Appendix A

The Harmonic Oscillator

Properties of the harmonic oscillator arise so often throughout this book that it
seemed best to treat the mathematics involved in a separate Appendix.

A.1 Simple Harmonic Oscillator

The harmonic oscillator equation dates to the time of Newton and Hooke. It follows
by combining Newton’s Law of motion (F = Ma, where F is the force on a mass
M and a is its acceleration) and Hooke’s Law (which states that the restoring force
from a compressed or extended spring is proportional to the displacement from
equilibrium and in the opposite direction: thus, FSpring = −Kx, where K is the
spring constant) (Fig. A.1). Taking x = 0 as the equilibrium position and letting the
force from the spring act on the mass:

d 2x
M= + Kx = 0. (A.1)
dt 2

Dividing by the mass and defining ω2 = K/M, the equation becomes

d 2x
+ ω0 x = 0. (A.2)
dt 2
As may be seen by direct substitution, this equation has simple solutions of the
form

x = x0 sin ω0 t or x0 = cos ω0 t, (A.3)

where t is the time and x0 is the maximum amplitude of the oscillation. The
angular resonance frequency ω0 is related to the cyclical resonance frequency F0

© Springer Nature Switzerland AG 2018 329

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
330 A The Harmonic Oscillator

Fig. A.1 Frictionless

harmonic oscillator showing
the spring in compressed and
extended positions

and period T (= 1/F0 ) of the oscillator by ω0 = 2π F0 = 2π/T where


ω0 = K/M. (A.4)

A.2 Energy

Without dissipation, the total energy in the oscillator at any instant in time is constant
and given by

1
Energy = (Mv 2 + Kx 2 ), (A.5)
2
where the first term is the kinetic energy of the moving mass (v = dx/dt) and
the second term is the potential energy stored in the spring. The energy oscillates
back and forth between those two forms during alternate half periods. At the turning
points, where the velocity of the mass is zero, the energy is stored in the spring. At
x = 0, where the velocity is maximum, the energy is all kinetic. If x = x0 sin ωt,
then v = dx/dt = ωx0 cos ωt and

1
Energy = [M(ω0 x0 cos ω0 t)2 + K(x0 sin ω0 t)2 ]
2
1 Kx0 2
= Kx0 2 (sin2 ω0 t cos2 ω0 t) = , (A.6)
2 2
where again ω0 = K/M and (sin ωt + cos ω−) = 1. Regardless of its kinetic
energy the particle is trapped in a potential well of the type shown in Fig. A.2. Since
the restoring force from the spring is

dU Mω02 x 2
F ≡− = −K1 x, the potential is U (x) = , (A.7)
dx 2
where x is defined with respect to the equilibrium position of the oscillator and
U = 0 at x = 0.
A The Harmonic Oscillator 331

Fig. A.2 Potential well for the simple harmonic oscillator. In this model the particle is bound,
regardless of its kinetic energy

A.3 The Damped Simple Harmonic Oscillator

In any real system there will be a “damping” force due to friction that at low
velocities takes the form
dx
FDamping = −Γ , (A.8)
dt
where Γ is a damping constant characteristic of the system. The total force acting
on the mass M now becomes F = Ma = −Γ dx/dt − Kx. Hence, the equation of
motion becomes

d 2 x dx
M Γ + Kx = 0. (A.9)
dt 2 dt
Dividing through by the mass and moving everything to the left side puts the
equation for the damped simple harmonic oscillator in the more convenient form,

dx dx
+γ + ω02 x = 0 (A.10)
dt 2 dt
where we have defined γ = Γ /M as the damping constant per unit mass. Exact
solutions to this equation may be obtained in the following straightforward manner.
Substituting a trial solution, x = eμt , gives a quadratic equation

μ2 + γ μ + ω02 = 0, (A.11)
332 A The Harmonic Oscillator

that has μ± roots of the form


 
γ γ 2
μ± = − ± iω0 1 − (A.12)
2 2ω0

where i 2 ≡ −1. The solutions without a driving term are of the form x = Aeμ+t +
Beμ−t , where A and B are complex constants determined by the initial conditions
(i.e., the displacement from equilibrium and the velocity at t = 0).
For (γ /2ω0 )2 < 1, which is typical of resonant systems, the roots have an
imaginary component that leads to free oscillation at a frequency
   
γ2 γ 2
ω ≈ ω0 − + Order ω0 (A.13)
8ω0 4ω0

that is damped at the amplitude decay rate γ . For γCrit = 2ω0, no oscillation occurs,
the damping is fastest and is said to be “Critical.” For (γ /2ω0 )2 << 1, one solution
is

x(t) ≈ x0 e−γ t/2 sin ω0 t, (A.14)

which corresponds to giving the mass an initial velocity kick at x = 0 and t = 0. In

this case the total energy in the oscillator decays with time as

Total Energy ∝ |x(t)2 | ∝ eγ t . (A.15)

Note that the energy decay rate is a factor of 2 faster than the amplitude decay
rate simply because the energy in this classical motion problem is proportional to
the square of the amplitude.

A.4 Resonance or Cavity Width and Quality Factor (Q)

Because the amplitude of the damped oscillator decays with time, the spectrum of
its motion is spread out in frequency about ω0. Regardless of the initial conditions
chosen, we know that the total energy decays as shown in Eq. (A.15). Since x(t) is
of the form exp(−γ t/2 + iω0t), the Fourier transform of the amplitude is of the
general form,
 ∞ 1
|x| ∝ x(t)e−iωt dt ∝ . (A.16)
0 |γ /2 + i(ω − ω0 )|
A The Harmonic Oscillator 333

Hence, the spectral distribution of energy in the resonant mode will be of the
form
1
|xω |2 ∝
2 , (A.17)
ω−ω0
|1 + γ /2 |

which is sometimes called a “Lorentzian” function of the frequency and, provided

(ω − ω0)2 >> (γ /2)2, is symmetric about the angular resonant frequency ω0 with
full cavity width at half-maximum energy response given by Δω = γ 2π ΔF . The
quantity

γ f c
ΔFcav = = (A.18)
2π π 4L
is sometimes called the cavity resonance width. In general, for standing-wave
resonances in a long cavity such as an organ pipe, γ = f (c/2L) where c is the
running wave velocity, L is the cavity length, and f is the round trip fractional
energy loss. For example, for a closed-pipe model of the first formant (F1 ) of the
vocal tract,

f
ΔFcav = F1 . (A.19)
π
The Quality Factor, or Q of a resonance, is defined traditionally by the relation

Energy Stored Energy Stored

Q = 2π = ω0 , (A.20)
Energy Lost per Cycle Rate of Energy Loss

where the rate of energy loss is γ × (Energy Stored). Noting that γ = Δω, an
alternative expression for the Quality Factor is

ω0 F0
Q= = , (A.21)
Δω ΔF
where F0 is the cyclical resonance frequency and ΔF is the full cyclical width at
half-maximum energy response, both measured in Hz.

A.5 Driven Damped Oscillation

When a sinusoidal or cosinusoidal driving term of the type

E(t) = E cos ωt (A.22)

334 A The Harmonic Oscillator

is added to the damped oscillator, the equation becomes

d 2x dx
+γ + ω02 x = E(t) (A.23)
dt 2 dt

where E(t) on the right-hand side represents the driving force per unit mass. This
equation has the closed-form solution

E
x= sin(ωt + Φ), (A.24)
D
where the denominator D is given by
 γ ω
2
D= (ω2 − ω02 )2 + (A.25)
M

and the phase angle is

γ

Φ = cos− 1 radians. (A.26)

MD
The introduction of damping causes the oscillator to lag in phase from the driving
term. Without damping, |x| → ∞ as ω → ω0 .

A.6 Electric Circuit Equivalent of the Damped Driven

Oscillator

The equation for the damped, driven oscillator has an exact equivalent in the series
LCR (inductance-capacitance-resistance) circuit shown in Fig. A.3. Written in terms
of the charge q on the condenser C, the equation for the voltage drop around the loop
becomes

d 2q dq q
L 2
+R + = E(t), (A.27)
dt dt C
where the current i = dq/dt, and we have made use of the basic relations for ac
voltage drops across the circuit elements: VL = Ldi/dt = Ld 2 q/dt 2 (Henry’s
Law), VR = Ri = Rdq/dt (Ohm’s Law), and VC = q/C (Faraday’s Law).
By comparison with the original equation for the oscillator, we find the equivalent
relationships in Table A.1.
One may determine the steady-state behavior of a linear resonant circuit using
complex ac circuit analysis. In this analysis, one usually assumes that the transient
solution has died down to a steady complex current of constant amplitude and phase
running through the various resistive and reactive elements of the circuit and that
A The Harmonic Oscillator 335

Fig. A.3 Series resonant

LCR circuit driven by a
sinusoidal EMF
(electro-motive force)

Table A.1 Electrical and AC circuit Mechanical oscillator

mechanical equivalent
parameters Charge, q ⇒ x Amplitude
Current, i ⇒ v Velocity
Inductance, L ⇒ M Mass
Resistance, R ⇒ γ Damping constant
Capacitance, C ⇒ 1/K 1/(Spring constant)
EMF, E(t) ⇒ F (t)/M = E(t) Driving force
AC circuit mechanical oscillator

the real part of the resultant voltage drops correspond to values one would measure.
Thus, if the current running through a circuit loop is of the form1

i = i0 ej ωt = i0 (cos ωt + j sin ωt), (A.28)

the voltage drops across the elements L (inductance), R (resistance), and C

(capacitance) may be written as
  
di q 1 t −j
VL = L = i(j ωL), VR = iR, and VC = = i0 e dt = i j ωt
.
dt C C ωC
(A.29)
Note that in the expression for VC we have made use of the fact that the charge on
the capacitance is the integral of the current over time; i.e., the current feeding the
capacitance is i = dq/dt, where q is the time-dependent charge on the capacitance.
In applying this convention, one makes use of Kirchoff’s Laws: (1) The sum of
the voltage drops around a closed loop equals the sum of the Emfs (“electromotive
forces”), which is a statement of the conservation of energy; (2) The sum of the
currents at any junction must be zero, which is a statement of the conservation of
charge. Using the methods and notation of complex steady-state ac circuit analysis
just discussed, the loop equation for the circuit in Fig. A.5 may be written

1 The relation eiθ = cos θ + i sin θ where i 2 = −1 is known as Euler’s formula and may be derived
from the infinite series for ex and those for cos θ and sin θ by letting x = iθ. This type of ac circuit
analysis was invented by Campbell, a research engineer at the A.T. & T Company in 1911, who
referred to it as “Cisoidal Oscillations”—an abbreviation for “cos i sin” Campbell (1911).
336 A The Harmonic Oscillator

  
1
E = i R + j ωL − . (A.30)
ωC

The magnitude of the current is

E
|i| = 
2 (A.31)
R 2 + ωL − 1
ωC

and it lags the driving voltage in phase by φ where

ωL − 1
(ω2 − ω02 ) (ω0 − ω)
tan− 1φ = − ωC
≈ ≈ . (A.32)
R ωL/R Δω/2

Here, the final approximation holds near resonance where ω is close to ω0 .

The power loss in the resistor is

E 2 /R
PR = |i|2 R = (A.33)
(ω2 −ω02 )
1+ (ωL/R)2

and is of resonant Lorentzian shape with full width at half maximum given by

Δω ≈ R/L ≈ 2π ΔF. (A.34)

The Q or “Quality Factor” of the circuit is then

Q = ω0 /Δω ≈ ω0 L/R. (A.35)

These properties are summarized in Fig. A.4.

A.7 A Different Approach

Normally, one uses ac circuit analysis to determine steady-state currents in the

presence of applied Emfs. However, useful results are obtained even when E = 0
in Fig. A.3. In that case, we can use the loop equation to determine the (complex)
frequency, an approach that gives the transient behavior of the circuit. Hence, if the
Emf is zero in Fig. A.3, we get
  
1
i R + j ωL − = 0. (A.36)
ωC
A The Harmonic Oscillator 337

Fig. A.4 Power resonance and phase shifts in the series LCR circuit, related to the properties of
the driven, damped mechanical oscillator

Assuming that i = 0 (i.e., that there is some initial charge or current flowing
in the circuit), the large square bracket must equal zero and we obtain a quadratic
equation in ω:

R 1
ω2 − j ω − = 0, (A.37)
L LC
whose roots are

 2
R R
ω=j ± ω0 1− . (A.38)
2L 2ω0 L

If we substitute this expression back into Euler’s formula, we get


2
R ±j ω0 1− 2ωR L t
ej ωt = e 2L t e 0 , (A.39)

where the first exponential describes damping of the current and the second
exponential gives the oscillatory solution through use of Euler’s formula with a real
frequency

 2
R
ω = ±ω0 1 − . (A.40)
2ω0 L
338 A The Harmonic Oscillator

Hence, the “steady-state “solution can actually give us the transient behavior of the
circuit. (Note that the sign on the real part of the frequency merely alters the phase of
the oscillation.) The damped oscillatory solution is, of course, the same one we got
earlier for the behavior of the damped mechanical oscillator using a trial solution of
the type eμ t. The present approach provides a less tedious way to analyze coupled
circuits, if you are comfortable with ac circuit analysis.

A.8 Coupled Oscillations, Wolf Tones and the Una Corda

Piano Mode

The double-strung pianos dating from the early work of Bartolomeo Cristofori
provide an interesting example of mode coupling with unusual potential for musical
expression. The physical processes involved were discussed qualitatively in Chap. 4.
We would like to show here that it is easy to derive the important quantitative
properties of the coupled system from an electric circuit model. As implied by
the term “double-strung,” each note on the piano has two strings tuned closely to
the same pitch, each of which can be represented as a damped harmonic oscillator.
We will only consider the fundamental resonances of the strings here, although the
present method could easily be extended to include harmonics of each fundamental
frequency. In the piano, the motion of the two strings is coupled through their
attachment to the bridge, which in turn transfers energy to the soundboard. We
will represent this situation by the pair of coupled circuits shown in Fig. A.5. For
simplicity, we will assume that the corresponding circuit elements are identical and
that each oscillator is initially tuned approximately to the same frequency given by
ω02 = 1/LC. (One might represent a string resonance and the other an air resonance
in a violin.)
The coupled circuits in Fig. 4.5 result in two loop equations:
  
1
i1 R + j ωL − − i2 R = 0 (A.41)
ωC

and

Fig. A.5 Coupled circuit

model for one note on a
double-strung piano. The two
current magnitudes may be
different
A The Harmonic Oscillator 339

  
1
− i1 R + i2 Rj ωL − = 0. (A.42)
ωC

Solving both of these equations for i1 / i2 , we obtain

i1 R R + j ωL − 1
ωC
=
= (A.43)
i2 R + j ωL − 1 R
ωC

Hence, we get an equation involving the complex frequency,

  
1 1
ωL − ωL − − 2j R = 0, (A.44)
ωC ωC

which has two distinct solutions for ω2 yielding roots for ω,


 2
R R
ω = ±ω0 and ω = j ± ω0 1− . (A.45)
L ω0 L

The first mode has no loss and oscillates at the resonant frequency, ω0 . The second,
lossy mode has the real component of its frequency shifted downward in magnitude
to

 
R 2
ω = ω0 1 − (A.46)
ω0 L

and an amplitude decay rate of R/L, which is twice as large as that for either LCR
loop by itself. Note that the Q of each resonant circuit is ω0 L/R 1. Hence,
the second term in the square root provides a small, real correction to the resonant
frequency, although the shift is larger than in either single isolated LCR loop alone.
The two coupled modes of oscillation may be interpreted in terms of the coupled
circuits shown in Fig. A.6. Here, the same current magnitude i flows in both loops,
but in different directions for the two different normal modes. The even-symmetric
mode on the left is the high-loss mode (with decay rate of R/L) since the two
currents flowing through the resistance are in phase. It corresponds to the normal
position of the hammer on the piano in which both strings are struck simultaneously.
Striking both strings in the normal hammer position is equivalent to depositing the
same charge on each capacitor at t = 0. The odd-symmetry mode at the right is the
low-loss mode and would correspond to a case in the piano where both undamped
strings were hit simultaneously with hammers moving in opposite directions. In the
circuit, opposite charges would have to be placed suddenly on the two capacitors
at t = 0. For the odd-symmetric case (the right side of Fig. A.6), the mode has no
loss at all because the two equal currents flowing through the resistor are 180°out of
phase. In the piano, that mode corresponds to periodic pulses from the two strings
340 A The Harmonic Oscillator

Fig. A.6 The even-symmetric (left) and odd-symmetric (right) modes of the coupled circuit (the
current magnitudes are all equal)

arriving at the bridge out of phase. A similar situation arises in the case of “Wolf
Tones” on bowed stringed instruments. (See Chap. 5.)
However, pure excitation of the odd-symmetry mode would not occur in a normal
piano because only the right-hand string is struck (in the double-strung piano) by the
hammer with the una corda pedal depressed. (The hammer is shifted to the right so
as to hit only one string, with both strings undamped.) Hence, to simulate the actual
una corda case, we need to take the difference between equal amounts of the two
normal coupled-circuit modes. Thus, initially,

iUnaCorda = iEvenSymmetric − iOddSymmetric . (A.47)

This combination localizes the current on the right side of the coupled circuit
when the hammer hits, after which the even-symmetric component rapidly dies
out leaving only the long-lived odd-symmetric mode still oscillating. Because the
odd-symmetric mode has the least loss (no loss at all for the idealized circuit in
Fig. A.6), the excitation would always tend to settle down in the odd-symmetry
mode with frequency ω0 . The apparent pitch goes up slightly as the high-loss mode
dies out. In reality there is some loss in the vibrating strings themselves due to air
resistance. That could easily be included in the electric circuit model by adding a
small resistance in series with each coil in the separate current loops. A reactive
component could also be added in series with the load resistance R representing
loss to the soundboard. Because equal currents flow in the two loops in the odd-
symmetric mode, the crescendo effect described in Chap. 4 after the string is struck
(produced by damping one string in the una corda mode after the sound decays)
would be limited to about 20Log10 2 ≈ 6 dB, or about the amount observed
experimentally.
A The Harmonic Oscillator 341

A.9 Harmonics and Subharmonics

As discussed throughout this book, most musical instruments produce waveforms

that are at least quasi-periodic and contain harmonics. These harmonics often result
from some nonlinearity in the oscillatory system, for example, the vibrating reed
in a woodwind, or the vibrating lips of the brass instrumentalist. In general, by
“nonlinearity” we mean that the output amplitude from the system does not rise in
simple direct proportion to the amplitude of the input signal as it would if the system
were perfectly linear (i.e., if the output simply varied as the first power of the input).
Instead, the output amplitude may depend on various powers of the input amplitude.
As a general rule, if we feed a sinusoidal signal into such a system, signals at
different frequencies that are harmonically related to the fundamental (i.e., are given
by integer multiples of the original frequency) will appear in the output. Further,
these harmonics are generally phase-locked to the fundamental frequency. This can
be understood in terms of basic trigonometry relations. For example, if you square
a sinewave, you will get a signal varying as the second harmonic of the input signal
which is in phase with the fundamental; if you cube the signal, the third harmonic
is produced; and so on. More strikingly, if you feed two signals with different
frequencies into the system, the output may contain extraneous signals at the sum
and difference frequencies of the two original frequencies.2 All of these effects can
be simply explained (albeit, tediously in some instances) through application of the
basic trigonometry identities such as those treated in high school.
Under some conditions, the output may even contain subharmonics of the
input frequency; i.e., frequencies that are lower than the input driving frequency.
Understanding how to produce subharmonics used to be a mysterious art. To quote
one of the pioneers who first demonstrated their existence in electromechanical
systems, “Most authors pass over this question in silence, as they take it for granted
that the occurrence of subharmonics is an impossibility.” (Pedersen 1934)3
It is useful to treat the simple mechanical oscillator model of Fig. A.1 with the
inclusion of nonlinear terms in the spring constant. As we will show, this model can
produce both harmonics and subharmonics, not to mention rectification.

