INFORMATION TO USERS

This manuscript has been reproduced from the microfilmmaster. UMI films die text directfy from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from aiy Q/pe of conqjuter printer.

The qnality of this reproduction is dependent upon the quali^ of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and iirproper aligument can adversefy affect reproduction.

In the unlikely event that the author did not send ^UMT complete mamiscript and there are missing pages, these will be noted. Also, if unauthorized coj^right material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., m^ps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscrit have been reproduced xerographically in this copy. Higher qualiQr 6" x 9" black and white photogr^hic prints are available for aiy photographs or illustrations appearing in this copy for an additional charge. Contact UMI direct^ to order.

UMI A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor. Ml 48106-134S USA 313/761-4700 800/521-0600

The Clarinet Reed: an Introduction to its biology, chemistry, and physics

Document

Presented in Partial Fulfilment of the Requirements for the Degree

Doctor of Musical Arts in the School of Music

The Ohio State University

By

Donald Jay Casadonte, B. A., M. A.

The Ohio State University

1995

Document Committee: proved by

James Pyne

Christopher Weait

Patrick Gallagher Advisor^ Department,Ç Music DHI Number: 9612138

Copyright 1995 by Casadonte, Donald Jay All rights reserved.

UMI Microform 9612138 Copyright 1996, by DHI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 Copyright by

Donald Jay Casadonte

1995 To My Parents and the Blessed Virgin Mary, with the hope that the effort of this work has honored them Acknowledgements

In completing such a complicated Interdisciplinary work It Is almost

Impossible to mention all of people to whom the author Is beholden. In the matter of the biology of the reed, the author thanks Dr. Frederick Sack, Clare

Lawton, Dr. Marilyn Veselac and Dr. Lisa Boucher for their discussions about

Arundo donax, John Mitchell for electron microscopy, Ann Osterfeld and Cathy

Wolken for optical densitometry plots, Richard Pearson for donating a rhizome.

Dr. Richard Swenson, Jose Diaz and LLoyd Lemmermann (?) for photographing the gross anatomy of the . Dr. Oliver Tuovlnen, Christine

Vanecko, Dr. Karl NIklas, Steve Lawton, and Matt MIslcka who gave valuable advise or provided equipment. Many other students and faculty of the OSD

Department of Biology conversed wXh the author or helped run tests. Their names are Important and deserve a more complete listing.

In regards to testing the chemistry, physics and material properties of the reed, the author thanks, from OSD: Dr. Patrick Gallagher, Ruth Anderson,

Gordon Renkes, Dr. Billy Culbertson, Dr. Alka Thakur, Paul Herman, Sandy

Jones and Dr. Jerry BIgham, as well as Dr. Ivan Goldfarb of Wright State

University, Dr. Helntz Melneker of the University of Akron, Dr. Andrew Summers of Miami University of Ohio, J. W. Lemmons of J. W. Lemmons and Associates,

Wayne T. Scheer. In addition, a dentlst/obolst’s at the University of Washington

(Dr. Robert ?) for suggested to the author the salic acid deposition from saliva.

Ill Many former students of Dr. Arthur Benade helped the author in acoustical matters: Dr. Douglas Keefe of the University of Washington Department of

Systematic Musicology, Dr. Walter Worman, Dr. Steven Thompson, Dr. Peter

Hoekje.

Thanks go to Tom Merrick, Martin (Marty) Haas, the site assistants at the

OSU computer labs, and Dr. Gary Kinzel for computer help. Dr. Michael

Trudeau and Dr. Joel Weaver of OSU for videotaping the reed vibrating in the mouth. Dr. Arthur Leissa and Dr. Anderson for discussing plate theory and critiquing results. Dr. Brian Harper for reading chapter five, Been-Der Yang and

Mohammed (last name?) for measuring the shape of reeds, Calvin Trefz in the physics department for providing physical testing equipment, and Dr. James

Grotberg, Dr. Guezennec, and Dr. Thorwald Herbert (each suggested the vortex shedding mechanism of reed excitation) for discussing fluid mechanics .

In regards to the musical aspects of the reed, thanks go to my past clarinet teachers, Charles Baker, and professor emeritis. Dr. Donald McGinnis, the document and general examination committees: Dr. Patrick Gallagher,

Christopher Weait, Craig Kirchhoff and Richard Blatti, Dr. Burdette Green for reading and correcting chapters one and two, Robert Sorton and James Hill for advise on reeds, members of ICS and IDRS, Ray Spielman, various posters on the internet for comments, and Dr. Kim Ellis for support. Special thanks go to the student reed rater: Dr. Bruce Curlette, Dr. Daniel Lochrie, and Dr .Melinie

Griffin, Amy Lammendola, Cathy Gardiner, Cathy Hope, Sarah Young, Jennifer

(Sutton) McDonald, Gail Letho, Debra Andrus, Wendy (Welsh) Harvey, Lora

Sabo, Seth Semmons, Beth Stimpert, Carmen Strine, as well as other former clarinetists of the OSU concert band.

iv Thanks goes to the OSU Graduate School (Sandy Walden and Bobbi Davls-

Jones), Elaine Moran of the ASA, and journalists Karen Schmidt, Bret Adkins, and Robert Cooke.

For emotional support, thanks go to Susan Sund, my brothers and sisters in the Third Order of Discaiced Carmelites, who have taught me more about life than I can ever hope to repay, and my brothers and sisters in the St. Micheal

Prayer group for support and prayers. There is not room enough to say all I would wish to my parents, Dominick and Billie Ann Casadonte, and brothers,

Thomas, Michael, and Dominick, for their care over the years. They know what difficulties the author has faced, and the finishing of this document is a testimonial to them.

This document would never have been finished without the extraordinary help of the author’s advisor, James Pyne. He originally invited the author to participate in his idea of forming a clarinet research group six years ago, and has, over the years, been instrumental in obtaining funding and instrumentation. He has coached the author through the rigors of a performance degree program, and taken care of ail of the administrative details which were necessary to obtain the degree. He has discussed many of his own acoustical research projects with the author, and has gone out of his way to be there whenever the author needed his help.

Finally, the author must thank Divine Providence for providing him with so many moments of grace over the last fifteen years while attending to musical matters at The Ohio State University. If this document gives witness to anything, it gives witness to His power to act through weakness, and to bring all things to completeness. Vita

June 19,1955 ...... Born - Cleveland, Ohio

1977 ...... B. A., Case Western Reserve University, Cleveland, Ohio

1982...... M. A., Cleveland State University, Cleveland, Ohio

1979-Present ...... Music Student, The Ohio State University.

Publications

1. "A Center-symmetric Potential Function for use in Liquid-State Modeling," senior research project," Case Western Reserve University, 1977.

2. “An Introduction to the Biology, Chemistry, and Physics of Woodwind Reed Material,” Journal of the Acoustical Society of America, vol. 90, no. 4, part October, 1991 (Abstract).

3. Cane That’s Musically Able (Interview) Science News, December, 14, 1991.

4. Reed Research Interview, Science News Update National Public Radio January, 1992.

5. “The Geometry of Melodic Structure: is there a strange attractor at the heart of Bach’s, Invention no. 4 in d minor?". Journal of the Acoustical Society of America, (abstract).

vi 6. The Hypergraphics of Musical Sounds, MA Thesis, Cleveland State University, 1982.

7. "The Hypergraphics of Musical Sounds", cited in the book. Results of the Chroma Foundation Search for New Music Notation, 1985.

8. “The Biology, Chemistry and Physics of A. donax," paper presented at the International Double Reed Society meeting, Towson State University, 1990.

9. “A tour through donax," (with James M. Pyne), presented at the International Clarinet Society meeting. University of Cincinnati, 1992.

10. "A Hypothetical Decompactiflcation Model for the Process of Humor," Paper presented to the Western Humor and Irony Membership Conference, Purdue University, 1988.

11. A Hypothetical Decompactiflcation Model for the Process of Humor," abstract in WHIM VIII, 1991, Purdue University Press.

12. "A Hypothetical Decompactiflcation Model for the Process of Humor," cited in the American Society of Aesthetics Newsletter, 1989.

13. "The Dynamics of Humor l-Decompactification Mathematics," and Humor in F: what's so funny about Mozart's Musical Joke?" Seventh International Conference on Humor, Hawahi, 1989.

14. "The Dynamics of Humor II- Chain Ordering and the Energy of Humor," Eighth International Conference on Humor, Sheffield, England, 1990.

15. "Transition Phenomena in Humor," Ninth International Conference on Humor, Ontario, Canada, 1991.

16. “Psychophysiological responses to Humor: a theoretical perspective," CORUUM/Tenth International Conference on Humor, Paris, France, 1992.

17. "Resolving the Mechanisms of Musical Humor: an analytical study of Mozart's Musical Joke," juried competition winning paper. Graduate Research Forum Competition, The Ohio State University, 1989.

18. An Introduction to the Biology, Chemistry, and Physics of ," juried competition winning paper. Graduate Research Forum vii Competition, The Ohio State University, 1991.

19. “The Geometry of Music: metric and phase space results," juried competition winning paper. Graduate Research Forum Competition, The Ohio State University, 1992.

20. DMA Solo Recital l-January, 26, 1990, The Ohio State University.

21. DMA Solo Recital ll-February, 1990, The Ohio State University.

22. “Who Killed Mrs. Myheart?" a musical mystery for Corps in Bb and Clarinet, © 1990, Donald Casadonte

Major Field: Music Minor Field: Chemistry

VIII Table of Contents

Dedication ...... ii

Acknowledgements ...... iii

Vita...... vi

List of Figures ...... xii

List of Tables ...... xii

Preface ; ...... 1

Chapter Page

I. Introduction ...... 5

Early Uses of Arundo donax in music ...... 5 Brief History of the Clarinet Reed ...... 9 Modern Manufacturing Techniques for Clarinet and Saxophone Reeds ...... 22 Reed Nomenclature ...... 33 Rationale of Current Study ...... 41

11. Arundo donax Anatomy ...... 50

The Root System ...... 51 The Stem System ...... 58 The Leaf System ...... 71 Ultrastructure ...... 79 Ground Tissue System ...... 95 Vascular Tissue System ...... 107 Cell Walls ...... 124

III. Clarinet Reed Chemistry ...... 132

The Chemistry of Arundo donax...... 132 Elemental Analysis ...... 132 ix Cellulose ...... 134 Degree of Crystallinity by Infrared Spectroscopy ...... 142 Extending the Infrared Testing Method ...... 153 Other Tests for Decrystallization ...... 155 Degree of Polymerization ...... 157 Hemicellulose ...... 159 The Role of Water and Hemicellulose ...... 163 Hemicellulose Degradation ...... 164 Infrared Identification of Hemicellulose ...... 174 Lignin Preliminaries ...... 175 Arundo donax Lignin ...... 180 Spectroscopic Methods of Lignin Determination ...... 181 Cell Wall Chemistry ...... 187 Chemistry and Material Properties ...... 189

IV. Clarinet reed Material Properties ...... 193 Introduction ...... 193 Descriptive Results ...... 196 Mass/Weight Properties ...... 208 Effects of Heat on the Clarinet Reed ...... 216 Effects of Light on the Clarinet Reed ...... 228 The Effect of Saliva on the Clarinet Reed ...... 229 Clarinet Reed Degradation ...... 232 Towards the Prevention of Reed Degradation ...... 239 Reed Hydroscopic Properties ...... 241 Changes in Physical Parameters with Hydration ...... 255 Mechano-sorptive Creep ...... 256 Damping ...... 256 Stress and Strain: Theoretical Considerations ...... 265 Stress and Strain: Empirical Tests and Results ...... 283 Biomorphological Variations ...... 286

V. Reed/Mouthpiece Dynamics ...... 298

Overview ...... 298 One-Dimensional Reed Models ...... 299 One-dimensional Unforced Models of the Reed ...... 302 One Dimensional Linear Models of the Forced Reed ...... 306 Two-Dimensional Reed Models ...... 318 Three-Dimensional Reed Mechanics ...... 325 X Reed/Mouthpiece Interactions: Diffuser Theory and the Origin of Reed Squeak ...... 336 Starling Resistor Model of Lip/Reed ...... 350 Reed Shape Effects ...... 359 Finite Element Simulation of the Clarinet Reed ...... 373

Appendicies A. Sample Reed Rating Instruction Set ...... 384 B. Reed Mode Shapes and Stresses ...... 392

Bibliography ...... 456

XI List of Figures

Figure Page

1.1. Early Reed flute and Pan Pipe...... 7

1.2. Early “Single Reed” and “Double Reed” Instruments...... 8

1.3. Historical Reed Development...... 10

1.4. Clarinet Reed Superellipse Paramenter (N) Variation...... 20

1.5. Effect of Changing N in the Superellipse Equation...... 21

1.6. Computer Superellipse of 1/2 Reed Tip (N = 1.786)...... 23

1.7. The Var Region of Southem France...... 24

1.8. Steps in Reed Manufacturing...... 25

1.9. A Biomorphological Miscellany ...... 29

1.10. Common Reed Nomenclature...... 34

1.11. The Reed Grid...... 35

1.12. The Wedge Profile...... 37

1.13. The Expanding Ridge (Crest) Profile...... 39

2.1. Arundo donax ...... 51

2.2. A Sample of Grasses at The Ohio State University...... 53

2.3. The Arunc/o donax Stem ...... 56

2.4. The Rhizome...... 60 xii 2.5. The Complete Arundo donaxCu\m ...... 62

2.6. Arundo c/onax Cross-section (Stele)...... 63

2.7. Arundo donax Intemodes and Nodes...... 64

2.8. Nodal Sites for Leaf Generation...... 65

2.9. The Processed Stem...... 68

2.10. The Three Tissue System...... 70

2.11. Arundo donax Leaf Pattern...... 72

2.12. Panicles...... 75

2.13. Panicle and Base Structure Showing Subdivisions of Aggregate....76

2.14. Spiklets...... 78

2.15 Spiklet, Glume, and Floret...... 79

2.16. Florets...... 80

2.17. Perianth...... 81

2.18. Three-Component Structure of Arundo donax Stem Material...... 83

2.19. Epidermis Schematic...... 85

2.20. Electron Dispersive X-ray Spectrum of Arundo donax Stem Epidermis...... 86

2.21. Electron Dispersive X-ray Spectrum of Arundo donax Stem Inner Tissue...... 87

2.22. A Tannin (Catechin) Structure...... 88

2.23. Subepidermal Sclerenchyma Tissue...... 90

2.24. Veselac’s Two Locations for Fiber Bands...... 93

2.25. Fibers and Sclereids...... 94

2.26. Bundle Sheath...... 96 xlii 2.27. Parenchyma Tissue...... 98

2.28. Parenchma I opology ...... 102

2.29. Parenchyma “Sheet”...... 104

2.30. Parenchyma Unit Crystal Approximation ...... 105

2.31. Vascular Bundle...... 108

2.32. Various Steles...... 110

2.33 An Explanation of the Vascular Bundle Power Law...... 111

2.34. Xylem Cells...... 113

2.35. Phloem Cells...... 115

T» 2.36. An Explanation of the Collateral Arrangement of Cells in the Vascular Bundles of Arundo donax ...... 117

2.37. Xylem Fiber...... 119

2.38. Companion Cells...... 120

2.39. Pits...... 122

2.40. Helical Thickening...... 123

2.41. An Explanation of Turgor Pressure...... 125

2.42. The Cell Wall...... 126

2.43. The Fibril Concept...... 128

3.1. 3-d-glucopyranose...... 135

3.2. Cellobiose...... 135

3.3. 1—>4 3-d-glucopyranose Bridge Structure...... 136

3.4. Cellulose Polymer Structure...... 137

3.5. Cellulose Crystal Structure...... 138

xiv 3.6. X-ray Detraction Spectrum of Clarinet Reed...... 141

3.7. Superimposed 020 Peaks for Spent and Pristine Reeds...... 146

3.8. Clarinet Reed infrared Spectrum...... 149

3.9. %C from Infrared Data for Selected Reed...... 152

3.10. Pristine and Spent Infrard Spectra...... 154

3.11. Differential Scanning Calorimetry Plot of Pristine and Spent Reeds...... 156

3.12. Arundo donax Hemicellulose...... 161

3.13. TGA Spectrum of Two Pristine Clarinet Reeds...... 166

3.14. TGA Spectrum of Two Spent Clarinet Reeds...... 170

3.15. Comparison of Differential Plots of Pristine and Spent TGA Spectra...... 172

3.16. Lignin Preliminaries...... 176

3.17. Infrared Spectrum of Arundo donax Stem...... 183

3.18. Infrared Peak Assignments for Arundo donax Spectrum...... 185

4.1. Overall Ratings for {R}...... 197

4.2. Descriptive Statistics for Reed Quality Rating...... 198

4.3. Timbre Ratings for {R}...... 200

4.4. Strength Ratings for {R}...... 201

4.5. Noise Ratings for {R}...... 202

4.6. Stability Ratings for {R}...... 203

4.7. Projection Strength Ratings for {R}...... 204

4.8. Summary Statistics for {R} ...... 205

4.9. Pearson Correlation Matrix for Performance Descriptors...... 205 XV 4.10. Mass Measurement Summary...... 211

4.11. Mass Statistics...... 210

4.12. Mass and Quality Descriptor Correlation Matrix...... 213

4.13. Contour Density Map of Oboe and Clarinet Reed...... 215

4.14. Optical Density Map of a Clarinet Reed Tip...... 217

4.15. Cross-sectional Optical Density...... 218

4.16. Stem Orientation...... 222

4.17. Thermal Expansion Curves for Arundo donax...... 223

4.18. Coefficient of Thermal Expansion for Arundo donax ...... 221

4.19. Shape of S. Epidermitis and its Migration in Reed Pit Cells...... 234

4.20. Bacterial Deposition on inner Xylem Wall...... 235

4.21. Bacterial Biomass...... 236

4.22. Bacterial Sheet (Heinrich’s “Fungus”)...... 237

4.23. Clarinet Reed Water Capacity...... 243

4.24. Water Adsorption on Cell Walls...... 245

4.25. Sorptive Hysteresis...... 247

4.26. Longitudinal and Transverse Reed Warp...... 249

4.27. Reed Warp Measurments, Pristine/Played...... :...... 250

4.28. % Warp for Pristine and Played Reeds...... 251

4.29. Water Transport Mechanism in Clarinet Reeds...... 253

4.30. Creep...... 257

4.31. Arundo donax Decay Curve (blow-up)...... 259

4.32. Arundo donax Decay Curve, Complete...... 261 xvi 4.33. Comparison of Wood and Arundo donax Decay Curves...... 264

4.34. Mass-Spring System...... 266

4.35. Cellular Spring Network...... 268

4.36. Young’s Modulus...... 270

4.37. Orientation of E and G Moduli...... 272

4.38. Two Diffemt Uses for Mohr’s Circle...... 275

4.39. Poisson’s Ratio...... 277

4.40. Viscoelastic Circuit Models...... 280

4.41. Typical DMA Plot ...... 285

4.42. Vascular Bundle Number in a Sample of Reeds...... 288

4.43. Vascular Bundle Statistics...... 289

4.44. Vascular Bundle/Quality Correlation Matrix...... 289

4.45. Different Types of Fiber Arrangements in Clarinet Reeds...... 291

5.1. A Comparison of Measured and Computed Reed Shapes...... 333

5.2. The Clarinet Moutnpiece and Reed...... 337

5.3. Flapping Flutter and Reed Squeak...... 345

5.4. The Starling Resistor Model...... 352

5.5. Reed Structural Factor: Heel Thickness...... 363

5.6. Reed Structural Factor: Uncompressed Shoulder Height...... 364

5.7. Reed Structural Factor: Compressed Shoulder Height...... 365

5.8. Reed Structural Factor: Tip Thickness...... 366

5.9. Reed Structural Factor: Thickness...... 367

5.10. Reed Structural Factor: 8mm, Left Thickness...... 368 xvii 5.11. Reed Structural Factor: 8mm, Right Thickess...... 369

5.12. Reed Structural Factor: Natural Frequency...... 370

5.13. Correlation Matrix for Structural Measures...... 371

5.14. Finite Element Grid Measurements...... 375

5.15. Modal Measurement Set-up...... 377

B.1. Reed Modal Shapes and Stresses...... 398

XVII! List of Tables Table Page

2.1. Comparative Size of Arundo donax Anatomical Elements...... 129

3.1. %C Data for Pristine and Spent Reeds...... 144

5.1. Statistics for Reed Structurai Measures...... 362

XIX Preface

The flow of fluids around structures, ranging from clarinet reeds to skyscraper, can cause destructive vibrations as well as useful motion...

-Robert Blevin, Flow-Induced Vibrations'^

A thorough search of the literature pertaining to the clarinet reed (see

Shea for such a list) reveals a large number of articles on practical issues, such as reed preparation and scraping, but few articles concerning scientific aspects of the reed. Neither the biology (except for Perdue’s introductory work and Veselac’s landmark dissertation), the chemistry

(indeed, there is no source outside of the chemical literature for information on reed chemistry), nor the physics (although, to be sure, in this area there have been a few articles with broad implications on topics such as reed resonance) have been covered to any great extent.^

At this point in history, the lack of information on the clarinet reed can no longer be dismissed with the remark that the undertakings necessary to remedy the situation are beyond the reach of current science. Advances in material science, physics, chemistry, and botany are such that many of the questions clarinetists would like to have answered about the reed and its

1 2 behavior are now within the capabilities of researchers to do so. This document is intended to be a first step in that process of the scientific study of the clarinet reed.

Because this task it has had to be interdisciplinary in nature from the outset, however, a few words are in order for both the musician and the scientist who, of necessity, must confront each other in the course of these pages. The author has decided to write each section, musical or scientific, at a level which is as close to a professional discussion of the research presented in the area as possible, while acknowledging that its principal readership will, in all likelihood, be drawn primarily from non-specialists in that area. This decision was made in order to avoid dwelling too extensively on background material.

Little explanation is given for such things as differential equations, for example, or the workings of the electron microscope, most standard laboratory equipment, or even how to hand-make a clarinet reed-to do so would simply have expanded the document well beyond the confines of reason. Because, however, there is probably no one reader who is well versed in all of the areas presented in this document, where it would increase ease of presentation, the author has provided “essential” background for some of the tests, chemical and reed nomenclature, etc. In addition, the author has introduced each new term used by bold-faced type to make them easier to find for the reader’s reference.

In order to bridge the gap between art and science, a discussion is made 3 of not only the scientific implications of the findings contained herein, but also a discussion of the musical implications as well. Although not all questions about the clarinet reed could be answered given the limitations of resources for this research, hopefully some of the questions which have been the source of frustration, guesswork and folk wisdom over the years will be answered herein, and in some cases, misconceptions corrected.

As simple as the clarinet reed appears to be, the scientific effort required to understand it is, nevertheless, not small. Many of the concepts needed to explain the reed behavior on a material and physical level are at the forefront of current scientific knowledge. It is hoped that this document will help to motivate other people, far better qualified than the author, to bring their own expertise to the problem of the clarinet reed.

The relationship between the musical arts and the physical sciences is very old. It is hoped that this document will help extend that relationship a little farther. Endnotes

1. Robert Blevins, Flow-Induced Vibrations (New York: Van Nostrand Reinhold Co., 1990): 1.

2. David Shea, “The Clarinet Reed: A Bibliographic Addendum”. The Clarinet Magazine (November/December, 1990): 28-29; Robert E. Perdue, Jr., “Arundo donax-Source of Musical Reeds and Industrial Cellulose.” Economic Botany, vol. 12, no.4 (Oct.-Dec). Industrial Cellulose”. Economic Botany, 12, no.4 (Oct.-Dec, 1958): 370-390; Marilyn Sue Warren Veselac. “Comparison of Cell and Tissue Differences in Good and Unusable Clarinet Reeds” (Ph.D diss.. Ball State University, Muncie, Indiana, 1979); Stephen Thompson, “The Effect of Reed Resonance on Woodwind Tone Production,” Journal of the Acoustical Society of America, 66, no. 5 (May-June, 1979): 1299-1307. Chapter I

Introduction

Early Uses of Arundo donax in Music

Of the three essential elements in the tone production of single-reed woodwind musical instruments-the resonating tube, the mouthpiece, and the

"reed", the most temporary and variable is the reed, because is usually consists of some form of plastic or semi-plastic (bio)polymeric substance which is subjected to both chemical and mechanical degradation during its use. The material still most commonly used to produce the characteristic sounds of the clarinet and saxophone families of musical instruments is naturally occuring--the stem of the monocotyledonous grass, Arundo donax L ,

(commonly, giant reed plant) .

Before it was used in the construction of woodwind reeds, however,Arundo donax, a relative of the family of (although definitely not bamboo, as is sometimes thought) was used primarily to form the entire body of primitive types of musical instruments. It was originally utilized as it occurred, in situ, by cutting portions of the long hollow cylindrical stem (having a cross-sectional diameter on the order of one inch) to a suitable length for the body of the instrument. The earliest known examples of such musical adaptations of the plant stem date back to the prehistoric forms of reed flutes and Pan pipes (an instrument consisting of many hollow tubes of different lengths coupled together in a row, functioning much as a row of glass bottles of different lengths-each pipe sounding a different pitch depending upon which segment is blown into (figure 1.1)).1

Some time before 3000 B. C. two new classes of musical instruments were developed in which some part of the musical instrument itself was set into motion by air passing from the lings into the air column (the direction of flow is indicated by the letter U in figure 1.1). In the double reed variety of instruments, the vibrating portion was separate from the body of the instrument, and may have been developed in primitive cultures as a result of a simple experiment in placing two thin reed slivers together and blowing. It is not our purpose here, however, to develop the long and complex history of double reed instruments.

The details of early double reed instrument history will be deferred to the many fine texts on musical organology currently available.^

The single reed variety of these early reed-driven instruments consisted of the hollowed tube of Arundo donax having a long rectangular segment cut into one end near the opening of the tube which acted as a vibrating tongue or

"squeaker" activated when the player blew through an open end (figure 1.2).

The instrument was often cut with several tone holes and mounted tandem to a second tube which sounded a single drone note.

Scholars are uncertain which of the two prototypical classes of single or double reed instruments developed first.3 Reed Flute Mouthhole Ô a o D Arundo donaxintemode

Pan Pipes

Arundo donaxintemode Frequency increasing U (= air flow' vector) ^Coupling Box

Figure 1.1-Early Reed Flute and Pan Pipe 8

Arundo Donexlntemode

U D

Prototype Double Reed Instrument

Arundo Donaxintemode Sque^er Flap ( "Reed") O CD CD O Qf

Drone■

Prototype Single Reed Instrument v ith Drone

Figure 1.2-Early “Single Reed” and “Double Reed” Instruments Although the double reed variety of such instruments flourished historically, assuming a variety of guises from bagpipes to racketts to early oboes, single reed instruments remained in their undetached squeaker form, except for some curious examples where reed slivers were mounted to tubes of ivory or metal which served as a holder.4 Despite being used for two thousand years, a clear-cut detached single-reed system does not seem to have developed at all before the late seventeenth century.s

A Brief History of the Clarinet Reed

The use of detached single reeds placed over a mouthpiece opening appears to have developed simultaneously with the first chalumeau (late seventeenth century). In fact, the name itself, "chalumeau", derived from the

Latin word, calamus, meaning, "a small reed", attests, somewhat, to this fact.e

According to a surviving chalumeau produced about 1700 by the famous

Nuremberg instrument maker, J. C. Denner (which may even possess a contemporaneous reed), the first "clarinet" reeds were very wide-1.5 cm. vs.

1.35 cm. for modern reeds-being mounted to equally wide mouthpieces. They were essentially rectangular in shape, with no rounding nor tapering at the tip (it was, essentially, a flat, rectangular plate), had a longer vibrating length relative to the clamped portion than modem reeds (figure 1.3), and were somewhat thicker than modern reeds- a feature accommodated by the less restricted acoustic requirements of earlier instruments. The wider, flatter, and thicker reeds allowed a looser, more flexible embouchure to produce a sound.7 10 1.5 cm Top 1.35 cm Superelliptical View Roundijig 1 1.2 cm Square Tip (N=oc) (N«1.5-3?) I V 2b 3.0cm ^ 3.35 cm {

1.2cm

} 2.1 cm { 3.4cm a. 1:3 Era Reed (1 7 2 0 -1 8 20) b. 2:3 Era RsecK 1820-1860) Cross-sections I c. 3:3 Era Reed (I8 6 0 -) 1 . C = 3 Ground Tissue Epidermis—

2 - W E■Vibratingw,k Clamped" 3. O Surface

Side View Front Vamp Back Vamp Front Vamp Back Vamp

Superelliptical Tip Rounding:

(2Lf”+ (lf = 1 a b

Figure 1.3-Historical Reed Development 11

Although the Encyclopédie of Diderot and D'Alembert (1751-1765) contains an engraving of an early clarinet reed, most of the evidence as to their construction and shape is scant and inferential.8 The first single reeds made specifically for the clarinet appear to have been little more than thin rectangular slabs about 1 mm. thick. The rectangular shape is probably derived from the

rectangular "squeakers" of the older single-reed pipes. This conjecture is supported by the fact that when these earliest plate-iike reeds are afixed to the mouthpiece the similarly to the reed squeaker arrangement of figure 1.2 is unmistakable.9

A simpie development would have been to remove the rectangular squeaker completely from the body of the squeaker-borne instrument and attach it to a separate mouthpiece, retaining, in the process, both the squeaker (now, detached) and the window opening through which the squeaker (reed) could move. Early mouthpieces were little more than tube extensions of the chalumeau body, being practicaliy continuous with the instrument, making the resemblance to squeaker-driven instruments evenc le a r e r.io

In retrospect, the design for the early reed, in which the lengthwise tapering common to all modern reeds is lacking, may seem strange to players of today, accustom as they are to having a thin tip from which to produce the tone, however, given the mouthpiece design of the earliest clarinets, with the interior chamber showing very little taper compared to modern mouthpieces, flat plate reeds would have been quite stabie (the book by Baines contains an interesting x-ray of an early clarinet, showing the interior of a mouthpiece in detail).n 12

We shall show in chapter five that the thickness variation of a reed is a direct function of the volume variation inside of the mouthpiece chamber. Since modern mouthpieces contract towards the tip, the reed must be made thinner at the tip to balance the forces generated inside of the mouthpiece, which decrease as one approaches the tip of the mouthpiece (although at the very tip other physical forces, such as Bemoulli forces, increase the pressure over this small region). The earliest mouthpieces, being almost constant and cylindrical in chamber shape would require reeds equally as uniform in thickness. Thus, the simple plate-like shape of the earliest reeds would have been appropriate for the instrument of the time. In any event, this primative reed shape makes vibration analysis very simple, as it is easily approximated by a simple rectangular beam or even a thick plate (chapter five).

The earliest reeds, specifically designed for the clarinet, had a playing surface

(the region of exposed cane) about 3 millimeters (mm.) shorter than modern reeds, judging from photographic evidence of early mouthpieces contained in organological references, such as the The New Grove's Dictionary of Musical

Instruments.^^ In the aforementioned work, in the article entitled, "Clarinet," sample photographs of old and modern mouthpieces and reeds are presented.

Since the sizes of both modern mouthpieces and reeds are known precisely, these values may be used to scale the other photographic measurements of earlier mouthpieces and reeds to correct size.

Using this approach, the vibrating surface of the reed, which is the portion placed in the player's mouth (called the front vamp or, more commonly. 13

simply, the vamp), Is 3 cm. for the earliest reeds (ca. 1760) vs. 3.35 cm. for a

modern reed (figure 1.3).

These measurements are under the assumption that the vibrating portion of

the reed is coincident with the opening (called thewindow ) in the clarinet

mouthpiece. This may or may not be a valid assumption. It does seem to be

supported by iconographie evidence, such as from VneEncyclopedie^ mentioned

above. F. Geoffrey Rendall, the clarinet historian, has also advocated this

photographic method of inferring reed dimensions from available contemporary

mouthpiece s h a p e s js indeed, since early reeds are all but nonexistent, this

inferential approach may be the only general way to derive even approximate

information about the development of the clarinet re e d . 14

The epidermal or unexposed clamped part of the earliest reeds (we call this the back vamp) was very short, gradually lengthening as time went on: 1.2 cm.

(ca. 1760) vs. 2.1 cm. (ca.1850) vs. 3.4 cm. (modern). The reason for this

difference in length is that the table, i.e., the flat portion of the mouthpiece against which the back vamp rested, was much shorter on early mouthpieces than modern ones, only gradually reaching down the full length in the late

nineteenth century.

The evolution of the mouthpiece allows one to draw some conclusions about the evolution of the clarinet reed. Denote the ratio of the back vamp length to the

front vamp length as rbf. Acording to measurements of the tip shape and table

lengths of photographs of old mouthpieces (realizing that the conclusions

drawn are based on a limited sample), the development of the clarinet reed may 14 be divided into three somewhat arbitrary but convenient eras according to gradual change in the rbf parameter: the 1:3 era, the 2:3 era, and the 3:3

(1:1) era, respectively (figure 1.3).

The 1:3 era (1720 to 1820) was characterized by ultra-short back vamps (1/3 of the length of the front vamp). There is some variation in the rbf in measured drawings, with one by Geoffrey Rendall having a 1:4 ratio.15 |n any case, this period has the largest overall value of the rbf parameter (typically, 1:3, more or less). The front vamps of the earliest reeds, in addition, were shorter and wider than modern reeds. Reeds of this era were played against the upper lip, and were either rectangular or (later on in the era), rounded at the tip. A mouthpiece made in 1780 in Paris shows unmistakable rounding at the tip.i6

The detailed profile of the reed varied enormously during this period. The earliest reeds (1720-1750), as mentioned, were fairly thick, flat plates ofArundo donax. After 1760, several other shapes began to emerge. Baines reports that the Tutor by Backofen (1803) is the first to give a detailed account of the reed shape. 17

According to Backofen, at the turn of the nineteenth century the tapered reed tip had not yet gained universal aceptance. The older flat plate reed, about one millimeter thick at the tip, coexisted along with the newer tapered tip reed. The tapered reed of the era, being shorter, thicker at the tip, and thinner than a modern reed, had a correspondingly shorter moment arm. Such a reed would probably feel "stiff if played on modem instruments (interestingly, the Vandoren

Black Master reed comes close to this historical profile). In any event, the 15 mouthpiece and air column of the time would have compensated for this resistance.

The transverse (width-wise) cross-section of the reed could be, according to

Backofen, uniform and rectangular (as in the flat plate) or convex on both the top and the bottom of the reed or even flat on the top and concave underneath

(since the reeds of this era were played "upside-down" compared to modern reeds, the flat "top" refers to the part facing the mouthpiece, so that the cross- section is equivalent to the modern reed shape), as in figure 1.3.

Whatever the cross-section, the reeds were generally thicker than modern reeds. The contemporary clarinetist, Muller, however, comments that such reeds were hard to c o n tro l.18 |n addition, such thick reeds required that the lay be very long to create the forces necessary to the physics of the reed (chapter five). When thinner reeds were accepted, the mouthpiece requirements were relaxed, allowing for the shorter lay which is used today, and consequently, greater control of the physical motion of the reed. The reed of this era, with a thinned tip, is probably the type which was used by professionals during and after Mozart's time.i9

Along with the variations in length and cross-section during this era, the width of the reed also changed- from 1.5 cm. in the earlier part of the eighteenth century to as small as 1.2 cm. in German reeds of the later part of the century (even today, German reeds tend to be narrower than other reeds, maintaining the 1.2 cm. width).so These changes are reflected also in the 16 narrowing of the mouthpiece window, especially at the bottom of the window, so that a quasi- "v" shape, with a top window opening of .93 cm. and bottom window opening of .33 cm. was formed (as one representative example).The value of rbf also changed as the table became longer, so that by the end of the century, 1:2.5 ratios were common. By 1800 this value was at 1:2 or smaller.

The date of the development of the rounded tip reed is impossible to establish, although, based upon mouthpiece designs, it can be established before 1780. Both round-tip and square-tip mouthpieces coexisted during this period. In fact, judging from photographic evidence, this situation existed as late as 1800, and so, presumably, did round-tipped and square-tipped reeds.

These early reeds, although mostly made of Arundo donax, were also made of material from the pine tree or fir tree, which must have been very thin in order to compensate for the difference in material properties, (as discussed in chapter four). In any event, they deteriorated easily. Materials such as fishbone, or possibly metal and ivory were also tried.2i

The reed was generally hand-made during this era (mass production began in the middle of the nineteenth century-by 1870, at the latest); however, there were small shops as early as 1770 which produced reeds for commercial consumer u s e .22

The reed was attached to the mouthpiece during this period by means of waxed thread or a silkencord.23 Rings were cut into the mouthpiece to allow for this. The tone was said to have good flexibility and sonority in the lower octave.

We shall see in chapter five why such a situation might be explainable 17

based upon the vibrational characteristics of the reed, and the tendency for the

ligature to act as a high-pass or low-pass filter.

The 2:3 era (1820, about the time of the invention of the metal screw

ligature by Ivan Muller, to 1860) was characterized by an back vamp length about 2/3 that of the front vamp, with the length of the front vamp being about the same as in the 1:3 era. The heel of the 2:3 era reed (see figure 1.10 for an explanation of reed structure terminology) was thinner than modern reeds, and the tip was definitely rounded. The earliest reeds from this era were played with the mouthpiece turned up or down, so that the reed was against either the upper or lower lip. As time went on, the reed down position became standard (it was supported by some players as early as 1784, with about fifty percent of them using the window down position by 1 8 2 4 ).24 The Paris Conservatory adopted the reed down position as standard practice in 1831.25

The variation in the rbf was rather large during this pre-standardization period-from about 2:3.4 for the reeds used on the European continent, to about

2:5.2 for reeds used in England (according to the mouthpieces shown by

Rendall).26 In addition, the width of the reed varied enormously; many of them were still hand-made. The average width was slightly narrower than modern

reeds, 1.2 cm. vs. 1.35 cm. The reed became wider as time progressed, instrument bores changed, and mass production of reeds began, eventually

reaching the modern width for French cut clarinet reeds.

The 1:1 era (1860 to the present) is characterized by a reed having a back vamp and front vamp of about equal length, being slightly wider (1.35 cm.) and 18 thicker at the heel (approximately, 2.5 cm. to 3.5 cm.) than in the 2:3 era (figure

1.3). The tip was rounded, and the playing position was with the mouthpiece turned down.

It is not clear what caused the historical development of a rounded reed tip and mouthpiece (which became universal certainly after about 1820), but the need for the alteration was possibly empirically deduced by manufacturers.

One may describe the rounded part of the reed as a semi-ellipse, whose length

(semi-major axis) is 1.3 cm. and width (semi-minor axis) is .25 cm. This is a tedious composite shape to model for physical calculations (players often modify the shape by hand, anyway) and so an approximation may be useful, which, coincidentally, allows for a natural description of the historical development of the tip, as well. The reed shape may be approximated mathematically by a plane figure known as a supereliipse.27

A superellipse is a generalized form of an ellipse. The formula for an ellipse, the reader may recall, is

(x/a)2 + (y/b)2 = 1 ( 1.1)

where a is the length of the semi-major (long) axis, and b is the length of the semi-minor (short) axis. A superellipse is defined by the formula:

(x/a)2N + (y/b)2N = 1 ( 1.2)

There are several useful limits of the superellipse equation useful in modeling the tip rounding of the clarinet reed. When N =1, the simpie ellipse equation is recovered, so that if the tip were a perfect half ellipse, it couid be 19 described as a superellipse with N = 1 on the half plane. If N = <», then the superellipse degenerates into a rectangle, and describes the case of the 1:3 era reeds, with their rectangular tip. Thus, the 1:3 era reed tip "rounding" may be described as a superelliptical shape with N =°o. For values between 1 and oo, the ellipse becomes gradually more squared (figure 1.4 and figure 1.5).

Based upon empirical measurements, the modern clarinet reed, with its approximately elliptical tip, may be described as a superelliptical shape with

N = 1.786. This value was determined by rewriting equation 1.2 as:

1 2N / xb . 2N b - ( — )

(1.3)

Since this is an implicit equation, N cannot be calculated directly. We rewrite the equation as:

(1.4)

A reasonable “seed’ value of N2 is chosen and from the known values of a, b,

X, and y, a value of Ni is calculated. One continues the iteration until Ng and Ni converge. This is the approximate value of N.

A computer simulation of one-half of a reed tip which uses this value of N is \

—1 - .8 — .6 - .4 " .2 .8 1

Figure 1.4-Clarinet Reed Superellipse Parameter (N) Variation 21 1

.8

A

.2

0 1 .6 0 5 1 N=1 N=2

N=3 N=4

Figure 1.5-Effect of Changing N in the Superellipse Equation (N = 1.78 for Clarinet Reed) 22

shown in figure 1.6. The graphing of only one-half of the reed tip is due to

computational difficulties, as non-integral values of N yield some non-elliptic

regions near -1 on the x axis, and this region has been omitted for clarity. The

profile looks quite close to that of a modern clarinet reed (the scale is arbitrary).

A comparison of the historical shapes mentioned in this chapter, inferred from

available historical mouthpiece designs, is shown in figure 1.3. A comparison

of typical vibration patterns of the first mode of a struck reed withN = ©o and N =

1.786 is given in chapter five.

While it is clear that the evolution of the mouthpiece influenced the

development of the clarinet reed, the influence of changes in the air column on

mouthpiece evolution is poorly understood, and must await a better

understanding of the physical relationship between the two.

Modern Manufacturing Techniques for Clarinet and Saxophone Reeds

Arundo donax is a subtropical to warm/temperate plant, produced commercially for use as musical instrument reeds chiefly in the southeastern

portions of France (figure 1.7) near the Var river, which is seventy-five miles

long, entering the Mediterranean four miles southwest ofN ic e .28 The towns of

Cogolin, Frejus, and Le Lavandou are often mentioned in this regard.29 Other

Mediterranean countries near France also possess a climate suitable for its

cultivation. The plant has been exported to many other regions of the world,

such as India, China, southern Africa, England (although there appear to be

relatively few pockets of the plant left in that country), Spain, the West Indies, 23

0.7

0 6

0 5

0.2

0.1

0.2

Figure 1.6-Computer Superellipse of 1/2 Reed Tip (N = 1.786) 24

Paris

Orleans

Dijon

Limoges

lordeaux (Provence)

Nice

Figure 1.7-The Var Région of Southern France 25

Australia, and the more tropical regions of North America-most notably,

California and the lower fringe of the United States: Texas, Florida, and up the

coast to near Washington D.C.so

Despite the wide geographicai distribution and ecological tolerance of the

plant, the Var region has enjoyed a century-long reputation as the only region

capable of producing high-quality reed material. Neither the nature of French

cane as compared to cane grown in other countries, nor the soil quality nor

growing conditions in the Var region have ever, to the author's knowledge,

been subjected to rigorous scientific studies (excluding possible proprietary

research done by the reed companies themselves). The comparison of quality

between French cane and other national canes, as well as the study of other

agricultural matters, are projects which definitely need to be done in the study of clarinet reeds, but they must be left to those who have the resources available to do so.31

Despite the lack of scientific scrutiny, however, many writers and reed

dealers continue to maintain the exclusivity of the quality of French cane, and

indeed, reed companies often cite the “rich soil and moderate climate” as the

reasons for the quality of their cane in their advertisements.32 Be that as it may, there simply is no evidence available at this time to decide if these claims are justified. In any case, the discussion may be moot, considering Perdue's

prescient remarks on reed research made nearly thirty-five years ago: 26

Any American cane made available at a price and quality competitive with the French product will still have a great psychological disadvantage. A majority of musicians are thoroughly convinced that only cane from France is suitable for their work. This belief can be attributed to the dismal failure of non-French cane as well as the good reputation that French cane has long enjoyed. A careful study of atmospheric and soil conditions under which cane is produced in southeastern France should point the way to the most satisfactory American environment. Such an environment appears to be in the southwestern or western United States.Careful study of the botanical and agronomic characteristics of the plant, in connection with the the detailed study o f the physical and musical quality of the cane, would provide a firm basis on which attempts to establish the industry could be made [italics, mine]. It is questionable whether the industry could profitably support the expense of the necessary research.33

Currently, clarinet and saxophone reeds are the result of a long, tedious, growing, curing, and manufacturing process (figure 1.8) which begins with the planting of the root of the plant-the rhizome (these and other morphological terms which will be described in more detail in the next chapter), typically, in

June.34 It is a myth that Arundo donax grows from seeds, even though they often develop a plume-like structure(panicle) near the top of the stem which is often mistakenly thought to contain seeds.

According to Perdue, these large, very fleshy roots are placed in rows, about two to three meters apart (Vandoren reed company literature places the spacing of the rows at twenty-five feet), and planted about ten centimeters deep (figure

1.8). 35 The plants are watered and hoed as needed (usually, very

rarely), and cultivated very simply-by removing weeds or immature (small diameter) versions of the plant.

