Uploaded by edrummond

Man kann einen jeden Begriﬀ, einen jeden Titel, darunter viele Erkenntnisse geh¨ren, o einen logischen Ort nennen. Immanuel Kant [258, p. B 324]

This book’s title subject, The Topos of Music, has been chosen to communicate a double message: First, the Greek word “topos” (τ ´πoς = location, site) alludes to the logical and o transcendental location of the concept of music in the sense of Aristotle’s [20, 592] and Kant’s [258, p. B 324] topic. This view deals with the question of where music is situated as a concept— and hence with the underlying ontological problem: What is the type of being and existence of music? The second message is a more technical understanding insofar as the system of musical signs can be associated with the mathematical theory of topoi, which realizes a powerful synthesis of geometric and logical theories. It laid the foundation of a thorough geometrization of logic and has been successful in central issues of algebraic geometry (Grothendieck, Deligne), independence proofs and intuitionistic logic (Cohen, Lawvere, Kripke). But this second message is intimately entwined with the ﬁrst since the present concept framework of the musical sign system is technically based on topos theory, so the topos of music receives its topos-theoretic foundation. In this perspective, the double message of the book’s title in fact condenses to a uniﬁed intention: to unite philosophical insight with mathematical explicitness. According to Birkh¨user’s initial plan in 1996, this book was ﬁrst conceived as an English a translation of my former book Geometrie der T¨ne [340], since the German original had sufo fered from its restricted access to the international public. However, the scientiﬁc progress since 1989, when it was written, has been considerable in theory and technology. We have known new subjects, such as the denotator concept framework, performance theory, and new software platforms for composition, analysis, and performance, such as RUBATO or OpenMusic. Modeling concepts via the denotator approach in fact results from an intense collaboration of mathematicians and computer scientists in the object-oriented programming paradigm and supported by several international research grants. v

vi Also, the scientiﬁc acceptance of mathematical music theory has grown since its beginnings in the late 1970s. As the ﬁrst acceptance of mathematical music theory was testiﬁed to by von Karajan’s legendary Ostersymposium “Musik und Mathematik” in 1984 in Salzburg [190], so is the signiﬁcantly improved present status of acceptance testiﬁed to by the Fourth Diderot Forum on Mathematics and Music [365] in Paris, Vienna, and Lisbon 1999, which was organized by the European Mathematical Society. The corresponding extension of collaborative eﬀorts in particular entail the inclusion of works by other research groups in this book, such as the “American Set Theory”, the Swedish school of performance research at Stockholm’s KTH, or the research on computer-aided composition at the IRCAM in Paris. Therefore, as a result of these revised conditions, The Topos of Music appears as a vastly extended English update of the original work. The extension is visibly traced in the following parts which are new with respect to [340]: Part II exposes the theory of denotators and forms, part V introduces the topological theories of rhythms and motives, part VIII introduces the structure theory of performance, part IX deals with the expressive semantics of performance in the language of performance operators and stemmata (genealogical trees of successively reﬁned performance), part X is devoted to the description of the RUBATO software platform for representation, analysis, composition, and performance, part XI presents a statistical analysis of musical analysis, part XII concludes the subject of performance with an inverse performance theory, in fact a ﬁrst formalization of the problem of music criticism. This does however not mean that the other parts are just translations of the German text. Considerable progress has been made in most ﬁelds, except the last part XIV which reproduces the status quo in [340]. In particular, the local and global theories have been thoroughly functorialized and thereby introduce an ontological depth and variability of concepts, techniques, and results, which by far transcend the semiotically naive geometric approach in [340]. The present theory is as diﬀerent from the traditional geometric conceptualization as is Grothendieck’s topos theoretic algebraic geometry from classical algebraic geometry in the spirit of Segre, van der Waerden, or Zariski. Beyond this topos-theoretic generalization, the denotator language also introduces a fairly exceptional technique of circular concept constructions. This more precisely is rooted in Finsler’s pioneering work in foundations of set theory [153], a thread which has been rediscovered in modern theoretical computer sciences [4]. The present state of denotator theory rightly could be termed a Galois theory of concepts in the sense that circular deﬁnitions of concepts play the role of conceptual equations (corresponding to algebraic equations in algebraic Galois theory), the solutions of which are concepts instead of algebraic numbers. Accordingly, the mathematical apparatus has been vastly extended, not only in the ﬁeld of topos theory and its intuitionistic logic, but also with regard to general and algebraic topology, ordinary and partial diﬀerential equations, P´lya theory, statistics, multiaﬃne algebra and o functorial algebraic geometry. It is mandatory that these technicalities had to be placed in a more elaborate semiotic perspective. However, this book does not cover the full range of music semiotics, for which the reader is referred to [361]. Of course, such an extension on the technical level has consequences for the readability of the theory. In view of the present volume of over 1300 pages, we could however not even make the attempt to approach a non-technical presentation. This subject is left to subsequent eﬀorts. The critical reader may put the question whether music is really that complex. The answer is yes, and the reason is straightforward: We cannot pretend that Bach, Haydn, Mozart, or Beethoven, just to name some of the most prominent

vii composers, are outstanding geniuses and have elaborated masterworks of eternal value, without trying to understand such singular creations with adequate tools, and this means: of adequate depth and power. After all, understanding God’s ‘composition’, the material universe, cannot be approached without the most sophisticated tools as they have been elaborated in physics, chemistry, and molecular biology. So who is recommended to read this book? A ﬁrst category of readers is evidently the working scientist in the ﬁelds of mathematical music theory, the soft- and hardware engineer in music informatics, but also the mathematician who is interested in new applications from the above ﬁelds of pure mathematics. A second category are those theoretical mathematicians or computer scientists interested in the Galois theory of concepts; they may discover interesting unsolved problems. A third category of potential readers are all those who really want to get an idea of what music is about, of how one may conceptualize and turn into language the “ineﬀable” in music for the common language. Those who insist on the dogma that precision and beauty contradict each other, and that mathematics only produces tautologies and therefore must fail when aiming at substantial knowledge, should not read such a book. Despite the technical character of The Topos of Music, there are at least four diﬀerent approaches to its reading. To begin with, one may read it as a philosophical text, concentrating on the qualitative passages, surﬁng over technical portions and leaving those paragraphs to others. One may also take the book as a dictionary for computational musicology, including its concept framework and the lists of musical objects and processes (such as modulation degrees, contrapuntal steps) in the appendices. Observe however, that not all existing important lists have been included. For example, the list of all-interval series and the list of self-addressed chords are omitted, the reader may ﬁnd these lists in other publications. Thirdly, the working scientist will have to read the full-ﬂedged technicalities. And last, but not least, one may take the book as a source for ideas of how to go on with the whole subject of music. The GPL (General Public License1 ) software sources in the appended CD-ROM may support further development. The prerequisites to a more in-depth reading of this book are these. Generally speaking, a good acquaintance with formal reasoning as mathematics (including formal logic) preconizes, is a conditio sine qua non. As to musicology and music theory, the familiarity with elementary concepts, like chords, motives, rhythm, and also musical notation, as well as a real interest in understanding music and not simply (ab)using it, are recommended. For the more computeroriented passages, familiarity with the paradigm of object-oriented programming is proﬁtable. We have not included the appendix on mathematical basics because it should help the reader get familiar with mathematics, but as an orientation in ﬁelds where the specialized mathematician possibly needs a speciﬁcation of concepts and notation. The appendix was also included to expose the spectrum of mathematics which is needed to tackle the formal problems of computational musicology. It is by no means an overkill of mathematization: We have even omitted some non-trivial ﬁelds, such as statistics or Lambda calculus, for which we have to apologize. There are diﬀerent supporting instances to facilitate orientation in this book. To begin with, the table of contents and an extensive subject and name index may help ﬁnd one’s keywords. Further, following the list of contents, a leitfaden (on page xxix) is included for a generic navigation. Each chapter and section is headed by a summary that oﬀers a ﬁrst orientation

1A

legal matter ﬁle is contained in the book’s CD-ROM, see page xxx.

viii about speciﬁc contents. Finally, the book is also available as a ﬁle ToposOfMusic.pdf with bookmarks and active cross-references in the appended CD-ROM (see page xxx for its contents). This version is also attractive because the ﬁgures’ colors are visible only in this version. In order to obtain a consistent ﬁrst reading, we recommend chapters 1 to 5, and then appendix A: Common Parameter Spaces (appendix B is not mandatory here, though it gives a good and not so technical overview of auditory physiology). After that, the reader may go on with chapter 6 on denotators and then follow the outline of the leitfaden (see page xxix). This book could not have been realized without the engaged support of nineteen collaborators and contributors. Above all, my PhD students Stefan G¨ller and Stefan M¨ller at o u the MultiMedia Laboratory of the Department of Information Technology at the University of A Zurich have collaborated in the production of this book on the levels of the L TEX installation, the ﬁnal production of hundreds of ﬁgures, and the contributions sections 20.2 through 20.5 (G¨ller) and sections 46.3 through 46.3.6.2 (M¨ller). My special gratitude goes to their truly o u collaborative spirit. Contributions to this book have been delivered by (in alphabetic order): By Carlos Agon, and G´rard Assayag (both IRCAM) with their precious Lambda-calculus-oriented presentation e of the object-oriented programming principles in the composition software OpenMusic described in chapter 51, Moreno Andreatta (IRCAM) with an elucidating discourse on the American Set Theory in section 11.5.2 and section 16.3, Jan Beran (Universit¨t Konstanz) with his contria bution to the compositional strategies in his original composition [49] in section 11.5.1.1, as well as with his inspiring work on statistics as reported in chapters 43 and 44, Chantal Buteau (Universit¨t and ETH Z¨rich) with her detailed review of chapter 22, Roberto Ferretti (ETH a u Z¨rich) with his progressive contributions to the algebraic geometry of inverse performance u theory in sections 39.8 and 46.2, Anja Fleischer (Technische Universit¨t Berlin) with her short a but critical preliminaries in chapter 23, Harald Fripertinger (Universit¨t Graz) with his ‘killer’ a formulas concerning enumeration of ﬁnite local and global compositions in sections 11.4, 16.2.2 and appendix C.3.6, J¨rg Garbers (Technische Universit¨t Berlin) with his portation of the o a RUBATO application to Mac OS X, as documented in the screenshots in chapters 40, 41, Werner Hemmert (Inﬁneon) with a very up-to-date presentation of room acoustics in section A.1.1.1 and auditory physiology in appendix B.1 (we would have loved to include more of his knowledge), Michael Leyton (DIMACS, Rutgers University) with a formidable cover ﬁgure entitled “Dark Theory”, a beautiful subtitle to this book, as well as with innumerable discussions around time and its reduction to symmetries as presented in chapter 47, Emilio Lluis Puebla (UNAM, Mexico City) with his unique and engaged promotion and dissipation of mathematical music theory on the American continent, especially also in the preparation and critical review of this book, Mariana Montiel Hernandez (UNAM, Mexico City) with her critical review of the theory of circular forms and denotators in section 6.5 and appendix G.2.2.1, Thomas Noll (Technische Universit¨t Berlin) with his substantial contributions to the functorial theory of a compositions, and for his revolutionary rebuilding of Riemann’s harmony and its relations to counterpoint, Joachim Stange-Elbe (Universit¨t Osnabr¨ck) with a very clear and innovative a u description of his outstanding RUBATO performance of Bach’s contrapunctus III in the Art of Fugue in sections 42.2 through 42.4.3, Hans Straub with his adventurous extensions of classical cadence theory in section 26.2.2 and his classiﬁcation of four-element motives in appendix M.4, and, last but not least, Oliver Zahorka (Out Media Design), my former collaborator and chief programmer of the NeXT RUBATO application, which has contributed so much to the

ix success of the Z¨rich school of performance theory. To all of them, I owe my deepest gratitude u and recognition for their sweat and tears. My sincere acknowledgments go to Alexander Grothendieck, whose encouraging letters and, no doubt, awe inspiring revolution in mathematical thinking has given me so much in isolated phases of this enterprise. My acknowledgments also go to my engaged mentor Peter Stucki, director of the MultiMedia Laboratory of the Department of Information Technology at the University of Zurich; without his support, this book would have seen its birthday years later, if ever. My thanks also go to my brother Silvio, who once again (he did it already for my ﬁrst book [328]) supported the ﬁnal review eﬀorts by an ideal environment in his villa in Vulpera. My thanks also go to the unbureaucratic management of the book’s production by Birkh¨user’s lector Thomas Hempﬂing and the very patient copy editor Edwin Beschler. All a these beautiful supports would have failed without my wife Christina’s inﬁnite understanding and vital environment—if this book is a trace of humanity, it is also, and strongly, hers.

Vulpera, June 2002

Guerino Mazzola

Contents

I Introduction and Orientation 1

3 4 6 9 10 11 12 12 12 13 14 14 16 16 17 17 17 19 21

1 What is Music About? 1.1 Fundamental Activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Fundamental Scientiﬁc Domains . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Topography 2.1 Layers of Reality . . . . . . . . . . . . . . . . 2.1.1 Physical Reality . . . . . . . . . . . . 2.1.2 Mental Reality . . . . . . . . . . . . . 2.1.3 Psychological Reality . . . . . . . . . . 2.2 Molino’s Communication Stream . . . . . . . 2.2.1 Creator and Poietic Level . . . . . . . 2.2.2 Work and Neutral Level . . . . . . . . 2.2.3 Listener and Esthesic Level . . . . . . 2.3 Semiosis . . . . . . . . . . . . . . . . . . . . . 2.3.1 Expressions . . . . . . . . . . . . . . . 2.3.2 Content . . . . . . . . . . . . . . . . . 2.3.3 The Process of Signiﬁcation . . . . . . 2.3.4 A Short Overview of Music Semiotics 2.4 The Cube of Local Topography . . . . . . . . 2.5 Topographical Navigation . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

3 Musical Ontology 23 3.1 Where is Music? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2 Depth and Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4 Models and Experiments in Musicology 4.1 Interior and Exterior Nature . . . . . . . . . . 4.2 What Is a Musicological Experiment? . . . . 4.3 Questions—Experiments of the Mind . . . . . 4.4 New Scientiﬁc Paradigms and Collaboratories xi 29 32 33 34 35

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

xii

CONTENTS

II

Navigation on Concept Spaces

37

39 40 44 45 47 48 50 55 58 61 61 63 66 67 67 69 70 76 79 85 89 93 99

5 Navigation 5.1 Music in the EncycloSpace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Receptive Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Productive Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Denotators 6.1 Universal Concept Formats . . . . . . . . . . . . . . 6.1.1 First Naive Approach To Denotators . . . . . 6.1.2 Interpretations and Comments . . . . . . . . 6.1.3 Ordering Denotators and ‘Concept Leaﬁng’ . 6.2 Forms . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Variable Addresses . . . . . . . . . . . . . . . 6.2.2 Formal Deﬁnition . . . . . . . . . . . . . . . . 6.2.3 Discussion of the Form Typology . . . . . . . 6.3 Denotators . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Formal Deﬁnition of a Denotator . . . . . . . 6.4 Anchoring Forms in Modules . . . . . . . . . . . . . 6.4.1 First Examples and Comments on Modules in 6.5 Regular and Circular Forms . . . . . . . . . . . . . . 6.6 Regular Denotators . . . . . . . . . . . . . . . . . . . 6.7 Circular Denotators . . . . . . . . . . . . . . . . . . 6.8 Ordering on Forms and Denotators . . . . . . . . . . 6.8.1 Concretizations and Applications . . . . . . . 6.9 Concept Surgery and Denotator Semantics . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Music . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

III

Local Theory

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

103

. . . . . . . . . 105 106 108 109 114 118 121 122 127 129 135 137 139 154 158

7 Local Compositions 7.1 The Objects of Local Theory . . . . . . 7.2 First Local Music Objects . . . . . . . . 7.2.1 Chords and Scales . . . . . . . . 7.2.2 Local Meters and Local Rhythms 7.2.3 Motives . . . . . . . . . . . . . . 7.3 Functorial Local Compositions . . . . . 7.4 First Elements of Local Theory . . . . . 7.5 Alterations Are Tangents . . . . . . . . 7.5.1 The Theorem of Mason–Mazzola 8 Symmetries and Morphisms 8.1 Symmetries in Music . . . . . . . . 8.1.1 Elementary Examples . . . 8.2 Morphisms of Local Compositions 8.3 Categories of Local Compositions .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

CONTENTS 8.3.1 8.3.2 8.3.3 8.3.4 8.3.5 Commenting the Concatenation Principle . . . Embedding and Addressed Adjointness . . . . Universal Constructions on Local Compositions The Address Question . . . . . . . . . . . . . . Categories of Commutative Local Compositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xiii 161 163 166 169 171 175 178 181 184 185 185 188

9 Yoneda Perspectives 9.1 Morphisms Are Points . . . . . . . . . . . . . . . 9.2 Yoneda’s Fundamental Lemma . . . . . . . . . . 9.3 The Yoneda Philosophy . . . . . . . . . . . . . . 9.4 Understanding Fine and Other Arts . . . . . . . 9.4.1 Painting and Music . . . . . . . . . . . . . 9.4.2 The Art of Object-Oriented Programming 10 Paradigmatic Classiﬁcation 10.1 Paradigmata in Musicology, Linguistics, and 10.2 Transformation . . . . . . . . . . . . . . . . 10.3 Similarity . . . . . . . . . . . . . . . . . . . 10.4 Fuzzy Concepts in the Humanities . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

Mathematics . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

191 . 192 . 196 . 198 . 200 203 203 204 205 205 207 216 217 219 221 225 228 231 232 238 241 243 247 258 259 262 268 271 272 272

11 Orbits 11.1 Gestalt and Symmetry Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 The Framework for Local Classiﬁcation . . . . . . . . . . . . . . . . . . . . . . 11.3 Orbits of Elementary Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.1 Classiﬁcation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.2 The Local Classiﬁcation Theorem . . . . . . . . . . . . . . . . . . . . . 11.3.3 The Finite Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.4 Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.5 Chords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.6 Empirical Harmonic Vocabularies . . . . . . . . . . . . . . . . . . . . . . 11.3.7 Self-addressed Chords . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.8 Motives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Enumeration Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4.1 P´lya and de Bruijn Theory . . . . . . . . . . . . . . . . . . . . . . . . . o 11.4.2 Big Science for Big Numbers . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Group-theoretical Methods in Composition and Theory . . . . . . . . . . . . . 11.5.1 Aspects of Serialism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.2 The American Tradition . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.6 Esthetic Implications of Classiﬁcation . . . . . . . . . . . . . . . . . . . . . . . 11.6.1 Jakobson’s Poetic Function . . . . . . . . . . . . . . . . . . . . . . . . . 11.6.2 Motivic Analysis: Schubert/Stolberg “Lied auf dem Wasser zu singen...” 11.6.3 Composition: Mazzola/Baudelaire “La mort des artistes” . . . . . . . . 11.7 Mathematical Reﬂections on Historicity in Music . . . . . . . . . . . . . . . . . 11.7.1 Jean-Jacques Nattiez’ Paradigmatic Theme . . . . . . . . . . . . . . . . 11.7.2 Groups as a Parameter of Historicity . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

xiv 12 Topological Specialization 12.1 What Ehrenfels Neglected . . . . . . . . . . . . . . . . . 12.2 Topology . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.1 Metrical Comparison . . . . . . . . . . . . . . . . 12.2.2 Specialization Morphisms of Local Compositions 12.3 The Problem of Sound Classiﬁcation . . . . . . . . . . . 12.3.1 Topographic Determinants of Sound Descriptions 12.3.2 Varieties of Sounds . . . . . . . . . . . . . . . . . 12.3.3 Semiotics of Sound Classiﬁcation . . . . . . . . . 12.4 Making the Vague Precise . . . . . . . . . . . . . . . . .

CONTENTS 275 276 277 279 281 284 284 291 294 295

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

IV

Global Theory

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Triadic and . . . . . . . . . . . . . . . . . . . . .

297

. . . . . . . 299 300 307 308 310 314 316 317

13 Global Compositions 13.1 The Local-Global Dichotomy in Music . . . . . . . . . . . . . . . . 13.1.1 Musical and Mathematical Manifolds . . . . . . . . . . . . . 13.2 What Are Global Compositions? . . . . . . . . . . . . . . . . . . . 13.2.1 The Nerve of an Objective Global Composition . . . . . . . 13.3 Functorial Global Compositions . . . . . . . . . . . . . . . . . . . . 13.4 Interpretations and the Vocabulary of Global Concepts . . . . . . . 13.4.1 Iterated Interpretations . . . . . . . . . . . . . . . . . . . . 13.4.2 The Pitch Domain: Chains of Thirds, Ecclesiastical Modes, Quaternary Degrees . . . . . . . . . . . . . . . . . . . . . . 13.4.3 Interpreting Time: Global Meters and Rhythms . . . . . . . 13.4.4 Motivic Interpretations: Melodies and Themes . . . . . . . 14 Global Perspectives 14.1 Musical Motivation . . . . . . . . . . . . . . . . . 14.2 Global Morphisms . . . . . . . . . . . . . . . . . 14.3 Local Domains . . . . . . . . . . . . . . . . . . . 14.4 Nerves . . . . . . . . . . . . . . . . . . . . . . . . 14.5 Simplicial Weights . . . . . . . . . . . . . . . . . 14.6 Categories of Commutative Global Compositions 15 Global Classiﬁcation 15.1 Module Complexes . . . . . . . . . . . . . . . . . 15.1.1 Global Aﬃne Functions . . . . . . . . . . 15.1.2 Bilinear and Exterior Forms . . . . . . . . 15.1.3 Deviation: Compositions vs. “Molecules” 15.2 The Resolution of a Global Composition . . . . . 15.2.1 Global Standard Compositions . . . . . . 15.2.2 Compositions from Module Complexes . . 15.3 Orbits of Module Complexes Are Classifying . . 15.3.1 Combinatorial Group Actions . . . . . . .

. 318 . 326 . 331 333 333 334 341 343 345 347 349 350 350 353 355 356 356 358 363 364

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

CONTENTS

xv

15.3.2 Classifying Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366 16 Classifying Interpretations 16.1 Characterization of Interpretable Compositions . . . . . . . . 16.1.1 Automorphism Groups of Interpretable Compositions 16.1.2 A Cohomological Criterion . . . . . . . . . . . . . . . 16.2 Global Enumeration Theory . . . . . . . . . . . . . . . . . . . 16.2.1 Tesselation . . . . . . . . . . . . . . . . . . . . . . . . 16.2.2 Mosaics . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2.3 Classifying Rational Rhythms and Canons . . . . . . . 16.3 Global American Set Theory . . . . . . . . . . . . . . . . . . 16.4 Interpretable “Molecules” . . . . . . . . . . . . . . . . . . . . 369 370 372 374 376 376 378 380 382 385

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

17 Esthetics and Classiﬁcation 387 17.1 Understanding by Resolution: An Illustrative Example . . . . . . . . . . . . . . . 387 17.2 Var`se’s Program and Yoneda’s Lemma . . . . . . . . . . . . . . . . . . . . . . . 392 e 18 Predicates 18.1 What Is the Case: The Existence Problem . . . . . . . 18.1.1 Merging Systematic and Historical Musicology 18.2 Textual and Paratextual Semiosis . . . . . . . . . . . . 18.2.1 Textual and Paratextual Signiﬁcation . . . . . 18.3 Textuality . . . . . . . . . . . . . . . . . . . . . . . . . 18.3.1 The Category of Denotators . . . . . . . . . . . 18.3.2 Textual Semiosis . . . . . . . . . . . . . . . . . 18.3.3 Atomic Predicates . . . . . . . . . . . . . . . . 18.3.4 Logical and Geometric Motivation . . . . . . . 18.4 Paratextuality . . . . . . . . . . . . . . . . . . . . . . . 19 Topoi of Music 19.1 The Grothendieck Topology . . . . 19.1.1 Cohomology . . . . . . . . . 19.1.2 Marginalia on Presheaves . 19.2 The Topos of Music: An Overview 397 397 398 400 401 402 402 406 412 419 424 427 427 430 434 435 439 439 442 442 443 444 445 446 446 448 448

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

20 Visualization Principles 20.1 Problems . . . . . . . . . . . . . . . 20.2 Folding Dimensions . . . . . . . . . . 20.2.1 R2 → R . . . . . . . . . . . . 20.2.2 Rn → R . . . . . . . . . . . . 20.2.3 An Explicit Construction of µ 20.3 Folding Denotators . . . . . . . . . . 20.3.1 Folding Limits . . . . . . . . 20.3.2 Folding Colimits . . . . . . . 20.3.3 Folding Powersets . . . . . . 20.3.4 Folding Circular Denotators .

. . . . . . . . . . . . with . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Special Values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

xvi

CONTENTS 20.4 Compound Parametrized Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . 449 20.5 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451

V

**Topologies for Rhythm and Motives
**

. . . . . . . . . . . . . . . . . . . . Time Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

453

. . . . . 455 455 456 457 459 461 465 466 468 469 472 472 473 475 477 479 479 481 482 484 487 490 491 493 496

21 Metrics and Rhythmics 21.1 Review of Riemann and Jackendoﬀ–Lerdahl Theories 21.1.1 Riemann’s Weights . . . . . . . . . . . . . . . 21.1.2 Jackendoﬀ–Lerdahl: Intrinsic Versus Extrinsic 21.2 Topologies of Global Meters and Associated Weights 21.3 Macro-Events in the Time Domain . . . . . . . . . . 22 Motif Gestalts 22.1 Motivic Interpretation . . . . . . . . . . . 22.2 Shape Types . . . . . . . . . . . . . . . . 22.2.1 Examples of Shape Types . . . . . 22.3 Metrical Similarity . . . . . . . . . . . . . 22.3.1 Examples of Distance Functions . 22.4 Paradigmatic Groups . . . . . . . . . . . . 22.4.1 Examples of Paradigmatic Groups 22.5 Pseudo-metrics on Orbits . . . . . . . . . 22.6 Topologies on Gestalts . . . . . . . . . . . 22.6.1 The Inheritance Property . . . . . 22.6.2 Cognitive Aspects of Inheritance . 22.6.3 Epsilon Topologies . . . . . . . . . 22.7 First Properties of the Epsilon Topologies 22.7.1 Toroidal Topologies . . . . . . . . 22.8 Rudolph Reti’s Motivic Analysis Revisited 22.8.1 Review of Concepts . . . . . . . . 22.8.2 Reconstruction . . . . . . . . . . . 22.9 Motivic Weights . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

VI

Harmony

499

23 Critical Preliminaries 501 23.1 Hugo Riemann . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502 23.2 Paul Hindemith . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 23.3 Heinrich Schenker and Friedrich Salzer . . . . . . . . . . . . . . . . . . . . . . . . 503 24 Harmonic Topology 24.1 Chord Perspectives . . . . . . . . 24.1.1 Euler Perspectives . . . . 24.1.2 12-tempered Perspectives 24.1.3 Enharmonic Projection . 505 506 506 512 514

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

CONTENTS 24.2 Chord 24.2.1 24.2.2 24.2.3 24.2.4 Topologies . . . . . . . . Extension and Intension Extension and Intension Faithful Addresses . . . The Saturation Sheaf . . . . . . . . . . . . . . . Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xvii 518 518 520 523 526 529 530 532 532 534 537 538 540 543 545 546 551 552 553 553 555 556 558 560 560 561 563 564 565 565 568 571 574 576 581 586 586 587 590 591

25 Harmonic Semantics 25.1 Harmonic Signs—Overview . . . . . . . . . . . . . . . . 25.2 Degree Theory . . . . . . . . . . . . . . . . . . . . . . . 25.2.1 Chains of Thirds . . . . . . . . . . . . . . . . . . 25.2.2 American Jazz Theory . . . . . . . . . . . . . . . 25.2.3 Hans Straub: General Degrees in General Scales 25.3 Function Theory . . . . . . . . . . . . . . . . . . . . . . 25.3.1 Canonical Morphemes for European Harmony . . 25.3.2 Riemann Matrices . . . . . . . . . . . . . . . . . 25.3.3 Chains of Thirds . . . . . . . . . . . . . . . . . . 25.3.4 Tonal Functions from Absorbing Addresses . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

26 Cadence 26.1 Making the Concept Precise . . . . . . . . . . . . . . . . . . . 26.2 Classical Cadences Relating to 12-tempered Intonation . . . . 26.2.1 Cadences in Triadic Interpretations of Diatonic Scales 26.2.2 Cadences in More General Interpretations . . . . . . . 26.3 Cadences in Self-addressed Tonalities of Morphology . . . . . 26.4 Self-addressed Cadences by Symmetries and Morphisms . . . 26.5 Cadences for Just Intonation . . . . . . . . . . . . . . . . . . 26.5.1 Tonalities in Third-Fifth Intonation . . . . . . . . . . 26.5.2 Tonalities in Pythagorean Intonation . . . . . . . . . . 27 Modulation 27.1 Modeling Modulation by Particle Interaction . . . . . . . . 27.1.1 Models and the Anthropic Principle . . . . . . . . . 27.1.2 Classical Motivation and Heuristics . . . . . . . . . . 27.1.3 The General Background . . . . . . . . . . . . . . . 27.1.4 The Well-Tempered Case . . . . . . . . . . . . . . . 27.1.5 Reconstructing the Diatonic Scale from Modulation 27.1.6 The Case of Just Tuning . . . . . . . . . . . . . . . . 27.1.7 Quantized Modulations and Modulation Domains for 27.2 Harmonic Tension . . . . . . . . . . . . . . . . . . . . . . . 27.2.1 The Riemann Algebra . . . . . . . . . . . . . . . . . 27.2.2 Weights on the Riemann Algebra . . . . . . . . . . . 27.2.3 Harmonic Tensions from Classical Harmony? . . . . 27.2.4 Optimizing Harmonic Paths . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Selected . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . Scales . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

xviii 28 Applications 28.1 First Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.1.1 Johann Sebastian Bach: Choral from “Himmelfahrtsoratorium” 28.1.2 Wolfgang Amadeus Mozart: “Zauberﬂ¨te”, Choir of Priests . . o 28.1.3 Claude Debussy: “Pr´ludes”, Livre 1, No.4 . . . . . . . . . . . e 28.2 Modulation in Beethoven’s Sonata op.106, 1st Movement . . . . . . . . 28.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.2.2 The Fundamental Theses of Erwin Ratz and Jrgen Uhde . . . . 28.2.3 Overview of the Modulation Structure . . . . . . . . . . . . . . 28.2.4 Modulation B G via e−3 in W . . . . . . . . . . . . . . . . 28.2.5 Modulation G E via Ug in W . . . . . . . . . . . . . . . . . 28.2.6 Modulation E D/b from W to W ∗ . . . . . . . . . . . . . . 28.2.7 Modulation D/b B via Ud/d = Ug /a within W ∗ . . . . . . 28.2.8 Modulation B B from W ∗ to W . . . . . . . . . . . . . . . 28.2.9 Modulation B G via Ub within W . . . . . . . . . . . . . 28.2.10 Modulation G G via Ua /a within W . . . . . . . . . . . . . 28.2.11 Modulation G B via e3 within W . . . . . . . . . . . . . . . 28.3 Rhythmical Modulation in “Synthesis” . . . . . . . . . . . . . . . . . . 28.3.1 Rhythmic Modes . . . . . . . . . . . . . . . . . . . . . . . . . . 28.3.2 Composition for Percussion Ensemble . . . . . . . . . . . . . .

CONTENTS 593 594 595 598 600 603 603 605 607 608 608 608 609 609 610 610 610 610 611 613

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

VII

Counterpoint

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

615

. . . . . . . 617 617 619 620 620 622 625 626 629 630 634 635 637 637 641

29 Melodic Variation by Arrows 29.1 Arrows and Alterations . . . . . . . . . . . 29.2 The Contrapuntal Interval Concept . . . . . 29.3 The Algebra of Intervals . . . . . . . . . . . 29.3.1 The Third Torus . . . . . . . . . . . 29.4 Musical Interpretation of the Interval Ring 29.5 Self-addressed Arrows . . . . . . . . . . . . 29.6 Change of Orientation . . . . . . . . . . . .

30 Interval Dichotomies as a Contrast 30.1 Dichotomies and Polarity . . . . . . . . . . . . . . . . 30.2 The Consonance and Dissonance Dichotomy . . . . . . 30.2.1 Fux and Riemann Consonances Are Isomorphic 30.2.2 Induced Polarities . . . . . . . . . . . . . . . . 30.2.3 Empirical Evidence for the Polarity Function . 30.2.4 Music and the Hippocampal Gate Function . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

31 Modeling Counterpoint by Local Symmetries 31.1 Deformations of the Strong Dichotomies . . . . . . . . . . . . . . . . . . . . . . 31.2 Contrapuntal Symmetries Are Local . . . . . . . . . . . . . . . . . . . . . . . . 31.3 The Counterpoint Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

645 . 645 . 647 . 649

CONTENTS Some Preliminary Calculations . . . . . . . . . . . . . . . . . . . . . . . Two Lemmata on Cardinalities of Intersections . . . . . . . . . . . . . . An Algorithm for Exhibiting the Contrapuntal Symmetries . . . . . . . Transfer of the Counterpoint Rules to General Representatives of Strong Dichotomies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4 The Classical Case: Consonances and Dissonances . . . . . . . . . . . . . . . . 31.4.1 Discussion of the Counterpoint Theorem in the Light of Reduced Strict Style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4.2 The Major Dichotomy—A Cultural Antipode? . . . . . . . . . . . . . . 31.3.1 31.3.2 31.3.3 31.3.4

xix . 649 . 651 . 651 . 655 . 655 . 656 . 657

VIII

**Structure Theory of Performance
**

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

661

. . . . . . . . . 663 665 666 667 667 668 672 674 675 679 681 681 681 683 685 686 687 689 690 691 693 695 696 697 699 701 701 703 704 706

32 Local and Global Performance Transformations 32.1 Performance as a Reality Switch . . . . . . . . . . . . . . . . 32.2 Why Do We Need Inﬁnite Performance of the Same Piece? . 32.3 Local Structure . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3.1 The Coherence of Local Performance Transformations 32.3.2 Diﬀerential Morphisms of Local Compositions . . . . . 32.4 Global Structure . . . . . . . . . . . . . . . . . . . . . . . . . 32.4.1 Modeling Performance Syntax . . . . . . . . . . . . . . 32.4.2 The Formal Setup . . . . . . . . . . . . . . . . . . . . 32.4.3 Performance qua Interpretation of Interpretation . . . 33 Performance Fields 33.1 Classics: Tempo, Intonation, and Dynamics . . . . . . 33.1.1 Tempo . . . . . . . . . . . . . . . . . . . . . . . 33.1.2 Intonation . . . . . . . . . . . . . . . . . . . . . 33.1.3 Dynamics . . . . . . . . . . . . . . . . . . . . . 33.2 Genesis of the General Formalism . . . . . . . . . . . . 33.2.1 The Question of Articulation . . . . . . . . . . 33.2.2 The Formalism of Performance Fields . . . . . 33.3 What Performance Fields Signify . . . . . . . . . . . . 33.3.1 Th.W. Adorno, W. Benjamin, and D. Raﬀman 33.3.2 Towards Composition of Performance . . . . . 34 Initial Sets and Initial Performances 34.1 Taking oﬀ with a Shifter . . . . . . . 34.2 Anchoring Onset . . . . . . . . . . . 34.3 The Concert Pitch . . . . . . . . . . 34.4 Dynamical Anchors . . . . . . . . . . 34.5 Initializing Articulation . . . . . . . 34.6 Hit Point Theory . . . . . . . . . . . 34.6.1 Distances . . . . . . . . . . . 34.6.2 Flow Interpolation . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

xx 35 Hierarchies and Performance Scores 35.1 Performance Cells . . . . . . . . . . . 35.2 The Category of Performance Cells . . 35.3 Hierarchies . . . . . . . . . . . . . . . 35.3.1 Operations on Hierarchies . . . 35.3.2 Classiﬁcation Issues . . . . . . 35.3.3 Example: The Piano and Violin 35.4 Local Performance Scores . . . . . . . 35.5 Global Performance Scores . . . . . . 35.5.1 Instrumental Fibers . . . . . .

CONTENTS 711 711 713 714 718 718 722 723 728 728

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hierarchies . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

IX

Expressive Semantics

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

731

. . . . 733 734 737 741 745 747 748 749 751 752 753 755 756 757 758 762 763 765 769 771 773 774 775 776 777 777 780 782 783

36 Taxonomy of Expressive Performance 36.1 Feelings: Emotional Semantics . . . . . 36.2 Motion: Gestural Semantics . . . . . . 36.3 Understanding: Rational Semantics . . 36.4 Cross-semantical Relations . . . . . . . 37 Performance Grammars 37.1 Rule-based Grammars . . . . . . 37.1.1 The KTH School . . . . . 37.1.2 Neil P. McAgnus Todd . . 37.1.3 The Zurich School . . . . 37.2 Remarks on Learning Grammars

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

38 Stemma Theory 38.1 Motivation from Practising and Rehearsing . . . . . . . . . . . . . 38.1.1 Does Reproducibility of Performances Help Understanding? 38.2 Tempo Curves Are Inadequate . . . . . . . . . . . . . . . . . . . . 38.3 The Stemma Concept . . . . . . . . . . . . . . . . . . . . . . . . . 38.3.1 The General Setup of Matrilineal Sexual Propagation . . . 38.3.2 The Primary Mother—Taking Oﬀ . . . . . . . . . . . . . . 38.3.3 Mono- and Polygamy—Local and Global Actions . . . . . . 38.3.4 Family Life—Cross-Correlations . . . . . . . . . . . . . . . 39 Operator Theory 39.1 Why Weights? . . . . . . . . . 39.1.1 Discrete and Continuous 39.1.2 Weight Recombination . 39.2 Primavista Weights . . . . . . . 39.2.1 Dynamics . . . . . . . . 39.2.2 Agogics . . . . . . . . . 39.2.3 Tuning and Intonation . 39.2.4 Articulation . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . Weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

CONTENTS 39.2.5 Ornaments . . . . . . . . . . . . 39.3 Analytical Weights . . . . . . . . . . . . 39.4 Taxonomy of Operators . . . . . . . . . 39.4.1 Splitting Operators . . . . . . . . 39.4.2 Symbolic Operators . . . . . . . 39.4.3 Physical Operators . . . . . . . . 39.4.4 Field Operators . . . . . . . . . . 39.5 Tempo Operator . . . . . . . . . . . . . 39.6 Scalar Operator . . . . . . . . . . . . . . 39.7 The Theory of Basis-Pianola Operators 39.7.1 Basis Specialization . . . . . . . 39.7.2 Pianola Specialization . . . . . . 39.8 Locally Linear Grammars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxi 783 785 787 788 789 791 792 793 794 795 797 801 801

X

RUBATO

805

40 Architecture 807 40.1 The Overall Modularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808 40.2 Frame and Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809 41 The 41.1 41.2 41.3 41.4 41.5 RUBETTE Family MetroRUBETTE . . . MeloRUBETTE . . . . HarmoRUBETTE . . . PerformanceRUBETTE PrimavistaRUBETTE 813 814 816 819 824 831 833 833 834 835 835 839 841 841 842 849 850

. . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

42 Performance Experiments 42.1 A Preliminary Experiment: Robert Schumann’s 42.2 Full Experiment: J.S. Bach’s “Kunst der Fuge” 42.3 Analysis . . . . . . . . . . . . . . . . . . . . . . 42.3.1 Metric Analysis . . . . . . . . . . . . . . 42.3.2 Motif Analysis . . . . . . . . . . . . . . 42.3.3 Omission of Harmonic Analysis . . . . . 42.4 Stemma Constructions . . . . . . . . . . . . . . 42.4.1 Performance Setup . . . . . . . . . . . . 42.4.2 Instrumental Setup . . . . . . . . . . . . 42.4.3 Global Discussion . . . . . . . . . . . .

“Kuriose Geschichte” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

XI

Statistics of Analysis and Performance

853

43 Analysis of Analysis 855 43.1 Hierarchical Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855 43.1.1 General Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855

xxii

CONTENTS

43.1.2 Hierarchical Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 857 43.1.3 Hierarchical Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . 858 43.2 Comparing Analyses of Bach, Schumann, and Webern . . . . . . . . . . . . . . . 860 44 Diﬀerential Operators and Regression 44.0.1 Analytical Data . . . . . . . . . . . . . . 44.1 The Beran Operator . . . . . . . . . . . . . . . 44.1.1 The Concept . . . . . . . . . . . . . . . 44.1.2 The Formalism . . . . . . . . . . . . . . 44.2 The Method of Regression Analysis . . . . . . . 44.2.1 The Full Model . . . . . . . . . . . . . . 44.2.2 Step Forward Selection . . . . . . . . . . 44.3 The Results of Regression Analysis . . . . . . . 44.3.1 Relations between Tempo and Analysis 44.3.2 Complex Relationships . . . . . . . . . . 44.3.3 Commonalities and Diversities . . . . . 44.3.4 Overview of Statistical Results . . . . . 871 873 874 874 877 880 880 881 881 882 883 884 897

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

XII

Inverse Performance Theory

903

905 . 905 . 906 . 909 911 911 912 916 916 917 918 919 921 922 925 927

45 Principles of Music Critique 45.1 Boiling down Inﬁnity—Is Feuilletonism Inevitable? . . . . . . . . . . . . . . . . 45.2 “Political Correctness” in Performance—Reviewing Gould . . . . . . . . . . . . 45.3 Transversal Ethnomusicology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Critical Fibers 46.1 The Stemma Model of Critique . . . . . . . . . . . . . . . . . . . . . 46.2 Fibers for Locally Linear Grammars . . . . . . . . . . . . . . . . . . 46.3 Algorithmic Extraction of Performance Fields . . . . . . . . . . . . . 46.3.1 The Inﬁnitesimal View on Expression . . . . . . . . . . . . . 46.3.2 Real-time Processing of Expressive Performance . . . . . . . 46.3.3 Score–Performance Matching . . . . . . . . . . . . . . . . . . 46.3.4 Performance Field Calculation . . . . . . . . . . . . . . . . . 46.3.5 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.3.6 The EspressoRUBETTE : An Interactive Tool for Expression 46.4 Local Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.4.1 Comparing Argerich and Horowitz . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Extraction . . . . . . . . . . . . . .

XIII

Operationalization of Poiesis

931

47 Unfolding Geometry and Logic in Time 933 47.1 Performance of Logic and Geometry . . . . . . . . . . . . . . . . . . . . . . . . . 934 47.2 Constructing Time from Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . 935 47.3 Discourse and Insight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 937

CONTENTS 48 Local and Global Strategies in Composition 48.1 Local Paradigmatic Instances . . . . . . . . . . 48.1.1 Transformations . . . . . . . . . . . . . 48.1.2 Variations . . . . . . . . . . . . . . . . . 48.2 Global Poetical Syntax . . . . . . . . . . . . . . 48.2.1 Roman Jakobson’s Horizontal Function 48.2.2 Roland Posner’s Vertical Function . . . 48.3 Structure and Process . . . . . . . . . . . . . . 49 The 49.1 49.2 49.3 49.4 Paradigmatic Discourse on presto The presto Functional Scheme . . . . Modular Aﬃne Transformations . . . Ornaments and Variations . . . . . . . Problems of Abstraction . . . . . . . .

xxiii 939 940 940 941 941 942 942 943 945 945 948 949 952 955 956 956 956 958 959 963 964 967 968 969 969 970 970 971 971 971 972 973 973 973 975 978 982 984 986

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

50 Case Study I:“Synthesis” by Guerino Mazzola 50.1 The Overall Organization . . . . . . . . . . . . . . . . . . . . . 50.1.1 The Material: 26 Classes of Three-Element Motives . . . 50.1.2 Principles of the Four Movements and Instrumentation . 50.2 1st Movement: Sonata Form . . . . . . . . . . . . . . . . . . . . 50.3 2nd Movement: Variations . . . . . . . . . . . . . . . . . . . . . 50.4 3rd Movement: Scherzo . . . . . . . . . . . . . . . . . . . . . . . 50.5 4th Movement: Fractal Syntax . . . . . . . . . . . . . . . . . . . 51 Object-Oriented Programming in OpenMusic 51.1 Object-Oriented Language . . . . . . . . . . . . 51.1.1 Patches . . . . . . . . . . . . . . . . . . 51.1.2 Objects . . . . . . . . . . . . . . . . . . 51.1.3 Classes . . . . . . . . . . . . . . . . . . 51.1.4 Methods . . . . . . . . . . . . . . . . . . 51.1.5 Generic Functions . . . . . . . . . . . . 51.1.6 Message Passing . . . . . . . . . . . . . 51.1.7 Inheritance . . . . . . . . . . . . . . . . 51.1.8 Boxes and Evaluation . . . . . . . . . . 51.1.9 Instantiation . . . . . . . . . . . . . . . 51.2 Musical Object Framework . . . . . . . . . . . 51.2.1 Internal Representation . . . . . . . . . 51.2.2 Interface . . . . . . . . . . . . . . . . . . 51.3 Maquettes: Objects in Time . . . . . . . . . . . 51.4 Meta-object Protocol . . . . . . . . . . . . . . . 51.4.1 Reiﬁcation of Temporal Boxes . . . . . . 51.5 A Musical Example . . . . . . . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

xxiv

CONTENTS

XIV

**String Quartet Theory
**

. . . . . . . Finscher . . . . . . . . . Humanists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

991

. . . . . 993 994 994 995 996 997

52 Historical and Theoretical Prerequisites 52.1 History . . . . . . . . . . . . . . . . . . . . . . 52.2 Theory of the String Quartet Following Ludwig 52.2.1 Four Part Texture . . . . . . . . . . . . 52.2.2 The Topos of Conversation Among Four 52.2.3 The Family of Violins . . . . . . . . . .

53 Estimation of Resolution Parameters 999 53.1 Parameter Spaces for Violins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1000 53.2 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1003 54 The 54.1 54.2 54.3 Case of Counterpoint Counterpoint . . . . . . Harmony . . . . . . . . Eﬀective Selection . . . and . . . . . . . . . Harmony . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1007 . 1007 . 1008 . 1009

XV

Appendix: Sound

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1011

. . . . . . . 1013 . 1013 . 1014 . 1018 . 1028 . 1028 . 1029 . 1031 1035 . 1036 . 1036 . 1037 . 1037 . 1041 . 1042 . 1044 . 1046 . 1049 . 1049 . 1051 . 1052 . 1052 . 1053

A Common Parameter Spaces A.1 Physical Spaces . . . . . . . . . . . . A.1.1 Neutral Data . . . . . . . . . A.1.2 Sound Analysis and Synthesis A.2 Mathematical and Symbolic Spaces . A.2.1 Onset and Duration . . . . . A.2.2 Amplitude and Crescendo . . A.2.3 Frequency and Glissando . .

B Auditory Physiology and Psychology B.1 Physiology: From the Auricle to Heschl’s Gyri . . . . . . . . . B.1.1 Outer Ear . . . . . . . . . . . . . . . . . . . . . . . . . B.1.2 Middle Ear . . . . . . . . . . . . . . . . . . . . . . . . B.1.3 Inner Ear (Cochlea) . . . . . . . . . . . . . . . . . . . B.1.4 Cochlear Hydrodynamics: The Travelling Wave . . . . B.1.5 Active Ampliﬁcation of the Traveling Wave Motion . . B.1.6 Neural Processing . . . . . . . . . . . . . . . . . . . . B.2 Discriminating Tones: Werner Meyer-Eppler’s Valence Theory B.3 Aspects of Consonance and Dissonance . . . . . . . . . . . . . B.3.1 Euler’s Gradus Function . . . . . . . . . . . . . . . . . B.3.2 von Helmholtz’ Beat Model . . . . . . . . . . . . . . . B.3.3 Psychometric Investigations by Plomp and Levelt . . . B.3.4 Counterpoint . . . . . . . . . . . . . . . . . . . . . . . B.3.5 Consonance and Dissonance: A Conceptual Field . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

CONTENTS

xxv

XVI

**Appendix: Mathematical Basics
**

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1055

. . . . . . . . . . . . . 1057 . 1057 . 1058 . 1058 . 1062 . 1062 . 1063 . 1066 . 1066 . 1068 . 1069 . 1069 . 1070 . 1071 1075 . 1075 . 1077 . 1080 . 1080 . 1080 . 1081 . 1081 1083 . 1083 . 1084 . 1085 . 1085 . 1087 . 1087 . 1088 . 1090 . 1090 . 1091 . 1091 . 1093 . 1096 . 1097 . 1098 . 1099 . 1099 . 1101

C Sets, Relations, Monoids, Groups C.1 Sets . . . . . . . . . . . . . . . . . . . . . . . . . C.1.1 Examples of Sets . . . . . . . . . . . . . . C.2 Relations . . . . . . . . . . . . . . . . . . . . . . C.2.1 Universal Constructions . . . . . . . . . . C.2.2 Graphs and Quivers . . . . . . . . . . . . C.2.3 Monoids . . . . . . . . . . . . . . . . . . . C.3 Groups . . . . . . . . . . . . . . . . . . . . . . . . C.3.1 Homomorphisms of Groups . . . . . . . . C.3.2 Direct, Semi-direct, and Wreath Products C.3.3 Sylow Theorems on p-groups . . . . . . . C.3.4 Classiﬁcation of Groups . . . . . . . . . . C.3.5 General Aﬃne Groups . . . . . . . . . . . C.3.6 Permutation Groups . . . . . . . . . . . . D Rings and Algebras D.1 Basic Deﬁnitions and Constructions . . . . . D.1.1 Universal Constructions . . . . . . . . D.2 Prime Factorization . . . . . . . . . . . . . . D.3 Euclidean Algorithm . . . . . . . . . . . . . . D.4 Approximation of Real Numbers by Fractions D.5 Some Special Issues . . . . . . . . . . . . . . . D.5.1 Integers, Rationals, and Real Numbers

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

E Modules, Linear, and Aﬃne Transformations E.1 Modules and Linear Transformations . . . . . . . . . . . . . E.1.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . E.2 Module Classiﬁcation . . . . . . . . . . . . . . . . . . . . . . E.2.1 Dimension . . . . . . . . . . . . . . . . . . . . . . . . E.2.2 Endomorphisms on Dual Numbers . . . . . . . . . . E.2.3 Semi-Simple Modules . . . . . . . . . . . . . . . . . E.2.4 Jacobson Radical and Socle . . . . . . . . . . . . . . E.2.5 Theorem of Krull–Remak–Schmidt . . . . . . . . . . E.3 Categories of Modules and Aﬃne Transformations . . . . . E.3.1 Direct Sums . . . . . . . . . . . . . . . . . . . . . . . E.3.2 Aﬃne Forms and Tensors . . . . . . . . . . . . . . . E.3.3 Biaﬃne Maps . . . . . . . . . . . . . . . . . . . . . . E.3.4 Symmetries of the Aﬃne Plane . . . . . . . . . . . . E.3.5 Symmetries on Z2 . . . . . . . . . . . . . . . . . . . E.3.6 Symmetries on Zn . . . . . . . . . . . . . . . . . . . E.3.7 Complements on the Module of a Local Composition E.3.8 Fiber Products and Fiber Sums in Mod . . . . . . . E.4 Complements of Commutative Algebra . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

xxvi E.4.1 E.4.2 E.4.3 E.4.4 Localization . . . . Projective Modules Injective Modules . Lie Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CONTENTS . . . . . . . . . . . . . . . . . 1101 . 1102 . 1103 . 1104 1107 . 1107 . 1108 . 1110 . 1111 . 1112 . 1112 . 1113 . 1114 1115 . 1115 . 1116 . 1117 . 1118 . 1120 . 1120 . 1122 . 1125 . 1126 . 1127 . 1127 . 1129 . 1130 . 1131 . 1131 . 1135 . 1137 1145 . 1145 . 1145 . 1146 . 1147 . 1147 . 1148 . 1148 . 1148 . 1150 . 1150

F Algebraic Geometry F.1 Locally Ringed Spaces . . . . . . . . . . . . . . . F.2 Spectra of Commutative Rings . . . . . . . . . . F.2.1 Sober Spaces . . . . . . . . . . . . . . . . F.3 Schemes and Functors . . . . . . . . . . . . . . . F.4 Algebraic and Geometric Structures on Schemes F.4.1 The Zariski Tangent Space . . . . . . . . F.5 Grassmannians . . . . . . . . . . . . . . . . . . . F.6 Quotients . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

G Categories, Topoi, and Logic G.1 Categories Instead of Sets . . . . . . . . . . . . . . . . . . . . . G.1.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . G.1.2 Functors . . . . . . . . . . . . . . . . . . . . . . . . . . . G.1.3 Natural Transformations . . . . . . . . . . . . . . . . . . G.2 The Yoneda Lemma . . . . . . . . . . . . . . . . . . . . . . . . G.2.1 Universal Constructions: Adjoints, Limits, and Colimits G.2.2 Limit and Colimit Characterizations . . . . . . . . . . . G.3 Topoi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G.3.1 Subobject Classiﬁers . . . . . . . . . . . . . . . . . . . . G.3.2 Exponentiation . . . . . . . . . . . . . . . . . . . . . . . G.3.3 Deﬁnition of Topoi . . . . . . . . . . . . . . . . . . . . . G.4 Grothendieck Topologies . . . . . . . . . . . . . . . . . . . . . . G.4.1 Sheaves . . . . . . . . . . . . . . . . . . . . . . . . . . . G.5 Formal Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . G.5.1 Propositional Calculus . . . . . . . . . . . . . . . . . . . G.5.2 Predicate Logic . . . . . . . . . . . . . . . . . . . . . . . G.5.3 A Formal Setup for Consistent Domains of Forms . . . . H Complements on General and Algebraic Topology H.1 Topology . . . . . . . . . . . . . . . . . . . . . . . . . H.1.1 General . . . . . . . . . . . . . . . . . . . . . . H.1.2 The Category of Topological Spaces . . . . . . H.1.3 Uniform Spaces . . . . . . . . . . . . . . . . . . H.1.4 Special Issues . . . . . . . . . . . . . . . . . . . H.2 Algebraic Topology . . . . . . . . . . . . . . . . . . . . H.2.1 Simplicial Complexes . . . . . . . . . . . . . . . H.2.2 Geometric Realization of a Simplicial Complex H.2.3 Contiguity . . . . . . . . . . . . . . . . . . . . . H.3 Simplicial Coeﬃcient Systems . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

CONTENTS

xxvii

H.3.1 Cohomology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1150 I Complements on Calculus I.1 Abstract on Calculus . . . . . . . . . . . . . . . . I.1.1 Norms and Metrics . . . . . . . . . . . . . I.1.2 Completeness . . . . . . . . . . . . . . . . I.1.3 Diﬀerentiation . . . . . . . . . . . . . . . I.2 Ordinary Diﬀerential Equations (ODEs) . . . . . I.2.1 The Fundamental Theorem: Local Case . I.2.2 The Fundamental Theorem: Global Case I.2.3 Flows and Diﬀerential Equations . . . . . I.2.4 Vector Fields and Derivations . . . . . . . I.3 Partial Diﬀerential Equations . . . . . . . . . . . 1153 . 1153 . 1153 . 1154 . 1155 . 1156 . 1156 . 1158 . 1160 . 1160 . 1161

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

XVII

Appendix: Tables

1163

1165 1167

J Euler’s Gradus Function K Just and Well-Tempered Tuning

L Chord and Third Chain Classes 1169 L.1 Chord Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1169 L.2 Third Chain Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1175 M Two, Three, and Four Tone Motif Classes M.1 Two Tone Motifs in OnP iM od12,12 . . . . . M.2 Two Tone Motifs in OnP iM od5,12 . . . . . M.3 Three Tone Motifs in OnP iM od12,12 . . . . M.4 Four Tone Motifs in OnP iM od12,12 . . . . . M.5 Three Tone Motifs in OnP iM od5,12 . . . . 1183 . 1183 . 1184 . 1185 . 1188 . 1195

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

N Well-Tempered and Just Modulation Steps 1197 N.1 12-Tempered Modulation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1197 N.1.1 Scale Orbits and Number of Quantized Modulations . . . . . . . . . . . . 1197 N.1.2 Quanta and Pivots for the Modulations Between Diatonic Major Scales (No.38.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1199 N.1.3 Quanta and Pivots for the Modulations Between Melodic Minor Scales (No.47.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1200 N.1.4 Quanta and Pivots for the Modulations Between Harmonic Minor Scales (No.54.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1202 N.1.5 Examples of 12-Tempered Modulations for all Fourth Relations . . . . . . 1203 N.2 2-3-5-Just Modulation Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1203 N.2.1 Modulation Steps between Just Major Scales . . . . . . . . . . . . . . . . 1203 N.2.2 Modulation Steps between Natural Minor Scales . . . . . . . . . . . . . . 1204 N.2.3 Modulation Steps From Natural Minor to Major Scales . . . . . . . . . . 1205

xxviii N.2.4 N.2.5 N.2.6 N.2.7 Modulation Steps From Major to Natural Minor Scales Modulation Steps Between Harmonic Minor Scales . . . Modulation Steps Between Melodic Minor Scales . . . . General Modulation Behaviour for 32 Alterated Scales . . . . . . . . . . . . . . . . . . . . .

CONTENTS . . . . . . . . . . . . . . . . . 1206 . 1206 . 1207 . 1208 1211 . 1211 . 1211 . 1212 . 1213 . 1214 . 1216 . 1217 . 1218

O Counterpoint Steps O.1 Contrapuntal Symmetries . O.1.1 Class Nr. 64 . . . . . O.1.2 Class Nr. 68 . . . . . O.1.3 Class Nr. 71 . . . . . O.1.4 Class Nr. 75 . . . . . O.1.5 Class Nr. 78 . . . . . O.1.6 Class Nr. 82 . . . . . O.2 Permitted Successors for the

. . . . . . . . . . . . . . . . . . . . . . . . . . . . Major

. . . . . . . . . . . . . . . . . . . . . . . . . . . . Scale

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

XVIII

References

1221

1223 1255

Bibliography Index

CONTENTS

XV. Appendix: Sound I. Introduction and Orientation II. Navigation on Concept Spaces III. Local Theory IV. Global Theory

Leitfaden

XVI. Appendix: Mathematical Basics

XVII. Appendix: Tables

XVIII. References Bibliography Index

CD-ROM

XIII.Operationalization of Poiesis

V. Topologies for Rhythm and Motives

VIII. Structure Theory of Performance

VI. Harmony

VII. Counterpoint

IX. Expressive Semantics

X. RUBATO®

XIV. String Quartet Theory

xxix

XI. Statistics of Analysis and Performance

XII. Inverse Performance Theory

xxx

CONTENTS

ToM_CD

ToposOfMusic.pdf legal software GPL.pdf README.txt presto documentation examples music hbdisk README.txt czerny_chopin mystery_child synthesis programs prestino.prg presto README.txt sources README.txt source1 source2 rubato macrubato examples program source nextrubato documentation source performances contrapunctus_III kuriose_geschichte traeumerei alptraeumerei takefive audio_files stemmata audio_files presto_files

rubato README.txt

Part I

Introduction and Orientation

1

.

It is a universal phenomenon of symbolic and physical.2. refer to [361]. not deal with the full reality of music. Our selection of corresponding fundamental scientiﬁc domains is not meant as a qualitative judgment over other scientiﬁc or artistic domains. music appears in a wide range of human realms. but for the time being. We shall. as it appears in psychological. Therefore. physiological. The basic scope of the systematization as described in this book is declared. For a semiotically motivated systematization of music which comprises these aspects. systematic and historic presence. In fact—as a result of a lengthy concept development—an explicit concept framework for musicology will only be achieved in section 19. the very concept of music. However. developing a concept of music should not get oﬀ with a deﬁnition ex machina. The need for such a support reveals a strong distinction from chemistry. physics or other natural sciences. religious and political contexts. see [361] for a detailed discussion of the “unscientiﬁc status of musicology”. to navigate within a safe concept framework. formal and emotional. social. It is one of the main goals of this work to develop such a framework. The point is not that these sciences are dispensed from the fundamental question of what they are about. –Σ– This book deals with the topos. it merely names the scientiﬁc pillars of the musical realm. however. 3 .Chapter 1 What is Music About? Musik ist das ganze Leben. but oﬀer propaedeutic orientation tools in order to make the reader understand why certain conceptual mechanisms or deﬁnitions are built. we just notice that musicology still lacks a stable concept framework. “Doing” musicology is much less easy. Later in this book we shall discuss some of the reasons for this deﬁciency. individual and social. to the “working scientist” paradigm: Realization of one’s scientiﬁc status by doing science. Rather the characteristic diﬀerence to musicology—and other humanities—is that natural sciences oﬀer a fast access to eﬀective activities. This chapter describes the overall extension of music-related activities in the spacetime of human existence. The historical overhead in musicology is an index of the eﬀort requested to get oﬀ the ground. Rudolf Wille [578] Summary.

or make an expressive performance. We recognize that developing a consciousness of their presence in music sheds a particular light on this subject.1: The four fundamental activities in music are visualized as sides of a tetrahedron. documentation. reception. and this on any level or reality. such an intertwinement is an exotic aspect. The following classiﬁcation of music-related activities probably applies to many cultural ﬁelds. or write or let a computer software ‘write’ a composition in a mental reality. For the sake of coherent representation. we represent these activities as sides of a tetrahedron. For example. see ﬁgure 1. But we cannot ‘make’ an analysis in this sense. These activities testify to a universally ramiﬁed presence of music in culture. since this kind of activity receives a given musical body. For example. → Reception refers to whatever is recognized as being on the side of ‘taking’ something from music on any level or reality. but musicology has to deal with it and this makes the case a very special one.2. In most sciences. and communication. and even arts.1.1 Fundamental Activities Summary. and communication. and then their foundation upon established scientiﬁc research ﬁelds. documentation. The four activities are as follows: production. –Σ– Perception Production Communication Documentation Figure 1. and because of the discussion in section 1. reception. we may hear a sound or let a machine decompose . we can make a sound on an instrument. The musical realm distributes among four types of artistic or scientiﬁc action and reﬂection: production. This is due to the fact that music traditionally relies on a strong connection between artistic facticity and intellectual reﬂection.4 CHAPTER 1. WHAT IS MUSIC ABOUT? In this vein we want to start the preliminary orientation by presenting the fundamental activities related to music. 1. These concepts are understood as follows: → Production refers to whatever is recognized as being on the side of the ‘making’ of music.

and Imagination/Arts. reason. Documentation/History is verbally the same as in our list. the arts express a creative activity of making a work from one’s imagination. documentation is not a necessary evil. 25] which go back to the global geography of science as designed by Francis Bacon in [34]: Reason/Philosophy. understand. documentation. But we may not perform a composition in the sense of a creative restatement. Now. imagination.1. Production leads to documents. In the “Encyclop´die” of Denis Diderot and Jean e le Rond d’Alembert. live production of musical sounds by improvising musicians is perceived by other musicians of the band.1. it deals with what in German is better conceived as “Vernunft”. and documentation are worthless abstracta. we can state that the main concerns in the classical encyclopedia of the Age of Enlightenment are congruent to the four basic music activities. or make an analysis of a musical score—by hand or supported by a computer. but they are broad enough to evidence the immense variety of perspectives when dealing with music. It is further remarkable that the synopsis of these equally important music activities draws a picture of perfect encyclopedic universality. above all. and so on. Let us make this clear. documents are retrieved and interpreted by instrumentalists or musicologists. And. → Communication is related to whatever happens when the three preceding activities are put into relation with each other. FUNDAMENTAL ACTIVITIES 5 it into its partials. a basic fact that we shall discuss in depth in chapter 5. the genuine encyclopedic concern is communication of the global orbit of human knowledge. And we insist on the fact that all these activities are of comparable relevance. it is substantial. Whatever is on the side of a production of some new musical entity is understood as being on the productive side. This is not only a comfortable raison d’ˆtre for music as a cultural phenomenon. Musical notation. It evokes the fundamental question of identity of an artistic work: Can we say that Beethoven’s Fifth is that determined text? Or is performance intrinsically part of the Fifth’s identity? Trying to understand music under permanent abstraction from any one of these aspects cannot succeed. What we have not inserted so far is the activity of communication. The same holds for reception. they are activities and not passive contemplation. receive. observe something which is presented to your intellect. and reception. data base systems for musical objects. which means to perceive. Summarizing. Since music is an art where time plays an omnipresent role. we have three basic conceptual coordinates [14. On the other hand. → Documentation deals with whatever is related to bringing musical facts into more or less permanent sign systems. Reason/Philosophy is the counterpart of reception. Communication deals with the omnidirectional processes between production. all this is subject to music documentation. Such a body must comprise a variety of . If we compare these entities to the music activities. then answered. its physical production in performance cannot be separated from the product as a preliminary task. the question of how and on which media ethnomusical performances should be stored. We do not contend that these four fundamental activities represent the only possible classiﬁcation scheme. Without this driving motor. tradition of musical texts. the making of music. they correspond as follows: Imagination/Arts is the counterpart of production. it also has deep consequences e for the construction of a viable concept framework. Documentation/History.

and psychology. diachronic semiotics is the adequate research ﬁeld. and psychology. The relation between the two lists of activities and domains is discussed. and which in this way mediate between mental and psychic contents. the historic dimension of music seems to be neglected. but not vice versa. psychology is the predestined science. physics. The fundamental scientiﬁc domains relating to the activities described in the preceding section include semiotics. such as music.2 Fundamental Scientiﬁc Domains Summary. physics. WHAT IS MUSIC ABOUT? ontological1 perspectives and cannot refrain from contents in favor of formal virtuosity. Let us explain this selection. And to understand the psychic contents. and this evokes mathematics as a directly 1 Ontology is the philosophical discipline dealing with the concept of “being”. We have to show that the four disciplines are suﬃcient to cover the ﬁeld of music and that no one of these sciences is superﬂuous in that it either is not involved at all. we have to accept such general uncertainties and will proceed in full consciousness of the provisional state of the art of epistemological classiﬁcation. mathematics. To deal with such a system. we have to make clear what “fundamental scientiﬁc domains” should signify. mathematics. At this point. whereas general history would be too generic in view of the given subject. First. They have a strong properly mental (“symbolic”) existence. . The system of predicates to be developed in chapter 18 will account for this speciﬁcation. blurred or otherwise unclear by the very deﬁnition of basics of particular disciplines. and here. However. For instance. physics are indispensable. And every natural science is dominated by mathematics. Is this selection a “basic” one as claimed above? There are several critical points which we should clarify. This is explicated and justiﬁed—in particular with regard to the only apparant elimination of the historical perspective which dominates traditional musicology. a kind of ‘basis’ of disciplines which are necessary to cover the subject. Now. semiotics has a strong competence in diachronic study of a sign system. chemistry is dominated by physics. Music history is prominently the history of a system of signiﬁcant signs (much like diachronic linguistics). Deﬁnition 1 (First provisional version) Music is a system of signs composed of complex forms which may be represented by physical sounds. Musical forms are not only mediated via physical representation. within the hierarchy of disciplines. In this sense. though this is only a ﬁrst approximation in order to delimitate the scientiﬁc overall extension. To represent these forms on the physical level. one is looking for a minimal set. –Σ– To begin with. To describe these forms. or that it is only involved via a proper sub-discipline. if one has to locate a special research subject. Many of the hierarchical relations are quite problematic. are such a ‘basis’. we cannot circumvent a provisional description of what music is about. We may view the collection of sciences as being a hierarchy of disciplines with its particular relations being deﬁned by a substantial dependency. 1. we argue that four sciences: semiotics. semiotics are naturally evoked.6 CHAPTER 1. mathematics is the adequate language.

The psychological aspect is less substantial. Perception Production Physics Psychology Communication Semiotics Mathematics Documentation Figure 1. Also pedagogical problems (music education) are subsumed under the label of psychology since it is the human psyche which is educated. Finally. FUNDAMENTAL SCIENTIFIC DOMAINS 7 involved research aspect. and the physical storage techniques. We come back to this theme in chapter 2. To complete the picture. of a tetrahedron. If we pay attention to the obvious semantics of this conﬁguration. vertices. and communication. and never will hope or hypothesize reductionism whatsoever. It is an extremely important point to recognize the relative autonomy of these four research ﬁelds in their descriptive force.2 and leave it to the reader to judge our selection.1 by a basic science. are visualized as sides. . documentation. it turns out that every triangular activity side is “spanned” by three of four vertices.1.2: The four fundamental activities. resp. There is no interest to view psychological reality as a surface of neurophysiological depth structure. we do “absorb” computer science and sound engineering within physics. but also in computer science which is concerned with data representation and processing in music—be it on the generic level of software or on the level of advanced digital sound synthesis and analysis. our selection is not pragmatic but ontological. Hence. In this sense. and this not only with respect to sound production (classical acoustics and physics of musical instruments). their formal representation. Let us simply give one example: Documentation is concerned with signs. we have to place the labels in such a way that each activity is delimited by the three most relevant sciences for its execution. This means that we do not. If we “label” each vertex of the tetrahedron shown in ﬁgure 1. Physics has a prominent role in music. We propose the labeling drawn in ﬁgure 1. together with the mathematical formalisms in theoretical computer science. resp.2. reception. social aspects of music are viewed as objects of social psychology on one. their counterparts of basic sciences in music. we should try to ﬁx the relative positions of the four basic sciences with respect to the four fundamental activities of production. It is evident that activity is more or less related to some of the four scientiﬁc disciplines. and socio-semiotics on the other hand.

where the question “What is music about?” can be initiated. . not overdress this elegant conﬁguration as a “magic” tetrahedron.8 CHAPTER 1. however. And it is also a very practical orientation scheme for designing future music research policy and strategy. the visualization merely serves as a concise and relatively well positioned synopsis of the overall situation. We do. WHAT IS MUSIC ABOUT? whence documentation is “spanned” by the respective vertices.

has meaning and mediates on the physical level between its mental and psychological levels. It is well known that the precision of mathematical results. This chapter deals with an ontological orientation in the subject of music. a Was wir hier sind. und wie der Berge Gr¨nzen. . it is proposed to set up from the very beginning a “three-dimensional” ontology. The local nature of this orientation scheme is discussed. It was already contended in chapter 1 that music is communication. an ontological system which tells rather where the concept of music subsists than what its being 9 . The topography involves three mutually independent dimensions: communication. On the other hand. and semiosis. i. Friedrich H¨lderlin: Die Linien des Lebens. kann dort ein Gott erg¨nzen a Mit Harmonien und ewigem Lohn und Frieden. (1843) [469] o Summary.Chapter 2 Topography Die Linien des Lebens sind verschieden Wie Wege sind. In the following “topography of music” we shall sketch an ontological model without any dogmatic or unmodiﬁable character. Such an orientation is topographic in nature since it oﬀers a number of ontological “dimensions” and “coordinates” to proﬁle musicological discourse and helps avoiding misplaced or blurred arguments. logical. mathematical or computational methods in music does not necessarily have to be an expert in music philosophy.e. . reality. these methods require a suﬃciently reﬁned orientation within the complex ontology of music in order to avoid erroneous interpretations of results obtained by use of these exact methods. Nonetheless. the extensive concept framework which we want to set up cannot even be approached without a ﬁrm pointer to musical ontology. together with a poor knowledge about the delicate ontology of music may provoke a dogmatism for which mathematics is unjustly made responsible. In view of the generic (and ﬁrst provisional) delimitation of the concept of music (deﬁnition 1). though constantly available as a powerful coordinate system to locate the problems within the organism of music. –Σ– Whoever is dealing with formal.

The argument is correct. 2. The dimension of musical reality involves physical. 2. In musical acoustics it is often claimed that—according to Fourier’s theorem—a sound “is” composed of “pure” sinoidal partials. –Σ– Music takes place on a wide range of realities. Following Immanuel Kant [258. TOPOGRAPHY is like.10 CHAPTER 2. but not eliminated. To understand music as a whole. Ontology then questions the place of the concept. Diﬀerentiation of realities is crucial for avoiding widespread misunderstandings about the nature of musical facts. These coordinates are • REALITY • COMMUNICATION • SEMIOSIS1 We shall hereafter (sections 2. and its communicative extension.2. psychological. a concept is a local place. B 324]. Whether the overall ontological question is settled by this procedure is not answered here. We do. a topos in the knowledge space. . and mental layers. we are not concerned with music as such (a Kantian “Ding an sich”) but with music that we localize and conceive within our knowledge system. However. It might be argued that this topographical reduction of ontology simply delegates the difﬁcult question to the ontological coordinates: instead of asking for one ontological speciﬁcation. It is not question of reducing one of these realities to the others: Either of them has an autonomous existence which can at most be transformed into others. a theorem of pure mathematics. Let us shortly deviate on this philosophical line to justify the somewhat particular approach.1 Layers of Reality Summary. you have to specify simultaneously its levels of reality. Being a fact of music means having these three perspectives or ontological coordinates. the space-coordination of the music topos. in other words. there is no 1 Semiosis refers to the fact that music deals with a system of signs. we have to examine three partial speciﬁcations. Omitting any one of these determinants is an abstraction (though not an aberration) from the full ontology. Dealing with semiosis without being concerned with its level of reality (and vice versa) is a huge advantage.5.1. Giving the ontology a topographical “turnaround” simpliﬁes the problem. p. roughly stating that every periodic function is a unique sum of sinoidal components. Its a priori status is a mental one.3) introduce them in detail. stress three ontological coordinates as an intrinsic unity pointing at the concept of music. A representative example of this problem is Fourier’s theorem. Let us instead begin with a closer inspection of the three proposed coordinates—we shall come back to this subject after that in section 2. but it is also true that these partials are simpler. The great majority of reality-speciﬁc phenomena cannot even be translated into external layers without substantial deﬁciencies. more elementary. 2. its semiotic character. mental and psychological levels. They may be grouped into physical. To be clear.

3. see [127.1. All physical sounds are equally natural. or by the tape patchwork of musique concr`te. To give the claim a physical status. its path through multiple relays stations of the brain stem.1 Physical Reality Music is essentially manifested as an acoustical phenomenon. Fourier’s statement is just one of an inﬁnity of mathematically equivalent orthonormal decompositions based on “pure” functions of completely general character. But ontologically. a computer driven synthesizer via loudspeakers. The use of physical sounds is a function of very special devices for synthesis and analysis. Physical reality of music is only relevant as an interface between ‘expressive’ and ‘impressive’ dynamical systems. reference to particular sound representation models is good for synthesis options.1).and archicortical centers for auditory processing and memory. VI]. The speciﬁc phenomenon within the psychological topos corresponds to another phenomenon within the physiological topos.1. the neo. such as the cochlea of the inner ear (see appendix. see appendix B. be they produced by a live violin performance. however. e On the other hand. there is no reason nor is it ontologically possible to reduce one reality to others. The problem is rather that classiﬁcation of musical sounds is arbitrary without reference to their semantic potential. 2. Without a speciﬁc link to physics. it is not known which analysis of the musical sounds takes place on the higher cortical levels. Nonetheless. We now give an overview of the three fundamental topoi of reality and their speciﬁc characters. see section 12. the phenomena do not collapse. This extremely complex physiological system is far from being understood. conserve the psychological ontology of the phenomenon. which is physically sensitive to the ﬁrst seven partials in Fourier’s sense. The problem is rather to describe the transformation rules from the manifestation of a phenomenon in one reality to its correspondences within the others.1. This means that on the cognitive level. its acoustical characteristics are less—if at all—condensed within a unique physical sound quality than in the physical input-output systems for sound processing. This is not due to missing synthesis or analysis methods and techniques. there are various digital sound synthesis methods.3. Therefore. but not as a ﬁrm reference to human sound processing. and for speculative models of cognitive science [292]. . it would be necessary to refer to a concrete dynamical system. This is reﬂected in the fact that to this date. Methodologically. Even though some insights into the dynamics of the cochlear subsystem do exist. auditory nerve. such as Heschl’s gyrus and the hippocampal formation.2.2. see section 12. human sound analysis is not yet understood. the central receptive system for music is the human auditory system: Peripherical and inner ear. It is not even clear how the elementary pitch property of an ordinary tone is recognized [72]. section B. To be clear. made by means of special instruments and perceived by humans. there is no generally accepted classiﬁcation method for musical sounds. see appendix A. ch. a neurophysiological transformation (“explanation”) of a psychological phenomenon does not. Besides classical analog sound synthesis methods as they are realized on musical instruments. LAYERS OF REALITY 11 physical law to support this claim.3 for a thorough discussion.

they may be complex and interlocked in concrete situations. Obviously. They reﬂect the fact that music is composed as well as analyzed on a purely mental level. TOPOGRAPHY 2. Like Molino’s scheme. Such an emotional reality of music is neither subordinate nor abusive. see ﬁgure 2. be it a “parler peinture” or “parler musique”. Molino’s scheme partitions the communication process into three instances: creator. It is a common misunderstanding that musical notation is an awkward form to designate physical sound entities. Scores are mental guidelines to an ensemble of musical objects. Communication is the second dimension of our topography.1. Like other realities of music.12 CHAPTER 2. we describe the tripartite communicative character of music. irreducible ontology. Its structure is comparable to Oskar B¨tschmann’s “grosses abstrakt-reales Bezugssystem der Auslegung” which he proposed as a a hermeneutical framework for the analytical discourse in ﬁne arts [43]. and ending on the receptive side of the listener’s esthesic perception.2 Mental Reality Just like mathematical. and esthesis. and socially. the psychological dimension cannot and needn’t be reduced to others. cannot succeed without considering their communicative structure. musical creations germinate as autonomous mental entities. on the contrary: to music lovers the emotional response is a dominant aspect. To these. the phenomenological surface of music is linked to mental schemes which we call scores: oral or written text frames of extra-physical speciﬁcation. 2. logical and poetical constructions. playing a chord on a piano corresponds to drawing a triangle on a sheet of paper. This was already known e to Pythagoreans [541] and deﬁned as a central issue of music by Ren´ Descartes [126]. see [120]. 2. music fundamentally expresses emotional states of its creators and emotionally aﬀects its listeners. scores do point at physical realization. neutral level. B¨tschmann’s construction issued from the insight that a dispassionate discourse on arts. Being a trace of intrinsically human activity.2 Molino’s Communication Stream Summary.1. Molino’s scheme is an abstraction to the essentials.3 Psychological Reality Besides its physical manifestation and its mental framework.5. It is not by chance that song titles such as “It don’t mean a thing if it ain’t got that swing” de facto reduce musical meaning to an emotional category. traversing the mediating neutral datum of the virtual work. it is a manifestation of an autonomous. it splits into a three-part stream starting from the poietic instance of the creator of a work of art. In this sense.1. and listener. Following Jean Molino. three types of analytical discourse correspond: poiesis. In . individually. but only as a projection of a mental stratum into physical reality. A fact of harmony or counterpoint is an abstractum much the same as an ideal triangle in geometry. –Σ– According to Jean Molino [377]. the feeling of swing. work. see section 2. These concepts are technical terms which have to be explicated in the following.

But the latter is not automatically inscripted in a vibration. a creator instance subsumes all factors which are essential and suﬃcient for the production of a musical work. the creator instance may as well be the musician or the performing artist. the improvisational aspect is a genuine making of the music. particular. worse.2 2. bears a strong poietical ﬂavor—quite contrary to the naive and widespread belief. “documentation” relates to “neutral level”. and communication. And. and the performance of a classical piece of music is a creational act. In jazz. documentation. Poiesis is a sharp analytical instrument. Poiesis is concerned with the individual condition of the creator as a result of its history of development as well as with the role of the broader socio-cultural frame of the work. for example. 2 We should like to relate these concepts to the activities of production. it usually results from the sound construction as a product of a periodic vibration of a determined frequency and of an envelope (see appendix A. Neither frequency of the periodic vibration nor the envelope are uniquely determined by their product.1: The three-part scheme of music communication following Molino. For example pitch (logarithm of frequency) is no invariant of the sound. We have nonetheless maintained these concept frameworks since Molino’s technical terminology does not reﬂect the aspect of overall activity in music. reception. “reception” is on the side of “esthesic level”. In speciﬁc cultural contexts or in a more reﬁned discourse about art production. Fourier’s theorem only states uniqueness of its coeﬃcients if the fundamental period is given.1). This level describes the “sender” instance of the “message”.2.1. not even the frequency of an ideal periodic vibration is uniquely determined from the vibration as such. classically realized by the composer. Activities are instances of what here is denoted by “communication”.1. MOLINO’S COMMUNICATION STREAM 13 Creator Production Work Reception Listener Poiesis Neutral Niveau Esthesis Figure 2.2. it can relate to completely “objective” physical or mathematical facts.2. In fact. According to the Greek etymology (πoιειν = to make). . or receptive passivity of the listener towards work and creator. such as pitch.1 Creator and Poietic Level To begin with. Molino does not prejudice such a thing as causality between creator and listener by means of the work. Such parameters are part of poiesis: the way of making the sound. etc. “poiesis” relates to the one who makes the work of art. What we call “poietic level” corresponds to “production activity”. not the “neutral” sound object. The corresponding analytical discourse of the work’s poiesis has been introduced by Etienne Gilson [183]. as described in section 1.2. This shows that poiesis is not restricted to psychological aspects of “how to fabricate a composition”. Already the very deﬁnition of a sound by means of its deﬁning parameters. color. and the activity of communication corresponds to the transition processes between all the three levels in Molino’s system.

In music. and neutral analysis is the variety of all these perspectives. A momentous analytical perspective is just one among many possible accounts of the work. This e means that. Second example: The improvisational culture of jazz—which in its making only marginally relates to traditional western scores—is based on the concept of the interior score (“partition int´rieure” [487]).14 CHAPTER 2. In this context.g. or the beat of waves on the open waters of the sea. Decisions about including sound color or tuning prescriptions in Bach’s “Wohltemperiertes Klavier” do in any case redeﬁne a new and diﬀerent work. exact methods will play an eminent role in realizing such a task.2. the deﬁnition of a work is left with ample variants. the analytical discourse which—independently of the selection of the tools—is strictly oriented towards the given work is termed neutral level. the score concept is also adequate for describing traditions which are more towards ethnomusicological antipodes. there are diﬀerent score instances. there is an interior reference system of lexical character. or for the hayashi notation systems for ﬂutes and drums.3 Listener and Esthesic Level The listener is the instance which perceives and interprets a given work. no matter: Neutral level means propaedeutic analytical precision work preceding any valuation whatsoever.2 Work and Neutral Level The work is the turning point of musical communication. Such a work may be the divine creation of a medieval tone system. a work of computer music which is explicated as far as its acoustical details. e. This relates to the dramatic diﬀerence between music and ﬁne arts. Its identiﬁcation depends upon the contract of creator (sender) and listener (receiver) on the common object of consideration. together with a selection code which guides performance. 2. It subsumes whatever determines directly the production of sound events within a speciﬁc context. for vocal utai music denoted in melodic units (fushi) to the right of texts.2. First example: In the music of Noh theater [269]. such a decision transcends the strict deﬁnition and establishes a valuation of the work in the light of historical parameters of Bach’s creativeness. Coordinates of a listener will vary from case to case and . TOPOGRAPHY 2. Evidently. Following Molino. Molino’s concept has only attracted controversies since it—faultily—seems to imply the preference of a particular analytical tool/method to the exclusion of other possible approaches. a score may be viewed as being the organizational scheme of a work. And above that. It is subject to the same type of determinants as the creator. As a matter of fact. In particular. physical realization of music is always based upon mental schemes which we call scores: oral or written text frames of extra-physical speciﬁcation. a painting is determined down to its humidity and temperature—at least from the moment it has gained ‘eternal’ appreciation and allocation in a professional museum. Apart from classical western traditions where this fact is evident. even for free jazz improvisers. Musicology has diﬃculties with the work concept since its determinants seem to be of fuzzy nature. the historical justiﬁcation for a work’s delimitation is not part of the neutral level. But such a thing as the “unique ideal interpretation” is precisely the famous unicorn of hermeneutics—a superﬂuous illusion. In fact. The fundamental fact behind the basic role of the score concept for music is that human organization in a complex time-space of acoustical and gestural nature cannot be executed without an interpersonal spiritual orientation common to the responsible participants.

Esthesis deals with perceptive valuation of a work by the listener. the study of beauty. Only in very rare cases can we expect the conservation of results from a poietical discourse within the esthesic point of view. In this way. And vice versa. into an esthesic discourse. esthesis is integral part of the analytical elements constituting the existence of a work. the creator’s existence is ﬁxed whereas the number of the listeners’ existences grows as a function of those who are engaged in the work. This latter question is a completely diﬀerent topic and cannot be identiﬁed with the former. retrograde poiesis is embedded in esthesis. retrograde structures may be recognized as objective facts within the neutral level of the score without being either constructed be the composer or consciously perceived by . The psychological question of whether a retrograde can be perceived is rather this: “Can a retrograde be distinguished from random?” Finally. but quite generally. This visualizes a well-known insight from ﬁne arts theories that the artist/creator is the ﬁrst observer. such a dogmatic phantom as “the ear” does not exist. The distinction between poietical and esthesic levels is fundamental for a sound conceptualization in musicology. The latter is a result which may be generated by very diﬀerent processes. Whereas the creator produces a work. This fact becomes manifest when we imagine the work’s production process as a ﬁctional unfolding in a retrograde movie. A classical conﬂict concerning the role of retrograde in music arises from the observation that this construction “cannot be heard and thus is a problematic feature”. Within the communication process. This valuation of the work’s attributes is realized as a function of the listener’s individually variable position. Poietics cannot be taken as the ideal access to a music work. More precisely. From the moment of a work’s existence. the creator above all diﬀers from the listener by his/her position in time. the esthesic perspective of the retrograde is concerned with the question whether and how clearly such a construct can be decoded by the listener. such as “pitch”. And this not only regarding the artistic performance. There is no thing such as a causal relation between creator and work. Like poiesis. On one hand. MOLINO’S COMMUNICATION STREAM 15 coincide only punctually with each other or with the creator’s coordinates. and vice versa. the listener perceives and interprets an already existing work. the role of the retrograde as an organizational instance is not a function of its perceptibility as an isolated structure. On the other hand. On the other. [537] the analytical discourse on the listener is e termed esthesis.2. following the Greek term αισθησισ (perception). we observe again and again the faulty trial to implant historically and systematically clearcut poietical concepts. Nonetheless. The characteristic diﬀerence between listener(s) and creator is the time arrow towards the common work. This interpretative valuation from a determined perspective is not less active than the creator’s activity in the creation of the work. in order to avoid confusion with esthetics. the ear fails as a metonymy of cortical music processing: there is a huge transformation process between the auditory cortex and the ear’s cochlear system.2. the esthesic discourse may be restated as a manifold of variants of imaginary retrograde poieses. The famous paradigmatic justiﬁcation of such a malfunction of analysis is known under the dictum that “the ear is the highest musical instance”. Communication coordinates make this discussion more transparent: The retrograde construction as a poietic technique is a common compositional tool. Example 1 The problem of symmetries in music is a good illustration of the communicationsensitive aspect. is incompatible with the multiplicity of listening cultures. It ﬁts into the toolkit of contrapuntal constructs for organizing the compositional corpus. According to a proposal of Paul Val´ry.

built upon abstract and neutral identiﬁcation. give a hint. .3.3. but see section 32. To begin with. This yields at least one justiﬁcation for an incessant performance practice in classical music. Albeit more complex than linguistic syntax.3. This coincides with the latin etymology of sign: signare = to point at.2. 2. Independently on what is communicated.16 CHAPTER 2. and 2. musical meaning is distributed over a sequence of semiotic subsystems. We give a short review of dichotomies of structuralist semiotics. valuation-free analyses. 2. –Σ– By use of a highly developed textuality of musical notation as well as by the very intention of musical expression. This ﬁrst identity is then enriched to yield the neutral identity as a result of the multiplicity of neutral. music is structured as a complex system of signs. and then (2. and on which level of reality this takes place. they express something in the sense of semiotics. It consists of a signiﬁcant expression. The esthesic identity is built from all these esthesic valuations and their mutual relations.3. spiritual contents or gestural units. The generic setup of semiotics understands a sign as being a tripartite object. we envisage a work that is given before any analysis is performed. 2. the signiﬁcate. Besides and beyond musical notation. see section 2.4. Musical semiosis reveals a complex concatenation of meta.1 Expressions Already the earliest medieval music notation is motivated by the very nature of the graphical neumes: etymologically as well as substantially they are gestural hints pointing at movements in pitch and rhythm.4) review the overall semiotic perspective of music. music intrinsically involves complex signiﬁcation processes.2. and thereby producing the expression’s content.1. the identity of a work is triply stratiﬁed. the communicative coordinates help localizing and thereby making more precise the musicological discourse. 2. Accordingly. Rather than being absent. We shall ﬁrst present the elementary sign character of musical objects (subsections 2. We call this data its abstract identity.3).and connotation-layers in the sense of Louis Hjelmslev.3 Semiosis Summary.3. inducing the signiﬁcation act of translation. TOPOGRAPHY the listener. This identity is the basis for a variety of esthesic—in particular: retrograde poietic—valuations.3. music is often viewed as an expression of emotions. In any case.2 for more details on performance and identiﬁcation.1 The Problem of Identity From the preceding communication-theoretic considerations. Neumes are aliquid pro aliquo. the musical syntax shares some of its characteristics. Summarizing. is completed.3. a work is only identiﬁed when the inﬁnite process of esthesic valuation. music has a phenomenological surface that is organized in a spatio-temporal syntax. This is not only a marginal aspect: Music is one of the most developed non-linguistic systems of signs.

which points from the expressive surface of the mediator (the signiﬁcant). It is responsible for the transformation from the expressive surface to the hidden meaning. Slightly deviating from Saussure’s original approach. i.e. . But mathematicians do associate it with a content. to the underlying meaning (the signiﬁcate). we deal with diachronic analysis. 2.3 The Process of Signiﬁcation Signiﬁcation is the most important instance for the realization of a sign. and signiﬁcate layers (Saussure’s signiﬁant/signiﬁcation/signiﬁ´). we deal with synchronic analysis. the concept of a moment really means a short duration wherein the system remains ﬁxed.3.4 A Short Overview of Music Semiotics Semiotics of music is a complex subject which cannot be dealt with in this context. This stratiﬁcation articulates the process of production of meaning e (the signiﬁcation). Like every existing semiotic system. For musical signs. The remarkable aspect of Hanslick’s approach is that it associates musical content with mathematical content. For a full account. Hanslick’s characterization is a minimal semantic setup but some kind of content can be identiﬁed. Of course. however.3. As a matter of fact. we review these items in the following section 2. This does. The fact that performance is such a central issue of music gives a strong proof of the qualiﬁed presence of signiﬁcation in music.3. We pursue this project according to Andr´ Martinet’s principle of e relevance [318]. we learn that: “Der Inhalt der Musik sind t¨nend bewegte Formen. SEMIOSIS 17 2. signiﬁcation. Let us recall that Barthes’ position does not mean that music is interpreted as being a type of language. 471].2. If. We shall describe music semiotics from the perspective of structuralist semiology as it has been sketched by Roland Barthes [41] as a generalization of the linguistic theory of Ferdinand de Saussure [471]. If we investigate the system’s state at a determined moment. On the contrary. we refer to [361]. the existentiality of the signiﬁcate is that of a concept. music is not what it is meant to be. a triangle is a mathematical object that essentially reduces to form. the latter being Saussure’s preferred signiﬁcates). It states that selection of materials and methods has to be justiﬁed by relevant criteria in the sense of points of view towards the actual object—music in our case. Without the laborious eﬀort of turning signiﬁcant expressions into the reality of their meaning. instead. this semiotic process bears a highly diﬀerentiated structure which is sensitive to Ferdinand de Saussure’s dichotomies [41. it is part of a “lexical body”. usually with a platonic entity addressed by abstract symbols or by drawings with precise quantiﬁcations. in contrast to the “real” things or to the psychological imagination.3. not dispense us from a short overview of this subject since there is no hope to understand music without a minimum of semiotical prerequisites. the system’s objects are observed in their temporal development.” This evidences that the notated complex of musical graphems is o not the content but points at some kind of sounding content: they mean something. music is embedded in history. To say that a system is semiotic means that it is articulate according to the fundamental stratiﬁcation of signs into signiﬁcant.4. we shall repeatedly stress signiﬁcant diﬀerences between the music system and linguistics.2 Content According to the famous dictum of Eduard Hanslick [206]. of a system of sayable things (λεκτ oν in the stoic tradition. 2.3.

If two signs are adjacent in space or time they are called contiguous. A standard means for diminishing arbitrariness is the use of system immanent logic or of instances exterior to the system which are at disposal as given evidences. For example. sychronicity and diachronicity are not strictly separable since—just as with physics—the space-time continuum of music is indecomposable. “now”) are exceptional signs. a rich deixis cannot succeed without a rich lexical support.2. is the standard situation. alto.g. But this does not mean that the lexical part of the system is marginal. such as emotional signiﬁcation. a texture of four parts (soprano. Signiﬁcation is arbitrary if it is deﬁned by pure convention. The latter is the case if our system lives in the context of other systems (e. we allow a multiplicity of simultaneous concatenations. A famous example of system immanent logic is Roman Jakobson’s poetical function. the emotional signiﬁcate of music can neither be “lexicalized” by decree. shifters (such as “I”. the concept of a sonata form is lexical. whereas the signiﬁcation of a tremolated kettledrum beat towards the signiﬁcate “thunder” is motivated as onomatopoeia. signiﬁcation. Saussure distinguishes between arbitrary and motivated signiﬁcation. for example.6. no further abstraction is needed. In linguistics. the etymology of the concept of consonance is a diachronic subject. A chain or network3 of signs in successive contiguity is called syntagm. whereas looking at semantics of the augmented triad in Franz Liszt’s “Faust” symphony points at a synchronic analysis. bass) or a discantus which is set against a cantus ﬁrmus voice. If two signs are similar then they are called in apposition.18 CHAPTER 2. or signiﬁcate. As a consequence. the signiﬁcation “highten a pitch by a semitone!” of the sharp # is pure convention. tenor. It is evidently a diﬃcult business to “hit” the unsayable by use of an arsenal of lexical approximation pointers. a huge number of semiotically well-motivated music software has started to break this artiﬁcial obstacle to the world of music. In music deixis.1. The special class of shifter or deictic signs is very important in music and is characterized as follows: Their signiﬁcate transcends the lexical reality and penetrates the unsayable existence of the system’s user. syntagmatic articulation of music looks quite diﬀerent from linguistics. tones which are perceived with loudness equal to that of a reference tone deﬁne the paradigm of loudness of this tone. we come back to this important diﬀerence to the syntagmatic contiguity in section 49. Notice that the arbitrariness of musical notation is a major source of the elitarian music culture. Whereas linguistic syntagm is one-dimensional (in time).) Similarity may regard any of the three layers—signiﬁcant. On the contrary.4. Two such tones do not necessarily yield the same loudness perception (see appendix B. It is a fact of concrete nature. “here”. in linguistics also: syntagmatix axis. fortunately. Whereas. A progression of degrees or a contrapuntal interval sequence are syntagmata in time.1). not only one chain such as the case in ordinary language. As a whole. It requires a high degree of abstraction. 48. As a function of the type of signiﬁcation process. TOPOGRAPHY For example. (Martinet uses the term “opposition” in linguistics but in music this binary term is too narrow. The collection of all signs which are in apposition to a given one deﬁne the paradigm of that sign. these are syntagmata in pitch space. software environments) which may help to operationalize it. nor verbalized whatsoever—contrary to numerous ﬂopped attempts.2)! 3 This extension transcends the usual deﬁnition. It enables production of signiﬁcates by means of “projection of the paradigmatic axis into the syntagmatic axis” (see below and sections 11. the music syntagm extends simultaneously in several dimensions. Contiguity is “in praesentia”. Apposition is a relation “in absentia”. For example. . the system has sign character but it is composed from single signs. Evidently.

the topographic cube is deﬁned. While each of the described dimensions splits into three “coordinate” values. Generalizing Hjelmslev’s approach. a given semiotic system can appear as a sign-theoretic instance (signiﬁcant. and semiosis. Here. For an overview of the entire music system articulation in the sense of Hjelmslev. The Hjelmslev articulation of music [361] shows that musical meaning is not concentrated on a semiotic spot. we shall adopt the dichotomy of (music) norm /(music) process (corresponding to language/speech). musical meaning is the result of a complex development of signs. this cube’s orientation is of local nature. On the other hand. in Hanslick’s terminology. This dichotomy shows a strong diﬀerence to the linguistic dichotomy in so far as such a thing as a music norm is neither a synchronic nor a diachronic constant. conversely.2. However. it is on the contrary distributed among several system layers of denotation and connotation. 2.4. it does not provide a global point of view. signiﬁcation. see [361]. In other words. The last of the famous Saussurean dichotomies—it is also decisive for distinguishing music from language—concerns language against speech (Saussure: langage/parole. Putting together the three dimensions of musical topography. This . we may view these contributions as coordinate axes of a more complete musical ontology. If it appears as a signiﬁcant layer. metasystemic. it grows from simple categories of signiﬁcation to more and more all-embracing and even transcendental categories as they appear in highly spiritual musical expression. The signiﬁcant system is the termed denotative level. In general. “sound speech”(“Klangrede”) as they used to call it. we speak about connotation. the cube involves 33 = 27 distinguished topographic locations. This means the contrast between general reglementation and concrete syntagms e and paradigms as far as dialects and revolutionary poetical invention. or signiﬁcate) of another such system. And emotional signiﬁcates of music by deﬁnition forbid any normative ﬁxation. the musical denotation level in classical score notation is extremely normative. one should also envisage the case where a system appears as the signiﬁcation layer of its supersystem. where as the superior system is termed connotative level. communication. For example. Their role in guiding a sophisticated discourse on music is discussed and exempliﬁed.4 The Cube of Local Topography Summary. The analphabet diﬀers from the score illiterate in so far as production of musical meaning has very little to do with score notation whereas competence in linguistic articulation is intimately related to lexical competence. we have the double articulation Note ⇒ Ton ⇒ t¨nend bewegte Form. the double articulation in language: grapheme ⇒ acoustical signiﬁcate ⇒ conceptual signiﬁcate can also be observed in music. grammar of language and harmony condivide their metasystemic character [70. In order to avoid confusion with the linguistic context. ?]. semiotic systems are complex interlockings of connotative. a system occupies the signiﬁcate layer of another system. On the other o hand. THE CUBE OF LOCAL TOPOGRAPHY 19 According to Louis Hjelmslev [226]. –Σ– After presenting the ontological dimensions of reality. and as such a source for elite formation whose control is often confused with musicality. If. and signiﬁcation layers. Barthes: structure/proc`s). they are music processes. whereas the inferior system represents the objectlevel or objectsystem. the superior system is termed metalevel or metasystem.

we have to specify its three coordinates in reality. it may as well occupy any subset of the cube. communication. Er lebt die Musikst¨cke in u sich.2. This produces a set of 33 = 27 possible topographic locations. Each coordinate may take one of three values as listed above. we present a classical text of music critique. p. the 27 points are just the elementary positions from which more complex ontological situations are composed. To illustrate this topography. Significate Semiosis Signification Significant Mental Reality Psychic Physical Creator Work Listener Communication Figure 2. • communication: creator–work–listener • semiosis: signiﬁcant–signiﬁcation–signiﬁcate This suggests a geometric representation of the ontological variety as a “topographic cube” of musical ontology. communication. die er vortr¨gt. To describe the ontological position of a musical object. cited following Knapp [265. It is the review of a concert by Franz Liszt written by Ludwig Rellstab in Berlin’s Vossischen Zeitung .2: The cube of musical ontology. Jetzt wird ein neuer Geist in ihm lebendig. see ﬁgure 2. each of them being articulated in three “values”: • reality: physical–psychological–mental. it is not necessary that a general object be localized at a single location. was bisher von irgend jemand einzelnem bezwungen worden .86]: Unter dieser Erweckung der vorteilhaftesten Eindr¨cke setzte er sich an das Inu strument. TOPOGRAPHY means that an ontological localization can be interpreted as a point in a three-dimensional cube spanned by the axes of reality. and semiosis. Of course.20 CHAPTER 2. W¨hrend er mit erstaunensw¨rdigster Gewalt der Mechanik a a u eigentlich alles leistet. and semiosis. So.

—Expressions such as “Aﬀekt seines Spiels”. “B¨ndnis mit dem Geist”. sie bleibt nicht getrennt von ihm. und außerdem noch ein ganzes F¨llhorn neuer Erﬁndungen. and “Tatsache des Inneren”. sondern wirkt in dem m¨chtigen B¨ndnis mit dem Geist. “Geist den u Formen einhauchen”. der sie erzeugt. –Σ– Observe that the topographic speciﬁcation of a fact of music is a local resp. “fesselnd”.5 Topographical Navigation Summary. “geistige Bedeutsamkeit”. “Aﬀekt”. “einhauchen”.—The neutral level is addressed in passages such as “Musikst¨cke vortragen”. Diese geistige Bedeutsamkeit seines Kunstwerkes pr¨gt sich aber auf das lebendigste in seiner Pers¨nlichkeit aus. The local and recursive nature of such a ramiﬁed navigation is described and exempliﬁed. u The communicative dimension speciﬁes in its poietical coordinate with expressions such as “tr¨gt vor”. “Kunstwerk”.)—Signiﬁcation is traced in expressions such as “lebt die Musikst¨cke”. the signiﬁcate of a sign in a metasystem—by deﬁnition—is an entire sign of the objectsystem. v¨llig ungekanu o nter Eﬀekte und mechanischer Kombinationen vor uns aussch¨ttet. For example. “mechanische Kombination”. den er diesen Formen einhaucht. das bei weitem anzieu hendere. “Aﬀekte der leidenschaftlich aufget¨rmten Seele”. recursive one in the following sense which one may visualize as a conceptual zoom-in eﬀect: • Parts of a sign may be entire signs of their own right. . Hence the corresponding discursive navigation receives a ramiﬁed path structure: A priori. Die a o Aﬀekte seines Spiels werden zu Aﬀekten seiner leidenschaftlich aufget¨rmten Seele u und ﬁnden in seiner Physiognomie und Haltung den treuesten Spiegel. each topographic location may open or participate in another topographic variety. a u ¨ “Geist”. “Haltung”. and “Uberﬂ¨gelung der Erwartung und Forderung”. “Gewalt der Mechanik”. and “in u dem m¨chtigen B¨ndnis mit dem Geist”. u u o and “fesselnd”.—The mental level is addressed in expressions such as “neuer Geist wird lebendig”.—Finally. “haucht Geist ein”. “Form”. so daß die aufs u h¨chste gespannte Erwartung und Forderung sich weit uberﬂ¨gelt sieht: bleibt doch o ¨ u der eigent¨mliche Geist. and “lebt die Musik in sich” refer to psychological reality. and “Haltung”. “Erﬁndung (F¨lla u u horn)”. signiﬁcates are addressed in “Leidenschaft”. (Observe that “Haltung”. “vor uns aussch¨ttet”. “Erwartung (der H¨rer)”. The semiotic system is instantiated when Rellstab talks about the signiﬁcant coordinate in expressions such as “Musikst¨ck”. “geistige Bedeutsamkeit”. “anreu gend”.2.5. The topographic cube oﬀers a local and recursive orientation. “Pers¨nlichkeit als Gef¨ß f¨r die u o a u Bedeutung”. a u 21 Rellstab addresses all three levels of reality. Geist “erzeugt” k¨nstlerische Leistung”. Words such as “Instrument”. for instance. “Geist wird den Formen eingehaucht”. simultaneously addresses physical reality and the semiotic signiﬁcant. “anregend”. TOPOGRAPHICAL NAVIGATION ist. Seine k¨nstu lerische Leistung wird zugleich eine Tatsche des Inneren. and “Kunstwerk”. anregendere und fesselndere Element. “Physiognomie”. u 2. The esthesic coordinate is addressed by allusions like “wecket Eindr¨cke”. and “Physiognomie” refer to physical reality. “Forderung”.

In other words. the performing artist is a creator for the auditory.22 CHAPTER 2. each knot being loaded by a localization within the cube. In particular. regarding levels of reality. the topographic cube yields a local conceptual orientation. • Third. the topographical description of a fact of music may open a complex tree of ramiﬁcations. in the communication chain. such as real numbers for parameter values. there is no consistent ontology without such a self-referential regression. TOPOGRAPHY • Also. but its description refers to mental instances. Supposedly.g. and by regression. and being communicated through the score. harmony syntactics) since a metalanguage precisely means recursiveness on the level of the signiﬁcate. but he/she includes an entire communication process. it is not necessary to introduce such a thing as a “topographic metacube” for the description of metamusical facts (e. . an acoustic sound is essentially a physical entity. starting from the composer.

Rather modern mathematics and logic is an issue of conceptual explicitness. We do not condivide such a credo for the following reasons. Further. –Σ– It is a common argument of traditional musicologists against cognitive and computational methods that the full extent of musical being cannot be grasped by these methods. depth and transcendence are not missed by computational and cognitive approaches. It is claimed that the very depth and transcendence of music are beyond any analytical and quantitative reasonment. cognitive and computational methods are neither restricted to quantiﬁcation. Such a discussion is substantial for a reconciliation of traditional musicology and innovative perspectives in cognitive and computational musicology. This chapter introduces the diﬃcult subject of musical ontology. it is not a question of specifying the music’s whatness as a Kantian “Ding an sich”. Hans Heinrich Eggebrecht [101. That is. nor to simple analysis. topology. 23 . a logical place. This provokes a crucial question of ontology: Can the problem of whatness be tackled by a diﬀerentiated discussion of the problem of whereness? At this point. Rather we can observe an increasing distribution of aspects of depth and transcendence among the conceptual topography. and geometric logic in the ﬁrst half of the twentieth century. Recall [417] that platonic ideas live in an ontological fundus. Above all. asking for the concept of music involves a topos in a particular ‘concept space’. simple numerical quantiﬁcation is a very special subject of old-style mathematics that has been overridden by structure theories since the development of set theory. In fact. p. modern algebra. semiotics of general sign systems has proven that formal explicitness on the levels of signiﬁcants and signiﬁcation processes does not prevent signiﬁcates from important added value of depth and transcendence. which is an explicit topographic site: Relative position and hierarchical location of its instances serve as basements for the whatness of ideas. but of how we conceive music: “What is the concept of music?” But a concept is—again in Kant’s words [258]—a topos. In other words. we should make more precise the ontological problem. the hyperouranios topos.192] Summary.Chapter 3 Musical Ontology Musik ist ohne Begriﬀe.

but science is not about comfort.1 Where is Music? Summary. nor is it exclusively in the composer. to mental compositional principles. cit. The diﬃcult localization of musical existence is due to three factors: (1) the interweaved usage of topographically distinct locations in the classical ontological discourse. but diﬃcult relation between thinking and making music. From the very beginning of European music theory in the Pythagorean tradition. playing an instrument (the monochord in those early days) was a method to gain evidence of the transcendent truth of music. Questioning this concept pairing without topographical diﬀerentiation (of reality coordinates in our case) yields a confusion of meanings and creates bunches of pseudoproblems (see loc. (2) the tendency of classical musicologists to override topographical complexity by all-embracing breviloquent approaches in order to uphold the discourse. it is necessary to distribute a concept over its topographical specializations. A most dramatic breviloquent repression strategy within traditional musicology was proposed by Hans Heinrich Eggebrecht on various occasions [134. We discuss and exemplify these factors. It relates to psychic categories.). (3) the genuinely related activities of thinking and playing—music is one of the best operationalized ﬁelds of human cognition. it is about truth. This procedure at ﬁrst sight destroys comfortable unity and simplicity. In our days. where to localize music ontologically. the concept pairing of consonance and dissonance is of complex reality (see Appendix B. after establishing a diﬀerentiated and adequate terminology. return to unity by conceptual constructions which include complexity rather than repress it. this is still a . and to physical reality of hearing and cortical sound processing. in the composer. 541]. Eggebrecht rightly observes that music is not exclusively in the work. This approach was not just a comfortable way to understand music. and by playing the monochord. Centered around the fundamental question. It cannot be discussed in detail. It was contended that a metaphysical tetractys symbol was the core of musical ontology. Whence he draws the erroneous conclusion that only a kind of abstract ego may comprise the totality of music. but it is for nothing. One may then. he concludes that music resides in the ”I”. 135]. etc. But it is evident that the foundation of musical ontology in a whatever abstract subject opens any issue of subjectivity and prevents scientiﬁc discourse whenever it suits corresponding needs. Historically. A third diﬃculty for locating musical ontology is a fundamental. a kind of super-breviloquent subject icon. how radical a declaration of scientiﬁc bankruptcy such a consequence does represent. –Σ– A closer look at musicological terminology reveals that many concepts are topographically blurred since their breviloquent function inhibits diﬀerentiation. MUSICAL ONTOLOGY It is therefore a legitimate procedure to open the ontological discourse by a view of the conceptual topography of music.3). For example. etc. it was the unique solution to gain understanding. A musicological discourse which is founded in the subject can go on forever. one can understand such a diﬃculty as a capitulation in view of exorbitant complexity. but systematically.24 CHAPTER 3. 3. This conclusion is deduced from a discussion of where music could be found elsewhere: in the work. we could access that transcendent being [330.

for example. On the other hand. In other words. just as it is an error to make physical experiments without quantum mechanical uncertainty and complementarity. we have already discussed this issue in example 1 of section 2. ineﬀable. To begin with. is the wrong strategy. the dominant part of traditional musicological language. this aspect is completely disjoint from the question of ineﬀability. to rephrase it with the words of Diana Raﬀman [432].2 Depth and Complexity Summary. It simply stresses the need for a communicative diﬀerentiation when dealing with music. 3. it is fundamentally so! This is. And it is wrong to believe that music is a special issue of science in that it deals with objects that point at non-mental strata of reality. neutral niveau. But we have to establish a sophisticated system of signs in order to grasp the signiﬁcates. But we may very well conceive them in a system of knowledge. the retrograde movement has a completely diﬀerent existentiality as a compositional (poietic) tool.” Not only is it admitted that a ‘thinking music by examples’ is useful. Thinking music is thinking in music. however in significantly diﬀerent ways. one of Raﬀman’s advanced arguments for ineﬀability of music. Summarizing. Sticking to common language. A famous defective argument which stems from the confusion of communicative coordinates is the judgment of contrapuntal symmetries in music. music is substantially tied to its communicative dimension: Unlike physics. in the line of Eggebrecht’s belief that music is without concepts. model their behavior with remarkable success for our cognition. deals with emotions. p. mathematicians speak of a “deep theorem” or a “deep . traditional musicology and mathematics. no. Psychology. this problematic stems from a poor localization on the semiotic axis. The latter is identiﬁed as a rhetoric chain of external reference pointers. The ramiﬁed discourse based on orientation via local topographic cubes is described as a tool for generating controlled complexity in understanding music. beyond words. It is not recognized that signs of non-verbal type—mathematical formalisms. For instance. of course. as the following proposition by Helga de la Motte [121. The claim that something is deﬁnitely ineﬀable presupposes knowledge of any possible language. than when discussed as a (esthesic) fact of perception. Forgetting about the making of music. –Σ– Both. for example—could eﬀectively resolve simple verbal ineﬀability. and esthesis. can be perfectly formulated in the mathematical language of vector ﬁelds.2. No non-trivial insight in modern physics could be obtained via common language—not to speak of modern mathematics. common language is not the tool for a musical concept space. music was always pronouncedly distributed over poiesis. The fundamental pointer character of this tree structure is compared to pretended depth in traditional approaches. boiling it down to an object independently of its making and perceiving is a basic error. use the word “depth”. we can state that most of the musical objects cannot be treated properly outside their topographically speciﬁed ontological modality. an absurdity.3. We shall see in part VIII on performance that performance nuances. thinking music by its direct invocation is ﬁrst of all the proof of a defective language. However. DEPTH AND COMPLEXITY 25 strong belief among musicologists. physics deals with elementary particles.232] shows: “Musikalisches Denken ist grunds¨tzlich Denken in Musik. All these objects share aspects that transcend human conceptualization.2.3. On the one hand. Music is not less and not more accessible than physics.