2 To see how these extraneous frequencies come about, just consider the trigonometric identity,
2 sin(A) cos(B) = sin(A+B)+sin(A−B). Let A = 2π F0 t and B = 2π F t be the higher frequency
resonance. Multiplying the two waves together yields “sidebands” at F = F0 . If F = N F0 , where
N = 1, 2, 3, 4, etc., “sidebands” spaced by F0 will occur throughout the entire spectrum, even if
there is no strong acoustic wave at F0 .
3 “Hi-Fi” enthusiasts will want to note that Pedersen got interested in this subject because his Jensen

high-fidelity loudspeaker was producing both first and second subharmonics of sinusoidal tones.
342 A The Harmonic Oscillator

A.10 The Nonlinear Oscillator

In general, one expects the oscillator equation will become nonlinear at large enough
amplitudes. For example, when x is sufficiently negative the spring will compress,
the coils will touch and a large repulsive force will occur. At large positive values
of x, the spring constant will begin to decrease and the restoring force will saturate.
Assuming that the restoring force from the spring is a continuous function of x with
continuous derivatives, it can be expanded in a power series of the form

F = −[K1 x + K2 x 2 + K3 x 3 + . . .] (A.48)

where x is the departure from equilibrium position at x = 0. K1 is just the normal

spring constant and may be expressed in terms of ω0 . Then, the driven oscillator
equation may be rewritten

d 2x dx x2 x3
+γ + ω0 x +
2
+ = E sin ωt (A.49)
dt 2 dt Xq Xc

where γ is the damping term from friction per unit mass and the coefficients Xq
and Xc have been defined to have dimensions of length and length-squared and
describe the amounts of quadratic and cubic nonlinearity. We will ignore terms
higher than cubic in x. Here, E is the driving acceleration per unit mass at frequency
ω that would be supplied by the Bernoulli effect in the real case. The quadratic
term will yield both even harmonics and subharmonics, as well as rectification;
the cubic term provides odd harmonics; and with both terms present, additional
sum and difference terms arise. The term on the right-hand side of the equation
represents the driving force per unit mass. It could be sinusoidal at the normal
resonant frequency ω0 . To investigate subharmonic production, it is useful to replace
the driving frequency by 2ω0 . Then the subharmonic appears at ω0 . Stable solutions
to nonlinear equations have very specific phase relationships that are determined
by the parameters of the system. Closed-form solutions to this type of nonlinear
equation have been investigated in the past, but are extraordinarily tedious to derive
and usually only hold for very small driving terms.4 However, it is relatively easy to
solve the equation numerically with a computer.

4 Lord Rayleigh (1877) investigated low-amplitude solutions to nonlinear differential equations

of this type in closed form. Also see, Hartley (1939), Hussey and Wrathall (1936) and Pedersen
(1934) for approaches to the closed-form solution of similar nonlinear equations.
A The Harmonic Oscillator 343

A.11 Potential Wells for the Nonlinear Oscillator

As shown in Fig. A.2, the potential well for the ideal linear oscillator has even
symmetry about the equilibrium point (x = 0) and can trap the mass regardless
of its kinetic energy. The situation changes drastically when we add a quadratic
nonlinearity.
The potential curve shown at the left in Fig. A.7 is for an oscillator containing
both linear and quadratic terms. Here, the restoring force is of the form,
 2
dU 2 x x3
F ≡− = −K1 x − K2 x , hence U (x) = ω0
2
+ , (A.50)
dx 2 3Xq

where again U ≡ 0 at x = 0, but we have written U as the potential per unit

mass. As can be seen from the figure, the quadratic force term reduces the potential
barrier on one side of equilibrium, while increasing it on the other. The result is
that the mass can now escape from the well at the left if it acquires enough kinetic
energy from the driving force to reach the top of the potential bump.5 (The top of
the bump occurs at x = −Xq and is of magnitude ΔU = ω02 Xq2 /6.) Consequently,
the oscillator with only the addition of a quadratic force term tends to be unstable,
except at very low driving amplitudes where the kinetic energy per unit mass is kept
less than ω02 Xq2 /6. With a large enough damping constant, the loss of energy inhibits
the escape of the mass over the potential bump. However, if γ is made very large,
the harmonic (and subharmonic motion) is damped out as well.
Finally, the potential well on the right in Fig. A.7 includes the effect of both
quadratic and cubic force terms. Here,
 2
dU 2 x x3 x4
F ≡− = −K1 x − K2 x − K3 x and U (x) = ω0
2 3
+ + .
dx 2 3Xq 4Xc2
(A.51)
The cubic term in the spring constant introduces a stabilizing term to the potential.
Hence, with the cubic force term present, stable solutions are obtained at very much
larger driving amplitudes than with the quadratic term alone. Surprisingly, with the
cubic term present, finite solutions exist at resonance for large driving amplitudes
even without the presence of damping. However, in that case, the oscillator never
settles down to a steady-state solution.

5 This same phenomenon occurs in the dissociation of diatomic molecules.

344 A The Harmonic Oscillator

Fig. A.7 Potential wells for nonlinear oscillators containing quadratic (left) and both quadratic
and cubic forces (right)

A.12 Numerical Solutions to the Non-linear Oscillator

Equation

The driven oscillator equation with nonlinear terms is exceedingly tedious to solve
in closed form. However, numerical solutions with a computer are straightforward
using a computational method previously developed by the author (Bennett 1976,
p. 200). Here, the main trick is to minimize numerical errors that tend to build up in
successive integration intervals. That objective may be accomplished by expanding
the acceleration in a Taylor series in the time and integrating that series term-by-term
over an interval, t, that is small compared to the period of the driving oscillation.
First, we rewrite the basic equation with the acceleration on the left-hand side.
For example, the equation for the driven, damped oscillator with quadratic and cubic
nonlinear terms given above can be rewritten

d 2x dx x2 x3
a= = E sin ωt − γ − ω0
2
x + + . (A.52)
dt 2 dt Xq Xc

Note that we now have an explicit closed-form expression for the acceleration
that can be differentiated as many times as we want to obtain the coefficients
in the Taylor series. Thus, after the small time interval t, the new values for the
acceleration, velocity, and position will be given by

t2 t3 dv
a
= a + bt + c + d + Order(t 4 ) = (A.53)
2! 3! dt

t2 t3 t4 dx
v
= v + at + b + c + d + Order(t 5 ) = (A.54)
2! 3! 4! dt
A The Harmonic Oscillator 345

t2 t3 t4 t5
x
= x + vt + a + b + c + d + Order(t 6 ) (A.55)
2! 3! 4! 5!
where we have integrated the successive equations term-by-term and have included
constants of integration (the values of the quantities x, v, and a at t = 0) in each
case. The errors in a
are now of order t 4 and those for the new position of the
particle x
are of order t 6 . Hence, by making t sufficiently small, we can make the
results as precise as we wish.
Putting these expressions into a reiterative computer loop where at each succes-
sive step one sets the initial values of a, v, and x to be the values computed for a
, v
,
x
at the end of the last interval t yields x as a function of time to high accuracy. One
could, of course, add more and more terms to the Taylor series for a, but terminating
the series as shown above is more than adequate for the present examples.
The equations for a, v, and x are especially adaptable to inclusion in a computer
program loop of the type used in BASIC or FORTRAN. One can simplify the
numerical equations further by adopting a time scale in which the time increment t is
defined to be unity. In that case the important program steps for solving the nonlinear
oscillator may be written as follows. First, we define some constants dependent on
the parameters of the oscillator.

Driving frequency Resonant frequency

W=ω W0 = ω_0
W2 = W ∗ W W02 = W0*W0
W3 = W ∗ W2
Damping constant per unit mass
G=γ

In terms of these computer variables, the equation for the acceleration becomes

AE ∗ SI N (W ∗ T )G ∗ V W 02 ∗ (XX ∗ X/XqX ∗ X ∗ X/Xc)dV /dT . (A.56)

For example, choosing the oscillator period to be 60 time units (or ω0 = 2π/60),
means that a step size of t = 1 leads to an error ≈ 5 parts in 101 0 in the calculation of
x over one increment. One typically wants to do the calculation over a time interval
Tmax ≈ 5 oscillator periods for the transient solution to settle down to a steady-state
solution.
The initial conditions on position and velocity are:
X=V =0
The reiterative loop equations then take the form in BASIC:6

6 Or in FORTRAN, DO T = 0,Tmax,1 followed by END DO instead of NEXT T.

346 A The Harmonic Oscillator

FOR T = 0 TO T max ’ w i t h i m p l i e d s t e p s o f T = 1
S1 = SIN (W ∗ T ) ’W = D r i v i n g F r e q u e n c y . S1 and C1
are d e f i n e d to avoid
C1 = COS(W ∗ T ) ’ r e c a l c u l a t i n g t h e SIN and COS w i t h i n
each i t e r a t i o n
A = E∗S1−G∗V−W02∗ (X+X∗X / Xq+X∗X∗X / Xc )
B = W∗E∗C1−G∗A−W02∗ (V+2∗X∗V / Xq+3∗X∗X∗V / Xc )
C = −W2∗E∗S1−G∗B−W02∗ (A+2∗(X∗A+V∗V ) / Xq+3∗(2∗X∗V∗V
+X∗X∗A ) / Xc )
D = −W3∗E∗C1−G∗C−W02∗ (B+2∗(V∗A+X∗B+2∗V∗A ) / Xq )
D = D−W02∗ 3∗ ( 2∗V∗V∗V+4∗X∗V∗A+2∗X∗V∗A+X∗X∗B ) / Xc
X = X+V+A/ 2 +B/ 6 +C/ 2 4 +D/ 1 2 0 ’ Use V t o f i n d X b e f o r e
changing V
V = V + A + B / 2 + C / 6 + D / 24
REM P l o t or P r i n t r e s u l t s h e r e NEXT T
NEXT T
Note that A = dV /dt, B = dA/dt, C = dB/dt, D = dC/dt and that the time
scale was defined so that t = 1 in the Taylor expansion for A and step size in the
loop on T .

A.13 Examples of Nonlinear Oscillator Solutions

Solutions of the nonlinear equation containing both quadratic and cubic force terms
are shown in Fig. A.8, where it was assumed that Xq = Xc = 10 and the driving
frequency was tuned to the normal resonance frequency, ω0 . A damping constant
γ = 0.5ω0 was chosen for this illustration.
Time increases from t = 0 at the left of each trace in Fig. A.8 and the full
transient response of the oscillator for the mass starting from rest at x = 0 is
shown for increasing values of the driving amplitude, E. The driving term is shown
in dotted lines and the oscillator response is in solid curves. For E = 0.01 (the
lowest curve), the response is nearly sinusoidal. However, as the driving amplitude
increases, various even and odd harmonics develop in the waveform. The behavior
of the oscillator becomes very complex when the excursion of the mass reaches
the critical region where x ≈ Xq = Xc = 10. For the conditions assumed, that
point occurs at E ≈ 0.1. Although the different Fourier amplitudes are continuous
in their dependence on E, there are discontinuities in slope that appear when new
harmonics cross threshold. Many of these discontinuities result in a reversal of
the direction of the dependence of a given harmonic on excitation level. As a
result, small changes in E can produce large changes in the spectral distribution.
Even harmonics are present that also oscillate with increasing excitation level, but
decrease in relative importance above E ≈ 0.05. The cubic characteristic dominates
at large excitation levels where one sees a series of peaks in the different odd
A The Harmonic Oscillator 347

Fig. A.8 Solutions to the nonlinear oscillator equation when driven at the normal oscillator
resonance frequency with γ = 0.5ω0, and Xq = Xc = 10. E increases from bottom to top.
The driving wave (amplitude E) is shown dotted. A histogram of the relative spectral amplitudes is
shown at the right in each case, normalized to the response at the driving frequency (the first term
represents the response at DC)
348 A The Harmonic Oscillator

harmonics. The waveform amplitude saturates above E ≈ 0.5, although more

and more ripples occur with increasing excitation. That behavior is characteristic
of many reed and brass instrument, not to mention the human voice. Above the
saturation level, the sensation of further loudness arises primarily due to the increase
in harmonic content rather than output power.
The relative harmonic amplitudes for N = 0, 1, 2, 3, . . . , 10 normalized to the
fundamental component (N = 1) are shown at the right in Fig. A.8 as a histogram,
as determined by Fourier analysis. (N = 0 corresponds to a DC rectification term.)
Rectification, harmonics, and even subharmonics are an automatic consequence of
the nonlinearities assumed. Although subharmonics are not shown in the Fourier
spectrum of Fig. A.8, their production in the case of a strong quadratic nonlinear
term is discussed in the chapter on the human voice.

A.14 Dynamic Chaos

At very high values of the driving force E, the nonlinear oscillator begins to exhibit
characteristics of chaos in its motion. As an extreme example, consider the case
where there is no damping force at all in the presence of large quadratic and cubic
terms in the spring constant with high values of a periodic driving force. Here, even
though the solutions are still bound and have strong spectral components at the
driving frequency and its higher harmonics, the solutions are not periodic (i.e., they
don’t repeat themselves from one period to the next). Hence, the oscillator motion
becomes unpredictable in detail from one cycle of the driving frequency to the next.
An example of this behavior is shown in Fig. A.9. It is not clear that this regime
of the nonlinear mechanical oscillator has any direct relevance to the behavior of
the real human voice or other musical instruments. But some laryngologists have
suggested that dynamic chaos may be present in vibrations of the vocal fold in the

Fig. A.9 Example of chaotic behavior of the nonlinear oscillator. Here, the damping has been
eliminated altogether (γ = 0) and the oscillator is driven at a very high excitation level (E = 1).
As before, the driving term is at the normal resonance frequency and indicated by dotted lines.
Note that the waveform does not repeat itself precisely from one period to the next
A The Harmonic Oscillator 349

singing voice. (See Sataloff et al. 1998 and Hawkshaw et al. 2001.) However at the
present time, these suggestions seem largely to be speculation drawn by analogy to
other biomechanical systems where chaos has actually been observed.

A.15 Subharmonic Solutions

There is some indication that subharmonics may be present in some musical

instruments. For example, the subtones produced by violinist Mari Kimura through
use of very large bowing pressure were interpreted by her as subharmonics. (See
Chap. 5.) Similarly, the extremely low frequencies produced by the Gyütö monks
and the Tuva throat singers might also be due to subharmonic generation. (See
Chap. 6 on the singing voice (Fig. A.10).)
Figure A.11 shows a subharmonic solution to the nonlinear oscillator when it
only contains a quadratic nonlinear term in the spring constant. As discussed above,
the oscillator is highly unstable under these conditions. A very weak potential well
exists to trap the mass and the driving force has to be exceedingly small (or the

Fig. A.10 Variation of

Fourier coefficients for
nonlinear oscillator solutions
of the type in Fig. A.9. As a
function of driving amplitude
350 A The Harmonic Oscillator

Fig. A.11 Subharmonic solution to the forced nonlinear oscillator containing a quadratic term in
the spring constant

damping very high) to prevent the oscillator from flying apart, which will happen if
|x| ≥ Xq .
To illustrate a subharmonic solution, it is useful to assume that the driving
acceleration is applied at twice the resonant frequency of the normal linear oscillator.
Hence, we will assume

E(t) = E sin(2ω0 t) (A.57)

as indicated by the dotted lines in Fig. A.11. The subharmonic then occurs at ω0 . The
damping constant γ for the solution in the figure was chosen to be much smaller than
the value (2ω0) for critical damping in order to enhance production of the suboctave
harmonic. (Although adding a large damping term would increase the stability of
A The Harmonic Oscillator 351

the oscillator, it would suppress the subharmonic component.) The solution shown
includes the transient response when the driving force is initially turned on with
the particle at rest at x = 0. The oscillator amplitude (solid curve) decays into
the steady-state solution at the extreme right of the figure, where the fact that it is
periodic in half the driving frequency is easily seen by eye. The first two Fourier
coefficients and their phases are shown in Fig. A.11, where C(1) is the subharmonic
amplitude and C(2) is the amplitude of the response at 2ω0 .
The phases of the two components (Φ1 , Φ2 ) in Fig. A.11 vary in a complex
manner with the driving amplitude. As previously noted, the oscillator with a
quadratic nonlinearity acts as a rectifier. That is, a significant DC (“Direct Current”)
offset occurs in the solutions. With the human voice, the latter might correspond
to a net average opening of the glottis which increases with driving amplitude. As
discussed above, the presence of a cubic force term results in a stable potential well.
A cubic (or at least an odd-symmetric force term) is also to be expected with most
mechanical oscillators on a physical basis. As the spring in the mechanical oscillator
compresses, the coils will eventually touch each other, creating a strong repulsive
force back toward the equilibrium position (x = 0). An analogous situation occurs
in the larynx when the glottis completely closes. At the other extreme, a large
opening of the glottis would result in increased contact with other tissue in the
larynx and provide a similar repulsive force back toward equilibrium. Hence, there
is justification for an odd-symmetric force at large amplitudes and a cubic term is
the simplest one to incorporate. However, the presence of the cubic term produces
odd harmonics and reduces the subharmonic content.

A.16 The Relaxation Oscillator

A relaxation oscillator was used in the Dudley VODER discussed in the chapter on
speech synthesis and a mechanical analog to the relaxation oscillator occurs in the
grab-slip phenomenon in bowed strings discussed in the chapter on bowed strings.
A very simple version is shown in Fig. A.12.
In Fig. A.12, a battery or other DC (“Direct Current”) voltage supply at the left
charges the capacitor C by means of a current i = dq/dt flowing through the
resistor R. Ignoring the neon tube at the right of the circuit for the moment, the
charge q flowing to the capacitor would obey the equation,

Fig. A.12 A simple

relaxation oscillator circuit
352 A The Harmonic Oscillator

dq q
E=R + , (A.58)
dt C
where E is the battery voltage (or “electromotive force”). Dividing the equation
by R and using the integrating factor e(t/RC) , it is seen that the voltage across the
capacitor is

q(t)  
V (t) = = E 1 − et/RC , (A.59)
C
where we have assumed that there was no initial charge on the capacitor before
connecting the battery. Expanding the exponent,
 
t t 2
V (t) = E + Order . (A.60)
RC RC

Hence, for t RC, the voltage V (t) rises linearly with time.
It is a characteristic of neon bulbs that once the voltage reaches a threshold value
called the “ignition voltage,” a value somewhat above the ionization potential of
the neon atom (≈ 21.6 V) but dependent on the pressure and electrode geometry,
a discharge occurs through the gas. The discharge current persists until the voltage
drops below the “extinction voltage,” a value somewhat below the first excited state
of the atom (≈ 11.5 V), but again modified by the geometry and pressure. Hence,
the capacitor, C, will be repeatedly charged up through the resistor R and then
discharged through the neon bulb. The period for this process will be in the order
of RC (a time constant characteristic of the circuit), but will be modified by the
magnitude of the power supply voltage.
If the discharge through the neon bulb occurred instantaneously, the output
voltage V (t) in the limit that t«RC would approach an ideal sawtooth waveform,
which has harmonic amplitudes that decrease as 1/n, where n is the harmonic
number; i.e., the power spectrum falls off as 1/n2 . For example, if the RC time
constant were adjusted so that the fundamental frequency of the oscillator were
120 Hz (typical of the vocal cord resonance in an adult male and the source
frequency used in the Dudley VODER and VOCODER), the power spectrum of
the oscillator would be distributed as shown by the solid line in Fig. A.13. In
practice, the discharge time is not zero and is limited by drift mobilities of ions
and electrons and other collision processes in the neon tube to values somewhat
less than a millisecond (but also modified by pressure and tube geometry). The
finite discharge time can reduce the harmonic output of the relaxation oscillator
significantly, as shown by the histogram in Fig. A.13. The histogram was computed
by Fourier analysis for an assumed discharge time of 2/3 of a millisecond. The
minima at the harmonics for N = 13 and 26 correspond to multiples of the ratio of
the oscillator period to the discharge time (≈13:1).
The limitations of the simple oscillator in Fig. A.12 may be overcome by using
more complex circuitry such as that shown in Fig. A.14. For example, the capacitor
A The Harmonic Oscillator 353

Fig. A.13 Power spectrum for a relaxation oscillator with finite discharge time compared with the
spectrum from an ideal sawtooth waveform. The histogram was computed from the square of the
Fourier coefficient amplitudes for the waveform

Fig. A.14 Relaxation oscillator used in the Dudley VODER (a manually operated synthetic
speaker) first exhibited at the 1939 San Francisco Exhibition and New York World’s Fair.
Reproduced from Dudley (1939) by permission of Lucent Bell Laboratories, Jean Dudley Tintle
and Richard Dudley. The spectrum produced by this circuit closely matched that for the Ideal
Sawtooth shown in Fig. A.13

could be charged by a constant current generator to provide greater linearity during

the charging cycle and the discharge of the capacitance could be achieved through
a triggered, high- mutual-conductance tube or avalanche diode. Dudley’s oscillator
was actually the inverse of the circuit shown in Fig. A.12. He used a gas triode with
a control grid and filament to charge the capacitance (in ≈ 0.3 ms). The capacitance
was then allowed to discharge by itself with roughly a 0.8-ms time constant. The
period of his oscillator (nominally ≈ 10 ms) was varied by biasing the control
grid on the gas tube. Because the discharge pulse occupied a small fraction of the
period, the power spectrum was very close to that for the ideal sawtooth shown in
354 A The Harmonic Oscillator

Fig. A.13.7 (Contemporary VOCODERS such as the MAM Model VF-11 use linear
sawtooth generators that do not depend on gas discharge tubes.)