The plants develop vertically (botanists cali upward growth, primary growth) for the first year, and then, once achieving full height-about ten to 27 2. Primary 1. Planting Rhizome Internode[ Growth Top Soil t Clyr. Primary Growth / / S i t e s \ ' Node— ^ ÎIÎ!-' Rhizome Primary Thickening / Adventitious Roots

3. Drying Crab 4. Cutting and Stacking

(2-4 months)

(1 yr.to 2yr.) Racks

|5. Splitting and Cutting Quarter Cut :o: Blank

o;

6. Finishing] 7. Shave Vamp and Round Tip (Flatten Bottom)

Figure 1.8-Steps in Reed Manufacturing 28

twenty feet-the cylindrical stem of the plant stops growing upward and begins to thicken outwardly by the addition of new parallel lengthwise (not horizontal)

layers, being allowed to develop, typically, for another year.

The material being added to the thickness of the stem is not properly referred to as "wood", in a botanical sense, since this is reserved for a type of plant material having a different type of 1) cell wall structure and 2) growth pattern than Arundo donax and most other grasses possess. Unlike the grass stem of

Arundo donax, once a true tree has finished its primary or vertical growth phase

(chapter two), it begins to grow outward in a clearly horizontal direction (botanist call this type of growth, secondary growth). This phase is accomplished partially by the addition of new layers of cells which push outward horizontally

(figure 1.9) each new growing season from cells which form a ring, known as a growth ring, or vascular cambium. These growth rings or vascular cambia form anew at the beginning of each season from the position where the growing stopped in the prior season.

The other mechanism by which the expansion of the cells outward is accomplished is by addition of secondary inner walls to the existing primary (or single thickness) cell walls of the already present plant cells as the new cells in the growth ring simultaneously begin to expand outward (figure 1.9).36 The

presence of these secondary walls is what charaterizes true wood cells. Grass cells do not have them.

The thickening whichArundo donax undergoes, by contrast, is not due to the

horizontal outward growth of cells, but, rather, is due to the addition of new Cell Growth Orientation Plasmamembrane (variable size and shape) [simplified ^ee"ce11l Secondary Growth

Primary Primary Growth Cell Vail Direction Secondary Trees and Grasses Cell Vail Cell Growth (Initial (upward) Orientation Nucleus growing phase) ^ olgi Apparatus

Plastid

Primary Thickening Grass Stems Cross-section Plasmamembrane (variable size and shape) (S = Secondary Wall)

Primary Cell Wall(P)

Nucleus Cross-section Golgi Apparatus

Plastid

Cell Wall Simplified Grass Stem Cell Slime Plug Filling a "Leak In a "Food" Conducting Cell

Figure 1.9-A Blomorphologlcal Miscellany 30

parallel vertical layers of cells next to the preexisting vertical layers of cells in the stem formed during primary growth (figure 1.9). This type of outward growth,

derived from vertical additions, is more properly referred to, botanically, as

primary thickening, since it uses the same mechanism as in the primary or vertical growth phase of the plant to achieve its thickening.

This primary thickening reaches one to four centimeters thick in the case of

Arundo donax. This is much less of an increase in thickness than in a true tree of similar initial diameter because of, among other reasons, the lack of secondary cell walls in Arundo donax, which would otherwise result from secondary growth mechanisms, as in true woods. This lack of a secondary wall structure gives Arundo donax less structural support, limiting its buckling strength. This biomechanical limitation is one cause for the smaller diameters and less thickness of the grasses in general.

After two years, the material is han/ested, typically, in winter (December,

January, or February), bundled in groups of twenty stems and placed in an x-shaped configuration called crabs (figure 1.8), which resembles the inner frame of an Indian te p e e .3 7 Once dried, the reeds are stripped of their leaves,cut into four-foot segments, and placed horizontally on low supports to dry in the sun for at least one, and possibly as long as two years.

Given the wide variability in the vibrational responses of manufactured reeds, it is unlikely that the current market pressure for clarinet reeds has allowed a curing schedule that is optimal (forcing the shorter, rather than the longer lengths of time), yet sunning is essential in stabilizing the reed ultrastructure 31 and histochemistry, since during this period chlorophyll, contained in the cell walls, is changed into other chromophores (color producing compounds) which

impart the characteristic golden color of fine c a n e .3 8 The rigidity of the cell walls of the plant (and subsequent reed) likewise stabilizes, being far too pliable for musical use when first harvested.

One point must be emphasized: the material used for woodwind reeds is made from an excised section of a dead plant. Although this point may seem obvious-that the reed material is no longer alive when used for musical purposes, nevertheless this is not trivial from a biostructural viewpoint. The biological and histochemical properties of live and dead plant tissues are quite different from each other. A living plant cell, as an example, has a plasmamembrane (a semi-liquid wall of lipid material inside of the rigid plant cell wall, functioning much as the inner tube in a tire) which hold the fluid-filled contents of the plant cells (figure 1.9). Within these plasmamembranes the plant metabolizes carbon dioxide, excretes wastes, stores nutrients, titrates chemical growth regulators, controls water, respiratory, and nutrient flow, and

participates in a host of other metabolic processes.

When the giant reed plant is harvested for the production of reeds, the cutting of the stems of the plant causes those plasmamembranes closest to the site of laceration to rupture, releasing their contents. This "wounding" of the plant causes an immediate mobilization of biochemical life-saving processes in the plant, such as the formation of sllme-plugs In the food conducting tissue

(similar to the formation of clots in animal injury) which severely alters the 32

original biochemical composition of the plant material (figure 1.9).

Eventually, these severe biochemical processes subside as the slime-plugs

slow down the leakage of material, but what remains for musical use is the

shell of the plant-partially filled with a severely altered histochemical

plasmamembrane environment relative to the original tissue. In time, the

remaining plasmamembranes are drained or evaporated, leaving the form, but

little of the substance, of the original material. It is important, then, to realize in the discussion to follow that the reed material used by musicians is in the final

and relatively fixed stage of a long series of physiological processes occurring

in the plant both before and after harvesting.

Once sun-dried, the stems are processed. The stems (now calledstokes or

batons) are split into quarters by means of a hammer and fluted spike. The quarters are then given a flat bottom. At this stage, the reed is called ablank, and consists of a rectangular slab of reed material. The blank is then precision cut into the shape of the traditional clarinet reed as the final step (figure 1.8),

possibly using diamond-tipped cutting machines, due to the high silica content

of the plant, which would quickly ruin standard stainless-steel blades. The

slope of the reed, as it tapers to the tip, is of paramount importance in the vibrational response of the reed. Different manufacturers have developed

different slope profiles over the years, trying to find the best profile for the widest

number of players.

Many excellent books and articles show the clarinetist how to make a reed

starting with the raw internode (i.e., one tube-like section of the plant), and we 33 have nothing new to add to the procedure, although we shall have much more to say about the final reed shape in chapter five. Rather than summarize the various methods of reed making, we refer the reader to the literature.39

Reed Nomenclature

At this point, it is useful to introduce the nomenclature scheme for the parts of clarinet reed which have become part of the common language of clarinetists, and a separate nomenclature scheme which will be useful in more technical discussions.

A schematic of a clarinet reed is shown in figure 1.10. The clarinetist's terminology for the parts of the reed are given, in part, according to Opperman's terminology.40 As may be seen, the terminology has been derived from a mixture of sources, some more derivative than others. The references to the human body are obvious.

The problem with using this terminology is that, while descriptive of some structural markers of the reed surface, it cannot easily describe with any precision the location of biomorphology and elements within a specific manufactured reed shape. To allow more specificity in the description of the reed surface, an x-y grid in millimeters may be overlaid on the reed surface with the zero vertical line (x axis) passing through the middle of the reed and the zero horizontal line (y axis) passing through the interface between the front and back vamp. We shall always denote the region containing the elliptic tip as one separate subdivision, one coordinate line, T (figure 1.11). 34

—Rounding Point /P oint of Resisting Heart Front Vamp (Vibrating) -Crest (Ridge) i \ Right - Shoulder L e ft------—Fiberous Region Shoulder “ Shoulder Underside ^Stock Back Vamp (Clamped) — Epidermis

Top Viewl \ / iBack View ------Heel (Butt) '------

Front Vamp Back Vamp ^side Profile

Left Edge Side View

Figure 1.10-Common Reed Nomenclature 35

- 4 0 6 (mm) . T Li ne 16 mm rff i------

Epidermal Interface^ 2 mm 0 mm 2 mm

16 mm

Figure 1.11-The Reed Grid 36

Looking down on region T, it is a half-ellipsoid shaped region containing the tip, with a semi-major axis of (typically) 1.3 cm. and a semi-minor axis of .25 cm. The slope (the thickness variation) is very slightly positive (i.e., increasing ), with a rectangular cross-section of slightly increasing width as one proceeds from the initial tip inward to the beginning of the rectangular sides.

The remainder of the front vamp is of complex profile. It is in this region that different reed companies show the greatest variability in manufacturing.

One company (Rico) has a simple rectangular cross-section which is 1.3 cm. in width with a linear increase in thickness along the length of the regions (figure

1.12). Such a "wedge" shaped profile (we shall call it thewedge profile) allows the reed to approximate a gradually thickening beam or plate, and makes the mathematical analysis of its vibration (chapter five) fairly tractable.

The reed is generally reported to require little pressure to initiate vibrations.

Another popular reed shape (Vandoren, Olivieri, Glotin) contains an expanding ridge or crest in the front vamp. This expanding ridge area makes analysis much more complicated due to the variation in a mathematical property known as the radius of gyration, k (which is the moment of inertia,

I, divided by the area function. A, both of which vary along the length of the reed with this shape-more detail on the radius of gyration is given in chapter five).

The cross-sectional geometry of such an expanding ridge profile (we shall refer to it as simply the ridge profile, when this is unambiguous, since there are many possible ridge profiles) may be further divided into four horizontal subregions (approximate values in cm.): 3r, 3rp, 2rg, Ire. as indicated in 37

-1 -2

T ~ )

Cross-section Cross-section

Slope Slope (Thickness) (Thickness)

Figure 1.12-The Wedge Profile 38

figure 1.13, based on variations in structure and thickness (which are responsible for the difficulty in assigning a simple value for k).

3r is the region just beyond region T, and is still dominated by the rectangular cross-section of the T region. In the dynamics of thin beams and plates, the concept that a change in thickness is equivalent to a change in material strength (rigidity), may be important in the thinner portions of this region, since scraping the reed here amounts to large changes in the local bending modulus of the material in this region.

Moving down the reed in region 3r, as the ridge begins to emerge, it does so first by the addition of a small protrusion or "bump" in the middle of the otherwise rectangular cross-section of region 3r to form a compound cross- sectionalshape, which is sufficiently different enough to denote as a separate region: 3rp (i.e., the region 3 centimeters from the front vamp/back vamp interface with rectangular cross-section and a protrusion stacked on top).

Progressing down the reed, the protrusion grows both higher and wider, spreading out beyond half the reed width. At this point the protrusion may no longer be viewed simply as a perturbation of region 3r, as it eventually assumes a size nearly equal in width to the rectangular portion. This is region

2rg, in which a more or less Gaussian-shaped cross-section, which almost reaches the sides of the reed, is attached atop the rectangular cross-section.

The protrusion and the Gaussian cross-sections along with the underlying rectangular base cross-section in the 3rp and 2rg regions, taken together. 3 9

Cross-section

Cross-section

Slope Slope (Thickness) (Thickness)

Figure 1.13-The Expanding Ridge (Crest) Profile 40 comprise the "heart" of the reed.

Finally, in regions 1re and 0, the Gaussian shape extends to the entire width of the reed, and the sides fill out, forming a true half-elliptical cross-section on top of the rectangular cross-section. The thickness of both the rectangular and the elliptical layers continue to increase linearly along the reed length towards the epidermis.

Such an expanded ridge reed profile is generally reported to offer more resistance to the initiation of vibrations than the wedge profile, no doubt due to the added mass of the ridge near the tip (in the 3r subregion), but is also reported to provide more "depth" to the sound.

The amount of variation in the thickness in each of these four regions of sections 0, 1,2, and 3 cm. differs from reed company to reed company, with the

Olivieri company preferring a steeper Gaussian curve in region2^g than either

Vandoren or Glotin (the base rectangular cross-section is thinner in the same regions for the Olivieri shape compared to the other two, with the result that the reed is somewhat more flexible, but also somewhat more sensitive to scraping).

Other theoretical shapes besides the wedge shape or ridge shape are possible in these two regions, although the wedge and the ridge profiles are the two most common shapes. These other shapes may also lead to other simple or complicated subdivisions as well. We shall discuss other theoretical profiles in detail in chapter five.

The back portion of region 2, and all of region 1 (figure 1.11) is a continuation of region 2re, but the thickness has reached its maximum and 41

stabilized. Region 0 (figure 1.11) commonly displays thick-walled fibrous tissue

(called sclerenchyma cells) exposed just under the epidermis. It has a

semielliptical cross-section stacked on a rectangular base, as in regions 10 and

2re. This region, while not placed into direct vibration, does participate in the dissipation of energy through the reed structure, especially at higher frequencies (as we shall show later).

The influences of reed shape on the vibration of the manufactured clarinet

reed, is profound. The influence of the shape on both musical quality and the biophysics of the reed will be studied in detail later.

Rationale of Current Study

Despite the fact that Arundo donax has been the preferred material for the reeds of most orchestral woodwinds for over two-hundred years, a careful scientific study of the materiai properties of the stem (the portion used for clarinet reeds) is still lacking. There have been, to be sure, a number of studies that treat certain aspects of the plant material, but a unified study of the biological, chemical, and engineering properties of this material as it relates directly to the area of musical acoustics is lacking. We shall review the published studies, such as they are, in due course. Most of the remaining

knowledge about the properties of woodwind reed material is anecdotal and of

uneven reliability. This folk wisdom will be assessed in the light of the findings of the current study.

As mentioned earlier, this study cannot include definitive work on the naturally 42

occurring plant Arundo donax, because this would require a substantially more

comprehensive approach to many botanical aspects (such as national

variations in the soil, growth patterns, resulting material properties, etc.) than

befits the context of this work. Rather, the purpose of the current study is to

begin the first large-scale investigation of the biological, chemical, and physical

properties of this plant material in the processed form of the clarinet reed, and to examine some possible types of synthetic replacements.

There are many reasons for undertaking such a study of the processed reed:

1 ) to know the basic engineering parameters of the plant material that is selected and processed as clarinet reeds, 2) to understand reed material degradation over playing time, 3) to measure the reed material variability. Both usable and unusable reeds occur within a random sampling of commercially produced reeds, supposedly of the same strength and constitution, 4) to define the best ranges of the material parameters for use in the manufacture of the clarinet reeds, 5) to correlate these properties with synthetic material substitutes, 6) to test the validity of certain types of folk wisdom concerning reed

use and selection among clarinet players, 7) to study the effects of shaping the

reed (i.e., scraping the reed with a reed knife) on the sound quality and vibrational characteristics of the reed, 8) to obtain a better understanding of how the reed vibrates in the environment of the mouthpiece (the so-called scund-structure Interaction problem), 9) to study the effects of the oral cavity environment on the reed material properties and vibrations.

To address these questions relating to the biology, chemistry, and physics of 43 the processed clarinet reed, a number of different experiments and machine tests are run. Each experiment or experimental machine set-up will be described in detail in the appropriate section following in this document, and the resuiting data collected and analyzed.

As mentioned in point number three above, the reed material exhibits large variations in properties from plant to plant. These variations are often passed on to the processed ciarinet reeds in the form of variable vibration response, water capacity (the abiiity to absorb and retain moisture), and a whole host of other subtle parameter responses. The first goal of this study, therefore, is to measure some of the biological, chemical, and physical parameters of pristine (i.e., reeds taken fresh out of the manufacturer's container) processed clarinet reeds in an attempt to define the range of natural variation within the parameter sets and correlate them with desirabiiity and undesirability (i.e., "good" reeds vs.

"bad" reeds).

In order to study the pristine state of the clarinet reed, a method of obtaining a set of reeds rated in terms of playing quality is needed. One-hundred reeds of different representative brands were given, ten each (i.e., one sealed box of reeds), to ten different students of roughly similar professional or preprofessional ability in the clarinet program at The Ohio State University with the instructions to rate the reeds on a number of different quality and performance criteria, without tampering with the reeds in any way, other than simply playing the reed for some time (however long was needed to make the initial determination of reed quaiity). We call this set of rated reeds, {R}, 44 throughout this document. It will form the basis for much of the material tests to follow.

This study also examines the vibrational characteristics of the clarinet reed which depend upon the natural variations in selected parameters of the pristine reed material. To date, the most complicated model of the clarinet reed developed was by Summerfeldt and Strong, who considered the clarinet reed to be a cantilever beam with linearly varying thickness.4i In reality, this study will show that a detailed analysis of the clarinet reed is best done by considering the reed to be a plate, not a beam, since the material thickness to length ratio is on the order of 1/34 at the tip (a plate has thickness to length ratios usually of 1/20 or less). In fact, a full analysis of the clarinet reed shows it to be a superelliptically-shaped, variable thickness, thin-thick, inhomogeneous plate with moving boundary and load conditions involving sound-structure interaction with the mouthpiece chamber. The reality of the clarinet reed is many orders of magnitude more complex than a simple cantilever beam, and even the situations mentioned above vary as the reed ages. We shall take up each of these topics in turn.

Much of this vibrational analysis will be theoretical in nature using the capabilities of several different computers. The machines will model as realistically as possible the vibration of a typical reed as it is supposed to occur in the mouth of a player. Scraping the reed surface, as many players do, will also be modeled. So as not to be completely devoid of empirical comparisons, the theoretical results will be correlated with stroboscopically slowed video 45 data obtained via fiber optic camera methods of reeds actually being played in the mouth of a skilled musician (the author). In addition, acoustic measurements of the vibrational modes of the reeds in air will be compared with calculated frequencies. It is hoped that such studies will greatly expand the knowledge of how the clarinet reeds function musically.

The second global area of study, and the second goal of this work, is to attempt to ascertain a realistic "playing ecology" of the reed. Reeds are subjected to many different force of a chemical and physical nature during the course of their playing life. By studying the changes in various chemical, physical and biological parameters which occur in the reed as a result of playing, the hope is that the causes of the eventually occurring "reed death" may be more precisely defined. As in the case of the pristine reed, the exact experimental protocols for the study of changes in the different aspects of reed material over time will be presented as warranted.

The results of this study may help to improve the quality of manufactured clarinet reeds, and even help to find a suitable synthetic substitute (although, due to the difficulty of exactly mimicking the complex biological structure of plant cell walls by polymeric formulations, this author doubts that plastic reeds will ever supersede the best natural reeds. Nevertheless, the development of a good substitute, useful, for all but the most demanding acoustical situations may be possible).

It necessary to point out that many of the new results presented in this document must be accepted as provisional. Much of the testing involved only 46

limited samples, due to inevitable economic limitations involved in doing

interdiscipiinary research in no less than nine different, highly specialized fields:

music, chemistry, material science, botany, physics, engineering, medical

pathology, computer science, and optical imaging. Multiple sample tests and

reproducibility studies were, thus, impossible in many cases. It would be of great value to independently confirm the findings presented herein, because they relate substantially to the body of knowledge available to both musicians and scientists. There is already too much pseudoscience in the area of reed research to inflict more upon the reader.

In addition, since much of the research had to be done in areas outside of the author's fields of expertise, important aspects of the research might have been misinterpreted or overlooked. Where possible, the author has solicited expert opinions and help in the testing procedures. Any oversights, omissions, or errors in interpretation are solely the responsibility of the current author. 47

Endnotes

1. Robert E. Purdue, Jr, "Arundo donax--Source of Musical Reeds and Industrial Cellulose," Economic Botany ^2, no.4 (Oct.-Dec. 1958): 375-376.

2. Anthony Baines, Woodwind instruments and Their History (New York: Dover Publications, Inc., 1991); Anthony Baines, Musicai instruments Through the Ages (Middlesex, England: Penguin Books, Ltd., 1976).

3. Perdue, “Arundo...,” 376.

4. F. Geoffrey Rendall, The Clarinet: some notes upon its history and construction^ 3rd ed., revised and with some additional material by Phillip Bates (New York: W.W. Norton and Co., 1971).

5. Most or all of the early single-reed folk instruments used squeaker type excitors.

6. Jack Brymer, The Ciarinet (London: McDonald and Jane's Publishers, Limited, 1976).

7. Nicholas Shackleton, "Clarinet," in The New Groves Dictionary of Musicai instruments, Stanley Sadie, ed. vol. 2 (New York: Grove's Press, 1975): 397.

8. ibid., 397.

9. Perdue, “Arundo...,” 374

10. ibid., 374.

11. Baines, Woodwind Instruments..., plate xxix.

12. Shackleton, “Clarinet...,” 396.

13. Rendall, The C/ar/nef...,56.

14. Another way would be to rebuild some of the old clarinets and attempt to make reeds which match known historical descriptions. This might give a clue as to the exact dimensions which produce the best sounds on the instruments. 48

15. Rendall, The Clarinet.., 6 (upper left mouthpiece)

16. Baines, Woodwindlnstruments..., plate vil (no. 8).

17. ibid., 300.

18. ibid., 300.

19. ibid., 300.

20. Shackleton, “Clarinet...,” 396 (illustration d).

21. Baines, Woodwind Instruments...,30^.

22. Rendall, The Ciarinet...,57.

23. ibid., 8.

24. Shackleton, “Clarinet...,” 397.

25. ibid., 397.

26. Rendall, The Clarinet..., 6.

27. C. M. Wang, L. Wang, and K. M. Liew, "Vibration and Buckling of Superelliptical Plates," Journal of Sound and Vibration 171 no. 3 (1994): 301-314.

28. Webster's New Geographical Dictionary (Springfield, Mass.: Merriam-Webster, 1988): 413, 1285.

29. Rendall, The Ciarinet...,57.49

30. Perdue, “Arundo...,” 370.

31. For instance, the British obois/botanist, Clare Lawton has reported work comparing Australian and French cane (private communication).

32. La Canne a Musique. Vandoren Reed Company brochure: 4; The Making of Greatness. Vandoren Reed Company brochure: 2.

33. Perdue, “Arundo...,” 391.

34. ibid., 380. 49

35. ibid., 381; The Making of Greatness..., 2.

36. A discussion of the morphological and ultrastructural differences between grasses and woods may be found in many standard plant anatomy texts, such as, Katherine Esau, The Anatomy of Seed Plants (New York: John Wiley and Sons, 1977).

37. Perdue, “Arundo...,” 382.

38. Frederick Sack,Lecture Notes, Plant Anatomy 656, The Ohio State University, 1991.

39. Kalmen Opperman, Handbook for Making and Adjusting Single Reeds: for all Clarinets and Saxophones (New York: Chappell and Co., Inc., 1956); a series of three articles by Lee Livengood appear in the pages ofThe Clarinet Magazine : 1 ) Lee Livengood, "A Study of Clarinet Reed Making. Part 1 : a case for clarinet reed making," The Clarinet Magazine 19 no. 3 (May/June 1992), 2) Lee Livengood, "A Study of Clarinet Reed Making. Fait 2: a method for making reed blanks from tube cane,"The Clarinet Magazine 19 no. 4 (July/August 1992), 3) Lee Livengood, "A Study of Clarinet Reed Making. Part 3: selected bibliography," The Clarinet Magazine 20 no. 1 (November/December 1992).

40. ibid., frontispiece

41. Scott D. Sommerfeldt and William J. Strong, “Simulation of a Player-Clarinet System,” Journal of the Acoustical Society of America, 83, no. 5 (May, 1988): 1908- 1917. Chapter II

Arundo donax Anatomy

introduction

The structural elements of theArundo donax plant may be classified according to three natural size ranges: gross (= .1 m-8 m), ultrastructural (=.001 m-.l m), and molecular (<.001 m). This chapter will examine the structural elements of the giant reed plant contained in the first two of these categories.

Issues involving the structure of the plant on the molecular level will be discussed in the context of the histochemistry of the reed (chapter 3).

Gross Anatomv

Morphologically, Arundo donax belongs to the generic botanical family for most grasses-Gramineae (its subfamily \sArundinoideae)J Within

Arundinoideae, the plant is in tribe 12, Festuceae (bamboo, a plant for which clarinet cane is often mistaken, is in another\r\be-Arundinaria ). The genus is Arundo, and the species, donax. A complete biotaxonomic classification is given in figure 2.1. For a listing of the specific traits which comprise each taxon the reader is referred to any standard botany textbook.

The gross anatomy of the giant reed plant, as indeed most grasses, may be

50 51

Arundo donax Classification

Kingdom; Plant I Subkingdom: Embryophta I Phyium: Tracheophyta I Subphylum; Pteropsida I Class; Angiospermae I Order; Monocotyledoneae I Family; Gramineae I Subfamily; Arundinoideae i Tribe (12); Festuceae 1 Genus; Arundo i Species; donax

Figure 2A-Arundo donax taxonomy 52 divided into three distinct functional/structural systems; 1 ) the fool system- responsible for stability of the plant underground, nutrient acquisition, and reproduction, 2) the stem system-responsible for the above-ground structural stability of the plant, and the conduction of water, glucose, and other nutrients, and 3) the leaf system-responsible for photosynthesis, transpiration, and some hydroscopic phenomena. The large-scale anatomy of each of three systems is considered separately.

Figure 2.2 shows several grasses that are structurally related toArundo donax

These specimens were found locally at The Ohio State University. The gross morphological similarities among these particular examples of grasses are quite evident-broad leaves, or plume-like inflorescence, for example, as well as their resemblances to the stem of theArundo donax plant (figure 2.3).

The Root System

The root system of Arundo donax is composed of two major structural elements: 1) the rhizome-a large, fleshy, cylindrical structure, which lies in a predominantly horizontal direction (i.e., perpendicular to the stem direction) just beneath the soil, and from which the stem sprouts, and 2) the well-developed, thin, creeping adventitious roots which extend from the rhizome and reach deep into the soil, allowing the plant to maintain hydration in times of drought by extending down into the water table. The term "adventitious" refers to the development of a structure-roots, in this case-in a location on the plant not normally expected. A photograph of the root system, showing a rhizome, a 53

Figure 2.2-A Sample of Grasses at The Ohio State University 54

Figure 2.2 (cent.)

i 55

Figure 2.2 (cont.) 56

Figure 2.3-The Arundo donax Stem 57

Figure 2.3 (cont.) 58

tangled adventitious root structure, and even an emergent stem Is shown In figure 2.4.

It Is a common misconception that Arundo donax grows from seeds. The reproduction of the plant actually takes place by means of the rhlzome.2 Once planted (two to three meters apart to start within a typical mass-productlon setting, at a depth of about ten centimeters), the rhizome quickly extends adventitious roots and begins to grow an emergent shoot (figure 2.4) that rapidly rises above ground to become a stem. Once the stem begins to grow, the cane Is moved farther ap a rt.3

Typically, new rhizomes are acquired by dividing (cutting up) old rhizomes and replanting them. Occasionally, growers plant large stems sideways underground to acquire rhizomateous matter.4 This last method leads to the development of young plants at the nodes of the stems, which are then transplanted to provide new root systems.

The Stem Svstem

The stem of the giant reed plant Is a long (essentially hollow) tube. The tube, in toto, is usually called a culm. The culm may grow as tall as two to eight meters

(although, about three-five meters Is the average size range of plants used In the production of clarinet or saxophone reeds, judging from commercial photographs obtainable from the reed manufacturer, Vandoren).s

A year old culm of Arundo donax with leaves still attached, 2.77 meters tall, and still In fair condition after harvesting and transport, Is shown In figure 2.5 59

Figure 2.4-The Rhizome. Symbol Key: R-Rhizome, ES-Emergent Shoots 60

Figure 2.4 (cont.) 61 with the author (who Is approximately seventy inches or 1.778 meters tall) standing next to it for comparison. This particular specimen from The Ohio

State University Arboretum has an inner diameter of 1.65 cm. and an outer diameter of 2.35 cm. at its base, yielding a thickness of .7 cm. (figure 2.6).

These sizes are typical for a slightly immature plant, as in this case.

The stem is divided up into smaller cylindrical subunits, called Internodes

(figure 2.7), which are joined together to make up the total length of the stem.

The points at which the internodes are joined, callednodes, are typically regions of a thick-walled, fibrous nature, resembling elbows. There is an alternating N-l-N-l-N-l-N... (N = node, 1= internode) arrangement of the stem parts, as shown in figure 2.7. Typically, according to Perdue, the internode lengths may be between approximately 12 to 30 centimeters.e Measurements on the present specimen indicate the nodes are about one-tenth of those dimensions (1.5 cm. to perhaps 4 cm.).

In addition, the nodal regions are the sites from which leaves sprout. The sites are called leaf gaps, and are holes in the nodes through which the leaf shoots pass (figure 2.8).

In general, the taller the plant, the wider the outer diameter of the stem, which may range from less than 1 cm to 4 cm or more. Since the range of lengths for stems is from two to eight meters, and the range of diameters is from about one to four centimeters, a convenient, "rule-of-thumb" is that:

= Stem Length/Stem Diameter = 200

(2.1) 62

Figure 2.5-The Complete Arundo donax Culm 63

Figure 2.6-Arundo ofonaxCross-section (Stele) 64

Figure 2.7-Arundo donax Intemodes and Nodes. Symbol Key: I - Intemode, N - Node, T -Tannin Deposit 65

Figure 2.8-Nodal Sites for Leaf Generation. Symbol Key: L-Leaf Gap Figure 2.8 (cont.) 66

Intemode

Lee/ Gap

Node ' ■"

Emergent Leaf Intemode 67

Generally, wide diameters are used for larger reed (bass clarinet, etc.). Plants

having diameters in the range of two to three centimeters are usually used for standard Bb clarinet reeds (figure 2.9).

The large size/diameter ratio of thegiant reed plant is explained in terms of the differences in growing patterns between plants classified as trees and grasses. The vertical growth of a plant is termed primary growth, while growth which is horizontal, girthward or outward, which usually occurs in tandem with or after primary growth is finished, is called secondary growth.

True trees typically show both primary growth and secondary growth.

Arundo donax, as well as grasses in general, do not have secondary growth

(there is no seasonal horizontal expansion of the plant), but rather, as the plant grows taller, new parallel vertical layers are added to preexisting layers. This process is called primary thickening, since the thickening of the stem uses only the mechanism of upward (primary) growth. These plants are said to have a primary thickening meristem (a meristem is a site of continuous growth).

The cellular distinctions which allow for secondary growth in trees (via the formation of secondary cell walls), but not in grasses (which have no such inner cell walls ) will be discussed later.

Arundo donax Is a very fast growing stem. A typical growth rate is .3 to .7 mm./week for several months, with the stem being essentially of a mature

(inner) diameter as it grows.7 The primary growth begins once the maximum height is reached. 68

Figure 2.9-The Processed Stem. 69

Figure 2.9 (cont.). Symbol Key: C-Ciarinet Reed (French Cut), B-Bass Clarinet Reed (French Cut) 70

iiliito»

Figure 2.9 (cont.). The Three Tissue System. Symbol Key: T-Tannin, P- Parenchyma, VB-Vascular Tissue, E-Epidermis 71

The Leaf Svstem

The leaves of Arundo donax are two ranked, which means that the leaves grow in simple alternation from side to side along a straight line up the stem

(figure 2.10). The leaves are typically 5 to Scentimeters across at the base, and taper to a tip.

The leaf sheaths typically wrap around the stem, covering it (figure 2.11).

The sheaths may stay attached to the stem long after the leaves have fallen off.

Many times the leave sheath will rot (black spots in figure 2.11), according to

Perdue, discoloring the stem undemeath it (figure 2.7). This causes the brown mottling often seen on the epidermal portion of clarinet reeds, in reality, this discoloration to be due to the presence of tannin, a substance found on the exterior of many plants (see the discussion of the epidermis, below), and Is not the result of a pathological process.3

The inflorescence, or flowering portion of the reed plant, occurs at the tip of the stem, and resembles a plume (figure 2.12). This plume is, in reality, composed of about three to five plum-like bundles which join together to form the larger plume (figure 2.13). Each individual plume bundle is called a panicle.

Each panicle is composed of a grouping of small, three to five millimeter long, V-shaped flowers, resembling wheat stalks, calledspiklets (figure 2.14).

Each spiklet on a panicle is composed of two very small outer leaf-like structures, called glumes (figure 2.15), which enclose another pair of leaves. 72

Figure 2.10-Arundo donax Leaf Pattern (Entire Stem and Close-up) 73

Figure 2.10 (cont.) 74

ti^jr

Figure 2.11-The Leaf Sheath: Black Spots are Areas of Decay on the Leaf Sheath 75

Figure 2.12-Panicles. Symbol Key: P-Panicles 76

Figure 2.13-Panicle and Base Structure Showing Subdivisions of Aggregate 77

Figure 2.13 (cont.)

1

I 78

Figure 2.14-Spiklets 79

Figure 2.15-Splklet (S), Glume (G), and Floret (F) 80

Figure 2.16-Florets 81

Figure 2.17-Perianth. Symbol Key: P-Perlanth, S-Spikiet 82 called florets (figure 2.16). Other more detailed parts of the Inflorescence, such as the perianth (figure 2.17), which are undeveloped In this plant, need not concern us.

Ultrastructure

In the remainder of this document, we shall only be concerned with the stem portion of the reed plant.s This Is not to Imply that the other gross structural elements are not Important to the finished clarinet reed, but a separate study of these elements Is warranted only after the stem characteristics which yield preferable clarinet reeds have been studied and Isolated.

Anatomically, the ultrastructure of the reed stem Is composed of threetissue systems: 1) the dermal, 2) the ground (or supportive), and 3) the vascular

(figure 2 .9 ).io These three tissue types corresponding somewhat loosely to skin, muscle/bone and vein tissue In animals, respectively. Using this classification scheme It Is possible to view the reed stem as an engineering material, and In this context It a three-component material with two of the components Infused and Interlarded (the ground tissue and vascular tissue) while the third component (epidermis) Is layered on top (see figure 2.18).

Dermal Tissue System

The dermal tissue Is contained In theepidermis, the hard outer covering of the reed stem, which Is Initially a greenish color, but Is usually a golden yellow in the mature plant (figure 2.7). The epidermis Is schematically shown In 83

^ p i dermis

************* *************** **************** Vascular Tissue Parenchyma

Figure 2.18-Three-Component Structure of Arundo donaxStem Material 84

figure 2.19. The "shininess" of the surface of the back vamp of the reeds in figure 2.9 is due to a thin layer of wax, which acts as a moisture and pathogen barrier in the living plant. The hardness of the epidermis is due to the unusually high silica content of the outer cells. This silication is shown in the electron dispersive x-ray (EDS) spectrum in figure 2.20, which gives a measurement of the elements present in a sample as it is scanned with a variable energy x-ray or electron beam (this technique is explained in detail in chapter three).

The first spectrum is of the epidermis. The only significant constituent is silicon

(Si). Palladium (Pd) and gold (Au) are coatings placed on the reed sample to allow higher energy electrons to be used, and hence, higher resolution, for the scanning electron microscope. By contrast, the interior cells (figure 2.21) show a much more diverse elemental analysis typical of biochemical processes, where

Chlorine (01), Sodium (Na), Magnesium (Mg), and Calcium (Ca) are most often used in energy transfer mechanisms within the plant.

The discoloration of the reed surface, mentioned earlier, is due to the increased concentration of a tannin, one compound in a whole class of dark- colored (usually brown) fatty acids stored as an energy reserve material for plant metabolism in certain types of cells called tannin cells. The structure of a representative tannin is shown in figure 2.22.

The rotting of the leaves could not be the cause of the brown flecks seen on the epidermis of clarinet reeds, as Perdue suspects, because of the presence of tannin beneath the epidermis, which is seen as thin dark brown lines in 85

Silica Bodies Epiculicuiar Wax Cutin / \

•Wax Pectic - î Material Height

Ceiiuiositic Material (Cell Walls) (Longitudinal - Length Cross-section)

Figure 2.19-Epidermis Schematic 86

C-SU SCfï'MRKl ELECTRON MICROSCCP,' U% i t i 094 m -9 i 15:10 Cursor: 0.200keV = 3

_U_.

0.080 • ."'-L W S ?. 15384 . 3.120 100 • E?IC€P«I£

Figure 2.20-Electron Dispersive X-ray Spectrum of Arundo donaxStem Epidermis. 87

OSLl SCfiM'JING ELECTRON MIO%5COPY LAB TUE 26-JRN-40 05:18 Cursor: 0 000k.eV : 0

11 i

0.000 VF5 : 4096 5.120 m

Figure 2.21-Electron Dispersive X-ray Spectrum oi Arundo donaxStem Inner Tissue. 88

OH OH

Figure 2.22- A Tannin [Catechin] Structure 89 figure 2.9. Since the epidermis is virtually impenetrable, no tannin, if formed by leaf decay, could pass through it into the interior tissue. Neither it is likely that tannin would adhere to the wax cuticle of the epidermis. The tannin discoloration is an indigenous part of the plant cell metabolic process, coming from within the plant as it ages, it exists within the epidermis, not on top of it. It is often thought by clarinetists that better cane has a higher amount of the brownish tannin flecks on the epidermis. This conclusion is debatable

The rigidity of the epidermis is due in part to the silica infusion, and partially to the presence of thin, thick-walled subepldermal sclerenchyma cells found just under the surface of the material (figure 2 . 2 3 ) . Dimensionally, these rod-like cells are 3 to 9 microns thick(1 micron = 1 micrometer = .001 mm.), compared to 1 to 3 microns for soft tissue cells (such asparenchyma, to be discussed later). The inner diameter in a random sampling varies from 4-10 microns (c.f. 50-70 microns for soft tissue cells), while the exterior diameter varies from 7-22 microns (c.f. 50-75 microns for soft tissue cells). Without access to whole stems, it is impossible to estimate the length of these cells, but limited observations from indigenously grown plants indicates that they extend considerable lengths along the stem.

These cells provide enough structural support to allow the giant reed plant to grow to unusually large heights for a grass stem. Sclerenchyma cells began to develop in their embryonic state as water conducting tubes (thexylem, to be discussed momentarily), but at some point in their development specialized into a load-bearing tube, developing thicker walls and smallerpit cells 90

Figure 2.23-Subepideral Sclerencyma Tissue. Symbol Key: E-Epidermis, 8- Sclerencyma 91

Figure 2.23 (cent.). Symbol Key: P-Pit 92

(figure 2.23) than a typical water conducting tube.

Veselac points out a structure in the stem of Arundo donax which she calls

"fiber bands" which seems to be in the same location directly under the

epidermis as the schlerenchyma cells mentioned above, surrounding the stem,

forming an inner band or ring (figure 2.24).12 She also says that the same fibers

surround the nutirient conducting vascular bundles of the plant. So as not to

confuse the reader with what may seem to be differing anatomical terminology

between her study and this one, a bit of explanation is In order.

Sclerenchyma is a generic term for the thick-walled supportive cells

(singularly) or tissue (collectively) of a plant stem. These cells may occur as

long slender tubes, typically called fibers, or as shorter tube-like units, called

sclereids (figure 2.25). Thus, Veselac is making the assumption that the sclerenchyma under the epidermis is essentially composed of fibers. We prefer a more generic term.

To be sure, other types of supportive tissue also exist within the plant, such as the thinner-walled parenchyma cells (figure 2.27), which are the major component of the inner tissue of Arundo donax. When adapted specifically for

load-bearing purposes, such as those parenchyma cells near the epidermis, they may develop cell walls thicker than is typical, and are then said to be

sclerifled.

The "fiber bands" under the epidermis to which Veselac refers are probably a

mixture of sclerifled parenchyma and sclerenchyma. We prefer the more

generic term, subepldermal sclerenchyma (or simply sclerenchyma), to 93

T ib e r Band"

Epidermis

Ground Tissue A: Stem

Xylem Fiber Band"

Metaxylem Pole

Phloem

B: Surrounding Vascular Bundle

-Igure 2.24-Veselac’s Two Locations for Fiber Bands (c.f Figure 2.23 and 2.26) 94

^yietn

^^ppottive)

Pits

'^"'‘<°^^'>fl,lenyKinil,l

2.2s.F,ben ® a w SclersUo 95

refer to them, since there are many types of fibers, such as phioem fibers, xylery fibers, etc., which may be scattered throughout the stem, and the overuse of this term could lead to confusion, especially when the same cells may be labeled by other, less ambiguous n a m e s . 13

Veselac alludes to a second structure (which she also and confusingly calls

"fiber bands") which surrounds the outside ofvascular bundles. These are probably sclerifled parenchyma cells, and not true fibers. It is typical to find paraenchyma cells surrounding vascular bundles in plant stems. To increase their protective ability, they may sometimes develop thicker, sclerifled wails, as do the parenchyma near the epidermis. To avoid confusion, we shall refer to the cells which surround vascular bundles by the generic name, bundle sheath, as they is often called, since the cells "sheath" the vascular bundle, as the name implies (figure 2.26).

Ground Tissue System

The ground or supporting tissue of the Arundo donax stem (c. f. muscle in animal tissue) is composed, essentially, of long chains of one cell type (figure

2.27). These parenchyma cells are typically tetrakaldecahedral in shape, having 14 sides partitioned into one heptagonal face, four hexagonal faces, five pentagonal faces and four square faces (figure 2.28). For a long time it was thought that this shape represents what is known in topology as aminimal surface-one which keeps the volumeas large as possible with the minimum surface area; however, it is known that there are some slight variations 96

m

Figure 2.26-Bundle Sheath. Symbol Key: X-Xylem, Ph-Phloem, B-Bundle Sheath (LIgnlfled), S-Sclerlfled Parenchyma 97

Figure 2.26 (cont.) 98

:..^k.*»>«iw- ;4#^ Mk.## f - . < # Tfcf • * f- •■■•■': ; t

-.Hir'KS^irar'iS?

K % : r % V.Va=cu,ar 99

Figure 2.27 (cont.) 100

Figure 2.27 (cont.). Transverse Section.

^ 101

Figure 2.27 (cont.). Transverse Section. Symbol Key; P-Parenchyma Cell, L- Lignin, W-Cell Wall 102

Figure 2.28-Parenchyma Topology 103

statistically from this idea condition in plant stems. 1 <

Dimensionally, approximating the parenchyma cell as a parallelpiped induces only a small error. The length in a random sample varied from 85.44 to

126.75 microns (pm), with a variation in the "width" of 50 to70 pm. These more accurate measurements were conducted real-time on a specially adapted JOEL scanning electron microscope (SEM) with a computer-controlled digital micrometer, while the other values were obtained from ruler measurements of

SEM micrographs. The cell walls are typically 1 to 3 pm thick (2.6 pm, for a typical example, measured by SEM).

As a tissue, parenchyma occurs in long stacked sheets which do not maintain exact alignment of the cell wall orientations or lengths from sheet to sheet

(figure 2.29). In many ways this stacked sheet arrangement gives the ground tissue, which is very simple in structure (almost a girder network), a unit crystal morphology (figure 2.30). To a first approximation the ground tissue may be modeled as an extended imperfect lattice, such as one might see in a crystal.

This crystalline-like structuring (normal wood is much more amorphous in the internal structure of the ground tissue), in some ways makesArundo donax easier to model as a vibrating entity than wood.

The unit cells (parenchyma cells) of this quasi-crystalline ground tissue are either hollow or filled with a gummy protective substance called lignin (figure

2.27), and because of the simple, nearly regular ordering pattern of these hollow/filled unit cells, we suspect that energy is much more efficiently dissipated throughout the ground tissue ofArundo donax than in more 104

Figure 2.29-Parenchyma “Sheet” 105

i l l H

Figure 2.30-Parenchyma Tissue Unit Crystal Approximation 106

amorphously structured wood ground tissue. This efficiently is likely one

reason for the unusually high damping coefficient of Arundo donax (more than ten times that of wood)--the energy resulting from an impulse applied to the

ground tissue is siphoned away from the initial site of impulse at a very high

rate. This has enormous implications for musical tone generation as we shall see.

As an aside, lignin is a highly complex molecule (there are many varieties of

lignins, as discussed in the next chapter), formed from dead cells, which acts as an all-purpose filler substance in stems, gives water protection to vulnerable cells, acts as a pathogen barrier, a parasite growth inhibitor, etc. It is

hydrophobic (water repelling), and tends to shrink and stabilize the cell wall thickness of the parenchyma cells because less water is able to penetrate them.15 The older the section of the plant, the more likely lignin is to be found as the cells become more brittle and more in need of reinforcing.

Because of its large molecular shape and extended polymeric formulation, lignin is a very viscous type of material. The cell walls of the parenchyma cells, by contrast, are essentially elastic, giving the reed a "springiness". The vascular tissue, to be discussed next, is somewhere in-between a viscous and an elastic material (it is usually classified as a viscous material). It is customary to refer to a material having a combination of resistive and elastic components as a viscoelastic material. We shall discuss these mechanical terms in the next two chapters. 107

Vascular Tissue System

Interspersed In the matrix of ground tissue (figure 2.27) are both the food and water conducting (vascular) cells of the plant, which are long tube-llke structures. These two distinct types of vascular cells or tissue do not usually occur In Isolation, but typically together in distinct collections of vascular tissue

(called, vascular bundles) which are Incorporated into the ground tissue as shown In figures 2.26 and 2.31. Vascular bundles In the Interior of the stem average from 250 to 300 p,m In diameter, while those near the epidermis average from 50 to 100 pm.