This capsule can—however—not be opened. As a basic rhetoric device. The discourse of important parts of that handbook relies on void pointers to never ending chains of implicit (!) external references. What. This is a prototypical deep statement since the reader is supposed to know that Eggebrecht is a scientiﬁc authority who has written a large number of books and papers on music.e. First. this kind of void pointer has been used in one of the most ambitious handbooks of traditional musicology [103]. At ﬁrst sight. information ﬂow breaks down. the citation is one of seven ﬁnal theses in a book entitled “Was ist Musik?” which Eggebrecht and the other German pope of traditionalist musicology.26 CHAPTER 3. his humble pupils and admirers have to accept that music is an unreachable secret and He. For example.4. in fact elementary. Second. p. but it is a very compact. He. MUSICAL ONTOLOGY theory” to indicate that it marks a result of a complex development. the musicological encapsulation sticks to a pointer to insight. Such a type of “depth” is completely standard in musicology. Semiotically speaking. has stated that “music is without concepts” (see the catchword heading this chapter [101. the capsule’s key has been lost. then.4 and ﬁgure ??) oﬀers an eﬀective tool of complexity which can help hammer out depth in the mathematical sense. You give an encapsulated statement and prevent any access to the (presumably) hidden complexity. the pupils. Carl Dahlhaus. We. the access to complexity is possible and its realization eventually yields insight. it simply states the verdict. and that this result gives a compact. it is pretended encapsulation without the license to access contents. You cannot obtain a follow-up. the topographic cube (see section 2. and veriﬁcation of the claim is not oﬀered on the base of explicit complexity. but alas! has to report that—ultimately— the real thing is beyond conceptual comprehension. does rightly encapsulate a deep insight. an authority of traditionalist German musicology. we have already 1 This theorem is one of the most famous and hardest results in the history of mathematics. Eggebrecht terminates by stigmatizing music with mysterious weirdness. In section 2. The Master. by the pronounced authority. Eggebrecht’s statement provokes a contradiction: How can somebody write hundreds of pages about music and simultaneously claim that conceptualization must fail? The solution is depth: We have to recognize that Eggebrecht has gone an incredibly long and audacious way in search of the ‘real thing’. the encapsulation remains locked. mathematical depth deals with simple signiﬁcants that are built from complex. Rather is the dark path to the hidden (pretended) rationale ornamented by metaphors. is the diﬀerence to depth in musicology? Strictly speaking. Let us make this clear in the following example. If applied with all its ramiﬁcations. statement. however. Fermat’s last theorem1 is a deep result since it involves an incredible complexity of mathematical structures. The text [101. its surface shows a simple appearance. without void encapsulation. In our case. pregnant insight into the addressed complexity. In fact. but rather in an encapsulated form. In this sense depth involves complexity. the pointer points into the void.192] does not give any hint.192]). no trace down to the very proof of the pretended conceptual failure is at hand. musicology is not doomed to void pointer depth. Fortunately. and this is mandatory for two reasons. It was proven by Andrew Wiles in 1994 and states that any non-trivial integer solution of the equation xn + y n = z n implies n = 1 or n = 2. thus turning depth into mystery—and science into fabulation. viz ramiﬁed and recursive signiﬁcation to connotation and/or metasystem levels. p. i. The Master has experienced the horriﬁc failure of concepts and we. Eggebrecht. Whereas in mathematics. have written to discuss the ‘ultimate and deepest’ questions about music. . should just accept and admire this report from hell.

but we should open it with respect to what is going on when understanding a performance of a work of art. The cube of topography is a local coordinate system for such perspectives. In this role. introduced the local character of ontological topography. But it does not mean that this recursion from one local perspective to the next tends to more elementary and eventually “atomic” ontology.1. the performing artist is on the esthesic coordinate with respect to the composer. In this description.1: The topographic cube of musical ontology has local character. it oﬀers orientation.and decapsulation processes on the communicative dimension.3. The result is a new type of work of mental and/or psychic reality which the listener will communicate internally or externally in a further processing of the musical experience. Etc. In this perspective—the inner dialog of the listener with his or her consciousness—the acoustical performance is again transformed into a mental structure and interpreted along the personal lines of musical insight. but no roots. see ﬁgure 3.2. the performer stays on the poietic side and produces a work of physical reality. but the reader can easily extend this ﬂip-ﬂop of perspectives on any other topographic . This complex identity of the performer is then encapsulated and taken as a new communication unity of the concrete performance. Example 2 On the string of communication. the listener is still encapsulated as an esthesic instance. only perspectives. Ontological atomism is neither desired nor even possible as a foundation of knowledge: Being has no elements. This latter is an expression of the artist’s reﬂection and will now transmit the message to the listeners of a concert hall. DEPTH AND COMPLEXITY 27 Figure 3. He or she receives the work and analyzes it as a mental entity. It means that the decision to attribute to a given situation speciﬁc topographic coordinates is no obstruction to other coordinates in a successive step of reﬁned analysis. A concept of music (represented by balls in this ﬁgure) may open diﬀerent localizations with successively reﬁned concept analysis. namely the sounding performance.– We have just sketched some en. say.

It is one of the main tasks of part II to develop a profound and detailed navigation formalism through music.28 CHAPTER 3. But a word of caution is imperative: Our task cannot be attained without a considerable amount of technical machinery—nihil ex nihilo. MUSICAL ONTOLOGY dimension. . but it is a clariﬁcation and gives ample orientation which void pointers deﬁnitely do refuse. It becomes clear now that opening ontological perspectives is by no means an approach to ontological roots. a formalism which provides encapsulation and reliable pointers to hidden complexity at the same time.

This is not only due to the fact that music belongs to humans and no longer appears as a revelation of divinities of whatever ﬂavor. fundamentally interactive understanding of human knowledge. We discuss the parallel to the epochal Galilean change from speculative to experimental natural sciences. it is necessary to review the overall epistemology of musicology since the preceding ontological topography has questioned traditional standpoints in a measure that does not allow of uncritical takeover of epistemological fundamentals. methods in musicology.Chapter 4 Models and Experiments in Musicology Plato’s pessimistic picture of empirical observation caused him to deny the validity of physical models and was largely responsible for the eclipse of empiricism for 2. eﬀective. Such a perspective is basic to all computational. and any lack of precise or explicit information is immediately blamed. it is the requirement of free access to (encapsuled) complexity that creates a boundless analytical attitude: It is no longer possible to ‘cultivate’ private. Peter Wegner [563] Summary. Above all. –Σ– In the ﬁnal chapter of this ﬁrst introductory part. The global knowledge space has become an ocean which is in permanent metamorphosis. It is also due to the massively improved arsenal of information and communication technology where the knowledge space becomes a concretely accessible site. inaccessible regions of knowledge. knowledge is constantly updated and extended. resp. This chapter introduces the paradigm of experimental humanities. This situation enforces a new. The paradigm of contemplative science which was essentially promoted by religious constraints [363. 29 . on which we navigate and experiment in a spirit of dynamical space-time [362].000 years. 477] has to be abandoned.

1: Galileo’s method concentrated on observation and experimental interaction with nature. as it was understood by medieval scientists such as Nicholas Oresme. In this preliminary discussion. In other words. Figure 4. Galileo’s concern was not just another speculative concept but resulted from intense observation of the movement of physical bodies on an inclined plane.30 CHAPTER 4. This detail of a fresco by Giuseppe Bezzuoli (Museo di Storia della Scienza. Measuring the body’s mechanical impact invokes a momentous property of that body.395]. Galileo was concerned with the deﬁnition of instantaneous velocity. Recall that. to have a closer look at the development of modern experimental science which prepared the Galilean revolution. MODELS AND EXPERIMENTS IN MUSICOLOGY We shall make these points more explicit in the following sections. however. changing discontinuously their constant velocity after a short duration. The case of Galileo Galilei is important for musicology since it is intimately related to a common problem: deﬁnition of musical tempo and physical velocity. Galileo’s approach was essentially built upon observation and measurement. could only view it as a succession of portions of uniform motion. for historical reasons. His proposals (dated around 1604) rather aimed at measuring (by impact) local velocity in a determined moment of time. He had to develop the very concept of instantaneous velocity since the traditional concept of uniformly accelerated motion. Florence) shows the famous experiment of a ball running down an inclined plane (right) and the traditional consultation of Aristotelian works (left). Though Galileo could . The turning point here is precisely the passage from speculative encapsulation to explicit accessibility by ‘doing’ science. he could not yet refer to calculus and deﬁne velocity as derivative ds/dt of space s(t) as a function of time t. it did not rely on abstract speculative reﬂections. we refer to Isabelle Stengers’ article [485. It is useful for the understanding of new paradigms in musicology. p.

it becomes clear that he was constantly dialoguing with his experiments. tables.m. The rectangular regions deﬁne regions of constant tempo (visualized here by reciprocal duration values). traditional musicologists deny the musical relevance of such a concept. musical tempo. according to the respective time window. and diagrams—and not with Aristotle’s authoritative writings—in order to construct a homogeneous concept space of physical movement. The opening and explication of these void pointers are left to musicians and programmers of music software. see ﬁgure 4. we encounter a concept of tempo which corresponds to Oresme’s setting: Tempo is the quotient of musical duration (measured in quarters) and physical duration (measured in minutes). shows an astonishingly parallel concept history—though nearly ﬁve hundred years delayed from physics.2: The ﬁgure shows a typical musicological ‘digitized’ version of the tempo concept. The interesting point in this procedure is a fundamentally operational method: thinking by doing. the author obtains diﬀerent averaged tempo regions.1. p. This parallels physics to musicology in a basic issue. and accelerandi. see ﬁgure 4. if one views Galileo’s twohundred-page notices written around 1604.1 for a detailed discussion of tempo). Many of the tempo-related concepts. as used by Hermann Gottschewski [110]. It is essential that this concept does not include instantaneous tempo in the sense of a derivative. Since no instantaneous tempo concept is given.31 not make this property explicit on mathematical terms. modern working . and theorists would not know how to give a workable description of such tempo phenomena. We have insisted on this episode since the musicological analog of velocity. Figure 4. This refutation stems from the completely speculative handling of tempo in traditional musicology. number lists. In fact.2. such as fermatas. he succeeded in discovering a physical one-to-one correspondence to instantaneous velocity. are encapsulated locked objects. ritardandi. Instead of learning from Galileo’s experimental observations theorists insist on a discrete tempo fantasm a la Oresme—despite the rich language of modern mathematics. in quarters per minute (see section 33. this yields the classical M¨lzel metronome a m.317]. In traditional European musicology [110. Precisely as Galileo.

Let us give two examples: In mathematics. According to a traditional opinion condivided by Immanuel Kant [258]. it is admitted that the Peano axioms for number theory1 are trivial creations of the human mind. consequently. MODELS AND EXPERIMENTS IN MUSICOLOGY musicians and engineers are at the cutting edge of explicitness and leave irrelevant speculations to feuilletonism where they are well-placed. i. (4) 0 is not a successor. (3) if a+ = b+ . (5) if a property φ for natural numbers is such that 1) φ(0) and 2) φ(a) implies φ(a+ ). only the ﬁttest concepts can survive the experimental struggle. Reasons for the categorical distinction between interior and exterior nature. The inner human nature is a vast ﬁeld of instances which we do not any better access than exterior objects.32 CHAPTER 4. have since been recognized as erroneous idealizations. Can human thought be an experimental ﬁeld? –Σ– Experimental sciences are classically related to empirical exterior nature. 2 Conjecture established by German mathematician Christian Goldbach. Just as medieval medicine was traditionally split into speculative academics (without license to practice!) and artisanal surgery (practised by charlatans). These two characteristics. this was evidenced by uncertainty and complementarity principles of quantum mechanics. Scientists such as physicist Galileo and the French surgean Ambroise Par´ had to ‘dissect’ the concepts and—hitherto tabooed—human bodies to explicate e complexity [463]. As with genetics. 1690-1764. there are no platonic ideas which we have to retrace from empirical reality. independently of the interrogator’s subject. then the property holds for all natural numbers. for every a. (2) every natural number a has a unique successor a+ among the natural numbers. Subjectivity is only a minor part of the inner nature.1 Interior and Exterior Nature Summary. But the impression vanishes immediately when abording such a simple question as the famous and still unsolved Goldbach conjecture2 : Can every even natural number be written as the sum of two prime numbers? This makes evident that the complexity of 1 (1) The number 0 (zero) is a natural number. no such a thing as objective facts. We learn from this double experimental revolution that it is not banausic artisanship that enforces opening of concept capsules.e. This is exactly Galileo’s method. Each response to an experiment may alter the theoretical position and the concepts which drive the experiments. But the argument of subjectivity is misleading. Galileo’s revolution was the answer to pretended depth of rhetoric discourse. 4. passivity and objective response. Nature is passive and will answer to the hearing in an objective way. experiments are understood as interrogations of a passive witness. then a = b. and to drive out at last void rhetorics. physics was split into mechanical engineering and speculative philosophy. the concepts are in incessant metamorphosis and mutate as a result of experimental cognition. It is deﬁnitely not true that we have a more indepth control of inner nature because of subjectivity. In what respect is then inner nature so fundamentally diﬀerent to exterior nature? It is said that inner human nature is subjective and that there is. . Physical experiments are substantially interactive processes. But physical nature is also responsive on a more conceptual level. We shall now take a closer look at the epistemological implications of this revolution. but the mere desire to redeem conceptual pointers.

4. Sch¨nberg’s treatise on harmony did rather stress than solve o those problems. or analysis.4. Candidates for experimental layers. As a piece of uncontrolled nature. performance. His or her contribution is but a germ of an incredible complexity that implies combinatorial processes. The Pythagorean tradition and the general concept of a music instrument. it has been saved by composers. . But it is an illusion to believe that the richness of such a creation can be controlled by the composer. WHAT IS A MUSICOLOGICAL EXPERIMENT? 33 our mental creations is not more under control than physical objects. the experimental scientists of music. we typically deal with the creation of a score. –Σ– We know from section 3. It was a kind of doing on the ground of prefabricated thoughts. It is not by chance that Boulez demanded to blow up all opera houses. what is the diﬀerence between exterior and interior nature? It relies on the more apparant. autonomy of exterior ‘natural’ phenomena. Giorgy Ligeti. the insistence on the aspect of activity in thinking music has survived all speculative assaults from tetractys to Keplerian harmony. However.2. What happens is that these forerunners are left alone by a speculative void-pointed musicology. a composition is quite the same as a piece of exterior nature. Bach’s “Art of the Fugue” is not fundamentally diﬀerent from a supernova out in interstellar space. playing the monochord was nothing more and nothing less than a passive veriﬁcation of already ﬁxed ideas. 3 This does not contradict his book on harmony [478] which is much less a warmed-up catechism than a critical review of failures of harmony. in fact the metaphysical tetractys. But this fails. Testing the congruence of a scientiﬁc model with corresponding material. such as with Pierre Boulez. or Iannis Xenakis. We believe that the Peano axioms are under control. But in their immediate conceptual vicinity. No longer can it be found under the auspices of divinities. interpretation strategies.1 that Pythagorean tradition oﬀers a type of “thinking by doing”. such as a composition. The so-called “emancipation of dissonance” is among others the emancipation of experimental musicology from speculative rhetorics. So. In other words. Nonetheless.2 What Is a Musicological Experiment? Summary. But a second view reveals an inner nature—especially on the level of man-made universes such as mathematics and music—which is not less complex and uncontrollable than exterior nature. But in their best performance. the part of subjectivity vanishes in front of an exorbitant structural variety. The shape in which we still encounter this method has of course changed. Already did Arnold Sch¨nberg quit the theoretical background in view of the speculative o overhead that could not cope with requirements of eﬀective conceptualization in harmony3 . In musical composition. though not fundamental. there are statements which do escape our mental power. they really appear as prophets of a future type of experimental musicologists. semiotic stratiﬁcation and so on. This is somewhat exaggerated since composers often condivide the status of medieval charlatans. Can we view this paradigm as an archaic version of Galileo’s experimental approach? We could if it provided us with a conceptual laboratory where concepts could be fabricated while making music.

Necessary and suﬃcient conditions for questions to be accessible on the level of the experimental setup. In other words. It is by no means clear that such a category can subsist within inner nature. and corresponding hypotheses and (text) theories as a background of orientation on the other.3 Questions—Experiments of the Mind Summary. Such a dialog can only succeed if the transformation channels between thinking and playing are explicit. painting. But such a generic restatement of the concept of an experiment has erased any speciﬁc reference to exterior nature. and a set of analytical tools living in the same concept or knowledge space. Measurement of physical events includes a transformation into data streams within a mental knowledge space. and under use of statistical orientation tools and theoretical hypotheses to be veriﬁed or falsiﬁed. literature or whatever can be subsumed under the general structure of a verbal or nonverbal semiotic ‘text’. and all kinds of symbols.e. the question underlying the experiment appears as a search mission towards a statement which ﬁts the theoretical orientation deﬁned by hypotheses and evaluation tools. But in the form of mutual inspection. But this is not what an experimental setup is dealing with. ﬁgures. Navigation on such a text crystallizes in form of a search for structural coherence between the variety of the text’s syntax and semantics on one side. Experience then boils down to a complex navigation process along this data stream. This will be dealt with in the next section. it should neither be a passive soniﬁcation of given speculations nor should it reduce to speculation without musical realization. What is accessible to scientiﬁc reﬂection is already a body of information in a well-structured concept space. Observe that the . So the decisive criterion is a bidirectional dialog between a ﬂexible conceptualization and an intelligent sonic realization. MODELS AND EXPERIMENTS IN MUSICOLOGY So what is a musicological experiment? It must clearly include both. Separating speculative questions from musicological experiments. the navigation being driven by two ‘dialog partners’: the data stream and the hypotheses and evaluation tools. When put together with the foregoing approach to the experimental setup. These requirements are necessary.34 CHAPTER 4. Navigation may thus be restated as interactive interpretation of a given human work. this yields a concept of a dialog between a set of data that are embedded in a concept space. We are still missing the adaptation of the crucial concept of experience to the context of humanities. an experiment may naturally occur in the context of humanities. We just deal with navigation on a data stream issued from a speciﬁc research object and a bunch of analytical orientation devices. –Σ– The naive view of experience in natural sciences evokes a bunch of facts in the space-time of exterior nature. but not suﬃcient to conceive such a thing as a musicological experiment. 4. the experiment turns out to be a search mission while navigating on a concept space. thinking and playing. This entire process is located on a (mental) knowledge space. already in its very data processing are the ‘naked facts’ codiﬁed in numbers. operational and highly eﬃcient. Improvement of scientiﬁc conceptualization by constraints from experimental options. reﬂection and action. be it music. Under this perspective. In this perspective. i. Experience in the humanities.

A theory may completely mutate under the inﬂuence of new directions in artistic creativity. if viewed as a question of the mind. Socially restated. Doing musicological experiments on a distributed laboratory. Galilei could not work with the medieval velocity chimaeras since they evidently provided no navigation through the attentive observation of balls running down the inclined plane. this demands the structure of collaboratories4 . NEW SCIENTIFIC PARADIGMS AND COLLABORATORIES 35 concept of empirical experience has been absorbed by the concept of navigation on a stream in a concept or knowledge space. an encapsulated locked concept of dominant. tonic and subdominant in function harmony cannot resolve any task of harmonic analysis for a concrete score since it asks for harmonic function values of any possible chord that may theoretically appear in the given score. the communicative network of collaborative science enforces a radical conceptual precision. We should stress that the experimental dialog is strongly interactive in that there is no priority of theories over ‘data streams’. and the very nature of dialog enforces data exchange on powerful channels. and time is replaced by an abstract ‘curve parameter’ of the navigation trajectory. as introduced by the US Department of Energy. void pointer concepts are immediately ruled out. On the level of institutions. the data stream cannot be navigated and the experiment breaks down. it is important to review the role of speculative void pointer concepts under the experimental paradigm. . not as an autistic reﬂection on locked concepts. For example. it is in constant renewal. Experiments are excellent exterminators of empty pointers: Their dialog immediately lays bare deﬁciencies hidden in locked concept pointers. Mutual inﬂuence of communicative networks and conceptual precision. Moreover. In the situation of an explicit dialog between given data and theoretical instances. The parameter of external physical time is no longer relevant. –Σ– From the preceding discussion it is straightforward that musicology (and humanities in general) in its experimental proﬁle cannot subsist in the private ambience of classical humanities. Redeﬁning scientiﬁc competence from the communicative point of view. not merely in noble and non-binding conference small talk. This being the case. see [363] for further information. rather as a dialoguing navigation.4 New Scientiﬁc Paradigms and Collaboratories Summary. And conversely. In cases where the function value(s) cannot be determined. The experiment. We insist on communication while doing research. such a type of science has to be realized in a strongly collaborative style. This is not only about the formal constraints when dealing with computer programs. navigation unfolds on an ‘abstract’ space. Collaboration is meant as a working style. not as a title for informal politeness.4.4. Just as physicists could not survive in isolated research units. musicologists have to initiate intense communication within research in order to succeed. it 4 The term “collaboratory” was coined by Bill Wulf in 1993 and is merged from “collaboration” and “laboratory”. can help us understand what happens when we build concepts and theories. the inner nature is no “nature morte”. 4. The experimental navigation dialog is demanding for all partners.

MODELS AND EXPERIMENTS IN MUSICOLOGY is really about communicating as a working method. scientiﬁc competence must then be rebuilt on the fundament of conceptual communicativeness while private regions of knowledge will lose their historically grown honorability. We should keep in mind that the entire navigation and orientation metaphor in the preceding discussion of the knowledge or concept space is not merely general knowledge science slang. Navigation happens to move along (local) coordinates of topographic cubes. . Evidently.36 CHAPTER 4. Learning by doing science can only succeed on the basis of unlimited access to encapsulated complexity. and we shall learn in part II that the general metaphor of a concept space can be realized in form of a veritable geometric space. but more concretely an adequate expression of the topographical (in fact: topological) setting which was developed in chapter 2.

Part II Navigation on Concept Spaces 37 .

.

Building concepts then amounts to conceiving this data. Both reductions cannot cope with what knowledge deals with. Music is an excellent ﬁeld to exemplify such knowledge spaces. including associated concept spaces. a coordinate system where we may place and retrieve substance. i. II. Knowledge has an object.e. Well-conceived information produces knowledge. pure information is substantial. but it is uncontrollable without a conceptual form. p. und dies selbst dann.Chapter 5 Navigation Verlassen wir das strenge Labyrinth der zementierten Begriﬀe. This enforces the development of concept-oriented access modalities to given information. viz building concepts of something. wenn wir reden. observe that knowledge involves two components: information and its mental organization. More precisely. 0 and 1. 1} of two substance values. –Σ– We initiate this part by a reﬂection on conceptual navigation since the concept spaces which will be deﬁned in the subsequent chapters are strongly motivated by universal orientation paradigms while surﬁng on encyclopedias of inhomogeneous music-related information and knowledge. and why therefore the digital paradigm is completely irrelevant to the yoga of electronic age: Digital substance. reduced to the BIT = {0. This is precisely why so-called digital information has nothing to do with knowledge whatsoever. and it reaches that object via its concept. the zero state of philosophy as exposed in the ﬁrst pages of [214]. Conceiving means being able to navigate in a conceptual coordinate system to attend that something. und ergehen wir uns ungezwungen in improvisierten Architekturen. It has become virtually impossible to navigate through music information without developing powerful concept formats. or OFF 39 .78] Summary. Pierre Boulez [60. organizing the access to information and doing this by use of well-deﬁned access modes. abstraction from information towards pure mental organization would mean getting stuck on Georg Wilhelm Friedrich Hegel’s germs of logic. Whereas boiling down knowledge to mere information would mean getting drowned in a sea of amorphous substance. In other words. and this amounts to having well-structured access modes. But conceiving means building concepts.

It is essential for a successful concept navigation methodology in music to realize a highly interactive and non-authoritative. It suggests an antagonist to the score-dominated way of making music. Boulez must have felt that “cemented concept labyrinths” no longer cope with the requirements of advanced musicology. as encoded knowledge. But it is much more and much more important than a way of making music. autonomous orientation and decision strategy. Refraining from cemented concept architectures means that the concept architecture has become dynamic. hard and insensitive to whatever. The problem is how to realize such a dynamic concept architecture without loosing rigor and reliability. Concrete is a rigid material. The digital age is not centered around “Bits and Bytes” but around their accessibility and handling.e. ever-changing and soft. NAVIGATION and ON. And from the preceding discussion regarding depth and complexity (section 3. i. be it on the creative. interactive and adapting to changing demands.40 CHAPTER 5. In no case is it any longer acceptable to be navigated by dogmatic magisterial guides. In fact. and this means developing a language for the open-ended navigation if ever this is the state of the art. i. Boulez’ vision is that of a ﬂexible concept environment. decoding information is needed in order to control and manage its knowledge potential.e. Now. the purely receptive view. We discuss the navigation problem on music data in relation to the general concept of an encyclopedic knowledge space. a digital record is worthless without being organized in a concept form.e. i. The key word is “ cemented”. Consequently. The upgrading is characterized by three changes: (1) the static cosmos is replaced by a dynamically developing data organism in space-time.. Like improvisation it should be interactive—this is to be retained. but it cannot per se be responsible for any kind of knowledge. For the time being. This is what the Boulez citation heading this chapter suggests: To create a conceptual environment that allows quasi-improvisational and free-will driven discourse on music. This latter idea turns out to be one of the deepest concerns of music in the making of our own world. In particular we should reconsider navigation as an interactive interpenetration of knowledge agents and data bases. is just a minimal substance set. We shall propose such an architecture in chapter 6.1 Music in the EncycloSpace Summary. the name does not matter. or be it on the reﬂexive side. the term “improvisational” in Boulez’ statement is somewhat misleading. the EncycloSpace. (2) the passive “speculum mundi”. i. Rather than ending in obscure concept mist conceptual navigation should lay bare semantical lacunae. and permission to navigate on any possible “concept path”. The latter is introduced as an upgrading of the classical concept of an encyclopedia.2) we deduce that navigation must have access to any encapsulated concept. rather than improvisational it should be termed dynamic. as conceived by the French encyclopedists Denis Diderot and Jean le Rond d’Alembert. (3) the textually oriented alphabetic ordering is generalized to a universal navigational orientation. let us concentrate on the generic navigation paradigm as suggested by Boulez and stressed in the ongoing transformation from energy to information society. From the above we learn that a powerful concept system for any ﬁeld of knowledge must provide us with a thorough navigation method that works on an extensive concept space. is replaced by an interactive instrumental relation to the data organism. ‘Understanding’. encyclopedic navigation splits .e. 5. in short: data navigation.

the Encyclop´die is recognized e as being a faithful realization of the medieval idea of a “speculum mundi”: “L’encyclop´die e assume par sa ﬁxit´ relative le rˆle d’un miroir du monde naturel. dass man sich eines Tages entsinnte.1: The principles of classical encyclopedia: completeness (complete circumference of the circle). is coupled to human knowledge production in an interactive and ontological way. Let us shortly recall that deﬁnition and then explain the concept’s genealogy and ingredients: Deﬁnition 2 EncycloSpace is the topological corpus of global human knowledge which evolves dynamically in a virtual space-time. the unity of philosophical synopsis. and allows of unrestricted navigation according to universal orientation within a hypermedially represented concept space. called EncycloSpace. and discoursivity (alphabetical discourse along the circle’s line). According to Sylvain Auroux’ analysis of Denis Diderot’s Encyclop´die [25]. ? Und so kam es.5. was given. In a critical review of these encyclopedic characteristics. wie es da stand für mmer so sein würde und dies ohne je etwas daran ändern zu können A. and the discoursivity of a mathematically inspired ordered representation (the alphabetic ordering of words). MUSIC IN THE ENCYCLOSPACE into a receptive and a productive variant. B. symbolized by the . a general deﬁnition of an encyclopedic knowledge space. see ﬁgure 5. –Σ– 41 In [363]. tel qu’il peu s’oﬀrir aux e o sujets connaissants.1.” This insight speciﬁes the attribute of completeness. the encyclopee dic principles of the Age of Enlightenment can be viewed as a combination of the completeness of a dictionary. C Figure 5. unity (the circle’s perfect shape). dass alles so.1.

add new knowledge and relativize old-fashioned approaches. Concepts are no eternal entities who live out in a platonic sky (hyperouranios topos). such a snapshot cannot claim to represent a world where knowledge is rapidly and incessantly growing. . They rather represent operational units with an ontology that is rightly deﬁned by the very accessibility of conceptual components. However. See ﬁgure 5. is not adequate. The development of knowledge is not an accumulation of essentially immutable and isolated time-slices. and subjected to laws of coupled synchronic and diachronic nature (see section 2.3.2: The EncycloSpace characteristics: Universal orientation in a dynamic knowledge space-time which is interactively coupled to human knowledge production. One could then replace the metaphor of a “speculum” by that of an “instrumentum”. Much like political and civilization dynamics or brute continental shift. Consequently. We have to deal with a dynamic universe of knowledge embedded in virtual space-time. We do not only look at preﬁxed things. the metaphor of a “speculum”. NAVIGATION orbit. the encyclopedic body is an open system which is incessantly reshaped in synergy with human knowledge production. the change of this dynamic system evidently does not happen by an autonomous activity: It is a result of a constant and substantial interaction of humanity with the corpus of knowledge. This aspect is substantial since it also questions the very nature of knowledge. similar to the photography of a stationary system. we do also deﬁne and redeﬁne them. This is a traditional static world view.2 for the following discussion. Instrument Virtual Space-Time Orientation Figure 5. as a ﬁxed cosmos. This latter is perfectly adequate to the computer as a bidirectionally active interface between humans and knowledge bases.e.4 for these semiotic concepts). i. Further. We have to accept that the purely spatial coordinates of a virtual cosmos (in Diderot’s words: “connaissances ´parses sur la surface de la terre”) must e be completed by the time coordinate. a passive vision of given things.42 CHAPTER 5.

) when asked to prove your understanding (to yourself or to others.” This killer sentence which terminates Ludwig Wittgenstein’s “tractatus logico-philosophicus” [580] collapses in the age of hypermedia discourse. MUSIC IN THE ENCYCLOSPACE 43 (In a more radical perspective. To be clear. Summarizing. navigation on the EncycloSpace must regain the original activity. Though “navigation” stems from Latin “ navigare”. since EncycloSpace has been determined as an interactively handled body of knowledge. such as pictures. Finally. the brute reduction to the deﬁnitely textual binary code encompasses virtually every possible information record. More in-depth. Such more general data types emerge in a natural way in hypermedia documents including visual and acoustic instances. etc. ontology of concepts is essentially topology: a study of the topoi where concepts subsist.1. acoustic/musical. Rather do we envisage the in-depth question of the structure of the space where concepts live. and determine the concept as explained in the above deﬁnition 2. a body which is not built. that the spatial metaphor for the ontology of concepts is crucial for any eﬀective discussion of the system of concepts. boils down to the way you activate its components (roof. far from being an ordering in the technical sense. This is also due to the fact that the concepts are not cast in a formal language and hence. Therefore. we have reviewed the concept of Diderot’s Encyclop´die and updated its e three attributes to meet the needs of a dynamic. But it is a diﬀerent business to understand information—this is deﬁnitely not in the reach of the binary code. We know from Plato’s allegory of the cave [417]. for example. the original . it is the geographic orientation given by the Encyclop´die’s cross-reference e system [25. sounds. the latter is rather a partial relation of “pointers”.321 f]. it is the alphabetic ordering dictated by the traditional textual representation. This point is not just a marginal note concerning formal aspects of knowledge. for example. and universally oriented knowledge society. It should be replaced by a recommendation of visual/geometric. In this theoretical sense. “Wor¨ber man nicht sprechen kann. alphabetic navigation is not suﬃcient. altered and developed by mysterious forces but by our genuine agitation upon the object. walls. conceptual navigation must provide tools and paradigms which transcend the textual alphabetism and include generic principles of answering to primordial discursive questions: Where do I come from? Where am I? In what direction do I proceed? From a more operative and practical point of view. diagrams. the order of representation of knowledge within the Encyclop´die is twofold: On e the very surface.5. beyond the alphabetic ordering no intrinsic organization is visible on the representational level. there is no diﬀerence). Navigation orderings have to cope with hypermedia orientation paradigms for practical and for theoretical reasons: There is no reason why the completely arbitrary textual reduction should and could grasp the intrinsic ordering of concepts and thoughts. dar¨ber u u muss man schweigen. such as geometric spaces. p. the present meaning of “navigation” is restricted to the passive steering of a vehicle which is already moved by some motor.) In this vein. or haptic/gestural alternatives to textual dead ends. or spaces of set collections (see chapter 6 for precise deﬁnitions). Whereas the former is strictly linear. understanding the concept of “house”. it could be argued that concepts are operators. windows. However. In this sense. Thus. which means giving motion to a ship. the alphabetic ordering inherent in the linguistic text paradigm must be completed by orderings which are genuinely related to non-textual data formats. music or dance scores. and this one from “navis” plus “agere”. from Aristotle’s “Topic” [20] and from Immanuel Kant’s comment on Aristotle’s ”Topic” in “Kritik der reinen Vernunft” [258]. we may baptize this update by the name of EncycloSpace. active.

. The main requirement in this situation is an optimal orientation environment. and generically representable on the visual interface. But this does not cover less trivial search problems. Given such a universal ordering. if we are looking for all chords in the score of Schumann’s “Tr¨umerei”. For instance. It deals with universal orderings to ﬁnd and represent knowledge for optimized understanding. onset. Such a space requires diﬀerent approaches to ordering—and chords do so a fortiori. They should a then be able to order these chords according to universal ordering principles since we cannot (1) redeﬁne from scratch orderings among EncycloSpace objects and (2) build individual orientation frames for each particular conﬁguration. and loudness. –Σ– Receptive navigation is the common situation in classical encyclopedias. Whatever are the searched for objects. the alphabetic ordering is adequate: We just leaf through the linear alphabetic ordering of words.—This will be an important feature of the denotator system to be introduced in chapter 6. In case of a simple name search. say. We would typically start on the level of composer names. At this point. Once arrived on the level of notes. Universal orderings among EncycloSpace objects is a central feature for receptive navigation because it yields orientation and thusly prepares the ﬁeld for recruitment of added knowledge which is to be created from productive navigation. every chord could be retrieved according to corresponding intuitive orientation paradigms. it requires ordering principles for chords as special types of note sets. duration. representing ordered lists of chords requires non-alphabetic ordering principles.44 CHAPTER 5. piano notes are points in a space of four dimensions. and ﬁnally to “Tr¨umerei”. productive navigation does change the body of knowledge. The EncycloSpace features should comprise a representation of all chords built from the 463 notes of the “Tr¨umerei”. universal navigation concepts are essential in innovative musicology. One is looking for a bunch of information in a more or less well-determined. then—once “Robert Schumann” has been found—descend to piano works. we will have a to surf through diﬀerent types of musical score signs. Receptive navigation means moving around within the body of knowledge without changing it. We recognize that inspecting such an ordered chord list is only the last link of an extended chain of encylopedic objects which ﬁnally leads to the required chords. Since musical data are of very inhomogeneous and general types. In contrast.2 Receptive Navigation Summary. We will discuss these two modalities in the following sections. NAVIGATION etymology of “navigation” is restituted: We shall distinguish between receptive and productive navigation. then to “Kinderszenen”. it turns out to be crucial for valid receptive navigation to be provided with universal ordering principles: orientation must be generic. Surﬁng through the set of chords in a determined score is a new situation. In fact. pitch. 5. and immutable ﬁeld of knowledge. Receptive navigation leaves the EncycloSpace unaltered. the alphabetic orientation a crashes.

Besides the technical problem of how to realize such a functionality without losing control. it is not probable that all such lists of chords drawn from scores of the piano music repertoire are incorporated in the EncycloSpace. but it diﬀers from receptive navigation: Navigation which is deﬁned by the question and terminates upon its answer produces a (tiny) extension of the EncycloSpace. which (3) are dominant in F a major. On a less trivial level. PRODUCTIVE NAVIGATION 45 5. Suppose that such a concept is named “Non-Standard-Motif”. Boulez’ idea of an improvisational concept navigation provokes fundamental questions of communication which are well known from music when improvisation started to penetrate the music market and musicology as a science. such as the number of (1) chords in the “Tr¨umerei” which (2) contain more than three notes. we supposed the existence of such a list on the EncycloSpace. and incessantly producing new data to be added to the given EncycloSpace. such as a special type of musical motives which meet non-standard conditions. More probably most of what is required by an interested user will not be allocated. On a still more invasive level. Productive navigation is a central issue in extending musicological and musical knowledge. such an extended functionality adds to the original repertory character of the EncycloSpace the character of a laboratory where questions are transformed into mental experiments as introduced in chapter 4. the calculation of complex data. –Σ– In the above example of an ordered list of chords. We do not want to be excluded from navigation because this concept is not yet part of the EncycloSpace’s vocabulary! In other words. The EncycloSpace is a repertoire as well as a laboratory. we are also confronted with the level of accessibility of extended EncycloSpace contents by third parties. can require too much calculation time to allow forgetting about the result. and which (4) contain a diminished third. we see that productive navigation responds to the demand for dynamic navigation.g. the mere possibility to access ad inﬁnitum an improvised piece of music initiated a discourse about music which was never . most jazz productions essentially rely on sound tracks which override pure score or songbook data. Summarizing. In fact. it can be a numerical information we want to obtain.: “How many quarter notes are contained in Thelonious Monk’s original tune “Blue Monk”?” This information usually is not part of the knowledge base.3 Productive Navigation Summary. but it can easily be calculated. we try to save the result and access it as soon as new navigation trips ask for this information. it may turn out that we have built new concepts. Searching for the answer is a trivial extension of knowledge.5. Boulez’ requirement of an “improvisational” discourse/navigation must include free vocabulary extension and then the operation upon the extended vocabulary. In this case. interacting with the body of knowledge. In the simplest level. However. The introduction of sound ﬁles as a major trace of and reference to the musical creativity is responsible for the very existence of the present jazz culture. This can be the case on very diﬀerent levels of knowledge. Evidently. Productive navigation interacts with the EncycloSpace such that the latter is enriched by a surplus value of added knowledge. Though technologically low level. But this is not a question of fundamental ordering.3. e. rather technology should be concerned with the social aspect of navigation-induced knowledge extension.

‘Improvisation’ within knowledge can now be traced and integrated in the global stream of knowledge production. hard-coded knowledge data bases in books and encyclopedias. we experience the same eﬀect: Classical ‘scores of knowledge’. . We would like to insist on this background throughout the present book since it exerts an incredible impact on how we recognize important from irrelevant aspects of formal conceptualization in future musicology.e.46 CHAPTER 5. are being confronted with dynamic records of ever moving knowledge streams. No need to stress the fundamental transformation of knowledge sociology under such perspectives. On the level of conceptual navigation. NAVIGATION possible before. And it enforced the relativization of classical scores as an identiﬁcation of the musical work. i.

Their semiotic structure represents an elaborate denotative baseline and resides in a purely mental level usually attributed to mathematical objects. such as to include attributes with character strings for names.Chapter 6 Denotators C’est peut-ˆtre dans la fa¸on de se e c repr´senter l’esprit qu’on pourrait d´chiﬀrer l’esprit. In particular it was not possible to deﬁne concepts by recursion to already given concepts. III] e Summary. not include deeper semantic layers which are related to physical. Let us instead describe the genealogy of the denotator concept. They were centered around a special type of mathematical structures (essentially ﬁnite point sets in particular parameter spaces) which could not comprise less theoretical objects. This framework was built upon so-called local and global compositions. e e Paul Val´ry [538. 2. However. it was believed that the concept framework of mathematical music theory was suﬃciently generic to handle formal issues in computational systematic musicology. Denotators generalize the structures of local compositions in mathematical music theory as well as the data model used in the RUBATO software. Therefore. –Σ– This chapter introduces the formal framework of the entire book. They do. the theoretical and program design groundwork for the analysis and performance platform RUBATO taught the developers that these concepts had several severe drawbacks: 1. 47 . social or religious meaning. and in connection with the development of the composition software presto [338]. Also were they ‘hard-coded’ in the sense that introduction of new concept types was not possible. psychic. This chapter introduces the universal data format of denotators to describe musical objects. After the publication of the German precursor [340] of this volume. We believe that the preceding chapters have given a huge motivation to elaborate a formally rigorous concept framework. however. or other non-numerical data. we shall no longer comment on the general purpose of this issue.

and then extending to the universally valid generic setup. psychic. and open to unrestricted modular data recombination. at least regarding to ‘real’ meaning of physical. completeness. Denotators are conceived according to the general criteria of EncycloSpace navigation. The reader should postpone semantic issues to chapter 18. The so-called functorial point of view was not taken seriously enough. social or religious character. –Σ– From the discussion of encyclopedic characteristics: unity. 5.48 CHAPTER 6. the present discussion is uniquely concerned with the universal denotation formalism. of extensive ramiﬁcation. Discussion of three principles of universal data format typology: unity. These are realized by a typology which is recursive. The mathematical ﬂavor was not compatible with programming requirements for data bases. The denotator concept is modeled following the Aristotelian pairing of substance and form or of the geometric paradigm of point and space.1). starting with a more down-to-earth version of the denotator framework (section 6. denotators are radically pre-semantic. and meets the requirements of productive navigation. But it will also require a rather complex concept scheme. we deduce three denotator attributes which cope with these characteristics. The denotator concept framework will eliminate these defects. 6. This hindered recombination of object selections—which now is possible. and discursivity. We discuss the fundamental pointer nature of points as basic concepts of axiomatized semiotics. Intuitively. Also we should take care of abstraction from any deeper semantic loading with denotators.1 Universal Concept Formats Summary. DENOTATORS 3. respectively. encyclopedic characteristic unity completeness discoursivity =⇒ =⇒ =⇒ =⇒ denotator attribute recursive typology extensive ramiﬁcation modes order and recombination . completeness. The naming policy did not allow clear distinction between objects and names (in a strictly semiotic sense). Notwithstanding their formal power. so observe that we do not follow mathematical tradition but navigation methodology. The objects were not well distinguished from the spaces they live in. including ‘mathematical navigation’ as a special case. We shall develop the details in several steps. 6. and discoursivity. 4. Each of these components: substance and form are arranged according to the structure of signs. this means that we had to stick to one ﬁxed ‘ontological’ point of view.

This ﬂow chart visualizes the corresponding denotator attributes. Construction of new and recombination of given denotators is non-restrictive. discoursivity is met by means of free construction and recombination of denotators and by universal orderings among these objects. Unlike with usual data base management systems [551] there is no ﬁxed ensemble of possible denotator constructs. Figure 6. .6.1.1 illustrates the denotator ‘ﬂow chart’.1: Denotators are designed to cope with universal encyclopedic characteristics. Finally. and discoursivity. Each ramiﬁcation mode ensures a particular way of stepping downwards in recursivity. such as unity. type ramification mode ramification knot types given by recursion type type type Figure 6. A type refers to other types in a recursive way through diﬀerent ramiﬁcation modes as indicated in the ramiﬁcation knot. One builds new denotators from old ones by systematic reuse. The system of denotators can be ordered in a universal way. completeness. Completeness must be guaranteed by extensive ramiﬁcation modalities when stepping down along recursive paths. UNIVERSAL CONCEPT FORMATS 49 The idea is that unity in creating concepts is achieved by means of recursive constructions.

5 These simple domains will be vastly extended. see Figure 6. consisting of Form-Name.5.′quotation marks reduce the word to a character string.2: Like a real thing.3. Later. together with a proper formspace. such as ′Note′2 . (6.. Fn of forms. By deﬁnition. the sequence of characters without any further meaning. but no fundamentally diﬀerent ideas will be added. DENOTATORS 6. the type is either compound or simple. based upon this concept.1 First Naive Approach To Denotators Summary. For products and coproducts..e. FLOAT.. . powerset {}. synonymy is superﬂuous and may be mimicked by a product with one single factor.1) The Form-Name is any character string. Compound types are ramiﬁcation modes. Rather is it necessary that each point carries its space with it like a snail carries its shell.. the present selection is made with regard to practical and programming-oriented usage. simple types are recursive start points. By deﬁnition. the substance-points are deﬁned. denoted in this way: F orm-N ame → T ype(Coordinator). four5 simple types are possible: STRING. INTEGER. the coordinator is a ﬁnite sequence F1 . the coordinator is one form F . a denotator is a substance-point. the naive approach is related to the topos Sets of sets. The naive deﬁnition1 of a denotator is split into a recursive deﬁnition of the form-space.2. we shall generalize the types. coproduct .. Their coordinators look as follows: 1 In the mathematically rigorous framework of form semiotics as described in appendix G. BOOLE. and Coordinator.. 3 These are just the ones which can be easily understood within this naive introduction.50 CHAPTER 6. + substance-point form-space = real thing Figure 6. Type. and synonymy Syn. The recursive structure of a form is a triple. There are four3 compound types: Product . –Σ– The universality of our claim implies that we cannot just set up a generic space and take the objects as being “points” in such a space. we view a denotator as being a substance-point within its form-space. 4 In the rigorous framework. A ﬁrst approach to denotators is given without requiring further mathematical background.1. For powerset and synonymy4 . i. and then. 2 The special ′. Referring to Aristotle’s description of real things as a combination of substance and form.

. products and coproducts or powersets and synonymies. the speciﬁc diﬀerence will emerge when we introduce the substance-points on the respective form-spaces. see appendix G. in this naive setup. Duration) (6. Duration). if we have the product form ′Piano-Note′ → (Onset. 8 In the general approach to denotator theory. With this naming convention in mind we introduce the following simple forms as a ﬁrst set of ad hoc examples for this ‘naive’ introduction: • Onset→FLOAT. respectively.5. 7 We do not consider inﬁnite decimal digits.. 51 • FLOAT: The set R of (ﬂoating point) decimal numbers7 .. Since coordinators are uniquely determined for simple forms.. the latter one a single form as coordinator. then we want to give it the shorthand name Piano-Note. ±3. etc. names are also denotators. b. Pitch. The general naming policy for forms is that two diﬀerent forms should bear diﬀerent names. (6. ±1. such as letters a. A rigorous expression for such a form would read ′Onset′→ SIMPLE(R).2) with coordinators Onset.1. ﬁgures 0. . and other characters. .. etc. The former couple has a sequence of forms. INTEGER(Z) simpliﬁes to INTEGER.0157. • INTEGER: The set Z = {0. Observe that presently. and the form itself would be called Onset. i. Loudness. 1}. UNIVERSAL CONCEPT FORMATS • STRING: The set CHR of character strings6 . .3. • Loudness→STRING. are not further distinguished. 1. 2. Pitch. 6 We may take the standard set of ASCII characters. such as 234. and Duration. Loudness. However.. we shall omit the coordinator bracket in these cases.1415926 . ±2. for example. such as π = 3. but from an informal point of view.3) if no confusion is likely8 .6. as illustrated for the Piano-Note form in Figure 6. • Duration→FLOAT.. In this sense. If we deal with complex compound forms and wish to represent them in a more graphical way. • Pitch→INTEGER... −25.} of integers. Pitch. the symbolism can be shown in a vertical arrangement. such as brackets. This enables us to identify a form with its name (the name is a “key” in the terminology of database theory [551]). . Loudness. and the present distinction is obsolete. just think of usual symbols. . etc.3. • BOOLE: The set BIT = {0.e. . we often adopt the bracketless denotation Piano-Note → (Onset.19022879. For example.

writing C or C Pitch(60) instead of ′C′ Pitch(60). though not very common usage of the ellipsis symbol “. .52 CHAPTER 6. denotator names are of secondary importance. by deﬁnition.e. ”: it means that one has started with a sequence of symbol combinations which follows an evident law. . This is so because denotators usually occur in large numbers (such as notes in a score). Fn . fn of denotators with forms F1 . sometimes also denoted by ! 10 In this naive setup. . In our naive setting. if Type = INTEGER then Coordinates is an integer number. including the recursive explication of its coordinator forms Onset.4 for concrete examples of simple denotators. The evidence is built upon the starting unit. For example. or a1 + a2 + . a denotator is a triple. . . i. for example. and Coordinates is an object which is determined according to the form’s type. . (6. we deﬁne Coordinates as being any sequence f1 . a denotator for pitch middle C on a keyboard could look like ′C′ Pitch(60). . respectively.4) Again. but there. Loudness. . . we need conceptual circularity on the level of forms and denotator namings. Coordinates is an element of the coordinator of the form. i. For denotators we cannot however stick to the naming policy of forms since in general. Very often it will even be reasonable to reduce the denotator name to the empty string ′′9 . Form is a form. an . an element of Z. if there is no risk of confusion. and then . This is however diﬀerent from the formal setup.e. such as “1. Again. Let Type be compound10 . . For example.3: The form Piano-Note in graphical representation. For product. . n. and if11 Coordinator = F1 . 2. it is denoted by Denotator-N ame F orm(Coordinates). . . . We have preferred this naive approach to ease understanding in a ﬁrst approach. and the coordinates Coordinates. . consisting of the denotator name DenotatorName. we shall identify a denotator by its denotator name. DENOTATORS coordinator forms Piano-Note P form name type optional Onset FLOAT — Pitch INTEGER Ÿ Loudness STRING CHR Duration FLOAT — Figure 6. such as 1. and we are forced to distinguish them by their coordinates rather than by names. f1 9 Be careful in distinguishing the empty string ′′ from the string ′ ′ consisting of one empty space. see Figure 6. the form Form. coordinates are also denotators. 11 In this book.” or “a1 +” in our examples. and Duration. Fn . Pitch. we make a logical. Denotator-Name is a character string. If the type is simple then. .

. “4. (6. which is nonsense. but exceptions may occur. for n = 3.. 12 If possible.. l. it is a meta-sign referring to the inductive oﬀset..5 shows the graphic representation of the Piano-Note form (see ﬁgure 6. . If Type is coproduct. D). . For example. i. L Loudness(l). + an is not correct. 2. fn = fn -Name Fn (fn -Coordinates). UNIVERSAL CONCEPT FORMATS 53 "L"~>Loudness(mf) X X F BOOLE "HiHat-Open"~>HiHat-State(1) STRING FLOAT INTEGER b 4 j j &b b 4 X X X X X X bb 4 & b4 X X X X X "E"~>Onset(0.4: Examples of simple denotators. The ellipsis means that the building law is repeated. and as such. and with form Piano-Note. Z. 2. D Duration(d). H. such as “2. . we name denotators with small initials. Duration and Coordinates e. for an index i = 1. .. such as “n” or “an ” in our examples.5) Figure 6.” or “a2 +”. n. h.” or “a3 +”. E Onset(e). = f1 -Name F1 (f1 -Coordinates).. D with forms Onset.. Loudness. L. fi = fi -N ame Fi (fi -Coordinates). n.. it would imply a notation such as 1. 3 or a1 + a2 + +a3 . H Pitch(h). in complicated indexing situation. or a1 + a2 + . . CHR. R respectively: myNote Piano-Note(E. such as “3.6) the following unit. we deﬁne Coordinates as being any one denotator fi of form Fi . . . the coordinates are four denotators E. L. . (6.1. . Moreover.6. if we are given a denotator named ′myNote′12 .” or “a4 +”. and if Coordinator = F1 .3) with the above denotator myNote inserted. the common notation would be overloaded.e. H. Pitch. . . In the limit. d in the coordinators R. Fn . and then inducing the following units to be denoted. etc. Therefore the more common notation 1. .625) "H"~>Pitch(70) b 4 j &b b 4 X X bb 4 & b4 X Figure 6. until the sequence is terminated by the last symbol.

Hence. Then. fk }. let P ianoSelector (Onset. we usually omit the curled set brackets in this notation and simply write {}(f1 . a chord in the classical terminology. DENOTATORS myNote ~> coordinates P types E coordinates ~> Onset H ~> Pitch L ~> Loudness D ~> Duration FLOAT INTEGER STRING FLOAT coordinators e ~> — h ~> Ÿ l ~> CHR d ~> — Figure 6.A). then a denotator with denotator name ′myNewName′ is myNewName New(f ).. F4 = Onset. P itch. . F3 .. . a denotator named ′mySelection′.. then a denotator myChord Chord(S) is deﬁned by a set S of pitches. fk ) instead of {}({f1 . all having one and the same form F . with this form.e. Loudness. and A Pitch(69) would be denoted by IIC {}(D. . and if Coordinator = F . is deﬁned by the selection of either f Onset(φ) or f Pitch(φ) or f Loudness(φ) or f Duration(φ). R.. F Pitch(65). Duration) be the corresponding coproduct form. if the denotator’s form is New→Syn(F ). Essentially.. F2 . i. respectively. Z.. and if Coordinator = F ... If Type is powerset. Pitch. In the above example with F1 . . this means renaming f 13 We shall.F. For example. For a ﬁnite set S = {f1 . a second degree chord13 IIC in C major consisting of pitch denotators D Pitch(62). CHR. For example. fk }). φ being an element of R. Duration.. if we consider the form Chord→{}(Pitch). If Type is synonymy. Loudness. of course. we deﬁne Coordinates as being a denotator f with form F . we deﬁne Coordinates as being a ﬁnite set S of denotators. given by ′mySelection′ PianoSelector(f ).54 Piano-Note CHAPTER 6.5: A product denotator named ′myNote′ with its coordinates in a recursively complete graphic representation. give rigorous deﬁnitions of all musicological terms later in this book.

Notice. nothing really happens—unfortunately.7) instead of ′′ AF orm(myCoordinates). etc. only the reference type is explicit. the form of a denotator if this speciﬁcation suﬃces. The underlying form concept is self-referential: In general. viz mathematics and computer programming data types (numbers. walls.1.6. e. We point at components of the concept. or. This is essential for understanding the construction history of concepts. if empty. But we do not point automatically at any possible knot of the entire concept tree! If the concept “house” appears in a discourse. “house”: We say that a house consists of a roof. strings). Similarly with denotators. Understanding such a concept means navigating on its recursive ramiﬁcation tree. this is a very common situation in the humanities. we penetrate such components. Circularity.1. This is the way we usually execute our understanding of a concept. to encapsulation of object data. we only delve into a minimal level of explicitness. this means that we just take the name. and this is a fundamental extension of concept building rules with respect to the 14 Usually a pointer to some memory address. UNIVERSAL CONCEPT FORMATS 55 by ′myNewName′ and doing this within the form which is essentially F . Besides a renaming of the space and the point. You only “unpack” what is really needed. This approach is similar to the paradigm of object-oriented programming. more precisely. 6. the new denotator is recursively later than the old one.g. Recursion is supposed to stop at a simple form.e. etc. Interpretations of the above concepts and comments on non-trivial implications are given. They have an identity14 . but the contents are not unveiled until really needed. To begin with. i. digits. . that synonymy is not symmetric. but with a new name ′New′. Only in case some request is made for more details do we unveil deeper concept levels and follow the path down to coordinators and coordinates.2 Interpretations and Comments Summary. –Σ– Pointer Character. a set of windows and doors. If necessary. But we shall not introduce analogous notation for empty-named forms since form names are essential as an identiﬁcation instance. If a denotator of form AForm and coordinates myCoordinates is given the empty name ′′. We contend that this scheme of recursive pointers is fundamental in human concept construction. let us stress the pointer character of the denotator concept. and prosecute their ramiﬁcation trees. we shall denote it by AF orm(myCoordinates) (6. we are left with a selection of substance from a common repertoire. the rest remains either referred by name or even completely hidden. Such a pointer scheme realizes a very economic handling of concepts. and there. however. In our setup. Denotators have the outstanding property that they are open to circular deﬁnitions. a form is deﬁned by another (bunch of) form(s).

this problem is solved in the following way: The basic form is termed ′book′.. subchapters. Nonetheless. and the denotator is completely determined. In the context of denotators. title 3. Evidently. Meeting Encyclopedism. this is the yoga of pointers. This may happen anywhere in the recursive regression of such a denotator and therefore. We want to deﬁne the concept of a book with chapters. number 2. and the depth of sectioning and subsectioning a book is not limited. Recursive typology meets unity since we are given a uniﬁed construction principle of concepts.′ → IN T EGER... title. We come back to this issue in section 6.56 CHAPTER 6. text 4. a set of chapters The problem with this deﬁnition is that we have to declare the “set of chapters”. chapters). text. Let us therefore complete the above list by a number: 1.7. text. It is. We have made self-reference to the form book within the powerset factor chapters of the four-fold product. ′N o. we may give it the empty set ∅ of coordinates: myChapters chapters(). and a set of chapters. We thereof deduce a natural ordering of appearance of the particular chapters. this circularity of the form book is not necessarily inherited by substance-points within the form-space. each chapter again consists of the same data type. But we can say that a book always consists of title. circularity of forms and denotators may occur in a much less inoﬀensive way which takes the very basis of concept construction into task. except that we have to enumerate the chapters. subsubchapters etc. and it has this structure: ′book′ → (N o. However.8) has any set of chapters. Since a book-formed denotator myBook book(. ′title′ → ST RIN G. Let us recapitulate the encyclopedic characteristics and their realization on the denotator level. ′text′ → ST RIN G. In this case. Let us give a common example to make the point explicit. no further speciﬁcation is needed. myChapters) (6. not .. a book can have an arbitrary number of chapters. any ﬁnite book tree can be captured in the sense we expect from practice. of course. DENOTATORS apparently founded recursivity of denotator structures. ′chapters′ → {}(book). In turn.

And discoursivity is guaranteed by two measures: First. But in view of the rigorous formal setup to be exposed in section 6. we only need to replace them by synonymous denotators which point at numerical pitch values. Even operations of semantic completion are feasible in the following sense. This is precisely the need to replace or enrich the string ′F ′ by more involved information. e. no further contents are available. (6. This provides us with an ‘omnipresent’ orientation for the discourse on denotators. say—this ‘symbolic’ data should be realized in a speciﬁc tuning. It may happen in a more in-depth research development that the pitch structure can be speciﬁed beyond strings. coproducts generalize to colimits. Suppose. (6. Pitch(66)) as related to keyboard middle C Pitch(60). for example. see section 6. The ﬁrst form relates every denotator pitch string to its numerical value. The second measure to meet discoursivity is a universally deﬁned ordering principle on denotators. (U nknownP itchT ype. but no special value in terms of frequency is speciﬁed. the old form UnknownPitchType can be replaced by the new KnownPitchType form in all forms which make use of it—just as in genetic engineering.8 for the rigorous version) this topic in the following section.2. it may happen that a pitch is just named ′F ′. complete freedom of form and denotator construction guarantees recombination. We shall introduce (still naively. whereas the second form is a restatement of pitch data via a synonym of the already given Pitch form. Then the form UnknownPitchType may be replaced by KnownP itchT ype → or even by form KnownP itchT ype → Syn(Pitch). Whatever procedure is selected. although the latter is being used in theoretical computer science. For example. 16 It 15 Products . that an ethnological pitch-related form has not yet been made fully explicit and is merely ‘sketched’ in a simple. is astonishing that database theorists have not yet learned to make systematic use of mathematical category theory. together with the epistemological factotum of synonymy. And UnknownPitchTypeformed denotators can keep their names.10) where Pitch is the integer-valued form discussed in formula (6. in a more explicit situation—of instrumentation.6.1. UNIVERSAL CONCEPT FORMATS 57 possible to demonstrate in a formal sense that our ramiﬁcation modalities are complete.g. KnownPitchType(′F ′.9) will then generalize to so-called limits. STRING-typed form U nknownP itchT ype → STRING within its recursion tree. gather everything which is known to be relevant for formation of mathematical and database structures16 . Observe that this ordering system does not restrict to lexical alphabetic ordering but extends to more natural geometric constructions. Pitch). coproducts15 and powersets. Then. This is a common situation in music theory.2). we can say that the most general mathematical construction principles of products. The denotators of form UnknownPitchType have only a string of characters.

if we look at simple coordinators. Then we can easily extend the ordering lexicographically with priority on <∗ : ′D1-N ame′ F orm1(Coordinates1) < ′D2-N ame′ F orm2(Coordinates2) if and only if either F orm1(Coordinates1) <∗ F orm2(Coordinates2) or F orm1(Coordinates1) = F orm2(Coordinates2). .2 for this concept). and preceding visualization. This is above all due to the informal presentation of ideas—no database concepts were available in 1750. • A denotator search algorithm has to be deﬁned in a universal way. To begin with. We contend that the ‘EncycloSpace of denotators’ should be provided with a linear ordering < (see section C. (6. To deﬁne a linear ordering among denotators D with D-N ame D-F orm(D-Coordinates). It was an interesting conﬂict within Diderot’s Encyclop´die that the alphabetic ordering was so e low level with respect to the order of ideas. since search engines must a priori be able to invoke this algorithm. Here are three points why such a requirement is indeed fundamental: • One should be able to have a universal orientation.58 CHAPTER 6. we are provided with well-known linear orderings: The coordinator CHR of STRING has the lexicographic ordering realized in every dictionary (see example 64 in appendix C. integers and decimal numbers are given their standard orderings.2 for its deﬁnition). We introduce the linear ordering on denotators and their forms by recursion. –Σ– The question of what d’Alembert called the “encyclopedic ordering” is fundamental to any encyclopedic enterprise. • Orderings on denotators should be a germ for their representation as ‘points in coordinator spaces’. we shall stick to non-circular forms. something that could be termed ‘leaﬁng’ on the denotator EncycloSpace. To simplify the discussion. The coordinator BIT of BOOLE is ordered by 0 < 1. DENOTATORS 6. We give a naive approach to the universally deﬁned linear orderings among denotators and illustrate the results for musically meaningful denotators. by inheritance from already deﬁned orderings on the coordinates and their coordinators.1.11) we proceed as follows: Suppose that we have deﬁned linear order relation <∗ for couples of denotators regardless of their names: F orm1(Coordinates1) <∗ F orm2(Coordinates2).3 Ordering Denotators and ‘Concept Leaﬁng’ Summary.e. i.

So we may forget about names and concentrate on the structure T ype(F1 . The same holds for forms: If we have settled forms without name. . UNIVERSAL CONCEPT FORMATS and in the latter case. . First. . We now turn to compound denotators and make the general recursion hypothesis: On coordinates. For the coproduct . ..gn) (f1. let us order the types as follows: BOOLE < INTEGER < STRING < FLOAT < Syn < Then.fn) F2 H X X X X X E X X X X X X X X E X X X X F1 Figure 6. and we are left with the order relation F orm(Coordinates1) <∗ F orm(Coordinates2).Gm ) iﬀ F1 .7. This settles the order relation between forms.Fn ) < T ype(G1 .. T ype(F1 .Fn ) of a form. we set17 Fn < < {} (6.g2.. . we take the lexicographic ordering: (f1 .. . Intuitively.. . . we set (fi ) < (gj ) iﬀ either i < j or i = j.. this is the situation we know from a library. For Syn. we have an example of piano notes in their linear ordering indicated by the dotted arrow. Again. CHR. if the type is ﬁxed. and R. The simple forms are already settled by the introductory remark on the standard orderings on BIT..6. ′D1-N ame′ < ′D2-N ame′ 59 in the lexicographic ordering on CHR..12) (g1.. gn ) iﬀ we have fi < gi for the ﬁrst index i where the coordinates diﬀer. fn ) < (g1 . see Figure 6. ..Gm (6. we can concentrate on couples of denotators regardless of their names.. . .6.f2. 17 “iﬀ” is mathematical shorthand for “if and only if”.13) for the lexicographic ordering on the words on the alphabet of forms.1. For .... Z.. we just inherit the ordering on the coordinator. we proceed lexicographically with priority to the type. see Figure 6.. To the right. . the lexical naming ordering just reﬁnes the nameless ordering.6: Linear ordering on product type. a linear ordering is deﬁned. Therefore. and then fi < gi .Fn < G1 .

the usual running through key-words being extended to typically leaﬁng through geometrically shaped coordinator spaces with many coordinate axes (like vector spaces). the user can “dive” into deeper coordinator layers. Finally. gm ) of given form F is deﬁned as follows (see Figure 6. the maxima being taken on the linear ordering of the form F . fn ) and T = {}(g1 . Further. Within each book. DENOTATORS F1 <1 < F2 <2 < . and M ax(S − T ) < M ax(T − S). The books Fi are numbered from 1 to n and arranged in this order on the shelf. . < Fn <n Figure 6. we have a given linear ordering of entities.8): S<T iﬀ either S is a proper subset of T or both diﬀerence sets S − T and T − S are non-empty. Intuitively. . .. proposition 65. for a proof that the poweset ordering is linear. the powerset ordering18 between diﬀerent sets S = {}(f1 .7: Coproduct ordering is similar to the ordering within a library. we may interpret the linear ordering among denotators as being a tool for conceptual leaﬁng. 18 See appendix C.. . though never getting lost within hierarchical ramiﬁcations since linear ordering is a much stronger ordering principle than hierarchies. . .60 CHAPTER 6. .

not change the surface of our terminology. Max. and the so-called space functor identiﬁer of the form. we have introduced the form Pitch → INTEGER of simple type INTEGER = INTEGER(Z) and corresponding denotators which look like c-Pitch Pitch(c). the naive perspective will be called the zero (address) perspective if a clearcut distinction from the functorial perspective is necessary. c ∈ Z. In section 6. Denotators and forms are the naive concepts for those readers who have no necessity to view the generic setup. however. The coordinator diagram is deﬁned and discussed in detail.2. or powerset. limit. They are deﬁned in a recursive way. Simple types deﬁne the beginning of recursive ramiﬁcation.1. We shall.2). XXXX XX X X X X X X k X Max.1. The functorial approach is introduced as a variable address question. Let us start with an elementary example. XXXX XX X X X X X X k X S & ? ## ## ∞ ## ## E. 6.2 Forms Summary. the coordinator.2. a diagram of recursive references. Forms are introduced as the universal format of spaces. synonym.6. We shall nonetheless adopt the standard attitude of modern algebraic geometry and talk about “points” even though the space where the points are positioned is not a ﬁxed one but may vary as a function of an entire parameter system of ‘addresses’. colimit. T-S S-T ## X X X & X X XX XX X X X ? ## XXXXE S<T & ? T Figure 6. The ramiﬁcation type is either simple. –Σ– The rest of this chapter is devoted to the generic and formal setup for forms and denotators. we have S < T.1. FORMS 61 & ? ## X X X X X X XX X X X X ## XXXXE ∞ E. For mathematical reasons which will become clear soon. The latter is intimately related to the so-called functorial point of view (see appendix G. 6. a recursive ramiﬁcation type. .8: The linear order relation between two sets S and T of notes. carry a name.1 Variable Addresses –Σ– Summary.

In order to express this reﬁnement. the aﬃne homomorphisms f = ea · b : Z → Z in Z@Z are in one-to-one correspondence with couples (a.9. why we say that such a denotator has “zero address” is that the set 0Z @Z of aﬃne homomorphisms from the zero module 0Z to the integers Z (appendix E) reduces to the translations in Z and therefore is in bijection19 with Z. ∼ . are the aﬃne images of the unit arrow 0 ⇒ 1 in Z. if the denotator 66pitch were attached to the 6th semitone above middle C. there are important approaches in mathematical music theory which already make use of variable addresses. on a well-tempered20 keyboard. Intuitively speaking. b = f (1) − f (0). if we take Z instead of 0Z . In musical terms.9: Two diﬀerent Z-addressed points deﬁne the musical notations F and G . there are deep reasons relying on universal constructions of new denotators from given ones: Without the functorial perspective some of these constructions would not be possible. a Z-addressed object f ∈ Z@Z gives rise to two data: the “base pitch” a = f (0) and the “shifted pitch” f (1) = b + f (0) staying b semitones above the base pitch. see Figure 6. our example 66-pitch could be reﬁned to the pitch alteration F by the arrow object f = e65 · 1 19 This is true in complete generality: For any module M . each z ∈ Z giving rise to the translation ez : 0 → z. This would be suﬃcient to grasp the chromatic pitches if we took the integers for labels of semitone steps.3. DENOTATORS The reason. For example. In other words.3. as “arrow” objects in Z.62 CHAPTER 6. an adequate nonzero address provides us the necessary data. But let us give a simple example as an ad hoc justiﬁcation. But this is a somewhat reduced view of pitch in music. depending upon the tonal context. 20 Recall that all musicological terms will be introduced with care later. Second. we have a canonical bijection 0 @M → M (appendix Z E. The above Pitch form describes integer-valued pitch denotators sitting on the zero address. the Z-addressed objects 0 1 Ÿ Ÿ 65 F 66 66 G 67 Ÿ Figure 6. b) ∈ Z2 where a = f (0). Here we only use examples for motivating the denotator concept. And these denote precisely what we were looking for: A base pitch (= arrow tail) and a shifted pitch (= arrow head). see chapters 24 through 26. In fact. we shall discuss this issue in section 8.3). In score notation. respectively. What is the interest in generalizing this zero address which presently looks like an algebraic overhead to catch the elements in modules? First. we not only have the eﬀective key number but additional information about alterations stemming from tonal frames. say. for example in harmony. Whence the expression “zero address” to specify the set of denotators c-pitch with coordinates in c ∈ Z. this could be denoted by F or G .

the relations between diﬀerent addresses add canonical connections to the objects which are introduced on diﬀerent address levels. 21 Again. if we look at the zero-addressed points tail := e0 · 0 and head := e1 · 0. Moreover. the unique aﬃne homomorphism zero : Z → 0Z transforms every zero-addressed point z : 0Z → Z into the composed Z-addressed point z ◦ zero : Z → Z. IF ) where (i) NF is a string of ASCII characters. including name. On the other hand. This means that we have reﬁned the zero-address object 0 to the Z-addressed “arrow” object 0⇒1 for F or 0⇒1 → 67 ⇒ 66 → 65 ⇒ 66 → 66 for G . The deﬁnition of a form is given.6. a Z-addressed point f = ea · b yields f ’s tail point f ◦ tail and f ’s head point f ◦ head. z ◦ zero is the arrow object which sends the arrow 0 ⇒ 1 to the zero-length arrow c ⇒ c. . Summarizing. zero-addressed and Z-addressed points are canonically related: On one hand. it is called the name of F and denoted by N (F ). Recall also that for functors F u in Mod@ . Recall ﬁnally the subobject classiﬁer Ω in the topos Mod@ 107. If z stands for the denotator c-pitch. Recall the notation M @F u for the value of F u at address M .3. T F. Deﬁnition 3 A form22 F is a quatruple F = (N F. respectively21 . and the space functor identiﬁer. CF. respectively.2. the Z-addressed points of M . FORMS 63 or to the pitch alteration G with arrow object f = e67 · −1. Here.2 Formal Deﬁnition Summary. i. 22 A completely general deﬁnition of a form which implies names which are also denotators. and a morphism f : M → N of modules is called an address change. see appendix E.e. This generalization is necessary for more ﬂexible “global” name spaces.2. adding variable addresses do enrich the expressivity of denotators. coordinator.5. we stick to a naming convention which restricts to ASCII names. type. a module M which is an argument of F u is called an address of F u. for any module M . the elements of Z@M . –Σ– Recall (appendix G) that Mod@ denotes the category of contravariant functors F u : Mod → Sets from the category Mod of modules to the category Sets of sets. are in bijection with the “arrows” m ⇒ n in M . is given in appendix G. 6.

.24 3. 4. for Syn. 4. To denote a form F . whereas its domain F u is called the space (functor) of F and denoted by F un(F ). 5. for Colimit. the naming formalism is identical with the naive one. 2.14) Identif ier X in Mod@ .e. for a family (Fi )i of forms. X = F un(CF ). 23 They turn out to symbolize ﬁve types of operators. 24 In the generic deﬁnition of a form.64 (ii) TF is one of the symbols23 1. 2. Power. we just need the symbols. For Simple. X = colim(D). the character strings. for Power. Syn. Evidently. CHAPTER 6. B. CF is a diagram D of forms. with this data: Naming. We shall also adopt the policy that diﬀerent forms should bear diﬀerent names. but in the deﬁnition. 5. X = ΩF un(CF ) . X = lim(D). the synonym form type is superﬂuous. for Limitand Colimit. The diagram D is a diagram of functors F un(Fi ). and Power. as deﬁned in (iv). i. it is called the identiﬁer of F and denoted by I(F ). We however maintain this type for semiotic reasons: synonymy is a proper type of reference which is meant to be diﬀerent from any other reference mode. CF is a module M . 3. . (6. C. we inherit the notation of the naive setup and add the identiﬁer below the arrow: N ame −→ T ype(Coordinator). Colimit. For Simple. for Syn. Simple. CF is a form. for Limit. The codomain of the identiﬁer is called the frame space of the form. it is called the type of F and denoted by T (F ). X = @M . (iv) IF is a monomorphism of functors IF : F u 1. DENOTATORS (iii) CF is one of the following objects according to the previous symbols: A. it is called the coordinator of F and denoted by C(F ). Limit.

product. this cannot be guaranteed. Coordinator. A basic comment on the structure of deﬁnition 3 is necessary. As to the other types. In mathematical deﬁnitions. B. Z. For synonymy and power types. circularity that modules can have any commutative or non-commutative coeﬃcient rings! the identiﬁer problem arose when transcribing the software-oriented PrediBase data base management system [589] to an abstract formalism. the identiﬁer’s domain designates the concrete structure. Identiﬁer. INTEGER. i.2. quite often the identiﬁer is not only mono but even iso.3. a remarkable diﬀerence in that we add the new instance of an “identiﬁer” to the naive setup. 25 Observe . simple types were restrained to a selection of four sample coordinators (CHR.5. BOOLE.0. Coordinator. In general. The real typological diﬀerence resides in the last three types. The latter is taken care of by the identiﬁer monomorphism whereas the functor F un(F ) of the form. we are given a module. simple type means selecting any coordinator module M .4. you make precise which kind of zero representation you have taken. we may encounter several onedimensional real vector spaces V. you sort of point at the particular 0. Indeed. say. and this is what our monomorphism takes care of: It identiﬁes a representative with “the abstract object”. The typology is somewhat more complicated: For the naive approach. for decimal numbers (ﬂoats).2. the last two share circularity: They refer to the form concept to be deﬁned. The concept of a “form” refers to four ingredients: Name. Type. BIT. i. Z. synonymy looks quite the same as with the naive approach: We just have to go back one step in recursion.e. Such an ∼ isomorphism I(F ) : V → R is denoted by the identiﬁer26 .2.e.1 for this subject. whereas a diagram would contain a number of “arrows” between members of such a sequence. and Identiﬁer. and R) whereas here. FORMS 65 Typology. mathematics. the default value will tell what kind of representative of the domain you have selected. instead of one of the four simple coordinators in the naive setting. We are going to describe all the types and their characteristics in section 6. 2.0 as a default value. For example. as well as (iv). For limits and colimits. Why? In the naive setup. In fact. See appendix G for the concept of a diagram in a category. There is. but essentially. What you are really doing is pointing at a particular representative of “the” 0R ∈ R. however. In our context. this means that the basic substance spaces are now taken from the category of modules and not merely from four types STRING. see appendix G. BIT.2. It is plausible that the vast domain of modules25 is large enough to cover once for all the needs of musicology. they are all isomorphic to R. and Colimit which generalize the naive types powerset. Power. and FLOAT. R). so we have to integrate them without losing their structural identiﬁcation. to 5. isomorphisms would create serious existence problems for circular forms. The coordinator looks quite familiar for simple type. 26 Historically. This initialization process means that you have to write down a ﬁrst concrete value to be instantiated without further activities from the user’s part. but see appendix G. If you write down 0. and computer science for grounding space substance. and coproduct. We shall make this more explicit in section 6.0: 0R → 0. a form diagram has its vertexes in the set of form names and evaluates to the corresponding form functors. at least in the cases (iii). we are given a diagram of forms. object-oriented programming languages ask for default values of instance variables in order to initialize objects. Whereas the ﬁrst two speciﬁcations are inoﬀensive. However.6. Philosophically. If you cannot stick to a unique domain of values.3 for a rigorous formalism. This generalizes the naive setup where we just had a ﬁnite sequence of forms in the coordinator. Limit. many equivalent modules or higher constructs (functors) may intervene. and C. we had a unique choice of simple spaces: the four coordinators (CHR. we have the same data as in the naive setting.

e. if h : F u → Gu is a natural transformation.2. we set a ∗ 0 = 0. the by the following subfunctor S ˆ subfunctor S is deﬁned by ˆ N @S := {(u ∈ N @M.15) is also natural in F u. (6. we deﬁne a natural transformation ˆ : 2F u → ΩF u ? (6. i. and a functor monomorphism F un(F ) F un(C(F )). But in set theory.1) in M . the module M (one of four sample modules) was all we needed to describe simple type forms. let F u be a functor of Mod@ and consider the functor 2F u of Mod@ . Syn.16) ∼ It is easily seen that the transformation (6. if one is given an ordinal number associated with every instance of the concept in question that appears within the deﬁnition. . Since in the naive setup. this means that F u is representable.3. F u is isomorphic to a sieve (appendix G. we use the set map f @F u : M @F u → N @F u to deﬁne f @2F u : M @2F u → N @2F u : S → f @F u(S) =: S · f . a functor F un(F ) in Mod@ .5. In “normal” set theory. the following canonical diagram is commutative: 2F u − − → Ω F u −− h h Ω ˆ ? ˆ ? 2 (6.3.66 CHAPTER 6.3.17) 2Gu − − → ΩGu −− 27 If the identiﬁer is iso.e. For a formal deﬁnition of logically consistent domains of forms. For a given address N . To connect the naive powerset type and the general one. Power.5. For example. –Σ– Simple. a coordinator module M . v ∈ N @F u)|v ∈ S · u}. the powerset of the F u-value at address M . and that it is represented by I(F ).15) ˆ ⊂ @M × F u for S ∈ M @2F u . We discuss the characteristics of the form types and compare them to the naive variants. In general. examples 99 and 97) that ΩF u evaluates to M @ΩF u → Sub(@M × F u) at address M . circular sets are excluded by the axiom of regularity [281]. see appendix G. the product of natural numbers a ∗ b is deﬁned according to the size of b: For b = 0. This form type adds a new name N (F ) to an already known coordinator form C(F ). 6. DENOTATORS is admitted only if it is embedded in a recursive deﬁnition structure. we did not have to change addresses. and a ∗ (b + 1) is deﬁned to be (a ∗ b) + a. a set Z deﬁned by Z := {Z} is a problematic object. To relate the naive powerset type to Power.3 Discussion of the Form Typology Summary. For an address change f : N → M . since no ordinal “size” number is given. We come back to this issue in section 6. deﬁned by M @2F u := 2M @F u . i. and an identiﬁer monomorphism27 I(F ) : F un(F ) @M . A form F of simple type consists of a name N (F ). such non-regular. Recall (appendix G.

n . a denotator is denoted by N ame : Address F orm(Coordinates) (6. the morphisms being the morphisms of the forms’ functors. an object’s existence is equivalent to the possibility to think the objects without provoking any contradiction to classical principles of logic: identity. (ii) FD is a form. F D. 29 This is a plural in singular mode. We would like to stress that all these constructions presuppose that the reference to the coordinator’s forms and their functors is possible. it is called the name of D and denoted by N (D). A “diagram of forms” is just a diagram in the category Mod@ . The formal concept of a denotator as being a form plus a substance point is introduced.2).. that these objects all do exist28 . Implicitly. (iii) CD is an element of M @F un(F (D)). . Equally. via the functor F un(F ) of a form F (see appendix G.3. contradiction.. so the terminology “diagram of forms” is also formally correct. –Σ– 6. that this transformation connects naive powerset denotators to those of Power form. and excluded third.n . According to the naive setup.6...3.2. item “Coordinates”. Limit.2. CD) where (i) ND is a string of ASCII characters. DENOTATORS 67 We shall see in section 6. A M -addressed denotator is a triple D = (N D. Deﬁnition 4 Let M be an address. 6. Based on the deﬁnition of a form (subsection 6. We come back to this problem in section 6.1. deﬁnition 151 for diagrams in categories). i. we deﬁne denotators. Colimit. it is called the coordinates29 of D and denoted by CT (D).3.1..1 Formal Deﬁnition of a Denotator –Σ– Summary.3 Denotators Summary.e.18) 28 Following Paul Finsler [153].. a coproduct Fi of forms is a colimit for a discrete diagram whose vertexes are the indices i of the product’s family (Fi )i=1. The relation to the naive setting is that a product Fi of forms is a limit for a discrete diagram (no arrows) whose vertexes are the indices i of the product’s family (Fi )i=1. the point concept is also recursively deﬁned by its reference to the recursive structure of the underlying form space that was introduced in the preceding section. We can also speak of the category of forms. it is called the form of D and denoted by F (D).5.

DENOTATORS So. the coordinator is a module N . the coordinates CT (D) identify with an element of M @F un(C(F (D))). the coordinates CT (D) identify with an element of M @ΩF un(C(F )) → Sub(@M × F un(C(F ))). In the case of the zero address M = 0Z . or even AF orm(myCoordinates) if myAddress is also clear (in fact. 30 In (6. Again. But we know from the discussion of form typology in section 6. We also see that an M -addressed naive coordinate set S ⊂ M @F un(C(F )) is just a set of “points” at a general but ﬁxed address M . For Simple. 2.20) (6. as already described for the naive setup. i. The naming policy is identical with the naive situation. Passˆ ing to the functorial setup means switching to an entire system of sets ?@S. Coordinates. 3. So the coordinates are described by recursion to the coordinator C(F (D)).19) but this clumsy writing is only used if absolutely unavoidable. however.2. (6.3).3 that each set S ∈ M @2F un(C(F )) of ˆ M -addressed points in F un(C(F )) gives rise to a subfunctor S of @M ×F un(C(F )). the zero address in the naive context). So the zero-addressed coordinates in the naive sense give rise to coordinates in the general setup. Naming.e. i. For Syn. and the coordinates CT (D) are identiﬁed (via the identiﬁer!) with an element of M @N . Address. if the denotator name is empty (which happens often). where ˆ S is a ‘functor of subsets’ in M @ΩF un(C(F )) . in the full notation. . The form of a denotator englobes the whole recursion information as well as the form’s functor. It is the latter which contains the coordinates. such a morphism clearly identiﬁes with a ‘real’ element of N (appendix E. a denotator D with form F is symbolized as follows: N (D) : Address N (F ) Identif ier −→ T ype(Coordinator)(Coordinates). as if there were just one ambient space instead of an entire space functor. a morphism CT (D) : M → N of modules. an element of M @F for a functor F ∈ Mod@ is called an “M -valued point of F”. Recall that we can have the empty denotator name ′′. a subfunctor of @M × F un(C(F )).e. we shall write myAddress AF orm(myCoordinates) for a denotator of form AF orm and coordinates myCoordinates. Form.21) ∼ algebraic geometry. Let us look at the shape of coordinates as a function of the particular form: 1. we shall not overstress it since in the everyday language we speak of denotators independently of their address.68 CHAPTER 6. The coordinates are one “point”30 or element of the form’s functor at the given address M . For Power. The address module M is an important generalization.

but ﬁxed address. generated by the relation x ∼ y. 5. but ﬁxed address M . we have a natural isomorphism M @lim(D) → {x ∈ i ∼ M @F ui | m(f )(xtail(f ) ) = xhead(f ) . 6. In other words. x ∈ M @F ui and y ∈ M @F uj . In the naive setup this was less hidden because of the missing identiﬁer (the identiﬁer was ‘set to identity’). theorem 60) that for a given address M . Recall (appendix G. So—up to general addressing—the coordinates of a Colimit denotator are those of the coproduct modulo an equivalence relation deﬁned by the diagram’s arrows. a denotator bears a fundamental pointer character: It can only be understood by pointing down to the coordinates which are distributed among the various recursive forms. For Colimit. Therefore the general setup does resemble the naive one as being focused on a general. we should comment on this decision: Why is the module structure a good choice? There are mathematical and musicological. Recall (appendix G. as well as natural transformations m(f ) : F utail(f ) → F uhead(f ) for arrows f in D. theorem 60) that at a given address M . Again. all arrows f of D} so that the coordinates of Limit denotators are special tuples in the product of all M @F ui .6. we have the equivalence relation ∼ on the coproduct i M @F ui . –Σ– Since modules are anchor structures for denotators.1. j = head(f ).2. let us inherit the notation of the preceding situation. but we recognize from the above discussion of the coordinates that a denotator’s coordinates are just as recursive as the involved forms. If f : F → G is a morphism of forms.2. It presents a justiﬁcation for choosing modules and aﬃne transformations as basic space types.4 Anchoring Forms in Modules Summary. and if D : X F (C) is a denotator. and m(f )(x) = y.4. This section deals with recursive foundation of forms in simple spaces. practical . selecting a general. suppose that the diagram D of forms has vertex forms Fi and vertex functors F ui := F un(Fi ) at vertexes i.1. iﬀ there is an arrow f in D with i = tail(f ). ANCHORING FORMS IN MODULES 69 4. and again. For Limit. Superﬁcially the denotator deﬁnition is not recursive. First examples are presented and discussed in the light of musical and musicological requirements. we canonically have the f -image of D f (D) : X G(X@f (C)). We then have a natural isomorphism M @colim(D) → i ∼ M @F ui / ∼ and recognize that the naive coproduct is just the basic space of the Colimit denotator space before dividing through the equivalence relation ∼. the limit denotators are canonically related to the product denotators from the naive setup.

and not what happens in any case.70 CHAPTER 6.618036 is the famous golden section. Fibonacci numbers31 with Karlheinz Stockhausen [218]. such a requirement would be too restrictive since mathematics should make available to music what may possibly happen in music. we need not only consider “points in spaces” but also algebraic operations which relate points to each other and allow creation of new points out of given ones. –Σ– after Italian mathematician Leonardo Fibonacci (1180-1250 (?)). the Fibonacci sequence (xi )i=1. let us shortly digress on the relation between mathematical structure and musical meaning. we have also compiled a number of common parameter spaces for musical objects in appendix A.2. In this sense. We have transferred the purely mathematical description of a module to appendix E. 6. a mathematical structure may serve as a vehicle of the sound material’s organization that otherwise has never been used. We presuppose the knowledge of module structure and of the examples from the appendices A and E.4. mathematics must furnish a minimal conceptual and theoretical region where the possible applications can be located.. Musicological aspects of the algebraic structure of a module and of aﬃne transformations among modules are discussed. Rather does it catalyze the fundamental dialectic process of how music sees itself. They are the adequate structure to grasp this formal mechanism of building new from old points. Before looking at concrete examples. we refer to these chapters and concentrate on the justiﬁcation discourse. for example.. or fractals with Mesias Maiguashca [220]. is deﬁned by x1 = x2 = 1 and recursively by xi+2 = xi +xi+1 . but we will pick up some of them to illustrate their role.. The limit number limi→∞ xi /xi+1 ≈ 0. To begin with. In other words. It cannot be decided a priori which mathematical structures carry musical meaning. e. and this insight is the reason for introducing modules. It is however not true that reﬂecting musical and musicological thinking via mathematical structuring may lead to “foreign inﬁltration in music” as it was argued in [394]. In order to handle denotators in music. more concretely: in the process of generating a determinate composition. a module structure is but an adequate formal frame for an empirical material which presently guarantees a broad ﬁeld of applications—in the future it may even be enlarged.3. such as text elements. In particular with regard to poietics.1 First Examples and Comments on Modules in Music Summary. Here. We should also keep in mind in the following that it is not necessary (and sometimes not possible) to look at all addresses for simple denotators. In music. let us view the general point of taking modules. We shall encounter cases where mathematical structure overhead seems to appear. 31 Named . DENOTATORS and theoretical reasons. We then obviously have to reﬂect on possible musical or musicological interpretations of this overhead. aleatorics with John Cage [277]. This is a delicate question since it may lead to contrary answers depending on its topographic context. We want to expose them generically and by use of common examples. One could require that a determinate structural instance (such as additive closure of modules) has to play a musical or musicological role in all situations where the structure is present. sound objects are not understood as being isolated points in amorphous spaces but parts of an entire relational system.g. However. We also illustrate the use of modules for more general denotators which may intervene for not strictly musical purposes.

BIT .23) and this means that the operator D(a. More generally. example 75. F ormList(S) := N =form name in S aN N . 0) is a reset to the state a of the switch. We ﬁrst build the free monoid A of words ′x1 x2 . pauses. The structure of Z ASCII as a Z-module is more than the simple names of denotators and forms—for example—require. This set is also known as BIT = Z2 = Z/2Z. bar-lines. see appendix D.6. xn ′ from a character alphabet (the set ASCII of ASCII characters in our case) is not a module.24) We obtain the eﬀective calculation of the score’s form list by successive addition: If S splits into two disjoint parts S1 and S2 . when compared to T . and we have CHR ⊂ Z ASCII . the denotators D(a. Its construction presupposes an alphabet A (generalizing A = ASCII) and a commutative ring R of ‘coeﬃcients’. etc. i. In fact. (6. time and key signature. 2.1). Further. if F ormList(S) and F ormList(T ) are two such lists. we have F ormList(S) = F ormList(S1 ) + F ormList(S2 ). 1)(s) = s if a = 0. xn ′ over the alphabet A. the words w = 1w are elements of the monoid algebra. the diﬀerence F ormList(S)−F ormList(T ) in general produces negative coeﬃcients which measure the relative amount of form names in S. 32 The symbol ¬s means the negation of state s. The monoid algebra R A consists of all formal sums w∈ A rw w with only ﬁnitely many coeﬃcients rw = 0. CHR. If we view the elements s of BIT as being the states of an ON-OFF switch on a musical instrument.e. Such a denotator D(a. . This deﬁnes a multiplicity-sensitive list of forms in S. In particular. the theory of switching circuits and automata [236] can be executed on denotators which are built upon BIT as a simple ‘module basis’. But the use of formal sums is straightforward: We can take the names of all forms which appear in a determined description of the denotators (notes. 1) is the a-translation on BIT and represents its additive structure. ANCHORING FORMS IN MODULES 71 1.1.4. . . 0)(s) = a.22) Id So a = 0 leaves the state s of the button whereas a = 1 changes it. the integers modulo 2. (6. look at the simple ‘self-addressed’ denotators D(a. In the naive setup. b) can be viewed as operators on the states of the switch: We have32 D(a. the set of character strings ′x1 x2 . b) : BIT Switch(ea · b) of form Switch −→ Simple(BIT ). . It has an additive structure which turns the elements into logical units used in music instrument editing. . having coordinates ea · b ∈ BIT @BIT . From this we learn that account of words from a determined alphabet can make use of the corresponding monoid algebra over some appropriate ring to perform its calculations. 1 + s = ¬s if a = 1. composer’s name.) of a score S and attribute to these names N the multiplicity aN ∈ Z of their appearance in denotators of S. (6.5. see also appendix G. Observe that we also have D(a. but it can be embedded in a module according to the very comfortable and common mathematical structure of a monoid algebra (see appendix D.

p-fold juxtaposition. . This is based on the principle that musical meaning is conserved after such ‘recombination’.25) as discussed in appendix A. we have p · (h(f ) − h(g)) = p-fold juxtaposition of h(f ) − h(g). Further. v. This transformation reﬂects the auditory experience that the frequency ratio f /g is perceived as a “pitch distance”: Fact 1 The additive structure of R[Q] is the basis of thinking in pitch distances. we interpret denotators M athP itch(h) as having the mathematical pitch coordinate h = h(f ) = u · ln(f ) + v associated with frequency f and a couple of normalization coeﬃcients u.3. • 1/q·(h(f )−h(g)) means that we seek a distance which—after a q-fold juxtaposition— yields the original distance h(f ) − h(g). What is the musical meaning of these three operations? • Multiplication of a pitch diﬀerence h(f ) − h(g) by −1 yields −1 · (h(f ) − h(g)) = h(g) − h(f ). q a positive natural number. In accordance with that discussion. DENOTATORS 3.e. In this view.1. the special case p = 0 means that we switch to the zero distance. the roles of the two pitches are exchanged. • a natural number p ≥ 0. the ratio h(f ) − h(g) = u · ln(f /g) of two frequencies f and g becomes a diﬀerence of MathPitch coordinates. and q-fold division of pitch distances. We want to view this module as being the coordinator of a simple pitch form M athP itch −→ Simple(R[Q] ).2. The real numbers R. Fact 2 Scalar multiplication in the pitch module splits into role exchange. scalar multiplication with a rational number a/b splits into a product of three special cases: multiplication by • −1. • 1/q.2. i.72 CHAPTER 6. we look from the second pitch to the ﬁrst instead of looking from the ﬁrst to the second. Remark 1 The signiﬁcation of a mathematical operation often becomes evident after a ‘split’ into special cases from which the general case results via appropriate conjunction. This is genuine musical thinking! The example of the contrapuntal operation of 33 In fact an inﬁnite-dimensional vector space. • For p ≥ 0. Id (6. Recall (appendix E. example 77) that R is a module33 over the subﬁeld of rational numbers Q which we denote by R[Q] .

Division 1/12 · o deﬁnes the distance for . and major third by role exchange. 1. the ﬁfth and major third (appendix A. Here. Therefore we have a corresponding morphism of forms E2M : EulerM odule → M athP itch (6.29) As in the preceding example.2. exchange of roles means that distances are juxtaposed ‘downward’ instead of ‘upward’. This module is introduced in appendix A.30) and the diﬀerence of pitch corresponds to the diﬀerence of Euler points. With the same transformation. (6. ANCHORING FORMS IN MODULES 73 “retrograde inversion” rightly demonstrates that the operation is only ‘understood’ via its decomposition into “retrograde” and “inversion”. every Euler point d = (r. We also view this module as being the coordinator of the simple form EulerM odule −→ Simple(Q3 ). and t = (0.28) induced by the above linear embedding and its canonical extension to the forms’ functors @ pv| : @Q3 → @R[Q] . and where the scalar product pv|x is the above pitch h(f ) associated with frequency f . juxtaposition and division.6. lnb (5)) is the “prime vector” relative to a logarithm base b. q = (0.3.3) are associated with the Euler points q − o and t − 2 · o. we have pv|x − pv|y = pv|x − y .27) of the Euler module in the pitch module. We come back to this very important subject in section 8. lnb (3). and this stems from music history. every diﬀerence x − y can be written as d = (r + s + 2u) · o + s · (q − o) + u · (t − 2 · o). The Euler module Q3 over the ﬁeld Q of rationals. 4. (6.32) and hence be generated from the octave. u) can be written uniquely as a linear combination d = r · o + s · q + u · t. (6. (6. according to fact 2. 0). 0). the additive structure becomes musically signiﬁcant via diﬀerences x − y of Euler points x and y (Euler points are just elements of the Euler module). In fact. s. where pv = (lnb (2). Hence.4. 0. 1) are the canonical base vectors.3 and represents one of the most classical mathematical structures in mathematical music theory.26) Recall that we have a Q-linear embedding pv| : Q3 → R[Q] : x → pv|x (6. If o = (1. 0. Id (6.31) On the other hand. But the Euler module has more in its structure. ﬁfth. we can carry over the musical signiﬁcation of scalar multiplication from pitch to Euler points. and not as such.2.

ﬁfth. DENOTATORS semitones in the 12-tempered system. limits without arrows.e.38) (6. q. N ) (6. Therefore there was interest in the diﬀerence between least common multiple of the numbers of octave steps and ﬁfth steps (= ﬁfth comma) as well as of third and ﬁfth steps module octave (=third comma). and t.3 that arbitrary small pitch diﬀerences can be produced by integer linear combinations of the three vectors o. In the tradition of just tuning. but they are very similar to simple forms.2. 5. these commata ﬁgures are not interesting since we know from appendix A. and a major third equals 4 semitone steps. If we work within just tuning. We can then Id build two forms F ×G F ⊕G −→ @(M ⊕N )→@M ×@N Id@(M ⊕N ) −→ ∼ Limit(M.74 CHAPTER 6. new forms are built from simple forms by products. upward or downward. a ﬁfth equals 7 semitone steps. i.40) Simple(M ⊕ N ) . These are no longer simple forms. Direct sums.39) (6. an octave equals 12 semitone steps. Commata will only become interesting in harmony of just tuning (see section 24. Let us make this precise for the prototypical case of the product of two forms F −→ Simple(M ) Id and G −→ Simple(N ) with coordinator R-module M and S-module N .35) (6.46 Cent third comma ≈ −21.33) (6. and major third. whereas 1/2 · (t − 2 · o) means whole tone distance between tonic and third in mediante tuning (see appendix A.34) Historically. any distance d = x − y can be produced by successive juxtaposition of octave. With these identiﬁcations we have: 12 ﬁfths = 7 octaves (semitone steps) 4 ﬁfths = 2 octaves + 1 third (semitone steps) One ﬁnds the following approximate values: ﬁfth comma ≈ 23. In this context.3).36) Mathematically. Often.51 Cent (6.1).2.37) (6. these commata arose from the 12-division of the octave in semitone steps. the following Euler point diﬀerences have special names [547]: ﬁfth comma ( = Pythagorean comma) Kq = 12 · (q − o) − 7 · o = 12 · q − 19 · o third comma ( = syntonic comma) Kt = (t − 2 · o) + 2 · o − 4 · (q − o) =t−4·q+4·o (6.

(6. For example. For instance. . (6. we can take OnM odm −→ Simple(Zm ) Id (6. If we have two module namings M := Z4 . With this in mind. its denotators OnP iM odm. then it can be stated that “M and N are identical modules”.n = OnM odm × P iM odn . but it is substantially more than that. and N := Z4 . the algebraic a priori isolation of its factors is overridden by a common module where the factors can be compared on an algebraic level. we may identify these forms and also identify34 F1 × .43) Recall from appendix A. the naming of forms is completely diﬀerent from naming mathematical modules. we may reinterpret the additive structure induced by direct sums of modules: If a product F × G of simple forms F and G is built as in (6. but it is also a relativization of mathematical structures in that they need to be enriched: Mathematics is only the ‘signiﬁcate’ of the form sign. .n (f ). whereas OnM odm can be viewed as the form whose coordinator Zm denotes the m-cyclic metrical onsets. But for two forms F −→ Simple(Z4 ) Id adopt the saying that “a compound form F simpliﬁes to a simple form G” each time we build a simple form G which is isomorphic to a compound form F . The mathematical anchoring of forms in modules is a clear creed to mathematics. we should review the role of mathematics and semantics in the form concept. These denotators S(f ) put into aﬃne relation both. 34 We . The paradigm of comparison will be furnished by the denotators S(f ).n is positioned on a torus and denotes a type of motivic points.3).44) which evaluate to aﬃne endomorphisms f of Zm ⊕ Zn . Fn if no confusion is likely. if n. Clearly.n (x.42) P iM odn −→ Simple(Zn ) Id with Z-modules Zn and Zm as coordinators. i. Now.6. and the type. and more in detail in section 22. ANCHORING FORMS IN MODULES 75 where the ﬁrst identiﬁer is the canonical isomorphism related to the universal property of direct sums of modules. Putting these two a priori autonomous aspects of a musical note together.e. .8.4.2. and thus initiate a comparative theory of motivic points and consequently of motives as we shall see in chapter 11. The coordinator of the form OnP iM odm. we obtain a common representation of onset and pitch within the module Zm ⊕ Zn .4. m are two positive natural numbers.3. Concluding this section. onset and pitch data. y) are points of what musicology considers in motif analysis (see section 7. these forms are isomorphic under the identity of @(M ⊕ N ). Under this canonical isomorphism.3 that P iM odn can be viewed as the form whose coordinator Zn denotes the pitch classes in n-tempered tuning. The conceptual genealogy can be anchored in mathematics.41) (6. identiﬁer and name are equally important to build the entire sign. section 11. Fn with F1 ⊕ . the algebraic connection of onset and pitch data in this symbolization becomes evident from the self-addressed denotators S(f ) : Zm ⊕ Zn OnP iM odm.39). . Consider the simpliﬁed form OnP iM odm.2.

5 that there are fundamental diﬀerences between forms and straight mathematics. and if ω is the successor of all the coordinators’ levels. (iv) A form is called circular iﬀ it is not regular. (iii) A form is regular iﬀ there is an ω such that the form is regular of level ω. The simplest example of circular synonymy is a form of this shape: ′SynCirc′ f :F u −→ Fu Syn(SynCirc).e. Such a form is identiﬁed by its name and identiﬁer. Their very existence can be veriﬁed only via their name’s function as a sign’s signiﬁcant. They transcend simple naming conventions in that form names are the only access to forms in circular deﬁnitions. The names are essential. or if such a recursion is not possible. f ) of a character string.5 Regular and Circular Forms Summary. Let us look at some catastrophes which may or may not occur in these cases: Circular Synonymy. but in most cases. (ii) Suppose that regular forms of all levels µ < ω have been deﬁned. This can however not work for circular forms. Here is the deﬁnition of a regular form: Deﬁnition 5 Let ω be an ordinal number35 . Then a form is regular of level ω iﬀ all its coordinator forms are regular of levels µ < ω. (6. –Σ– As already announced.76 CHAPTER 6.45) where a form SynCirc is synonymous to itself and therefore the identiﬁer is a monomorphism f of the form’s functor F u. 35 See [281]. 6. we may just think of ω being a natural number. F u. (i) A form is regular of level ω = 0 iﬀ it is simple. We have to distinguish two cases: Regular and circular forms. and a mono endomorphism of this object. i. We discuss the existence problem of a form if in its deﬁnition recursion is noncircular. we cannot say that “F and G are identical forms”.e. F −→ Simple(M ) and F −→ Simple(N ) are identical Id Id forms: The naming diﬀerence on the mathematical level is irrelevant. DENOTATORS and G −→ Simple(Z4 ). However. So the recursive deﬁnition of a regular form is built upon its level. and we may deﬁne regular forms without any further complication. not from mathematics. we could Id say that. terminates on the level of simple forms. . with the above notations. an object in Mod@ . it reduces to a triple (′SynCirc′. the existence question of a form is not trivial. i. We shall see in section 6. and this stems from semiotics.

.47) (6. clearly.6. Again.49) From this example we learn that there are inﬁnitely many36 realizations—on the level of f or of F u—of the above circular limit form. for an address M . g0 ). Circular Colimits.54) But we have the canonical monomorphism f = sg = ˆ · {} : F u ? 36 There 37 If 2F u ΩF u (6. a diagram has no arrows and is ﬁnite. The prototype of a circular form of limit type is this: ′LimCirc′ −→ f :F u F u×F un(G) 77 Limit(LimCirc.52) f : Fu → Fu which is induced by the cofactor isomorphisms f |i = Id : F un(G)i → F un(G)i−1 ∼ Id : F un(G)i → F un(G) ∼ for i > 0. This one can be inspected on a prototype of shape ′P owerCirc′ −→ f :F u ΩF u Power(P owerCirc). G). (6. M @f :M @F u →M @F u × M @F un(G) (gi )0≤i → ((gi+1 )0≤i . for i = 0.53) F un(G)i being the cofactor of index i in the coproduct. This form can be realized F u := N ∼ F un(G). (6. (6.5.55) are even inﬁnitely many mutually non-isomorphic realizations. ∼ (6. Circular Power. If we deﬁne the form’s identiﬁer by F u := F un(G)N .50) ∼ f :F u→F u F un(G) with the disjoint union by of functors in its identiﬁer codomain. there are inﬁnitely many realizations of the above circular colimit form. REGULAR AND CIRCULAR FORMS Circular Limits. we may look at this form37 : ′ColimCirc′ −→ Colimit(ColimCirc. where. we may denote the factors as for products in the naive approach. In complete analogy to the preceding limit case.46) where the circular form LimCirc appears in a product with another (supposedly inoﬀensive) form G. and F un(G). and f : F u → F u × F un(G). (6.51) (6. (6. G).48) (6.

64) (6. Simple(Z ASCII ).59) (6.1.58) (6. we consider ′book′ −→ Limit(N o. where F in(F u) is the subfunctor of 2F u of all ﬁnite subsets of N @F u at address N . there was just the zero address.2. Let us look at the example 6.67) D = • → P . But notice that we do not have uniqueness of such a solution! However. the above limitation technique to prevent the power sets to appear in full is not always suﬃcient to avoid non-standard situations with circular forms. But observe ? that this situation forbids that the identiﬁer becomes an isomorphism. In our situation. lemma 91) and ˆ (see equation (6. −→ @Z ASCII (6.2. chapters) f (6. title.) × F un(title) × F un(text). ΩF u −→ @Z ASCII ′chapters′ −→ F in(F u) Power(book). and this is guaranteed by the second isomorphism in proposition 103 in appendix G. The point is that with the naive setup. ∼ (6. DENOTATORS which is composed from the canonical singleton morphism {} (appendix G.78 CHAPTER 6. text.) × F un(title) × F un(text) × F un(chapters) (6.′ ′title′ ′text′ −→ F un(N o..65) (6.60) (6. and which canonically injects to ΩF u .62) (6. where g P −→ Power(F ) and Id N @g : • → N @F un(F ) ∈ Sub(@N × F un(F )) .) F un(title) F un(text) @Z F un(N o. In fact.61) Simple(Z ASCII ).66) (6.56) with identiﬁer f : Fu and the factor forms ′N o.57) Simple(Z). we have to ﬁnd a solution which is functorial. f (6.15)) and serves as identiﬁer for any functor F u.63) this would guarantee that any text data (codiﬁed in H) and any ﬁnite set of chapters (codiﬁed in F in(F u)) can be reached by a book-form denotator. such an isomorphisms f cannot exist since the calculus of cardinalities [281] categorically excludes set injections of type 2A A for any set A. Consider the example deﬁned by these data: F −→ Limit(D). we want to deﬁne a functor F u such that f is an isomorphism f : F u → H × F in(F u). To restate that example in the formal context. Setting H := F un(N o.8 from the naive setup.

and crescendo.70) (6. Id Loudness −→ Simple(R). we shall add more and more examples to meet the speciﬁc problematic.71) (6.69) (6. This would even not be the philosophy of the denotator formalism. Id .74) P itch −→ Simple(R). It cannot be the scope of this chapter to present an exhaustive list of musicologically relevant regular denotators. The lesson from this chapter is above all to learn the technique of building concepts via denotators. RegDen-1. Id Crescendo −→ Simple(R).72) (6. Id Glissando −→ Simple(R). we should present a fairly representative list of denotators for a reasonable arsenal of musicological concepts. They are structurally characterized by non-circular recursivity: These denotators do not make recurrent reuse of reference denotators—forms or coordinates—within their coordinators. pitch.6. including most of the common object types encountered in music scores and in music-theoretical contexts. These are three basis parameters onset.73) (6. Id Duration −→ Simple(R). After all. A common space functor for each of them is the real numbers R (viewed as a real vector space). We have to insist that denotators can and should be introduced again and again according to the speciﬁc need. As theory evolves. We deduce the functorial set equation N @f : N @F un(F ) → {N @F un(F )} ∼ 79 (6. 6. –Σ– Deﬁnition 6 A regular denotator is a denotator of a regular form. We then have this arsenal of simple forms: Onset −→ Simple(R). we need a common numeric representation of elementary mental tone parameters as they appear in common European music score notation. Throughout the following explications the reader should bear in mind the prerequisites about common spaces in musicology as discussed in appendix A. Id (6.6. To begin with.68) which is fulﬁlled by any circular set C = {C} and the constant functor N @F un(F ) := C. glissando. There is no reason whatsoever to believe that a terminology should be limited in its concrete exempliﬁcations: Music is a dynamic ﬁeld of human culture and cannot be understood upon any historically ﬁxed conceptual ﬂashes. This section is devoted to a ﬁrst example set of denotators. REGULAR DENOTATORS for the ﬁnal functor • and an address N .6 Regular Denotators Summary. and loudness and three corresponding pianola parameters duration. Nonetheless. denotators have been introduced to cope with all the options of dynamic conceptualization.

g. for example.57 or even Onset = 3. then we may view it as a rational denotator and simply write Onset|Q = 123/1 = 123 since no confusion is likely. etc.57 for an Onset denotator of coordinate 3. the canonical (di)morphism R[Q] ∼ R deﬁnes an isomorphism (6. e. in the notation “A ”. if we restrict pitch to integer values. DENOTATORS We inherit the notation from naive denotators and write Onset(ON ).D.25).C for denotators of forms Onset.4. Id ∼ (6. Loudness. Observe that an inclusion T S of subspaces of @R implies inclusions of respective forms Onset|T Onset|S Onset. S @R (6. the address is always zero. we shall simplify any direct product of the above forms to the associated simple form. Without further comment. Glissando. for example. A quantitative pitch may be associated as a function of speciﬁc tuning paradigms.80 CHAPTER 6. In particular.1. Also we shall often use the names E.3.G. With the identiﬁcations discussed in section 6. Pitch. e.57 ∈ 0Z @R. . RegDen-2.L.g. etc. Thus we need a “neutral” symbolic notation which we realize by the obvious form P itchSymb −→ Simple(Z ASCII ).79) 38 We often omit the “functor” speciﬁcation and more intuitively speak about “spaces” when we mean “space functors”.77) M athP itch → P itch|@R[Q] with the mathematical pitch space deﬁned in (6. item 5 in mind. the middle C pitch of the MIDI code is P itch|Z = 60 or H = 60 if the form P itch|Z is clear.78) (6. Crescendo and also write E = 3. Onset × P itch × Loudness × Duration → Onset ⊕ P itch ⊕ Loudness ⊕ Duration. if we are given a denotator Onset|Z = 123.1 we know that a form morphism also transplants the respective denotators.76) From section 6.75) For example. the latter being a form with coordinator R4 . etc. Symbolic pitch is often used. we simply write P itch|Z −→ Syn(P itch) since the other details are clear.H. for an onset with coordinate ON if the address is clear. In our situation. we specify the subspace38 S of @R and then write Onset|S −→ Syn(Onset). Duration. (6. But pay attention in distinguishing this sloppy notation from an equation between forms! If we want to restrict the space functors of such simple forms. In this notation.

musical tone parameters in the physical domain which correspond to the above mental parameters are required. RegDen-5. (6. Onset. though not so commonly used.80) (6. We denote them by the following self-explanatory forms: P hysOnset −→ Simple(R).86) This gives these product forms: AbsDyn −→ Limit(DynSymb. onset. we may set P ause −→ Syn(Onset ⊕ Duration). Thus.6. Bar-lines are onsets with no other properties. Id P hysCrescendo −→ Simple(R).87) (6. synonymy guarantees the common ground of onset which yields the bar-line concept.90) and get an isomorphic form.4 that the quantitative meaning of absolute and relative symbolic dynamics is quite non-trivial.85) P hysP itch −→ Simple(R).84) (6. A pause is determined by its onset and duration. This type of sophistication is speciﬁc to the formalism of forms and allows musicologically adequate constructs.6. Onset) RelDyn −→ Limit(DynSymb. Duration) (6. and duration parameters.82) (6. Id P hysGlissando −→ Simple(R). For symbolic absolute or relative dynamics such as “mf” or “crescendo”. we have to take a dynamic symbol (word).89) Id and the reader will recognize that we could have set BarLine∗ −→ Simple(R).91) . the only speciﬁcity is the name.88) for absolute and relative dynamics. REGULAR DENOTATORS 81 RegDen-3. Id Corresponding restrictions as shown in the mental situation above are obvious.81) (6. However. Id (6. Id P hysLoudness −→ Simple(R). we therefore reject the unspeciﬁc form BarLine*. Id (6. RegDen-6. First. Id (6.83) (6. We shall see in section 18. So we have BarLine −→ Syn(Onset). RegDen-4. deﬁne the form for dynamic symbols: DynSymb −→ Simple(Z ASCII ) Id (6. Id P hysDuration −→ Simple(R). For performance purposes.