A.17 The Helmholtz Resonator

A number of acoustical problems can be analyzed in terms of Helmholtz Resonators.

The diagram in Fig. A.15 is useful in deriving the resonant frequency of such a
volume resonator.
We first want to calculate the change in pressure in the large volume that occurs
when the gas (not necessarily just air) contained in the small cylinder is pushed into
the sphere of initial volume V0 . Since a small volume of gas is added and we are
ultimately going to consider changes in which that volume of gas goes in and out of
the sphere, it is appropriate to think of the process as one in which there is no net
heat flow in or out of the system. For such an “adiabatic” process involving an ideal
gas, it is shown in thermodynamic texts that8

P V γ = constant, with γ = Cp /CV , (A.61)

where P is the pressure of gas occupying volume V and γ is the ratio of specific
heats for the gas at constant pressure CP and constant volume CV . The ratio γ varies
appreciably depending on the molecular complexity of the gas. (See Table A.2.)
Now suppose a piston of area A is pushed through the cylinder of length L in
Fig. A.14 forcing the gas from the small volume

ΔV = A × L (A.62)

Fig. A.15 Schematic

diagram of a Helmholtz
oscillator of volume V0

7 See Dudley 1939; Figs, 7, 10, and 11.

8 See, e.g., Zemansky (1951, Section 6.7.)
A The Harmonic Oscillator 355

Table A.2 Values of γ for different common gases near room temperature
Gas Wet air Helium Argon Hydrogen Nitrogen Carbon dioxide Methane
γ 1.37 1.67 1.67 1.40 1.40 1.29 1.30
Source: Zemansky (1951, pp. 129, 130). Zemansky notes that a method for measuring γ was
devised by Rüchhardt in 1929 in which the mass of air in the Helmholtz cylinder was replaced
by a small metal ball fitting snugly in the tube and measuring the oscillation frequency when the
apparatus was oriented with the tube in the vertical direction

into the sphere. Differentiating Eq. (A.60) yields

ΔP V γ γ P V γ −1 ΔV = 0, (A.63)

where ΔP is the change from the initial pressure P in the sphere due to the change
in gas volume ΔV . Rearranging that equation,

ΔP = γ P ΔV /V . (A.64)

Hence, the change in pressure in the sphere is

γ P AL
ΔP = (A.65)
V0

which results in a force pushing back on the hypothetical piston per unit length of
the cylinder (i.e., the effective “spring constant” K for the system) given by

γ P A2
K= . (A.66)
V0

One thus has a situation where a pressure fluctuation pushing in on the cylinder
of gas with mass ρAL activates harmonic oscillation. Comparing these results with
the basic harmonic oscillator equations [Eq. (A.1) through Eq. (A.4)], the oscillation
frequency at resonance is seen to be
 
γ P A2  γPA
ω0 = 2πf0 = / ρAL = . (A.67)
V0 ρLV0

The resonant frequency can be rewritten in terms of the velocity of sound c in the
gas as
 
c A γP
f0 = where c = (A.68)
2π LV0 ρ
Appendix B
Vibrating Strings and Membranes

B.1 The Wave Equation for the Vibrating String

Just as with the harmonic oscillator described in Appendix A, the differential

equation describing the motion of a vibrating string can be derived by a simple
application of Newton’s law of motion, F = ma. Here, it is especially important to
realize that both the force F and the acceleration a are vectors (i.e., have direction
as well as magnitude).
Consider the element of string under tension T between positions x and x + Δx
along the horizontal axis as shown in Fig. B.1. We assume that the magnitude of the
tension is constant throughout the length of the string and that the deflection y(x) in
the vertical direction is small. By small, we mean specifically that the angle θ that
the curve makes with respect to the x-axis is small enough so that

∂y
sin θ = tan θ = (B.1)
∂x
at each point on the curve. The net force on the string in the vertical (y) direction is
then given by the difference in y components (each of form T sin θ ) of the tension
between x and x + Δx. Defining μ to be the mass per unit length and using
approximation B.1, the net vertical component of Newton’s equation acting on the
differential string element Δx with mass Δm = μΔx becomes
   
∂y ∂y ∂ 2y
Fy ≈ T − = Δmay = μΔx . (B.2)
∂x x+Δx ∂x x ∂t 2

(The horizontal components of force cancel within the present approximation.)

© Springer Nature Switzerland AG 2018 357

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
358 B Vibrating Strings and Membranes

Fig. B.1 Forces on a differential element of a vibrating string

Next, we expand the function y/x in a Taylor series about the point x, noting1
     
∂y ∂y ∂ ∂y
= + Δx 1 + Order(Δx 2 ). (B.3)
∂x x+Δx ∂x x ∂x ∂x x

Substituting Eq. (B.3) into the left side of Eq. (B.2) gives
 
∂ ∂y
Fy ≈ T Δx 1 + Order(Δx 2 ). (B.4)
∂x ∂x x

Substituting Eq. (B.4) into Eq. (B.2) and taking the limit as Δx → 0 yields

∂ 2y ∂ 2y
T = μ (B.5)
∂x 2 ∂t 2
which is the same as the wave equation,

∂ 2y 1 ∂ 2y
2
= 2 2 (B.6)
∂x c ∂t

1 The Taylor series permits evaluating any well-behaved function f(x) at a point displaced to x +Δx

by the infinite series

∂y ∂2y ∂3y
f (x + Δx) = f (x) + ∂x Δx 1 + 1
2 ∂x 2 x
Δx 2 + 1
3! ∂x 3 x
Δx 3 + . . . Here, we simply
x
let f (x) = ∂y/∂x.
B Vibrating Strings and Membranes 359

given in Chap. 1, provided the velocity of the wave is


T
c= (B.7)
μ

Note that μ is in units of mass per unit length.

B.2 General Solution of the Wave Equation

Equation (B.6) is linear and can be solved by separating the space- and time-
dependent variables through a substitution of the type

y(x, t) = X(x)T (t) (B.8)

where X(x) is only a function of x and T (t) is only a function of t. Substituting this
definition into Eq. (B.6) and dividing by y(x, t) yields

1 ∂ 2 X(x) 1 ∂ 2 T (t)
= =≡ −k 2 = constant. (B.9)
X(x) ∂x 2 c2 T (t) ∂t 2

That is, the only way the left side of the equation (which depends only on x) could
equal the right side (which depends only on t) for all values of x and t is to have both
sides equal to the same constant, chosen here to be −k 2 to insure that it is a negative
value. (If we had chosen a positive constant at this point the solutions would not be
oscillatory.) Equation (B.9) implies two separate differential equations,

∂ 2 X(x) ∂ 2 T (t)
+ k 2
X(x) = 0 and + k 2 c2 T (t) = 0. (B.10)
∂x 2 ∂t 2
These equations are both of the Harmonic Oscillator type discussed in Appendix A
and have solutions of the form

X(x) ∝ sin kx and T (t) cos ωt or T (t) ∝ sin ωt (B.11)

provided k 2 c2 = ω2 . (The solution X(x) ∝ coskx, although perfectly valid, would

not satisfy the boundary conditions on x in the present problem.) The requirement
that X(x) = 0 at both x = 0 and x = L in Eq. (C.11) is satisfied by

kn = nπ/L where n = 1, 2, 3, . . . , hence, ωn = kn c = nπ c/L. (B.12)

The cyclical resonant frequencies are then given by Fn = ωn/2π = nc/2L as

we showed in Chap. 1 by two different methods. Note that the spatial boundary
360 B Vibrating Strings and Membranes

conditions in Eq. (B.12) are what actually determine the resonant frequencies of
the string. Again, because the wave equation is linear, any linear combination of
solutions of the type described by Eq. (B.11) is a solution to Eq. (B.6). The most
general solution is of the form originally given by Daniel Bernoulli,
∞
 ∞

y(x, t) = An sin(nπ x/L) cos(2π nF0 t) or An sin(nπ x/L) sin(2π nF0 t).
n=1 n=1
(B.13)
The form on the left containing cos(2π nF0 t) is most useful when the shape of
the string is known at t = 0 (as in the case of the plucked string). The second
form containing sin(2π nF0 t) is most useful when the initial conditions involve the
velocity at some point on the string, because there one needs to evaluate ∂y/∂t ∝
cos(2π nF0 t) at t = 0 (as for a string struck by a piano hammer, or bowed on a
violin).2

B.3 The Plucked String

Here, the first form of the solution given in Eq. (B.13) is the most useful. At t = 0,
the string is distorted into a triangular shape of the type shown in Fig. B.2. [But
keep in mind that the solutions given above apply for sufficiently small deflections
of the string such that the approximation in Eq. (B.1) is satisfied. Figure B.2 greatly
exaggerates the relative size of the deflection for the sake of illustration.]
Consider what happens when a string is plucked at t = 0 at some specific point
x = P0 along its length, L. If the amplitude of the string at t = 0 at the plucking
point is A, we see from Eq. (1) that

Fig. B.2 Shape of a plucked

string at t = 0 (the amplitude
is greatly exaggerated)

2 The pointin both cases is, of course, that cos(2π nF0 t) = 1 at t = 0, which leaves we with
y(x, 0) = ∞ n=1 An sin(nπ x/L).
B Vibrating Strings and Membranes 361


Ax/P 0, for 0 ≤ x ≤ P0
y(x, 0) = f (x) = (B.14)
A(L − x)/(L − P0 ), for P0 ≤ x ≤ L.

From Eq. (B.13) at t = 0,

∞

f (x) = An sin(nπ x/L). (B.15)
n=1

Here, we can use the same orthogonality properties of the sine function summarized
in Chap. 1 to obtain the values of the coefficients An , yielding
 L
2
An = f (x) sin(nπ x/L)dx. (B.16)
L x=0

The integral must be broken into two parts in accordance with the form of f (x)
given in Eq. (B.14). The spectral amplitudes are then determined from
 P0  L
2 2
An = (Ax /P0 ) sin(nπ x/L)dx+ [A(L−x)/(L−P0 )] sin(nπ x/L)dx.
L x=0 L P0
(B.17)
Integrating by parts and collecting terms, one obtains

2AM 2
An = sin(nπ/M) (B.18)
(M − 1)n2 π

for the spectral distribution, where M = L/P0 is a measure of the plucking point
and it is assumed that n = 1, 2, 3, . . . is an integer. (M does not have to be an
integer.) Note that for M = 2 (plucking at the mid-point), one only gets odd
harmonics and that An = 0 for n equal to integral multiples of M.
The time-dependent motion of the string is then obtained by substituting the
amplitude coefficients An from Eq. (B.18) back into the original equation for y(x, t)
given at the left of Eq. (B.13). Examples of the motion are shown in Chap. 3.

B.4 The Struck String

Some instruments (especially, various forms of the piano and a few Hungarian
instruments such as the Cembalom used by Kodaly) hit the string with a hammer. In
this case, the right-hand solution in Eq. (B.13) is the appropriate form to start with
because the boundary condition at t = 0 is one on velocity. Hence, we will start
with a solution of the form,
362 B Vibrating Strings and Membranes

∞

y(x, t) An sin(nπ x/L) sin(2π nF0 t). (B.19)
n=1

The velocity distribution over the string is obtained by taking ∂y/∂t, giving

 ∞
∂y(x, t)
v(x, t) = = 2π F0 nAn sin(nπ x/L) cos(2π nF0 t). (B.20)
∂t
n=1

In principle one could integrate the equation over some finite pulse duration during
which the hammer was in contact with the string. However, we shall just assume
here that a velocity distribution is suddenly imparted to the string by the hammer
before the string has a chance to move. The approximation will be best for the
low notes on the instrument where the vibrational periods are longest. In that
approximation the velocity distribution at t = 0 is given by
∞

v(x, 0) = V (x) = 2π F0 nAn sin(nπ x/L). (B.21)
n=1

Then, in analogy with Eq. (B.16), the spectral coefficients are given by3
 L
1
An = V (x) sin(nπ x/L)dx. (B.22)
nF0 L 0

As an example, consider a hammer that imparts a rounded velocity distribution to

the string of the type shown in Fig. B.3, for which the velocity distribution is given
by
  2 
P0 − x
V (x) = V0 1 − for P0 − R ≤ x ≤ P0 + R (B.23)
R

and V (x) is assumed to be zero everywhere else. V (x) has a rounded maximum
value of V0 at x = P0 , with an effective width of 2R. (Roughly speaking, R
corresponds to the hammer radius at the tip.) Although the shape was arbitrarily
assumed, it is not unlike that found at the top of felt hammers currently used in
grand pianos.
For the assumption in Eq. (B.23), the spectral coefficients in Eq. (B.22) become


L L/2, for m = n.
3 i.e., noting that for m, n integers, 0 sin(mx/L) sin(nx/L)dx =
0, for m = n.
B Vibrating Strings and Membranes 363

Fig. B.3 Shape of the

velocity pulse given by
Eq. (B.23)

    
2H 2 H +M H −M
An = V0 4 4 − cos nπ + cos nπ
n π F0 MH MH
    (B.24)
nπ H −M nπ H +M
− sin nπ − cos 2π
H MH H MH

where n is the harmonic number, H = L/R (the string length divided by the
hammer radius) and M = L/P0 (the ratio of the string length to the striking point,
in analogy to the case of the plucked string). As in the case of the plucked string,
Eq. (B.24) gives only odd harmonics when M = 2, or the string is struck in the
middle. Also, An = 0 for n = M (or the string is struck at a node for the M th
harmonic). In this approximation, voicing the hammer corresponds to adjusting the
value of H .
The shape of the string as a function of time is obtained by putting the values of
An from Eq. (B.24) back into the right-hand side of Eq. (B.13). Immediately after
the hammer hits the string, a narrow pulse (of width 2τ at the left of Fig. B.5) pops
up at the striking point (x = P0 , occurring at t = 0 in the figure). This narrow
pulse consists of two equal-amplitude, oppositely-directed running waves. As time
increases, the initial pulse broadens until the wave running to the left bounces off
the support at x = 0, where it undergoes a “hard” reflection and changes sign.
Now negative, it travels back in the +x direction, canceling out its previous positive
portion. Meanwhile, the running wave initially moving to the right has continued
on its way. The result of adding these two running waves together is an isolated
broader, positive pulse (at the right in Fig. B.4) that runs the length of the string.
The rise and fall times (τ ) of this wider pulse are each equal to half of the initial
narrow-pulse duration. However, the breadth of the wide pulse is determined by the
time delay taken for that half of the initial pulse that bounces off the support at
x = 0 to get back to the striking point at x = P0 . Hence, as indicated in the figure,
the broad pulse time duration is 2P0 /c, where c is the velocity of the wave. After
reaching the point x = L, two “hard” reflections (in succession for each running
wave) occur which send an inverted pulse back toward x = 0. Examples of these
solutions, together with their spectral distribution, are shown in Chap. 4.
364 B Vibrating Strings and Membranes

Fig. B.4 Pulses launched on the string by the striking process. Left: The initial pulse at the striking
point. Right: The broader pulse running down the string after the first reflection

B.5 The Bowed String

As described in Chap. 5, the excitation of a violin string consists of a stick-slip

process which is repeated at the fundamental round-trip frequency (c/2L) and
causes pulses to run back and forth on the string of length L. The problem is different
from the plucked- and struck-string problems treated above in that the string at the
contact point (which we will again take to be x = P0 ) is forced to move at about
the constant velocity of the bow until slipping occurs. When the string does slip,
it returns rapidly to its initial point where it is again grabbed by the bow and the
process repeats. The string displacement thus executes a sawtooth motion at the
point P0 . Helmholtz (1885, pp.384–386) presented an approximate solution to the
problem in which he assumed straight-line motion in the two halves of the stick-
slip cycle. He then used Fourier analysis of this sawtooth motion and expressed the
spectral amplitudes in the solution to the string equation in terms of those Fourier
coefficients. The idealized motion is illustrated in Fig. B.5, where will assume the
amplitude varies from −1 to +1 in the vertical direction and note that the vertical
displacements are centered about the time axis.
As in the case of the struck string waveform, we will take the general solution
for the shape of the string to be
∞

y(x, t) = An sin(nπ x/L) sin(2π nF0 t). (B.25)
n=1

But here, the boundary condition that determines the expansion coefficients An is
on the time-dependent saw-tooth motion at the bowing point x = P0 shown in
Fig. B.5. In Helmholtz’s formulation of the problem, he expressed one cycle of the
motion in a general Fourier series including both sine and cosine terms in the time.
He then shifted the time axis to obtain a result involving a series of sine terms only.
Although that approach is perfectly valid and straightforward, there is a lot tedious
algebra required to keep track of all the terms. It is simpler to note at the start from
symmetry that, if we choose the origin of the time axis to be centered in the slip
cycle as shown in Fig. B.6, one only needs sine terms in the Fourier series. Thus, the
B Vibrating Strings and Membranes 365

Fig. B.5 Slightly more than two idealized Stick-Slip cycles, corresponding to up-bowing. The
total period is T = T1 + T2 , where T1 is the “Stick” time and T2 is the “Slip” time

Fig. B.6 Choice of time

origin to simplify the
calculation of the Fourier
series

series for the motion at the bowing point may be written

∞

y(P0 , t) = Cn sin(2π nF0 t). (B.26)
n=1

The Fourier coefficients are then given by

 +T /2
2
Cn = sin(2π nF0 t)dt
T T /2

2
 −T2 /2  T

= t+ sin(2π nF0 t)dt
T −T /2 2
   
2 +T2 /2 T2 T2 2
+ − t+ sin(2π nF0 t)dt
T −T2 /2 2 2 T2
366 B Vibrating Strings and Membranes

2
 T /2  T1

T2

2
+ − + t− sin(2π nF0 t)dt,
T T2 /2 2 2 T1

which simplifies to
    
2 T T T2
Cn = − 2 2 sin nπ where T = T1 + T2 . (B.27)
n π T1 T2 T

One then substitutes Eq. (B.27) in Eq. (B.26) and compares the result with
Eq. (B.25) for x = P0 . It is then seen that
∞
   
2 T T sin(nπ T2 /T )
y(x, t) = −A sin(nπ x/L) sin(2π nF0 t),
n2 π 2 T1 T2 sin(nπ P0 /L)
n=L/P0
(B.28)
where A has been introduced as an amplitude scaling factor and Eq. (B.28) was used
to compute the figures illustrating the Helmholtz method in Chap. 5. The singularity
that would occur if L/P0 were an integer is avoided by setting the partial amplitude
to zero for that harmonic. (That particular harmonic would not be excited because
the bow would be at a node.) For more general behavior such as that reported by
Pickering for real strings and discussed in Chap. 5, one must use the general form
of the Fourier series including cosine terms in which the Fourier coefficients are
computed numerically by methods equivalent to those discussed in Appendix C.

B.6 The Torsional Wave Equation4

As discussed in Chap. 5, the generation of torsional waves is important in the case

of large diameter bowed strings under large bowing force. In this case, a form of
the wave equation analogous to that in Eq. (B.5) also applies, but with very different
wave velocity.
First, consider a hollow uniform cylinder of radius r and thickness dr stretched
in the x-direction with the angular rotation at point x along the cylinder given by
∂ϕ/∂x. The shear force on the cylinder material is given by nr∂ϕ/∂x where n is
defined as the “rigidity” given by

Y Lateral Contraction
n≡ where , (B.29)
2(μ + 1) Longitudinal Extension

Y is Young’s modulus, and by definition

4 Nearlyall treatises on mechanics ignore the torsional wave equation. The present derivation is
based on one given by Lord Rayleigh (1877, pp. 243–254.)
B Vibrating Strings and Membranes 367

Actual Length − Natural Length

Longitudinal Extension ≡ . (B.30)
Natural Length

The quantity n lies between Y /2 and Y /3 for different materials.