The arrangement of vascular tissue within the ground tissue of a stem varies from plant type to plant type, and one often refers to the collection of the supportive and vascular tissue In a stem as astele, from the German word meaning stem. There are many different types of stele arrangements, and several of these arrangements are shown In figure 2.32. InArundo donax, the vascular bundles are scatter randomly throughout the stem, giving this plant the designation of an atactostele.

As one progresses from the Interior tissue towards the epidermis, the number of vascular bundles Increase according to some power law (figure 2.33):

Nvb = No (R/Rinner )^ (2.2)

where No is the number of vascular bundles along the Inner radius encompassing the annulus, Rjnner + Rvascuiar Bundle- Rinner Is the inner radius, R Is 108

Figure 2.31-Vascular Bundle. Symbol Key: PX-Protoxylem, MX-Metaxylem, PP- Protophloem, MP-Metaphloem, PO-Metaxylem Pole, P-Parenchyma, B-Bundle Sheath 109

Figure 2.31 (cent.) 110

Black = Vascular Tissue Whit e = Gro un d Tiss u e AtactostelefA'undo donax)

Subcaiagories Cross-section Protest el 8 (No Ground Tissue): \/^Halptostele Siphonostele: ( Actinostele Ectophloic 'x V . Plectostele O Miphiphlolc J

Subcatagories Subcatagories

\ Siphonostele: r Dictyostele Eustele V (Amphlphloic)

3-d1mens1ona1 example: Siphonostele

O

Figure 2.32-Various Steles 111

outer

Inner

+ R inner YB

There is an increase in the number of vascular bundles as one moves from the inner radius to the outer radius which is governed by a power lav of theform: Nvb (vb/mm) inner

R inner '^oiier

Figure 2.33-An Explanation of the Vascular Bundle Power Law 112 the radius of interest (Router ^ R ^ Rinner). Router is the outer radius, and P is an

exponential (power law) scaling factor. From emeprical measurements, No is typically on the order of 4 bundles/cm. and P is typically on the order of 1.96.

The xyiem (figure 2.34) is the name for water conducting tubes (composed of smaller tube-like subunits calledtrachiery elements joined end to end) which occurs in the vascular bundle, and the phloem (figure 2.35) is the food conducting tube (composed of smaller sieve tube elements, similarly linked together).

Xylem have diameters of 50 to 100 pm in the inner portion of the stem, and 20 to 40 pm near the epidermis. Phloem has very irregular cell construction, which makes diameter measurements difficult, however, an approximate range of 5 to

20 pm may be given. The length of each sieve tube ranges from 150 to 250 pm.

The particular arrangement of the vascular tissue (xylem and phloem) in a vascular bundle is one type of distinguishing characteristic of a plant. The type iorArundo donax is known as acollateral arrangement (figure 2.31 and figure

2.36). The phloem region lies above the xylem region separated by an imaginary line (hence the designation collateral) running between them. The phloem region is abaxial relative to the xylem. The term abaxial signifies that the phloem is closer to the epidermis or surface than the xylem region. The cells sandwiched between the xylem and phloem are the parenchyma type cells mentioned above.

In discussing the development of a typical vascular bundle, the imaginary collateral line is a useful structural device. It corresponds to a region of 113

Figure 2.34-Xylem Cells. Symbol Key: X-Xylem Cell (Tracheid), P-Pit, W-Cell Wall 114

Figure 2.34 (cent.). Transverse Section. 115

i

Figure 2.35-Phloem Cells. Symbol Key: P-Phloem Cells, C-Companlon Cells 116

Figure 2.35 (cent.). Longitudinal Section. 117

Bundle Sheath X Old (EaMy)-'Proto- Protophloem

Metaphloem

Ph oem- Imaginary New="Meta- Collateral ■ ■ "L in e g Xyem

Metaxylem Old (Early) ="Proto Metaxylem Pole

Protoxylem

In a typical vascular bundle, the food c on ductin g tiss u e (phloem) lies above an imaginary canter(collateral) line, when seen in this orientation. Thewaterconducting tissue (xylem) lies belowtheiine. Earlier tissue is pushed away from the coliateral (meristematic) line, so that the earliest (proto) tissueisnearthe outside of the bundle,andthe younger(meta) tissue is nearerthe center.

Figure 2.36-An Explanation of the Collateral Arrangement of Cells In the Vascular Bundles of Arundo donax 118

generation where xylem is forming and pushed away below the line, and phloem is formed and pushed away above of the line. Starting from the imaginary dividing line, the first phloem to develop (protophloem or "first" phloem) is pushed away from the center line as later(metaphloem or "middle" phloem) cells develop. Thus, the vascular bundle elements are older the farther away they are from the collateral line. The designations proto- and meta- simply refer to whether the phloem developed at the beginning of the life of the vascular bundle or later on, respectively. The same situation occurs for the xyiem, leading to the designations, protoxylem and metaxylem. The result is show schematically in figure 2.36.

As the growth stresses of the developing vascular bundle push the early cells farther and farther away from the collateral line, they begin to impact against the outer circular region of thick lignified parenchyma cells(bundle sheath) which encases the vascular bundle (figure 2.36). The forces are so great that often the earliest phloem and xylem are crushed up against the two half-walls of the vascular bundle.

Typically, the xylem in the vascular bundles are surrounded by smaller, immature, and somewhat thicker types of tissue which resemble xylem in appearance (figures 2.37 and 2.25). These tissues surrounding the xylem are called xylem fibers (or simply fibers) and provide support and protection for the xylem. The phloem tubes occasionally have such support as well, but it is more characteristic to find small phloem-likecompanion cells coupled to the phloem cells to form a phloem/companion cell pair (figure 2.38). 119

Figure 2.37-Xylem Fiber 120

r ,

n KV X 2 I

Figure 2.38-Companlon Cells 121

Standing often at one apex of a triangle formed by the two metaxylae in a

vascular bundle is the smaller metaxylem pole ( typically, 65% the size of

the xylem size in the vascular bundle-figure 2.31). This lacuna is generally

nonfunctioning.

The near-epidermal vascular bundles tend to be smaller (typically 1/3 to 1/5

as large) than those lying in the interior of the stem and have some elongation

of the abaxial side of the bundles, giving them an egg-shaped appearance

(figure 2.31).

It is an interesting fact that xylem tissue actually kills itself shortly after

reaching maturity so as to be useful in water conduction (only the shell is

important). This often precipitates the formation of lignin around it. Phloem cells engage in energy transfer and storage within the plant, and must remain alive, so there exists the odd case of the functioning xylem being dead and the functioning phloem being alive.

From the standpoints of physical stress-strain properties, water retention, and its eventual breakdown as a viable vibrational medium, it is important to mention two other anatomical elements of the vascular tissue, specifically elements of the xylem tubes. The first are small (1 p,m by 2 pm) ellipsoidal perforations, called pits (figure 2.39), that iie along the outside of the xyiem tubes and allow materials such as proteins, ions, etc., to be pumped into (or out of) the xylem tube in order to be transported upstream or downstream via the streaming fluids inside of the tube. These pits align themselves in a helical pattern (called helical thickening). A typical coil of the helix is 6.3 pm 122

% I I

Figure 2.39-Pits 123

Figure 2.40-Helical Thickening 124 thick. A scanning electron micrograph of an exposed helical section of a xylem tube if shown in figure 2.40. The helical thickening strengthens the xylem cell, much as a spring mounted inside of a tube of cloth might.

Cell walls

Each cell of the reed, when alive, is filled with a complex fluid ( called the plasmamembrane) that provides a hydrostatic pressure inside the membrane and pushes outward against the rigid cell wall that contains it. This turgor pressure is very much like the function of air pressure inside of an inner tube-providing support and shock resistance (figure 2.41).

This turgor pressure requires the development of thick cellulositic cell walls capable of resisting the immense hydrostatic pressures of the plasmamembranes. In the case ofArundo donax, as in most grass stems which have only primary growth, these cell walls may be separated into two parts, having approximately a 2:1 ratio in thickness: the part common between and connecting any two adjacent cells-the middle lammela, ( .875 pm) and the part which beiongs exclusively to the individual wall of a cell, the primary wall

( 1.75 pm), as shown in figure 2.42. The chemistry changes radically from the middle lammela to the primary wall-a point we shall return to in the next chaper.

Arundo donax is a grass stem, and as such, lacks some of the other major components of typical wood cell walls. Chief among the missing elements is a secondary wall which forms interior to the primary wall, and is responsible for 125

Cell Wall

"VSÿ "Deflated" plasmamembrane (small pressure y against cell wails)

Lov

Turgor Press I High

Inflated plasmamembrane Ê (large pressure against cell wails)

Figure 2.41-An Explanation of Turgor Pressure .126

0002 20K0- X

Figure 2.42-The Cell Wall. Symbol List: P-Primary Wall, M-Middle Lamella, L- Lacuna 127

the horizontal thickening of most trees. BecauseArundo donax lacks this secondary cell wall structure, it characteristically grows very tall but not very wide.

The fact that Arundo donax grows slightly thicker horizontally than most grasses is because of a horizontal growing region(primary thickening meristem) in the outermost portion of the internodes which resembles in some respects the so-called secondary (or horizontal as opposed to primary or horizontal) thickening found in normal woods. Although most xylem cells with secondary cell walls have pits, some prior researchers ofArundo donax have mistaken the existence of pits inArundo donax xylem as evidence of secondary cell walls. In fact there is more than one kind of pit and some of the simpler types of pits (so-called simple pits, which are the type found in Arundo donax) are quite often found in cells having only primary walls.

At an even smaller level, these cell walls are composed of long, layered chains of cellulose called fibrils with an alteration of crystallite regions and relatively amorphous regions within a fibril (figure 2 .4 3 ) .16 These fibrils form a network of cellulose into which a number of other chemical substances are interlarded, as we shall discuss in the next chapter.

Before we turn to a discussion of the basic chemical components of the clarinet reed, all of which exist solely in the plant cell walls since the plasmamembranes are dried out during processing, we summarize this section of anatomy by showing the relative sizes of the different anatomical components mentioned in this section (chart one). 128

Middle Lamella Primary Wall

lary

FIbrila

Magnified

Elementary Fibril

Magnified VfW {«VVLV.------/C e llu lo s e Polymer m > Lattice

Amorphous Cellulose

Figure 2.43-The Fibril Concept 129

Table 2.1-Comparative Size of Arundo donax Cells and Their Components

Element Size Comparative Size illustration

Stem 2-8 M 20,000 X larger than xyletn

Leaves 5 -8 cm 40 xsmalierthan the stem

Parenchyma 80-130(ji lO x less volume than length xyiem cells 50-70|a radius

150-250 n Xylem Approximately the length thickness of human hair 20 -100 [A radius

150-250 n Phloem 1/4 the thickness of length human hair 5-20n radius

Sclerenchyma 7-22JX 1/4 the radius of xyiem, radius yet 3 xthickerthan xyiem

Fibril < .Oln > 200,000 xsmalierthan stem 130

Endnotes

1. Purdue, Arundo donax...page 368.

2. ibid., page 381; the author gratefully acknowledges the help of Richard Pearsonin obtaining this specimen.

3. ibid., page 381.

4. ibid., page 380.

5. Vandoren Reed Co., The Making of Greatness... page

6. Purdue, Arundo donax...page 369.

7. ibid., page 369.

8. It seems strange that Purdue, an experience botanist, would make this assertion, when a partial understanding of plant flavinoids (tannin, for example) already in existed before his article was written.

9. For a thorough discussion of all of the aspects of plant ultrastructure, the author can do no better than to recommend, Esau, The Anatomy of Seed Plants...

10. Fredrick Sack, Lecture Notes, Plant Anatomy, 656...; tissue is defined here as a large collection of cells which function as a unit.

11. The author is grateful to Fredrick Sack for a discussion about proper nomenclature for these cells, as there is some confusion in the reed literature.

12. Veselac, A Companson...page 34

13. This terminological ambiguity has been the source of numerous arguments between researchers in the reed field.

14. See Esau, The Anatomy of Seed Plants...tor a discussion of parenchyma morphology. 131

15. Fengel, Dietrich, and Wegener, Gerd. Wood: chemistry, ultrastructure, reactions (New York: Waiter de Gruyter, 1984): chapter 6.

16. Wegener and Fengel,Wood, page 97. Chapter III

Clarinet Reed Chemistry

The Chemistry of Arundo donax

As mentioned in chapter one, a clarinet reed is nothing more than the shell or

remnant of cell walls of the giant reed plant after the plasmamembranes have

been removed due to cutting, curing, and shaping. The manufacturing process

leaves the basic cell wall “building blocks” of the clarinet reed intact in a

relatively fixed form so that the entire chemistry of clarinet reed material is contained within them (figure 2.42). We shall be concerned primarily in this chapter with the molecular composition within the cell wall lattice structure, as this forms the basis for many of the engineering properties of the clarinet reed material.

Elemental Analysis

Electron dispersive x-ray analysis may be used to study the elemental composition of a material in a qualitative manner. This method takes its idea from the studies of the early twentieth-century physicist, H. G. J. Moseley, who found (1913) that when an element is bombarded by energetic electrons inner shell electrons may be dislodged from the atom, radiating x-rays of a frequency

132 133 particular to that element according to the formula:

3cR

(3.1) where c is the speed of light, R is the Rydberg constant, and Z is the atomic number of the element."!

If such high energy electrons (on the order of 20,000 electron volts) are directed at the surface of material composed of a single unknown element the surface will radiate x-rays whose frequency uniquely identifies the element.

Typically, one starts with some low energy of electrons and slowly increases the energy until a maximum transmitted radiation is found. One may then identify the element by its characteristic frequency, since these have been precisely determined (in fact, Moseley’s method was used to confirm the discovery of new elements after about 1930). In practice, since energy and frequency are related by the Plank relationship (E = hF , where, h = Plank’s constant), typically most

EDS-capable machines vary the energy of the incident electron beam, and then calculates the frequency from the known energy of the beam. For a composite material, one finds a series of peaks in the x-ray spectrum, each peak being characteristic for one of the elements in the sample.

Figure (2.20) shows a typical EDS analysis from the epidermis of a clarinet reed. The high silica (SiOa) content is reflected by the very large peak in silicon

(Si) present in the spectrum. The palladium (Pd) and gold (Au) peaks are artifacts of their use as coating agents to allow the sample to withstand the high 134 energy electron bombardment without deterioration.

By contrast, the interior cells (figure 2.21) show a much more diverse elemental analysis, typical of organic and bio-organic compounds: carbon, hydrogen, and oxygen (the basic building blocks of organic chemistry and biochemistry), as well as chlorine (Cl), sodium (Na), magnesium (Mg), and calcium (Ca), which are the most often used elements in energy transfer mechanisms within a plant.

This technique is not generally quantitative (although some systems will allow for semi-quantitative analysis), but does indicate that the interior of the plant has most of the general elemental characteristics of a (once) living biological material, whereas the epidermis is more closely related to a glass-like inorganic material in composition.

Despite the fact that there are hundreds or thousands of compounds which may be synthesized by the living reed plant from only the elements mentioned above, to a large extent there are only three major organic chemical compounds found in the reed cell walls after the plasmamebranes are removed.

Two of these compounds are carbohydrates, and one is a phenolic derivative.

Each of these three compounds will be discussed in turn.

Cellulose

The first and most important of the compounds contained in the cell walls of

Arundo donax is cellulose, which, according to Fengel and Wegener, comprises 40%-50% of the total plant material for bamboo. This is 135

approximately the same percentage for Arundo donax, according to Perdue.s

Celluiose is the name given to a six-member ring structure of carbon and oxygen (cellulose therefore belongs to the branched hexose or pyranose family

of carbohydrates). Its technical name is p-d-glucopyranose (figure 3.1):

OH

H OH

Figure 3.1-|3-d-glucopyranose

Cellulose is rarely found as a single carbon ring, but rather, is usually found in a paired arrangement with another p-d-glucopyranose ring connected by an oxygen bridge (as shown in figure 3.2):

HH OH OH

OH

Figure 3.2-Celiobiose

This oxygen bridge (the formation of oxygen bridges is one thing which renders carbohydrate chemistry unique), incidentally, is what renders cellulose 136

indigestible to mammals. Glucose, the most common simple sugar, differs from

cellulose only in the position of this oxygen bride.

Numbering the carbon atoms on the two rings with the aldehyde containing

carbon (i.e., the carbon to the left of the oxygen as seen from outside the ring)

always numbered one, this linkage is between the 1 carbon of ring 1 and the 4

carbon of ring 2.3 The designation 1->4 is often used to denote where the

carbon “trusses” of the bridge are in the joined rings (figure 3.3):

CH2OH H OH

OH H OH H

H OH CH-OH

Figure 3.3-1—>4 (3-d-glucopyranose Bridge Structure

Thus, this is a 1-->4 p-d-glucopyranose bridge structure. This two molecule version of cellulose has a special name:cellobiose.

These cellobiose units link together into long continuous chains (the oxygen bridge alternates in the up and down position). If cellulose is called a monomer (literally, “mono = one, mer” = unit, in Latin), and cellobiose is a dimer (i.e., two unit), then extended chains of cellulose molecules are known as polymers: 137

H Q1 H m M « H «

Figure 3.4-Celluiose Polymer Structure

The most common polymeric form of cellulose in woods is cellulose I (there are four major forms of cellulose, and several subcategories). Large-scale order is generated not only by the stringing together of cellulose molecules in a line (forming a linear polymer), but also because some of the hydrogen atoms on one ring of a cellulose molecule can weakly bond across space to other hydrogen atoms on another ring (so-called, hydrogen bonding). This hydrogen bonding (there is bonding both within a cellobiose molecule.

Intramolecular hydrogen bonding, and across space between other cellobiose molecules, intermolecular hydrogen bonding) allows bonding between the cellulose molecules in a three-dimensional sense, which permits cellulose macromolecules to conserve energy and space by arranging themselves along the sides of a cubic-like arrangement, called monoclinic a unit cell (figure 3.5).6

In the monoclinic unit cell, the three vertices, a, b, c are of different lengths, and the three angles between them, a, p, y are related by a = (hence the term, monoclinic, which means “single angle", referring to the single nonequal angle). Each unit cell structure is a single unit in the three-dimensional lattice of the extended hydrogen bonded polymer chains (figure 2.43). These extended 138

Hydrogen Bond nonoclinic Uni t Arrangement of £*JJ______Celiufose Atom* 002 020r 200 ' X /

...... %

Miller Indieies for Cellulose Unit Crystal Lattice

Figure 3.5-Cellulose Crystal Structure 139

lattice chains grow into a macromolecular structure which resembles thin fibers

-- elementary fibrils. These elementary fibrils, in turn, form the fibrils

mentioned in chapter two, which then fuse together to form the ceil walls of the

plant.

There are different planes which pass through the monoclinic cell, and they

are identified by a set of indices called Miller indices. The origin of the

indices is not hard to see in figure 3.5. A point (usually a corner) on the unit cell

is assigned as an arbitrary zero point (0,0,0) for the cell in the x, y, and z

directions. The planes passing from the origin to the points on the cell are

noted. In the monoclinic unit celi, there is one plane, for instance which passes

through the cell exactly at the midpoint of the c vertex (c = 1/2). Since it does

not intersect the a orb axis, one is free to give these coordinates any vaiue, and

they are typically assigned to infinity (o o ). This particular plane of the monclinic

unit cell intersects the iattice at 0,0,0 and (o°,1/2,oo)v The Miller indices are

generated by taking the reciprocal of the point on the unit cell other than the origin which intersects the plane, in this case,( 1/o o , 2, 1 /o o ) = (0,2,0). This plane

is designated then the 020 plane, and it is the major plane in cellulose. Since the unit cell has symmetry, the planes 020, 002, and 200 are equivalent. The

major planes are shown in figure 3.5.

When an x-ray beam, which has a wavelength on the order of the spacing

between edges of the unit cell for a given substance impacts with the unit cell, depending upon the size and shape of the unit cell (there are fourteen different types of three-dimensional unit cells, the so-called Bravais Lattices), the 140 x-rays are scattered at characteristic angles off of the different planes of the unit cell.7

An x-ray diffractogram of a clarinet reed is shown in figure 3.6. This diffractogram (henceforth, this method of obtaining the x-ray diffraction information will be abbreviated as XRD) is made by placing a portion of the clarinet reed in a flat holder and mounting it on a rotating turret. A fixed x-ray beam, using a Ca ka radiation source is then shined at the holder as the holder is slowly rotated about 360 degrees. Depending upon how the molecules are oriented to the beam, scattering occurs when the x-ray beam impacts a natural plane of the unit cell and the intensity of the reflected x-ray beam is recorded by use of a scintillation counter as counts per unit time-in this case, ten seconds.

We shall represent this quantity as counts/10 sec. or cts, for short. The angle of rotation of the holder at that point tells at what angle that particular natural plane is found in the unit cell.

The large peak in the diffractogram occurs at 22.2 o. This is the principle scattering angle for the 020 plane of the cellulose unit cell. There are two smaller “halos”, the doublet at 16.0° and 17.6o and the singlet at 38.4o, which are the 101 (1 0 1 ) and the 040 planes respectively (the underscoring indicates an axis perpendicular to the other axis).

A knowledge of these angles may allow one to reconstruct the unit cell fairly well, however, a better method to determine the unit cell parameters of cellulose

( the angles and the lengths of the cell) is the single crystal method, in which a small single crystal of the cellulose material is exposed to an x-ray beam for 141

Figure 3.6-X-ray Detraction Spectrum of Clarinet Reed 142 extended periods at different rotation angles and the diffraction pattern is recorded on photographic film. Such detailed methods were not available for this study, however, and the bulk or powder x-ray (XRD) method mentioned above was used.

Degree of Crvstallization bv XRD

The cellulose matrix is only partially ordered (the monoclinic extended lattice). Some of the cellulose molecules are not bonded in this matrix, but exist as collections of individual molecules which tend to be amorphous in structure.

These amorphous regions exist in combination with the crystalline cellulose

(figure 2.43). The amount of ordered cellulose (0% to 100%) relative to the total amount of cellulose is reflected by a quantity known as thedegree of crystallization (or crystallinity index), which we shall write as, “ %C ”. In other words, the degree of crystallization measures the percentage of the cellulose which belongs to the ordered monoclinic lattice network.

There are two methods which are used to determine this quantity. The first method uses the intensity of the 020 line of the XRD spectrogram. The second method uses selected absorption intensities of bands in the infrared spectrum.

The infrared method tends to overestimate the %C, but is easier to obtain.

In the XRD method, the intensity of the 020 peak is taken to be a measure of the amount of crystallized material trapped in the extended monoclinic cell structure.8 Since some of the cellulose molecules are not bonded in this matrix, it is assumed that this amorphous cellulose forms a background noise in the 143 spectrum. The %C is calculated by subtracting the 020 intensity from the amorphous background and taking the ratio according to the formula:

% C = [ lo20"lam /lo20 ] X 100% (3.2)

A place in the spectrogram which is not influenced by the monoclinic lattice is used as a reference for the amorphous background intensity. Most often, this value is 19°.

A typicai calculation of the %C proceeds as foilows: the maximum intensity of the 020 is seen to be 992 counts/10 sec. at 22.053°. The amorphous background intensity is 290 at 19°. Subtracting the amorphous intensity from the 020 intensity and taking the ratio gives:

992 cts.-290 cts

%C = ______X 100% = 70.8% (3.3)

992 cts

This %C value is common for piant cellulose, which varies between 60% to

80%.

As an experiment, we selected three pristine reeds (reeds taken out of an unopened box of Vandoren V I2 reeds, strength 3 1/2), and four spent reeds

(supplied by students and faculty), and obtained spectrograms of each, centering on the theta values encompassing the 020 peak. The %C values given in table 3.1 were obtained by two slightly different methods, which use: 1) maximum intensity measurements, 2) intensity measurements at a fixed point in 144

R eed Label lnt.22.2G Max. Int. Amoro. Int. %C (22.20) %C(Max. IP 984 984 290 70.528 7 0 .5 2 8 2P 863 889 286 66.859 67.829 3P 959 963 2 9 3 69.447 6 9 .5 7 4 IS 1055 1077 3 5 4 66.445 67.130 2 8 979 1008 343 64.964 65.972 3 8 1058 1076 3 6 7 65.311 65.892 4 8 867 868 295 65.974 6 6 .0 1 3

1P 2P 3P 1S 2S 3S 4S

Reed Label

Table 3.1-%C Data for Pristine and Spent Reeds 145 the 020 peak (22.2°). The results are essentially the same (although pristine

reed two is shifted to a higher %C value). A superimposed plot of the spectrograms are presented in figure 3.7. The label p stands for “pristine” and the label s stands for, “spent.”

Several things are worthy of note. The %C values of the pristine reeds

(average, %C of 69.6%) are slightly higher than the %C for spent reeds

(average, %C of 66%). Also, the pristine reed maxima are shifted on average to lower theta values (22.05 vs. 22.2) relative to the spent reeds, indicating a larger lattice spacing, 4.027 Angstroms vs. 4.00 Angstroms The normally reported values for the lattice spacing of the 002 to 002 planes in the unit cell is approximately double these values (7.9 Angstroms is the value usually reported) because the measurements indicated here are for only half length, i.e., the distance of the 002 to 020 axis, instead of 002 to 002 axis (which is the complete span of the unit cell).

The lower %C values of spent reeds may be explained in a number of ways.

The first possible explanation is that the cellulose extended lattice of a spent reed breaks up into slightly smaller units than that found in a pristine reed due to vibrational stresses on the reed cell walls. This hypothesis would be in line with findings pointed out by Carleen Hutchin that violin wood tends to decrystallize over time due to vibrations of the violin body as the instrument is played, s

Another, more chemically-based explanation is decreased hydrogen bonding in the interlattice matrix. As we have seen, it is hydrogen bonding 146

R I

- Ri

n

(spuBsnoqj_) SpuODBS 01 Sfunoj

Figure 3.7-Superimposed 020 Peaks for Spent and Pristine Reeds 147

which gives cellulose its supermolecular lattice structure, and we shall shortly

present evidence that a class of compounds called hemlcelluloses are

removed from the reed cell wall over time due to saliva. Hemicelluloses tend to

hydrogen bond to the exterior of the cellulose lattice forming a parallel sheet to

it. This hemicellulose sheet allows water to get trapped between itself and the

cellulose lattice. The combined hydrogen bonding of both the hemicelluloses

and the water to the cellulose matrix increases the apparent crystallization of the cellulose matrix. When the hemicelluloses are removed, one of these two hydrogen bond sources (hemicelluloses) is removed (water continues to bond to the cellulose material), and the lattice shrinks, fractures or holes appear, lowering the order in the cellulose extended matrix.

A third possible explanation is that the results are somehow artifactual. The reed must be placed in a holder, and if the reed is not placed in precisely the same spot each time, the reed, being a wedge as opposed to a flat plate, will then allow varying degrees of material thickness to be exposed to the x-ray beam, which could influence the intensity of the beam which is diffracted, and possibly the resulting %C, however, since it is the ratio of crystalline to amorphous peak intensities which yield the %C, both peaks should be influenced by the positioning of the reed to approximately the same extent, cancelling the effect of thickness changes, leaving the ratio unchanged.

It seems, then, that this small, but measurable decrystallization effect is real.

The data presented, while limited, is provocative, and warrants a more detailed study with larger sample numbers. Nevertheless, spent clarinet reeds appear to 148 decrystallize slightly, perhaps due to vibrationally induced stresses or the loss of hemicellulose in the cell wall matrix. The lattice spacing also appears to be decrease slightly, which suggests a contraction of the cellulose matrix after

removal of the hemicellulose.

Decree of Crvstallization bv Infrared Spectroscopy

Another method by which %C may be calculated, introduced by Nelson and O’Connor, uses the ratio of certain band intensities which occur in the infrared spectrum of cellulose to determine the degree of crystallization of the cellulose lattice.10

Infrared spectroscopy is a method of determining molecular structure based on the interaction of infrared light with molecules. Figure 3.8 shows a typical infrared spectrum of a clarinet reed (we shall return to the matter of interpreting the peaks in this spectrum later). For the purposes of calculating %C, it is only necessary to focus on two transmission (%T) bands in the spectrum, those at

1372 cm-1 and 2900 cmwhich are due exclusively to the cellulose molecule of the reed, whereas most of the other bands are influenced by the other chemical constituents. The band at 1372 cm 1 corresponds to C-H bending in cellulose, while the band at 2900 cm-i corresponds to a relatively stable (i.e., unaffected by crystallization of the polymer lattice) C-H or G-H 2 stretching mode.

The method of determining the percent crystallinity of cellulose as proposed by O’Connor and Nelson is given below.11 As an example, the “troughs” in the infrared transmission spectrum in figure 3.8 at 1372 cm-i and 2900 cm 1 are oim pads peJBJ^ui peey )auuB|0 - 8 G © Jn S y

Transmission Intensity (100% = 1) o o o P - o N5 U1 -A Ol Kj U1 4 0 0 0

3 8 7 0

3 7 4 0

3 6 1 0

3 4 8 0

3 3 5 0

3 2 2 0

3 0 9 0

2 9 6 0

2 8 3 0

2 7 0 0

2 5 7 0

2 4 4 0

• 2 3 1 0

2 1 8 0

2 0 5 0

1 9 2 0

1 7 9 0

1 6 6 0

1 5 3 0

1 4 0 0

1 2 7 0 to

1 1 4 0

1010

8 8 0

7 5 0 150

isolated. These troughs or wells measure the percent of infrared light at that

particular frequency (in cm-i) which passes through (i.e., is transmitted by) the substance without interacting with the molecular structure of the material. When

light of a particular frequency hits a molecule, it can cause portions of the molecule to vibrate, much as hitting a row of connected springs at one particular point may cause "localized" vibrations of the spring system. The amount of infrared light transmitted is that amount of light which does not interact with the molecule at that particular frequency.

O’Connor and Nelson prefer to work with absorption peaks (%A), which is simply the amount of infrared light which is absorbed, rather than transmitted, by the sample at a given frequency (%A = 100% - %T). Because of this relationship, if a particular frequency shows a trough in the transmission spectrum, it occurs as a peak in the absorption spectrum (and vise versa). Both types of measurements are used in infrared spectroscopy. For the sake of clarity, we shall base the discussion on transmission spectra, trusting that the : relationship to O’Connor and Nelson is obvious.

Since no baseline exists to measure the well depths at 1372 cm i or

2900 cm- 1, one is drawn for each of the peaks (figure 3.8), using a line drawn from the peaks at 1315 cm-i to 1372 cm i as a level line from which to measure the 1372 cm 1 well depth. Similarly, a straight line is drawn across the top of the well at 2900 cm-i.

Next, the depth of each well is determined on some convenient scale at the two frequencies (we simply measure them with a ruler in mm). Finally, the ratio 151 of the depths of the two wells is calculated and the percent crystallization is determined by the simple formula:

%C = Di372cm-1/D2900cm-'l X100% (3.4) where, is the well depth measured from the baseline at 1372 cm-i, etc.

The resulting value is a rough qualitative measure of the degree of crystallization, since the method works on a scale much courser than x-ray radiation, and hence only bulk average molecular properties are used. Since infrared spectroscopy averages over molecular states, the values for %C derived are not always accurate, depending upon the configuration of the particular molecule being studied.

A typical calculation of the %C from infrared data by the O’Connor-Nelson method is:

Di 372/D2900 x100% = 17.5mm/23.5mm x 100% = 74.5%

(3.5) which is in agreement with the values obtained via XRD.

The results of calculating %C from a small group of infrared spectra of processed ciarinet reed samples is shown in figure 3.9. The differences in degree of crystallization are quite noticeable. The %C of reeds number 1 and 5, are quite high (on the order of 60%-70%). By contrast, the %C for reeds number 2, 3, and 4 are much lower (on the order of 40%). Such a wide variation (20% or more ) in the %C makes this method somewhat tenuous for 152

T- 0 . 6 - E o o o O) CM Q

E o MCM eo û 0 .2 -

Reed Number

Figure 3.9-%C from Infrared Data for Selected Reeds 153 quantitative analysis. The infrared method is more sensitive to the loss of cellulose-like material, such as hemicelluloses, and the effects of cell wall molecular orientation, than XRD. More carefully controlled experiments than performed here need to be done to see if this method can accurately distinguishes between pristine and spent reed states of crystallization.

Extending the Infrared Testing Method

While obtaining the infrared spectra for the determination of the %C, the author noticed an interesting difference between pristine reed spectra and spent reed spectra: the presence of several additional bands in the spectra. A comparison of three spectra, two pristine, and one spent, (middle line at the start of the spectra) is presented in figure 3.10. The spectra are virtually identical except in the 1550 cm-i to 1700 cm-i and the 800 cm-i regions where the spent reed spectrum diverges from the pristine reed spectra. The additional peaks between 1550 and 1700 cm-i have been identified for the author by Andrew

Summers at the University of Miami (Ohio), using an infrared microscope as being of human origins-scrapings from the inside of the human cheek reproduce the bands seen in the spent reed exactly. This strongly suggests that the peaks at 1550, 1620, and 1680 cm -i are related to the N-H deformation

(amide II) and C=0 stretch of peptides (such as proline- rich proteins) in the lip which coat the reed over time.

It may be possible to use these bands as a way of measuring the degree of deterioration of a clarinet reed (we shall comment more on the effects of saliva 154

g

Figure 3.10-Pristine and Spent Reed Infrared Spectra 155 on the reed surface in chapter four).

Other tests for Decrvstallization

A third method which may reveal decrystallization of the cellulose matrix is known as differential scanning calorimetry (DSC). The idea behind DSC is that a sample containing many different compounds is heated, and the heat flow rate into and out of the sample (in milliwatt equivalent energy) is measured. From this data and the weight of the sample, it is possible to calculate basic thermodynamic properties of the material.

When a DSC spectrum of a spent and a pristine reed are compared, it is found that the vaporization temperature of the spent reed cellulose is lower by about fifty degrees than that of the pristine reed (onset temperature 475° C for pristine vs. 416° C for the spent reed), as in figures 3.11. These values are determined by extending a line from the slope and the straight line portion before and after a weight loss. The intersection gives the onset temperature of vaporization.

This lower onset vaporization temperature of the spent reed is consistent with a decreased %C in the spent reed compared to the pristine reed.

Thermodynamically, more energy (which is measured by the enthalpy of vaporization) must be put into the crystallized form of a material to vaporize it, in order to randomize or destroy the extra stability gained by the lattice ordering

(i.e., to overcome the lattice stabilization energy), than for a more amorphous form of the same material. 156

Curve îi ose File info, 00082 Fri Nov 22 16,35,IB 1991 Soap le Weight, 1.480 »S Clorlnet Reed Prietine TS # I Clcrinet Reed Prietine TS Hect Flo* (sV> S 10.00 9 2 Clorlnet Reed Ueed K Heat Flo* (mM) S 5.00 ■

-5.00 - S -10.00 -

-15.00 -

•2 0 .0 0

•25.00 -

•30.00

•35,00 -4— 50.0 250.0150.0350.0 450.0 550.0 10 C/aln in Stognent Air Temperature (*C) F.K. CelJaqher *0.0 c/»,n PERX2N-ELMER 7 Serlee Thermal Analysle System Hon Nov 25 13,16,02 1991

Figure 3.11-Differential Scanning Calorimetry Plot of Pristine and Spent Reeds 157

Thus, it Is possible that the Increased melting point of the pristine reed Is due

either to the extra stability of the crystallized cellulose, or possibly due to the

presence of hemicelluloses (which are missing In the spent reeds, as

mentioned above) In the cell wall matrix, which would affect the molar

composition of the cellulose and lead to boiling point elevation (as In the

familiar case of salted water). It Is not possible at this time to determine which of

these phenomena lead to the change In vaporization temperature.

Degree of polvmerlzatlon

If the degree of crystallization measures the amount of ordered cellulose

which exists In the total amount of cellulose material, then another number, the

degree of polymerization, measures the number of cellulose molecules

strung together In each extended cellulose lattice. This number Is determined

by dividing the molecular weight of the cellulose lattice by the weight of one

cellulose molecule (not celloblose, but a single molecule):

Dp = wt. polymer / wt. monomer

(3.7)

Typical values for the Dp range from 15,300 molecules of cellulose for

California cotton still In Its capsule, to 305 cellulose molecules for rayon fibers.

The other carbohydrate compounds which form the cell wall matrix, the

hemicelluloses mentioned above, by contrast have Dp values of about 150 for mature tlssue.i2 158

It has not been possible in this study to measure the Dp of clarinet reeds,

since this involves extensive laboratory preparation of the sample, (e.g.,

techniques such as vapor phase osmometry, viscometry, light scattering,

sedimentation, etc., which are typically used to measure Dp) and is something

which must be reserved for future studies. The Dp for the cellulose in the cell

wall matrix in Arundo donax, for this study, will be assumed to be approximated

by the value for cellulose, which is on the order of 700 cellulose molecular units.

Fibrils

Cellulose molecular lattices link together to form long chains, which can form even more complicated three-dimensional lattice structures due to hydrogen bonding. These extended lattice structures are calledmicelles, and there is a huge literature on micelles in biophysics, as this is a topic currently of much interest in polymer chemistry and physics, if these chains and resulting lattices are sufficiently ordered and extended, a larger visual macromolecular structure is often formed. This mesoscopic structure of cellulose reveals itself as long slender thread-like fibers, calledfibrils (figure 2 .4 3 ).13

Fibrils of this size, just above the scale of the extended polymer lattice micelles of cellulose (= .003 |i-.008 |x), are technically called elementary fibrils or microfibrils, and just as thread or twine contains smaller threads twisted together to form the larger threads, elementary fibrils join together to form larger thread-likefibrils (sometimes called macrofibrils). It is these fibriis which link together to form the matrix of the cell wall. 159

The fibril structure is more complex than implied above, however, because hemicelluloses (the other class of carbohydrate compounds in the cell walls of plants) tend to form sheets or chains parallel to the outer surfaces of the cellulose micelles of the fibrils, and hydrogen bond to the cellulose. On the opposite side of the hemicellulose sheet, glycoproteins cross-link the cellulose-hemicellulose elementary fibril to other fibrils, to form a quasi-lattice of elementary fibrils, creating a macrofibril. In addition, lignin bonds to the hemicellulose material in the macrofibril, stabilizing the cell wall matrix, by

“filling in the gaps in the matrix”, but also slightly lowers the elastic strength of the material.

There are several theories which have been presented to explain the formation of the cell wall from fibrils: the multinet theory, the hélicoïdal theory and an extension of these theories based onperiodic microstructuring of a composite material, by Niklas. We shall refer the reader to the endnotes for more detail, since such a discussion is outside of the scope of this document, even though an understanding of the influence of the properties of the cell wall fibrils of Arundo donax may have importance implications in the growing of the plant for clarinet re e d s . 14

Hemicellulose

The other class of carbohydrate compounds found in clarinet reeds are the hemicelluloses, so-called because at one time they were thought to be steps in the natural biosynthesis of cellulose. The particular generic hemicellulose 160 found in clarinet reeds is the same one found in many softwoods, namely, arabino-4-O-methyl glucuronoxyian (figure 3.12).The molecular structure of this compound is more complicated than simple cellulose, and is formed by the linking together of so-calledxylan units, more specifically,

(1->4)-linked-p-D"Xylopyranose (abbreviated, Xyl), to form a backbone.

The structure of this xylan is shown in figure 3.12.

From the parent backbone, two different chemical units attach or branch from different and sometimes variable points on the backbone:

L-araboRofuranose (abbreviated, Araf), and 4-0-methyl-a-D-glucuronic acid (abbreviated, 4-Me-GlcA), as shown in figure 3.12. Araf typically attaches at the so-called 0-3 position of the Xyl backbone, which means that an oxygen bridge connects the Araf molecule from carbon number 1 of its ring system to the carbon numbered 3 in the Xyl ring system (so, more exactly, one could call this a 1-0-3 bridge, but convention is to simply assume that the external attaching molecule does so from the number 1 carbon atom in the ring unless otherwise stated). 4-Me-GlcA has variable bridging positions within the

Xyl backbone, but is typically at the 0-2 position.

The generic structure of the primary hemicellulose in Arundo donax mentioned above (which we shall abbreviateAMGX, hereafter,) and shown in figure 3.23, consists of branching Araf and 4-Me-GlcA units, but the exact locations are not specifically known, so the structure indicated is to be taken as an approximation, consistent with known chemical knowledge obtained through 4-O-methyl-ct-D-glucuronic acid :00H L-arabi nofuranose OH /

OH OH OH

OH OH OH

( 1-->4)-1 i nked-p-D-xy lopy ra nose ( b a c k b0 n e )

Schematic Structure

j 0 -2 0 -3

(0 -2 etc., => oxygen linkage)

# =Xy1

= 4 - Me-01 uA

= Araf

Figure 3A2-Arundo donax Hemicellulose 162 degradation experiments.

Different softwoods (and grasses, such as Arundo donax) have different concentrations of the Araf and 4-Me-GlcA side-chains. For exampie, Fengel and Wegener summarize a number of sources which show that the ratio of Xyl to 4-Me-GlcA varies in different softwoods from 5 : 1 to 6 :1 in softwoods, although extremes of 4 : 1 and 3 : 1 are known; the ratio of Xyi to Araf varies from 6 : 1 to 10 : 1 in softwoods.i6 Taken together, they give an averaged ratio of Xyi to 4-Me-GlcA to Araf of 8:1.6:1. In addition, softwoods also contain higher concentrations of 4-Me-GicA than hardwoods. Using Sodium Hydroxide at varying concentrations as an extractant, Solomon et. al, were able to obtain hemicelluiose giobules and determine that the ratio of Xyl to Araf is 5.8 : 1, and

Xyl to 4-Me-GlcA is 10 : 1 in Arundo d o n a x .i7

As Arundo donax matures, the concentration of hemiceiluioses and hemicellulose components changes. Since it was known from previous work that the relative amount of mature tissue greatly increases from the apex of the

Arundo donax stem down to the bottom of the stem, in a relatively immature

(one month old), but fast growing stem, Joseleau and Barnoud excised tissue from only 1 to 3 cm from the top of the stem, representing the youngest portion of the plant, from 9 and 15 cm from the top, representing the middie stage in development, and from 21 centimeters from the top (i.e., the bottom of the stem), representing the oldest portion of the plant. They found that the total hemicellulose content changed from 34% to 44% to 25% in the early, middle, and late tissue, respectively. 18 163

More specifically, the Xyl concentration changed from 68% to 89% to 81 %, while the Araf concentration decreased as the plant matured. They interpreted these results to the increased used of hemicellulositic material in the formation of the primary thickening of the stem, which typically occurs after the initial primary (vertical) growth period of the plant, and before stabilization of the stem size.19

The degree of polymerization of AMGX also changes as the plant matured: from 63% to 91% to the stable 151% mentioned earlier in this chapter during the rather general discussion of Dp. This corresponds to the observation that as the size of the vascular tissue increases, more of the hemicellulose is polymerized into the cell walls.

A comparison of the concentration of Xyl in different sections of a maturing

Arundo donax indicates that there is variation not only over time, but also within the different cell types. It has already been mentioned that the Xyl concentration changes from 34% to 44% from young to middle-aged tissue. More specifically, this is true for vascular tissue. Parenchyma tissue shows an even larger change: from 24.3% to 51.2%. In both types of tissue, the Araf concentration fell off dramatically: 15.9% to 2.6% for vascular tissue, and 7.3% to 2.6% for parenchyma.20 These results are also consistent with the primary thickening meristematic growth of the stem mentioned above.

The Role of Water and Hemicellulose

The purpose of hemicelluloses in Arundo donax, and woods in general, is to aid in the absorption of water. Hemicelluloses tend to be hydrophillic in nature. 164

and allow for the absorption of water molecules in the interstitial spaces of the

cellulose crystallite lattice. A related class of molecules, pectins, serve this

function even more in most woods (not, however in the grasses, since the

concentration of pectin is low).

Xylans may be isolated and grown into extended crystal structures. These

crystals may be hydrated, and show the relative hydrophyllicity of AMGX or

similar xylans. Upon reaching 100% humidity the crystal size of a

representative xylan crystal changes 5% in width and 1% in height.2i Partially

because of the steric hindrance of the side-chains in Xyl and other xylans, the type of hydrogen bonding which is seen in the cellulose unit crystal is not possible, and it is the addition of water molecules, with their propensity for hydrogen bonding, which is thought to be responsible for enabling the xylan crystal structure to form.22 The “piggyback” nature of the borrowing of the hydrogen bonds of water which stabilizes the xylan unit crystal (although, probably not as much as the “purer” hydrogen bonding in cellulose), and as well as the presence of the protruding side-chains of the molecule, explains the lower Dp of the hemicelluloses compared to cellulose.

Hemicellulose Degradation

It is sometimes possible to obtain qualitative data on the composition of large molecular entities in a material by the use of a technique known as thermogravimetric analysis (TGA), and we use it here to study the effect of saliva on hemicellulose in the reed material. The idea behind TGA is that, as a 165

sample containing many different compounds is heated, each compound will

boil off at its particular vaporization temperature, resulting in a drop in weight as

the particular compound is released in the from of a gas. When samples of

pristine and spent reeds are placed in TGA chambers and slowly heated and the weight loss recorded, the various volatile compounds contained in the reed

can sometimes be identified.