It is. For example. a generic form of the following type OrchSet −→ F in(Z ASCII ) Ω@Z ASCII Power(InstruN ame) (6. . RegDen-9. . to the form P iM od12 introduced in (6. y. in section 8. O2 . Id (6. To denote instrument names.97) are evidently viewed as 12-pitch classes.3. e) means that the time signature x/y has been positioned at origin e.93) and we are in state of deﬁning an orchestral instrumentation n myOrchestra : 0Z n InstruN ame( i ai Instri ) (6.96) is adequate. we may construct a generic naming form InstruN ame −→ Simple(Z ASCII ). Id (6. Later (e. Whereas zero-addressed denotators P C12 (r) : 0Z P iM od12 (r) (6. a string quartet has the orchestral instrumentation denotator myStringQuartet = 2′V iolin′ + 1′V iola′ + 1′V ioloncello′. but leave it to the user to be more speciﬁc. however.95) If we do not want to specify one orchestral conﬁguration. see chapter 25 for a detailed discussion of this theory.82 CHAPTER 6. The time signature x/y is deﬁned by numerator x and denominator y and has a determined onset origin. They were extensively studied by Thomas Noll [400] in the frame of generalized function theory. with ai copies of instrument named Instri . “self”-addressed denotators [a.99) . The ﬁrst small example relates to pitch modulo octave.98) (6.42).94) or shorter myOrchestra = i ai Instri . not a mathematical fraction. RegDen-8.g.e. (6.92) where a denotator T imeSig(x. DENOTATORS RegDen-7.4). b] : Z12 P iM od12 (ea · b) (6. more generally addressed denotators deserve special interpretation. usually identiﬁed with the ﬁrst bar-line onset. This study concentrated on Z12 . we shall devote in depth discussions to this subject. since the denominator relates to metrical meaning. i. . Onset). Ok } consisting of k orchestra conﬁgurations.resp. A zero-addressed denotator then is a ﬁnite set myOrchSet = {O1 . We thus may set the product form T imeSig −→ Limit(Duration|Z . Let us give two “small” examples for non-zero-addressed denotators. Duration|Z .

Later. i. We have an injection !@Z12 : 0Z @Z12 Z12 @Z12 which induces P C12 (r) →[r. We shall see there that self-addressed pitch denotators appear also in the context of Euler modules (6. y/x). b ∈ Z12 . y/x). the second is a kind of onset diﬀerence. (b.26). more precisely: int((a. v)(a. and the law is (b. ea · b ∈ Z12 @Z12 .e.e. The ﬁrst coordinate is a zero-addressed onset in Onset. (6. REGULAR DENOTATORS 83 for a. In other words.e. (b. x). the above interval action will be recognized as an address change.6. x) are just self-addressed onset denotators t : R Onset(t).102) .100) (6. The second small example relates to the onset space Onset and its rational restriction Onset|Q as considered in the theories of David Lewin [300] and Dan Tudor Vuza [552]. y)) = ((b − a)/x. Self-addressed pitch classes generalize the 12-pitch classes by the unique projection ! : Z12 → 0Z of Z-modules. These are pairs (a. i.2).6. (b. This change-of-address technique is the germ of a far-reaching generalization of harmony. we should rewrite it under the form eb · y = ea · x · ei · p = ea+xi · xp. This is the known law of composition of aﬃne transformations. i. y)). In other words. x). pitch classes are interpreted as constant self-addressed pitch classes. (b. as denotators of shape [a.e. x). But this is clear since eb · y = ea · x · ei · p EulerM odule(ea · b) (6. y)) = int((u. v)(b. Lewin’s time spans t = (a. but viewed as a multiplicative quantity of relative increase. y)) is invariant under time dilatation is a special case of the fact that for any aﬃne automorphism f = eu · v : R → R of the ambient space R of the Onset form. x)(i. 0]. see chapter 24. Given two time spans (a. they are compared by the “interval” which connects them. x) ∈ R × R+ in the sense of time objects. x). when we know morphisms between denotators (8. xp). the ﬁrst time span (a. y). p) = ((b − a)/x. x). Lewin considers time spans. In [300].101) In other words. p) = (a + xi. b] : Q3 with ea · b ∈ Q3 @Q3 . the function int((a. (u. the interval does not change if applied to the transformed objects. whereas Lewin’s remark that the interval int((a. x) is transformed into the second one (b. i. y) = (a. y) by an interval int = (i. The structural meaning of this object is best understood via the transformation rules set up by Lewin.

(6. we can introduce the corresponding physical sound form by F ourierSound −→ Limit(D). P hysOnset = e. To describe a Fourier decomposition of a periodic sound of frequency f . . we may consider the simple form F ourier −→ Simple(R ⊕ R2×N ). 1] and set Envelope −→ Simple(C 0 [0. 1] of continuous functions on the real unit interval [0.106) For Envelope = V . For example. RegDen-10. essentially ASCII words.108) 39 With the necessary convergence conditions [127]. the instrument names are only symbols. Combining elementary data such as onset.105) of physical time t. F ourier. Id (6.84 implies CHAPTER 6. Fourier synthesis. and that the physical nature of the denoted instruments are not included.103) We then have zero-addressed denotators myF ourier = (f. duration. P hi )0≤i ) which represent the periodic sound function39 ∞ (6. Often. see ﬁgure 6. RegDen-11.4. Notice that denotators do not represent the physical meaning of an information unit. DENOTATORS f · eb · y = f · ea · x · ei · p. (Ai . Id D = (P hysOnset. for all automorphisms f .10. we may use the real vector space C 0 [0. RegDen-12. 1]).104) sd(myF ourier)(t) = 0≤i Ai sin(2πif t + P hi ) (6. Anticipating the notation of morphisms between denotators in (8. Envelope). e + d] and has its support between e and e + d if V has its support between 0 and 1.2). The transformation from “symbolic” to physical or technological meaning relates to the paratextual signiﬁcation processes described in section 18. the above equation rewrites as 1/int(f eb · y) = f ea · x. P hysDuration. one also normalizes a periodic sound function to a maximum value M ax(sd(myF ourier)) = 1. The same holds for the F ourier form.d (t) = V ((t − e)/d) for t ∈ [e. To denote an envelope function on the unit interval. and envelope. the required equation with transformed objects and the same interval int = ei · p. Id (6. e + d].107) which vanishes outside [e. and non-vanishing duration P hysDuration = d we obtain the deformed envelope Ve. 0 else (6.

56). A denotator myF ourierSound = (e. An M -addressed denotator myBook of .109) 6.10: Aﬃne deformation of an envelope denotator V on the unit interval.d · sd(myF ourier).d t 0 1 e e+d t Figure 6.7 Circular Denotators Summary.7. In this section we shall take care of a more eﬃcient way of conceptualization: circular forms and denotators. CircDen-1. myF ourier. This second set of denotators is devoted to important constructs of musical objects which are the foundation of powerful musicological concepts. The inductive repetition of partials was not quite the most economic way to think of forms. d. V ) is associated with the physical sound function sound(myF ourierSound) = Ve.d “starts” at onset e and “ends” at onset e + d. The deformed envelope Ve. (6. They make use of circular deﬁnitions which do reuse forms or denotators on diﬀerent recurrent coordinator levels. Recall the book form discussed in (6.110) (6. it was felt that there was a sort of superﬂuous conceptualization. CIRCULAR DENOTATORS 85 V Ve. –Σ– Already for the example of Fourier denotators. More precisely we have Deﬁnition 7 A circular denotator is a denotator of a circular form.6.

DENOTATORS myBook : M book(myN umber.2 for historical and technical discussion) which looks much more complicated than the Fourier situation if . Let us have a second look at the Fourier form. (6. we have to reenter in the book form until we reach the situation where the chapters build the empty set.112) ccsd(myCcF ourier) = 0≤i Ai sin(2πfi + P hi ). title. . . myN umber : M N o.111) We know from 6. where F u = F un(CcF ourier).). myT ext : M text(T x : M → Z ASCII ). and that for each collection of chapters. P hi ))0≤i . and P artial −→ Simple(R3 ). CcF ourier). myT itle. This means that it is not necessary to refer explicitly to the space functor which guarantees the form. P = F un(P artial). the numerical information is the same as for the non-circular concept of the Fourier decomposition. the book is completely determined. And this is what we know from category theory. chapter1 : M book(. We just have to know that it exists.2. fi . the circular form is much simpler—once we have settled the functor existence question. . The inﬁnite partials can also be deﬁned by the following product construction: CcF ourier f :F u→P ×F u −→ ∼ Limit(P artial. Thus. the naive setup is justiﬁed by the categorical background for determined classes of forms which are covered by the categorical existence theorems. . This means that we have to specify number.(N : M → Z). .113) For the special values fi = if we get back the classical case.. and this can be interpreted by the Fourier expansion ∞ (6. We want to look at the FM synthesis (see appendix A. CircDen-3.1. Of course. .86 this form looks as follows: CHAPTER 6. Id (6. myChapters). myChapters : M chapters(chapter1 . chapterk : M book(. and text explicitly. but conceptually. .. CircDen-2. .). Denotators myCcF ourier of form CcF ourier are identiﬁed by sequences myCcF ourier = ((P artial(Ai . myT itle : M title(T t : M → Z ASCII ).47 that such functors exist. myT ext. chapterk ). If we know this recurrent information until its empty roots.

CIRCULAR DENOTATORS 87 deﬁned without circular forms. For each F M -Object denotator myF M Object. etc. The point is that the modulator typology is not ﬁxed and may vary from stage to stage.F.sin(2pF+Ph+?) (A.114) Id We need to solve the functor equation F → F in(P × F ).7. a solution F always exists. We may. it essentially consists of a sinoidal “carrier” function which in its argument adds other sinoidal “modulator” functions which in turn refer to other modulator functions. and then an arrow from each knot vertex to all the knot vertexes of its modulator. F M -Object). however.6. FM-Object S Knoti {} Knot modulator p carrier ∼ FM-Object Partial Simple A.2. Let us ﬁrst deﬁne the FM form and then make a picture of the form. . The form is called FM-Object and looks as follows: F M -Object ∼ −→ ΩF K Power(Knot) with f :F →F in(F K) F = F un(F M -Object). but according to proposition 103 in appendix G. its graph Γ(myF M Object) is deﬁned as follows: Draw a vertex for each knot. F K = F un(Knot) and the product form Knot −→ Limit(P artial. (6.11: The graphical representation of an FM-Object denotator. introduce directed graphs (so-called “algorithms” (A. etc.2)) but the spirit is more obscured than evidenced by such a construction. by induction (including circularity!). In fact.Ph) Œ —3 Figure 6.1.

etc. . . . Recursion here means that after a ﬁnite number of downward steps from FM-Object factor to FM-Object factor. .. To begin with..g. We only have to replace the partial by an event of a particular type. . a vector space R4 of the piano events).88 CHAPTER 6. and the sound function is well-deﬁned. myM akroi for Knoti . . However. . Makro coordinates myEventi resp.118) . we end up with the empty factor. We introduce a ﬂattening operation and deﬁne the M akroBasic -formed denotator F latten(D) = {D}. let Basic be a form which describes a sound event type. Basic). CircDen-4. . These forms follow the same schema as discussed for FM-Object forms. for example the piano speciﬁc event type Basic = Onset ⊕ P itch ⊕ Loudness ⊕ Duration introduced in (6.117) The general deﬁnition of the ﬂattening operation for a denotator M akroBasic = myM akro = {D1 . In music.115) where myM odulatori is the FM-Object factor of Knoti . myM akro) with non-empty Makro coordinate myM akro = {Knot1 . the sound function is recursively deﬁned as follows: F M sound(myF M Object)(t) = n Ai sin(2πFi t + P hi + F M sound(myM odulatori )(t)) i (6.g. Knotr } and knot events resp. ∅) with the denotator myEvent in the event space. i = 1. Each knot has it partial denotator which deﬁnes the sinoidal function of the knot. Id (6. myM akroi ). .k F latten(Di ). each knot has its modulator which is another FM sound obtained like the basic carrier. inﬁnite descent may also yield well-deﬁned sound functions. let a denotator KnotBasic = D of form KnotBasic have coordinates (myEvent. Dk } is F latten(myM akro) = i=1. (6. We want to deal with this phenomenon in deﬁning Makro forms. F K = F un(KnotBasic ) and the product form KnotBasic −→ Limit(M akroBasic . for example when dealing with arpeggios or trills. we set F latten(D) = {(myEvent + myEventi .78).116) How must we interpret this construction to get ‘real’ events back from this grouping structure? Suppose that we have a module structure on the space of the Basic form (e. Starting from F M sound(∅) = 0. an FM-Object denotator myFMObject is a ﬁnite set of knots Knoti which form the basic carrier of the FM sound.. in R4 in the above piano form. under appropriate convergence conditions.11). In addition. Put generically. (6. r}. e. DENOTATORS What does this mean for the sound synthesis function? Intuitively (see ﬁgure 6. For a denotator KnotBasic = D = (myEvent. We then set M akroBasic f :F →F in(F K) ∼ −→ ΩF K Power(KnotBasic ) with F = F un(M akroBasic ). it is common to consider sound events which share a speciﬁc grouping behavior.

These tools are applied to representation and retrieval problems of general musical objects. (linear and partial) orderings on systems of denotators can be deﬁned. F latten∞ (myM akro).6. founded on these types of modules.8 Ordering on Forms and Denotators Summary. . –Σ– . .5.2. i.e. we may view this as a ﬁnite set of Basic-formed events. 6.8.12 for a ﬁrst intuition about Makro events. In a number of important foundation modules. Since each element of F latten∞ (myM akro) has empty Makro coordinate. These orderings induce canonical (linear and partial) orderings on denotators. If a Makro event myMakro ends up with empty Makro coordinates after a ﬁnite descent. We shall come back to this construction in section 21.3 and section 39. ORDERING ON FORMS AND DENOTATORS 89 Pitch Flatten once Onset Figure 6. The option is fundamental in deﬁning receptive navigation tools. becomes stationary after an n-fold concatenation: F lattenn (myM akro) = F lattenn+1 (myM akro) = . i. the ﬂattening operation also ends up.e. refer to ﬁgure 6..12: The ﬂattening of a macro-event in the mental onset-pitch domain. a local composition in the terminology to be introduced later in chapter 7.

12). we would like to interpret them as four-letter words T p Co Id N m with the lexicographic ordering41 as above.90 CHAPTER 6. Since names are strings of characters. So whenever we are given already ordered forms. we want to look for generalizations of the naive order relation among forms and denotators. example 64. BOOLE. see section 6. Next. The types are ordered by Simple < Syn < Limit < Colimit < Power. synymous naming is of course still allowed. we consider exclusively regular forms here.1 for such an ordering. As in the “naive discussion” in section 6. In this spirit. we have all the modules as coordinators. We suppose to be given such an ordering <Mod which extends the naive ordering. the letter of the ﬁrst form is smaller than the corresponding letter of the second.119) similar to the ordering deﬁned in (6.1. we had F1 < F2 iﬀ for the ﬁrst position in the “words” T p1 Co1 N m1 and T p2 Co2 N m2 where the letters diﬀer.8. But we do not want to do this in a completely arbitrary way. Functorial forms F = N m −→ T p(Co) Id are a bit more complicated. if two forms F1 = N m1 −→ T p1 (Co1 ) and F2 = N m2 −→ T p2 (Co2 ) were given. more precisely: on all morphisms of this category.1.. we need an order relation <Mod on these modules. The diﬀerence is that for Simple type.3.e. This is part of the basic. (6. and coordinator. This means that we are allowed to deﬁne order relation recursively on the level40 of forms. then the form names should also follow that ordering. The name N m poses no problem. Recall from section 6. we want to look at the identiﬁer. diﬀerent forms must bear diﬀerent names. Cm of forms in the compound case and the respective module name in the simple case. Z2 <Mod Z <Mod Z ASCII <Mod R. i. if the ﬁrst three letters are ﬁxed. i.120) In other words: The simple forms are ordered according to their coordinator modules. DENOTATORS In this section. . and R. the ordering must be read from the forms’ names. we only had four modules CHR. INTEGER. they are linearly ordered. thereby englobing the modules as identities. . i. purely module-theoretic “level of reality”. Recall that we require that form names be a key to forms. (6. This is supposed to extend the “naive” ordering. Z. whereas in the naive case. we have the lexicographic ordering on usual ASCII strings. BIT. Let us inspect the diﬀerent orderings on the four positions of T p Co Id N m. This is the general policy behind the following procedure. The form names should reﬂect the recursively given orderings of the form components: type.3 that naive forms were ordered lexicographically as three-letter words. Since we now have all the modules of Mod as coordinators of simple forms. We want to deﬁne orderings on forms relating to their names in a generative way.5. identiﬁer. and we may order forms according to their names. lexicographic orderings. and FLOAT..e.e. We understood that the coordinator was itself presented as a word C1 C2 . see appendix C. supposing by construction that we have ﬁxed a form type T p and a coordinator Co. giving rise to the simple types STRING. However. . This means that the form’s identiﬁer is a monomorphism 40 See 41 Concerning deﬁnition 5 in section 6. The naming policy then means that if ever the three ﬁrst letters deﬁne an ordering among two forms.

. Cm ) was ﬁnite and linearly ordered by its indices. j) d d d is the same for both diagrams.2 .j .0 C2. the vertex forms are already linearly ordered by their names—let us index them for notational comfort by Ci . we further diﬀerentiate according to the involved diagram arrows. In view of the components N (D) : A(D) F (D)(CT (D)) of a denotator D.j in their lexicographic order. we have a similar situation as with forms. the coordinator is a form of smaller level. iﬀ there is a (necessarily mono) morphism g : F u1 F u2 such that f2 · g = f1 . Co) is the universal functor construction associated with T p and Co.1 .6. we set Id1 < Id2 iﬀ the subobjects verify f1 < f2 . . . The naive case appears here as related to the identity identiﬁer. addresses i.0 mor0. is reﬂexive and transitive.j : We build the words C1. and we are done with the ordering of forms. We are going to build a word with the morphisms f : Ci → Cj of our diagrams. that the latter relation is neither antisymmetric nor total in general.8.i and C2. We again order denotators lexicographically according to the four-letter wording F (D)A(D)CT (D)N (D). forms. If the ﬁxed type is either Syn or Power. . modules. in the given double index ordering: mord = mor0. It is 1 2 now clear how to compare the words mor and mor lexicographically.j. .e. If we have two forms of ﬁxed form type T p and coordinator Co. . i. and we are done by recursion. Cm of forms.e. one should however respect the identiﬁer orderings. 2. Comparing diagrams in the naive case was reduced to the lexicographic comparison of words C1 . We further suppose that the set of morphisms f : Ci → Cj is linearly ordered for all vertex indices i and j. d = 1. Fix a pair i. we suppose that • by recursion. and names are already ordered.j = fi. .e. Observe. We are left with the ordering of the coordinator in case it is not a module (compound type). and therefore suppose that the arrow number s(i. where F u = F un(F ) and the frame space U f = U ni(T p. i. To begin with. . see appendix C. . We may thus suppose that V (D1 ) = V (D2 ) = (Ci )0≤i . Upon this ordering. however. and therefore. we have two monomorphisms f1 : F u1 U f and f2 : F u2 U f and these are canonically compared by the subobject relation42 . and compare them lexicographically. ORDERING ON FORMS AND DENOTATORS 91 f : Fu U f . . 2 d d d whose letters are the mori. it is only a partial ordering43 . both elements of 42 See 43 It appendix G. j of vertex indices are ordered lexicographically. sort these words by the number of their arrows. The naive setup was concerned with discrete diagrams where the vertex family V (D) = (C1 . This suﬃces to terminate the construction. we can build a word for each diagram Dd . This is always possible since the form genealogy is a question of ﬁnitely many forms. be the word of diagram morphisms d fi. We are thus left with a diagram coordinator D. . . According to the lexicographic ordering.2 . .1 . When generating new form names. mori.s : Ci → Cj .j . . Since by the preceding. and C2. To generalize that situation. we may suppose that identiﬁer ordering is carried over to name ordering. Since the couples i. s = 1. Let us now look at possible orderings among denotators. Let mori. . we are left with the question of ordering coordinates.1 . j of vertex indexes. .j.0 C1. d = 1. 1 1 We may now arrange the words mori. s(i. we may thus look at two denotators D1 and D2 with identical form F and address M and with coordinates Co1 and Co2 .j.1 fi. . where the ordering among the forms Ci was supposed to be given by recursion. j) in the given linear morphism ordering of diagram Dd . We are therefore able to compare two such diagrams D1 and D2 with respect to their vertex forms C1.

For the frequently used subfunctor F in(F un(C)) ΩF un(C) . we deﬁne the ordering on the colimit set as the partial ordering generated by the image relation of the coproduct’s ordering.2. the subobject ordering can be reﬁned to an ordering which is also linear in case the ordering on F un(C) is. we have a surjection M @F un(Ci ) i M @Colim(D) deﬁned by the equivalence relation that is generated by the diagram’s arrows. If the contrary is not stressed we shall always use the ordering on F in(H) for a functor H ∈ Mod@ whose value sets are ordered. we know from appendix G. in other words. we have M @ΩF un(C) = Sub(@M × F un(C)) which is naturally provided with the ordering among subobjects as already discussed above. i. Whether this ordering on F in(F un(C)) can be extended to a partial ordering on all of ΩF un(C) that includes the subobject ordering. Since the latter functor is identiﬁed to a subfunctor of the universal construction U = U (T ype.92 CHAPTER 6. and M @ΩF un(C) for a module N . under the hypothesis that the index set is also linearly ordered. The left set carries a canonical ordering as described in the naive discussion in section 6. and we are done by recursion. If we are to carry over this data to the colimit. However. or a coordinator form C.2.1 that M @Lim(D) ⊂ i M @F un(Ci ) with the above notation of the diagram’s vertexes Ci . DENOTATORS the functor value set M @F un(F ). We have to investigate the situations M @N . in general! This terminates the construction of orderings among denotators as far as this is possible. For the power type ΩF un(C) . M @F un(Syn(C)). this ordering is not related to the recursively given ordering on (the value sets of) F un(C). In fact. and we may consider the lexicographic ordering on the product.e. a diagram D. For Syn type. we have to turn the above projection into an order-preserving map. is an open question. Coordinator) of the form. The ordering reﬁnes the inclusion ordering among ﬁnite subsets. by proposition 65 in appendix C. For Limit type.1.3. restricted to the limit subset. M @Colim(D). We deﬁne it by use of the linear ordering among the coordinator forms of D. In case of simple type. and by the recursively given orderings on the functor sets M @F un(Ci ). In fact. the natural inclusion F in(F un(C)) ΩF un(C) with the subobject orderings admits a reﬁnement on F in(F un(C)) which is linear if the ordering on F un(C) is. The functor values M @F un(Ci ) are ordered by recursion. We are left with the Colimit type which is less comfortable to handle than the other types. M @Lim(D). M @N bears an ordering by hypothesis on the category Mod. every set A@F in(F un(C)) is ordered by a partial/linear ordering induced from the partial/linear ordering on A@F un(C). see example 64 in appendix C. we have M @F un(Syn(C)) = M @F un(C). But observe that this will—in general—be something much larger than the image relation since the latter is not transitive. we can restrict our question to orderings on value sets of such universal construction functors. .

This is not so much a purely mathematical exercise. When building compound forms and denotators on these forms. it is not necessary to care about its order position. we start the skeleton by the few examples of hitherto basic modules and 44 See appendix G.8. –Σ– Modules and Morphisms. In this sense we should also understand the following set of applications. After all. This perspective stresses the “natural” ordering and not an artiﬁcial one which may be built exclusively upon the axiom of choice. example 87. and then the objects within each isomorphism class.e. In other words. i. . we may apply existing orderings on the already used modules for the compound structures. The only presence of choice in our context is the historical appearance of concepts. the recursive construction of partial orderings is also dynamic. This is taken into account on the level of the ordering within the fundamental category Mod of modules and aﬃne morphisms. it should persist for ever. The only essential point here is that one should not reorder modules after dynamical extension: Once an order relation M <Mod N is deﬁned. as an ordered coproduct of ordered isomorphism classes. Before we really need a module. For the purpose of ordering modules.8. of forms and denotators. 6. For example. mathematicians know that any set can be well-ordered.6. The point is much more a constructive approach which lends itself to data base management systems and the related successive growth of denotator arsenals—as a function of a dynamically expanding knowledge base. we want to observe the following principle: Principle 1 Order ﬁrst modules from a skeleton44 of Mod. We give concrete order relations and applications to musical situations. Otherwise. in the development of theory and empirical material. Then order modules lexicographically according to the skeleton and the member of the isomorphism class. Only if a new module intervenes in the knowledge extension process is it necessary to insert it relatively to the already ordered module set. but see chapter 18 for an extensive study of “what is the case” and what is not. i.1).e. Modules in Mod appear in two functions: as coordinators of simple forms and as addresses of denotators. ORDERING ON FORMS AND DENOTATORS 93 We should make a ﬁnal comment on the sense of ordering forms and denotators. reusable retrieval on database management systems break down. We start this section with a discussion of possible order relations on the the category Mod of modules and their aﬃne morphisms. at least in acceptance of certain common axioms (see appendix C.1 Concretizations and Applications Summary.

<Mod Z <Mod Z ASCII <Mod Q <Mod Q2 . A-addressed denotators D : A F ×G(Co) of F ×G with Co ∈ A@M ×A@N are ordered lexicographically following A@ <M and then A@ <N . . (6. . As usual in linear algebra. DENOTATORS ∅ <Mod 0Z (= Z1 ) <Mod Z2 . We have to select speciﬁc modules and give them individual orderings (without regard to compatibility with aﬃne transformations such as automorphisms).e. i. (6. a minimum of universality should be observed since one should not produce more incompatibilities than necessary. the vectors of this module are viewed as columns. and this includes ordering sets of aﬃne homomorphisms between modules which are also. .94 order them as follows45 : CHAPTER 6. <Mod Zn <Mod Zn+1 . under certain conditions. according to the preceding discussion. N with respective orderings48 A@ <M .4. <Mod Rn <Mod Rn+1 . . . . For example. 45 In this sequence. So we have to take care of orderings on modules M and their direct sums ﬁrst. n × 1 matrices. . <Mod R <Mod R2 . A@N whenever we construct an ordering on a direct sum A@(M ⊕ N ). . . inserted between (classes of) modules Rn−1 and Rn+1 : R1×(n−1) <Mod Rn×1 <Mod Rr1 ×s1 <Mod . 0@M identiﬁes to M . . that switching from Limit to Simple type changes the order position of the underlying form. . R1×n <Mod R(n+1)×1 .1 should be compatible with order relations47 . i. 48 The following symbolism A@ < for an ordering on a value set A@F of a functor F in Mod@ does not mean F that the ordering is functorial. Ordering on individual modules is quite problematic since nothing functorial can be expected.122) Before further discussing orderings on forms we have to look at orderings on individual modules since orderings on diagrams depend on orderings on sets of morphisms between functors. To begin with.e. <Mod Qn <Mod Qn+1 . we have to order coeﬃcient rings49 . In this spirit. . 46 see appendix E for more details on this somewhat delicate subject! 47 Observe. However. . if these denotators are identiﬁed with A-addressed denotators in F ⊕ G. and then. we look at a ﬁxed positive power Rn . we should order the direct sum A@(M ⊕ N ) by the lexicographic ∼ ordering of its summands under the canonical bijection A@M ⊕ N → A@M × A@N .121) To give an example of the ordering within an isomorphism class. A@ <N Id at an address A. modules46 . . we arrange the factorizations n = r × s of n by increasing second factors 1 < s1 < s2 < . but n n [Z] we avoid this clumsy notation whenever possible. i.e. the identiﬁcation of forms related to direct sums of modules in point 5 of section 6. however. Then. the modules Z are viewed as Z-modules. . We thus write n = n × 1 and identify Rn with Rn×1 . n and obtain an ordering among the following matrix R-modules isomorphic to Rn . the functor does not necessarily evaluate on the category of ordered sets. . The common case is zero address. we should write more correctly (Z ) . 49 Mathematicians should pay attention not to confuse the orderings in this section with the ordering concept for ordered rings. Hence. . Suppose therefore that we have a product form F × G of two simple forms F −→ Simple(M ) Id and G −→ Simple(N ) over modules M. We therefore have to take the lexicographic ordering from the summands A@M.

N ) with Rr×s ⊕ Rs . Generalizing the above case with Z ASCII . p = prime and k = 1. instead of R. With the previous constructions we are in a state of ordering diagrams of forms having functors which are represented in free modules over rings which already carry an ordering. 3. Z. 50 See 51 By appendix E. • For general n. Ordering on Morphisms. . Zpkr . . r 1 On this direct sum. then the lexicographic ordering as deﬁned in appendix C.122).8. as well as for each index i ∈ I a module Mi with an ordering <i . Matrix Modules. via the identiﬁcation50 of Z-modules Z ASCII = Z(W ) . A and M . instead of R. : i < j iﬀ i < j. . Yoneda’s lemma. and this will be done in the canonical way: Identify Rr×s canonically with (Rr )s . . Here. The same ordering procedure works for any matrix module over a ring R which is provided with an ordering. Direct Sums. well-ordered by <. we suppose given a family of indices I. . Then we have the Sylow ∼ decomposition (appendix C. The latter are strictly related to the identity and construction of speciﬁc modules. Q. . In particular. we have ordered isomorphic copies of Rn for diﬀerent factorizations n = r × s as matrix modules Rr×s . ordered as a direct sum of the ordered module Rr×s with the ordered matrix space Rs . see appendix G. then the morphisms F → G correspond canonically51 to aﬃne morphisms en ·f : M → N . Again we should pay attention to the fact that isomorphic modules may not have isomorphic orderings.2. we take the lexicographic ordering based on the previous deﬁnition of the orderings on the direct factors. And the last construction yields orderings on functor values A@M at address A if both. Then the direct sum I Mi is ordered lexicographically according to the index ordering. • R: usual ordering. let n = pk1 · · · pkr be the (unique) prime decomposition with increasing r 1 prime factors and non-zero exponents (see appendix E. we canonically identify M @N = eN · Lin(M. pk − 1}. In (6. • Z: usual ordering. 2. R = Zn . if M and N are the vector spaces M = Rr and N = Rs .2). example 64.121) and will now deﬁne orderings on these prototypes: • Z0 and Zpk = {0. • Z ASCII : Take the lexicographic ordering on the basis set W of ASCII-words. If we are given two forms F and G with functors which are represented by modules. • Q: usual ordering.6. are free over the coeﬃcient ring which is provided with an ordering. The same construction is valid for any free modules M = Rr and N = Rs over a ring R which is provided with an ordering. order the column summands Rr and then the total sum as deﬁned above for direct sums.4) and a canonical isomorphism Zn → Zpk1 ⊕ . ORDERING ON FORMS AND DENOTATORS 95 We have already listed some important rings in (6. . . we have to order these copies themselves. F u(F ) = @M and F u(G) = @N . . This includes the ﬁnite direct sum Rn of n copies of an ordered ring. R for example.3.

96 CHAPTER 6.5 for chord classiﬁcation.1) (1. see section 7. Example 4 Ordering Note Groups from a “Chord Perspective”. We 52 See section 11.0) (2. as shown in ﬁgure 6. This means that we do not only consider them as sets of tone events. If instead we look at self-addressed denotators of P iM odn . .13. Its A-addressed denotators live in the set A@Zn and we have the ordering of Zn on the zero-addressed denotators. we have the self-addressed tones studied in [400].2. e11 · 11. Intuitively speaking. x mod 3). we have to consider the elements (b.1) (2. For n = 12. we have the prime decomposition 12 = 22 3 and therefore inherit the lexicographic ordering of Z4 ⊕ Z3 via the canonical bijection Z12 → Z4 ⊕ Z3 : x mod 12 → (x mod 4. .1) (0. In case of n = 12. but with respect to their harmonic or metric content. Let us develop on this scope by use of forms.0) (3. this look like four arpeggiated augmented triads52 . . . the last being the inversion between B and C. orderings on forms and the corresponding denotator systems. e11 · 0 and ends with the inversion tones e0 · 11 < e4 · 11 < .2) Z12 : 0 < 4 < 8 < 9 < 1 < 5 < 6 < 10 < 2 < 3 < 7 < 11 which starts from 0 with two major third steps.1) (3. But we may also look at these objects from a harmonic or metrical point of view. .0) (0.2) (2.0) (1. it is a good exercise to apply the mathematical constructions to concrete musical examples and to give them interpretations in terms of musical meaning. This gives the following ordering on zero-addressed denotators Z4 ⊕ Z3 : (0. Consider the form P iM odn of n-pitch classes as deﬁned in (6. For the sake of transparency we only look at two parameters. a) of the ordered module Zn ⊕ Zn which parametrize the aﬃne morphisms ea · b. The ordering thus starts with the so-called constant self-addressed tones e0 · 0 < e4 · 0 < .1. DENOTATORS This being true. If played in notes from middle C on.2) (1. ∼ & X X # X # X X X X # X # X # X nX X Figure 6. Example 3 Ordering Zn . we may look at (ﬁnite) sets of notes on a score and try to order them for lexical retrieval or/and for listing them in a data base management system.41).2) (3.3.13: The ordering on 12-pitch classes according to the standard procedure described in this section. then jumps by one semitone and repeats the major third steps from there etc. onset and pitch.

128) (6. Id Onset ⊕ P itch −→ Simple(R2 ).126) (6. We order the points as follows: • ← •.125) (6. Id Id (6. For the ﬁrst form L(Dp1 ). D1 ∈ A@F in(@R ) = F in(A@R ). C1 = F in(Onset ⊕ P itch) in Dp1 and C0 = F in(P itch).127) −→ F in(@R) F in(Onset ⊕ P itch) Power(Onset ⊕ P itch). F in(@R2 ) Corresponding to the above projections and by lemma 90 in appendix G. we ﬁrst need an ordering on the diagram’s vertexes and arrows. D1 ) with these speciﬁcations: D0 ∈ A@F in(@R) = F in(A@R).130) (6.2.134) . −→ Ω@R2 (6. we have 0 1 1 2 2 1 C0 = F in(Onset).1. A@F in(p1 )(D1 ) = D0 . We then have limit forms deﬁned by these diagrams: L(Dp1 ) −→ Limit(Dp1 ). Let us consider order relations of denotators in this context. C1 = F in(Onset ⊕ 2 2 1 1 P itch) in Dp2 . we have two morphisms of forms p1 : Onset ⊕ P itch → Onset and p2 : Onset ⊕ P itch → P itch. 2 2 (6. Id Corresponding to ﬁrst and second projection p1 : R2 → R and p2 : R2 → R. and then it is clear.132) (6. (6. an A-addressed denotator D ∈ A@L(Dp1 ) looks like this: We have D = (D0 . the latter diﬀer only by their names. Power(P itch).8. in fact. According to the general theory.e. Our diagram scheme is just two points and one connecting arrow.6: Onset −→ Simple(R). (6. F in(p1 ) : F in(Onset ⊕ P itch) → F in(P itch).124) Recall that F un(Onset) = F un(P itch) = @R and F un(Onset ⊕ P itch) = @R2 and consider the forms of ﬁnite sets in the three preceding forms: F in(Onset) F in(P itch) −→ F in(@R) Ω@R Ω@R Power(Onset). Therefore. and therefore denotators on L(Dp1 ) precede denotators on L(Dp2 ). the vertex words satisfy C0 C1 < C0 C1 since forms verify Onset < P itch. we have two projections F in(p1 ) : F in(Onset ⊕ P itch) → F in(Onset). ORDERING ON FORMS AND DENOTATORS need the three corresponding forms introduced in section 6. i. This implies that L(Dp1 ) < L(Dp2 ).129) and corresponding diagrams Dp1 and Dp2 .123) (6. Id 97 P itch −→ Simple(R).131) L(Dp2 ) −→ Limit(Dp2 ).6.133) (6.

Its two one-dimensional projections onto the Onset and P itch forms reveal metrical/rhythmical and harmonic aspects of the plane denotator. respectively. 28. 6 71 67 65 64 62 59 p2 Onset ª Pitch 55 50 43 Pitch 0 1 2 3 4 5 6 7 8 9 10 p1 Onset Figure 6. 28.98 CHAPTER 6.14: A bar of Fr´d´ric Chopin’s Pr´lude op. No. No. . DENOTATORS X X X X X X X X X X X X nX XX # X ?# X & ## Frédéric Chopin Prélude op. 6 is projected to an Onset ⊕ P itche e e formed denotator in the plane.

9. We have therefore two denotators from one subset D1 in the onset-pitch plane. The second tells about the pitch conﬁguration of D1 via second projection p2 (D1 ) to the pitch axis. Each of these orderings happens to be the type deﬁned according to proposition D2 65 in appendix C. it is not necessary to misrepresent or over-interpret the pitch data of a note in an ethnological context in favor of the European lexemata of pitch when denotating an Indian melody. it is still possible to map the character strings to numbers.6. 6. i = 1. D1 ⊂ R2 such that D1 projects onto D0 when we forget about the Pitch coordinates. we have analogous explications). In this case. i i Consider now two A-addressed denotators Di = (D0 . whereas the second deals with its harmonic aspects. x2 ) ∈ (A@R)2 which stays in D1 − D1 .135) To understand the meaning of such denotators. Recall that D1 D1 means that on the onset axis either D1 ⊂ D1 or max(D1 − D1 ) < 2 1 max(D1 − D1 ). Correspondingly. or onsets verify x1 < y1 . and after they have been identiﬁed. There may be many diﬀerent conﬁgurations with the same metrical/rhythmic aspect. The possibility to apply character strings as coordinates opens the way to incomplete semantics. the denotator D ∈ 0Z @L(Dp1 ) consists of two ﬁnite sets D0 ⊂ R. let us—as usual—start with zero-addressed denotators and see ﬁgure 6. D1 ) stays in the same mathematical space but veriﬁes the relation A@F in(p2 )(D1 ) = D0 .9 Concept Surgery and Denotator Semantics Summary. According to the general 1 2 2 1 discussion of Limit and Power orderings. MIDI numbers or whatever may . the ﬁrst deals with metrical or rhythmical aspects of D1 . Roughly speaking. y2 ) in D1 − D1 such that either its onsets coincide and pitches verify x2 < y2 . in the second form L(Dp2 ).14 for an illustration. 2 of form L(Dp1 ) (for the second projection form L(Dp2 ). a denotator D ∈ 0Z @L(Dp2 ) consists of two ﬁnite sets D0 ⊂ R. D1 ⊂ R2 such that D1 projects onto D0 when we forget about the Onset coordinates. What does it mean that D1 < D2 when their metrical/rhythmic aspects coincide? We then 1 2 1 2 have D1 = D1 and D2 D2 . This shows that denotator semantics must be understood as an open-ended speciﬁcation. we have D1 < D2 iﬀ either D1 D1 or D1 = D1 and 2 1 D2 . when building a pitch model for that context. –Σ– Denotators are very useful for formally repertorizing ethnomusicological data and facts. a denotator D = (D0 . So we are left with comparison of denotators within one ﬁxed form.2. Our order principles imply that metrical or rhythmical aspects precede harmonic aspects if ever we are going to allocate such denotators in a library or in a data base system. there is 2 1 y = (y1 . However. This means that we ﬁrst look at the ordering among metrical/rhythmic aspects of denotators. we turn to the tones behind these 1 2 1 2 1 2 aspects. For example. where the latter is induced by the lexicographic ordering on 2 1 2 A@R . Denotators can be used as dynamic tools for concept production. such as European pitch classes. D1 ). This means that for any x = (x1 . (6. CONCEPT SURGERY AND DENOTATOR SEMANTICS 99 Correspondingly. The ﬁrst gives information about the relation between the onset conﬁguration of D1 via ﬁrst projection p1 (D1 ) to the onset axis.

e. such a morphism can be deﬁned if we know just what happens on the zero address. Suppose that the symbols we are interested in stem from the set CHR of ASCII-character strings. This substitution process is understood as yielding a ‘decoding’ method for ‘deepening meaning’ of signs. i. with an intrinsic character of a non-resolved semantics.. AR @ R = R R. Whereas a European score symbol for dynamics is recorded as a character string coordinates by ′mf′. for any address change c : B → A over the scalar restriction γ : S → R. Since we have a unique address change ! : B → 0S of S-modules. Id Giving the symbolic denotators of form EuroDyn a more concrete meaning in terms of form Loudness means deﬁning a “meaning” or “modeling” morphism of forms m : EuroDyn → Loudness. we get the map B@m by commutativity of the diagram S R × CHR −S − 0S @R − → − = R × CHR − − → B@R −− B@m 0 @m !@R (6.e. i. we must have a commutative diagram R γ R × CHR − − → A@R −− R×Id B@m A@m c@R (6. A more concrete parametrization of loudness would be the simple form Loudness −→ Simple(R) (6. This is given by circular forms! Let [CHR] be the constant functor54 of CHR and R the Yoneda functor of the ring R. This procedure may be generalized to the end of substituting character strings by entire denotators in order to ‘ﬁll up symbolic strings’ with more explicit denotators.. But until that realization. ′pp′.69). the denotators keep their full potential of “European ethnological context of notation”. applied to the coeﬃcient ring of a given address AR . DENOTATORS be relevant to the given scientiﬁc project53 . We want to consider European dynamic signs with a minimum of superﬂuous information. For example. quite the contrary: Without such an open denotator system. pp etc.1. more precisely. Let us explicate the technical aspects.3.138) S 53 This is precisely what the RUBATO analysis and performance platform makes possible in the case of European score denotators such as dynamics mf. the retrieval of ethnomusicological data would fail from the beginning since the original information would be distorted in an irreversible way if the record was meant to be a reliable ﬁrst order source. the user may interpret these coordinates by numerical evaluation for performance purposes. We consider the form ′EuroDyn′ Id −→ R×[CHR] Syn(EuroDyn). This means deﬁning a map A@m : R R × CHR → A@R which is functorial in address A over the ring R. To begin with.5 for details. see section 41. . This is no luxury for ethnomusicological research. We still stick to our EuroDyn example to make the general idea clear.100 CHAPTER 6. (6. one is looking for forms that express the above ‘symbolic’ representation of semantically open data.136) Next. let us consider the technical aspect of modeling this symbolic status on more concrete parameters. 54 See example 93 in section G.137) S R × CHR − − → B@R −− of set maps.

deﬁned by commutative diagram (6. ∼ But we have an isomorphism of functors R × [R] → 0? @R on the category Rings.6. we are given a morphism model : ArchaicF orm → M odelF orm. Denote by Dmodel the diagram deﬁned by this morphism.142) BetterF orm − − − − −→ G . model · g) of morphisms into Dmodel : G g − − → ArchaicF orm −− model pmodel g Better (6. BetterF orm −→ Limit(Dmodel ) Id (6. In other words. we should also be capable of making their signiﬁcation more precise on demand in the language of denotators. We want to complete this technique in order to have the generic procedure at hand. The set maps in proposition 1 are precisely what one would deﬁne in an ethnomusicological modeling of dynamics. Deﬁne a new form. (6. the latter evaluating to 0S @R at ring S. we easily see that m is deﬁned by the product Id R × m0 for a set map m0 : CHR → R. restricted to the zero addresses deﬁnes a morphism between set-valued functors on Rings. We want to start with the general situation of a given corpus of forms. This example shows that though it is necessary to work with incomplete concepts such as EuroDyn. Since the identity is the only endomorphism of R.140) related by model · parchaic = pmodel . We then replace every morphism of forms g : G → ArchaicF orm by the universal morphism Betterg : G → BetterF orm deﬁned by the pair (g. CONCEPT SURGERY AND DENOTATOR SEMANTICS 101 and therefore. But then.139) which displays two projections parchaic : BetterF orm → ArchaicF orm. the morphism m. we have this result: Proposition 1 The modeling morphisms m : EuroDyn → Loudness are in one-to-one correspondence with the set maps m0 : CHR → R. Suppose now that a particular form ArchaicF orm is connected to a “better” form M odelF orm by means of a “modeling” morphism.e. we replace every morphism g : ArchaicF orm → G of forms by its composition Betterg = g · parchaic . what happens is that we have mapped a circular denotator into a regular one and thereby deepened the semantic short circuit of circularity into regular meaning in terms of real numbers. any modeling transformation is determined by its data on the zero addresses. The functorial extension is for free in this case.9. i.138). All these forms are constructed by use of more or less large coordinators which are diagrams of forms if they are compound.141) BetterF orm − − → M odelF orm −− Similarly. pmodel : BetterF orm → M odelF orm. and we obtain a commutative (!) diagram H h − − → ArchaicF orm −− g Betterg h Better (6. What follows will be called concept surgery since it deals with replacement of given forms by “better” ones. built upon other forms. i.e. Conceptually.

for example.143) deﬁned by the surgically inserted BetterF orm from the model morphism. i. and emotional. Despite this semantic surgery. we have excised the ArchaicF orm from all its appearances in other forms. In fact. and replaced its role by that of a new BetterF orm. This is the reason why denotators must be deepened to yield this existential aspect underlying music. (6.e. meaning. resp. the denotator system has its very strict limits and requires a supersystem of connotation to grasp deeper layers of meaning. we may now replace every form OldF orm where ArchaicF orm is involved somewhere in its recursive tree by a new form model OldF orm. . In this surgical intervention.102 CHAPTER 6. denotators are not sensible to what really exists in music. denotators are far from realized in their complete potentiality. Beethoven’s Fifth Symphony is not just one of a large repertoire of given denotators! Its existence has farreaching consequences on the theory and practice of music and its meaning. Set more formally. DENOTATORS of “bad and better” morphisms of forms for any morphisms g : ArchaicF orm → G and h : H → ArchaicF orm. So this surgery makes it possible to improve the whole concept body of forms and denotators once one form has found an improved version. Moreover. Already with the above example of ethnomusicological signiﬁcation it became evident that the denotator system is only a signiﬁer surface pointing at interpretative. performative. social. this is an essential fact about music: In the real world. We shall develop this subject in full detail in chapter 18. as a member of the repertoire of a given culture. g · h = Betterg · Betterh which means that the “deviation” of morphisms to BetterF orm does not disturb any of the given factorizations through the old ArchaicF orm.

Part III Local Theory 103 .

.

The attentive reader could observe that the category of forms has been deﬁned. II. meters.e. . The chapter concludes with a discussion of tangent objects in local theory. ordinary and fractal chords.66] Summary. –Σ– The previous chapter was devoted to the universal concept and navigation formalism of forms. so elementar ich sie auch immer denke: ich bin bereits gezwungen. The next chapter will be devoted to a thorough discussion of the morphisms among local compositions. It is shown that all denotators may be transformed into local compositions. A special type of local compositions can be deﬁned from a ﬁxed set of denotators of a given “ambient space”. More downto-earth speaking. There are two directions of generalization which must be observed. rhythms. and motives—are presented. The basic vocabulary as well as an introductory list of common local compositions—such as scales. ) Pierre Boulez [60. This imposes a deeper theory of non-objective functorial local compositions. um sie niederschreiben zu k¨nnen o – im Grunde f¨llt der Lokalstruktur auch a eine alphabetische Rolle zu (. we want to set up the objects of the category Loc of local compositions. i. our attention to relations among denotators has not been set up.Chapter 7 Local Compositions Ich kann ein lokales Objekt. sie zu kodieren. p. With these so-called objective local compositions the problem of universal construction of new concepts from given ones cannot be solved. but no categories of denotators were considered. denotators and orders among these objects. Local compositions are introduced as elementary objects of music. of the type of denotators which are the core of musical conceptualization. It is however not straightforward in which generality such objects should be established. . a concept framework which leads to alterations and related results by Mason and Mazzola. denotators are—until now—entities which are only related by orders and not by more in-depth relations which take care of the precise positions of denotators within their space functors. In this chapter. They derive from powerset denotators. eine lokale Struktur nicht abstrakt schaﬀen. The ﬁrst one relates to the dichotomy 105 .

v. i. w) =(0.4. it is known from harmony of the 18th century that single notes were only harmonically signiﬁcant as material parts of triads. This chapter will only deal with the local aspect and introduce and justify the global one in chapter 13. v. we did already introduce the basic role of modules insection 6. We illustrate this rigidity requirement by an example. LOCAL COMPOSITIONS of ﬁxed versus variable addresses and has been introduced previously.1) (7. 1) corresponding to (0. −4. letters of the musical alphabet be reduced to single events. In this spirit. A .2) . triadic chords. 4.1 The Objects of Local Theory Summary. have to be linked in a ‘rigid’ way. For example. e. e. 7) corresponding to (0. the single sounds. w) but only the following relations among u.1. c ) (c. w. tones are “the material of music”. We agree with Boulez’ encoding requirement. B) (c. we will certainly require from an elementary system of sounds that it is viewed in a common context which enables us to determine the relations of the constitutive sounds. But it is not true that a singular tone is the only elementary unit which one will encounter. The second relates to the dichotomy of local versus global structure. A . In this context. p ∈ Z. intervals. We cannot prescribe that elementary objects.e. –Σ– According to Arnold Sch¨nberg [478]. What Pierre Boulez refers to in this chapter’s header is an alphabet of music. 7. −4. and w: |u − v| = 4. Then there are four solutions (u. The gestalt’s parts. Besides singular ‘points’. −7) corresponding to (c. We then consider not explicitly selected triples (u. a triad is determined by three pitch numbers u.106 CHAPTER 7. motives or rhythms can also be ‘elementary’ as musical units. But it should be recognized that single tones do not constitute what one would recognize as a musical unit. It represents pitch c by p = 0. o modules and the form space functors deduced by universal constructions are the mathematical spaces where this material is surveyed. i. 4. at zero-addressed denotators p : 0 EulerM odule(p/12 · o). encoded entities. Look at the 12-tempered tuning in the Euler module pitch space. etc. built from local. Musical material needs an elementary structuring in order to build elements of music. and u = 0. After a heuristic introduction. v.e. c by p = 1. In the language of gestalt psychology [136] one could restate our project in that “musical gestalts” should be deﬁned which are elementary in the sense that no relations among their material points remain indeterminate. g) (c. F ) (7. v. On the other hand. |v − w| = 3. objective and functorial local compositions are introduced. −1) corresponding to (0. and we contend that positioning musical material in the shape of a denotator-point within its form space functor at a determined address does meet encoding.

1).7. a local composition is a subfunctor x → @A×F un(S). we often write D : x → @A × S or D : X → A@S resp. it would be diﬃcult if not impossible to relate the concept of an elementary musical gestalt to relations as above in the general case. ﬁnite. Such a conﬁguration would not be termed “elementary”. However. w) whose ends are tied by hinge v whereas hinge u is ﬁxed at 0. the coordinates1 x is called its support. and the third stays three semitones from the second. We then obtain four solutions as shown in the quasi-mechanical ‘hinge’ system. The coordinator S of form F is called the ambient space of D. with x = X. In particular. The local composition D is called objective iﬀ x ∈ 2F un(S) i. . D = (x.2 for the concept of cardinality. Evidently.1. One only asks that the ﬁrst pitch be zero whereas the ﬁrst stays four semitones from the ﬁrst. with triads which verify our conditions (7. Intuitively speaking.1. Up to names. A@S). Non-objective local compositions are also called functorial. Deﬁnition 8 A local composition is a denotator D : A F (x) whose form F is of Power type.e. By these relations. iﬀ there is ˆ X ⊂ A@F un(S). inﬁnite. sometimes we also use the common language if the strict terminology sounds too far-out. It is composed of diﬀerent loosely related parts. iﬀ its cardinality is zero. the triad is not uniquely determined. ﬁnite. @A × S) and D = (X. it looks like a hinged mechanical system which consists of two rigid rods (u. The cardinality card(D) of an objective denotator D is the cardinality2 of its coordinates set X. or inﬁnite. THE OBJECTS OF LOCAL THEORY 107 7 4 1 0 4 0 0 -1 -4 0 -4 -7 Figure 7. 2 See appendix C. see also ﬁgure 7.1: Example of a non-elementary pitch conﬁguration in 12-tempered pitch (parametrized by integers) with non-ﬁxed pitches. v) and (v. and an objective local composition is a subset X → A@F un(S). A bit more sloppily. 1 Recall that “coordinates” is a plural in singular mode and as a such covers the muliplicity of coordinate values within such an object. an objective denotator D is called empty. Therefore. in this case X is also called a support of D. we shall restrict the concept of such a structure to the really ‘indecomposable’ conﬁguration of a local composition and introduce to non-elementary structures later.

but not in the much too narrow frame of simple pitch transposition. its science. they provide us with the basic vocabulary of music where concepts such as scale. Whoever tries to establish a standard of basic musicological concepts is unavoidably confronted with the ambiguity of the existing usage and with the problem of arbitrarily normative and materially overrestrictive or even dogmatic “standard language”. motives in the language of objective local compositions. But ambiguity of musicological concepts is not so much a defect but rather expression of a complex situation. A@S) seems to be quite poor. We shall introduce this requirement in extenso. 7. We introduce and discuss concepts of scales. the classical Ehrenfels criterion of “invariance under transpositions” [136] seems to be lacking. rhythms.108 CHAPTER 7. should be declared. the latter is a very elementary special case of general symmetries which are embedded in the concept of morphisms of denotators. but we shall not systematically exploit this rigid approach.1. and time-invariant laws of music. Not least is it historical development of music which does force an opening of the symmetry invariance. Before entering into the very generality of functorial local compositions. the construction of LocF is a kind of “generic” since any local composition of ambient space F can be derived from LocF by synonymy and subfunctors. LOCAL COMPOSITIONS If we are given a form F . chord etc. This impression is faulty and stems from the fact that we have packed a lot of information in the preparation of this concept. Further.2 and 18. Ehrenfels’ criterion is but one of a large number of possibilities to deﬁne symmetry invariance for the gestalt concept. we then consider the objective local composition {D} in LocF . Recall that this includes: • selection of the ambient space and address • the recursively hidden module structures to the ambient space • the relations among the elements ξ of the support X.7. In fact. chords.1 for more in-depth discussions of the historical dimension of music. we should like to get acquainted with the objective local compositions and try to understand the yoga of this concept. but see sections 11. Evidently. Under this general perspective. At ﬁrst sight the semantic potential of a simple structure such as D = (X. Should we—for example—include in the concept of “scale” any temporal . Fixing a symmetry (such as transposition) would cast the concept of gestalt in a too rigid framework. –Σ– Musical resources are indispensable for the musicological discourse.2 First Local Music Objects Summary. any denotator D of this form can be wrapped as a singleton {D} of the form named LocF and deﬁned by LocF IdΩF un(F ) −→ Power(F ) We shall always apply this wrapping formalism if talking of a general denotator D of form F as being “wrapped as local composition” or simply “wrapped”. to be introduced in chapter 8.

3 Precision in fact regards the naming. e. And this is done in full consciousness of the variability of topographic perspectives (discussed in chapter 2). e. tuning. For the backing principle of “Yoneda philosophy”. play a secondary role. .7. Momently. FIRST LOCAL MUSIC OBJECTS 109 order of its tones? Is it then c. 7. if we abbreviate F= F un(EulerM odule). a. This provides us with the possibility to build successively operative concept ﬁelds from precise aspects and to realize inner coherence by precise relations. And if you play in C-major. sound color etc. the concept of paradigmatic classiﬁcation (chapter 10) will lead to a more systematic treatment in chapter 11. We do take into account this ambiguity by usage of a variety of forms and denotators and therefore local compositions (later also global ones). We therefore make this deﬁnition: Deﬁnition 9 A chord is a ﬁnite local composition with EulerM odule as ambient space. This procedure has. we refer to chapter 9. Duration. more precisely3 .2. g. roughly stating that knowledge comes through the sum of aspects or perspectives.5 and chapter 5 and realized more down-to-earth in chapter 40.2. b for the “ascending” C-major scale and b. It should be stressed that this scheme is in perfect harmony with the knowledge and data base requirements alluded to in section 2. d. such as chords and scales as local compositions and gives examples from music and music theory. f. it is a denotator of form EulerChord −→ F in(F ) ΩF Power(EulerM odule) (7. onset. d. we say that it is an n-chord if its cardinality card(Cr) = n.3) If Cr : A EulerChord(Cr) is a chord at address A. Thissection deﬁnes pitch-related local compositions. etc. the essential is that intuitively it is a ﬁnite collection of pitches. however. c for the “descending” variant? And which “c” do we intend? Any “c” or ‘the’ “c”? How long does such a tone last? What is its loudness. the methodological advantage to exhibit a general scheme of musicological conceptualization by which the reader can easily build new concepts according to individual needs. these concepts are presented in an unsystematic way. g. does this mean that all tones are contained in the C-major scale? And do they even have to be played in the order indicated by the ascending or descending scale?—We immediately recognize that it is easier to tell what a scale is not than what it is. This scheme will be fully developed in the frame of the global theory (chapter 13).1 Chords and Scales Summary. –Σ– Example 5 Chords Although there are many ways to view a conﬁguration of tones as being a chord. f. nothing more. a. Let us now introduce some basic concepts which are elementary in the sense that they pertain to the language level of objective local compositions. The price for this gain in conceptual profoundness and universality is a certain slowness of ascent from elementary to more complex concept of practical usage.

8) xjust : A EulerP lane(xjust ). setting F = F un(p-EulerClass). we are taking Q3 /Z · p.5) more precisely. One then works in the quotient Euler module4 Q3 /Z · p and then deﬁnes as follows: Deﬁnition 11 Let p be a period in the Euler module. period p may be any non-zero denotator p : 0Z EulerM odule(p). and the second is the Euler plane. Id (7. xjust ) which live in the corresponding space forms: xtemp : A p-T empClass(xtemp ). Bernhard Stopper’s tuning [512] or contemporary jazz composition principles [359].110 CHAPTER 7. a chord event is not a local composition. Id (7. Often.4) where limit means product of the two factor coordinators. but: Exercise 1 Redeﬁne chord events as being special local compositions with ambient space Onset ⊕ EulerM odule. this period is the octave o : 0Z EulerM odule(o). (7.e. one adds onset to pitch and sets Deﬁnition 10 A chord event is a denotator of form EulerChordEvent −→ Limit(Onset. [Z] . EulerChord) Id (7. More generally. Thus. Hence the points x : A p-EulerClass(x) of a o-class chord consist of two components x = (xtemp . i. see (6.7) where the ﬁrst is the circle of rational cosets modulo o. it is a denotator of form p-ClassChord −→ F in(F ) ΩF Power(p-EulerClass). with form EulerP lane −→ Simple(Q · q ⊕ Q · t).9) be precise. Usually. The p-class chord is a ﬁnite local composition with ambient space p-EulerClass −→ Simple(Q3 /Z · p). and therefore we have a 0-chord for each address! To distinguish chords from their instantiation in time. and we have this representation of a direct sum of two modules: ∼ Q3 /Z · o → Q · o/Z · o ⊕ (Q · q ⊕ Q · t) (7. Id 4 To (7.6) The octave period o is particularly common. with form o-T empClass −→ Simple(Q · o/Z · o). LOCAL COMPOSITIONS Observe that—as always with denotators—the address is relevant. the underlying module is a Z-module.69) for the Onset form. pitches of chords are considered modulo a pitch period p. in order to include Wyschnegradsky’s non-octave-periodic constructions [131].

e.41) and notation from chapter 6. Cr2 of A-addressed chords or class chords. 0Ÿ @ Ÿ12 Cr = {PC12 (5). [1. if a o-class chord has all its points in the ‘tempered’ ambient space o-T empClass (resp. The diﬀerence here is that we have embedded the pitch classes in the general Euler module context. [6. For every couple Cr1 . We shall often identify the underlying modules and the associated forms. below a self-addressed 12tempered class 3-chord. w ) diﬀerent (coordinate sets for) w-tempered class chords (resp. a zero-addressed 12-tempered class 3-chord. in the ‘just’ ambient space EulerP lane) we say that it is tempered (resp. Of course. deﬁned by a positive natural number w. a 0-addressed w-tempered class n-chord can be . We then look at the tempered ambient subspace w-P itchClass −→ Simple(Z · Id 1 · o/Z · o) w (7. and a tempered class chord is called w-tempered if its points are contained in the ambient space w-P itchClass.2: Above. On the (octave-)tempered ambient space T empClass.1]. we can build their Boolean combinations: union Cr1 ∪ Cr2 . the submodule ∼ 1 Z · w · o/Z · o → Zw appears as a regular w-sided polygon. Whereas there are inﬁnitely many (coordinate sets for) just class chords. FIRST LOCAL MUSIC OBJECTS 111 Accordingly. PC12 (11)} 0Ÿ PC12 (11) PC12 (5) Ÿ12 PC12 (8) Ÿ12 @ Ÿ12 Cr = {[10.1]} [10.1] Figure 7. one often considers pitch classes in a regular subdivision of these pitch classes. n-element class n chords). one may also build its complementary chord Cr = χw − Cr.1] [1. For a w-tempered class chord Cr. RegDen-9. and diﬀerence Cr1 − Cr2 .11] [6. i.7.41). just). the ambient space w-P itchClass is isomorphic to the form P iM odw introduced in (6.2. PC12 (8). there are only 2w (resp. When representing its circle module coordinator as a circle. With the identiﬁcation from (6. We conclude with a remark on diﬀerent addresses for chords. the diﬀerence from the w-chromatic class chord χw of support Zw .10) of T empClass. intersection Cr1 ∩ Cr2 .11]. If there is no danger of confusion we shall omit the octave period in this context and simply speak of pitch classes or class chords when we deal with points or chords in ambient space o-EulerClass.

But it is not true in general that each non-empty p-class chord Cr deﬁnes a scale. a p-class chord. Then. In fact. the scale S may be recovered from its p-class chord by inverse image: S = mod−1 (modp (S)). We have Lemma 1 Let F = F un(EulerM odule). i. . p see also ﬁgure 7. one has to ﬁx a periodicity in order to express the repetition of scale tones under a particular period.11) which is induced by the canonical projection π : Q3 Q3 /Z · p. for addresses A. One may rightly deﬁne a form which comprises exactly the scales as its denotators. To deﬁne a scale. . We shall work in the EulerM odule form as an ambient space.e. For the same reasons as explained for chords in example 5.112 written as a set CHAPTER 7.e. LOCAL COMPOSITIONS Cr = {P Cw (r1 ). i.12) . for a critical address such as A = Z12 . the map A → Sc(A) = {S ∈ A@2F | S is a p-scale} deﬁnes a subfunctor of 2F . This associates with every objective local composition L : A → A@EulerM odule its pointwise deﬁned projection modp (L) : A → A@p-EulerClass. there is no point x : Z12 EulerM odule(x) which projects to the canonical injection ξ : Z12 o-EulerClass(ξ) in the tempered part.2 shows two class chords with diﬀerent addresses. Conversely. Exercise 2 Give a proof of lemma 1. . we select a non-zero period denotator p : 0 EulerM odule(p). bn ])} with rn . Figure 7. b1 ]. an . We then deﬁne Deﬁnition 12 Given a period p. .3. and such that (i) its projection modp (S) is ﬁnite. [an . . S is p-periodic. P Cw (rn )}. a p-scale is a non-empty local composition S of ambient space EulerM odule. The associated p-class chord is called the scale’s chord. (7. bn ∈ Zw . We have a morphism modp : EulerM odule p-EulerClass (7. . Example 6 Scales This concept regards special conﬁgurations in pitch spaces. (ii) S = ep S. and a self-addressed w-tempered class n-chord can be written as set Cr = {[a1 .

5. w In this case. According to the above.1. a zero-addressed scale can be represented as a subsets of this polygon. Again.4 for common scales in 12-tempered tuning.13) and we have a morphism scalemodp : p-Scale → p-ClassChord of forms. and the vertical one is the third axis. With lemma 1 we can introduce the scale form p-Scale −→ Power(EulerM odule) Sc ΩF (7. w-tempered and just scales. is said to be wtempered iﬀ the scale’s class chord S is w-tempered. When representing zero-addressed pitches on the EulerP lane.2.2. see ﬁgure 7. A scale. We therefore stick to the latter in the following discussion of common scales. FIRST LOCAL MUSIC OBJECTS 113 third axis period vector p fifth axis octave axis Figure 7. . we encounter two main types of o-scales.2.2 Just Scales A scale is said to be just if its class chord is just. In practice.7. This also means that the scale is contained in the w-tempered scale space w-T emperedScale −→ Simple(Z · Id 1 · o). this morphism allows us to recover all p-scales by inverse images of non-empty p-class chords. We show three common scales via their class chords in ﬁgure 7.1 w-Tempered Scales Let w be a positive natural number. 7. the horizontal axis is the ﬁfth axis.3: A scale with period p in Euler space. these scales will refer to the octave period. 7. We may then represent the scale by a class chord living in the EulerP lane ambient space.1. represented by its class chord S.

6. There is a strange belief that either a concept encompasses the intended phenomenon—or it cannot and should not be deﬁned5 ! We come back to this special issue in section 10. This insight becomes even more dramatic if we visualize the 12-tempered and just chromatic scale pitches within one octave.2 that this is a wrong impression. Thissection deals with onset time speciﬁc local compositions.114 CHAPTER 7. It can be shown that both. . the 12-tempered and the just octave conﬁgurations. . such as local meters and local rhythms.2 Local Meters and Local Rhythms Summary. This lack is not only due to the traditional preponderance of harmonic considerations since the Pythagorean school was founded. but it is also due to a remarkable (and not forgivable) lack of corresponding theory in European musicology. related to pitch c = 0 which is shown as the highest point of the polygon. admit a distinguished inner symmetry each. Whereas the 12-tempered conﬁguration looks like a pearl chain.11/12. see ﬁgure 7.4: Some of the most common 12-tempered scales. . 5 See the problem of deﬁning a motif in [444. the other numbers 1. We shall learn in section 8. 7. Messiaen Ÿ12 3. the just conﬁguration seems to be completely random. 2. .4. Messiaen Ÿ12 C-major Ÿ12 Natural C-minor Ÿ12 Melodic C-minor Ÿ12 Harmonic C-minor Figure 7. 11 follow in clock-wise orientation. for example. footnote]. p. and that this symmetry is intimately related with the theory of consonances and dissonances (see part VII). This gives a ﬁrst impression of how diﬀerent tempered and just tuning look from the geometric point of view. –Σ– Before discussing the precise concepts we should recall that meters and rhythms are among the most ﬂuﬀy concepts in musicology. LOCAL COMPOSITIONS Ÿ12 Chromatic Ÿ12 Whole-tone Ÿ12 2. it is also an eﬀect of the incapability of musicological theorists to construe complex concepts from simple ones in order to grasp complex phenomena. This is partly due to the extremely complex phenomenon of time-based regularities.2.

7 for an illustration. The diﬀerence b − a is called the length of L. FIRST LOCAL MUSIC OBJECTS 115 thirds a b f d e c a b g e f d fifths C-chromatic thirds thirds fifths fifths natural C-minor C-major Figure 7. e) (6.2. We start this concept framework with the elementary concept of a local meter. b]} (7. as represented by their class chords in the EulerP lane. y. • The beat meter is deﬁned by beat period Duration(1/y).5: The just C-major. b] = [−∞. if we are given a time signature T imeSig(x. The diﬀerence is the class of the syntonic comma. and C-chromatic scale. origin Onset(e). .34). and interval numbers [a. and interval numbers [a. b] = [−∞. ∞]. origin Onset(e). the latter according to Martin Vogel [547]. Observe that natural C-minor contains another pitch class for b than C-chromatic. the uniquely determined duration p is called the period of L. respectively. let p : 0Z Duration(p) and o : A Onset(o) be a positive duration period and an onset origin. see also ﬁgure 7. ∞]. two canonical local meters are associated: • The barline meter is deﬁned by barline period Duration(x/y). Deﬁnition 13 Given an address A. see (6.14) of A-addressed onsets. natural C-minor.7. Then a (A-addressed) local meter is an objective local composition L → A@Onset of ambient space Onset deﬁned as the p-periodic interval L = {etp · o| t ∈ [a. in case of positive length. In particular. Let −∞ ≤ a ≤ b ≤ ∞ be two extended integers.92).

metrical structures deal with onset weights. We may thus view the concept of a local meter as a basic account of the weight of selected onsets. which means that local meters are only the elements of a compound time structure. periodic repetition of onsets. In the understanding of Hugo Riemann [453]. Later. This approach has a psychological interpretation insofar as it suggests that elementary perception and/or expression of temporal stress or “weighting” relies on a repeated appearance of the same object in regular distances. From the music(ologic)al point of view. we shall introduce ﬁnite length meters for bars and beats to take care of the limits of concrete scores.6: One octave of the 12-tempered chromatic C-scale (above) and the just (Vogel) chromatic scale (below) The images seem to be very diﬀerent but share an important symmetry each.1 which deals with software for metrical analysis.4. LOCAL COMPOSITIONS 12-tempered C-chromatic thirds one octave e e d d c fifths octaves b b a a g g f f b d g e just (Vogel) C-chromatic a c f thirds b d a e fifths octaves Figure 7. A local meter is . We shall deal with global meters insection 13. The second one deals with “globalization” issues. There are several generalizations and complexiﬁcations of this conceptual germ which build a viable tool to understand metrical and rhythmical phenomena.3.116 CHAPTER 7. more in depth in chapter 21. a local meter is the minimum of time regularity that can be observed: a simple. Observe that the barline meter is independent of the time signature representation of the fraction x/y whereas the beat meter isn’t. and more concretely in section 41. the selection being driven by a period of regular distribution over time. The ﬁrst generalization concerns local rhythms and will be dealt with now.

Riemann describes rhythm as being related to a periodicity of higher level which means that a melodic movement may involve short onset diﬀerences and durations but can nonetheless produce a “rhythmical” period of larger extent. a local rhythm is a local meter whose recurrent unit has been blown up to a whole set of points with additional parameters. period and length are no longer uniquely determined from the datum of R alone. Deﬁnition 14 Given an address A.7: Three local meters as appearing in a local composition in pitch-onset space. But normal sounds do have lots of additional parameters to make them “audible”. Observe that we will not consider local meters within the given local composition if their onsets are not realized as projections of events to the onset axis. Given a local composition germ → A@(Onset×P ara) in a product ambient space of onset and some additional “parameter” space P ara. And these parameters are responsible for producing rich shapes which are beyond pure onset structures. Intuitively speaking.2. we may let ep act on the onset component of D. see also ﬁgure 7. In fact. (7. FIRST LOCAL MUSIC OBJECTS Pitch 117 ? Onset length = 3 period length = 2 period period no local meter! Figure 7. Carrying over this action point by point to objective local compositions with ambient space Onset × F . . we have a shift ep · D of D by ep as above. observe that for a denotator D : A (Onset × F )(D) in a product space of Onset with a form F and a period p : 0Z Duration(p).15) Observe that now. a more general and powerful one will be added when morphisms between local compositions are available. since ∼ A@(Onset × F ) → A@Onset × A@F . We want to give this idea a ﬁrst conceptualization.7. let p : 0Z Duration(p) be a positive duration period and −∞ ≤ a ≤ b ≤ ∞ two extended integers. Then a (Aaddressed) local rhythm is a local composition R of ambient space Onset × P ara of shape b R= t=a etp · germ. and this is it. In [453]. only concerned with periodicity on the onset axis.8. we obtain a shift operation on such local compositions. To do so. leaving the F -component invariant.

however. crescendi. be more precise in order to make the ideas transparent: .3 Motives Summary. Let us. it is already now possible to construct local “polyrhythms” from given collections of local rhythms.2. Whereas we shall extensively develop the global perspectives and prove that this satisﬁes many of the present urges. and length is 1. The parameters which are added to onset are numerous.106. Exercise 3 Give an exact description of the space P ara in ﬁgure 7. Motives are introduced as a type of local compositions which share (among others) pitch and onset time in their coordinators. including duration. loudness.118 CHAPTER 7.4.75 to Nr. and accents. legati. LOCAL COMPOSITIONS germ e8/4. see section 7. Evidently. 75-76. The germ consists of the entire bars Nr. bars Nr. the purely local approach to meter and rhythm is unsatisfactory since it is felt that real phenomena of “time grouping” are much more involved.78. Exercise 4 Deﬁne a metronome as a local rhythm with space P ara = Loudness. –Σ– In a rather generic setting we could deﬁne a motif as a local composition whose ambient space projects to a product of onset and pitch.8: A local rhythm Beethoven’s op. 7. period is 2 = 8/4. Allegro.8.germ rhythm of period 8/4 and length 1 Figure 7.

FIRST LOCAL MUSIC OBJECTS 119 Deﬁnition 15 If A is an address.9: Three examples of motives: M1 . pOnset : F → Onset is the canonical projection. j j # # X X XX X X X X. for any form P ara of additional “parameters”. their relevant ‘shape’ for motivic analysis is centered around onset and pitch.2.9. a (A-addressed) motif M is an objective local composition M → A@F whose ambient space F = Onset ⊕ P itch × P ara. Whenever we want to make the . and one local composition. Though these local compositions live in spaces with possibly many parameters. X X X X X X X # X X n # XX X #XX X XX X X X X X & X # X nX X X jX X X #X X ## X ? M0 NO! M1 YES! M2 YES! X X # XX # X X X ¥ X X X X X X #X X X X M3 YES! abstraction from secondary parameters Pitch Onset Figure 7. To see examples of motives. nOnset (n) = pOnset to the onset space are diﬀerent. M3 . The reason is that naming tradition views meters and rhythms as being something complex whereas motives are conceived as being “cells” of melodic structures. If the cardinality of a motif is c we say that it is a c-motif. projections mOnset = pOnset (m). which is not a motif.7. refer to ﬁgure 7. n of diﬀerent elements in M . M0 . motives are not attributed by “local”. say. Notice also that in contrast to local meters and rhythms. M2 . Notice that this concept is very restrictive in that it does not allow for motives which contain chords or several contrapuntal voices. and such that for any pair m.

Naively. So it is useful to study a less elementary address situation. A prototypical case is a two-voice counterpoint with cantus ﬁrmus and discantus. i. see also ﬁgure 7. DP itch .10: A motif at address Z can grasp contrapuntal structure of ‘arrows’ targetting from cantus ﬁrmus notes to discantus notes. j X j X X X X X X X # XX # X X X X X X X X X #X X X X #X X X X X X X X Pitch discantus cantus firmus Onset Figure 7. the arrows OND ⇒ OF FD and ONE ⇒ OF FE are diﬀerent.10. Musically speaking such a motif consists of ‘intervals’ of zero-addressed denotators such that no two such intervals intervene twice with same ‘ON’ and ‘OFF’ values.1.e. a Z-addressed denotator D consists of the ON -denotator of the cantus ﬁrmus and the OF F -denotator of the discantus voice. a motif at this address means a set of denotators D : Z (DOnset .120 CHAPTER 7. They . The motif condition of separation of motif elements by their onset projection then means that for any two denotators D. As insection 6. this denotator DOnset can be symbolized by an arrow OND ⇒ OF FD in Onset space. the attribute “local” is inserted. Let us shortly comment on the address question here. E of the motif. the latter being added simultaneously to each note of the cantus ﬁrmus voice. we always think of zero address.2. either their ‘ON’ values or their ‘OFF’ values are diﬀerent. If we take A = Z. DP ara ) from which the onset coordinate is an aﬃne homomorphism DOnset : Z → R. LOCAL COMPOSITIONS situation unambiguous. In this case.

3. Then S ⊂ S. we may associate to each functorial local composition S → @A × F at address A and in ambient space F its ˇ objective trace S.3. It is now evident that in general. Conversely.1 that an objective denotator D → A@F at address A in ambient ˆ space F is identiﬁed with its functorial ‘version’ D → @A × F . –Σ– Recall fromsection 6. .3 give ˆ f @T = {f } × T · f (7. To this end.2. let f : B → A be an address change and let S → @A × F be a local composition at address A and in ambient space F . Its objective trace is deﬁned as the following objective local composition: ˇ S = {s ∈ A@F | (IdA . (7. t)}.20) for an objective local composition T at address A and address change f : B → A.16) at address A and in ambient space F . We now want to make more explicit pictures of functorial local compositions in order to understand the speciﬁc diﬀerence to objective local compositions. s) ∈ A@S} (7. and S = S iﬀ S is objective. such as ﬁber products of local compositions.3 Functorial Local Compositions Summary. 7.17) The following is evident: Sorite 1 Let S → @A × F be a local composition at address A and in ambient space F . It settles problems of universal constructions. The associated objectivized composition qua functor will be deﬁned and denoted by ˇ S = (S) (7. and a motif just distinguishes its points by the OND -value. The generalized conceptual framework of non-objective. the notation of section 6. In particular. Hence OND ⇒ OF FD is the ‘zero’-length arrow OND ⇒ OND . Deﬁnition 16 Let S → @A × F be a local composition at address A and in ambient space F . which do not always exist in the frame of objective local compositions. local compositions are not objective.18) and call this set the f -slice of S whence B@S = f :B→A f @S. (7. proper functorial local compositions is presented.11 for the visual representation which we shall adopt henceforth.7.19) see ﬁgure 7. FUNCTORIAL LOCAL COMPOSITIONS 121 diﬀer in pitch but not in onset coordinates. Then we set f @S = {s ∈ B@S| s = (f.

Take a simple form of functor @M and the zero morphism z : A → A. f @S = {f } × A@F for all f : A → A whereas f @S = {f } × A@F · f . associated module of a local composition. the full local composition S = @A × F gives the cartesian product A@S = A@A × A@F at address A.122 CHAPTER 7. For example. see chapters 15 and 19—we cannot refrain from non-objective constructs. LOCAL COMPOSITIONS B@F f@S B@S B@A f:B A Figure 7. general position of points.e. Then we have z@S = {z} × A@M · z = {z} × eM · z. The elementary concepts in local theory are discussed: cardinality. 7. sub-compositions.4 First Elements of Local Theory Summary. as evaluated at address B with its ‘slice’f @S at address change f : B → A. Since ﬁber products are fundamental tools of global theory and also in all questions of Grothendieck topologies. products. i. –Σ– In the following discourse we deal with many diﬀerent local compositions at a time and with their recombinations under speciﬁc operations.2.3.11: Graphical representation of a functorial local composition S at address A. dimension. coproducts. the constant morphisms in A@M which is much less then the full set A@M of aﬃne morphisms. projections and injections. Since full local compositions play an essential role in the theory of coordinate functions for musical parameters of global compositions—the deep question of realization of abstract compositional ideas. i. Other examples of limitations for objective local compositions will appear in questions of universal constructions such as ﬁber products. From the perspective of rigorous denotator .e. this is one further motivation to transgress the frame of objective local compositions which a decade ago were the core of mathematical music theory [340]. seesection 8. Boolean constructs.

for a thoroughly unambiguous discourse. (−∅) is objective while ¬∅ = @A × F is ˆ ˆ not.4. V ∈ 2A@F and two functorial local compositions D. The proof is left as an exercise. we have the Boolean algebra 2A@F of subsets of A@F . (D ∨ E) = D ∨ E . Also. intersection.1. and not only the names of the ambient spaces of these denotators. in other words R = R · IdA = U · IdA ∩ (A@F − U ) · IdA = U ∩ (A@F − U ) = ∅. we have ˆ ˆ U ⇒V ⊆U ⇒V.e. This also shows that the inclusion in point (vi) above is strict in general. On the other hand. R ⊂ A@F .5. We ﬁx an address A and an ambient space F . Hence we shall not specify these names and just name denotators and ambient spaces. i. for two objective local compositions U. ˆ ˆ (iv) Union preserves objectivity. Observe that negation. E ∈ Sub(@A × F ) we have: ˆ ˆ (i) If U ⊂ V then U ⊂ V .e. However. (iii) If D ⊂ E then D ⊂ E . Also it is necessary in a conceptually full-ﬂedged musicological context where names are essential.g. Rather would this hide what we are doing than clarify anything. FIRST ELEMENTS OF LOCAL THEORY 123 theory. It makes sense to invoke the Heyting algebra Sub(@A × F ) of subfunctors6 of @A × F . this would be an ‘overdressed’ formalism. ˆ But U ∧ (A@F − U ) is clearly not empty in general.7. (v) Union preserves associated objective local compositions. Heyting Constructions. U ∪ V = U ∨ V . ˆ (vii) We have ¬U ⊂ −U . equal to R. then we would have ˆ ˆ ˆ IdA @R = IdA @(U ∧ (A@F − U ) = IdA @U ∩ IdA @(A@F − U ) . 6 See appendix G. For example. Here are some facts which relate the two algebras: Lemma 2 With the preceding notation. as announced above. within an implementation of the theory in data base management systems (viz. ˇ ˇ (ii) If D ⊂ E then D ⊂ E. ˆ ˆ (vi) We have U ∩ V ⊂ U ∧ V . respectively. . one should always specify the names of forms where these denotators belong. (viii) With respect to the Boolean and Heyting implications. if U ∧ (A@F − U ) were objective. i. and implication behave in a rather complicated way. However. e. This is a very special case of a ﬁber product which is not objective. the strict formalism must be followed. for the description of our recombination tools. the implementation for the RUBATO software described in chapter 40). proposition 113.

Observe that we take the new ambient space with the name ′F × G′.2. Then we have an isomorphism of sets f @(S × T ) → fA @S × fB @T ∼ (7. appendix G. an equation which is functorial in the domain C of address change f : C → A and induced by the canonical injections iF : F → F G.124 CHAPTER 7. If we are given two local compositions S → @A × F and T → @B × G. 9 See appendix G. LOCAL COMPOSITIONS Products. This formula also deﬁnes two projections pS : S × T → S and pT : S × T → T induced by the ﬁrst projections pA : A × B → A. the product functor7 S × T can be identiﬁed with a local composition in the following way. pG : F × G → G. including the name ′F G′ of the coproduct ambient space.3.1.8: take M = 0 and C = 0 in theorem 54. The coproduct functor9 S T is a subfunctor of @A × F @B × G which in turn is canonically isomorphic to @A × (F G). S × T is a subfunctor of (@A × F ) × (@B × G).e. called the product ambient space. Coproducts of local compositions S and T can be built in case they pertain to the same address A = B. But we have a canonical isomorphism8 of func∼ tors @A × @B → @(A × B).21) which is functorial in the variable address C.23) T −− − − − − −→ subfunctor @B × G which yield a model for general morphisms between local composition to be introduced in chapter 8. In other words.1. pF : F × G → F and the second projections pB : A × B → B. Let fA : C → A and fB : C → B be the two components of f . the f -slice of S × T is calculated as follows. We shall henceforth take this local composition when speaking of the product S × T of local compositions. i. We also ∼ have f @(S T ) → f @S f @T . S × T is isomorphic to a local composition in @(A × B) × F × G. for each address A.2. section E. as with the product we identify S T with the corresponding local composition of @A × (F G). More concretely. if f : C → A × B is an address change. Coproducts. Therefore.22) S −− − − − − −→ subfunctor subfunctor @A × F S × T − − − − @(A × B) × F × G − − −→ @p ×p pT B G (7. but observe that the new address is also the product A × B of the old ones. we have two commutative diagrams of functors S × T − − − − @(A × B) × F × G − − −→ @p ×p pS A F subfunctor (7. 8 See 7 See . Since limits and colimits of set-valued functors in Mod@ are calculated as corresponding constructs in Sets. S → @A × F and T → @A × G. Projections and Restrictions.

25) S T − − − − @A × (F − − −→ subfunctor G) which also anticipate the concept of morphisms between local compositions. But in the special and important case of simple ∼ ambient space F with F un(F ) → @M for a module M over a commutative ring R. Suppose that {s − x| s ∈ S. Such objective local compositions are called commutative (over R if this speciﬁcation is required). and S is called ﬁnite.y} 10 See 11 See s∈S−{x. sorite 13 in appendix E. We again deduce two commutative diagrams of functors: S −− − − − − −→ subfunctor 125 iS @A × F Id×i F (7. we can say much more about objective local compositions S which are addressed at R-module ˇ A and their objective trace S is contained in the subset A@R M of R-aﬃne10 morphisms f : A → M .7. s = x} is a set of linearly independent vectors. inﬁnite.3. We then have a subset S ⊂ A@R M of the 11 R-module A@R M . so dy ⊂ dx . Identify S with its objective trace S. and let s∈S−{y} λs (s − y) = 0 be a vanishing linear combination. Proof.4.24) S T T − − − − @A × (F − − −→ −− − − − − −→ subfunctor subfunctor G) iT @A × G Id×i G (7. Let us ﬁx a commutative local composition S over ring R. y ∈ S be two elements of S (more precisely: of S). and ambient space ˇ @M . iﬀ its cardinality is so.y} section E. A@R M ) be a non-empty commutative local composition over ring R ˇ and let x.3 for notation. are equal. etc. address A. . Module of a commutative local composition. FIRST ELEMENTS OF LOCAL THEORY iG : G → F G. In general. = − λx (y − x) + s∈S−{x. Cardinality. Lemma 3 Let (S. respectively. We then have 0= s∈S−{y} λs (s − y) = s∈S−{y} λs ((s − x) − (y − x)) λs (s − x) − ( λs )(y − x). the non-zero elements of dx form a basis of dx iﬀ the non-zero elements of dy do so. The cardinality of a local composition S at address A is deﬁned by card(IdA @S) = ˇ card(S). We have s − y = (s − x) − (y − x). Then the sub-R-modules dx and dy of A@R M generated by the sets dx = {s − x| s ∈ S} and dy = {s − y| s ∈ S} of diﬀerences to x and to y. the space functors of local compositions do not carry algebraic structures. and a symmetric argument shows the other inclusion. Further.

and that S is embedded iﬀ S ⊂ R.S. by lemma 3. together with the rhythm’s period. in fact on a horizontal line. Loudness general position of germ Onset Loudness special position of germ Onset Figure 7. we have λs = 0 for s ∈ S − {x. this statement is independent of the chosen element x. . an Rmodule R.S = A@R M . its points are in general position iﬀ its module is free of rank card(S) − 1. Above. and therefore also λx = 0. A@R M ) over ring R. y}. 12 Recall 13 Including that in the category Mod the empty module is admitted. the zero-dimensional case dx = {0} and R. Sorite 2 If S is a ﬁnite commutative local composition. and thus giving general position and evidence of the germ. the germ is no more retractable from the local rhythm. we see the new germ with ﬁrst event in higher loudness. The symmetric argument yields the other implication.S we say that the points of S are in general position13 . Since the germ’s elements are in special position. Deﬁnition 17 For a commutative local composition (S. LOCAL COMPOSITIONS By linear independence of the elements in {s − x| s ∈ S. dx − {0} forms a basis of R.26) for any element x of S if S is non-empty. see appendix E. otherwise we set R.∅ = ∅. If for a non-empty S. It is called the module of S.126 CHAPTER 7.3.12: Below.S = dx with dx = {s − x| s ∈ S} (7. we have a germ of a local rhythm in Onset ⊕ Loudness space. We say that S is generating iﬀ R.S is deﬁned by R. QED. By lemma 3.S = 0. this module12 is well-deﬁned. s = x}. Again.

(2.S is free of rank card(S) − 1. Evidently. the key it denotes is just f. b) = (0. The condition is necessary by deﬁnition. M modules over commutative rings. 10)} in ambient space Onset ⊕ Loudness. Let us consider the zeroaddressed local three-element composition S = {(0. putting the germ events into general position makes a hidden (purely mental) idea—in this case: the periodicity and the germ—evident. this cannot be done. If we use this local composition as a germ for a local rhythm with period 3. ALTERATIONS ARE TANGENTS 127 Proof. 10)}. the characterization of projective ﬁnitely-generated modules in appendix F.(1. 10). they do point to a reﬁned situation: We have to keep in mind two notes: the origin and the amount of shift from the origin. To describe alterations we consider the dual number restrictions i : R → R[ε] : r → r + 0 p : R[ε] → R : a + bε → a and set M [ε] = R[ε] ⊗R M and f [ε] = R[ε] ⊗R f for corresponding functor values on objects ∼ and morphisms in ModR . 7. Let T → @A × @M . take a piano note denoted by “e-sharp”. Associated transformations between tangents and base-points are introduced. –Σ– The technique of alterated notes is more than a notational formalism. guarantees that the points are in general position if the module R.2. and no deeper reason exists to write “e-sharp” instead of “f” to deﬁne a particular key and pitch. Musically. audible. But for variable temperament such as is present on string instruments. and the points of S are not in general position but aligned on a horizontal straight line of height 10. 0) ⊕ R. 10). We should therefore give one preliminary illustration to make evident the musical meaning of general position. However. 18). By use of dual numbers.3. (1. Local commutative compositions with points in general position play a fundamental role in classiﬁcation theory and for understanding performance strategies. Here.e.S = R.7. QED. R. For example. Recall from appendix E. Refer to ﬁgure 7. −8).(1. (E. theorem 58. to speak geometrically. Whatever the concrete meaning of such alterations. From appendix E.2 that over R. i. if we want to stress the period and play the ﬁrst event of this germ louder than the others.2. 10). and this is a general position situation. Conversely. On a well-tempered piano. this yields a uniform sequence of events. we obtain a germ S ∗ = {(0.12 for the following comments. Musically..S = R.5). (1. 0). alterations of tones in pitch or any other coordinate are interpreted as ‘tangent’ objects. a).5. we know that any restriction r : R → S of commutative scalars gives rise to a functor S⊗R ? : ModR → ModS : N → S ⊗R N which carries an R-aﬃne transformation f = ek · f0 : N → K to the S-aﬃne S ⊗R f = e1⊗k · S ⊗R f0 . we have R.(1. at value 18 say. we have to play a regularly spaced sequence of three notes of equal loudness. M [ε] = M + εM → M 2 such that scalar multiplication on dual numbers is given by ε(a. (2. In this section we concentrate on (not necessarily objective) A-addressed local compositions in simple ambient spaces @M .5 Alterations Are Tangents Summary.

Conversely. points in dual modules M [ε] are eﬀectively identiﬁed with Zariski tangents. this will be done in chapter 8. for example. Deﬁnition 18 Transformation α+ and the corresponding natural transformations @α+ is called sweeping orientation. then the zero-addressed x = 60 − ε2 codiﬁes C . we may view x as being a ‘real’ pitch xred plus an ‘pitch alteration quantity’ xalt .29) α± T − − − − − − −→ 14 In @A × @M algebraic geometry. Musically speaking.23). in the above example. and call Tred the image of T under the projection Id × @p : @A × @M [ε] @A × @M . there is need for more systematic treatment of such relations among local compositions. M = P itch.27) Tε − − − − @A × @M [ε] − − −→ of functors. both diagrams remind us of the previous diagrams which we have encountered in (7. Clearly.22) and (7. To recover ‘real music events’ from this formalism. Call Tε the image of T under the embedding Id × @i : @A × @M @A × @M [ε] such that we obtain a commutative diagram T −− − − − − −→ subfunctor subfunctor iT @A × @M Id×@i (7. consider two R-linear projections α+ . deﬁned by α+ (x) = xred + xalt and α− (x) = xred − xalt . p. let T → @A × @M [ε]. the reduced component xred is the meaningful musical event whereas the alteration component xalt symbolizes a shift from the reduced value. Evidently. the transformation α− and the corresponding natural transformation @α− is called hanging orientation. α− : M [ε] → M . . whereas α− gives 62. xalt : A → M . and 60 to middle C. if pitch is codiﬁed such that a semitone corresponds to real number 1.1 and [386. respectively. Again. This suggests that we view alterations as tangents14 . These concepts will become central in the counterpoint theory of part VII. this is no longer the case with dual ambient space @M [ε]. see appendix F. We view this artifact as follows. Then we have a corresponding commutative diagram T pT − − − − @A × @M [ε] − − −→ Id×@p subfunctor subfunctor (7. α+ gives the altered pitch number 58 = α+ (60 − ε2) from the dual representation.4.28) Tred − − − − − − −→ @A × @M of functors.128 CHAPTER 7. If. For instance. it can be written as x = xred + εxalt with two components xred . If x : A → M [ε] is a denotator in this ambient space. we have corresponding commutative diagrams on A-addressed local compositions T in dual ambient space M [ε]: T − − − − @A × @M [ε] − − −→ Id×@α subfunctor subfunctor ± (7.25]. LOCAL COMPOSITIONS be such an object for the ring R. If we suppose that the elements of a local composition with ambient space @M are musically reasonable entities.

The use of alterated pitch is a standard argument in music analysis to associate neighboring tones. we say that T is elementary. In particular. . Deﬁnition 19 Let T be an A-addressed alteration of ambient space M [ε]. The following results deal with this problem and give a general answer to the question of how broad an alteration process can become in terms of ambiguities. the theorem gives information about the signiﬁcance or insigniﬁcance of alteration arguments in traditional analysis. This means that there is a bijection Tred → α+ T which is induced by two bijective projections ∼ ∼ T → Tred . t ∈ Z. p(k)} = ∅. alterations are very frequent. –Σ– In classical music theory. This bijection consists in an m-shift of one element p(k) to α+ (k) = m + p(k). one could argue that this technique may be abused to ‘demonstrate’ anything by construction of appropriate alterations.13. In the worst case. ∼ F . i.1R . as represented by the basic diatonic scale. . a so-called “di-alteration”. If we have an elementary shift E following is then immediate15 : 15 Here. very rarely do we ﬁnd the pitch arsenal within a determined diatonic scale. we again write E F and say that F is a shift of E.0 : A → R. and pitch 67 as α+ (66 + ε1). positive and negative alterations is a well known—though not very frequent—tool to produce two alterated pitches from one ‘abstract’ datum. thereby reinterpreting given pitch as an alteration of a diﬀerent ‘pitch origin’. The ∼ the summand eZ1R .0 (constant shift). Ei−1 Ei = F of elementary shifts. . viewing simultaneously pitch 65 as α− (66 + ε1). . Mason’s theorem is generalized to nchromatic classes. If there is a sequence E0 = E E1 . This is often inevitable since in real scores. E 1 E2 . Most chords are loaded with alteration signs. ALTERATIONS ARE TANGENTS 129 with α± T being the image of T under the ‘ambient morphism’ Id × @α± .e. In the following discourse we shall only deal with commutative local compositions in ambient spaces @M of modules M over a ring R and the associated dual version. Mason [321] concerning alteration with 12-tempered pitch classes. we write E F and say that F is an elementary shift of E. Deﬁnition 20 If E and F are two local compositions such that there is an elementary alteration T over ring R with E = Tred and F = α+ T .5. and let 0 = m ∈ A@R M . Commutative local compositions of ambient space M [ε] are also called alterations. The theorem of Mason deals with the richness of possible alterations within 12-chromatic pitch classes between determined scales. see also ﬁgure 7. if M = R and m = e±1 . 7. we also have an associated bijection E → F . and T → α+ T . Our results generalize the theorem of Robert M. The consideration of both. whereas the other elements remain ﬁxed.7. Musicologically. But it is nevertheless “clear” that one essentially stays in a given tonality.1 The Theorem of Mason–Mazzola Summary.0 relates to the constant shift morphisms et. Then T is called m-elementary iﬀ it contains one element k = kred + εm such that T − {k} = (T − {k})red and T − {k} ∩ {α+ (k).5. The problem of such an approach is that reinterpreting any given set of pitch values as being the result of a speciﬁc alteration process provokes a huge amount of polysemy.

From the proof of theorem 1 we will also obtain Mason’s result: Proposition 2 (Mason. See ﬁgure 7. where [−l. Theorem 1 Let n be a positive natural number. k] as deﬁned in the theorem. the special case m = 7 yields k = 3. we can choose |k − l| ≤ 1 in the above theorem. In particular. The following results which were solicited by Mason’s investigation [321] show that the existence of shifts is a strong condition on couples of equipollent16 local compositions. l = 3. k] = {−l. l be as in the theorem. k]. and let E. . l = 2 or k = 2. This means that any scale with seven-element class chord F can be shifted to any other equipollent scale with class chord E by an alteration K whose elements have at most three sharps and two ﬂats or at most two sharps and three ﬂats. We can even choose K as being deﬁned by a sequence of elementary shifts Ei Ei+1 which all satisfy the condition imposed on K.0 such that ∼ ∼ the bijection E → F associated with the shift is deﬁned by two bijections T → Tred = E and ∼ T → α+ T = F . This fact is very important to judge musicological argumentations built upon alterations. 16 Equipollence means that they have equal cardinality. Then there are two non-negative integers k. Let k. k] for the x-shift of the interval [−l.m e+k Figure 7.13: An m-elementary alteration T controls a shift between its reduced and its alterated images. Sorite 3 If there is a shift E F .130 CHAPTER 7. and let x ∈ Zn . l with k + l = n − m and a shift E F which is induced by an alteration K ⊂ Zn + ε[−l. k − 1. Write [l|x|k] = ex · [−l. It could happen that reinterpreting a chord as a shift of another. LOCAL COMPOSITIONS A@eM T k p kred a+ A@M e. For classical European music. chord (such as one sitting in a particular scale as above) is always possible and therefore does not mean anything to the actual statement! For the proof of theorem 1 we need some notation. k}.14 for an illustration. there is a unique alteration T in A@R + εeZ1R . . [321]) For n = 12 and 7 ≤ m. any class chord Ch with card(Ch) ≤ 7 is the shift of another equipollent class chord in any ﬁxed diatonic scale. distinguished. . F be two equipollent zero-addressed local compositions in Zn with cardinality m. . −l + 1.

t+ ]×[s− . 1. Lemma 4 Let E and let s ∈ E. Look at the subset [t− . 17 Mathematicians’ short form of “without loss of generality’. So suppose 5 ≤ n − m. ALTERATIONS ARE TANGENTS 131 eŸn e·k { { { }=K }=K }=K Ÿn a+ e·-l Figure 7. x. x+ − 1. . F be a shift whose alteration K veriﬁes the theorem’s conditions with k. . s) ∈ ∆. If we had a t. (ii) If σ is a ‘second’ such that α+ p−1 s = s+ and s|σ|s+ .15. If k. 2. l are clear we also write x+ = x + k and x− = x − l. but it cannot hit ∆.14: The condition imposed on on alteration K in theorem 1. 4 are easily veriﬁed under the stronger hypothesis |k − l| ≤ 1. and it makes sense to say that an element y ∈ [l|x|k] is to the right or to the left of x according to whether y appears before of after x in the sequence x− . and then WLOG17 we can suppose that z = n − 1 ∈ E ∪ F .5. Consider now the ‘second’ σ = {y.7. x+ ]. Suppose ﬁrst that E ∪ F Zn . s+ ] of Zn ×Zn and refer to ﬁgure 7. s) stays to the left of ∆. s+ ]. 3. it cannot reach the bottom right edge (t+ . Proof of lemma. The proof scheme is the same for both cases. QED. . we write x− |σ|x. x+ . s− ). . whence [l|x|k] = [x− . and we may restrict to case (i). if y + 1 stays to the right of x we write x|σ|x+ . Under elementary shifts the couple (t. [l|x|k] has exactly n − m + 1 elements. The diagonal of Zn ×Zn is denoted by ∆. which is forbidden by statement (i) of the lemma. l a+ (i) If σ is a ‘second’ such that α+ p−1 s = s− and s− |σ|s. the couple (t. t+ ] × [s− . . then there is no t ∈ E such that α+ p−1 t = t− and t− |σ|t. . Since according to our premises. . x− + 1. s) is shifted one unit in horizontal or vertical direction within [t− . then evidently (t. y + 1}. then there is no t ∈ E such that α+ p−1 t = t+ and t|σ|t+ . We make an induction on n − m and observe that the cases n − m = 0. Proof of theorem. Since k + l = n − m < n. If y stays to the left of x.

. Take a point z ∈ Zn where z − 1 ∈ F and z ∈ E − F . ˜ To begin with. z + 1} be the ‘second’ in Zn−1 as above from eliminating z. we shift z. LOCAL COMPOSITIONS s+ s D s s- t- t s t+ Figure 7. z ∈ E whereas z + 1 ∈ E. we cannot have simultaneously a shift of s ∈ E to s− and a shift of t ∈ E to t+ with s− |σ|s and t|σ|t+ . we have z. We may therefore suppose that E. According to lemma 4. Let σ = {z − 1. l . we denote the points s− . Point z is unshifted and s− |σ|s. Point z is shifted and neither s− |σ|s nor s|σ|s+ . . l = l . At worst. and we are done. 0} in Zn−1 .132 CHAPTER 7. . 3. By induction hypothesis. t+ ] × [s− . 2. The case E ∪ F = Zn−1 is a bit more involved. s+ with respect to k . . which shift the elements s ∈ E ˜ what happens to the shift quantities if we reinsert z ‘in the middle of’ σ. there are non-negative integers k . σ). . There are six cases: 1. . F ⊂ Zn−1 by contracting the ‘second’ {n − 1. Ei−1 Ei = F of elementary shifts in Zn−1 which shift every element s ∈ E within [s− . The point here is to understand and ending in F . z + 1. s+ ] and the relative position of (t. s+ ]. l with k + l = n − m − 1 and a sequence of elementary shifts starting from E ˜ within [l |s|k ]. z by one unit to the right and obtain E. 0} to point 0 ∈ Z and creating the ‘second’ σ = {n − 2. Point z is unshifted and neither s− |σ|s nor s|σ|s+ . If s ∈ E. By induction hypothesis. in the second case we have a shift with k = k + 1. z + 1. E 1 E2 . Suppose that here. After reinsertion of z we have a required shift with k = k . l = l + 1 in the ﬁrst case. . l with k + l = n − m − 1 and a sequence E0 = E E1 . . . either some s are shifted to s− with s− |σ|s or some t are shifted to t+ with t|σ|t+ .15: The product set [t− . ˜ there are 0 ≤ k . s) with respect to the diagonal ∆ and the ‘second’ diagonal (σ. .

After reinsertion of z there is no change in the shift intervals in cases 1. [l |s|k ] for (unshifted) elements s ∈ E in the ﬁve possible cases. we have (s − (l + 1))|τ |s. and 2. Remark 2 It is not known whether the choice |k − l| ≤ 1 is always possible. Set τ = {z.7. z +1}. and therefore. since z does not intervene in those intervals. 5. [l − 1|s|k + 1] 3. QED. It remains to be shown that case 3. in case 4. and in case 5. [l |s|k ] 2. We shall do the following: Whenever an element s is shifted to jump from z −1 to z +1 in σ or conversely within one of the existing elementary shifts. we have s|τ |(s + (k + 1)). [l + 1|s|k ] 4.5. we get (s−(l +1))|τ |s. σ cannot be to the right of s. . Point z is unshifted and s|σ|s+ . 6. 133 The last possibility cannot happen since s − (z + 1) + k ≤ m − 1 + n − m − 1 = n − 2. ALTERATIONS ARE TANGENTS 4. and 4. we have these total shift intervals 1. In case 3. [l |s|k + 1] 5. are mutually exclusive. we shall replace this elementary shift by the succession z − 1 z z + 1 or z + 1 z z−1 of two elementary shifts. If we go back to the original positions. But this follows rightly from lemma 4. Point z is shifted and s− |σ|s. So we are left with the question of what happens to the possible total shift intervals. Point z is shifted and s|σ|s+ . versus cases 2.

.

The core process of symmetry leads to structure-conserving transformations which we rebuild in the context of local compositions and the underlying forms. it must be a concept of “morphological” comparison which is able to grasp these correspondences as expressions of underlying gestalts.” 135 . The modern concept of symmetry is in fact a semiotical one: Symmetry bears the character of a sign. die ein gewisses Unbehagen vor dem Worte Symmetrie. Such a type of comparative paradigm leads to the concept of symmetry as it is used in its modern form [574] and may be rephrased1 more generically as follows: Deﬁnition 21 Symmetry is correspondence of parts as an expression of a whole. –Σ– The basically ‘rigid’ structure of a local composition as it was discussed in section 7. The form (the signiﬁcant) of a symmetry is a mathematical transformation which speciﬁes 1 Rudolf Wille’s deﬁnition in [574. die darin Formalismus. on the other it recognizes the semantical function of symmetry and is no longer limited to formal aspects. On one hand.444] is this: “Symmetrie ist Gleichheit von Teilen als Ausdruck eines Ganzen. das zu phantasievollen Spekulationen Anlaß gegeben hat. Schematisums sehen und es o am liebsten ganz aus der ‘g¨ttlichen’ Vernunft verbannen m¨chten. Such a concept of comparison must then conserve the interior correspondences among parts of the local compositions.Chapter 8 Symmetries and Morphisms Hoﬀentlich habe ich diejenigen beschwichtigt. this deﬁnition by far exceeds simple axial symmetry. o o Wolfgang Graeser [194] Summary. their relations are discussed and formalized. p.1 requires an additional ingredient of conceptualization to be able to compare such ‘rigid’ objects. Thereby. nicht verbergen k¨nnen. symmetry is a key concept—however in its modern version which by far exceeds traditional axial or rotational symmetries. After having introduced the objects of local theory.

p.1 for this subject). only correspondence! In a symmetry of moral qualities. We insist on replacing the word “Gleichheit” (equality) in Wille’s deﬁnition by “correspondence” in the above deﬁnition since there is no a priori replication under correspondence. The point is that • First. • Third. Kepler’s second law states that a planet sweeps out area at a constant rate2 . and only appear on the phenomenological surface when investigated with powerful tools. If the great composers and musicians participate in a sphere of high or even divine sophistication it is only logical to approach their works by use of adequate. but they do correspond. we should admit that deep musical facts cannot be understood by use of trivial tools. symmetries often appear as broken or hidden symmetries which means that they are not immediately visible. The physical symmetry of forces means life to the ropedancer. SYMMETRIES AND MORPHISMS the correspondence. We therefore hope that the reader will not reject conceptually involved investigations because of a conﬂicting belief that there should be easy ways to understand the genius in music. p. this would be as absurd as to develop physical theories without inﬁnitesimal calculus. For example.524]. and music which debates that symmetries would cast the artistic freedom of shaping highly sophisticated works of human expression into a trivial or even fascistoid Procrustean bed and thereby fail to explain the very essence of artistic expression. 2 See 3 It [307. The concept of symmetry does in fact not imply equality of symmetrically related parts. this fact is well known and—h´las—responsible for exorbitant costs of particle accelerators: Fundamental laws e of physics are virtually always driven by hidden symmetries which are only visible under extremely high energies. For example. The visible symmetry of a “perfect” circle motion is replaced by a more abstract symmetry where “distance” is replaced by “area per time unit”. for example. In elementary particle physics.136 CHAPTER 8. The meaning of planetary circle motion is lifted to a more involved physical concept.523] . the dynamical symmetry of a ropedancer produces an equilibrium of his body which prevents him from crashing down and risking his life. the stratiﬁcation of symmetries into form and content initiates complex relations of form and content. good and evil are not equalized. is the law x ∧ p = const. of conservation of the angular momentum but see [307. local-global strategies are at least as important (see also chapter 13. literature. area per time unit relates to angular momentum3 . equally skilled tools of human insight. • Second. Hence it is not the distance between a planet and the sun which remains constant but the area which is swept out by the straight line connecting planet and sun. We also feel obliged to contradict the stubborn simplistic argument against symmetries in ﬁne arts. and this is essential to avoid the widespread belief that symmetry is an ‘equalizer’ concept. symmetries are not the only instance of artistic expression. whereas the function (signiﬁcation) of a symmetry consists of a production mode of meaning (the signiﬁcate) from the relation between meanings (signiﬁcates) of the form’s parts (expressions) which are transformed into each other under the given transformation. As a ﬁnal argument for using non-trivial symmetries in music.

This introductory section reviews the multiple and sometimes controversial presence of symmetry in music. On the other hand.2) who in 1924 applied symmetries with great success to Johann Sebastian Bach’s “Kunst der Fuge” [194]. Arnold Sch¨nberg’s (ﬁgure 8. Questioning the role of symmetries in music must always begin with the topographic initial “Where?”. neutral signiﬁcation of symmetries is explicitly given in Arthur von Oettingen’s [406] and in Sigfrid Karg-Elert’s [259] modulation theories. SYMMETRIES IN MUSIC 137 8. The discussion is complemented by reference to Jakobson’s poetic function which reveals a dominant role of symmetries in every poetic eﬀort. Such a claim fails to grasp the very concept of symmetry which in this case only refers to a privileged relation between the original and the retrograde version. . From the historical point of view there is an interesting coincidence of musicological and mathematical development regarding symmetries. Following music theorist Wolfgang Graeser we may further ask whether the incessant structural fascination of Bach’s music cannot be explained from a network of locally present symmetries. Graeser. retrograde. realizing the thoroughly o symmetric dodecaphonic composition principle (see example 12 below).1.8.1.1. In music psychology it is traditionally argued that the retrograde (see example 9 below) is not recognized (heard) as being essentially the “same” underlying melody. example 12.4. Whatever its vagueness. from the neutral and poietic point of view. see ﬁgure 8.1. Here is his summary: 4 Observe that fractals are generated by symmetries in the sense of iterated application of one ﬁxed transformation. We come back to this aspect in chapter 11. In 1928. –Σ– A classical “counterexample” against symmetries in music is the retrograde of a melody. as compared to any random melody. The question would rather be whether the listener feels a privileged relation between melody and its retrograde. the paradigm of symmetries grew into its ﬁrst order prominence in music as well as in mathematics. Around 1924–1929. 31 had its premiere. symmetries have been applied (among many others) from Johann Sebastian Bach in fugal composition to Pierre Boulez in serial techniques. Nowadays. for precise deﬁnitions) of a kernel motif (the composition’s fundamental series) was consciously designed to deﬁne a motivic paradigm. Let us just remark here that ﬁrst. symmetries are widely used in musical composition.1 Symmetries in Music Summary. We give a short historical account of the quasi-simultaneous appearance of explicit symmetry paradigms in music and mathematics in the works of Sch¨nberg. and explicitly in the author’s presto (see chapter 49) and in Opcode’s MAX or in music software with fractal4 composition tools as realized by Mesias Maiguashca [220] or by Peter Stone on his ingenious “Symbolic Composer” software. it was the aforementioned young Wolfgang Graeser (ﬁgure 8. the problem whether symmetries can be “heard” is above all an esthesic one and has—to our knowledge—not been investigated systematically in music psychology. It is very likely the ﬁrst time in the history of western music that an entire mathematical group orbit (the orbit of a given series under the full contrapuntal aﬃne group generated by inversion. symmetries play a prominent role. otherwise the eﬀort is futile. This example reveals the essential role of the topographic approach to symmetry in music in the sense of section 2. for instance in the various transformation tools of sequencers. and transposition (see subsection 8. Second. o and Noether around 1928. In music analysis.3) “Orchestervariationen” op.

