistic processes and a form of probabilistic notation. I was also initially
tempted to draw a parallels between unmeasuredness and stochastic processes, similar to Popoff's
analysis of Cage's Number Pieces. However, this idea was rather naive if applied to Couperin's
preludes. After all, the work differs from contemporary open scores and pieces of music, and as
reminded by the harpsichordist and music theorist Philip Chang, "an in nite number of realizations
[of Couperin's preludes] may exist, but not all of them will be tasteful and appropriate."117 In fact,
the rhythmic freedom of the performer is very bound up and contextualized in the inexpressible
French bon goût or baroque musical conventions, requiring a meticulous harmonic analysis of the
piece by the performer

7.4.2. Musicological grounds of unmeasured preludes

7.4.2.1. De nition of a prelude

A prelude is "une composition libre, où l’imagination se livre à tout ce qui se présente à
elle".118 In his analysis of Couperin's unmeasured Prelude 7, Dan Tidhar interestingly highlights the
similarities between Northern Indian Ālap music which helps establishing the "tonality" (the Rāg)
of a performance and baroque preludes.119 Indeed, a prelude introduces the key tonality of
proceeding movements, pragmatically this enables a performer to test the tuning and condition of
the instrument (lute, viola, or harpsichord),120 allowing the performer to warm up while also
preparing the listener to listen. And again similarly to Ālap's improvisatory setting, Tidhar also
reminds that preludes are generally performed with a certain amount of rhythmic-metrical
looseness/ exibility, granting the preluding performer an important role and freedom of
interpretation.121

Philip Chih-Cheng Chang. Analytical and Performative Issues in Selected Unmeasured Preludes by Louis Couperin, University of
117

Rochester, New York, 2011, p.30.

118 This is François Couperin (Louis's nephew)'s de nition of a prelude: "a free composition where the imagination abandons itself to
all that comes to it" François Couperin, L’Art de toucher le clavecin, A Paris, 1717, p. 61.

119Parallel made by Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured
Preludes. ibid. p.60.

120 ibid. p.118.

121 ibid., p.60.

79
fi
fl

fi

fi
 7.4.2.2.Speci city of unmeasured prelude
The unmeasured preludes express this freedom of the performer through a particular system
of notation, mainly developed by Louis Couperin. This notation delineates sequences of pitches
with semibreves on the staves or more occasionally vertical bar lines; and groups pitches in further
harmonic sub-collections with slurs. Couperin's "artistry and mastery of unmeasuredness"122 has to
do with the way his music suggests ordered groups of notes, articulations of harmonic events, and
(which groups of) notes may be stressed or held — only vaguely indicating the length/duration of
each note. Thus the performer de ning the rhythmical pacing of the piece, adding silences,
suspensions or ornamentation

Fig. 7.4: One page of Couperin's prélude non mesuré (André Bauyn's manuscript transcription)123

122Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured Preludes, ibid.,
pp.30-31.

123 Couperin, Louis. Manuscrit Prélude non-mesuré, Manuscrit Rés.Vm7 674–675 (ancienne cote : Vm7 1852 & Vm7 1862), dit
manuscrit Bauyn [Prosper Bauyn d'Angersvillers], Bibliothèque nationale de France, copiste inconnu, daté d'après 1676 : 122 pièces
pour clavecin, 4 pour orgue, 5 pour orchestre. Available on Musicologie Jean-Marc Warszawski, "Biographie Louis Couperin."2004.
https://www.musicologie.org/Biographies/c/couperin_louis.html .
80
fi
.

fi
 7.4.2.2.1. Unmeasured prelude interpretations: "harmonic rhythm"124
The tempo, phrasing, and agogic accents of Couperin's unmeasured preludes primarily
depend on an interdependence between the speci c timbre of the instrument: its limited resonance
and the short decay of each struck note, and the harmonic structure of each groups of notes found in
the score. Indeed, it is unanimously acknowledged that harmony is at the core of any interpretation
of the unmeasured preludes. Thus, in the words of Dan Tidhar, "any decisions on part of the
performer regarding groupings of notes and distribution of emphases will necessarily be perceived
as some sort of metre providing orientation and notions of structure and direction of the music."125