Two typical TGA spectra, spectra pi, and spectra p2, from two different

excised samples of a single pristine reed (about five mg per sample) are shown

in figure 3.13. The author is extremely grateful to Dr. Patrick Gallagher of The

Ohio State University department of chemistry/material science, who performed the TGA (and DSC) analysis, for donating his time, expertise, and his own equipment in this research.

The plots show a striking resemblance to the TGA curves for cellulose, as should be of no surprise, since most of the reed material is cellulositic.23

Differential plots (which show the change in slope of the original spectra and makes the weight loss easier to see) are overlayed in each case shown in figure 3.13. A tentative analysis of the loss in weight for the different molecular entities in the differential plot for the two pristine material is as follows (some form of evolved gas analysis would be needed to definitively identify the compounds); 166

^>va 11 TGA Fila Info, T383B Mon Nov 25 C5,25,47 1991 Soaplt Walght, 4.342 "g Clcrinat Raad Prlatlna TS f : Clorlnat Raad Prlattna TS WolghC (Wt. %) 100.0 -F # 2 let Darlvotlva (Z/eln x 10 ) 10.0

90.0 0.0

80.0

70.0 •30.0 b ±? 60.0 4 "40.0 & 50.0 4 e •50.0 40.0

30.0 ■; ■70.0

0.0 200.0 3CÔ.0 500.000Ô.0 700.0 800.0 900.0400.0 10 C/eln Sn Ar Tmparotura (*C1 P. K. Gollodwr îSSi: Ut: i Îiæ: feg — 5'§ÏÏS^oi/^=iy.i.syrt« M— M.W. oc ne.cT. 90 raoi '

Figure 3.13-TGA Spectrum of Two Pristine Clarinet Reeds 167 Figure 3.13 (cont.)

Curve ]i TGA File Infoi T383S Fri Nov 22 18,54,09 1091 Sample Weight, 4.528 mg Prlmtlne ftemd TS 0 I Prletinet Roed TS Weights (Wt. V § 2 let Derlvotlvm (Z/sln>

80.0 0.00

-5.00 2 70.0 Q •10.00 é •15.00 > & 50.0 •20.00 e 40.0-}

30.0 •25.00

20.0 -30.00

10.0 •35.00

0 .0 0.0 200.0 300.0 500.0 10 C/min in Air Teperoture C*D PERKIN e&S ? fnat tS tiS 7 Seriem^ThermalSeri______System Fri Nov 22 16,56,31 168

100° C-water, (approximately 5% loss by weight ( = Ibw)), 2200 C-AMGX (15% Ibw), 2800 c-LlgnIn, and some types of non-AMGX hemicelluloses, possibly MGX (approximately 18% Ibw) 3100 c-Llgnln (present In pristine reed sample p i. Not present In used reeds, si, Sa) 3500 C-cellulose (approximately 55% Ibw).23

Table 3.3-TGA PeakAssignments24

The presence of the additional peaks in spectra pi (noted above) may be due to a different sampling environment (i.e., having been chosen possibly from a region richer or poorer in vascular tissue than the material in spectrum 2 p), or due to different machine settings, as the tests were conducted on two different dates.

When a clarinet reed is initially selected and played, most musicians notice a sweet taste to the reed. We believe this taste is due primarily to the presence of

AMGX in the reed (having side-chain structures related to many other sugar-like glucosides). The reed is also extremely hydrophillic (we shall discuss the water capacity of clarinet reeds in chapter four).

As the reed is played upon, the sweet taste gradually diminishes, and the reed becomes gradually less hydrophillic, until some stable point is reached.

This is the point at which the reed is generally said to be “broken in.” These effects are probably due to the loss of hemicellulose in the reed cell wall matrix, and thus, it is this chemical degradation which is mainly responsible for the breaking in of a reed.

Considering that the reed is being subjected during the course of playing to something, which, as a chemistry laboratory procedure which could be 169

described as, “immerse in a water-filled, alkali/enzyme bath and shake

vigorously,” it is not difficult to understand why this type of material, which is

soluble In mild bases, is leached out of the reed cell walls.

Saliva is stimulated by the infusion of Ca2+ ions reaching the Acinar cells of

the salivary gland, and possibly the slightly alkali nature of these ions, along

with the action of such glucose degrading enzymes as amylase in solution,

along with the physical excitation of the reed, is enough to remove the AMGX in

the pristine reed. This is not to say that saliva is alkali, in and of itself (it is not,

except when the flow rate is sufficiently high, as explained in chapter four), but

rather that AMGX is sensitive to the presence of several compounds and ionic

species contained within saliva, some of which, such as Ca2+ andHCO 3- have

alkali properties and which can react with the AMGX when near it.

The loss of AMGX is shown in the TGA spectra, Si and sa in figure 3.14 (each

sample excised from the same single spent clarinet reed, but, as earlier, run

under different machine settings). These TGA spectra may be compared with the pristine samples pi and pa shown earlier (figure 3.13). A comparison of the

differential plots is shown in figure 3.15. The loss of the peak at 220 oC in the

spent samples indicates the absence of AMGX in the spent reeds. Lignin,

however, is unaffected, as both reeds show a peak at 280 oC.

As mentioned, MGX (4-0-methyl glucuronoxyian) is also present in the reed,

as identified in the spectra above at 310° C, and this peak also appears to be

removed by saliva, as it is missing in the spent reed as well. 170

Curv« It TGA Fll« Infoi T3B38 Fri Nov-22 38,55,37 3883 SapU Volghti 5.273 mg Clorlnmt Rood U##d K Clorlmmt Rmmd Uwd K

3D0.fi *

8fi.fi - fi.fiO

8fi.fi • • -5.00 7fi.fi - R é 5fi.fi

4fi.fi *

3fi.fi -

•25.00 3fi.fi- fi.fi - •30.00 fi.fi lfiO.fi 2fifi.fi 3fifi.fi 5fifi.fi 10 C/mln In Air Tmqwrotur# CO « % d è s g «ai: fcg «R :"•=

Figure 3.14-TGA Spectrum of Two Spent Clarinet Reeds Figure 3.14 (cont.) 171

Curve 11 TGA File Info, T3639 Mon Nov 25 :1,50,58 199: Sample Weight, 4.941 eg Clarinet Reed Ueed K # 1 Clorlnet Reed Ueed K Weight: (Wt. Z) 0 2 let Oerlvotlve (Z/mln x 10 ) 10.0 90.0 0.0

00.0 •10.0 70.0 Q ÿ 60.0 •30.0 ft 50.0- •40.0 40.0 - •50.0 30.0

20.0 •70.0 10.0

0.0 100.0 200.0 300.0 500.0 C 700.0600. 000.0 900.0 10 C/mln in Ar Temperature <*C) P^^CoIlagher iin lUTi,* ia.fl e/i 7 Serlee Thereol Analvele Syetem Mon Nov 25 12,29,00 1991 ' 172 Curve 11 TS.* Fila info, T3B36 Fri Nov 22 Î8i55,37 199: Soaplo Weight,5.273 eg Clarinet Reed Ueed K U I Clcrinet Reed Ueed V. let Zerlvetive C/e in) § 2 Prietine Reed TS let Oerlvotlve C/eln) C. 00

e -10.00 I

100.0 300.0 400.0 500.0 600.0 700.0 900.0 10 c/eln In Air Teeperoture (*D P.K. Colleger PERXIN-ELMER asi: I «%: 8:8 7 Serlee Thereel Analvele Syetea Kon Nov 25 12,49,12 %%] ^

Figure 3.15-Comparison of Differential Plots of Pristine and Spent TGA Spectra 17 3 Figure 3.15 (cont.)

Curv# Il ToA fllQ infoi T3S38 Ken Nov 25 C9i25i47 1S81 Seapl« Vsightt 4,342 mg Clsrinat Read Pristina TS tf ! Clorinst Raad Pristina 75 1st Oerlvctivs (I/sin x 1C 10. C -i § 2 Clorinst Rasd Ussd K 1st Oorlvotlva CZ/aln % 10*^) 0.0

„ - 10.0 1 - -20.0 4 ; ! - -30.0 -i I ? •"“■“1 5 -50.0*1 I -60.0 J

-90.0

;cô.o 200.0 300.0 400.0 600.0 700.0 600.0 600.0 10 c/min in Ar Ttmpsroturs <*C> P. K. Golloohsr I SATCI* 1 0 .0 C/mir PERKIN-£U€R g 7 Sariss Thonsol Anolysis Systom Mon Nov 25 12, 42i 45 I%1 174

Infrared Identification of Hemicellulose

Earlier in this chapter we discussed the possible role of infrared

spectroscopy in determining the degree of crystallization of the reed material.

Infrared spectroscopy may also be used to identify compounds by their

characteristic absorption peaks.

Marchessault and Liang have done extensive analysis of the infrared spectra

of xylan structures.25 On the basis of their research into the MGX, it is possible

to identify some of the peaks in the infrared spectra ofArundo donax as being

of AMGX origin. A complete identification of the peaks in the spectra are given

in the section on lignin later in this chapter.

Interestingly, the infrared spectra of spent reeds indicate that AMGX is

present in the same amounts in both spent and pristine reeds (figure 3.10), if

one assumes that the band between 1000 and 1150 cm i is due to the presence

of AMGX, as the work by Marchessault and Liang seems to suggest. This

discrepancy with TGA is something which must be explored further, as the

decrystallization results and TGA results seem to indicate some change in the

composition of the reed hemicellulose due to saliva immersion. It is possible that the reaction with saliva is more complicated than the simple leaching model would indicate. In addition, there are biomorphological differences in infrared

spectra depending upon concentration of ground tissue and vascular tissue, which have different chemical environments. Most of the work done in this chapter was using infrared microscopy, which would be more sensitive to these tissue/environment differences. 175

To summarize, later harvesting of Arundo donax leads to stable concentrations of AMGX In the plant stem (about 25%). Alternately, younger stems, sectioned only near the base of the plant, show the same trend.

Ultimately, AMGX appears to be leached out of the material relatively early on In the playing life of the reed (a detailed time-dependent study via TGA or Infrared spectroscopy of the rate of loss of AMGX due to saliva Immersion time Is still wanting).

Llanin Preliminaries

The last major chemical component (20% to 40%) of Arundo donax Is lignin

(from the Latin word for wood). This highly complex gummy filler substance is formed from dead plant cells, and usually occurs as a result of the aging process of plants. Lignin protects older plants from pathogens and water, stabilizes the cell wall matrix, and gives viscous shock protection and some plasticity to the plant.

There are three major types of lignin which exist In different hardwood, softwoods (and grasses): guaiacyl lignin, syringyl lignin, and, guiacyl-syringyl lignin (for reason which will be explained, momentarily).26

Each type of lignin Is formed from the Incredibly complex polymerization of one or more of three simple molecular units of propyl alcohol (PrA) which have a different phenol ring derivative attached at the third carbon: p-coumaryl alcohol (PA), coniferyl alcohol (CA), and sinapyl alcohol (SA) (figure

3.16). It should not be Inferred that there Is a one-to-one correlation between 176

OH- /y ^ CH2CHCÜOH

NHg tyrosine

i4 \ CH^OH CH OH CHgOH I CH CH CH II II II CH CH CH

s\ OCH 3 OCH OCH I OH OH

p-coumaryl coniferyl si napyl alcohol alcohol alcohol PA CA SA

e e OCH OCH C9 OH OH •O p-hydroxyphenol yuiacayl S i napyl 1 unit (H ) unit (G) unit (S ) ë N c « a HC=0 HC=0 HC=0 2 £

OCH OCH OCH OH OH OH p-hydroxybenzaldehyde vanillin syri ngaldehyde H (y) V S (y)

Figure 3.16-Llgnln Preliminaries 177 the three alcohols and the three above-mentioned lignins. That there are three

of each is purely coincidental, as the three lignins mentioned above each

contain mixtures of ail three alcohols, and derive their names from the

by-products of chemical analysis of the lignins (to be explained shortly).

The polymerization of the building-biocks mentioned above results in a

molecular structure composed of many linked benzene/phenoi ring systems, which has not yet been completely worked out for any of the lignin types, and estimates or models of the the number of rings involved for a single molecule of a prototypical lignin range from sixteen rings to ninety-four rings! Such aromatic entities, having large numbers of carbon atoms, typically appear as gummy substances, and lignin is no exception, as mentioned above.

It is not possible to go into the incredible amount of biosynthesis which creates each of the three types of lignin, nor their polymerized forms, and for such a discussion the reader is referred to either Fengel and Wegener, or

B ra u n .2 7 In a simple-minded manner, however, it is possible to imagine that lignin is formed in most true woods (i.e., those plants having secondary growth) from glucose, which is converted to a compound called shikimic acid (this method of biosynthesis of lignin is called, not surprisingly, the shikimic acid pathway), which eventually works its way through different biological reactions to one of the three alcohols mentioned above, which then begin to polymerize.

Grasswoods, on the other hand, arrive at these alcohol precursors by a different route than the shikimic acid pathway, starting instead with tyrosine,

(which is known also as the phenylpropanoid, p-hydroxyphenylalanine). 178

Of the three types of generic lignlns, the gualacyi llgnin, which comprises most softwoods, is derived primarily from coniferyl alcohol. Hardwoods contain mostly guiacyl-syringyl lignin, which is a copolymer comprised of the coniferyl and sinapyl alcohols in ratios of 4 : 1 to 2 : 1 . 28 Sinapyl lignin is found mainly in compression wood, which is the type of wood which results when a plant is subject to bending or is injured, it contains large amounts of p-hydroxyphenol elements, and is sometimes alternately called p-hydroxyphenyl lignin.

These are several ways to classify the three types of lignins mentioned above, as might be expected from their complex molecular composition. If one ignores the propanol unit which is common to all of the three precursors mentioned above, and is only concerned with the remaining aromatic substituent, then one may call these guaiacyl units (from coniferyl alcohol without the propanol subunit, i.e., CA without PrA), syringyi units (from sinapyl alcohol without the propanol subunit, i.e., SA without PrA), and p-hydroxyphenyl ( from the p-coumaryl alcohol without the propanol subunit, i.e., PA without PrA ). These aromatic subunits are abbreviated,G, 8, and H, respectively (i.e., G =CA without PrA, S=SA without PrA, H=PA without PrA).

When a lignin is reacted with nitrobenzene in a process called nitrobenzene oxidation, the biopolymer is fractured into the three basic constituent molecules mentioned above (G, 8, H), only converted to their aldehyde form.29 The guaiacyl ring (G) in its aldehyde form is simply vanillin

(i.e., G->V). The aldehyde form of the syringyi unit is syringyialdehyde (i.e.. 179

S-->Sy). Finally, the aldehyde form of the p-hydroxypenyl subunit is p-hydrcxybenzaldehyde (i.e., H—>Hy)). Since there is a one-to-one ratio between the number of moles of the aldehydes formed through the nitrobenzene oxidation process (which is often easy to measure), and their original aromatic subunits counterparts (G ,S,and H), measuring the number of moles of aldehyde formed suffices to determine amount of the corresponding aromatic forms. The reader should not be surprised that the two labeling systems are essentially the same, since they refer to related compounds, with V being preferred for the aldehyde form of G, and Sy, and Hy being use in this document (the literature uses S, and H for the aldehyde forms of S and H, as well but this might confuse the reader) for the aldehyde derivative.

Depending upon which lignin is used, different ratios of these three subunits,V, Sy,Hy (or G, S, H) are obtained by nitrobenzene oxidation. As mentioned earlier, most softwoods contain large amounts of CA, and hence it comes a no surprise that nitrobenzene oxidation yields large amounts of vanillin, indicative of the guaiacyl subunit, which is derived from coniferyi alcohol (CA). Hardwoods, as mentioned above contain both CA and SA and yield primarily V (G) and Sy upon nitrobenzene oxidation. Grasswoods contain primarily CA, but also sizable amounts of both SA and PA, so it should come as no surprise that V,Sy, and Hy products are formed by nitrobenzene oxidation.

Based upon these results, one may refer to softwood lignins asG-iignins, hardwood (dicot) lignins as GS(y)-iignins, and grasswood lignins as

GS(y)H(y)-lignlns (the usual literature omits the y). 180

Arundo donax Lianin

With these preliminaries in mind, results of studies of Arundo donax lignin may be summarized.

All three lignins covalently bond to most of the various types of hemiceiluloses found in plants. Since the ratio of Araf to Xyl is higher in the lignin obtained via the residue when the mill wood process is applied to a stem of Arundo donax than that reported for the hemicellulose of Arundo donax,

AMGX, Diesheng Tai, et al., concluded that that the bonding between the GSH lignin of Arundo donax and the major hemicellulose, AMGX, occurs typically at the site of the side-chain unit, Araf, which was mentioned above (figure 3.12) in the discussion of hemicelluose.30

This group also found that the ratio of V, Sy, and Hy (i.e., G, S, H) were,

1 ; 1.14: .34 (10.87%, 14.98%, 2.95%, by weight of the milled wood process residue) respectively, confirming that this lignin is a composite GSH-lignin. The molecular weight, indicative of the biopolymer, is between 1000 and 10000

Daltons, which is similar to most grass lignins, and indicate that up to 100 carbon rings may be found in the complex biopolymer, as has already been mentioned.

Joseleau et al. found, as to be expected, that there is also a correlation between age and lignin concentration.3i The concentration of lignin increases about ten-fold as one goes from the top (younger portion) of the stem to the bottom (older portion). This group also found that although the concentration of

G is always large, as the plant matures, the amount of Sy increases. 181 presumably reaching the roughly 1 : 1 ratio mentioned above by Diesheng et al.

We shall discuss the implications of these findings on the biomechanics of the clarinet reed later, since the presence of the rather amorphous lignin in the relatively flexible cell wall structure of Arundo donax may be responsible for changing the plant stem from one in which there is only one naturally dominant plane of material strength (a situation common in true woods) into one in which each of the x, y, and z planes are of equal strength.

Spectroscopic Methods of Lianin Determination

There is no iaboratory chemical process which will precisely determine the concentration of lignin in a given sample of wood, although nitrobenzene digestive methods are commonly used. In the area of machine testing, there are several ways in which lignin may be studied non-destructively, which yield additional information. The three most common are infrared spectroscopy, ultraviolet spectroscopy, and nuclear magnetic resonance spectroscopy.

The easier technique to use is ultraviolet spectroscopy, in which the absorption or reflectance of ultraviolet light of various wavelengths

(frequencies) is measured. Lignin, containing many aromatic rings, shows an absorption maximum at wavelengths of 280 nm (nanometers), and a shoulder in the 230 nm wavelength ran g e.32 Another very large peak occurs in the 200 to 208 nm range. Deisheng et al. found that the ultraviolet spectrum of Arundo donax showed the peaks at 280 nm and an additional peak at 310 nm (it is not reported if they found the peak at 200 nm, although they almost certainly did).3i

The position of the 280 nm peak changes as the lignin type changes, thus 182 allowing for confirmation of lignin types which may be present in a sample.

Softwood lignins have the second peak at 280 or higher nm, while hardwood lignin shows the peak at lower wavelengths (270 nm-275 nm), which is indicative of the larger amounts of the Sy units in hardwood, which have higher symmetry than the the G units found in softwoods (recall that hardwoods are

G-Sy lignins, and softwoods are G-lignins). Interestingly, since the amount of

Sy increases as Arundo donax ages, it should be possible to qualitatively determine the relative maturity of samples in the growing field by the shift in the

280 nm peak location as the stems mature. Also, the peak at 310 nm may also show variation with age, since it is primarily due to the presence of Hy units in the stem.

Infrared spectroscopy, mentioned earlier, is also useful in revealing the presence of lignin in a woody sample, although the interpretation of which type of lignin is present is not straightforward.33 a typical infrared spectrum is shown in figure 3.17. As mentioned earlier, the troughs in the spectrum corresponds to the decreased transmission of infrared radiation at that frequency (or equivalently, wavelength, since the two are related by a simple formula). When

IR radiation is absorbed, it sets into motion the internal modes of oscillation of a molecule. Different wavelengths effect different modes of vibration (such as bending of certain elements on the molecule, or stretching of some of the atoms, etc.).

A list of the internal modes of oscillation of the three principal components of

Arundo donax (C = cellulose, H = hemicellulose, L = lignin) corresponding to uieiSXBUop opunjy ;o lunjpads paJBJjU|-/|. s ajnSy Transmission Intensity (100% = 1) o 0 o o o o o to CO Ul 01 OJ CO U l 4000 3876

3752

3628

3504

3380

3256

3132

3008

2884

2760

2636

2512 o 3 2388 2264

2140 s 2016

1892

1768 O 1644

1520

1396 1272 W 1148 W 1024

900 to 776

£81. 184 the peaks in the IR spectrum in figure 3.17 is given in table 3.2. This table was compiled by examining the reported transmission peaks of the three components, separately.34 For the most part, the peaks are distinct to each species, allowing precise identification of bands to be given. For instance, the bands at 833, 1510, and 1600 cm-i are diagnostic for lignin, while the broad band at 1000-1120 cm-i is diagnostic for AMGX (although cellulose peaks do occur here also, but are masked). The peak at 1600 cm-i may sometimes be used to indicate the presence of bound intercellular water, if the concentration of lignin is sufficiently low to allow the peak to be expressed. The fact that the peaks at 1600 and 1510 cm-i are so strong indicates that this is truely a

Guaiacyl (G-)lignin from grasses and softwoods, as opposed to a

Syringyl-dominated GS(y)-lignin as found in hardwoods.

The shoulder at 1550 is problematic. In spent reeds, this may indicate the presence of amide N-H deformation, however, it is so masked by the lignin antipeak that this is difficult to see. This amide peak may possibly be the identity of the peak at 1530 cm-i in figure 3.29 in the bottom spectrum. The

1600-1700 cm-1 region shows the presence of either 0 = 0 stretching in lignin (a single peak) in pristine reeds, N-H deformation (triplet of peaks) of the primary amides in the proteins in saliva in spent or played reeds, or adsorbed water in the reed cellulose matrix. We have indicated these possibilities by labeling the peak OLA. Several of the peaks are common to all three species (such as the

3200-3400 cm-1 peak), and we have so indicated this on the spectrum.

It may be possible to use the intensity of the peaks to determine various 185 Symbol Frequency fcm'^) Çomments(see endnote 34)

LI 833 Aromatic C-H out-of-plane deformation L2 870 Same as LI HI 897 C-1 group frequency or ring frequency H2 990-1120 C -0 stretch (cellulose peaks buried in region also) H3 1130 Antisymmetric in-phase ring stretch H4 1160 Antisymmetric bridge C-O-C stretching L3 1240 Guaiacyl ring breathing with CO-stretching (may be 0 -H in-piane bending in Hemicellulose) L4 1320 Syringyi ring breathing with C-O stretch L5 1370 C-H deformation (symmetric)

C l 1420 C-H2 symmetric bending L6 1460 C-H deformation (asymmetric) L7 1510 Aromatic skeletal vibrations (signature for lignin) AL 1550 L =Lignin “antipeak” , A = secondary amine N-H deformation (an unidentified peak may occur instead at 1530 cm'^ in some (pristine?) reeds) LB 1600 Aromatic skeletal vibrations (also, indicates prescence of

intercellular H2 O in cellulose at 1595 cm"^) OLA 1650 C=water of hydration (cellulose), or L=carbonyl stretching (para-substituted aryl ketone), A=N-H deformation, 1° amide (1620-1650 cm'^), or C = 0 stretch, 1^ amide (1620-1670 cm"^) H6 1720 C = 0 stretching (acid) CO2 1780-2700 Air (carbon dioxide), unidentified peak at 2100 cm"^

L.H,C 2780-2880 L=OH-stretch in methyl and methylene groups (2880, 2940 cm""'), H=C-H stretching (2873,2914 cm'"*),

CH2 antisymmetric stretching (2935 cm'^ ), C =CH 2

symmetric stretch (2851 cm"^), C-H stretch (2907 cm'^) H,C,L 3250-3500 H =0-H stretch, C = 0 -H stretch, L = O-H stretch

Figure 3.18-lnfrared Peak Assignments for Arundo donax Spectrum 186

properties (such as the degree of polymerization and the maturity of a given

specimen) of the three species, such as lignin, of Arundo donax, since

Kawamura and Higuchi have shown that the ratios of G to S to H may be

classified according to IR peak information, although such studies have not yet

been undertaken for Arundod o n ax.35

What is reasonably clear from the data presented in figure 3.20 shown earlier

is that Arundo donax lignin is not changed appreciably over the course of the

playing life of the reed. There appears to be little difference in peak intensities

from pristine to spent reeds in general, and no profound changes in peak

locations or intensity (except for the water peak at 1600, and 1650 cm-i, which is

expected to change intensity due to drying) or the manifestation of new peaks

(other than the amine peaks from saliva, which do affect the reed cellular matrix,

as mention in chapter four). Thus, the reed appears to be remarkably stable over its playing life. The increase in bacteria cellulose in the reed due to oral

microflora (see chapter four), also has not been measured.

In regards to the final spectroscopic technique of studying lignin (and the reed

in general), Deisheng et all. report a very complex Nuclear Magnetic

Resonance spectrum, involving some 35 differentp eaks.36 We shall not attempt to summarize their work, as it is quite complicated, and we refer the reader to the original article. Using this information, however, they were able to confirm that the G and S units predominate over the H unit concentrations, as

discovered through chemical degradation independently by Joseleau.37 187

Cell Wall Chemistry

In terms of the relative contribution of the three principal classes of clarinet

reed components which occur in the cut plant stem: cellulose, hemiceiluloses

(among which, AMGX, is the most abundant), and GS(y)H(y)-lignin, the

research is summarized by P urdu e.3s On average, the components contained

in the stems five different plant stems analyzed by five different researcher

digested to yield 42-50% cellulose, 20-24% hemiceiluloses, and about 10-20%

lignin. The amount of cellulose in Arundo donax is comparable to that in wood

(40-50%), while the lignin percentage is iower than wood (20-40%), higher than

other grasses (12-19%), and approximately the same as other types of cane,

such as Bamboo (14-32%). As Pointed out earlier, the percentage of

hemicellulose is roughly the same as for the cane grasses (such as Bamboo),

hardwoods and grain grasses (15-30%), but more than for most softwoods

(8-14%). The ash content (approximately 4%), compares favorably to other grasses, but is less than for woods (= 1%). The total silica content of the stem

(epidermis and soft tissue) was 1-2%.

Although it is not possible to examine in detail the chemical contents of the cell walls of Arundo donax within the confines of this document, some broad conclusions may be drawn from work on related wood species. Softwoods tend to show a larger concentration of lignin ( 60%) than cellulose (14%) in the compound middle lamella (middle lamella and primary wall). In the secondary walls the situation is reversed, with more cellulose (60%) than lignin (27%).

The amount of cellulose in the secondary wall (most noticeably the 82 layer) 188 gradually Increases as the plant ages and more cellulositic material is required in secondary growth.

Since Arundo donax does not a have a secondary wall structure, it is reasonable to assume that most of the lignin in the stem is deposited in the middie lamella, and that most of the cellulose and hemiceiluloses are deposited in the primary wall, with the concentration of hemicellulose gradually changing over time due to growth requirements.

Unfortunately, at this time, there is no quantitative way of measuring the total lignin concentration of a plant stem, since chemical extrative methods tend to adulterate some portion of the lignin. One method, suggested by the work of

Goring, and reported by Fengel and Wegener, is to thin-section the material and study the cross-sections by means of ultraviolet microscopy.39 Since lignin is ultraviolet active, the exact location of the p-alkylphenol entities contained within the lignin in the thin-section of stem cell walls is made to fluoresce using a monochromatic 280 nm light source (i.e., the aromatic constituents, such as G,

8, and H).

This type of study may be crucial in the classification of the quality of Arundo donax cane and the resultant clarinet reeds; however, since the location of lignin throughout a clarinet reed must be determined without tampering with the reed (something not possible with the ultra thin sectioning required for UV microscopy), another possible method for revealing the lignin structure in clarinet reeds must be developed. One possible method is based upon the birefringence under UV radiation of the 280 nm wavelength of lignin phenyl 189 groups which are attached tc different sides cf the hemicellulose unit crystal.

Such a techniques has not yet been developed, but if successful, should prove a reliable method for studying the entire lignin concentration variations across a whole clarinet reed, without having to section the material.

Chemistry* and Material Properties

With this basic understanding of the three major chemical components of the clarinet reed finished, the next step is to consider the natural progression to the bulk material properties of the reed, i.e., how the molecular building blocks mentioned in this chapter affect such properties as damping, modulus of elasticity, water capacity, etc., of the reed. Since these topics form a separate body of material, we shall consider them in the next chapter. 190

Endnotes

1. James A. Richards, Jr., Francis Weston Sears, M. Russell Wehr, and Mark. W. Zemansky, Modern University Physics (Reading Mass., Addison-Wesiey, 1964): 807.

2. For and excellent discussion of nomenclature, see, R. D. Guthrie,introduction to Carbohydrate Chemistry, 4th ed. (Oxford, England: Clarendon Press, 1974).

3. Guthrie, ...Carbohydrate Chemistry..., 9.

4.Eero Sjostrom, Wood Chemistry: fundamentals and applications (New york. Academic press, 1981).

5. Purdue, “Arundo...”, 392.

6. Fengel and Wegener, Wood..., 84.

7. Alvin Hudson and Rex Nelson, University Physics (New York: Harcourt Brace and Jovanovich, 1982): 839.

8. M. L .Nelson and R. T. O’Connor, “Relation of Certain Infrared Bands to Cellulose Crystallinity and Crystal Lattice Type. Part II. A New Infrared Ratio for Estimation of Crystallinity in Cellulose I and II,” Journal of Applied Polymer Science 8: 1325-1341; “Relation of Certain Infrared Bands to Cellulose Crystallinity and Crystal Lattice Type. Part I. Spectra of lattice Types I, II, and III and Amorphous Cellulose,” Journal of Applied Polymer Science 8: 1311-1324;

9. Panel discussion at the San Franscico meeting of the Material Research Society, April, 1994.

10. Nelson and O’Connor, "Infrared Bands...Part II,” 1336.

11. ibid.,1336

12. Fenger and Wegener, Wood..., 109.

13. Karl Niklas, Plant Biomechanics: an Engineering Approach to Plant Form and Function (Chicago: University of Chicago Press, 1992): 245 ff. 191

14. ibid., 247 ff.

15. Jean-Paul Joseleau and Fernand Bernoud, “Hemiceiluloses ofArundo donax at Different Stages of Maturity,” Phytochemistry, 14 (1975): 71

16. Fenger and Wegener, Wood..., I l l

17. 8. Soiomon, G. H. Rozmarin, and Cr. Simioneseu, “Etude Chemique des hemiceiluloses du Roseau. I. Separation et Fractionnement des Hemicelluloses,”Ce//u/oseChemistry and Technology, 2 (1968): 291-304.

18. Joseleau et al., “Hemiceiluloses...," 72.

19. ibid., 72.

20. ibid, 72.

21. Fenger and Wegener, Wood...,113.

22. ibid. 114.

23. ibid. 321.

24. ibid. 323

25. R. H. Marchessault and C. Y. Liang,"The Infrared Spectra pf Crystalline Polysaccharides VIII. Xylans.” Journal of Polymer Science 59 (1962): 357-372.

26. Fenger and Wegener, Wood..., 149.

27. Friedrich Emil Brauns and Dorothy Alexandra Brauns, Chemistry of Lignin, suppliment volume, covering the literature for the years 1949-1958 (N.Y. Academic Press, 1960).

28. Fenger and Wegener, Wood..., 150.

29. ibid. 142.

30. Diesheng Tai, Weixin Oho, and Wenlan Xi, “Studies ofArundo donax Lignins," 4th International Symposium on Wood and Pulping Chemistry (Paris: Jacques Poncet Bresson, 1987): 13-17. 192

31. Jean-Paul Joseleau, Gerhard E. Miksche and Seiichi Yasuda, “Structural Variation of Arundo donax in Relationship to Growth," Holzforshunsg 31 no. 1 (1976): 19-20.

32. Fenger and Wegener, Wood..., 157.

33. ibid. 161.

34. See, Tsuboi Masamichu,"Infrared Spectrum and Crystal Structure of Cellulose,” Journal of Polymer Science, 25(1957): 159-171; C. Y. Liang and R. H. Marchessault, “The Infrared Spectrum of Crystalline Polysacchrides, Ill-Native Cellulose in the Region from 640 to 1700 cm Journal of Polymer Science 34 (1959): 269-278; N.L. Nikitin, Chemistry of Cellulose and Wood (Jerusalem: Israel Program for Scientific Translations, 1966): 58-59, Ward Pigman and Derek Horton, eds.. The Carbohydrates: chemistry and biochemistry (New York: Academic Press, 1980): 1400; R. H. Marchessault and C. Y. Liang,"The Infrared Spectra pf Crystalline Polysaccharides VIII. Xylans.” Journal of Polymer Science 59 (1962): 357-372; Friedrich Emil Brauns and Dorothy Alexandra Brauns, Chemistry of Lignin, suppliment volume, covering the literature for the years 1949-1958 (N.Y. Academic Press, 1960).

35. Fengel and Wegener, Wood..., 163.

36. Tai, et al., “Studies of Arundo donax...," 16.

37. Joseleau et al., “Structural Variation...," 18.

38. Purdue, “Arundo donax," 392.

39. Fenger and Wegener, Wood..., 228. Chapter IV

Clarinet Reed Material Properties

Introduction

Given a basic understanding of the molecular components ofArundo donax, we now turn to a consideration of some of the bulk mesoscopic properties of the processed plant material. It is these properties which most affect the clarinet reed in terms of its mechanical responses.

In this chapter we summarize the results of empirical tests made to determine some of the more elementary material properties of Arundo donax in the form of processed clarinet reeds. Where possible, we also relate these properties to reed quaiity.

In order to have a valid test set, as mentioned in chapter one, a series of tests were performed on an experimental set of 80 reeds, which we shall abbreviate as {R}, consisting of fifty Vandoren V12 reeds (of which forty were strength 4, and ten were of strength 3 1/2), ten Biack Master brand reeds (Vandoren Reed

Co.) of strength 3 1/2, ten Olivieri tempered brand reeds (Olivieri Co.) of strength 4, and ten Grand Concert brand reeds (Rico Reed Co.) of thick blank and strength 4. In designing this expérimentai set of reeds, two different subsets were intended to be created-an intrabrand test set, consisting of the

193 194

forty Vandoren reeds of strength 4, and an interbrand set, consisting of the

Black Master, Olivieri, Grand Concert, and Vandoren (strength 3 1/2) reeds.

The first subset of Vandoren V I2 strength 4 reeds, which we call subset,

R2.intra, allowed for the ratings and variables within a single brand to be

compared, while the second subset of different reed types and strengths, which

we call subset R2.lnter, allowed for the ratings and variables between brands

to be compared. There were forty reeds in each subset. In large measure, the

results of the quality and performance criteria ratings were independent of

brand, while the structural measurements, which we shall describe later, varied

greatly, and showed some interesting trends. Because of the independence of the quaiity and performance criteria to brand , to reduce the amount of graphs

and statistics (which could grow quite large, otherwise), the author will present, where possible, only the results of the composite set {R} of eighty reeds. The structural statistics will be dealt with in chapter five.

Admittedly, other reed brands should have been included, such as Glotin brand reeds, and Queen brand reeds, but the brands included in the test set were those actually being used by the members of the clarinet section of The

Ohio State University Concert Band at the time of the tests, and seemed to be a balanced, although not thorough, representation of the distribution of brands of

reeds used by students throughout the country. The author produced the experimental set by asking the players their particular brand and strength, and then supplying each of them with a test box and a free box as a remuneration for participation in the testing. Overall, ten students rated reeds, and the set {R} 195 of eight boxes of reeds was selected before actual rating started to create the two subsets mentioned above. The two other student's reeds were used primarily as additional data In testing the correlation between various physical parameters, although, where appropriate their quality reports were added to the overall test set, {R} when the maintenance of the subset structure was unimportant.

The reeds were rated on a one to seven scale, since It was determined that such a rating scale best conformed to standard statistical measurement practices for human judgment tests. The reeds were rated by students ranging

In experience from advanced college undergraduates to doctoral candidates In clarinet performance. The rated Items were (In the following listed pairs, the descriptor associated with the rating of one Is listed first, while the descriptor at the other end of the scale at seven Is listed secondly. I.e., dark vs. bright means dark = 1 to bright = 7): 1) overall quality (i.e., bad-good), 2) dark vs. bright (I.e., timbre), 3) soft vs. hard (strength), 4) buzzy vs. fluid (noise), 5) squeaky vs. stable (stability), 6) weak-bodled vs. full-bodied (acoustic strength). A sample instruction set Is shown In appendix a.

The results were tabulated and compared to physical measurements made on each of the reeds Including shape measurements (to be described In detail

In the next chapter), measurements of volume fraction of vascular tissue, structural damping measurements, natural mode frequency measurements, and measurement of reed mass, both In Its normal hydrated state and oven dried to constant weighing. In addition, selected reeds were measured for their 196 extensional modulus of elasticity, bending modulus of elasticity, and material damping properties (mainly those reeds lying at the extremes of overall quality rating).

We shall summarize the measurements and findings individually in the sections to follow. Matters relating purely to shape theory, however, will be postponed until next chapter, where the effects of shape on reed vibration will be discussed in detail.

Since the reed material is a biological material, the range of variation in the measured properties is quite large, as to be expected (according to Niklas, these results show a scatter reminiscent of aWeibuil distribution).'' In some cases, largely due to expense and difficulty of obtaining research equipment, only a few exploratory tests of a variable could be made (such as thermal expansion and reed bending modulus). In these cases, the goal was to establish a general order of magnitude for the variables, with the realization that further testing is warranted.

Descriptive results

The reported values of the overall quality rating obtained are plotted as a function of reed number in figures 4.1 (a rating summary is also plotted). The descriptive statistics for the reed quality ratings, generated using the statistical package SYSTAT, are given in figure 4.2 (C.V. = cumulative variance, a measure useful in ANOVA statistics): 197

B M G COÜV V(3.5) B M 7

5 Q QQ- ? 41 > 3

■ee-

0 20 40 60 100 Reed Number

« 3 I O'

12 3 4 5 6 7

Rating Number for Overall Quality

Figure 4.1-Overall Ratings for {R} 198

N OF CASES 80 MINIMUM 1.000 MAXIMUM 7.000 RANGE 6.000 MEAN 4.350 VARIANCE 2.661 STANDARD DEV 1.631 STD. ERROR 0.182 SKEWNESS(GI) -0.259 KURTOSIS(G2) -0.487 SUM 348 C.V. 0.375 MEDIAN 4.000

Figure 4.2-Descriptive Statistics for Reed Quality rating

The positive bias ( mean = 4.350) in the data towards the "good" range of reed performance (48.75% of the reeds lie in the 5 to 7 “good quality” range, but only

26.25% lie in the 1 to 3 “poor quality” range, with another 25% at exactly 4) is probably due to one of three causes: 1) the reeds which were rated were high performance reeds, made for professional players. It is quite possible that the adage, “you get what you pay for,” is true in this case, 2) the players rating the reeds were among the finest student players in the country, and although it may seem that the attainment of such high levels of skill should make these clarinetists more demanding in terms of reed quality (and certainly, it does), it also allows the players to adapt more easily to the vagaries of the reed material and still obtain good results, 3) it may be that there is a minimum set of parameter values which will permit a reed to respond in a high quality and a 199 student quality manner (in fact, in chapter five, we shali argue this exact point), and that, by trial and error (assisted by the concomitant refinement of mouthpieces) these parameters have been understood by the reed companies in a general way. Thus, the overall higher than average ratings of the reeds in the experimental set may be a testimonial to the intuitive and empirical understanding of the reed makers themseives.

The five different performance criteria also show interesting statistics. The reported ratings and distributions are piotted in figures 4.3 to 4.7. The summary statistics are given below (figure 4.8). The pair abbreviations are DA-BR =

Dark-Bright, SO-HA = Soft-Hard, BU-FL = Buzzy-Fluid, SQ-ST =

Squeaky-Stable, WE-FU = Weak-Bodied-Full-bodied, where the first pair member is at the lower end of the scale. The names of the statistical measures

are abbreviated where possible to conserve space. The reason for the slightly larger number of cases for some of the statistical measures is that data from an additional student (Vandoren Biack Master strength 3 1/2) were included where possible.

The ratings show that in a random sampling of high performance clarinet reeds, the reedsare more than likely to be; 1) generally playable, 2) bright sounding, 3) slightly hard, 4) slightly more fluid in response, 5) slightly more stable in general, and 6) full sounding.

A Pearson correlation matrix was created in order to test the correlation between the performance descriptors with the overall reed ratings. The results are summarized in figure 4.9. BMGC Oliv BM 2.00 8

6 000- 0

0 X 3 OOO o o o ■e-e- CD- 0( >■ - QO o» L mI 4 03 03 O QD w u 3 0 0 0 ' O O □O OO00 n û

2 ■Œ - 0-0 o o q>-

0 0 20 40 60 80 100

Reed Numbers

2 3 4 5 6

Rating Number for Dark-Bright

Figure 4.3-Timbre Ratings for {R} B MGC Oliv VT3.51 B M 2 0 1 8

6 ■Q~

■(D-QQ QQ" q j I Q Q "2 I?

? 4 (DO ( 30- ■0— 0 4 [D

w ■O-OCD

2

0 0 20 40 60 80 100

Reed Number

19 > «r I

2 3 4 5 6

Rating Number for Soft-Hard

Figure 4.4-Strength Ratings for {R} BM G C Oliv VT3.51 BM 2 0 2 8

6 e» Q- oc

OQQ-O

4 - J-

Q Q Q &

2 ■ ^ ■ O O Q

■œ

0 0 20 40 60 80 100

Reed Number

3 4 5

Rating Number for Buzzy-Fluid

Figure 4.5-Noise Ratings for {R} BMGC Oliv 8 203

7

6 (D an >QQ a e —o ù Of 5 ■ee-

? 4

3

1

0 0 20 40 60 80 100

Reed Number

& 5 ; O" 41 L

3 4 5 6 7

Rating Number for Squeaky-Stable

Figure 4.6-Stability Ratings for {R} VWoren, V 12 (4) BM GC O liv V{3 .5) BM 204

■a Q C» - C '

20 40 60 80 100

5 y b.L

12 3 4 5 6 7

Rating Number for Veak-Full

Figure 4.7-Projection Strength Ratings for {R} 205

QUAL DA-BR SO-HA BU-FL SQ-ST WE-FU CASES 80 80 90 90 90 80 MINIMU 1.000 1.000 1.000 1.000 2.000 1.000 MAXIMU 7.000 7.000 7.000 7.000 7.000 7.000 RANGE 6.000 6.000 6.000 6.000 5.000 6.000 MEAN 4.350 4.050 4.556 4.189 4.756 4.200 VARIANC 2.661 1.947 • 1.800 2.627 1.917 1.858 ST. DEV 1.631 1.395 1.342 1.621 1.385 1.363 STD. ER 0.182 0.156 0.141 0.171 0.146 0.152 SKEW -0.259 -0.146 •0.077 -0.149 -0.245 -0.064 KURTO -0.487 -0.843 -0.590 -0.805 -0.928 -0.398 SUM 348 324 410 377 428 336 C.V. 0.375 0.345 0.295 0.387 0.291 0.325 MED 4.000 4.000 4.500 4.000 5.000 4.000

Figure 4.8-Summary Statistics for {R}

QUAL DA-BR SO-HA BU-FL SQ-ST WE-FU

QUAL 1.000 DA-BR -0.399 1.000 SO-HA -0.131 -0.284 1.000 BU-FL 0.623 -0.298 ■ -0.011 1.000 SQ-ST 0.455 -0.102 0.019 0.431 1.000 WE-FU 0.731 -0.593 0.092 0.534 0.309 1.000

NUMBER QF OBSERVATIONS: 70

Figure 4.9-Pearson Correlation Matrix for Performance Descriptor 206

As a rule of thumb, a .7 (or 70%) correlation coefficient is accepted as a threshold for true correlation in the physical sciences. The soft sciences accept lower thresholds (we have seen some psychologists present data claiming that

.3 correlation coefficients are evidence of correlation!). In this document, since the reed material is of biological origin, we shall not be bound so closely to the exact correlation threshold requirements of the physical sciences, but we shall not be so loose as to generate vague associations. We shall hold to the requirement that a .5 correlation coefficient is the minimum threshold for true correlation in this document.

Under this criteria, the correlation matrix given above is particularly interesting. There appears to be a mild correlation (-.399) between the timbre of a reed and its rating. Since there is a negative sign, this implies the correlation is the lower the DA-BR value (i.e., the darker the reed), the higher the quality rating. There is no significant correlation between reed strength and rating, although there is a slight trend to associate harder reeds with higher rating.