Needless to say that o symmetries play a crucial role in contemporary mathematics and therefore in mathematical physics. and this is remarkable since conversely. written in 1929 [398].e. symmetries play a prominent role. an erster Stelle betrachtet zu werden.1: Peter Stone’s “Symbolic Composer” [511] software includes a large number of composition operators. i. Let us shortly digress on poesy whose structural substrate has a strong musical character. it was Emmy Noether who accomplished a persuasive theory of general symmetries. This means that associations between concepts in poesy are controlled by strict correspondence within a rhythmically segmented thread of language. In his famous statement. Wir werden in der “Kunst der Fuge” ihre fast uneingeschr¨nkte Herrschaft besonders deutlich erkennen.138 CHAPTER 8. Dante Alighieri [207] o or Fran¸ois Villon [312]. According to Jakobson. Die Eigenschaft der Symmetrie spielt in der Musik eine so ungeheure Rolle. the theory of group representations in vector spaces. a In modern abstract algebra. and since by these facts the apparently unpoetical nature of symmetries is defeated. the poetical moment in music—as it is mediated by meter and rhythm—historically grew from the linguistic poetical form. the symmetries which appear in these conﬁgurations of correspondences are a dominant means of poetical structuration [248]. in the paper “Hyperkomplexe Gr¨ßen und Darstellungstheorie”. modern poetology has indeed discovered an astonishing density and c . In its study of classics such as Friedrich H¨lderlin [247]. Here. daß sie verdient. in particular for fractal techniques.. SYMMETRIES AND MORPHISMS Figure 8. Roman Jakobson has deﬁned poeticity [245]: The poetical function projects the principle of equivalence from the axis of selection to the axis of combination.

Before starting the general theory of morphisms between local compositions. transformation etc. which encompasses a large portion of what can be thought of being a symmetry phenomenon in music. Applying all translates eno . we come back to Jakobson’s poetical function in subsection 11. . the octave class of x. .1.5.—Also in traditional European musicological analysis. For such an octave class modo (x) and an octave class k-chord Ch : 0 o-ClassChord(c1 . and we have et · x = t + x. SYMMETRIES IN MUSIC 139 Figure 8. 8. it may be identiﬁed with an element x ∈ Q3 . we should include a number of elementary examples of symmetries in music. –Σ– A word of caution: In the course of these examples we use words such as symmetry.2 for an extensive discussion in symmetry-oriented musicological analysis as cultivated in the American tradition of “musical set theory”. in an informal way. even dominance of symmetrical constructions.2: Wolfgang Graeser (1906– 1928) was the ﬁrst music theorist to apply systematically symmetry groups to music analysis. we are given a “translation vector” t ∈ Q3 which deﬁnes a translation transformation et : Q3 → Q3 and therefore a transformation of functors @et : @Q3 → @Q3 which transforms any EulerModule-denotator x : A EulerM odule(x) into its t-translate et (x) : A EulerM odule(et · x).1. Only in section 8.6. n ∈ Z of integer multiples of the octave o to x yields modo (x) : 0 o-EulerClass(eZo · x). The symmetry of transposition is fundamental for thinking in terms of pitch classes. .2 will we introduce the corresponding technical terms. ck ). Figure 8.8. If x is even a zero-addressed point in the EulerModule space. .1. Example 7 Transposition.1 Elementary Examples Summary. c2 .3: Arnold Sch¨nberg (1874– o 1951) was the ﬁrst composer who consciously applied full orbits of symmetry groups to musical composition. see section 11. symmetries have been recognized with growing success [395]. More precisely.

. retrograde. 2.2.7)) we can refrain from adding the class attribute here. SYMMETRIES AND MORPHISMS we can consider the translate emodo (x) · Ch = {emodo (x) · c1 . translation of a local composition G in an ambient space S = Onset ⊕ P ara by et . Translation et on the simple space Onset of onsets means something very diﬀerent from the same mathematical transformation on pitch domain. Example 9 Inversion.140 CHAPTER 8. 4. i. 4.2. 8.2. the translation et · G in the sense of rhythm theory as introduced in section 7. We have interiorized this symmetry to such a degree that we do not even realize it as being essentially diﬀerent from identity. 10}. Intuitively. periodic repetition of a given “percussive pattern” as a generation principle for rhythmical structures. the translate emodo (11) · Cmaj is equal to the 12-tempered B-major scale Bmaj = {11. da capo. 5. retrograde inversion. 9.2. its visual shape depends on the selected pitch space. . means playing G a second time. 6 Pay attention to distinguish the symmetry of inversion from chord inversion which just means moving some of their notes one octave up or down. • We can also repeat parts of a composition within the composition in the sense of a da capo. 7.e. In the technically ideal case. .. 6. Let us have a closer look at the highly instructive analysis of these concepts and their multiple meaning and corresponding mathematical interpretations. These are classical symmetries in composition and theory. rhythm. emodo (x) · c2 . CD. this is a point where Sergiu Celibidache [476] was right.. However. it is often and erroneously felt that any such reproduction is the piece itself. . 11}. the just B-major scale (chord) is obtained from the just C-major scale (chord) via a translation eq+t = eq · et of a ﬁfth (class5 ) q plus a major third (class) t. • Finally. 3. • A special case of a repetition within one and the same composition is the duplication of a melody in the canon construction. etc. i. canon. 1. emodo (x) · ck }. inversion6 is a reﬂection of the pitch axis at a ﬁxed pitch. repeating it later in the course of world time. We have four diﬀerent common meanings in this case: • If we view the onset axis as being a time line. every replay is a translation on the physical time axis.e. whereas in just temperament. If for example Ch is the 12-tempered C-major scale Cmaj = {0. LP. Example 8 Repetition. This seemingly trivial fact is the case if we replay a composition on a media such as tape. we may view the looping. Everyone of the hitherto discussed pitch spaces P tch 5 Since the ﬁfth and third coordinates are not touched by factorization through the octave in just classes of the Euler plane (see equation (7. this was already introduced in section 7.

below: bar 33) represent the characteristic motif of this fugue. w Then inversion is an aﬃne symmetry of shape Us = es · −1 which sends a denotator D : A P tch(x) to Us D : A P tch(Us (x) = s − x). For ﬁxpointless inversions between pitch g and g . the ﬁxpoint is indicated. EulerM odule has the Q-vector space coordinator Q · o ⊕ Q · q ⊕ Q · t = Q3 . shows an inversion at g in 12-T emperedScale. were of simple form: P itch has the R-vector space coordinator R.e. and w 1 ∼ w-P itchClass has the Z-module coordinator Z · · o/Z · o → Zw . for example. in Johann Sebastian Bach’s “Wohltemperiertes Klavier” I. followed by a shift es . the inversion is a reﬂection in the . BWV 851. The symbol Ug is written in the traditional notation. EulerP lane has the Q-vector space coordinator Q · q ⊕ Q · t = Q2 . SYMMETRIES IN MUSIC 141 Ug Pitch Figure 8. 1 has the Z-module coordinator Z · · o. the w-tempered scale from origin 0. we have 2t = s and 2x0 = 0.. w-T emperedScale. otherwise.8.4: This example from Fuga 6.1. whereas in case of w-T emperedScale. For real or rational coeﬃcients in the ﬁrst four forms this yields a unique solution 1 t = 2 s and x0 = 0. M athP itch has the Q-vector space coordinator R[Q] . The inversion has a ﬁxpoint x = Us (x) iﬀ with the unique decomposition x = et · x0 into translation and linear factor. This is inversion at the origin of the coordinator module. The grid shows semitone steps and quarter notes onset units. that notation would yield. i. The motivic note groups shown here (above: bar 29. 1 and there is one iﬀ s is an even multiple of w . we have at most one solution. D minor.

showing zero-addressed denotators again for simplicity. duration. If we try to invert time naively as it is done by inversion of a tape. On . Z2 ). Intuitively. or in the middle of one mesh side (as in the case of this ﬁgure). this idea is not quite clear. we show two inversions U3 and U4 on zero-addressed denotators in 12-P itchClass. the ﬁxpoint will stay either on one vertex of the integer grid. the retrograde symmetry is a reﬂection of time’s run down. 1 1 middle between two adjacent points η w o. or in the middle of one mesh. Exercise 5 Calculate all ﬁxpoints of an inversion Us for self-addressed denotators D : Z12 12-P itchClass(x). something diﬀerent from musicological “retrograde” happens. the condition s ∈ 2Zw is not automatic. To the right we have inversion Uq on the EulerP lane space. The second 4 inversion is U4 has ﬁxpoint 2 = 2 and would traditionally be denoted by Ud = Ug . To begin with. Let us select the mental onset space Onset with simple coordinator R. we are dealing with aﬃne di-homomorphisms in x ∈ A@Zw . If w = 2q is even. Time intervenes in onset.5 show the picture of an inversion in the pitch space 12-T emperedScale and in pitch class spaces for 12-tempered and just tuning. If w is odd. t.5: To the left. So let us have a more detailed look at the retrograde phenomenon. If the inversion’s shift vector p has integer coordinates with respect to the grid’s basis q. and we are looking for numbers s ∈ 2Zw and di-linear homomorphisms x0 : A → Zw with 2im(x0 ) = 0. onset is concerned—independently of whether we consider mental or physical time. 2 has an inverse in 1 Zw . and envelope of a sound.4 and 8. sound color. This symmetry can be viewed as a 180◦ -rotation around the ﬁxpoint. Figures 8. (η + 1) w o.142 CHAPTER 8. It leaves invariant the middle q/2 between origin and q. The ﬁrst has no ﬁxpoint and is also denotated by U1/2 or Uc /d in traditional notation since its virtual ﬁxpoints are between c and d as well as between g and g . For form w-P itchClass. Since several attributes are related to time. SYMMETRIES AND MORPHISMS t Ÿ12 U3 Uq Ÿ12 U4 q Figure 8. and ﬁxpoints are given by t = 2 s and the zero-di-homomorphism x0 = 0. then obviously. whereas the di-linear factor must map into the ∼ subgroup qZw → Z2 and is parametrized by Lin(A.

s − onset}. and section 6. we have to work in the two-dimensional space OD = Onset ⊕ Duration with coordinator R2 as deﬁned in formulas (6. usual retrograde concerns onset as well as duration.6. ks acts on onset. duration1 ) 7 The letter “k” stands for German “Krebs”. In order to include the role of duration in deﬁning retrograde symmetry. see bottom part of ﬁgure 8. ks (oﬀset)} = {s − oﬀset. the transformed couple will have a disturbing eﬀect because ks (oﬀset) will not end when ks (oﬀset) starts. such as Duration D2 D1 Duration Ks(D2) Ks(D1) ks(D2) ks(D1) duration2 duration1 onset offset Onset Onset reflection Duration ks ks(D2) transvection ks(D1) ? ? Onset Figure 8. if the duration of the second event at time oﬀset is longer than onsetdistance. in drum patterns. oﬀset}. . If we apply ks to such a couple {onset.6. oﬀset} = {ks (onset). In fact. SYMMETRIES IN MUSIC 143 this space. retrograde symmetry means the complete analogy to pitch inversion and can be written7 as ks = es · −1 for a shift number s. and the “retrograde” of the second event will start onsetdistance before the “retrograde” of the ﬁrst event.39) ﬀ. this one-dimensional symmetry is somewhat strange if we look at its eﬀect on two events at times onset and oﬀset = onset + onsetdistance which are onsetdistance apart. Whereas Us has the same formula and acts on pitch. However.1. we obtain ks {onset.8. However. this transformation is not what usually happens when a score is played from the right to the left in reversed order. So we have to take two events OD(onset.6: In contrast to pitch inversion. It is composed from a horizontal reﬂection (resulting in the lower conﬁguration) and a horizontal transvection (shown in the right upper part). Such a symmetry may be interesting if we disregard durations.

Observe that sound colors are not altered by this symmetry. duration1 ) and OD(ks (oﬀset) − duration2 . for example! Figure 8. see ﬁgure 8. retrograde inversion KUs. We ﬁrst perform ks on their onset parameters and obtain an intermediate transformation OD(ks (onset). duration2 ) as shown in the lower part of ﬁgure 8. and duration. SYMMETRIES AND MORPHISMS and OD(oﬀset. It involves three parameters: pitch.t is deﬁned as composition of a retrograde symmetry Ks and a pitch inversion Ut . a Finally. Examples of this duration-sensitive symmetry are abundant in classical European literature. d) → (s − o − d. and we 8 See appendix E.6. Putting all together we have deﬁned a duration-sensitive retrograde which is given by the aﬃne map Ks : Q2 → Q2 : (o.7.3.1) and extends functorially to any A-addressed denotator D : A OD(D : A → Q2 ) if we set Ks (D) : A OD(Ks ·D). Bach: retrograde canon from “Musikalisches Opfer”. We do obtain what is required in a fugue for piano.144 CHAPTER 8. d) which means Ks = e(s.7: J.6.4 .6. duration2 ) as shown in the upper right part of ﬁgure 8. BWV 1079 (with kind permission of B¨renreiter Publishers). duration1 ) and OD(ks (oﬀset). Ths latter is a horizontal transvection or ‘shearing’ operation8 .S. we go one step further and shift the intermediate onset by the event’s duration to the left and get the events OD(ks (onset) − duration1 . see left upper part in ﬁgure 8. duration2 ).0) · −1 −1 0 1 (8. onset. To avoid overlaps due to durations.

and which we have dealt with above.t = e(s. We stick to the sound description in Fourier synthesis space F ourierSound −→ Limit(P hysOnset. KUs. Envelope) Id introduced in (6. Then the transformed sound function evaluates to ∞ pβσµ (t) =βV (((σt + µ) − e)/d) · 0≤i Ai sin(2πif (σt + µ) + P hi ) (βAi ) sin(2πi(f σ)t + (2πif µ + P hi )). It is reported that Beethoven.2) 0 0 1 which means Ks ·Ut = Ut ·Ks if these symmetries are canonically lifted to the three-dimensional space OP D. to name a typical operation.1. Let us consider this latter situation to review some details of what happens when a tape is reversed. µ of real numbers. d.8.0) · 0 −1 0 (8. we obtain a new time function pβσµ which at physical time t evaluates to pβσµ (t) = βp(σt + µ). since one normally groups the basis parameters and then adds the pianola parameters. . (8. This type of transformation on the sound level is a standard processing tool for analog and digital audio sampling technology. SYMMETRIES IN MUSIC 145 may consider the typical three-dimensional space9 OP D = Onset ⊕ P itch ⊕ Duration with coordinator Q3 . For all triples β. We then obtain the formula −1 0 −1 KUs. myF ourier. P hysDuration. in one of those then famous competitions among pianists. Let the Fourier denotator be given by myF ourier = (f. Applying symmetries to abstract sound descriptions as they occur in score notation. Example 10 Sound transformations. Beethoven had just turned around 180◦ Steibelt’s string quartet score and played that symmetric transformation of the composition—evidently turning it into garbage (in those days it very probably was garbage). Let us look at the eﬀect of this symmetry type on the underlying Fourier denotators. is completely diﬀerent from applying symmetries directly to sound events on the physical level or reality.108).3) For σ = −1. V ) which give rise to physical sound function p = sound(myF ourierSound).t. More practically interpreted. (Ai . σ = 0. F ourier.t is a rotation of 180◦ on the score plane. forced his contrahent nicknamed “Tremolo Steibelt” to leave the salon steamed up. =V ((t + µ−e d )/ ) · σ σ ∞ 0≤i 9 We choose this sequence of parameters instead of P OD. we obtain a reversed run though time and therefore the well-known tape reversion. The space contains denotators myF ourierSound = (e. P hi )0≤i ). for example.

The prize for this invariance is in particular that duration may become negative for negative σ. This is the case where the sample will run in reversed direction. V ) σ σ (8.8: A sound sample and its transform by a symmetry Qβσµ . The major-minor problem is one of the classical schisms in occidental music theory. 2πif µ + P hi )0≤i . i.146 and we have the aﬃne transformation CHAPTER 8. pβσµ as being zero-addressed denotators of form 0 Sound −→ Simple(Ccp ) Id 0 where the coordinator Ccp is the real vector space of functions f : R → R which are continuous and have compact support10 . d. showing time inversion. say. which is the situation in various sound synthesizer drivers.1. We are not going to shed philosophical light on this conceptual ﬁeld but will concentrate on the concrete and most 10 The 11 See support supp(f ) is the topological closure of the set {x ∈ R| f (x) = 0}. The problem is to understand the relation between the “major” and “minor” paradigms.3. V ) → ( µ−e d . The loudness change is solely attributed to the Fourier sum expression.8.4) which relates the symmetry on the sound level to the underlying parametrization. Let us make this a bit more transparent for later constructions. sound(myFourier) sound(myFourier)bsm Figure 8. P hi )0≤i . (Ai .. We also see that the envelope is left invariant by Qβσµ which means that it can even be taken as an envelope normed to maximal value 1. Example 11 The major-minor problem. (f. see also ﬁgure 8. the map sound : myF ourierSound → sound(myF ourierSound) commutes with the above symmetries. appendix C. (f σ. (βAi .5) which means that sound is an equivariant map11 for the two symmetry actions Qβσµ and ?βσµ . . we have sound(Qβσµ (myF ourierSound)) = sound(myF ourierSound)βσµ (8. . If we interpret the sound functions p. SYMMETRIES AND MORPHISMS Qβσµ : (e. example 70 for this concept.e. With this.

e.1.3. Exercise 6 Describe all aﬃne endomorphisms f of EulerM odule which transform the major chord into the minor chord. 14 German: “Tr¨ bung”. Carl Dahlhaus [100] has interpreted this fact as contributing to the erosion of the concept of a harmonic function13 . We should however ask whether turbidity has nothing to do with symmetry and must therefore be structurally inferior to dualism as being founded on a purely emotional level. This view is charged by the emotional connotation of gloom as being conveyed by the minor triad. Already Johann Wolfgang von Goethe did not accept this psychologically charged theory which had its roots long before Hindemith.6) 0 1 12 We shall discuss the concept of cadence in more detail in chapter 26. g} into the minor class chord {c. This symmetry extends to a transformation of the C-major scale into the C-minor scale: Uq (C-major) = C-minor. t − 2 · o. Here. q − o). Let us consider the C-major scale (scale chord. e. −t + o.and Uq (g) = c. To discuss this subject in simpler terms. We thus obtain the symmetry −1 −1 A = eq · (8. d}. let us look at the situation in the EulerP lane space. for example) to the minor third (c − e in {c. e . The key to this unexpected symmetry lies in the inner symmetry of the major scale itself. g}. I = {c. This contradicts the function of dominant and subdominant in Riemann’s harmony in that the fourth degree in minor should have a subdominant function whereas the ﬁfth degree in minor should have dominant character. see ﬁgure 8.e. q − o) becomes a minor chord EulerChord(0. t − 2 · o. we have the inversion symmetry Uq which converts the major class chord {c. e . e . a . g}. This fact was used by Arthur von Oettingen [406] and by Hugo Riemann [450] to deduce the minor cadence from dualism. g}. V = {g. It states that the minor triad is derived from the major triad by a “turbidating” lowering shift of the major third (c − e in {c. g}). more precisely. to be precise) to ﬁx the ideas.8. u 13 We . the major chord EulerChord(0. b.5 to evidence this fact. g} of major triadic degrees is transformed into the sequence I = {c. e. Then we add a shearing by 45◦ in direction −q. a. IV = {f. e . b . In fact. e. −t + o. he realized that in just tuning. come back to Riemann’s function theory in section 25. V = {g.9. We want to show that this is not true: The theory of turbidity is based on symmetry just as dualism is. it was Gioseﬀo Zarlino [591] who initiated the subject by the observation that there is a dualism between major and minor. Under the “dualism” Uq the major cadence12 sequence I = {c. Uq (e) = e .. Uq (c) = g. I = {c. q − o)) = EulerChord(0. i. q − o) after a symmetry of EulerM odule. The theory of dualism is confronted with the theory of turbidity14 which was prominently forwarded by Paul Hindemith [224]. d}. c}. The inner symmetry of C-major runs as follows: It is a reﬂection of the EulerP lane space at the vertical line through q/2. g} of minor triadic degrees. c}. f (EulerChord(0. SYMMETRIES IN MUSIC 147 common context of just temperament where this discussion is incessantly virulent. g}. Historically and theoretically. IV = {f. but see ﬁgure 7.

followed by a horizontal transvection of 45◦ . and a ⇔ b. And this does the following: Uq · A(c) = c Uq · A(e) = e 15 We Uq · A(g) = g Uq · A(a) = a Uq · A(f ) = f Uq · A(b) = b . SYMMETRIES AND MORPHISMS inner symmetry refl ecti on sv tran on ecti Figure 8. there is only the above A as an “inner” symmetry15 of C-major. the above inversion U is termed real inversion. example 60) how real and tonal inversion relate to each other.7) shall make this statement precise in section 8. f ⇔ d. then A · Uq evidently deﬁnes another symmetry from C-major to C-minor. We shall see in connection to the formalism of alterations (section 49. if Uq deﬁnes the dualism. In music theory such a reﬂection is called tonal inversion since the transformation is performed within the given “tonality” if we think of the white keys as being equidistant (which they are not). Uq · A(d) = d (8. and is the “tonal inversion” at the scale’s third from the tonic on the keyboard (e in C-major). Besides identity. . it turns out that we obtain a reﬂection of white keys (C-major) at key e. To distinguish the terms.3.148 CHAPTER 8. The black keys are so to speak eliminated. If we execute this symmetry on a keyboard.9: The inner symmetry of a major scale in just tuning. The interesting fact about this non-trivial symmetry is that this one mediates between the theories of dualism and turbidity! In fact. with A(C-major) = C-major where e remains ﬁxed whereas we have exchanges c ⇔ g. This horizontal reﬂection at a skew axis is composed of a horizontal reﬂection at a vertical axis through q/2.2.

In the framework of dodecaphonic composition. Summarizing: Scholion 1 The schism between dualism and turbidity is only apparent. o abuse of language with respect to denotator formalism is fairly acceptable. 16 from 17 This German: “Zw¨lftonreihe”. SYMMETRIES IN MUSIC 149 We recognize that A · Uq lowers exactly the three tones e. n)-series is called n-phonic series. n)-series. Ser1 . and we have learned that the unique inner symmetry A of major mediates between dualism and turbidity.43) and look for local compositions which are motives in this ambient space: Deﬁnition 22 Let k.2 op. 28. Figure 8. each showing diagonal or/and codiagonal symmetry. They deﬁne the paradigm associated with a dodecaphonic series16 . If these coordinates are pairwise distinct. This makes evident that turbidity is just the other symmetry A · Uq from major to minor. . symmetries play a fundamental role. An (n. we invoke the ambient space P iM od12 from (6. 17. string quartet op. n be positive naturals. Berg. in particular a dodecaphonic series for n = 12. Serk−1 ) with coordinates17 Seri = Ser(ei ). We shall also encounter the fundamental inner symmetry A when developing modulation theory in chapter 27. o op.10: Three dodecaphonic series by Webern. . This one mediates between dualism Uq and turbidity A · Uq so that we are allowed to call both approaches equivalent from the symmetry point of view. To make this compositional framework more precise. var. . This means that they generate a ﬁeld of twelve-note series which are deduced from a basic form of the series. 30. and Webern. A (k. the serial motif is called a (k. n)-serial motif is a Zk−1 -addressed denotator Ser : Zk−1 P iM odn (Ser0 . b by a semitone. Example 12 Sch¨nberg.8.1. a. and this is the process which turbidity theory describes. . It reduces to understanding the inner symmetry A of the major scale.

8) where. this axis will not be present in a straightforward form within a concrete score.150 CHAPTER 8. “Grundgestalt”. Webern’s and Berg’s twelve-note como position is a poietical and mental principle.n .1. We shall come back to the systematic aspects of this method in section 10. simplify notation. Joseph Rufer has exposed this approach in detail in [465]. A typical element g = (h. In our generic setup. n)-serial motifs is denoted by SERMk. The 48 versions are deﬁned by the set D12.2. pitch inversions Us . it may however occur simultaneously in a chord. We discuss this issue in more detail in section 11. SYMMETRIES AND MORPHISMS The set of (k. The subset of (k. See ﬁgure 8. this group is the direct product group Dk. a note with larger onset index should not appear earlier than a note with smaller index. 19 To 18 German: . n)-series is denoted by SERk.Ser)i = f (Serh(i) ).9) running over a set {Serr } of representative series. onset retrograde revk . There is an unsolved problem in this approach: It is not clear how far the poietical approach in Sch¨nberg’s method may be transferred and evidenced on the neuo tral and/or esthesic levels. and we abbreviate SERMn. and their combinations19 . Semiotically.12 } (8. Essentially. In general.n = SERMn . The genealogical principle in twelve-note music starts with a particular dodecaphonic series Ser—called basic series18 —and then creates 48 transformed versions according to a complete group of symmetries. it gives you a procedure to fabricate a composition. Olivier Messiaen’s composition technique [370] intends to treat all parameters of time and pitch equally.n = SERn .10 for historic examples. and where revk is the ‘retrograde’ address involution deﬁned by revk (ei ) = ek−1−i .1. however.n = revk × T In where T In = eZn −1 . As a matter of fact.12 · Ser = {c. The coordinate index i is viewed as being an ‘abstract’ onset time line which controls the compositional process. the historically problematic and poor reception of dodecaphonism shows that a poietical paradigm cannot only be practised. also called “paradigms” SER12 = r D12. According to the general theory of group actions (see appendix C.5. the dominant role of symmetries is made explicit. f ) of the group acts on a serial motif Ser by (g.12 have the task to build a paradigmatic ﬁeld around the basic series. the symmetries from D12. The musical meaning of symmetries in Sch¨nberg’s. several elements may coincide.12 · Serr (8. its protagonists should also take care of mediating it for the sake of public’s recognition and acceptance on the esthesic level. and we abbreviate SERn. Here we merely make evident the structural context and the role of symmetries. example 70). the set SER12 of all twelve-note series splits into a disjoint union of twelve-note orbits. A general rule in twelve-note composition is that within a local realization of the basic series in the particular score. This group consists of transformations known from classical counterpoint: transpositions es .3.Ser| c ∈ D12.n . Example 13 Messiaen modi and non-invertible rhythms. we denote Us and k11 without adding the identity components in the respective other coordinate.

i. among which Messiaen for obscure reasons omits M9 = {c. c . f. From the classiﬁcation theory in section 11. For a given mode M . e. Messiaen introduced modes with limited transpositions. g . b } 4. c . e. f . mit dem Unterschied. d. g . g . Further. M3 = (augmented triad) = {c. f . 2. b} = {c. 7 as listed by Messiaen: 1. g. d . d . the smallest candidate for such an invariant transposition is t = 1. Diese Modi lassen sich in symmetrische Gruppen teilen. d. 3.1. ¨ ohne daß man wieder in dieselben T¨ne hineinger¨t (enharmonisch gesprochen). These are special scales in Z12 . d . M4 = {e. c . a. M1 = whole-tone scale = {c. Let us anticipate the discussion by the original text [370]: Diese Modi realisieren im Vertikalen (Transposition). c . We come back to this selection problem when dealing with local classiﬁcation in section 11. f . g. diese Rhythmen auch.5 we learn that—up to transposition—there are ten such scales. o a ebenso k¨nnen diese Rhythmen nicht r¨ckl¨uﬁg gelesen werden. d. d. if M has this limited transposition. f. g . d. M5 = {c. The explicit selection criterium which is recommended and felt by Messiaen is limited transposition. SYMMETRIES IN MUSIC 151 In the dimension of pitch. ﬁnally. g } 6. c . 4. e. M6 = {d . g. a. M7 = (tritone) = {c. u Die Analogie ist also vollkommen. and M8 = {c. diese Rhythmen k¨nnen nicht umgekehrt werden. i = 1.8. dass die Symmetrie der rhythmischen Gruppen eine r¨ckl¨uﬁge ist. b } 3. f. weil sie in o sich selbst keine Umkehrungen enthalten.3. a. d. g. c . Here are the remaining seven scales Mi . b . the way Messiaen uses them to introduce non-invertible rhythms is quite disquieting. b} = {c. Messiaen’s list includes seven modes which are however not selected according to any evident logic. g. d . weil sie sich — ohne Polytonalit¨t — in der o a modalen Atmosph¨re mehrerer Tonarten zugleich bewegen und in sich keine Transa positionen enthalten. e. a} as well as its inversion M10 . . f. g . This means that there is a nonzero pitch class t ∈ Z12 such that a Messiaen mode M → 0@P iM od(12) remains invariant under the transposition by t: et · M = M .3.e. the full chromatic “mother” scale M = Z12 can be omitted as a trivial case. a}. . M2 = (diminished seventh chord) = {c. g . a} 5. Diese Modi k¨nnen nicht transponiert werden.. f . in 12-tempered tuning. 6. .5. a. Endlich hat jede Gruppe dieser Modi jeu a weils ihren letzten Ton gemeinsam mit dem ersten Ton der folgenden Gruppe. its complement M shares this property. Tats¨chlich k¨nnen diese Moa o di nicht uber eine gewisse Anzahl von Transpositionen hinaus transponiert werden. f . b } Whereas Messiaen’s modes are not problematic to deﬁne. ohne daß man geo u a nau dieselbe Anordnung der Werte wiederﬁndet wie in der Grundform. f . g. . was die nicht-umkehrbaren Rhythmen im Horizontalen (Umkehrung) realisieren. f . und die Gruppen dieses Rhythmus umrahmen einen f¨r beide gemeinsamen Zentralwert. e. b } 7. e. and we may concentrate on M having 6 ≤ card(M ). c . b } 2. c .

followed by dilatation by 2.2.11. The least we can say is that we have P ara = Duration⊕P ara0 . There is not the least reason to invoke “perfect analogy” here. However. SYMMETRIES AND MORPHISMS The last sentence is as typical for mathematically oriented phrasings by musicians as it is erroneous. And this symmetry is very diﬀerent from the simple transposition as we know from the preceding discussion. Pierre Boulez. Let us start with the underlying modules for modes and rhythms. it is not suﬃcient to consider exclusively durations since we have to include pauses in Messiaen’s rhythms. Serialism is derived from the idea that Sch¨nberg’s series should be extended to all sound o parameters. each provides exchange of pitch and onset as an inner symmetry. see ﬁgure 8.2). non-invertibility is deﬁned in terms of onset and duration. a fact which is responsible for establishing criteria of compositional esthetics. and therefore. Figure 8.1): It is the symmetry of “reading from right to left what is normally read from left to right”. As instances of the compositional process. In fact. Messiaen seems satisﬁed with the fact that there is symmetry at all.152 CHAPTER 8.10 shows three such series. Kagel. As a point of departure for this approach we have to consider Messiaen’s construction of non-invertible rhythms as a codex for durations in analogy to his scales (see Example 13). such√ serial transformations include rotation by 90◦ or rotation by 45◦ . Maurizio Kagel. Karlheinz Stockhausen. The analogy is anything but perfect. non-invertibility means this: Messiaen limits himself to indicate just one germ of a rhythm. not only the rhythm concept but also the concept of symmetry includes onset and duration. Stockhausen. Now. respectively. This more general type of symmetries leads to the general concept of morphisms of denotators to be introduced in section 8. For modes. Under these conditions. Further. symmetries have been considered by exponents of the serial school. or Stockhausen’s “Kontra-Punkte”. i. For instance [138]. Boulez’ “Structures pour piano”. serialists start from treating all parameters equally. The musical meaning of symmetries is again of poietical and mental topography and serves—like dodecaphonism— the generation of a structurally speciﬁed paradigmatism of musical gestalts wherein hitherto . the coordinator where Messiaen’s rhythms are deﬁned is Onset ⊕ Duration ⊕ P ara0 . retrograde transformation is also a symmetry in the mathematical sense (to be discussed in detail in section 8. germ = Ks · germ.2. and Boulez. in other words: to elaborate Messiaen’s analogy into a symmetry of the parameters’ roles. we cannot stick to local meters L → 0@Onset of period p since durations play an important role with Messiaen.2 where the parameter space P ara includes duration as well as possibly some additional characterization to diﬀerentiate pauses from non-pauses. and to ask from germ that it be symmetric under a retrograde symmetry. no such analogy as suggested by Messiaen can be discovered.e. Example 14 Serial techniques: Eimert. But to deal with rhythms. Another example is Webern’s symmetrization of the roles of onset and pitch in his twelve-note series. It has been extended to the degree of admitting more general geometric transformations on parameter spaces in order to generate new sound conﬁgurations from given ones. and other composers. Apart from the fact that like transposition. This does nothing else than reconﬁrm the dominant role of symmetries in Messiaen’s approach. and this means that we deal with rhythms in the technical sense of deﬁnition 14 in subsection 7.. The idea of a parameter exchange was adopted by Herbert Eimert. Albeit such an analogy would be feasible. we clearly have pitch classes referring to coordinator Z12 . Examples of compositions using such methods include Kagel’s “Transici´n o II”. The retrograde which Messiaen refers to in [370] is precisely the one which we have denoted by Ks in formula (8.

Could it be that in special situations several parts correspond to one single part? From traditional mathematical terminology.1. a chord Cr : A EulerChord(Cr) projects onto a class chord modo (Cr) in o−classChord.11). it is allowed to play a selected part P of the fundamental series Ser or of its transforms c(Ser) as a chord. It results from a counter-clockwise √ rotation by 45◦ . . In twelve-note composition. no two elements of P are identiﬁed under π2 .. In this context. In deﬁnition 21 of symmetry as given at the top of this chapter. correspondence of parts is mentioned.) Example of a serial transformation described by Eimert in [138]. i. simultaneously. SYMMETRIES IN MUSIC 153 Figure 8. symmetry transformations are understood as being one-to-one. separate musical parameters become freely interchangeable. But the dodecaphonic paradigm precisely means expressing the compositional parts (such as chords) as instances of a whole. but elements of the original chord which are multiples of the octave apart are identiﬁed under modo . Clearly. Let us in fact give two easy examples of symmetries which do not involve one-to-one correspondences. followed by dilatation by 2.8. ©1964. In particular. Example 15 Non-invertible symmetries. Here.e. But this is not mandatory. symmetry acts as an identiﬁer and is far from one-to-one. The correspondence p → π2 (p) for p ∈ P does not conserve temporal order.11: (From [138]. with kind permission of Universal Edition. The second example is the nearly invisible symmetry of projection modo from the pitch space EulerM odule onto the classical octave pitch class space o−EulerClass (7. The ﬁrst one is projection. and this is rightly done by projection in this case: the projection π2 (P ) is in correspondence with P (and indirectly with Ser) as an expression of a unifying principle. There is one point which we must make clear before leaving the level of examples. But it is not speciﬁed what should be the nature of this correspondence. but we can no longer tell the temporal order of the original elements. Vienna. the correspondence under modo creates a semantic added value since on the side of o − classChord. and we consider its projection π2 (P ) ⊂ P itch which is a chord. it is not clear whether such a correspondence should be one-to-one. we have a subset P ⊂ c(Ser). Nonetheless. Mathematically.

This section introduces the formal deﬁnition of morphisms between local compositions. We shall now give the formal deﬁnition of a morphism between local compositions on two levels of abstraction: ﬁrst for objective local compositions. • The shape of this correspondence must rely on a generic type of structural transformation which operates on the ‘background’ of the local composition’s ambient space. is understood that the notation K stands for K → A@F and has to be explicated if ever necessary. We come back to this issue in chapter 47.1. we use the usual shorthand A@F to denote A@F un(F ). we shall show how to integrate morphisms among completely general denotators in this concept framework. In a last step.154 CHAPTER 8. There may exist many different background transformations which yield the same correspondence. This is a delicate subject since it makes visible interface problems between what is a constructive background and what is on the surface of a given composition. a morphism has to deal with a correspondence between the denotator ‘points’ which are involved in local compositions. We already saw in the example of dodecaphonic composition that the relation between compositional parts on the score and poietic genealogy (preliminary and subsidiary constructions by use of symmetries) is of fundamental importance to the communication of the composers message. Deﬁnition 23 (Morphisms for objective local compositions) Let K → A@F and L → B@G be two objective local compositions20 at addresses A and B and ambient spaces21 F and G. harmonic progressions only become visible under such a correspondence. the essential is only that we ﬁnd at least one of them which induces the visible correspondence. We shall then establish the connection between the two levels and show that everything ﬁts well. 21 With 20 It . It is not the same to write down an explicit correspondence among elements of two local compositions and to write down a mathematical transformation rule which is responsible for that correspondence.1. • The background transformation is only subsidiary and should not be confused with foreground correspondence which really identiﬁes the morphism. their functors being extracted for the functorial calculations. –Σ– From the representative examples in subsection 8. second on proper functorial local compositions.e. The deﬁnition is motivated by representative examples from music as discussed in subsection 8. 8. This short discussion yields another problem which underlies symmetry theory: the codiﬁcation of a symmetry transformation within a determined correspondence. In fact.2 Morphisms of Local Compositions Summary. SYMMETRIES AND MORPHISMS harmony furnishes structural insight which is not visible on the original chord.1 we have to draw three essential construction basics for general morphisms between local compositions (and whatever will be derived from this nucleus): • As a realization of the general idea of symmetry.1.. i.

10) Lα − − → A@G −− commutes (Lα being the image of L under the functorial map B@G → A@G. also denoted by α if no confusion is possible).11) if the underlying ambient spaces are clear. the diagram K − − → A@F −− α. (8.1. Let K → A@F . MORPHISMS OF LOCAL COMPOSITIONS 155 respectively. such that there exists a morphism of forms h : F → G with α.1 and give interpretations in terms of the above deﬁnition. QED.g (8.f A@h (8.f is induced by h : F → G.. we also write f instead of f /1. Explicate all possible underlying symmetries.gf = α(β. L.g) which is uniquely determined by β.g) β. f. i.8.f : K → Lα → M βα which is induced by A@h · A@k = A@h · k. Proof. Exercise 7 Review the examples of symmetries in subsection 8.g)α. and suppose that α. This morphism is called the composition of f /α and g/β. Any morphism h of forms which gives rise to f /α as above is called an underlying symmetry of this morphism. L. Lemma 5 If f /α : K → L and g/β : L → M are two morphisms of objective local compositions. an address change α : A → B and a set map f : K → L. naturality of k with respect to α deﬁnes a set map α(β.12) M β − − → M βα −− commute. L → B@G. α) composed of K.g) A@k M βα − − → A@H −− and hence the composed set map βα.2. whereas β.g and by α and makes the diagram L − − → Lα −− α(β.f A@h Lα − − → A@G −− Lα − − → A@G −− α(β. A morphism from K to L is a quadruple (K. This morphism is denoted by a fraction f /α : K → L.e. We also have these commutative diagrams K − − → A@F −− α. If the morphism’s address change denominator α is the identity 1 = IdA .g is induced by k : G → H. and M → C@H be the complete data.f = A@h|K . The following is straightforward from the above deﬁnitions: . To begin with. then so is g/β · f /α = gf /βα : K → M .

and where f : K → L is a natural transformation such that there exists a morphism of forms h : F → G such that the diagram K − − → @A × F −− f (8. Let us next look at morphisms in the functorial setup. then we have associativity of composition: (w/γ · v/β) · u/α = w/γ · (v/β · u/α) = w/γ · v/β · u/α. As with objective local compositions such a morphism is denoted by f /α : K → L. we have 1L · u/α = u/α = u/α · 1K . the deﬁnition of a morphism between objective local compositions boils down to the classical deﬁnition of morphisms of local compositions in [340]. these local compositions can be viewed as subsets of their coordinator modules. 22 See appendix G. (ii) If u/α : K → L. For simple ambient spaces and zero-addressed local compositions. In fact.2.2) are precisely the aﬃne homomorphisms between these coordinator modules inducing set maps in the classical deﬁnition. B be two addresses. v/β : L → M . associativity) (i) For an A-addressed objective local composition K. α) where α : A → B is an address change.13) @α×h L − − → @B × G −− commutes.156 CHAPTER 8. Deﬁnition 24 (Morphisms for functorial local compositions) Let A. (iii) If u/α : K → L.1 for ﬁber products or pullbacks. and let K → @A × F and L → @B × G be two (functorial) local compositions. the identity 1K = 1/1 = IdK /IdA : K → K is a morphism. we could ask for a natural transformation fα : K → Lα where Lα is the ﬁber product22 subfunctor of @A × G deﬁned by the pullback diagram Lα − − → @A × G −− @α×Id L − − → @B × G −− and where we ask for an underlying h : F → G such that K − − → @A × H −− @1 ×h A G fα Lα − − → @A × G −− commutes. and w/γ : M → N are three morphisms. SYMMETRIES AND MORPHISMS Sorite 4 (Existence of identity. and the underlying morphisms by the Yoneda lemma (appendix G. . Exercise 8 Equivalently. A morphism from K to L is a couple (f.

It remains to be shown that f is well-deﬁned. But we know that A@h(k) ∈ Lα. This functorial morphism K → L is denoted ˆ ˆ by f . and canonical factorization for functorial local compositions) The statements of sorite 4 are literally true if we replace local compositions and morphisms by their functorial homonyms. Of course this terminological duplication is not accidental and we can in fact establish a complete embedding of the objective setup in the functorial one as follows. k·g) = (αg. i. Therefore the right vertical arrow is uniquely determined by f .. and we have deﬁned a map ˆ : f /α → f /α ˆ ? (8. ˆ So let us calculate the image of an element (g. then their composition g/β · f /α : K → M is deﬁned by g/β · f /α = gf /βα.14) ˆ α@L ˆ − − − → αx@L −−− surjection with surjective horizontal arrows. for any address change x : X → A. the fact that the composition is also a morphism is immediate.e.3 that every objective local composition K → A@F has its functorial counterpart K with ˆ slice f @K = {f } × K · f for any address change f : X → A. Moreover. k · g) ∈ g@K for g : X → A. the evaluation of f at IdA sends IdA @K to α@L and yields α.e. Suppose that we are given a morphism f /α : K → L of objective local compositions with L → B@G and an underlying h : F → G. Further. We have X@α × X@h(g. However.8. l·αg).e. ˆ ˆ ˆ ˆ i.15) which transforms an objective morphism into a functorial one. We already know from section ˆ 7. this is already evident from the deﬁnition of composition of morphisms: Deﬁnition 25 If f /α : K → L and g/β : L → M are two morphisms of functorial local compositions. that the embedding of K. followed by the product @α×h. We have a similar sorite to the above one for objective local compositions: Sorite 5 (Existence of identity. this is an element of αg@L. i.e. associativity. i. In ˆ ˆ ˆ fact. A@h(k) = l·α.. factorizes (necessarily ˆ uniquely) through L. it is less elegant than the one we gave in deﬁnition 24. This time.f on the second coordinates of these sets. (A@h(k)) · g). we have a commutative diagram ˆ −−− ˆ IdA @K − − − → x@K ˆ α. Therefore (A@h(k))·g = l·αg and therefore X@α×X@h(g. X@h(k · g)) = (αg.. MORPHISMS OF LOCAL COMPOSITIONS 157 This latter deﬁnition reminds us more of the previous deﬁnition for objective local compositions. a subset of X@L. k · g) = (αg. ..2.f f surjection (8. we have a factorization f /α = 1/α · fα /1 via the ﬁber product composition Lα and the morphism fα deﬁned in the previous exercise. depends only on f and not on h. The obvious claim is that this same h and α induce a commutative diagram ˆ −− K − − → @A × F @α×h ˆ −− L − − → @B × G ˆ of functors.

In that context.α and α−1 /α : L.α → L.1. 24 See appendix G. if α is an isomorphism. As already introduced in special situations (see also example 4 in section 6. imply a corresponding factorization on the objective side. In other words.17) What does it mean to have a morphism of two such singleton local compositions? We are just given two elements D1 ∈ A@F orm1 and D2 ∈ B@F orm2 and a form morphism h : F orm1 → F orm2 as well as an address change α : A → B such that we have the equation23 hD1 = D2 α. The formal deﬁnition of the category ObLoc of objective local compositions and morphisms is given and discussed with respect to its musicological relevance. 8.3. A semantic model of general symmetries over the n-dimensional integer module Zn is discussed. However.16) and then deﬁne the singleton objective local composition {D} : Address F ormF orm (D).158 CHAPTER 8. one has the factorization f /α = α−1 /α. (8. where fα /1 : K → L. the facˆ torization of the image f /α does not. that the map ˆ deﬁnes a full functor from objective morphisms ? to functorial ones. –Σ– After having prepared all necessary prerequisites. Therefore. Before turning to the overall discussion of categories of local compositions we should show how to integrate more morphisms for general denotators. We shall denote this type of morphisms of singleton local compositions by !/α and also write !/α : D1 → D2 to ease notation. and an address change α. This means that the category of local compositions generalizes the category of elements of each of its forms. and the category of elements is Mod F . Prove this statement.2. but it is not faithful. we have h = IdF for the underlying functor F = F un(F orm) : Modopp → Sets. we have considered denotators as “elements” and morphisms between such “elements” D1 . SYMMETRIES AND MORPHISMS Exercise 9 For objective local compositions.fα /1.8) a denotator D : Address F orm(Coordinates) could be wrapped in the following local composition: We ﬁrst set up a new Power-typed form F ormF orm −→ Power(F orm) Id (8. .fα /1 into a pure address change morphism and a morphism with ﬁxed address. We shall see later (proposition 3 in subsection 8. D2 as couples of a unique set map of singletons ! : {D1 } → {D2 } which is induced by a functorial morphism of underlying spaces. we are ready to deﬁne the category of objective local compositions: 23 We are a bit sloppy in identifying the denotator names with their coordinates and also omitting the address in the natural transformation h. although it reﬂects isomorphisms. in general. but this makes notation less clumsy. We now recognize that this is a standard concept in category theory: the category of elements24 for set-valued functors.3 Categories of Local Compositions Summary.2). there is no factorization f /α = 1/α.

25 We should apologize for maintaining the mathematical terminology of “composition” of morphisms. CATEGORIES OF LOCAL COMPOSITIONS 159 Deﬁnition 26 The category ObLoc of objective local compositions has as its object collection the set of all objective local compositions as declared in deﬁnition 8. and that any concatenation.3.5 and E. The composition of morphisms is that declared in lemma 5. S.e. Of course we know that underlying symmetries are only the background transformations for visible correspondences on the supports of local compositions. the translation in the direction of the second coordinate axis. but also that any ‘wild’ morphisms may intervene. 1. The set 1 ObLoc of morphisms is the disjoint union of the sets ObLoc(L1 .e. . This manifold of symmetries can be described in a very reduced language by the following theorem which is proven in appendix E. Therefore we are going to discuss a semantical paradigm for symmetries between modules which play an important role in practice. 0 ObLoc Mathematically. Let us therefore digress on a semiotic paradigm for general symmetries. But we have to ask whether symmetries—as a structural potential for possible morphisms—could be musically signiﬁcant—independently of the possible points which could be moved around between such spaces..3. n)} of aﬃne endomorphisms on Zn which consists of the following elements: 1. tn ) and matrix P = (Pij)1≤i. and which we restate in terms that ﬁt with the musical context: Theorem 2 Consider the set MusGen = {T. K. modules for Simple Type. aﬃne transformations. . 0). we believe however that no confusion will arise thereby. 0. Very often practice deals with gauged and then digitized parameters such that we may suppose that they are integers. . composition25 of morphisms is allowed—in fact a critical musical and musicological point. on ‘prototypical’ modules Zn which underly morphisms of objective local compositions in such ambient spaces. The identity morphism of an objective local composition L is the morphism 1L as declared in point (i) of sorite 4. L2 ) of morphisms f /α : L1 → L2 as declared in deﬁnition 23. i. Ps (s = 2. viz. see chapter 49. we may suppose n = m without loss of generality. . This question becomes of primordial importance not later than when local compositions are transformed by automatic processes as they are implemented in music software. . there is no further point in this deﬁnition. K.. i. In delicate contexts we switch to the synonym “concatenation” to make things more distinct. Dm (m ∈ N).) if no ambiguities are likely.3. In this case. . .. T = et . For this case we should have a certain theoretical guarantee for the musical meaning of what is being done. . such as in presto . etc. . .j≤n . i. However. They are usually denoted by their names (often capital letters L. once its ingredients have been introduced. Further any such symmetry can be interpreted as an endomorphism of a large power ZN . M1 . for example by viewing both spaces as direct summands of a suﬃciently large superspace. t = (0. The remark is that the categorical framework not only suggests general ambient spaces. We may therefore work on Zn and consider symmetries f : Zn → Zm .e.8.6. there is one delicate remark to be made as to the musical signiﬁcation of the entire construct. a symmetry et · P : Zn → Zn is described by n + n2 integers via vector t = (t1 .

. . If m ∈ N. d11 = d12 = d22 = . 3. . But it should be stressed that this selection is a function of the selection of the special generator set MusGen. . ss. S = (dij ) the transvection or “shearing” of the second coordinate in the direction of the ﬁrst coordinate axis. 2. . The subject here is not coverage of all possible meanings but evidence of the very existence of musical meaning behind such a priori constructs from mathematics. w · o. es all factors being taken from MusGen. but it states. each of the generators from MusGen is concerned with aﬀecting at most two coordinates. . a classical technique in counterpoint. In particular. s = 2.1. On this space. parameter exchange of ﬁrst and sth coordinates. they aﬀect only the ﬁrst two coordinates! The following concatenation principle yields a semantical interpretation of the preceding result. 4. every symmetry f : Zn → Zn can be written as a product f = e1 · e2 · .e. i. Except generators of type parameter exchange Ps . and the space P ara has coordinator Z duration. In order to turn this purely mathematical fact into a musicological one. Before discussing this principle in subsection 8. dij = 0 for ij = 11. . . let us apply it to theorem 2. i. . dilatation by −1 or reﬂection in the ﬁrst coordinate. SYMMETRIES AND MORPHISMS 2. 0 ≤ m}.1. K = D−1 . Ps = (dij ).1). and dij = 0 else. d22 = . .1) by one pitch unit upwards in w-tempered pitch space.160 CHAPTER 8. whatever. . the invertible symmetries) on Zn can be written as a concatenation of elements of MusGen − {Dm . i.1. the generating symmetries from MusGen have the following musical meaning: 1. Accordingly. dnn = 1. 3 ≤ s. Then every aﬃne endomorphism on Zn can be written as a concatenation (composition) of some of the elements of MusGen (including repetitions of the same element). This means d11 = m. a classical musical process. This principle can obviously not be demonstrated in full generality (this would be a test of an inﬁnite number of musical situations—God beware!). no m-fold dilatations for 0 ≤ m. whereas d1s = ds1 = d33 = . dij = 0 else. Translation T means transposition (example 7 in subsection 8. let Dm = (dij ) be the m-fold dilatation in the direction of the ﬁrst coordinate axis.2.e. and for i = j and ij = 1s. a cornerstone of generic musical semantics: Principle 2 Concatenation of two musically meaningful symmetries or morphisms yields a musically meaningful result. 5. Reﬂection D−1 means retrograde of onsets (example 9 in subsection 8.e. .1. dnn . in its very generality. n.. and that there are many other semantical interpretations of symmetries. dnn = 1.. . Aﬃne automorphisms (i.e.3. we have to view the modules Zn as a coordinator for a reasonably chosen form space. s1. Let us take a direct sum space of shape Onset|Z ⊕w-P itch⊕P ara where w-P itch is the w∼ tempered scale space w-T emperedScale (see section 7.1) with canonical identiﬁer @Z → @Z· 1 n−2 for any ‘accessory’ parameters such as loudness.

The only explanatory access to this symmetry is its decomposition qua concatenation of retrograde K and inversion U ! In this interpretation. Dilatations Dm are classical augmentations in onset directions: distances of onsets are augmented by factor m.1. This concept. .8. and parameter exchange Ps . retrogrades. this was nothing more than destruction of its very intention. Operations Ps of parameter exchange are related to classical compositional techniques or established musical thinking. see chapter 49. –Σ– A. Already in the middle ages (to which Messiaen refers [156]) we can observe a symmetry and exchangeability of treatment of modes in time and pitch. one considers translations. We give some musicological comments on the concatenation principle by means of two examples. Counterpoint. But there is more: Musically. inversions. this does not make any sense in classical terms of European music theory. 8.1. concerning exchange of onset and pitch we had already seen Messiaen’s—unfortunately not very consistent—exchange principle (example 13 in subsection 8. By such symmetries. Further we should recall the P2 -symmetry in twelve-note series from example 12 in subsection 8. 3. This collection meets the requirement that concatenation (composition) of two of its members is musically meaningful. In classical counterpoint. P4 etc. is less investigated by composers and theorists. retrograde K.1). rotation was—in those days—simply ridiculous.2. every local Aaddressed composition K in ambient space Onset|Z ⊕ w-P itch ⊕ P ara is embedded in its orbit − → − → GL(Zn ) · K = {F (K)| F ∈ GL(Zn )} which we call K’s general aﬃne paradigm. in fact a semiotical symmetry. a systematic application can however only be traced in nowadays computer-assisted compositional environments. retrograde inversion is composed by a reasonable action in time and one in pitch.3. P2 is not applied on a given compositional material but as structural ‘isomorphism of meaning’. there is no genuine . All relations among their tones are preserved without loss. n. In other words. s = 2. the retrograde inversion (see example 9) KU isn’t understood as a 180◦ -rotation. If Beethoven rightly operated on Steibelt’s score by such a rotation. 5. . Shearing S in onset direction means arpeggio since the notes of a chord instance at a determined onset are played one after another from low to high pitch. as well as rotation by 90◦ which can be viewed as concatenation of retrograde and parameter exchange.1. CATEGORIES OF LOCAL COMPOSITIONS 161 3.1 Commenting the Concatenation Principle Summary. − → Let us conclude by a remark on the group GL(Zn ) of invertible symmetries on Zn which is generated by transposition T . Serial technique makes use of such symmetries in some cases [550]. together with speciﬁc selections of (parametrized families of) subgroups of the full symmetry − → group GL(Zn ) will lead to a far-reaching conceptualization of paradigmatic phenomena in music and music history. 4. . and retrograde inversions. Here. see section 10.3. arpeggio S. Parameter exchange of onset with other parameters in P3 . local compositions are always transformed in a ‘reversible’ way.

B. twelve-note composition oﬀers an interesting example. This is quite a progress with respect to classical counterpoint. Figure 8. Concerning semantics of symmetries on Z2 . so wie kein Wort allein. enth¨llen sie ihre wahre a u Bedeutung nur durch ihr Zusammenwirken. Variation und Entwicklung die hohere Entwicklungsstufe musikalischer Formtechnik. in dem musikalische Gedanken dargestellt werden.) This graphics shows complex combinations of the fundamental series and its variations in twelve-note technique. The musical space where these transformations take place is described as follows by Sch¨nberg o [479]. [550]). a We pick up from this very modern text that Sch¨nberg thinks at once in two dimensions and o not twice in one dimension. ohne Beziehung zu anderen Worten. Obwohl die Elemente dieser Gedanken dem Auge und dem Ohr getrennt und unabh¨ngig voneinander erscheinen. einen Gedanken ausdr¨cken kann.12: (With kind permission of Reclam Publishers. Dodecaphonism (see also [465]. sondern auch in allen Richtungen und Fl¨chen und zeigt seinen Einﬂuß selbst an entfernten Stellen. The concatenation principle precisely generalizes this classical mechanism of musical semantics. With this example not only semantic relevance of symmetry groups is exempliﬁed. Alles was an irgend u einer Stelle dieses musikalischen Raumes geschieht. Es wirkt nicht nur in seiner eigenen Ebene.162 CHAPTER 8. It seems that . SYMMETRIES AND MORPHISMS theory and semantics in two-dimensional space Onset ⊕ P itch. Sch¨nberg’s exegete Joseph Rufer como ments on the esthetic principles of dodecaphonism [465]: Die beiden an der Entstehung einer musikalischen Form vornehmlich beteiligten Gestaltungsprinzipien sind: die Wiederholung und die Variation. Wiederholung ist das Anfangsstadium.12: Der zwei. see also ﬁgure 8. hat mehr als nur lokale Wirkung.oder mehrdimensionale Raum. one also learns to understand the communicative problem implied by dodecaphonism. ist eine Einheit.

.8. This insight is intimately related to Einstein’s principle that physical information—for instance on a reference frame’s time—can only be transmitted by physical support. The correspondent semantic paradigmatics. on the neutral level. . The sonata principle has been so successful since its basic components—exposition. en−1 (.) → e(K) = L must be evidenced. their product e = e1 .3. It is shown that the embedding of ObLoc in Loc is full and reﬂects isomorphisms. . The entire chain of intermediate compositions must be integrated into the composition to become part of communication. . The category ObLoc of objective local compositions is embedded in a category Loc of functorial local compositions which admits suﬃciently general universal constructions. as inner symmetries. em does not automatically bear musical evidence. . even more prominently. CATEGORIES OF LOCAL COMPOSITIONS 163 dodecaphonists had realized that symmetry groups could unfold their presence to the listener only ‘subcutaneusly’. This compositional process is strictly poietic. . The concatenation principle attributes musical meaning to a concatenation of symmetries or morphisms if the factors share such a meaning. Here the composer has to bring symmetry relations into audible shape. especially in twelve-note and serial composition. . em are immediately evident to the listener. .3. must be turned into neutral musical ‘material’ in order to become musical reality. This does not provide us with a justiﬁcation automatism for twelve-note composition. such as light. . We conclude that the concatenation principle is a metamusical thought. Explicating a musical thought within a musical composition. e1 (K) . The fact that the hoped-for emotional echo has mainly failed does not prove conservative auditory habits (though this is true to a certain degree) but rather a deﬁciency of compositional strategy. and. it could as well be hidden to the listener. evidence of the underlying factors since they are not at all unique. –Σ– . The turning point from construction to audible evidence is the evidence of concatenation.2 Embedding and Addressed Adjointness Summary. From this we have to abandon hope to transmit the message more than in a diﬀuse emotional sphere. In fact. Let us make this more explicit. the temporal course of a composition does not allow the listener to reﬂect symmetry relations. However. development. is no compromise to the listener but a genuine ingredient of whatever art claims to communicate. for instance by means of compositional parts which make these symmetries evident. . Neither the “ideal” listener nor diligent analysis are in a state to decode the massive ambiguities and intertwinements in twelve-note composition. A step-by-step construction of a local composition L from K via K → e1 (K) → e2 (e1 (K)). Each media has to unfold (communication) in its proper terms. . It is as if you were listening to a spoken sentence in a foreign language but you do not know its word grouping. . The contrapuntal symmetries which de facto embed the basic series in its paradigm is instantiated by symmetries which as musical thoughts exist on the poietical and—at best—neutral level (the latter via analytical activities). and recapitulation—build a thoroughly communicative concept. 8. if the factors e1 . . . say. Somebody has to make the grouping evident by means of prosodic tools like stress or pauses. i. and therefore at most at light’s speed.e.

we are ready to deﬁne the category of local compositions: Deﬁnition 27 The category Loc of local compositions has as its object the collection 0 Loc the set of all local compositions as declared in deﬁnition 8 of chapter 7. To see that the functor reﬂects isomorphisms. The set 1 Loc of morphisms is the disjoint union of the sets Loc(L1 . etc. They are usually denoted by their names (often capital letters L. g/β : L → M two ˆ morphisms of objective local compositions.15) we know that objects and morphisms of ObLoc are mapped into objects and morphisms of Loc under a map ˆ Recall from section 7. SYMMETRIES AND MORPHISMS After discussion of some topics pertaining to objective local compositions. Take (q. Functoriality of ˆ Let (K. Take any morphism f /α : K → L. k · q). g =(βαq.α − − → A@G −− . so f is also determined. f (k) · αq)) ˆ ˆ g =ˆ((αq. the objective trace K is given by the set IdA @K. look at the ˆ ˆ ˆ evaluation of f /α at IdA @K: We have f (IdA . Proof.. ˇ that for a functorial local composition K at address A. K. The identity morphism of a local composition L is the morphism 1L as declared in point (i) of sorite 5. L2 ) of morphisms f /α : L1 → L2 as declared in deﬁnition 24 of this chapter. (M.19) A@h (8. and functoriality is established. and look at the composed map on the q-slice g·f ˆ ˆ ˆ q@K = {q} × K · q.α) which uniquely determines α.3 and deﬁnition 16 ?.f and α. Its evaluation f /α : IdA @K → α@L deﬁnes a commutative diagram K − − → A@F −− s (8.164 CHAPTER 8. So we are done if we can show that ˆ ˆ ˆ ˆ the functor is full.) if no ambiguities are likely. M1 . k · q)) =ˆ((αq. f (k). A@F ).e. α : A → B. ˆ ˆ = gf .18) L. Let q : X → A be a point at address A. The composition of morphisms was declared in deﬁnition 25 of this chapter. (L. g(l) · βαq) =(βαq. k · q) ∈ q@K. C@H) be three objective local compo?: sitions. B@G). From deﬁnition 8 and formula (8. g(f (k)) · βαq) =gf (q. We now have this embedding proposition: Proposition 3 The map ˆ : ObLoc → Loc ? is a full functor which reﬂects isomorphisms. β : B → C two address changes. and f /α : K → L. k) = (α. But if α is an isomorphism. We have to show that g /β · f /α = gf /βα. then g (f (q. l · αq)) for an l ∈ L. i.

we have an objective trace functor ˇ@A : Loc@A → ObLoc@A ? ˇ ˇ ˇ ˇ which carries K to K and f /IdA : K → L to f /IdA : K → L via IdA @f : IdA @K → IdA @L. The deeper reason why we have built the functorial point of view upon a universe of objective local compositions is a double one. CATEGORIES OF LOCAL COMPOSITIONS 165 of set maps where the right vertical arrow is the natural transformation h underlying α. This gives us an evident restriction of the embedding functor and the addressed objective trace functor ObLoc@A 26 See ˆ@A ? ˇ@A ? Loc@A (8. and it is precisely this type of universal construction which fails within the purely objective setup.2.e. from the geometric and logical point of view. =(αq. Then ˆ f /α(q. one needs to compare local compositions by use of universal constructions such as limits. or colimits..f . To achieve such a goal.8. i. Fix an address A. However. We contend that f /α = r/α. In other words. by functoriality of h.. we have to restrict to a ﬁxed address. h(k) · q). Let us therefore deﬁne addressed comma categories as follows. we shall see in the global concept frame work of part IV that gluing local objects to global patchworks is genuine musicology. But there is also a more musicological reason for our extension policy: Musical objects are not only local in their spatial nature.3. the unique morphism !/IdA =!/1 : K → @A.r. Take ˆ any r : K → L such that s = α. r and we are done with the claims. this construction does not carry over to morphisms. we denote by ObLoc@A and Loc@A the comma categories over the local composition @A where the structural address change is the identity on the address. k · q) ∈ q@K. Further. the mathematical description of global objects has to make use of ﬁber products of local compositions. i. k · q). every local composition at address A can be ‘frozen’ to the ‘identity point’ IdA and thereby reduced to its objective trace at A.e.αq) =ˆ/α(q. s(k) · q) =(αq. it does not matter. . So the extension to functorial local compositions is seamless and can be ‘traced back’ to the objective framework. On Loc@A . h(k · q)) =(αq. QED. ∼ We have X@(A × @00 ) = X@A × {0} → X@A.20) appendix G. k · q) =(αq. r(k). With this convention. When talking of @A as of a local composition we identify it with the functor—in fact the objective local composition— @A × @00 .1. ˇ Although we can build the objective trace K for any object in Loc. We shall see in chapter 18 that predicate logic on the topos theoretic level needs substantially universal constructs. the objective local compositions can be classiﬁed up to isomorphism in their proper category ObLoc or in Loc. Let (q. First. These not only have generic interest from category theory but are essential to model logical derivates of given ensembles of sounds and other musical objects. the terminal26 object of Mod. where 00 is the zero module over the zero ring.

By the surjectivity of the left vertical x-arrow. By deﬁnition 16 in section 7. i.3 Universal Constructions on Local Compositions –Σ– Summary. SYMMETRIES AND MORPHISMS since the basis object @A and the structural morphisms !/IdA remain ﬁxed under both functors.3 and proposition 3. ˆ Consider the eﬀect of a morphism f /1 : K → L at a point x : X → A. there is a good chance that local compositions too admit such universal constructions.166 CHAPTER 8. ˆ So all we have to show is that every morphism f /1 : K → L over A factorizes through L . We have adjointness for ﬁxed address. L ) ˆ HomLoc@A (K. see for example [314]. ? Proof. and in particular admits universal constructions such as ﬁnite limits. . Recall from appendix G. L).e. we have an isomorphism ˇ ∼ ˆ HomObLoc@A (K. the image of x@f also lives in x@L . limits and colimits are “calculated pointwise”.1 for the concept of adjointness.2.2.. We want to investigate the speciﬁc construction in this subsection. IdA @L = IdA @L . ‘addressed’ adjointness: Proposition 5 The morphisms ˆ@A and ˇ@A build an adjoint pair ˆ@A ? ? ? ˇ@A . we have an injection ˆ HomLoc@A (K. 27 See appendix G. So we have the commutative diagram @f ˆ IdA− IdA @K − − → IdA @L − x x (8. It is well-known [314] that Mod@ is a topos. By deﬁnition. a subject which is central for the topos? ? logical approach. We terminate this subsection with a short discussion of adjointness 27 of the embedding functor ˆ@A and the objective trace functor ˇ@A .1 that in Mod@ . QED.21) ˆ x@K −−→ −− x@f x@L But the very deﬁnition of the embedding ˆ (see (6..3. colimits and a power object. 8. L ) and by the injection L L from sorite 1. Proposition 4 The addressed restriction functor ˇ@A is a left inverse to the addressed embed? ding ˆ@A ? Exercise 10 Give a proof of the preceding proposition.e. L) → HomLoc@A (K. the application ˆ x@f : x@K → x@L. Since the denotator space functors are in Mod@ . i. We show the ﬁnite completeness of Loc and its interpretation in musical terms.16)) implies that the image of x : IdA @L → ? x@L stays in x@L .

8.23) (8.25) has its functors K. D = A ×C B. M and their morphisms f. suppose that f..26).e. and the morphisms q . L. Since the cartesian diagram (8. k : G → H. Given three local compositions K → @A × F. and we can replace P by its image ˜ P = im(P ). g/β : L → M . Then we have two cartesian diagrams in Mod@ : s R −−→ G −− (8. p from P . p by the corresponding morphisms q.25) K −−→ M −− ˜ with P = K ×M L as set-valued functors.3. i. the ˜ ˜ ﬁber product P clearly injects into @(A ×C B) × F ×H G.22) (8. without ˜ ˜ ˜ losing the ﬁber product properties of the functor P . we are looking for a local composition P → @D × R and two morphisms q/κ : P → K.24) r k F −−→ H −− with R = F ×H G and ˜ ˜ − p− P −−→ L g q ˜ h α λ f /α p/λ (8. From appendix E. the product diagram @(A ×C B) × F ×H G − − → @B × G −− @β×k @κ×r @λ×s f (8. g as subfunctors of the corresponding ambient space functors in diagram (8. 167 Proof. L → @B × G.8 we know that the underlying address changes have a ﬁber product D −−→ B −− β κ A −−→ C −− of modules and di-aﬃne morphisms in Mod. we have a commutative . CATEGORIES OF LOCAL COMPOSITIONS Theorem 3 The category Loc has ﬁber products. M → @C × H and two morphisms f /α : K → M.26) @A × F − − → @C × H −− @α×h is also cartesian.3. In other words. But then. Further. p/λ : P → L such that we have a pullback diagram P −−→ L −− g/β q/κ K −−→ M −− of local compositions. g are induced by the functor morphisms h : F → H.

2) that a category has ﬁnite limits iﬀ it has ﬁber products and a terminal object. (ii) f and α are mono. and this one evidently induces t. QED. (iii) implies (ii). To this end. Therefore (i) and (ii) are equivalent.e.26). Corollary 2 For a morphism f /α : K → L in Loc the following statements are equivalent: (i) f /α is mono. p · t = v and κ · ρ = τ. we have a space functor morphism @ρ × z : @X × W → @D × R. Conversely. s. and this in turn means that f and α are mono. λ · ρ = ν.2. Proof. QED. Proof. Let 00 be the terminal object (see appendix G. i. We use the fact (see appendix G. the required characterization of mono diaﬃne morphisms is given. has ﬁnite limits. Corollary 1 The category Loc is ﬁnitely complete. . The local composition 1Loc = @00 × @00 then is terminal since any A-addressed local composition K has a unique morphism !/! : K → 1Loc . If z : W → R is the universal arrow induced by r. By proposition 86 in appendix E.27) f1 /α1 (8. Then we have a uniquely determined functor morphism t : T → P and a uniquely determined di-aﬃne morphism ρ : X → D such that q · t = u. it is uniquely determined and we are done. where u/τ is induced by m : W → F . Together with theorem 3 we can apply the criterion for ﬁnite limits. give any commutative diagram T −−→ L −− g/β u/τ K −−→ M −− of local compositions with T → @X × W . this is equivalent to f1 = f2 being an isomorphism.2.2) that f /α is mono iﬀ in the cartesian product ∆ −−→ K −− f /α f /α f2 /α2 f /α v/ν (8. Clearly. QED.2. SYMMETRIES AND MORPHISMS diagram of local compositions with ambient spaces from diagram (8. (iii) X@f is injective for every address X and α is diinjective.28) K −−→ L −− both projections f1 /α1 .2 that a natural transformation f in Mod@ is mono iﬀ its evaluation at every address X is injective..168 CHAPTER 8.1) in Mod. But clearly. f2 /α2 coincide and are isomorphisms. and where v/ν by n : W → G. we know from appendix G. We now have to show that this diagram is cartesian. It is well-known (see appendix G.2. So if t/ρ is a morphism of local compositions. and α1 = α2 being an isomorphism.

Take M = A@F . i. L → M . see also ﬁgure 8. Let us denote this latter category by SinLoc. L are the same. Let us now evaluate the ﬁnite completeness of Loc in more concrete terms and with regard to the subcategory of objective local compositions. From an overall point of view. We approach categories as point spaces which are sorted and structured by addresses. In other words.. objective and functorial local compositions and the related address navigation in order to understand the overall ‘geography’ of addresses. –Σ– We should now review the entire universe of denotators. To begin with. and x@T ∼ K so that the morphism = x : IdA @T → x@T cannot be surjective and the ﬁber product T not objective. 8. We shall in fact exhibit a pair f /α : K → M. general ﬁber sums do not exist in Loc.3. The dual situation is less simple.2. To understand this phenomenon we want to construct a ﬁber product which does not fulﬁll the following evident requirement for objective local compositions: For any objective local compositions K at address A. Suppose that K ∩ L = ∅.13. It extends the full subcategory ObLoc of objective local composition which are essentially sets of denotators of address-ﬁxed “ambient space”. This implies that we cannot glue together arbitrary local compositions to a new local composition. g/β : L → M of objective local compositions for address changes α : A → C.1. we should discuss an example where the ﬁber product of objective local compositions is a proper functorial local composition.8. in fact. CATEGORIES OF LOCAL COMPOSITIONS 169 Remark. In other words. This means that we take sets of intervals i = i0 ⇒ i1 . Then the evaluation ˆ ˆ ˆ of the ﬁber product T = K ×M L gives IdA @T = ∅. we have built a comprehensive category Loc of local compositions. The latter include the singletons which are the very general denotators and take care of the classical construction of “categories of elements”. We make the address concept explicit: Variation of addresses in music and musicology.4 The Address Question Summary. more precisely that Kx = Lx.e. but the interval arrow heads i1 never coincide. ‘global compositions’ as a result of gluing together local compositions present a proper extension of the local concept framework. β : B → C such that the ﬁber product map u : IdA×C B @(K ×M L) → u@(K ×M L) is not surjective for a selected address change u : W → A ×C B. corresponding to morphism ei0 · (i1 − i0 ) as in subsection 6. that the base points of the intervals in K.3. This is the mathematical reason for the entire global theory to be developed later in part IV. So we have the chain SinLoc → ObLoc → Loc of full subcategories. the objective perspective at address A yields no common points whereas the zero-addressed functorial perspective at x shows common base points. (8. and for any X-valued point x : X → A. Consider two non-empty objective local compositions K. L ⊂ A@P itch in the pitch form with ambient space F = @R. Take A = B = C = ZZ and take the zero point x = 0 : 0Z → A.29) . the map x : IdA @K → x@K is surjective. and look at the natural inclusions K → M.

30) which is indeed a functor. This deﬁnes a map Ad : Loc → Mod : D → Ad(D) (8. we have diﬀerent ﬁbers which are of interest in determined contexts.e. i.13: Two objective local compositions K. The functorial ﬁber product ˆ ˆ ˆ K ×M L identiﬁes these common ‘base points’ whereas the objective ﬁber product (in fact the intersection) is empty. For example.. L in pitch space M at address Z with identical ‘perspective’ at the zero-addressed point x = 0 : 0Z → Z. and its ‘identity’ subcategory IdA with the identity of A as its only arrow. since morphisms f /α of local compositions include address change α. then we have two corresponding ﬁber categories . SYMMETRIES AND MORPHISMS Interval heads (pitch) K L ŸŸ x 0Ÿ Interval tails (pitch) Figure 8. Within this framework. every denotator D has an address Ad(D).170 CHAPTER 8. a module object in Mod. if we consider the full subcategory End(A) → Mod whose only object is A. the address functor. On the other hand.

Viewed as a functor it is not one naive set but an entire connected collection of sets. 8. Even an objective local composition is solicited by an underlying space functor whose evaluation at the composition’s address produces the foreground set structure. But what part of a functor can be seen at a determined address is not only mathematically diﬀerent. In this spirit we can make the point as follows: Principle 3 The ﬁbration of Loc and its subcategories via the address functor constitutes a ‘bundle of ontologies’. viz all its evaluations at addresses X. in our example from ﬁgure 8. esthetics and in compositional contexts. Musicologically this perspective traces an ontology of diﬀerent ﬂavor compared to the poor reduction to base points under the zero address. it also deals with ontology.13. The indicated ﬁxations of A are only the most obvious constructs. the passage to its associated functorial ‘copy’ ˆ K → @A × F is indeed an extension of the ﬁxed address to variable address perspectives. The ˆ evaluation of K at a morphism x : X → A is a kind of ‘K’s view as seen from address X ˆ ˆ under the perspective x’. CATEGORIES OF LOCAL COMPOSITIONS 171 SinLoc@A ObLoc@A Loc@A IdA −−→ −− SinLocEnd(A) ObLocEnd(A) LocEnd(A) End(A) −−→ −− SinLoc subcategory ObLoc Loc −−→ −− −−→ −− full and reﬂexive (8. But the address question is not only a matter of relativization of perspectives. Intervals are not just an ordered couple of zero-valued points! For instance. they have a radically diﬀerent perceptual quality in music psychology. one for each address—or even for each subcategory of Mod.5 Categories of Commutative Local Compositions Summary. We shall see in chapter 9 that the Yoneda lemma essentially instantiates this ‘philosophy of perspectives’: We look at an address A from all other addresses and try to understand what’s happing at A by collecting all these relative views. the ﬁber of X@K at point x (see also ﬁgure 7. What is seen is the ‘slice’ x@K. This subsection deals with the structure of the practically very important category .11 for this interpretation). The evaluation at a determined address generates a set which is speciﬁcally related to this address.3. In principle. For instance.3. we have Z-valued points which reﬂect concepts around intervals in pitch space. Address change therefore is an expression of an ontological shift and thereby constitutes a fundamental reconstruction tool of musical reality from a bundle of addressed ontologies. Let us ask for the existentiality of a local composition. The functor can only be ‘seen’ under such an evaluation. But there is more: If we are given an objective local composition K ⊂ A@F in the A-ﬁber ObLoc@A . every subcategory of Mod deﬁnes an associated theory of musical objects on its Ad-ﬁber.8.31) −−→ −− −−→ −− Ad −−−− −−−→ subcategory −−−− − − − → Mod subcategory deﬁned by this evident system of cartesian squares. it is also ontologically diﬀerent.

the ﬁrst assertion is clear since. by deﬁnition..172 CHAPTER 8.(g · f ) = R. Consider the following subcategories: • The full subcategory ComLocemb of ComLocA whose objects are the embedded commuA tative local compositions. Proof. A-addressed local compositions28 as deﬁned in section 7. but this is a straightforward aspect which may be suppressed for the majority of mathematically oriented considerations.x (8.K and R. it is clear that f → R. Then we denote by ComLocA the subcategory of ObLoc@A whose objects are commutative.f is functorial. SYMMETRIES AND MORPHISMS of commutative (objective) local compositions. if a morphism f : K → L in ComLocA has underlying morphism h : M → N of the supporting coordinator modules. • The full subcategory ComLocgen of ComLocemb whose objects are the generating comA A mutative local compositions. and the result lives in R.k0 = h. 28 Recall that we often suppress form names and simply write (L. Once the ﬁrst assertion is proven.L.33) Rh (8.L : x → h0 . Lemma 6 With the preceding notations. so h does not really matter.∅ = ∅.1K = 1R. A@M ) or (L. R. (8. it only depends on f . A@ F ) for a simple ambient R space F with Functor F un(F ) = @R M for such an object since the names do not intervene for classiﬁcation (= determination of isomorphism classes in a category). Hence a morphism f : K → L is a set map f such that there exists an R-aﬃne morphism h : M → N of the respective coordinator modules such that the diagram K − − → A@R M −− A@ f L − − → A@R N −− commutes29 .x = h0 . 29 The notation A@ h with index R means that di-morphisms are restricted to the identity of the coeﬃcient R ring R.K → R.e.g · R. and if h0 is the R-linear part of h.f is functorial. They however play a role for identiﬁcation. We therefore have a covariant functor R : ComLocA → ModR called the module functor. But if k0 ∈ K.4. then the R-linear application R.32) is well-deﬁned.k − h0 . –Σ– Deﬁnition 28 Let A be an address which is an R-module for a commutative ring R. which means that no address changes are admitted. i. R.34) .f is called the linear map associated with f .f . and h0 . The association f → R.k0 = f (k) − f (k0 ). If K is empty. we have x = k − k0 . Its morphisms are all morphisms f = f /1 of ObLoc@A . The application R.f : R. QED.k − h.

L). there are a natural number n and two direct decompositions M = S1 ⊕ T1 = S2 ⊕ T2 such that: 30 See 31 See i appendix E.3. and L is evidently embedded.f |K .K → R. we are done. CATEGORIES OF LOCAL COMPOSITIONS 173 • The full subcategory ComLocin of ComLocemb whose objects have injective R-modules A A as coordinators. an equivalence. Then 0 any underlying symmetry F : R.4.K) and 0 0 a morphism f : (K. . and ingen is full. we may suppose without loss of generality that F is linear.K − − → I(R. 0R @R. we have the following results: (i) The embedding ComLocemb 0 ComLoc0 is an equivalence of categories. M ) be two objects in ComLoc0 such that their coordinator has ﬁnite ∼ length30 l(M ).e. i. Then each isomorphism f : K → L is induced by an invertible symmetry − → − → F ∈ GL(M ). by injectivity of I(R.K) −K − F H i ∼ (8.L). a local composition (K. Therefore ingen is fully faithful and surjective.K) and gen(f : K → L) = R@f according to appendix E. Let f : K → L be a morphism in ComLocgen .. M ) and (L.L extends to a symmetry H : I(R. (ii) Let ingen : ComLocin → ComLocgen be the functor which maps (K. we have an isomorphism e−k0 : K → e−k0 (K) = L.K − − → I(R.. there is a faithful functor gen : ComLocemb → ComLocgen A 0 deﬁned by gen(K. lemma 77). I(M )) for the injective envelope31 I(M ) of M .2. M ) to (K.35) R.L) −L − commutes. According to Fitting’s lemma (appendix E. ingen is surjective. (ii) Since every generating (K. the map iL · F factorizes through injection iK .e. In fact. M ) → (L. the orbit GL(M ) · K describes all local compositions with coordinator M which are isomorphic to K.3. i. (i) If K is an object of ComLoc0 . lemma 82. Since R@f is an extension of f to the compositions’s modules. M ) is the image of (K. A@R M ) = (K. R. in particular f = R. Proof. Since the embedding of embedded objects is fullyfaithful.. With this notation. (iii) Clearly. 0R @M ) may be identiﬁed with (K. Sorite 6 For zero address A = 0R .8. M ).K) → I(R. ingen is faithful.e.4 appendix E. Fact 3 For an address A. (iii) Let (K. we may suppose that K and L are embedded and that the isomorphism ∼ ∼ f : K → L and its inverse f −1 : L → K are induced by linear symmetries G and H. Then ingen is an equivalence of categories. and then. the diagram of modules and R-aﬃne morphisms R.4.2. N ) to R@f as deﬁned in the above fact 3 for embedded local commutative compositions. i. and if k0 ∈ K.

the theorem of Krull–Remak– ∼ Schmidt (appendix E. For an A-addressed commutative local composition (K. a ﬁnite direct product of commutative ﬁelds (see appendix E.36) into the automorphism group of K. but for a ambient module M of ﬁnite length.2. By a symmetric argument on S2 we conclude that ∼ G(HG)n |S1 : S1 → S2 .L ⊂ S2 . By deﬁnition. then the categories ComLoc0 and ComLocgen are equivalent. (GH)n |S2 ∈ GL(S2 ). RK ⊂ S1 . 0 Proof. and we obtain an invertible extension of G(HG)n |S1 to all of M . R. . (i) and (ii).e.. Since in this case. so it is also in GL(S1 ). and the 0 0 corollary follows from sorite 6. the homomorphism is surjective. and by a symmetric argument. (HG)n |S1 ∈ GL(S1 ). if we denote T riv(K) = Ker(r) the group of those symmetries of K which act trivially (as identity) on K. Since (GH)n kills T2 . (iii). This is neither surjective nor injective in general. by sorite 6. H(GH)n |S2 maps S2 into S1 .37) if K’s ambient module M is of ﬁnite length. we have ComLocin = ComLocemb . every module is injective. we have a group homomorphism r : Sym(K) → Aut(K) : h → h|K (8. we denote by Sym(K) the symmetry − → group of K of all invertible symmetries h ∈ GL(M ) which induce an automorphism of K. But be aware that the isomorphism class of Sym(K) is not an invariant of the class of K while the automorphism group Aut(K) clearly is. 3. G(HG)n |S1 maps S1 into S2 . Since l(M ) < ∞. CHAPTER 8. SYMMETRIES AND MORPHISMS 2. But the composition H(GH)n |S2 · G(HG)n |S1 = (HG)2n+1 |S1 has its nth power in GL(S1 ). Statement (iii) of sorite 6 can be restated in the following way. T2 = Ker(GH)n . T1 = Ker(HG)n . i. QED.2). also G(HG)n |K = f . we have Sym(K)/T riv(K) → Aut(K) ∼ (8. Corollary 3 If R is a semi-simple commutative ring. and in this case.5) implies that T1 → T2 .174 1. A@R M ) in ComLocA .

there is no lowest level. 175 . this lemma introduced a revolution in understanding structures of general types. thinking in this style is bottomless and creates existence from the very fact of performing logically consistent mental processes of conceptualization. et qu’il faut faire les nombres chaque fois qu’il faut les penser. the lemma serves as a background to rebuild the very concept of a point as an elementary structure of generic geometry. This chapter reviews and completes the inherent paradigm change from “objectbased” to functorial mathematics which was initiated by category theory. Beyond an apparant technical innocence. In algebraic geometry as it was developed by Alexander Grothendieck in his Paris School. But the deeper impact of this lemma is a radical review of the nature of human conceptualization. And it does this in an approach of inﬁnite recursive descent: a point is a point is a point. and transcending pure mathematics. This is a modern form of Paul Finsler’s principle [152]: 1 The contribution of Yoneda to this lemma is not completely clear.Chapter 9 Yoneda Perspectives Kant m’apprit qu’il n’y a point de nombres. and this means that there is no basic level of concepts.. This is what we should keep in mind when tracing the role of what is called the “Yoneda philosophy” since its introduction in mathematical music theory [332].. Its implications touch general hermeneutics as well as esthetics of art. but the name has been commonly accepted. and—quite paradoxically!— principles of “object-oriented” programming. –Σ– At ﬁrst sight Nobuo Yoneda’s lemma1 is a technical tool of category theory. The drama behind the Yoneda revolution in mathematical sciences is that it proposes a thorough geometrization of human concepts—without limitations. and completed in the celebrated Yoneda lemma. [335]. Alain [13] Summary.

3 He used to call the formal games “paper science”. And ∅new is such an object for the concept type “newset”. We could then start with the “newemptyset” ∅new deﬁned by “for every newset X. the deﬁnition of a circular set M = {M } is acceptable as long as it does not create logical inconsistencies. though in a more “platonic” mental reality. i. The only set which (in the usual setting. However. see [281]. which mathematicians often prefer for shear comfort of passing to their beloved game of empty forms. together with the overall negation of the element relation. This is a circular concept: We ask “What is S?” And you answer: “It is the collection of those sets X which are elements of S. mathematical objects were entities that existed like physical objects. To do so. These spaces are isomorphic. and the Power type is nothing Id Id but the element relation in its category-theoretic transﬁguration. X ∈ ∅new ”. A and Non-A are mutually exclusive and a third possibility is excluded. and this was not only the usual concept framework: Finsler was extremely concerned with circular deﬁnitions as an important device for creating basic objects of mathematics.176 CHAPTER 9. we need examples of sets which are generated by special deﬁnitions or deﬁnition schemes.1) exists from scratch is the empty set ∅: It is deﬁned by the statement that “for every set X. no memory storage device can replace mental activity since the latter is semantic. To him. see C. It resembles Murray Gell-Mann’s “Eleventh Commandment” which states that whatever is not forbidden is mandatory. the set concept is circular: Set theory is deﬁned by a fundamental concept “set” and a binary element relation “∈” among instances of the set concept. no arsenal of strings. But this is not the point since we are not dealing with formal languages which are mapped to some models in a given reality: The question is about the very concept building: What do we think when we think of mental objects named “sets”? The point then is that we think of self-referential. we have the well-known identiﬁcation of a set S: It is a reference to the elements of S. . the other in the form named “newset”. we are not aware that conceptual circularity is unavoidable. YONEDA PERSPECTIVES Principle 4 Mathematical existence of an object means that it can be thought of in a logically consistent2 way. X ∈ ∅”. to him. To Finsler (see ﬁgure 9. and deﬁne a “newset” theory with everything except the name unaltered. It is the reference to all instances of the concept. we have to investigate a universe of such objects which can be built without logical inconsistency. Usually. to bring set theory to existence.” Your set S is only identiﬁed via a pointer to a well-deﬁned collection of instances of the same concept. From the perspective of mathematical model theory. mathematics was not the usual formal game with empty symbols3 . we could as well restart set theory with a new name for the basic concept: Call the new objects “newset”. Existence of such “logical” objects meant consistence with absolute logic. It is not important which organism does the job—a computer could as well—but it is not legitimate to “delegate thoughts to philosophers”. The set ∅ is such an object for the concept type “set”. There is no substitute for really performing thoughts. To make this evident. the second is newset −→ Power(newset). while doing so. for example. we would say that the newset theory is just another model for set theory. For instance. or shorter: “S refers to no element”. These reﬂections make clear 2 To Finsler logic was not a formal game but “absolute logic” which reaches as far as the three fundamental laws: A is identical to A. it has to happen somewhere and sometimes. The ﬁrst form is set −→ Power(set). those sets X such that X ∈ S. This is exactly the situation of denotators and their forms: Up to form names.e.1). the diﬀerence is reduced to their names. circularly deﬁned objects and that. the denotators are the same—one species lives in the form named “set”. and we adhere to this verdict. The empty set is the pure reduction to the concept.

We shall discuss this topic in section 9.1: The mathematician Paul Finsler (1894–1970) was a pioneer of circular and nonformal concept design. and I admit that his anti-formalist point of view never failed.. it is not obvious that this methodology gives you back all you need to know about an object. that we understand concepts as being pointers. and so forth.177 Figure 9. Their semantics is nothing more than this thorough instantiation of the coordinates of your concepts.4 below and hope that the reader may now feel and then understand that the previous remarks about concept architectures are only apparently limited to mathematics. i. We should however not enter a more detailed discussion after this bunch of provocations of mathematical catechism4 without stressing that Yoneda’s basic epistemological insight is fundamentally the same as it is recognized in the theory of ﬁne arts and of musical performance. he had a unique instinct in unveiling hidden formalia. structured recursive references with possibly circular recurrence. I have participated in the last Finsler seminar before his death. But this is Yoneda’s kick: It is indeed suﬃcient to know about the pointer chains. . Quite radically.e. Now this is very much the same idea tah you learn from Yoneda’s lemma: It tells mathematicians that all they can understand about mathematical objects is from the way they point at other objects. from these to still others. Of course.. And this in turn means: Thesis 1 Concepts are nothing more than this recurrent pointing. Replacing a “real” object X by its functor @X means forgetting about the “point as a such” and being exclusively concerned with its reference to other points. This is a thorough geometrization of the concept structure: Concepts are points that point to other points. viz its isomorphism class. there is nothing behind concepts except the paths and ramiﬁcation modes leading down to their referenced ‘coordinate concepts’. 4 We should remark that Finsler’s point of view was passionately attacked by the establishment of “paper math” logicians [153].

and there. The latter is deﬁned by the circular image “to pierce slightly with a sharp point”6 . The second lacuna is the variety of arrows that is oﬀered in Euclid’s geometry. nothing else. a line has an attribute and a surface has two of them. and surfaces have an identity. Reviewing Euclid: points. –Σ– In the ﬁrst book of his elements [215]. to show. The fact that Euclid did virtually not use his fundamental deﬁnitions in his axiomatic system of geometry is not our argument. and at least intuitively acceptable. 5 We are not discussing synthetic geometry here where points as well as lines are formal objects of an axiomatic system. Why is it imaginable. This does not mean that a point is nothing. to prick. a point would have its coordinates as parts. but fascinating. What can be the identiﬁcation criterion if the intended point object has no part or attribute which can be used to make its identity more explicit? In analytical geometry. Thesis 2 Identiﬁcation of what has no parts is that pointing gesture. And deﬁnition ﬁve reads: “A surface is that which has length and breadth only. and we were done. acute (sharpened) instrument. but Euclid’s attempt to introduce meaningful objects via attributes. line. addresses. In the etymology of the word “point”. “Raum” is explained as “Menge von Dingen. the identiﬁcation act seems a secure process which is not further problematic. and arrows. First. the pointing subject was uniquely there to ‘throw the identiﬁcation arrows’. review of the theory of polynomial equations which eventually ended in modern algebraic geometry. lines. it was hidden as a subsidiary instance behind the targeted points. The turning point came through a complex. two particularities were unspeciﬁed: The pointing subject and the variety of identiﬁcation types. In the vein of analytical geometry as initiated by Descartes. we indeed ﬁnd a strong argument for the acceptance of the said identiﬁcation mechanism: “Point” stems from latin “punctum”.” In contrast to a point. mathematical objects were cast somewhere in a Platonic topos and one had to recognize them. its identiﬁcation is established by a pointing action—the English expression “to point at” is essential: a point is what we point at. a point p is a compound concept.” So a line has a part: length. The important fact is that a point is a result of a movement with an arrow-like. and surface. that a point object could have an identity and nothing else? Evidently. but points have no further speciﬁcation of their identity whereas lines and surfaces have5 .178 CHAPTER 9. y. die Punkte genannt werden”. but here. Deﬁnition one reads: “A point is that which has no part. the past participle of “pungere”. the concept is irreducible. YONEDA PERSPECTIVES 9. It is the result of a pointing gesture which in German reads “zeigen”. had to hold a quiver ﬁlled with identiﬁcation arrows. Points. For a long time. it is identiﬁed via its numerical coordinates: p = (x. to describe their eternal properties. z). So we are confronted with the identiﬁcation of points. the problem of identiﬁcation arrows was not recognized. Euclid deﬁnes point.” Deﬁnition three: “A line is breadthless length. The explanation of the German Brockhaus is not less circular: “Punkt” refers to “Geometrie”/“Raum”. . No other irreducible targets are considered. In Euclid’s identiﬁcatory approach. 6 This is Webster’s explanation.1 Morphisms Are Points Summary. and that two lines are equal if their lengths are. From the naive Platonic point of view. Why morphisms are elementary. There is only the ‘point’ type of arrows. pointers. but no part called “breadth”.

Z 2 + 1) is isomorphic to the √ domain C of complex numbers via the identity on the reals and z → i = −1. Z) → f (x. but created a new one uniquely through the homomorphism e.2. Y. Y. Y. Y. but these values are no longer visible in the naive context of real coordinates. Z]/(X. Y. Thus we have x = y = 0 and z 2 = −1. Y. the Euclidean pointing subject and the variety of arrows are instantiated. Z] → R[X. Z − z intervene in the following restatement: We look at the ring homomorphism7 evp : R[X. So the coordinates of p are recovered by generators of Ker(evp ). Z]). But the algebraic approach suggested a radical review of this view. Z] → B with codomain any commutative ring B and identify it with a kind of ‘generalized point’. these arrows are called “B-valued points of R[X. Z]/(X. In this setup. Z − z) generated by the three polynomials X − x. this ‘generalized point’ e is something very exotic! We have not found a classical coordinate triple. see ﬁgure 9.1. This point has coordinates x = e(X). MORPHISMS ARE POINTS 179 In the classical setting. To us the only point is that algebraic geometry views our homomorphism e as a morphism Spec(e) : Spec(B) → Spec(R[X. B) on the ‘addresses’ B. In algebraic geometry. y = Y (mod X. 8 See 7 See . We may now look at homomorphisms e : R[X. either we have the coordinates or we don’t. y = e(Y ). y. This seemingly complicated restatement has a far-reaching advantage. In algebraic geometry. Y.1. z = e(Z). the coordinates are ‘given’ numbers. z = Z (mod X. In other words.1. The three linear polynomials X − x. What is the key argument here? The introduction of ‘points’ via homomorphisms allows solution of equations which were not solvable in the previous real-valued coordinate context. Z 2 + 1). Z 2 + 1). the spectra Spec(R) of rings R. the essence is that such an object is really controlled by its functor Rings → Sets : B → Rings(R. Clearly the third equation cannot be fulﬁlled on R. Y. Y − y = 0. Y.1) on the ring of polynomials f (X. This means that we can restate a point p by the evaluation homomorphism evp . appendix D. 9 See appendix F. So the creation of such generalized points identiﬁes with the construction of solutions of algebraic equations. Y. Y − y. The point p can be seen as a solution of three polynomial equations X − x = 0.2.9. ﬁnally. So the Euclidean subject is instantiated by this address selection whereas the Euclidean ‘arrow variety’ is precisely the set of arrows from Spec(B) to Spec(R[X. there is no question of inventing whatsoever. z) (9. In our above case. Z 2 + 1). Y. Y. Y. Y. Z 2 + 1) with x = X (mod X. Y. Z − z. Y − y. Y − y. Z) with real coeﬃcients in the variables X. Z]) which in the spirit of our general denotator terminology starts at ‘address’ B instead of the usual address R of real coordinates. Z] → R : f (X. Y. Z − z = 0. Let us see why this is completely natural! Take the canonical factorization homomorphisms e : R[X. the relativization of the point’s subject domain appendix D. the quotient ring B = R[X. In other words. Z. Z]”. a homomorphism d : A → B of rings is read as an arrow with reversed direction Spec(d) : Spec(B) → Spec(A) of associated geometric objects9 . And this is a crucial situation in the development of modern mathematics: Construction of solution spaces for algebraic equations! The complex numbers are the solution space whose points are built around the ‘square root of −1 point’. The point p is then associated with the kernel of evp which is the (maximal) ideal8 (X − x.

It is not the place here to pursue in depth the deep consequences of this change of paradigm in mathematics which Alexander Grothendieck introduced around 1963. points are arrows of spectra. From now on. Unlike classical topology. Y = y. General rings B in the role of ‘point addresses’ yield solutions for general algebraic equations. Z = (-1)1/2 complex solutions Spec(¬) Spec(—) "addresses" p: X = x. YONEDA PERSPECTIVES Spec(B) solution of general equations at "address" B p: X = 0. a Grothendieck topology (see appendix G. if its codomain Y is the domain of another arrow g : Y → Z. This is completed in the 1963 revolutionary approach to mathematics of William Lawvere in joint work with Myles Tierny [290]: He started his successful foundation program to lay the basis of mathematics not on set theory but on category theory. y. The reality of real numbers is only one of a huge universe of ontological addresses. corresponding to ring homomorphisms in reversed direction. Y = 0. In electromagnetic theory and engineering. Z = z real solutions Figure 9. the elementary objects are no longer sets as given by their element points. What then is left in this new paradigm from the old pointing gesture? Exactly what was the essence in concept construction: You have a domain address for your arrows and for any arrow f : X → Y . At this point it also becomes evident that the view of addresses as ontological parameters such as was described in principle 3 is completely canonical for mathematical ontology: New numbers provide new ontologies of mathematical objects. adds new points to the usual ones from naive real geometry. z. Complex points deﬁne a diﬀerent space which is no more visible on the naive level and is built upon the solution of equation Z 2 + 1 = 0. you can concatenate the arrows to X → Y → Z and end up with a composed pointing gesture g · f : X → Z. we should however notice that the very concept of a space was relativized according to the address variable. the Grothendieck concept works on the point concept which we have introduced above: Points are arrows in categories. In fact. but arrows which replace the old points and are now parametrized by their ‘domain addresses’.2: In modern algebraic geometry. Real-valued points can be drawn in usual three-space by three coordinates x.4 and our more speciﬁc coverage of the subject in chapter 19) on a category C is a collection of point sets where points are arrows on all possible addresses in C with special properties. this is standard: things such as imaginary currents are common practice.180 CHAPTER 9. This axiomatics of category theory reduces to the very nature of ‘pointing at’: the only thing which can be done in the universe of .

With the notation of section 8. @Y ) : f → @f. 39)}. see also ﬁgures 9. Then the applications 0C an object. 41). Onset ⊕ P itch) with each three elements: M1 = {(0. 39). M2 = {(−1.10 9. (M2 . Let us make an example to learn what such perspectives can teach us about X.. γ → (f : Y → X → f @F (γ)) 0C (9. 40).2.3 and 9. M1 and M2 look the same when viewed from S. i.e.3. we try to understand its contents. sorite 6. F ) → X@F : η : X@F → Hom(@X.2) (9. intuitively. To begin with. if Y ∈ g → g(X)(IdX ).2 Yoneda’s Fundamental Lemma –Σ– Summary.4) η : X@Y → Hom(@X. and F ∈ 0C @ a contravariant set-valued ι : Hom(@X. 43)}.2 for the formal statement. To this end. See also appendix G. take two threeelement motives (M1 . YONEDA’S FUNDAMENTAL LEMMA arrows is to put them together if ever possible. This is the statement: Lemma 7 Let C be a category. The ﬁrst address is the singleton motif S. take the category ComLoc0R of commutative zero-addressed local compositions over the reals. We also call such arrows f ‘perspectives of X viewed from Y ’. The arrows c : S → Mi correspond to the elements in motives Mi .. the singleton perspective would already do: The perspectives from the singleton just deﬁne the cardinality of a set.4. and corollaries.9. In particular. what does it mean to replace an object X by its functor @X? We are indeed creating an overall perspective to X since @X means selecting all possible ‘addresses’ Y in C and then pointing at11 X via all possible Y -addressed arrows f : Y → X. we have a bijection (9. 11 This is the reason for selecting @ as a symbol for the arrows f : Y → X in Y @X . We want to give a comprehensible interpretation of the last statement concerning the Yoneda functor. F ) : are inverse to each other.5. The formal statement. Since we have three of them in each target motif. the Yoneda functor @ : C → C @ : X → @X is fully faithful. Let us look at diﬀerent ‘addresses’ from which we look at our motives. the arrows are in one-to-one correspondence. So. and why this terminology is adequate. (1. (0. X ∈ functor on C. Observe that if we were dealing with set theory. reduced to its very essence.3) is another object in C. (2. Onset ⊕ P itch). and this means ﬁxing 10 The reduction of semantics to ‘pure arrows’ in cognitive science is approached in a remarkable work by Daniel Dennett [124]. (1. a comprehensive interpretation. There is no more than pointing and pursuing the arrows’ path as long as there is a way out. 181 Thesis 3 This is semantics. 40). Here.

do not yield any visible diﬀerence between the motives. . M1 and M2 look the same: For both. If we restrict to the singleton perspective. In other words. M2 and points addressed in the singleton and in a two-element motif. a suﬃciently general variation of the perspective can reveal subtler structures of arrow point sets. but not the full aﬃne geometry of the category of local compositions. M1 and M2 .3. Exercise 11 Another argument would have been to inspect the module functor introduced in section 8. as well as six injective arrows r from the two-element motif. So let us try a reciprocal confrontation of these motives. The next standpoint of perspective is a typical two-element motif as shown in ﬁgure 9. YONEDA PERSPECTIVES Pitch M2 40 r Onset p c r Onset 40 0 0 c Figure 9.5. However. Then we have injective arrows hitting two diﬀerent points in the target motives. On this level.3! We have a bijective morphism p · s : M1 → M2 through the projection p onto the ﬁrst coordinate and the shearing s in pitch direction (transvection). and we get six arrows which again.182 Pitch M1 CHAPTER 9.3: Two three-element motives M1 . Use the fact that a bijection in this situation must be an isomorphism and the fact that a functor maps isomorphisms to isomorphisms. We have two types of arrows r here: First those which factorize through the singleton: r = c · p. But the points in M1 are not collinear and we have no bijection. From this easy example from motif theory we learn that replacing an object X (a motif in our case) by its functor @X does not yield a full picture of the object at any one of its arrow perspectives. These can be any set-theoretically deﬁned maps.3. you can never see all of M1 from M2 . But in the converse direction. we only see the set-theoretic aspect. see ﬁgure 9. the isomorphism class of the set! But we are not in set theory and want to vary perspectives. unfortunately. deﬁnition 28. we have the three arrows c from the singleton and three arrows factorizing through the singleton from the two-element motif. every morphism must leave the collinear elements of M2 collinear. This breaks the symmetry of roles between the candidates.

The horizontal bijections are correspondences among point sets for any addresses A. In fact. whereas a single ‘address’ may blur a lot of a particular object. every possibility seemed too steep to be attacked by mountain climbers of those early days. . The common view from Zermatt showed no evident path. we now have the crucial statement of Yoneda’s lemma: Passing from the objects to their functors yields not only a rich system of perspectives but the entire information ∼ we need to tell whether objects are isomorphic or not. For instance: Is it possible to reconstruct X from a bunch of photographs Y1 .4: Whereas M2 can be pointed-at from M2 . a change of the ‘dogmatic’ perspective unveiled the solution to Whymper: Viewed from the eastern perspective of the Theodul glacier. Yn without knowledge of the perspectives pi : Yi → X? A historically famous example for this situation is the discovery by Edward Whymper of an easy route to the Matterhorn’s top.9. The moral of this lesson is that. YONEDA’S FUNDAMENTAL LEMMA 183 M1 q M1 s M2 bijection impossible! Figure 9. . ξ : @X → ∼ @Y is an isomorphism of functors iﬀ ξ stems from an isomorphism x : X → Y. there is no bijective M2 -valued point at M1 : In its totality. for every ‘address change’ g : A → B. With this in mind.2. . and the vertical arrows connect these 12 It B@ξ (9. If we view such maps as arrows (in an appropriate category) p : Y → X. we may ask about the information on the 3D object X which is obtained through its photographs Y under the viewpoint p. . In a more intuitive setup12 .4). ξ = η(x). we may look at 2D and 3D objects and perspective maps from 3D objects X to 2D objects Y which occur in painting and photography.5) A@ξ can be made rigorous by use of projective geometry. Mathematically. However. variation of address and perspective may aﬀord a more complete view of our object. but this is beyond our concern. M1 is literally ‘invisible’ to M2 . B. the northern crest appeared as much less steep than they believed from the common ‘knowledge’ down in Zermatt. the isomorphism ξ means that we are given a commutative diagram B@X − − → B@Y −− g@Y g@X A@X − − → A@Y −− with horizontal bijections. by bijection (9.

all we know from an object is determined via its behavior (its functorial point system). “The functorial point of view is the geometric one”. This intuitively means that if we give a system of bijections between points or ‘perspectives’ at objects X. We further know from Yoneda’s isomorphism (9. Y .4) that classiﬁcation of objects is equivalent to classiﬁcation of their functors.3 The Yoneda Philosophy Summary. Perspective and truth. if M is the address of D. But it is essential to understand its impact on musicology as it is exposed in the present viewpoint based upon functors and points in the form of spaces and denotators. In . As a corollary of Yoneda’s lemma we now understand why we always called “points in a space functor” the denotators. it is completely manageable on the behavioral level. we cannot explain all the proper mathematical consequences of the Yoneda lemma. the denotator’s D coordinates CT (D) is an element of the denotator’s functor F (D) ∈ 0 Mod@ . however. if the functor F ∈ C @ is not representable (see appendix G. We shall call the following body of insights the “Yoneda philosophy” because it is more than a technical approach. proposition 97) that it is the colimit of a diagram of representable functors Fι = @Xι . But even if a functorial deﬁnition fails corresponding to a real object.2). we can be sure that they indeed generate a ‘real’ isomorphism between X and Y . it can be shown (see appendix G. Whatever the ‘real ontology’ of an object. see appendix G. –Σ– As this is not a book on mathematics.1. YONEDA PERSPECTIVES bijections of variable address points. retrieved among ‘real’ objects. we can discover everything about an object via its point spaces on variable addresses: The object is recognized through variation of perspectives. By Yoneda’s statement 9. objectivity and dialog.5). which copes with diagrams of type (9. we have this statement: Corollary 4 The coordinates of denotator D is a point of the denotator’s space F (D) with values in the denotator’s address @M . we have an ontology in the sense that the point space of this address is a speciﬁc layer of spatiality. i. This philosophy is built around the restatement of the elementary character of Euclid’s points in terms of arrows: points are addressed perspectives.184 CHAPTER 9.e. and on each address. And therefore. The Yoneda revolution in mathematics: functors instead of objects. Knowledge is the result of a network of (ontological) perspectives. In particular. i.2. it is a paradigm of far-reaching power which reshapes not only mathematics but also concept architectures and in particular software engineering as well as methodologies in ﬁne and other arts. 9. an object can be ﬁrst deﬁned by its functorial properties. we may identify CT (D) with the arrow η(CT (D)) : @M → F (D).4. and then. Indeed. if required. an object may also be deﬁned by means of its behavior! In mathematical terms. that we are not given all possible ‘addresses’ to be able to calculate F (D) from its denotators since our addresses @M are only the representable functors. Notice. Therefore.e.2.