7.4.2.2.2. Functions of slur

Couperin's unmeasured notation always remained mysterious, rst, in its historical origins
and the accuracy of ability to transmit through time, but also as a precise guide for interpretation.
Yet three main functions of the slurs are commonly admitted among performers and music analysts.
First, slurs may indicate sustained notes, which are also harmonically the most crucial. These type
of slurs determine the relative length of each note-head. The second type of slurs simply indicate
grouping of notes, in a melodic context. The third enhances a separation or breath between
successive notes or successive groups or notes. Deciphering the precise functions of these slurs in
Couperin's music always requires a thorough analysis, illustrated in several extensive studies of
individual Preludes.126

7.4.3.Computational analysis of unmeasured preludes

Among them, Dan Tidhar's (2005) analysis of Couperin's unmeasured Prelude Seven127
particularly caught my attention. Tidhar used the tools to from computer science, music cognition
and music analysis to build a computational grammar from the Prelude 7 and emulated one
appropriate performance of the piece according to this grammar. Contrary to Popoff's analysis to
Cage's Number Pieces as stochastic processes, using probabilities to emulate numerous
performances of the piece and derive the piece's statistical analysis, Tidhar's methodology is

124Expression taken from Dan Tidhar, Hierarchical and Deterministic Approach to Music Grammars and its Application
toUnmeasured Preludes, ibid. p.61.

125Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured Preludes, ibid.,
pp.61-62.

126 See Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured Preludes,
ibid. See also Chang, Philip Chih-Cheng. Analytical and Performative Issues in Selected Unmeasured Preludes by Louis Couperin,
University of Rochester, New York, 2011.

127 Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application to Unmeasured Preludes, ibid.
81
s

fi
fi
 deterministic and cognitively motivated. His dissertation tries to illustrate that cognitively structural
notes are those that are the most appropriate in terms of interpretation of the music. For him, "the
[unmeasured prelude] score serves as a model of the composers intentions and as a model for its
interpretation. The resulting music serves as a model of the interpretation (now a certain synthesis
between the composer's and the interpreter's intentions); and this also serves as a model for the
aesthetic process which is involved in the listening experience."128 I nd this way of considering
notation and interpretation especially relevant in for blandine and maciej, as much as the way
Tidhar highlights the role of the performer. In my opinion, an 'unmeasured preluder' has a very
similar role to contemporary performers of open scores, such as Maciej in my piece.
Tidhar's approach to listening experiences is also one which intrigues me. To support his
view upon the unmeasured preludes as a model for the "aesthetic process [involved] in the listening
experience," Tidhar develops the interesting notion of note impression. It seems that the latter very
much resonates with some compositional ideas I had for my piece. A note impression "describes the
psycho-acoustical event in the listener's mind upon listening to an acoustic relation of a note. The
cognitive content of a note impression [depends on] the listener's musical skills, their acquaintance
with the piece, their momentary concentration level, etc." This notation considered as "a
representation of the impressions of all the notes that [a piece] includes."129 So a note impression is
a mental image of a previously heard note or sound. I believe I have played a lot with note
impressions in my own piece, thanks to the partially sequential combination of the harpsichord and
electronic parts.

Based on this notion, Tidhar rst nds in the Prelude the most "structurally signi cant
notes," those which have the highest propensity to be retained by listeners. To do so, Tidhar goes
through a detailed harmonic and structural analysis of the Prelude's notes, putting an emphasis on
the bass notes, their modalities, how they suggest different harmonic units. Tidhar then compared
the complexity and similarities of these harmonic events. All these rules found by the researcher to
analyze the music allow him to deduce a "grammar," i.e. a Parse Tree, supposed to "[capture] the
essential interpretational act of assigning different durations to the identically written notes"130.
Indeed, at the last stage of his dissertation, Tidhar combines this Parse Tree with an interpretation
model for deriving durations and this results in a Weighted Tree. In brief, each note is assigned a
xed duration according to how much it is considered musically signi cant. Typically, the more

128Dan Tidhar, A Hierarchical and Deterministic Approach to Music Grammars and its Application toUnmeasured Preludes, ibid.
p.31.