The strong correlations of the last three descriptors indicates that whatever psychological processes are used to arrive at a rating for a clarinet reed, these three descriptors are given high significance. The degrees of buzziness of the reed is over the threshold for true correlation, which means that it is likely that if a reed sounds buzzy, it is automatically given a lower rating. This is interesting, in that, because of reed shape differences, high performance reeds are less prone to asymmetric torsional motion (the principle cause of reed buzz) than student quality reeds. It is the correlation between buzziness and rating, and 207

not the quality of cane, which is why these reeds are regarded as inferior.

Reed buzz is primarily a function of transverse (side-to-side) structural factors, as we shall show in chapter five. Squeakiness, excluding the sudden

loss of lip damping (as expiained in chapter five), is primarily a function of

longitudinal (lengthwise) structural factors. It also shows a strong, but slightly smaller correlation with reed quality. The more stable (less likely to squeak) a

reed sounds, the more highly rated it is. It is amazing that this correlation isn’t

higher, but since the mean is of the stability pair descriptor is so skewed towards “stable” in the test set, there probably were not enough squeak-prone reeds to see the correlation fully.

The depth of the sound produced by a reeds is the most highly correlated of all of the performance descriptors with reed quality. The correlation coefficient is over the .7, which is the threshold for meaningfulness the physical sciences.

This means that, in any given population of reeds, the most highly prized property of a reed is its ability to sound full throughout the range. Unfortunately, although there are some interesting one-dimensional theories, there is no clear understanding of the coupling of reed acoustics to air column acoustics (chapter five). The problem is complicated by the existence of a mouthpiece, whose

design invites highly non-linear fluid mechanical responses. We shall outline the problem in chapter five and offer a partial solution.

In terms of the relationship between descriptors, we note two. There is a negative correlation between brightness and fullness of sound (meaning a darker sound is correlated with a fuller sound) and a positive correlation 208 between a fluid sound and a full sound. Apparently, the same process which

leads to a fluid sound also allows for a larger power output from the reed

(probably fewer energetic loss factors, such as higher degrees of freedom, i.e., energy being siphoned into torsional modes, friction, etc.).

Since the last three descriptors are so highly correlated with reed quality, we may ask if there is an equation which will allow one to predict the reed quality from a measurement of these descriptors. If we assume a simple linear regression model:

QUALITY = a [BU-FL] + p [SQ-ST] + Ô [WE-FU] + E

(4.1) then a least-squares fit of the data gives the coefficients: a = .265,a = .203, a =

.652, a = -.398, with an R2 = .635. The fit is not so good as we should like(R2 =

.635), but it is a beginning in delimiting the actual (and perhaps intuitive) criteria people use to rate reeds.

Mass/Weight Properties

Density is defined as the amount of mass (usually, kilograms, kg, in the

System system of measurement), per some unit volume (usually, cubic meters, m3, in the S.i. system):

p= m / V (4.2) 209

By taking a small parallelplped of clarinet reed material and measuring its weight using a microbalance, a representative density of the reed material was computed to be 461 kg/m 3. In the older, but still used cgs system, this is 0.461 g /c m 3 . Summerfeld arrived at a slightly higher value of 0.5 g/cm 3, although he probably used only an approximation for his work. Bamboo, by contrast has a much higher density of 660 kg/m 3, although the extensional modulus in the plane of the vascular tissue is in the same range asArundo donax. This moderate density yet high modulus ( for a plant material) may account for some of the characteristics which makesArundo donax unique as a reed material.2

The empirical value quoted above is obviously representative of the plant material on the average because of the effects of: 1 ) material inhomogeneity

(i.e., different cell types are in different concentrations in different regions of the material), 2) pressure and temperature dependency (these measurements were made at ambient room temperature (= 24 °C) and ambient pressure (= 30 mm/Hg)), 3) the age of the stem (it is impossible to tell in what stage of development the reed material was when harvested).

The test was repeated with two other samples, which yielded densities of 497 kg/m3 and 412 kg/m3, respectively. To be safe, a variation of ±10% should be allowed in the total density range of the material. This may seem to be a wide variation, but given the biological nature of this material, such a range seems prudent.

To study the variation of reed mass, prior to distributing the reeds to the student players to be rated, the masses of the reeds were measured on a 210 standard laboratory (Mettler) microbalance. A plot of the mass measurements for all of the reeds is shown in figure 4.10. Selected statistics are given in figures 4.11.

N OF CASES 80

MINIMUM 0.682

MAXIMUM 1.018

RANGE 0.336

MEAN 0.862 (gr)

VARIANCE 0.005

STANDARD DEV 0.074

STD. ERROR 0:008

SKEWNESS(GI) -0.063

KURT0SIS(G2) -0.316

SUM 68.988

C.V. 0.085

MEDIAN 0.860 (gr)

Figure 4.11-Mass Statistics

A breakdown of the mean by brand is ( all measurements ± .005 gr.): Vandoren

V I2 (4) = .876 gr, BM = .752 gr, GC = .896 gr, Oliv = .891 gr, V I2 (3) = .856.

The mass data shows some interesting variations both within and between 211

Vamdoren V12 [A] BM GC O liv V[3.5] BM 1.1

o o 1.0 0 n-f 0 0 ^ ■a o o n o <0 % « CD Ch ^ 0.9 c _ o 0 % ° ? ° % O 0 p % n ° 1 0 °o °°o° „ o o o ( > O S 0.8 (S> z 0 ° 0 <1 (1 0 %

0.6 20 40 60 80 100

Figure 4.10-Mass Measurement Summary for {R} 212

brands of reeds. The Vandoren subset shows a difference between the

average mass of the two different reed strengths (3 1/2 and 4). This is

significant in the reed selection process, since most reed companies, we

assume, grade the strength of reeds according to tip bending resistance and not

simple reed mass.

The difference in reed strength(3/12 and 4) in the Vandoren sets above

implies that for some reason there is a modest correlation between tip bending

strength and reed mass. One may be tempted to simply conclude that whatever

bending tests the Vandoren reeds were put through to grade them simply

discriminated between thicker and thinner reed tips, thus accounting for the

greater mass of the strength 4 reeds; however, this assumption would be wrong,

as the Vandoren 3/12 strength reeds in S2 show almost identical mean tip thickness as those of strength 4 (strength 3 1/2 = .135+ .013 mm, strength 4 =

.134 ± .009 mm).

In fact, a large portion of the mass difference comes not from the tip, but from a slightly thicker midsection for the strength 4 reeds (that the heel thickness is not a contributing factor is reflected by the fact that the heel thickness of the two strengths are virtually identical: strength 4 = 3.136 mm, strength 3 1/2 =

3.138 mm). For strength 3 1/2 reeds the midsection (measured in the center at

16 mm from the tip) thickness is 1.114 mm, while for strength 4 it is 1.346 mm. It is probable that it is the increase in midplane bending resistance of the thicker strength 4 midsection which largely accounts for the difference in reed strength

rating. Such a large thickness difference in the midsection also is much more 213 likely to account for the mass variation seen between the two strengths than a tip thickness difference might.

The interbrand mass differences are also striking. The Black Master reed set, having a 3 mm shorter length, is correspondingly lighter in mass (.752 ±

.047 gr). In fact, all of the Vandoren reeds are lighter than any of the other reed brands measured. This difference may be accounted for simply by thicker heels, etc. in the other reeds.

A Pearson reduced correlation matrix of the mass to quality rating descriptors is given in figure 4. 12. The complete matrix of coefficients is not given because the descriptor coefficients are given in the submatrix in figure 4.10.

Mass Mass 1.000 Qual -0 .1 4 2 DA-BR 0.190 SO-HA -0.131 BU-FL -0.2 7 6 SQ-ST 0.068 WE-FU -0 .0 8 7

Figure 4.12-Mass and Quality Descriptor Correlation Matrix

There is, apparently, no significant correlation between the mass of a previously unused reed and the quality of that reed. The modest negative correlation with mass and buzziness is intriguing and complicated to explain (we shall do so in chapter five). 214

So far, we have talked almost exclusively about mass, when we would really like to include information about reed density, especially as it changes over the reed surface, due to different amounts of vascular tissue in different regions.

The problem is that due to the irregular shape of the reed, it is difficult to measure density of the reed precisely for any given reed, let alone portions of the reed.

In an attempt to grapple with this matter, the author was led to the Campus

Electron Optical Facility in the medical school at The Ohio State University (the author is grateful to Dr. Richard Swenson in the department of zoology for this suggestion). At this facility there is an optical densitometer, which is a device used in engineering, primarily, to measure density variations on the surface of a metal. We have'adapted it here to look at density variations on the reed surface.

The idea is that light passes through a material based upon the density of the material, so that a recording of the percent transmission of light through the material should correlate with the thickness or local density of the material. In the case of a metal, one substitutes electrons passing through the metal and uses, instead of an optical sensor, as we used, an electron detector. Both techniques are handles by the same computer program. We used light because this allowed one to see the sample, and required nothing more than a specialized camera and video detection equipment. A sample optical densitometry plot is shown in figure 4.13.3 The plot, of an oboe reed, shows the unmistakable density variations induced by scraping the reed. The blow-up of 2 1 5

Figure 4.13-Contour Optical Density Map of Oboe and Clarinet Reed 216 the tip of a clarinet reed in figure 4.14 shows a very light transparent vascular bundle.

The computer also allowed for cross-sections along a line to be plotted. In figure 4.15 we show such a cross-section. In this case, the vascular bundles are quite easy to pick out as the depressed areas of the graph (= lower transmission values).

The reason that this technology cannot be used to quantitatively study density variations within a reed, however, is that while there is a linear relationship between density and light (electron) transmission in metals, the exact relationship between light transmissibility and the reed material is unknown, and expected to be nonlinear. In addition, the signal to noise ratio is quite low, resulting in the vascular tissue, which is darker than surrounding tissue, looking much thicker than surrounding tissue, when in fact it is merely darker (due to greater lignification). These considerations, plus problems with scattering and internal reflectance, introduce errors which make this a method for qualitative use only. Nevertheless, as a pedagogical tool for teaching scraping techniques, it is unrivaled, as the plot of the oboe reed suggests.

Effects of Heat on the Clarinet Reed

Many external agents, such as heat, light, and oral environment attack the clarinet reed over time, and it is worth spending some time examining some of these activities. There are many folktales associated with the effects of these agents, as we shall point out, and hopefully the information presented in this 2 1 7

: DMEl Int:18 Artt:B,8 512,512 . Gfi£Y: 512 HAG r Ix CflH N T = 8 SCALE 1:1.5234

? g

: :'0

Hoii«: zooa ptn zoom doun up

Figure 4.14-Optical Density Map of a Clarinet Reed Tip 2 1 8

Figure 4.15-Cross-sectional Optical Density 219 section will help to correct the material passed on from teacher to pupil about the preservation of reeds.

Heat, within the normal physiological (32-37 oC) and ambient temperature ranges has almost no deleterious effects on the normal clarinet reed matrix.

The results of both TGA (recounted in chapter three) and TMA (Thermal

Mechanical Analysis, see below) tests show that there is little or no measurable chemical degradation and nominal dimensional changes of the reed material within this temperature range. Temperatures in more "hostile" environments do have pronounced effects on both the dimensional stability of the reed and the chemical integrity of the reed matrix. Elevated temperatures within the range of

40 oC-100 °C are the most common temperatures normally encountered by clarinetists in the course of deliberate reed work (such as in scraping and filing the reed, which can induce thermal hot spots on the back side of the reed well over 50 °C due to friction, or as in soaking the reed in nearly boiling water to clean the reed).

If the heating is done while wet instead of dry, studies on wood reported by

Fenger and Wegener imply (and we assume that the same qualitative behavior applies to Arundo donax) that the cellulose matrix, instead of decrystallizing, will actually recrystallize, and the percent crystallization will be prevented from decreasing as in the case of heating dry wood.4 In addition, the presence of water will prevent chain-splitting or "scissoring" (i.e., cutting) of the cellulose polymer within the range of 120 oC-160 oC, thus preventing chemical breakdown. Beyond these temperatures, the cellulose begins to decompose 220 into elementary furans, becoming amorphous in structure.

While lignin in wood is generally stable to 200 oC, nevertheless, the hydrogen bonds which are loosely attached to the molecule begin to breakdown at 60 °C-80 °C, and between 100 oC-180 oC the aromatic quality of the molecule is lost.s In addition, lignin softens in this temperature range

(bamboo lignin, for instance, which is surely similar to that of Arundo donax, softens at 162 oC).6

Heat in this hostile temperature range also has profound effects on the dimensional properties of the reed material. A material property, the coefficient of thermal expansion, is defined as the volume change per unit degree of temperature change of the material, and measures these effects:

a = 1/V(aV/3T)p

(4.3)

In order to measure the coefficient of thermal expansion in the various axes of the processed plant material, small samples of rectangular cross-section

(approximately, 5 mm in length) were excised from several clarinet reeds and submitted to Dr. James Culbertson of the Ohio State University department of dentistry and his postdoctoral assistant. Dr. Alka Thakur, for testing. The reed samples were cut from different orientations of the reed material, so that the affect of temperature on the expansion of each of the three standard material plane (L, R, and I , see below for nomenclature) could be studied.

A word needs to be said about the natural axis in cylindrical plant stems.

Engineers identify three such axes (unfortunately, botanists use a different 221 scheme using similar names, and it is often confusing to know which is which):

1) the longitudinal axis which is parallel to the direction of growth, 2) the radial axis, which is directed inward, in the direction one might push the long end of a razor blade from outside of the epidermis (in the direction of elongation) to the inner, soft tissue, 3) thetangential axis, which is the axis one exposes when the top is cut off of the stem. These orientations are summarized in figure 4.16.

A computer-controlled test device called a thermomechanical analyzer

(Du Pont TMA 2940 Thermal Mechanical Analyzer), which measures the change in length of a sample (in microns) with increasing temperature, was used to determine the thermal expansion curves as a function of temperature.

The results are presented in figure 4.17. Unfortunately, during the test, we used the biological instead of the engineering axis nomenclature, so that the plots are in biological nomenclature and the results below are in engineering nomenclature. The correspondence is: transverse(b) = longitudinal(e), tangential(b) = radial(e), and radial(b) = transverse(e). The expansion coefficients (engineering nomenclature) at room temperature in the physiological temperature range (25 oC to 35 °C) for the three axes are:

Longitudinal 18.3 pm/m oC (-7.90pm/m oC) Radial 33.5pm/m oC Transverse 65.2pm/m oC

Figure 4.18-Coefficents of Thermal Expansion for Arundo donax

The longitudinal, radial and transverse directions all show expansion upon 222

Fiber

/

Longitudi nal Radial Tangential

(note; the TMAsampiea usethe biological nomenclature instead of engineering, ao that longitudinal = tangential, radial = radial, and transverse = tangential)

Figure 4.16-Stem Orientation 2 23

Sample: REED-TRANSVERSE F ile : C: REEO l.007 sue: 3.1710 mm TMA O p e ra to r: ALKA THAKUR M ethod: TMA FOR REED SAMPLE Run D a te: 1 4 -A g r-9 5 18: 27 Comment: THICK SAMPLE

oH 4 1 .5 8 % o— 8.01pm/ffl*C I

5 “5"

- 1 0 4

-1 5 -5 0 50 100 150 200 250 Temperature (%) TMA V5.1A DuPont 2100

Figure 4.17-Thermal Expansion Curves for Arundo dona) Figure 4.17 (cent.) 2 2 4 Sample: flESD-TANGENTlAL (A) rile : C: REEDl.004 Size: 1.7300 mm TMA O p e ra to r: ALKA THAKUR M ethod: TMA FOR REED SAMPLE Run D ate: i4 -A p r- 9 5 15: 46 Comment: THICK SAMPLE

0 -

- 1 0 - 1 5 0 .2 2 *0 a"-37.9pm/m*C

-2 0 -5 0 50 100 ISO 200 250 Temperature (*C) TMA V5. lA DuPont 225 Figure 4.17 (cont.)

Sample: RSEO-AAOIAL F i le : C: REEOl.OOG Size: 2.2130 mm TMA Operator: ALKA THAKUR Method: TMA fo r AE50 SAMPLE Run D a te : iA -A p r -9 5 17: 34 Comment: THICK SAMPLE

0 -

- 1 0 -

cn - 2 0 -

-50 50 100 200 250 Temperature ("Cl TMA VS.lA OuPont 2100 226

heating. The radial and transverse axes, which have less structural strength

(lower moduli, etc.) than the longitudinal axis, show the effects of heating to a

more pronounced extent than the direction of axial growth (the longitudinal

axis). This is to be expected, as heating is a form of stress, and the wood reacts

as with any applied stress, showing greatest resistance to stress in the

longitudinal axis.

The interesting observation here is that at physiological temperatures,

Arundo donax actually undergoes a brief contraction period (for reasons which

are still not clear). This value is shown in parenthesis in figure 4.18. This should not be interpreted to mean that the material actually contracts inside a

players oral cavity, because the expansion induced by hydration due to saliva

more than compensates for the small contraction seen due to thermal effects.

All three graphs show the same characteristic three regions of activity from

0 oC to 250 OC. In region one, approximately 0 oQ to 50 oQ, the material expands

roughly linearly as many other materials do. In region 2, approximately 50 oQ to

150 oQ, the material loses water of cellular hydration, and begins to shrink. Each axis shows a drop of about 13 pm in size, which amounts to a .4 % (L), .75 %

(R), and .63 % (T) for each of the three axis. In region three, approximately 150 oQ to 250 OQ, the water had been removed, and the hemicelluloses begin to break down into elementary furans. The removal of that amount of hydrogen bonding which was due to the presence of water allows the cellulose/lignin lattice to re-expand outward a bit. Consequentially, the material begins to expand at a rather rapid rate. 227

The mechanical properties of Arundo donax change in response to heat.

Bodig and Jayne suggest that the effect of a moderate (± 100 oC) change of

temperature on the moduli of elasticity is to cause them to decrease for

increasing temperature and decrease for decreasing temperature.^ This

relationship may be expressed, in many cases, by a simple linear equation of

the form:

(E or G),2 = (E or G)ti [1+ (3i,n (t2-tl)] (4.4)

where E and G are the extensional and bending moduli, respectively at either the higher ( subscript, t2) or lower (subscript, t1) temperature in Celsius, and Bn

is an empirical coefficient for each of the different types of moduli and their

assooiated directions: 1 = extensional, 2 = bending, in each of the different

directions, n= L, R, T. Ba.L, for instance is the constant which measures the change in bending modulus in the longitudinal direction. Typical values of B

are on the order of .001 to .01 (mostly near .001), and the sign is negative, since the effect is to decrease the original modulus value with increasing temperature.

Using this equation, it is easy to see from the data supplied by Bodig and

Jayne for wood (and we assume that the range of values is roughly the same for

Arundo donax) that the temperature change from ambient room temperature

(20 oQ) to oral cavity temperature (37 oQ) causes at most a 17% change in the

tangential modulus, and only a 5% or less change in the modulus in the

longitudinal direction. By contrast, the effeot of water causes changes almost

five times as great. 228

The Effects of Light on the Clarinet Reed

Light in the visible range of the spectrum induces color changes in plant material, which is dependent upon the moisture, temperature and chemical

components (nitrogen, etc.) of the atmosphere. Arundo donax, as mentioned in chapter one, changes from a green to a golden color due to the conversion of chlorophyll (green colored) to a different chromophore (a chormophore is a color generating compound), probably xanthophyll. Most reed growers strive to achieve a golden yellow color in the sunning process. The precise colorimetric changes may be monitored by UV-Visible spectroscopy, although the author has not done so, since this would require cultivating reed material.

It is not obvious if reed color by itself is a good indicator of reed quality.

Since the reed color may be precisely determined by the UV-Visible spectroscopic methods mentioned in chapter three, at least the interior material color (the epidermis color is variable due to the presence of tannin), it is possible to test this hypothesis with actual clarinet reeds. This is an experiment which still needs to be done.

In the case of wood, the penetration of UV radiation is only 75 |im and

200pm for visible light.s If this is also true forArundo donax (a permissible extrapolation, we expect), then this means that the tip of the clarinet reed

(approximately 100pm thick) is the most vulnerable to degradation by light; however, generally either long exposure times, or high radiation levels, or special atmospheres are required to induce rapid change in the cellular 229 structure, with cell wall shrinking and microcracking the most common result. It is doubtful (although not tested) that short-term, low level radiation from light bulbs or fluorescent lights will have any significant affects on the reed cellular matrix. The effects on the ionizing radiation may be influenced by moisture content and temperature, and is a matter which needs study.

With high levels of UV radiation, some changes in cellulose may occur, such as a loss in weight, a lowering of the Dp, or a change in the cellulose components (a loss of a-cellulose, for example), accompanied by the formation of many free radicals (again, we may assume that the results transfer to Arundo donax to some degree).^ The in weight loss is even more pronounced if the sample of isolated cellulose is hydrated.

The presence of lignin seems to protect the cellulose matrix, however, since, lignin is ultraviolet active and preferentially absorbs the UV radiation, mainly in the chromophoric sites of the molecule, such as phenyl hydroxy groups, double bonds, and carbonyl g ro u p s .10 Again, with the relatively high lignin content of Arundo donax, one may assume that short-term exposure of ambient radiation (light bulbs, etc.) does little to damage the reed material compared, at least to many other more severe chemical effects.

The Effect of Saliva on the Clarinet Reed

What are the chemical effects likely to be observed in the reed/oral cavity environment? We have already speculated that the hemicellulose is gradually leached from the reed material, probably due to the alkali content of saliva. 230

Salivai pH varies depending upon the rate of salivai flow from a pH of approximately 6 (no flow, a condition known as xerostomia) to a pH of approximately 8 in full flow.n The pH is sensitive to flow rate mainly because this determines the amount of salivai bicarbonates in solution (since bicarbonates generally raise the pH into the alkaline levels in solution). The faster the flow rate, the more bicarbonates are delivered into the saliva stream.

Since saliva pH is dependent upon flow rate, several observations are

possible. In the course of a concert, if one suffers from dry mouth (xerostomia), the salivai pH is expected to be below 7, and be therefore, acidic. A complex buffering system in saliva, again, partially due to the presence of bicarbonates, keeps mouth pH from falling below 6 and eating though the oral cavity. Since the least change for the reed is probably induced in the pH range of neutral water (pH 7), perhaps using a salivai flow stimulant at concerts would help increase reed life.

Nervousness, often induced by concert playing, besides leading to xerostomia, may also lead to rapid breathing, which tends to flush CO2 out of the system, producing a condition called respiratory alkalosis (whereas holding one's breath leads to an increase in CO2 and respiratory acidosis).

The drop in CO2 reduces the amount of available bicarbonates (which are synthesized from carbon dioxide), lowering the saliva pH.

By contrast, during a rehearsal, when a player is salivating and also partially holding their breath ( increasing the amount of CO2) during the course of playing, the saliva pH is expected to be slightly alkaline. Thus, the mouth pH 231 varies considerably during the playing life of a typical clarinet reed. One would expect, on average, that rehearsal conditions would be encountered more frequently than concerts, so that the mouth pH stays slightly alkaline most of the time (the author has never had the chance to test this hypothesis, but based on the arguments presented above, it seems plausible). This slight degree of alkalinity probably enhances the leaching of hemicellulose from the reed matrix

(hemicelluloses are soluable in mild alkali solutions). The affect of these changes in pH on the cellulose matrix and lignin is relatively slight.

As pointed out earlier, of more importance than simple saliva pH are the contaminants in saliva which infuse into the reed matrix. The presence of amines in the infrared spectra of spent reeds indicates the presence of epithelial cells from the mouth as well as glycoproteins contained in saliva: salivary mucosins (MG1, MG2), Proline-rich Glycoproteins (PRG), a-amylases (a-Am), peroxidase (Px), Carbonic Anhydrase (Anh),

Fucose-rich glycoproteins (FRG), Immunoglobulin (IgA), Kallikrein

(KIk), Lactoferrin (Lf) and Fibronectrin ( F n ) .i2

Several other chemical barriers to water absorption, in addition to the loss of absorption due to the decrease in hemicelluloses, are created due to the evaporation of saliva on the reed surface. The existence of hydrophobic (water avoiding) pockets in MG1 (mucin glycoprotein 1) has been demonstrated, and is one such barrier (playing on a reed while having a cold which causes a great deal of mucous discharge from the throat and lungs, will very quickly render a reed less waterabsorbing).13 232

In addition, oral bacteria quickly remove the sugar containing elements of the glycoproteins of saliva once it leaves the mouth and rests on the reed surface

(the bacteria themselves are yet another source of reed deterioration to be discussed momentarily).The removal of sugars from saliva results in a protein precipitate which is rich in calcium and of low solubility.is

Finally, a-amylase, among other glycoproteins, contains significant levels of bound sialic acid, which is quickly released from the salivai solution by enzymatic hydrolysis. Once released it forms a water insoluble precipitate.16

Combined, the barrier presented by these salivary artifacts may be quite formidable, and may keep the reed from fully hydrating as it ages, resulting in higher stiffness and brittleness.

Clarinet Reed Degradation

The effects of heat, iight and oral environment all contribute to the breakdown of the clarinet reed. So far, the author has identified five different major routes by which the reed deteriorates over time in a typical playing situation. These breakdown pathways are summarized below. It should be pointed out that remedies or preventives potentially exist for each of these deteriorative processes, and we shall comment on each in course.

To begin with, contrary to foiklore, the clarinet reed material does not wear out as it ages. A comparison of pristine and spent reeds show the same number of microcracks in the cell wall matrix of both. Under SEM no large tears were observed in either type of reed. Any mechanical degradation seems to be 233 limited to the slight decrystallization of cellulose in the reed matrix mentioned earlier.

Chemical breakdown is of three types: 1) loss of hemicellulose due to alkali leaching of saliva (which, although a type of breakdown, is responsible for the

“breaking in” of the clarinet reed, as mentioned in chapter three), 2) contamination of the reed due to salivai artifact (amines, glycoproteins, sialic acid, etc.). These breakdown pathways render the material, in effect, more brittle and less hydrophyllic, and, 3) in addition to leaching hemicelluoses and coating the reed, saliva contains ammonia and other alkali, and these have been shown to induce plasticization in wood. The result is that the modulus of the material falls (the reed becomes “softer”), and the damping increases (the reed does not maintain sound easily).

These three process, as well as the mass increase due to the contamination of the saliva and oral microflora, change the vibrational charateristics of the reed, as we shall show in chapter five.

Finally, while examining several clarinet and saxophone reeds under the scanning electron microscope, the author noticed a single strain of bacteria which lined some of the vascular tissue and parenchyma cells (figures4.19 -

4.22). The size of the bacteria proved to be on the order of one micron and was coccoidal (spherical) in shape (figure 4 .1 9 ).

Heinrich points out in his dissertation on bassoon reeds that he observed a type of bacteria which infused itself into reed material as ita g e d j s The author maintains that both bacteria are the same. 2 3 4

«

Figure 4.19-Shape of S. Epidermitis and its Migration in Reed Pit Cells 235

Figure 4.20-Bacterlal Deposition on Inner Xylem Wall 236

Figure 4.21-Bacterial Biomass 237

00^5 15KU X6,0I00 tN«i WD14

Figure 4.22-Bacterial Sheet (Heinrich’s “Fungus”) 238

Typically, the bacteria enters though the pit cells of the xylem (figure 4.19) and deposit themselves there via a glocosidic lining (figure 4.20). The bacteria grow to considerable biomass, eventually forming thick sheets on the reed

(figure 4.21) which resembled a whitish coating (figure 4.22).

The effect of the bacteria on the clarinet reed is three-fold: 1) it adds mass to the clarinet reed tip, 2) it reduces the range of motion of the torsional and bending modes, and 3) it changes the overall shape of the reed tip. Contrary to

Heinrich’s assertion, this particular bacteria does not harvest the reed matter for food (it is one of the types of bacteria which gets its nutrients through the extraction of sugars in saliva, as we mentioned earlier), and so the degradation caused by the reed is on the physical response of the reed and not on the material itself.

This bacteria has been identified for the author by pathologist Dr. Leona

Ayres of The Ohio Sate University Hospitals/department of Medicine. She did several detailed wipes of both pristine and spent reeds in order to recover and culture bacteria for identification. The pristine reeds were almost entirely bacteria-free (to the relief of the author and reed players everywhere), or contained a small number of harmless and non-growing bacteria, at least one of them being of a soil-borne variety, as the author had suspected. The pristine reeds also showed minute traces of some common oral microflora. None of these bacteria either grow in or interact with the reed matrix.

By far the largest bacteria type present, and the identity of the bacteria which the author found in the SEM study (and we suspect of Heinrich’s as weil) is 239

Staphalocccus Epidermitis -the common bacteria found, as the name implies, on the mouth and hands of virtually everyone (the bacteria is innocuous). Only this type of bacteria is of any consequence in the breakdown of the clarinet reed vibrational life, as no other common bacteria grows under typical playing conditions.

As mentioned above, S. Epidermitis is one of the bacteria which removes the sugars from saliva (causing the deposition of sialic acid as a by-product) as it feeds on these salivai nutrients. The bacteria becomes dormant without water, and so, one way to slow down the growth of the bacteria it to keep a reed cool and dry. This will not kill the bacteria, however, but only arrest their growth.

Because the bacteria attach to the cell walls through a glucosidic monolayer, even if one could kill the bacteria by a strong enough preparation of alcohol or hydrogen peroxide which some players use to sterilize their reeds (preparations which themselves also harm the reed cell walls), neither would fracture the glucosidic monolayer. The result would be a collection of dead bacteria within the reed, with more to appear with each new use. The end result is no gain for the player. What is needed is either a bacterial resistant coating for the reed or an antibacteria agent which is non-alcohol based. Fortunately, both of these already exist or soon will, and only need to be developed.

Towards the Prevention of Reed Degradation

There is little one can do to salvage a reed once salivai glycoproteins and bacteria have been deposited. Prevention is therefore the best alternative. 240

A water-based antibacterial rinse for the reed is needed (alcohol rinses may cause dehydration of the cell wall matrix). Dr. Ayers, mentioned above, and the author have such a candidate (this candidate is a rinse already currently on the market, but which requires a prescription), which not only kills the bacterial, but breaks the monolayer attaching the bacteria to the reed. As the rinse must be made in non-prescription strength, and pass an FDA review, no further comment on its effectivness is warrented in this document. Our purpose here is merely to state the nature of a proper solution to the bacterial contamination problem, in contrast to home remedies currently being used by some players.

We have not tested commercial preparations made specifically for reeds, and cannot comment on their effectiveness, although, from what the author has been told of them, most do hot appear to deal with the bacterial attachment problem.

The second method to prevent reed degradation is a rather novel approach, which, if successful, should be a great help to the clarinet community. Certain polymers exist as solids at room temperature and pressure, but are such that at slightly elevated temperature quickly enter the gas phase. By placing a material in a chamber and introducing the polymer in the gas phase, the polymer may be caused to coat the material (this process is called gas phase infiltration).

Heart valves and electronic components are coated in this way.

A particularly inert polymer, parylene, has the above properties and is often used for gas phase deposition of material, such as heart valves and rare books

(in fact, the United States Library of Congress uses this polymer to coat its rare 241 books).19 Because the material is deposited as a gas, the thickness deposited is quite small (about lOpm). It is virtually invisible, impervious to mild chemical and bacterial attacks, and already has the approval of the FDA.

In the author’s opinion, this polymer material has the potential to change the reed industry, since the reed remains somewhat water absorbing, yet chemically inert. Since it is applied in a large chamber, it may be applied to a bulk quantity reeds at one time. The only thing not known at the present time is the amount of material which may be deposited without changing the reed vibrational characteristics, but it is hoped that such research will be finished soon.

The only current proper retardant for reed degradation other than those still experimental ones mentioned above (and excluding some commercial preparations for double reeds mentioned above which the author has not had the advantage of testing and whose effectiveness is unknown) is to keep the reed in a cool, dry place, such as a refrigerator (not a freezer) if condensation is not a problem, otherwise on a cool, dry shelf or reed holder with a desiccant

(calcium sulfate, for example). Until more modern solutions have been perfected and researched, current rinses and ultrasound treatments, etc. should not be expected to restore a spent reed, nor prevent degradation.

Reed Hvdroscopic Properties

The clarinet reed is rarely played in a dry state. Since the amount of water saturation in the cell wall matrix when the reed in the oral cavity is quite high. 242

the influence of water and saliva on the reed structure is crucial in

understanding the subsequent vibrational characteristics of the clarinet reed.

There are two types of water typically stored in a clarinet reed: 1) the water of

hydration of the cell wall matrix, i.e., the more or less permanent water which is

bound to the cellulose-hemicellulose-lignin matrix (this is called, after Bodig,

bound water) and 2) the ephemeral water loading within the cellular cavities

and vascular tissue capillaries during salivai hydration during playing (this is called, free w a te r ) .20

In order to determine the amount of water bound to the cell wall matrix, thirty

reeds in {R} were oven dried using an oven at 110 oC in the Department of

Chemistry at The Ohio State University for two hours (the author especially thanks Ruth Anderson, who is in charge of the undergraduate analytical laboratory for her help). Ten of these reeds, in addition, were further heated overnight at 60 °C oven to make sure that the original two hour drying was sufficient to reach a constant reproducible dry weight (it was, as the weights measured overnight and after the first two hours were virtually identical). The reeds were weighed before drying, and after drying. A plot of the masses in grams is shown in figure 4.23. A plot of the mass percent of bound water (the so-called water content, Gwa) is also given. The average of the Gwa is 5.98 %.

To determine the amount of free water, the author weighted three pristine reeds and then placed them in his mouth for a length time equal to that encountered during warm-up prior to playing. The reeds were not "dripping water”, but were moist, which is the condition in which most clarinet reeds are 243

L

0 .8 -

0.7 0 10 20 30 Reed Number Mass, Raw (grs.) Mass, Dry (grs.)

7

6

& <9 5 u

4

0 10 20 30 Reed Number Gwa

Figure 4.23-Clarinet Reed Water Capacity 244 maintained during playing. The reeds were then immediately reweighed. An average mass change of 3.75 % occurred due to this hydration. This may seem small compared to the percentage of hydration of the reed cell wall matrix, but the larger surface area of the cell wall surface, and the proximity to hydrophillic agents accounts for this difference. The overall hydration of a typical reed from these two sources is 9.73 % or about 10 % by mass.

Mechanisms of Hvdration

The cell wall matrix becomes hydrated primarily due to the hydrogen bonding of water to the hemicellulose as mentioned in chapter three and allows the hemicellulose to bond effectively to the cellulose fibril. In addition, water is hydrogen bonded to the cellulose fibrils themselves in a pattern of cross-linking

(figure 4.24).

The partial pressure of water in the air constantly changes. As the atmospheric partial pressure increases, adsorption, the wetting of the material, is increased. When the partial pressure decreases, desorption

(drying) occurs. It is instructive to take a pristine clarinet reed and record the weight change of the unplayed reed once per day for a month to prove that the weight of the reed actually changes daily by a small amount each day due to the presence of larger or smaller amounts of bound water in the intercellular matrix as the atmospheric vapor pressure (related to the relative humidity of the air) changes (we shall not reproduce the results here). This moisture cycling does not occur smoothly, however, in both directions, and according to Bodig a 245

R H H--H H--H H--H H

jp :' iPx H H-H H-H H-H H

Figure 4.24-Water Adsorption on Cell Walls 246 hysteresis curve results (figure 4.25), in which the rates of adsorption and desorption are different (see reference 20 for details).

Shrinking and Sweiiino

The amount of shrinking (swelling) in the reed matrix as bound water is removed or added may be calculated from the thermal expansion curves given earlier, figure 4.17, yielding: .4 % (L), .75 % (R) and .63 % (T), as reported. We assume that the expansion of the material is negligible compared to the shrinking due to water loss, so that these values are almost entirely due to water loss.

Swelling of the reed in the presence of free water is an order of magnitude greater than that for bound water, but still shows anisotropy in the direction of the vascular tissue length. To determine swelling values, the thickness in each of the three planes was carefully measured for reeds taken from a new box of reeds (the reeds were not oven dried, since this would have changed the properties of the material slightly). The samples were then immersed in water and measured again. The degree of swelling was 4.5 % (L),16.8 % (R), 7.5 %

(T) for the three axes. The value for longitudinal swelling is an average of two different measurements due to the uncertainty in measuring the samples, and is more suspect than the other two values.

This anisotropy in swelling and shrinking is due in large measure to the fact that the cellulose microfibrils of the cell walls show a pronounced orientation in the longitudinal direction, and the displacement of the crystalline polymer lattice 247

H Z 111 H Z o Ü DESORPTION 111 OC D H

O ADSORPTION z

VAPOR PRESSURE

Figure 4.25-Sorptlve Hysteresis 248 of cellulose is resisted in this direction by the relative rigidity of the long chains of cellulose. Also, hydrogen bonding of water to the intercellular matrix occurs most easily in the hemicellulositic portions of the cell wall, and these are usually perpendicular to the cellulose lattice axis (the transverse and radial directions).

This anisotropy in swelling rates and extents is the cause of another phenomenon familiar to every clarinet player: reed warp. Actually, there are three types of reed warp which may be distinguished: 1) warping which occurs along the longitudinal axis of the reed. We shall call this l-warp, 2) warping which occurs along the transverse axis of the reed (i.e., side to side). We shall call this type of warping, t-warp, and 3) reed “wrinkling” at the tip. We shall call this w-warp. We illustrate these three types of warping in figure 4.26.

In order to study the effects of l-warp on clarinet quality rating, the heights of the reeds in experimental set {R} were measured at the center of the shoulder

(i.e., at the beginning of the epidermis). Since the reeds were bowed upward slightly (l-warp), the measurements were repeated again with finger pressure applied to compress the reeds sufficient to completely remove the bowing, thus leaving the reed completely flat in its underside. The results are plotted in figure

4.27.

After the reeds had been rated by the students, these measurements were repeated to see if the playing and moisture cycling of the reed caused any change in the degree of l-warping of the reed. The results of the second series of measurements are shown in figure 4.27. The percentage of warp (%W) of both the first and second set of measurements is shown in figure 4.28, and is SHOULDER 249

BACK FRONT T VAMP EPIDERMIS VAMP

\ UNDERSIDE WARPED FLAT REED REED

L-WARP

T-WARP HEEL

Tip Tip

W-WARP T-WARP (Greatly Exaggerated)

Figure 4.26-Longitudinal and Transverse Reed Warp Pristine 3.5 250

3 .4 -

3 .3 -

I 3.2 -

41 z

3.0 -

2.9 -

2.8 0 20 40 60 80 Shoulder (U) Reed Number Shoulder (C)

Played 3.6

3 .4 -

1 s 3 .2 -

4» z

3 .0 -

2.8 0 20 6040 80

-o Shoulder (U) Reed Number ♦ Shoulder (C)

Figure 4.27-Reed Warp Measurements, Pristine/Played 251 4

2

>

0

0 20 40 60 80 100

95 V (Pristine) Reed Number % V (Played)

Figure 4.28-% Warp for Pristine and Played Reeds 252 calulated by taking the differenced between the uncompressed and compressed reed and dividing by the value for the compressed reed.

The reeds definitely are more bowed (l-warped) by a small amount after playing. The readings indicates an average D value of .0278 mm between the compressed and uncompressed reeds before playing, but a .0389 mm difference after playing. The average compressed reed thickness before playing is 3.182 ± .005 mm, and after playing it is 3.576 ± .005 mm. (due to uncertainty in measurement of thickness, and the length induced changes due to moisture content, etc.). The warping accounts for a .874% dimensional change in the reed before playing, but 1.088% after playing. Thus, the reed appears to be more warped after playing than before. We shall discuss the effect of l-warp on player ratings in chapter five.

We have no data on t-warping (the effect is difficult to separate from l-warp, although it can be done) but this type of warping appears to be the major cause of reed squeak, as explained in chapter five.

Mechanisms of Water Transport in the Clarinet Reed

There are two different mechanisms by which water is transported along the longitudinal axis of the clarinet reed, and they depend upon the particular cellular cavities which are being filled up (figure 4.29). This is very clear in a video microscope study which the author has done of the motion of dyed water moving along the cells of the reed. By setting a 1 millimeter mark on the reed, it was possible to measure the total number of frames of video for the passage of Water Drop 253

Osec. îBBünoHBnnnnnnnonnnnnc Mnnnconnnonnnnnnnonnc ~ P«?nohym&Matnx

■■■■■DnCQCDDnonDODDCSDI .025 sec. ■■■■■□DDDDDDDQdnnQQDCt mODDDDDDDDDDDDDDDt

■■■■■■■■■■E3DDDDDDQDDDa .05 sec. lODDDDDDDDQDC IDDDDDDDOnDDC

.075 sec. £?N

.1 sec. I â t d ? r

Mechanism 1 : Diffusion Controlied Transport

Water Parabolic Velocity Profile \ / _____

— XylemCeli j r r ( p i - p a ) 8 ijL Mechanism 2: Capiiiary Controlled Transport

Figure 4.29-Water Transport Mechanism in Clarinet Reeds 254 water through the two different cell types (vascular and ground). For the ground tissue, it took 3 frames, at 1/30 second (= .1 seconds) to fill in the one millimeter mark (i.e., the water moves at a speed of 10 mm/sec.). For the vascular tissue, it took 53 frames (= 1.76 seconds or .57 mm/sec.) to fill in the one millimeter mark

(the motion of the water down the xylem is quite clear).

Parenchyma cell cavities are filled in very rapidly by what we take to be a diffusion-controlled process, from observation of the video of water hydration of the material. The governing equation is expected to be similar to a Fick's second law equation;

* ,4 .5 , where Ng is the mass concentration of saliva, and D is a dimensionless number called the diffusion coefficient.

Chirkova et al. report that the diffusion of water into the cellular lattice proceeds by two stages: 1) a rapid stage, where the liquid flows in the longitudinal direction filling in the anatomical cavities (we referred to this diffusion-controlled, Fick’s second law process above), such as the parenchyma voids, and 2) a slow stage, where the liquid diffuses into the cell wall. We have not been able to measure this rate.21

The second type of transport is specific to the capillary-like vascular tissue

(xylem), and is essentially a slow transport process (actually, it is 18 times slower than the diffusion of saliva into parenchyma cell cavities), and follows the 255

Hagen-Poiseuille Law:

_ (P 1-P 2 ) 8 V L

(4.6) where 4)3 is the velocity of the water through the capillary cell cavity, r is the radius of the capillary tibe, pi and p2 are the pressure differences at the entrance and exit of the tubes, T| is the viscosity of the liquid, and L is the length of the tube.

Chances in Phvsical Parameters with Hvdration

As the amount of water in the intercellular matrix and the cellular cavities increases, most of the material properties change as well, such as modulus of

elasticity, damping, etc. Wakefield found that a simple power law of the form:

P = a (4.7) described the effect of moisture on the mechanicalp ro p e rty .22 Here, m is the moisture content below saturation, a, and b are empirical constants for a given material (the values of a and b for Arundo donax are still unknown). Thus, the modulus of elasticity is expected to fall off exponentially as the reed is moisture saturated. We shall show later that the modulus does indeed fall with increased moisture. 256

Mechano-sorotive Creep

As a material adsorbs water, stresses are built-up in the material matrix, which are typically relieved by the swelling of the material. When the water is removed, the material typically returns to its initial structural state. Reed material, and many engineering material, however, contain amorphous regions within the elastic matrix. These viscous regions, when stressed, dissipate their energy of hydration partially as heat. The result is that the recovery of the material, once the water is removed is not perfect, there being some residual amount of deformation due to swelling (figure 4.30). After a number of adsorption-desorption cycles, this deformation becomes permanent and large enough to be detected. This type of permanent non-recoverable deformation, due to the imperfect recovery of a viscoelastic material, is called creep. Since the stress which initiates the creep is due to water adsorption-desorption, this particular type of creep is called mechano-sorptive creep.

The principle effect of mechano-sorptive creep on the clarinet reed is to change its dimensions after repeated playing. The usual method of creep measurement is to measure changes in strain under constant loading. The exact extent of creep behavior of Arundo donax is a matter still in need of investigation, however, the results of the l-warp test above suggests that this is a phenomenon of consequence for the clarinet reed.

Damping

When a material is subjected to a sudden impact, vibrations are typically set 25 7 1. An instantaneous fotoe Is applied to a cantilever beam, bending it.

2. As the force is held constant, overtime the rate of bending slows. This is one way to I measure creep.

3. When the force is removed, a portion of the bending remains permanently. This is creep.

Figure 4.30-Creep 258

up in it, the frequencies of which depend upon the shape, temperature and

restraint of the material.23 The vibrations eventually die down (damp) when the

energy of the impact is dissipated through the material, usually in the form of

heat, due to frictional forces in the material.

There are two constants which are typically used to measure the rate of

damping of a material. Consider a simple damped sine wave, characteristic of

a large class of damped quasi-linear systems. Arundo donax damps almost

exactly as a sine wave, as in figure 4.31. A generic equation for this response

curve is:

Y(t)= Uo sln(cùt + (t))e- k t (4.8)

where, 'F is the amplitude at time t, Uq is the initial amplitude, co is the frequency

(in radians), (j) is the phase angle, and k is the decay constant.