Understanding “equations” and “solutions” really is a deep problem: What does a particular address unveil about the object in question? What problem does a particular ‘geometric’ space perspective address? This agenda item drives us towards the approach of viewing the Yoneda philosophy as a fundamental concern of understanding things in spaces which are created to make problems evident. We shall see later. Since denotators are basically deﬁned by recursion to ramiﬁcation instances (via universal constructs such as limits etc. Understanding painting and music is a synthesis of perspective variations.1 Understanding Fine and Other Arts Painting and Music Summary. i. this is the philosophy of denotators. In order to understand an object via its functor. We have seen that in the theory of local compositions. such a statement is more general if we are not only looking at instances of geometric provenance. singleton perspectives are not classifying. the very object is less important than its behavior! 14 “Der funktorielle Gesichtspunkt ist der geometrische!” [175] . they are even necessary on a basic building level since rooted axiomatics can not yield all the conceptual foundations which are needed to think the concepts—instead of just playing around without knowing what instance is really doing the hardware of thinking.4. and Val´ry’s esthetic principles in view of the a e Yoneda philosophy. UNDERSTANDING FINE AND OTHER ARTS 185 other words: In the colimit.9. 9. all behavioral deﬁnitions are reachable via ‘objective’ deﬁnitions. there is a basic research agenda to pursue: • Exhibit a minimal selection of ‘ontological addresses and perspectives’ which are suﬃcient to classify an object (in which case we say that they are “classifying”). B¨tschmann’s. • Understand what are “equations”. In traditional architecture it is known that three mutually orthogonal projections are classifying. in the global theory. Of course. and what their “solution” means on a particular address.4 9. representable functors13 . The year 1954. circular constructs in concept architectures are not only possible.4.).e. It essentially states that the behavioral reality is the essential one! In music this has another interpretation which we shall review in section 9.4. Interpretation as an identiﬁcation process. to what degree local compositions are classifying. building concept points is also a concatenation of points qua arrows. Points are not only mathematical details. they are also technical realizations of concept constructs. –Σ– 13 In mathematics it has become common to use this technique to introduce objects since often. In particular. The Yoneda embedding as functorial review of objects yields a bundle of addressed ontologies which are by no means abstract: “The functorial point of view is the geometric one!”14 This proﬁled statement by Alexander Grothendieck’s pupil Peter Gabriel claims that geometry cannot terminate on the limits of everyday’s poor intuition and experience. Comparison of Adorno’s.

the musical work as we understand it in modern times. To the contrary: Following the composer’s indications has for a long time been a must and was even a severe criterion of success in the European tradition of music critique. how we all should see and accept it!” On the other.” Performance is a substantial part of a o 15 I remember that a computer-aided graphical analysis of Raﬀael’s School of Athens during the Darmstadt symmetry exposition [329] was vehemently attacked by conservative historians because the computer simulation included arbitrary variation of the original frontal central perspective.5 and 9. this would obscure the relative position of the work of art in a variety of possible alternatives. the consciousness of a particular view has always been part of the art because of its technical requirement to map a higher reality onto a canvas or another plane surface. The selection of a determined perspective. the prescription of a ﬁxed authoritative perspective was explication of a rigid ontology: “This is the world.186 CHAPTER 9. We come back to this subject in chapter 45. initiated a new approach to understanding musical composition. But there is one important point which broke this music tradition. This freedom of manipulation was felt as a serious disrespect against Rome’s Sancta Ecclesia. 16 Here. walking around the work which is positioned in free space is mandatory to understand it16 . A perspectivic view was accepted. symmetry and devotion. this is. interpretation and understanding. On one hand. Perspective variation was—on the contrary—a well-protected tabou of presentation. It is virtually never created without its message to an esthetic instance. Whereas in sculpture. We refer to our computer-aided graphical analysis of Raﬀael’s “School of Athens” which is documented in [329] and which was highly controversial because the computer simulation admitted free perspective variation and thereby destroyed the hitherto dogmatic frontal view of the monumental fresco in the Vatican’s sacred rooms. And it is always clear that the multiplicity of views and interpretations is either desired or at least accepted by the artist. but not as a possibility. the explicit point of view became even more prominent in that the divine position “sub specie aeternitatis” was replaced by an explicit human place. With the development of the central perspective in the early renaissance. was an expression of a weltanschauung or dogma for the humble addressee15 . The work is never neutral and independent of its perception. According to Adorno [6] we have to recognize this fact: “Die Idee der Interpretation geh¨rt zur Musik selber und ist ihr nicht adkzidenziell. be it in painting or sacral or political ceremonies. see also ﬁgures 9. perspective variation is practically forbidden in musical compositions of classical European tradition.The musical work is the discovery of the central perspective in music. Yoneda’s statement that the object is understood via the totality of its perspectivic views—its functor—is completely obvious. YONEDA PERSPECTIVES In the ﬁne arts. But it was much less evident and acceptable to the artists and even less to their religious and political patrons that a work of ﬁne art cannot be understood except through a complex and long-term process of variation of the perspective. human composition. It is about the same time when in music too.6. took place. . It is a commonplace to conceive the artist as the ﬁrst observer or listener. the passage from the representation of metaphysical harmonies as conveyed from Pythagorean tradition to the individual. You cannot really understand if you cannot see the special choice among a variety of—possibly worse or questionable—alternatives. rather as a mandatory presentation of dignity. above all in painting. and which must be seen in the context of Yoneda’s lemma: It was Theodor Wiesengrund Adorno who recapitulated the question of identity of a musical work with respect to performance.

UNDERSTANDING FINE AND OTHER ARTS 187 Figure 9. We could still argue that the best one is the only admitted one. He does a accept mutual contradiction within a set of interpretations which—each in itself—are consistent 17 ”Argumentationsgemeinschaft”. “ens causa sui” which appeared as one and the same object at any time and to everybody.4. B¨tschmann not a only presents a picture of multiple interpretations where the artist’s is only the very ﬁrst. and in this respect. This is nothing less than the requirement of a coherent network of single interpretations. but their concern is much deeper.5: Original frontal perspective of Raﬀael’s “School of Athens” in a computer-aided simulation. a critical point in B¨tschmann’s system. performance is not only a perspective of action but instantiation of understanding. Of course. In other words. It reveals hidden symmetries of the work of art. we must inspect a variety before selecting the optimal alternative. Val´ry’s approach was a rupture with the classical tradition that e viewed a work of literature as an autonomous unit. p.160]. In poetry.9.3. Adorno’s insight was anticipated by Paul Val´ry’s reﬂections on literature which e culminated in the famous dictum: “C’est l’ex´cution du poeme qui est le po`me. he also argues for an interesthetic dialog in a community of argumentation17 . however.” According e e to Hans Robert Jauss [250].6: Perspective from the right side of Raﬀael’s “School of Athens” in a computer-aided simulation. This means that the work of art is not fully existential before or without performance. In the hermeneutics of ﬁne arts. of interpretation of given structures. [43. Jauss concludes that a work of art only unveils its substance in the course of its historical life. this addendum is a critical one since now. Oskar B¨tschmann has explicitly thematized the mula tiplicity of interpretative esthesic perspectives in his systematic treatise [43]. . in fact the functorial interconnectivity of the collection of address ontologies. It seems that Adorno and Val´ry have uniquely stressed the exterior performance activity e in music and poetry. see [329] for details. There is. the Adorno and Val´ry approach is e classiﬁcatory on the level of functorial representation of works of art. This is quite the same as to say—with Adorno—that the literary work realizes its identity in the light of a time chain of performance perspectives. Figure 9. We should keep in mind that performance is strongly tied to its rhetoric function as a means to express understanding. but this is no longer the unique one. to use our terminology from Yoneda philosophy as developed in section 9. composition’s identity. performance with all its variations would no longer provide us with a unique perspective but with an a priori inﬁnite bunch of alternatives which share the composition’s essence. The simulation includes 58 human ﬁgures and architectural essentials.

methods. 9. This is exactly a hundred years after the equally simultaneous emergence of the local/global paradigm as proposed by Hanslick and Riemann. 18 But they will intervene in the chapters concerning design and implementation of software. –Σ– Concluding this chapter.4. it should not be argued that we already know everything about classical works of art. see parts X and XIII 19 The dot-notation for method actions on objects in the Java language is quite near to this category-theoretic formalization. Let us summarize that interpretation in the sense that functorial point perspectives distributed among a variety of ontologies is not a subsidiary task of understanding objects of the ﬁne arts or of mathematics or even conceptual constructs. Bartok’s compositions. and will for a long time not be. For instance. So an object has an interior structure which is hidden to the program’s context. be this Bach’s.2 The Art of Object-Oriented Programming Summary. identiﬁcation by behavior. It is not necessary to reject the total set in this case. object-oriented programming is essentially a categorical access to programming. they pertain to the object’s privacy and must be accessed via a special method. There. van Gogh’s. . Methods can be formalized in the language of categories [358].. we feel obliged to complete the circle from concept frameworks through mathematics and philosophy of ﬁne arts to the core of our knowledge society and its cultural impact: computer science. if an object describes a point on the graphical interface. Scriabin’s. Without explaining technicalities which are of no interest in this chapter18 we can characterize object-oriented programming languages as being built on programming units called objects. Characteristics. The complete identiﬁcation is not yet. a method is an arrow from the given object’s space functor (technically speaking: its class) to a second object’s space functor which is responsible for the method’s output19 . Picasso’s paintings. Raphael’s. These have an identity (in fact a pointer to their data structures) and are accessible through so-called methods which can be performed in a messaging action. nor is it of any particular quality if an interpretation is materially complete and consistent. i. We would in particular like to discuss the most advanced constructive paradigm of theoretical informatics: object-oriented programming. such as encapsulation. This is the encapsulation principle. Poe’s. Benn’s poetry. settled as long as the full functorial poly-ontology of point arrows is not developed to maturity. the point’s coordinates are not automatically accessible. you can only understand the object by its response to method messaging. roughly speaking. inheritance. or Villon’s. it is the very identiﬁcation process which is solicited in so doing. YONEDA PERSPECTIVES in argumentation and method. Conversely. A priori. We cannot accept such a “tolerance” which in fact denies any overall system of comprehension. namely around 1954–1956. All you know about objects is through their response upon messaging to their methods.188 CHAPTER 9. Rather should the fact of contradiction within a system of interpretations (and not within the proper art work where contradictions are common) be a source of reﬁned research.e. It is a remarkable fact that Adorno’s approach was virtually simultaneous with Yoneda’s. In contrast to its name. and class or instance variables do realize what the Yoneda lemma suggests: to replace program entities by the behavior they can show under determined conditions.

It is remarkable that the implementation of the denotator theory in the Java-based RUBATO software environment (see chapter 40) has realized the full functorial point of view exposed in this chapter. it is just an arrow to an address where you can access further data. it will only give you further information upon request of coordinate arrows which—in this programming paradigm—correspond to the object’s methods. .4. In terms of denotator theory. UNDERSTANDING FINE AND OTHER ARTS 189 Object-oriented programming is also an application of the point arrow approach to semantics which we discussed in the context of concept architectures: As such the identity of an object is fairly insigniﬁcant. the identity is the denotator’s arrow @A → F un(F orm).9.

.

we want to give an overview of the principles of local classiﬁcation and the corresponding references to sciences which are sensitive to this type of classiﬁcation. linguistics. We give motivation of the paradigm concept from musicology. Roman Jakobson [245] Summary.Chapter 10 Paradigmatic Classiﬁcation The poetic function projects the principle of equivalence from the axis of selection to the axis of combination. whatever this may signify—we have to review the diﬀerentiations of this subject. in practice. and mathematics. be they precise or only fuzzy. and mathematics. 191 . We are going to discuss the actual types of fuzziness and their genealogy in musicology. Here. Paradigmatic classiﬁcation deals with formation of classes of objects which belong to speciﬁc paradigms. they often appear in mixed form. In a more general setting than in mathematics. We do not deal with lexical classiﬁcation here. Rather is this onto-logical cornerstone a necessary evil which in virtually all cases is overridden by grouping of large sets of objects whose identity is much less relevant than some common features. In a mathematical perspective. from concept building and abstraction to concrete gestural actions. Its essence is very clear: In most human activities. Since chapters 11 and 12 will have more technical ﬂavor. paradigms are ﬁelds of equivalence or association. Rather are we concerned with paradigmata. semiotics and poetology. Concept taxonomy following classical Porphyrean trees and similar constructs are completely absorbed in that discussion.e. the second by topology.8 dealing with orders on denotators. equivalence is not necessarily understood as being a transitive relation. –Σ– This chapter is a preliminary discourse on local classiﬁcation theory. they can be incorporated into exact reasoning by means of a reﬁned paradigmatic reconstruction. This means that fuzzy concepts in the humanities are not a priori useless. Our taxonomy yields two types of paradigms: by transformations and by similarity—however. identity is not easy to get under control. i. “ﬁelds of equivalence”. this was already discussed in section 6. the ﬁrst type is covered by group theory.

it shows a multiple speciﬁcation which is essential to the word’s explanatory power. We call this meaning the topological one. and counterpoint. Its neighborhood to my own position is not given from an a priori ambient space but by . The ﬁrst suggests a space where neigborhood makes sense. We shall vaporize these mystiﬁcations in the following chapters. all I do is to pick it from a multitude of available objects in that neighborhood. Characteristic combination of transformation and similarity in poetics. so that we can better understand our point when recognizing similar points. the second yet seems already partly settled from our previous discussion on the Yoneda philosophy: classiﬁcation in the sense of a determination of isomorphism classes is nothing else than the overall task of understanding an object as we have identiﬁed it in the functorial approach in the spirit of Adorno. We could also say that the model object and its neighboring specimens are taken into account because they are similar to the point where I am standing. traditional musicology has only rarely been able to control the variety of their objects. model. and e B¨tschmann in chapter 9. but see our discussion in section 9. Already this Greek root has a double meaning for the relation of that thing: On one hand it is “side by side”. It is however open whether the object is in my neighborhood because I am pointing at it or vice versa: whether I am pointing at something because it is in my neighborhood anyway. that thing side by side α which you show or point at. PARADIGMATIC CLASSIFICATION We should however stress that classiﬁcation is highly controversial in musicology for three reasons: • It is commonly believed that classifying musical objects on whatever level is contrary to the individual expressivity of compositions as they have been cultivated since the renaissance. a 10. melodic contours. Group theory and topology: the poetics of mathematics. This ambiguity is after all not a defect.4.1 Paradigmata in Musicology. Val´ry. and Mathematics Summary. the model object is situated in a spatial neigborhood without my intervention. The diﬀerence between “side by side” and “point at” as explanatory roots is that in the ﬁrst case. • Classiﬁcation is misunderstood as a purely bureaucratic activity of list compilation. Linguistics. • Due to a catastrophical lack of technical tools. Transformation and similarity in music: Neumes. the model object is “side by side” because I have shown it in a pointing gesture. the second is just a connecting action from me to that thing. A disdain of detailed technical work which is psychologically comprehensible must scientiﬁcally be blamed for a major scientiﬁc retardation even with respect to other humanities such as linguistics. –Σ– The etymology of paradigm is Greek παρ` δειγµα: example. on the other. you “point at” it. In the second case.192 CHAPTER 10. variations. De Saussure’s paradigmatic and syntactic axes.

it is a well-deﬁned functional value. The word neume is rooted in Greek νε˜µα. This gestural system was then abstracted in the neume system with its diﬀerent writing styles and developmental completion. It is understood that the precise identity of a melody is less important than its—however ill-deﬁned—similarity class which may be termed “contour” to indicate some type of topological paradigmatics1 .8.10. and by this very etymology υ alludes to a movement of the voice that is only deﬁned in the vague similarity paradigm. Figure 10. After a long period of hidden or even patent refusal to work out explicit deﬁnitions of terms such as (melodic) “motif” 2 . To begin with. We call this meaning the transformational one. PARADIGMATA IN MUSICOLOGY. the cheironomic practice of hand gestures served as a paradigmatic sign system to indicate melodic movements in choral singing.1. Ms. before written notation of European music was developed. hint.1. 11th century). tracing articulation. LINGUISTICS. The examples can help understanding of our point because they are related under a speciﬁc operation. the resulting model object is not random as in the topological case. In this case. the vagueness of the contour concept has been attacked 1 Contour theory is an interesting object for the problematic building of valuable paradigmatic concepts in musicology. accentuation and diastematic movement in Gregorian choral. Bibl. The explanatory power here results from a transformation of my position to the target position. 10673.1: Neumes from the Beneventan writing (Rome. AND MATHEMATICS 193 a construction of this arrow whose head is my model object. . lat. This similarity approach in comparison of musical objects is widespread. Let us ﬁrst see how both these paradigmatic explications arise in music. The topological type means that we are given a musical object which is similar to the model object.e. Vat. Fact 4 Both topological and transformational paradigms help understanding of the proper position by the formation of classes of equivalent objects: the “example set” of what is to be explained. the model is a transformed object issued from my operation upon the given proper position. This rooting of melodic conceptualization is not only historic. i. we shall deal with it in several contexts and can restrict here to a characteristic summary. see ﬁgure 10. 2 See our discussion of Rudolph Reti’s theory of motives in chapter 22. it is—strangely enough—still a fundamental attitude towards melody: Music theory has not yet been able to really deﬁne a musical motif or melody apart from a priori vague concept ﬁelds.

a correspondence that cannot be similar at random. In structuralist linguistics. paradigmatic relations intervene in the basic dichotomic system. retrograde inversion. is just a restatement of a structure with neighboring values. without comparison of “equivalent” chords. We shall come back to this approach but should remark here that fuzzy relations are a poor imitation of what general topology. A huge ﬁeld of extremely topological paradigmatics opens in the subject of sound colors.1. see ﬁgure 10. and even theory of pseudometric spaces. harmony is based upon these classiﬁcations in the pitch domain.194 CHAPTER 10. the theory of turbidity (see our discussion in example 11 of section 8. So much the more is it remarkable that the entire Riemann program of functional harmony [100] which was designed to include the entire zoo of chord classes did not incite musicologists to write down complete lists of chord classes once for all. from major to minor tonality. pitch classes or chord classes are far from similar in the topological sense. an answer need not be a faithful image of the original. To Saussure. Note that. Clearly. The consciousness of similarity is also present in variational methods. On a more basic level. see the introductory section 2. say.1) is indeed thought in this topological spirit. the imprecise as a scientiﬁc category was the real thing which musicologists seemed to wait for. harmony would be an impossible task.1.3. in general. retrograde. or augmentation. In contrast.2. the decision of whether sound S1 is more similar to a reference sound than sound S2 is very diﬃcult and in fact points at one of the most complex problems in sound synthesis and analysis as well as associated topologies. What Saussure had called an associative ﬁeld has been reviewed in diﬀerent theories by . but they are essential in the understanding of chord or pitch concepts and structures. the contrapuntal technique of fugue and canon constructions typically recurs to the second meaning of transformational paradigmatics: The construction of a comes answer from the dux’ thematic germ makes use (among others) of contrapuntal symmetries. In this sense a tonal variant of a melody or harmony. could grasp much better. it can as well be a poor shadow. These secondary instantiations of the original object are transformations and not similar objects. The paradigmatic axis covers relations between language units. Variation does not mean to apply a determined transformation but to move around in a given topological neigborhood. This is also completely logical from the construction guideline of a fugue or canon: Answering to the dux melody is a “mental reﬂection”.2.4. For example. On the contrary: It is not true that these “models” are topologically similar to the original.3. but these relations are not present in the textual syntagm. We shall discuss it more carefully in section 12. But it has to be a functional consequence and not just something vaguely deformed. PARADIGMATIC CLASSIFICATION by what lends itself to the imprecise: fuzzy theory [428]. However. we have already explained this fact in the course of the general symmetry discussion in section 8. such as inversion. as sketched by Ferdinand de Saussure in his famous “Cours de Linguistique” [471]. The interesting point is that the fashionable expression “fuzzy” was necessary to introduce instances of topological reasoning to musicology. Also in the vein of the transformational paradigm is the classiﬁcation of chords according to transposition and/or inversion: A chord B is a model of a given chord A if it is deduced from A by a transposition and/or pitch inversion. A retrograde theme may look very dissimilar from the original. language is spanned within a twofold dimensionality: the syntagmatic and the paradigmatic axes. they occur on a more abstract level “in absentia”. the formation of pitch classes also creates a pitch paradigm by transpositions of multiples of an octave. Classiﬁcation is often based upon transformational paradigms.

Understanding means identifying not the individuum but its role. semantical. This position can be realized in two ways: topological deformation and transformation. for example. i. Observe that the paradigmatic relation is deﬁned on diﬀerent levels: phonological. In each. There are still uninformed persons—often in the humanities—who believe that mathematics is this type of mechanical quantiﬁcation business. numerical and functional quantiﬁcation. In mathematics. PARADIGMATA IN MUSICOLOGY. originally termed “analysis situs”. deals with invariants of geometric objects under shape deformation: What is preserved if we stretch a rubber band? What is the common property of all kinds of holes? Transformations are commonly dealt with in group theory: We want to describe the nature and essence of systematically performing transformations upon a set of objects: Is it the concrete mechanism or are we given a general context which is present each time we proceed to a transformational activity? Is it the case that there are no plane ornaments with ﬁvefold rotational symmetry? Why are there exactly ﬁve platonic. Topology. Both situations do occur: Grammatical paradigmatics is transformational in the sense that the inﬂection of a word.10. its position in the overall language environment. is topological. But it should be recognized that paradigmatic theories have not only introduced a strong qualitative proﬁle of mathematical concerns. Understanding a word as a member of a language is virtually never a question of abstract and isolated identiﬁcation. paradigmatic processes and methods are well known but not in the semiotic interpretation.1. The diﬀerentiation among these theories also reﬂects the double meaning of “paradigm”. anti-poetical science. . for example. AND MATHEMATICS 195 Figure 10. LINGUISTICS. grammatical. as Roland Barthes has rightly criticized in [41]. In a more transformational spirit3 . it is also true 3 Observe that we in no sense allude to Noam Chomsky’s transformational theory here. you learn rules of the language game which determine what you can do to an instance of the game without altering its essence. The classiﬁcatory function of paradigmatic activity is perhaps best understood in the linguistic context. rather it is a basic mathematical activity of classiﬁcation to boil down the overwhelming multitude of individual mathematical objects and structures. subsequent research.e. It was about geometric. relations among instances of one paradigm are termed “in opposition” and suggest a binary operationality which of course does not cover the general phenomenology. obeys the transformational paradigm whereas phonological vicinity in “minimal pairs” [12] such as “vine”/“ﬁne”. etc. perfectly regular bodies? Before the advent of these paradigmatic theories mathematics was a thoroughly sober.2: A visualization of the paradigm of the word “enseignement” in Saussure’s “Cours de Linguistique”.

Regular and degenerate rules: symmetries.2 Transformation Summary. Everything is already given by inherent structures. [477] and [363] for this background. As may be expected. be it on the phonological or on the semantic level. You simply have to recognize the model object as compared to the original object. We shall deal with this combined paradigmatics in chapter 22 on motif gestalts. music. Such an entanglement of mathematical and poetical roots induces a reversal of perspective: introduction of poetical aspects in mathematics. 4 This 5 See has been popularized in [229]. Even among conservative mathematicians it is commonly felt that works of art such as Cornelis Maurits Escher’s or Johann Sebastian Bach’s or Dante Alighieri’s approaches are a kind of incarnation of mathematics in the ﬁne arts. typically realized in rhyme constructs. a contrapuntal answer may be the inversion of the leading theme followed by a tonal alteration to a new scale. e Fact 5 The poetical character of qualitative theories in mathematics is a restitution of a common root of artistic and scientiﬁc thinking in the tradition of renaissance universalism as issued from medieval constructivism5 which goes back to the Pythagorean tradition of operationalized thinking in music and mathematics. –Σ– There is a fundamental and far-reaching diﬀerence between topological and transformational paradigmatics. such as shown in ﬁgure 10. But once a general acquaintance with the underlying ambient space topology is present. A very intuitive application of mixed paradigmatics is realized in Escher’s metamorphosis works. morphisms. specializations. nothing has to be added beyond perception of what is there. and that this is a substantial correlation4 . Of course this one need not be completely evident. For example.3 10. The former not only refers to an ambient space topology but also remains much more passive than the latter. to accumulate paradigmatic connectivity. and fractals. the most common syntactical symmetry transformation. alterations. Transformation as a transfer of association to generative rules. this projection introduces group theory in poetology since the functional connection of paradigmatics to syntagmatics uses syntactical symmetries. In fact. The recent impact of fractal theory and quasi-crystal theory on mathematical rationales of esthetics has conﬁrmed Hermann Weyl’s symmetry-oriented approach to the arts [564] as well as Ren´ Thom’s topological catastrophe theory of cognitive processes [526]. for example in Jakobson’s identiﬁcation of the poeticity as a projection from the paradigmatic axis to the syntagmatic axis [245]. . the double meaning of the word paradigm is often realized in combined form. and the rest is a plain veriﬁcation of similarity. acts on the association of rhyming units. PARADIGMATIC CLASSIFICATION that linguistic poetology has strongly proﬁted from the group-theoretic approach. the only ‘activity’ which is left to the observer is a cognitive act. The model object is then either a topological deformation of a transformed specimen or vice versa: a deformed version of a transformed specimen. and literature.196 CHAPTER 10. Rhyme periodicity. the psychology of topological pattern recognition is not trivial.

the projection of a part of a dodecaphonic series onto a chord is .V. its ‘objective trace’ no longer enables the transformation’s reconstruction. a transformation is a process which unfolds on a metalevel. For example. ending up with the garden gnome ﬁguration. double reﬂection restores the starting point. As a transformation becomes irreversible it is important to know the underlying transformation. Transformational paradigmatics is much more complex.10. whenever transformational paradigmatic association is required. Also the reﬂected object is a faithful image.2. not only Saussure’s “absence” is felt (in contrast to the syntagmatic presence of manifest contiguity). the objects which are acted upon will only be results of this activity. the level of understanding is also quite diﬀerent from the topological one. TRANSFORMATION 197 Figure 10. – Baarn – Holland. In other words: Fact 6 On the level of transformational paradigmatics. The generating cube is positioned to the left middle. There are several degrees of complexity of generative rules as related to their eﬀect upon objects. and therefore the phenomenon is degraded to a secondary level which can only be understood by a competence on the generative rules rather then their eﬀects. All rights reserved). the relation is evident and the operation as such is not diﬃcult to retrace. In the simplest case of an axial symmetry. then iterated according to a plane ornament symmetry. Not only are we asked to retrieve the actual arsenal of admitted transformations and therefore recall a non-automatic competence. the ornament is then deformed to organic shapes. In fact. the simple association of manifest objects is transmuted to a complex construction activity of the subject who is concerned with the paradigm. the activity as such pertains to a hidden operating system.3: Detail from Maurits Cornelis Escher’s “Cycle” (©2002 Cordon Art B. So. but also the highly demanding activation of our transformational competence.

1 for this subject. Paradigmatic methods in counterpoint and serial composition are transformational techniques which create ways of understanding the musical ideas as paradigmatic germs of an organic universe of sounds. a + a + x. where the classical logical diﬀerentiae speciﬁcae are reﬁned by a distance measure of fuzzy validation. . In the theory of alterations which we introduced in section 7. Here. or fuzzy relations in musical contour theory [428]. It may seem that the entire discussion of transformational paradigms is mainly an analytical aﬀair. . 6 The set is named after French mathematician Gaston Julia [254]. comparison of objects which pertain to ﬂoating paradigms is much more delicate. see section 47. . . the naive visibility of symmetry transformations has completely disappeared or at least been disguised in a fascinating esthetics. without its generating transformation rule the knowledge of the Julia set per se is completely mysterious. This is why the morphisms have become the core of structure theories. f n+1 = f · f n . on the contrary. x. transformations which are associated with alterations virtually disappear in the relative local compositions.2 for the corresponding discussion. The same spirit controls the theory of fuzzy sets [37]. as the attentive reader will have noticed.3 Similarity Summary. PARADIGMATIC CLASSIFICATION a “degenerate map” and destroys the temporal order of the series: the chord is a very special ‘shadow’ of the original series. f 2 = f · f. a + . The set of points which approach inﬁnity is separated from the rest by a boundary known as the Julia set6 . i. Similarity as a quantiﬁcation of comparison. . Anton Webern’s compositional techniques following Goethe’s organic principles realize these techniques. 10. . fc (x) for n → ∞ from those n x which have all their iterations fc (x) in a ﬁnite region of the complex numbers. but of an operation: Id.. . A transformation of type fc : C → C : z → z 2 + c is used to distinguish those complex numbers 2 n x which go to inﬁnity with successive iteration x. . see section 17. In psychometrics this problem is classically tackled by numerical quantiﬁcation of similarity. This trace of the hidden transformation—an iterated application of a ﬁxed function—shows no simple evidence of the transformation. altered local compositions are nearly topological deformations. .e.198 CHAPTER 10. . and which shows a breathtaking beauty. fc (x). f 3 = f · f 2 . see ﬁgure 10. . . a composer has to pay especial attention to the mentioned vanishing degree of transformations because the communicative evidence of a composition depends strongly on the emergence of the operational metalevel within the unfolding composition material. The extremal situation of degeneracy is fractal theory. In view of this. a + x. but it is equally concerned with composition and poiesis. Metrical and topological similarity. Whereas exhibiting logical diﬀerentiae speciﬁcae is an easy task. Quantities and qualities: What is a parametrization? –Σ– The scientiﬁc goal of comparison is a diﬃcult when the speciﬁc diﬀerence of the comparanda has to be objectivized. . for example by Charles Osgood’s semantic diﬀerentials [409]. f = f · Id. . fc (x). . a + x. such as category theory and its crystallizations in the theory of local composition as presented and commented on in chapters 8 and 9.4. . . So we are not looking at a repeated shift of a point. . .5.

However. i. In fact. We shall make substantial use of this type of non-metrizable spaces in the theory of musical motives. In this context. and this “analysis situs” has revealed an axiomatic construction where no reference to metrical data is left.e. “thick” and “thin” points of any extension. SIMILARITY 199 Figure 10. a paradigm of highly complex overall appearance. see chapter 22. Whereas a thick point may have many thin points in its closure. music is not a branch of psychometrics.4: The Julia set. the original concern of paradigmatic neigborhood was to understand what it means to stay in the vicinity of some given point. This approach is completely natural since experimental veriﬁcation is primarily concerned with numerical measurement. is a trace of a transformation fc = z 2 + c on the complex plane which completely disappears in favor of the visible result. But there are important classes of topological spaces which are far from metrizable. with topological spaces which can be deduced from the metrical neighborhood paradigm.10. there is a theoretical branch which deals with metrization theorems. see appendix H. for example Zariski topologies in algebraic geometry. of which a detail is shown here. These “exotic” topologies share the remarkable property that there are points of diﬀerent “size”.3.1. Powers of fc produce chains of associated points which tend to inﬁnity or stay in a ﬁnite region. The topological neigborhood concept has shifted to a pure quality where the naive “epsilon neigborhood of a point x” is replaced by an abstract open set containing x. these thin points do not have the thick point in their closures. and metrical distance functions are the right thing to invoke when such quantiﬁcation is required. metrical similarity is a very special realization of the topological paradigm. and no allusion to metrical measurement of vicinity was implicit. This is an important asymmetry which . In mathematical topology the generic concept of “vicinity” or “neigborhood” has been analyzed.

So topological paradigmatics is a huge extension of traditional metrical similarity. see appendix H for more details. • uncontrolled paradigmatics. In the humanities. the equilateral triangle is a very special example which we should never draw in classroom since equilaterality is not the typical property of a triangle. it is a scientiﬁc reﬁnement of the common language concept of comparison. we can mutually separate them by suﬃciently small disjoint neighborhoods. i. It is contended that polysemy is in contradiction to precision. fuzzy concepts are a frequent phenomenon. –Σ– There are two objections to precise terminology in the humanities: • historically and culturally generated polysemy. . or the diﬀerence between generic and special motives in a musical composition. parameters which include but go far beyond numerical quantizations. It nevertheless remains within the ﬁeld of objectivized description of paradigmatic vicinity. PARADIGMATIC CLASSIFICATION would not happen in metrical contexts: distances are symmetric quantities. This is the decisive diﬀerence between old-style and new-style precision. On the basis of metrical spaces one could never arrive at non-symmetrical neighborhoods and therefore never grasp the phenomenon of special and generic examples of a paradigm! For example. So the idea is to create spatial parameters for paradigmatic ﬁelds. Parametrization is an auxiliary measurement system but not necessarily by numbers.200 CHAPTER 10. Topology is the qualitative theory of neighborhoods. if we have to give examples of the triangle concept. The common denominator of topological paradigmatics is to distribute comparable objects in a space which allows the objects • attributes of spatial location which share the property of • being objective. explicit. and highly diﬀerentiated. and whenever we have two diﬀerent points in a metric space. giving examples of equilateral triangles would never produce other special types of triangles. topological character. This is an important advantage of non-metrical spaces which can be used in. adequate. topological motif theory. Polysemy has been a major objection to scientiﬁc discourse beyond what Hermann Hesse called “feuilleton science” [221]. rather by precise concepts of spatial. Paradigmatic reconstruction of such concepts by use of precise similarity and/or transformation concepts can help in validating amphibological discourses instead of rejecting them.4 Fuzzy Concepts in the Humanities Summary.e. this is a piece of mathematics without a priori quantiﬁcation which helps building powerful concepts beyond measurement. 10. The characteristic difference lies in the renouncement from plain quantiﬁcation. and is a central feature of. To understand this diﬀerence. topologies of the “exotic” type are necessary. We should however notice that the absence of immanent quantiﬁcation has its prize: Classiﬁcation of topological spaces is a hard program which is described in the so-called algebraic topology.

The reason for this success is that the signiﬁcation mechanism speciﬁed in the equation f (x) = 0 is explicit and precise. Hence it is not the problem to eliminate polysemy. In collaboration with the Institute for Architecture of the Technische Hochschule Darmstadt and the Fraunhofer-Gesellschaft computer graphics working group. we want to give a short illustration of how failure of paradigmatic tools can create defectous results in the humanities: In the course of an exhibition project on the Darmstadt Mathildenh¨he [329] and with respect o to the “School of Athens”. FUZZY CONCEPTS IN THE HUMANITIES 201 But there is no logical argument against a precise description of polysemy. The theory of global compositions developed in part IV is an example of adequate scientiﬁc formalization of polysemy for the analysis of musical texts. spheres.e. Usually this happens for one or more of three reasons: • blurred consciousness of the layer of reality or—more generally—the topographic location of the bunch of phenomena. But the central perspective in which this masterpiece was demonstrably painted permits more precise conclusions on the form of the geometric conﬁgurations in the painting. a solution X of an equation f (x) = 0 is a sign having multiple signiﬁcates. a smeared topographical analysis. It came to light that the star does not comprise equilateral but right-angled triangles which conﬁgure . themes and melodies. but to analyze it in a powerful language. blocks.e. • taking blurred reﬂections for blurred facts. Raphael’s famous fresco in the Vatican. However. Here. Quite radically. including the 20 most important among the total of 58 human ﬁgures.10. But any attempt to build a well-deﬁned paradigma from the vague concept ﬁeld fails. In the argumentation the defectuous concept tools are not recognized. a computer graphics reconstruction was made. detection of the third is more delicate. i. The more serious obstruction to scientiﬁc precision in the humanities is a fuzzy concept in a non-technical sense. or the addressee is persuaded that an uncontrollable inﬁnity of cases and variants simply rules out any conceptual eﬀort. [539]. A star of David must comprise two regular triangles. • defectuous concept tools for building well-deﬁned paradigmata. this semantic was never examined with respect to the precise star shape. To date it had been assumed that the star represented a star of David [149]. In short: the complexity of the phenomenon transcends human intellectual power.e. We shall discuss this latter attitude at length in section 22. is a successful branch of mathematics. The description of the variety of solutions. the collection of all valid semioses of the signiﬁer “solution X of f (x) = 0”.4. These are typically produced as follows: Initially. viz algebraic geometry. i. i.).2) is precisely the organized ensemble of polysemies on the analytical level. In general. a speciﬁc research yields a ﬁeld of conceptual variants which seems to point to a coherent phenomenon. Bramante star etc. the question concerning the signiﬁcance of the star being constructed by the Bramante ﬁgure in the fresco’s lower right was raised. Whereas the ﬁrst two reasons are simple defects in analytical mastery.8 on Rudolph Reti’s concept sketches dealing with musical motives. A well-known example of an exact theory of polysemy is the mathematical theory of algebraic equations. as well as important geometric elements (steps.2. uncontrolled paradigmatics. Rather does the discourse either seek a deep mystery in the phenomenon itself. Visualization of the object conﬁgurations from all perspectives with the aid of computer graphics software permitted new insights due to visualization of previously concealed features. the neutral identiﬁcation of a musical work (see section 2.

202 CHAPTER 10. . and that only rigorous elaboration of these tools by use of adequate computer software and geometric background theory led to a reliable understanding of the star paradigm in the given context. Result 1 The moral of this anecdote is that the paradigmatic tools for analyzing the deformed version of the star in the spatial paradigm of central perspective were not available in the traditional interpretations. a central ﬁgure of the composition. and in sharp contradiction to previous interpretations. At the same time it turned out that the points of these triangles are set with amazing precision in relation to the base-points of the principal human ﬁgures in the work [328]. PARADIGMATIC CLASSIFICATION in relation to the lateral faces of Platonic bodies and thus also create a link to Plato. Thus a new immanent semantic has become apparent in the fresco by means of precise geometric analysis.

12 Harald Fripertinger [169] 100 000 000 000 The average number of stars in a galaxis. The statement is 203 . Group orbits and the gestalt concept. This chapter deals with groups of symmetries. Enumeration theory of orbits of local compositions in ﬁnite Z-modules—including traditional pitch class sets and motives—is presented and discussed for its implications towards a “Big Science” in music. their action and orbits as musicological and mathematical concepts. self-addressed chords. we still recognize the melody. The concept is characterized by transposability and super-summativity. Here we concentrate on the ﬁrst point. We come back to this second point in chapter 12. including a review of the American tradition and recent developments.Chapter 11 Orbits 2 230 741 522 540 743 033 415 296 821 609 381 912 The number of isomorphism classes (orbits) of 72-element motives in Z2 . Elementary local compositions—chords. Transformational paradigmatics is addressed in Christian von Ehrenfels’ concept of a gestalt as proposed in [136]. Orbits are tools of conceptual abstraction. and motives are classiﬁed under group actions. Hubert Reeves [436] Summary. Follows a discussion of grouptheoretical methods in composition and theory.1 Gestalt and Symmetry Groups –Σ– Summary. Ehrenfels’s famous example of a melodic gestalt explains these attributes: If a melody is transposed in pitch. –Σ– 11. the single notes. and—according to Ehrenfels—this also demonstrates that the melody “is more than the sum of its parts”.

But such exceptional situations will be expressely mentioned.3. this invariant is a transformational paradigm under translation symmetries in the pitch domain (see example 7 in section 8. This means that the orbit set {et (M )| t ∈ ambient space of M } is an attribute of the gestalt of M .4 and section 8. call the category of these commutative local compositions.3. we recognize paradigmatic classiﬁcation as a classical subject of the humanities. So our commutative local compositions are addressed at an R-module. here: cognitive science in music and music psychology. The melody’s gestalt is something which is identically the same in all the melody’s transpositions. In terms of the general denotator theory of form semiotics exposed in G. Ehrenfels’ approach can only be a prototypical one. symmetries. together with R-aﬃne module morphisms ComLocR . set-valued functor . i. 11. changes after transposition. Using the aﬃne R tensor product as deﬁned in appendix E.1). something remains invariant. The very concept of gestalt refers to the entire set of transpositions of a given melody object.. as an individual local composition.e.5 for these local compositions and their categories). We delimit the categories of local compositions where one presently has signiﬁcant classiﬁcation results.1. we deﬁne a new contravariant. since the question of which “transpositions”. acting on a melody are really responsible for the cognitive construction of a gestalt cannot be answered ante rem.1. or for musicological analysis. we are working over the topos Mod@ instead of the larger topos Mod@ . it is obvious to identify the set of all transpositions et (M ) of the melody M with an attribute of this gestalt of M . Principle 5 So mathematical music theory has to furnish the orbits for all possible—or at least all reasonable—applications. address changes including diﬀerent R commutative rings may occur.2 The Framework for Local Classiﬁcation Summary. However. Again. as discussed in chapter 10. We should also notice a canonical technique of ‘address killing’ which occurs in the following situation: We are given a functor F ∈ 0 Mod@ and an address B ∈ 0 ModR . those having Rmodules as simple spaces and identify them whenever reasonable with their objective trace (see also section 7.e. –Σ– In this chapter we make the general hypothesis that all coeﬃcient rings R of modules are commutative.1) and could as well be extended to temporal translation (see example 8 in section 8.5. The concrete selection of the acting symmetry group is not the concern of our calculations. But then. for compositional purposes.204 CHAPTER 11. before cognitive experiments are performed to answer empirically to the empirical question in cognitive science. We also restrict to commutative local compositions. i. these must be justiﬁed by other criteria. But the a priori calculations of orbits is a technical prerequisite without which no eﬀective cognitive science can emerge.. In Ehrenfels’ example. be it for cognitive science. ORBITS that although the melody. say. This motivates the conclusion that building classes of local compositions under actions of determined groups of symmetries is a conceptual abstraction tool for constructing the gestalt concept. and this is an attribute of the melody’s gestalt.

. it has to be attributed to one of the list’s representatives. Y ∈ and f : X → Y ∈ 1 ModR . even if we had a theoretical solution of a classiﬁcation problem. and of small cardinality motives. If ever any random local composition is given. i. So the following sections are only the state of the art and not complete classiﬁcation. To begin with. and other categories as displayed in diagram 8. We therefore call it the B-address killer of F . Let X. and this requires understanding the candidate in depth! So what we suggest is at least a passive lecture on the following classiﬁcation discourse in order to trace the conceptual path and to memorize it in case deeper investigations should occur. classiﬁcation is far from settled in the module categories which underlie local compositions. M ∈ 0 ModR . we repeat it. However. We have added such lists in the appendices and the working musicologist may consult such lists if she or he does not want to plug in these technicalities.1 Classiﬁcation Techniques The ﬁrst step in any classiﬁcation program is to turn the given objects into points of a space where isomorphism classes are controllable by canonical parameters.11. Third. there is a considerable number of cases where complete classiﬁcation is feasible. an interesting problem of computer algebra. sorite 14) this functor has the property that ˜ ∼ 0R @B @F → B@F . classiﬁcation is a multi-threaded task.1) (11. we get the well∼ ∼ ˜ known identiﬁcation X@B @M → (X B)@R M → X@R (B@R M ).3.2) By the equation 0R B → B (appendix E. For F = @R M . ComLoc. it gives us back the B-addressed points of F as zero-addressed points. in particular for zero-addressed and self-addressed chords and rhythms. ∼ (11.3. we shall exclusively deal with ﬁnite. as we shall see. see appendix E. 11. non-empty commutative local compositions here. and the latter is essential for local compositions. ˜ f @B @F = f IdB @F. ˜ X@B @F = X B@F.31.e. Second. Already the strict enumeration of isomorphism classes is. lemma 79 and proposition 83. At present we are far from a complete classiﬁcation for three reasons: First. even if module category had solved the classiﬁcation problem the algebro-geometric classiﬁcation problem would be open. as we shall see in section 11. ORBITS OF ELEMENTARY STRUCTURES ˜ on R-modules B @F ∈ Then @ 0 ModR 205 0 ModR as follows.3 Orbits of Elementary Structures Summary. let us .4. 11. This section deals with diﬀerent common types of local compositions and their classiﬁcations. Finally. classiﬁcation is much more than the naked list of representatives. –Σ– In complex categories such as Loc. This means that musicology disposes of important complete lists of representatives of isomorphism classes. However. even on the algorithmic level. it would be an unsolved problem to ﬁnd algorithmic solutions for concrete cases.

the operation is K → h. . L ⊂ B@R N be two objects of ComLocgen End(B) . .. . So we are dealing with the simultaneous action of the − → − → groups GL(B) from the right and GL(M ) from the left on the set ComLocgen. But since both K. this is also deﬁned by a singleton (denotator) K : R B @M ˜ way: Write the elements of K as a sequence k = (k0 . . we look at local compositions K ⊂ B@R M of positive cardinality card(K) = n + 1.e.K. and this is h. .L → R. i. we denote this point n ˙ ˜ . α ∈ GL(B).. . This implies that we may select one representative of every (linear) isomorphism class of ModR and then concentrate on the local compositions within one ﬁxed such module M as ambient space.M of objects of End(B) − → − → gen ComLocEnd(B) with ambient space M .e.(L. . . Identify the symmetric group Sn+1 with the group of aﬃne automorphisms set by R @B @M of Rn which permute the canonical aﬃne basis (e0 = 0. So k is a zero-addressed R B@R M → R @R (B@R M ) → (B@R M ) = (0R @B @M ˜ point of (B @M )n+1 which is identiﬁed with an injective Rn -addressed denotator. 1)) of Rn . one by Sn+1 and one by GL(B). we are given an automorphism B@R h which sends the constant elements 0R @R M = M bijectively onto the constant elements 0R @R N = N . n ˜ (k) in the following But then.e.α. an Rn ˜ addressed generating2 point in B @M with pairwise diﬀerent coordinates.206 CHAPTER 11. We ﬁx an address B ∈ 0 M odR and an ambient module M ∈ 0 M odR . ∼ This means that we have an associated linear isomorphism R.e. . We further concentrate on generating local compositions. Then the address-change orbit k. i.K → R.. Let K ⊂ B@R M.α. 0).31. − → Putting these actions together. but with ﬁxed coeﬃcient ring R. as follows.α − − → B@R N −− with α an automorphism of B and an underlying morphism h ∈ M @R N of ambient spaces1 . we have a left action of GL(M ). kn ) and observe that Rn @B @M = ∼ ∼ n n n+1 ˜ )n+1 . i. .. and if K is such an object. we extend from the category ComLocgen as B deﬁned in section 8.(α. So the local (n+1)-element compositions ˙ ˜ K correspond to the elements of the orbit space Rn @B @M/Sn+1 . e1 = (1. however admitting all endomorphisms of B as address changes.f ) : R. 0.L.3) 1 And we have a canonical linear isomorphism R. An isomorphism f /α : K → L in this category is given by a diagram K − − → B@R M −− B@ α.e.α) since the action x → x. en = (0. points in ambi˜ ent spaces. (11.α is linear and α is ∼ auto. L are generating. 2 i.5 to ComLocgen End(B) as discussed in diagram 8. . i.4) We next transform these local compositions into singleton denotators. .f L. We view K as a zero-addressed local composition in ambient space B @M .Sn+1 of the denotator k corresponds 1-1 to K since it abolishes the arbitrary choice of indices of K-elements. and two commuting right − → actions. This gives the following theorem: ∼ Rh (11. If h ∈ GL(M ). h is an aﬃne isomorphism..3. ORBITS recapitulate the isomorphisms we will deal with. . 0. k deﬁnes a generating local composition.

3. B.6 commutes. n. M ) × Sn+1 → X(R.End(B) be the set of isomorphism classes of (n + 1)-element generating commutative local compositions in ComLocgen. k0 (0R @R M )) (11. –Σ– The next step is concerned with geometric parameters for the denotator orbit space DenOrb(R.M .5) 11. M ) = GL(M ) \ Rn @B @M/GL(B) × Sn+1 of denotator orbits. We have a projection of the denotator orbit space into the orbit space X(R. Classifying orbits of group actions on such standard morphisms. n. W ). ∼ (11. M )/Sn+1 ˙ ˜ in the following sense. M ). B.M n+1. n.11. . M ) = {(V. let X(R. ORBITS OF ELEMENTARY STRUCTURES 207 Theorem 4 With the above notation. all 0 ≤ i < j ≤ n. The geometric framework. (11. the linear part (α@g)0 of the aﬃne map (α@g) : B@R M → B@R M is α@g0 .End(B) → DenOrb(R. −1 −1 X(R. and there are R-linear isomorphisms such that diagram 11.2 The Local Classiﬁcation Theorem Summary.7) where σ0 is the linear part of σ. B. B. can. we have an element −1 drap(k) = (Ker(k0 ). M ). Lemma 8 With the above notation. n. B.6) (11. Before proving the next proposition we need a technical lemma. n. σ0 W ) iso. W )| V ⊂ W submodules of Rn with ei − ej ∈ V. iso.} W/V − − → 0@R M −− can. n.M n+1. n. B. let ComLoClassgen. Then we have a canonical bijection ComLoClassgen. M ). M ) : ((V. For k ∈ Rn @B @M . σ) → (σ0 V. Deﬁnition 29 With the preceding notation. n.8) of X(R.3. B. Rn /V − − → B@R M −− and deﬁne a right action of the symmetric group Sn+1 . and End(B) consider the set − → − → ˙ ˜ DenOrb(R. B.

. M ) − − → X(R. n. .208 CHAPTER 11.e.11) Rn − − → B@R M −0 − of linear parts according to lemma 8 also commutes. g ∈ − → GL(M ) such that the diagram Rn − − → B@R M −− α@g σ Rn − − → B@R M −− commutes. whence the claim. n. we have σ0 · drap(k) = drap(l). M ) −− (11.9) drap dr X(R. the kernels of k0 . Proposition 6 With the above notation. n. and the action of GL(B) = 1 becomes trivial. Rn ). ∼ Last. So k and l have the same orbit in DenOrb(R. Since the right arrow of the latter diagram is iso. Let us ﬁrst see that dr is welldeﬁned. and if f ∈ B@R M . Further. the subspace 0R @R M of con−1 −1 stant maps in B@R M is invariant under α@g0 . M ). B. α@g(f ) = (ee · (α@g0 ))(f ). For the rest of this section. we want to see that dr is injective for B = 0. If p(k) = p(l). l0 verify σ0 (Ker(k0 )) = Ker(l0 ). B. that if they have same orbit in DenOrb(R. To begin with. we suppose that B is the zero-address3 . and therefore. Suppose we are given two denotators k. ORBITS Proof. B@R M → M . n.10) l k 0 (11. and we want to present algebro-geometric classiﬁcation results in this case. then their images drap(k) · − → Sn+1 and drap(l) · Sn+1 coincide. This means that q(drap(k)) = q(drap(l)). α ∈ GL(B). l ∈ Rn @B @M . there is a (necessarily unique) surjective map dr such that the diagram p ˙ ˜ Rn @B @M − − → DenOrb(R. n. then α@g(f ) = eγ ·g0 ·f ·α whereas γ α@g0 (f ) = g0 · f · α is linear in f and α@g(0) = eγ . Therefore. Therefore. consider the short exact sequence 0 − − → M − − → M n+1 − − → M n − − → 0 −− −− −− −− ∆ d l (11. QED. B. there is a triple σ ∈ Sn+1 .12) 3 It is an open problem to ﬁnd a canonical generalization of the drap map for general address B such that dr becomes a bijection. the corresponding diagram Rn − − → B@R M −0 − α@g σ 0 q k (11. If eγ ·g0 is the decomposition of g. ˙ ˜ Proof. M ). if for −1 σ ∈ Sn+1 . σ0 induces a linear automorphism of M by the quotient automorphism from σ0 . σ0 (k0 (0R @R M ) = l0 (0R @R M ). In this case. i. B. M )/Sn+1 −− with canonical horizontal surjections commutes. Surjectivity of dr follows since drap and q are evidently surjective. For the zero-address we have several simpliﬁcations of the above formalism. B. In other words. The map dr is bijective if B is the zero-address. drap(k) − → specializes to (Ker(k0 ).

Then for all indices i = 1. . let m. Moreover: Lemma 10 The canonical map of orbits − → GL(M )\M n+1 /Sn+1 → GL(M )\M n /Sn+1 (11. Therefore we may observe the linear action of GL(M ) and of Sn+1 on M n . The left orbit set (and therefore also the right one) identiﬁes to the set of isomorphism classes of (n + 1)-element generating local compositions in M . both sets are invariant under the above actions. by ∼ the canonical isomorphism M n → LinR (Rn . Let m. M ) of LinR (Rn . . i. this action identiﬁes to the left-right actions of these groups on LinR (Rn . Clearly.11. the canonical left action of GL(M ) and the right action of Sn+1 on Rn via a group homomorphism D : Sn+1 → GLn (R) which on transpositions (1. Proof.14) −1 −1 . . 1 D(1. we have ni − n0 = g(mσi − mσ0 ) = g(mσi ) − g(mσ0 ) and hence the aﬃne equation ni = (en0 −g(mσ0 ) · g)(mσi ). mσ1 . . The map of orbits is also surjective since d : M n+1 → M n is surjective. . Let M n be the subset of M n consisting of the n-tuples with pairwise diﬀerent and nonvanishing coordinates and such that these coordinates generate the module M . and this yields the following zero-address version of theorem 4. m1 . . mn ) · σ = (mσ0 . −1 0 1 with −1 entries overriding the n × n identity matrix on row i − 1. n. . and d projects M n+1 onto M n . σ ∈ Sn+1 with g · d(m. . mσn ) which leaves ∆(M ) invariant and therefore induces a linear action on the quotient M n . in the left orbit set coincide. ) · σ = d(n.e. We also have the left − → diagonal actions of GL(M ) on M n+1 and of GL(M ) on M n . The ﬁrst claim is true by construction. For the second. and this shows that the orbits of m. . . m. . . −1 .3. Clearly. . Then d(eu · g(m. M ). )) = g(d(m. i) = (11. n. . . Set M n+1 = d−1 M n .13) is a bijection. . . We have a right linear action of Sn+1 on M n+1 deﬁned by (m0 . .. mn − m0 ). i) is given by the matrix 1 0 0 1 . )) since u cancels out by the diﬀerence homomorphism d. Further: Lemma 9 The projection d is equivariant with respect to the above actions of the symmetric − → group Sn+1 and the canonical group homomorphism GL(M ) → GL(M ). . m) and the diﬀerence formula d(m0 . ∈ M n+1 such that there are g ∈ GL(M ). . Proof. . m1 . . . . M ). ). . mn ) = (m1 − m0 . M ). . ORBITS OF ELEMENTARY STRUCTURES 209 of R-modules and linear maps with the diagonal embedding ∆(m) = (m. . and n. . QED. This reduction from aﬃne to linear structures is only possible for the zero-address. Through this identiﬁcation M n identiﬁes to a stable subset LinR (Rn . . ∈ M n+1 and em · g ∈ − → GL(M ). The last claim is clear by construction.

O We now proceed to the linear analogy of the ﬂag space X(R. M ) q −−→ −− Ker M Xn (R) p (11.M R → GL(M )\LinR (Rn . M ) have kernels which diﬀer by a permutation. n. n+1. Let ComLoClassgen R be the set of all isomorphism classes (any ambient R-module M n+1. ei − ej ∈ V. an action which we recognize as being the linear analog to the above action on X(R. M )/Sn+1 . B.210 Theorem 5 There is a canonical bijection CHAPTER 11. Since the horizontal map Ker is Sn+1 -equivariant. M ) for zero-address. We know from theorem 5 that this class set is in bijection with the orbit set GL(M )\LinR (Rn . Then we have a disjoint union M Xn (R) = Xn (R) (11. In the diagram LinR (Rn . B. n+1. ∼ M Clearly. QED. (11. then their quotients which identify to M are related by an automorphism of M and therefore. n. via −1 V ·σ = D(σ )·(V ). and we may factorize through the GL(M )-orbits.16) {V | V ∈ Xn (R) and there is an R-linear isomorphism Rn /V → M }. the factorization through f is deﬁned. l : Rn → M in LinR (Rn .17) M ∈M (most of these co-factors are empty. the kernel of g does not change under an automorphism of M .O Proof. M ). Then the above lemma gives g . M )/Sn+1 . and f is surjective since both. if two Rn -addressed denotators k. q(k) = q(l). Ker and p. Xn (R) and Xn (R). of course) and its corresponding orbit space union Xn (R)/Sn+1 = M ∈M M Xn (R)/Sn+1 . ORBITS ComLoClassgen.15) (11. there is a unique bijection f which makes the diagram commute. the symmetric group Sn+1 acts from the right on both.18) M Lemma 11 The orbit set Xn (R)/Sn+1 is in canonical bijection with the class set ComLoClassgen. In fact. so f is mono.M R .O allowed). Let M be a system of representatives of isomorphism classes in ModR . all 1 ≤ i < j ≤ n}. M )/Sn+1 − − → Xn (R)/Sn+1 −− f with canonical vertical arrows and the upper horizontal arrow Rn → M → Ker(g). are. We set Xn (R) = {V | V a submodule of Rn with ei . M Xn (R)= ∼ (11. Finally.19) M GL(M )\LinR (Rn .

every such composition is isomorphic to a generating one4 and we may forget about this speciﬁcation and even concentrate on ambient spaces of coordinator Qr . and ﬁnally. i..e.e. generating. Before delving into techniques of algebraic geometry we should try to understand the geometric aspect of theorem 6 and its musical interpretation.O We therefore may concentrate on the left space and look to situations which may lead to geometric classiﬁcation spaces.K). Qr ) by a Qn addressed denotator k : Qn → Qr means that we really have an instance of a more general approach which is based on standard local compositions (see ﬁgure 11. The local composition ∆n is called standard simplex composition of dimension n. by the drap r 4 Simply because every local composition (K. n+1.11. F.. The variability of K has been transferred to that of k : ∆n → Qr . we denote by ∆n the zero-addressed commutative local composition of cardinality n + 1 in ambient space R⊕n consisting of the canonical aﬃne base. Observe that for Q. ∼ D1 e2 D2 e2 D3 e0 e1 e0 e1 e0 e3 e1 –1 –2 –3 Figure 11. i. the denotator k : Qn → Qr corresponds 1-1 to the morphism k∆ : ∆n → Q with generating image. replacing a generating local (n + 1)-element composition (K. zero-addressed local compositions over ring R) There is a canonical bijection Xn (R)/Sn+1 → ComLoClassgen R .3. 2. . ∆n = {ei | 0 ≤ i ≤ N } (11. So we have abstracted from K and concentrated on one distinguished local composition: the standard simplex composition ∆n . To begin with.1: Local standard simplex compositions over Q of dimension 1. ORBITS OF ELEMENTARY STRUCTURES 211 Theorem 6 (Classiﬁcation of n+1-element. where the classes appear as points on appropriate geometric spaces.20) with the standard notation e0 = 0 and ei = (δij )1≤j≤N for i > 0 and the identiﬁcation of R⊕n with 0R @R⊕n . M ) over a ﬁeld F is isomorphic to (K.1): Deﬁnition 30 Given the ring R and a natural number n. Let us concentrate on the intuitive situation of local compositions over the rationals. and K identiﬁes to the orbit k∆ · Sn+1 . and 3. With this in mind.

Observe also that the orbit V · S n+1 is contained in the Sn+1 -invariant set of the n − rdimensional subspaces which are complementary to all W · σ. this passage can be viewed as a realization of a symbolic composition in special. We shall come back to this classiﬁcation approach which also solves the problem for global classiﬁcation in section 15.1. the more important point is that the passage to kernel spaces via drap turns the seemingly dispersed appearance of local compositions in their ensemble into a geometric situation because subspaces can easily be compared with each other by continuous movement. This means that we may now look for class-invariant geometric relations among 5 See appendix F. For the time being. and deﬁned by the variable “coordinate system” Qn = V · σ ⊕ W which works for all orbit members.s V‡ s W Figure 11. generic composition” ∆n and to ﬁnd a set of coordinates to realize a special variant of this generic data. projection pV. this process means to give an “abstract.212 CHAPTER 11. musically meaningful coordinates. that of k has been reduced to the kernel V = Ker(k0 ).σ that is complementary to a given subspace W in Qn gives rise to a speciﬁc projection of the standard composition ∆n whose image deﬁnes a special instance of our ensemble of local compositions.1. together with an ensemble of projections pV. By construction.1.1. This means that we deal with one “generic” local composition whose points are in general position6 and then project this generic specimen onto specialized images deﬁned by selection of special coordinate functions. Ontologically. 6 See appendix H.e.1.5. This may be viewed as a variable coordinate system and leads to musical interpretation of the classiﬁcation technique.1. This is a stable open set in the corresponding Grassmannian variety (appendix F.. there is a single submodule W ⊂ Qr which is a linear complement for all elements V · σ of this orbit: Qn = V · σ ⊕ W . In other words: the ensemble of local n+1-element local compositions K with dim(K) = r is transformed into one single “standard representative” ∆n . K = pV. p V.2: Each kernel space pV. But what does this restatement mean? We have a ﬁnite orbit V · Sn+1 and therefore5 .2.σ indexed by V and the symmetric group. We shall work out this now somewhat artiﬁcially blown up interpretation in 15.σ (∆n ).5). .σ : V · σ ⊕ W → W gives us K as the bijective image of the standard composition. i. but want to mention it already now since this sheds light on the entire classiﬁcation business. ORBITS function. Musically speaking. see ﬁgure 11. all σ ∈ Sn+1 .

their module R. Further.r (R). a scope which allows a very reﬁned comparison of musical structures—such as motives or chords—in concrete compositions. But scalar extension commutes with formation of exterior powers ([63.4. if x ∈ Rn . the present eﬀorts aim at a geometrically sophisticated understanding of local compositions.r (R) of the Grassmann functor Grassn. 8]) and so ∧r 1 ⊗ x = 0 with respect to the direct summand S ⊗ V of S n . x ∈ V }. Denote by Mr the representative subsystem of modules in M r M which are locally free of rank r ∈ N and set Xn (R) = M ∈Mr Xn (R).x (f ) : Vn. ORBITS OF ELEMENTARY STRUCTURES 213 local compositions. applications of this view are given in section 11. To construct a geometric parametrization of classes. We can show the following: 7 See section F.2. letVn.r (R) → Grassn. r ˜r We need to deﬁne a subfunctor Xn (R) ⊂ Grassn. In particular. Xn (R) is not functorial in R. Prop. setting eij = ei − ej for i = j and eii = ei for the canonical basis r r r vectors ei in Zn .eij (R) = Vn.e.5 for the theory of Grassmann schemes. r Xn (R) ⊂ {V ⊂ Rn | Rn /V = locally free of rank r}. theorem 144).x (R) iﬀ ∧r x = 0 n for the direct summand V ⊂ R .x (R) = {V | V ∈ Grassn. Exercise 12 Prove this last claim with R = Z. . and the complement of their union r r Vn (R) = Vn. for example linear incidence among their points.III. to Vn (R).r (f ) : Grassn.22) r In fact.eij (R) 1≤i≤j≤n gives us the Sn+1 -invariant set r r Xn (R) = Grassn. and that we want to ﬁnd a functorial expression Xn (R) ⊂ Xn (R) of its open complement.3. In other words.r (R) is in Vn. §7. and to Grassn.1⊗eij (R). deﬁned by Vn.r (R). (11. M ) which are locally free of rank r.1⊗x (S).r (R) which is disjoint from Vn (R) and r r “essentially” gives us Xn (R).eij (R). we know from appendix E. r r r In contrast to the Vn. if f : R → S is a ring homomorphism. no.r : ComRings → Sets (11.r . The r above action of the symmetric group Sn+1 permutes the subfunctors Vn. we restrict to local compositions (K.. ch.x (R) → Vn. but the latter is by deﬁnition7 the evaluation Grassn. i.eij .6. In other words. we have subfunctors Vn.21) r at ring R.r (S) : V → S ⊗ V restricts to r r r Vn. the functorial map Grassn.K is locally free of rank r (see appendix F.11. More precisely.2. The idea is that Vn (R) is representable by a closed subscheme r ˜r of the Grassmannian.eij of Grassn.5. lemma 83 that V ∈ Grassn.r (R) − Vn (R).

the intersections (11. i in i.eij (R) → Grassn.r. Since the action of the symmetric group permutes the elements eij .kl ∈ Grassn. .r.ekl (R) identiﬁes to the set t of n × (n − r + 1)-matrices Mf.eij is covered by the ˜r open subfunctor Xn. Further. and every orbit in Xn is contained in an aﬃne r r r ˜ ˜r neighborhood. We know from appendix F.kl deﬁne the ﬁrst column of an invertible n−r+1 × n−r+1 -matrix over R. . (R) ∩ Xn.5 that the functor Grassn.eij is an open r subscheme of Grassn. .i.. (R) with the subfunctor Xn. QED.i. n − r and an (n − r + 1)th column ekl such that all (r + 1) × (n − r + 1)-determinants of Mf.r (R) are representable by closed subschemes r Vn.r.r − Vn. We then ˜r look for the intersection of Grassn. The inclusion Xn (R) ⊂ r Xn (R) follows from construction.5. . Lemma 13 The open subscheme ˜r Xn = 1≤i≤j≤n ˜r Xn.eij are covered by ˜r Xn.3. To see that the open complements Grassn.eij (R). This set n in aﬃne n−r+1 -space is open by proposition 96 in appendix F. . Let i. observe that if R is a ﬁeld. denotes the complementary index sequence of i. To see that this is an open subfunctor. The last statement is clear for a ﬁeld. Proof. .r.eij → Grassn. We have Xn (R) ⊂ Xn (R).r (R) consisting of the direct factors V ⊂ Rn such that ∧r eij : r V → r+1 Rn splits (has a right inverse). Restricted to such an open aﬃne subfunctor. i2 .2. (R) of those direct factors V ⊂ Rn such that Ri. ∩ Xn.ekl (R) since non-vanishing of the determinants creates a basis vector by the exchange theorem for vector spaces. ).214 CHAPTER 11. . to the n × (n − r) matrices with columns r (ei .II. ir be an increasing subsequence 1 ≤ i1 < i2 < . and the determinants are ˜r polynomial functions of the coordinates of Γf .r.i. ch. . then Grassn. If i. n−r+1 n n of Mf. Each of the open complement schemes Grassn.e. .r (R) is covered by the aﬃne open subfunctors Grassn. ORBITS r Lemma 12 The subfunctors Vn. f (ei )) for j = 1.r. . . This is a closed condition deﬁned by the ideal which is generated by these determinants.5. and hence for a semisimple ring which. by Wedderburn’s theorem is a ﬁnite direct product of ﬁelds (see appendix E. Ri.eij (R) ⊂ Grassn.i. observe that an R-valued point ˜r Mf. 4. Vn. i. f (ei ))t . b = 1. so the intersection Grassn. and if R is semi-simple we have Xn (R) = Xn (R).24) ˜r is invariant under the given action of Sn+1 .r − Vn. theorem 48). Proof. since the Grassmannian is projective. .eij .r .eij (11. .kl vanish. ir ≤ n.i. QED. it permutes the members of the intersection (11. = i1 .24).eij (R) (11.4]. (R) identiﬁes to the set of graphs Γf of linear maps f ∈ LinR (Ri.kl with n − r columns (ei .23) coincide with the complements of the r sets Vn.i. . .23) n is characterized by the property that the (n−r+1)×(n−r+1)-determinants db . . every ﬁnite ˜r ˜r orbit in Xn is contained in an aﬃne neighborhood by [199. projects isomorphically onto Rn /V .

sp of isomorphism classes of zero-addressed. If R is a semi-simple ring. that R(k − l) is a direct factor of any module which contains it.3. Proof.e. The exactness property follows from lemma 13 and [123. i.11.0R of isomorphism classes of zero-addressed.4. In fact. QED.III. all module homomorphisms .lf. If the coeﬃcient ring R is self-injective9 .4.. the condition that the linear homomorphism .3]. ORBITS OF ELEMENTARY STRUCTURES 215 ˜r Theorem 7 With the above notation.(k − l) : R → M : r → r(k − l) for any two diﬀerent elements k. local (not necessarily generating. n+1. no.0R locally free. ch.0R practice.(k − l) : R → M : r → r(k − l) is split boils down to the vanishing of the annulator of k − l: ∼ Ann(k − l) = 0.1.25) is characteristic for membership in Xn .3. §2. the set ComLoClassn+1. (11. appendix E.3.6. and generating (n + 1)-element compositions with any R-module as ambient space is in canonical bijection to the set of R-valued points of the scheme Cln = 1≤r≤n ˜r Xn /Sn+1 .sp in a special case which is useful in n+1. there is a quotient scheme8 Xn /Sn+1 . locally free. see theorem 59 in appendix F. . 8 In 9 See the sense of [123. l ∈ K are split injections.(k − l)) = Ann(k − l) and in that case.25) we say that it is split. no. we have Ker(. local.25) Deﬁnition 31 If a zero-addressed commutative local composition has the splitting property (11. ˜r The translation of the deﬁning property for the scheme Xn is that for such a generating local composition (K.6. §2.lf. ch. or split) (n + 1)-element compositions with any R-module as ambient space is in canonical bijection to the set of R-valued points of the scheme Cln . there is an exact sequence of schemes ˜r Xn × Sn+1 pr1 µ ˜r ˜r Xn → Xn /Sn+1 ˜r with group action µ and ﬁrst projection pr1 . M ).1]. split. Corollary 5 The set ComLoClassgen. R(k − l) → R means by appendix E.III. Let us have a closer look at the set ComLoClassgen. ˜r Exercise 13 Verify that the splitting property (11. proposition 88. Its R-valued points are the orbits of Xn (R) and of r Xn (R) in case R is a semi-simple ring.

a situation which is particularly important in music theory. Zpnr . there is a unique maximal element ((K1 × K2 . ORBITS 11. A AS AT × : ComLocin → ComLocinS × ComLocinT . Zn ) for 1 < N. Lemma 14 The above map × : ComLocA → ComLocAS × ComLocAT is a functor.1. AS @S MS ). For an address A in ModR . has well-deﬁned restrictions × : ComLocemb → ComLocemb × ComLocemb . K2 ).8. we will consider zero-addressed local compositions (K. 0 < N n. Recall the following from appendix D. Let R = S × T be a direct product ring. Denote by x = xS + xT the decomposition of element x ∈ M into its S-component xS = 1S x and T -component xT = 1T x. we then associate the pair ×(K.5. . and 11. We call them K1 -K2 -projecting. A@R M ) ∈ ComLocA . . we may apply sorite 6 in section 8. Since ZN is self-injective by appendix E. M an R-module. A@R M ) = ((KS .216 CHAPTER 11. (K2 . A@R M )) ∈ ×−1 (K1 .and of a T -module. A A A and all these functors are surjective on the objects.3. .3 The Finite Case –Σ– Summary. pnr is the prime factorization (see appendix D. With any object (K.3. So we ﬁrst have to deal with direct products of rings and corresponding categories of r local compositions. UT = K2 . MT = 1T M the corresponding decomposition of M such that M = MS ⊕ MT is a direct sum of an S. proposition 78. we may concentrate on the latter. We leave the proof as an exercise.3.2). So the ﬁber consists of all local subcompositions U ⊂ K1 ×K2 with US = K1 .3. AT @T MT ) → (LT . deﬁnition 129. we also have a corresponding canonical ∼ decomposition A@R M → AS @S MS ⊕ AT @T MT which carries over to local compositions in these spaces. Although a general geometric classiﬁcation of local compositions in the above sense is not accomplished. AT @T MT )) ∈ 0 ComLocAS × 0 ComLocAT . 11. AS @S MS ) → (LS . (KT . we have ZN → Zpn1 × r 1 1 . Recursive classiﬁcation on ﬁnite modules via socles and geometric classiﬁcation.1. AS @S NS ) and fT : (KT . We shall make this more explicit in sections 11. A@R N ) in ComLocA we have an obvious decomposition into two morphisms fS : (KS . . For a morphism f : (K.5 to classify such local compositions: Since the categories ComLocin and ComLocgen are 0R 0R equivalent.3. AT @T NT ) and therefore a map f → ×f = (fS . More precisely. AT @T MT )) ∈ ComLocAS × ComLocAT . and by MS = 1S M. fT ). . A@R M ) → (L. we may use geometric classiﬁcation to settle local compositions in ﬁnite ambient spaces. AS @S MS ). Observe that for a pair ((K1 . ∼ If N = pn1 · . A AS AT × : ComLocgen → ComLocgen × ComLocgen .7.

proposition 88. and we are in the case of the characteristic p prime ﬁeld R/Rad(R) = Zp (see appendix E. K2 of representatives of each class.r−s (R) and can be classiﬁed by the geometric method which is standard for Grassmannians. Recursion runs on the cardinality card(Zr n ) of our free rank r module over p the ground ring R = Zpn . the problem boils down to ﬁnite product group actions on ﬁnite set products. Since Rr is injective. the latter concerning the esthetic aspect of classiﬁcation. 1. it is free. Let us now return to the more substantial problem of classifying the factor objects. I → Rs . there is an injective envelope I ⊂ Rr of V . s < r.3.1) of I. a subject of pure combinatorics. Let V ∈ Ur (R).4. By appendix E. We want to settle this problem by means of a recursive 0Zp procedure that runs on the set of orbits of the set Ur (R) = {V | V a submodule of Rr } (11.11. objects in the category ComLocgenn . This process settles many of the practically important cases as we shall see in a moment. We shall however give several partial answers in the following sections. Let soc(V ) soc(Rr ).4.e. But we have to stress that we do not control the general geometric classiﬁcation with address changes—a fortiori we do not control the deeper musicological consequences of this general geometric setup.2. 11. Since the socle is invariant under any linear action. we may as well calculate the orbit of V /soc(Rr ) in Ur (R/soc(R)) which is recursively earlier. If these are ﬁnite.17. and in section 11. then V ∈ Ur (R/Rad(R)).3. This completes the recursion process.26) under the action of a subgroup G ⊂ Sr+1 which extends the known linear action on Xr (R) deﬁned in 11. i. see section 11.. This is the geometric situation which is settled in theorem 7. Then we have V ∈ Us (R). 3. and we have to calculate its G-orbit for smaller cardinality.1 on this subject. If soc(V ) = V . Let soc(Rr ) ⊂ V (see appendix E. i. it is a direct factor of Rr .6. This direct factor of Rr is an element of the Grassmannian Grassr. 2..3.4 for the socle concept). Let G = (Sr+1 )I ⊂ Sr+1 be the isotropy group (see appendix C.e. –Σ– .2. We discuss the problematic concept of dimension of local compositions. Then classiﬁcation of the set K1 K2 of K1 -K2 -projecting local compositions amounts to the calculation of the orbit set Aut(K1 ) × Aut(K2 )\K1 K2 under the obvious left action of Aut(K1 ) × Aut(K2 ).4 Dimension Summary. Observe that the latter groups are subsets of the symmetric groups of the underlying sets K1 and K2 . and since ∼ R is local. Let soc(V ) soc(Rr ) and soc(V ) V .4 for the radical concept).3. ORBITS OF ELEMENTARY STRUCTURES 217 Suppose that classiﬁcation is settled for the factors and that we are given a pair K1 .

3. ORBITS We already learned in section 6.218 CHAPTER 11. So in both cases. 3 anticipating classiﬁcation of motive residues in these ambient spaces (see section 11. − → − → but the corresponding operating group GL(Z2 ) is ≈ 691.26). and appendix M). e. If we want to classify motives modulo octave and in onset periods dividing 12. 12 which at ﬁrst glance is very surprising. 12 (typical to dodecaphonism) with those for m. and a 5-Sylow group. 12 against 45 classes for m. but in the ﬁrst case.2 times larger than GL(Z5 × Z12 ). We have Z2 → 12 ∼ Z2 × Z2 whereas Z5 × Z12 → Z5 × Z4 × Z3 . There is however strong evidence from practice that this could be the case in light of the above results: Among jazz musicians it is well known how diﬃcult it is to improvise in 5/8 time signature compared to 3/4.5 12 for the calculation). ﬁfth. − → − → The orders are card(GL(Z2 )) = 663 552. that the pitch space is not the seemingly one-dimensional real space but the subtler construction of Euler module which in the common setting (octave. 12. 6/8 or 12/8 time signatures. rhythmic phenomena are understood to share period 12 whereas in the second case. we have 26 isomorphism classes in case m. onset is taken modulo 12 and therefore.3. or 6/8.12 . Music theory thinks of octave.7 times more isomorphism classes for the 5-periodic motive residues. But if we proceed to an identiﬁcation of integer onsets modulo m and integer pitch modulo n—a completely common practice. the 4 3 second has a 2-.. look at a zero-addressed motif (7. This corresponds to the general aﬃne groups − → − → − → ∼ GL(Z2 ) → GL(Z2 ) × GL(Z2 ). the period has to be 5. To ﬁx the ideas.and a 3-Sylow group. we compare the residual motives for m. To understand what happens. There are 144 = 487 344 zero-addressed three-element motive residues in OnP iM od12.12 than in OnP iM od5. 3/4.7 times easier than to do so modulo octave and in onset period 5. We have ≈ 14.2. It is of course not known—but to ﬁnd out would be an important experiment in cognitive psychology—whether and to what degree humans do perform classiﬁcation tasks in the above sense. one agrees that the dimension of the underlying space is the rank 2 of a space isomorphic to Z2 . 12 4 3 − → − → − → − → ∼ GL(Z5 × Z12 ) → GL(Z4 ) × GL(Z3 ) × GL(Z5 ). This is even more delicate for progressive abstraction as it happens for pitch and onset classes modulo some positive periods. third) has three dimensions over the rationals although it injects into the real line. n = 5.12 3 whereas there are only 60 = 34 220 three-element motive residues in OnP iM od5. This diﬀerence has a remarkable interpretation in terms of music theory and even music practice.2 more motives in OnP iM od12. formula (6. n = 5. This means that our cognitive eﬀort to identify motive classes is nearly double for 5/8 time signature compared to the other situations. i.1. in 5/8 time signature. In this case.44)) shares a more complex dimensionality.8. 4/4. And this has a dramatic impact on the musical consequences of classiﬁcation. and card(GL(Z5 × Z12 )) = 960 (see appendix C. ∼ The surprise vanishes if we look at the Sylow groups of these modules.eg in 2/4. then the residual motif M ⊂ OnP iM odm. n = 12. Onset|Z ⊕ P itch|Z) with integer coordinates. pitch is taken modulo octave. However. 12 This is due to the larger number of Sylow groups in the second module and therefore produces ≈ 1.4. coherent. For example. bar-exceeding improvisation in Desmond’s Take Five is much more diﬃcult than improvisation in a waltz such as Sherman’s .e. and restriction to rational scalars does the job in the mathematical sense.n (deﬁned in (6. The ﬁrst space has a 2.12 .3) (M. we recall it—. and third as being three independent musical directions. ﬁfth. n = 12. 4/4. this is ≈ 1. a 3-.

The systematic investigation of dimension as a degree of freedom in a musical system is undoubtedly a central issue for the description of global strategies in musical composition and cognition. more information concerning conjugacy classes of endomorphisms has been added. is accessible in appendix L. no systematic research on classiﬁcation of chords has ever been undertaken. but the fact that 5 and 12 are relatively prime. –Σ– We shall not discuss historic questions of this basic classiﬁcation issue here. The numbering ﬁrst reﬂects the cardinality. . together with additional determinants and class invariants. and so forth until the six-element classes.5 Chords Summary. So the increase in diﬃculty is not due to odd against even time signature. 5-periodic rhythms). and conjugacy classes of endomorphisms in 12-chromatic pitch class space Z12 . we concentrate on the description of the information which speciﬁes the isomorphism classes.. Structural/cognitive dimension would be a measure for the amount of structural/cognitive independence. diminished. 11. the empty class. These classes have a star exponent with their class number. seven.11. so-called chord dictionaries are published which are anything but complete. So when counting the isomorphism classes. Classiﬁcation of zero-addressed chords in P iM od12 . eleven. all others count double. whereas 3 and 12 are not.3. This does not only testify to a fragmentary practice but much more (as a foundation of practice) a fragmentary theory. exceptions to this rule: There are six-element classes which are isomorphic to their complements..2 on the American musical set theory tradition for more detailed remarks. Classes with more than six elements. But we should stress right here that. There are. starting with the special case of the full. Even in recent times [38]. however. major. and the zero-element. This gives us a total of 157 non-empty classes. then the one-element class. since they appear as complements of the explicitly listed classes. automorphism groups. those 18 classes with star count only once. The systematic problem behind these observations is to investigate a general understanding of what is structural “independence” among musical attributes and how this one could be related to cognitive “independence”. their symmetry groups.. The list has ﬁve columns: • The ﬁrst column denotes the class number. This list was ﬁrst discussed in a series of University lectures [327] in 1981. These observations suggest that cognitive dimension in music is rather related to Sylow groups than to the more obvious length of a ﬁnite ambient space (which would be 6 vs. In [400]. 12-element chord. European classical musicology and music theory has only dealt with some important chords (such as minor.3. but refer to section 11. then the two-element classes. are not listed explicitly. and published in [328]. As the chord class list can be calculated by a computer program. for example number 82*. besides this American tradition. and augmented triads) and has not dealt with classiﬁcation in the sense of determination of isomorphism classes under any non-trivial group beyond transposition. 3 in 12-periodic vs.5.. The complete list of isomorphism classes of zero-addressed chords in P iM od12 . ORBITS OF ELEMENTARY STRUCTURES 219 Chim Chim Cheree.

the star exponent in the class number The only exceptions are four pairs of classes which are not distinguished: • 18. and if we recall that all isomorphisms of local compositions in this ambient space are induced by aﬃne isomorphisms. Although. 68* • 75*. we have • • ◦ • ◦ ◦ ◦ ◦ ◦ ◦ ◦◦ < • ◦ • • ◦ ◦ ◦ ◦ ◦ ◦ ◦◦ According to this order. Conjugacy class of the symmetry group 3. In particular. and b is the corresponding number for the complement. i. the classes of the complements (circle representatives) are ordered according to the order of the bullet representatives. 82* . the ﬁrst representatives of an orbit are taken. Cardinality 2.e. the symmetry group is unique up to conjugation. • The third column shows the symmetry group of Sym(K) of the representative K. • The conjugation class of the symmetry group from column three is indicated in the fourth column. Pair of numbers a|b of numbers of conjugacy classes of endomorphisms 4.5.8. symbolized by numbers as explained in the list’s preamble.3. Autocomplementarity. and the numbering of the classes for given cardinality is the ordering among the ﬁrst representatives. . The complements are shown in circles. ymin 0 1 2 of the diﬀerences X − Y and Y − X instead of the maximal ones to deﬁne the ordering. • The second column shows the ﬁrst representatives as discussed in the previous item. • The ﬁfth column shows the pair a|b where a is the number of conjugacy classes of endomorphisms modulo automorphisms of the given local composition. This gives the lexicographic ordering of subsets of Z12 where a bullet is prior to a circle. 19 • 40. as drawn in column two of the list. 11. this group is not an invariant of the class.. 41 • 64*. Contemplation of this list shows that the following data nearly deﬁnes a complete set of invariants: 1. in general. and we take the minimal elements xmin .220 CHAPTER 11. . indicated by sets of bullets. as deﬁned in 8. if we stick to local compositions in Z12 . ORBITS The numbering for ﬁxed cardinality is as follows: We order chords according to the powerset order deﬁned in 6. For example. but with two (more historically motivated) diﬀerences: The order ¯ on Z12 is the clock order ¯ < ¯ < ¯ < .

so essentially. including enharmonic identiﬁcation. Our examples split into three lists. . these (22) classes are shown in ﬁgure 11. It is. however be addressed in more specialized and musically motivated chapters. The ﬁrst list (from [193..41). the pitch parameter was only relevant from the denotator perspective here. i.3. The following exposition—initiated by Thomas Noll in [402]—is a ﬁrst sketch of this project. we have simultaneously classiﬁed all zero-addressed local compositions which implement the same ambient space.2. but it already shows some remarkable results. Unfortunately. chapter IV]) refers to 85 examples which boil down to 23 chord classes. local compositions in 12-cyclic metrical onsets. to this day no systematic repertorization of chords in classical literature has been realized. The material in [402] was taken from Nikolai Garbusow’s work [193] on harmony. not a serious problem and this repertorization should absolutely be realized in order to know the empirical distribution of chord types. We shall see in chapter 41.e. We cannot discuss the richness of the chord classiﬁcation here. Except for class 78 which has a top number of 23 conjugacy classes of endomorphisms.3 that this project is also due to some conceptual diﬃculties concerning the very deﬁnition of what can be identiﬁed as a chord in a given score. We should instead make a ﬁnal remark on the musicological range of the above list.3. According to deﬁnition 14 in section 7. Evidently. we have classiﬁed 12-periodic local rhythms in the (reduced) onset domain. however.2.6 Empirical Harmonic Vocabularies –Σ– Summary. The diﬀerence of Grabusow’s chord notation to the original works he refers to is not relevant to our objective since we only look at pitch classes in P iM od12 . Mathematically. for example local compositions of form OnM od12 as deﬁned in formula (6. The systematic classiﬁcation is compared to some empirical material.11. 11. only the ﬁrst two couples are indistinguishable with the above data. ORBITS OF ELEMENTARY STRUCTURES 221 We shall learn in chapter 30 on counterpoint theory that the classes of the last two couples can also be separated by the span and diameter invariants. this will.3.

one of which is a 4-chord of class 29 with maximal cardinality of endomorphism conjugacy classes. The total range of the latter cardinalities is shown as a dark bar.4. and 7-chords and are shown in ﬁgure 11. .4. 6-. The second list refers to 28 examples from Alexander Scriabin’s sonatas 6 to 10.. have large numbers of endomorphisms. Conj. Classes Figure 11. symmetry group conjugacy classes (ordinate axis). ORBITS 1 2 6 8 15 2 10 8 16 4 15 6 14 3 chords 8 # Endo.5. Classes 1 2 4 6 8 4 42 47 50 56 59 62 8 12 16 20 58 5 chords # Endo.3: The 22 chord classes from Garbusow’s list from [193. Classes 1 2 3 4 6 8 13 17 2 25 17 22 28 30 29 32 36 37 4 6 8 10 35 26 4 chords 12 # Endo. all classes range among the top numbers. and cardinality of endomorphism conjugacy classes. i.e. We recognize that except for class 42. These examples boil down to 15 classes. Conj.222 CHAPTER 11. The other 14 classes are all 5-. chapter IV] in their distribution among cardinalities 3. Conj.

4 The third list refers to 45 examples from 20th century composers. 26 . 82.3.1: Examples from 20th Century Composers Debussy: “Ib´rie” e Debussy: “L’isle joyeuse” Ravel: “Ma M`re l’Oye” e Prokoﬁev: “Sarcasmes” Prokoﬁev: “Le sacre du printemps” Sch¨nberg: “Erwartung” o Casella: “Sonatine” 62. (2 ) 40 . 88 40 . Classes 7 chords 56^ 60^ 58^ 4 8 12 16 20 24 54^ 59^ 28 # Endo. 62 (8). Classes 6 chords 82 85 12 16 20 24 # Endo. 10 . 56. 62 . . Conj. 53 . 80 . together with the composers. 78. 12 .6.1 shows the class list. 78. 33 . 70 20 . 86 54 . Again. Classes 1 4 6 8 4 8 54 60 59 62 12 16 20 # Endo. 80 . table 11. 54 . Classes 1 3 8 4 73^ 8 Symmetry Group Conj. 17 . (1). many classes have large numbers of endomorphisms.4: The 14 chord classes of Garbusow’s list from Scriabin’s sonatas 6 to 10. 36 .11. Conj. 47 42 . Our third “statistics” of classes is shown in ﬁgures 11. 65 . Conj. 85. Classes 78 58 5 chords 223 Symmetry Group Conj. Table 11. 58. Classes 1 4 6 Figure 11. 54 .5 and 11. 39 . 45 . 62. 62. 10 . 26 . 54 . 85 . ORBITS OF ELEMENTARY STRUCTURES Symmetry Group Conj.

4. Classes 7 chords 47^ 42^ 53^ 45^ 62^ 4 8 12 16 20 24 28 # Endo. Classes 1 2 8 39^ 40^ Figure 11. We again have large numbers of endomorphisms. Classes 4 6 8 4 8 56 62 12 16 20 # Endo. 11.5: The 5-.6: The 8.1. 9 chord classes of Garbusow’s list from 20th century composers listed in table 11. we see that there is a strong tendency in this empirical repertoire to be placed where the numbers of conjugacy classes of endomorphisms for a given symmetry group within a determined cardinality is high. and 11. Conj. We again have large numbers of endomorphisms. Classes 58 CHAPTER 11. Conj.6. .5. Classes 1 3 13 14 4 20^ 17^ 36^ 33^ 8 12 16 20 24 28 32 # Endo. ORBITS 5 chords Symmetry Group Conj.1. Classes 1 6 Figure 11. 6-.3. Classes 54^ 82 79 78 85^ 6 chords Symmetry Group Conj. From the ﬁgures 11. Conj.224 Symmetry Group Conj. Conj. Classes 1 6 8 9 10 12 18 4 65^ 85 80^ 70 86 88 8 12 16 20 24 # Endo. Classes 9 chords 10^ 12^ 14^ 4 8 12 16 20 24 28 32 36 # Endo. 11. This observation is an empirical hint to the approach introduced in [402] which interprets endomorphisms as generalized tones. Classes 26^ 25^ 8 chords Symmetry Group Conj. Symmetry Group Conj. Conj. 7-chord classes of Garbusow’s list from 20th century composers listed in table 11.

11. we have learned that sets of aﬃne endomorphisms F = et · s : Z12 → Z12 (modulo certain automorphisms) are relevant to important chords in European tradition. and if 1 < card(K). the situation of Z12 -addressed chords in P iM od12 which will be treated in section 11. An objective local composition is called self-addressed if its elements are. If 1 = card(K).K → R. If M is indecomposable of ﬁnite length. address change makes sense.α). classiﬁcation of self-addressed compositions at R also boils down to the above orbit calculation since we are then essentially dealing with each local factor of R which is indecomposable and therefore 10 In ∼ [400]. but for endomorphisms this does not make sense. if we are given a ring R which is of ﬁnite length and which is a product of local rings. For local compositions. In this section we want to discuss classiﬁcation of these endomorphisms. But then. A functorial local composition is self-addressed (at M ) if it is a subfunctor of the product functor @M × F . The reﬂections from the previous section 11. since the linear part of the isomorphism on M must be nilpotent. given two A-addressed local compositions K ⊂ A@M. we know from Fitting’s lemma 77 in appendix E that F must ∼ also be an isomorphism.—So we want to discuss classiﬁcation of ﬁnite local compositions K ⊂ M @M .7 Self-addressed Chords –Σ– Summary.3.6 suggest that endomorphisms of ambient spaces could play a major role in harmony.3. ORBITS OF ELEMENTARY STRUCTURES 225 In our language this is. it can only be isomorphic to L under a non-invertible endomorphism if it is a constant. it can be shifted to the constant L by a translation isomorphism. local compositions and self-addressed denotators or local compositions. In fact.3.7. we are then left with − → the calculation of orbits under the left and right action of GL(M ) on M @M . As before. we conclude that F0 is an isomorphism. This idea was pursued in [400] and will be taken up in our part VI on harmony. More precisely.αf : R. Quite generally. Now. of course. let f /α : K → L be an isomorphism with underlying endomorphism F = em ·F0 of M . . Observe the conceptual diﬀerence between endomorphisms of ambient spaces. L ⊂ A@M . 11.(L.3. looking at the associated linear isomorphism R. clearly. such objects were called “fractal” because of evident self-similarity structure. This approach also suggests that harmony could be based to the diﬀerentiated richness of Z12 -addressed chords in P iM od12 which are canonically associated with zero-addressed chords. a denotator S : M F (s) is called self-addressed (at address M ). In order to restate this context we introduce a type of denotators in simple forms. we look for a classiﬁcation for self-addressed local compositions at the ﬁxed address M . We discuss classiﬁcation of self-addressed local compositions. Since the denominator α is an isomorphism by deﬁnition. in particular chapters 24 through 26. the self-addressed denotators10 : Deﬁnition 32 If F is a simple form with coordinator module M .

. . l are arbitrarily invertible. . . .t. τi = ti − t1 . ORBITS endomorphisms are induced by automorphisms. This is what is required when classifying selfaddressed local compositions at address Z12 . .k.(x. t1 . . u. . Again.(x.0) · Gv.k. s2 . . meaning (A. τn ) where σi = si − s1 . 1 < i ≤ n. s2 . 0. τ2 . since the elements u. . . y).k. . . u. also restricting to the sequences with n pairwise diﬀerent couples (si . . l) ∈ R × R× represent an automorphism eu · v and a base change ek · l of R. our action is also described11 by Gu = e(u. by deﬁnition of T 2 (R). y)t )t with transposed rows. . . ti ). or at address Z12 [ε] for counterpoint theory. . . . . . . .k. sn . tn ) and the (2n − 1)-tuples in the right module by (s2 .0) (Gv. . t2 .) via orbits under Sn (and as before.l). s2 . it has linear part v k Gv. ﬁx a positive natural number n and look at the short exact sequence 0 − − → R − −1→ R2n − − → R2n−1 − − → 0 −− −− −1 − −− of R-modules deﬁned as follows. tπ1 .. tn ). . 0. If we let the symmetric group Sn act linearly from the right on R2n by (s1 . l. . . . tn ) = (σ2 . ∆ d (11. σn . sπ2 . .t. their action eu · v · es · t · ek · l is represented by (u + v.t. .l is in the aﬃne triangular group T 2 (R). Let es · t ∈ R@R be a self-addressed point which we identify with the row vector (s. v) ∈ R × R× and (k..k. .k. t2 .3. Then the sequence is deﬁned by ∆1 (u) = (u. . t1 . t1 . v.t) v. sn .0) · Gv. for example. respectively.s + v.e. . tn ) · π = (sπ1 . 1 < i ≤ n. the diagonal 11 We make the common abuse of language and write a matrix multiplication A. . t) → e(u. sn .2. sπn .27) . We denote the 2n-tuples in the middle module by (s1 . If (u. .l → − → − where e(u.l (s. tπ2 . Taking up the methods of section 11. The group action is the following. . we get local compositions from sequences (s. . . sn . i.s + k. In the following discussion we ﬁrst concentrate on local rings R of ﬁnite length. tπn ) for π ∈ Sn . . t)) = (u + v. τ 1 = t1 . τ1 .l = 0 l in the group T2 (R) of upper triangular 2 × 2-matrices over R (with invertible diagonal coeﬃcients). t. .l : (s. t1 . k are arbitrary and v. . So. t2 . 0) and d1 (s1 .226 CHAPTER 11. . t) ∈ R2 . t2 . We are given local compositions K ⊂ R@R of ﬁnite cardinality n.

. v. l. this one has the same property. we have an induced action on the cokernel R2n−1 .3.. the symmetric group acts from the right and we are in a similar situation as for the ﬁnite classiﬁcation discussed in section 11. and the invariants from this algorithm are parametrized by S-rational points of quotient schemes from actions of ﬁnite subgroups of Sn on Grassmann schemes. τn ) with (σi .k.δ1 v. .k. let us suppose that R is selfinjective. (11. s2 .σ2 + k. . .τn ).l (σ2 . of Rn . is invariant under this action and therefore. we have Gv. When ﬁxing one of these spaces.τn . τn ) = (v. . . This again gives monogenous subspaces as invariants. and if GV ⊂ Sn is its isotropy group. 0). and we therefore have classiﬁed the self-addressed local compositions in ZN [ε] which will be used in counterpoint. . . only those generators (σ2 .30) correspond to local compositions.(s1 . We therefore have a ﬁrst invariant: the monogenous subspaces R. .τ. Summarizing. and denote the projection pr2 V onto the second to last coordinates. it induces a linear action of the upper triangular group T2 (R) on the cokernel R2n−1 . then there is an algorithm which classiﬁes self-addressed local compositions of cardinality n in R. . More precisely.τ1 . we have the induced action of v ∈ R× on the quotient space Rn−1 /pr2 V . . say.28) and this means the following: First. If we ﬁx a representative V from this classiﬁcation. t2 .e. . we are left with the action of this group on monogenous spaces W ⊂ Rn−1 /pr2 V . with δ1 = (1. σn . Let G = Gu . and recursion can go on until the algorithm stops. if we have a ﬁnite ring ZN . . 0.δ1 .). . tn ) leaves the diagonal image R. i. .σn + k. on the second half Rn of R2n−1 which. see chapter 29. we have these invariants of the left action of the triangular aﬃne group: 1 Un = {(V. In particular. . according to appendix F. τ2 . or an inﬁnitesimal extension R[ε] of a local. τi ) = (0.11.τ2 . the action of the invertible elements l ∈ R× generates orbits R× . is equivalent to considering monogenous subspaces R. . This may be summarized in Theorem 8 If R is a local. (11.τ. σn . V .29) (11. we ﬁrst have to classify the monogenous subspaces of Rn under the action of Sn .. selfinjective commutative ring R of ﬁnite length. Again. We have the diagonal left action of the aﬃne triangular → − group T 2 (R).τ. . lemma 87. where the occurring rings S = R/mr are quotients of the ground ring R.3. τj ). 1 < i < j (11.τ2 . . . Then G. W ⊂ Rn /pr2 V both monogenous}. Then we may proceed by literal copy of the recursive algorithm from 11.3. ORBITS OF ELEMENTARY STRUCTURES 227 image Im(∆1 ) = R. sn . τi ) = (σj . Moreover.l invariant and commutes with the action of the symmetric group. τ1 . l. 1 < i and (σi .3. the translational → − part of the action of T 2 (R) has orbits which include R. If Gv. τ1 .3.31) On this space.l ∈ T2 (R). . W )| V ⊂ Rn . ⊂ Rn .k. self-injective commutative ring of ﬁnite length and maximal ideal m. . τ2 . .δ1 -cosets . the contribution of the τ -component to the ﬁrst half is only a function of these monogenous spaces. . t1 . . Moreover. l. Of course.

36) (11. y) Dopp (1 3)(x. so it is useful to pay them a short moment of attention. 12 12 12 So we are left with the the calculation of X2 (Z12 )/S3 . as 12 12 a quotient of Z2 is also a submodule12 of Z2 . Let us ﬁrst describe the representation Dopp : S3 → GL2 (Z12 )opp : π → Dopp (π) = D(π −1 ) of the right group action in terms of 2 × 2-matrices over Z12 : Dopp (Id) = 1 0 0 1 Dopp (Id)(x. y) = (y. zero-addressed local compositions K in OnP iM od12.3 that we may concentrate on three-element 12 objects in ComLocgen .4. K ⊂ Z2 with card(K) = 3. −x − y) Dopp (2 3)(x.6. We shall also need this kind of list in sections 11.228 CHAPTER 11.34) (11.3.3. x) Dopp (1 2 3)(x.4. Conversely.0Z 12 object K has a subspace V ⊂ Z2 such that Z2 /V → Z12 .3. y) = (−x − y. The list which we shall calculate now is shown in appendix M.3. The following case of three-element motives is far from general classiﬁcation but it gives a complete description of the algorithms and. and therefore.2.33) (11. yields explicit lists of representatives of isomorphism classes. y) = (−x − y. x) Dopp (1 3 2)(x.3. y) = (y. K is already realized in Z2 .K.e. .35) −1 −1 1 0 0 1 −1 −1 (11. ORBITS 11.37) Let us now calculate the S3 -orbits of subspaces V of Z2 . But we know that 12 Z2 is injective. 11. y) Dopp (1 2)(x. i. y) = (x..2.12 .6. –Σ– The classiﬁcation techniques which we discussed in the previous sections will now be applied to describe explicitly classiﬁcation of small motives. mathematically speaking. y) = (x. see section C. and therefore we know from 11. −x − y) (11. We want to classify the three-element commutative. and we may start by the calculation of orbits of the larger space U2 (Z12 ) under a subgroup of S3 . Classiﬁcation of motives of small cardinality.32) ∼ Dopp (1 2) = Dopp (1 3) = Dopp (2 3) = Dopp (1 2 3)= Dopp (1 3 2)= −1 −1 0 1 1 0 −1 −1 0 1 1 0 (11. any such 3. the module Z12 . But then. 4 12 This follows from the main theorem on ﬁnitely generated abelian groups.8 Motives Summary.K. in contrast to the more abstract performance of enumeration theory which is discussed in 11. according to the theory which yields theorem 6.

2) Z2 4 Z4 × 2Z4 Isotropy Group S3 S3 D D S3 D opp opp opp ∼ semi-simple semi-simple (1. 0). we are in the semi-simple situation since the socle is invariant and we have to look for the orbits of subspaces in Z2 /2Z2 . But then. and the S3 -orbit of the one-dimensional space 4 V = Z4 . and we are done with the subspaces.3. Z3 .3. Summarizing. 0). Z3 .(1.V = 0. Z2 . Z3 .x → Z4 since otherwise. So we have V = 0 for 4 dim(V ) = 0. Suppose 2. It therefore is a Z2 -vector space. e2 = ((0. and e1 − e2 . . Exercise 15 Calculate the representatives of motives in that table from the spaces. According to the group action table above. 0). Z2 . Z3 .(2.(0. we have these subspace orbit representatives. including their isotropy groups which we need later: Representative 0 2Z2 4 Z4 (2. 2) Representatives 0. 2) We have the following orbits of subspaces of Z2 under the two isotropy groups of the above list: 3 Isotropy Group S3 Dopp (1. Recall that these kernel spaces Ker(f ) stem from canonical linear maps f : Z2 → Z12 . This gives us the list of kernel spaces in column three of table M. 2Z2 ⊂ V . V = 2Z2 for dim(V ) = 2. 1). we have to select those products V4 × V3 ⊂ Z2 × Z2 which do not contain 4 3 the vectors e1 = ((1. 0) Z4 (1.(1.3 along the lines of the previous example. 2) (1. Then the injective envelope W of V in Z2 must be either free of rank one 4 or rank two by proposition 90 in appendix E. So we have three orbits of 4 semi-simple submodules. (1. V would be semi-simple. 1) 3 0. one for x = (1. Exercise 14 Recapitulate the general algorithm described in 11. 4 In the rank one case. 1) 3 From this repertory. i. The zero space has already been recognized 4 4 above.(1. 1).K which is 12 associated with the motif K. W = Z2 .3. 2). 0). and we claim that in this case.e. Hence we have the two cases V = Z4 × 2Z4 and V = Z2 .(1. one for x = (1. V ⊂ 2Z2 . otherwise. 2) Dopp (1. In fact. we have V = Z4 . 0)). ORBITS OF ELEMENTARY STRUCTURES 229 V is semi-simple. Z3 .. (0. we have two orbits. 4 4 ∼ there is X ⊂ W with X → Z2 and X ∩ V = 0 and W cannot be the injective envelope of V . 0) Z4 (1. 1)). In the rank two case. 2) semi-simple injective injective injective (1. 0) is the set of all one-dimensional spaces in 2Z2 .11.

since the elements of Ki are listed in a determined order (ordered representation is automatic in computer programs). we can ﬁx two of the four tones to one of the ﬁve representatives of twoelement motives of our table. If this is the case. the volume is an invariant among all local compositions in the same ambient space Rn .K in R is an ideal V ol(K) of R. the class weight in column four. It also completed the redundant list in [328] which applied the present methods. the orbit Ker(fi ) · S3 contains Ker(fj ) and we can proceed to the next candidate. if every ideal in R is principal. but without computer support. The ideal V ol(K) is called the volume of K. The procedure in [513] ﬁrst uses classiﬁcation of two-element motives to build a reasonably small set of motives which includes all isomorphism classes. 50055}. decides whether its class has already been found and then adds it to the the found class representatives or else throws it away and picks the next element. Then the image of the n-fold exterior power R. Therefore: Proposition 7 With the above notation. we also call this generator the volume of K.1. This work made use of the methods introduced in the previous sections and conﬁrmed the list of Egmont K¨hler [266] which was communicated without any o speciﬁcation of the used methods. Ki is a new representative. L ⊂ Rn are related by an invertible symmetry F . and the volume in column ﬁve. we concentrate on those representatives for the following comparison: For each four-element motif Ki ∈ List4 . The program picks each of the list’s elements in this order. and this leaves the ideal invariant. In particular. and if R is a product of local rings of ﬁnite length. The other two tones are free. If this data is not yet present in the list of class representatives. and if R is of ﬁnite length. the program calculates a set G(i) of generators of the kernel Ker(fi ) of the linear map fi : Z2 → Z2 which is canonically associated 12 12 with Ki in our theory. which is listed in table M. according to sorite 6 in section 8.3.12 by use of a software written in C. By the preselection . This number is included in the last column of table M. . otherwise. To decide upon class membership. the volume of a local composition is deﬁned as follows: Deﬁnition 33 Let K ⊂ Rn be a commutative local composition in the free R-module of rank n n n n. If the ring R is of ﬁnite length. . The elements of this list are ordered by an index: List4 = {Ki | i = 1.5. and we obtain a total list List4 of 5 · 142 · 141/2 = 50055 motives to be investigated. if volume and class weight is already present. In [513].3.230 CHAPTER 11. the generator of V ol(K) is unique up to invertible factors in R. Then the program calculates the orbit of G(i) under the action of the symmetric group S3 and checks whether an element of this orbit is included in the kernel Ker(fj ) of a motif Kj ∈ List4 which has already been selected as a representative and which has the same volume and class weight as Ki .5. . Whereas the class weight will be deﬁned and discussed later in section 14. This symmetry’s linear n n ∼ part yields an automorphism of the nth exterior power R →R R. Else. the program ﬁrst calculates the volume and class weight for a selected motif Ki ∈ List4 . two isomorphic local compositions K ⊂ Rn . ORBITS Table M. By abuse of language. Hans Straub has classiﬁed of four-element motives in OnP iM od12. From this classiﬁcation.3 also contains two more invariants. we have found a new representative and proceed to the next element Ki+1 .

Bach/Sarabande Nr.4. However. These Melodies were stratiﬁed into motives built from three.) 19 14 15 23 We see that there are major diﬀerences in the latter data.and fourelement groups of successive notes. the second number is not very reliable since the orbit cardinalities should be included in these ﬁgures. Johann Sebastian Bach/Sarabande Nr. John Lord13 /Sarabande.6. A. i. We review Harald Fripertinger’s work on this subject. Bach/Gigue Nr. 32 J.1-8. The classiﬁcation of these motives shows rather strong commonalities for three-element motif classes: Composer/Piece J. In [513]. it yields a reﬁned melody classiﬁcation method which can be used for stylistic classiﬁcation. –Σ– 13 John Lord is the organist of the rock group “Deep Purple”. 52 J.) 10 10 10 10 The pieces diﬀer signiﬁcantly for the reﬁned four-element motif stratiﬁcation: Composer/Piece J. This program also settles the question of ﬁnding the class of any given motif and therefore can and has been used for experimental purposes. Bach/Gigue Nr. 11.. This type of analysis should be investigated systematically and with statistical skill. A.11. Bach/Sarabande Nr.and four-element motif classiﬁcation was used to compare melodies.2. Lord/Sarabande W. 52 J. ENUMERATION THEORY 231 via volume and class weight. Lord/Sarabande W. The selection includes: Johann Sebastian Bach/Gigue Nr. We shall perform a detailed and poetological melodic analysis built on three-element motives in section 11. 32.S. Wolfgang Amadeus Mozart/KV 449/T. as it is pointed out in [513].4 Enumeration Theory Summary. Without doubt.S. . the orbit calculations happen less frequently and the program terminates quickly.e. Mozart/KV 449 Number of Occurring Classes 52 66 75 65 Most Frequent Class (Nr. 52. three. Enumeration theory is a quantitative approach to classiﬁcation of local compositions via actions of permutation groups. Mozart/KV 449 Number of Occurring Classes 13 12 15 14 Most Frequent Class (Nr. frequency should be divided by orbit cardinality.S. 32 J.S.

One looks o at powersets F P of ﬁnite sets F. 171]. motives. scholars often deal with enumeration of local and global compositions without clear conceptual distinction. In musical enumeration theory. knowing the class numbers is a strong prerequisite to apply probabilistic and statistical reasonment for classiﬁcation of large music repertories. These sets represent particular species of objects. appendix C.38) another approach to enumeration theory of local compositions in ﬁnite abelian groups. We gave a small account to this direction in the ﬁnal words of section 11. counting orbits is much less than listing them. a good part of the enumeration work automatically yields or eases explicit lists of representatives.1 P´lya and de Bruijn Theory o Summary.4. g) → f · g as above for P = Zn .3. To count the G-orbits in 2Zn . The groups consist of selected systems of musically interesting transformations. twelve-tone rows and the like. 170.. g) → χK · g. So the set of zero-addressed local compositions K ⊂ Zn is identiﬁed with the powerset 2Zn . Let us shortly digress on the general P´lya context to understand the ideas. If a − → group G ⊂ GL(Zn ) of aﬃne automorphisms acts on the set of zero-addressed local compositions K ⊂ Zn . see [204]. –Σ– This section mainly refers to Harald Fripertinger’s work [168. · 1036 isomorphism classes of 72-element motives in Z2 ? We shall deal with the more philosophical 12 aspect and the question concerning big science in musicology in section 11.232 CHAPTER 11. lemma 63.2. such as transposition or inversion. . (11. and second.e. we introduce weight functions w : 2 → R with values in a ring R which express special sets of local compositions. F = 2.8.1. Why is there any interest to know such numbers—for example the incredible number of 2 230 741 522 540 743 033 415 296 821 609 381 912 = 2. 169.2.4. We only deal with local enumeration theory here and postpone global enumeration to section 16. . 1}. 11. we can anticipate that ﬁrst. these structures are chords. A P´lya weight function is a o function w : F → R with values in a ring R which contains the ﬁeld of rational numbers. this action is reﬂected on the powerset via the obvious right action (χK . Here. Evidently. . P and canonical right group actions of permutation groups G ⊂ SP on F P via (f.23 . the set K is identiﬁed with its characteristic function15 χK : Zn → 2 with the identiﬁcation 2 = {0. It rests on combinatorics of ﬁnite group actions on ﬁnite sets14 . i. P´lya and de Bruijn theory yields a crucial tools for enumeration of classes of local o compositions in the case of ﬁnite abelian groups as ambient modules. or even describing them as rational points of schemes or similar classifying spaces of geometric nature. ORBITS Enumeration theory deals with counting of types of structures. orbits of ﬁnite group actions on ﬁnite sets. In musical enumeration theory. partitions of pitch class sets. This induces a product weight function Πw : F P → R : f → p∈P 14 For 15 See w(f (p)). In order to apply the P´lya theory to the o enumeration of isomorphism classes of zero-addressed local compositions K ⊂ Zn .