129 Dan Tidhar, p.25.

130 Dan Tidhar, p.25.

82
fi

fi
fi
fi
fi

fi
 signi cant a note in the Prelude is — according to Tidhar's grammatical rules — the longer they are
sustained in his emulation of the piece and reciprocally. This enables Tidhar to emulate one single
performance of the piece thanks to computational procedures and to assess/evaluate its musical
relevance with other early music experts. For me, this evaluation is rather questionable as it consists
in comparing the interpretation model with other rather vague types of time allocation for each
notes: equal durations, random durations according to a uniform probability distribution. That said,
for the time being, I will leave this aspect of Tidhar's analysis aside.

7.5. Comparing three methodologies and relations to unmeasuredness

To wrap things up, I will go through the different methodologies and relations developed
around the unmeasured notation encountered in this section, starting with Couperin's.
Couperin's compositional process began from his free improvisation, a stream of an ordered
set of notes transcribed into an unmeasured score, leading to a multiplicity of "free" interpretations
with no xed durations but constrained by musical conventions. Computations are obviously absent
from Couperin's music but the model that this notation proposes unleashes a multiplicity of possible
mathematical/algorithmic interpretations.

"Free" improvisation Ordered set of notes + Unmeasured notation

"Free" interpretations Analysis (harmonic rhythm)

Fig 7.5: A ow-chart representing Couperin's unmeasured prelude, from composition to

interpretatio

One such 'algorithmic interpretation' of the Prelude is found in Tidhar's analysis. The latter
starts from an unmeasured score, separating the notes into harmonic units, allowing a thorough
harmonic and structural analysis of the piece and its "note impressions." This analysis is also the
gateway for the derivation of mathematical rules and expressions from the Prelude, compiled into a
deterministic computer program able to generate one determined interpretation of the piece.

83
fi
fi
fl
n

 Ordered set of notes/ Analysis (harmonic rhythm)

Unmeasured notation
harmonic units

One "determined" interpretation Deterministic computer program

Fig 7.6: A ow-chart representing Tidhar's analysis of the Couperin's unmeasured prelude

Finally, my composition of for blandine and maciej, originally started from a computed,
probabilistic generation of notes and an ordered set of notes. These notes were transformed into
"notes impressions" by adding slurs and an unmeasured notation. The latter then led to a
multiplicity of "free" interpretations, with no xed, rigid pre-existing historical / aesthetic
constraints but still required doing a harmonic analysis of the piece to nd out its simple inner
workings. The piece also includes a set of simultaneous, "unordered" sustained sine tones, acting
like reminiscences of the harpsichord's "note impressions." Certainly, the interactions between the
instrumental and electronic parts is still to be explored and enhanced — particularly the moments
when the "notes impressions" on the harpsichord and "tones impressions" in the electronics
ambiguously merge.
Working further on this aspect of the piece would deepen a compositional wish of mine: to
create an intermediary space between the bare perception of harmonic relationships on the one
hand, and, on the other hand, to have mentally inferred knowledge about the music's structure based
on previous listening experiences.

Probabilistic computer (randomly) ordered set of (added) unmeasure

program (generating notes) notes notation

Probabilistic computer
"Free" interpretation(s) Analysis (harmonic rhythm)
program (generating tones)

Fig 7.7: A ow-chart representing for blandine and maciej, from composition to performance

84

fl
fl
d

fi
fi
7

 7.6. Conclusion

The piece for blandine and maciej illustrates the discovery of a compositional strategy, one
existing between probabilities and unmeasuredness, and it also exempli es a work that could be
expanded and deepened. For instance, my use of slurs could be nested together in a much more
harmonically complex and intentional way — reminiscent of Tenney’s use of nestedness in Arbor
Vitae's.131 These slurs could simultaneously allow different harmonic units to manifest themselves.
Moreover, I would like this compositional strategy to be experienced with several performers,
collectively deciphering the function of these slurs and complementing them with additional ones of
their own. This leads me to also articulate the most important nding in this work: a way of relating
with a performer. Beyond the music as such, I recognize that the musical collaboration is very
crucial to me; this is also an interest I suggest with the titles of my works. Not being a musician
myself creates an acute form of interdependency between the performers and I, one where we are
on equal terms, learning and discovering the piece together. I nd this way of working very
motivating and humbling.
Last, but not least, through its successive steps, versions, recordings, the piece was an
occasion for me (and Maciej) to retrace some memories of Blandine Verlet's
"justesse" ("attunement" or " ne-tuning") and "touché," as well as the way she continuously
refreshed and re ned her approach to playing the instrument and music throughout the years.