The rate at which the crests of the waves diminish is directly related to the rate

at which the material damps down the energy put into it. If we denote the

amplitude (height of the wave) at time to, as 'F(to), and the amplitude of the crest

immediately next to it as Y(ti) , then the ratio of the two amplitudes, which is called the logarithmic decrement. A, should give the rate at which the amplitudes are exponentially diminishing (due to the exponential term in equation 4.7, 'F(t-->°o)->0) . The ratio of amplitude of one crest to another is simply calculated: 25 9

Signal Display: rurdo 'SCO.Sx D-C sy Cur-.-f iLong ; ms : '

1 if .2hT~ I S0.*>00 ' Lfroth: 3.0v>? :00 :>C. | :Cot-:cr: Î.055 Zcô ffc. i ;Lfvfi: O.fi.'? Volts 1

.03^ Z.03^5 Z.o:£ 2.0^0

Figure 4.31-Arundo donax Decay Curve (blow-up) 2 6 0

Y (to)/Y (ti) = e-kto /e -k ti (4.9)

Taking the logarithm of both sides yields the logarithmic decrement, A:

A = In ['F(to) / T (ti)] = In [e-kto / e- kti] = k t,- k to = k(ti-to) = kp,

(4.10) since ti and to are one period, p, apart, it Is easy to measure the amplitudes of the sound waves from the sound pressure spectrum, so A may be easily determined, and from this calculation, the decay constant, k.

In order to determine the logarithmic decrement for the clarinet reed material, a small rectangular strip was prepared of dimensions 23.5 mm length, by 7.2 mm width, by .66 mm thickness. The reed was mounted as a cantilever (I.e., one side clamped, the other side free) and plucked. The resulting vibrations are shown In figure 4.31, and a blow-up of the end of the sound pressure recording

Is shown as well (figure 4.32). Using the results of the analysis above, the logarithmic decrement Is simply the natural logarithm of the ratios of peak n to peak n+1. In practice, to decrease the possibility of sudden error from one cycle to the next, the ratio Is usually taken between two cycles some distance apart, and the formula for A Is modified as;

A = (1/n) In [Y(to)/Y(tJ]

(4.11) where n represents the nth cycle.

For the plucked reed cantilever, the amplitude (voltage) of the decay curve Is 261

Signal Display: 'lon y*

“ ilf .:AT4 -, SO.vXX) kH:) ; Jk?0 f f C . j Cjr-jar: 2.650 CC0 . [ •.'0 1 * . : (Compressed)

!ïS ft

2.6T0 2.^=‘5 2.^10^ Z .7 y *

Figure 4.32-Arundo donax Decay Curve. Complete 262

Signal Display: M. d o n a * d e c a y c u r v e

|rl le MOD I S3.900 kHz! Length: S. 902 229 Sec. Cursor; 3.162 3C»d Sec. L e v e l : V o lt s (Compressed)

2 3 2.1S2 3 .2 3 23.

Figure 4.32 (cont.). A Blow-up of the end of the Curve 2 6 3

measured at two different points five cycles apart. The result shows a

logarithmic decrement of .1799 (being the log of the ratio of two numbers, the

logarithmic decrement is a dimensionless number). This compares with a value

for wood (a popsicle stick, in fact) of .089. This means that wood damps much

more slowly than Arundo donax, and because of this, retains some residual

amplitude, a “memory,” from cycle to cycle under typical musical vibration

frequencies (figure 4.33). This would lead to destructive interference from cycle

to cycle, and a very unclean sounding tone if used as a reed.

These values are dependent upon the frequency at which the experiment is

run, since, as we shall discuss later, the viscoelasticity of the material affects the

damping in a frequency-dependent way. The frequency here was 220 Hz, which was the natural frequency of the plucked sample.

There is a logarithmic decrement for each of the three axis of the reed, and we have presented results only for the longitudinal plane, since this is the easiest to measure in this simple fashion (obtaining sufficient material in the other two directions with a stem of this radius is very difficult).

The second common measure of damping for simple linear material is the damping constant, r, which is used in the differential equation of lumped parameter models of the clarinet reed (see chapter five), and is related to both the logarithmic decrement and the decay constant. It Is defined as: 264

Signal Display: o. sonu decay surue

|Pi tr KD '• 50.000 ‘•nri Length: S.002 220 Sec Cjrtar: 3.16c 3(W Sec | (Compressed) |Lew?l: vottt j

' A

' A A r. U::r:|4ÎrUuiTUv'J'V-vVv-v: -2 .0 . ;. . '.I •. ij J - 3 . 0 ; . ; ; L ' ' ^ V ' ' :.;S2 3.-02 3.222 3.2^2 3.262 3.2E2

'i0 n â l O i!»pl l e o e e ic le : t i c k 96mm» l^n m a S rw

2.-.'y 300 J«. .Cj-t:'"- i.a»f C'.V 3e;. • L f e . .'a r f (Compressed)

..yJ ( 111 : 1 < , j| r. '! I / 1; i ,l M s . f i' 1 . 1 . . 1 4 I -I.! ' M , V ,1 ;l ’•( « y » • • • ■'■■'li

1.51? 1.53? i .35? 1.57? 1.3?? 1.61? 1.633 I.65? 1.67? 1.6??

Signal Display:

I r t l e .Z ^ T i, I 5O..J0O k n ; , , luenoth: 3.000 300 5ec. ^ iCurtcr: !.?«? 6-0 ;ec. I Lewfl : »ol?t ! (Compressed)

Figure 4.33-Comparison of Wood and Arundo donax Decay Curves 2 6 5

r = — = 2mk n

(4.12)

where, cûq is the measured frequency, and m is the mass. The units are gr/sec.

A typical measured value for r is: 1,72 gr/sec.

Logarithmic decrement is related to a measure of damping specifically used

in viscoelastic material, the so-called tangent 6. We shall discuss this

relationship later.

Stress and Strain: theoretical considerations

In order to understand some of the stiffness properties ofArundo donax, a

brief introduction to the various measures of these properties is warranted.

Imagine a typical linear spring (figure 4.34). When the spring is pulled back by a force, the spring will tend to resist the action of the force by exerting a

"restoring force", F, which is proportional to the length that the spring is pulled (it is an empirical observation that the farther one extends a spring, the harder it is to pull farther). One may write this relationship by the formula:

F = -kx (4.15) where F is the restoring force, k is a constant of proportionality (called the spring constant) which is replated to the "stiffness" of the spring, and x is the amount of extension of the spring in the direction of pull. The minus sign is included to indicate that this is a restoring force, which acts in the opposite 2 6 6

^ F F --- ;— restoring extension

k X

Figure 4.34-Typlcai Linear Spring 2 6 7

direction of the applied force by the person pulling. According to Newton's third law, the restoring force must exactly equal the pulling force at equilibrium.

Imagine that a large number of springs are connected end-to-end, as in figure 4.35.

It may be shown that this is equivalent to a line of springs connected in "series". For n springs connected in a row, one may write n equations, showing the relationship of the springs pairwise.24 These may be rewritten by eliminating the simultaneous terms (a tedious process, which may be found in any elementary physics textbook) to yield:

F total — + 1/kn) (Xi -Xp) — -KtutalX

(4.16) where Ktotai = (1/k1 + 1/kn), and X = (Xi -x„).

This one-dimensional, infinitely thin series of springs behaves, in the continuum, as a single large spring with a single associated spring constant. To a first approximation, this model could be used to represent the behavior of a very slender bar on which a pulling force is applied at one end, while the other end is clamped.

The bar will extend in the direction of the pulling force, but will tend to resist, much as a spring does. Each "cellular spring" acts as a single spring connected in series to form the bar, and as the length of the springs approaches zero (one may think of them as being molecular or cellular sized), the bar approaches a realistic "solid". The pulling force is assumed to be perfectly aligned with the axis of the bar (i.e., only one degree of freedom). The value of X for such an extended spring "bar" is the same as the length of extension of the bar, L, being, that is, the difference of the two end lengths. 268

Large number of springs

Figure 4.35-Cellular Spring Network 269

The composite spring constant for the continuum series of springs representing the actual bar has a special name when the force is applied along an axis of the bar: the modulus of elasticity, bearing the symbol E . It is also called Young's modulus,

after Thomas Young, (1773-1829), who first discovered it. In the continuum limit, the pulling force, F (or the restoring force, since they are equal, but of opposite signs, by

Newton's third law) is given a specific name in engineering tests: stress, and the resulting extension, L, of the bar is similarly called the strain, or extension. Equation

4.16 may be rewritten as

Fstress — E L

(4.17) which is the more familiar engineering form for a real solid, and not the collection of springs which we used to motivate the discussion.

Since the value of the modulus of elasticity, E, is unique for any given material, and measures how responsive the material (i.e., how easy it is to stretch) is when an axial force is applied, this is a useful number to know. It may be determined simply, since, by formula 4.17:

E = F s(resg/L

(4.18)

To determine E, it suffices to apply a force to the material and measure the extension which results. If one plots the force as a function of the extension, the slope of the line in the linear (i.e., straight line) region is Young’s modulus (figure 4.36). Beyond a 270

! to ProportionaJ Limit

Strain

Figure 4.36-Young’s Modulus 271 certain point (the proportional limit), most materials no longer behave linearly.

A real bar, unlike a one-dimensionai bar, has a finite thickness and width. It then makes sense to speak of stresses F%, Fy, F% applied to any of these three faces and, similarly, of the resulting strains L%, Ly, L%. This allows one to define a modulus of elasticity, E%, Ey, Ez for each one of the three principal axes, x, y, and z, in the bar, containing opposite faces. The resulting measured modulus of elasticity of the material might be different in each of these directions. One should bear in mind that this type of modulus is measured strictly along a line in the plane of the opposite faces of the material, and should be called an axial or normal modulus. To represent this more precisely, we denote the moduli (as well as the stresses and strains), correspondingly) as E%x, Eyy, and Ezz, to remind the reader that the stress is applied across two faces of the bar lying across the x axis, the y axis or the z axis.

Stress does not have to be applied only across lengths of a material, but may be applied along the faces of the material as well. Thus, in general, one has to include the moduli (with corresponding stresses and strains), E%y, Exz, Eyz, Eyx, Ezy, Ezx, defined in the plane of each face of the bar, as well as the simple face-to-face planes moduli,

Exx, Eyy, and Ezz (figure 4.37).

Mohr's Circle

It was assumed that the axial stress used to determine the modulus of elasiticy in the preceeding section was applied at the midpoint of the rod edge or plate face.

Sometimes, however, a stress is applied at a different location on the face or edge (or. 272

E XX XX

zz

E

* 1 ^ / Y xy à 4" /' T /

Figure 4.37-Onentation of E and G Moduli 273

alternately, when the point of origin of the force remains the same, but the face has

been moved) or possibly when the face has been deformed, perhaps due to swelling

in the case of wood, and the question logically arises: how does one compute the

resulting strain on the structure?

A very elegant method was devised based upon the trigonometric relationships of

the principal stress vector (or, more properly, tensors). It may be shown (see, Bodig

and Jayne, pg. 81 ff., for details) that the sum of the stress components along the

principal axis (diagonal) of the face undergoing deformation remains unchanged

regardless of the deformation of the face (specialists in the mathematical discipline

known as tensor mechanics-a tensor being a generalized type of vector-say that the total stress component is invariant) .25

This discovery allows one to graph the constant total stress on a face as a radius of a circle. As the face of the material is deformed, the off-diagonal stresses change, but in such a way that the sum of the stresses remain constant (i.e., much one may vary the two sides of triangle and still maintain the same length of the hypotenuse).

Mohr's circle, as this graphical construction is known, allows one to determine the strain resulting from a stress applied to any deformed face (in the linear realm of

response), since the relationship between the stress components may be calculated.

This method is particularly useful when one is dealing with derivative states of a material, such, as mentioned earlier, swelling or heliotropically-induced extensions in wood.

Specifically, Mohr's circle is of interest in analyzing the stress-strain properties of certain clarinet reeds when: 1) the reed had been moved off-center on the mouthpiece. 274

as some players do, and 2) when reeds are machine cut so that the principal direction

of the fiber axis is tilted relative to the direction indicated by the cutting of the reed

(figure 4.38). Such cutting in the manufacturing process is not at all rare. The grain

angle (or fiber/cutting skewing angle)for the clarinet reed measures this

divergence between cutting and biomorphology.

Moving the reed off-center, to the right or to the left means that the bending stresses

which pass over the reed surface in the form of a pressure antinode propagating down

the mouthpiece from the bottom to the tip during the course of a vibration cycle is

impacting on a distorted (slanted) reed face. The effects of such skewing we shall

examine in chapter five.

Shear or Bending Modulus

The facial moduli, E%y, etc., which are proportional to the strain when a stress is

applied (the face is said to be "sheared") along a face of a bar are given a specific

name in engineering: shear moduli, (alternately, bending moduli, or moduli of

rigidity) and are denoted by the letter G (figure 4.37). There are six bending moduli,

G xy. G xz. Gyz, Gyx, G%y, G » , sod the associated stresses are calledshear stresses;the

strains are called shear strains. These moduli measure not only the deformation on the faces of the material, but equivalently, how difficult it is to induce off-axial bending

(shearing) of the material, as opposed to in-the-plane stretching, as the normal moduli,

Exx measures. This equivalence makes the determination of theG moduli straightforward, as one may imagine, for the purposes of testing the shear moduli, that a flat plate of a material is connected to some support by springs at each corner, so 275

XX

XX

After rotation, the newtjia read, which is the atreaa on the material. A similar Circle may be drawn for the strain. The resulting Evaluais read as the ratio of the transformed values

Grain Angle Reed Orientation g on Mouthpiece

Cutting

Figure 4.38-Two Different Uses for Mohr’s Circle 276 that facial stress tends to cause deformation in these edge springs. From this analogy, it should be clear that one simply needs to measure the bending force and resulting deformation on the various faces of the material. The standard test for this moduli is called a three point bending test. We shall have more to say about this later.

These nine moduli, three normal, and six shear, completely define the moduli of the material, but because of symmetry (E%y =Ey%,Exx = Exx etc.), only six moduli are really needed- -three normal, and three shear: E%x, Eyy, Ezz, Gxy. Gxz, Gyz.

If the magnitude of one of the normal moduli is much larger than that of the other two, such that one of the axis of the material has a much greater elongational strength than the other two (e.g., Exx ” Eyy > Ezz), then this sort of material is called orthotropic

(ortho = right, in Greek). This is the common state for wood material. If the three axial moduli are roughly equal, the material is called isotropic (iso = equal in Greek).

Complicated interactions involving the relationships between the facial and normal stress strains may be derived based on a mathematical formulation involving tensor equations (a tensor is a type of generalized vector), but as this is somewhat advanced material, and not particularly illustrative for this document, we shall refer the reader to

Bodig and Jayne for a detailed explanation.26

Poisson's Ratio

When a material is pushed together from opposite sides, it typically tends to expand in the direction perpendicular to the applied pressure (one need only think of pushing on the opposite sides of a panel of rubber-the width of the rubber increases as the pressure increases, as shown in figure 4.39). Similarly, if one pulls the material apart 277

-11

n=E,(E^

Figure 4.39-Poisson’s Ratio 278

from opposite edges, such as one might do to a rubber band, the width of tfie material

decreases or contracts. When a force is applied across one axis, the constant which

measures how much a material changes dimensions in the axis perpendicular to the

applied stress is called Poisson s ratio, after the famous French mathematician of

the same name, Simeon-Denis Poisson (1781-1840).27

Mathematically, if the deformation (strain) in the original axis of applied stress is En,

and the deformation (strain) in the axis perpendicular to it is E^, then Poisson's ratio is

defined as:

n =E|| / Ea

(4.19)

Since there are two axes which are be perpendicular to a given axis in Cartesian

space (e.g., both the y and z axis are perpendicular to the x axis), in general, an

applied stress in one axis has a resultant effect in either or both of the perpendicular

axes. Thus, one needs to specify in which perpendicular axis one Is measuring the

resulting deformation for when a stress is applied to the original axis. This

nomenclature problem is treated by the use of subscripts in the designation of a

particular Poison's ratio. Let us call the three axes, x,y, and z for simplicity (other coordinate systems, such as polar coordinates, use different terminology). One may then speak of rixy, andt |xz where, by convention, the first coordinate subscript is the axial direction, and the second coordinate is the perpendicular direction.

It may be shown (see, for example, Niklas) that for an isotropic material, the various moduli and Poisson's ratio are related by the simple formula: 279

G = E/2(1+ti)

(4.20)28

Elements of Viscoelastic Theory

Because of the composite nature of the biological material,Arundo donax (i.e., ground and vascular tissue), the material behaves in a manner different than steel, for instance, when a time-dependent load is applied. The amorphous material (typically concentrated around the vascular tissue in the form of lignin), is essentially a superviscous liquid, and stores the applied stress, much as a dashpot might store mechanical energy, or a capacitor store electrical energy. The crystalline or ordered cellulose lattice, on the other hand, is essentially elastic, responding as a spring might to mechanical stress, or a resistor might to an electrical stress.

How then, should one model a material composed of a composite of both viscous and elastic elements? The most-successful current models represent the viscous elements as dashpots, since they share the same response properties, and the elastic elements as springs, for the same reason (figure 4.40).

If the elastic spring has the response of equation 4.17:

Fe = E . L

and the viscous element obeys the law,

F V d t '

(4.16) 280

SPRING

Ü DASHPOT

MAXVELL CIRCUIT

VEICHART CIRCUIT

BURGER CIRCUIT (GENERALIZED LINEAR CIRCUIT) VOIGT CIRCUIT

Figure 4.40-Viscoelastic Circuit Models 281 where v is the viscosity of the dashpot, E is the elastic modulus, and L is the extension of the spring or dashpot, then any viscoelastic combination may be modeled as a mechanical circuit (or an electrical analogue).

The most common simple circuit used to model a viscoelastic material when it is extended a rapidly as possible, and then the stress is allowed to decay so that the material “relaxes” (this is the basis for the so-called stress-relaxation test) is a combination of a dashpot and spring in series, called a Maxwell circuit or model.

This model is in general to course to describe viscoelastic relaxation accurately, so it has been generalized to a finer network of parallel coupled Maxwell circuits, which is called a Wiechart circuit, or model. This generalized form allows one to use continuum equations and arrive at more realistic relaxation models. We refer the reader to Bland or Gross.29

To model creep, the simplest model is that if a spring and dashpot connected in parallel. This type of mechanical circuit is called a Voigt circuit or model. It also has an electrical analogue of a resistor and capacitor in parallel (the equations of the mechanical and electrical circuits are identical in form). This circuit also is too course to give accurate results in creep studies, and a generalization, the standard linear circuit (or Burger’s circuit), which contains n (very large) number of series connected Voigt elements, is used. Again, the reader is referred to the references above for more detail (see also, Broutman).30

Because of the viscoelastic nature of the reed material, a simple modulus generally does not capture both the viscous and the elastic nature of the material. Engineers have developed a modulus which is the sum of the elastic (storage) modulus, E’ and 282 the viscous (loss) modulus, E". It is called thecomplex modulus (an equivalent formulation exists for the shear modulus, substituting G for E):

* ' " E = E +iE (4.16)

This modulus should, properly speaking, be reported for any viscoelastic material, however, most tables of wood moduli give only the storage modulus, which is the familiar engineering modulus mentioned at the beginning of this chapter.

Because an imaginary number may be represented in trigonometric form by the transformation:

X + iy = r (cos 5 + i sin Ô)

(4.16)

It becomes possible to rewrite equation 4.23 in polar form:

E + iB" = E*(cos 5 + i sin S)

(4.16) where E’ = IE*I cos Ô, and E" = lE'l sin 5. The angle between the real and imaginary parts in the complex plane is simply the ratio:

E c E* sin 5 - = tan Ô = E iE cos (5 (4.16)

Tan 5 is taken to be equivalent to a viscoelastic damping constant. It is roughly related 283 to the logarithmic decrement by the relationship:

A = 7c tan 8

(4.26)

Stress and Strain: empirical tests and results

In order to determine the extensional and shear moduli, two separate tests were attempted. In the first test, a small strip was isolated the longitudinal axis of the stem and clamped, forming a cantilever. The strip was plucked, and the natural frequency measured. The extensional modulus was then calculated using the formula:

4 8 z ^ o ) ^ p A L'* E = ------(k .L )- I

(4.16) where © is the measured natural frequency, p is the density, A is the cross-sectional area (= bh, where b = width, h = thickness), kp is obtained by solving the transcendental equation,

COS(knL) COSh(knL) = -1

(4.28)

where L is the length of the cantilever (in this case, for the lowest mode, k,L = 1.875 ), 284

and I is the moment of inertia (= bh3/12 for a rectangular cross-section).3i Simplifying

equation 4.27 leads to;

E - P L ' ( k „ L ) “ h"

(4.16)

which is the one we used. Granted, there are refinements which must be made if the

sample is very thin (approximating a thin plate), but generally, the equation above is

simple to use and the error introduced is not too large. As an example, the frequency

of the thin strip used in the damping experiment earlier was measured (before it was

filed to make it thinner for the damping experiment). The length was 43 mm, thickness,

.66 mm, width, 7.2 mm, density (assumed) 461 kg/ms. The frequency measured was

269 Hz. Using this data in the above formula the value of E in the longitudinal

direction is found to be 10 x 109 Pa (wood = average, 8x109 Pa), which is low compared to bamboo (20 x 109 Pa).32 The wet sample had a frequency of 220Hz, or

7.46 X 109 Pa, a drop (here, of 25%), as expected for a hydrated natural material.

The second method used a differential mechanical analyzer (DMAT) to determine the G modulus, which is a device which allows for three-point bending of very small samples. This test allowed for the loss modulus, the storage modulus and the tangent 5 to be measured. A typical plot is shown in figure 4.41. The tangent 5 is seen to be about .25, which is considered very large, as expected for a material showing such high damping. 285

Curv# il DMA Tmo/TIm Scon in 3 Point Bonding rilo Inf 01 2o7tP Mon Jun 22 IBiSSiAB 1592 rroquoncyi 1.00 Kz Oynooie Scroiu 1.00o*0S Pa 2t7ptn P 1 2o7ptr Storeeo HeOuluo (Pa * 1 0 ) 3.00 i- 9 2 con 6 1.00 P 3 Lota Modwlv# (Po * 10A ■0.90

0 .8 0

^3 2.00 ' 0 .70

•0.60 I • 1.50 ' ■0.50 hO.40I

■0.30

0.20

0.10

I------1------1------^0.00 0.00 2.00 4.00 6.00 10.00 bed rood 90 dtg not Tim# (mlnut##) î iS c ■” * P*™** 10.0 C/«i PeWIH-EUCR

Figure 4.41-Typical DMA Plot 286

In terms of tangent 5, there Is a frequency dependence cn the damping. As the frequency is increased, this is equivalent tc lowering the temperature in terms of the response of the material tc damping. The material beings to resemble more and more the glassy state of the material (conversely, at low frequencies = high temperatures, the material is in its rubbery or elastomeric state). In this state, the material damping falls and the influence of the viscous portion of the material is diminished.

This means that the higher the frequency of the vibrating reed, the less the viscoelastic perturbations influence the vibrations, and the reed behaves more as an elastic material (spring). Thus, at lower frequencies, it is the high damping which helps to maintain the purity of the sound, whereas at high frequencies the smaller amount of audible partials in the more elastic response substitutes for the lower damping to maintain the purity of sound.

Biomorpholooical Variations

There are many different cellular structures which occur in Arundo donax, as has been shown in chapter two. There is a natural variation in the size of individual cells of any given type, and also a range of sizes between the different cell types. Let us call the first type of size variation the cell size variations (CSV), and the second type of size variation the type size variations (TSV). The chart at the end of chapter two gave a summary of the TSV and the CSV.

Veselac did the first extensive study to relate the size and morphological variation of selected cell types to the quality of clarinet reeds.33 Her research is worth dwelling on.

She took at least thirty-four pairs of reeds (one good, one bad) and examined the 287

morphology of the material by embedding the samples and microscopically examining

a representative cross-section. The commentary on her findings are extensive and well worth reading.

She found that good reeds have a greater number of vascular bundles near the epidermis which are twisted off-axis than undesirable reeds. Such twisting occurs in material taken near the leaf gap, which would imply that the material is harvested near a node. She also found that good reeds have a higher degree of complete bundle sheathing around the vascular bundles, indicating a higher degree of ligninfication than in undesirable reeds. Such lignification is indicative of longer growing periods.

The parenchyma cells were also found to be smaller with thicker walls in the good material, indicating a longer harvesting period, since cell wall deposition was greater in the good reed material, as indicated by the thicker walls (as we pointed out in chapter three, the amount of hemicellulositic material peaks and then drops as the plant matures, indicating that more cellulose than hemicellulose is deposited in mature cell walls).

She also found that the larger the grain angle, the worse the reed. We shall point out in chapter five that the skewing angle of the vascular tissue is partially responsible for the increase in buzziness of the clarinet reed.

In the current study, we counted the number of vascular bundles along the entire line at the rounding point of the reed (see diagram 1.10) using a dissecting scope to magnify the reed. The results are plotted in figure 4.42. The summary statistics for the count is given in figure 4.43: 288 VnndoreiL V12 Ml BM GC O liv V(3.5) BM 30

0

0 oo 9 2 o o o 0 ( 1

1 0 0 0 t < > 0 (D

0 0 O G O 0 0 o o 0 o 120 to 0 Q D 0 9 0 3 0 9 (D > 0 Œ ZD 0 01» o o (

0 0 0 0

O C 0

o

10 20 40 60 80 100

Reed Number

Figure 4.42-Vascular Bundle Number in a Sample of Reeds 289

VASBUNDL

N OF CASES 80 MINIMUM 13.000 MAXIMUM 27.000 RANGE 14.000 MEAN 20.263 VARIANCE 6.778 STANDARD DEV 2.604 STD. ERROR 0.291 SKEWNESS(GI) 0.442 KURT0SIS(G2) 0.549 SUM 1621 C.V. 0.128 MEDIAN 20.00

Figure 4.43-Vascular Bundle Statistics

The volume fraction of vascular bundles was approximately 9 % -13 % along the line at the tip of the reed, which is equivalent to measuring at Rjnner from chapter two.

Despite Veselac’s findings, this author found no correlation between the number of vascular bundles and any of the quality rating parameters which were measured:

MATRIX OF SPEARMAN CORRELATION COEFFICIENTS

VASBUNDLE QUAL DA-BA SOHA BU-FL VASBUNDL 1.000 QUA! 0.012 1.000 DA-BA -0.017 -0.376 1.000 SO-HA 0.013 -0.086 -0.251 1.000 BU-FL 0.238 0.631 -0.195 -0.001 1.000 SQ-ST -0.060 0.464 -0.088 0.063 0.438 WE-FU 0.104 ---

NUMBER OF OBSERVATIONS: 80 (70 for WE-FU) ’

Figure 4.44-Vascular Bundle/Quality Correlation matrix

As mentioned earlier, the largest correlation was between the number of vascular bundles and the degree of buzziness of the reed. In addition, there is a trend toward 290

the larger the number of vascular bundles and a fuller sound (albeit, a very slight one).

This is probably the source of the folk wisdom that the greater the number of vascular

bundles near the tip of the reed, the better the reed. In truth, this is a very misleading

statement, as there is no correlation with reed quality, but the fuller sound has been

interpreted to mean this. Overall, the vascular bundles apparently help to mitigate the

effects of energy dissipation during the reed vibrations, but may also induce off-axial

bending.

The function of the vascular bundles is essentially that of a ribbing or stiffener.

Since the fiber (vascular bundles) and matrix (parenchyma) together function as a

binary composite compound, the composite modulus (E) near the tip is given by the

simple rule of mixtures:

^xx ~ ^fiber^fiber ^matrix^matrix (430)

This equation only holds for unidirectional, continuous fibers with a skewing angle of

zero degrees, such as might hold near the tip, since the vascular bundles are long

enough to be considered continuous in this region.

Farther away from the tip, the vascular bundles become discontinuous and also

possibly skewed. If the bundles are skewed and discontinuous, an extension of the work of Mallick gives the following five equations relating two extensional and (in the

longitudinal, 5%% and transverse, Eyy directions), one shear modulus, and the major and

minor Poisson’s ratios (figure 4.45):34 2 9 1

Skewed, Continuous

Axial, Discontinuous

Skewed, Discontinuous (Arundo donax]

Axial, Continuous

Figure 4.45-Dlfferent Types of Fiber Arrangements in Clarinet Reeds 292

+ + t ( 4 G XX 11 '22 12 2v. % s i n ^ 2 e 11

1 sin^0 cos^0 1. 1 :— = + ------+ - ( - 4 'G yy 1 1 '22 12 2v- 12-)s in ^ 2 0 1 1

1 2v. — + 1 2 + - i G xy 11 1 1 '22 2v. 1 1 + 12 + )cos^20 1 1 1 1 ^ 2 2 ^ 1 2 293

V„ 1, 1 2v,

+ - -^)sin^2e ^ 2 2 ® 12

V . = XX

where,

p _ ^ ■*■ 2(Lf/Df)?7LVf M l - r ------. m

p _ ^ 2 rjjV f ^22 ------E m l-77TVf

Vi2 - VfVf + and, 294

(E ,/E J -1 (Ef/E„) + 2(L,/Df)

(Ef/E„) + 2

_ ( G f / G ^ ) - 1 (G f/G „ ) + 1

The subscript f refers to the fibers, and m refers to the matrix. The ratio, Lf / Df is calied the fiber aspect ratio. Despite all of these messy equations, the messages are

clear: 1) the larger the number of vascular bundles, the greater the composite moduli,

and 2) the larger the skewing angle, the smaller the composite moduli. Overail, effect one usually wins out, unless the skewing angle is huge.

We have seen in this chapter a beginning of the study of the material properties of

Arundo donax. There is still much work to go before the exact uniqueness of this material is characterized.

in the next chapter, we put ali of the preceeding chapters together, as we take a brief look at the very complicated physics of the clarinet reed. 295

Endnotes

1. Nkilas, Plant Biomechanics...,108.

2. Bamboo is much too stiff to use as a reed substitute, for example.

3. We thank the people at the Campus Electron Optics Center, especially Ann Osterfeld, and Cathy Wolken for their help in making the optical densitometry maps.

4. Fengel and Wegener, Wood..., 330.

5. ibid. 337.

6. ibid. 337.

7. Jozsef Bodig and Benjamin A. Jayne,Mechanics of Wood and Wood Composites (New York; Van Nostrand Reinhold, 1982): 570.

8. Fengel and Wegener, Wood..., 346.

9. ibid. 350.

10. ibid. 356.

11. Jorma O. Tenovuo, ed.. Human Saliva. Ciinical Chemistry and Microbiology, vol. 1(N. W. Bocca Raton, Florida: CRC Press, Inc., ): chapter 2.

12. ibid. Chapter 4.

13. ibid. Chapter 4

14. ibid. Chapter 4

15. ibid. Chapter 4

16. ibid. Chapter 4 296

17. We thank Dr. Tuovenen in the biology department at OSU for his help in identifying the shape of the bacteria and for discussion in general about the pathology of the bacteria.

18. Jean-Marie Heinrich states that the bacteria lyses (secretes an enzyme that degrades) the cell walls. This is probably not correct. While some bacteria do so, this particular bacteria probably does not.

19. Parylene, Poly (Para-xylxylene), is commercially available through the Nova Tran Corporation, a subsidiary of Union Carbide. It is located in Clear Lake Wisconsin, and the contact person, as of the time of this writing, is Bruce Humphrey.

20. Jozsef Bodig, “Moisture Effects on the Structural Use of Wood," in.Structural use of Wood in Adverse Environments, Robert W. Meyer and Robert M. Kellog, eds. (New Yor: Vna Nostrand Reinhold, 1982): 54.

21. E. Chirkova, A. Kreitusol, and I. Andersone, "Studies on Diffusion of Antiseptic Solutions into the Wood,” in Ligno-Cellulose: Science, Technology, Development, and Use, J. F. Kennedy, G. 0. Phillips, and P. A. Williams, eds. (New York: Simon and Schuster, 1992)

22. Bodig, “Moisture Effects...,” 58

23. Bodig and Jayne, Mechanics of Wood... 251 ff.

24. ibid. 81.

26. ibid. Chapters 3 and 4.

27. Niklas, Plant Biomechanics..., 68.

28. ibid. 74

29. D. R. Bland, The Theory of Linear Viscoelasticity (New York: Pergamon Press, 1960); B Gross, Mathematical Structure of the Theories of Viscoelasticity (Paris: Hermann et Cie, 1953).

30. Lawrence J Broutman, Richard H. Krook, eds.Modem Composite materials (Reading Massachusetts: Adams Wesley, 1967): 123.

31. Cyril M. Harris, Carles E Credes, Shock and Vibration Handbook, 2nd ed. (New York: McGraw-Hill Book Co., 1976): 3-44 297

32. S. H. Li, S. Y. Fu, B. L. Zhou, and X. R. Bao, “Reformed Bamboo and Reformed bamboo/Aluminum Composites,” Journal of Material sSciences, 29 (1974):5990-5996.

33. Veselac, Tissue Differences...

34. P.K. Mallick, Fiber Reinforced Composites (New York: Marcel Dekker, Inc., 1988): 108-111. Chapter V

Reed/Mouthpiece Dynamics

Overview

This chapter, the heart of the present work, is concerned with the question of how the clarinet reed actually vibrates in the mouth of the player, and how such things as shape and material affect this motion. As stated in the preface, much of the material in this chapter is necessarily advanced, as the vibrational analysis of the clarinet reed involves techniques at the forefront of mathematics, engineering mechanics, biomechanics, and physics. For the expert in these areas, the discussion should be self-explanatory. For those not familiar with the techniques employed in this chapter, references are given in the endnotes.

Where possible, we shall summarize the results of each of the sections in this chapter in a non-technical way.

It is not possible, except in a highly idealized way, to understand the physics of the clarinet reed apart from its interactions with the acoustical and flow-control properties of the mouthpiece (the sound (or fluid)-structure interaction problem in engineering). Essentially, the reed is a flexible plate lying on top of an enclosed, air-containing chamber of nonuniform shape. The mouthpiece and reed are, in addition, partially immersed in the player’s oral

298 299 cavity.

It is the purpose of this chapter to analyze the various aspects of reed dynamics and mouthpiece interaction. To do this, we shall divide the problem into separate one-, two-, and three-dimensional models of both the static reed and the aerodynamically forced reed. In each section, we shall present analytical and numerical results, where possible, and in the case of the three-dimensional models of reed and mouthpiece, and finite element simulations will be introduced. The results of the three-dimensional models will be compared to rare motion studies of the reed vibrating in a player’s (the author’s) mouth, which is stroboscopically slowed to allow observation of the motion. Finally, we shall discuss the physics of some of the complications to the basic reed and mouthpiece, such as the lip, reed degradation, scraping, mouthpiece lay, etc.

One-dimensional Reed Models

The earliest model of the clarinet reed belongs to Euler. Leonard Euler

(1707-1783) was Swiss by birth, but spent most of his life in Russia in the company of several other European mathematicians at the request of the Tzar of Russian. In two treatises and one famous letter he treats the nature and propagation of sound.

At the age of twenty (1727) he published his first treatise, Dissertatio Phvsica de Sono [Dissertation on the Physics of Soundj.i In this work Euler considers the nature of sound as arising from the compression of "air globules", and uses 300

this idea to calculate the speed of sound (arriving at an incorrect value). He

then takes up the vibratory action of strings, and derives from theory some of the

basic empirical results of Galileo and Mersenne. These theoretical results were

also derived around this same time by J. Bernoulli and the Englishman, Brook

Taylor (1685-1731). The correct solution of the problem of the vibrating string,

and not merely isolated conclusions, however, goes to d'Alembert.

Of concern with regards to the clarinet reed is the statement by Euler:

To this type of sound production must be referred also the sounds made by vibrating reeds inserted in tubes blown by the wind; although the later also pertains to the third mode of sound production...The wind, in seeking a passage for itself opens the reed like a valve [my italics]; however, in opening it stretches too much so that the valve closes again, attempting to restore its original state; it is then opened again, so that it imposes a vibrating motion on the air passing through it. It is necessary indeed that with the wind flowing evenly [my italics] the valve shall stay quiet and the sound shall stop. However, taking heed of this, the wind, while propelled by bellows [lungs] strikes the orifice of the device unevenly [author’s italics]. The wind with the help of the pipe and the inserted valve is rendered vibratory.^

This statement is extraordinary inasmuch as this is essentially the correct

physical mode of action for the clarinet reed behavior. The valving action of the

reed was rediscovered by Bouasse only in the 1920's, but not fully understood

or exploited until fifty year later.3 Euler’s theory for the onset of oscillations s

unclear, however. Oscillations are a result of the non-linear nature of the flow of

air through the mouthpiece aperture. It is not clear if this is what Euler means by

“even” and “uneven” air flow, but since vortex formation, etc. are types of

uneven air flow, we cannot ignore this possibility. In any event, Euler did 301 nothing further to solve the reed physics problem.

Helmholtz, in his, On the Sensations of Tones as the Basis of Musical

Sounds (1862) separates the different types of valving actions in musical instruments into two categories: those valves (reeds) which are pulled together when air first starts across them (the clarinet reed and mouthpiece--the distance between them decreases), and those valves which, when air flows through them are pushed apart (the lips in brass instruments).4 He calls the first case an inward beating reed, and the second case an outward beating reed.

Fletcher has developed the nomenclature that an inward beating reed is represented by a minus sign, and an outward beating reed by a plus sign.s He notes both the upstream response and downstream response of the reed by parentheses; (downstream value, downstream value). If the reed beats inwards in the downstream (positive or inflow) direction, and outward in the upstream

(reverse or outflow) direction (clarinet reed), it may be represented as (-,+). This is the situation for the clarinet reed-the reed is inward beating in the direction of forward air flow into the air column and outward beating in the reverse direction.

The lip reed or trumpet may be represented, conversely, as (+,-).

Despite the fact that the valving action of the reed was already known, and despite the fact that elementary two-dimensional beam deflection theory had been known for a century (originating with Euler, himself), the free, inward beating reed, in ail theoretical calculations until late in the twentieth century, was portrayed as a simple one-dimensional “mass-spring” system. When a complex system, having many degrees of freedom, is reduced to a single 302 composite unit of mass, m, and the vibrating parts similarly reduced to those of a composite spring, the resulting model is called, alumped parameter model, since all of the physical quantities are “lumped” together. The early one-dimensional mass-spring models of the clarinet reed motion were of this type, treating the reed as an equivalent mass, and the elasticity of the reed as an equivalent spring.

By contrast, when the reed is treated as a series of very small connected masses coupled by springs to give a solid structure, it is called adistributed parameter model, in that the system is made up of small mass-bearing elements (cells, etc.) which distribute the mass throughout the system.

One-dimensional Unforced Models of the Reed

The first reed calculations were those of Das, in 1929.6 His lumped parameter model is the prototype for subsequent reed dynamics models. The basic equation is that of a damped mass-spring system:

(5.1) where y is the displacement of the reed, m is the equivalent lumped mass of the system (here, simply the reed mass), ^ is the mass damping of the reed, and E is a constant of elasticity (Young’s modulus-in Das’s original equation, the damping term, ^dy/dt, is omitted). The function, F(p) describes the air flow in 303 the tube and mouthpiece, and although Das spends most of his paper discussing this, we shall ignore its exact form in this section, which deals only with the clarinet reed proper. This model has been called, the“standard

[lumped parameter] reed model”, by Keefe, and indeed, every reed researcher except for Sommerfeldt, et al.has used it in one form or another.?

Equation 1, without a forcing function (air-flow) has simple sine and cosine

(or exponential) terms for its solution, and tells nothing about the real reed motion itself, except for some simple facts related to frequency and damping.

We shall only mention a few details of this standard lumped parameter model, since its real strength is in modeling the reed-air column interactions, which is presented in the next section. The interested reader may find this equation discussed in more detail in any elementary text on differential equations.8

One may assume a solution to the homogeneous version of equation 5.1

(F(p) = 0 ) of the form y = e«*. Substituting this trial solution in the differential equation leads to:

[ma^+ Ça+ E] = 0

(5.2)

Since the bracketed term must be equal to zero (e“‘ ?!: 0), one may use it to solve for y. It is seen (see, for example, theShock and Vibration Handbook, chapter 2 , for a discussion), that there are several types of responses of this one-dimensional model of the clarinet reed depending upon the values and 304 sign of m, and E .9 A summary of results follows:

Case 1 : Undamped (F = 0, ^ = 0)

y = A sin .1— t + B cos ^ — t m V m (5.3)

In the undamped case, once disturbed from equilibrium, the reed vibrates forever at the angular frequency, cOn = (E/m)i/2 which is related to the natural frequency by fn = (1/2n) ©„•

Case 2: Positive Damping (F = 0, ^ > 0)

Let 2 (Ç m)i/2 = Ç^be defined as the critical damping coefficient, = (pbe the

fraction of

y sin co^t + B cos cn^t) critical

damping.

Depending upon the value of cp, three responses are possible:

2a: Positive, Less-Than-Critical Damping

(^>0 , (p< 1 )

(5.4)

2b: Positive, Critical Damping 305

(^> G, (p =1 )

y = + Bt)

(5.5)

2c: Positive, Greater-Than-Crtical Damping

(^> 0, (p >1 )

y = + Be““"^‘)

(5.6)

where, C0d= ®n(1- 92 )1/2.. For reeds operating below critical damping, the solution oscillates while damping to its equilibrium state, as in the famiiiar damped sine wave. The damping factor is given by ^t/2. In the other two cases (cp =1, cp >1), the damping is equal to or above the critical damping, and the solution damps to equilibrium without osciliating. If ^ were negative (negative damping), the solutions would grow without bound, with similar oscillatory or nonoscillatory solutions possibie. Negative damping is never encountered in the unforced reed case, although air friction acting on the reed is a form of negative damping, and its effects will be discussed later. There are many different books describing these damping cases, and we refer the reader to books on the interpretation of differential equations for more details.10 306

One-dimensional Linear Models of the Forced Reed

When the mass-spring model of the reed in equation 5.1 is forced by some periodic function, several responses are possible. Let the forcing fuction, F, be described by F = Fq sin (cot), where Fq is the amplitude, and co is the forcing frequency. When there is no damping, the solution is:

F /E y = A sin(co^t) + B cos(co^t) + ------^ — - sin(o)t) l-CO /CO, (5.7) where the first two sine and cosine terms are identical to the simple damped sine wave without forcing of equation 5.4. The initial transient (sine and cosine

COn terms) eventually damps out, leaving only the third term, which is a steady-state oscillation at the forcing frequency, co. co„ is the angular frequency of the reed.

Whent there is positive damping, the solution is:

^ Fn|- sin(g)t-e)______^

I + { I c p a l c o j + natural frequency terms (damped sin+cos)

(5.8)

where. 307

— '< ï 5 7 5 >

(5.9)

Usually, the natural frequency terms damp out as before, very rapidly, leaving the steady-state terms, so we have not restated them in the above. If the forcing frequency is sufficiently different than the angular frequency, «)„, then the solution above reduces to three cases:

y ~ ( "Æ ^)sin(ojt + 7t) = (^ \)s in (c o t) (o) »c o j CO Ç m c o

(5.10, a, b, c)

The first case is called the spring-controlled system, the second case is called the damper-controlled system, and the third easels called the mass-controlled system. Since it can be proven that the clarinet always operated below the resonant frequency of the reed, which is approximately equal to the natural frequency of the reed when damping is low (case one above). Thus, the clarinet is a spring-controlled system, and the reed responds with simple sinusoidal motion, up and down in the planewhen forced by a 308 periodic jet of air. This is the simplest model for the clarinet reed.

In place of the simple sine wave forcing above, Das attempted to model the forcing of the reed by an actual air column. Starting from the lumped parameter model of the reed (without damping!), and including the pressure difference across the reed (his nomenclature has been standardized to that used in this chapter):

d^y + Ey = A(P-p)

(5.11) where, y is the reed deflection, m is the reed mass, E is Young’s modulus, A is the area of the reed surface, P is the excess blowing pressure above atmospheric pressure inside the oral cavity, and p is the pressure spectrum inside of the mouthpiece. By assuming a velocity potential of the form:

0 (ct-x)

(5.12) where c is the wave speed of the pulse down the tube, he was able to show that the general form of the reed/air column coupling is:

A p c P d - ^ )

m l4 + Ey = A P ------dt pc-coa

(5.13) 309

From this equation he found three results. By assuming a solution of equation

5.13 of the form:

y = Yo + COS (n t)

(5.14)

(which solves the homogeneous form of equation 5.13) he was able to show that the natural frequency of the reed drops due to the coupling with the air column. In equation 5.14, n functions as an adjusted modulus (compare with B cos Ont = B cos ( E /m )i/2 in equation 5.7):

2 _ ApcPm n = E — (pc-û)a)^

(5.15)

which is lower than the true modulus, thus appearing to lower the natural frequency of the reed.