F. . P.6 for the explicit formulas. It gives a formal account of the cycles16 of the elements of G in P . Whence the product weight Πw (χK ) = xcard(K) . we can take the ring R = Q[x] of rational polynomials over the indeterminate x and the weight 1 if i = 0. . . y∈F w(y)2 . (11.41) and we see that this polynomial in Q[x] has the number of G-orbits of local. The P´lya theory oﬀers tools to calculate the conﬁguration counting series. The relation between the conﬁguration counting series and the cycle index is set up by the main theorems of P´lya’s enumeration theory: o Theorem 9 With the above notation. we have C(G.11. . 2). 2. (11. F. The central o object of the theory is the cycle index. .3. we have the orbit number card(2Zn /G) = Z(G)(2. 1) = card(F P /G) is the number of orbits and this evaluates to card(F P /G) = Z(G)(card(F ). . (11. (11.40) x if i = 1. zero-addressed k-element compositions K ⊂ Zn as its xk -coeﬃcient in the above case P = Zn . P. .45) appendix C. deﬁnition 127 for the concepts of cycles and the cycle type cyc(g) of a permutation g.4. w) = Z(G)( y∈F w(y). 16 See (11.39) For our above example. the cycle index is known. ENUMERATION THEORY 233 Clearly.42) g∈G For several common permutation groups.44) − → In particular. . if we have the above action of a group G ⊂ GL(Zn ) on the set of zero-addressed local compositions K ⊂ Zn .cyc(g) . if the weight w = 1 is the constant weight with value 1. F = 2. y∈F w(y)d ). see appendix C. card(F )). w) of this weight is deﬁned by C(G. F. the cycle index of the permutation group G is the rational polynomial Z(G) = card(G)−1 X. C(G. F. P. a rational polynomial which depends on the subgroup G ⊂ SP . . the product weight is invariant on a G-orbit and we may deﬁne the product weight on the orbit space via Πw : F P /G → R : f · G → Πw (f ). card(F ). w(i) = (11. The conﬁguration counting series C(G. .3. Then: Deﬁnition 34 With the above notation. w) = ω∈F P /G Πw (ω) (11.43) Corollary 6 With the above notation. P.6.

k odd. . 17 See (11. This is a straightforward application of cycle index formulas for these groups. x1 − xt ) n n with the evident kernel Ker(δ) = ∆Zn . j|gcd(n.k) j|gcd(n. The interspace sequence of K is deﬁned to be the image δ(K) of the sequence (K) under the linear endomorphism δ : Zt → Zt : (x1 .48) appendix. (11. for example in short exact sequence (11. 19 Terminology of Fripertinger [170]. kt ) for the ordered sequence (also called the sequence of K) of the t = card(K) elements of K in the canonical linear order 0 < 1 < 2 < .1.12). and where [t] denotes t’s integer part18 . .e. the number of G-orbits of zero-addressed local compositions in Zn is calculated from the coeﬃcient of the k-th power xk in theorem 9 and from the formulas for Z(Tn ) and for Z(T In ). C. For a nonempty local composition K ⊂ Zn .47) (n−2)/2 [k/2] where ϕ is Euler’s ϕ function17 . For the translation group Tn = eZn . . xt ) → (x2 − x1 . k even.k) ϕ(j) ϕ(j) n/j k/j n/j k/j +n +n n/2 k/2 ) ) n. . (11. . . We have Proposition 8 [170] The number of Tn -orbits of zero-addressed k-element local compositions in Zn is 1 n/j ϕ(j) .3.k) ϕ(j) k/j + n [k/2] 2n ( 1 2n ( 1 j|gcd(n.. ORBITS These results yield formulas for counting the number of classes of zero-addressed local compositions with respect to isomorphism types which are common in chord classiﬁcation. − → − → For the special case n = 12. . k2 . . .1.k) and the number of T In -orbits of zero-addressed k-element local compositions in Zn is n/j (n−1)/2 1( 2n ) n odd.1 0 1 1 1 1 1 1 1 2 6 6 5 3 19 12 9 4 43 29 21 5 66 38 25 6 80 50 34 7 66 38 25 8 43 29 21 9 19 12 9 10 6 6 5 11 1 1 1 12 1 1 1 Enumeration of Interspace Structures For special groups of aﬃne automorphisms of Zn .4.46) n k/j j|gcd(n.4. 18 The . . xt − xt−1 . and for the group of translations and inversions T In = Zn e −1 . . the cycle index Z(GL(Z12 )) and theorem 9 yield the GL(Z12 )orbits [170]. write (K) = (k1 .234 CHAPTER 11. largest integer ≤ t. the isomorphism classes. they realize a special instance associated with our above construction of diﬀerence structures. n − 1 on the canonical representatives of Zn . interspace structures19 of zero-addressed local compositions in Zn are useful. and we obtain this orbit number list: k= T12 T I12 − → GL(Z12 ) 11. n even. i.

yn ] : g → x g(i) yg(i) . Conversely. their orbits under the group T In are equal iﬀ [K] = [L]. yn ] : g → xn yg(i) (11.n → Q[x. Therefore. the number of interspace structures of t-element local compositions in Zn is given via the xn -coeﬃcient of the cycle index evaluation n−1 n−1 n−1 2 x2i yi . xt ) → (xt . Z(T It )( i=1 xi yi . . . . then (K) − ϕ(L) ∈ Ker(δ) = ∆Zn . yn ] : i → xi yi (11. . y2 . . and by the preceding case. y1 . ψ → T It .n ⊂ [1. Then the above weight function is invariant under the action ∼ of the above dihedral group ϕ. lemma 15 yields the following technique for enumerating classes of local compositions which have special interspace conﬁgurations. . . (11. . . n − 1] of positive integers. it = n. so K and L have same T In -orbit.51) . y1 . and since the set ISt. . Together with exercise 16. and we conclude that K is a translation of L. the orbits of such local compositions are identiﬁed with T In -orbits of sequences g in ISt. L ⊂ Zn are non-empty local compositions. y2 . Since we can identify the set of interspace sequences of t-element local compositions in Zn with ISt. consider the weight function w : [1. if K. .38. . i2 . . x1 . L are such that δ(K) = ϕ · δ(L). . −xt−1 . the T In -orbits of K and −L coincide. . Proof. We next consider three linear automorphisms of Zt : n ϕ : Zt → Zt : (x1 . xt ) → (xt . y2 . The interspace structure [K] of K is deﬁned ∼ as the orbit of δ(K) under the dihedral group ϕ. . y2 . xt−1 ) n n ∼ ψ : Zt → Zt : (x1 . it ) such that i1 + i2 + . . . .n . We then have Theorem 10 With the above notation. x1 ) n n ∼ ρ : Zt → Zt : (x1 .n under the same action. . with the formalism introduced in formula 11. .4. . We then have Lemma 15 [170] If K. we have the product weight Πw : F t → Q[x. Clearly. . . . n − 1]t of those sequences of positive integer t-tuples (i1 . . . yn are indeterminates.n is characterized by the sum of coordinates being n. . the translation e1 (K) and the inversion −K have the same interspace structure as K. . n − 1] → Q[x. we have a weight function W : T It \ISt. If δ(K) = ψ · δ(L). . . y1 . ENUMERATION THEORY 235 Exercise 16 [170] Consider the subset ISt. y1 .50) where the indices of the y give information about the intervals between successive elements in these local compositions. − x1 ) n n and observe that δ · ϕ = ϕ · δ and ϕ · δ · ρ = ψ · δ. xt ) → (−xt . xt−1 . If x. . . ψ → T It .11.n → Zt is a n bijection onto the set of interspace sequences of t-element local compositions in Zn . . the above equation gives δ(K) = ψ · δ(L) = ϕ · δ · ρ(L). i=1 i=1 t xti yi ).49) ∼ which is deﬁned on the interval F = [1. . Then the canonical map ISt. . . QED. .

(11. the coeﬃcient of x12 is 2 2 y1 y10 + y1 (y2 y9 + y3 y8 + y4 y7 + y5 y6 ) + y2 y8 2 2 3 + y2 (y3 y7 + y4 y6 + y5 ) + y3 y6 + y3 y4 y5 + y4 and we can calculate.. for example. Satz 2. those interspace structures with all intervals being at least k units by setting y1 = y2 = . . the group Dn. . Example 16 For example. Example 17 For example.5]. 2 n!! = n(n − 2)(n − 4) · .4.1 that a n-phonic series is a denotator Ser : Zn−1 P iM odn (Ser0 .1.12 -orbits of a dodecaphonic series is 9 985 920.53) A proof can be found in [168].n \SERn is as follows. . . . the number of orbits in Dn. We omit the details here and just reproduce the particularly interesting list of orbits for n = 12: . yk−1 = 0 and yi = 1 else. Sern−1 ) with pairwise distinct coordinates. More generally. Enumeration theory yields this number of orbits: Proposition 9 For 3 ≤ n. For k = 2 we get seven such structures. 1 Then the numbers are 1 ((n − 1)! + (n − 2)!!( n + 1)) 4 2 1 ((n − 1)! + (n − 1)!!) 4 if n ≡ 0 else.2 Enumeration of Series Recall from deﬁnition 22 in section 8. . So there is still some uncovered material for dodecaphonic compositions. As we know from the discussion of n-phonic series following deﬁnition 22. the Bruijn extension of P´lya’s enumeration theory yields formulas for oro bits of (k. mod 2.n acts on the set SERn of n-phonic series. (11. Set n(n − 2)(n − 4) · . . the number of D12. ORBITS i.52) if n ≡ 0 else.e.2. . the coeﬃcients of the monomials in the indeterminates yi of the xn -coeﬃcient are the numbers of interspace structures of given interval distribution. if we look for interspace structures of local 3-element compositions in Z12 . . mod 2. n)-series [170. 11.1.236 CHAPTER 11.

n) − − → −− D SERMn D SER(n−1. and for any aﬃne automorphism ∼ g : Zn → Zn .1. we have D · g = g0 · D.n on SERMn .54) deﬁned by D(Ser)i = Seri+1 − Seri . we invoke the context of deﬁnition 22 in section 8.e.n) − − → SERM(n−1. Consider the left action of the group Dn. Then. 0 < i ≤ k. 0 ≤ i < k. Call I(Ser) the integrated serial motif. . This projection has a right inverse injection I : SERMk. D · I = IdSERMk. D · revn = −revn−1 · D. Ser ∈ SERMk.n (11.n) −− of sets of serial motives with all horizontal arrows being inclusions.1. Seri−1 .n . the set of all-interval n-phonic series is the intersection set ALLSERn in the cartesian diagram ALLSERn −−→ −− SERn D−1 SER(n−1.11. call D(Ser) the derived serial motif.n . By deﬁnition. i. (11. For natural numbers 0 < k ≤ n. We have a projection homomorphism D : SERMk+1.n → SERMk+1. we have the set SERMk. (Ser + Ser )i = Seri + Seri for Ser.56) .n (11.. n)-serial motives which is an additive abelian group by pointwise addition.n of (k.55) deﬁned by I(Ser)0 = 0 and I(Ser)i = Ser0 + .4. and the reversion operation rev. . SERn is invariant.e.n → SERMk. i. ENUMERATION THEORY k 2 3 4 5 6 7 8 9 10 11 12 number of orbits of (k.. 12)-series 6 30 275 000 060 280 880 680 440 160 920 237 1 4 9 (dodecaphonic series) 9 2 14 83 416 663 993 980 985 To deal with all-interval series.

. . Therefore. . .2 Big Science for Big Numbers Summary. these properties of all-interval series.2] The number of orbits of zero-addressed local compositions K in OnP iM odn. we deal with purely enumerative aspects for local compositions K in OnP iM odn. musically indicated subgroups H ⊆ Gn . .3 Enumeration of Motives We have discussed the classiﬁcation techniques for zero-addressed local compositions (K. Complexity and depth in music revisited.6. F = 2.3.5. . In [170].m . 12 the entire formula is written down in appendix C.3. Theorem 11 [170. 11. 1 + xnm ). .5] card(G12 \ALLSER12 ) = 267. . − → For the special case n = m = 12 and H = GL(Z2 ). G = Zn ⊕ Zm . The reader is referred to this work for more details. 11. –Σ– .3. We are again in the situation exposed in theorem 9. in particular. we have Proposition 10 [170.n and even under the larger group where we take all aﬃne pitch automorphisms instead of transpositions and inversions only. We write down the ﬁrst few monomials of the cycle index polynomial in theorem 11: − → Z(GL(Z2 )(1 + x.3. . for n = 12. It shows that. the number of 72-element classes is 2 230 741 522 540 743 033 415 296 821 609 381 912 = 2.4. so there are still some motives to be introduced into musical composition!—In [170].23 . the enumeration of some other dodecaphonic structures is calculated. and the weight is again given by formula (11. 1 + x2 .n) is also invariant under Dn. .12 in section 11.3. Signiﬁcation of the exorbitant growth of numbers of classes of local compositions for computational complexity in musical analysis. however. Zn ) in N section 11.n and under the larger group Gn generated by the reversion and all aﬃne pitch automorphisms. and more concretely of local compositions K in OnP iM od12. · 1036 . 1 + x2 . as well as speciﬁc weight function (as considered above) are used to calculate the size of the orbit spaces of all-interval series under the actions of diﬀerent.m under a group H of aﬃne automorphisms of Zn ⊕ Zm is the coeﬃcient of xk in the cycle index polynomial Z(H)(1 + x. the evaluation has been calculated by use 12 of a Turbo Pascal program in [170]. Here.238 CHAPTER 11.1.40). ORBITS Therefore D−1 SER(n−1. Satz 2. In particular. 1 + x144 ) = x144 + x143 + 5x142 + 26x141 + . .8. So the set ALLSERn of all-interval series is invariant under Dn. as discussed in [170]. Satz 2.4. this time.

Big Science is recognized by the following characteristics: • Scientiﬁc language • Models and theorems • Experimental paradigms and operationalization • Universal collaboration and communication • Adequate laboratories and machines • Political acceptance and corresponding resources Why does the exorbitant number of isomorphism classes of local compositions which we have exposed in the previous sections suggest the research context of Big Science? Let us look at the above characteristics and their realization in centers such as CERN. Nowadays.7. 20 CERN 21 LEP = Centre Europ´en pour la Recherche Nucl´aire.7: (©cern) The 27 km long LEP ring on the CERN site. e e = Large Electron Positron Collider. With a total of roughly 10 000 employees. The ring is situated 100 m underground.11. see ﬁgure 11. viewed from above. ENUMERATION THEORY 239 Big Science is the type of science created by physicists since the Los Alamos nuclear bomb project during world war II. . an electricity consumption equivalent to that of the city of Geneva and a ﬁnancial volume of about one billion CHF [83] in 1995. among others a circular LEP21 . it is best illustrated by the research environment of Europe’s CERN20 near Geneva.4. Figure 11.

Further. Finally. Whatever the technological apparatus which becomes necessary to control this complex data.g. See also chapter 20 for the visualization of musical data. For example. or motives. etc. One also disposes of good operationalization techniques. And one could then compare this data to the models of harmony etc. see the annual ICMC22 meetings. The third item is also attainable. precisely because of class numbers such as the above number 2. in order to verify/falsify corresponding statements. or also the relevant parts and chapters in this book (e. and navigation tools. as we shall see in section 41. a classical sonata with its note size of order 104 . of local meters and so on within a determined material selection. · 1036 of 72-element motive classes. harmonic. of serial motives. it has become evident that depth now is no more a question of empty or censored encapsulation and knowledge hiding. parts X. ORBITS Evidently. And these must be set up in technologically sophisticated and collaborative environments.3. We would nonetheless like to stress that music is of comparable signiﬁcance to humans. the ﬁrst two items are within the reach of mathematical music theorists. rhythmic. see ﬁgure 11. at least in principle. . or 3D-caves. .6 and 11.23 . We are given an enormous repertoire of musical works and know how to do experiments on this material. . we cannot understand mental event systems having the quality of high-ranked music without adequate analysis. What is less evolved is universal collaboration.8. it is absolutely necessary to have auditory representations of whatever music or analysis data exists in order to control the complex material.240 CHAPTER 11. What is “adequate” in this case? There are several requirements to be fulﬁlled.8. and this justiﬁes much of such an eﬀort which is not likely to produce destruction devices such as nuclear bombs. 22 ICMC = International Computer Music Conference. However. must be based on powerful computers. To begin with. as well as political acceptance and corresponding resources. Second. We have already given small examples of such investigations in 11. Depth can now completely be recovered in its explicable form of a complex universe which is overwhelming but not disclosed from understanding. the calculation of harmonic paths is proportional to 2number of onsets of a composition . the negative extremum is the serious lack of adequate laboratories and machines.3. any serious processing of relevant musical material. for example 3D-tables. navigating on data bases of relevant musical works must be based on powerful visual and auditive representation paradigms. All this combines to require a huge apparatus for calculation and representation..3. and this is practically inﬁnity. and motivic analysis of such class data requires enormous combinatoric calculations. XIII). The situation is much the same as with the early days of anatomy where opening a human body was prohibited by authorities who feared that detailed and rational access to the subject would annihilate the mumbo-jumbo of Aristotelian and Christian mystiﬁcation. For example. representation. say. It is not clear and we would not hope that a musicological CERN would be as expensive and monumental as the physicists’ display. Now. can this be compared to the CERN equipment which deals with experimental necessities deduced from high energy conditions required to unveil hidden symmetries of suspected laws of external nature? But we have already discussed this topic in chapter 4: Experiments with inner nature require high spiritual energies in a very rational sense. one could calculate the distribution of isomorphism classes of chords.

this space is a reduction to the paradigmatically essential information. however. chords.11. 11. –Σ– The wide-spread presence of groups of symmetries in music has already been discussed in chapters 8 and 10. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 241 Figure 11.8: The 3D table [282] as a necessary virtual reality tool for visual navigation in music data bases. rhythms? De facto. once the paradigm and a concrete object have been related either by compositional or analytical eﬀorts.5. as the previous classiﬁcation discourse .5 Group-theoretical Methods in Composition and Theory Summary. Knowing that paradigmatic classiﬁcation is a fundamental poetic device that must be taken into account in the analytical perspective. too. what is the speciﬁc power of group orbits of musical structures such as motives. we want to relate the quantitative and qualitative aspect of orbit spaces to historically traced approaches in composition and analysis. In analysis. we have abstracted from contingencies which blur the core information. So what help is given by moving within a space of prototypes? First of all. refer to further symmetry-related harmonic analysis because this aspect will also be treated in due detail in part VI. and contemporary serialists. We do. This section reviews the group theoretic methods in composition from Bach to Sch¨nberg. But then. we are thrown into a space of “prototypes” as orbits may be called in a more intuitive language. approaches from Graeser to Forte and the o American set theory school are reviewed. Here. in many cases.

So here. combinatorics gives a feeling of divine power: Everything is reachable. This is one essential aspect: Disposing of an alphabet of structural ingredients reduces much of the work to a task of combination or recombination. In other words. In the tradition of medieval artistic poetics (such as Fran¸ois Villon’s poesy c [312]) or in the baroque combinatorial rhetoric (such as Athanasius Kircher’s classiﬁcatory work [477]) such lists of reasonable size are germs of combinatorial approaches to universal classiﬁcation or artistic construction.7 after a . the space speciﬁcs are not or only in part taken into account.. the resulting list is often ﬁnite and—for some classical cases such as chord classes— of manageable size. On the one hand. yields a deeper principle of uniﬁcation of the instance of these classes in a given composition.and diachronic development of the musical sign system. So we observe that each determined combinatorial setup on the basis of a classiﬁcation arsenal provokes the antithesis of progressive negation of the classiﬁcation’s basics. and this is one aspect which indeed is historically prominent as we shall see. i. For example.. Vocabularies tend to englobe small portions of space-time in musical cultures. history unfolds and lets us recognize a coupled syn. the roughly 1036 classes of 72element motives in Z2 . More precisely. We should nonetheless recognize that lists which turn out to be inﬁnite or so large that even the best computers cannot search them in reasonable time (e. it may also happen that either in composition or in analysis. this same “total control” not unexpectedly excites one’s boredom with a once-for-all limited alphabet of creativity.4. the analysis of historical chord repertories such as discussed in section 11. and not only in an insigniﬁcant list array. the overall situation is a massive shift of creative quality. the speciﬁc nature of this space lends itself to an enrichment of semantic depth. We shall come back to this overall phenomenon in section 11. To begin with. This fact means that the “game’s rules” may de facto vary without creating much ado. but the potential must be observed. Whatever the size of combinatorial arsenals. the negative aspect of a well-deﬁned classiﬁcatory arsenal is the limitation of the “game’s rules” in a rigid framework.3. Let us ﬁrst concentrate on the positive side: complete reachability of classes in an orbit space. see section 11.3) give a slightly diﬀerent approach to combinatorics: 12 We have to introduce statistical procedures since systematic combination would fail.g. Second. On one hand.1. ORBITS has shown.7 used the size of automorphism groups of classes as a measure for historical distinction between important and marginal chord classes. the enrichment of having placed classes in a coherent space. on the other.242 CHAPTER 11. of power and creative expansion.e. the whole universe at your ﬁngertips. synthetical and analytical creativity are projected onto a ﬁeld of combinatorial activity. the spoken “parole” in Saussure’s dichotomic setting. We know from music semiotics [361] that a signiﬁcant difference between the music sign system and the linguistic system is the weak conventional aspect termed “langue” vs. So there is no automatic integration of classifying space attributes. In other words. at least is this a fact without a priori negative consequences. the classes being points in a particular space type—such as a geometric parameter space for the complete set of class invariants—may be related to each other qua points and therefore induce an additional instrument of poetic construction or analytical investigation. In this dialectic dynamics of combinatorics and its negation. aleatoric components cannot be avoided. On the other hand. You can play God with the molecular combinatorics of chemical substances from a small number of roughly a hundred atom types. this is a plain added value. the paradigma as such.

a new one for each composition.5. retrograde. Only after the dispersion of harmony in the ﬁrst years of the 20th century was the counterpoint group rediscoverd by Sch¨nberg and his followers. In contrast to the baroque usage of such orbits. Moreover.5. So the logic of orbits forced a thorough inves23 Of course are there other characteristics than this of dodecaphonic composition technique. a procedure to build vocabularies. the orbit of themes under inversion. Moreover. inversion and their combinations in verse and rhyme construction. this time. The harmonic level was to be developed into a dominant syntactical force which evolved until its dispersion at the end of the 19th century for reasons we shall discuss in chapter 23.1. and genuine serialism.. 243 11. the artistic tradition of medieval poesy was highly sensitive towards construction of explicit correspondences between poetic instances: Recall Fran¸ois Villon’s poesy [312] which abounds c with retrograde. On the one hand. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY closer look at representative examples. But the contrapuntal tradition had prepared a generic procedure of relating voices by a “switch” between counterparts. i. So the really new message in Sch¨nberg’s attempt was o the exclusive basing of local compositional coherence upon the orbit of a given dodecaphonic series. So the progress against fugue tradition and the harmonic vocabulary was to make available a proper alphabet for each composition instead of a ﬁxed alphabet being perpetuated along a large number of compositions.1. On the other hand. Extending transformation groups as a developmental parameter: Counterpoint. the rhetoric scheme of a dialog between dux and comes is not speciﬁc enough to deﬁne a thematic vocabulary. Sch¨nberg and serialists in composition in o section 8. In this spirit of combinatorial artistry. We shall see in chapter 29 that this idea can be made completely precise in modeling Fux’ contrapuntal rule system. The common denominator of the dodecaphonic vocabularies was exactly this meta-vocabulary. . the corpus of such compositions was classiﬁed by a fundamental invariant: the orbit of the governing dodecaphonic series23 .12 · Ser of series Ser (see section 8. –Σ– We have already seen symmetries with Bach. This led to an apparantly secondary role of the contrapuntal dialog of voices during the harmonic homophony. dodecaphonic composition technique.1).1. transposition and their combinations was a perfect completion of the said empty rhetoric scheme.e. this was however not the only principle of compositional construction. punctus contra punctum meaning that two opposites are confronted. it only gives empty boxes to be ﬁlled with musical structure. Here.11.1 Aspects of Serialism Summary. the construction of themes was not completely explicit in the sense that the interrelation between thematic/motivic and harmonic/contrapuntal constraints had not been made evident as a composed phenomenon. the orbit deﬁned the a priori alphabet of local compositions to be distributed over the syntagm of the composition. This rediscovery followed Sch¨nberg’s o o declared program of rebuilding the compositional paradigm ab ovo and without any implicit interrelations to harmonic constraints. They are situated in the global conﬁguration of the alphabet deﬁned by the orbit D12. In the baroque fugue technique as completed by Bach. We shall come back this subject in chapter 13. we reﬂect their roles with regard to paradigmatic classiﬁcation via group orbits.

244 CHAPTER 11. It takes values in a parameter space form. Ser11 ) at the address Z11 which essentially deﬁnes a sequence or “row” of pitch values. 0 serial index row 11 Figure 11. form P iM od12 . In other words. . . In view of this canonical generalizatio. . loudness. for example. amounts to looking at its projection onto the corresponding form and to making the requirement explicit on this projected parameter space. .. and even failure. No systematic treatment of admitted series has ever been published. the deﬁciency has occurred on two levels: Object description and control of transformations together with the attached orbit spaces. for each serial index 0. then a generic series would be a Z11 -addressed denotator with values in a more general form such as Duration ⊕ P itch ⊕ Loudness ⊕ . from such a concept. 502. Then a dodecaphonic series is by no means a generic concept: There is no deeper reason for disclosing other musical parameters. More precisely. Asking such a series to fulﬁll the requirement of unique representation of a pitch. for example.9: A generic series as viewed by serialists can be interpreted as a discrete curve.n which is immediate from the universal denotator language. .. we notice a permanent struggle with the fundamental . i. . in the tubular form Duration ⊕ P itchm . onset. .9). The ﬁrst one could be observed when composers were dealing with the question of realizing a given series on the level of a concrete composition. 11.e. Under such aspects was it no surprise that Sch¨nberg even patented one of his series. Ser1 . or a variant with (partial) modular division Durationn ⊕ P itchm ⊕ Loudness ⊕ . . o We know that a dodecaphonic series is a denotator Ser : Z11 P iM od12 (Ser0 . 138. such as all-interval series and other ‘special’ types. 550]). whatever. . here. ORBITS tigation of classes of dodecaphonic series. The formal power of composers and musicologists (in serialism often in personal union) was evidently too poor for grasping the complexity of a systematic serialism. when reviewing the serial compositions and their interpretations and analyses (see for example [94. such as duration. the historical realization and handling of serial perspectives has been a poor attempt. We could also say that a series is a “discrete curve” with values in a determined parameter space (see ﬁgure 11. if a 12-series is a curve with values in a pitch domain. So. .

5 for the cardinality of general aﬃne groups over ﬁnite ﬁelds. treating it according to one’s inherited capacities.5. such as Fibonacci numbers or the mumbo jumbo of magic squares. in the line of Riemann. but important24 motif (ﬁgure 11.3. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 245 discipline of what parameter should be included. we shall discuss more extensively the computer program presto developed for this type of paradigmatic composition based on structural units and their innumerable transforms (in fact presto allows more than 126 billion aﬃne transformations in each parameter plane). 24 See 25 See the motivic analysis in [328].2.10) from Beethoven’s Hammerklavier sonata op 106. This motif is a local composition of form OnM od71 ⊕P iM od71 ⊕LoudM od71 ⊕DurM od71 and is transformed according to the general aﬃne group of Z4 . We have already mentioned this perspective in section 11. So there is no evidence for mental control of which transformations of series should be allowed or how they should be formalized. In this spirit of positive redeﬁnition of a crisis. The impression is that of a chaotic usage of undigested form and concept fragments from combinatorial and group theory. not to mention the question of classiﬁcation via group orbits. We come back to this subject in the introduction of global compositions in chapter 13. the situation is that of constructing a new musical instrument and having nobody who is capable of playing it in a controlled way.0445 · 1037 . say.1 by Jan Beran [49] which is the broad instantiation of transformations of a short. serialism has even been deﬁned as a precursor of post-serialism [502]. Sch¨nberg’s dodecaphonism was in a strong sense this kind of reaction to fragmentary and o deﬁcient harmony. comparable to the birth of modern score notation. And we should also mention the impressive Piano concert No. the very complexity of explicit control and manipulation of large ensembles of orbits and transformation groups requires more than just conceptual control. . respectively. no human memory can manage this amount of a priori elements in a single orbit. It seems that we are participating in a painful process of conceptualization. It should however be emphasized that there is no moral or artistic ground for refuting such a development. This chaotic approach is further alimented by an old and strong numerological tradition. related to others. Though adaptation of classiﬁcation techniques from mathematics as described above solves some of these deﬁciencies. that a fragmentary theory and blurred practice had led to its abandon in the name of (a true or desperate) creative innovation. This orbit deﬁnes the a priori local alphabet of Beran’s piano concert. A typical application of the computer-controlled paradigmatic composition will be discussed in chapter 50. In chapter 49. Clearly.4. Intuitively. So the chaos of concepts intertwines with the chaos of transformations and results in a wild—euphemistically termed “creative”— conglomerate of operations and listings. It rather turns out to be a typical eﬀect in the history of music: Already it had happened with classical harmony. but instead. The second deﬁciency is a consequence of the ﬁrst one: Lack of adequate conceptualization makes control of transformations virtually impossible. The computer is precisely the tool we are asking for. appendix C. History of music is also a history of failures to control results of creative extension. ordered.11. and then realized in a concrete composition or its analysis. a group of cardinality25 71 10 445 260 466 832 483 579 436 191 905 936 640 000 or 1. it requires auxiliary tools to operationalize mental handling.

) The vectors a and b were chosen such that repetitions of the same pattern occurred only after a very long period. l are positive integers in a certain range. bars 75 and 76.5. The starting points were deﬁned by (onset. not necessarily in the sequence given here: Step A: Selection of the basic theme M : The basic theme consists of the upper voice in Beethoven’s Hammerklavier sonata op. 11. see ﬁgure 11. 106. These consisted mainly of combinations of slight rotation.246 CHAPTER 11.1. See chapter 49 for technical details of the presto functionality and sound representation.10. ORBITS Figure 11. The note events were put as zero-addressed denotators in general (geometric) position in the fourdimensional space P = Onset ⊕ P itch71 ⊕ Loudness ⊕ Duration.1 Serial Strategies in Jan Beran’s Piano Concerto Nr. 1: an important motif from Beethoven’s Hammerklavier sonata op.10: The fundamental “series” in Beran’s Piano Concert Nr. . The structural principles were only used as a tool that helped ‘translate’ intuition into a large coherent composition. 106. The technical tool for the eﬀective realization of the composition was the composition application presto .) Step B: Basic pattern in P : A basic pattern (ornament) was created by copying theme M to a large number of starting points in the two-dimensional projection Po = Onset ⊕ P itchM od71 of P . Step C: Most frequent transformations: A set T of basic transformations (matrices) was deﬁned. stretching and shearing. pitch) = ja + lb where a and b are two (nonparallel) vectors in Po and j. (Recall that here all values of pitch are understood modulo 71. It should be noted that the equally fundamental ‘principle’ used for the composition was ‘musical intuition’.1 In the following. The construction consisted of the following steps that were partially carried out jointly and repeatedly. the main structural principles used for the piano concert are described brieﬂy. (The embedding of presto ’s half-tone pitch modulo 71 into a MIDI pitch space is discussed in chapter 49. and thus create a ‘skeleton’ that was a starting point for the actual musical composition. by assigning diﬀerent values of Loudness and Duration to diﬀerent notes.

since this is where the actual creative compositional work took place. p. Both of these works are divided in two formal parts. Our account will be split into a genealogical discussion (11. Step E: Instrumentation: Instruments were assigned (and created) to polygonal areas of the composition such that musically interesting structures in P are emphasized. [35]). . and David Lewin ([300. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 247 Step D: Application of transformations: The pattern created in Step B was varied by local application of transformations.5. –Σ– 11.1) and a contribution to a vocabulary switch (11.2) between the American set theory (for short: AST) and our present concept environment. The purpose of this section is essentially to account for and discuss some crucial contributions of the American tradition to the emergence and proliferation of what Babbitt termed ‘professional music theory’ [31].50]). The former work points to “the engagement by composers in fundamental music-theoretical explications” ([59. Of particular interest were transformations of projections that excluded either onset time or pitch. The transformations were applied to all two-dimensional projections of P . We may also suggest here that this distinction is not only relevant for an historical discussion on pitch-class set theory. from the important “Perspectives on Contemporary Music Theory” (1972. more or less explicitly.2. 301]). most of them being from set T . [59]) to the recent “Music Theory in Concept and Practice” (1997. meta-theory and methodology/compositional theory and historical and theoretical essays/analytical studies. This distinction appears. Transforming the pattern led to new musical motives that were used subsequently in the composition.2.11. respectively. Step F: Tempo curve: A detailed tempo curve was deﬁned. in many studies on the subject. Steps E and F were done subsequently together with a second application of Step D. Most of the eﬀort was spent with steps D through F. The sequence of the steps was as follows: Steps A through D were done using a piano score only. but that it also helps in understanding how this theory successively enlarged its ﬁeld of applicability thanks to important works by John Rahn [430].2 The American Tradition Summary.vii]).5. Perhaps one of the most fruitful approaches is based on the underlying dichotomy between an apparently more compositional attitude (Milton Babbitt) and a radically analytical perspective (Allen Forte) towards music theory.5. 11.1 Genealogy The impossibility of giving even a partial (ordered) description of the topics dealt with in the American music-theoretical literature since the 1950s leads us to look for historically and methodologically pertinent ‘segmentations’ in the domain of contemporary music theory. Robert Morris [380]. p.5.5.2. while the latter suggests that “the very fact that Forte is not himself a composer has changed the ﬁeld of theory considerably” ([35.

p.248 CHAPTER 11. Webern.ix]).xiv]) and suggested that the relevance and “the force of any ‘musical system’ was not as universal constraints for all music but as alternative theoretical constructs. . But even the idea of applying the mathematical concept of group for modeling musical systems can be regarded as one of Babbitt’s most fruitful intuitions26 .8). from Lewin’s GIS structure to Clough and Agmon’s modern theory of diatonicism. 507]. a “set” is an ordered collection of pitch classes and it is used as a perfect synonym for row and series. Partition problems connected with Babbitt’s original idea have also largely proved their relevance to mathematics with their natural embedding into the theory of groups [300.ix]). 178. p. In contrast.. A synonym for it is the term “collection”. springing from a truly fervent desire to 26 However anticipated in 1924 by Wolfgang Graeser in his study on Bach’s “Kunst der Fuge” [194]. and whose combinatorial characteristics are independent of the ordering imposed on this content” ([26. p. closed systems. 27. p. and Berg in the ﬁrst 20 years o of the 20th century [159]. 28. Suggestions for further reading in this area may be found in Rahn’s review of Lewin’s “Generalized Musical Intervals and Transformation” ([431]). Reﬂecting on the problem of “an adequately reconstructed terminology” ([30. ORBITS As pointed out by George Perle in his comprehensive study on serial and atonal music. inversional symmetry and complementation. for Babbitt. This ‘Babbittian’ presentation of twelve-tone problematics constitutes perhaps the most appropriate introduction to the composer and theorist Babbitt. experience. invariance under transformation. rooted in a commonality of shared empirical principles and assumptions validated by tradition. p.x]). has important compositional consequences which are “directly derivable from the theorems of ﬁnite group theory” ([28]. 29]. It is widely accepted that he ﬁrst provided “twelve-tone theory with a consistent technical vocabulary” ([414. as Babbitt’s four crucial articles make clear [26. This equivalence of structures. and experiment” ([59. particularly combinatoriality and partitioning [319.. p. 380]. The most important representative of the analytical approach in the American musictheoretical literature is Allen Forte who is the author of a theory of set complexes [157] and of a book primarily devoted to the atonal music of Sch¨nberg. the very predecessor of Allen Forte’s “pitch-class set” is Babbitt’s “source set”. provided that “the rules of formation and transformation of the twelve-tone system are interpretable as deﬁning a group element (a permutation of order of set numbers) and a group operation (composition of permutations)” ([30. properties of adjacency as compositional determinants (. “the most important inﬂuence of Sch¨nberg’s method is not the 12-note idea in itself. Forte’s main purpose is to “provide a general theoretical framework. repr. ﬁrst introduced in [27]. ﬁrst introduced by Lewin in [295] and widely discussed for its analytical pertinence in [296]. One cannot emphasize enough that.10]) is a necessary preliminary step for the description of structural characteristics of the twelvetone system. with reference to which the processes underlying atonal music may be systematically described” ([159. p. a set “considered only in terms of the content of its hexachords. aggregate construction. but along o with it the individual concept of permutation. 235.) and [31]. 45. 506. It is perhaps no exaggeration to see the introduction of groups by Graeser and then Babbitt as the ‘Copernican Revolution’ of modern music theory. Forte’s starting-point is “ﬁrmly analytical. especially if one considers the proliferation of group-theoretical methods applied to music. a vast body of American literature was devoted to the study of the speciﬁc properties of sets and collections. Subsequently.)” ([414.57n]).20]). Detailed discussions of Babbitt’s terminology also feature in ([26]. p.

p. and pitch-structure based on groups of operations acting on pitch-classes” ([380]. His very ﬁrst article concerning generalized approaches in describing intervallic collections of notes [295].. Lewin’s GIS model is algebraically equivalent to the structure of principal homogeneous G-sets ([553. Babbitt’s “general formative role”. which is almost in the spirit of Babbitt’s compositional attitude. Forte particularly emphasized this aspect. p. Bernard.. p.11. like the interval function.) enigmatic repertoire” ([56. The following section is dedicated to a detailed discussion of many of the topics dealt with by Morris by means of the theory of local compositions.44]). By taking G to be the additive group of a vector space. However.. p. aggregate completion.330]). Lewin’s most important abstract construction is probably the GIS (Generalized Interval System). Most of the areas developed by Lewin since the sixties ﬁnd their natural place in his fundamental treatise [300].5. like Forte’s interval vector or Regener’s common-note function [297]. stating that “The structure of atonal music (is) above all the study of a musical repertoire rather than a theoretical presentation” ([160. his work convinced many of the interest of a formal study of chromatic space” ([369. In his most recent article.271]). His work could also be regarded as one of the clearest attempts to go beyond the music-theoretical dichotomy of compositional vs.. p.50]). The study of such a structure leads to a natural generalization of Forte’s Set Theory. Lewin later studied the speciﬁc differences between his own “interval function” and other similar theoretical constructions. and isographies.xiii). a more elaborated version of his previous “Formal Interval System” or FIS [299].83]). together with Forte’s terminological heritage are widely recognized in the US-American sphere and the book has become a standard reference for further discussions in set theory. As deﬁned by Morris. aﬃne spaces) that are amply discussed in the present book. The book also prepares “its reader for the professional literature in the ﬁeld” ([430. An example 27 See also our discussion of global compositions in chapter 13. one is formally lead to the same family of objects (i. p. o As pointed out by Jonathan W. p. networks.v]) and gives accurate references for further specialized topics including advanced serialism and combinatoriality. A crucial point here is its intimate connection with Lewin’s transformational approach. analytical procedures. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 249 uncover the secrets of an (. makes it “the conceptual predecessor of the interval-class vector (. where the natural generalization of this and other global concepts in AST are treated.e. “compositional spaces are out-of-time structures from which the more speciﬁc and temporally oriented compositional design can be composed” ([381. An instance of this is Rahn’s pedagogically oriented introduction to some problematics common to the atonal and serial repertoire [430]. the fact that Lewin’s function deals with the interval content of a single collection of notes. to Sch¨nberg’s hexachordal pieces [296].) in Forte’s work” ([56. Like Babbitt and Forte. p. . p. Robert Morris for example developed some “relations between ordering. a topic27 recently developed by Robert Morris [382].49]) together with the elaboration of the “set complex”. also by means of a transformational model that includes transformational graphs. was soon followed by an analytical application of some formal constructions. Morris’ recent formalization of his so-called “compositional spaces” provides a new theoretical tool which compensates for some of the weaknesses of his original compositional model. one of his most striking merits was the introduction of a “consistent terminology for pitchclass collections based on the mathematical properties of the set” ([56..90]). Lewin’s contribution to the ﬁeld of music theory is both terminological and methodological. But it is probably true that “Allen Forte’s real success lies in the developments he inspired: beyond his theorization of atonal music.

?].42). We shall comment on the structure and performance of American Music Set Theory (the abbreviation AST also in honor of its characteristic abbreviation overhead) in section 11. 30].5. In fact.2.2 Concepts and Theory—A Vocabulary Switch This section is devoted to an embedding of the Musical Set Theory constructs of local structures into the theory of local compositions with their forms and denotators. 90].2. As suggested by Eytan Agmon in his recent comprehensive study in the subject [10]. Transformational networks and “their pertinence for music theory and analysis” are largely discussed in Lewin’s second book [301]. This explicitly includes the plain integer pitch module for n = 0. whereas n = 0 is called the linear case. . The global aspects will be dealt with in section 16. 42/2. Morris’ presentation is quite generic.2. including comments on the semantic background of AST constructs. 89. i. we keep track with Robert Morris’ lucid book [380]. A special issue of JMT28 conﬁrms the amount of interesting mathematical and musicological problems concerned with diatonic theory and the so-called Neo-Riemannian Theory. This would be in perfect tune with Babbitt and Lewin’s main concepts.4. The basic space form is n-modular pitch. The study of combinatorial properties of generalized well-tempered musical systems [178] leads in a natural way to a formalization of musical structures by means of group-theoretical methods [36] and the elementary theory of numbers [88. a synonymy to P itch|Z . The scheme of our proposed embedding is a translation of the AST concepts into forms and denotators. the second small example of 9. A ﬁrst large class of AST concepts is built around pitch spaces..6. dedicated to Babbitt and Forte. for example. explicit and also recognized among the AST theorists. To be consistent with conceptual cross relations. and if they are related by modular identiﬁcation with 0 < n. they may be traced back in the AST literature. see [82. by paraphrasing John Rahn. the modern theory of diatonicism originated in some problems ﬁrst raised by Babbitt [29. have transformed in the USA the ﬁeld of music theory into a modern mathematical study. Both. for the topic leads to important developments in the ﬁeld of mathematical music theory. We conclude this brief genealogy of the American music-theoretical tradition with the mention of a few noteworthy works on modern diatonic theory. But a vocabulary switch of the basics is the indispensable minimum for intercultural communication.250 CHAPTER 11. For further applications to Lewin’s transformational approach to the problem of voice leading in atonal music see [515]. Vol. ORBITS of the way in which Lewin’s theoretical constructions could be described in terms of forms and denotators was considered in section 6.e. this always works via the common projection modn : Z0 = Z → Zn and its associated natural transformation and form morphism modn : P iM od0 → P iM odn 28 Spring 1998.5.. 9. the form P iM odn = P iM odn −→ Simple(Zn ) Id deﬁned in formula (6. i. We do not claim coverage of theorems and deﬁnitions. Morris calls the cases n = 0 the cyclic cases. 11. If form spaces are deduced from these basic spaces. for other references.e. but we refrain from this identiﬁcation since we do not want to connotate too much of the context of P itch|Z .

. n − 1]. but only comparative pitch.. COM (R · Cont) = COM (Cont)ρ . 29 In the AST. evaluate to pitch numbers between 0 and n−1. Clearly. We do not inherit this unmathematical terminology.k (X) upon the comparison matrix is this: COM (I · Cont) = −COM (Cont). A contour for c-spacen (X) is deﬁned as a serial motif which takes values in c-spacen (X). i.11. i.e. Contour spaces just parametrize pitch in some generic context where precise pitch is not relevant. We denote the set of these contours by CON Tn. . . Serk−1 ). Contk−1 ) whose coordinates are cps. the index of a contour is an abstract substitute for onset time. A c-pitch (= cp. a denotator Cont : Zk−1 c-spacen (X)(Cont0 . The eﬀect of the K4 -action on CON Tn. i. n)serial motives or synonymous objects. As with dodecaphonism. i. COM (Cont)τ = −COM (Cont). The number k is the length l(Cont) of Cont.e. We shall deal with the general approach to motif theory in chapter 22. 0)-serial motives Ser : Zk−1 P iM od0 (Ser0 .. AST deals with (k. . consider (k. The musical background of such a space is encoded in the name “X”. Very often. C-Spaces. and this is numbered by integers from 0 to n − 1. . In this sense a contour formalizes an abstract motif structure. The AST group operations of contours are those in the Kleinian subgroup K4 of Dk.e. here. For a positive natural number n. In particular.57) so that the form name is characterized by the generic “c-space” head and the speciﬁcation of order and name X. its cardinality is nk .k (X).. the comparison matrix COM (Cont) associated with a contour Cont of length k. we just describe the AST approach. So the order does only aﬀect the possible denotators. plural: cps) of X a zero-addressed denotator cp : 0 c-spacen (X)(x) with coordinate pitch numbers 0 ≤ x < n. a contour space or c-space X of order n has form c-spacen (X) −→ Syn(P iM od0 ). GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 251 which will be assumed to be clear in the present AST context.5. A zero-addressed local composition in ambient space c-spacen (X) (whose elements are c-pitches!) is called a cpset29 . COM (Cont) is the integer k × k-matrix with coeﬃcients COM (Cont)i. . A pcset is always of cardinality at most n.0 (see the discussion of dodecaphonism following deﬁnition 22) generated by the address retrograde R = revk and the inversion I = en−1 · −1 which ﬂips the interval [0.j = sign(Contj−1 − Conti−1 ). The topological classiﬁcation paradigm for contours is deﬁned by an “abstract motif” (see chapter 22). and sequences are called ordered sets. we shall explicitly include serial motives with integer pitch values. . Id (11.e. COM is skew-symmetric. sets are called unordered sets. the material nature of the cps is hidden in this name. .

For positive natural numbers n. For a more precise pitch information. . the case M = 2. a p-pitch (or shorter: a pitch) of [M 1/n ] is a zero-addressed denotator x : 0 [M 1/n ](x) with coordinate pitch numbers x that stands for a physical frequency F0 M x/n . . M . the co-diagonal transposition κ. y} which is also written as an equivalence class of matrices according to the permutations of the pset elements. We have a commutative diagram CON Tn. In other words. ORBITS the 180-degree rotation ρ of the matrix around its center. P-Spaces. Segk−1 ) is called pitch segment and abbreviated by pseg.k (X) − − → K4 \Zk×k −− and can deﬁne the AST segment classes or contour classes as ﬁbers of the composed map.58) K4 \CON Tn. The ordered p-space interval ip < a.k (X) K \ 4 −−→ −− COM Zk×k K 4\ (11. the iterated interval succession is denoted byIN Tm (Ser) = Dm (Ser). the COM -map is K4 -equivariant for the above actions of K4 . AST uses the concept of p-space. .54). . Pitch cycles are the same as pitch segments. b} = |ip < a. see below.59) where the form name is characterized by the nth root of M which deﬁnes a physical connotation of the pitch numbers. b > is the integer a − b. A zero-addressed ﬁnite local composition in ambient space [M 1/n ] is called (unordered) pitch set and abbreviated by pset. Id (11. The interval content of a pset P is the function int : P × P → Z : (x. this is a typical situation in motif theory (see chapter 22). The interval succession of a pitch segment Seg is precisely its derivation IN T (Ser) = D(Ser) deﬁned in (11. More precisely. a pitch space or p-space [M 1/n ] has form [M 1/n ] −→ Syn(P iM od0 ). n = 12 being the 12-tempered tuning starting from F0 . the unordered p-space interval is deﬁned by ip{a.252 CHAPTER 11. the AST only distinguishes them by diﬀerent admitted group operations. a serial motif Seg : Zk−1 [M 1/n ](Seg0 . y) → ip{x. we denote the space of length k pitch segments in [M 1/n ] by SEGk [M 1/n ]. We have COM (IR · Cont) = COM (R · Cont)τ = COM (Cont)ρτ = COM (Cont)κ . b > |.

y} is the smallest nonnegative representative in Zn of the diﬀerences i < x. B)0 . The set of all pc is called the aggregate and denoted by Un or just U for the usual n = 12. rt (Ser)i = Seri+t . . i < y. B)11 ] with IV (A. y > is ic < k >”31 . B) = [ICV (A. B) = [IV (A. An ordered pc interval between pc x and pc y is i < x. the set {(x. . Rotation rt of a pitch segment Ser is the serial motif rt (Ser) deﬁned by cyclic permutation address change i → i + t mod k. The pitch class spaces in AST are thought of as being a modular derivative from pitch spaces. − → The group GL(Z2 ) is denoted by T T O. . The aﬃne transformations Tn Mm = en · (m) on the pitch coordinator Z deﬁne the set of canonical operators. B)6 ] with ICV (A. i. ICV (A. .. Ser0 ) of Ser. “i < x. Pc-Spaces. Zn @Zn . Z@Z. y >= y − x. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 253 The cyclic interval succession CIN T (Ser) is IN T (C(Ser)). An (unordered) pitch class set pcset is a local composition in ambient space pc-spacen . x >. as already with the p-spaces. . y) ∈ U 2 | i < x. Notation is again Tm and Mm for translation and multiplication. 12 multiplication M5 is notated by M . . y >. . and its 48 elements are called TTO operators. B)0 . We know from (11. . and the standard interpretation of passing from the p-space [21/12 ] to a 12-modular pc space makes it useless to introduce a more reﬁned terminology than just “the” n-pc space. i. In p-space.11. AST considers translation Tn = en and multiplication Mn = (n) of pitch.e. pc-spacen −→ Syn(P iM odn ) Id for positive natural number n. B the interval class content vector is the 7tuple in square brackets ICV (A. . whereas general inversions are of shape Tn I. ICV (A. Operations on pc-spacen are.. B)j = 1 card(A × B ∩ ic{j}) for 0 < j < 7. but useful. Serk−1 . The interval vector is the 2 12-tuple in square brackets IV (A. A pitch class (abbreviated pc) is a zero-addressed denotator x : 0 pc-spacen (x) in pc-spacen . ic < k > for the set {(x. B)j = 30 This 31 The is not AST terminology. AST writes ic{k} resp. Operations and TTOs (TTO = Twelve Tone Operators). y >= k} and says “i{x. the aﬃne transformations on the ambient space.5. AST ﬁxes n = 12. Given two pcsets A.. i. y} = k} resp. commutes with multiplications and has IN T (R(Ser)) = −R(IN T (Ser)).54) that IN T is invariant under translations.. y) ∈ U 2 | i{x. i. the unordered pc interval i{x. B)0 = card(A ∩ B). A set class SC is an orbit T T O · A of a pcset A. and I = (−1) represents inversion. IV (A. . mostly used for n = 12. diﬀerence between unordered and ordered interval notation is not AST. Retrogression R on pitch segments is deﬁned as with any serial motif. A pitch class segment pcseg is a serial motif with coordinates in pc-spacen .e. CIN Tm is just the iteration CIN T m .e. y} is ic{k}” resp.e. the interval succession of the “cyclic extension”30 C(Ser) = (Ser0 .

5. The cyclic interval succession CIN T (S) of a pcseg S is the derived serial motif IN Tm (C(S)) of its cyclic extension as deﬁned above for psegs. −n) and M U L(X. The G-prime form of a pcset is the ﬁrst representative of its G-orbit in the lexicographic order deﬁned in section 11. The Tn /I set class of a pcset is its orbit under T I12 . The “complement theorem” is the formula M U L(A . A) is denoted ICV (A). sym7 (A. Exercise 17 Prove that the last assertion is equivalent to Morris’ deﬁnition “EM B(A. A ) > which is an invariant of the T T O-orbit of a pcset. B)). sym11 (A. This subject will be dealt with in chapter 22. B . sym7 (A. The number of (categorical) subobjects deﬁned by a TTO f : A → B is the “embedding number”33 EM B(A. as above. B) is invariant under translations and exchanges its arguments under Inversion. 32 This 33 Terminology is not AST terminology. sym5 (A. also called the multiplicity of position j and denoted by M ul(A. B) = card(Symi (A. not explicit in AST. n) = M U L(A. The special case ICV (A. Y.e. Retrograde R and rotations and pitch operations from group T T O or the monoid of aﬃne operations are deﬁned mutatis mutandis as above for psegs. the mth interval succession of a pcseg S is the mth derived serial motif IN Tm (S) = Dm (S). For a given subgroup G of T T O. B). the interval class vector of a pcset A is invariant under inversions and translations. B) + 12 − (card(A) + card(B)) which is a trivial consequence of the facts M U L(X. j). Pcset Similarity. . A). sym1 (A. B) = {x = et · (i) ∈ T T O| x(A) ⊂ B} and symi (A. X. A ). A). ORBITS card(A × B ∩ ic < j >). n) = M U L(X. SC X is the abstract complement of SC Y iﬀ they are the orbits of complementary pcsets. B) = number of pcsets in the SC of A which are contained in a pcset member of the SC of B”. X. The interval vector IV (A. Pc Segments. As with psegs.254 CHAPTER 11. CIN Tm (S) is the mth iteration of CIN T (S). B. sym11 (A. Y. n) + M U L(X .. A). A pcset A is called invariant under a TTO K if it is a symmetry of this local composition. A). T T O contains a morphism f : A → B of local compositions. Then the invariance vector of A is the 8-tuple sym(A) =<sym1 (A.3. n) = card(Y ). A ). sym5 (A. Abstract inclusion of pcset A in pcset B means that there is a member in the orbit (SC) T T O · A being included in B. The order of Sym(A) is called the degree of symmetry of A. i. A ). Clearly. Deﬁne32 Symi (A.

More recent applications are.. The justiﬁcation for considering TTOs and not any general permutations is 12 precisely the aﬃne character ([380. and cycle representations of the T T O as a permutation group on aggregate U12 . i. and such that this is the lexicographically ﬁrst among these maximal interval right-hand pcs.3 Software for Musical Set Theory The thoroughly combinatorial and algorithmic character of AST has led to several implementations of its objects and operations on the level of computer applications. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 255 Given two pcsegs X. A segment class SGC of a pcseg S of length k is the orbit Dk. These constructs are used of compositional design. for example.e. self-addressed local compositions in pc-space12 (subsets of Z12 @pc-space12 ) consisting of TTOs are considered. the abstract conﬁgurations of pcs. the ﬁrst ones being already described by Allen Forte [158] in 1970.e. . p. start the sequence with the right pc with such a 12 maximal interval. SET-SLAVE computes several “normal forms” of zero-addressed local compositions c of pitch classes from AST: • Rahn’s normal form N F : This is a standard sequential presentation of c. and operators to build background structures for musical compositions. sets of operations. 11. OpStrgk−1 ).3 for more details on global aspects in AST. the formula for the inverse of a TTO. Take the smallest pc with these properties. TTOs. the CMAP package developed by Craig Harris together with Alexander Brinkman [208] (see also [82] for further references) and updated by Peter Castine for Macintosh in C++ [82]. . together with strings of operations.125]).e. the serial motives OpStrg : Zk−1 ˜ Z12 @pc-space12 (OpStrg0 . Finally. Chapter four of Morris’ book [380] is entirely devoted to the study of the group T T O = − → GL(Z2 ). appendix III]).. pcsets.5. i.5.11 Another implementation in LISP has been presented by John Amuedo [15].11. AST calls X (literally) included in Y iﬀ there is an order preserving address change monomorphism ι : Zl−1 Zk−1 : es → eι(s) such that X = Y · ι. His class hierarchy is shown in ﬁgure 11. The conjugacy classes of subgroups of T T O are discussed and listed ([380.. .2. See our discussion in section 16.—What a mess! . So abstract inclusion is a fractional morphism f /ι of denotators (viewed as singleton local compositions). pcsegs. including MIDI input.) 34 Take the largest interval of successive pcs on the circle Z . Y of lengths l = length(X) ≤ k = length(Y ). using a not very systematic cyclic permutation34 of the ordered sequence of c in [0. The package includes a computing environment SETSLAVE for general AST calculations in Common LISP for Macintosh computers.12 · S (see the discussion of n-phonic series following deﬁnition 22). see the technique of address killing deﬁned in section 11. Abstract inclusion means combining address-change monomorphisms with T T O-induced morphisms f on pitch space.2. The T T O description includes elementary facts about ﬁnite groups. 11]. Orbits generated by dodecaphonic series are called row-classes. i. (Taking the lexicographically ﬁrst sequence would be the better solution.

See chapter 25. . 3. i3 .2 for these relations). where the “bass note” is codiﬁed by the “rotation index” x for the power of the cyclic permutation (“rotation”) (1. • Straus’ zero normal form ZN F . represented by e−t N F. 2. (2) a BabbitObj class describing the basic objects of AST. . This package contains a CHORD-CLASSIFIER. visual inspection of scales.256 CHAPTER 11. • Forte’s interval class vector IV C. i6 with ij = card({ unordered intervals (x. using the lexicographically ﬁrst orbit representatives. i.. and its decimal representation DN F .x from ZN F . it also computes Forte’s “K” and “Kh” inclusion relations (see discussion in section 16. • Forte’s prime form P F : This is a standard notation of T I12 -orbits of zero-addressed local compositions of pitch classes. Recently a library speciﬁcally devoted to American Set Theory has been integrated in Ircam’s visual programming language OpenMusic based on CommonLisp/CLOS (see chapter 51 . Like CMAP. The SET-SLAVE is combined with three interactive MAX programs. i5 . y} = j} • Amuedo’s decimal normal rotation DN R. Amuedo also includes the evident binary codiﬁcation (BN F ) of this normal form via the characteristic function on Z12 . Amuedo’s thesis also includes software for real-time harmonic analysis. ORBITS Set Class Database SetClass BabbittObj Object pcSet OCTest BabbittTest InvarianceTest pcRow Figure 11. Castine’s principal subclasses are (1) a SetClassDatabase class managing database requirements.2 for more details. y) in c with i{x. a SCALE-FINDER.11: Built on the object-oriented Think Class Library.e. and a SCALE-MONITOR for analysis of chords. t = ﬁrst pc of N F . stated in terms of inversion of pitches (successive octave increase of a pitch) that yields a new bass note. This is the reﬁned ZN F. the sequence of cardinalities i1 . . This is just the transposition-orbits.—This is nothing but the distinction of a particular pitch. i4 . and (3) a BabbitTest class which gives a generic framework for operations to be performed on objects. and analysis resp. card(c)) of the ZN F sequence bringing the “bass note” pc to the ﬁrst position of the sequence. i2 .

Nonetheless.5.3 for the presentation of Vuza’s model of rhythm) and that will be continued with the study of the aﬃne group and of the symmetric group (in the pitch as well as in the rhythmic domain). However. with the possibility of taking into account any division of the octave in a given number n of equal parts. for example.. duration. the work of Dan Tudor Vuza or Anatol Vieru does heavily recommend such an extension. many valid lists of isomorphism classes. A speciﬁc function calculates the numbers of equivalence classes of k-chords (up to transposition and/or inversion). a mathematical. the index of a subgroup. e. The AST has never dealt with all these parameters in a global conceptual framework. the AST is a very special achievement. the Z-relation.3 for more details). but this is not what leads to a powerful theory of relations between local and/or global musical objects. is a generalization of a previous Lisp library of Janusz Podrazik (http://www. Although the theory of categories has been around since the early 1940s and is even recognized by computer scientists. The library.2. the prime form. glissando. We mention that the Dn library is part of a wider project on implementation of algebraic methods in OpenMusic which originated with the Zn library (devoted to the structure of the cyclic group of order n in the classiﬁcation of chords and of tiling rhythmic canons.2. Its concepts are thoroughly out of date from the point of view of 20th century mathematical conceptualization. Other set-theoretical concepts which have been implemented are the normal order. or the semi-direct product of translations and multiplications which explain a large portion of the T T O by the corresponding short exact sequence 1 → eZ12 → T T O → Z× → 1. Even the most standard concepts in group theory are ignored. or the full T T O group as well as conjugacy classes of subgroups of T T O have been established. such as onset.mracpublishing. translations and inversions. It would be important to apply the above vocabulary switch in order to adopt the ﬁndings of AST to workable mathematical formalism such as it has been used by Fripertinger in P´lya and de Bruijn enumeration theory. Mathematically.11. no attempt is visible in AST to deal with morphisms between pcsets. We have seen certain germs of this direction in the deﬁnition of abstract subsets.5. combinatoriality and partitioning. called Dn (from the Dihedral group of order 2n). just to name a few important ones. 12 Also the concept of a group action and corresponding elementary facts such as orbit cardinalities in relation to isotropy groups does not appear. the literal and abstract complement. and crescendo.com) on AST. the interval vector. GROUP-THEORETICAL METHODS IN COMPOSITION AND THEORY 257 for a presentation of this computer-aided system for music composition and analysis) by Carlos Agon and Moreno Andreatta. and a model-oriented one. such as chord classes under translations. a conceptual. loudness. see section 16. We also have to recognize .4 Comments We want to conclude this section with three short comments on the AST. 11.g. It also allows us to switch from the pitch perspective to the rhythmic content by mapping the circular representation into a ‘chord’ or into a voice (see section 51. o It would also be necessary to confront the AST approach with the many other parameters which deﬁne musical events. We shall deal with these directions in chapter 21 on metrical and rhythmical global compositions. the literal and abstract inclusion and the sub-complex Kh. The originality of the OpenMusic implementation consists in the manipulation of settheoretical operations in the circular representation. for example.

In turn. Whatever the status of a baby theory the AST concept framework might be. emotional or fuzzy-like poesy. perhaps best in the work of Herbert Eimert and Iannis Xenakis. could have the slightest connection with something as intuitive. Morris’ composition designs are a real enrichment to grasp the complex construction of precise sound aggregates when starting from pcsets and similar elementary local compositions. As groups pertain to mental reality. Group-theoretical classiﬁcation is a central issue of poetology. Most theorems of AST are of strictly combinatorial nature. It appears that modeling has predominantly been oriented towards and useful for compositional strategies. 11. –Σ– This subject is hard to understand from scratch. European music theory has only very rarely shown up in this domain. We are happy that ﬁnally. But we can help in channeling them and focusing attention on the critical points. Also the analytical use of the AST language is a considerable one. in Amuedo’s work).6 Esthetic Implications of Classiﬁcation Summary. But the mainstream of after world-war II European musicology had turned towards dialectic mumbo-jumbo and far-out aesthetics and transcendental black-box-theories. Perhaps the confusion between mental facts and their psychological or physiological correspondences is one of the important sources of irritation. neutral level.g. it is an indispensable reset of a rotten conceptualization in musicology where even the most elementary things are blurred. The composer may use certain tools to express esthetic categories. In many talks and discussions. these latter may or may not relate to actions on other levels of reality. ORBITS that the operationalization of musicological concepts has been completely realistic insofar as computer programs and algorithms have been provided. We shall discuss this subject in extenso in especially section 30. On the mental level.258 CHAPTER 11. The conceptual comment must take care of a dramatic need for precise musicological concepts as tools for dodecaphonic analysis and of its theoretical extension in atonal theory. But there remains a big lack of models in the sense that beyond descriptive tasks.. The question of theoretical modeling is a diﬃcult one. So we could summarize the AST achievement as a necessary but far from suﬃcient attempt to escape decadent and impotent European musicology. the AST language has very seldom led to musicological modeling. we do not discuss psychological or physiological aspects of the phenomenon but only mental perspectives.3.2. I have experienced the disbelief that something like abstract group theory. let us localize the problem on musical topography. . So Americans had to start from scratch with precise conceptualization of even the simplest concepts such as pcsets. pcsegs and their classes. To begin with. It is used as an esthetical tool in analysis and composition. We come back to such achievements in chapter 25. together with group operations and orbit set construction. This section is not written to replace passionate discussions about the legitimation of abstract algebra in poetology. and esthesis.4. there are three communicative allocations of poetic instances: poiesis. it was possible to simply talk in a precise jargon about analytical problems of atonal and also tonal music (e.

. message. Orbits which are generated via the multidimensional forms of local musical compositions (motives. even if we deal with so-called musical prose in the sense of Sch¨nberg or Wagner. Construction of poetic structure by use of orbits under symmetry groups..6. [457]. Roman Jakobson has isolated the poetical function as a central issue of esthetic structure in language. Understanding musical poeticity is far more than a business of evident transformations. but esthetic categories emerge on the human consciousness and therefore primarily pertain to the poietic or esthesic level to which we stick from now on. In fact. in the linguistic context. He has also pointed out that the poetic structure is not restricted to poetry. The message must be strongly constructed in absence of contextual and predeﬁned semantics. Immanent analysis versus imported knowledge. context. This is the crux and blessing of music. contact. Esthetic properties may also be deﬁned on the neutral level of the proper work.11. However. 11. it may be present in mnemotechique or publicity or political rhetorics as well. A strong esthetic category is poetic structure or poeticity..e. The turning point of Roman Jakobson’s poetic function in poetic analysis. prosaic esthetics could hardly ever be really esthetic in music. So poeticity is the function of language qua message. referentiality. paradigmatic transformations are not so rich and complex as with music. receiver. –Σ– In his famous paper [245].) can only be controlled by explicit group theory. The importance of group theory to poeticity is that it makes the instance of paradigmatic transformation equivalence precise.ihr prim¨res Kennzeichen ist das Abweichen von Normen klassischer musikalischer Metrik.”. and this is a core characteristic of poeticity of music. autonomous constructions is an essential procedure to elaborate the musical sign system.1 Jakobson’s Poetic Function Summary. And poeticity is one important issues of esthetics. we do not have automatic semantics and therefore. They characterize it as an antithesis to metrical regularity35 . In fact.. Jakobson deﬁnes poeticity—without further empirical deduction—as follows: 35 “.. He relates these structures to the functions of (in the same order) emotivity. poeticity. but o metrics is by far not the only instance of poeticity. musical prose can be perfectly poetic in the sense of the word as obtained by Jakobson’s revolutionary approach which we introduce below. After an introduction to Jakobson’s poetic function we want to perform a poetological analysis (Schubert/Stolberg) as well as a poetological (re)construction (Mazzola/Baudelaire). semantical enrichment by techniques of inner-systemic. Jakobson lists six structures: sender. chords etc. and code. ESTHETIC IMPLICATIONS OF CLASSIFICATION 259 and the listener or analyzer may apply appropriate tools to perceive esthetical categories. for example. and this is what perhaps causes another irritation among those who would prefer that the genealogy of poesy be as straightforward as the evidence of its power upon humans is manifest. conativity. It is interesting to note on which basic linguistic structures and functions poeticity is located. We shall stress the group-theoretical aspect and give a critical comment on this procedure. and metalanguage. i. together with emotional functionalities. phaticity.6. a . in music.

This enforces the construction of syntagmatic equivalence based on syntagmatic adjacency.e. ORBITS Deﬁnition 35 The poetical function projects the principle of equivalence from the axis of selection to the axis of combination. i. this function is poetical. The ﬁrst one is a given relation. Apart from imParadigm prose: random .12: The Jakobson function associates the paradigm with each syntagmatic unit. the second one is not. mediate adjacency. . For Brutus is an honourable man. The typical example is a periodic distribution of syllables in a verse form. The The “axis of combination” is the relation among signs in their linear ordering of the syntagm.. this function is “random” whereas in case of poetry. such as a hexameter. In case of prose.. The “axis of selection” is the paradigmatic axis of language. the sequence of ﬁve dactylic metrical units . The ﬁrst shares the associative relation of paradigmatic ﬁelds whereas the second shares the adjacency relation of syntagmatic chains.. no a priori relation is given. But it is also possible to start on a more elementary level of dactylic equivalende: We have aperiodicity of . the relation among its signs by topological or transformational equivalence as discussed in chapter 10.... mostly a periodic relation.e.. But Brutus say he was amitious. Syntagm Paradigm poetry: regular told you Cesar ws ambitious.. The principle of equivalence on the paradigmatic axis is association whereas this principle on the syntagmatic axis is relative position on the syntagmatic chain.260 CHAPTER 11. The terms in this deﬁnition have the following meaning. i. periodic with respect to a selected equivalence between syntagmatic positions. Syntagm Figure 11. So we are given two spaces (axes): the abstract ensemble of all possible signs and a concrete string of signs which are juxtaposed in a linguistic message.

. ..B.. but with prosaic texts. a “vertical” poeticity must be considered. This vertical perspective relates to the connections between diﬀerent semiotic aspects of signs when coupled by poetical functions.. verse A ending with “. So the rhyme on the ﬁnal syllable of corresponding verses is a poetical Jakobson function on the phonological level. Suppose. it shows no regularity whereas the function (by deﬁnition) realizes a poetical functionality if it is periodic with respect to a syntagmatic (in this case: periodicity) equivalence: Its values are identical for equivalent arguments on the syntagmatic axis.. It associates with each syntagmatic unit us its paradigm jak(us ) = paradigm of us .. Given such a syntagmatic periodicity (just to ﬁx a concrete syntagmatic equivalence relation).11.6. including “domicile”. equivalent positions in syntagm are the syllables on a determined column of the verse matrix. for each dactylic meter one unit. and that the poetical function is realized on the phonological level by a rhyme at the verse ends. we would have two of these words on corresponding arguments.B. for example. This Jakobson function jak is always deﬁned. . See [361] for this topic.. If we have the semantic paradigm of “house”. Jakobson wasn‘t precise either on the level of paradigmatic equivalence. But it is also possible to start on a more elementary level of dactylic equivalence: We have a periodicity of successive triples of syllables. but it does more: it correlates the semantic contents of the words “God” and “mob” in a stronger way than in normal prose.. It relates to the semantic production mode by symmetries which we have discussed in the beginning of chapter 8. see also ﬁgure 11. Vertical poeticity is an important tool for semantic enrichment. a fundamental feature of poeticity. that we are given a verse form equivalence on the syntagmatic axis.12. ESTHETIC IMPLICATIONS OF CLASSIFICATION (/ ) (stressed/unstressed or long/short) plus the ending trochean unit (/ / / / / / / / / . and then equivalence of metrical position means the ﬁrst. the Jakobson projection is related to a special function jak from the syntagmatic axis to the paradigmatic axis. / / / / ): 261 In such an arrangement of hexametric rows. such that besides the “horizontal” poeticity following Jakobson. God”. as may be shown in a more abstract “metrical versiﬁcation” scheme: / / / .A. “home” etc. Suppose that we have alternating verses A. We are given a poetic . and the next verse A ending with “mob”. second or third place within each dactylic meter. Roland Posner has observed [421] that diﬀerent levels of poeticity must be distinguished. This projection of the paradigm of the phonological class /o/ in the ﬁnal syllable onto the syntagmatically equivalent ends of two verses of type A is a ﬁrst poetical functionality.

is reﬂected in the soul of the singer. Wipfel/Zweige. semantics is enriched. not only stressed: We obtain an “added value” of meaning.. one of Schubert’s best performances. The way this song accomplishes this identity turns it into a singular artistic achievement [424]. a result which Moritz Bauer [44] has called a masterpiece. 11. and we shall do this for the transformational paradigm. By means of poetic coupling. in the second strophe. After this reference to traditional humanities. the soul moves rather passively “wie ein Kahn” “auf der Freude sanft schimmernden Wellen”. westlich/¨stlich. mainly in the sense of group actions. no poem realizes symmetry—in the sense of congruence between emotional disposition and esthetic form—so perfectly as these lines. each threefold. The objects of this world are arranged in a symmetric fasha ion around something like “meine athmende Seele”. This is a prototypical motivic analysis relating motives to dactylic metrics. the rigorous articulation catches our eye: three strophes of six verses each. The composer succeeds in establishing a fascinating identity between text and musical transformation. At ﬁrst sight. Classiﬁcation (building equivalence ﬁelds) as a prerequisite to perception of equivalence is a central reference of poeticity. Schubert’s setting (op. the dactylic meter achieves another equilibrated movement: the regularity of undulation.6. again a triple measure. The creative power of the poetic Ego incorporates its environment. It reveals a fundamental regularity in the distribution of three-tone motives in Schubert’s composition. What happens to the Ego in the exterior object world. Whereas in the ﬁrst strophe. f¨r meine u Agnes”. see ﬁgure 11.262 CHAPTER 11.” Summary. ending on identical rhymes. It is an added value to the prosaic meaning of syntagmatic units. ¨ namely in a series of four polar word couples: Uber/Unter. 425] has recognized that in the eighteenth century. we want to start our analysis with the textual poeticity and then try to recognize the corresponding poeticity in music. 72) translates the word’s movement into continuous movement of sound. we shall deal with a concrete and precise realization of the Jakobson principle. In the following examples. This semantic enrichment is a conditio sine qua non for poetic meaning. Whereas the identical rhyme produces the intended eﬀect of monotony. written in 1782 and set to music by Franz Schubert 1823. o Kalmus(s¨uselt)/Schein (winket). Science of literature [424.2 Motivic Analysis: Schubert/Stolberg “Lied auf dem Wasser zu singen. “atmet die Seele” more actively. 3. Poeticity is based on an “unfolding of correspondences in time” or “a tale of symmetries”.. –Σ– We want to analyze Leopold Stolberg’s poem “Lied auf dem Wasser zu singen. 2. This is the text: . The following three facts give an upshot of poeticity: 1.13 for the vocal score. ORBITS index to a semantic relation between God and mob.

the same activity (also identiﬁed on the ﬁrst position of the a verse) relates “die Seele” to the said “Kahn”. respectively.6. The turning points of “Heins” and “Scheins”. “westlichen”/“¨stlichen”. we have two phonological paradigms of (1) a trocheic meter / with its two identity paradigms /Wel/ and /len/ and (2) the stressed / with identity paradigm /Kahn/. Morgen entschwinde mit schimmerndem Fl¨gel u Wieder wie gestern und heute die Zeit. auf der Freude sanftschimmernden Wellen Gleitet die Seele dahin wie der Kahn. The ﬁrst relates to the ﬁrst strophe. We are given the syntagmatic equivalence of feminine and masculine ending syllables of verses 1. produce symmetry correspondences of the ¨ word couples “Uber den Wipfeln”/“Unter den Zweigen”. the third is a pairing of distinct sensorial modalities: visual-showing against auditive-whispering.5. Therefore. an animal. o Unter den Zweigen des ¨stlichen Haines o S¨uselt der Kalmus im r¨thlichen Schein. Denn von dem Himmel herab auf die Wellen Tanzet das Abendroth rund um den Kahn.3. a Ach. o Ach es entschwindet mit thauigem Fl¨gel u Mir auf den wiegenden Wellen die Zeit.5. a o Freude des Himmels und Ruhe des Haines Athmet die Seel’ im err¨thenden Schein.3. On verse 4. Bis ich auf h¨herem stralendem Fl¨gel o u Selber entschwinde der wechselnden Zeit.6. the latter pairing . this Kahn’s activity of “gleiten” (sliding) is compared to the same activity of “Schw¨ne” (swans). On verse 2.4. This metaphoric intensiﬁcation sets forth a strong semantic enrichment qua eﬀect of this poeticity. On these two classes of positions.4. ESTHETIC IMPLICATIONS OF CLASSIFICATION LIED AUF DEM WASSER ZU SINGEN.11. So three object categories: a boat. The second example relates to the second strophe. These two equivalences relate the corresponding verses and their corresponding units: The phonological equivalence /Kahn/ points to the semantic unit of the word “Kahn” (boat). 263 We discuss two examples of the Jakobson function. the second is a pairing of geographic antipodes. and 2. and 2. This turns out to be more than a syntagmatic coupling of unrea lated objects: Each pairing is a paradigmatic symmetry on the semantic level! The ﬁrst is a spatial localization pairing of vertical antipodes relating to the extremal positions on a tree. ¨ Uber den Wipfeln des westlichen Haines Winket uns freundlich der r¨thliche Schein. “(Schein) o winket”/“(Kalmus) s¨uselt”.6. A phonological poeticity is set up which is isomorphic to that described before: equivalence on verses 1. two diﬀerent object categories “Seele” and “Schw¨ne” are loaded with additional semantics: they slide in the same way as the “Kahn” a does. ¨ FUR MEINE AGNES Mitten im Schimmer der spiegelnden Wellen Gleitet wie Schw¨ne der wankende Kahn. and a psychic entity are moving in the same way.

This one will become evident in the third strophe.Winket... ¨ again coupled with the other vertical pairing: visual activity is from above (“Uber. in the last couple of verses.13: Voice part from Franz Schubert: Lied zu singen auf dem Wasser. The ensemble’s added semantical value is this: We recognize a group of instances being arranged around a central point of rotation which switches the polar positions into one another: west into east and vice versa. metaphor for the poetic Ego. u op. a force which turns things around. der Schein”). 72 [480]. ORBITS Figure 11. f¨r meine Agnes. in other words.. but a force which remains unnamed. Formally. This ensemble of a polar symmetries presents semantic equivalences. we are ﬁnally informed of the moving force: It is “die Seele” which “athmet” “Freude des Himmels” (from above!) and “Ruhe des Heines” (from below). the center of rotation is the breathing soul. the pairings are two-element orbits of positions which are arranged by the basic poeticity on the phonological level. etc.. and vice versa. However.264 CHAPTER 11.. Only there—at the end of a successive and dramatic . The poetical function sets up a center of rotation.S¨uselt der Kalmus”). above into below.. auditive activity is from below (“Unter.

. bk X Schim-mer der 14 Schwä-ne der 14 Freu-de sanft 10 .. Let us now look for poetical functionality on the musical level. we are led to work in form OnM od12 ⊕ P iM od12 . ESTHETIC IMPLICATIONS OF CLASSIFICATION unvealing process—are we provided with the basic.. 3 Denn von dem 14 tan-zet das 10 . The Schubert composition is written in 3/4 time. 11 X K k X @K@X Wel-len II Kahn III1 Wel-len III2 . Together with octave periodicity... So 12/16 time suggests that we introduce onset of form OnM od12 .. nominative form of the pronoun “Ich”.. 10 See-le da10 .11. 14 jX j X jX spie-geln-den 19 wan-ken-de 6 schim-mern-den 11 .. 10 Him-mel her10 A-bend-rot 10 . and we may capture all durations in integer multiples of 1/16 notes.... Relating to the dactylic text as discussed above..6... This gives a grid of 6 times 3 motivic units. 11 ab auf die 14 rund um den 10 .. Each motivic unit will be given . 10 Kahn IV2 ... 10 jXj X . corresponding to six verses per strophe and three dactylus meters per verse (the endings are not considered). IV1 X . Figure 11..14: The highly symmetric arrangement of motive classes in the dactylus grid of the voice score in Schubert’s setting of Stolberg’s poem “Lied auf dem Wasser zu singen.. V Wel-len VI1 Kahn VI2 .. bX j J KX hin wie der 11 . f¨r meine u Agnes”. 11 glei-tet die 3 . 265 I Mit-ten im 14 glei-tet wie 14 Ach auf der 23 . we want to group melodic parts according to the three dactylus groups in each verse.....

this is 37%. 10. it is a highly characteristic feature of a good composer to realize an individual brand of paradigmatic equivalences. but a whole variety of analyses). This “spectral analysis” would yield a poetic support of the parameter space in question (in fact not one analysis. and then to reﬁne the group choice until non-automatic poeticity emerges. − → In this vein.7% to hit a representative of this orbit36 . we take the three-element motives with eighth note onset distance. − → We take the full aﬃne group GL(Z2 ) as equivalence principle. rely on known paradigms.2. we should follow the line of this size hierarchy. In order to decide which group is the best one (if it is any unique one!). for example). on the level of group actions there is a hierarchy of orbit size or group size. not common or standardized. we have an entire bunch of paradigmata to be applied. consisting of all parameters where poeticity can be observed. The paradigm is extend to the total of 144 = 487 344 motives in Z2 contains 110 592 motives of the orbit of Nr.14 shows the result. Its 12 orbits are 26 in number and we may ask whether this equivalence relation yields structures beyond random functionality on the dactylic grid. the identity group will very improbably instantiate a poetical functionality (though we have exactly this situation on the phonological level described above). The problem is that the analytical work cannot. this is 12 3 − → 2 card(GL(Z12 ))/card(Aut(N r. the idiolect of an individual composer can vary considerably. More precisely. Figure 11. the orbit Nr. contrary to the linguistic situation. In music. On the other. 10 appears on 10 of the 27 positions. More systematically one may call poeticity the entire spectrum of Jakobson’s poetical functions jakπ : Syntagm → P aradigmπ for all paradigms of a certain parametrized arsenal of a priori candidates. complete identiﬁcation (by the full permutation group on Z2 ) 12 of all function values will clearly produce poeticity. we approach the easiest non-trivial case: the full aﬃne group GL(Z2 ). respectively. against a probability of 22. We just start an experiment (of the mind) and set up a ﬁrst approach to equivalence. However. conformal with the accentuation. This is by no means obvious 12 or mandatory. The Jakobson function of this paradigm has highly symmetrical (regular) properties.2. though not globally. ORBITS its natural structure of a three-element motif M in OnM od12 ⊕ P iM od12 . The Jakobson function has to be evaluated along the syntagmatic dactylus grid.14 shows the grid as well as the isomorphism class number of the three-element motives as listed in the class list of appendix M. but this is a trivial case which is deﬁnitely unspeciﬁc. and it isn’t necessary to copy the poietic standpoint to the esthesic perspective. 14 and Nr. And it shows a perfect axial symmetry in the dactylic grid.10)) = 663 552/6 = 110 592. The values have to be taken in a paradigmatic space which is. Where we have the repetitive sixteenth note motives (bars 2. So the idea is to start at this end where poeticity is most probable. The dominant Nr. in fact. 5. 11 orbits. Figure 11. 10 “spine” also shares a more reﬁned paradigmatic role in the following sense which will be made more precise in chapter 12. so we should discuss the choice now. This forwards an important methodological question: Where should one start on the paradigmatic analysis if no pre-selection is available? Is there the “right” selection against “wrong” alternatives? Globally seen. in this case. 36 The . Other symmetric conﬁgurations are realized for the Nr. Above all. we are not going to judge upon what would be the best analytical choice of paradigmatic equivalence in this piece of music. One one end.266 CHAPTER 11.

The Hasse diagram37 of this relation on the isomorphism classes of 12 three-element motives is shown in ﬁgure M. numeric functions which measure the presence of motif classes in the total motivic space of this music piece.15: The weight proﬁle of the three-element motive classes as a quantiﬁcation of the musical paradigm of motives.11. it dominates every other class.2. Figure 11. The importance of the spinal motif Nr. 37 The minimal diagram whose transitive closure is the given relation. 10 is also evidenced by use of motivic weights. 10 is the generic class. the spinal class is also the generic specialization paradigm. 10 I II III1 V IV2 IV1 III2 VI2 VI1 Figure 11. This relation is − → invariant under GL(Z2 ). and related to the corresponding textual instances of the poetic Ego. Under the specialization paradigm.15 shows the weight proﬁle of the motives over the dactylic grid. Athmet die Seel' im er röthenden Schein. and it appears with highly signiﬁcant overhead above a priori probability. 10 towards the end of the poem.2 of appendix M. deﬁnition 121.6. Therefore. in particular the dominant Nr. see appendix C. Details can be found in chapter 22 and section 41. POETICAL "EGO" Tanzet das Abendroth rund um den Kahn. Selber ent schwinde der wechselnden Zeit. Nr.2. ESTHETIC IMPLICATIONS OF CLASSIFICATION 267 dominance relation between equipollent motives: X Y (X dominates Y or Y is a specialization of X) iﬀ there is a surjective morphism X → Y of local compositions. .

composed and performed on CD [339] in 1990 by the author. ORBITS So these facts are a concentrated presentation of a highly poetical functionality together with a motif-weight dominance of the spinal class Nr. Rather is the question emerging from this analysis whether Schubert’s motif conﬁguration is an unconscious instantiation of a highly poetic expressivity. “Seele”.2. “Himmel”.A rhyme: LA MORT DES ARTISTES Combien faut-il de fois secouer mes grelots Et baiser ton front bas.6. The compactness of this analysis should however not entail that the propedeutical work of classiﬁcation and specialization data be underestimated or neglected. –Σ– This section deals with paradigmatic composition techniques applied in scherzo movement three “Poem of Wind” of the 45 minute concert for piano. no intention from the side of Schubert has been subtended.and 3. “Kalmus”. ˆ mon carquois.15 shows these words: “Abendroth”. “Abendroth”. Figure 11. To understand this analysis. In particular. 10. The words being associated with the spinal dactylus are nouns of the poetic subject: “Freude”. This analysis is fairly neutral: It is scarcely esthesic and limits itself to a motive structure which is established by means of a classiﬁcation list and its extension to the specialization Hasse diagram. “Zweigen”. a preliminary understanding of classiﬁcation is indispensable. The composition was realized on the composition software presto which will be dealt with in chapter 49.he-rem stra-len-dem Fl¨ -gel o u Sel-ber ent-schwin-de der wech-seln-den Zeit.B. “Poem of Wind” is a musical transformation of the ﬁrst and second strophes of Charles Baudelaire’s poem “La mort des artistes” from his abysmal “Les ﬂeurs du mal” (in the ﬁnal version of 1861). Within this perfection also appears the subject in its explicit form in verse V of the third strophe: Bis ich auf h¨. a locus where also the symmetry of Nr. “Seele”. The climax of the motivic weight is taken on the middle dactylus of the V I1 verse setting. Classiﬁcation in fact has a deep cognitive impact which we do not yet understand: In which way can poetic feeling be generated through group actions and the associated orbits of equivalence? A ﬁrst answer to this question will be given in chapter 30. and follow an embraced A.B. 10 and Nr. These strophes are centered around the poetic Ego whereas the third and fourth strophes are written in the third person.268 CHAPTER 11. Combien. This section describes the use of the 26 isomorphism classes of three-element motives in the construction of a composition on the basis of Baudelaire’s poem.3 Composition: Mazzola/Baudelaire “La mort des artistes” Summary. percussion and bass “Synthesis”. perdre de javelots? o . de mystique nature. 11. 14 becomes perfect. and “(Ich) entschwinde”. They are four verses each. and this is correlated with the middle of the three last verses of strophes 1. morne caricature? Pour piquer dans le but.

11. 39 The 38 The . a particular percussion sound so that we may distinguish the motives from their percussion sound range. etc.16).12). We started from the phonological paradigm realized on the level of letters and spaces. “Contemporary Percussion”.(/f/. In this setup. In this piano concert. The melody is a local composition M elodyGerm in Onset|Z ⊕ P itch|Z and the covering is a sequence of 26 subcompositions M otift ⊂ M elodyGerm such that each M otift reduces to a representative of class Nr.4). see the global theory.22).. (/m/.16 shows the covering and the germinal melody.48).0). each group being distributed along an interval of pitch values. Figure 11.19). This covering of a local composition by a system of small sub-compositions will turn out to be a crucial procedure for global structures in music. (/e/. These instruments are grouped into toms.23). and the inter-strophe onset diﬀerence was set to 96. (/n/. Then. 10. It was constructed such that it can be covered by representatives of these classes.17).1). we take care of the selection of a representative for each class.60).20). in particular the theory of “interpretations” discussed in chapter 16. i. and by the hierarchy of motive classes in the specialization Hasse diagram (see appendix M. The next step is the concrete realization of these representatives on the level of instrumental voices.24). (/m/. Each letter and inter-word space within a verse were given an onset with onset diﬀerence of 12 integer units. e Avant de contempler la grande Cr´ature e Dont l’infernal d´sir nous remplit de sanglots! e . In the next step. (/p/. we have letter /n/. For example. (/ /. as a local composition and in conventional notation.. starting like this: (/C/..18). (/j/.6).. the piano is accompanied by 122 diﬀerent percussion instruments39 . etc.17 shows the presto score as elaborated for the original piano score. 269 Compared to the Schubert piece. horizontal axis is integer onset time. (/o/. instruments are taken from Yamaha RX5 and TX802. and not to usual pitch. the bijection is this: (/e/. Vertical axis is presto onset (1-71).14).5). (/t/.2). we just choose a reasonable candidate.84).15). ﬁgure M. Each motif is given a speciﬁc pitch range. (/u/..36). (/o/. (/l/. (/b/.10).72). So we have an sequence of letter-onset couples. (/n/.96). we make use of a germinal melody which is responsible for the entire piano concert. (/i/. Each inter-verse space was deﬁned by an onset diﬀerence of 48 units. The small quadratic icons bijection is not unique. corresponding to motif class Nr. On each such onset.. (/r/. the paradigmatic realization of the 26 motif classes was much denser here.13). (/q/. and this is associated with the generic motif Nr. For the eﬀectively used 21 letters. bells. (/s/. Figure 11. and Roland R-8M synthesizers..6. ESTHETIC IMPLICATIONS OF CLASSIFICATION Nous userons notre ˆme en de subtils complots. (/f/. The most frequent letter is (as expected) /e/.12). a Et nous d´molirons mainte lourde armature. (/i/. we have snares for pitch values 8 to 17. each once. and “Mallets”. (/c/. the pitches are assigned to diﬀerent percussion instruments. This yields the candidates for representing all classes. The latter is charged with the sound chips “Jazz”. t modulo 12 in onset and pitch .11).e.3).11. relating to Onset|Z. To this end..26).25). In our context it is only an auxiliary tool to construct a kind of synopsis of all classes as being “charts” of a global connection. (/d/. (/g/. (/a/. (/b/.2). a motive class will be allocated following a bijection between the 26 classes and the 26 letters which is motivated38 by the frequency of letters in the two strophes. (/y/.. (/v/.

including the prosodic speciﬁcation by tempo which must be deﬁned here since the presto software is required to furnish a real performance of percussive parts such that the piano can interact as if the synthesizer output would be a human one. we see the tempo curve which determines the micro timing of the piece.12 . denote elements of the local composition. The melody is shown as a score and as local composition (magenta-colored points).270 CHAPTER 11. ORBITS 6 J & 8 X X X bX bX X X X bX bX bX X X X X X X X X bX X X bX bX X X X j X 15 24 14 25 19 12 9 3 13 11 6 16 26 22 23 5 4 10 1 2 8 21 17 18 7 20 Figure 11. percussion. each representing one motif class in OnP iM od12. we can just remark that the function of tempo—and this is considerable. Below the upper rectangular score. We recognize not only the letter-related onsets of the diﬀerent motives. . This subject will be treated in detail in section 33. So the overall philosophy of this musical construction is a one-to-one transformation of the linguistic text into a musical one. Here. and bass is a patchwork of 26 three-element motives. as one can recognize from the extreme variation of the presto tempo curve—is that of a musical prosody which is parallel to the linguistic one related to the poem’s words.1.16: The germinal melody of the Synthesis concert for piano. but also some ornamentation which was mainly added for echo-type percussion eﬀects.

to 4096 onsets/min. This problem arises when we question the program of neutral analysis.7.17: The original Synthesis score from the presto graphical interface and piano performance prescriptions. Whereas historicity seems abolished from the group-theoretical point of view. G¨nther Mayer u has pointed out this problem in [104]. In fact neutral analysis does not oﬀer tools for historical localization of works. This ideas originate from Jean-Jacques Nattiez’ paradigmatic theme. classiﬁcation appears to be an ahistorical approach. Mathematical structure and diachronic position are independent. in other words: Analytical work seems to be reduced to a mathematically oriented systemic process. 11. . the tempo curve below the score shows an extreme variability from 32 onsets/min. We want to discuss this problem in the light of structuralist theory of musical analysis as it has been developed by Ruwet and Nattiez. MATHEMATICAL REFLECTIONS ON HISTORICITY IN MUSIC 271 Figure 11.7 Mathematical Reﬂections on Historicity in Music Summary.11. –Σ– The group-theoretical classiﬁcation theory which we have dealt with in the preceding development is based upon categories of local compositions which are denotator constructs of seemingly exclusive mathematical nature. In this sense. apparently contradicts historical perspectives—which are essential to a musical dynamics that parallels the historical dynamics of our physical universe. it is evidenced that the diachronic line of compositional tools parallels—among others—the size of involved transformation groups. and as such.

For instance. deﬁned by a determined species of geometric transformations [393.1 Jean-Jacques Nattiez’ Paradigmatic Theme Summary.4. one could ask for a restriction to translations. i. in the words of Ruwet [466. par exemple. When comparing two hierarchical systems of units under a paradigmatic theme means admitting only the isomorphisms which stem from the system P = (PM )M of automorphism groups. The germ of such a reconciliation was laid down in the investigations of the Paris school of structuralist linguistics. a special subgroup of automorphisms PM ⊂ GL(M ) for each ambient module of musical parameters. On this level. Ruwet’s concept of neutral analysis means partitioning the work into a hierarchical system of units which have to be classiﬁed up to equivalence among each other and compared to units within other works. more − → speciﬁcally. However. equivalence means selecting a special subcategory of local compositions. e The totality of classiﬁcation criteria of a given analysis have been termed “paradigmatic theme” by Nattiez..134]: Les diveres unit´s ont entre elles des rapports d’´quivalence de toutes sortes. Mathematics oﬀers a wide range of structures. PM = eM . ORBITS 11. in particular no classiﬁcation tools are discussed on this level: the paradigmatic theme is only a local aﬀair.7. This theme is realized by a concept of equivalence among units. in Ruwet’s and Nattiez’ approach no systematic account is paid to the structure of this system.265]: Les unit´s paradigmatiquement associ´es sont ´quivalentes d’un point de vue e e e donn´ (le th`me paradigmatique). –Σ– The above singular “mathematical structure” is misleading. and the possibility to distribute this spectrum on the diachronic axis may initiate a reconciliation between structure and history. rarement identiques. et reli´es entre elles par des e e e transformations qui d´crivent les variants par rapport ` des invariants. de tel autre — et aussi des segments empi´tinant les uns sur les autres.7. e a The hierarchical system of units is in fact a special case of what we shall call interpretation of a local composition in the framework of global theory.e. This section discusses the structure of Nattiez’ paradigmatic theme in describing associations in the frame of Jakobson’s function.1 of Jakobson’s poetical function) to musical semiology and analysis.2 Groups as a Parameter of Historicity Summary. mainly in the works of Nicolas Ruwet [466] and Jean-Jacques Nattiez [393] regarding applications of methods and insights of Saussure and Jakobson (see our previous discussion 11. This section terminates the chapter with an abstract of the determinants of size in historical growth of paradigmatic transformation groups. –Σ– . des segments de longeur in´gale — tel e segment apparaˆ ıtra comme une extension. p. So we could redeﬁne the paradigmatic theme by the above system P .6. 11. see 13. e e rapports qui peuvent unir. ou comme une contraction.272 CHAPTER 11. p. possibly extending the latter to a subcategory of Mod which has P as the system of automorphisms.

7.11. MATHEMATICAL REFLECTIONS ON HISTORICITY IN MUSIC 273 By use of the preceding mathematical rephrasing of a paradigmatic theme P . On the contrary: Principle 6 The appearance of certain symmetries as transformations in analytical technique or compositional tools is a secure index of a new musical epoch. And this is by no means restricted to esthesic positions. In fact. a perspective which is equally aﬀected by historical dynamics. . We could even hypothesize that the paradigmatic theme has been monotonically growing since the last ﬁve hundred years of European music history. the paradigmatic theme becomes a determinant which traces historical dynamics. we may attach P as a diachronic index to a determined analysis. it can be applied symmetrically on the poietical side of a work. This means that the a priori variability of admitted isomorphisms—which by default is purely mathematical—could be used as a valuation criterion in view of historical localization. the presence of a paradigmatic theme in a deﬁned historical moment—be it in esthesic valuation or in poietic construction—is by no means an invariant of the diachronic axis.

.

In general. We make explicit the latter topic in a discussion of the problem of topological classiﬁcation of sounds. more details will appear in other chapters. topological considerations are of central importance since slight deformations of objects to neighboring objects are standard identiﬁcation concepts—though never handled with the necessary care. It deals with the general question of what it means to be in the vicinity of an object. musicology and also humanities have not yet understood the deep impact of genuinely topological reasoning. and we begin with a classic of shape conceptualization: Ehrenfels’ gestalt theory. Therefore any description of them must ﬁnally prove but approximations. transformations will conserve interior relations of objects. for instance in motive theory (chapter 22) or in inverse performance theory (chapters 45 and 46).12] Summary. This aspect is covered by the topological perspective. or site. The relatively abstract character of topologies may be a reason for their sparse usage in the humanities. The following discussion will in particular evidence the advantage of topologies for the delicate and still unsettled problem of sound classiﬁcation. In its infancy. in the ambient space. and in the semantic potential which topology induces. The point in making these structures precise lies in the sharpening of a fundamental descriptive tool. In music. but the abstract character is precisely the power of this approach: It is very helpful in creating concepts which are akin to usual fuzzy situations in the humanities. We want to make this plausible in this preliminary chapter. Topology has only penetrated the humanities on the metonymous level of metrical reasoning where topology is not at its best. general topology is a basic discipline. p. mathematical topology was in fact called “analysis situs”. Genuine topology is a radical antagonist to transformation or metrics.Chapter 12 Topological Specialization Musical phenomena come to existence in the constant ﬂuency and motion of compositional creation. but not their ‘absolute’ position. –Σ– Although in mathematics. Rudolph Reti [444. 275 .

e. Now. The former means that the gestalt of an object does not depend upon the identity of the object. super-summativity is akin to the old Aristotelian principle that the whole is more than its parts. And deformation is not transformation since we do not refer to any transformation rule whatsoever to say that the second melody is ‘similar’ to the ﬁrst one. the second being produced from the ﬁrst by a pitch transposition. Small deviations of the pitch relations and also small deviations in the temporal reproduction of the second melody against the ﬁrst one will not bother our gestalt identiﬁcation. We may replace this object by another specimen which is related to the original object by a determined transformation. even the most innocent transformation. but they are not suﬃcient. We rather need a concept of distance or neighborhood such that the second melody stays at this small distance or in the neighborhood of the ‘prototype’.276 CHAPTER 12. it is also an argument towards classiﬁcation in the sense of stable concepts in parameter spaces underlying denotator forms. We do not discuss this characterization here. We discuss the reasons. Without this stability aspect there is no workable concept of gestalt. –Σ– In Christian von Ehrenfels’ seminal work on gestalt [136] he has characterized gestalt by two attributes: invariance and super-summativity. In the musical example this means that the melodic gestalt is more than the enumeration of its tones.. Suppose that we have to play these melodies on an ordinary piano.1 What Ehrenfels Neglected Summary. no diﬀerence in the totality of relations is aﬀected. this invariance against small deformations is neither super-summativity nor transformational invariance. Let us discuss this topic on Ehrenfels’ example of a musical melody. i. Our present concern is rather that Ehenfels did not give a workable deﬁnition of gestalt since he did not specify the problem of gestalt stability. in other words: it is related to the relations among its parts rather than to their isolated presence. The same observation holds with super-summativity: If gestalt refers to relations among its parts. the transposition of the melody by some semitones does aﬀect its identity but not its gestalt. TOPOLOGICAL SPECIALIZATION 12. On the contrary. In fact. This is not only an argument from cognitive psychology. clearly. . Then. Small deformations really have to take care of the original position and small alterations of its coordinates. for example. such as a transposition. take into account the question of gestalt stability. So we may summarize that gestalt must be preserved under small deformations. does strongly alter the identity of the melodies constituents: pitch will change dramatically if we transpose by two octaves. however. it is equally true that it is stable under small variations of these relations. the conservation of the gestalt under more or less small deformations. In Ehrenfels’ example of a musical melody. Suppose that we are given two melodies of equal gestalt. The point is that not only perception and sensorial processing are subjected to stability requirements. Apart from a marginal remark he did not. see chapter 22. Ehrenfels’ concept of gestalt is characterized by invariance under transformations and super-summativity. we will not refuse the gestalt equality of these melodies even if the piano is not tuned in perfect equal temperament. To be clear: we contend that the two well-known Ehrenfels criteria are necessary. more will be said in the global theory and in particular in the topological theory of motives. The second attribute. But the precise quantity of alteration is not relevant. no speciﬁcation is required—except that it be ‘small’.

such that: 1. music involves two radically diﬀerent topological processes: comparison of given neighboring objects. and this is one of the valid arguments of Ren´ Thom’s catastrophe theory. The second requirement is not so.. there is a special neighborhood V of x such that it also is a neighborhood of every one of its elements y ∈ V . 4. All we need is a minimum of properties of neighborhoods. Both types of topologies are equally important in music and musicology. set-theoretic understanding. 3. then adding other points to these does not alter this speciﬁcation. the old title for topology in mathematics. the near ones are still there. We are given a set T op of ‘points’ of whatever nature and want to give an axiomatic account of what it means that we stay in a neighborhood of a selected point x. the modern topos theory is nothing else but the functorial globalization of this old approach. Here we suppose that the reader has this prerequisite in his/her mind or that he/she is willing to believe that the following discourse is provided with a rigorous mathematical theory. The third axiom is a ‘ﬁrewall’ against too general a situation as suggested in axiom 2: It says that a neighborhood contains a ‘core’ neighborhood around that point in the sense that this core neighborhood is also a neighborhood of all its elements. –Σ– We refer to appendix H. . V ∈ Nx . This need not be the case for an arbitrary neighborhood: it is not true that any member y of a 1 Not toposes in the modern technical sense but in the old. In other words. we ask that x ∈ U . Basically.e. 2. It means that if a neighborhood contains the ‘very near’ points to x.2. The cornerstone of topology is the concept of a neighborhood. e 12. then any larger set U ⊂ V is also a neighborhood of x. TOPOLOGY 277 Abstract conceptual processes are also built upon stability phenomena. We concentrate on the characteristics of topological argumentation. if we say that a gestalt of an object x is conserved under small deformations. Topology deals with the “logic of toposes”1 . However. called neighborhoods of x. For every such neighborhood U . Here are our requirements: Axiom 1 Every point x ∈ T op has a non-empty system Nx of subsets U ⊂ T op. see [526]. it is a conceptual framework for the “analysis situs”. If U. The ﬁrst requirement is completely natural from common language. So a neighborhood of x is a set of points ‘surrounding’ x and containing ‘small deformation’ points around x.2 Topology Summary. V ∈ Ny .12. and if we agree that a neighborhood of x should by deﬁnition (!) contain those small deformations. i. then a larger set a fortiori contains those small deformations and therefore still is a neighborhood.1 for the mathematical deﬁnition of a topological space. If U is a neighborhood of x. If U is a neighborhood of x. then also U ∩ V . and “degenerative” specialization of an object into a derived object. This section introduces the topological argument and its formalism.

In this setup.} consisting of the integer zero and all positive prime numbers. whereas all other primes are “closed” points.1) So we have a slight asymmetry: All primes are dominated by 0.278 CHAPTER 12. see ﬁgure 12. We also say that x dominates or specializes to z iﬀ z ∈ {x}− . a situation which is completely pathological if one thinks of naive neighborhoods in normal life (say a person sitting near you in a crowded subway will have the same feeling that you equally sit near this person). x is an arbitrarily small deformation such as z. the closure {x}− of x is deﬁned as the set of those points z such that x is a member of all their neighborhoods.1. In this language2 . We set x = 0 all 0 ∈ U ⊂ Spec(Z) with Spec(Z) − U = ﬁnite. 2. 2 Open sets are a common alternative to neighborhoods to write down the axioms of a topology. The example space is the set T op = Spec(Z) = {0. . in symbols: x z. it is not true in general that specialization is a symmetric relation. This suggests that we should consider globally ‘stable’ sets of points in T op: By deﬁnition. Intuitively speaking. but 0 is dominated by no prime.1.2 for this concept from algebraic geometry. The point of this construction is that it is by no means symmetrical with respect to the points of T op. topological space to illustrate the concepts. Nx = x = 0 all U ∈ N with x ∈ U. 13. the so-called open sets of the topological space T op deﬁned by the above neighborhood collection. 11. The fourth requirement is evident from common language. the prime spectrum3 of Z. So x is arbitrarily ‘near’ to z. see appendix H. namely the zero. Let us give one concrete and classical example of such a strange Spec(Ÿ) 0 2 3 5 7 11 Figure 12. these are the sets O ⊂ T op which are neighborhoods of all their elements. 0 (12. 17. Other examples will be given in the following sections. 5. . . This means that there may be points x which are arbitrary small deformations of points z. 7. the core neighborhood of x in axiom 3 is an open set (and neighborhood) built around x. but not vice versa. More precisely. . 3 See appendix F. TOPOLOGICAL SPECIALIZATION neighborhood of x has this neighborhood as its own neighborhood.1: The topology of the prime spectrum of the integers has a unique “generic” point. but there are small neighborhoods of x such that z will not appear as a small deformation of x in such neighborhoods. But the core neighborhood is stable: everyone of its members contains small deformations in this neighborhood. it is a paradigmatic framework sui generis! And we shall make full use of it. We insist on this abstract point of view since it shows that by no means can we associate topology with transformational paradigms. if we consider any points x in T op. 3.

–Σ– In the humanities.2 for more details. measuring something like similarity between the couple’s objects. Exercise 18 Prove all these claims concerning the preceding example Spec(Z). above all in psychometrics. For instance. So all primes in our space are closed whereas 0 is not. (Not to be confused with temporal metrics of music!) Metrics are quantiﬁcations of similarity and generate a very special—thoroughly intuitive—type of topologies. metrical comparison arises if a set S of objects is loaded with a non-negative. to introduce topologies which are far from the common intuition. Avoiding such pathological—but realistic—phenomena in psychometrics. 279 (12.12. let us assume that we have a distance function on a set S of denotators such that it deﬁnes a metric 4 on S. In general.1. for example we may have d(x. {x}− = {x} else.1 Metrical Comparison Summary.2. .2) One says that a point x in a topological space is closed iﬀ {x}− = {x}. and mathematically standard. 6 See appendix H. The idea of variation is problematic in this topological paradigm.3) appendix H. every prime doesn’t dominate any other point: We have Spec(Z) if x = 0. for example. this aspect is a neutral comparison of given objects and does not ask for generative procedures to derive objects from each other. real-valued distance d(x.1. x) > 0 or d(x. This example shows that it is very easy. if we measure the distance between two pitches in auditory recognition. this can be a dramatic function. Of course it is not a question of introducing artiﬁcial topologies. it dominates all other points and is therefore called generic. if S is a local composition in an ambient space which is isomorphic to an n-dimensional real vector space V . Essentially. the usual Euclidean metric on V deﬁnes5 a metric on S. such a quantiﬁcation of similarity can be arbitrarily wild. y) for every couple of elements x. Such a metric canonically deﬁnes a topology6 on S by the neighborhood sets Nx = {U | there exists a positive 4 See 5 It such that B x ⊂ U } (12. we should try to make the a priori point: topology is a very powerful tool for creating similarity paradigms without any allusion to naive distance concepts which are well known to psychologists in the wide-spread polarity proﬁles. But here.2. See appendix B. The zero is a very ‘thick’ point compared to the prime numbers which are all topologically closed. TOPOLOGY Further. x) if the pitches x and y are presented enough time apart. 12. and this one is completely natural as a mirror of well-known facts from number theory. y) = d(y. We shall introduce such natural topologies on spaces of motives. and if the presentation time is not a parameter of the events. For example. for example. y in S. Metrical comparison is based on geometric distances between objects within their ambient spaces. is however a non-trivial problem to recalculate a possible ambient space for embedding an arbitrary set S which is provided with a metric on S.

In particular. TOPOLOGICAL SPECIALIZATION where B x = {z| d(x. There are important properties which are intrinsically topological. For example. nothing is prototypical. paradigms are not disjoint partitions of the given space. there is no such thing as a pitch which can be deﬁned via psychological perception. But the associated topology wouldn’t have changed7 . the relation is symmetric. dominance is absent. 5)-just tuning8 are a dense subset of all pitches: there is no -paradigm of any pitch without just tuning pitches. they will be more similar. As a rule of thumb. this is an important fact for pitch paradigmatics. every point is closed. theorem 75 for background. But this means that the “paradigmatic -equivalence” around x deﬁned by being in B x is not a mathematical equivalence relation. For 1 such a serial motif Ser we set Ser 1 = k |Seri | and then d(Ser1 . this says that we lack basic techniques for generating hierarchies among the given objects. To put it more topologically: In a topology which is derived from a metric. if x is in the -ball neighborhood of y. Division by k guarantees that if the series get longer and have only few diﬀerences.0 of serial motives of length k with integer values. z) < } is the -ball (in S) around x. talking about prototypical objects and variants thereof in this context is a delicate point. memorizing a corpus of objects in this ﬂat world is utterly unattractive. then no pitch concept will emerge since equality will not be an equivalence relation. So why stick to topologies if one is given metrical functions? What is the added value? The added value is that in general. i. in the above example.2. appendix A. the pitches stemming from (2. However. one should observe that metrics and associated topologies are not good candidates for paradigmatic concept constructions. We have no control of how relata in a given paradigm could have been generated from each other. Distance does not generate prototypes. equivalent entities. It may happen that an object z is in two diﬀerent paradigms.2. A motivic analysis or a sound classiﬁcation which is built on metrical topologies is completely ﬂat. group-theoretical paradigms. Example 18 Take the set S = SERMk. then the same is true ith exchanged roles. Exercise 19 Check this with the above example. metrics have an artiﬁcial ﬂavor. The other way round. For example. it is of dramatic impact: If one tries to deﬁne pitch by perceptual equivalence via diﬀerences which are just below the famous “smallest just noticeable diﬀerence” (jnd). Therefore. y ∈ B x and z ∈ B y does not imply z ∈ B x. these topologies are the antagonist of functional relations in paradigms.3. the relation is not transitive: In general. In other words.. In this setup. we could have taken another distance function. and this contradicts an elementary requirement of conceptual identiﬁcation. This is a trivial mathematical fact. . However. the topological paradigms derived from metrics have the symmetry of paradigms in common with the transformational. and 7 see 8 See appendix I.1. and therefore. Ser2 ) = Ser1 − Ser2 1 .280 CHAPTER 12. one of a second x2 . starting from the Euclidean norm 1 Ser 2 = ( |Seri |2 ) 2 instead of Ser 1 . Summarizing. 3.e. one of x1 . but they diﬀer in that transitivity is broken. they are independent. they are independent of any chosen metric. but for musicological conceptualization.

the topology being the ordinary metrical topology. This is evidently violated by metrical deformation. Our present concern is something between transformation groups L K Figure 12.3. section 12. TOPOLOGY 281 composition (poietical work) from this perspective will never yield more than an associative chaos.6.2.2.12.2: Three (light) points in colinear position may lose this property while performing a very small displacement. associated with metrical structures. This is a problem which does not arise with aﬃne transformation groups since any set of points on a line will also be colinear after the transformation. we start working in the very small category of zero-addressed ﬁnite local compositions K in ambient vector space M = R3 . namely the relative position of points. Specialization is related to generative relations between “dominant” and “dominated” objects. The idea of motivic hierarchies is related to this type of topologies. We come back to this basic fact in the following discourse on sound classiﬁcation.2 Specialization Morphisms of Local Compositions Summary. and its signiﬁcation for the Schubert/Stolberg composition exposed in section 11. If we are to produce basic directives for compositional strategies. something may happen which is invisible from the metrical point of view. say. and metrical deformations: Transformations from transformation groups are invertible. 12. Then the set GL(R3 ) of invertible . Look at the topological situation within the space of aﬃne transformations et · m which is isomorphic to − → R3 ⊕R3×3 . The topologies which are instantiated in this context are radically diﬀerent from the above topologies.2. 12 –Σ– To make the ideas concrete and intuitive. see ﬁgure 12. We discuss the specialization graph of three-element motives in Z2 . and therefore sets of colinear points in local compositions are invariants of the group orbits. metrical topologies cannot be the motor of quality. While deforming such a local composition in the sense of metrical neighborhoods discussed in the previous section.2. by moving each point a bit away from its given position.

we encounter an interesting phenomenon. do intersect. see ﬁgure 12. But for vanishing λ. but the limit composition K(0) has strictly10 more colinearity relations than the other K(λ). we obtain the projection m(0) = p1.2 . 9 The 10 At determinant is a polynomial of these coeﬃcients. m(λ) of this curve to a given local composition K.. More concretely.K = R3 . det(m) = 0. and this is a continuous function. If λ = 0. R. except for the special value λ = 0. in fact dim(R. TOPOLOGICAL SPECIALIZATION transformations is open since it identiﬁes to those points et · m with nonvanishing determinant.3. The latter is a proper specialization of the general member of the curve (K(λ))λ of local compositions.282 CHAPTER 12. straight lines connecting pairs of points in K which do not intersect and aren’t parallel. This is a condition which remains true if we change the matrix coeﬃcients of m by − → a suﬃciently small quantity9 . we obtain a one-parameter family or curve of local compositions K(λ) = m(λ)(K) which approximates the two-dimensional local composition K(0) for λ → 0. on K(0). m(λ) ∈ GL(R3 ). we cannot retrace K from K(0). in other words. where we have a surjective morphism which is not iso. i. All colinearity relation of K are preserved since we are dealing with aﬃne transformations. However.3: A one-parameter family of local compositions. Whereas all K(λ) give us K back via K = m(1/λ)(K(λ)). now. specializing to the plane projection for value γ = 0. The point is that we have a curve of morphisms m(λ) : K → K(λ) which are isomorphisms. if we approach the boundary of GL(R3 ).K) = 2. take a curve m(λ) of linear transformations 1 m(λ) = 0 0 0 0 1 0 0 λ − → as a function of the real parameter λ. The ﬁgure shows that if we apply a transformation l=1 lÆ0 l=0 Figure 12.e. least for our K having its points in general position as in the ﬁgure.