… Tout au long de sa vie, on apprend, petit à petit, à toucher au plus juste

la façon de traduire ses émotions par un instrument et puis on a l'impression que
petit à petit on s'af ne; alors on a peut-être tord, mais on a l'impression qu'on
s'af ne, donc c'est très agréable de retraduire des musiques que l'on a touchées
pendant beaucoup d'années et d'avoir la sensation qu'on est plus apte à les
traduire.132

131 In that sense, without being able to articulate it clearly, I am intuiting a connexion between Couperin's slurs and Tenney's swells.

132 Marc Zisman, Blandine Verlet, une rencontre-podcast, Couperin "l'ami de toute une vie," 21 Mai 2012, Interview Audio, https://
www.qobuz.com/fr-fr/info/magazine-actualites/rencontres/blandine-verlet-une-rencontre89772 An approximate translation "…
Throughout one's life, we learn, little by little, how to "rightly touch" and translate our emotions through an instrument, and then we
get the feeling that little by little, we re ne ourselves or " ne tune"; so perhaps we are wrong, but we have the feeling that we re ne
ourselves, thus it is very pleasant to revisit some pieces of music that we "touched" for many years, and feeling that we are more able
to translate them."
85

fi

fi
fi

fi
fi
fi
fi
fi
fi

fi
 8. Conclusion (and opening

In September 2020, I remember Richard Barrett asking me what were my artistic intentions.
This question left me dumbfounded and intrigued, as I was not able to pinpoint the speci cities of
my work. From a very personal perspective, I could only intuit a question permeating to this project
and my music, and this was one about "harmony" and what it may be, where it may begin, and
where it may be found. Some answers were found gradually throughout the last two years, others
were only found in the last few days, and the answer to others questions will hopefully follow in the
future.

Importantly though, this project allowed me to take notice that my relation to harmony,
mathematics and listening is quite peculiar. As a non-musician composer, computing rational
intonation systems was my main entry to music theory and the only useful way I have found to
deepen my understanding of musical intervals and harmony. In addition, I have partially developed
my musicianship thanks to, or through, mathematics. Though still in its embryonic stages, my
relation to mathematics is profound. Finding more clarity in mathematics allows me to nd more
clarity in my music and reciprocally. Certainly, nding this clarity will be a very long (and never-
ending) process and this thesis was only a rst step. But it has been a fruitful one: as I have begun
to research and dissect some of the most simple, basic mathematical concepts involving
probabilities and this has already spawned many compositional ideas.

Similarly, this research con rms my strong interest in ef cient computational methods to
create reduced, logical structures that apply to harmony and sound. Contrary to Flynt’s assertion, I
believe that attempting to combine considerations for abstract, mathematical structures and the
sensory presence of sounds, is a compositional path worth pursuing. However, for me, this path
requires one to follow it with exibility; to circumvent being caught exclusively in either the
perceptual pole, or (and perhaps especially) in the conceptual pole. Whether attached to music or
mathematics, to quote the mathematician Albert Lautman, "our conceptions are never more than a
provisional arrangement that allows the mind to go further forward."133 In fact, this thesis is an
example of such a provisional arrangement, one which has been woven from and owes much to the
provisional arrangements made by others; thus provisional does not mean 'unimportant'.

133 Albert Lautman, Mathematics, Ideas and the Physical Real, Continuum International Publishing Group, 2011, p.88.
86

fl
fi
fi
fi
fi

fi
fi
 This leads me to the most signi cant realization of this project. Beneath my research
question, that being the transition between mathematical concept, musical concept and percept, I
had surprisingly missed an essential link in the chain. This missing link revolved around the
relationships developed with people whose musicianship, music, words or teachings inspire my
compositions. I have found in them precious friends and companions — ones who, sometimes
unexpectedly or unknowingly, offered an immense and ineffable guidance.

87

fi

 Bibliography

Aho, Al and Ullman, Jeffrey, Foundations of Computer Science: C Edition, W. H. Freeman,

Computer Science Press, October 1994. ISBN 978-0716782841.