Secondly, if one solves equation 5.13 by equation 5.14 and expands the solution in a Fourier series, one of the terms will expand as:

(5.16) 310

where s = (2L/%)(n/c), and r = 1, 2, 3... It is clear that when s approaches r,

the denominator biows up and the radiated power increases dramaticaliy. As

Das puts it, “The denominator renders it obvious that those harmonics of the

pipe which nearly agree with the vibration of the reed will have large

amplitudes.'"! 1 This remark anticipates Thompson’s work on reed resonance by

fifty years. 12

Finally, Das found that the reed vibrates about a center of rest which has a

sudden (my emphasis) shift alternately away and towards the mouthpiece. In fact, the reed does appear to swing directly from open to closed to open again

in video of reed motion the author has obtained in the mouth of a player;

however, this is only approximately true. Das shows that for each half-cycle, r =

1, 2, 3... the displacement is:

— (1 + + A cos (nt) E p c - c û â

(5.17)

This equation is only approximately correct because of the neglect of nonlinear terms in this essentially linear theory, which changes the sine or cosine term in equation 5.17 to the soiution of a nonlinear mass-spring problem of the Lienard type. A generic solution of such an equation (for example, the Van der Pol equation) is of the form: 311

A cos(nt) =>

a ,sin(co^t) + ajCosCco^t) + sin(3o)^t)- cos3(co^t)] +... (5.18) which shows the general character of the nonlinear equivalence of a cosine wave, called a limit cycle, of which the true solution must be (for those

unfamiliar with the concept of a limit cycle, the reference by Abraham explains it nonmathematically).i3 The solution reduces to the linear case for small damping.

In reality, studies by Bachus, Idogawa et al., and the current author have shown that the cosine-like oscillation of the clarinet reed only occurs at low blowing pressures, where the damping on the reed is small. At moderate and high blowing pressures, a plot of the displacement of the reed as a function of time shows inflections at both the top and bottom of the cosine wave. This non-sinusoidal motion shows clear evidence nonlinearity. In fact, under certain conditions, the motion of the reed may even degenerate to a type of quasi-random, “chaotic”, dynamics, of the kind currently under study in physics.15

The next person to include the air-fluid interactions with the reed was

Bachus, in 1963. In a seminal paper entitled, “Small Vibration Theory of the

Clarinet,” Bachus extended and refined Das’s theory considerably, although it applies only to low blowing pressure when the reed does not beat-the so-called quasi-linear regime of the instrument, since most of the 312 non-linearities are too small to be of importance at soft sound levels.

He found that the volume flow, U, through the mouthpiece slit is given by the empirical relationship:

2 4

U = 3 7 p 3 & 3

(5.19) where p is the pressure, and a is the slit opening (a perfectly inflexible reed would give a simpler exponent for p of 1/2, but the effect of reed bending as it moves and the curved lay of the mouthpiece act to modify the exponent). Later research has shown that this equation may need to be modified.16

Bachus also found an explicit expression for the minimum pressure needed to initiate vibrations to be:

p , .

(5.20) where, S is the reed stiffness,B q is the aperture opening at rest, Aq is the cross-sectional area of the tube portion of the instrument, RqJs the aperture resistance (= po / Uo), L is the tube length, p is the air density, c is the wavespeed, and Q is the quality factor (= % A, where A is the logarithmic decrement of the generated sound).

These relationships were monumental, as they provide the first really empirically derived results linking the reed and the air column. 313

Bachus’s work was extended and refined by two different groups of researchers: Wilson and Beavers (1974), and N. H. Fletcher (1 9 7 9 ).17 Wilson and Beavers considered a lumped parameter equation for the reed tip of the form:

d ^ a _ d a m- y + C— + K(a-a^) - - (P-p) d r d t

(5.21) where, C = reed damping constant, K = spring constant (= (P-p)/(a-ao)), P is the oral cavity pressure, p is the upstream pressure far from the mouthpiece, a(t) is the gap height between the reed and mouthpiece opening, and ao is the gap height at rest. They attempted to find the conditions under which equation 5.21 is unstable (i.e., oscillations begin). They assumed a solution of the form:

a(t) = (a„ - P/K) + ai(e‘"’'“ )

(5.22)

(essentially, a cosine wave), and attempted to find the conditions on co above which the solution is exponentially growing in amplitude (unstable).

They found that for lightly damped reeds (i.e., little or no lip pressure = small

K) the frequency generated was close to the natural frequency of the reed, and that for heavy damping (i.e., nominal lip pressure, large K), the reed operates at a frequeny near the resonance of the tube. These two regimes, the first high frequency, and the second low frequency, we may call the reed regime and air column regime, respectively. The first gives rise to reed squeak, the 314

second to musical sound.

The result that there are two regimes of operation of the clarinet reed is correct, but the analysis used to derive it is far too simple, and incorrect. A complete three-dimensional treatment of the air flow in the mouthpiece is needed to show the origins of the two regimes. We shall outline the essentials of the solution later, and show that the two regimes follow naturally from the geometry of the mouthpiece and the physics of the fluid (air).

The second researcher who examined the one-dimensional reed-mouthpiece problem and extended Bachus’s work wasN.H. Fletcher. He proved that the clarinet must function below the resonant frequency of the reed, and derived formulae for the deflection of the reed at a given pressure:

s

a=ao-(— ^)Po

(5.21) where, a is the reed deflection (ao is the initial opening at rest), Sr is the area of the reed, mr is the effective moving mass, and po is the blowing pressure, with the result that three regimes may be defined (c.f. equation 5.10): 315

a = ^ (increasing) for nû)«û)

-

a^ = -^a at resonance

a = — ^ (decreasing) for nû)>dO

(5.21)

which measures the reed deflection below, a t , and above resonance, which is useful in showing the amplification of the different partials of the measured sound spectrum, where ©r is the reed resonance frequency, © is the playing frequency, and n is the number of the partial in the overtone spectrum.

Returning to the matter of reed resonance, originally proposed independently by Bouasse and Das, Thompson found that if a partial of a given tone is within some range ( 5 %) of the reed resonance frequency, that partial is stabilized, and the tone seems to be more stable. Since there are a number of such partials for any given note, the interaction of the tube partials and the reed are complicated.

He also showed that for notes which contained only weak partials (small input impedance), such as the upper notes of the clarinet, the interaction of the impedance with the reed transconductance (radiated power) may act to amplify 316 the impedance peak and stabilize these notes. Thus, reed resonance plays an important part in stabilizing the upper notes of the clarinet.

The models above have all been feedback-controlled models of the air column and reed, but there are alternative models which need to be explored, such as models used to describe self-excited oscillations in collapsible tubes.is

In general, a more realistic one-dimensional model of the clarinet reed would include non-linear damping due to fluid frictional forces and air resistance.

One extension of the forced mass-spring equation of the last section is a modification of an equation used by Erneux et al. which is used to model the interaction of a mass-spring with a fluid:

■^+(y"-Pr)-^+(^'-p,)y-ey'=-(P-p)

(5.23) where y is the displacement, pr is the pressure exerted on the reed by the air flowing over it (this may be determined by aerodynamic piston theory as described by Dowell, et. al), ^ is the reed damping, P is the oral cavity pressure, and e is a damping function for the fluid.19 This equation is, in reality, a combined Van der Pol / Duffing equation.

The solution to the above equation is quite complicated and far beyond this work to summarize (the reader is referred to the original paper for details), however, it does seem that it may have possibilities of extending the work of

Wilson and Beaver by showing in a one-dimensional model the nonlinear 317 origins of the two regimes of the clarinet. It successfully predicts the existence of a threshold blowing pressure, pc beyond which a Hopf bifurcation occurs leading to a flutter solution (reed regime motion) of the clarinet reed. Specific bifurcation analysis shows that when the value of e < 0, i.e., the fluid friction acts as negative damping, then the displacement of the reed diverges (i.e., the reed moves inward toward the mouthpiece) until the critical pressure is reached, at which point flutter occurs. This is the correct behavior of the reed when it undergoes squeaking.

If the parameters are selected properly, as the amplitude of the fluttering increases, the frequency falls, as seen in the clarinet, when an increase in amplitude leads to pitch-flattening of the squeak. Finally, if the pressure is increased even more, a second state is reached which is back in the divergence range, and fluttering stops. This is the case when the blowing pressure increases to the point where the squeak is choked off in the instrument. This three stage bifurcation of the solution, from divergence to flutter to divergence as the pressure increases is called ^.-bifurcation by Erneux et al., since the bifurcation diagram resembles the g reek letter. A further study of the equation may show thata seconf low frequency regime is possible, simulating the air column regime.

We do not intend to present the above equation as a rigorous model of the reed/air interaction, but rather to illustrate that one may start from basic fluid-elastic equations and derive an equation which gives a reasonable model 318 of the motion of the clarinet reed. Hopefully, a good unified lumped parameter model Incorporating both the structure and the fluid will be forthcoming.

Two-dimensional Reed Models

Summerfeldt and Strong, In attempting to model a realistic player clarinet system took the step of updating the reed model to one approaching a more realistic case. They treated the reed as a cantilever beam (the term, cantilever, means that one end Is clamped and the other end Is free), such as If the reed were a diving board. The generic equation governing this type of free beam response Is the so-called biharmonic equation:

d ^ y d ^ y -EAK^|^=pA^,

(5.24) where E Is Young’s modulus, A Is the cross-sectional area (= bh, b Is the base and h Is the height of the cross-section), K Is the radius of gyration (K=(l/A)vz, where I Is the moment of Inertia, I = bhV12for a rectangular cross-section, where b Is the width and h Is the cross-sectional height), p Is the density, and y

Is the equilibrium displacement about the centroid axis of the beam (the centerline In a rectangular cross-sectlon).20 This equation, and the following are taken verbatim from Summerfeldt, except where noted.21

The basic solution of this equation Is: 319

y= A isin(( 3x) + A2Cos(px) + A3Sinh(px) + A4COsh(px)

(5.25) where, p = © n p A / E I , and ©„ is the frequency of the nth m o d e .22 |n fact, we used this equation in chapter four to determine the modulus of elasticity of the reed.

The particular values of the constants. A,, etc., depend upon the possible boundary conditions (free, clamped, simply supported, etc.). An illustrative example of finding the values of the A constants from experimental measurements of frequencies of the various modes is given in theShock and

Vibration Handbook, chapter seven .23 This reference also contains the mode shapes for the first four modes of a number of different boundary conditions.

In addition, Summerfeldt, et al. let the thickness of the reed vary from tip to midsection, by writing the area term in equation 5.24 as A = wb(x), where b(x) is the slope of the thickness increase, which Summerfeldt et al. found experimentally to be .081-.047x. Since A is no longer a constant, the form of equation 5.24 must be changed (adding a damping term, as well, where R is the damping coefficient);

|^[EK2.M x)|i]=pA(x)§+ ^

(5.26)

Adding damping only changes the amplitude and not the nodal positions.

The solutions to this equation must be determined numerically, as Summerfeldt 320 et al. have done.

As much of an advance as this beam deflection equation model is (we shall abbreviate It as BDE), there is room for improvement. Since most high-performance reed shapes do not maintain a rectangular cross-section, as is assumed in the BDE, (a situation true only for the wedge reed shapes, such as in Rico brand reeds), the radius of gyration, K, is no longer a constant, but rather a function of the length and possibly width along the reed. The BDE may be rewritten to reflect this:

-|^^[EK2(x)A(x)|J]=pA(x)^+

(5.27)

We shall refer to this as BDEK.

As an example of its use, we study the “wedge” shape reed (which is not really a wedge, perse ), which may be modeled as a semiellipse resting on top of a rectangular cross-section, the height of both changing as a function of x down to the tip (in other words, looking from the heel fonward, the basic heel shape height simply becomes thinner in both the elliptical and the rectangular portion).

Since moments of inertia (and the corresponding square of the radius of gyration) add linearly, one may then write: 321

kwedge(x)^ = kr(x)2 + k@(x)2

(5.28) where, kr(x) is the radius of gyration of the rectangular region, and k g (x ) is the radius of gyration of the decreasing elliptical section:

k = r 12

k = M S ^ ( 9 ;j 2 . 6 4 )

(5.29)

h r(x ) and h g (x ) are the functions relating the rate of change of the heights of the two sections, respectively.

One may substitute an area function for the reed, and the combined radius of gyration into equation 5.27 and obtain a fairly realistic mode! of the static vibration of the clarinet reed. The combined radius of gyration information is given above. One may assume a simple equation for the height increase of the rectangular and semielliptical cross-section of the form:

= \ , e '

(5.30)

A polynomial approximation for the height increase of each of the rectangular 322 and the semielliptical cross-sections of the wedge shape is given by

Summerfeldt’s equation, hr = h@ = 1/2(.081-.047x), i.e, both the rectangular portion and the elliptical portion increase in height as a function of distance, and we assume they do so at approximately the same rate, each contributing to half of the total height at any given point (the ridge shape has a more complicated polynomial approximation which we shall state later). In theory, one then simply solves the resulting equation numerically using a finite difference scheme such as Summerfeldt and Strong have used.

The principle problems with modeling the reed as a beam (wedge or other shape) by using the BDE are two-fold: 1) the ratio of the thickness (.1 mm) to the length (approximately 34 mm of the total 67 mm length) of the vibrating portion of the reed is much smaller than the 1/20 ratio normally allowed in standard beam deflection theory. As such, the tip portion of the reed is better modeled as a plate of varying thickness, 2) the reed undergoes a deflection of approximately 2 mm from equilibrium to the mouthpiece rail, which is twenty times the thickness of the reed. This means that significant stiffening of the plate occurs due to membrane forces in the midsection, and linear superposition of modes does not hold. The result is that even linear plate theory is no longer adequate to describe the mode shapes of the large deflection plate, and a more complicated, nonlinear two- or three-dimensional theory, such as the one due to von Karaman, must be used (a non-linear beam deflection theory might also be used, as the deflection is not toonon-linear).24 We shall return to this point, momentarily. 323

Two-dimensional Reed-Fluid Models

Summerfeldt and Strong added a forcing term to the BDE:

- Y A K - 0 * F . p a | ï * r |

(5.31) where,

( P t - P s ) w - [ p W r / 2 ] [ U r / A r ] ^ (forced)

F = Fr + Fb ={

0 (static)

(5.32) where F,- is the pressure in the mouthpiece, and Fb is the force due to the

Bernoulli effect (this effect says that for sufficiently small cross-sections the pressure in a tube or aperture increases as the velocity increases), Pt is the pressure in the mouthpiece, neglecting Bernoulli forces, Pg is the pressure in the oral cavity/vocal tract, w is the reed width, Wr is the mouthpiece aperture width, Ur is the flow rate through the aperture, and Ar is the aperture area.

This, in the Summerfeldt and Strong model of the total clarinet system, is the governing equation for the forced motion of a clarinet reed. This equation

(hereafter, known as the forced beam deflection equation,FBDE) reproduces the observed motion of the clarinet reed as reported by Bachus in 324

1962 (we may incorporate the Improvements mentioned in the last section to give a model which incorporated radius of gyration and area changes, as well, which we shall call FBDEK). The model simulation holds the reed closed as long as the governing equation showed that the motion of the reed carries it past the mouthpiece rail (a physical impossibility).

In the above equation, the Bernoulli force is given by the formula:

Fy = [pW r/2][U r/Ar] ^

(5.33)

Worman derived an explicit form of the Bernoulli force which incorporates reed/mouthpiece geometry:

Fy = ------^ ------2 w ta n 0 x ( t )

(5.34) where w is the reed width, and 6 is the angle formed by the inside of the mouthpiece (approximately 15° to 25 °).2 5 This led Worman to an equation similar in form to Summerfeldt and Strong’s, only in a one-dimensional form;

§+ gr J (X (t)-H ) =

2 w/i^S^ tan 6 x ( t ) 325

(5.35)

where h is the height of the resting reed, and gr is the half-power bandwidth

(see the Shock and Vibration Handbook, chapter two, for an explanation of this

term), a is +1 or -1 depending upon whether the reed is an inward beating reed

(clarinet) or an outward beating reed (lip).

This equation may be improved byinserting the Erneux, etal. Expression for

the reed:

■^+(y" - ^ - u(t)Z„)y - ey’

------2wu^S^tan(0)x(t) (5.35a)

Three-dimensional Reed Models

Given that a higher-order approximation to the motion of the free reed is

obtained via plate theory, there are still several wrinkles which must be

accounted for by a complete theory:

1) the reed undergoes large deflection, making the model non-linear, since both

lateral and in-plane forces must be considered, and these are necessarily

coupled. A model based on the large deflection theory of von Karaman model

is used here

2) the reed has a complicated tip shape, best modeled as a superelliptical shape. This reduces tip stress, but changes the edge excitation of the tip 326

3) the reed shape is often complex, and varies from a thin tip to a thick butt. It is

easier, in terms of modeling, to divide the reed into two coupled sections,

separated by the advent of the epidermis: thevibrating area (VA), and the

supporting (or clamped) area (SA). The vibrating area is a superellipse with

polynomially varying thickness in two directions (x,y). The supporting area is

essentially a semi-ellipse resting on top of a rectangular cross-section, being thick enough, in addition, to satisfy the requirements of beam theory, thus simplifying the mathematics. It may also be modeled as a half-lenticular region

(see Mansfield, however).26

4) the nonlinear plate theory presented here is only good for thin plates undergoing large deflections. Thick plates (which is what the far edge of the VA is, in essence) requires a more complex theory, such as theMidlln thick plate theory, since transverse shear and rotary inertia can no longer be ignored. To include both the effects of thin and thick plate theory, as well as large-scale deflection theory would require the use of every level of plate theory in existence. Since this document must be somewhat reasonable in coverage, we will have to make the simplifying (!) assumption that thin plate theory applies throughout the VA.

5) the boundary conditions are complicated. The portion of the VA behind the lip is simply supported on an inelastic foundation. The region of the reed between the lip and the tip is initally resting above the mouthpiece rails in empty air, but it gradually moves towards the mouthpiece rail as it is pulled inward by air pressure during flow-induced vibrations, changing it from a free to simply 327 supported boundary condition, as the reed reaches the rail. This is a called a moving boundary condition, as the boundary condition is not stable in front of the lip portion of the VA. The SA may be considered as having two simply supported ends on an Inelastic foundation

6) the lip must be considered as a Winkler type foundation, which is a spring attached to a fixed wall (in this case, the teeth is the wall and the lip is the springy membrane). The ligature is a more rigid version of the lip, since the ligature has very stiff elastic properties, usually (although string ligatures are an exception). Both lip and ligature exert forces in three-dimensions, which makes exact modeling difficult (but crucial!).

7) the tip of the reed is free until it impacts the mouthpiece, leading toHertzian contact deformation of the reed tip

8) the load on the reed varies with time after the onset of vibrations, as mentioned above. There are three types of loading on the reed surface: a) a laminar flow near the reed surface, b) a pressure varying wave, c) Bernoulli flow near the tip of the mouthpiece-reed. Each of these must be taken into account and this makes the shape of the mouthpiece critical. This combination of static and time-dependent loads causes the load across the reed face to vary in time, and is called the moving load problem in engineering.

9) the reed moduli change across the reed face, leading to “soft spring" spots and “hard spring” spots, in terms of the behavior. This variable moduli must be taken into account, since it is responsible, in part for the introduction of rotational motion in the reed 328

10) the reed may be treated as a laminar plate, consisting of different regions (in this case, vascular and supporting tissue). Although a truly realistic model must include the effects of different cell types, we have only discussed these effects in the context of viscoelasticity theory, and then only simply. We shall discuss the effects of fiber orientation however, as it does have an effect on the finer aspects of reed vibration.

11) the reed is water saturated, and inertial effects from water, as well as adjusted moduli must be included

12) water condensation occurs on the reed surface over time as water saturated air flows over the reed face. This condensation makes the mass of the reed vary. This is, admittedly, a second-order effect in the free vibration of the reed, but in the case of air flow over the reed surface, where roughness of the surface is influenced by the amount of condensation, among other things (including the natural roughness of the reed material), this added thickness and mass may be significant.

13) the reed is not perfectly clamped, and undoubtledly undergoes slipping as it vibrates. This, too, is a second-order effect.

Taking all of these complicating factors into account in a single theory of the vibrating reed is difficult, if not impossible in a document of reasonable length, however we shall comment on as many of these factors as possible.

Large Deflection Plate Model of the Clarinet Reed

Because both lateral and in-plane forces must be considered in modeling the 3 2 9 large deflection of a thin plate (unlike classical plate theory, which includes only lateral forces), a refinement which includes both of these forces has been developed by von Karaman, and is the basis for all modern large deflection theories.27

The large deflection of plates is covered in great detail in the book by Chia.28

The review and application of von Karaman’s equations by Chia is excellent in providing the background for the discussion to follow. One might also cite the classic work by Leissa as a summary of classical and extended plate theory.29

What we wish to do here is to model the large deflection of a plate whose height varies in the x and y directions. This will allow for the study of parameter variations in the vibrating areas of various commercial and hypothetical reed shapes.

Fortunately, Chia has provided such an equation for a skew plate (one in which the principal axis are not at right angles) having variable height in problem 1.4 in his book. With the skewing angle taken to be zero for both axes

(a =0,p=0), the equations presented there reduce to the desired result for use in modeling the vibrating area of a rectangular clarinet reed of variable height (we shall see below that such a rectangular reed is not too poor an approximation to the real reed, although we shall suggest refinements momentarily). These are a series of two nonlinear coupled equations:

^xxxx"^^^xxyy"^^yyyy^Y'^xx^yy ) 330

(5.36)

2) [^xxxx’^^^xxyy'^^yyyy

^[(hh^^+ 2 h \ )(w^^ + V Wyy )+ (hhyy + 2h2y X + Wyy )

+(hh^y+2h^hp(2(l-v)w^y)] =

^^g J--[P(t)-Po w„ +h(w^^Fyy+ -2w^yFxy)]- yw,

(5.37) where, F is the Airy stress function, E is Young’s mo(dulus, v is the Poisson ratio, h=h (x,y) is the height function,P(t) is the forcing function (as describecd in the last section, including reed forces and Bernoulli forces), and w is the deflection.

We have added the damping term, -yw,, to Chia’s original equation, to make the situation more realistic, wherey is the reed damping coefficient.

The above system of two coupled, nonlinear equations, is the starting point for the more exact model of the clarinet reed we shall consider in this chapter.

We shall refer to it as the reed-plate equation, RPE, henceforth. The equation for the height variation, h (x,y) which we wish to consider is the ridge shape, which is approximated (in units of meters) on an 84 by 29 unit grid to about 98% accuracy by the polynomial expression: 331

h ( x ,y ) = U\ g (Y) FCX q )

(5.38)

where,

F(X) = -2.2809x10-15x7 + 6.6053x-13x6 -7.0147x10-1 ^X^

4- 3.0129x10'^ X^ -9.9902xl0'^X^ -2.8696x10'^X^

4- 2.9681x10'^ X 4-2.4347x10^

G (Y ) = 2.11 7 3 x 1 0 '^^y7 - 4.8106x'^°Y^ 4-3.2191x10'^ Y^

- 9.8876x10-7 y '*' + i .5728 x 10'^Y^ -1.3319 x 10'"''Y^ 4- 5.9642x10-^Y -4.1182x10-5

(5.39)

and F(Xo) is the value of the polynomial for the longitudinal direction (x)

measured at the midsection of the reed (in this case, at a value of 40 out of 84).

This normalization is necessary because the polynomial fit we used for the

cross-sectional shape of the reed surface was generated across the 40 unit line

on the grid, and we have taken this to be the reference height.

To be more specific, to generate the polynomials above, a 2000 point grid

(actually, an 84 by 33 grid) of height measurements of an actual reed surface

were made using a coordinate measuring machine (we shall return to this point 3 3 2

in the section on finite element modeling). The measured surface (comprising

the front vibrating area of the reed) is plotted in figure 5.1. A line was drawn

through the center of the reed to give the center longitudinal height variation

(similar to what Summerfeldt and Strong did). A linear least squares

polynomial fit of the data produced the F(X) polynomial. Then, a cross-section

line along the middle of the reed was taken and a polynomial least squares fit was performed on it to generate G(Y). Since the middle of the reed was taken to be the origin (grid point 40 by 16), cross-sections above the line have greater thickness (height), and those below, leading to the tip of the reed, have less thickness than the origin. To normalize the thickness, the value F(X) polynomial was divided by the reference height at midpoint (X = 40 units) of the reed. The polynomials fit the data to a high degree of accuracy, as shown in figure 5.1.

These equations satisfy points one, three and four in the list of complications for the clarinet reed mentioned above for the vibrating area of the reed.

The supporting area, SA is modeled as the three-dimensional version of the combined rectangular / semielliptical combination mentioned earlier:

[E ((^+^^(97C -64))A (x,y)0] - OX 3 6 k ox

pA (x,y)0 + R |^ a t ot (5.39) 3 3 3

xIO"

100 30

X 10

80 100 30

Figure 5.1-A Comparison of Measured and Computed Reed Shape 334

where A(x,y) is the area function. Since the height does not change in the x

(length) direction, this reduces to A(x,y) = C+ A(y) = C + 1 - hm(y/w)i/z, where C is the height of the rectangle under the ellipse, and h^ is the height at midpoint

(i.e., semiminor axis) of the ellipse.

The reed tip in the RPE is modeled as a rectangular plate, which is not proper, since the clarinet reed tip is rounded. This is dealt with in a number of possible ways: 1 ) by considering the clarinet reed vibrating area to be a type of square plate with rounded corners, and thus amenable to conformai mapping, as Irie, Yomada and Sonoda have done for the case of a thin flat rounded plate.30 In this approach, the radius vector, r, of the square plate is transformed onto the unit circle by the transformation:

r— )

(5.40)

We wish to use an equivalent, but for our case, a more flexible approach, by considering the reed tip to be an ellipse in which the edges are relatively flattened (the opposite of the rounded square plate). We do"this because such an equation exists to describe a large class of variations on the elliptical shape, namely, the superelliptical equation from chapter one: 335

El b j

(5.43) where a is the length of the major axis, b is the length of the minor axis.3i The superelliptical equation reduces to the equation for the ellipse when N = 1, and ai=4a, bi=4b. The superellipse looks like a rectangle with rounded corners, depending upon which values of ai, bi and N. we saw this demonstrated in chapter one.

The idea is to transform the FPE for the rectangular vibrating area (VA) into a superelliptical equivalent region (i.e., the grid points on the square are substitued into the superellipse equation, instead). The resulting output at the far end (away from the tip) of the vibrating area is used as input for the lenticular supporting area (SA), and the resulting solutions matched at the boundary. The rounding of the back portion of the VA introduces some slight error into the VA which needs to be corrected. We shall not do this here.

To set up and solve such a problem would be an interesting problem in computational mathematics, but is far beyond the scope of the current work. We shall present a finite element approximation later on in this chapter, instead. In passing, we point out that a Rayleigh-Ritz approach might also be tried for the two sections. 336

Reed/Mouthpiece interactions: Diffuser Theory and the Origin of Reed Squeak

To reiterate what was stated at the beginning of this chapter, it is not possible, except in a highly idealized way, to understand the physics of the clarinet reed apart from its interactions with the acoustical and flow-control properties of the mouthpiece. In this section we consider such interactions, and attempt to provide a plausible origin for the excitation of the clarinet reed from a fluid mechanical standpoint, as well as provide two contrasting theories for the origins of oscillations in the double reed instruments. In addition, we shall also provides a plausible origin of reed squeak, and specifies those conditions under which it will occur.

To begin with, the mouthpiece may be divided into two basic sections, as shown in figure 5.2: a beak-like section, extending from the tip of the mouthpiece to, typically, 30-35 mm down the length (i.e., to the beginning of the table), and a section of cylindrical cross-section joined to the beak-like section which extends the mouthpiece from about 35 mm to 70mm (at which point the mouthpiece connects to the tuning barrel).

The two different regions of this two-stage mouthpiece design may be modeled as two simple types of engineering flow ducts: the beak-like region being, essentially, a modified converging-diverging plane diffuser (the flat bottom in such a “plane” diffuser is replaced by the clarinet reed, and the wall of the diffuser above the reed is curved), while the cylindrical section is a typical thick-walled cylinder. The clarinet mouthpiece is then, in a fluid mechanical sense, a two-stage converging-diverging plane walled diffuser coupled to an 337

Figure 5.2-The Clarinet Mouthpiece and Reed 338 extending cylinder.

Despite having thoroughly examined the literature, the author could not find a singie article which describes the clarinet reed mouthpiece in this fashion, which is a pity, since it might have provided insight into the excitation mechanism of the reed eariier. The closest approach is the one by Hirshberger, et al., who model the clarinet mouthpiece as a simpler Borda type mouthpiece.32

A diffuser, in general, is a type of duct used in fluid fiow control which has the purpose of modifying the iniet pressure and flow velocity in opposing ways. It was originally discovered in ancient Rome, where water flowing through the aqueducts was charged for by the diameter of pipe used. The designers found that if they angled the inner walls of the aqueducts so that they gradually increased in diameter, then they got the same water pressure but with smaller overall pipe diameters.

For a diffuser operated in the subsonic regime (i.e., beiow Mach 1) a converging (or focusing) diffuser is one in which the height decreases beyond the inlet, so that flow converges to the center of the duct. The converging diffuser has the effect of decreasing the inlet pressure, but increasing the inlet flow velocity. A diverging (or de-focusing) diffuser is one in which the height increases beyond the inlet, so that flow diverges away from the center of the duct. The diverging diffuser has the effect of increasing inlet pressure, but decreasing flow velocity. A converging-diverging diffuser is a duct which initially decreases in height (converges), and then increases in height 3 3 9

(diverges), perhaps due to a bend or sag in the duct. The resulting pressure changes depend upon which of the two types of slopes (increasing or decreasing) predominate. In the case of the beak-like region of the clarinet mouthpiece, the converging section occurs only at the very tip and is only 2 mm long, while the diverging section (=20° angling) dominates the remainder (30 mm) of the section. Thus, most of the large-stream flow in the mouthpiece is dominated by the divergent nature of the mouthpiece.

The shape of the inner upper wall curvature and side-to-side height asymmetry in the divergent section of the mouthpiece beak adds three-dimensional complications to the fluid flow, but do not appreciably change the physics of the reed-mouthpiece interaction. They have been added to the original straight-walled purely diverging (i.e., ramped increase in height) diffuser design of early clarinet mouthpieces due to experimentation by mouthpiece designers. The inner longitudinal curvature changes the point of flow separation (to be discussed later) while simultaneously enhancing vortex formation.

The practice of making one side of the mouthpiece interior higher (i.e., farther away from the reed) than the other side has the function of correcting an asymmetry in pressure in the body of the clarinet which results because air passing in the vicinity of each tone hole loses pressure and energy due to vortex formation and acoustic streaming. Since there are more tone holes on the right side of the clarinet than the left side (viewed from the tip down, with the speaker hole on the bottom of the clarinet), the pressure on the right side of the 340

tube emerging into the connecting portion of the mouthpiece is lower than on the left side of the tube. If the beak-like portion of the mouthpiece were perfectly

symmetric in height across the inner curvature, the right-left pressure difference

in the tube would be maintained in the flow as it emerges from the mouthpiece

into the oral cavity. If, in addition, the reed were also symmetric in height and

uniform in its bending moduli the pressure difference in the tube would normally

impart a slight rotational force to the reed surface, leading to off-axial torsional

motion of the reed, which would yield an audible “buzz" to the sound.

This pressure gradient situation is usually modified in one of two ways, either by modifying the reed, or by modifying the mouthpiece. Typically, both the reed and the mouthpiece are modified, however. One may make the side of the reed

(left side, as viewed from the top of the reed) which has the higher mouthpiece/air column pressure slighty thicker to offset the increased pressure on the reed underside (this is the reason for the right-left side thickness asymrnetry found in the heart region of almost all high performance reeds).

Alternately, as skilled mouthpiece craftsmen do, one may increase the mouthpiece chamber height on the side of the mouthpiece opposite to the side having the higher tube exit pressure. In this case, since the-^eft-hand side has the higher tube pressure, the right-hand side of the mouthpiece is raised. This has the effect of increasing the pressure on the right-hand side, since as has been mentioned, in a diverging diffuser, the higher the divergence angle, the greater the pressure increase. This heightening compensates for the tone-hole induced pressure gradient. The result is the restoration of an even pressure 341 across the reed surface, and a smoother sound.

The requirement for an even pressure gradient across the reed surface is, in fact, the governing principal for all of the shapes used in high performance reed design, and we state it as such (we shall have more to say about this later):

Principal One of Reed design: the requirement for the optimal reed design for a given clarinet and a given person is that the pressure gradient felt across the reed surface be essentially constant so that the reed may maintain essentially transverse motion with not too much damping.

Given this duct-reed system design for the clarinet mouthpiece mentioned above, a more detailed examination of the air flow through the mouthpiece indicates that there are three basic types of physical modes of air flow in the clarinet mouthpiece. In addition to the vibrationally induced motion of the clarinet reed and air stream (essentially a plane wave which propagates up and down the tube), the air flow near the surface of the reed is viscous, resulting in the formation of a laminar flow layer (a layer of air flowing parallel to the reed surface) near the reed face, which is on the order of the thickness of the reed (=

.1 mm). Under most conditions, this laminar layer is flat, and the flow may be considered quasi-steady. If the reed is too rough, the material too inhomogeneous, or the sides of the mouthpiece opening leak, the laminar layer undergoes a transition to turbulence (characterized by chaotic air flow patterns and vortex formation, perceived as white noise ( or a gurgling sound) 342

superimposed on the clarinet tone.

The third type of fluid response is the Bernoulli effect, in which the

pressure between two parallel plates (here, approximated by the reed and

mouthpiece tip) changes proportionally as a function of the separation distance and air speed between the two plates (as the height becomes larger at constant air speed, the pressure falls; as the air speed increases, the pressure decreases). This has the effect of pulling the reed closer to the mouthpiece opening, since the pressure between the two “plates” is lower than the pressure outside of the two plates. Worman has implicated this effect to be the one primarily responsible for the flow-induced vibrations in the mouthpiece.

Although the Hirschberg et al. article cited earlier has shown that there is some reason the question this assumption farther into the mouthpiece, it unquestionable that the effect exists near the mouthpiece tip.

Given the fact that the mouthpiece is a converging-diverging diffuser, it becomes possible to speculate on how the excitation of the reed might actually occur. White shows that for such a reentrant sudden contraction large amounts of flow separation occur, vorticies form at the entrance and subsequent losses as well. Rounding decreases the loss, which is one reason why the upper mouthpiece entrance is rounded. The reed, however, not being rounded, means that a vortex probably forms near the tip of the reed.As the air passes down a diffuser, the pressure increases, while the local velocity decreases. If the diffusing angle is sufficiently great (above, say, 7 degrees) then the combination of fluid friction and slowing velocity flow near the upper wall 343 causes the velocity to decrease to the point where it actuaily reverses direction near the upper wali, somewhere near the maximum curve of the upper mouthpiece wall. This flow reversal causes the air to “separate” from the upper wali (the mouthpiece diffuser forms what is known as an adverse pressure gradient on the interior of the mouthpiece, which is a pressure gradient supportive of flow separation).

This leads to the formation of a vortex of low pressure somewhere downstream in the mouthpiece at the point of separation, which is usually at the highest curved point in the mouthpiece chamber.33 Due to the converging nature of the mouthpiece tip, plus the motion inward of the reed as the pressure increases, and its reentrant nature, a second vortex probably forms near the reed tip. Such vortecies are actually observable in flow visualization studies of tubes having roughly the same shape as the mouthpiece.34

As the pressure increases, the size of the downstream vortex increases, and moves farther downstream towards the tube. The upstream (near the tip) vortex moves closer to the mouthpiece opening as the reed bends inward. When the threshold blowing pressure is reached, the upstream vortex (region of low pressure) finally reaches the opening of the mouthpiece. Since it is of lower pressure than the atmospheric pressure, when it is ejected, the pressure at the tip suddenly recovers, forcing the reed open. The downstream vortex is then ejected into the tube by the sudden increase in pressure at the tip, and the cycle begins again. Eventually, a plane wave is formed, and a steady-state osciilation is achieved. 344

The limit cycle quality of the reed motion, with the slight rebound at the end of each cycie, is caused most likely by the non-linear interaction of the fluid friction with the reed damping. This is a matter which needs to be examined further.

Reed Saueak

It has been found that for tubes with cuffs around them (much as an artery with a sphygmomanometer cuff around it), which is a good model for the lip surrounding a clarinet reed/mouthpiece, two vibrational regimes actually occur:

1) a low frequency regime, characterized most likely by vortex shedding, similar to what we have described for the clarinet reed excitation mechanism, and 2) a high frequency regime. The mechanism for the high frequency oscillations has been studied extensively, and a number of theories exist. One such theory, due to Grotberg and Reiss, is of particular application to woodwind reeds, as the parallels will be obvious (actually, their theory applies exactly for the excitation of double reeds, and explains the origin of reed squeak in single reeds).

Consider the bassoon reed in figure 5.3. The two halves of the reed are really shallow shells (or curved plates), and the motion of each half is described by the Donnell shell equations: 3 4 5

Bassoon reed Top Plate L

Bottom Plate C j L s . - : T=Top Shell Bassoon reed approximation U B=Bottom Shell (.ijr Velocity) /O rigin al position

■ ''Divergence (pulling together)

Bassoon Below critical airspeed I f Fluid friction V > in this part of the reed causes squeaks Above critical Clarinet airspeed Mouthpiece

Reed Flapping Flutter squeak _ No damping t or high fiuid friction Mouthpiece Mouthpiece

Clarinet reed Clarinet reed divergence

Figure 5.3-Flapping Flutter and Reed Squeak 346

D V ^w + = F K

V ^ c p - EhV^w = 0 K (5.44) where,w is the displacement, cp is the Airy stress function (given by a complicated series of equations which we shall not reproduce), and

D = - ^ 1 2 ( 1 - v 2 )

^ I 'f

v2 = J -rA (X â i_ ) + R AB ,?« R , A 5 a 9 1 3 R , B 5/3 (5.45) where A, and B are vector lengths, a and p are principal coordinates, and Ri,a are positional vectors (see Leissa for details).3s

The theory of Grotberg and Reiss proceeds as follows: irnagine that the shallow shells of the bassoon reed may be replaced by two parallel plates (an acceptable approximation in two dimensions), as shown in figure 5.3.36 The plates can move in a number of different modes, but we shall only examine the case, as do Grotberg and Reiss, of the symmetric and antisymmetric modes of the two plates. Since the air is a compressible fluid , three equations are 347

sufficient to describe the flow:

Continuity: = 0 di ds Shell wall condition: W. + 0 W - 0 = 0 (for z = W) t X X z

Unsteady Bernoulli equation: P = + -V 0 # V 0 + 2fO]

(5.46)

where Q = wall speed/speed of sound, 0 is the velocity potential, S = fluid

velocity/speed of sound, W is the displacement of the plates, P is the fluid

pressure. Pa is the driving pressure, and f is a scaled friction factor. These

equations are subject to the boundary condition in the middle between the two

plates,

= 0 at z = 0 (symmetric)

0 = 0 (at z = 0 (antisymmetric)

(5.47)

The fourth and final governing equation should be the displacement equation

for the parallel plates, W(x,y) but we may substitute equation 5.44 (the shallow

shell equations) with an additional term to model the added mass due to the lips

for the plate equation in Grotberg and Reiss’s treatment.

It would take a fairly advanced treatment to reproduce their arguments 348

(originally, for parallel plates) for the case of parallel shallow shells, but, nevertheless, the essential results would be the same. The end result is that for sufficient values of the ratio of fluid damping to reed damping, (Grotberg and

Reiss’s Y term), the two parallel shells will undergo a transition at some threshold blowing pressure from a static divergence (i.e., the two shells get close together with higher blowing pressure) to a type of symmetric flapplng-flutter of the two shells. This type of flapping-flutter is exactly what is seen in stroboscopically slowed motion studies of the vibration of an oboe reed which I have made. The two shells flap together in a rather complicated fashion, due to nonlinear flow, no doubt. The slight divergent diffuing nature of the oboe reed inner shell complicates matters, but we assume that this is a second-order effect. The Grotberg/Reiss model seems to do a very good job of describing, ab initio , the motion of double reeds.

We may extend this treatment to the case of the clarinet reed. The clarinet mouthpiece-reed combination presented above for the beak-like section of the mouthpiece is identical in structure to the case of the two parallel plates (the bottom plate being the reed, the upper plate being the mouthpiece upper inner wall), in which the upper plate, the mouthpiece upper wall, is immovable or inelastic (figure 5.3). In addition, the upper plate is placed at an angle to the lower plate. This diffusing (angling) of the mouthpiece inner upper wall shape

(plate) has the effect of changing the pressure along the upper wall plate

(increasing the pressure along its length).

The primary effect of the rigid mouthpiece upper wall and the diverging angle 349

is to change the boundary conditions, the Bernoulli pressure equation, and the

plate equation. Since the upper wall does not move, but is in a fixed shape to

begin with, the equations for the upper (W+) and lower wall (W-) are given by:

W+$ - 0 = 0 for z = W+ X X z W - +W-0 - 0 = Oforz = W - t XX z (5.48)

The Bernoulli equation is modified not only by the diffusing pressure gradient,

Pd, but also due to pressure loss along the sides of the mouthpiece, Pg. Pj is determined by the upper wall equation and the resulting 0. Pg must be determined empirically:

p = p - [0 + iv $ .7 0 + 2f$l - P a '■ t 2 '

(5.49) where, f is the fluid friction. W- may be from the Grotberg andReiss plate equation, or the more exact equation 5.37 above.

The solution of the above problem is very complicated, due to the pressure gradient in the diffuser, nevertheless, we expect the basic results as in the double reed case should hold, except that only the bottom plate undergoes longitudinal flapping-flutter (figure 5.3). This reed flutter occurs at high frequencies, due to the high modulus and damping of the material, and is very near the reed resonant frequency. This flapping-flutter is commonly called 350 reed squeakfor the clarinet reed. It may be caused by two different situations:

1 ) the damping of the lip goes to zero, suddenly (in fact, if reed damping is set equal to zero in the equations above, flutter occurs), 2) the reed shape is such that the fluid friction along the sides of the reed increase beyond a critical value.

This is the case of a reed which is warped inwardly. Thus, squeaks occur during performances when the player suddenly releases lip pressure ( while flattening the lip parallel to the reed surface), or if the reed warps. As an experiment, one may make a squeaky reed by deliberately bending the sides of a reed inwardly by applying thumb pressure to the sides of the reed.This causes the sides of the reed to press down on the mouthpiece rails, increasing the local fluid friction along the side. This should promote reed squeaking.

The opposite is true as well. If the reed is bent by hand (with not too much pressure!) so that the sides of the reed are slightly bent away from the rails, squeaking becomes much less likely (lip damping excepted). This is the basis for the Listoken method of bendingreed s.37

Starling Resistor Model of Up / Reed

In the discussion on reed squeak, we mentioned that the clarinet reed and mouthpiece could be modeled as two parallel plates, one elastic, the other fixed. In this section, we extend this analysis. There is a huge literature on parallel plate / fluid interaction, of which, apparently, the musical acoustics community is largely unaware.

A bit of nomenclature is appropriate. In fluid flow studies of flexible elastic 351

tubes, either end of the elastic tube Is typically placed over a rigid (Inelastic) tube and air Is let In through the first rigid tube and removed through the second

E rigid tube. Let us use the terminology, to denote the properties of the upper

and lower elastic plates In the parallel plate approximation to the thin-walled elastic tube, and similarly, y for the Inelastic plates of the rigid tubes (the upper

term of the fraction denotes the properties of the upper plate; the bottom term denotes the properties of the upper plate). The usual arrangement for thin-walled elastic tubes used In the blomechanlcal studies mentioned above. Is

I°E°I where the dots Indicate coupling In such a way that the tube ends rE°I exactly meet each other but one does not fit over (sheath, or overlap) the other.

Such a configuration of rigid and elastic tubes Is called an unsheathed or end-to-end Starling resistor In blomechanlcs (5.4).

Comparing the model of the Starling resistor to the clarinet mouthpiece and reed shows that the mouthpiece and reed, using the above nomenclature, form

I an y pair. The rigid Inlet tube In the Starling rerslstor model Is a first-order

approximation to the oral cavity, since the jaws make the cavity rigid.

The rigid outlet tube Is simply the cylindrical second stage of the clarinet mouthpiece mentioned earlier, which may be modeled, as In the other tube 3 52 •Starling Resistor-

,T u b e s . I ■Rigid Eiastic Rigid Rigid y

I + sheading ^ -sheathing ; o sheathing

Parallel Plate Approximation

k E -l

i-t-E-l

Figure 5.4-The Starling Resistor Model 353

cases, as two rigid plates in cross-section. Thus, a two plate approximation to

1° 1° I the complete clarinet system is represented by the system, go j • We call this

an unsheathed Starling half resistor (SHR) model, since the rigid upper wall of the mouthpiece is not a resisting structure, as it would be if it were elastic.

In a real clarinet system, the lips are not coincident with the mouthpiece, but rather go over them, which would be equivalent to the rigid inlet tube forming a sheath over the elastic tube without collapsing it.