We have already alluded to this type of dominance topology in our example 11. Here. Then the set ObLocR R of all these local compositions bears a n. L K is equivalent to K.0 NK = {U | [K) ⊂ U }.d) · 0 1 0 0 0 0 with ﬁxed onset and duration values o. d produces a chord out of a motif. The dominance graph (Hasse diagram shown in ﬁgure M. and K L. So the hierarchy of motive classes is given a remarkable presence in Schubert’s setting of Stolberg’s poem. the partial order relation is invariant under GL(R3 ) and we may pass to the quotient 3 dominance topology11 on ObLoClassR R . A typical example is the harmonic projection of a motif. This deﬁnition evidently works on n.0 much more general spaces. the dominance was meant as hierarchy among diﬀerent instances of three-element motives in Schubert’s composition.12 instead of R3 . This derivation process is also a standard rule in dodecaphonism when the generic series is instantiated in a score.0 dominance relation introduced in 12. The dominance topology is an antagonist to the metrical topologies in that it is highly unsymmetrical: the core neighborhoods of two local compositions are equal iﬀ the local compositions are isomorphic. and duration.3) visualizes all core neighborhoods: for each class of number x. we deﬁne a neighborhood system on ObLocR R by n. 11 See appendix H. − → Clearly. R3 ). the set of classes above x are equal to [x). Musically.2 turns out to be identical with the synonymous relation introduced in this section. it is not true that motive classes in Z12. Exercise 20 The core neighborhood of K is uniquely determined. 3 (12.12.2. namely [K). and it is also the dominant element in the dactylic grid as shown in 11. In general. L being isomorphic. If [K) = {L| L K} is the 3 closed halfspace above K. in this order.1.12 have a unique dominant top element.0. the class number 10 stands for the generic class.2. On both these dominance topologies.2 that the presence of the generic class is signiﬁcantly above probability. this means that we are given a hierarchy of objects which is deﬁned by ‘projection’ of generally positioned points onto points which can be thought as living in lower dimensional ambient spaces.0 reﬂexive partial order relation K L iﬀ there is a surjective (and therefore bijective) morphism K L. We already saw in 11. but the idea is best seen on this concrete example. TOPOLOGY 283 The intuition of a ‘dominance’ of K over the special position projection K(0) can be turned into a topological dominance relation by the following construction: Let us ﬁx a cardinality n of 3 our local compositions (K.4) and thereby obtain the dominance topology on ObLocR R .6.2 of appendix M. The point of these topologies is that dominance means existence of degenerate transformations specializing the dominant local composition into the dominated one. There. . If we of think our example space R3 as parametrizing onset.2 where we worked in the module Z12.6. then projecting into the second axis by 0 0 0 p = e(o.6. pitch. the topological n.

3. there is no good classiﬁcation without semantic constraints. Purely mathematical games are arbitrary and inﬁnite in number. As may be expected. such as the catastrophe of the “Fourier paradigm”. good mathematical candidates for metrical or dominance topologies must be valid with respect to meaning of sounds. Sound descriptions are not neutral. The most irritating problem is the topological classiﬁcation of sounds with respect to its semiotic potential. Third.3. There is no such thing as a universal approach. we give the descriptive level necessary to deal with the semiotic problem which is the third topographic dimension to be treated subsequently in section 12. In that discussion. a combination of metrical similarity and dominance topology will be introduced. –Σ– We have positioned this subject within the topological paradigmatics because it is above all a topological problem—besides the other aspects. thereby generating a synthesis of the antagonist views set up in this chapter.3 The Problem of Sound Classiﬁcation Summary. of communication and semiotics. So the antagonists are only preliminary stages of a universal topological similarity paradigm. the communicative determinants of sounds are not clear. In fact. . 12. the shape of such varieties changes dramatically depending on these prerequisites. The reasons are triple: First.284 CHAPTER 12. We discuss the two relevant dimensions: communication and reality. sound varieties are—independently of selected sound parametrization/representation— completely unclassiﬁed objects. they depend on the topographic perspective. 12. Second. –Σ– In the following sections. meaning on poietical and esthesic. This is a delicate discussion since its chronically underpinned relevance has led to serious errors in music psychology and psychophysics. on physiological and psychological levels. which are also important but cannot be solved without a thorough reﬂection on types and tools from paradigmatics. as a function of the chosen sound representation and parametrization. We then turn to the problem of describing varieties of sounds.3. TOPOLOGICAL SPECIALIZATION We shall develop the dominance topology of motives in chapter 22. the semantic charge of sounds is a substantial constraint for classiﬁcation. Sound classiﬁcation is one of the most complex unsolved problems in musicology. This coupling of semiotic constraints in sound classiﬁcation should be the central concern of any classiﬁcatory theory.1 Topographic Determinants of Sound Descriptions Summary. We start our discussion with an account on the topographic determinants of sound description.

the synthesis machine may be a classical instrument.7. Suppose we are given a process of Fourier synthesis.1. More concretely. see ﬁgure 12. 12. This diversity on the physical level may be understood in its mirror structures on the mental level. or an electronic synthesizer built on analog and/or digital principles.3. sound is produced by a speciﬁc synthesis machine and received by another analysis machine which is in a completely independent device environment.4. example . a microphone and a digital or analog analyzer device built on a speciﬁc dynamical system or a digital sample recorder. THE PROBLEM OF SOUND CLASSIFICATION 285 S ansin(Qn(t)) construction decomposition Úgnf(t)dt mental reality Sound synthesis analysis physical reality sender message receiver Figure 12.. this phenomenon corresponds to a connotation/metalanguage structure.3.1 Communication Summary. Poietic and esthesic descriptions are independent of each other and may intervene for completely diﬀerent reasons. –Σ– As a physical object.e. The transition from mental to physical reality is more complex than shown here. The analysis machine may be a human ear. The necessity of such a distinction is classically misunderstood in the ubiquitous “Fourier paradigm”. Semiotically speaking.12.4: The axes of communication and physical/mental realities of sound production (psychological reality is omitted here). or human voice. The technological codiﬁcation of the purely mental constructs of mathematics are an intermediate layer. like a violin. i. the summation of a series of sinoidal curves as described by a circular Fourier denotator as described in section 6.

This has dramatic consequences for physical and/or physiological mirrors of this mathematical representation. otherwise. then it has to map small deformations of sinoidal waves to sinoidal templates. Fourier ideology (above all on the metaphysical level of so-called “pure” sinoidal sounds. The task of this system is to present the English consonants (e. T ense) Id 12 See [12. The SP E is built on a multidimensional digital space. . for example a rotation in a plane spanned by any two of the sinoidal base curves. if the auditory system really works by Fourier analysis to build its cognitive performance (and not only on the cochlear level. Consonantal. Round. /b/. −. The resulting parameter representation of the curve message will then have diﬀerent coordinates which relate to the original ones by an arbitrary isometry. in our example. The resulting curve is a time function which can be decomposed by a series of functions which are derived from the sinoidal functions via any isometric linear transformation. Sonorant.g. . see appendix B for the detailed mechanisms!). /i/. Id Id All have the same Bit-type Coordinator Z2 and diﬀer by name. The synthesis as well as the analysis machine have to be robust under small deformations.g. In order to identify these two aspects. 0 are traditionally codiﬁed by +. . . or conversely: in every neigborhood of a sinoidal function there are an inﬁnite number of impure functions. Low) Id consisting of 17 factors of essentially identical form: Syllabic −→ Simple(Z2 )... V ovel) Id with a ﬁrst cofactor Consonant −→ Limit(Syllabic. /ε/) as points in a stable coordinate system. maintained by Stockhausen. one would have to look at the orbits of Fourier coordinates under the group of isometries in our function space. for example) does not take orbits but the identical Fourier coordinates. The values 1. High. etc. So communication of sound poiesis to sound esthesis is a topological problem: You have to take care of topologically stable attributes of sound poiesis in order to guarantee its communicability. Consonant −→ Simple(Z2 ). The stability has to be a mathematical one and one that can be mapped into acoustic reality. /m/) and vowels (e. Back. p. this point of view is also unstable: Any tiny rotation of the coordinate system will turn pure waves into impure coordinate sets. TOPOLOGICAL SPECIALIZATION 2.286 CHAPTER 12. Topologically. say. perception of ‘pure waves’ is illusory. A good example of a stable codiﬁcation of sound phenomena is the phonological system as proposed by Roman Jakobson and Morris Halle in [244] and elaborated to the Sound Pattern of English (SPE) system by Halle and Noam Chomsky in [85]. For example. More precisely12 . Low. But this is not the common approach. The second cofactor is a six-dimensional product V owel −→ Limit(Syllabic.103ﬀ] for details. we set up a form SP E −→ Colimit(Consonant. a rotation.

−) with its coordinates13 in the cofactor space Consonant. 14 MIDI = Musical Instrument digital Interface. We discuss the representations inscripted by chronospectra. After all. see [399.2 Reality Summary.e. +. And it is meant that by identical vocal tract dispositions among humans (at least English speaking individuals) the esthesic interpretation of the symbolism should be unambiguous. etc.12. e. wavelet methods. To move deeper into the sound classiﬁcation problem. the consonant θ : 0 SP E(−. So small displacements do not change the type of physical dynamics in the vocal tract. Here. But music is not linguistics. linguistics has been successful on this path. Id 287 of Bit-typed factors. physical.g. 12. From classical Fourier analysis on real instrumental sounds. +.. The variety of musical sounds is far from frozen in a low-dimensional space above Z2 . This latter approach is also interesting because it represents something like a prototype of psychologically oriented topological classiﬁcation: it is hoped that the continuous space of physical parameters may be split into disjoint regions where discrete values can be attached in a stable way.105]. see [379. and physical modeling. frequency modulation. However. +.3. −. for example). This justiﬁes the digital representation of acoustical dynamics. So communication will have to face dramatic changes on both sides of the message object of neutral sound. −.1. There have been attempts to give these numbers a certain standard meaning (General Midi. −. −. the attempt is oriented towards grouping ‘similar’ instrumental sounds (such as piano-like sounds in number 0 to 7) on neighboring places on the axis of natural numbers.. one has to open a poietic (construction) or esthesic (decomposition) method beyond pure convention. −. sound types are accessed via program change numbers and these are just numbers for predeﬁned sounds in synthesizers which understand the MIDI code. +. i. p..3. –Σ– Sound parameters can be symbolized on the string level of pure names. This implies that any such position is isolated with respect to the others in an adequate representation of the vocal tract dynamics. −. but this has nothing to do with the contents of sounds. −. This representation is mathematically discrete and the acoustical correspondence to the abstract digital symbols is meant to be produced by a distinctive position/movement of the vocal tract. 399] . This is the case for the MIDI14 code. Psychological. −. −. its name is the phoneme’s usual name. and symbolic descriptions may vary substantially. A phoneme is a denotator of form SP E. The levels of reality deﬁne a strong parameter for sound representations. −. it is standard to model the temporal unfolding of a sound 13 Of course. −. the distinctive nature can be stably replicated on the basis of the receiver’s knowledge on personal vocal tract topology. and there should be no smooth transition between any two of these dimensions. but the information is unaltered. a grouping of some of these coordinator spaces would have allowed for fewer dimensions and more values (2k per dimension). THE PROBLEM OF SOUND CLASSIFICATION which has the same type Syllabic −→ Simple(Z2 ). for example.

exactly like the envelope form Envelope described in section 6. see ﬁgure 12. TOPOLOGICAL SPECIALIZATION sec 5 4 dB 75 50 25 0 0. For instruments of the violin family. Since we shall deal with this subject in chapter 53 on parameters for string quartet. We have described a corresponding form F ourierSound in section 6. the form ChronoF ourierSound is enriched by two classes of parameters: bow application and vibrato. we can stick to a short sketch here.5 for an illustration. we may replace the factor form F ourier by a one-parameter family of such forms Chrono − F ourier. It is erroneous to think that this is the end of the classical sound description.0 4. This deﬁnes the form ChronoF ourierSound as a generalization of F ourierSound. example 11.6.0 0 1 2 3 kHz Figure 12.0 2. example 12.6.5: The chronospectrum of a trumpet sound.0 3. the amplitude is deﬁned by an envelope curve. Moreover.288 CHAPTER 12. Bow parameters are split into four groups: • Bow pressure • Bow velocity • Contact point bow to string • Bow angle. In order to generalize that form to variable Fourier spectra. this is only a basis of general instrumental characteristics. . as a function where Fourier components are not constant.0 5. the dark sections represents the Fourier spectrum at a given time of the sound’s execution. In terms of instrumental construction.5 1. played decrescendo. parametrized by a variable λ ∈ [0. 1] of the unit interval. The vibrato eﬀect is given by four numeric parameters: • Relative delay time from onset • Modulation frequency (frequency of ﬁnger displacement) • Pitch modulation (extent of ﬁnger displacement on string) • Amplitude modulation (contact point of ﬁnger-tip).

2)..2. though less complex than physical modeling. The physical modeling technique is related to this situation. We then take the − → orbit GL(R) • ψ and apply each of the orbit’s members as aﬃne deformations of the prototype. (12. Second. the FM-coeﬃcients cannot be retraced in the general case. not the abstraction via mathematical curve synthesis. the FM-graph represents a new type of combinatorial variable (with Fourier it was only a discrete set of points).7. classically called a musical instrument. at present. i. Wavelets are a last example in our overview. In principle. It generalizes the classical Fourier construction in that it introduces a hierarchy of sinoidal functions which act as relative modulators in a functional concatenation which can be displayed as a directed graph15 . the wavelet representation of f (t) is faithful and 15 In YAMAHA’s TX802 synthesizer technology. On this function. ψa. We have to ﬁx a generic wavelet ψ. THE PROBLEM OF SOUND CLASSIFICATION 289 We call these eight parameters the technical parameters of a violin sound.2 for this.1. See appendix A.e.2 for this problem. thus yielding a system of ‘coordinate’ wavelets.1. the wavelet method is comparable to the Fourier method in that it parametrizes a given time signal function f (t) by a distinguished system of ‘coordinate functions’. So FM-synthesis is strongly poietic. it is a strictly poietic construction methodology of instrumental sounds via software modeling of the complete dynamics of instrumental execution. However. a reconstruction of the deﬁning parameters from the message signal is a very complex mathematical task. The inversion formula for wavelets says that this representation has an inverse and that it is an isometry. the FM analysis is far from settled: Even if we know about the underlying FM-graph. If the Fourier construction is rather easy to decode since the Fourier coeﬃcients can be calculated by classical methods (see appendix A.3 for technical details. TX802 oﬀered 32 such algorithms.5) This furnishes the “wavelet” paradigm prototype of the method. b) = cψ f.1.2 for technicalities. there are two important diﬀerences in the zoo of these FM-sounds: First. Relating to Fourier construction.b is the scalar product Lψ f (a. It however goes one step further in that the Fourier chronospectrum is replaced by the simulation of the physical device.2.3.1. So it models the making of a sound. example 3. They are an additional variety of sound construction superimposed on a given chronospectrum. see appendix A. The coordinate of the test function f (t) with respect to a coordinate wavelet ψa. And that a corresponding analysis. only special graphs show solutions.3).2.12. it shows that the topology of such poietic objects can become quite complex.3. for example the Mexican hat function ψ(t) = (1 − t2 )e−t 2 /2 .2. but less hard than physical modeling. The high ﬁdelity to natural sounds which physical modeling makes possible has the drawback that the parameter space for sound description becomes increasingly complex. this graphical structure was called the “algorithm” of the FM synthesis. see appendix A.b = eb · a • ψ = |a|−1/2 ψ ◦ (eb · a)−1 . the aﬃne − → group GL(R) acts from the left by ψa.b L2 (see A. FM sound construction is associated with a form F M -Object introduced in section 6. in other words. . frequency modulation (FM) is rather hard. The complexity of this poietical perspective of physical sound production is in sharp contrast to what can be analyzed on the esthesic side. see next section 12. possibly with cycles.1.

It is. etc.290 CHAPTER 12. And it is absolutely unknown whether on the neocortical level. but we have to pay the price of two coordinate variables. The rationale of Lerdahl’s construction (see also ﬁgure 12. which is built on a wrong analysis model. is a very limited system of ramiﬁcation trees from that theory which enforces absurd hierarchies without necessity. “openness”.6: Two hierarchical trees from Lerdahl’s attempt to impose a linear order and hierarchy on vibrato and inharmonicity. And they are also very fuzzy as such. Let us now turn to more psychological sound representations.6) since there is not the least indication of a one-dimensionality in non-harmonic Fourier components (what would be the theorem to which this fuzzy construct refers?). and it is induced by systemic . TOPOLOGICAL SPECIALIZATION preserves metrical properties among functions. “vibrato”. The common characteristic of these attempts is a shift from the physiologically founded coordinates of the SPE system to frankly psychological perception coordinates such as “acuteness”. p. The vibrato space is at least four-dimensional. In the attempt to transpose the linguistic SPE model to music. the wavelet method also must be concerned with questions of topological stability. they can hardly transgress the communicative barrier to poiesis. Even worse is the linearization of “inharmonicity” (right part of ﬁgure 12. For example. and the vibrato types are much richer than an ordered set which is dominated by a “prototype” (left part of ﬁgure 12. no fundamental frequency is needed here. All these attributes being on the esthesic side. In fact. In contrast to the Fourier representation. This perceptual model.79ﬀ]) deals exactly with this problem. several authors have described spaces of sound colors. however not evident that the ‘cochlear wavelet’ yields a faithful wavelet representation of sound (relating to adequate wavelet frames). wavelet analysis is performed. Lerdahl’s “vibrato” is completely wrong as a one-dimensional attribute. An example is the similarity between diﬀerent variants of a vowel described by prolongational ramiﬁcations where no hierarchy is given. “brilliance”.6) Figure 12. “inharmonicity”. Even if we accept templates of vibrato. The universe of sounds (let us talk about the poietic construction universe discussed above) is much too broad to be captured by manifestly reductionist words such as “brilliance”. “smallness”. these would be trivially conventional for technical reasons of software preferences and could not claim prototypical roles in musical performance. Among others Wayne Slawson [489] and Fred Lerdahl [293]. is wrong on every single step. However. the theory of wavelet frames (see [308. in particular the reduction to discrete “frames” which represent the continuous variety of parameters.6). we have seen above that four dimensions are needed for a common description. The generative theory of tonal music is mathematically primitive since it lacks most elementary tools of topology.

whereas one is more interested in topological description of ‘dominant regions’. .. and do not cover any of the fundamental problems in sound description. THE PROBLEM OF SOUND CLASSIFICATION 291 constraints which are completely artiﬁcial in this subject. P hi )0≤i is diﬃcult to turn into a reasonable manifold. The reduction to hierarchical trees simply destroys facts in favor of graph theory for beginners. A serious look at these problems is exactly what Lerdahl i missed in his work [293] at IRCAM where he applied the CHANT program to deduce his broken sound color space. The topological richness would never enforce unique maximal elements since it does not enforce global hierarchies. there are no reasonable sounds relating these regions.e. classiﬁcation. i. the relation between diﬀerent such regions is not clear: Possibly.) of dimensions which are even exceeded by the linguistic system (17 dimensions for consonants in the SPE model)..g. As a function of the chosen representation. the inﬁnite-dimensional coeﬃcient space of the amplitude and phase spectrum (Ai . . have nothing to do with sound synthesis and analysis whatsoever. see also ﬁgure 12.g. three. 16 We 17 Supposing stick to discrete Fourier spectra for simplicity for simplicity and experience that no two successive amplitudes are equal. ak < bk < ck .bk . In fact. i. at most. the universe of sounds constitutes a more or less diﬃcult variety. Such terrible simpliﬁcations from psychology are in dramatic contrast to the complexity of sound description for realistic synthesis and even analysis on the physical (or technological) and physiological level. possibly are there many diﬀerent paths from one such region to another. So we are left with complex problems regarding subsets of reasonable candidates and the manifolds which are deﬁned by these subsets. e. have a ridiculously low number (two.c ..7. In fact. by inverse images of functional equations F (A. We can summarize that the present sound color spaces from psychology are reductionist in their construction.c1 ∩ Ua2 . j with a ≤ i < j ≤ b or b ≤ j < i ≤ c.3.b1 . Such a formant condition deﬁnes an open formant set Ua. If we agree that such neighborhoods are regions of reasonable sounds. .e. locally maximal coeﬃcients in the sense of formants which are also responsible for vowel recognition16 . We look at possible shapes of such varieties and their meaning for the classiﬁcation problem.ck of all local peak neighborhoods of an increasing sequence of index intervals a1 < b1 < c1 ≤ a2 < b2 < c2 ≤ . Ai < Aj for couples of indices i.2 Varieties of Sounds Summary.c2 ∩ .). .3. the naive Euclidean distance does not express more than local similarity. This means that one is looking for indices b such that the amplitudes have local peaks at index b. –Σ– Already for classical Fourier construction.12. Topological dominance can have as many maximal elements as necessary. upper limits A2 ≤ const. dominance. Uak ..b2 .. and each17 formant set is given by the intersection Ua1 . 12. . CHANT is exactly what would be necessary to make experiments which are based on serious theories of formant manifolds.b. see also Reinhard Kopiez’ review of this situation in music psychology [274]. It is not even clear whether these manifolds are of a determined dimension since boundary conditions could deﬁne closed submanifolds. and sound semantics.) = 0 or energy constraints (e. similarity.

What was only deﬁned by the selection of relevant candidates is intrinsically built in for FM objects. This splits the FM-sounds into disjoint sets and is not a workable environment for the totality of FM-sounds since we are not yet able to compare sounds of diﬀerent graphs. Instead of sticking to graphs Γ. we get the generalized FM-object myF M w which refers to Γ(myF M )w . deﬁned by algebraic or analytical constraints on energy. for the deﬁnition of the FM-graph Γ(myF M ). if it does not act. The point is that graphs are discrete objects which we should embed in a topological space in order to manage deformation of FM-objects. We have a double description level of these objects. a graph arrow a : vi → vj in Γ(myF M ) symbolizes that the partial at vi acts as one of the modulators on the partial at vj . In fact. The strength of this action is not variable.292 CHAPTER 12. We then have an obvious projection p : myF M w → myF Mw (12. So this is a digital reduction which generalizes in a canonical way. P hv ) and generates a normal (weight 1) FM-sound from a weighted object. For a general weight w on Γ(myF M ). we have a p-section σ1 of the set Γ1 of 1-weighted FM-objects of given graph Γ in 18 See section 6. a patchwork of open manifolds and closed connections. example CircDen-3. Fv . So the variety of FM-sounds is parametrized by the directed graphs. Fk . we weight their arrows by non-negative real numbers.7. An F M -Object-formed denotator myF M has its FM-graph Γ(myF M )18 and the partial coeﬃcients (Ak . the relevant sounds can deﬁne a subset which looks like an inhomogeneous variety. In other words. then the unweighted situation for FM-sound myF M was equivalent to the situation with Γ(myF M )1 . TOPOLOGICAL SPECIALIZATION Ai Figure 12. P hv ) at the tail v of arrow a by the partial (w(a)Av . P hk ) ∈ R3 for each vertex k of Γ(myF M ).7: In the inﬁnite-dimensional space of Fourier coeﬃcients. Fv . If we denote 1 the weight having values 1 for all arrows. spectral formant distribution etc. writing w(a) for the weight of arrow a We therefore consider weighted graphs Γw . . we simply omit it.6) which by deﬁnition replaces each partial (Av . and is an open set in R3vΓ for the vertex cardinality vΓ of the parametrizing graph Γ.

This means that the generic point of µaΓ · myF M dominates the generic point of µaΓ · myF M ∗ . a speciﬁc graph deﬁnes a species. and the subspace Γ1 is deﬁned by the closed condition w = 1. So we can integrate diﬀerent graphs into a generic graph in such a way that they are deﬁned by cutting adequate subgraphs. if we let the weight of arrow a of Γ(myF M ) vanish. see appendix F. chapter 10]. within each family. If myF M is an object in Γ1 .2 Γ1 or Γ2 if w(a1 ) → 0 or w(a2 ) → 0.2.9. a2 in the latter which yield 1.8: Cutting oﬀ arrows means specializing weighted graphs. two graphs Γ1 . Though these methods yield valid and precise topological classiﬁcation of FM-sounds. according the number 1-6 of root partials. see [340. This means that a vocabulary switch between the preceding topology and the psychological intuition on instrumental colors is needed. . THE PROBLEM OF SOUND CLASSIFICATION the space Γ− of weighted FM-objects of given graph Γ. Such a specialization is obtained. myF M ∗ which are dominated by the orbit of myF M . Within one species. see ﬁgure 12. we have the action of the multiplicative group µaΓ . for example. respectively. In this setup. and then.3. The composed map Γ1 − − → Γ− − − → Γ1 −1 − −− σ p 293 is the identity. That approach divides the 32 algorithms from TX802 into six families. here for two arrows starting at the same vertex. Γ2 are called kindred if they are derived from each other by moving one arrow. if we are given an object myF M . where aΓ is the arrow cardinality of Γ: An element (ta )a=arrow acts factorwise on the weight w.8. This idea was already introduced in the classiﬁcation of TX802 sounds in the ﬁrst software prototype MDZ71 of presto . 19 This classical algebro-geometric topology is also given by the prime spectrum of a ring. the specialized FM-object has no contribution from this arrow. This means that they are both specializations of a weighted object Γw in such a way that there are two arrows a1 . its orbit µaΓ · myF M is irreducible in the Zariski topology19 . the perspective is strictly poietic and mental/technological. But on Γ− . we may ask for objects w=1 wÆ0 w=0 Figure 12. the aΓ -th power of the multiplicative group of the base ﬁeld R. It is known that the FM construction is far from evident on the psychological level: Controlling the FM parameters when building such a sound is very diﬃcult if one aims at constructing a sound which one “has in mind”. Then. see ﬁgure 12.12.

The point is that from all possible topologies. metrics. It should however not be a selection which is dictated by technical incompetence— such as the broken two-dimensional space constructed by Lerdahl in [293]—it should be deﬁned by rational criteria and as a such a variable selection. They are a function of the type of musical compositions and their instrumentations. For example. To complete this short remark on sound varieties. . but see [293] for details. classiﬁcation may become easier and more akin to psychological or semantic requirements.). it is often reasonable to select subvarieties of relatively small dimension. Sound classiﬁcation has to cope with constraints of meaning of sound. –Σ– Topological classiﬁcation of sounds is not only a matter of mathematical description of sound varieties on speciﬁc levels of reality. a vocal ensemble. 12. depending on the context. etc. selecting a sound subvariety which contains the corresponding instrumental colors will do. This concerns above all the topological properties of sound varieties. let us mention an important technique which will be applied in the discourse on string quartet theory 53. They also depend on the semantic charge which the composer imposes on his/her message.3. TOPOLOGICAL SPECIALIZATION Figure 12. In these subvarieties. Although sound varieties are of inﬁnite dimension.9: Two graphs (from the TX802 algorithms) are in kindred relation if they are specializations of a dominant third graph and are mutually related by displacement of a particular arrow.3 Semiotics of Sound Classiﬁcation Summary.294 CHAPTER 12. for questions concerning a ﬁxed orchestration (a string quartet.

they are also the key to a deeper understanding of many problems of musical communication and—above all—semantics on all levels of reality: mental (composition). Unfortunately.4 Making the Vague Precise Summary. So we are confronted with a complex problem: To map topological sound varieties (varieties of timbral colors.. we know the timbral spaces. no signiﬁcant research has been done to investigate possible maps and associated timbral topologies. physical (auditory perception). though. Or to deﬁne metrics on these manifolds such that geodesics correspond to semantically shortest paths. –Σ– The foregoing reﬂections are far from systematic. It is also clear that the present state of our theory. Regarding Jakobson’s poetical function.12. but of deﬁning them in order to settle semantical constraints. this was a more mathematical ﬂavor. They are good if they reﬂect meaning of sounds. emotional quality.. Summarizing. topological considerations unveil deep problems of communication and semantics of music. the problem is related to the deﬁnition of topological paradigms of sounds which turn emotional syntax of musical compositions into poetical syntax. is not suﬃcient to cope with more reﬁned topological situations. we know that the emotional syntax is poetically instantiated on the sound colors of musical compositions.. “metrical rhythm”. Meaning on any level: auditory physiology.). and we shall see that in the theory of motives. We are at the very beginning of an exact theory of topological classiﬁcation of musical objects. we are interested in those which are able to carry a semantic function. and psychological (emotion). So we are well prepared to envisage this new challenge: What are musical manifolds? . 12. and to give good and precise answers to the intriguing question why the string quartet came up to primordial prominence so suddenly in the middle of the eighteenth century. and we still lack an intrinsically music-related approach to global views. The question is highly non-trivial since it is not only a problem of testing topologies. Topological considerations are not only good technical classiﬁcation tools. Transcending the evident structural task by far. We know the emotional syntax (not so sure. it is shown that general position of timbral points in the violin family can help to represent contrapuntal and harmonic structures of classical compositions by Joseph Haydn or Luigi Boccherini. global metrics. and in the string quartet theory. We encountered many situations where the word “manifold” or “variety” was the good pointer. such as “motivic gestalt”. these techniques can be successfully applied to construct basic concepts. or compositional/analytical function. In the example [293] of the string quartet. the name is irrelevant) to semantic manifolds—such as emotional landscapes—with their topologies. But we have a good technical basis. But we do not know which timbral topologies and/or metrics turn music into timbral poetics. however. in such a way that the mapping becomes continuous or even an isomorphism of topological spaces. MAKING THE VAGUE PRECISE 295 etc. perceptual category.4. namely the strictly local point of view. topological methods of musical paradigmatics are powerful tools for grasping blurred and ambiguous concepts in musicology.

.

Part IV Global Theory 297 .

.

and performance.e. The uncertainty relation in physics.4 that the identiﬁcation of a musical work involves an integration of multiple perspectives. analysis. The categories ObLoc and Loc are representations of local objects. . interpretation. –Σ– This chapter represents a major segmentation point of this book since it introduces the paradigm of global objects into the topos of music. including ecclesiastical modes. the shaping elements of a performance. The adequate concept of a global composition is deﬁned. However. Pierre Boulez [60. tridadic degrees. . cognition. all of them are immersed in an omnipresent dialectic of analysis into local parts and synthesis of a global whole. understanding.. und zwar von der morphologischen Mikrostruktur bis zur rhetorischen Makrostruktur. a “patchwork” of multiply zoomed 299 . However. of objects that are not composed of proper “parts”. motives and themes. The investigation of the musical object tends to be an integral part of the object. it is very diﬃcult to separate analytical or synthetical activity from the object of investigation. this relation becomes a dominant and even characteristic attribute of music as an expression of human nature. This is a mandatory step as the nature of music is an organically compound one from all topographic points of view. compared to the investigation of exterior nature. stating that every experimental interaction with nature is subjected to a complementarity between hidden and revealed information. all of them deﬁne works and musical systems as multiply varied perspectives of a whole. i. II.60] Summary. bei der Bestimmung der Großform wirkt uber den ganzen Eingliederungsprozeßhinweg ¨ die gleiche Denkweise. Composition. and the units of cognitive processes.Chapter 13 Global Compositions . the parts of an analysis. music mainly deals with compound objects. The corresponding vocabulary of elementary global music objects is described. This integration is in particular one of local points of view. meters. We know from our previous discussion of Yoneda perspectives in section 9. The composition’s germs. rhythms. p.

Composition. there are only ambiguous theorists. This concept generalizes that of local compositions in the sense that it takes care of and formalizes ambiguity. Most of the relevant concepts. The objects of the global theory are mainly global compositions. analysis. degree.1 The Local-Global Dichotomy in Music Summary. 13. tonal function. The possibility of crystallizing selective actions in mathematical objects encompasses their structural comparability. refer to objects which are composed of local ingredients.300 CHAPTER 13. it is the deﬁniteness of ambiguity. the precise setup of morphisms qua objects of comparative studies implies an objective (meta)discourse upon comparative studies. see 8. The power of the global approach is. The fact that we were only able to deﬁne very elementary objects in the previous theory reﬂects the musicological fact that ambiguity enters very early in music and its science. but as a precise methodology of language formation. This section motivates the local/global dichotomy in music. There is no way out of making the multiplicity of compositional and interpretative activities a part of the theory’s concerns. On the level of music performance. The following global theory will compensate for this defect. on the contrary. We also believe that musicology cannot survive in the present status quo where essential competences and problems regarding the intertwining of object and subject are not lifted to a scientiﬁc level but still reside in the lowlands of feuilletonistic entertainment. be an adequate one. and performance do use this dichotomy—virtually without clear-cut . Moreover. proportional to the power of the local components: If we had not prepared a universally valid local concept framework. This achievement is interesting for musicians and musicologists since a theory of ambiguity also favors mediation between various possible explications of a given work or analysis. the global theory would also fail.3.3) that there is a limit of the local approach which must give rise to a global context. Rather the problem is to set up an exact theory of ambiguity.” Music is only instantiated via its interpretation or performance. This is the methodological background of the theory of global compositions. etc. A theory which crashes on the ambiguity of its object in misunderstanding it as an indeterminacy cannot.. contrary to Carl Dahlhaus’ judgment [100]. We should stress that this global approach is not intended as a normative standardization of musicological language. score representation. But the huge variability of such views is not a drawback or defect of music and its theory. this general insight was pronounced by Theodor Wiesengrund Adorno [6]: “Musik interpretieren: Musik machen. GLOBAL COMPOSITIONS micro-views. such as tonality. In an important special case—where we talk about interpretations—such global objects precisely represent the result of interpretative activity. The widespread belief that ambiguity and polysemy in music is a ﬁrewall against precise science unmasks its representatives: There is no ambiguous theory. a possibility which will be dealt with in chapter 14 about morphisms between global objects. is substantial: Fact 7 Music is not the deﬁnitely ambiguous. Recall that we had pushed the local theory to the extreme of its potential and then recognized (regarding the non-existence of general colimits. however.

86]: Bei Webern ﬁndet sich der Keim einer ¨ußerst fruchtbaren Idee. We see local subcompositions A. Berg. deﬁning a total of 15 partial series. H of Boul and some transformations. it would be absurd to start from such immense local compositions. which consist of 104 to 105 tone events. the idea o of hierarchies of local compositions is made explicit by Boulez [60. drawn as subsets of points which are surrounded by rectangles. Boulez’ reﬂections refer to the local composition Boul in ambient space OnP iM od0. Rather does the composer start from ‘small’ local compositions. . This approach deals with the composer’s perspective. Graeser and Ruwet/Nattiez. Ehrenfels’ criterion of super-summativity introduces the dichotomy into psychology. string quartets. These subcompositions are not disjoint in general. chords. We come back to a more systematic and formal discussion of the covering conﬁguration of this example in section 14. A short historical account on the simultaneous appearance (1854) of these approaches in musicology and mathematics with the work of Eduard Hanslick and Bernhard Riemann is given. Diese isomorphen Figuren bilden die Basis von privilegierten Mengen. sie integrieren sich in die Conditio sine qua non der Zw¨lftonreihe: die chromatische Totao lit¨t. I. including Boulez/Webern. die a ihrerseits auf einer h¨heren Ebene wiederum das vorstellen. The question of a precise description of the diﬀerence between the “sum of the parts” and the “whole” in Ehrenfels’ criterion is addressed. for sonatas.g. p. die die a Reihe als Einigungsfaktor von Untergruppen und Obergruppen betrachtet. G. a musical composition is never created as a local composition. THE LOCAL-GLOBAL DICHOTOMY IN MUSIC 301 conceptualization or even consciousness. it is evidenced that speciﬁcation of parts in music objects is not accidental for understanding the very nature of music. Xenakis and other music program designers.13. More generally. and Webern.12 of integer onset and common pitch classes. enta sprechend den gegebenen Transpositionen und Umkehrungen.4. Durch Verkettung f¨gen sich u 1 German: ”Binnenstruktur”. Jackendoﬀ/Lerdahl. they deﬁne a rather complex overlapping covering of the given series. Boulez/Webern. rhythms and similar elements as a basis of the ‘creative combinatorics’ and then merges these parts or recombinations thereof by use of various transformations and deformations in order to build a compound whole. In his discussion of compositional principles with Sch¨nberg. –Σ– Let us introduce the subject by the summary of a (non-exhaustive) series of approaches to global structures in music(ology).. such as motives. A1 . themes.1. was die isomoro phen Figuren selbst innerhalb der Reihe bedeuten. A good example is Boulez’ example of a dodecaphonic series in [60.. Ude/Wieland and Marek. its composition from partial series and their transformations. We discuss the radical diﬀerence between the local/global dichotomies in music and mathematics.1. daß sie sich immer innerhalb der gleichen Ordnung abwickeln. I]: He describes a series together with its internal structure1 .e. see ﬁgure 13. Im Eﬀekt sind alle isomorphen Figuren einer Grundstruktur von der Tatsache abh¨ngig. or symphonies. Built on the Aristotelian tradition. i. e. Hofmann and Kaiser. Evidently. In classical European literature. B.

einer in gewisser Weise “h¨heren Reihe” zusammen.A1 K.G ( ). The art of performance is not accessory.U. . zu einer Ganzheit.und Obergruppen betrachtet werden. Die Grundreihe kann dann als o strukturelle Kraft der Vermittlung zwischen Unter. Darum geh¨rt a a o die Idee der Interpretation zur Musik selber und ist ihr nicht akzidentiell. der als Synthesis die Sprach¨hna lichkeit festh¨lt und zugleich alles einzelne Sprach-¨hnliche tilgt.A1 b) A G U.B1 U. welche festgelegte Privilegien besitzen.B1 A1 B1 K. GLOBAL COMPOSITIONS a) A1 ( ).U. K = retrograde).1: a) Internal structure of a dodecaphonic series following Boulez [60. b) representation of the series as a local composition in ambient space OnP iM od0.G 1 0 0 2 U. Frames are drawn around partial series as suggested by Boulez (part a) of this ﬁgure).A1 1 0 0 7 C 1 0 0 8 H ( ). this was already pointed out by Adorno [6]: Musikalische Interpretation ist der Vollzug. I]. Ude/Wieland and Marek.12 . Reihenformen. which are related to each other by symmetry transformations (U = inversion. with kind permission of Schott-Verlag.A1 = B1 Figure 13.302 CHAPTER 13. So it is evident that Boulez had learned from the second Viennese school that there is a strong local-global principle in serial composition. but essential in the constitution of a musical work.U.

elementary parts of the composition. it was the merit of Ernst Theodor Amadeus Hoﬀmann—above all in his famous review of Beethoven’s Fifth symphony in the Leipzig-based “Allgemeine Musikalische Zeitung” in 1810—of having given music criticism importance as a contribution to the esthesic identiﬁcation of the work. In fact. This level of grouping into local units is also essential for mnemotechnical purposes and in order to give the ﬁngering strategy a support. it would be bad piano playing to phrase against the ﬁngering strategy. the metrical grouping of the melody creates Figure 13. and has contributed to the self-estimation of music and to the history of its reception. Since this achievement. successive parts which induce the shaping in performance by dynamical and phrasing (legato/staccato) prescriptions. music a need not be decoded. in Ceslav Marek’s standard work “Lehre des Klavierspiels” [317]. to name some of the important contributors. Semiotically speaking. Here. A semantic which is independent of the poetical function is excluded here. Uhde seems to refer to Jakobson’s poetical function (see section 11. with kind permission of Atlantis-Verlag). built from local. After the foundation of music criticism by Mattheson.1) as a projection of the paradigmatic axis to the syntagmatic axis of the sign system. Hofmann and Kaiser. It establishes an articulated whole. In practical performance theory.1.2: Two grouping proposals by Marek for dynamics and phrasing at the beginning of the C-minor fugue of Bach’s Well-Tempered Piano I. 2 Uhde: 3 Uhde: “meinende Intention”. such as Eduard Hanslick or Joachim Kaiser. the meaning of “alles einzelne Sprach-¨hnliche” is a semantic moment2 . .6. We choose the example of dynamics and phrasing which is built upon criteria of verse poetics (see ﬁgure 13. This view of musical performance as a poetical oriented activity is essentially syntactical articulation and paradigmatic intertwining. music criticism has been cultivated by proﬁled critics. or Avison.. but requires “imitation” of itself 3 . based upon criteria of verse poetics (from [317]. e. THE LOCAL-GLOBAL DICHOTOMY IN MUSIC 303 According to Uhde’s and Wieland’s comment in [535].g. “erheischt Nachahmung ihrer selbst”.2).13. Rousseau. Thus. the basic insights of Adorno and Unde/Wieland are realized in the artisanal details. the signiﬁcate of the involved signs.

Bauen wir einmal ein kontrapunktisches Werk auf. On the opposite side of the composers stand the musicologists whose eﬀorts for an adequate analysis are characterized by the need to retrace the composer’s thoughts. recapitulation. degrees. such as exposition. GLOBAL COMPOSITIONS It happens only as nonprofessional side-eﬀect that this criticism is eager to celebrate the “uniquely valid” interpretation and performance. and end up with large local compositions. Da die Themen selber Mengen von . This is a central eﬀort to open the access of a broader public to the present work. then this a fortiori is the work of the critic—only on the level of esthesis. This principle of hierarchical organisms was explicitly put into evidence in 1924 by Wolfgang Graeser in his analysis of Bach’s Kunst der Fuge [194].17]): Bezeichnen wir die Zusammenfassung irgendwelcher Dinge zu einem Ganzen als eine Menge dieser Dinge und die Dinge selber als Elemente der Menge. approaches which add to the given ones new perspectives. Dies ist eine Zusammenfassung gewisser a T¨ne. He describes a contrapuntal form as follows ([194. which may result in tiny local eﬀects on agogics. Da haben wir zun¨chst ein Thema. some are eliminated. deren Elemente T¨ne sind. Immer wird dies Durchf¨hrung u u die Zusammenfassung gewisser Themaeins¨tze zu einem Ganzen sein. Every new listening changes or questions the relevant local compositions and their mutual relations. Das klingt etwas abstrus. the articulated listening to music which was founded by music criticism has been enriched by a new aspect: multiply repeated perception. p. The overall impression of a performance integrates knowledge. variations of explanatory power. contrapuntal and harmonic progressions. or else to lament its vanishing. also eine Menge.304 CHAPTER 13. Graeser and Ruwet/Nattiez. such as chords. CD. it is a never-ending objective in this ﬁeld of classical literature to set forth new approaches to the spiritual torso of these sonatas. With the technology of saving works on LP. development. organically composed hierarchies are common structures. and personal disposition. questioning and evaluating the singular approaches in the spirit of a work in progress towards understanding an inﬁnite evolution. others are added. prejudice. Although Kaiser may be right in his Beethoven book [257] that whereas it is a rare event in our days when a fundamentally new reading of the 32 Beethoven sonatas is presented. wir wollen aber gleich sehen. so bekommen wir etwa das folgende Bild einer kontrapunktischen Form: eine kontrapunktische Form ist eine Menge von Mengen von Mengen. In this respect it is not astonishing that in the analysis of musical works. tonalities. the music critics should do exactly the job of commenting on this process and of comparing. the listener accumulates a patchwork of elements of comprehension. dynamics. voice leading. Aus diesem Thema bilden o o wir eine Durchf¨hrung in irgendeiner Form. also eia ne Menge. In the media and concert business. was wir uns darunter vorzustellen haben. and coda in the sonata form. Successively. articulation. and tuning. tonal functions. Now. They start from small local elements. MC and other media. if the poietic work of the artist is expressed by articulation and correlation of the composition’s local parts. Listening repeatedly to a work on CD changes one’s articulation and grouping activity. deren Elemente Themen sind.

sondern die Verkn¨pfungen und u Verbindungen der T¨ne untereinander.1. like rotations and reﬂections. ist die Eigenschaft der Symmetrie. de tel ıtra autre — et aussi des segments empi´tinant les uns sur les autres. though less ﬂexible regarding the transformations which may be applied (Graeser limits his approach to transformations of the “rigid” geometry. above all in the investigations of Nicolas Ruwet [466] and Jean-Jacques Nattiez [393]. in such a space. 305 The explicit reference to set theory is historically interesting since set theory was a relatively new language in mathematics in 1924. o Das wichtigste Grundprinzip der festen K¨rper. However the geometric aspect—embedding tones in spaces which admit symmetry transformations—is more radical with Graeser. isometries). It deals with portions of notes which are heard as building a unit of hearing. e e rapports qui peuvent unir. Jackendoﬀ/Lerdahl. o Es wird uns interessieren. wir k¨nnen also sagen: eine Menge von Mengen von o Mengen.13. These units are associated with each other by certain equivalence relations: Les divers unit´s ont entre elles des rapports d’´quivalence de toutes sortes. The text is somewhat misleading since it suggests that tones are abstract objects. and he also recognizes the role of symmetry transformations. ob wir gewisse Analogien zwischen den Gebilden. THE LOCAL-GLOBAL DICHOTOMY IN MUSIC T¨nen sind. denn deren o spezielle Beschaﬀenheit spielt gar keine Rolle. deren Elemente u Mengen von Mengen sind. und die Geometrie ist nichts o anderes als das Studium der festen K¨rper. This is however not the case: Graeser views tones as points in a geometric space. e This language resembles the one which Graeser seems to aim at. des segments de longueur in´gale — e tel segment apparaˆ comme une expansion. rarement identiques. par exemple. und unseren aus T¨nen o hergestellten erkennen k¨nnen.13]): Gegenstand der Untersuchungen sind aber nicht die T¨ne selbst. e The equivalence relations are not arbitrary but realized by speciﬁc transformations: Les unites paradigmatiquement associ´es sont ´quivalentes d’un point de vue e e donn´ (le th`me paradigmatique). ein kontrapunktisches Musikst¨ck ist die Zusammenfassung u gewisser Durchf¨hrungen zu einem Ganzen.e.. i. Und eine kontrao u punktische Form. In this . Explicit grouping concepts are described by Ray Jackendoﬀ and Fred Lerdahl in [243]. et reli´es entre elles par e e e des transformations qui d´crivent les variants par rapport a des invariants. Grouping is described from an esthesic point of view of music psychology. also ein Menge. so ist die Durchf¨hrung eine Menge von Mengen. p. die man in der Geometrie aus Punkten aufbaut. The method of neutral analysis which was developed by these authors starts from a hierarchical ordering into units and subunits which are a function of the given work. ou comme une contraction. transformations which may be applied to alter sets of tones or to compare diﬀerent tone-sets ([194. o The Paris school of structuralist linguistics has applied results of semiology after Saussure and Jakobson to musical analysis.

The American music for tape movement was initiated by Otto Luening and Vladimir Ussachevsky (ﬁgure 13. .17].286] it follows that these authors have no concept of the transformation groups which may be adapted to speciﬁc contexts. e concerto des ambiguit´s in 1950. It is also less paradigmatic than Graeser’s. a group consists of strictly timeadjacent notes and cannot be restricted to proper subsets within time-slices which would be deﬁned by voice splitting or parametric splitting of pitch or loudness.7. This approach resembles Graeser’s concept of a contrapuntal structure which is also a hierarchical grouping of parts. From the remarks in [243. Ruwet’s and Nattiez’ because no signiﬁcant statements are made concerning the association of groups under symmetry transformations. Their achievement was applied by John Cage and others in the Project of Music for Magnetic Tape. together with Morton Feldman. also based on tape as a ﬂexible medium of e syntactical combination and recombination. for example. David Tudor. Also this grouping is a nearly perfect hierarchy under inclusion: Overlapping neighboring groups may only contain one common onset and are treated as very special situations in this theory.4).1 for this concept). p.e. Pierre Schaeﬀer and Pierre Henry initiated the “musique concr`te” movement (ﬁgure 13.4: Pierre Henry produced his ﬁrst work in musique concr`te. Figure 13. a group is always deﬁned by a time interval. GLOBAL COMPOSITIONS approach. Otto Luening and Vladimir Ussachevsky.3: The founders of American music for tape movements. But it is precisely the psychological claim of grouping—even in its strictly hierarchical appearance of the Jackendoﬀ-Lerdahl theory—as a cognitive basic which undermines the fact of global structures in music. in Paris.. i. but Graeser’s idea was more general insofar as it did not strictly ask for time slices. Figure 13.306 CHAPTER 13. A more general “web of motivic associations” would be beyond the theory because it is not hierarchical [243. p.3) and applied by John Cage in the early ﬁfties4 since his ﬁrst music for tape Imaginary landscape. The e photograph shows him while realizing this music to the ballet Voyage au coeur d’un enfant by Maurice B´jart e 4 In the Project of Music for Magnetic Tape. and Christian Wolﬀ. Schaeﬀer and Cage. as it is explicitly proposed by Ruwet and Nattiez under the ﬂag of “paradigmatic theme” (see 11. At the same time. the new technology of tape music became a paradigm for local-global constructs. In the last years of the 1940s of the 20th century.

1 Musical and Mathematical Manifolds Summary. became an explicit and primordial feature of poiesis. this is a problem which touches all levels of the communicative axis.1. Cage realized compositions. But it does provoke the question how much more the added value exceeds the sum of the parts. etc. in fact the only interesting point of super-summativity. In the light of these rich traces of a local-global paradigm in music. Its fascination stems from gluing together the ends of o an ordinary ‘belt’-shaped strip after a rotation of half the full circle of one end. And it is a problem which involves interpretative activity and its innate ambiguities. such as his concert for piano and orchestra (1957). At present. and an intended ambiguity of identifying such parts. similar to geographic atlases.1. The local parts are termed tracks. The common ground of both approaches is that locally trivial structures can add up to esthetically valid conﬁgurations if glued together in a non-trivial way. Hanslick added that these forms are by no means elementary but o composed in an artistical way. it also changed the concept of a score. –Σ– The historical point of creating and pronouncing mathematical and musicological concepts of global gestalts is situated around 1854 when Eduard Hanslick (ﬁgure 13. notation. In particular. In the same year.5) published his famous treatise ”Vom Musikalisch Sch¨nen” [206]. The above examples show that this may be a very complex question. it is largely extended and reﬁned in various software for musical composition.13. Ehrenfels’ approach to gestalt which stresses super-summativity (in fact a warmed-up version of the Aristotelian principle that “the whole is more than the sum of its parts”) does not look very original. though having identical parts? How can we compare wholes if we suppose that their parts are comparable? Fact 8 As long as no precise structure theory of wholes qua constructs from local parts is available. We shortly discuss the question of how music and mathematics diﬀer in their understanding of local and global structures. and postproduction as compared to hard disk recording. global and local scores. where the modular structure of the composition. For example. These are understood as being patchworks of locally cartesian charts.6) conceived a far reaching generalization of the mathematical concept of space to so-called manifolds [448]. But the tape music movement had not only consequences for music software. . and building a unity within the manifold. This was a starting point for an entire group of technological realizations of localglobal patchworks. as restated by music theorist Hugo Riemann. We shall review one of these approaches in chapter 49 when discussing the composition software presto . By the arsenal of the preceding examples. parts. showing a number of relatively autonomous local parts. How is the whole constructed from the simple collection of its parts? When are two wholes diﬀerent. the mathematician Bernhard Riemann (ﬁgure 13. no real understanding of music is possible. 13. THE LOCAL-GLOBAL DICHOTOMY IN MUSIC 307 Tape as a new medium had become a tool for concretely cutting and merging time-slices of music. A simple and well-known example of such a global shape is the M¨bius strip. Musical content was recognized as being o “t¨nend bewegte Formen”.

Figure 13.6: Bernhard Riemann introduced global structures (manifolds) as compound mathematical spaces. the M¨bius strip is realized by the harmonic strip of triadic degrees within a o diatonic scale. we may go to the colimit of all atlases. nobody will complain that the geographic identity of the globe has changed. or periods. If I add a city map.5: Eduard Hanslick described music as a compound structure. formally captured by the concept pairing of charts and atlases. Mathematically speaking. This is a “patchwork” of (objective) local compositions. motives. The colimit is not allowed. Intuitively speaking. such as chords. is essential to the identiﬁcation of the composition. the individual interpretational activity is an integral part of the object’s identity. to use Ehrenfels’ concept. This more technical section gives the precise deﬁnition of an objective global composition. a standard structure in diﬀerential and algebraic geometry. the sphere of our globe. The characteristic diﬀerence between musical and mathematical manifolds is that musical manifolds are deﬁned with a ﬁxed atlas whereas mathematical manifolds are not tied to ﬁxed atlases. After these technicalities. GLOBAL COMPOSITIONS Figure 13. 13. –Σ– After the preceding propaedeutic reﬂections. we want to give a precise technical deﬁnition of a global composition built upon the objective local compositions as local charts.2. see section 13. two diﬀerent coverings change the composition qua global structure or gestalt. this is completely diﬀerent: A given covering of a composition by a determined set of charts. is the same if we add new small or large charts as long as they are compatible with the given ones. In music. we give comments on more intuitive aspects.2 What Are Global Compositions? Summary. in musicology.308 CHAPTER 13. . viewed as a mathematical manifold. built artistically from parts.

B. An A-addressed objective global composition is a covering I of G (often abbreviated by GI ) together with an equivalence class of A-addressed atlases for GI . (vi) for each couple s.G by the retrograde 5Φ ∼ ∼ Ψ has index set TΦ TΨ . A@Ft )t∈T of A-addressed local objective compositions. (v) a bijection φt : Kt → It for each t ∈ T . 0@OP ) and the evident atlas Φ via the identities φ1 : K1 → A. A@Fs ) → (φ−1 (Is ∩ It ). and the atlas corresponding to the coproduct surjection TΦ TΨ → I. Two A-addressed atlases Φ.2. . WHAT ARE GLOBAL COMPOSITIONS? Deﬁnition 36 A (global) objective composition is deﬁned by the following data: (i) A set G and a ﬁnite. If no confusion is likely. we could replace the identity chart Id : K. G.. .1. We consider the covering I = {A. 0@OP ). . A@Ft ) t s of local compositions. For instance.13. Clearly.t /1 : (φ−1 (Is ∩ It ). We select as the supporting set the support Boul of the zero-addressed local composition (Boul. .t := φ−1 ◦ φs : φ−1 (Is ∩ It ) → φ−1 (Is ∩ It ) t t s (restricted to the respective domains and codomains) deﬁnes an isomorphism φs. 0@OP ). by just naming the support set G of the objective global composition. The bijections φt (or—by abuse of language—the local compositions Kt ) are called the charts of the atlas Φ. (iii) a family (Kt . non-empty covering I of G. the induced bijection φs. (iv) a surjection I? : T → I : t → It . φ15 : K15 → H on the chart supports. we may abbreviate the entire data by saying that we are given an objective global composition G. . .12 . . Ψ for the covering I of G are called equivalent iﬀ their disjoint sum5 Φ Ψ is an A-addressed atlas for the covering I of G. i. and which express the internal structure of the series.1). . H} by the 15 subsets described in ﬁgure 13. ∼ ∼ 309 The data (iii) to (v) are called an A-addressed atlas Φ for the covering I of G. Example 19 Let us start with an illustration of this deﬁnition by an example which we have already alluded to: the internal structure of Boulez’ dodecaphonic series (see the Boulez/Webern approach in section 13. These subsets of Boul deﬁne corresponding zero-addressed local compositions K1 = (A.12 ) in that example and abbreviate OP := OnP iM od0. the gluing condition (vi) is veriﬁed. 0@OnP iM od0. .e. Evidently. . K15 = (H.1 and ﬁgure 13. K3 = (G. (ii) an address A. this is not the most economic atlas since many of the local compositions are ∼ isomorphic. K2 = (B.G → K. 0@OP ). t ∈ T such that Is ∩ It = ∅.

(C. which we distribute in three-space R3 . 8 This is a typical abuse of language: identiﬁcation of the chart with its codomain. In his treatise on harmony [479] Sch¨nberg talks about the harmonic strip9 between two chords o which have one or several tones (pitch classes) in common. Recall that N (GI ) is a union of aﬃne simplexes |σ| associated with the (abstract) simplexes σ ∈ n(GI ) such that for any two diﬀerent simplexes |σ|. For the moment. the (abstract) nerve or simplicial complex n(GI ) I of G . (A1 . the only condition is that they be distinct if their charts8 J ∈ I are.. If we view chords as being charts in the framework of harmony (see also section 13. i. we look at all triples J. it is not important where to place these points.4. 0@OP ) of this isomorphism. |J | as vertices.e. the interior of an aﬃne 1-simplex is the straight line minus the two endpoints. J |. we should immediately introduce a visualization tool: the nerve of an objective global composition. Now. 0@OP ). J |. (B.1 The Nerve of an Objective Global Composition Since the covering GI is a ﬁxed data of an objective global composition. 0@OP ).2 for these concepts. 9 “Harmonisches Band. |σ|o ∩ |τ |o = ∅. For example. however.. J with non-empty intersection.2. we start with the zero-dimensional skeleton N0 (BoulI ) which is a set of 15 points. see 13. (B1 . and therefore introduce a new chart domain (G. |τ |. one goes one step further. see 13. musicology has been aware of such a construction. In our next step. And the last step consists in joining each group of four appendix H.7 b). 0@OP ) (G.” 7 The 6 See . one for each chart J ∈ I. deﬁning a second. |J | are connected by a straight line. 0@OP ). In fact. (H. We then add 2-simplexes |J. this construction will be extended. o In combinatorial topology. 1-simplexes are precisely the formalization of Sch¨nberg’s harmonic strip. 0@OP ). see ﬁgure 13.G. the 0-simplexes |J|. 0@OP ).7 a) for this procedure.7 c). Their 0-simplexes |J|.310 ∼ CHAPTER 13. Exercise 21 Show that the relation of atlas equivalence in deﬁnition 36 is in fact an equivalence relation 13. we look at all couples J.e. Example 20 For our example BoulI . i. 0@OP ) 02 together with 15 corresponding isomorphisms ψi . a 1-simplex |J. There is no deeper reason to stop at 1-simplexes and to proceed with a test for common tones in three or more charts. 0@OP ). Given a global objective composition GI . interior of an aﬃne simplex is the simplex minus its faces. triangular surfaces with the 0-simplexes |J|.2 and chapter 27). equivalent atlas Ψ for BoulI . |J |. J . This second atlas makes evident the typology of the internal structure of Boul. We can proceed in this way until we obtain a minimal set of eight chart domains (A. This visualization idea stems from combinatorial topology. (( 1 0 ) · G. as well as its geometric realization N (GI ) are deﬁned6 . but we should look at its most elementary aspect as soon as possible. J of mutually distinct charts such that they have common elements. J . their interiors7 are disjoint. In the course of the theory. GLOBAL COMPOSITIONS isomorphism K : G → K. we are only looking at the codomains in the covering I.

Observe that in general. which is called a discrete interpretation.2.A1 1 0 0 7 ( ). a) Every chart corresponds to a point.1. The shape of the nerve of an objective global composition is a good measure for the complexity of the global conﬁguration. WHAT ARE GLOBAL COMPOSITIONS? 311 a) b) c) K·U·B A1 d) A G U·A1 B1 A1 U·A1 U·B1 ( ). There are two extremal situations of this perspective: On one hand. c) Three diﬀerent charts having common points are connected by triangular surfaces. i..7 d). distinct 0-simplexes |J|. and we essentially boil down everything to counting notes.7: Construction of the nerve of the Boulez series Boul for its internal structure deﬁned in ﬁgure 13. reduced to the 0-dimensional skeleton. b) Two points are connected by a straight line iﬀ the corresponding charts have non-empty intersection. This is quite silly (though not superﬂuous for accounting purposes).7 d). see 13. this will completely destroy its gestalt. |J |. d) four diﬀerent charts with common tones deﬁne a full tetrahedron. and the connectivity of the global construction is trivial. So the internal structure of Boulez’ series appears as a complex intertwining of local charts which overlap as shown by the geometric nerve in 13. no ﬁve charts have common notes. This happens if we just draw disjoint groups of notes on a given composition (as principally suggested by Jackendoﬀ/Lerdahl).e. and we . |J |. a nerve may be discrete.G 1 0 0 2 B K·G C H Figure 13.U. there are only isolated local objective compositions around. In this case. this procedure will not stop in three-dimensional simplexes. Only here.13. The extreme case of such a “strategy” happens if we just draw circles around every tone of a composition. |J | by a full tetrahedron if the intersection J ∩ J ∩ J ∩ J is non-empty.

Both extrema are no intelligent solutions for human cognition: Good global structures are somewhere in the middle between intractable monsters and insigniﬁcant atomized data. Suppose we are given a family of Aaddressed objective local compositions RelΦ = ((Ki. A@Fj ) → (Kj. The GTTM is built on a competence which is both. iﬀ the relation ∼ is an equivalence relation.j . and the non-empty intersections ki (Ki.j ◦ φj. GLOBAL COMPOSITIONS shall call such a global composition a silly interpretation.j (x) = y.1. Often. then the colimit.j → U . which is identiﬁed10 to U = Kj. T a set of indexes.j → U . Then the canonical maps kj : Kj. QED. and the claim follows. A@Fj ))j∈T . together with a family IsoΦ = (φi.j . formally and semantically. Let U be the colimit of the system of set maps φi. i.j is non-empty.j /1 : (Ki.j . A@Fi ))i. the relation ∼ is an equivalence relation. On the other hand. also since classiﬁcation of large local compositions is quite intricate (see our discussion in chapter 11).j deﬁned by x ∼ y iﬀ x ∈ Ki. (Kj. we have inclusion morphisms ρi.j for some index couple i.312 CHAPTER 13.j .j . more specially. Theorem 12 Let A be an address. and φi.j .j∈T ∼ (13.j . φi.2) of isomorphisms in ObLoc whenever Ki. φi. And we assume that each (Ki. A@Fj ))j∈T and intersections (Ki. Proof. This will be a very diﬃcult. cryptic object. deﬁnes injective maps (!) kj : Kj.i = Id. A@Fi ) which are isomorphic under the morphisms of the system IsoΦ. an objective global composition is not given in advance. conversely.j ) correspond to the local subcompositions (Ki. and in this case. . out of the reach of the GTTM. 10 See appendix G. we may just take one single local chart to cover the entire composition. and therefore boils down to an interface between psychology and music theory which does neither solve the formal inconsistency of music theory nor observe the variability of the psychological grouping: a bad glue of two unresolved components. But it would be nonsense to set up artiﬁcial limits (like Jackendoﬀ/Lerdahl [243]) to the grouping formalism because it is one of the most important objectives to make all grouping perspectives accessible and formally tractable. and that for all indexes. Consider the binary relation ∼ on the disjoint union of the sets Kj. If the canonical maps deﬁne an A-addressed objective global composition with atlas ((Kj.2.j∈T of forms Fj . A@Fj ))i.i = Id. but only results from a compatible gluing of local charts. we have a huge set of notes which does not observe all the more local similarities or symmetries or other relations. If. A@Fj ).i . The breakdown of the Jackendoﬀ/Lerdahl Generative Theory of Tonal Music (GTTM) happens where their grouping levels have to rely on traditional music theory in order to work.. and.j and ρi.j → U deﬁne an A-addressed objective global composition with atlas ((Kj.e.j / ∼. A@Fj ).j . The next theorem describes a necessary and suﬃcient condition for a system of local compositions to become an atlas for a global composition.j . A@Fj ) is a sub-composition of the diagonal element (Kj. then the relation ∼ is the inverse image of the equivalence relation of equality on U for the surjection Kj.i .i ) ∩ kj (Kj.j .j ⊂ Kj.j /1 : Ki. A@Fj ) on the charts.1) (13. j. we call it the indiscrete interpretation.

c3 . 1) in R2 and set Ki. K3.j = ∅ else. not refrain from naming or sweep names away in favor of anonymy. In fact. 0). 0@R2 ).2 = K2. 0@R2 ). 0). i = 1. 2.i = ({c1 . Ki. K1. the ambient spaces are no longer uniquely determined. The index set is T = {1. c4 }.1 .2 = ({c1 . On the level of local supports. c3 = (1. univocal naming would not be a good idea since on the overlaps of two diﬀerent covering sets. In fact. 0@R2 ).13. what name should one select? Homonymy is unavoidable for global compositions.2.1. The concept is a proper extension to the local framework. Global naming is interchangeable.3 = ({c3 . Exercise 22 Let us deﬁne a zero-addressed objective global composition by data corresponding to the above theorem. c4 }. the construction of a global composition makes chart names exchangeable: it is ‘homonymous’. c2 = (0. however. 1). φ1. c2 }. 3.3 = IdK3.2 = e1. More precisely. c4 }. φ2. We take a constant space form Fi = Onset ⊕ P itch with functor @R2 .1 · −1 0 0 −1 . we have as many names for a given point x of the support as we have chart names on any of the compatible atlases. K2.3 = φ3. We shall develop necessary and suﬃcient criteria for covering constructs in section 16. We should observe a hidden subtlety of the global composition context: Global ambient spaces. But if we are given an atlas of local compositions.1 = ({c3 .1 . c2 . 2.1 = ({c1 .1 that this global composition cannot be constructed by a covering of a local composition. but not irrelevant: Deﬁnition 37 The name N ame(x) of an element x ∈ G of an objective global composition G is the set of names of all the local chart forms (in any compatible atlas) which hit this element. WHAT ARE GLOBAL COMPOSITIONS? 313 Observe that objective global compositions need not be derived from given local compositions by coverings as in the above Boulez example. 3}. c4 = (1. every local composition is connected to an ambient space form F .3 = K3.2 = IdK2. 0@R2 ). Show that the conditions of the theorem are satisﬁed and try to draw a picture of the global composition. 0@R2 ). The family IsoΦ is given by φ1. One should. We shall see in section 16. c2 }. and the general construction method is described by the above theorem. So the name of the ‘ambient space’ of x is an entire collection of local chart names. The family RelΦ of local compositions is deﬁned as follows: Consider the four (zero-addressed) points c1 = (0. K1. .

the (functorial) local compositions in the category Loc. (v) an isomorphism of functors φt : Kt → It for each t ∈ T . Example 21 The ﬁrst and immediate example for this deﬁnition is the functorial global comˆ position G associated with a given objective global composition G. (ii) an address A. (vi) for each couple s.3 Functorial Global Compositions Summary. (iii) a family (Kt . the induced isomorphism φs.e. An A-addressed functorial global composition is a covering I of G (often abbreviated by GI ) together with an equivalence class of A-addressed atlases for GI .314 CHAPTER 13. The bijections φt (or—by abuse of language—the local compositions Kt ) are called the charts of the atlas Φ. The data (iii) to (v) are called an A-addressed atlas Φ for the covering I of G. i. by just naming the support G of the global composition. If no confusion is likely. @A × Ft ) t s of functorial local compositions. and the atlas corresponding to the coproduct surjection TΦ TΨ → I. non-empty. . 11 Φ ∼ ∼ Ψ has index set TΦ TΨ . @A × Fs ) → (φ−1 (Is ∩ It ). in particular we have an atlas (Kt . Ψ for the covering I of G are called equivalent iﬀ their disjoint sum11 Φ Ψ is an A-addressed atlas for the covering I of G. To construct this object. Corresponding to functorial local objects. @A × Ft )t∈T of A-addressed functorial local compositions. –Σ– The deﬁnition of functorial global compositions is the transposition of the objective case with everything made functorial: Deﬁnition 38 A (global) functorial composition is deﬁned by the following data: (i) A functor G ∈ Mod@ and a ﬁnite. Two A-addressed atlases Φ. if the latter has some adequate properties. A@Ft )t∈T of local compositions. i.t := φ−1 ◦ φs : φ−1 (Is ∩ It ) → φ−1 (Is ∩ It ) t t s (restricted to the respective domains and codomains) deﬁnes an isomorphism φs. (iv) a surjection I? : T → I : t → It ...t /1 : (φ−1 (Is ∩ It ). generating set I of subfunctors of G. suppose that we are given an A-addressed objective global composition G with the notation and data of deﬁnition 36. composed of (functorial) local compositions. t ∈ T such that Is ∩ It = ∅. GLOBAL COMPOSITIONS 13. G = I.e. global functorial compositions are introduced as global objects by an atlas of charts. we may abbreviate the entire data by saying that we are given a global functorial composition G.

as well as a system of subcomposition inclusion monomorphisms ˆ ˆ ρs. the canonical squares ρt. since the evaluation of the inductive system of functors at the morphism IdA (the slice at IdA ) yields the ˆ original objective conﬁguration. the maps A@Fs → B@Fs are injective for all indexes s.x. ˆ In general.t ˆ ˆ ˆ of functorial local compositions. @A × Ft )t∈T of functorial ˆ local compositions.e. for example.t /1 : Kt. although the chart morphisms are compatible in the sense of fulﬁlling the conditions of theorem 12.t = φ−1 (Is ∩ It ) for any two diﬀerent indexes s. ˆ deﬁned by the inverse images Ks. x@Ks. the generators It are diﬀerent iﬀ the original covering elements ˆ ˆ ˆ It are.s − − → −− ˆ ρs.3. @A × Ft )t∈T . t.t ˆ ˆ Ks i s (13. an atlas (Kt . if A = 0Z . in fact. @A × Ft ) ⊂ (Kt . we have the formula ˆ ∼ ˆ ˆ x@G → colim(x@Ks . injective for all addresses. This means that we will not be able to extend the gluing procedures to the ambient spaces. this completes the construction.t . the slice x@G is also denoted by G. or if Fs is constant. @A × Ft ). Exercise 24 Establish a theorem for functorial compositions which corresponds to theorem 12.t : (Ks.t satisﬁes the conditions of theorem 12. Then it follows that the system of isomorphisms and inclusions φs. ρs. Because of the right colimit expression.t . Let G be the colimit functor of the system φs..t ) = colim(Ks . we have a family of functorial local compositions (Kt .4) i ˆ ˆ Kt − − → G −t − ˆ ˆ ˆ are cartesian. i. the canonical morphisms ˆ t → G are not even injective. and a surjection I? : T → I. Ks.t .s ˆ ˆ Kt. we suppose that for all address ˆ K changes f : B → A. G is not a global functorial composition. FUNCTORIAL GLOBAL COMPOSITIONS 315 ˆ To begin with. So we have obtained a ﬁnite covering I of G. Exercise 23 Explicate this construction in all the details. This is the case.3) ˆ on the slices of the colimit.t .t of isomorphisms and inclusions. ρs. no compatibility of the underlying functor morphisms on the ambient spaces is required.s → Ks. and therefore.t ◦φs. in other words: global objects are really global. To obtain a global composition.13.x.x) (13. For an address change x : B → A. Therefore G is generated by the isomorphic images It of Kt . Further. the canonical morphisms ˆ ˆ it : Kt → G are mono. no common ambient space is available in general! . Remark 3 Observe that. We also t have induced isomorphisms ∼ ˆ ˆ ˆ φs. Furthermore.

they are not ‘isomorphic’ to any possible interpretations (we shall make these remarks precise later). But observe that the intersection Kι1 ∧ Kι2 is not objective in general. i. Here is the general construction method of such global compositions which we therefore call interpretations. Then the covering K I . We shall therefore introduce a special construction method of global compositions which take into account this important technique as an interpretational basic. Boulez’ series of example 13. music analysis and composition explicitly deal with local compositions. but this coarse perspective is not suﬃcient: implicitly. A@F )i∈I . We give standard examples of such global objects. however. A@F ) be an A-addressed objective local composition. together with the obvious atlas (i. GLOBAL COMPOSITIONS 13. this type of chart speciﬁcation is not temporary. . Then we have an interpretaˆ ˆ ˆ tion I = (Kι )ι of the local functorial composition K since unions and ˆ commute by lemma ? ˆ ˆ 2.e. 466]) also use this kind of local part selection. deﬁnes an objective global composition. For example. As already stressed. @A × F ) be an A-addressed functorial local composition. 12 The functor K is generated by the subfunctors in I.g. This construction is a subtlety that tends to escape to common interpretative activities.316 CHAPTER 13.4. Then the functorial covering K I . It is. Many analytical texts (e. together with the obvious atlas (i.1 needs a covering by 15 local parts in order to be fully understood. A@F ) be an objective local composition. –Σ– Very often. and there. Let (K. and let I be a non-empty covering of the support set K. one speciﬁes local parts of this data and diﬀerent overlapping relations among such parts. a basic prerequisite for every interpretative activity in music. Let (K. the interpretation K I of K associated with the covering I. This is why this type of global construction is called interpretation. This is however the case for the zero address A = 0Z . the colimit construction and this one coincide. [243. We shall see in chapter 16 that interpretations deﬁne a special case: Global compositions may be far from ‘interpretable’. @A × F )i∈I . and let I be a nonempty covering12 of the support functor K. section 7. and take an interpretation K I by an atlas I = (Kι )ι of subcompositions. 393. Deﬁnition 39 Let (K. A large set of (objective) global compositions is constructed by “interpretations” of given (objective) local compositions. the interpretation K I of K associated with the functorial covering I. deﬁnes a functorial global composition.. Example 22 A second example of functorial global compositions is associated with interpretations of objective compositions. it is a substantial attribute of the analytical work.4 Interpretations and the Vocabulary of Global Concepts Summary.

. . We then introduce this sequence of forms: Gi+1 −→ 2F un(Gi ) ΩF un(Gi ) G0 = F. A very classical ‘inﬁnite interpretation’ is the one induced by the nerve of an interpretation. .e. . I. and then inductively wt+1 (x) = u∈stt+1 (x) w(u). we can give an interpretation by a covering set I of subcompositions (Kι . the star of x st(x) = {U ∈ I with x ∈ U } is an element of n(I). and the star st2 (x) = st(st(x)) of st(x) is an element of n3 (I). In particular. if x ∈ K. taking the covering I of K. . if x ∈ K. 2.) which indeed is in Limint(F ).. together with coverings I of K. A@F ) is an A-addressed objective local composition. . Here. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 317 13. the union operator is deﬁned functorially as a transformation i : Gi+1 → Gi . . In fact. Interpretation is not a one-step process. This deﬁnes a sequence n∞ (K I ) = (K. Power(Gi ). Then. i = 0. Then. it may be iterated on an inﬁnity of levels.. sti (x) ∈ n1+2i (I). . I = K. i. . . Then each covering set U ∈ I can be given a numerical weight. Suppose that each element x ∈ K is given a weight w. in other words. an inﬁnite sequence of interpretations of constantly increasing level of complexity. .13. Of course there are inﬁnitely many such denotators even if we ﬁx an initial sequence of ﬁrst k positions. etc. we get an initial sequence (K.. with the initial evaluation 1 (I) = K.1 Iterated Interpretations Summary. I. if we start as above. if we take the above formula w(u) = ( x∈u w(x) − 1)p to calculate recursively the weight of elements of Gi . We present the formal framework and examples. .4. The simplicial complex n(I) is an element of G3 and can be iterated ad libitum by nt+1 (I) = n(nt (I)).). This associates a power series λw (x) = 1 + 1≤t wt (x) t T . –Σ– If (K.5) whose denotators are local compositions K as above. n(I). nt (I). .) and we can deﬁne a corresponding limit form Limint(F ) −→ Limit(D) Id (13. the interpretation I is an A-addressed objective local composition in G1 whose union is K. So we have an inﬁnite diagram 1 2 3 D = (G1 ← G2 ← G3 ← .2) or some other musically motivated evaluation. A@F ) of K. . i = 1. . n2 (I).4. for example a power w(U ) = ( x∈U w(x)−1)p as in the metrical theory (see section 21. t! . This means that we are given an inﬁnite succession of interpretations and interpretations of interpretations etc. and coverings J of I etc.. . . for example the constant weight w(x) = 1 or some weight stemming from an analysis of K as in motivic or metrical theory. we have a weight w1 (x) = w(st(x)) = U ∈st(x) w(U ). .

–Σ– Recall from section 7. . consider the two standard cases: 12-tempered and just scales. 13. The reﬁnement speciﬁes determined notes or groups of notes which give the scales an interior proﬁle. 1) log(3) M12 m12 mjust = e (-1. see left half of ﬁgure 13. Sk } in P iM od12 or in EulerP lane. Here.0) . For the 12-tempered case. we have the major third M12 = e4 and the minor third m12 = e3 translation.1 that we may identify an octave-periodic scale S by its class chord S : A o-Scale(S1 . Triadic and Quaternary Degrees Summary. .8: Left: Major and minor third transpositions in just tuning pitch classes. 1) Figure 13. . Sk ). x mod 3). If the pitch class space is identiﬁed with its Sylow decomposition: ∼ @Z12 → @(Z4 ⊕ Z3 ) : x → (x mod 4. respectively. To understand this construction. Ecclesiastical Modes. This means that inﬁnite interpretations can be boiled down to yield numerical ‘coordinates’ of points in compositions. .8 for the torus representation of P iM od12 . See right half of ﬁgure 13.2. GLOBAL COMPOSITIONS The reader may easily generalize this idea to obtain a numerical evaluation of points in inﬁnite interpretations.8.4.2. whereas m12 identiﬁes to e(−1. log(5) Mjust = e(0.318 CHAPTER 13. The construction deals with “reﬁnement” of scales qua local compositions. Many common scales and chords are classiﬁed by use of the so-called third chain construction.2 The Pitch Domain: Chains of Thirds. .1) and mjust = e(1. and therefore quantitative measures for further processing within performance or analytical contexts. we have the corresponding translations Mjust = e(0. . With these translations in mind.1). we are eﬀectively working in the ambient pitch class space P iM od12 with module functor @Z12 . More elaborate interpretations of scales concern triadic and quaternary degrees for concerns of harmony. Right: The third transpositions on the third torus representing 12-tempered pitch classes. M12 identiﬁes to M12 = e(0. we can deﬁne a third chain: Deﬁnition 40 A chord S = {S1 . of cardinality k is called a third chain iﬀ the elements of S can be ordered in such a way that .−1) .1) . Chains of thirds and modes are a classical domain of interpretation. On the EulerP lane ambient space (see example 5 in section 7.

the set of third chains containing a given zero-addressed chord S with minimal cardinality are called the minimal third chains of S. the mode . and augmented. but not necessarily in a third chain. To meet this requirement. The point here is less the tuning but the construction of an interpretation from a given scale (chord). the chromatic 3-chord S = {0. 2} has 23 minimal (7-element) third chains.2.13.1). If we admit positive and negative translation in the minor third direction mjust . This set can grow quite dramatically for nonstandard chords. and this set is called the third chain closure of S. Lemma 16 Every zero-addressed chord in P iM od12 is contained in a third chain. . INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 319 Sij+1 = Tj (Sij ) for all j = 1. Lemma 17 Every zero-addressed chord in EulerP lane is contained in a weak third chain. Exercise 25 Give a proof of the preceding lemmata. we obtain the concept of weak third chains in just tuning (see also right half of ﬁgure 13. For example.4. The (objective) interpretation 0@P iM od3Chains of 12 0@P iM od12 by the covering 3Chains is called the third chain interpretation of the pitch class space. such as selected degree chords. the scale and the tonic singleton.3). it is denoted by 3Chain(S). respectively. According to the position of the tonic. we make use of the 12-tempered space P iM od12 . the enrichment of the scale by a determined tonic f .. This is an interpretation Xf = X If of X by an atlas If of two charts: I = {X. 4. 7. The ﬁrst interpretation is the mode Xf . the other modes are be derived by evident transposition. The set 3Chains of all zero-addressed third chains in P iM od12 is known and has been used in computer software presto (see section 25. See appendix L.8). etc. 2. i. as well as in RUBATO ’s HarmoRubette (see chapter 41. Mjust ) or the minor major translation m12 or mjust . 11}. {f }}. although other pitch class spaces would also do the job. In order to describe ecclesiastical modes. minor.2 for this list. 5. 1. a mode may and will be enriched by other structural aspects. as well as the ﬁnalis or ﬁnal f . 9. In the 12-tempered case. . To describe modes. including third chains which start with three of the classical triads: major. .e. Observe that this is not a full-ﬂedged concept of tonality. one selects two elements in X: the tenor or recitation tone t. k − 1 and Tj is either the major third translation M12 (resp. Our modeling of ecclesiastical modes will follow the (zero-addressed) scale X = C-major = {0.

CHAPTER 13. but they are the basis of the modern major-minor system where the tenors have disappeared (and we only consider the modes Xf ). 6. the ﬁnal tone being renamed to tonic.p in table 13. interpretations are important which associate 3-chords with special harmonic functions. the octave where the modal melody may move. We are going to deﬁne the triadic degree interpretations for X = C.a 2.p 6.t = {X.a 1. i. however.4.4. For the authentic mode we have either t = f + 7 or t = f + 8 (sixths). Observe however.2.p Table 13.a 3. that the name of such a scale. In harmony. we have either t = f + 3 or t = f + 4 or t = f + 5 (thirds or the fourth)..A mode and plagal variant ﬁnalis f Dorian Hypodorian Phrygian Hypophrygian Lydian Hypolydian Mixolydian Hypomixolydian Aeolian Hypoaeolian Locrian Hypolorian Ionian Hypoionian d d e e f f g g a a b b c c tenor t a=d+7 f =d+3 c=e+8 a=e+5 c=f +7 a=f +4 d=g+7 c=g+5 e=a+7 c=a+3 g =b+8 d=b+3 g =c+7 e=c+4 The next step introduces an ecclesiastical mode13 as a reﬁned version of the simple mode interpretation: An ecclesiastical mode on scale X is the interpretation Xf. 1. {f. the latter being marked by the preﬁx “Hypo”. for the plagal one. 13 The ambitus.a 5.4). as well as the melodic and harmonic minor scale xm and xh (see ﬁgure 7. The modal pair Nr. The name is only historically loaded and not in our structural setting. we shall discuss them in the sequel for the 12-tempered and just tuning. such that we now have aeolian =cantus mollis (Ca for scale C). they are scarcely documented.A) and the plagal (.e.p 2. GLOBAL COMPOSITIONS Nr.2.2. {f }.a 7. does not imply the selection of a tonic.a. we again take the major scales X. For every ﬁnal f .4. and ionian = cantus durus (Cc for scale C). is omitted here because we work in octave classes. there are two modal variants: the authentic (.a 6.A).p 7. .a 4.t with atlas If. Corresponding interpretations of scales are called triadic interpretations. 6.p are listed for completeness.A. The aeolian and ionian modes were only introduced in the 16th century by Glarean.p 4.p 5. the others being deduced by transposition. t}}.4.2. for example.a in table 13.320 has the names as listed in table 13. For 12-tempered pitch classes. We shall see below that the examples to be discussed here have special automorphism properties. X = C. This is why we call them triadic degree interpretations.t = X If.p 3.

{II ↔ V }. More precisely. dim. .e. Referring to the automorphism groups of these scales. minor. b. •◦◦•◦◦◦•◦◦◦◦ ◦◦•◦◦•◦◦•◦◦◦ ◦◦◦•◦◦◦•◦◦◦• •◦◦◦◦•◦◦•◦◦◦ ◦◦•◦◦◦◦•◦◦◦• •◦◦•◦◦◦◦•◦◦◦ ◦◦•◦◦•◦◦◦◦◦• Observe that not all major. aug. {III ↔ IV }. The harmonic strip has seven simplexes of dimensions zero and two. dim. maj. Following Sch¨nberg’s proposal [478]. maj. {V II} xm : {I ↔ V }.13. and fourteen simplexes of dimension one.B cm − •◦•← •◦•◦•◦•◦• | = Ug tpe. min. we have these degree orbits in the three scales: X : {I ↔ V I}. The arrows in row two are for the alteration shifts in the melodic and harmonic minor scales against the major scale. IIch . {V II} where the third is so since the automorphism group of harmonic minor is trivial. the triad {a . •◦◦•◦◦◦•◦◦◦◦ ◦◦•◦◦•◦◦◦•◦◦ ◦◦◦•◦◦◦•◦◦◦• •◦◦◦◦•◦◦◦•◦◦ ◦◦•◦◦◦◦•◦◦◦• •◦◦•◦◦◦◦◦•◦◦ ◦◦•◦◦•◦◦◦◦◦• tpe. min. For example. | = Ud deg. or augmented triads in these scales are automatically degrees. . . VC . etc. (3) (3) The geometric nerves N (X (3) ). cm is deﬁned by (3) seven charts (3m ) = {Icm . The 1-skeleton is exactly Sch¨nberg’s harmonic o strip. . V IIC } 321 of three-element charts.2. I II III IV V VI V II •◦◦◦•◦◦•◦◦◦◦ ◦◦•◦◦•◦◦◦•◦◦ ◦◦◦◦•◦◦•◦◦◦• •◦◦◦◦•◦◦◦•◦◦ ◦◦•◦◦◦◦•◦◦◦• •◦◦◦•◦◦◦◦•◦◦ ◦◦•◦◦•◦◦◦◦◦• ch − − •◦•← ← •◦•◦••◦◦• tpe. dim. d} is not a degree in harmonic minor ch . IIC . V IIcm }. min. {V I}. The precise values for all our charts are shown in table 13. N (xh ) of each of these global objective zero-addressed compositions are M¨bius strips. {III}. {V }.. .2. dim. writing simply I for the ﬁrst degree. xh is “rigid”. we observe that the degree atlases (3). {IV }. (3m ). V → I → IV → V II → III → V I → II (→ V ). The non-trivial automorphisms of the major and melodic minor scales are shown with the notation Ux = inversion (German “Umkehrung”) at pitch x. {III}. min. Table 13. {V I ↔ V II} xh : {I}. dim. min. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS The triadic degree interpretation C (3) is deﬁned by a seven element atlas (3) = {IC . (3h ) are all invariant under the respective automorphism groups. V IIch }.9. {II}. The order of the degrees on the strip’s boundary is the so-called ﬁfth sequence. we omit it. as listed in the chord classiﬁcation table appendix L. x + 1. min. i. aug. IIcm . (3) C •◦•◦••◦•◦•◦• autom. The analogous notation works for minor scales. IIIC .1. {II ↔ IV }. min. IVC . and ch by (3h ) = {Ich .4. V IC . maj.4. If ever the scale is clear. maj. and Ux/x+1 = inversion between neighboring pitches x. maj.4. maj. as shown in ﬁgure 13. N (xm ). o o we call it the harmonic strip. maj. .B. . diminished.

see ﬁgure 13. 13.4.2. N (xh ) is a M¨bius strip. we observe that the problem here is not so much one of tunings. N (xm ). although this is only a diminished ﬁfth (tritone) for the seventh-fourth passage.9: The nerve N (X (3) ). as was discussed by Carl Dahlhaus in [100]. Before starting the discussion.1 Orientation in Riemann Function Theory We should add a remark on the failure of Riemann’s attempt to build a global function theory of harmony. GLOBAL COMPOSITIONS II IV VII V VI I triads III Figure 13. the boundary is connected.322 CHAPTER 13. but rather one attached to the non-orientability of the triadic interpretation . The central concept of this theory is the function which is to be attached to every (!) possible chord. In other words. unlike with a normal strip where we have two connected components. This property has consequences for Riemann’s harmony. The mutual position of the boundary changes after a round-trip. (3) (3) Figure 13. Following Sch¨nberg’s proo o posal.10. we call it the harmonic strip. not only to the chords of common usage.10: The harmonic strip is not orientable. This is due to the lack of orientation on the harmonic strip: One walk around the entire strip changes your upside to downside.

each. which can be attributed to chords. ﬁfth. i. + τ (IV ) = S are diﬀerent according to the ﬁfth steps V → I → IV . + τx (VX ) = D. for each pitch class x: + 0 τx : Ch → T DS (x-major tonality) τx : Ch → T DS (x-minor tonality) According the the selected tonality. and “Subdominant” (S). this is the ﬁfth sequence. This concept is an obscure one. and minor.: + τx (IX ) = T. p.13. This procedure can be termed “musical logic” in the sense of Riemann. Riemann’s idea was to deﬁne “tonality” by use of a function with three possible values:“Tonical” (T ). but this is the equivalent to ours since the sequence is cyclic. the fundamental sequence along the harmonic strip’s boundary14 . “Dominant” (D). as Dahlhaus has criticized in [100] relating to an erroneous proposal. as Dahlhaus has rightly recognized. For example. d} is dominant in C-minor tonality”. b. Rather is the problem of function theory to extend the values of tonality functions from common ﬁrst. S} of harmonic function values. d}) = D which means “{g. + τ (I) = T. we have to distinguish between two sub-categories of tonal functions: major. When they deﬁne tonality.” According to Dahlhaus. 14 Dahlhaus starts the sequence in degree I. Therefore we want to stick to the harmonic strip as discussed above (this nerve will reappear for the triadic interpretation in just tuning which we discuss below). Moreover. to make the ideas precise). b. However. and fourth degree triads to any chords in such a way that the harmonic coherence is reﬂected. or 0 τc ({g.e. So we have to deal with a function τ : Ch → T DS deﬁned on the set Ch of all (zero-addressed objective) chords (in P iM od12 . der ‘harmonischen Logik’. This sequence lays the basis of the idea of “diﬀerent” (German:“diﬀerente”) degrees: the values + τ (V ) = D. d}) = T which means “{g. + τx (IVX ) = S. b. d} is tonical in G-major tonality”.4. with values in the three-element set T DS = {T. special chords of the triadic degree interpretation X (3) are given special values . D. ist also mit einer Regel uber ¨ die Reihenfolge der Stufen verbunden. In order to proceed. b.96]:“Die Bestimmung der Akkordbedeutungen. one would like to have + τg ({g. it is not true that the function concept is not a mathematical one. It is however possible to shed some light on this approach [100. . Riemann’s program is to attribute a speciﬁc value to given chords. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 323 independently of tuning speciﬁcations. the other four triadic degrees must obtain one of the existing values.

For example. the parallel degree is πY = III. the phenomenon is founded in a very precise fact: the non-orientability of a the M¨bius strip.324 CHAPTER 13. The orientation being only a local one. In fact. On the harmonic strip.11): Let δY be the successor of Y in the ﬁfth sequence. If we move along the harmonic strip in such a way that the strip stays to the right when moving from Y to δY . We also have πI = V I. das sie erkl¨ren soll. then πY is to the right in front of us. The second musicological idea relates every degree Y to its “parallel degree” πY . wo auch das Ph¨nomen. The wording “parallel” is also geometrically correct since we look for the degree which is parallel to the present position. This contradiction can also be read as follows: When applying the parallel function to every degree on the harmonic strip.11: The local orientation on the harmonic strip yields the parallel degree as the one staying in front and to the right when moving in ﬁfth sequence direction such that the strip stays to our right. . we then must have D= + τ (V ) = + τ (πV ) = + τ (III) = + τ (πIII) = + τ (I) = T.” In reality. p. if we move from Y = I to δY = V . the musicological requirements can only be fulﬁlled locally and do not glue to yield a global function! Therefore we cannot follow Dahlhaus [100. ins Vage und Unbestimmte a a ger¨t. GLOBAL COMPOSITIONS Y dY pY Figure 13. the latter is deduced from the ﬁfth sequence in the following way (see ﬁgure 13. and therefore the tonal values cannot be diﬀerent on the vertices of a 2-simplex. the connectedness of the boundardy (=the ﬁfth sequence) leads to the parallelism ππY = δY (the “Gegenklang” relation). The contradiction in function theory comes out from the requirement in function theory that parallel degrees should have equal function values: + τ (Y ) = + τ (πY ). Then. πIV = II. πY is the third member of the 2-simplex containing Y and δY . Any attempt to deﬁne a function despite this fact must fail for mathematical o reasons.102] when he says that function theory: “gerade dort versagt.

The triadic degrees are not completely clear here.13. and these have an empty intersection. With such a construction. In fact.e.2 Just Triadic Degree Interpretations 325 For this situation.12. see ﬁgure 13. The problem is that one would like to respect certain pitch relations. VII with I =]I[= {]c[.2. The second degree now can be realo ized by selecting one variant of the third comma equivalent just degrees (tone by tone.4. II ∗ = {d∗ . ambiguities from just tuning disappear. f ∗ }. V II ∗ = {b. i. The justest scales can then be deﬁned as local compositions by the quotient module JKt = Z2 /ZKt. If we want to have a minor chord for degree II which has two tones in common with degree IV . with ambient space @JKt .2. if we want it to be built from two minor thirds d − b and f − d. In this setup. V II. and thereby obtain the interpretation C(3) with degrees I. recall the just C-major scale in the EulerP lane from section 7. IV. VI. In the same sense.12: The list of just triadic degrees. d∗ − d = Kt. Except for degree II. a}. ]e[. we need to take the tone d∗ = −2q + t instead of d = 2q. see also ﬁgure 13.2. this group is generated by .33) for the deﬁnitions. not as a whole!). see formulas (6. we may transport the triadic degrees of just tuning C (3) to C-major in justest tuning .. II. We then have the alternative degrees I II II* III IV V VI VII VII* Figure 13. f. and the nerve N (C(3) ) is a M¨bius strip. In order to harmonize this irritating situation—which stems from pitch-distance requirements on degrees—one proceeds as follows: We no longer work with pitch classes x but with third comma classes ]x[= x + ZKt. refer to the second f -variant f ∗ = f − Kt. For the C-major scale. a space which can be identiﬁed with the Pythagorean tuning subspace @Zq in the EulerP lane. etc.13. d.4. the third comma—more precisely: its pitch class. in contrast to the canonical degrees II. V II we obtain major or minor chords..e. together with the variants II ∗ . ]g[}. the seventh degree must. we have Z2 = Zq ⊕ ZKt. III.1. i. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 13. V II ∗ which meet standard requirements of interval distances. The atlas of degrees of a just triadic degree interpretation also remains invariant under the automorphism group of the just scale. V.

We describe macros for rhythmic germs.2. the skew reﬂection A = eq · −1 −1 0 −1 (see also example 11 in section 8.13: The third (or Pythagorean) comma classes cover all just pitch classes when starting from the Pythagorean subspace. {IV ↔ V }.14. We shall see however in chapter 27. We would like to include formally the . {III ↔ V I}.6 that modulation in just tuning is also easily modeled without necessarily building justest tuning constructs. {II ↔ V II}.2 that a local rhythm is deﬁned as a union of translates of a local ‘germ’ composition in an ambient space Onset × P ara. we want to give an account of global time constructions deduced from this type of phenomena.1. diﬀerent local metrical structures mostly coexist in the same portion of music. 13. In this section. Global meters and rhythms as interpretation by atlases of local rhythms and meters.1. –Σ– In music.4. Recall from section 7. one is allowed to select any one of the possible representatives within one comma class. GLOBAL COMPOSITIONS ]x[ third comma Pythagorean tuning subspace Figure 13. Third comma classes solve the problem of the second and seventh degrees. The orbits of the degrees look as follows: {I}.326 CHAPTER 13.1). as for example illustrated in ﬁgure 13. Exercise 26 Verify the invariance and the orbit structure as described above.3 Interpreting Time: Global Meters and Rhythms Summary.

4.2. Deﬁnition 41 With the above notation.14: Example of a simultaneous presence of local meters in Beethoven’s op. We have an obvious projection endomorphism pmeter : Rhythm(P ara) → Rhythm(P ara) which is the identity on cofactor Onset and projects onto the ﬁrst factor on R(P ara).b]p G. in the sense of section 7. the projection pmeter (L) is called the associated metrical rhythm. The right hand plays eight quavers while the left hand plays six triplets per meter. The ﬁrst requirement is met by the form Rhythm(P ara) −→ Colimit(Onset.106. bars 209-210. For a local composition L in Rhythm(P ara). on the other. A local P ara-rhythm deﬁned by a singleton germ G = {g} is called a local P ara-meter. Allegro. a local (A-addressed. we have the evident translation action ep on Rhythm(P ara). objective) P ara-rhythm is an A-addressed local composition R deﬁned by R= t∈[a.b] etp G and denoted by e[a. case where the rhythm is a local meter (empty parameter set for space P ara) on one hand. and where the parameters may vary. and the intersection L ∩ Onset is called the metrical component of L. P ara). a local P ara-meter is again called a local meter if its . and therefore on the local P ara-rhythms. This means that an objective A-addressed local comId position Rh in ambient space Rhythm(P ara) is a subset of A@Onset A@Onset × A@P ara (with corresponding functors of these forms). Conformal with the known terminology.13. For any period p : 0 Duration(p). b] of extended natural numbers. R(P ara)) Id with R(P ara) −→ Limit(Onset. for a rhythm germ G ⊂ A@Rhythm(P ara) and an interval [a.2. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 327 Figure 13.

So g. This implies that either g. periods. Let R = [b. h are both onsets. while b = l(M ) its length. . we may assume the ﬁrst case. the empty germ will do it).b]∆ ω. it is said to be (P ara-)metrical iﬀ every chart is a local (P ara-)meter. Suppose that R ∩ S is not empty (otherwise. Proof: Exercise. The second requirement is met by suitably generalizing the form P ara: Suppose that we want to take local rhythms from diﬀerent parameter spaces P ara1 . For a local P ara-meter M . suppose it lies to the right. i. Lemma 19 For two local A-addressed P ara-meters R. Clearly. These deﬁnitions englobe Vuza’s concepts of rhythms and of canons developed in his works [552. P aran . we have ∆ = (s − s∗ )p = (t − t∗ )q. With this terminology.b]∆ ω. η ∈ R. Then we evidently can increase b until we reach y. . P ara2 . h have the same linear parts and if their translation parts are γ. So. So the intersection is this local meter. Take any element y = s∗∗ p + γ = t∗∗ q + η in R ∩ S − e[a. So the point lies to the right or left of the set e[a. 15 WLOG = without loss of generality. g]. and take x = esp g = etq h. g]. GLOBAL COMPOSITIONS germ lives in the cofactor Onset and is equal to its metrical component. we abbreviate this object by M = [b. We may suppose that ∆ is the smallest possible positive number of this type. there are such numbers a. . Else. with nonnegative b. respectively.e. and origins. . A ﬁnite local P ara-meter M with germ {g} may be written as M = e[0. q. Then we may collect these options to P ara −→ Colimit(P ara1 . and this is a contradiction. . 556. we are done with length zero. we shall always assume the period to be also zero.328 CHAPTER 13. . then l(R) ≤ l(S). For length zero. A-addressed local composition G in ambient space Rhythm(P ara) is said to be (P ara-)rhythmical iﬀ every chart is a local (P ara-)rhythm. g is called the origin onset. h ∈ A@R are A-addressed onsets. it is called local rhythm if its germ lives in the cofactor R(P ara). 0] is in R∩S. WLOG. Then we claim that the maximal integer interval a ≤ b such that e[a. the interval [0. if R ⊂ S. 555. or they have the same P ara-component. Proof. h] be a representation by germs. 554. S = [c. g. WLOG15 .b]∆ ω ⊂ R ∩ S equals R ∩ S. the now uniquely deﬁned period of a local P ara-meter M is denoted by p(M ). QED. Take a maximal interval. S is a A-addressed local P ara-meter. b. If there is no other common point. It is evidently unique. S. Id and the second problem is also solved. we have the equation sp+γ = tq+η of real numbers. It cannot lie between two elements of the local meter e[a. P aran ). if there is another common value ω = s∗ p + γ = t∗ q + η. 557]. p.. Lemma 18 The intersection of any two A-addressed local P ara-meters R. the interpretation GI of an objective.b]∆ ω since this would contradict minimality of ∆. p. . P ara2 .b]p {g}.

Observe that by lemma 19. We deﬁne the inverse images levi = lev −1 (i) by recursion: • levo = M ax(I). Its charts are called the canonical (local A-addressed) P ara-meters of X. This interpretation is denoted by X M etP er[P ] . the set of (set-theoretically) maximal members of I. the set M axM et(X) of all ﬁnite intersections of elements of M ax(X) consists of local A-addressed P ara-meters which form a base of the so-called maximal meter topology on X. The level lev(x) is deﬁned to be the level of the (uniquely deﬁned) smallest canonical P ara-meter containing x.e.13. With the above notation. 1. QED. we have lev(x) ≤ sp(x). M ax(X) = M axM et(X)|0. σ ⊂ τ implies τ ⊂ σ.e. For any interpretation K I of a ﬁnite local composition K.1.4. and if < is the dominance relation16 on the maximal meter topology. for each i. if x. Consider the set M ax(X) of all maximal local A-addressed P ara-meters contained in X. the situation is easier. i. We then The above example of M axM et(X) has the atlas M ax(X) for its interpretation X I|0 . 4. 16 See appendix F. whence the claim. This interpretation is denoted by X M etLg[L] . We identify it with the corresponding covering and also denote it by the interpretation symbol X M axM et . Select the set of all local A-addressed P ara-meters U ⊂ X such that their periods p(U ) are at most equal to a limit P . 3. the simplicial metrical weight sp(x) of a point x ∈ X is deﬁned to be the dimension of the simplex Sp(x) of all maximal (=level zero) canonical P ara-meters in X M axM et which contain x. In fact. When taking (P ara-)meters. and we set I|i = evidently have an interpretation K I|i . y are points in the local composition X. we have a surjective map of partially ordered sets : n(X M axM et|0 ) → X M axM et : σ → σ. The open set U (x) = Sp(x) is the minimal neighborhood of x. 2. • levi+1 = lev0 (I − k≤i levk )..6) i. Proof. it is not easy to decide how to interpret a local composition by (P ara-)rhythms since the decision between large repetition numbers and large germs has no evident rationale. . The charts in levj are called charts of level j. The dominance relations in this topology read as follows: Lemma 20 With the above notation. we may introduce a level function Lev : I → N as follows. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 329 In general. then we have y < x iﬀ Sp(y) ⊂ Sp(x) iﬀ U (x) ⊂ U (y). We have several basic examples of (P ara)metric interpretations of an A-addressed objective local composition X in Rhythm(P ara). Take the atlas of all local A-addressed P ara-meters U ⊂ X such that their lengths l(U ) are at most equal to a limit L. (13. k≤i levk .2.. Proposition 11 With the above notation.

Musically speaking. . . but not necessarily vice versa. . ∼ we have17 (A@M )n → A⊕n ⊕ Rn−1 @M and we can parametrize an n-element germ G by the n-tuple (G) of A-addressed points. Proposition 12 The irreducible closed sets of the maximal meter topology are the closures of their points with maximal simplexes in the nerve. .7. Then the elements of G are in A@M . the generic points are those which participate in maximal collections of local P ara-meters. There are however two approaches in order to circumvent such a problem.3. . . For example.330 CHAPTER 13. 18 The present ﬂattening operation is just the inﬁnite one from the previous discussion. In this case. . QED. as discussed in section 6. U (y) intersect. . . n − 1. n A = (A) ⊕ R The second solution to ambiguities in germ deﬁnitions uses macros and a ﬂattening operation18 . We start from the hypothesis of the preceding discussion and therefore may have the germ G living in a simple form indexM akroBasic Basic of module M = R ⊕ P ARA where—for the sake of mathematical simplicity—the ﬁrst factor parametrizes the Onset coordinates. . . If y is any point of F . . . We have y < x iﬀ x ∈ U (y) iﬀ U (x) ⊂ U (y). 13. and lemma 20 implies that Sp(x) = Sp(z). We need the circular form KnotBasic : 17 The bijection takes the n-tuple (g = eti ·g t i i.3. This point dominates x and y. and we suppose that the coeﬃcient ring for A is R. ιi (a) = (0.0 )i into e 1 ·g.6).) = gi. . g(0. and where P ara → @P ARA is simple with a module P ARA. and take care of the lengths of the local P ara-meters as well as of the periods and the positions of the points within the maximal local P ara-meters in chapter 21.) = tj+1 − t1 for j = 1. .) = g1 (aa ). 0. . . . where M is one of the modules R. By maximality of the simplicial metrical weight of x. . and g(0.). We shall introduce reﬁned metrical weight functions which generalize the above simplicial weight.8). ai . The injections ιi : A n A are these: ι1 (a) = (a. QED. there is no unique maximality of germs. the minimal neighborhoods U (x). This means that the germ has coordinates in a module. . the simplicial metrical weight of z is equal to the weight of x. whence Sp(y) ⊂ Sp(x). . and the generic points are the elements of the images of the maximal simplexes under the map (13. The irreducible components correspond to the maximal simplexes of the nerve n(X M axM et ). . by the construction of ﬁber sums of modules (appendix E. R ⊕ P ARA. ai = a. . 0. 0) for 1 < i. and then by the corresponding point (G) with address n n−1 .4. GLOBAL COMPOSITIONS Proof. 1n+i−1 . the closures of points of maximal simplicial weight are irreducible components. . The rest is clear. . so take a z in that intersection.0 (ai ) for 1 < i. Proof. The ﬁrst deals with address change. . . .1 Macros for Rhythmic Germs We have seen that for general rhythmic germs. so x dominates y. where g(a1 . and that the metrical special case is built upon the singleton germ set. . 1n+j . . Suppose we are given a ﬁnite germ G which is an A-addressed local composition which is completely contained either in the sub-form Onset or in ∼ the sub-form R(P ara) of Rhythm(P ara). Suppose F is an irreducible closed set and x is a point having maximal simplicial metrical weight among the elements of F . and this means that they are present in a maximal collection of regular patterns in onset time. . . whence x ∈ U (x) ⊂ U (y) and therefore U (x) ⊂ U (y).

neither in the delimitation against each other. F latten∞ (D) = i b + F latten∞ (Knoti ). We shall deal with this aspect in chapter 22.4 Motivic Interpretations: Melodies and Themes Summary. In this language. in particular. a global rhythm is an interpretation of a local composition in ambient space KnotBasic by local P ara-rhythm of macro type. A theme is a subcomposition of a composition which plays a certain role in the entire construction of the composition qua expression of meaning. For example. the subtle interpretative activity in motivic analysis.e. ‘ﬂat’ germ G is the ﬂattened version of the former. a perspective which we shall reconsider in chapter 21 when discussing grouping rules in Jackendoﬀ-Lerdahl’s GTTM. F K = F un(KnotBasic ). we may take a germ which contains ornaments as M akrobasic -formed denotators. m). But there is an evident relation of this approach to the Schenker idea. we deﬁne ep (G∗ ) = (ep (b). m) deﬁned by 1. We discuss the local/global dichotomy in this subject.4. 13. On the other hand. if ep is a period translation in time. . –Σ– Concepts such as “motif”. but includes a semiotic perspective. the very concepts are not clear. and that we are far from having established any systematic treatment of such objects. Motives are considered as being local ingredients of melodies. . if m = {Knot1 . nor in the intrinsic attribution.4. and. Knotk }. An A-addressed local P ara-rhythm can be paraphrased by the local compositions in ambient space KnotBasic . i. and they constitute a hierarchical shaping of the ﬂat rhythmical germs.. they are local compositions of the shape e[a. M akroBasic ) Id with M akroBasic f :F →2F K ∼ −→ ΩF K Power(KnotBasic ) and F = F un(M akroBasic ).13. “melody”. INTERPRETATIONS AND THE VOCABULARY OF GLOBAL CONCEPTS 331 KnotBasic −→ Limit(Basic. F latten∞ (D) = {b} if m = ∅. how to decide when two motives are “similar”. G = F latten∞ (G∗ ). it is not clear how to compare motives. and if the macro germ is G∗ = (b. . So global macro rhythms can be projected to global rhythms by ﬂattening. for example. We have the ﬂattening operation on KnotBasic Denotators D = (b. and/or drum patterns or the like as ‘satellites’ of the macro. We should stress that these reﬂections are merely ‘germs’ of a theory of macro-events. 2.b]p G∗ . only that it operates exclusively on the basic event in the knot macro. Then the idea is to deﬁne a macro germ G∗ such that the given. We should also stress that a theme is not necessarily related to the motif . the concept of a theme or motive is not only a structural one. There are three main sources of such fuzziness: On the one hand. built upon macros G∗ . m). The translation in time is the same as before. “theme” are among the most fuzzy of musicology. Thirdly.

. In this section. Every contour or gestalt or shape concept is derived from the motif structure by elaboration of speciﬁc motif aspects. . such as interval sequences.e. In fact. If we cover a motif by characteristic motifs. We have already introduced the motif concept in deﬁnition 15 (subsection 7. not the structure. the motivic interpretation which deﬁnes the melody upon a motif has a more elementary—but nonetheless dramatic—function which we discuss now. We are urged to set up the basic tools for comparison of global compositions. melodies. it is a derived attribute which comes out from a complex and not uniquely deﬁned abstraction process.332 CHAPTER 13. Gestalt is a central invariant of melody. as expressed by the melody’s nerve. the covering may split into several disconnected (disjoint) motives. a melody is a local composition with successive tones in time. For example. a theme may as well be a harmonic or rhythmic object. e. But even if we are given two ﬁxed motivic interpretations. Before the gestalt aspect comes in.or inter-dependent parts. p. these units are not neutral data. a melody is also associated with the Ehrenfels gestalt qualities of super-summativity and transformational invariance. So this elementary interpretation activity precedes reﬁned considerations: it deals with the initial statement of which are the parts and which is the whole.2. In particular: Deﬁnition 42 A melody is deﬁned as a motivic interpretation of a motif. onset-pitch-interval angles. the melody’s motives. it is not clear how one should compare two such objects. We should therefore deﬁne a “motivic” interpretation as being an interpretation of a local composition in a determined motif ambient space (having onset and pitch factors) such that every chart is a motif in this space. i. and this is what we shall do in the next chapter..g. We shall deal with this ramiﬁcation in chapter 22. etc. But it is more: it is understood as a compound object with “characteristic” melodic ingredients. Gestalt is a concept of equivalence which should be built upon comparison devices for melodies. or motives may have just one tone in common. they are results of interpretative interventions.3) as an objective local composition with onset and pitch coordinators in its ambient space. but it is not the intrinsic concept. diastematic ambitus. they depend upon the interpretative. Its characteristic property is the semiotic role. this action reﬂects a disintegration of the original underlying melody motif into several relatively small parts which may intersect in a rather loose way. length ratios. The only thing which is clear now is that melodies are not neutral objects. the article “Melodie” in the Riemann Musiklexikon [457. analytical. we shortly focus our attention on the problem of elementary and compound motivic structures. or poietic interaction. and such that the onset projection is bijective. In traditional musicology (see. In other words: they are a covering of the motivic tone material of the melody by speciﬁc selected submotives. GLOBAL COMPOSITIONS concept. The AST has dealt with this aspect in the contour theory. Now. The crucial question is whether we should keep in mind the original melodic motif (the entire set of tones) or whether the splitting activity is really meant as a decomposition in henceforth in.554]).

at least on the level of a reliable scientiﬁc language. We discuss this issue with regard to comparative analysis of melodic themes in a composition. We describe the combinatorial aspects of globality and associated functors. We introduce this subject together with its musical motivation and append the formal deﬁnition of morphisms among global compositions. This leads to the categories ObGlob of objective and Glob of functorial global compositions. as well as corresponding geometric classiﬁcation tools: nerves and simplicial weights. The musicological fuzziness of global structures appearing in melodic analysis—which we have made precise via motivic interpretations—makes it impossible to take a step further in the comparative methodology of global structures. –Σ– 14. Robert Frost (1874–1963) Summary. and deep. requiring their comparison is quasi-automatic. it is intuitively accepted that one deals with smaller and larger units. It turns out that the latter is virtually non-existent. this is not in the tradition of musical analysis. Since global compositions are of musical and musicological interest. Beethoven’s compositions. on the contrary. However. –Σ– Mathematically. In the typical case of a comparison of two melodies. But I have promises to keep.Chapter 14 Global Perspectives The woods are lovely. dark. or in the discussion of the motivic-thematic work in. it is straightforward how to extend the categories of local compositions to categories of global compositions. And miles to go before I sleep. say. this is not a common situation. But musicologically. and in the more general context of comparative analysis of musical works. and that comparison 333 . Global perspectives deal with relations among global music objects. And miles to go before I sleep.1 Musical Motivation Summary.

isomorphisms or composition of morphisms. f : G → H is a set map. s of indexes which correspond under the map ι (i. A@Et )T for G and (Ls . α) where 1. there is a rich connectivity. if we take the chart isomor∼ ∼ phisms φt : Kt → It and ψs : Ls → Js for some pair t. And. This is the trivial case where the added value is naught. So the added o value is complex: it lies in the single charts (the triads) as well in their intersections (thirds or single notes) and also in the combinatorial conﬁguration. how strongly is the harmonic strip related to the original scale? Are there many scales with essentially equal harmonic strips? Same situation for motivic analysis: If we take the main theme of Bach’s Art of Fugue. So Ehrenfels’ requirement of an “added value” to the “sum of parts” against the “whole” is not understood: How much more do we need to get the whole? How can we compare this speciﬁc diﬀerence. ι(It ) = Js ). colimits. For example. So the comparison reduces to counting notes. a comparison of gestalts is a fortiori impossible. e. In fact. So. . On the other hand. in the nerve. Then a morphism from GI to H J is a triple (f. But in general. ι : I → J is a set map such that f (i) ⊂ ι(i) for all covering sets i ∈ I. ι.2 Global Morphisms Summary. Special aspects. synonymy. and all kinds of simple forms. the denotators of musical works. This section introduces the technical deﬁnition of morphisms between global compositions. α is a family α = (αi : A → B)I of address changes. the cardinality of the original local composition.g. The concept of interpretable global compositions is presented and illustrated. for example with the harmonic strip. B@Fs )S for H. The category theory for general denotators is still embryonal. without answering these questions. In this chapter we shall introduce only comparative tools (morphisms) for global compositions. such as European scores. then the induced maps ft : Kt → Ls deﬁne morphisms ft /αi : Kt → Ls of objective local compositions. and there is no connection left. Comparison of the gluing of parts is not conceived. if we have split a local composition into isolated singleton charts (the ‘silly’ interpretation). 14.. 3. the disintregation is extremal: We essentially have to count the tones. 2. is it the 8-tone theme or the 12-tone theme? This still debated question depends on the development of a sophisticated technique of motivic analysis of melodies or themes. are discussed. such that 4.e. see the contribution of Mariana Montiel Hernandez [378] concerning a standard form for piano scores. –Σ– The ﬁrst deﬁnition regards objective global compositions: Deﬁnition 43 Suppose we are given two objective global compositions GI at address A and H J at address B. for any atlases (Kt . cannot be grasped by compositions (in the technical sense). in fact the M¨bius strip. but it is not clear how the relations of parts within a ﬁrst melody (for example) should be taken over to relations among parts of a second melody. GLOBAL PERSPECTIVES should take into account comparison of parts.334 CHAPTER 14. they deal with more general forms of all possible typologies: limits. ﬁnally. overall comparative analysis of musical works of course transcends the theory of local or global compositions..