The AmbientFox, "Catherine Christer Hennix - The Electric Harpsichord [1 of 2]," 2002.
Accessed May 10, 2022. YouTube video, 15:00. https://youtu.be/eXxobmct4xY

Charles Amirkhanian, Morning Concert: Composer Jim Tenney, KPFMA-FM, 1976, Accessed
May 10, 2022, https://archive.org/details/MC_1976_01_12, 63'00"-65'00."

Arditti String Quartet "Herma" In Iannis Xenakis: Chamber Music 1955 - 1990, Believe Music,
2016. Accessed May 10, 2022, YouTube video. https://youtu.be/HWDLoc3eCDE 7:23.

Barlow, Clarence. "Algorithmic Composition, illustrated by my own work: A review of the period
1971-2008". In Proceedings of Korean Electro-Acoustic Music Society's 2011 Annual
Conference (KEAMSAC2011), Seoul, Korea, 22-23 October 2011.
———————. On Musiquantics, Musikinformatik & Medientechnik, Musikwissenschaftliches
Institute der Johannes Gutenberg-Universität Mainz, Report No.51, translated by Deborah
Richards, Jay Schwartz, Mainz, 2012.
———————. "On Ramifications of Intonation". In KunstMUSIK, No. 16, Cologne, Germany,
2014, ISSN: 1612-6173.
———————. "Glossary of Terms with Respect to Intonation". In KunstMUSIK, No. 17,
Cologne, Germany, 2015.
———————."Approximating Pi," In Compositions by year, 2010. Accessed May 10, 2022.
http://clarlow.org/compositions-by-year/approximating-pi-8ch15/ 15:14.

Bento, Pedro Manuel Branco dos Santos. The Harpsichord: Its Timbre, Its Tuning Process, and
Their Interrelations, University of Edinburgh, 2013.

Boon, Marcus. "Basically One to Infinity: an Interview with Catherine Christer Hennix." Blank
Forms Journal, Vol. 2: Music From the World Tomorrow, Blank Forms Editions. New York,
2018, pp.121–142.

Cage, John. Silence, Lectures and Writings by John Cage, 50th Anniversary Edition, Wesleyan
University Press, Middletown, 2012.

Caron, Jean-Pierre. "On Constitutive Dissociations as a Means of World-Unmaking: Henry Flynt

and Generative Aesthetics Redefined". e-flux Journal, Issue #115 ,February 2021.
https://www.e-flux.com/journal/115/374421/on-constitutive-dissociations-as-a-means-of-world-
unmaking-henry-flynt-and-generative-aesthetics-redefined/

88

Chaitin, Gregory J. The Limits of Mathematics: A Course on Information Theory and Limits of
Formal Reasoning, New York: Springer, 1998.
———————. Exploring Randomness, New York: Springer, 2001.

Chang, Philip Chih-Cheng. Analytical and Performative Issues in Selected Unmeasured Preludes by
Louis Couperin, University of Rochester, New York, 2011.

Couperin, François. L’Art de toucher le clavecin, A Paris, 1717. Available on the online Catalogue
Générale of the Bibliothèque Nationale de France.
https://gallica.bnf.fr/ark:/12148/btv1b53059259q.image

Couperin, Louis. Manuscrit Prélude non-mesuré, Manuscrit Rés.Vm7 674–675 (ancienne cote :
Vm7 1852 & Vm7 1862), dit manuscrit Bauyn, Bibliothèque nationale de France, copiste
inconnu, daté d'après 1676.

Fallowfield, Ellen. "Multiphonics: Basics", Cellomap, https://cellomap.com/multiphonics-basics/.

Fonville, John. "Ben Johnston's Extended Just Intonation: A Guide for Interpreters." In Perspectives
of New Music, Vol. 29, No. 2. (Summer, 1991), pp. 106-137.

Fiore, Giacomo. "Tuning Theory and Practice in James Tenney’s Works for Guitar." In Music
Theory Spectrum, Vol. 0, Issue 0, pp. 1–19, 2018, Oxford University Press.
DOI: 10.1093/mts/mty022 .
——————— "Heterophonic tunings in the music of Larry Polansky." In Tempo, Vol. 68, Issue
267. January 2014, pp. 29–41. DOI: 10.1017/S0040298213001319 .