For later purposes, a nomenclature for the sheathing phenomenon mentioned above needs to be developed. The clarinet reed-mouthpiece is sheathed by the lips, as mentioned above. If we denote the sheathing of the n + 1 plate by the nth plate by a plus sign, and the sheathing of the nth plate by the n + 1 plate by a minus sign, then the clarinet reed / jaw / tuning barrel

I+E° I combination may be represented as: g o j , since the diffusing part of the

mouthpiece merges smoothly into the cylindrical second stage of the mouthpiece as described above (the reason for the o coupling). There are, of course many different types of possible arrangements of plates and connection, and these form a natural algebraic group, whose representation we leave for the motivated reader. It may, in fact, be possible to use group theoretic representations and symmetry classes to predict which arrangements may be used as musicai exciter. 354

This structural classification is also naturally represented in graph theoretic terms. Let a filled in dot represent an inelastic plate, a hollow dot represent an elastic plate, a straight vertical line represents a upper plate to lower plate coupling, and a horizontal straight line represents coupling between plates of different tubes. An outward curving line between plates represents the first plate sheathing the second plate (we call this a positive sheath), and an inward curving line represents the sheathing of the first plate by the second plate (we call this a negative sheath). Thus, the symbol.

represents an inelastic plate positively sheathing an elastic plate (as In the case of the lip and clarinet reed). An example of the clarinet reed / lip Qaw) coupling is:

(Positive Sheath) I n e l a s t e------_ M o u th p ie c e J a w ^ ^ "Plate” "Plate" ^ m centerline t ..... Î Inelastic Elastic S t e " (PosiMve Sheath)

We may, of course, also use the less graphical nomenclature presented above to indicate the same thing, where a positive sheath is indicated by a plus 355 sign and a negative sheath is indicated by a minus sign. Thus, the situation depicted graphically above is also represented algebraically as: where

the division sign represents the centerline of the upper and lower plates.

Another example is a sheathed Starling full resistor (two elastic plates in the middle),

# . P Q i The Starling resistor = I-E + I

l-E + l

The nomenclature and graphical methods should be straightforward enough to use without further discussion. In this nomenclature scheme, the clarinet

1+1° I structure is ■ as mentioned above.

The above digression on nomenclature now allows for the algebraic or diagrammatic classification of any type of music reed (cane or lip reed). The distinction of inward and outward beating reeds mentioned at the beginning of the chapter is easily modeled by these methods. The clarinet reed, as

I + r I mentioned has the structure: - g- . The trumpet (lip reed) on the other hand. 356

r E - i has the configuration; jo g i j , since the oral cavity is coincident with the inside

of the lip (the o symbol means that the two plates are continuous and do not sheath each other), while the inelastic trumpet mouthpiece sheathes the two elastic lip plates.

If the upper portion of the clarinet mouthpiece were elastic, then the symmetry between the two cases would be more apparent: j (clarinet, or

I ° E “ I inward beating reed) vs. p -g-_ j (trumpet, or outward beating reed), or

diagrammatically (the straight lines indicate the two plates join in a straight line without sheathing, as the symbol o also indicates):

t % JaVj Clarinet Reed/Meuthaieee, Lever Meutheieee Stase

_ P t ■ c 5 ‘ “ “ ' é Jaw, Lips, Trumpet Mouthpiece

if we use the shorthand of simply noting the values of the couplings, then we 357

+ 0 = _ may represent the clarinet as: and the trumpet (lip reed) as — , or by

symmetry, simply 2 (+, o ) and 2 (o, -), for the clarinet and trumpet, respectively.

As air flows from one model tube section to another (i.e., lip to mouthpiece, etc.), if the walls (or plates) of the n+1 tube deflects inwardly for positive sheathing we call this a negative equivalence between the sheath and the motion (+ sheath = - deflection, or - sheath = + deflection), since the signs are opposite (i.e., multiplied by -1). If a positive (negative) sheathing produces an outward (inward) deflection, we call this a positive equivalence (+ sheath =

+ deflection, -sheath = - deflection).

Thus, if the inward deflection is represented by an I, and the outward deflection by a 0, then for negative equivalence (as in the clarinet and trumpet),+ sheathing implies an I deflection, and - sheathing implies a 0 deflection. Neutral sheathing (the case where, instead of +, or - there is a o) produces no deflection (we shall symbolize this by 0 ).

For the clarinet, the situation is (I, 0 ) in the direction of forward flow (the first I is the inward deflection of the oral cavity to reed coupling response, and the second 0 describes the response of the reed-beak coupling to the cylindrical lower half of the mouthpiece). For the trumpet, it is 0( , 0). For reverse flow, the signs are reversed.

Since the n to n+1 plate sheathing models for reed/lip interaction mentioned above yields (I, 0) for the clarinet in the downstream direction, and (0,0) in the 358 upstream direction, the full description is actually [(I, a), (0 ,0 ))]. This is equivalent to (-,+) in the Fletcher notation mentioned earlier in this chapter, since he is only interested in the oral cavity to reed transition, and the second mouthpiece to outlet tube transition (0 in both cases) is ignored. It should be noted that our use of + and - signs are for different purposes (they represent sheathing) than Fletcher’s, and the reader should not be confused by the using of the same symbols to mean two different things (our sheathing and his deflection). For the trumpet, one has: [(0 , 0), (0 , 1)] = (+,-) in Fletcher’s notation, as predicted.

Thus, the reason for the different responses of the inward and outward beating reeds are seen to be consequences of the symmetry in the coupling responses between the sheathing and plate approximations introduced above.

Since the clarinet case is the negative of the trumpet parallel plate structure, the physics is symmetric across the imaginary axis (a point we shall not prove, however, Fletcher makes this point clear), yielding the empirical and theoretical observation that the clarinet reed vibrates below resonance, and the trumpet lip reed vibrates above resonance.38

The methodology introduced above allows for the consideration of any

theoretical type of reed / lip structure. The bassoon reed, for instance, is represented diagrammatically as: 359 t: T : :

Jaw, Lips, and Staple of the Bassoon

It should be possible to deduce Fletcher’s equations for reed excitation from

these diagrams, but we shall leave this for future research.

Reed Shape Effects

In this section we shall briefly examine the various influences of reed shape on the clarinet sound. We shall compare these parameters to the rated quality of the clarinet reeds in the experimental set {R} from chapter four.

To begin with, the question should be asked as to why modem clarinet reeds have only the two shapes, wedge and ridge, that they do. The answer is to be found in the discussion of diffuser theory above. In a diverging diffuser, such as the clarinet mouthpiece, pressure increases as one progresses down the diffuser. It makes sense that if one wishes to keep a constant pressure gradient over the reed surface, the thickness of the reed must increase as a function of the pressure increase inside the mouthpiece, so that higher pressure is met by stiffer (thicker) material.

All high performance reeds have the ridge shape of some sort as a necessary consequence of trying to keep the force on the reed at any given point roughly constant down the length of the mouthpiece. Thickness must increase as the 360 mouthpiece pressure increases to compensate.

The evolution of the clarinet reed has been such that this is the universal design principal in clarinet reed shapes. Historically, this has been borne out. The earliest reeds were coupled to mouthpieces with very small diffusing angles. The result is that the optimal design was essentially a flat plate. As the mouthpiece angle grew larger, the thickness down the midplane of the reed grew larger as well. A negative impression casting of the inside of a mouthpiece would reproduce (all things being equal) the optimal shape for the associated reed for that mouthpiece (not to scale, of course). All things are not quite equal, however, and this forces modifications from this negative Impression design principal.

The ridge shape typically has thinner sides compared to the middle. From the discussion on reed squeak, the reason for this design should be obvious. The thin side design lowers the fluid friction along the sides of the reed (less mass implies less resistance) and lowers the likelihood of reed squeak. Of course, if the sides are too thin, then lip contact might be lost, resulting in the lip damping going to zero, producing squeaks in this manner. The balancing of the thick middle with the thin outer sides is an art.

We have already commented on the left-right side thickness asymmetry in modem reeds in the introduction to this chapter. This asymmetry attempts to compensate for frictionally-induced losses in the air column which would normally produce asymmetric loading of the reed surface and undo torsional motion, resulting in reed buzz. Different reeds show different degrees of asymmetry, as we shall show shortly. It is possible to remove the asymmetry and use a balanced reed if the ridge area is made longer to 361

distribute the stress more evenly across the reed surface. This is the reason why

Grand Concert reeds, which show almost no asymmetry, nevertheless, do not buzz.

These three principles, negative impression shape design, side friction reduction,

and load asymmetry reduction, are the three governing principles for optimal reed

design, and are shown in all high performance reeds.

Are better reed designs possible? This question may only be answered when the

associated question, “are better mouthpiece designs possible,” has been answered.

We shall not attempt to answer that question in this document on reeds, however.

In terms of a statistical sample of reeds, the variables were: 1) heel thickness (mm),

2) shoulder (uncompressed), 3) shoulder (compressed), 4) tip, 5) 16 mm midpoint

thickness (mm), 6) 8 mm left thickness, 7) 8 mm right thickness. The statistics are

given in table 5.1. There are interesting interbrand variations in terms of structure,

shown in the scatter plots in figure 5.5 to 5.12. Vandoren Black Master reeds are 3 mm

shorter than the other reeds (64 mm vs. 67 mm), and they also have the greatest

degree of asymmetry. 8 mm (L) - 8 mm (R) = .692 mm (mean) for Black Master. By

contrast, Grand Concert Reeds have almost no asymmetry with a mean asymmetry of

only .0223 mm. As we have pointed out, however. Grand Concert reeds also have a very broad ridge which cancels out the rotational inertia of the air column pressure

asymmetry.

In terms of heel thickness. Grand Concert has the thickest heels (as its “thick”

designation might imply), at 3.14 mm, while Black Master has the thinnest, at 3.034

mm. The same pattern is evident for tip thickness, with Black Masters having a

.129 mm mean tip thickness, while Grand Concerts has a .155 mm thickness. 362

Summary statistics for Summary statistics for Summary statistics for Heel Shoulder(U) Shoulder (C)

Mean 3.1228625 Mean 3.2096250 Mean 3.1818500 Median 3.1540000 Median 3.2025000 Median 3.1660000 Numeric 80 Numeric 80 Numeric 80 Skewness -0.68836235 Skewness -0.15937069 Skewness -0.I7I99808 Kurtosis -0.17612425 Kurtosis 1.21 13954 Kurtosis 0.98803471 StdDev 0.14059572 StdDev 0.09654833 StdDev 0.10143940 Range 0.61300000 Range 0.57500000 Range 0.57900000 Variance 0.01979528 Variance 0.00932158 Variance 0.01028995 Min 2.7600000 Min 2.8680000 Min 2.8430000 Max 3.3730000 Max 3.4430000 Max 3.4220000

Summary statistics for Summary statistics for Summary statistics for Tip 8mm (L) 8mm (R)

Mean 0.13642286 Mean 0.38132500 Mean 0.32952500 Median 0.13600000 Median 0.38300000 Median 0.33550000 Numeric 70 Numeric 80 Numeric 80 Skewness 1.1558933 Skewness 0.33919626 Skewness -0.21152404 Kurtosis 4.1815250 Kurtosis 0.17129777 Kurtosis -0.92688029 StdDev 0.01348840 StdDev 0.02411144 StdDev 0.02900575 Range 0.09100000 Range 0.12000000 Range 0.11000000 Variance 0.00018194 Variance 0.00058136 Variance 0.00084139 Min 0.10400000 Min 0.33600000 Min 0.27200000 Max 0.19500000 Max 0.45600000 Max 0.38200000

Table 5.1-Statistics for Reed Structural Measures 363

3.4

3.2 O O O (D oo

ci 3.0

Oo

X 2.8-

2.6 020 40 60 80 1 0 0

Reed Number

Figure 5.5-Reed Structural Factor: Heel Thickess 364

3.5

o 3.4

E 00 E 5 3.3 H T3 O o M0) Or ° O O S 3.2 « P o ,p " o o ® Oo a. o® ° o O o oo E o o o O c 3.1 ~ 3 o 2 3.0- 3 £o (/) 2,9-

I I T 20 40 6 0 80 100

Reed Number

Figure 5.6-Reed Structural Factor: Uncompressed Shoulder Height 365

3.5

3 .4 -

E E 3 .3 - •0 10a V) 3 .2 - £ a 003 oé> E Oo o Shoulder (C) o U 3.1 - o> 2 3.0 O3 £

2.8 0 20 4 0 60 80 100

Reed Number

Figure 5.7-Reed Structural Factor; Compressed Shoulder Height 366

0.20

0.1 8-

0.16-

Tip (mm) o o Oo 0.14 Oo O 00 o

0.10 0 20 40 60 80 100

Reed Number

Figure 5.8-Reed Structural Factor: Tip Thickness 367

1.3

E E

a. 1.2

E oo S

E (OE 03 S 01 (P o 0 0 3 !

20 40 60 80 100

Reed Number

Figure 5.9-Reed Structurai Factor: Thickness 368

0.50

0.45

E E o o

0.40

E coE SP o o 0 .3 5 - CD oo

0.30 0 20 4 0 60 80 100

Reed Number

Figure 5.10-Reed Structural Factor: 8mm, Left Thickness 369

0.40

00 0.3 8-

0.36 - 0

E E 0.34

S 00 a 0.32- Ë O o E 00 0.30

0.2 8-

0.26 0 20 40 60 80 100

Reed Number

Figure 5.11-Reed Structural Factor: 8mm; Right, Thickness 370

2300

o o

2200 o o o o O CD o o o o Oo o o o o o o o cr 2100 2 o o Ü. o o 00 a o o 0 0 0 00 o o CD o o o CD 2000 o o o o o o o o o 1900 T“ 20 4 0 60 80 100.

Reed Number

Figure 5.12-Reed Structural Factor; Natural Frequency 371

Despite all of these variations, the three governing principals for reed high performance reed design are adhered to by each of the brands of reeds tested, since each has a thick midsection, asymmetric sides (except where compensated for by broadening of the ridge), and thin sides.

In terms of correlation with quality, the following correlation matrix Is found;

Pearson Product-Moment Correlation Heel Sh(u) Sh(c) Tip 16mm Qual D-B S-H Buzz Sque Weak Asym Heel 1.000 Sh(u) 0.2 40 1.000 Sh(c) 0.2 32 0.987 1.000 Tip 0.1 90 -0.147 -0.141 1.000 16mm ■0.316 -0.108 -0.108 -0.1011.000 Quai -0.116 -0.282 -0.288 0.061 0.071 1.000 D-B 0.074 0.119 0.106 0.088-0.002 -0.334 1.000 S-H 0.014 -0.065 -0.054 -0.137-0.226 -0.109 -0.376 1.000 Buzz -0.233 -0.269 -0.268 0.134 0.101 0.6 35 -0.199 -0.018 1.000 Sque 0.131 -0.110 -0.120 0.247 -0.142 0.466 -0.116 0.080 0.461 1.000 W eak -0.105 -0.309 -0.313 0.095 -0.085 0.739 -0.593 0.137 0.548 0.347 1.000 Asym -0.066 -0.044 -0.063 -0.356 0.252 0.211 0.0 50 0.1 0 7 0.274 0.087 0.005

Figure 5.13-Correlatlon Matrix for Structural Measures

Interestingly, the result that the thicker the heel, the thinner the midsection seems to be a trend (the correlation between heel thickness and thickness at 16 mm Is -.316).

There Is a weak negative correlation between thickness of heel and degree of buzzlness (thinner heels are less prone to buzzlness). This tallies with the negative correlation that thinner shoulders also yield less buzzlness.

Thicker tips seem to be more stable and less prone to squeaking. This Is a finding well known among clarinetists. The reason Is that the degree of fluid friction necessary to Induce flapplng-flutter Is greater with Increased reed mass at the tip. Also, In high performance reeds, there seems to be a trend of the thicker the tip the less asymmetric 372

the reed.

In terms of reed quality, there seems to be a mild negative correlation between

thinner shoulders and higher quality, although the correlation is very small. Also, there

is a mild correlation between reed asymmetry and quality, so that the more asymmetric

the reed, the better the quality, although, this too, is a small correlation. It is gratifying

to see that, although all of the correlations between reed structure parameters and

performance rating criteria are small, the largest is between asymmetry and buzziness,

as predicted.

In terms of buzziness, one more parameter, the skewing or grain angle, must be

mentioned. Veselac found that there was a correlation between large grain angles

and poor reed performance. One reason for this correlation may be found in the

computer simulations of Liu and Huang for the free vibrations of thick cantilever

laminated plates with step-change thickness.39 The clarinet reed, as mentioned in chapter two, to a first approximation, may be modeled as a series of stacked crystal-like plates, as in the Liu / Huang study. In that study they found that if the

laminated regions contain fibers (as the clarinet reed does), then increasing the skewing angle of the fibers increases the frequency of the first torsional mode, but decreases the frequency of the first bending mode.

In a perfectly symmetric clarinet reed, one would expect that as the skewing angle increased, the amount of torsional motion of the reed should decrease and the reed should be more stable in the bending modes (less buzzy), and yet, the correlation between skewing angles greater than zero and reed buzz persists in all of the measurements made in this study. A possible reason is however, that in the Liu / 373

Huand study, the first bending mode always occurs at a lower frequency than the first torsional mode, while in the case of the clarinet reed, the lowest bending modes (at

760 Hz and 790 Hz) are distorted due to the asymmetry of the reed to resemble torsional modes. We call these pseudo-torsional modes, and they lie below the first major pure bending mode, which is at about 2000 Hz.

What probably happens in the case of the clarinet reed is that as the skewing angle increases the frequencies of the lowest pseudo-torsional modes decrease as they would if they were pure bending modes. The first real torsional mode, which occurs at about 4000Hz probably shows the increase in frequency which the Lui / Huang study predicts, but the pseudo-torsional modes, being really bending modes decrease in frequency, which is a sign that they are easier to excite. This is one way in which the correlation between skewing angle and reed buzz may be explained.

To summarize, it does appear that the data supports, if mildly, the design criteria for high performance reeds suggested in this chapter. Human variables, such as embouchure pressure, and lip thickness could not be added to this study.

Finite Element Computer Simulation of the Clarinet Reed

In general, the Navier-Stokes equations which model the full fluid flow in the mouthpiece, and the Midlin Thick Plate equations which model the reed deflection cannot be solved analytically. This naturally leads to the use of numerical methods.

In the final technical section of this chapter, we develop a very powerful computational model for clarinet reed deflection based upon finite element technology.

We shall briefly describe the computational procedure, and then display the calculated 374 mode shapes of a struck clarinet reed without lip damping which lie within the range of human hearing. The calculated frequencies of these modes are compared against empirical measurements of the reed vibrational frequencies.

In order to create the computer model of the clarinet reed, the 84 by 33 grid of measured points for the clarinet reed VA mentioned above were entered into the finite element program, ANSYS. These grid points (called nodes in the software package) were used as vertices of soiid rectangular elements which were joined together to form a solid model approximation to the clarinet reed (figure 5.14). In fact, thé model of the reed and mouthpiece used earlier to explain the excitation mechanism of the clarinet was generated using this software. We have added a lip and teeth to the model (figure

5.15) using material properties appropriate for each (the lip is a water-filled membrane, and the teeth were given material strength from handbook data on teeth).

In calculating the mode shapes for the cantilever reed shown in appendix b, we have removed the lip and teeth, since they give rise to very low frequency (1-2 Hz) lip modes as weil as reed modes.

The reed was subjected to a sudden impact on the computer, and the resulting frequency response and mode shapes were calculated (figure 5.16). The calculated mode shapes which are in the audible range of hearing are shown in appendix b.

As mentioned earlier, and as shown in appendix two, the lowest two vibrational shapes for the reed appears to be pseudo-torsional modes. The reason that these are pseudo-torsional modes is that it has been shown that for a rectangular plate the first bending mode must be lower than the first torsional mode.^o These lowest two modes resemble torsional modes, but the side to side motion is an artifact, we believe. 375

Reed Vibrating Area (VA); 2904 point grid

X 10'

2.5 H

i l .5 H 0 1

0.5 H

too 40 X grid

y grid

Figure 5.14-Flnlte Element Grid Measurements 3 7 6

Clarinet Reed: basic model

[Note: colors are altered for printing purposes]

epidermis \ V \ \ /

sclerified re gion teeth

applied force (arrows)

elliptic- parabolie reed shape

Figure 5.15-Flnite Element Model of the Reed 3 7 7

Hammer

Microphone """T——

Ligature

Cut Mouthpiece Spectrum AnaJizer

Figure 5.16-Modal Measurement Set-up 378 induced by the asymmetry of the reed, which we measured in chapter four. This would seem to be confirmed by the very large acoustic strength of the third mode, which is known to be purely bending.^i

Many other questions such as the affects of scraping a reed at different point, the changes in mass due to bacterial, etc. may be studied using this technology. We can only summarize the results of computer experiments briefly. Bacterial growth

(modeled as a density increase over parts of the reed over time to mimick the growth of the bacteria) adds mass to the reed. This has the effect of lowering the modulus of elasticity of the reed. This tends to make to reed feel softer, but also destroys the extra stability gained by the reeds ability to resonate with the air column modes (due to the shift to lower frequencies as the modulus decreases). For a while this effect is compensated for by the increased brittleness due to salivai coating of the reed.

Eventually, however, the bacterial mass wins out and the reed becomes permanently softer and thinner sounding.

Scraping the reed has different effects depending upon where the reed is scraped.

It has been emperically observed that scraping the back of the reed “frees up" the lower notes on the clarinet, while scraping the tip tends to affect the upper notes. To scrape the reed on the computer, one simply lowers the thickness of the reed at the point of scraping. When this is done to the back portion of the reed, since the thickness is lower, the modulus and natural frequency increases, shifting the nodal lines forwards towards the tip of the reed. The result is that the tip of the reed vibrates more, and this equalizes the motion throughout the whole body of the reed, including the tip, which gives power to the lower modes of the reed, which stabilizes the lower 379 notes of the instrument, since these notes use mostly only the lowest modes of the reed,

It is hoped that shortly a full computer simulation of the fluid dynamics of the reed mouthpiece interaction will be available using this technology, so that we may study some of these question in more detail.

In the next chapter we summarize the results of this document. 380

Endnotes

1. Leonard Euler, Dissertatio Physics de Sono , selections, in Benchmark Papers in Acoustics, R. Bruce Linsey, ed.

2. ibid.

3. Stephen Thomson, “Reed Resonance...,” 1299.

4. Douglas Keefe, ‘Tutorial on Physical Models of Woodwind Instruments, Part l-lntroduction,” preprint. Department of Systematic Musicology, University of Washington, 1989: 3.

5. N. H. Fletcher, “Autonomous Vibrations of Simple Pressure-Controlled Valves in Gas Flow," Journal of the Acoustical Societry of America 93, no. 4 (Apr 1993): 2172-2179.

6. Panchanon Das, ‘Theory of the Clarinet,” Indian Journal of Physics, 6 (1931): 229-232.

7. Keefe, Tutorial...,2.

8. See, for example, William E. Boyce and Richard C. Diprima, Elementary Differential Equations and Boundary Value Problems, 5th ed. (New York: John Wiley and Son, Inc., 1992).

9. Harris and Credes, Shock and Vibration Handbook, chapter 2.

10. For example; D. W. Jordan, and P. Smith, Nonlinear Ordinary Differential Equations (Oxford: Clarendon Press, 1988)

11. Das,‘Theory...,” 230.

12. Thompson, “Reed Resonance..”.

13. Ralph H. Abraham and Christopher Shaw, Dynamics-The Geometry of Behavior (Santa Cruz: Aerial Press, 1982). 381

14. Tohru Idogawa, Tokihiko Kobata, Kouji Komuro, and Masakazu Iwaki, Nonlinear Vibrations in the Air Column of a Clarinet Artificially Blown, Journal of the Acoustical Society of America, 93 no. 1 (January, 1993): 540-551; Bachus, "Vibration...", Donald Casadonte, An Introduction to the Biology, Chemistry, and Physics of Woodwind Reed Materia, Journal of the Acoustical Society of America 90, no. 4 (October, 1991), abstract.

15. See, for example, Gregory L. Baker and Jerry P. Gollub, Chaotic Dynamics: an Introduction (New York: Cambridge University Press, 1990).

16. 14. John Bachus, “Smali-Vibration Theory of the Clarinet” Journal of the Acoustical Society of America, 35 no. 3 (March, 1963): 305-313, but see, Douglas Keefe, “The Influence of Clarinet and Saxophone Reed responses on Sound Production,” paper delivered at the Acoustical Society of America Meeting, Octovber, 1993, Denver, Colorado.

17. Theodore A. Wilson and Gordon S. Beavers, “Operating Modes of the Clarinet,” Journal of the Acoustical Society of America, 56 no. 2 (August, 1974): 653-658; N. H. Fletcher, “Excitation Mechanisms in Woodwind and Brass Instruments,”Acoustica 43 (1979): 63-71.

18. See, for example, R. D. Kamm and T. J. Pedley, “Flow in Collapsible Tubes: a Brief Review,” Journal of Biomechanicai Engineering, 111, no. 3 (August, 1989): 177-179.

19. Thomas Emeaux, Edward reiss, J. F. Magnani, and P. K. Jayakumari, “Nonlinear Stability Control and (^-Bifurcation,” SIAM Journal of Applied Mathematics, 47 no. 6 (December, 1987): 1163-1175.

20. Scott D. Sommerfeldt and William J. Strong, “Simulation of a Player-Clarinet System," Journal of the Acoustical Society of America, 83, no. 5 (May, 1988): 1908- 1917.

21. ibid.

22. Bodig and Jayne, Mechanics... 262.

23. Harris and Credes, Shock and Vibration Handbook, chapter 7.

24. Chuen-Yuan Chia, Nonlinear Analysis of Plates (New York: McGraw Hill, inc., 1980): 1.

25. Douglas Keefe, “On Sound Power Production in Reed-Driven Wind Instruments,” preprint. Department of Systematic Musicology, University of Washington, 1990: 4. 382

26. E. H. Mansfield, The Bending and Stretching of Piates (New York: Cambridge University Press, 1989): 150.

27. Chia, Noniinear Anaiysis...

28. ibid.

29. Arthur W. Laissa, Vibration of Plates, NASA SP-160 (Washington D. C.: Office of technology Utilization, National Aeronautics and Space Administration, 1969).

30. T. Irei, G, Yamada, M. Sonoda, “Natural Frequencies of Square membranes and Square Plates with Rounded Comers," Journal of Sound and Vibration, 86, no. 3 (1983): 442-448.

31. Wang and Wang, “Vibrations...of Superelliptic Plates."

32. A. Hirshberg, R. W. A. van de Laar, J. P. Marrou-Maurieres, A. P. J. Wijnands, “A Qyasi-Stationary Model of Air Flow in the Reed Channel of Single-Reed Woodwind Instruments,” Acousf/ca 70 (1990): 146-154.

33. For and introduction to diffuser theory, see, Frank M. White,Fluid Mechanics (New York: Meg raw Hill, Inc., 1979): 366 ff.

34. C. D. Betram and T. J. Pedley, “Steady and Unsteady Separation in an Two-dimensional Indented Channel," Journal of Fluid Mechanics, 130 (1983): 315-345.

35. Arthur W. Leissa, Vibration of Shells, NASA SP-288 (Washington D. C.: Office of technology Utilization, National Aeronautics and Space Administration, 1973).

36. J. B. Grotberg and E, C. Reiss, “Subsonic Fiapping Flutter,”Journal of Sound and Vibration, 92, no. 3 (1984): 349-361.

37. Listoken

38. Fletcher, “Excitation...", 65. For a disoussion of the Starling resistor, see, Ascher H. Shaporo, “Steady Flow in Collapsible Tubes,” Transactions of the AME, Journal of Biomechanicai Engineering, 99, (August, 1977): 126-147.

39. Witt Liu and C. C. Huang, “Free Vibration of Thick Cantelever Laminated Plates with Step-Change of Thickness,Jouma! Of Sound and Vibration, 169, no. 5 (1994): 601-618.

40. Authur Leissa, private communication. 383

41. This is the mode most often mistakenly measured as the lowest mode of the clarinet. More recent measurements by Keefe have shown the existence of the two lower pseudo-torsional modes. Chapter VI

Summary

in chapter one we raised several reasons for wishing to undertake a study of the

clarinet reed. In this final chapter we shall see how much progress has been made in this study in dealing with these issues.

The first reason for studying the clarinet reed was the simple need to know the material parameters of the reed material which are selected and processed as clarinet

reeds. In this regards, we have made a start, but only a start. We have some idea of the order of magnitude of the elastic moduli, but not complete knowledge of the orthotropic nature of the material. We know a great deal about the chemistry of the material, and some things about the thermal and hydrative properties of the material.

Much more work is needed, since there are many ways to analyze a biopolymeric substance, and we have only used a few.

The second reason, the tendency for reed material degradation over playing time has been explained by a number of different causes. We have seen that the cellulose lattice structure seems to decrystallize in spent reeds to a small extent, either to vibrational stresses or due to chemical activity on the lattice elements. Hemicelluloses may be leached out of the reed material (different tests imply that this may or may not be the case) due to the action of saliva on the reed, which varies on pH over the

384 385

playing life of the reed. This variable pH tends to act as a plasticizer on the reed,

lowering the modulus of the reed over time. The glycoproteins in saliva coat the reed

by three different routes, forming a water-insoluable barrier. Finally, the growth of S.

Epidermitis in the reed causes an inhomogeneous mass distribution which increases

off-axial motion (buzziness) as well as lowers the modulus and decreases the natural

frequency of the reed (removing the effective stabilization of reed resonance). These

effects combine to alter the motion of the reed and its response to the air collumn in

such a way that the power of the sound is lost over time, making the sound more

thinner and buzzier. Preventatives do exist and have been discussed.

The large reed material variability has been precisely recorded by the response of students to a variety of perfomance measurements of the reed material. The perfect

reed for an individual seems to involve an overlap of a number of different parameters, and we have begun to delineate a few of them.

It has not been possible in this study, with its limited resources to define the best ranges of the material parameters for use in the manufacture of the clarinet reeds.

Indeed, the mechanism of biosynthesis of cellulose in the cell wall is not precisely known, so that the only way to insure consistent quality in reeds at the present time is to resort to genetic engineering methods, either by simple cross-breeding stems to achieve the right gene pool, or to actually alter the cellular structure of the plant by recombinant DNA methods. We have not been able to examine either of these two methods in detail.

Nevertheless, although a precise control of the growing pattems of the material have not yet been achieved, there are a few things which growers may do to help 386 increase the likelihood for a good quality of reeds. The most important ones seem to be increased sunning time of the reed, making sure that the reed is cut with a grain angle as close to zero as possible, and better adapting the reed shape to a given mouthpiece configuration.

In attempting to correlate the material properties of the reed with with synthetic material substitutes, the author has met with some sucess. A review of nine thousand polymer formulations which are commercially available has shown none to contain the same properties as Arundo donax, however, the one which comes the closest seems to be a nylon 6, 6 material, since the extensional modulus is the same asArundo donax, and nylon has the highest degree of water absorbancy of any polymer material, mimicing the water absorbing capability of Arundo donax (nylon absorbs from 10% to

20 % by weight, the same as Arundo donax)). Although the author has not been able to develop a synthetic reed, due to limitations of resources, the way to do so seems clear-select a brand of nylon 6, 6 which is slightly lower in modulus than Arundo donax, and then add a fiber reinforcer of somewhat higher modulus to bring the composite modulus to the same range as that forArundo donax.

Such a material will not have the natural inhomogeneity of the plant material, and hence will be prone to off-axial motion. The inhomgeneity of Arundo donax acts to diffuse the uneven pressure gradient of the air column such that high modulus areas may, at random, be in places of high pressure , etc. These are good reeds. Poor reeds have their modulus distribution different, so that a place which need resistance does not have it. These thing happen entirely at random and vary from reed to reed

With the synthetic material, unless the designer is careful to allow a certain 387 inhomgeneity in the material, the reed will be too uniform in property and will be of little help to statistically counteract the pressure gradient in the air column.

We have tested the validity of certain types of folk wisdom concerning reed use and selection among clarinet players. We have found that the number of vascular tissue at the rounding point is mildly correlated with fuller sound. We have not found any evidence that either temperature or moisture affects the reed in a seasonal way, and that the wisdom about using thinner reeds during the summer, while correct, is explained by factors not related directly to the reed material.

The shape of the reed has been rationalized as a reaction to the shape of the mouthpiece. Both historical evidence and physical theory tend to support the idea that the reed shape is a response to the pressure variation inside of the mouthpiece chamber, and is designed in such a way as to keep the bending stress of the reed constant over the reed surface.

Along with the effect of reed shape, we have shown that a study of the mouthpiece is essential to obtain a better understanding of how the reed vibrates in the environment of the mouthpiece. We have proposed thoeries of both the low frequency air column-driven motion of the reed, and the high frequency flapping-flutter behavior of the reed during reed squeak.

Finally, although we have discussed the oral cavity environment within the context of salivai attack of the reed, the exact nature of the fluid dynamics inside of the mouth is still wanting. It is likely that there are many vortex motions generated at the covering points of the lip and the mouthpiece, but these, as well as the matter of lip elastance are matter for further study. 388

We began this work with a quote by Robert Bievin. The study of the clarinet reed has been both rewarding but also exasperating for more reasons than one. It is with the hope that we may have done some good for the musical community and little harm, that we paraphrase Dr. Blevin’s admonition (in our case, to the reed researcher) to end this work:

This book is dedicated to people who will never know my name or yours. They are the people who play the clarinet, create beautiful sounds, and warm the soul. They motivate our work, and they are our first responsibility. Appendix A: Sample Reed Rating Instruction

The following is the actual questionare used by students to rate the reeds in {R}. The instructions are self-explanatory.

389 Reed Rating Instructions 3 9 0

You should have received two boxes of reeds of the same brand and strength. The box with the alphabetical markings (A, B, etc.) is the experimental set, and is to be rated and returned. The other unopened box is yours to keep for participating in this experiment. The experimental reeds are numbered on the back, 1 to 10 (A1, A2, 81, etc.). This will help organize the rating process. They have been subjected to a number of measurements which will help determine if some of the current reed wisdom is correct. The experiment in brief...all of the reeds have been weighed, and about 40% were then heated in an oven set to 250°F to drive off all of the trapped water (this process may or may not have destroyed the quality of the reeds. One of the things to see is if drying out reeds really makes them unplayable). The drying will allow me to determine the reed's true weight as well as the amount of water trapped in the reed matrix (botanists call this the water content of a plant material). I can then see if there is a correlation between either the “raw” weight, the desiccated (dry) weight, or the water content and reed quality. In addition, the reeds have all been measured on a very accurate digital micrometer at a number of points (see diagrams), such as the heel, the tip, the shoulder, etc. The idea is to see if there is any correlation between these measurements and read quality. In addition, the amount of lengthwise reed warp (the type everyone always tries to get rid of with a file) has been determined for each reed, so that I can see if the “unflattness” of the reed has an influence on quality. As you know, reeds are cut asymmetrically, with the left portion of the front vamp being slightly thicker than the right side, as seen with the reed face forward (as in the diagrams). From these measurements I can determine if the amount of reed asymmetry really has an effect on reed buzz. The reeds will also be measured for their moduli of elasticity and damping to see if these have any effect on reed quality. If you want to know how the experiment turned out, I will summarize the results as a handout for clarinet class in about a month.

Instructions for Rating

Simply take the ten reeds and play them for some extended time (practicing, long tones, etc.) so that you have a chance to fairly rate the reeds in such a way that you feel your rating would not change more than one point if you were to try the reeds again in a few weeks. Play the reeds long enough to “break them in", so that the reeds stabilizes. If you find a reed you like, by all means play on it until I collect it. P le a s e , do not alter the reed in any way such as filing or scraping the reed. The rating is done on a seven point scale, with the extremes on either side of the scale. The criteria are: 1 ) overall rating (the most important), 2) brioht-dark. 3) soft-hard. 4) buzzv-fluid. 5) soueakv-stable. 6) full-bodied-weak. There should be six ratings per reed, for a total of sixty ratings. I will pick up the reeds in band in about a week (Thursday or Friday, April 6, 7) to give me time to conduct other experiments and examine the results. Again, thank you for your help, and if you have any questions, feel free to ask. Overall Rating Reed Designation (write the alphabetical code from box top, A, B, etc.): Brand and Strength of Reeds: 391

Very B a d ------Medium------Very Good

Reed:

1.

2.

3.

4.

5.

6.

7.

8 .

9.

10. 3 9 2 Bright-Bark (Timbre) Reed Designation (write the alphabetical code from box top, A, B, etc.): Brand and Strength of Reeds:

Very D a rk------Medium------Very Bright

Reed:

1 .

2.

3.

4.

5.

6 .

7.

8 .

9.

10. 393 Soft-Hard (Strength) Reed Designation (write the alphabetical code from box top, A, B, etc.): Brand and Strength of Reeds:

Very S o ft ------Medium Very Hard 1------2------3------4------5------6------7

Reed:

1 .

2.

3.

4.

5.

6.

7.

8 .

9.

10. 394

Squeaky-Stable (Stability)

Reed Designation (write ttie alptiabetical code from box top, A, B, etc.): Brand and Strength of Reeds:

Very Squeaky------Medium------Very Stable 1------2------3------4------5------6------7

Reed:

1.

2.

3.

4.

5.

6.

7.

8 .

9.

10. Buzzy-Fluid (Clarity) 395

Reed Designation (write the alphabetical code from box top, A, B, etc.): Brand and Strength of Reeds:

Very B u zzy ------Medium------Very Fluid 1------2------3------4------5------6------7

Reed:

1.

2.

3.

4.

5.

6 .

7.

8 .

9.

1 0 . 396 Full-Bodied-Weak (Depth of Sound and Character)

Reed Designation (write the alphabetical code from box top, A, B, etc.): Brand and Strength of Reeds:

Very W e a k------Medium------Very Full 1------2------3------4------5------6------7

Reed:

1.

2.

3.

4.

5.

6 .

7.

8 .

9.

10. Appendix B: Reed Mode Shapes and Stresses

The following (figure B. 1) are the calculated mode shapes and stresses for the first twenty modes of the isolated (i.e., not attached to a mouthpiece) clamped clarinet reed (see figure 5.16 for the experimental set-up). For each mode there are three plots: 1) simple displacement, 2) contour plot of displacement (symbol = USUM), 3) contour stress map

(symbol = SINT).

The key for the abbreviations is as follows: ANSYS = finite element software name, Step = load step, Sub = mode number, Freq. = frequency

(Hz), RSYS = machine coordinates system, DMX = displacement maximum

(machine coordinates), SMX(B) = stress maximum (internal scale), SERC = energy norm (a relative measure of mesh accuracy), DSCA = internal scaling, xv, yv, zv, = viewing coordinates, dist = distance from object, xf, yf, zf = focal point, A-ZS = rotation off-set, precise hidden = hidden line method.

3 9 7 ANSYS 5.0 A MAY 22 1995 23:03:33 PLOT NO. 1 DISPLACEMENT STEP=1 SUB =1 FREQ=774.257 T1 _ *ZF =0.0185 Ü) A-ZS=-94.41 zr PRECISE HIDDEN "O0) % m (X œ

U)1 (D

8 I OOi T1 cq’ ANSYS 5.0 A C MAY 22 1995 CD 23:19:09 PLOT NO. 1 00 NODAL SOLUTION STEP=1 SUB =1 FREQ=774.257 USUM RSYS=0 DMX =307.539 SEPC=47.959 SMX =307.539 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0 34.171 68.342 102.513 136.684 170.855 205.026 239.197 273.368 307.539 iS ' ANSYS 5.0 A MAY 22 1995 23:55:45 i PLOT NO. 1 00 NODAL SOLUTION STEP=1 SUB =1 FREQ=774.257 SINT (AVG) DMX =307.539 SMN =0.615E-15 SMX =0.599E+12 SMXB=0.774E+12 274832 .326438 296488 041133 048241 .010494 0185 4.41 HIDDEN 615E-15 666E+11 133E+12 200E+12 266E+12 333E+12 399E+12 466E+12 532E+12 599E+12

è! o ANSYS 5.0 A T1 MAY 22 1995 cq’ 23:03:37 c PLOT NO . 2 (D DISPLACEMENT DO STEP=1 SUB =2 FREQ=843.701 RSYS=0 DMX =485.038 SEPC=60.331 DSCA=0.848E-05 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 % *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN %

è ANSYS 5.0 A T1 MAY 22 1995 cq' 23:19:15 C PLOT NO. 2 (B NODAL SOLUTION 00 STEP=1 SUB =2 FREQ=843.701 USUM RSYS=0 DMX =485.038 SEPC=60.331 SMX =485.038 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN _ 0 53.893 107.786 161.679 215.572 269.465 323.359 377.252 431.145 485.038 ANSYS 5.0 A MAY 22 1995 TI < n ' 23:55:46 c PLOT NO. 2 cB NODAL SOLUTION STEP=1 00 SUB =2 FREQ=843.701 SINT (AVG) DMX =485.038 SMX =0.164E+13 SMXB=0.505E+13 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0 0.182E+12 0.364E+12 0.546E+12 0.728E+12 0.910E+I2 0.109E+13 {MB 0.127E+13 0.I46E+13 0.164E+13

S ANSYS 5.0 A TI cq ‘ MAY 22 1995 c 23:03:40 3 PLOT NO. 3 DISPLACEMENT D3 STEP=1 SUB =3 o o FREQ=1925 3 RSYS=0 DMX =204.227 SEPC=37.423 DSCA=0.201E-04 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 \\ PRECISE HIDDEN

\ V \ ■

I ANSYS 5.0 A MAY 22 1995 (Q 23:19:21 C PLOT NO . 3 (5 NODAL SOLUTION CD STEP=1 SUE =3 FREQ=1925 USUM RSYS=0 Dt4X =204.227 SEPC=37.423 SMX =204.227

XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YP =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0 22.692 45.384 68.076 90.768 HM 113.46 136.151 r " " " i 158.843 181.535 204.227 o en ANSYS 5.0 A (O MAY 22 1995 C 23:55:47 (B PLOT NO. 3 00 NODAL SOLUTION STEP=1 SUE =3 FREQ=1925 SINT (AVG) DMX =204.227 SMX =0.178E+13 SMXB=Ü.426E+13 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0.198E+12 0.397E+12 0.595E+12 0.793E+12 0.991E+12 0.119E+13 0.139E+13 0.159E+13 0.178E+13

è O) ANSYS 5.0 A ~n MAY 22 1995 cq‘ 23:03:43 c PLOT NO. 4 c3 DISPLACEMENT CD STEP=1 SUB =4 o FREQ=2148 o RSYS=0 Z3 DMX =426.88 SEPC=55.698 DSCA=0.964E-05 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 % PRECISE HIDDEN

§ ANSYS 5.0 A MAY 22 1995 (Q 23:19:27 C PLOT NO. 4 cB NODAL SOLUTION ro STEP=1 SUE =4 o FREQ=2148 o USUM RSYS=0 DMX =426.88 SEPC=55.698 SMX =426.88 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 47.431 94.862 142.293 189.724 237.155 284.586 332.017 379.449 426.88 È ANSYS 5.0 A T1 MAY 22 1995 cq’ 23:55:48 C PLOT NO. 4 (B NODAL SOLUTION 00 STEP=1 SUE =4 FREQ=2148 SINT (AVG) DMX =426.88 SMX =0.289E+13 SMXB=0.431E+13 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0 0.322E+12 0.643E+12 0.965E+12 0.129E+13 0.161E+13 0.193E+13 0.225E+13 0.257E+13 0.289E+13

O CD ANSYS 5.0 A MAY 22 1995 (Q 23:03:47 c PLOT NO. 5 (B DISPLACEMENT CD STEP=1 SUB =5 FREQ=2361 RSYS=0 DMX =564.998 SEPC=67.962 DSCA=0.728E-05 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN

% A O ANSYS 5.0 A (Q MAY 22 1995 C 23:19:32 s PLOT NO. 5 00 NODAL SOLUTION STEP=1 o SUE =5 o FREQ=23 61 3 USUM RSYS=0 DMX =564.998 SEPC=67.962 SMX =564.998 XV =0.274832 YV =-0.326431 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN 0 62.778 125.555 188.333 251.11 313.888 376.665 439.443 502.22 564.998 ANSYS 5.0 A MAY 22 1995 (O 23:55:50 c PLOT NO. 5 (3 NODAL SOLUTION 00 STEP=1 SUE =5 FREQ=2361 SINT (AVG) DMX =564.998 SMX =0.568E+13 SMXB=0.168E+14 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN _ 0 0.631E+12 0.126E+13 0.189E+13 0.252E+13 0.316E+13 0.379E+13 0.442E+13 0.505E+13 0.568E+13 ANSYS 5.0 A *n MAY 22 1995 (Q‘ 23:03:53 c PLOT NO. 6 m DISPLACEMENT CD STEP=1 SUB =6 FREQ=4473 8 RSYS=0 ZJ or4X =522.65 SEPC=65.658 DSCA=0.787E-05 XV =0.274832 YV =-0.326438 ZV =0.296488 *DIST=0.041133 *XF =0.048241 *YF =-0.010494 *ZF =0.0185 A-ZS=-94.41 PRECISE HIDDEN

4^ CO ANSYS 5.0 A TI MAY 22 1995 (Q' 23:19:37 C PLOT NO. 6