α is a family α = (αi : A → B)I of address changes. Observe that this is the precise generalization of the morphism concept from objective local compositions. deﬁning the composition of morphisms of global compositions is straightforward. as deﬁned in deﬁnition 36. Show that the identity morphisms are in fact left and right identities in the sense of category theory. by f /α : G → H. their composition h/γ = g /β · f /α : GI → K L is deﬁned by the triple (h = g · f. where β · α is the family (βι(i) · αi )I . γ = β · α). whereas the set 1 ObGlob of morphisms is deﬁned by deﬁnition 43. Exercise 27 Show that composition of global morphisms is associative whenever corresponding factors are deﬁned. f : G → H is a natural transformation. IdA ). @A × Et )T for G and (Ls . if no misunderstanding is possible. K L . 3. Deﬁnition 46 The category of objective global compositions ObGlob has the objective global compositions as its object set 0 ObGlob.e.2. This morphism will be also be denoted by f /α : GI → H J or. as deﬁned in deﬁnition 38 whereas the set 1 Glob of morphisms is deﬁned by deﬁnition 44. the identity morphism is the triple (IdG . the deﬁnition of morphisms runs correspondingly: Deﬁnition 44 Suppose we are given two functorial global compositions GI at address A and H J at address B. It is useful to view functorial global compositions as a system of objective global compositions as follows: ι κ ζ κ ι ι . α) where 1. for any atlases (Kt . ι. IdI . ι(It ) = Js ). Deﬁnition 45 Given three objective or functorial global compositions GI . then the induced maps ft : Kt → Ls deﬁne a morphism ft /αi of functorial local compositions. a morphism f /α : GI → H J . if no misunderstanding is possible. For functorial global compositions. @B × Fs )S for H. 2. ι : I → J is a set map such that we have subfunctor relations im(f |i) ⊂ ι(i) for all covering functors i ∈ I.. if we take the chart ∼ ∼ isomorphisms φt : Kt → It and ψs : Ls → Js for some pair t. such that 4. If we have GI = H J . H J . s of indexes which correspond under the map ι (i. by f /α : G → H. ζ = κ · ι.14. Then a morphism from GI to H J is a triple (f. and a morphism g /β : H J → K L . The category of functorial global compositions Glob has the functorial global compositions as its object set 0 Glob. GLOBAL MORPHISMS ι 335 This morphism will be denoted by f /α : GI → H J or. This being the case.

also denoted by ι. i. xi+4 . and which has the atlas deﬁned by the second factors in the charts f @Kκ = {f } × Kκ. For example. . More generally. If the local composition is a major or minor scale scale X as discussed in section 13. we have a B-addressed objective global composition f @GI which is deﬁned on the subset of B@G which is locally covered by the images f @Iκ of the chart subsets f @Kκ of any atlas. 7}. (2) = {{xi . Fix any address change f : B → A.2) which has. the triadic interpretation is deﬁned by the atlas (3) = {{xi . . also with indexes mod 7. δ7 ) : (i) → (i + 1) denotes the choice of δi ∈ BIT = {0. . δ2 . One starts with the singleton interpretation X (1) ..336 CHAPTER 14. The entire construction is well deﬁned by lemma 21. we have a chain of reﬁnements K I|0 → K I|1 → . 2. xi+4 }| i = 1. . . Deﬁnition 47 For every address change f : B → A and a functorial global composition GI at address A. all i ∈ I. with B@Iκ ∩ B@Iλ coincide. i.λ /α on the inverse images of the intersection have denominator α = IdA . We denote this covering of f @G by f @I. With this setup. it may be reinterpreted as being one of two charts of the successive reﬁnements. xi+2 .e. In fact. and therefore. and consider the subsets f @Kκ ⊂ B@Kκ . xi+6 }| i = 1. . The identity slice IdA @GI is denoted by (GI )∨ . the parts related to f correspond. We have discussed the triadic interpretation X (3) . More precisely. together with the given transition isomorphisms. . ending with the unique morphism to the local scale composition: X (1) → X (2) → X (3) → X (4) → X δ δ δ ! (14. GJ of an objective local A-addressed composition G. . . a map ι : I → J with i ⊂ ι(i). Proof. but this is just one possibility. . x2 . and the tetradic interpretation X (4) which is important in jazz harmonics is deﬁned by the atlas (4) = {{xi . In particular.2. and . 1} for the ith value. . the intersections are the same since the induced isomorphism φκ. xi+2 . the third interpretation X (2) is deﬁned by charts of thirds. of course an important musicological meaning: If ever we are given a chart of atlas (i). . if Z i is the degree Z in X (i) .. respectively. and a reﬁnement map. Then the intersection of their images under B@φκ and B@φλ . set X = {x1 . where (1) = {{xi }| i = 1. GLOBAL PERSPECTIVES ∼ Lemma 21 Let GI be a functorial global composition at address A. we have 27 reﬁnement maps for each succession of the above atlases.1) ι of an A-addressed ﬁnite objective local composition K. . 7}. and call the objective global composition the f -slice of GI .4.. 7}. any interpretation GI yields a unique reﬁnement morphism ! : GI → G.3. . .. So we have this succession of reﬁnements. Suppose that φκ : Kκ → Iκ ∼ and φλ : Kλ → Iλ are any two charts for GI . we have two embeddings Z i ⊂ W i+1 depending on whether we take Z i as the lower or higher part within W i+1 . Example 23 Given two interpretations GI . then the identity IdG and ι induce the reﬁnement morphism IdA /IdA : GI → GJ . and the reﬁnement map δ = (δ1 . 2. . The lower embedding is denoted by 0.4. the higher one by 1. . we may consider the diﬀerent versions of degrees. 2. . 7}. etc.f . x7 } in the order of increasing position on Z12 . . f @Kλ ⊂ B@Kλ . with the notation of 13. xi+2 }| i = 1. . i.e. K I|n → K I|n+1 → K I (14.e. 2.

This germ is essentially an ascending and then descending .1: The four interpretations of a major or minor scale. The general harmonic principles of this sonata yield a motivic germ G ⊂ Onset × P itch|Z[1/2] with coeﬃcients in the localization ring Z[1/2] of fractions with denominators that are powers of 2. and associated maps between nerves (see section 14. more radically. this documents the vast ambiguity in elementary harmony of degrees! For the corresponding nerves. The absence of orientation on the harmonic strip vanishes on the tetradic interpretation of jazz. in fact a union of tetrahedra. So the absence of orientation of the harmonic strip vanishes on the tetradic jazz harmony.4 for this association). up to the tetradic interpretation for jazz harmony. This may not only ease jazz harmony. see ﬁgure 14. there is more freedom of choice. more improvisational ﬂexibility—at the cost of unambiguous harmonic syntagmatics. is more complex. GLOBAL MORPHISMS II VII V III VI IV I singletons 337 third intervals triadic degrees scale as a whole four note degrees Figure 14.1. it throws the context of jazz harmony one dimension higher: Moving in a three-space such as the torus. Every interpretation uniquely maps to the ‘uninterpreted’ local scale composition shown to the right. The harmonic strip of the triadic interpretation is embedded in the tetradic torus as a strip which is ‘entwined’ along the torus’ interior circle. the intersection conﬁguration around any ﬁxed tone. Observe that the tetradic interpretation has a nerve which is a full torus.14. Example 24 This example refers to the analysis in [328] of the ﬁrst movement of Beethoven’s ‘Hammerklavier’ sonata Op. We have successive reﬁnement morphisms. 106. This ambiguity is also dramatic with regard to the geometric conﬁguration of the nerve geometry. starting with the singletons.2.

Qualities are compared lexicographically: (s1 . In [328] it is shown K Figure 14. Of course. see also ﬁgure 14. p2 ) iﬀ either s1 < s2 or s1 = s2 and p1 < p2 . p1 ) < (s2 . GLOBAL PERSPECTIVES G(2) G e6G e12G Figure 14.3: The right-hand melody which appears in bars 75-76. To begin with. the problem is not the existence of such interpretations since the silly interpretations always ﬁt in such a framwork. that a large number of motivic charts can in fact be interpreted such as to become isomorphic to interpretations of local subcompositions of G(t). chromatic scale of three half-tone steps ‘octave’ period. We therefore look for good quality interpretations. 106 is essentially an ascending and then descending chromatic scale of three half-tone steps ‘octave’ period. Deﬁnition 48 Let K I be an interpretation of a local composition K.t]6 G. 106.2: The germ G of motivic work in Beethoven’s ‘Hammerklavier’ sonata Op. we look at the right-hand melody which appears in bars 75-78 (ﬁgure 14. and 77-78 of the ﬁrst movement in op. With this in mind. So the best quality is (1.338 CHAPTER 14. p). The claim of our analysis is that the motivic work (German: “motivisch-thematische Arbeit”) relates to interpretations of the P itchrhythms of type G(t) = e[0. The quality of K I is the pair (s. we look for best possible qualities in our interpretations. and p is the cardinality of connected components of N (K I ). 1). Here we show the P itch-rhythm with three copies of the germ.3). where s = card(I). .2 for this ‘motivic zig-zag’. The (projected) representation K in Onset × P itch space is shown to the score’s right.

GLOBAL MORPHISMS 339 K A. The sample. every interval is isomorphic to {(0. 0). is shown to the left in ﬁgure 14. we see that this melodic part is isomorphic as a local composition (quality (1. a twelveelement motif.10. 1)}. L2 . In ﬁgure 14. (0. So the isomorphism of the interpretations by submaximal charts is no guarantee for the isomorphism of the original local compositions. and the pitch inversion. Evidently. The ﬁrst chart L1 equals B · K1 with a transformation B = eq · 2 9 0 −1 where the linear part is an onset dilatation by 2. 0)} isomorphism class Nr. L3 .14. (0. 0).13. L = {(0. zero-addressed three-element motives in OnP iM od12.4: The right-hand melody in bars 75-76 and 77-78 is isomorphic to a local subcomposition of the motivic ‘zig-zag’ under a vertical arpeggio A. L(2) by the three two-element ‘interval’ subcompositions. an isomorphism since its determinant −2 is invertible in Z[1/2]. In ﬁgure 14. 1). followed by a ninefold horizontal arpeggio. 1)) to a subcomposition of G(2). To the right we see the (projected) representation and interpretation K of this motif in Onset × P itch space.6 we see the subcomposition L of G(2). respectively.K Figure 14. The isomorphism is given by the matrix A = eq 1 2 0 1 representing a vertical arpeggio.5. 11)} isomorphism class Nr. we have K (2) → L(2) . 0). . This interpretation is isomorphic to the interpretation K I under the following isomorphisms: L2 and L3 are translations of K2 and K3 . 1). as well as its interpretation LJ by three disjoint charts L1 . The second sample needs a proper interpretation of quality (3. (1. (0.2. 3). (3.4. and consider the interpretations K (2) . Example 25 Consider the two following non-isomorphic.12 : K = {(0. and since the intersections are single∼ tons.

2. Exercise 28 Consider the following zero-addressed local compositions in P iM od12 : C = {0. f (10) = 5. Y = I. 4}. 0}. 10} and call the associated interpretation X (3) . IIX = {7. 10. GLOBAL PERSPECTIVES K Figure 14. Exercise 29 Show that the ecclesiastic modes (section 13. Show that we have an isomorphism f /1 : X (3) → C (3) which is deﬁned by f (0) = 7. . 4. IVX = {10. Show that this example fails if we do not allow isomorphisms with ﬁfth or fourth transforma∼ tions. IIIX = {3. 4. V IIX = {4. i. V IX = {2. 3. f (7) = 2. isomorphism class Nr 38 . 0. 9} C-major scale. f (3) = 4.5: A twelve-element alto-voice motif K in bars 79-80 in the ﬁrst movement of op. VX = {0. 3. 2}. it is even true that any two diﬀerent ecclesiastic modes ι ∼ . V II.2) are rigid. 2. 7. 2. 4. Take the triadic interpretation C (3) and the following triadic interpretation of X: IX = {11. 11} isomorphism class Nr 50 . if only inversions and transpositions are allowed. 106. f (2) = 9. then we have C → X as soon as (3) ∼ (3) C →X .. ι(YX ) = YC . 7. 8.t ) = 1. 11. . f (4) = 11. 11}. 5. 3}.4. 7.340 CHAPTER 14. f (11) = 0. which are not isomorphic.. i.e. they have trivial automorphism group: Aut(Xf. X = {0. 7}. 10.e. .

It is a fundamental question in every category to ﬁnd out whether some determined subclass of arguments of the functor @X is already rich enough to determine X up to isomorphism. X.3. i. dorian ↔ dorian phrygian ↔ ionian lydian ↔ locrian mixolydian ↔ aeolian and this means that modes become rigid when transformed into their ecclesiastic castings. The morphisms which are addressed in local compositions separate morphisms of global compositions. on the same scale are not isomorphic. @T ) → Hom(@S|Af f . .6: The interpretation of the alto voice in bars 79-80 of quality (3.3 Local Domains Summary.if S. We discuss the musical consequences. we have a bijection Hom(@S. Y ) → Hom(@X. in fact. more precisely. @T |Af f ). –Σ– The Yoneda lemma (see our discussion in section 9) tells us that we may replace any global composition X in Glob by its functor @X ∈ Glob@ and we shall not lose any information about its isomorphism class.. we have a canonical ∼ bijection Hom(X. Y . In algebraic geometry. Ud (Cf )= Cb .. Ud (Cg )= Ca . 14.e. Ud (Ce ) = Cc . we have a permutation of these modes (we take the authentic mode names): Ud (Cd )= Cd . for any two global compositions. i. 3). We shall now show that we have a weaker result in the theory of global compositions. a completely traditional eﬀect.14. T are two schemes. since Aut(C) = Ud . LOCAL DOMAINS L KI K1 K2 341 LJ K3 L1 L2 K3 Figure 14.e. This is false for modes Xf . the local arguments are ‘separating’. @Y )... it is well known ([198]) that the restriction @S|Af f of the scheme functor @S of a scheme S to the subcategory Af f of the ∼ aﬃne schemes is classifying.

t @X) of the restriction morphisms Ws.e. we have Y @loc X = Y @X = Hom(Y. We show that u is uniquely determined ι by its restriction to local arguments. Let u : @X I → @Y J be a natural transformation. respectively.. the contravariant functors @loc X . If X I = X is a local composition.t @Y ) which is a subset of W @Y . respectively. GLOBAL PERSPECTIVES Deﬁnition 49 Let X be a global composition in Glob. Also. In fact. j∈J @loc Yj . respectively. B. and ι the restriction f · hi on the covering (Xi ) of X determines f . @loc Y J ) is injective. @loc Y J ) → i∈I 1 For I ∼ Hom(@loc Xi . consider the canonical embeddings hi /IdA : Xi → X of the i ι i ι(i) subfunctors Xi of the covering I of X. the canonical restriction map loc : Hom(@X I . the address change α = (αi )i is determined. then loc is a bijection. ι:I→J i∈I Proof. QED. the map on the atlases for a local domain Y reduces to the indication of one chart Ii or chart index i in the target composition X. Therefore f /α and its functorial ι counterpart u = @f /α is uniquely determined by the local restrictions. By u. We have to extend u to a natural transformation U : @X → @Y J . local domains are separating. i. @Y J ) → Hom(@loc X I . respectively. @loc Y J : Loc → Sets. Y J are local compositions at addresses A. Since a local composition has just one chart. Lemma 23 If X I . the kernel is strictly smaller than the morphism set because of the index map! . then there is a bijection ∼ l : Hom(@loc X I . By @loc X we denote the restriction of the Yoneda functor @X to the subcategory Loc of local compositions. Therefore.t → Wt . and we are done.t Ws. By deﬁnition. where αi is the address change of the ith covering subfunctor. this kernel is mapped into the J I J kernel Ker( t Wt @Y J s. Let now X be local and let u : @loc X → @loc Y J be a natural transformation. Then the diﬀerence kernel Ker( t Wt @X s. the map ι : I → J is determined by the local evaluation at Xi . are isomorphic to i∈I @loc Xi . B. Hence. Lemma 22 If X I .342 CHAPTER 14.t → Ws identiﬁes1 to W I @X. By the Yoneda lemma we know that u = @f /α for a ι ι i morphism f /α : X → Y . Therefore we write i f /α : Y → X to indicate elements of Y @loc X. To this end. Proof. X) for any local (functorial) composition Y . Ws. Let W I be a global composition with charts Wt . Hom(@loc X I . So let us show that f /α is determined by the restriction loc(u) to local arguments. @loc Y J ) → Xi @Yι(i) . Y J are local compositions at addresses A.t Ws. @loc Y J ) global codomains. We have u(hi /IdA ) = f /α · hi /IdA = f · hi / αi .

respectively. @loc Y J ). i2 ∈ I. Exercise 30 Prove these statements. Y J are local compositions at addresses A. By Homloc (@loc X I . NERVES and by lemma 22 and Yoneda’s lemma Hom(@loc Xi . B. @loc Y J ) is well deﬁned and bijective. To each global composition is associated a combinatorial structure. QED. In other words. Lemma 24 With the above notation a natural transformation u : @loc X I → @loc Y J stems from a morphism X I → Y J of global compositions iﬀ its image l(u) = (fi /αi : Xi → Yι(i) )i∈I with index map ι : I → J has the property that for every couple i1 .14. This functorial association retrieves important information on the global object. We are however far from completely understanding the nature of the local domain functors and their category loc Loc@ . then the canonical restriction map loc : Hom(X I . fi2 coincide on the intersection functor Xi1 . in fact the simplicial complex of the covering of the composition by its charts. So we have this proposition: Proposition 13 If X I .4.i2 = Xi1 ∩ Xi2 . the transformations fi1 .. i. Y J ) → Homloc (@loc X I . we denote the set of these natural transformations and call them localizable. the nature of the patchwork of the local aspects is crucial. and this is why we look at the nerve functor now. But we know that above the local domain argument. 14. we may replace the study of isomorphism classes of global compositions X I by the study of the isomorphism classes of their local domain functors @loc X I under localizable natural transformations (which deﬁne a non-full subcategory loc Loc@ on the local domain functors). @loc Y J ) → Xi @loc Y J → j∈J ∼ ∼ 343 Xi @Yj whence our claim follows from the distributivity laws for products and coproducts of set-valued functors. its nerve.4 Nerves Summary.e. With the restriction that the functors of local perspectives only reﬂect isomorphisms of global compositions if these isomorphisms are localizable. –Σ– . morphisms on local compositions. The musicological meaning of these results is that the Yoneda philosophy for global musical structures may be restricted to local perspectives.

a natural transformation f : G → H which stems from a morphism of global compositions.e.. iﬀ f = g.2. appendix H.1. the geometric nerve can be realized 2 as a polyhedron in R2m+1 if dim(n(GI )) ≤ m.2. So.2. In particular. we have Proposition 14 Any two mathematically equivalent morphisms f /α. example 103.2. Clearly. the geometric nerve N (GI ) is deﬁned as the space |n(GI )| of n(GI ).344 CHAPTER 14. i.3. but see appendix H. i. 3 See 2 See ι ι ι ι κ ι κ ι κ . is called a mathematical morphism between global compositions. N : ObGlob → Top into the category Top of topological spaces and continuous maps3 . For functorial global compositions. for two mathematically equivalent morphisms f /α. every morphism f /α : GI → H J of objective or functorial global compositions gives rise to a morphism n(f /α) : n(GI ) → n(H J ).. by lemma 96 in appendix H.2. We shall come back to mathematical morphisms in section 19.1. GLOBAL PERSPECTIVES Nerves of objective global compositions were introduced in 13. the simplicial complex whose vertexes are the covering subfunctors Gi .e. the corresponding simplicial maps are contiguous4 . Deﬁnition 50 Two morphisms f /α. Since I is ﬁnite. and whose simplexes are the (ﬁnite) vertex sets with nonempty intersection.. g /β : GI → H J induce homotopic maps on the corresponding geometric nerves. we have the same procedure.1. theorem 74. Exercise 31 Draw a picture of a homotopy between the equivalent morphisms δ : X (2) → X (3) from the third interval to the triadic interpretation (yielding the harmonic strip) of a diatonic scale as described in the diagram 14. i..2. The abstract nerve n(GI ) of a global functorial composition is the covering’s simplicial complex. g /β : GI → H J .2. and this one generates the continuous map N (f /α) : N (GI ) → N (H J ). 4 See appendix H.e. the functor N of geometric nerves generates topological invariants for isomorphism classes of global compositions.e. i. everything in a functorial way. An equivalence class of mathematically equivalent morphisms. we have two functors N : Glob → Top. appendix H. Evidently. g /β : GI → H J between global compositions are called mathematically equivalent iﬀ they coincide on the functors.

3. for every ksimplex σ = {Gi0 . ∼ Make ∩σ ⊂ G into a local composition by the induced isomorphism φij . its nerve has less vertexes and these may collapse. We shall discuss such an example relating to the functorial interpretation associated with the harmonic strip in section 25. . SIMPLICIAL WEIGHTS 345 Proposition 14 means that the variation of the embedding of a chart—of a reﬁned interpretation. –Σ– Lemma 25 Let GI be an A-addressed objective or functorial global composition. However. Take the limit (gs : L → Bs )s of this system of address changes over A (it exists according to appendix E.σ : Kij .3. Proof. there is an f -slice f @GI such that the canonical simplicial map n(GI ) → n(f @GI ) is well deﬁned and surjective. The case of a one-dimensional simplex is just the axiom of gluing charts together by transition isomorphisms of local compositions. So take address changes fs : Bs → A. and take a k-simplex σ = {Gi0 . we have the f -slice f @GI . .e.5 Simplicial Weights Summary. This means: Theorem 13 For every global functorial composition GI . Gik } of n(GI ). This yields a tool for local and global classiﬁcation. to be treated in the next chapter. . one for each maximal simplex σs . QED. they form an atlas for the (local) composition imposed on the intersection ∩σ. In general. But observe that ∅ = f @ ∩ σ implies ∅ = f · g@ ∩ σ for any address change g : C → B. The structure of a nerve induces weights on the nerve’s simplexes with values in isomorphism classes of local compositions. it appears as a specialization morphism by collapse of particular vertexes.4.5. i..14. . This can be checked on an adequate slice of X I . an objective global composition. Also some simplexes could vanish. This local composition (viewed as a global composition with one chart) is again denoted by ∩σ. this transition situation restricts to the intersection of any higher number of covering functors. Consider the subobject (set or functor) ∩σ ⊂ G and ∼ the chart isomorphism φij : Kij → Gij of a G-chart Kij for the ij th vertex of our simplex. example 61. For every address change f : B → A from the address A of a global functorial composition GI . Clearly. 14. .σ → ∩σ on the inverse image Kij . such that all intersections fs @ ∩ σs are non-empty. Then we have one unique address change h = fs · gs : L → A with nonvanishing intersections in every simplex of the functorial composition.σ = φ−1 (∩σ). On the geometric nerves. say—in one particular candidate of a larger chart of a coarser interpretation is related to the topological operation of homotopy if we consider the associated geometric nerves. . The nerve n(X I ) may be signiﬁcantly larger than the nerve n(X I ). . Gik } of its nerve. there is an address change f : B → A such that f @ ∩ σ = ∩j f @Gij = ∅. this means that reinterpretation of an embedding is a kind of deformation on the geometric level. We present the elementary example of simplicial motive weights. Intuitively. Then all the charts deﬁned by any of the simplex vertexes are ij compatible.8). For an interpretation X I of a local objective composition X. we have seen the construction ˆ ˆ of the global functorial composition X I associated with the covering of X by I.

Example 26 We consider the harmonic strip of X (3) of a major scale X.7: The class nerve of the triadic interpretation of a major scale. we have a geometric nerve N (GI ) and a class weight iso∩ for each global composition.12 .1. This is a very intuitive invariant object which we want to illustrate in two examples. The numbers are the class numbers from the classiﬁcation list in appendix L. the simplicial weight of GI .346 CHAPTER 14. Example 27 Consider the motif situation.1. Clearly. The class nerve CN (X (3) ) is shown in ﬁgure 14. more precisely.. So we have only the exchange of the boundary around the seventh degree. Summarizing. we call this object the class nerve of GI and denote it by CN (GI ).3) (with the usual identiﬁcation of local and one-chart global compositions). The classes are numbered according to the classiﬁcation in appendix L. in fact the induced automorphism from the uninterpreted local scale. by the three 2-element ‘interval’ charts in the motif. if two ∼ ∼ . i. For example. We look at the ‘face’ interpretation M (2) by maximal subcompositions. we have a contravariant functor ∩ : n(GI ) → Loc : σ → ∩σ (14. this means that this information can be visualized by a polyhedron with a class weight attached to each simplex.e. GLOBAL PERSPECTIVES If we view the abstract nerve n(GI ) as a category by the inclusion morphisms iσ. the zero-addressed local threeelement compositions M in OnP iM od12. 2 2 2 II 5 10 10 2 2 2 5 10 IV VI 2 2 6 2 10 2 6 10 I 15 VII 5 2 V 2 2 6 10 III 2 5 Figure 14. This functor induces a map iso∩ : n(GI ) → Loc/ → from the abstract nerve to the set Loc/ → of isomorphism classes of local compositions which we call the class weight (function) of the global composition GI . this nerve has seven 2-simplexes with weight number 2 for the unique zero-addressed singleton class in P iM od12 .7.τ : σ → τ among its simplexes. We see that the only automorphism of this interpretation must ﬁx the seventh degree. If we parametrize the isomorphism classes of local compositions by a determined set of symbols.

14. –Σ– We ﬁx an address module A over a commutative ring R. The classiﬁcation list in appendix M. If we consider global objective A-addressed compositions which are glued together by commutative local charts. The category ObLocA of commutative local compositions introduced in section 8.3. The four-element motives show counterexamples. .5 that the category of commutative local compositions refers to the ﬁxed address module A. however. Recall from deﬁnition 28 in section 8. and the class weights (2. The motif classes which are not uniquely determined by volume and class nerve are indicated by a star.5 has a global counterpart which is discussed here. Figure 14.12 . together with the motif’s volume (see section 11.4 which was calculated by Hans Straub [513] shows the class weights on the tetrahedra CN (M (3) ) of the face interpretation of the four-element motives in OnP iM od12. the morphisms are of form f /IdA with f being induced by an R-aﬃne morphism of the Coordinator R-modules. we see that the class nerve is not classifying. Denote this category of commutative global composition at address A by ComGlobA .8).3.3. this means that the transition morphisms are isomorphisms of commutative local compositions.3. 2. motives are isomorphic. So the class nerve yields invariants of the local classiﬁcation. 22 2 2 6 2 M (2) CN(M ) 2 6 (2) Figure 14.1. From the table in appendix M.3. but. For motives with larger cardinality this is.14. And it means that morphism between commutative global compositions are induced by R-aﬃne morphisms on the ambient modules of the local charts. the chart class numbers of the intervals from appendix L.6.6 Categories of Commutative Global Compositions Summary. CATEGORIES OF COMMUTATIVE GLOBAL COMPOSITIONS 347 class nr . 6) on the class nerve. the class nerves are also. as well as the interpretation.8: A three-element motif of class number 22 and its class nerve. the class nerve yields a complete set of class invariants.8 shows a motif of class number 22 in the classiﬁcation from appendix M. no longer true.

.

So it is a kind of essence of the overall eﬀorts of the art of music. Nun. no more general theory has been elaborated and we prefer dealing with situations where concrete results are available. in particular esthetics. Global classiﬁcation relies on two concepts: aﬃne functions and resolutions of global compositions. this is one of the most diﬃcult chapters since the relation of classiﬁcation and musicology. commutative) global compositions.19] Summary. 349 .Chapter 15 Global Classiﬁcation Wir sehen also. distances and angles (bilinear and exterior forms). These constructs are discussed and exempliﬁed. but presently. der eine Sammlung von Polyedern anlegte. o Mir jedenfalls will die Notwendigkeit solcher pataphysikalischer Spekulationen nur schwerlich einleuchten. es mag dahinstehen. –Σ– This chapter deals only with objective (more precisely. This is probably not always necessary from the technical point of view. or understanding. From the musicological point of view. But classiﬁcation is a deep concern since it reveals the a priori extent of a structural framework and therefore its power as an expressive tool of artistic activity. Pierre Boulez [60. It is therefore possible to view “molecules” as being global compositions with additional constraints. daß in der Musik die Mehrzahl der sogenannten “wissenschaftlichen” Geister fast so naiv ist wie Monsieur Achras – eine Figur von Jarry –. II. performance. is quite implicit. ob in unserem Fall Polyeder wirklich von unersch¨pﬂichem Interesse sind. p. their musical meaning is discussed. We derive classifying spaces and compare them to the situation in the Dreiding–Dress–Haegi theory of molecules: The latter are deduced from global compositions by additional structures concerning orientation. be it in composition.

Their role is that of generalized coordinate functions for musical compositions. 15. we ﬁx a commutative ring R. if N (GI ) has c connected components.1. Example 28 Since any morphism of global compositions f /α : GI → H J yields a natural transformation n(f /α) : n(GI ) → n(H J ). the constant module complex of M is the complex with M (σ) = M for all simplexes and identity transition. addresses A are R-modules. This approach makes it possible to separate the core process of composing music from its realization in instrumental parameter spaces.3) M : n(GI ) → R Mod (15.1) into the category R Mod of R-modules and R-aﬃne morphisms. As usual in sheaf theory.350 CHAPTER 15. Deﬁnition 51 An (R-)module complex over GI is a covariant functor (a coeﬃcient system. –Σ– . Given a global composition GI . If we do not stress the contrary. Example 29 If M is any R-module. with ι ι ι ι ι f /α M (σ) = M (n(f /α)(σ)). we put ΓM = limn(GI ) M (σ) and call this the set of global sections. where Mσ. every module complex M over H J induces a module complex on f /α M over GI . GLOBAL CLASSIFICATION 15. see section 14.5. Global aﬃne functions are patchworks of aﬃne functions on charts of atlases. in particular. Module complexes describe systems of global aﬃne functions in a classical homological language.τ : M (σ) → M (τ ) are transition morphisms of modules for the simplex inclusions (morphisms) σ ⊂ τ . Observe that its global sections are in bijection with the set M c . –Σ– For this chapter. see appendix H. we only consider R-modules and R-aﬃne morphisms between R-modules.1 Module Complexes Summary.1 Global Aﬃne Functions Summary. we recall the category n(GI ) of its abstract nerve.

5). although it is only determined up to an isomorphism of module complexes1 on GI . We call such a morphism an A-addressed function on the simplex. If we have a morphism σ ⊂ τ . M )(σ) = LinR (R.. The set of global functions of the function complex is denoted by Γ(GI ). Therefore. we select one ambient space A@R Fσ for every simplex σ of n(GI ). we have an embedding of module complexes N/C 1 By GI∗ (15. The subcomplex C = CGI of constant functions is deﬁned by C(σ) = {f ∈ nΓ(GI )(σ). i. we have an R-linear homomorphism R. a morphism f /IdA : ∩σ → A@R R may be identiﬁed with the supporting map f : ∩σ → A@R R.∩(σ) of A@R Fσ (see deﬁnition 17 in section 7.3) deﬁnition of module complexes. ∩ (τ ) → R.1. Then we have the submodule R. . of the full R-module A@R R. A@R R) = GI∗ and call this the module complex of A-addressed forms on GI . ∩ (τ ). M ) on the homomorphism modules for any given module M .f : R. Since an aﬃne function on ∩τ is the restriction of an aﬃne morphism A@h : A@R N → A@R R. f = constant on ∩ σ}. ∩ (σ). ∩ (σ) which induce these homomorphisms Linσ.e. we get the quotient complex M/N with values M/N (σ) = M (σ)/N (σ). M ). Suppose that we are given an Aaddressed commutative global composition GI . Exercise 32 Show that sums and scalar multiples of A-addressed functions is again an Aaddressed function. M ) → LinR (R. For an aﬃne function f : ∩σ → A@R R. this is an isomorphism of functors. if N is any module complex with C ⊂ N ⊂ nΓ(GI ). Let A@R R be the A-addressed local composition . This module complex is denoted by LinR (GI ..e. we shall work on categories ComGlobA of global commutative compositions—except when explicitly mentioning the contrary.τ (M ) : LinR (R. (rf )(x) = rf (x). it evaluates to LinR (GI . For a simplex σ of n(GI ). Lemma 26 The kernel of RGI is the module CGI of constant functions. we abbreviate LinR (GI . we deduce a R-linear homomorphism R. i. r ∈ R. and we have proven that the transition morphisms by restriction of aﬃne functions are surjective. and the map f → R.15. f evidently extends to the restriction A@h|Kσ .4). Suppose that we have an inclusion of simplexes σ ⊂ τ of n(GI ). Then the ambient spaces of the charts of these simplexes are the same. ∩ (σ) → A@R R (see lemma 6 in section 8. The set nΓ(σ) of these functions is provided with the structure of an R-module by the usual addition and scalar multiplication of function values: (f + g)(x) = f (x) + g(x).f is R-linear and functorial in σ. For the following construction.2) If we have a subcomplex situation N ⊂ M on GI . The corresponding complex of aﬃne functions is denoted by nΓ(GI ). ∩ (σ). Take a chart ∩(σ) of ∩σ in A@R Fσ . In the particular case M = A@R R.3. M ). we have a morphism of module complexes on GI : RGI : nΓ(GI ) → GI∗ (15. the inclusion of local compositions ∩τ ⊂ ∩σ are in bijection with an inclusion of local compositions Kτ ⊂ Kσ ⊂ A@R N for a speciﬁc module N . MODULE COMPLEXES 351 In the sequel of this chapter.

Deﬁnition 52 Let P be a property of modules. Let f /IdA : GI → H J be a morphism of A-addressed commutative global compositions. If M is any subcomplex of nΓ(H J ) which contains the constant complex CH J . QED. and if nΓ is injective (in particular. Therefore. GLOBAL CLASSIFICATION Proof. The function f is in the kernel iﬀ its values R. if M is any subcomplex of nΓ(H J ). we have the short exact sequence 0 → CH J → M → M/CH J → 0 and therefore the corresponding short exact sequence ι ι ι 0 → f /IdA CH J → f /IdA M → f /IdA M/CH J → 0 . and therefore. QED. ι ι ι ι ι (15. the map is R-linear. Since we have the zero address. we have an isomorphism nΓ/C → GI∗ . its induced complex f /IdA M is mapped R-linearly onto what is called the retracted module complex M |f /IdA ⊂ nΓ(GI ). and this in turn means that f is constant on ∩σ. Lemma 27 If A is the zero address A = 0R .4) In particular. Proof. If M is a module complex over GI . Moreover. we have CH J |f /IdA ⊂ CGI . if R is a ∼ ﬁeld). Take a simplex σ in n(GI ). any ∩σ is isomorphic to an embedded local composition in the same ambient module. each restricted morphism f |∩σ /IdA : ∩σ → ∩σ gives rise to a map f /IdA nΓ(H J )(σ) = nΓ(H J )(σ ) → nΓ(GI )(σ) by right composition with this restricted morphism. the morphism RGI is surjective. In particular. and any linear map l : R.f (si − sj ) = f (si ) − f (sj ) vanish. M is called injective (projective) iﬀ the module M (σ) is so for every simplex σ.352 CHAPTER 15. if M = CH J . it is said to share property P iﬀ all its values M (σ) do so. ∩ σ → R extends to a map on the ambient module. and its image σ under the associated simplicial map. Then.

Consider the module complex BN deﬁned on simplexes σ by the expression BN (σ) = LinR ((N (σ)/C) . Suppose that N is a module complex of A-addressed aﬃne functions on GI which contains the constant complex C = CGI . and orientation in Euclidean three-space.7) . In particular. there is a canonical projective system of R-linear maps Γ(H J ) → f /IdA nΓ(H J )(σ) → nΓ(GI )(σ) and therefore we have a canonical R-linear map Γ(f /IdA ) : Γ(H J ) → Γ(GI ) which is a contravariant functor Γ : ObGlobA → R Mod.6) 15.1. N (σ)/C)) (15.2 Bilinear and Exterior Forms Summary. –Σ– Until now. no considerations regarding angles and similar properties from Euclidean geometry have been made in the theory of local or global compositions.1. Bilinear and exterior forms capture the language of classical geometry of angles. This formalism is built on the systems of global functions. ι ι (15. we have this surjective R-linear homomorphism f /IdA nΓ(H J )/CH J → nΓ(H J )/CH J |f /IdA .15. ι ι (15.5) Finally. Here is the formal framework for such perspectives. for M = nΓ(H J ). distances. by construction of global sections. MODULE COMPLEXES which projects to an exact sequence on GI as follows: ι ι ι 353 0 E f /IdA CH J E f /IdA M E f /IdA M/CH J E 0 c c ι ι c c c c ι 0 E CH J |f /IdA c E M |f /IdA c E M/CH J |f /IdA c E 0 0 c E CGI ι c E nΓ(GI ) c E nΓ(GI )/CGI E 0 including the deﬁnition of M/CH J |f /IdA .

2. GI∗ (σ)) → LinR (A@R R. ∩ (σ) → (A@R R. we also speak of the bilinear form x.3 in mind we have a canonical R-linear map LinR ((N (σ)/C) . GI∗ (σ)) and the right side is LinR ((LinR (R.8) The right side describes the bilinear forms on the space of A-addressed points in the module R. respectively. Take global sections β ∈ ΓBM . GLOBAL CLASSIFICATION with X being the R-linear dual of X. ∩ (σ) ). ∩ (σ) ) . We therefore associate a bilinear form β(x) with each linear map x : (N (σ)/C) → N (σ)/C. Let f /IdA : GI → H J be a morphism of A-addressed commutative global compositions. Let M. Then we have a canonical map Bf : BN → BM (15.10) . the pair (GI∗ . and therefore a map LinR ((GI∗ (σ)) . N be submodules of nΓ(GI ).3. γ ∈ ΓBN . ∩ (σ) → (A@R R. even when we cannot instantiate an associated form β(x) (which is the case for general address A). Deﬁnition 53 With the above notation. With the injection 15. and with the evident transition homomorphisms. u : A@R R. A@R R)) . ∩ (σ) ) . ∩ (σ). (15. lemma 81 identiﬁes to LinR ((A@R R. ∩ (σ). LinR (R. ∩ (σ). ∩ (σ)) ). (A@R R. ∩ (σ).3. nΓ(H J ). respectively.9) N/C|f /IdA M/C of quotients modulo constants. N (σ)/C)) → LinR ((GI∗ (σ)) . both containing the constants. and in this framework. (A@R R.354 CHAPTER 15. Then appendix E. We now assume that A = Rn . ∩ (σ)) ) which gives an R-linear map βσ : BN (σ) → LinR (A@R R. 0 ≤ n natural. Suppose that we have an inclusion of the ι ι restriction of N under f /IdA in M : ι N |f /IdA and therefore an inclusion ι M (15. according to appendix E. proposition 85 yields linear maps d : A@R R. ∩ (σ)) . A@R R)) which.2. A@R R. β) with a global section β ∈ ΓBN on GI∗ is called an N -formed global composition. ∩ (σ).

So this short deviation is not only an application of the general formalism of global compositions. and orientation. .3].j x1 . xn . We shall see below (16. 2 The Gram form veriﬁes the identity Ω(x . the (zero-addressed) points of GI are viewed as being momentary positions of atoms of a molecule in R3 . ι ι 15. . . (15. and for the zero address. . VII.9). “Molecules” Summary. If Ω ∈ ΓΛn (N/C) is a global exterior n-form. As with bilinear forms.1. y ) = det((β(x . with the evident nth exterior power morphism whenever we have the above f inclusion (15.e. They emerge from the latter by adding bilinear and exterior forms. For this situation. it also shows that enriching the inner structure of a global composition may cause substantial diﬃculties in the understanding of the corresponding category. γ) if ΓBf (γ) = β. i. are very similar to representable global compositions. . Given a submodule complex N of the complex of aﬃne functions on GI . Geometrically this means that compositions may be “deduced” from molecular structures by abstraction from angles. we ask that • dim(nΓ(σ)/C) ≤ 3 for all charts σ of GI . we have an evident complex Λn (N/C) of exterior n-forms.15. . • we are given an exterior 3-form Ω which is β-normed. as they are considered in the Dreiding–Dress–Haegi theory. β) → (H J . we say that the morphism f /IdA is a morphism of formed compositions f /IdA : (GI . y )) . The atom species are parametrized by natural numbers. we call (GI . . Ω) an oriented global composition. . and therefore of molecular structures. special formed and oriented global compositions may represent molecular structures such as have been considered in [129]. We consider the speciﬁc diﬀerence of such an abstraction in the musical perspective: What does music gain after adding “molecular” information? –Σ– For the coeﬃcient ﬁeld R = R of real numbers. Molecule structures. distances.1. MODULE COMPLEXES ι 355 over f /IdA and therefore one of global sections ΓBf : ΓBN → ΓBM . see also [196.4) that classiﬁcation of formed and oriented global compositions. . Ω) → (H J . • GI is formed by a symmetric bilinear form β which is positive deﬁnite on all charts. . we have the concept of a morphism f /IdA : (GI .11) ι Deﬁnition 54 With the preceding notation and hypotheses. . containing the constants.3 Deviation: Compositions vs. . . x )Ω(y . ∆) of oriented global compositions if ΓΛn (∆) = Ω. is signiﬁcantly more diﬃcult than classiﬁcation of global compositions.. Moreover. yn of elements in the chart vector space. for all sequences n n 1 1 i j i. y1 . the Gram identity Ω2 = det(β) is valid2 . and their distribution on GI is deﬁned by a marking application α : G → N.

15. and by a theorem which tells us how to rebuild compositions as quotients from such special modules.2.e. The second device for classiﬁcation is the global standard composition. canonically associated with the composition’s nerve. an α-marked. The second method is the construction of quotient compositions from given modules of aﬃne functions. Combining these methods. χ)-deﬁned in the sense of [129]. and Ω-oriented (zero-addressed) global composition GI is called a global molecule. we set ai = (0.8) that there is an isomorphism A n → Rn ⊕ An+1 . section 11. β. the standard composition is a geometric realization of the composition’s nerve and thus depends only on combinatorial information. GLOBAL CLASSIFICATION Deﬁnition 55 A quadruple (GI . β-formed. We denote the canonical basis of Rn by (e1 . There is a natural projection from the standard object onto the generating composition. . The standard compositions are objects representing compositions with “notes in general position”.1 Global Standard Compositions Summary. 15. it is a free object in the category of global compositions and helps standardize the various special objects via corresponding special module complexes of aﬃne functions. we shall classify global compositions by the classiﬁcation of special module complexes of aﬃne functions in free objects. Ω) with the above properties. . their conﬁguration is as ‘free’ as possible from ‘occasional’ coincidences. . . Interpretable molecules are canonically associated with distinguished structures in the theory of Dreiding–Dress–Haegi.. –Σ– We ﬁrst want to broaden the provisional concept of a standard local composition given in deﬁnition 30. A global molecule which is isomorphic to an interpretation of a local molecule3 is called interpretable. the proof may be omitted in this context: Theorem 14 The molecular structures which are associated with interpretable molecules are (d.2. 0) for the 3 One with a single chart. Special module complexes on such standard objects allow the reconstruction of the original composition. en ). α. on the interpretation’s charts. bilinear forms and the orientation are induced from the local data. a. Recall from its construction (appendix ∼ E..2 The Resolution of a Global Composition Summary. A morphism between global molecules is a morphism of global compositions which respects the additional structures as discussed above.e. . .3.356 CHAPTER 15. In fact. i.3. . we denote by A n the n + 1-fold coproduct n+1 A of A. and for any element a ∈ A and 0 ≤ i ≤ n. i. . . –Σ– This section introduces two methods which are used in classiﬁcation theory: The ﬁrst is the global standard composition. . Given a module A in R Mod and a natural number 0 ≤ n.

and if s. tc } ⊂ [0. In fact. m = card(G) − 1} of natural numbers. . we have the canonical injection iq : A ∆c → A ∆m via σj → σtj . n0 (GI ))). sn ) is any sequence of A-addressed points in M . ai ) = aﬃne if i > 0. Show that the assignment A → A ∆n and α → Id/α : A ∆n → B ∆n is a functor ∆n : c Mod → ObLoc on the category c Mod of modules over commutative rings and diaﬃne morphisms. and the formula si = f · σi is immediate. for any subset q = {t0 . example 102. For n . . . indexed corresponding to increasing values. .2. By construction. we deﬁne the global standard composition A ∆n at address A by the interpretation of the local standard composition A ∆m which is given by the present covering of [0. n. m] of c + 1 elements. m].0 .0 (a) (linear) for i ≥ 0. . . we denote by Id : A ∆n → B ∆n the “identity” σi → σi . with associated local composition S = {s0 . . then there is exactly one morphism of local compositions (s. sn } ⊂ A@R M . we shall represent the nerve n(GI ) by an isomorphic standard representative nerve n induced by a covering of the natural interval [0. . m] = {0. The universal property of this global standard composition reads as follows. Take the category Covens of coverings of sets4 .14) This morphism is in fact deﬁned by the universal property of the coproduct and is mediated by the following aﬃne function f : A n → M : Write si = eti · si. Exercise 33 If A. = (s0 . f (ei ) = ti − t0 (linear) for i > 0. with linear if i = 0. . (15.) : A ∆n → S : σi → si for i = 0.2. respectively. 2. .1. . . the zero element is denoted by e0 . We are also given a standard atlas of A ∆n . (15. . . and consider the covariant functor A Covn : ObGlob@A → Sets : GI → HomCovens (n . THE RESOLUTION OF A GLOBAL COMPOSITION 357 n + 1-tuple in An+1 having a at its i + 1-th position and zero else. S M od. Then we have f (e0 ) = t0 . f (ai ) = si. σi (a) = (ei . . A-addressed composition A ∆n ⊂ A@R A n which is called the A-addressed local standard composition of dimension n. Consider base changes α : A → B. B are two addresses. This deﬁnes the standard atlas. (15. sitting in R M od. it has the following property: If M is any R-module. (G.15) Then we have this result: 4 See appendix H. To deﬁne global “free” objects among the A-addressed objective compositions with ﬁnite charts. We have the inclusion morphisms σi : A → A for 0 ≤ i ≤ n.13) n (15.15. . 1.12) This deﬁnes a local. 3.

In particular. not an isomorphism in general. however. due to the universal property of the global standard compositions.18) The following deals with the reconstruction of the category ObGl@A from its subcategory of free objects. every morphism f /IdA : GI → H J can uniquely be lifted to a corresponding morphism res resolutions to make the diagram res A ∆ GI − − − → ∆ H J −−− res /Id A ι ι f /Id ι ι f /IdA of resGI H J /IdA (15.19) (15. (G. no (GI )). This complex is used to reconstruct the generating composition from the standard composition. In particular.2. this object and the morphism resGI /IdA being called the resolution of GI ..358 CHAPTER 15.17) GI −−→ −− f /IdA HJ commute. if we take the standard nerve n = n (GI ) ∼ of the nerve of GI and then the corresponding ‘identity’ morphism Id : n → (G. The proof is left as an exercise. 15. We therefore have a resolution functor res@A : ObGlob@A → ObGlob@A and a natural transformation δ@A : res@A → IdObGlob@A .e. –Σ– 5 It is. Clearly. The projection of the standard composition onto its generating composition canonically induces a module complex of global aﬃne functions on the standard composition. n0 (GI ))) → HomObGlob@A (A ∆n . we have a bijection HomCovens (n . i. (15.16) ∼ with the notation ∆GI = A ∆n (GI ) .2 Compositions from Module Complexes Summary. GLOBAL CLASSIFICATION Proposition 15 The functor A Covn is representable by the standard global composition A ∆n . we obtain a corresponding bijective5 morphism resGI /IdA : ∆GI → GI (15. . GI ) which is functorial in the A-addressed composition GI . the associated simplicial morphism n(resGI /IdA ) : n (∆GI ) → n (GI ) is an isomorphism.

the dotted points σi : A → L deﬁne the universal map HL : A n → L . ˙ The problem is that the evaluation is not a morphism of local compositions in general.20). Exercise 34 Give a proof of the commutativity of diagram (15. Moreover.2. containing the constants C.21). ˙ ˙ f /IdA (15.21) .e. i. Next. We have to investigate suﬃcient conditions for the existence of a morphism. We then have a commutative diagram S − − → A@R LS −− A@|f T − − → A@R LT −− where |f is the R-dual of the canonical linear map |f : LT → LS . Then.20) of the retracted resolution complex of H J in the resolution complex of GI . LT . Deﬁnition 56 Let S ⊂ A@R U be an A-addressed objective local composition in the R-module U . this assignment commutes with the morphism of the resolution functor. Call this complex the resolution complex of composition GI .. ˙ ˙ This construction yields a morphism f˙/IdA : S → T of local compositions in the ambient spaces LS . and that we would like to construct a kind of “quotient” composition whose aﬃne functions are those of M . With this technique. for a morphism f /IdA : GI → H J we have a canonical inclusion nΓ(H J )|f /IdA ⊂ nΓ(GI ) ι ι (15. T ⊂ A@R V and a morphism f /IdA : S → T. R) of L is deﬁned by s(a)(l) = l(s)(a). suppose we are given two local compositions S ⊂ A@R U. let S = A ∆n ⊂ A n . together with a module LT ⊂ Γ(T ) whose retract LT |f /IdA is included in a module LS ⊂ Γ(S). The next step deals with the reconstruction of GI from nΓ(GI ) and the related question of classiﬁcation of global compositions by use of the resolution complex which is suggested by the functorial relation (15. THE RESOLUTION OF A GLOBAL COMPOSITION 359 The resolution functor res@A and its associated natural transformation δ@A give rise to a module complex of aﬃne functions ∆nΓ(GI ) = nΓ(GI )|resGI /IdA in nΓ(A ∆GI ). The generic situation from the preceding constructions is that we are given a module complex M ⊂ nΓ(GI ). we may associate a global composition with a module complex N ⊂ nΓ(A ∆n ) of aﬃne functions in the standard composition A ∆n of a standard covering n . we have this guarantee: In fact. For a submodule L ⊂ Γ(S) of aﬃne functions on S. We ﬁrst look at the local situation. the evaluation map ˙ : S → A@R L into the A-valued points of the dual module L = LinR (L. and we have interpreted ˙ ˙ : A ∆n → A@R L as a morphism of local compositions. In the special case which is of interest. respectively.15. for each global A-addressed composition GI .

Therefore. the diagram (15. there is a function of N on this chart which separates these points. Deﬁnition 58 We call this composition A ∆n /N the N -quotient of A ∆n → A ∆n /N from diagram (15. suppose we are given such a module complex N of aﬃne functions.23) has bijective horizontal arrows. are all surjective. With this. the dot map σ → A@R N (σ) is injective. A ∆n . i. τ ⊂ σ. This is certainly the case for retracted function modules from resolution morphisms.. and with LS = A ∆c(σ) ⊂ A@A N (σ). then we have injective vertical arrows in the corresponding commutative diagram ˙ −− A ∆c(σ) − − → A@R N (σ) (15.21) to the situation where S = c(σ) .360 CHAPTER 15. and the images ∩σ are injected into the limit A ∆n /N . the colimit diagram (15. we have the resolution complex ∆nΓ(GI ). the dot maps are all bijective onto the images since the vertical maps in diagram (15. the resolution complex is separating iﬀ the . if N is separating. if resGI : A ∆GI → GI is the resolution of the A-addressed composition GI . this means that for any pair of points on any chart.23) is denoted by /N . if we apply the construction from diagram (15. we have a commutative diagram of sets A ∆c(σ) −−→ −− ˙ A ∆n ∩σ A ∆n (15.22) inclusion A@res A ∆c(τ ) − − → A@R N (τ ) −− ˙ where res is the restriction map. Setting A ∆n /N = colimn ∩ σ. LT = N (τ ). So. iﬀ for every zero-simplex (chart) σ ∈ n0 (GI ). Deﬁnition 57 Call a module complex N ⊂ nΓ(GI ) separating. For a simplex ∼ σ of the nerve n(A ∆n ) → n we note c(σ) = card(∩q∈σ q) − 1.22) are injective. if in the above situation N ⊂ nΓ(A ∆n ) is separating. we have constructed a canonical global composition and a bijective morphism from the free object to a global composition which is deﬁned by the functions of N . We write ∩σ = A ∆˙c(σ) .23) −− − − − − −→ /N =colim˙ /N induced by the dot morphisms of local compositions. we shall tacitly assume that all module complexes of aﬃne functions have surjective transition morphisms.e. So these images cover the limit and the images of the zero-dimensional simplexes build a canonical atlas of a global A-addressed composition. and suppose that the restriction transition morphisms N (τ ) → N (σ).23) becomes a bijective morphism of A-addressed global compositions. In the following discussion of classiﬁcation. GLOBAL CLASSIFICATION To this end. The morphism In particular. Intuitively. and T = A ∆c(τ ) ⊂ A@A c(τ ) for simplexes τ ⊂ σ of A ∆n . and therefore get a surjective morphism of diagrams of local compositions A ∆c(σ) → ∩σ over n . So. Clearly.

and we may take a linear form h ∈ U such that h(s(a)) = h(t(a)) which means that the induced aﬃne form l = A@h|σ separates s from t.24) which stem from morphisms of global compositions except—possibly—for f . Proof.. if s = t are two points in the chart σ ⊂ A@R U .15. Proposition 16 If the A-addressed composition GI admits a projective atlas. in which case we also say that GI is separating. sn } ⊂ A@R U of cardinality6 n + 1. .4. QED. this is a property which is invariant under bijective morphisms among A-addressed compositions. see appendix E.2. i. an atlas whose charts have projective R-modules. Then we are given a commutative triangle of covering set isomorphisms A ∆GI /∆nΓ(GI ) © I A ∆GI /∆nΓ(G ) f d d resGI d d d E GI (15. THE RESOLUTION OF A GLOBAL COMPOSITION 361 complex of aﬃne functions nΓ(GI ) is. . we have the corresponding commutative triangle of set bijections: A ∆n /∆Γ(S) © A ∆n /∆Γ(S) f d d˙ d d d E S (15. For the module Γ(S) of aﬃne functions g : S → A@R R on S.2. Now. .e. . we may concentrate on the local situation.25) which becomes a triangle of morphisms of local compositions if we can derive f from an aﬃne morphism F : Γ(S) → U of ambient modules. then GI is separating. the canonical bidual map U → U is injective for each ambient space U of a projective atlas. Proposition 17 Suppose that GI is separating. Of course.26) U 6 Recall − − → (U @R R) −− b that in this discussion. On the level of ambient modules. In this case. This means that we are given a separating local composition S = {s0 . we are considering global compositions with ﬁnite charts. there is a ∈ A with s(a) = t(a). we have the following commutative square of aﬃne morphisms: A n −−→ −− S ˙ S Γ(S) q (15. So there is a ∆nΓ(GI )-quotient if the complex of aﬃne functions nΓ(GI ) is separating. To look for conditions when f is a morphism.

A suﬃcient condition for such inverse morphism is that the surjection q : U @R R → Γ(S) has a section. This condition is certainly satisﬁed if Γ(S) is projective. Therefore. the map f in diagram (15. Moreover. one has IdΓ(S) = q · p .. In fact. respectively.25). Exercise 35 Verify the commutativity of diagram (15.e. If U is ﬁnitely generated and projective.e.. GI is separating. in this case. i. i. Proof. GLOBAL CLASSIFICATION where the b(u)(t) = t(u) is the extension of the bidual. S. and by the local diagram (15.28) so that we are now left with the question of characterizing those module complexes of aﬃne functions in A ∆n which could give rise to compositions having this free object as their resolution. where q : U @R R → Γ(S) is the canonical ˙ ˙ surjection. ∆nΓ(GI ) = nΓ(A ∆GI /∆nΓ(GI ))|/∆nΓ(GI ) (15. then the horizontal arrow f in the factorization diagram of proposition 17 is an isomorphism. a ﬁnite product of commutative ﬁelds.e. the bidual U → U is an isomorphism (appendix E. the factorization diagram of proposition 17 is a diagram of global compositions. The factorization morphism f has a chance to become an isomorphism if we succeed in constructing an ambient space morphism in the other direction. and we may take p · b to go back from U to Γ(S) . a left inverse p : Γ(S) → U @R R.26) by use of the universal property of the (n + 1)-fold coproduct A n . the retracted module can also be recovered from the quotient composition. and the dual r : (U @R R) → U of the inclusion U → U @R R deﬁnes a right-inverse i ofb. .24) is a morphism of global compositions. i.27) which yields the required morphism for f in diagram (15. QED. applied to the charts of this situation. So we have this result: Theorem 16 If GI has a ﬁnitely generated projective atlas and projective functions. This means that Theorem 15 If the composition GI has a ﬁnitely generated.362 CHAPTER 15. S are the universal morphism associated with S. projective atlas7 . This means that in this case.4..2). This condition is evidently satisﬁed if R is semi-simple. This implies that on the duals. we are given the commutative triangle A ˙ S © Γ(S) i·q n d d S d d d E U (15. and where S. by proposition 16.27). 7 All charts have ﬁnitely generated projective modules. we are able to reconstruct GI from its retracted aﬃne functions on the resolution. We say that the global composition has projective functions iﬀ its complex of aﬃne functions is projective on the zero simplexes.

–Σ– Let us ﬁrst recall the overall situation of resolutions. if our isomorphic global compositions GI . i. H J have a ﬁnitely generated projective atlas and projective functions.2. as described in section 15. their retracted function complexes NGI .. Since the charts of the quotient have ambient spaces N (σ) . By retraction (see (15. the quotient composition has charts Γ(S) which are again ﬁnitely generated projective and separating. and we have NGI = NH J |resf .1. ORBITS OF MODULE COMPLEXES ARE CLASSIFYING 363 Clearly. Secondly. we may suppose that it is ﬁnitely generated projective.31) (see deﬁnition 58).n of the standard composition A ∆n identiﬁes to a subgroup of the symmetric group Sm+1 of permutations of A ∆n if the standard covering is deﬁned on the integer interval [0. What are the functions of the quotient? Since the . we have a bijective morphism r : A ∆n → A ∆n /N (15.15. the automorphism group SA. the quotient has a ﬁnitely generated projective atlas.3. under these conditions. the ﬁrst condition on such a module complex is that it is separating.30) f So.e.n of representative module complexes on A ∆n ret : RepA. we have the functorially associated standard resolution ∆GI = A ∆n with the standard covering n of GI . So we proceed to the analysis of module complexes N ⊂ nΓ(A ∆n ) which are separating. 15.n : (N. The second requirement is the one we had in theorem 16. NH J on the standard composition A ∆n are representative. and contain the constant functions. Conversely. g) → N |g. (15. we obtain a classiﬁcation frame for global compositions. this group acts from the right on the set RepA. we have a commutative diagram with horizontal isomorphisms: resf −− A ∆ n − − → A ∆n res J res I (15. if we are given a representative module complex on A ∆n . which we now abbreviate by f if no confusion is likely. If a global composition H J is isomorphic to GI via a morphism f /IdA : GI → H J . its modules on the zero-dimensional simplexes are so. By means of the representation of a composition via its module complex of functions on the purely combinatorial standard composition. The ﬁrst requirement is in particular the case for the global compositions in theorem 16 above since the charts have ﬁnitely generated projective ambient modules.3 Orbits of Module Complexes Are Classifying Summary. ﬁnitely generated projective. For a global composition GI .n → RepA. The third and last requirement is obvious: N should contain the constant functions. and the resolution bijection resGI : ∆GI → GI . Also.29) G H ι GI − − → H J −− By the universal property of the standard compositions.4)). m] as discussed above. call these complexes representative.n × SA.

and since N is ﬁnitely generated projective. a ﬁnite direct product of commutative ﬁelds (see appendix E.32) and. the module complex of aﬃne functions on A ∆n reads as follows: We recall from section 15. In particular.n . So we are given a representative module8 of aﬃne function N ⊂ Γ(A ∆n ) in a local standard composition. To begin with. Since this retraction is isomorphic to the function module of the quotient. Conversely. . and on an argument a ∈ A evaluates to f (σi )(a) = t + n(σi )(a) (15. We discuss the action of the automorphism group of the standard composition on module complexes since it induces isomorphism classes of global compositions.2. we may absorb the constant t in n.n with N = M |g. in one direction.2. This bijection is induced by the retraction of the function module complex to the resolution A ∆n . N ∈ RepA. and we have proven this: Theorem 17 The orbit space RepA. we may write F = et · n .3. in the other.3. i. which means that nΓ(A ∆n /N )|r) = N . An aﬃne function f on the image ∆ ⊂ A@R N is induced by an aﬃne morphism F : N → R. and is separating.1 that the standard covering n 8 So N is ﬁnitely generated. So conversely. the latter is also representative. this classiﬁcation result is valid for the global compositions having as their address a module A over a semi-simple commutative ring R.n is in bijection with the set of isomorphism classes of A-addressed global compositions with projective functions and ﬁnitely generated projective atlases which have a covering complex isomorphic to n . contains the constants. then we have a commutative diagram with isomorphism on the horizontal arrows A ∆n rN −−→ −− g A ∆n /N − − → −− r A ∆n M (15. and an automorphism g ∈ SA. So the retraction of the induced function f on σi . we are interested in a more explicit description of the action of the automorphism group SA.n on the set RepA.n of representative module complexes on the standard composition A ∆n . projective.e.n /SA. theorem 48). isomorphic representative module complexes give rise to isomorphic quotients.33) A ∆n /M by the factorization of the resolution map g · rM through the quotient A ∆n /N . –Σ– By the above classiﬁcation theorem 17. GLOBAL CLASSIFICATION simplicial complexes are isomorphic. it is suﬃcient to consider the local situation on a chart. 15.1 Combinatorial Group Actions Summary.364 CHAPTER 15. and by the quotient composition on a given representative module complex on A ∆n . if we are given two representative module complexes M. with the bidual n of an element n ∈ N . since the constants are in N ..

. an automorphism of the underlying global standard composition boils down to a permutation of the interval [0. So we have a direct equivariant decomposition (A@R R)c+1 = ∆(A@R R) ⊕ (A@R R)c which carries over to the submodules: N (σ) = A@R R ⊕ N (σ)red .3. ﬁnitely generated projective as well as separating in dimension zero. m] by subsets (the zero simplexes) σ = {t0 . we have a commutative diagram ∆(A@R R) Id c ∆(A@R R) ⊂ ⊂ ∼ E N (τ ) resτ. Moreover. m] which is compatible with the covering. with N (σ)red ⊂ (A@R R)c . we have a corresponding local standard composition A ∆c(σ) . for any two simplexes τ ⊂ σ.σ pr c c c c ⊂E N (σ)red (A@R R)c(σ) .12) in section 11. . there is an element n ∈ N (σ) such that its coordinates ni and nj diﬀer.3. and for any pair i = j of indices between 0 and c(σ).σ c c E N (σ) ⊂ E (A@R R)c(τ )+1 pr ⊂ c c E (A@R R)c(σ)+1 with surjective vertical arrows induced by the projection to the right side. In particular. . Further. On each simplex σ of this covering. What does it mean that it is separating? It means that for any zero-dimensional simplex σ. tc(σ) } of c(σ) + 1 elements. ORBITS OF MODULE COMPLEXES ARE CLASSIFYING 365 may be given in form of a covering of the integer interval [0.3.2: 0E d E A@R RE ∆ (A@R R)c+1 E (A@R R)c E E E 0 E which is equivariant for the given permutation group action. and its function module is Γ(A ∆c(σ) ) → (A@R R)c(σ)+1 whereas the constants correspond to the diagonal submodule ∆(A@R R) ⊂ (A@R R)c(σ)+1 . on every module N (σ) acts by induction from the permutation of components on the supporting direct sum module (A@R R)c(σ)+1 . Recall the diagonal embedding and its factorization from formula (11.14) of section 11.2. this is a ﬁnite group action which.15. The action on the reduced factor has been described in formula (11. So the above projection diagram reduces to N (τ )red ⊂ E (A@R R)c(τ ) resτ. and therefore by the standard atlas injections iσ : A ∆σ → A ∆m : σj → σtj . and our module complex N must be a module ∆(A@R R) ⊂ N (σ) ⊂ (A@R R)c(σ)+1 for each simplex σ.

c ⊕ (A@R R)i−1 ⊕ 0 ⊕ (A@R R)j−i ⊕ 0 ⊕ A@R R)c−j−1 if i = j. See [63] for scalar extensions on duals.j = ∆A. ∼ (15.r (S) = {V ⊂ S ⊗R WA. in the local situation of a local standard composition A ∆c of dimension c. We have these quotients S ⊗R WA.4. if s : R → S is an R-algebra.c : (A@ R)i−1 ⊕ 0 ⊕ (A@ R)c−i if i = j.j vanishes on the module A. j entries are equal. This section is devoted to the classiﬁcation theorem: There is an algebraic scheme whose rational points represent certain isomorphism classes of global compositions.m+1 (S). i pi.c V ∈ Xr (s). R R Vi.c → (S ⊗R A)@S S = (15. on a ﬁnitely generated projective module complex Nred = N/C and on the reduced group action.2 Classifying Spaces Summary.2. consider the diﬀerence projections pr if i = j. 15.c . we have A. we have to deal with the separation property.c /S ⊗R Vi. –Σ– To begin with.35) prj − pri if i = j.c . So. The separation fact. We are in a similar situation as in the discussion of the local classiﬁcation in section 11. i.i. S ⊗R (A@R R) → S ⊗R (A ⊕ R) → (S ⊗R A) ⊕ S → (S ⊗R A)@S S. while the other entries vanish.c Xr (s) = GrassWA.34) Since9 S ⊗R WA. Deﬁne these submodules of WA.c = (A@R R) .i. Let us assume henceforth that A is locally free of rank m.j : S ⊗R WA. let ∆A. this functor parametrizes S-modules of aﬃne functions on S⊗R A ∆c which have locally free10 quotients of rank r.2)..21)) with respect to a selected point (formula (11. where pri is the projection onto the i-th factor. Here.3. Then we have S ⊗R Vi. 10 The property “locally free” is equivalent to “ﬁnitely generated projective”. if we make the general hypothesis that A is projective of ﬁnite type.j.c | S ⊗R WA. 9 In ∼ ∼ ∼ ∼ .e.c be the submodule of WA.c → ((S ⊗R A)@S S)c .j.j ∈ GrassWA.3. For any pair 1 ≤ i ≤ j ≤ c. Next. A.22)).c whose i. GLOBAL CLASSIFICATION It is known that N (σ) is projective of ﬁnite type iﬀ A and N (σ)red are so (see appendix E. The scheme is a ﬁne moduli space in the sense of David Mumford.366 CHAPTER 15. where we had to deal with subfunctors of the Grassmannian (formula (11.j → (S ⊗R A)@S S. The separation property means that no diﬀerence projection pi. the S-valued ambient module is the scalar extension S ⊗R WA. Writing WA.c we consider a (covariant) Grassmann functor Xr : ComRings|R → Sets on all commutative c R-algebras.2. For two indices 1 ≤ i = j ≤ c.c /V = locally free of rank r}. see appendix F. theorem 58.c . we may concentrate on the reduced part of the module complex.

an action which is induced by the automorphisms of the underlying standard composition. theorem 59. the separation property deﬁnes an open subscheme of the Grassmannian which is also invariant under all permutations of the indexes.2. A.c In other words. 2. .j for all 1 ≤ i ≤ j ≤ c. pr1 E E C A. r. This evidently is a closed. . Since all combinations of coranks are possible. i = A.5.c latter condition deﬁnes a subfunctor Xr.c.j scheme DrapA. that the quotient scheme of orbits of the ﬁnite group action of SA. σ2 .. In particular.15. ORBITS OF MODULE COMPLEXES ARE CLASSIFYING 367 condition for V then means that we do not have V ⊂ S ⊗R Vi.c A. The A.m+1 It is a closed subscheme since a section of structural morphism of the Grassmannian is a closed immersion. we need to know that N (σ1 )|ρ = N (σ2 )|ρ . In fact.c of the scheme GrassA. see appendix F.3. equivariant condition and we have shown that the representative module complexes over the R-algebra S which are of any rank conﬁguration on the respective simplexes deﬁne the S-valued points of a locally closed subscheme C A. So if ρ is any simplex whose vertexes are σ1 . A.n µ .j .i. = r. we know from appendix F.c 1..c deﬁned by the m + 1-codimensional ﬂag component being ﬁxed to S ⊗R Vi. = i=1.i. A. there is a subscheme J n of a projective Spec(R)-scheme of ﬁnite type such that its S-valued points J n (Spec(S)) for an R-algebra S are in bijection with the classifying orbits of module complexes N in S⊗R A ∆n which are locally free of deﬁned co-ranks on the zero-simplexes of n . So locally.i. k.k Ori i of the ﬁber product of Grassmannians over Spec(R). To obtain the eﬀective candidates.n ×Spec(R) SA.6.n exists.. this theorem gives the classiﬁcation of any global composition which is addressed in a ﬁnitely generated R-module A. Or.c over Spec(R). lemma 86 for details.c.n in the sense of the diﬀerence cokernel of the group action µ and the ﬁrst projection pr1 C A. see appendix F. This means that we have this result: Theorem 18 For an address A which is locally free of rank m over the commutative ring R.6).n of a projective scheme over Spec(R). our module complex is given by an open subscheme Or.c A. if the ground ring R is semi-simple.c. and since any ﬁnite number of points is contained in an open aﬃne subscheme in this situation (projective schemes. we have to take V in the (open) complement scheme Or of all closed subA. . So if the zero-dimensional simplexes of the global standard composition are A ∆ci . we have to consider coincidence on restrictions.j of Grassr . we should also take the coproduct OA.c schemes Grassr.j of Xr which is represented by a closed subscheme GrassA. it is the closed subscheme of the ﬂag r r. of all these open subschemes on which the permutation (automorphism) group of the covering acts.

.

unsere gegenw¨rtigen Schwachstellen a zu beheben.71] Summary. 369 . Pierre Boulez [60. In contrast to general classiﬁcation. This will be dealt with in the next chapter. This chapter exposes criteria for characterizing interpretable compositions in terms of classifying spaces. the meaning of the classiﬁcation techniques for musicology cannot be underestimated. so this entire approach from algebraic geometry opens the subject of germinal vs. and this is the view we shall deal with in this chapter. p. However. It is the turning point between thinking and making music. interpretable molecular structures are diﬃcult to classify. it should be kept in mind that this subject is far from pure mathematical exposition. We also review global enumeration theory as well as global American Set Theory. derived ideas in musical composition. the question of those compositions which can eﬀectively be played on ensembles of musical instruments is embedded in the perspective of algebro-geometric specialization. This is a central issue since interpretability yields access to instrumental parameters for the physical ‘rendering’ of a compositional structure. sie zum Vorbild zu nehmen. between mental construction and physical realization. war mir daran gelegen. die zur Zeit die am weitesten entwickelte Methodologie besitzt. In particular. This is why aﬃne functions are so important: they yield the entire potential of parametrizing music in acoustic realms. –Σ– We are not going to give further musicological comments and interpretations of the preceding and the following classiﬁcation results.Chapter 16 Classifying Interpretations Weil die mathematische Methode die Wissenschaft ist. The interpretation of isomorphism classes in terms of points of a scheme also lead to a comparative theory of global compositions: We can now deal with the problem of which compositions are more generic than others. das uns helfen kann. II.

where ξ. if a global composition GI is interpretable by the covering J of a local composition K ⊂ A@R M . ∼ we have free function modules Γ({x}). and function modules which are locally free of deﬁned co-ranks on the zero simplexes of their resolutions. Take the ring R = Z of integers. and y = eη · u. a free module I = Rd of deﬁned rank d.2. quite pathological things may happen. this means: Lemma 28 Interpretable global compositions have ﬂasque function module complexes. and we consider the category ComGlobA of A-adressed commutative global compositions. This section presents a condition for a composition to be interpretable. we again assume that the address A is a module over a commutative ring R. –Σ– Pursuing the path of the last chapter. Then no aﬃne function on M can separate x from y since their diﬀerence is annihilated by any aﬃne function. Suppose that its function module is ﬂasque. This condition regards the restriction behavior of aﬃne functions which relates to ﬂasque sheafs. for an R-module M . and take a non-zero prime ideal m = (p) in R. and that the global functions are separating. Moreover. it should also be possible to characterize the interpretable compositions by their function module complexes. η ∈ R/m are two diﬀerent elements.2. and function modules which are locally free of deﬁned coranks on the zero simplexes of their resolutions. we have M @R R → eR · I since all linear forms on R/m vanish. {y}. and the (global) functions of K. essentially because the ambient module is not projective. by retraction yield all the functions of GI on their charts (=zero-simplexes). We therefore have an interpretation with locally free function modules and charts whereas the global functions are not separating. Consider a local composition L = {x. Γ({y}) → R ⊕ I of rank d + 1. if we consider the interpretation of L by the two singleton charts {x}. then we have a bijective morphism f /IdA : GI → K. Clearly. But we have the following: Proposition 18 Take GI in ComGlobA with A locally free of rank m. Consider the R-module ∼ M = I ⊕ R/m. CLASSIFYING INTERPRETATIONS 16. by the orbits of the retracted function module complexes. and where u : A → I is a surjection.370 CHAPTER 16. Moreover. the global functions are separating. Deﬁnition 59 A module complex on a global composition is called ﬂasque iﬀ every section on a zero-simplex stems from a global section. As we were able to classify those commutative global compositions in ComGlobA with A locally free of a deﬁned rank. . Example 30 Let us study such an example. and any address A. y} ⊂ A@R M with x = eξ · u. if GI is interpretable by a projective module M . If a local composition has ambient module M which is not projective. For example. this follows from proposition 16 of section 15. Then GI is interpretable.1 Characterization of Interpretable Compositions Summary.

Then an interpretation of GI which is guaranteed in proposition 18 can be constructed in an ambient vector space of dimension µ. 0). x2 .2.2 that an atlas of GI is given by the duals N (σ) of the functions on the given charts σ of GI . x4 . c = (1. U2 = {x1 .σ)) the maximum of dimensions of the modules associated with the charts σ of GI . U3 = {x3 . We shall discuss a further example of this type in chapter 17. x4 → c. x4 → c. but with a more semantic orientation.3.3.1. But in the general case. if the ring R is semi-simple. x4 }. x6 }. and R be an inﬁnite ﬁeld.16. We know from the quotient construction of global compositions from function modules which backs the classiﬁcation theorem 18 in section 15. x2 . x5 . and we have reconstructed GI as an interpretation. 1). In particular. x6 }. So on the one hand. d = (0. If we visualize this conﬁguration by means of a subdivision of each chart as a union 1 The ∼ image is a direct summand. x6 }. Here. x5 . x3 .. The three covering charts are all in bijection with a ‘square vertex’ composition S = {a = (0. and three charts for its covering. On the other. 1)} ⊂ 0@R R2 as follows: U1 → S : x1 → a. x6 → b. x3 → b. ∼ U2 → S : x1 → a. we have a split1 injection iσ : N (σ) → Γ(G) into the dual of the global function module since the latter is ﬂasque. QED. Exercise 36 Let A be the zero address.1. It has six points. CHARACTERIZATION OF INTERPRETABLE COMPOSITIONS 371 Proof. x5 . 0).e. so that the intersections of any two of these charts yield an isomorphism of a two-point composition in R2 . U3 } with U1 = {x1 . Take µ = maxσ (dim(R. x5 → c. and over the ﬁeld of real numbers. Before terminating this classiﬁcation discourse. see ﬁgure 16. x2 → d. G is injected into A@R Γ(G) with ambient module Γ(G) . and σ ⊂ A@R N (σ) are isomorphic. So the two charts σ ⊂ A@R Γ(G) . we should add examples of non-interpretable global compositions. G = {x1 . Example 31 We discuss a non-interpretable global composition GI in the zero address. ∼ U3 → S : x3 → b. x2 → d. the function module N (σ) and therefore its dual is locally free of deﬁned rank. a necessary and suﬃcient condition for module complex classes to yield interpretable global compositions is not known. b = (1. . we obtain a complete characterization of interpretable global composition via their function modules in the resolution given in theorem 18 in section 15. x2 . we simply give the examples as facts of the classiﬁcation discourse. x3 . x5 → a. i. and since by hypothesis. U2 . x6 → d. x4 . I = {U1 .

2 = f4. Therefore f5. x6 ) can be separated by f . Here.j = fi − fj . 3.1. and f5. we obtain a M¨bius strip.1: A non-interpretable composition and its resolution. Let f : G → R be an aﬃne function (we suppress the zero address and just work in the respective ambient spaces). 2. The resolution of GI in the o same surface representation is also shown.5 . 16.6 = f6. so 0 = f1. the charts are no longer plane compositions but tetrahedra in three space. CLASSIFYING INTERPRETATIONS 3 6 4 1 5 DG 1 p 2 4 G1 6 3 1 5 2 Figure 16.1. 5. x2 ). x4 ).372 CHAPTER 16.3 = f6.6 .5 . 4.2 = f5. it is suﬃcient to see that the global aﬃne functions on this composition do not separate points.3 . set f (xi ) = fi for i = 1. on chart U2 : f1.2 = f4.3 = f6. see ﬁgure 16.5 = 0.6 + f6. (x5 .1 Automorphism Groups of Interpretable Compositions Summary. we have on chart U1 : f1. –Σ– . We show that every ﬁnitely generated abelian group can be represented by the automorphism group of an interpretable composition. 6 and fi. of two triangular surfaces.5 . Since f is aﬃne on each chart. on chart U3 : f4. Let us show that GI is not interpretable. by deﬁnition. (x3 . and no one of the pairs (x1 . By the above results.

Lα.βi . Intuitively. . We shall construct a disjoint union M r Mm1 . as well ∼ as Aut(Ei ) → Zmi . .β ).β ) deﬁned by the arrow tails and heads: D Ex(D. and Aut(Mmi ) → Zmi . Lα.e. CHARACTERIZATION OF INTERPRETABLE COMPOSITIONS 373 Theorem 19 Let H be a ﬁnitely generated abelian group. . β} = {γ. i ∈ Z and arrows ai : vi → vi+1 .β .1) What are the automorphisms of such an extension? They are related to an underlying automorphism of the arrow set of D. For a positive integer m. and correspond one-to-one to the arrows of the graph D.β ) = Id. Aut(Lα. r. α. 1} ⊂ 0@R R for any pair 0 < α < β < 1 of real numbers such that α + β = 1. Then such an extension can be realized as an interpretation in R3 . i. and therefore we have this fact: Lemma 29 The automorphism group of Ex(D.δ are isomorphic iﬀ {α.2) does the job. .β are mapped to the graph vertexes corresponding to the arrow heads. Zmk . We have ∼ Aut(D∞ ) → Z. i ≤ r.r+k ∼ ∼ Ei (16. Clearly. i.β by D. . Lα. We then consider the extensions Ei = Ex(D∞ .. Consider the directed graph D∞ with vertexes vi . Aut(GI ) → Zr × Zm1 × . and Ei = Ex(Dmi−r . Mmi with Aut(M r ) → Zr . i = 0. . Clearly.β is rigid. . Lα.16. Lγ. . this means that we replace each arrow by a copy of Lα. . The cell of our interpretations is a zero-addressed local composition Lα. . . Mmk of mutually non-isomorphic connected ∼ ∼ interpretations M r . And on each chart Lα. the group of translations of Z. Map the elements 0 in the copies of Lα. i = 0. r < i ≤ r + k. Given a directed graph D without zero-loops2 .β for a given arrow. βi as indexes of mutually non-isomorphic cells Li = Lαi . i = r + 1.β = {0. Zmk . β. m − 1. Aut(Dm ) → Zm .. . Then there is a zero-addressed in∼ terpretation GI over R = R with H → Aut(GI ). .β . δ}. take the global composition Ex(D. . Suppose that we have H → Zr × Zm1 × . Zmk .β . Li ). . . any two Lα. 2 No arrows with identical domain and codomain. ∼ where we close the polygon by deﬁning vm = v0 . m − 1 and arrows ai : vi → vi+1 . Moreover. Take the group Zr × Zm1 × . We are now ready to deﬁne the interpretation with the required automorphism group. The proof idea runs as follows. consider the regular polygon graph Dm with vertexes vi . . Li ). the extension of Lα. and ﬁx r + k mutually diﬀerent pairs αi .. r + k. .. i ∈ Z. the chart morphism must be the identity since the charts are rigid. The graph D canonically identiﬁes to the global subcomposition of Ex(D. Lα. No ∼ two of these interpretable compositions are isomorphic and we have Aut(Ei ) → Z. Identify these elements iﬀ their images under this map coincide.β ) (16.e. Exercise 37 Give a proof of this lemma. QED. So the automorphism must also conserve heads and tails of arrows.1. . No two extensions are isomorphic if their cells aren’t.β to the graph vertexes corresponding to the arrow tails.. . which has the following structure: Its charts are all isomorphic to Lα. i = 1. .β ) is isomorphic to the automorphism group of the directed graph D. Suppose that the graph D has no multiple arrows (independent of their direction). It is now immediate that the disjoint union GI = i=1. with the 0 on the tail and the 1 on the head. whereas the elements 1 in the copies of Lα.

374 CHAPTER 16. Denote by Sk n(GI ) the set of singular k-simplexes of the nerve n(GI ). The vanishing of df means that we have f (i1 . It is well known that di+1 ·di = 0. and in congruence with the general cohomology theory of coeﬃcient systems (see appendix H. f (i. Take such a leaf. The second statement is obvious by induction on the cardinality of the nerve: There are leaves.. and hence the interpretability of the composition. we have the usual cohomology modules H k (GI ) = Z d /B d with Z d = Ker(dk ). –Σ– If GI is a global composition at address A.3. we have the module C k (GI ) = s∈Sk n(GI ) nΓ(s) (16. We derive a necessary condition for the vanishing of the ﬁrst cohomology and for the complex of the module complex of aﬃne functions to be ﬂasque. ik+1 ). For the vanishing of ﬁrst cohomology.. We then conclude the existence of a 0-chain g = (g(i))i∈I ∈ C 0 (GI ) with dg = f by the following lemma (QED). i2 ) + f (i0 . i2 ) − f (i0 .2 A Cohomological Criterion Summary. j))(i.1.3) of k-cochains of functions. i1 )|i0 ∩i1 ∩i2 = 0 for any singular simplex i0 . i1 . We have a linear diﬀerential coboundary map for each k: dk : C k (GI ) → C k+1 (GI ) where for any singular k + 1-simplex s = (i0 . and therefore. j). i1 . Proposition 19 Suppose that the A-addressed global composition GI has a nerve n(GI ) which is a ﬁnite acyclic graph.1. In particular. Proof.k+1 (16. Module complexes of deﬁne cochain complexes and their cohomology modules.. . i) = 0 for any singular 1-simplex (i. and nΓ(GI ) is ﬂasque. . . vertexes which are connected to (at most) one other vertex. We may suppose that there is a tree (a directed acyclic graph) T such that its undirected image |T | is isomorphic to n(GI ). i) = 0. .1). we set dk (f )(s) = j=0.5) where sj is the face of s after omitting vertex ij . For each k ≥ 0. and extend a function on the rest to the omitted leaf. i2 . i. omit it. everything over the commutative ring R.j)∈S1 n(GI ) in Z 1 . and therefore. B d = Im(dk−1 ) for positive k.4) (−1)j f (sj ) (16. which evidently identiﬁes to the module Γ(GI ) of global sections. CLASSIFYING INTERPRETATIONS 16. we may deﬁne the cochain complex C (GI ) as follows. we have f (i.. j) + f (j. Then H 1 (GI ) = 0.e. Denote by AT the arrow set and by VT the vertex set of T . and H 0 (GI ) = Ker(d0 ). we may suppose that the nerve is connected and take a cochain f = (f (i.

The following examples (which are also exercises) show that we are far from understanding the connection of cohomology of global compositions to the problem of interpretability. Consider the subgraph T of T obtained after removing i0 and A . n(G1) G1 G3 —2 G2 Figure 16. whence H 1 (GI ) → R3 . Since T is a tree. i). g(i) = 0.2: An interpretation with non-vanishing ﬁrst cohomology.e. j) ∈ AT . G3 as shown in ﬁgure 16.j)∈AT Γ(i ∩ j). We may add any global section d to the system (gi )i∈AZi without altering the diﬀerences which yield g (i) − g (j)|i∩j . in fact a forest with possibly several maximal subtrees. j) for all (i. ∼ ∼ ∼ ∼ check! . we have Γ(GI ) → Γ(Gi ) → Γ(Gi ∩ Gj ). It is a not necessarily connected graph. i0 ) ∈ A deﬁnes exactly one connected component Zi of T which contains i. take a i0 ∈ VT which is the head of a non-empty set A of arrows. CHARACTERIZATION OF INTERPRETABLE COMPOSITIONS Lemma 30 Let f ∈ (i. set gi = f (i0 . then there exists g ∈ g(j) − g(i)|i∩j = f (i.. 3 Please. therefore ∼ ∼ ∼ ∼ B 1 (GI ) → R6 . i∈VT 375 Γ(i) (= C 0 (GI )) such that Proof. QED. connected components. We have C 0 (GI ) → (R2 @R)3 . we have the global functions ∼ ∼ Γ(GI ) → H 0 (GI ) → R3 . and where no arrow tails exist.1. and choose g(j) = h. i. Induction on card(AT ) ≥ 1. and g(i0 ) = 0. Extend f (i. j. however. j) ∈ AT . Example 32 Consider the interpretation GI of the zero-addressed local composition G ⊂ R2 by three charts G1 . Since nΓ(GI ) is ﬂasque. suppose that f resides on i. H 0 (GI ) → R3 . o It is not interpretable since the functions are not ﬂasque. and3 Z 1 (GI ) → (R2 @R)3 → R9 . This does the job. j) for all (i. The ﬁnal deﬁnition of g runs as follows: Take g(i)) = g (i) for all vertexes except i0 . and H 1 (GI ) = 0. there is a cochain g ∈ i∈VB Γ(i) such that g (j) − g (i)|i∩j = f (i. j) to h on i. we may therefore choose g (i) freely. By induction. For card(AT ) = 1. every arrow (i. For the general case. Example 33 This example is the “M¨bius” strip composition introduced in example 31 above.2 Since every intersection Gi ∩Gj is generating.16. G2 .

Edwin Hewitt. Recall ∼ from sorite 6 that any isomorphism f : U → V among two such charts may be extended to an aﬃne automorphism of M .2. Fripertinger’s recent classiﬁcation of canons are an exception. Dan Tudor Vuza. So this is an interpretation by mutually isomorphic charts which can even be transformed into each other by a transformation in G. David Lewin.3: A non-interpretable composition with acyclic nerve and vanishing ﬁrst cohomology. Contributions by George Halsey. with discrete nerves.3 The G1 3 2 Ÿ3 1 4 n(G1) Ÿ3 G2 Figure 16. and Harald Fripertinger are discussed.1 Tesselation Summary. –Σ– Enumeration theory deals with the calculation of isomorphism classes of interpretations of local compositions in ﬁnite abelian groups.2 Global Enumeration Theory Summary. If these charts (U.5) is discrete and has constant . see ﬁgure 16.. has a strong root in the American tradition. the techniques of global classiﬁcation presented in chapter 15 do not apply here. in other words. deﬁned as a gluing of two copies of Z3 = 0@Z3 Z3 along two common points. CLASSIFYING INTERPRETATIONS Example 34 Take the zero-addressed global composition over the ﬁnite ﬁeld R = Z3 . such as tesselations.e.M) are all isomorphic under G. Since these are never projective. only very special non-trivial coverings have been classiﬁed: Global enumeration theory is mostly restricted to zero-dimensional nerves. i. Enumeration of interpretations in special cases. composition cannot be interpreted since the three points in each card are colinear. We review the work by George Halsey and Edwin Hewitt [204] on the tesselation of ﬁnite Z-modules (abelian groups). we call M I a G-isotypic tesselation of M .. 16. or canons. and we may consider its interpretations M I with mutually disjoint (non-empty) charts. Fix a subgroup G ⊂ Aut(M ).376 CHAPTER 16. but it is not possible to maintain this property in a common ambient Z3 -vector space while distinguishing the two non-common points of the charts! 16.e. i. –Σ– A ﬁnite Z-module M is a zero-addressed local composition. to this date. Further. the class nerve CN (M I ) (see section 14. in ﬁnite Z-modules.

e. We have the canonical bijective tesselation morphism M I → M of zeroaddressed compositions. S3 → Z3λ . we denote by Γt (M.4. n = pqrs.3. For any natural number j. In [204] a tesselating chord is deﬁned as an isomorphism class of tesselation morphisms on M under the group G = eM of translations on M . n = pα q for positive natural α. They are the groups of order o n = 1. §12]).. Theorem 20 [204. j) the number of tesselating chords in M which represent tesselations by charts of cardinality j.10] Let m(2) be the number of direct cyclic summands in the 2Sylow group of M . Then Deﬁnition 60 A ﬁnite abelian group M is called a Haj´s group if either M is trivial. o Based on an algorithm similar to Sands’ algorithm. and if card(M/S3 = d). q. We have two formulas for special cases: one for j = 2.15] If the 3-Sylow S3 group of M is cyclic. i. . o o see [204.13) in [204]. GLOBAL ENUMERATION THEORY 377 weights under G. and n = p2 qr. Their list is given in table (12. All ﬁnite cyclic Haj´s groups have been classiﬁed ([204. and let c be the number of elements of odd order in M . 3 8 In principle. Satz 11. 5 See appendix C. n = p2 q 2 . n = pqr. we have 1 3 Γt (M. Then we have Γt (M. §12] for all cyclic groups of order ≤ 24 (they are automatically Haj´s o by the above). n = pα . 4 Instead. 2) = 1 (card(M ) + 2m(2) − c − 1). j) can be calculated by Sands’ algorithm for all ﬁnite Z-module M which are Haj´s groups (this property is related to Haj´s’ solution of Minkowski’s problem. The classiﬁcations in [204] deals with these tesselation morphisms: A G-isomorphism between such tesselation morphisms M I → M . r. In particular.16. Z72 is the smallest cyclic non-Haj´s group. for any distinct primes p. we could equivalently consider the interpretation which adds the chart M to the atlas I and then just look for G-isomorphisms of such interpretations of M . the number Γt (M. denote T rans(W ) for the translation symmetries of W . and M J → M is an element g ∈ G which deﬁnes this commutative diagram MI − − → M −− g g MJ − − → M −− of global compositions4 .2.6) Theorem 21 [204. 2 ∼ (16.2. §12] for further references). either T rans(U ) or T rans(V ) is not trivial. 3) = 1 + (d2 − 1) + d2 (32(λ−1) − 1). the numbers of tesselation chords have been calculated in [204. s. and one for j = 3 if the 3-Sylow group5 is cyclic. M ). Satz 11. or for o every tesselation M = v∈V ev (U ) by translates of a subset U . For a local composition (W.

42). X[ n ] ).k (16. in this case the calculation of isomorphism classes ∼ is not restricted to translations. the set M osG of G-mosaics in Zn n is partitioned into the subsets M osG of k-element partition mosaics for k = 1. With this. k]|Xi = esi )|Xi = 0 ∂Xi (16. with n = card(P ). n]Zn . i. Theorem 1] With this notation.1. Then we have the following enumeration theorem. According to the general combinatorial P´lya methodology applied by Fripertinger. and we may identify the mosaics with the orbits of the action of the direct product S[1. . [1. P ) is replaced by a value ξi . if each variable Xi in Z(G. and that we are only interested in the ﬁbers. . The notation is related to the second variable P = Zn . Isomorphisms among partitions P art(I).2. . i. formula (11. n.9) . n] under the left action of the symmetric group Sn . Also.k subsets correspond to the function sets of functions p : Zn → [1. and i si = i(X1 + . zero-addressed discrete interpretations of ﬁnite cyclic groups.8) (16. These n. As usual in P´lya theory.378 CHAPTER 16. . we put Mk = Z(G. . Again. and G ⊂ SP which we omitted in the previous notation of Z since P was ﬁxed.. Zn |Xi = ∂ )Z(Sk . CLASSIFYING INTERPRETATIONS 16. We review the work by Harald Fripertinger [172] on the enumeration of mosaics. Xn . .n] × Gopp on the function set [1. whence the action of the symmetric group on the value domain.e. Put M0 = 0.e. Of course. and then being evaluated at zero). the isomorphism g ∈ G acts from the right on p. P art(J) are n deﬁned by commutative diagrams as for tesselations.4. but the partition is not necessarily isotypical. So the ambient module is more special than in section 16. We ask for g ∈ G that the diagram ZI − − → Zn n − − g g ZJ − − → Zn n − − commutes.2 Mosaics Summary. On the other hand. The reason is that the ﬁbers of the function p deﬁne a partition of Zn . P corresponds to an orbit Sn · p. . Write [ n ] for the greatest integer ≤ n/i.7) for 1 ≤ k ≤ n (diﬀerential operators acting on the second factor. a partition P is identiﬁed to an orbit of a function p : Zn → [1.. 2. as in that context.6. –Σ– This classiﬁcation deals with discrete interpretations ZI of Zn by card(I) = k non-empty n charts of arbitrary size. Then we have card(M osG ) = Mk − Mk−1 n. . we need to calculate the cycle index Z(G) = Z(G. a rational polynomial in the indeterminates X1 . P ) as deﬁned o in section 11. the o calculation of mosaic numbers requires the action of a ﬁnite group on a set of objects which represent the partitions. P |Xi = ξi ). we write Z(G. we are selecting a subgroup G ⊂ Aut(Zn ) → eZn · Z× . n I and we say that a k-partition is the canonical morphism P art(I) = Zn → Zn of a discrete interpretation ZI with card(I) = k. n] with card(Im(p)) = k. With this index. To this end. i Theorem 22 [172. an orbit of partitions is called a G-mosaic. no matter which values they stem from.

. [1. n]) n. and ﬁx once and for all a partition Λ of [1.λi ] . n]|Xi = iXi )|Xi = 0. So we clearly have two linear conditions i λi = k.e. i iλi = n for λ. λ2 . n]. but also the block type. n]) of bijections f : Zn → [1.3. and we obtain the canonical identiﬁcation M osG → HΛ × Gopp \ Bi(Zn . If we are given a partition P art(I) of Zn . GLOBAL ENUMERATION THEORY and therefore card(M osG ) = Mn . ∂Xi (16. f ) → g · f identiﬁes to the set of partitions of type λ.11) ∼ Since the cycle index of a wreath product6 of groups can be deduced from the indexes of its factors. we add the well-known right action of G. n]) → Bi(Zn . gΛ deﬁne the same partition iﬀ g −1 · f stabilizes Λ. Let HΛ ⊂ S[1.λ of the set M osG of G-mosaics of type λ with the orbits of the two-sided action of G and HΛ . and two such inverse images fΛ .2. . [1.10) A calculation by use of the computer program SYMMETRICA in [172] yields this table of mosaics for three important groups: k = T12 T I12 − → GL(Z12 ) 1 1 1 1 2 179 121 87 3 7254 3838 2155 4 51075 26148 13730 5 115100 58400 30121 6 110462 56079 28867 7 52376 26696 13835 8 13299 6907 3667 9 1873 1014 571 10 147 96 63 11 6 6 5 12 1 1 1 − → This yields a total of 351773 T12 -mosaics. The numbers of T12 -mosaics of all types have been calculated by the SYMMETRICA program and yield this table: 6 See appendix C.λ Using this identiﬁcation. n] which is of type λ. n]) : (g. Zn |Xi = n. .16.2.λ ∂ )Z(HΛ . not only the cardinality card(I) = k is invariant under isomorphism of the group G. take the set Bi(Zn . We again replace sets by functions. Theorem 2] The number of G-mosaics of type λ is given by the formula (due to de Bruijn) card(M osG ) = Z(G. n] of left action HΛ × Bi(Zn . i.. 179307 T I12 -mosaics. To get the mosaics of this type. [1. Given a type λ. λn ) where λi is the number of charts in I with cardinality i. This method can be reﬁned to yield ﬁner classiﬁcations of mosaics. we have Theorem 23 [172. n 379 (16. Then the inverse image fΛ = f −1 Λ of Λ under a bijection f deﬁnes a partition of Zn . We summarize one such reﬁnement. [1.n] be the stabilizer of Λ. Then the orbit set HΛ \ Bi(Zn . the sequence λ(P art(I)) = λ = (λ1 . and since the stabilizer is isomorphic to S[1. i the formula is controllable. [1. . n.i] S[1. and 93103 GL(Z12 )-mosaics. [1.

–Σ– Following Vuza’s context [552]. 2) 2 6 29 29 340 35 424 1820 386 2330 3500 297 792 11580 4463 7740 5890 3510 2792 6 16. 4. 5) (22 . 3. 7) (13 . except the onsets which we also restrict to the rationals. 3. 25 ) (110 . 24 . 3) (19 .3 Classifying Rational Rhythms and Canons Summary. 11) (1. 22 . 7) (1. we consider rhythms without any further parameters. 2. 52 ) (13 . 2. 6) (23 . 4) (34 ) 2 2 (1 . we work with zero-addressed objective local and global compositions on the space Onset|Q of rational onsets. 4) (23 .2. 5) (17 . 6) (16 . 5) (1. 2 . 5. 3) (12 . 4) (24 . 2. In Vuza’s theory. CLASSIFYING INTERPRETATIONS λ (12) (3. 2. 3. 9) (1. We give a short comment on the classiﬁcation of rhythms and canons on rational onsets. 5) (43 ) 4 (1 . 22 . 4. 2. 4) (14 . 2. 7) 2 (1 . Therefore. 8) (12 . 6) (1. 2. 10) (4. 5) 2 (1 . i. 3. 32 ) (17 . 5) (22 . 3) (16 . and positive period p. 24 ) (112 ) 1 12 85 38 510 236 2320 610 1170 3510 3510 2915 5890 1170 610 2610 424 340 2325 1 λ (1. 3. 6) (2. 4) (1. 4. 6) (13 . 4) (13 . 42 ) (12 . 23 . 52 ) (1. Formally.. the translation automorphisms of R form the group epZ . 3. 2. 4. 42 ) (1. 32 . 42 ) (13 . 5) (15 . 2. 6) (3. 8) (14 . 5) (1. 7) (62 ) 2 (3 . 22 .e. 32 . 10) (13 . 6) (12 . 2.∞]p G over a ﬁnite germ G ⊂ 0@Onset|Q . 32 ) (15 . 4) (16 . 33 ) (14 . 7) (1. 2. 3. R is a zero-addressed objective local composition in Onset|Q which has a non-zero translation in its automorphism. 3. 2. 0 < p being called the period P er(R) of R. 33 ) (16 . 5) (12 . 4) (15 . 2. 22 . 9) 2 (1 . 3. 23 ) 1 30 84 170 340 610 781 50 3480 6960 708 2347 5890 2325 29 6005 8725 12 645 λ (2. a (periodic) rhythm is a zero-addressed local rhythm R = e[−∞. 4) (18 . The group T rans of rational translations acts on the set RV uza of Vuza rhythms. Call Vuza rhythm such a local rhythm. 22 . 32 . 3. 3. 3. 5) (13 . 5) (14 .380 CHAPTER 16. we are working on the additive group of rationals. 23 . 9) (22 . 6) (12 . 8) (5. The determination of Vuza classes runs as follows. In other words. 32 ) (13 . and deﬁnes the Vuza classes as being the translation isomorphism classes or orbits in T rans \ RV uza of Vuza rhythms. on the Z-module QZ . Given a Vuza rhythm R. 6) (14 . 2. 3) (14 . 4) (12 . 4. 8) (1. we write 1 Ru = P er(R) e−r R the uniﬁed rhythm with period one which is obtained by contraction after a . 6) (1. 32 ) (1. 22 ) 6 12 140 340 38 610 645 386 1170 1170 38 4470 17370 8860 713 17630 11623 554 84 λ (1 . 4. 8) (2. 3) (26 ) 8 (1 . 22 . 7) (15 . and we suppose that it is non-empty since otherwise everything is known. 42 ) (2. and which is locally ﬁnite. 7) (2. Equivalently. 23 .

a non-empty ﬁnite set of Vuza rhythms of the same translation class. p DJ I J are equivalent canons of period one. Vuza [555] calls the common chart class the ground class groundclass(C I ) of the canon C I . §6]. Vuza [555] considers “unending rhythmic canons” which are deﬁned as ﬁnite.3.R ⊂ QZ is ﬁnitely generated. and Ru is embedded. . After a R dilatation by the invariant nR .3. GLOBAL ENUMERATION THEORY 381 shift by the minimal non-negative element r of R.2. we have a recursive procedure to enumerate the classes in question. Hewitt. We leave it as an exercise to go through that procedure with the translation group instead of the full automorphism group.Zn . we may also suppose that the resulting classes are embedded. Clearly. Since the Z-module Z.16. have period nR under the group of integer translations.12) DJ − − → D −− as above for tesselations.3. we have a commutative diagram CI − − → C −− t e et (16.. In [204. this has been explicated.. DJ are equivalent iﬀ they have the same period p and the contracted canons p C I . and not necessarily covering the entire ambient module. but this module is QZ instead of a ﬁnite abelian group.e. 0 minus those which have non-trivial translation groups H ⊂ Tn . as discussed by Halsey. it must be of shape n1 Z. it is monogeneous. two Vuza rhythms of ﬁxed period p have the same class iﬀ their uniﬁed rhythms do so. with uniquely determined positive integer nR . So this situation is a generalization to non-discrete isotypic coverings. D having this special period. whereas the Vuza class of the union C is called the resultant class resclass(C I ) of the canon. non-empty subsets of a Vuza class. But this classiﬁcation is clearly equivalent to that of the objects in ObLocgen. i. although no further parameters are considered. Hence we may concentrate on classifying embedded rhythms with period one modulo translations. and as it contains Z. Z) which 1.. Based on this local situation. we are left with the classiﬁcation of zero-addressed objective local compositions (R. the period of a Vuza canon is a welldeﬁned positive rational number given by P er(C I ) = P er(groundclass(C I )). 2. and Fripertinger. a Vuza canon is the interpretation C I of a zero-addressed objective local composition C ⊂ 0@Onset|Q by a T rans-isotypic covering set I consisting of Vuza rhythms. i. The classiﬁcation itself can be performed along the lines of our exposition for the ﬁnite case in section 11. As with the local situation of one Vuza rhythm. and the latter are recursively gen. Since these classes are the classes of ObLocgen. Moreover. DJ are called equivalent iﬀ the resulting rhythms are translation equivalent under a translation et which is compatible with the respective atlases. also called the voices of the canon. Two Vuza canons C I .4.Zn which have trivial translation automorphisms (they are “maximal” in 0 the terminology of Halsey and Hewitt [204]). We have P er(Ru ) = 1. i.R = Z. so we may restrict our discussion to C .e. we subtend that the Vuza context must be interpreted as a projection of a Para-rhythm as discussed in section 13. are generating: Z.e. Two Vuza canons 1 1 C I .Zn /H determined as the classes stemming from ObLoc0 . Equivalently.

but see [173]. we have to look at interpretations cι . D ⊂ Z. Lewin. By deﬁnition. The algorithm has been implemented in o OpenMusic by Carlos Agon and Moreno Andreatta. i.D. in which the second part is concerned with the so-called Pitch-Class Set Complexes.C ⊂ Z. We give a short account to this sector. For a detailed discussion on Vuza’s algorithm in the perspective of the theory of nonHaj´s groups and the Minkowski conjecture. In a recent work. Since these classes are translation-equivalent. We do not know how far these invariants are away from being classifying. with the invariant resulting divisor nC = nD = n as already discussed above in the local case. So. we have the generating local compositions n. such as the (translation) class nerve and the cohomology groups. given a pc set class X. So modulo this period. A special family of Vuza canons are the regular complementary canons of maximal category (for short: RCMC-canons). see [17] for a complete list of solutions. we obtain the same module 1 Z.. . are “translation rigid”. zero-addressed objective compositions by translation-isotypic translation rigid charts. So our classiﬁcation problem reduces to the classiﬁcation of “canons” in the cyclic residue groups Zn . Inclusion relations are basically of two type: the K and the Kh relations. CLASSIFYING INTERPRETATIONS C ⊂ Z. The local/global dichotomy is already present in Allen Forte’s book [159]. i. where both T rans((U ) and T rans(V ) are trivial (a non-Haj´s group.2. a pc set class Y is a member of the set complex about X iﬀ Y can contain X or can be contained in X (or the corresponding for the complement of X). but the charts are not necessarily generating. American Set Theory has developed a number of “combinatorial” structures which relate to interpretations of pitch class sets. see [16]. sets of sets associated by virtue of the inclusion relation. called the nexus.C. Fripertinger has also classiﬁed such canons.D ⊂ Z with a period which divides n.D = n Z. as it was developed by Forte. see deﬁnition 60). –Σ– As mentioned in section 11. with some preliminary conditions on the cardinality of X and Y which are: 1.3 Global American Set Theory Summary. The classiﬁcation of such interpretations of local compositions in Zn cannot be settled by the known resolution theorems from chapter 15 since the involved function modules are not projective.5. and Rahn. 16. However. Morris. In [557]. By deﬁnition. if we dilatate this situation by the resulting divisor.. Vuza presents an algorithm which enables the calculation of any tesselation M = v∈V ev (U ) of a group M by translates of a subset U . the inequalities 2 < card(X) < 10 and 2 < card(Y ) < 10. d ⊂ Zn by equipollent translationisotypic atlases ι. he o proves that six is the minimal number of voices of an RCMC-canon and nR = 72 is the shortest period. we have a number of numerical invariants. κ which consist of card(ι) = card(I) (!) not necessarily generating local chart compositions which do not have non-trivial translations in Zn . n.2. And the classiﬁcation goes by translations on the ambient space Zn which carry over to the interpretations. and which are covered by local chart compositions which are all (strictly) n-periodic. In particular. dκ of local compositions c.e. part of what we called the American tradition focused on global instead of local musical properties. these are interpretations of generating.e.382 CHAPTER 16.C = Z.

The main diﬀerence compared to Forte’s relations consists of the fact that K relates two pairs of set classes whereas KI simply relates two set classes. B are said to be in the KI relation iﬀ A ⊆ B. The following example in ﬁgure 16. A couple of (abstract) complementary set classes A/ − A is said to be the Kh nexus of the SC-comp list X if for all couples of complementary set classes Y / − Y of X the couple A/ − A is in the Kh relation with Y / − Y . As pointed out by Forte. 3. 2. A stronger condition. whereas abstract relations are among collections of pc sets related by some equivalence relation (usually transposition and/or inversion). the set class 6-5 corresponding to the pc set {0.g. In this representation. called the Kh relation. In the case of a self-complementary hexachord. Literal relations are among unordered collections of pitch classes. “the rule of set-complex membership yields aggregates of considerable size” [159. Note that this new deﬁnition drops Forte’s original condition by assuming that a couple Y / − Y may be equal to the Kh nexus A/ − A. 6. The new family of sets is called the subcomplex of a given pc set class X. Notice that the abstract inclusion relation has been independently theorized by Rumanian composer Anatol Vieru by means of the concept of “modal structure”. In the case of the abstract inclusion.5. By deﬁnition two set classes A. 3. As mentioned in section 11. For a discussion of some diﬃculties arising in the analytical application of Morris K and Kh relations as well as of a possible generalization of the SC-list concept in relation to Forte’s concepts. 4. In [545. p. this hexachord is listed alone.3. p. the inequalities card(Y ) = card(X) and card(Y ) = card(−X).285-288 and pp.2. Two sets that belong to the set complex about a given pc set class X are said to be in the K relation. Forte’s Kh relation may be reformulated now in terms of SC-comp lists. Forte’s K and Kh relations are examples of abstract relations. a modal structure is an equivalence class of a pc set under simple transposition and without taking inversion into account. The following ﬁgure 16. Morris suggests to represent K and Kh relations as lists of set classes displayed in complementary pairs which are called