Flynt, Henry. "Essay: Concept Art," 1961, http://www.henryflynt.org/aesthetics/conart.html

Fulton, Nora. "Sybilline Gray On Sibylline Gray." In The Poetry Project #260, Feb/March/April
2020.

Gerhardt, Spencer. "Domains of Variation: Choice Sequences, Continuously Variable Sets, Remarks
on the Yellow Book." In Blank Forms Journal, Vol. 4: Intelligent Life, Blank Forms Editions.
New York, 2019, pp.95-150.

Goree, Sam. Structure and Randomness in Iannis Xenakis’ Analogique A, Musical Studies,
Capstone Thesis May 14, 2017.

Hennix, Catherine Crister. Poësy Matters and Other Matters, New York: Blank Forms Editions,
2019.

Josel, F. Seth and Tsao, Ming. The Techniques of Guitar Playing, Kassel: Bärenreiter, 2014.

Kossak, Roman and Philip Ording (Eds). Simplicity. Ideals of Practice in Mathematics and the Arts,
New York: Springer, 2017.
89
Lautman Albert, Mathematics, Ideas and the Physical Real, Continuum International Publishing
Group, Translated by Simon B.Duffy, 2011. ISBN 978-1-4411-4433-1.

Montague, Eugene C. The Limits of Logic: Structure and Aesthetics in Xenakis's Herma, University
of Massachusetts, Amherst, 1995,
http://www.moz.ac.at/~sem/lehre/lib/mat/text/montague-herma/

Muirgene Leonore Gourgues, "Approximating Omega," Edition Wandelweiser Records, 2019,

Accessed May 10, 2022, SoundCloud audio, 33:22.
https://soundcloud.com/muirgeneleonoregourgues/approximating-omega .

Perce, Marcus, Rohrmeir Martin, "Musical Syntax I: Theoretical Perspectives." in Springer

Handbook of Systematic Musicology, Berlin, Springer, pp. 487-505, 2018.
———————."Musical Syntax II: Empirical Perspectives" in Springer Handbook of Systematic
Musicology, Berlin, Springer, pp. 473-486, 2018.

Pelodelperro, "John Cage - Five," by Paul Hillier, Terry Riley, Theatre of Voices, Alan Bennett,
Andrea Fullington, Paul Eliott, John Cage: Litany for the Whale, Harmonia Mundi, 2002.
Accessed May 10, 2022. YouTube video, 5:04.
https://www.youtube.com/watch?v=1W57Eaq4e7g

Polansky, Larry, Alex Barnett and Michael Winter. "A Few More Words About James Tenney:
Dissonant Counterpoint and Statistical Feedback." In Journal of Mathematics and Music,
Vol.5:2 (2011): 63-82, https://doi.org/10.1080/17459737.2011.614732.

Popoff, Alexandre. "John Cage's Number Pieces as Stochastic Processes: a Large-Scale

Analysis." In arXiv: Physics and Society, 2013.
———————. "John Cage, Five - A 365 days project," Alpof, September 2020,
https://alpof.wordpress.com/2020/09/05/john-cages-five-a-365-days-project/
———————. "The Number Pieces of John Cage (7)," Alpof, January 2017,
https://alpof.wordpress.com/2017/01/13/the-number-pieces-of-john-cage-7/

Quatuor Bozzini, Arbor Vitæ, CQB 0806_NUM, January 1, 2008. Accessed May 10, 2022,
Bandcamp Audio, 13:17. https://collectionqb.bandcamp.com/album/arbor-vit .

Rohrmeier, Martin, Musical Expectancy. Bridging Music Theory, Cognitive and Computational
Approaches, ZGMTH 10/2, pp.343–371, 2013. https://www.gmth.de/zeitschrift/artikel/724.aspx

Sabat, Marc and Wolfgang von Schweinitz. "Intonation — An Experimental Application of

Extended Rational Tuning." In Johann Sebastian Bach RICERCAR Musikalisches Opfer 1, 2001.

Stein, Charles. "Being= Space x Action." In Blank Forms Journal, Vol. 5: Aspirations of Madness,
Blank Forms Editions. New York, 2020, pp.89-150.

Tan Siu-Lan, Peter Pfordresher and Rom Harré. Psychology of Music, Psychology Press, New York,
2010.