The moment of inertia dI for a hollow cylinder of length dx is given by

dI = ρ2π r 3 drdx

where ρ is the mass density. The net change in resisting torque from shear over
that length is

∂ 2ϕ dI ∂ 2 ϕ
ΔTorque = n2π r 3 drdx 2
=n dx
∂x ρ ∂x 2

where we have made use of a Taylor expansion to show that

   
∂ϕ ∂ϕ ∂ 2ϕ
− = dx
∂x x+dx ∂x x ∂x 2

Applying Newton’s law relating torque and angular acceleration, we get the
torsional wave equation,

dI ∂ 2 ϕ ∂ 2ϕ ∂ 2ϕ 1 ∂ 2ϕ
n dx = dI dx or nx2dxdI t2dxor = . (B.31)
ρ ∂x 2 ∂t 2 ∂x 2 n/ρ ∂t 2

Hence, the torsional waves have a velocity given by


n
cT = . (B.32)
ρ

Note that the radial dependence of the moment of inertia cancelled out in these
equations. Therefore, the result for cT is independent of the radial mass distribution
as long as it has axial symmetry. One, of course, needs to know the values of the
mass density ρ and of n from Eq. (B.30) for the string material in order to compute
the velocity.

B.7 The Vibrating Membrane as a Two-Dimensional String

The extension of the wave equation to two-dimensional form follows in a straight-

forward manner from the derivation of the vibrating string equation given at the start
of this appendix. Here, we will consider a thin membrane extending in the x and y
directions with small vibrational amplitude in the z direction.
368 B Vibrating Strings and Membranes

Fig. B.7 View looking down

on the membrane

Here, we use a two-dimensional extension of the argument given in connection

with Fig. B.1. (See Fig. B.7.) The constant tension T will now be per unit length
along the x and y directions and we are concerned with the net restoring force in
the z-direction acting on a differential mass element Δm = μΔxΔy, where σ is the
mass density per unit area of the membrane. Applying Newton’s Law to the motion
in the z-direction for this differential membrane element, we get
        
∂z ∂z ∂z ∂z
Fz ≈ Tx Δy − + Ty ΔX − Δx
∂x x+Δx ∂x x ∂y y+Δy ∂y y

∂ 2z
= Δmaz = σ ΔxΔy .
∂t 2
As with the one-dimensional case, we can get the component of force normal to
the surface by expanding the functions dz/dx and dz/dy in a Taylor series about
the point x, y noting that
       
∂z ∂z ∂ ∂z ∂z
= + Δx 1 + Order(Δx 2 ) and
∂x x+Δx ∂x x ∂x ∂x x ∂y y+Δy
   
∂z ∂ ∂z
= + Δy 1 + Order(Δy 2 )
∂y y ∂y ∂y y

Substituting in the force equation and neglecting quadratic terms in Δx, Δy, we get
   
∂ ∂z ∂ ∂z
F z ≈ Ty Δx + Tx 1
Δy 1 + Order(Δx 2 , Δy 2 )
∂x ∂x x ∂y ∂y y

where we have assumed that Tx might not be equal to Ty for the sake of generality.
(That permits having different wave velocities in the two orthogonal directions.)
Substituting in the force equation and taking the limit as Δx, Δy → 0, we get a
non-isotropic form of the two-dimensional wave equation
B Vibrating Strings and Membranes 369

   
1 ∂ 2z 1 ∂ 2z σ ∂ 2z
+ = .
Ty ∂x 2 Tx ∂y 2 Tx Ty ∂t 2

Of course, if Tx = Ty , this expression reduces to the more usual isotropic form of

the two-dimensional wave equation,

∂ 2z ∂ 2z 1 ∂ 2z
+ = (B.33)
∂x 2 ∂y 2 c2 ∂t 2

for which the wave velocity, c, is the same in both directions. That is,

T
c= for T = Tx = Ty . (B.34)
σ

With the isotropic case, the variables are easily separable and we can write

Z(x, y, t) = X(x)Y (y)T (t) (B.35)

which substituted in the wave equation yields

1 ∂ 2 X(x) 1 ∂ 2 Y (y) 1 ∂ 2 T (t)

+ = ≡ −k 2 = contant. (B.36)
X(x) ∂x 2 Y (y) ∂y 2 c2 T (t) ∂t 2

The solutions must be valid for all values of x, y, and t and the only way that can
happen is for the separate terms to equal constants. Here as in the one-dimensional
case, we take a negative definite constant for the time-dependent part of the solution
to insure stable oscillatory solutions. The first two terms on the left of the equation
may be rewritten

1 ∂ 2 X(x) 1 ∂ 2 Y (y)
= −k 2
x , = −ky2 , where kx2 + ky2 = k 2 (B.37)
X(x) ∂x 2 Y (y) ∂y 2

where kx and ky are constants. This substitution gives rise to solutions of the form

X(x) ∝ sin kx x and / or cos kx x

Y (y) ∝ sin ky y and / or sin ky y (B.38)
T (t) ∝ cos ωt and / or sin ωt

where Z(x, y, t) is given by the product of the three.

Spatial boundary conditions determine the resonant frequencies. For example, if
the membrane is clamped on all edges, the solutions must have zero amplitude there
and the appropriate spatial functions are

X(x) ∝ sin kx x and Y (y) ∝ sin ky y (B.39)

370 B Vibrating Strings and Membranes

with

kx = nx π/Lx and ky = ny π/Ly where nx , ny = 1, 2, 3, . . . (B.40)

The oscillation frequencies of these modes are then given by

   2
 nx 2 ny
ωn = c kx2 + ky2 = π c + . (B.41)
Lx Ly

The equations for the homogeneous membrane give rise to modes of the type
illustrated in Fig. B.8.
Solutions for the nonisotropic membrane are of interest because of their relation-
ship to those in thin soundboards (for example, in harpsichords and violins) in which
the velocity of wave propagation is quite different in the two principal orthogonal
directions. The solutions in that case may be obtained in exactly the same way using
the principle of separability. However, it is easier to note that the solutions for the
non-isotropic case may be obtained from those for the isotropic case by a simple
coordinate transformation of the type

x
= Ty x and y
= Tx y. (B.42)

Hence, the case of nonequal velocities in the two orthogonal directions has solutions
that are equivalent to the isotropic case with different relative dimensions of the
rectangular membrane.

Fig. B.8 Modes for a rectangular membrane (or drumhead) in which the tensions (wave velocities)
are equal in both coordinate directions and the sides are in the ratio of 2:3
B Vibrating Strings and Membranes 371

As noted above, the equation of motion in this case is of the form

     
1 ∂ 2z 1 ∂ 2z σ ∂ 2z
+ = (B.43)
Ty ∂x 2 Tx ∂y 2 Tx Ty ∂t 2

Again the equation is separable in the time and space variables by the substitution

Z(x, y, t) = X(x)Y (y)T (t). (B.44)

Hence
     
1 ∂ 2 X(x) 1 ∂ 2 Y (y) σ 1 ∂ 2 T (t)
+ =
Ty X(x) ∂x 2 Tx Y (y) ∂y 2 Tx Ty T (t) ∂t 2
= −ω2 = constant (B.45)

where a negative constant was again chosen to assure a stable oscillatory solution
with angular frequency ω. For the two terms on the left of the equation to add up to
a constant, each must separately be constant. Hence,

∂ 2 X(x) ∂ 2 Y (y)
2
+ ωx2 Ty X(x) = 0 and + ωy2 Tx Y (y) = 0 (B.46)
∂x ∂y 2

with solutions
 
X(x) ∝ sin ωx Ty x and Y (y) ∝ sin ωy Tx y, (B.47)

where
 
ωx Ty = nx π/Lx and ωy Tx = ny π/Ly where nx , ny = 1, 2, 3, . . . (B.48)

That is, the modes are determined by the spatial boundary condition that the
vibrational amplitudes are zero on the boundaries; that is, we assume the membrane
is clamped on the edges. The oscillatory frequency for a given mode is then given
by
 
ω = ωx2 + ωy2 = π n2x /L2x Ty + n2y /L2y Tx (B.49)

and they, in general, will not be harmonically related.

372 B Vibrating Strings and Membranes

B.8 Circular Membranes

Drums utilize circular membranes in most cases. These are easiest to treat using
circular (or cylindrical) coordinates. Consider a coordinate system where the radius
vector r is in the xy plane at angle φ with respect to the x axis (Fig. B.9). (The new
and old z-axes are identical.)
It may be shown that the two-dimensional wave-equation operator transforms as5
 
∂ 2 A(x, y) ∂ 2 A(x, y) 1 ∂ ∂A(r, ϕ) 1 ∂ 2 A(r, ϕ)
2
+ 2
= r + 2 . (B.50)
∂x ∂y r ∂r ∂r r ∂ϕ 2

Hence, the wave equation in circular coordinates becomes

  
1 ∂ ∂A 1 ∂ 2A 1 ∂ 2A T
r + 2 2 = 2 2 where c = (B.51)
r ∂r ∂r r ∂ϕ c ∂t σ

c is the running wave velocity (assumed to be the same in all directions), T is the
constant tension per unit length around the circumference, and σ is the mass density
per unit area.
Again, we can separate the variables in the form

A(r, ϕ, t) = R(r)Φ(ϕ)T (t), (B.52)

where T (t) cos(ωt) or T (t) = sin(ωt).

Fig. B.9 Relation of

rectangular and cylindrical
coordinates

5 See, e.g., Weatherburn (1951), pp. 15 and 16.

B Vibrating Strings and Membranes 373

Then we get

∂ 2Φ
+ m2 Φ = 0 where Φ(ϕ) = cos(mϕ) or Φ(ϕ) = sin(mϕ). (B.53)
∂ϕ 2

Because Φ(ϕ) must be periodic in 2π so that the function closes on itself in one
revolution about z-axis, we must have m = 1, 2, 3, 4, . . . This condition means that
 
∂ 2R 1 ∂R ω2 m2
2
+ + 2
− 2 R = 0, (B.54)
∂r r ∂r c r

which is known as Bessel’s Equation and has solutions of the form

∞
 (x/2)m+2p
Jm (x) = (−1)p . (B.55)
p!(m + p)!
p=0

where Jm (x) is a Bessel function of real argument of order m.6 Hence, the radial
solutions for the circular membrane are of the form
ω

R(r) ∝ Jm r = Jm (kr). (B.56)

c
A particular time-dependent solution is then

A(r, ϕ, t) ∝ cos(mϕ)Jm (km,n r) cos(2πfm,n t). (B.57)

The nodal diameter for a particular allowed solution is determined by the value of
m and arises from the periodicity requirement on Φ(ϕ). The allowed values of the
parameter k are then determined by the requirement that the rim of the circular drum
corresponds to a root n of the Bessel function of that particular order, m. Thus, the
resultant values of k depend on two integers, m and n. The frequency of the mode
of vibration then is given by

km,n c
fm,n = (B.58)
2π
To illustrate a particular mode for a circular kettle drum of radius, R0 , it is
necessary to scale the argument of the Bessel function so that it goes through zero
for a particular root when r = R0 . Thus, a given kettle drum mode distribution
would be proportional to

6 During the Great Depression, mathematicians were hired by the WPA to compute tables of Bessel

Function up to order 600 or more and in high precision. These tables, which are now totally
obsolete if any rudimentary computer is available, were stored in the basement of Low Library
at Columbia University.
374 B Vibrating Strings and Membranes

Table B.1 Values of km,n (roots) for the Bessel function Jm

m, n 1 2 3 4 5
0 2.40483 5.52008 8.65373 11.7915 14.9309
1 3.83171 7.01559 10.1735 13.3237 16.4706
2 5.13562 8.41724 11.6198 14.796 17.9598
3 6.38016 9.76102 13.0152 16.2235 19.4090
4 7.58834 11.0647 14.3725 17.616 20.8269
Computed by the author

 
r
cos(mϕ)Jm km,n . (B.59)
R0

Approximate values for the first few roots of the Bessel function Jm are given in
Table B.1. (Such modes are sometimes called “eigen functions” and the quantities
km,n are “eigen values.”) Representative kettledrum mode shapes are shown in
Fig. B.10 in 3-D.
 B
Vibrating Strings and Membranes

Fig. B.10 The first few kettledrum modes from Eq. (B.59) and Table B.1 are shown in stereoscopic pairs. Note that for M = 0, there are no nodal diameters;
for M = 1, there is one nodal diameter; and for M = 2, there are two nodal diameters. Similarly, for N = 1 there is one circular node (at the rim of the drum);
for N = 2 there are two circular nodes; and for N = 3 there are three such nodal circles. To see these figures stereoscopically, hold the paper as near to your
eyes as permits comfortable focusing. For each pair, you should see four images—two with each eye. Make the two center images coincide in your brain. The
375

amplitudes are exaggerated for clarity (The figures were computed by the author)
Appendix C
Fourier Analysis

We will include here some of the mathematical details dependent on calculus that
were omitted in Chap. 2, together with a working program for discrete Fourier
analysis. We will restrict the discussion to “well-behaved” periodic functions which
obey

V (θ + 2π ) = V (θ ), (C.1)

for which Fourier (1822) showed that the function V (θ ) may be written
∞
 ∞

V (θ ) = C0 + An sin nθ + Bn cos nθ. (C.2)
n=1 n=1

We wish to evaluate the DC constant C0 together with the harmonic coefficients An

and Bn .
Determining the DC constant in Eq. (C.2) is particularly easy. Recalling the fact
that the definite integral is just equal to the area under the curve,1 it is apparent
that the integrals of all the sine and cosine terms in Eq. (C.2) over one period
will identically vanish; i.e., the sine and cosine functions have equal areas above
and below the horizontal axis. Hence, in averaging Eq. (C.2) over one fundamental
period, all of the sin nθ and cos nθ terms will drop out, leaving
 2π
1
C0 = V (θ )dθ (C.3)
2π 0

1 Thefact that the area under the curve in the integrand is given by the definite integral between
two points on the horizontal axis is sometimes called “The Fundamental Theorem of Calculus.”

© Springer Nature Switzerland AG 2018 377

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
378 C Fourier Analysis

where, as indicated using a notation invented by Fourier himself, the definite integral
runs from 0 to 2π radians.
Determining the coefficients An and Bn for the waveform is a little harder. Here,
we will make use of something known as the “orthogonality” of the sine and cosine
functions over one period.2 You can show from trigonometric identities that3
 2π
sin mθ cos nθ dθ = 0 (C.4)
0

for all integral values of m and n.

Similarly, it can be shown that4
  
2π 2π
π, for m = n.
sin mθ sin nθ dθ = cos mθ cos nθ dθ = (C.5)
0 0 0, for m = n.

where m and n are integers.

To evaluate the general coefficient Am , multiply both sides of Eq. (C.2) by sin mθ
and integrate from 0 to 2π . All the terms in the Fourier series then vanish except for
 2π
1
Am = V (θ ) sin mθ dθ. (C.6)
π 0

Similarly to get Bm , multiply Eq. (C.2) by cos mθ obtaining

 2π
1
Bm = V (θ ) cos mθ dθ. (C.7)
π 0

Then, as shown in Chap. 2, we may rewrite V (θ ) as

∞

V (θ )C0 + Cn sin(nθ + ϕn ), (C.8)
n=1

2 The term arises because the sines and cosines can be regarded as projections of orthogonal
“vectors” in a multi-dimensional space in the sense that their generalized dot- (or scalar-) products
vanish.
3 Since

2 sin mθ cos nθ = sin(m + n)θ + sin(m − n)θ

and m and n are integers, the integrals of the sine functions on the right side of this identity
over one period must vanish.
4 E.g., consider the identity

2 sin mθ sin nθ = cos(m + n)θ cos(m − n)θ.

If m = n, integrals over one period on the right side both vanish. But if m = n, the second term
on the right side of the trig identity is cos(0) = 1. Hence, for m = n,
 2π 
1 2π
0 sin mθ sin nθdθ = 2 0 dθ = π
C Fourier Analysis 379

where the constants are given by


Cn = A2n + Bn2 for n ≥ 1 and ϕn = arctan(Bn /An ). (C.9)

The procedure for determining the Fourier coefficients then consists of computing
the integrals in Eqs. (C.3), (C.6), and (C.7), which generally must be done
numerically from digital samples of the waveform over one period. Then one uses
Eq. (C.9) to determine the net harmonic amplitude and phase. There is one pitfall to
avoid in computing the phase from the arctangent relation in Eq. (C.9). Computer
programming languages and pocket calculators do not always have provision for
automatically putting the angle whose tangent is (B/A) in the correct quadrant. If
the computer only deals with the numerical result from evaluating x = B/A, it
cannot tell what the quadrant should be from the statement φ = ATN(x). If A is
positive, the answer for φ will be correct and the angle will fall in either the first
or fourth quadrant. However, if A is negative, the answer will be wrong and should
fall in the second or third quadrant. The problem is solved in BASIC by including
statements such as

ATN(B/A)
(C.10)
IF A < 0THEN φ = φ + π.

The FORTRAN programming language has a single arctangent statement of the

type ATAN2(B/A) that takes care of the problem. If any phase angle were off by
π , reconstruction of the waveform from the Fourier coefficients would give the
wrong shape. Of course, the spectrum of harmonic amplitudes wouldn’t be altered.
Reconstruction of the original waveform is, of course, obtained by substituting
the coefficients Cn and φ back into Eq. (C.8). If the reconstructed waveform does
not match the original, you have either made a mistake or have not taken enough
harmonics into account. (See Fig. 2.12.)

C.1 Energy Distribution in a Fourier Series

It is generally true in classical physics (i.e., non quantum-mechanical problems) that

the energy in a vibrating system is proportional to the square of the amplitude of the
disturbance. That fact is shown in detail for the harmonic oscillator in Appendix A.
What we are interested in here is the average value of the square of the amplitude
over the fundamental period of the Fourier series. The relative values are usually of
main concern rather than the absolute values. Suppose we have a waveform of the
type
380 C Fourier Analysis

∞

y(t) = Cn sin(nωt + ϕn ). (C.11)
n=1

Then, the average energy over one cycle is5

 2π ∞

1
Energy ∝ y(t)2 dt = Cn2 (C.12)
T 0 n=1

where T = 1/f = 2π/ω is the fundamental period of the waveform. Hence, the
relative energy in the nth harmonic is simply proportional to Cn2 .

C.2 Program for Discrete Fourier Analysis6

The following program is written in BASIC, but could easily be rewritten in

FORTRAN or another language. In what follows, REM means a remark that is
ignored by the computer. Explanatory comments after the apostrophe (’) are also
ignored when the computer runs.
REM D i m e n s i o n s t a t e m e n t s 7

DIM A( 5 0 ) , B ( 5 0 ) , C ( 5 0 ) , P ( 5 0 ) ’ For F o u r i e r
coefficients
DIM V( 2 5 5 ) ’ To s t o r e t h e Data
P i = 4∗ATN ( 1 . 0 ) ’ P i = Greek \ p i

REM E n t e r d a t a f o r one p e r i o d = P8
READ P

5 To prove the 2
statement, note that y (t) ∞could be written
y(t)2 = ∞ n=1 nC sin(nωt + ϕ n ) m=1 Cm sin(mωt + ϕn ).
Hence,  ∞ 
Energy ∝ T1 ∞
T
n=1 m=1 Cn Cm 0 sin(nωt + ϕn ) sin(mωt + ϕn )dt
where the terms for m = n vanish because of orthogonality of the sine functions.
6 Note: In order to save running time, it really pays to compile the program.
7 Arrays are just subscripted variables; e.g., A(N) is the same as the mathematical variable AN.

Dimension(DIM) statements tell the computer to set aside the maximum number of elements that
could appear in the array; e.g., DIM A(50) means that there could be 50 elements (harmonics) in
array A(N).
8 The READ statement assigns the next previously unread number within DATA statements in

the program to a given variable. E.g., here P = 109, which is the total number of points in the
following DATA statements and V(1) = 3. The loop structure, FOR I = 1 TO P permits using
the same statement, READ V(I), within the loop over and over for P times. The loop is closed by
the NEXT I statement. Note that the waveform is periodic. (The first and last points are both equal
C Fourier Analysis 381

FOR I =1 TO P
READ V( I ) NEXT I

REM Mode−Locked Garden Hose ( n o r m a l i z e d t o ± 1 0 0 0 )

REM P= Number o f p o i n t s i n P e r i o d = t h e f i r s t datum
DATA 109
DATA 3 , 2 0 , 4 3 , 6 8 , 1 1 1 , 1 7 6
DATA 2 7 3 , 4 2 6 , 6 3 4 , 8 6 1 , 1 0 0 0 , 9 8 9
DATA 8 6 4 , 7 0 5 , 5 6 3 , 4 5 2 , 3 5 8 , 2 7 3
DATA 1 9 9 , 1 3 6 , 8 5 , 4 5 , 1 7 , − 1 4
DATA −40 , −63 , −85 , −105 , −125 , −129
DATA −151 , −162 , −165 , −165 , −165 , −165
DATA −170 , −165 , −170 , −170 , −170 , −173
DATA −173 , −170 , −162 , −159 , −156 , −153
DATA −151 , −142 , −139 , −136 , −131 , −128
DATA −128 , −128 , −125 , −119 , −114 , −108
DATA −102 , −94 , −91 , −91 , −91 , −91
DATA −91 , −91 , −91 , −91 , −91 , −91
DATA −85 , −82 , −82 , −80 , −74 , −74
DATA −74 , −80 , −80 , −80 , −80 , −80
DATA −82 , −82 , −82 , −85 , −88 , −91
DATA −91 , −91 , −91 , −91 , −91 , −85
DATA −82 , −80 , −74 , −71 , −68 , −65
DATA −60 , −57 , −45 , −37 , −28 , −14
DATA 3

REM F u n d a m e n t a l f r e q u e n c y i n Hz
DATA 3 0 7 . 6 9 2 ’ Not u s e d i n t h i s program

REM D i s p l a y o f t h e Waveform 9
Xmax=550 ’ Xmax and Ymax a r e t h e maximum number
Ymax=500 ’ o f p i x e l s a v a i l a b l e and v a r y w i t h t h e
computer
REM P l o t H o r i z o n t a l A x i s
Pen =0
X=0
Y=Ymax / 2
GOSUBS PLOT

to 3.) The DATA were all taken with an A- to-D (Analog-to-Digital) converter with 1 part in 1000
resolution, hence are in the range 1000.
9 Xmax and Ymax are the maximum number of points (“pixels”) used on the plotting device. Pen is

a “Pen- Lift” variable. (Imagine plotting on an old-fashioned xy-recorder.) See Subroutine PLOT:
for its meaning.
382 C Fourier Analysis

Pen =1
X=Xmax
GOSUB PLOT
REM P l o t Waveform
Pen =0
FOR I =1 TO P
Y= 0 . 5 ∗ Ymax∗(1+V( I ) / 1 0 0 0 )
X= ( I −1)∗Xmax / ( P−1)
GOSUB PLOT
Pen =1
NEXT I

REM C h o s i n g t h e maximum number o f h a r m o n i c s 10

PRINT ‘ ‘Nmax ’ ’
INPUT Nmax
REM Harmonic A n a l y s i s 11

A0=2∗ P i / ( P−1) ’ A0 i s a c o n s t a n t u s e d i n t h e l o o p
on N .
FOR N=0 TO Nmax ’ N=0 c o r r e s p o n d s t o t h e c o n s t a n t o r
DC t e r m
A(N) = 0
B (N) = 0
FOR I =1 TO P
A(N) =A(N) +V( I ) ∗ SIN (N∗A0 ∗ ( I −1))
B (N) =B(N) +V( I ) ∗COS(N∗A0 ∗ ( I −1))
NEXT I
A(N) =A(N ) ∗ 2 / ( P−1)
B (N) =B(N ) ∗ 2 / ( P−1)

10 Normally one could simply estimate the maximum number of harmonics (Nmax) from the
number of wiggles in one period of the waveform and INPUT (enter) that number from the
keyboard. The present case is tricky because the waveform is a sharp pulse without wiggles. One
way is to make a guess (e.g., Nmax = 20) and see how closely the waveform reconstructed from
the Fourier coefficients agrees with the original. Then increase (or decrease) Nmax as needed. See
Fig. 2.12.
11 This part of the program takes most of the running time. A(N) and B(N) correspond to the

Fourier amplitudes given by the integrals in Eq. (C.6) and (C.7). They are done here numerically
by a method equivalent to drawing straight lines between the successive points and adding up
the areas (the “trapezoidal Rule”). Each variable A(N) and B(N) is initially set to zero and the
increments repeatedly added within the loop FOR I = 1 TO P-1 in a series of the typeA(N )0.5(y1 +
y2 )Δx0.5(y2 + y3 )Δx + . . . + 0.5(yP −1 + yP )Δx= (y1 + y2 + y3 + . . . + yP −1 )Δx + 0.5(yP −
y1 )Δxwhich simplifies because yP = y1 (i.e., the series is periodic.) Here, e.g., yi = V (I ) ∗
SIN(N ∗ A0 ∗ (I − 1)) and Δx = 1. The net Fourier amplitude C(N ) and phase P (N ) are
computed using Eqs. (C.6), (C.7), and (C.8).
C Fourier Analysis 383

C (N) =SQR(A(N) ∗A(N) +B(N) ∗B(N ) )

IF N>0 THEN P (N) =ATN( B (N ) / A(N ) ) ’ a v o i d DC t e r m
IF A(N) <0 THEN P (N) = P (N) + P i ’ arctangent
problem
NEXT N
REM DC or c o n s t a n t t e r m n e x t
C( 0 ) =B ( 0 ) / 2
REM P l o t H i s t o g r a m o f t h e C o e f f i c i e n t s
REM F i n d Maximum Harmonic c o e f f i c i e n t , Cmax
Cmax=−1E30 ’A n e g a t i v e number b e l o w t h e
s m a l l e s t l i k e l y C (N )
FOR N=1 TO Nmax
IF C(N) >Cmax THEN Cmax=C(N)
NEXT N
FOR N=1 TO Nmax
X= (N−1)∗Xmax / Nmax
Y=0
Pen =0
GOSUB PLOT
Y=C (N) ∗ Ymax / Cmax
Pen =1
GOSUB PLOT
NEXT N

C.3 Additional Waveforms

Additional musical instrument waveforms are presented below in the same format
as that given above for the garden hose for those who might like to study them
and for use elsewhere in this book. These data were taken in Davies Auditorium
at Yale, using a high-quality Sennheiser MKH104 omnidirectional microphone
with uniform response (±1 dB) over the range from 50 Hz to 20 kHz and the
Hewlett-Packard equipment shown in Fig. 2.9. The 10-bit A-to-D converter used
had a dynamic range of about 60 dB. The instruments were played without vibrato
by professional musicians to whom the author is greatly indebted. The brass
instruments were played by James Undercoffler; the violins, by Syoko Aki; the
oboe, heckelphone, krummhorn and rohr schalmei by James Ryan; and flute and
piccolo by Leone Buyse. I am indebted to the late William Liddell for the use of
his krummhorns, to Richard Rephann for the loan of the historic brass instruments
from the Yale Instrument Collection, and to Robert Sheldon of the Smithsonian
Collection in Washington, DC for playing the serpent. The heckelphone was loaned
by the Yale Concert Band and the rohr schalmei was borrowed from the author’s
personal pipe organ.
384 C Fourier Analysis

F-Cornet
DATA 107
DATA − 4 , 4 4 , 1 1 4 , 1 6 3 , 2 1 2 , 2 4 9
DATA 2 8 0 , 3 0 1 , 3 1 1 , 3 1 9 , 3 2 1 , 3 2 1
DATA 3 1 1 , 2 9 0 , 2 5 9 , 2 2 3 , 1 6 6 , 1 1 1
DATA 57 ,5 , −47 , −83 , −106 , −122
DATA −124 , −119 , −98 , −83 , −67 , −52
DATA −41 , −47 , −62 , −83 , −109 , −150
DATA −192 , −231 , −259 , −285 , −301 , −303
DATA −293 , −269 , −228 , −166 , −96 , −13
DATA 7 8 , 1 7 4 , 2 7 5 , 3 7 0 , 4 7 7 , 5 7 8
DATA 6 8 1 , 7 7 7 , 8 7 0 , 9 4 8 , 9 9 5 , 1 0 0 0
DATA 9 7 4 , 8 8 6 , 7 6 4 , 6 1 7 , 4 3 3 , 2 4 4
DATA 67 , −101 , −238 , −345 , −412 , −448
DATA −461 , −448 , −415 , −383 , −337 , −301
DATA −272 , −259 , −251 , −269 , −301 , −332
DATA −365 , −399 , −425 , −446 , −453 , −456
DATA −446 , −425 , −399 , −383 , −368 , −355
DATA −347 , −337 , −329 , −301 , −262 , −241
DATA −210 , −166 , −117 , −65 , −4
REM C o r n e t F r e q u e n c y i n Hz
DATA 3 1 6 . 0 7 5
Piccolo Trumpet
DATA 168
DATA −5 ,97 ,195 ,280 ,367 ,450
DATA 530 ,610 ,680 ,755 ,807 ,857
DATA 898 ,935 ,967 ,982 ,997 ,1000

P i c c o l o Trumpet DATA −755 , −720 , −673 , −623 , −563 , −500

DATA 168 DATA −433 , −355 , −272 , −193 , −110 , −5
DATA −5 ,97 ,195 ,280 ,367 ,450 REM P i c c o l o Trumpet
F r e q u e n c y i n Hz
DATA 5 3 0 , 6 1 0 , 6 8 0 , 7 5 5 , 8 0 7 , 8 5 7 DATA 597.672
DATA 898 ,935 ,967 ,982 ,997 ,1000
DATA 9 9 5 , 9 8 7 , 9 6 5 , 9 4 0 , 9 1 0 , 8 7 5
DATA 8 3 5 , 7 9 7 , 7 4 2 , 6 9 5 , 6 3 7 , 5 8 5
DATA 5 2 7 , 4 7 5 , 4 1 7 , 3 5 7 , 3 0 2 , 2 4 7
DATA 187 ,130 ,85 ,37 , −10 , −50
DATA −80 , −110 , −140 , −160 , −180 , −193
DATA −203 , −200 , −200 , −193 , −170 , −155
DATA −130 , −93 , −55 , −12 ,35 ,77
DATA 1 2 5 , 1 7 7 , 2 2 7 , 2 7 7 , 3 2 5 , 3 6 7
DATA 4 0 8 , 4 5 0 , 4 9 0 , 5 1 7 , 5 4 5 , 5 6 7
C Fourier Analysis 385

DATA 5 8 0 , 5 8 7 , 5 9 0 , 5 9 0 , 5 8 0 , 5 6 0
DATA 5 4 0 , 5 1 5 , 4 8 5 , 4 5 0 , 4 0 8 , 3 6 7
DATA 3 2 7 , 2 8 5 , 2 4 0 , 1 9 0 , 1 4 7 , 9 7
DATA 55 ,15 , −35 , −65 , −100 , −133
DATA −155 , −180 , −200 , −215 , −230 , −233
DATA −245 , −245 , −242 , −245 , −242 , −240
DATA −238 , −235 , −238 , −235 , −235 , −242
DATA −253 , −260 , −265 , −282 , −300 , −320
DATA −343 , −363 , −390 , −413 , −440 , −463
DATA −490 , −513 , −535 , −557 , −583 , −603
DATA −623 , −643 , −660 , −680 , −690 , −715
DATA −733 , −753 , −773 , −793 , −810 , −830
DATA −845 , −857 , −873 , −875 , −875 , −882
DATA −873 , −870 , −857 , −835 , −810 , −785
DATA −755 , −720 , −673 , −623 , −563 , −500
DATA −433 , −355 , −272 , −193 , −110 , −5
REM P i c c o l o Trumpet F r e q u e n c y i n Hz
DATA 5 9 7 . 6 7 2

REM V i s u a l r e c o n s t r u c t i o n from t h e F o u r i e r
c o e f f i c i e n t s 12

Pen =0
FOR I =1 TO P
V( I ) =C ( 0 ) ’ c o n s t a n t o r DC t e r m
FOR N=1 TO Nmax
V( I ) =V( I ) +C(N) ∗ SIN (N∗A0∗ I +P (N ) )
NEXT N
Y= 0 . 5 ∗ Ymax∗(1+V( I ) / 1 0 0 0 )
X= ( I −1)∗Xmax / ( P−1)
GOSUB PLOT
Pen =1
NEXT I

REM Dummy I n p u t t o r e t a i n d i s p l a y
INPUT Q$

12 Here, we reconstruct the Fourier Series to compare with the original waveform initially stored
in V(I). (The contents of the array V(I) are changed when the program runs.) Note that without
the constant or DC term, the reconstructed waveform may be offset vertically from the original
waveform. An effective DC offset can sometimes arise from the presence of low-frequency
background noise that is unrelated to the waveform of interest. Several languages also permit
constructing the sound of the waveforms, in which case the fundamental frequencies contained
in the last DATA statement is important.
386 C Fourier Analysis

END

SUBROUTINE13

PLOT

RETURN

REM Pen =0 p l o t s p o i n t s a t (X, Ymax−Y) , w i t h o r i g i n

a t ( 0 , Ymax )
REM Pen =1 Draws l i n e from l a s t p o i n t t o c o o r d i n a t e s
(X, Ymax−Y)
IF Pen =0 THEN POINT (X, Ymax−Y)
IF Pen =1 THEN LINE (X, Ymax−Y)
Note: Some programming languages permit constructing the sound of the
waveform. The fundamental frequencies for each waveform are included in Hz in
the DATA statements for that purpose.
French Horn (Loud)
DATA 151
DATA 21 ,112 ,188 ,242 ,273 ,318
DATA 356 ,387 ,399 ,394 ,385 ,371
DATA 349 ,309 ,264 ,200 ,133 ,76
DATA 38 ,21 ,17 ,17 ,26 ,52
DATA 67 ,95 ,121 ,147 ,159 ,171
DATA 176 ,166 ,157 ,143 ,140 ,138
DATA 128 ,124 ,124 ,124 ,133 ,143
DATA 147 ,147 ,143 ,140 ,135 ,133
DATA 119 ,105 ,93 ,76 ,62 ,26
DATA −12 , −52 , −95 , −138 , −181 , −214
DATA −252 , −283 , −299 , −314 , −309 , −295
DATA −268 , −249 , −240 , −233 , −214 , −195
DATA −166 , −128 , −76 , −10 ,64 ,121
DATA 176 ,216 ,245 ,273 ,285 ,295
DATA 292 ,292 ,302 ,314 ,333 ,340

13 The instructions POINT (X,Y) and LINE (X,Y) are intended to plot a point at coordinates X,
Y whenPen = 0 and a line from the previously plotted point to coordinates X, Y if Pen = 1,
respectively, assuming the origin is in the lower left-hand corner. Plotting commands of this type
are contained in most languages, but with varying designations. Most languages now put the origin
in the upper left hand corner of the plotting device, in which case one needs to replace Y by
Ymax-Y as has been done in the subroutine here. One may also need scaling factors dependent
on the maximum number of pixels available in the x and y directions. The convention on naming
subroutines varies from one version of BASIC to another. In some versions the term PLOT has a
different, specific meaning built into the operating system.
C Fourier Analysis 387

DATA 3 4 9 , 3 4 7 , 3 4 7 , 3 4 9 , 3 7 5 , 4 2 8
DATA 5 0 8 , 5 7 5 , 6 2 7 , 6 5 6 , 6 7 2 , 6 7 9
DATA 6 8 4 , 6 8 4 , 6 8 9 , 6 8 4 , 6 7 0 , 6 4 1
DATA 5 9 6 , 5 5 1 , 5 0 8 , 4 5 1 , 3 8 0 , 3 0 4
DATA 214 ,128 ,45 , −43 , −164 , −314
DATA −461 , −589 , −670 , −703 , −717 , −743
DATA −791 , −846 , −895 , −931 , −960 , −988
DATA −1000 , −971 , −924 , −886 , −855 , −838
DATA −836 , −841 , −838 , −829 , −774 , −698
DATA −613 , −527 , −437 , −335 , −219 , −95
DATA 21
REM F r e n c h Horn F r e q u e n c y i n Hz DATA 222.488
French Horn (Soft)
DATA 151
DATA 4 , 4 6 , 8 7 , 1 2 1 , 1 5 4 , 1 8 3
DATA 2 1 2 , 2 3 1 , 2 6 2 , 2 7 2 , 2 9 1 , 2 9 7
DATA 3 0 6 , 2 9 7 , 2 9 7 , 2 9 1 , 2 8 3 , 2 6 8
DATA 2 6 4 , 2 4 7 , 2 3 7 , 2 2 9 , 2 1 6 , 2 0 6
DATA 2 0 0 , 1 9 1 , 1 7 9 , 1 6 6 , 1 5 8 , 1 4 1
DATA 1 3 1 , 1 1 6 , 1 2 1 , 1 1 2 , 1 0 6 , 1 0 4
DATA 1 0 0 , 9 8 , 9 6 , 8 3 , 8 7 , 7 9
DATA 7 3 , 7 1 , 6 2 , 5 6 , 4 8 , 4 0
DATA 4 0 , 4 2 , 4 6 , 5 4 , 5 8 , 6 2
DATA 6 4 , 7 1 , 6 7 , 7 3 , 7 1 , 7 1
DATA 7 5 , 8 1 , 9 6 , 1 0 4 , 1 1 4 , 1 3 3
DATA 1 5 6 , 1 6 6 , 1 8 3 , 2 1 2 , 2 2 5 , 2 4 7
DATA 2 7 2 , 2 9 7 , 3 2 2 , 3 4 9 , 3 7 2 , 3 9 9
DATA 4 2 2 , 4 4 1 , 4 6 2 , 4 8 4 , 4 9 9 , 5 1 6
DATA 5 2 8 , 5 4 1 , 5 5 5 , 5 6 3 , 5 6 3 , 5 7 2
DATA 5 7 0 , 5 6 3 , 5 5 5 , 5 4 5 , 5 2 8 , 5 1 1
DATA 4 8 6 , 4 6 4 , 4 2 4 , 3 9 1 , 3 5 6 , 3 0 8
DATA 2 6 2 , 2 1 8 , 1 6 4 , 1 1 6 , 5 6 , − 1 0
DATA −71 , −135 , −210 , −279 , −345 , −407
DATA −482 , −541 , −607 , −661 , −719 , −773
DATA −825 , −867 , −911 , −940 , −965 , −979
DATA −996 , −1000 , −994 , −983 , −973 , −950
DATA −925 , −892 , −857 , −811 , −767 , −717
DATA −659 , −603 , −536 , −468 , −403 , −343
DATA −285 , −233 , −175 , −127 , −71 , −25
DATA 4
REM S o f t F r . Horn F r e q u e n c y i n Hz DATA 223.294
Ophicleide
DATA 196
388 C Fourier Analysis

DATA −4 , −117 , −186 , −235 , −283 , −324

DATA −348 , −393 , −421 , −421 , −377 , −324
DATA −247 , −170 , −73 ,40 ,126 ,170
DATA 158 ,142 ,113 ,45 , −32 , −81
DATA −138 , −202 , −259 , −287 , −312 , −340
DATA −360 , −364 , −364 , −377 , −364 , −348
DATA −316 , −308 , −287 , −275 , −291 , −324
DATA −348 , −377 , −393 , −413 , −405 , −389
DATA −364 , −340 , −328 , −324 , −324 , −312
DATA −291 , −259 , −259 , −259 , −267 , −287
DATA −291 , −283 , −279 , −247 , −170 , −73
DATA 4 9 , 1 5 8 , 2 5 9 , 3 7 2 , 4 6 2 , 5 2 6
DATA 5 7 9 , 6 0 7 , 6 1 1 , 6 0 7 , 5 7 5 , 5 1 8
DATA 4 6 2 , 4 1 7 , 4 0 1 , 3 8 5 , 3 6 0 , 3 3 2
DATA 3 0 8 , 3 0 0 , 2 6 7 , 2 3 5 , 1 7 8 , 1 2 1
DATA 7 7 , 3 2 , 2 4 , 3 2 , 6 1 , 8 1
DATA 8 9 , 9 3 , 1 0 9 , 9 7 , 9 7 , 1 0 5
DATA 1 3 4 , 1 9 4 , 2 5 1 , 2 5 5 , 1 8 6 , 7 7
DATA −24 , −85 , −101 , −97 , −85 , −101
DATA −101 , −40 ,40 ,105 ,142 ,142
DATA 1 5 4 , 1 8 6 , 2 2 3 , 2 7 1 , 3 1 6 , 3 8 1
DATA 4 2 1 , 4 2 1 , 4 0 5 , 3 6 4 , 3 2 8 , 3 0 4
DATA 3 1 6 , 3 6 4 , 4 2 1 , 4 7 8 , 5 1 8 , 5 1 4
DATA 486 ,413 ,304 ,142 , −89 , −348
DATA −632 , −911 , −1219 , −1510 , −1741
DATA −1915 , −1960 , −1911 , −1781 , −
DATA 1603 , −1381 , −1154
DATA −927 , −700 , −462 , −219 ,28 ,235
DATA 429 ,595 ,749 ,879 ,955 ,996
DATA 1000 ,1000 ,972 ,964 ,935 ,866
DATA 777 ,688 ,599 ,514 ,437 ,364
DATA 304 ,255 ,211 ,170 ,158 ,142
DATA 126 ,121 ,130 ,126 ,130 ,142
DATA 142 ,126 ,89 , −4
REM O p h i c l e i d e F r e q u e n c y i n Hz
DATA 85.103
Serpent
DATA 203
DATA −5 ,53 ,113 ,167 ,207 ,237
DATA 233 ,223 ,207 ,170 ,140 ,90
DATA 47 , −7 , −47 , −87 , −107 , −113
DATA −107 , −80 , −43 ,10 ,73 ,133
DATA 187 ,223 ,250 ,263 ,263 ,250
DATA 223 ,200 ,157 ,130 ,73 ,20
C Fourier Analysis 389