90
Tenney, James. From Scratch. Writings in Music Theory, Urbana, Chicago, and Springfield:
University of Illinois Press, 2015.

Tidhar, Dan. A Hierarchical and Deterministic Approach to Music Grammars and its Application to
Unmeasured Preludes. Technishe Universitaät Berlin, Berlin, December 2004.
ISBN 3-89825-996-X.

Walter, Caspar Johannes. "Meantone Circles, Texts,

https://www.casparjohanneswalter.de/texts/meantone_circles

Wannamaker, Robert A. "Structure and Perception in Herma by Iannis Xenakis". In Music Theory
Online, Society for Music Theory, Vol.7, No. 3, May 2001.

Warszawski, Jean-Marc. "Biographie Louis Couperin."2004. Accessed May 10, 2022.

https://www.musicologie.org/Biographies/c/couperin_louis.html

Weber, Rebecca. Computability Theory, Student Mathematical Library, Volume 62, Providence:
American Mathematical Society, 2012.

Winter, Michael. "On James Tenney’s Arbor Vitae for String Quartet". In Contemporary Music
Review, Vol. 27, No. 1, Routledge, February 2008, pp.131 - 150.
———————. Structural Metrics: An Epistemology, University of California, Santa Barbara,
2010.
———————. Approximating Omega, unboundedpress, 2010.

Whitehead, Alfred North. "Mathematics as an Element in the History of Thought". In The world of
mathematics, Vol.1, J.R.Newman, Simon & Schuster, New York, 1956 Reprint by Tempus,
Microsoft Press, 1988.
———————. Modes of Thought, The Free Press, New York, First Edition, 1968.

Xenakis, Iannis. Formalized Music, Thought and Mathematics in Music. Harmonologio Series
No.6, Revised edition, Pendragon Press, Stuyvesant NY, 1992. ISBN 0-945193-24-6.

Zalamea, Fernando. Synthetic Philosophy of Contemporary Mathematics, Urbanomic x Sequence

Press, Translated by Zachary Luke Fraser, 2012.

Zenil, Hector. Randomness Through Computation: Some Answers, More Questions, World
Scientific, 2011.

Zisman, Marc. Blandine Verlet, une rencontre-podcast, Couperin "l'ami de toute une vie," 21 Mai
2012, Interview Audio, https://www.qobuz.com/fr-fr/info/magazine-actualites/rencontres/
blandine-verlet-une-rencontre89772

91
 Appendix

1_arborVita
- 1.1. Quatuor Bozzini, Arbor Vitæ, CQB 0806_NUM, January 1, 2008
- 1.2. The reconstruction of the piece's algorithm on SuperCollider
- 1.3. A synthesized realization of the piece made on 17.02.2022.

2_forEdgar
- 2.1. The recording of the piece performed by Edgars Rubenis on 09.04.2021.
- 2.2. Tuning of the strings on the guitar
- 2.3. Tuning scheme with sine tones.
- 2.4. The score of the piece
- 2.5. A video of a rehearsal of the piece at Studio Loos, April 2021.

3_diverseExamples
- 3.1. Iannis Xenakis's Herma, by Claude Helffer in Iannis Xenakis: Chamber Music 1955 - 1990,
Believe Music, 2016
- 3.2. John Cage's Five by Paul Hillier, Terry Riley, Theatre of Voices, Alan Bennett, Andrea
Fullington, Paul Eliott, in John Cage: Litany for the Whale, Harmonia Mundi, 2002
- 3.3. Clarence Barlow's Approximating Pi by Clarence Barlow, Self-release, 2010
- 3.4. Catherine Christer Hennix's The Electric Harpsichord performed by Catherine Christer
Hennix, Die Schachtel, 2010.
- 3.5. Michael Winter's Approximating Omega by Muirgene Leonore Gourgues, "Approximating
Omega," Edition Wandelweiser Records, 2019.

4_forBlandineMacie
- 4.1. The score of the piece
- 4.2. An excerpt of the harpsichord part (not tuned) by Maciej Skrzeczkowski.
- 4.3. An excerpt of the harpsichord part (tuned)by Maciej Skrzeczkowski.
- 4.4. The electronics soloed
- 4.5. The recording of the full piece performed by Maciej Skrzeczkowski at the Discussion
Concert on 30.03.2022.

92
s