DATA −40 , −97 , −147 , −200 , −240 , −267

DATA −287 , −280 , −267 , −240 , −203 , −180
DATA −163 , −133 , −123 , −127 , −120 , −123
DATA −127 , −123 , −130 , −133 , −127 , −133
DATA −133 , −113 , −107 , −73 , −30 ,37
DATA 1 0 3 , 1 6 0 , 2 1 0 , 2 3 7 , 2 4 0 , 2 5 0
DATA 2 3 7 , 2 0 7 , 1 7 0 , 1 2 7 , 9 0 , 7 3
DATA 6 0 , 5 0 , 4 0 , 2 3 , 2 0 , − 1 3
DATA −47 , −80 , −120 , −150 , −187 , −193
DATA −193 , −173 , −150 , −133 , −107 , −80
DATA −57 , −50 , −43 ,0 ,60 ,143
DATA 2 3 7 , 3 2 3 , 4 4 0 , 5 3 3 , 6 2 0 , 6 8 7
DATA 7 4 0 , 7 4 7 , 6 4 0 , 4 6 0 , 2 5 3 , 2 0
DATA −353 , −737 , −1053 , −1333 , −1497
DATA −1527 , −1440 , −1273 , −993 , −667
DATA −327 , −20 ,210 ,340 ,340 ,260
DATA 1 6 0 , 8 0 , − 7 , − 2 0 , 5 0 , 1 4 7 , 2 7 7
DATA 4 2 0 , 5 5 3 , 6 6 7 , 7 7 3 , 8 5 3 , 9 4 3
DATA 9 9 7 , 1 0 0 0 , 9 7 0 , 9 2 7 , 8 1 0
DATA 6 7 3 , 5 0 0 , 2 7 3 , 4 0 , − 1 9 3
DATA −387 , −507 , −560
DATA −540 , −447 , −337 , −217 , −93 ,23
DATA 8 3 , 1 3 0 , 1 4 7 , 1 4 3 , 1 4 0 , 1 4 0
DATA 1 4 7 , 1 5 3 , 1 6 7 , 1 7 3 , 1 5 3 , 1 1 7
DATA 77 ,23 , −40 , −60 , −100 , −133
DATA −173 , −203 , −240 , −267 , −280 , −280
DATA −300 , −287 , −273 , −257 , −257 , −230
DATA −227 , −213 , −207 , −213 , −200 , −193
DATA −163 , −150 , −110 , −70 , −5
REM S e r p e n t F r e q u e n c y i n Hz
DATA 6 1 . 7 0 2
Andreas Amati Violin (G String)
DATA 171
DATA 14 , −153 , −330 , −507 , −641 , −732 , −
DATA 813 , −900 , −967 , −1000
DATA −967 , −880 , −737 , −603 , −498 , −450 , −
DATA 426 , −431 , −440 , −426
DATA −402 , −354 , −306 , −330 , −335 , −306 , −
DATA 273 , −177 ,53 ,211 ,364
DATA 474 ,545 ,593 ,608 ,641 ,641 ,603
DATA 574 ,493 ,431 ,416 ,431
DATA 512 ,603 ,651 ,651 ,584 ,459 ,316 ,
DATA 187 ,86 ,0 , −100 , −177
DATA −234 , −273 , −239 , −177 , −100
390 C Fourier Analysis

DATA 3 3 , 1 3 4 , 2 0 1 , 2 2 0 , 2 4 9 , 2 7 8
DATA 3 3 5 , 3 9 2 , 4 6 9 , 5 0 7 , 4 6 9 , 3 8 3 , 2 9 7
DATA 1 9 1 , 1 2 4 , 7 2 , 9 1 , 1 2 4
DATA 1 4 8 , 2 2 5 , 3 5 4 , 5 5 0 , 7 7 5 , 8 9 5
DATA 8 8 0 , 6 8 9 , 3 9 7 , 1 4 4 , − 1 0 , 2 9
DATA 1 4 4 , 2 3 9 , 3 1 1 , 2 7 8 , 2 0 6 , 1 6 3
DATA 2 3 4 , 3 3 5 , 3 8 3 , 4 4 0 , 3 9 2
DATA 258 ,33 , −62 , −24 ,57 ,115
DATA 96 ,29 , −124 , −234 , −249
DATA −182 , −38 ,57 ,110 ,115 ,124
DATA 1 4 4 , 2 3 0 , 3 4 0 , 4 4 0 , 4 7 4
DATA 402 ,301 ,196 ,67 , −62 , −201
DATA −344 , −445 , −464 , −388
DATA −273 , −144 , −38 ,129 ,244 ,392
DATA 5 0 7 , 5 3 1 , 5 1 2 , 4 3 5
DATA 3 7 8 , 3 5 4 , 3 4 4 , 3 7 8 , 4 3 5 , 5 3 6
DATA 6 4 6 , 7 3 7 , 7 6 6 , 6 9 9 , 6 2 7
DATA 5 4 5 , 5 1 2 , 4 7 8 , 4 7 4 , 4 5 9 , 4 0 2
DATA 3 7 8 , 3 5 4 , 3 3 5 , 3 4 0 , 3 4 4
DATA 3 4 0 , 3 0 1 , 2 4 9 , 1 6 7 , 7 7 , 1 4
REM Amati V i o l i n F r e q u e n c y i n Hz
DATA 1 9 6 . 7 1
Ordinary Violin (G String)
DATA 130
DATA 76 ,389 ,686 ,854 ,945 ,980
DATA 872 ,600 ,290 ,27
DATA −75 ,7 ,168 ,325 ,456 ,566
DATA 662 ,748 ,796 ,761
DATA 644 ,491 ,336 ,226 ,137 ,111
DATA 157 ,254 ,332 ,403
DATA 496 ,606 ,690 ,690 ,597 ,440
DATA 254 ,77 , −102 , −215
DATA −296 , −354 , −407 , −460 , −513 , −566
DATA −637 , −746 , −878 , −976
DATA −1000 , −912 , −743 , −584 , −454
DATA −341 , −186 ,49 ,270 ,440
DATA 531 ,518 ,458 ,389 ,350 ,305
DATA 237 ,166 ,115 ,53
DATA −18 , −62 , −82 , −111 , −146 , −146
DATA −122 , −58 , −2 ,31
DATA 53 ,53 ,27 ,0 ,7 ,49
DATA 124 ,212 ,281 ,369
DATA 467 ,549 ,566 ,451 ,283 ,53
DATA −104 , −170 , −146 , −86
C Fourier Analysis 391

DATA −53 , −33 ,31 ,119 ,181 ,243

DATA 310 ,296 ,184 , −31
DATA −283 , −442 , −515 , −546 , −540
DATA −558 , −580 , −617 , −642 , −666
DATA −639 , −577 , −584 , −692 , −808 , −878
DATA −816 , −569 , −199 ,76
REM V i o l i n F r e q u e n c y i n Hz
DATA 1 9 4 . 4 1
Oboe (Lauré)
DATA 194
DATA 2 , −181 , −397 , −581 , −717 , −809
DATA −839 , −829 , −777 , −697 , −596 , −486
DATA −370 , −261 , −159 , −79 , −10 ,45
DATA 8 2 , 1 0 7 , 1 2 7 , 1 4 9 , 1 6 1 , 1 7 4
DATA 1 7 9 , 1 7 9 , 1 7 4 , 1 6 9 , 1 5 6 , 1 4 9
DATA 1 3 4 , 1 2 4 , 1 0 4 , 8 4 , 5 5 , 1 5
DATA −35 , −79 , −112 , −144 , −159 , −164
DATA −161 , −159 , −154 , −151 , −159 , −161
DATA −164 , −169 , −159 , −144 , −114 , −84
DATA −50 , −10 ,37 ,84 ,127 ,179
DATA 2 1 8 , 2 4 6 , 2 6 1 , 2 5 3 , 2 2 3 , 1 7 6
DATA 127 ,69 ,10 , −40 , −77 , −122
DATA −159 , −189 , −223 , −251 , −270 , −283
DATA −288 , −280 , −273 , −263 , −258 , −258
DATA −258 , −258 , −258 , −258 , −251 , −243
DATA −241 , −231 , −218 , −208 , −191 , −169
DATA −154 , −139 , −124 , −122 , −122 , −129
DATA −141 , −154 , −159 , −159 , −156 , −151
DATA −139 , −129 , −122 , −124 , −129 , −134
DATA −151 , −169 , −179 , −199 , −208 , −231
DATA −243 , −253 , −258 , −258 , −251 , −238
DATA −213 , −191 , −161 , −139 , −104 , −74
DATA −40 , −5 ,27 ,55 ,77 ,89
DATA 9 7 , 9 7 , 9 4 , 8 7 , 7 4 , 6 0
DATA 37 ,15 , −20 , −60 , −104 , −156
DATA −213 , −280 , −347 , −409 , −474 , −526
DATA −566 , −588 , −586 , −558 , −499 , −422
DATA −330 , −221 , −109 ,15 ,136 ,256
DATA 3 7 7 , 5 0 1 , 6 2 0 , 7 3 9 , 8 3 1 , 9 1 3
DATA 9 6 0 , 9 9 3 , 1 0 0 0 , 9 9 5 , 9 8 3 , 9 6 3
DATA 9 3 8 , 9 1 3 , 8 8 3 , 8 5 6 , 8 2 1 , 7 8 9
DATA 7 5 2 , 7 1 5 , 6 6 3 , 6 0 0 , 4 9 6 , 3 5 5
DATA 1 7 6 , 2
REM Oboe F r e q u e n c y i n Hz
392 C Fourier Analysis

DATA 2 5 9 . 5 8 1
Heckelphone
DATA 130
DATA − 8 , 1 4 2 , 2 5 0 , 3 0 8 , 3 0 0 , 2 8 3
DATA 2 6 2 , 1 9 6 , 5 8 , 3 3 , 4 2 , 4 2
DATA 29 , −8 , −54 , −104
DATA −225 , −321 , −417 , −475 , −488
DATA −417 , −242 ,0
DATA 2 1 7 , 2 2 5 , 3 4 6 , 6 0 8 , 8 6 7 , 1 0 0 0
DATA 942 ,642 ,25 , −50 , −108 , −267
DATA −475 , −650 , −667 , −471 , −217 , −204
DATA −167 , −25 ,171 ,317 ,417 ,425
DATA 329 ,308 ,229 ,92 , −50 , −142
DATA −208 , −267 , −358 , −425 , −458 , −454
DATA −425 , −388 , −304 , −221 ,13 ,142
DATA 2 6 2 , 3 4 6 , 3 9 2 , 4 2 5 , 4 4 2 , 4 2 5
DATA 238 ,158 ,75 , −33 , −125 , −242
DATA −367 , −450 , −388 , −317 , −233
DATA −142 , −75 ,13 ,112 ,192 ,225
DATA 2 5 8 , 2 9 2 , 2 9 2 , 2 5 0 , 1 7 5 , 4 6 , − 1 0 4
DATA −254 , −292 , −321 , −317 , −292 , −25
DATA −188 , −121 ,50 ,196 ,396 ,612
DATA 7 4 6 , 7 2 5 , 6 0 8 , 4 7 9 , 2 6 2 , 1 6 7
DATA 50 , −117 , −283 , −400 , −417 , −350
DATA − 6 7 , 1 0 8 , 2 5 8 , 3 6 7 , 4 4 2 , 4 7 5
DATA 500 ,458 , −17 , −321 , −600 , −817
DATA −892 , −808 , −671 , −538 , −192 , −8
REM H e c k e l p h o n e F r e q u e n c y i n Hz
DATA 1 2 9 . 7 7 7
Krummhorn
DATA 174
DATA −59 , −91 , −157 , −216 , −100 , −78
DATA −162 , −176 ,333 ,1000 ,623 , −130
DATA −412 , −355 , −100 ,422 ,853 ,850
DATA 490 , −100 , −623 , −760 , −760 , −603
DATA −150 ,255 ,157 , −169 , −277 , −196
DATA −250 , −309 , −150 ,113 ,206 ,279
DATA 377 ,331 ,108 , −74 , −51 ,15
DATA 145 ,353 ,510 ,507 ,333 ,164
DATA 93 ,152 ,243 ,341 ,431 ,439
DATA 419 ,373 ,257 ,230 ,186 ,59
DATA 64 ,135 ,201 ,206 ,86 , −78
DATA −194 , −196 , −142 , −83 ,47 ,103
C Fourier Analysis 393

DATA 39 , −32 , −91 , −167 , −199 , −120

DATA 0 , 3 4 , 2 7 , 5 4 , 8 3 , 1 1 8
DATA 1 2 5 , 1 3 7 , 1 9 6 , 2 3 0 , 1 7 4 , 1 1 5
DATA 83 ,20 , −120 , −316 , −471 , −412
DATA −277 , −81 ,152 ,211 ,47 , −137
DATA −267 , −353 , −319 , −176 , −69 , −25
DATA 1 0 , 4 7 , 9 3 , 1 0 0 , 1 3 2 , 1 4 2
DATA 1 4 2 , 1 6 2 , 1 5 7 , 1 0 5 , 2 7 , − 4 4
DATA −78 , −110 , −118 , −110 , −78 , −44
DATA −32 , −51 , −74 , −98 , −103 , −137
DATA −137 , −110 , −78 , −88 , −113 , −123
DATA −78 , −34 , −5 , −20 , −88 , −196
DATA −353 , −451 , −453 , −512 , −534 , −578
DATA −534 , −446 , −397 , −324 , −306 , −358
DATA −306 , −245 , −140 , −61 ,69 ,96
DATA 1 6 2 , 1 5 2 , 1 4 2 , 1 7 2 , 2 1 3 , 2 8 9
DATA 2 9 4 , 2 9 4 , 3 0 9 , 2 2 5 , 1 4 5 , − 5 9
REM Krummhorn F r e q u e n c y i n Hz DATA 1 9 2 . 5 2
Rohrschalmei (Laukuft)
DATA 129
DATA 4 , −94 , −132 , −186 , −172 , −224
DATA −256 , −202 , −146 , −116 , −68 , −20
DATA −36 , −64 , −64 , −104 , −168 , −136
DATA −98 , −16 ,82 ,126 ,212 ,158
DATA 158 ,152 ,20 , −80 , −188 , −282
DATA − 2 0 6 , 7 2 0 , 6 9 2 , 4 4 4 , 1 0 0 0 , 6 4 8
DATA 694 ,230 ,68 , −228 , −938 , −826
DATA −726 , −830 , −474 , −168 ,126 ,322
DATA 3 7 6 , 5 6 2 , 4 0 8 , 2 8 0 , 3 1 6 , 2 0 6
DATA 40 ,16 , −44 , −96 , −178 , −168
DATA −170 , −194 , −242 , −402 , −568 , −488
DATA −584 , −754 , −604 , −410 , −370 , −32
DATA 1 4 4 , 4 7 6 , 5 4 2 , 6 7 8 , 8 4 8 , 8 1 6
DATA 638 ,272 ,146 , −176 , −320 , −420
DATA −482 , −450 , −400 , −202 , −43 , −132
DATA −40 ,56 , −16 ,0 , −36 ,88
DATA 1 7 2 , 2 4 6 , 4 1 4 , 4 0 4 , 3 6 6 , 4 3 0
DATA 364 ,276 ,88 , −48 , −124 , −282
DATA −272 , −238 , −314 , −306 , −282 , −306
DATA −266 , −136 , −124 ,12 ,52 ,128
DATA 2 1 2 , 1 6 6 , 2 6 2 , 2 2 0 , 1 9 2 , 1 6 4
DATA 1 5 2 , 7 8 , 4
REM R o h r s c h a l m e i F r e q u e n c y i n Hz
DATA 2 5 9 . 6 7 4
394 C Fourier Analysis

Flute
DATA 192
DATA 18 , −54 , −75 , −101 , −134 , −165
DATA −178 , −247 , −281 , −291 , −335 , −376
DATA −428 , −425 , −464 , −438 , −446 , −425
DATA −402 , −412 , −330 , −299 , −250 , −160
DATA − 8 2 , 2 1 , 9 3 , 1 8 0 , 2 1 6 , 3 1 7
DATA 3 2 5 , 3 5 8 , 3 2 0 , 2 8 9 , 2 5 5 , 2 2 2
DATA 1 4 4 , 1 1 9 , 1 1 1 , 8 5 , 5 , 2 6
DATA 5 , −34 , −52 , −57 , −106 , −67
DATA −64 , −75 , −34 , −34 ,8 , −5
DATA 3 6 , 5 9 , 1 0 1 , 1 1 1 , 9 0 , 1 1 9
DATA 1 1 9 , 9 8 , 3 9 , 5 7 , 1 5 , − 2 6
DATA 3 , 1 5 , 0 , 8 , 8 , − 4 1
DATA −62 , −77 , −106 , −126 , −183 , −232
DATA −260 , −240 , −253 , −247 , −191 , −157
DATA −103 , −67 , −26 ,28 ,108 ,131
DATA 2 5 0 , 3 4 0 , 4 3 0 , 5 0 3 , 5 9 5 , 6 3 4
DATA 6 6 0 , 6 9 8 , 6 7 5 , 6 1 9 , 5 2 6 , 4 4 8
DATA 3 6 9 , 3 4 0 , 2 0 6 , 1 2 4 , 8 8 , − 2 6
DATA −82 , −178 , −250 , −291 , −302 , −371
DATA −407 , −410 , −436 , −464 , −521 , −559
DATA −582 , −652 , −765 , −881 , −920 , −972
DATA −912 , −910 , −838 , −678 , −577 , −479
DATA −353 , −247 , −134 , −21 ,62 ,121
DATA 1 5 5 , 2 1 6 , 2 0 4 , 1 9 3 , 2 2 2 , 2 0 1
DATA 2 0 4 , 2 1 6 , 2 3 7 , 2 9 1 , 3 7 6 , 4 2 8
DATA 5 8 5 , 7 9 9 , 8 4 0 , 8 9 2 , 9 5 9 , 1 0 0 0
DATA 9 0 7 , 7 4 7 , 5 8 8 , 4 3 0 , 2 2 2 , − 4 6
DATA −289 , −376 , −518 , −598 , −655 , −655
DATA −621 , −526 , −459 , −327 , −188 , −26
DATA 9 0 , 1 7 0 , 2 6 8 , 3 6 1 , 4 5 4 , 4 5 4
DATA 4 4 3 , 4 8 5 , 4 9 5 , 4 2 3 , 4 0 7 , 3 4 8
DATA 2 9 1 , 2 2 7 , 1 4 4 , 7 7 , 6 2 , 1 8
REM F l u t e F r e q u e n c y i n Hz
DATA 2 6 2 . 0 5
Piccolo
DATA 170
DATA −11 ,34 ,101 ,100 ,218 ,276
DATA 333 ,391 ,451 ,508 ,556 ,590
DATA 631 ,643 ,679 ,695 ,706 ,707
DATA 706 ,695 ,667 ,652 ,631 ,604
DATA 590 ,542 ,525 ,487 ,460 ,422
C Fourier Analysis 395

DATA 3 8 8 , 3 6 2 , 3 4 1 , 3 2 6 , 3 0 7 , 2 9 5
DATA 2 7 6 , 2 6 9 , 2 4 7 , 2 4 0 , 2 4 7 , 2 3 7
DATA 2 2 5 , 2 1 8 , 1 9 2 , 1 8 9 , 1 6 8 , 1 5 1
DATA 1 3 4 , 1 0 6 , 9 1 , 7 4 , 5 3 , 3 6
DATA 36 ,14 , −10 , −29 , −34 , −62
DATA −77 , −86. −115 , −115 , −125 , −129
DATA −139 , −146 , −149 , −146 , −153 , −146
DATA −146 , −137 , −137 , −127 , −127 , −118
DATA −108 , −89 , −72 , −55 , −38 , −10
DATA 2 4 , 4 8 , 9 4 , 1 2 5 , 1 7 0 , 2 0 6
DATA 2 4 7 , 2 8 5 , 3 1 7 , 3 5 3 , 3 8 8 , 4 0 8
DATA 4 3 6 , 4 5 8 , 4 6 0 , 4 8 4 , 4 9 4 , 4 7 7
DATA 4 8 2 , 4 6 8 , 4 5 8 , 4 5 1 , 4 2 9 , 4 1 2
DATA 3 8 8 , 3 6 9 , 3 4 5 , 3 3 1 , 3 2 1 , 2 9 3
DATA 2 6 6 , 2 4 9 , 2 1 6 , 1 9 9 , 1 7 3 , 1 6 1
DATA 115 ,86 ,46 ,22 , −14 , −70
DATA −120 , −182 , −245 , −317 , −386 , −480
DATA −556 , −616 , −686 , −751 , −815 , −859
DATA −897 , −935 , −959 , −981 , −993 , −990
DATA −990 , −1000 , −993 , −993 , −964 , −954
DATA −959 , −930 , −914 , −887 , −887 , −856
DATA −811 , −763 , −731 , −671 , −631 , −568
DATA −499 , −444 , −367 , −317 , −261 , −175
DATA −118 , −62,−11
REM P i c c o l o F r e q u e n c y i n Hz
DATA 5 9 3 . 8 7 7

C.4 Program for Discrete Fourier Analysis

The following program is written in MATLAB using much of the same code from
the original BASIC program by Bennett.

C.4.1 Contents

• Load waveform data for one period

• Variables
• Compute the Fourier Coefficients
• Plot the Waveform
• Plot Histogram of the Coefficients
• Reconstruct the Waveform from the Fourier Coefficients
396 C Fourier Analysis

• Plot the Waveform

clear all;

C.4.2 Load waveform data for one period

Call load_waveforms.m
load_waveforms;

C.4.3 Variables

% Choose the maximum number of harmonics

Nmax=20;

% Select which waveform to analyze

waveform = ’gardenhose’;

switch waveform
case ’gardenhose’
V=gardenhose;
strLegend=sprintf(’Garden Hose\nf_{n=1}
=307.692 Hz’);
case ’fcoronet’
V=fcoronet;
strLegend=sprintf(’F-Cornet\nf_{n=1}
=316.075 Hz’);
case ’piccolotrumpet’
V=piccolotrumpet;
strLegend=sprintf(’Piccolo Trumpet\nf_{n=1}
=597.672 Hz’);
case ’frenchhornloud’
V=frenchhornloud;
strLegend=sprintf(’French Horn (Loud)\nf_{n=1}
=222.488 Hz’);
case ’frenchhornsoft’
V=frenchhornsoft;
strLegend=sprintf(’French Horn (Soft)\nf_{n=1}
=223.294 Hz’);
case ’ophicleide’
V=ophicleide;
C Fourier Analysis 397

strLegend=sprintf(’Ophicleide\nf_{n=1}
=85.103 Hz’);
case ’serpent’
V=serpent;
strLegend=sprintf(’Serpent\nf_{n=1}
=61.702 Hz’);
case ’amativiolin’
V=amativiolin;
strLegend=sprintf(’A.Amati Violin (G)\nf_{n=1}
=196.71 Hz’);
case ’violin’
V=violin;
strLegend=sprintf(’Ordinary Violin
(G)\nf_{n=1}=194.41 Hz’);
case ’oboelaure’
V=oboelaure;
strLegend=sprintf(’Oboe (Laure)\nf_{n=1}
=259.581 Hz’);
case ’heckelphone’
V=heckelphone;
strLegend=sprintf(’Heckelphone\nf_{n=1}
=129.777 Hz’);
case ’kurmhorn’
V=krummhorn;
strLegend=sprintf(’Krummhorn\nf_{n=1}
=192.52 Hz’);
case ’rohrschalmei’
V=rohrschalmei;
strLegend=sprintf(’Rohrschalmei (Laukuft)\
nf_{n=1}
=259.674 Hz’);
case ’flute’
V=flute;
strLegend=sprintf(’Flute\nf_{n=1}=262.05 Hz’);
case ’piccolo’
V=piccolo;
strLegend=sprintf(’Piccolo\nf_{n=1}
=593.877 Hz’);
end
398 C Fourier Analysis

C.4.4 Compute the Fourier Coefficients

Call fouriercoef_trapezoid.m

[A0 A B0 B C0 C PHI]=fouriercoef_trapezoid(V,Nmax);

C.4.5 Plot the Waveform

figure;
plot(V./max(abs(V)),’Color’,’k’,’LineWidth’,3)
line([0 length(V)],[0 0],’Color’,’k’,’LineWidth’,1)
set(gca,’LineWidth’,2,’FontSize’,14)
set(gca,’YLim’,[-1.1 1.1],’YTick’,[0]);
set(gca,’XLim’,[0 length(V)],’XTick’,[]);
xlabel(’Time’,’FontSize’,14);
ylabel(’Microphone Signal’,’FontSize’,14);
pbaspect([2 1 1])
Microphone Signal

Time

C.4.6 Plot Histogram of the Coefficients

figure;
stem([1:Nmax],C./max(C),’Color’,’k’,’LineWidth’,3,
’Marker’,’none’)
set(gca,’YLim’,[0 1],’YTickLabel’,[]);
set(gca,’XLim’,[0 Nmax],’XTick’,[0:2:Nmax]);
set(gca,’LineWidth’,2,’FontSize’,14)
xlabel(’Harmonic Number, (n)’,’FontSize’,14);
C Fourier Analysis 399

ylabel(’Fourier Coefficients, C(n)’,’FontSize’,14);

pbaspect([2 1 1])
text(12,.8,strLegend,’FontSize’,14,
’HorizontalAlignment’,’left’);
Fourier Coefficients, C(n)

Garden Hose
fn=1=307.692 Hz

0 2 4 6 8 10 12 14 16 18 20
Harmonic Number, (n)

C.4.7 Reconstruct the Waveform from the Fourier Coefficients

P=length(V);
for I=1:P
V2(I)=C0;
for N=1:Nmax
V2(I)=V2(I)+C(N)*sin(N*A0*(I-1)+PHI(N));
end
end

C.4.8 Plot the Waveform

figure; hold on;

plot(V./max(abs(V)),’Color’,’k’,’LineWidth’,3)
plot(V2./max(abs(V2)),’--’,’Color’,’b’,’LineWidth’,3)
line([0 length(V)],[0 0],’Color’,’k’,’LineWidth’,1)
set(gca,’LineWidth’,2,’FontSize’,14)
set(gca,’YLim’,[-1.1 1.1],’YTick’,[0]);
set(gca,’XLim’,[0 length(V)],’XTick’,[]);
400 C Fourier Analysis

xlabel(’Time’,’FontSize’,14);
ylabel(’Microphone Signal’,’FontSize’,14);
clear strLegend;
strLegend{1}=’Recorded Waveform’;
strLegend{2}=sprintf(’Reconstructed Waveform,
Nmax = %u’, Nmax);
legend(strLegend);
pbaspect([2 1 1])

Recorded Waveform
Microphone Signal

Reconstructed Waveform, Nmax = 20

Time

C.5 Additional Waveforms

Additional musical instrument waveforms are presented below in the same format
as that given above for the garden hose for those who might like to study them
and for use elsewhere in this book. These data were taken in Davies Auditorium
at Yale, using a high-quality Sennheiser MKH104 omnidirectional microphone
with uniform response (±1 dB) over the range from 50 Hz to 20 kHz and the
Hewlett-Packard equipment shown in Fig. 2.9. The 10-bit A-to-D converter used
had a dynamic range of about 60 dB. The instruments were played without vibrato
by professional musicians to whom the author is greatly indebted. The brass
instruments were played by James Undercoffler; the violins, by Syoko Aki; the
oboe, heckelphone, krummhorn and rohr schalmei by James Ryan; and flute and
piccolo by Leone Buyse. I am indebted to the late William Liddell for the use of
his krummhorns, to Richard Rephann for the loan of the historic brass instruments
from the Yale Instrument Collection, and to Robert Sheldon of the Smithsonian
Collection in Washington, DC for playing the serpent. The heckelphone was loaned
by the Yale Concert Band and the rohr schalmei was borrowed from the author’s
personal pipe organ.
C Fourier Analysis 401

C.5.1 Contents

• Mode-Locked Garden Hose (normalized to +/- 1000)

• F-Cornet
• Piccolo Trumpet
• French Horn (Loud)
• French Horn (Soft)
• Ophicleide
• Serpent
• Andreas Amati Violin (G String)
• Ordinary Violin (G String)
• Oboe (Laure)
• Heckelphone
• Krummhorn
• Rohrschalmei (Laukuft)
• Flute
• Piccolo

C.5.2 Mode-Locked Garden Hose (normalized to +/- 1000)

%DATA 109 % P = Number of points in Period = the first

datum gardenhose=[...
3,20,43,68,111,176,...
273,426,634,861,1000,989,...
864,705,563,452,358,273,...
199,136,85,45,17,-14,...
-40,-63,-85,-105,-125,-129,...
-151,-162,-165,-165,-165,-165,...
-170,-165,-170,-170,-170,-173,...
-173,-170,-162,-159,-156,-153,...
-151,-142,-139,-136,-131,-128,...
-128,-128,-125,-119,-114,-108,...
-102,-94,-91,-91,-91,-91,...
-91,-91,-91,-91,-91,-91,...
-85,-82,-82,-80,-74,-74,...
-74,-80,-80,-80,-80,-80,...
-82,-82,-82,-85,-88,-91,...
-91,-91,-91,-91,-91,-85,...
-82,-80,-74,-71,-68,-65,...
-60,-57,-45,-37,-28,-14,...
3];
% Fundamental frequency in Hz
%DATA 307.692
402 C Fourier Analysis

C.5.3 F-Cornet

%DATA 107
fcoronet=[...
-4,44,114,163,212,249,...
280,301,311,319,321,321,...
311,290,259,223,166,111,...
57,5,-47,-83,-106,-122,...
-124,-119,-98,-83,-67,-52,...
-41,-47,-62,-83,-109,-150,...
-192,-231,-259,-285,-301,-303,...
-293,-269,-228,-166,-96,-13,...
78,174,275,370,477,578,...
681,777,870,948,995,1000,...
974,886,764,617,433,244,...
67,-101,-238,-345,-412,-448,...
-461,-448,-415,-383,-337,-301,...
-272,-259,-251,-269,-301,-332,...
-365,-399,-425,-446,-453,-456,...
-446,-425,-399,-383,-368,-355,...
-347,-337,-329,-301,-262,-241,...
-210,-166,-117,-65,-4];
% Cornet Frequency in Hz
%DATA 316.075

C.5.4 Piccolo Trumpet

%DATA 168
piccolotrumpet=[...
-5,97,195,280,367,450,...
530,610,680,755,807,857,...
898,935,967,982,997,1000,...
995,987,965,940,910,875,...
835,797,742,695,637,585,...
527,475,417,357,302,247,...
187,130,85,37,-10,-50,...
-80,-110,-140,-160,-180,-193,...
-203,-200,-200,-193,-170,-155,...
-130,-93,-55,-12,35,77,...
125,177,227,277,325,367,...
408,450,490,517,545,567,...
580,587,590,590,580,560,...
C Fourier Analysis 403

540,515,485,450,408,367,...
327,285,240,190,147,97,...
55,15,-35,-65,-100,-133,...
-155,-180,-200,-215,-230,-233,...
-245,-245,-242,-245,-242,-240,...
-238,-235,-238,-235,-235,-242,...
-253,-260,-265,-282,-300,-320,...
-343,-363,-390,-413,-440,-463,...
-490,-513,-535,-557,-583,-603,...
-623,-643,-660,-680,-690,-715,...
-733,-753,-773,-793,-810,-830,...
-845,-857,-873,-875,-875,-882,...
-873,-870,-857,-835,-810,-785,...
-755,-720,-673,-623,-563,-500,...
-433,-355,-272,-193,-110,-5];
% Piccolo Trumpet Frequency in Hz
%DATA 597.672

C.5.5 French Horn (Loud)

%DATA 151
frenchhornloud=[...
21,112,188,242,273,318,...
356,387,399,394,385,371,...
349,309,264,200,133,76,...
38,21,17,17,26,52,...
67,95,121,147,159,171,...
176,166,157,143,140,138,...
128,124,124,124,133,143,...
147,147,143,140,135,133,...
119,105,93,76,62,26,...
-12,-52,-95,-138,-181,-214,...
-252,-283,-299,-314,-309,-295,...
-268,-249,-240,-233,-214,-195,...
-166,-128,-76,-10,64,121,...
176,216,245,273,285,295,...
292,292,302,314,333,340,...
349,347,347,349,375,428,...
508,575,627,656,672,679,...
684,684,689,684,670,641,...
596,551,508,451,380,304,...
214,128,45,-43,-164,-314,...
-461,-589,-670,-703,-717,-743,...
404 C Fourier Analysis

-791,-846,-895,-931,-960,-988,...
-1000,-971,-924,-886,-855,-838,...
-836,-841,-838,-829,-774,-698,...
-613,-527,-437,-335,-219,-95,...
21];
% French Horn Frequency in Hz
%DATA 222.488

C.5.6 French Horn (Soft)

%DATA 151
frenchhornsoft=[...
4,46,87,121,154,183,...
212,231,262,272,291,297,...
306,297,297,291,283,268,...
264,247,237,229,216,206,...
200,191,179,166,158,141,...
131,116,121,112,106,104,...
100,98,96,83,87,79,...
73,71,62,56,48,40,...
40,42,46,54,58,62,...
64,71,67,73,71,71,...
75,81,96,104,114,133,...
156,166,183,212,225,247,...
272,297,322,349,372,399,...
422,441,462,484,499,516,...
528,541,555,563,563,572,...
570,563,555,545,528,511,...
486,464,424,391,356,308,...
262,218,164,116,56,-10,...
-71,-135,-210,-279,-345,-407,...
-482,-541,-607,-661,-719,-773,...
-825,-867,-911,-940,-965,-979,...
-996,-1000,-994,-983,-973,-950,...
-925,-892,-857,-811,-767,-717,...
-659,-603,-536,-468,-403,-343,...
-285,-233,-175,-127,-71,-25,...
4];
% Soft Fr. Horn Frequency in Hz
%DATA 223.294
C Fourier Analysis 405

C.5.7 Ophicleide

%DATA 196
ophicleide=[...
-4,-117,-186,-235,-283,-324,...
-348,-393,-421,-421,-377,-324,...
-247,-170,-73,40,126,170,...
158,142,113,45,-32,-81,...
-138,-202,-259,-287,-312,-340,...
-360,-364,-364,-377,-364,-348,...
-316,-308,-287,-275,-291,-324,...
-348,-377,-393,-413,-405,-389,...
-364,-340,-328,-324,-324,-312,...
-291,-259,-259,-259,-267,-287,...
-291,-283,-279,-247,-170,-73,...
49,158,259,372,462,526,...
579,607,611,607,575,518,...
462,417,401,385,360,332,...
308,300,267,235,178,121,...
77,32,24,32,61,81,...
89,93,109,97,97,105,...
134,194,251,255,186,77,...
-24,-85,-101,-97,-85,-101,...
-101,-40,40,105,142,142,...
154,186,223,271,316,381,...
421,421,405,364,328,304,...
316,364,421,478,518,514,...
486,413,304,142,-89,-348,...
-632,-911,-1219,-1510,-1741,...
-1915,-1960,-1911,-1781,-1603,-1381,-1154,...
-927,-700,-462,-219,28,235,...
429,595,749,879,955,996,...
1000,1000,972,964,935,866,...
777,688,599,514,437,364,...
304,255,211,170,158,142,...
126,121,130,126,130,142,...
142,126,89,-4];
% Ophicleide Frequency in Hz
%DATA 85.103
406 C Fourier Analysis

C.5.8 Serpent

%DATA 203
serpent=[...
-5,53,113,167,207,237,...
233,223,207,170,140,90,...
47,-7,-47,-87,-107,-113,...
-107,-80,-43,10,73,133,...
187,223,250,263,263,250,...
223,200,157,130,73,20,...
-40,-97,-147,-200,-240,-267,...
-287,-280,-267,-240,-203,-180,...
-163,-133,-123,-127,-120,-123,...
-127,-123,-130,-133,-127,-133,...
-133,-113,-107,-73,-30,37,...
103,160,210,237,240,250,...
237,207,170,127,90,73,...
60,50,40,23,20,-13,...
-47,-80,-120,-150,-187,-193,...
-193,-173,-150,-133,-107,-80,...
-57,-50,-43,0,60,143,...
237,323,440,533,620,687,...
740,747,640,460,253,20,...
-353,-737,-1053,-1333,-1497,...
-1527,-1440,-1273,-993,-667,...
-327,-20,210,340,340,260,...
160,80,-7,-20,50,147,277,...
420,553,667,773,853,943,...
997, 1000,970,927,810,...
673,500,273,40,-193,...
-387,-507,-560,...
-540,-447,-337,-217,-93,23,...
83,130,147,143,140,140,...
147,153,167,173,153,117,...
77,23,-40,-60,-100,-133,...
-173,-203,-240,-267,-280,-280,...
-300,-287,-273,-257,-257,-230,...
-227,-213,-207,-213,-200,-193,...
-163,-150,-110,-70,-5];
% Serpent Frequency in Hz
%DATA 61.702
C Fourier Analysis 407

C.5.9 Andreas Amati Violin (G String)

%DATA 171
amativioling=[...
14,-153,-330,-507,-641,-732,-813,-900,-967,-1000,...
-967,-880,-737,-603,-498,-450,-426,-431,-440,-426,...
-402,-354,-306,-330,-335,-306,-273,-177,53,211,364,...
474,545,593,608,641,641,603,...
574,493,431,416,431,...
512,603,651,651,584,459,316,...
187,86,0,-100,-177,...
-234,-273,-239,-177,-100,...
33,134,201,220,249,278,...
335,392,469,507,469,383,297,...
191,124,72,91,124,...
148,225,354,550,775,895,...
880,689,397,144,-10,29,...
144,239,311,278,206,163,...
234,335,383,440,392,...
258,33,-62,-24,57,115,...
96,29,-124,-234,-249,...
-182,-38,57,110,115,124,...
144,230,340,440,474,...
402,301,196,67,-62,-201,...
-344,-445,-464,-388,...
-273,-144,-38,129,244,392,...
507,531,512,435,...
378,354,344,378,435,536,...
646,737,766,699,627,...
545,512,478,474,459,402,...
378,354,335,340,344,...
340,301,249,167,77,14];
% Amati Violin Frequency in Hz
%DATA 196.71

C.5.10 Ordinary Violin (G String)

%DATA 130
violin=[...
76,389,686,854,945,980,...
872,600,290,27,...
-75,7,168,325,456,566,...
408 C Fourier Analysis

662,748,796,761,...
644,491,336,226,137,111,...
157,254,332,403,...
496,606,690,690,597,440,...
254,77,-102,-215,...
-296,-354,-407,-460,-513,-566,...
-637,-746,-878,-976,...
-1000,-912,-743,-584,-454,...
-341,-186,49,270,440,...
531,518,458,389,350,305,...
237,166,115,53,...
-18,-62,-82,-111,-146,-146,...
-122,-58,-2,31,...
53,53,27,0,7,49,...
124,212,281,369,...
467,549,566,451,283,53,...
-104,-170,-146,-86,...
-53,-33,31,119,181,243,...
310,296,184,-31,...
-283,-442,-515,-546,-540,...
-558,-580,-617,-642,-666,...
-639,-577,-584,-692,-808,-878,...
-816,-569,-199,76];
% Violin Frequency in Hz
%DATA 194.41

C.5.11 Oboe (Laure)

%DATA 194
oboelaure=[...
2,-181,-397,-581,-717,-809,...
-839,-829,-777,-697,-596,-486,...
-370,-261,-159,-79,-10,45,...
82,107,127,149,161,174,...
179,179,174,169,156,149,...
134,124,104,84,55,15,...
-35,-79,-112,-144,-159,-164,...
-161,-159,-154,-151,-159,-161,...
-164,-169,-159,-144,-114,-84,...
-50,-10,37,84,127,179,...
218,246,261,253,223,176,...
127,69,10,-40,-77,-122,...
-159,-189,-223,-251,-270,-283,...
C Fourier Analysis 409

-288,-280,-273,-263,-258,-258,...
-258,-258,-258,-258,-251,-243,...
-241,-231,-218,-208,-191,-169,...
-154,-139,-124,-122,-122,-129,...
-141,-154,-159,-159,-156,-151,...
-139,-129,-122,-124,-129,-134,...
-151,-169,-179,-199,-208,-231,...
-243,-253,-258,-258,-251,-238,...
-213,-191,-161,-139,-104,-74,...
-40,-5,27,55,77,89,...
97,97,94,87,74,60,...
37,15,-20,-60,-104,-156,...
-213,-280,-347,-409,-474,-526,...
-566,-588,-586,-558,-499,-422,...
-330,-221,-109,15,136,256,...
377,501,620,739,831,913,...
960,993,1000,995,983,963,...
938,913,883,856,821,789,...
752,715,663,600,496,355,...
176,2];
% Oboe Frequency in Hz
%DATA 259.581

C.5.12 Heckelphone

%DATA 130 [JLR probably a typo, should be 138]

%[JLR] Note: There are two typos in the manuscript
which are fixed here heckelphone=[...
-8,142,250,308,300,283,...
262,196,58,33,42,42,...
29,-8,-54,-104,...
-225,-321,-417,-475,-488,...
-417,-242,0,...
217,225,346,608,867,1000,...
942,642,25,-50,-108,-267,...
-475,-650,-667,-471,-217,-204,...
-167,-25,171,317,417,425,...
329,308,229,92,-50,-142,...
-208,-267, -358,-425,-458,-454,...
-425,-388,-304,-221,13,142,...
262,346,392,425,442,425,...
238,158,75,-33,-125,-242,...
-367,-450,-388,-317,-233,...
410 C Fourier Analysis

-142,-75,13,112,192,225,...
258,292,292,250,175,46,-104,...
-254,-292,-321,-317,-292,-250,... %[JLR] last point
changed to -250
-188,-121,50,196,396,612,...
746,725,608,479,262,167,...
50,-117,-283,-400,-417,-350,...
-67,108,258,367,442,475,...
500,458,-17,-321,-600,-817,...
-892,-808,-671,-538,-192,-8];
% Heckelphone Frequency in Hz
%DATA 129.777

C.5.13 Krummhorn

%DATA 174
krummhorn=[...
-59,-91,-157,-216,-100,-78,...
-162,-176,333,1000,623,-130,...
-412,-355,-100,422,853,850,...
490,-100,-623,-760,-760,-603,...
-150,255,157,-169,-277,-196,...
-250,-309,-150,113,206,279,...
377,331,108,-74,-51,15,...
145,353,510,507,333,164,...
93,152,243,341,431,439,...
419,373,257,230,186,59,...
64,135,201,206,86,-78,...
-194,-196,-142,-83,47,103,...
39,-32,-91,-167,-199,-120,...
0,34,27,54,83,118,...
125,137,196,230,174,115,...
83,20,-120,-316,-471,-412,...
-277,-81,152,211,47,-137,...
-267,-353,-319,-176,-69,-25,...
10,47,93,100,132,142,...
142,162,157,105,27,-44,...
-78,-110,-118,-110,-78,-44,...
-32,-51,-74,-98,-103,-137,...
-137,-110,-78,-88,-113,-123,...
-78,-34,-5,-20,-88,-196,...
-353,-451,-453,-512,-534,-578,...
-534,-446,-397,-324,-306,-358,...
C Fourier Analysis 411

-306,-245,-140,-61,69,96,...
162,152,142,172,213,289,...
294,294,309,225,145,-59];
% Krummhorn Frequency in Hz
%DATA 192.52

C.5.14 Rohrschalmei (Laukuft)

%DATA 129
rohrschalmei=[...
4,-94,-132,-186,-172,-224,...
-256,-202,-146,-116,-68,-20,...
-36,-64,-64,-104,-168,-136,...
-98,-16,82,126,212,158,...
158,152,20,-80,-188,-282,...
-206,720,692,444,1000,648,...
694,230,68,-228,-938,-826,...
-726,-830,-474,-168,126,322,...
376,562,408,280,316,206,...
40,16,-44,-96,-178,-168,...
-170,-194,-242,-402,-568,-488,...
-584,-754,-604,-410,-370,-32,...
144,476,542,678,848,816,...
638,272,146,-176,-320,-420,...
-482,-450,-400,-202,-43,-132,...
-40,56,-16,0,-36,88,...
172,246,414,404,366,430,...
364,276,88,-48,-124,-282,...
-272, -238,-314,-306,-282,-306,...
-266,-136,-124,12,52,128,...
212,166,262,220,192,164,...
152,78,4];
% Rohrschalmei Frequency in Hz
%DATA 259.674

C.5.15 Flute

%DATA 192
flute=[...
18,-54,-75,-101,-134,-165,...
-178,-247,-281,-291,-335,-376,...
412 C Fourier Analysis

-428,-425,-464,-438,-446,-425,...
-402,-412,-330,-299,-250,-160,...
-82,21,93,180,216,317,...
325,358,320,289,255,222,...
144,119,111,85,5,26,...
5,-34,-52,-57,-106,-67,...
-64,-75,-34,-34,8,-5,...
36,59,101,111,90,119,...
119,98,39,57,15,-26,...
3,15,0,8,8,-41,...
-62,-77,-106,-126,-183,-232,...
-260,-240,-253,-247,-191,-157,...
-103,-67,-26,28,108,131,...
250,340,430,503,595,634,...
660,698,675,619,526,448,...
369,340,206,124,88,-26,...
-82,-178,-250,-291,-302,-371,...
-407,-410,-436,-464,-521,-559,...
-582,-652,-765,-881,-920,-972,...
-912,-910,-838,-678,-577,-479,...
-353,-247,-134,-21,62,121,...
155,216,204,193,222,201,...
204,216,237,291,376,428,...
585,799,840,892,959,1000,...
907,747,588,430,222,-46,...
-289,-376,-518,-598,-655,-655,...
-621,-526,-459,-327,-188,-26,...
90,170,268,361,454,454,...
443,485,495,423,407,348,...
291,227,144,77,62,18];
% Flute Frequency in Hz
%DATA 262.05

C.5.16 Piccolo

%DATA 170
piccolo=[...
-11,34,101,100,218,276,...
333,391,451,508,556,590,...
631,643,679,695,706,707,...
706,695,667,652,631,604,...
590,542,525,487,460,422,...
388,362,341,326,307,295,...
C Fourier Analysis 413

276,269,247,240,247,237,...
225,218,192,189,168,151,...
134,106,91,74,53,36,...
36,14,-10,-29,-34,-62,...
-77,-86.-115,-115,-125,-129,...
-139,-146,-149,-146,-153,-146,...
-146,-137,-137,-127,-127,-118,...
-108,-89,-72,-55,-38,-10,...
24,48,94,125,170,206,...
247,285,317,353,388,408,...
436,458,460,484,494,477,...
482,468,458,451,429,412,...
388,369,345,331,321,293,...
266,249,216,199,173,161,...
115,86,46,22,-14,-70,...
-120,-182,-245,-317,-386,-480,...
-556,-616,-686,-751,-815,-859,...
-897,-935,-959,-981,-993,-990,...
-990,-1000,-993,-993,-964,-954,...
-959,-930,-914,-887,-887,-856,...
-811,-763,-731,-671,-631,-568,...
-499,-444,-367,-317,-261,-175,...
-118, -62,-11];
% Piccolo Frequency in Hz
%DATA 593.877

C.6 Harmonic Analysis

This part of the program takes most of the running time. A(N) and B(N) correspond
to the Fourier amplitudes given by the integrals in Eqs. (C.6) and (C.7). They are
done here numerically by a method equivalent to drawing straight lines between the
successive points and adding up the areas (the “trapezoidal Rule”).

function [A0 A B0 B C0 C PHI]

=fouriercoef_trapezoid(V,Nmax)

% The net Fourier amplitude C(N) and phase PHI(N)

are computed using
% Eqs (C.6), (C.7), and (C.8).

% Check for input errors

error(nargchk(1, 2, nargin));
if (nargin < 2); Nmax = 20; end;
414 C Fourier Analysis

P=length(V);

A0=2*pi/(P-1); % A0 is a constant used in the loop on N.

for N=1:Nmax % FOR N=0 TO Nmax

% N=0 corresponds to the constant or DC term,
we will calculate it later

A(N)=0;
B(N)=0;

for I=1:P % FOR I=1 TO P

A(N)=A(N)+V(I)*sin(N*A0*(I-1));
B(N)=B(N)+V(I)*cos(N*A0*(I-1));
end; % NEXT I

A(N)=A(N)*2/(P-1);
B(N)=B(N)*2/(P-1);
C(N)=sqrt(A(N)*A(N)+B(N)*B(N));

%IF N>0 THEN P(N)=ATN(B(N)/A(N)) % avoid DC term

PHI(N)=atan(B(N)/A(N));
%IF A(N)<0 THEN P(N)=P(N)+Pi ’arctangent problem
if (A(N) < 0); PHI(N)=PHI(N)+pi; end;

end; %NEXT N

% DC or constant term (A0 calculation above)

B0=sum(V)*2/(P-1);
C0=B0/2;

Published with MATLAB7.12

Appendix D
The Well-Tempered Scale

The numbers here were computed to 0.06 ppm and rounded off to the nearest
0.01 Hz assuming the 1936 international convention that A4 = 440.0000 Hz, using
21/12 = 1.05946310. Note as a check: 1.0594631012 = 2.00000012

Frequency in Hz
A 27.50 55.00 110.00 220.00 440.00 880.00 1760.00 3520.00 7040.00 14080.00
A 29.14 58.27 116.54 233.08 466.16 932.33 1864.66 3729.31 7458.62 14917.25
B 30.87 61.74 123.47 246.94 493.88 987.77 1975.53 3951.07 7902.13 15804.27
C 32.70 65.41 130.81 261.63 523.25 1046.50 2093.00 4186.01 8372.02 16744.04
C 34.65 69.30 138.59 277.18 554.37 1108.73 2217.46 4434.92 8869.85 17739.69
D 36.71 73.42 146.83 293.66 587.33 1174.66 2349.32 4698.64 9397.28 18794.55
D 38.89 77.78 155.56 311.13 622.25 1244.51 2489.02 4978.03 9956.07 19912.13
E 41.20 82.41 164.81 329.63 659.26 1318.51 2637.02 5274.04 10548.08 21096.17
F 43.65 87.31 174.61 349.23 698.46 1396.91 2793.83 5587.65 11175.31 22350.61
F 46.25 92.50 185.00 369.99 739.99 1479.98 2959.96 5919.91 11839.83 23679.65
G 49.00 98.00 196.00 392.00 783.99 1567.98 3135.96 6271.93 12543.86 25087.72
G 51.91 103.83 207.65 415.30 830.61 1661.22 3322.44 6644.88 13289.75 26579.51

© Springer Nature Switzerland AG 2018 415

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
Solutions

Problems of Chap. 1

1.1 The oil would spread out very rapidly over the water, reducing its surface tension
and thereby preventing wave motion at the short wavelengths characterized by the
ripples.
1.2 (a) 3.11 m. (b) 615 THz.
1.3 Answers: The frequency is 233.08/(1.441 × 1017 ) = 1.62 × 10−15 Hz. The
period is 6.183 × 1014 s = 19.59 million years.
1.4 (a) 633 nm. (b) 43.6 ft.
1.5 (a) 3.8. (b) 2.7
1.6
From the original length of the string, one expects frequencies at nc/2L where n
= 1,2,3,. . . . But because of reflections from the kink you also get odd harmonics of
3c/8L and of 3c/4L. Note that the string is free to move up and down at the kink,
so that the frequencies on either side of the kink are determined by only one “hard”
phase shift per round trip, and the resonances have the same form as those for a
closed pipe.
1.7 (a) Multiples of 110 Hz. (Neglecting phase shift and time delays in the amplifier,
the running wave phase shift would be 2πfL /c per trip around the loop.) (b) It would
oscillate at some other frequency if there were significant phase shift in the amplifier
(e.g., from tone controls or a large gain variation with frequency.)
1.8 Answer: The pipe is closed at the top, initially by the flap valve, and then by
the water flowing down the pipe. The fundamental frequency of the pipe starts out
at c/4L = 46 Hz, with odd harmonics at 138, 230,. . . Hz. As the water goes down
the pipe, the acoustic length shortens and the pitch for each harmonic goes up.

© Springer Nature Switzerland AG 2018 417

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
418 Solutions

1.9 The main frequencies are 523 Hz, 622 Hz, 740 Hz, 932 Hz, 1570 Hz, 1865 Hz,
2217 Hz, and 2797 Hz. Starting at C above A=440 Hz, the notes are C, E, G, B,
G, B, D, and F. (See figure for musical notation.) The first two intervals are minor
thirds, giving the whistle its characteristic, mournful sound. The bottom half might
be regarded as a half-diminished seventh. The second half is just the first chord
raised by a twelfth. (The lower chord would resolve on a diminished seventh and
then on a B minor chord) (Fig. S.1).
1.10 Answers: Hard phase shifts occur at both the ceiling and the floor, hence
one might expect the main resonances to be given approximately by fn = nc/2L.
Resonances do occur at those frequencies (multiples of 55 Hz.) But because the floor
is in the focal plane of the concave ceiling, resonances also occur at fn = nc/8L.
(The running wave from a point source on the floor makes eight transits to the ceiling
and back before it closes on itself.) Assuming the radius of curvature of the ceiling is
20 ft and twice the ceiling height, the main resonances would occur every 13.75 Hz
throughout the audio band. (Speech is totally unintelligible, but the mumbling sound
is impressive.)
1.11 One reason is that the open end of a closed pipe radiates as a monopole source
(equally in all directions), whereas an open pipe acts like a dipole with the maximum
radiated intensity going in the vertical direction (unless the pipe is turned 90◦ ).
1.12 By listening to the full Doppler shift on the bells at grade crossings, he or she
could determine the median pitch and the total fractional change in frequency on the
Well Tempered Scale. From that, one could calculate the speed.
1.13 Answers: (a) 38.2 mph. (b) About one whole step.
1.14 The Doppler effect from sound reflected by the rotating blades creates a
warbling effect.
1.15 (a) 676 Hz. (b) 786 Hz. (c) 738 Hz.
1.16 +715 Hz. (Note that the velocity of the image of the radar gun is twice the
speed of the car.)
1.17 Answer: c/4L = 1040/(4 × 6.9) = 37.7 Hz.
1.18 Answer: Since f = 65.4 Hz, λ = c/f = 1087/65.4 = 16.6 ft.

Fig. S.1 The steam engine

whistle
Solutions 419

Problems of Chap. 2

2.1 ΔF ≈ 2π Δt ≈ 0.032 Hz.

2.2 (a) To go through 50 Hz would require at least
Δt ≈ 2π1ΔF = 2π150 = 318 ms.
Therefore, to go through 10,000 Hz, it would take a minimum of about 64 s, or
(b) a scanning rate of 156 Hz/s.
2.3 See Eqs. (2.7) and (2.8)
2.4 10 dB
2.5 20Log10 (80 × 5280/10) + 60 = 152 dB
2.6 Serber was about 15.4 miles away. Los Alamos was about 250 miles from the
test site. The blast would have been about 24 dB louder at Serber’s position.
2.7 20Log10 [3000/1] + 60 ≈ 129 dB (or about as loud as a rock band in a typical
concert setting).
2.8 About 60 dB (≈ 20Log10 1024.)
2.9 About 96 dB (≈ 20Log10 216 = 320Log102). But as shown by Bennett (1948),
one gets an extra 3 dB by averaging the quantization errors in most systems.
2.10 Answer: The sound intensity per unit area would be attenuated by a factor
of 36, which corresponds to a decrease of 16 dB. Hence, the sound level at the
neighbor’s house should be about 64 dB (something between normal conversation
and shouting at 4 ft).
2.11 Answer: (a) 20Log10 (4) ≈ 12 dB. (b) 10Log10 (4) ≈ 6 dB.
2.12

Frequency (Hz) 31.5 63 125 250 500 1000 2000

Signal (dB) 65 78 88 80 70 67 50
Amplitudes (÷1000) ≈ 1.8 7.9 25 10 3.2 2.2 0.3

2.13

Frequency (Hz) 31.5 63 125 250 500 1000 2000 4000 8000
Signal (dB) 50 52.5 52.5 55 60 67.5 60 57.5 52.5
Amplitudes (÷100) ≈ 3.2 4.2 4.2 5.6 10 23 10 7.5 4.2
420 Solutions

2.14

n= 1 3 5 7 9 11
dB 0 −9.5 −14 −17 −19 −21

2.15 A closed organ pipe.

2.16 The spectral components would have side bands at ±6 Hz.

Problems of Chap. 3

3.1 If only one coil located at x0 were used, nulls in the spectrum would occur
at nπLx0 = mπ where m = 1, 2, 3, . . . hence, at the harmonics n = L/x0 ≈
9, 18, 27, . . .
3.2 If the output voltages from the two coils in Fig. 3.10 were added, the result
would be 
V (t)∝ ∞ n=1 An {sin[nπ(x
 nπ x0  0 +nπa)/L]
 + sin[nπ(x0 − a)/L)]} fn (t)
= 2 ∞ n=1 An sin L cos a
L fn (t) and the spectrum would have minima
at n = m xL0 and n = (2m − 1) 2a L
where m = 1, 2, 3 . . . which for the dimensions
given in the text would occur at n ≈ 9, 18, 26, 27, 36, 43, etc.
3.3 The separation between the negative and positive peaks for the 56.8 Hz
resonance should correspond to half the wavelength of the surface wave along the
grain direction. Hence, for the long dimension assumed above, λ/2 ≈ 28.1 inches
and the surface wave velocity would be about
cSurface = λf ≈ (56.2)(56.8) = 3192 in./s ≈ 266 ft/s, or about 1/4 of that for
the velocity of sound in air.
3.4 The frequency is F = 261.6 Hz. The tension = 6.64 106 dynes = 14.9 lbs.
3.5 We want the density per unit length to be the same. Hence, the cross-sectional
area of the wire should be 1.4/7.83 = 0.1788 times that of the gut string, or the
diameter of the steel wire should be 0.423 × 0.029 in. = 0.012 in.

Problems of Chap. 4

4.1 Answer: About 9.4 times for the Stein prellmechanik action, 9.2 times for the
Broadwood action, and 9.5 times for the Streicher action.
Solutions 421

4.2 Answer: Remove all the dampers for notes on that chord before dropping it.
(Don’t try that with a good piano!!! If the cast iron frame breaks, it is virtually
impossible to repair it.)
4.3 Answer: The highest loss (symmetric) mode occurs when all three strings are
in phase. Intermediate loss modes occur when one string is 180◦ out of phase with
the other two. The lowest loss (odd-symmetric) mode occurs when the outer strings
are 180◦ out of phase and the middle string is not vibrating.
4.4 Answer. The strength of the wire increases with its cross-sectional area, but
so does its density per unit length. Hence, the ratio of the breaking tension to the
density per unit length is about constant and that ratio determines the wave velocity,
hence the pitch, at the breaking point.
4.5 The wavelength is λ ≈ 2 × 72 = 144 in.; hence, the wave velocity c = λf =
144 × 87.3 = 12,571 in./s ≈ 1048 ft/s.
4.6 The tension is about 9 × 107 dynes ≈ 202 pounds of force.
4.7 About 29 lbs, or about 6 lbs more than to bring the string up to normal pitch.
(It takes about 1.26 × 202 ≈ 255 lbs of force to break the string. The mechanical
advantage of the tuning hammer is 10/0.1125 ≈ 8.9)
4.8 About 388 ft/s. (The maximum and minimum are separated by half a wave-
length.)
4.9 For F2, about 3.001; for C9, about 6.8.

Problems of Chap. 5

5.1 Answer: E= 3 × 440/2 = 660 Hz, D= 2 × 440/3 = 293.33 Hz, G = 2 ×

293.33/3 = 195.55 Hz. On the WTS, E = 659.25 Hz, D = 293.66 Hz, and G =
196.00 Hz.
5.2 For the A string, c = 2 × 33 × 440 = 29,040 cm/s; for the G string, c =
12,906 cm/s.
5.3 μ = T /c2 = 2.838 × 10−5 g/cm3 .
5.4 195.55 Hz.
5.5 391.1 − 330 = 61.1 Hz, or B two octaves below middle C.
5.6 The difference frequency is an octave below the D, so the “Tartini Tone” doesn’t
stand out the way it would for a double third or fourth.
5.7 1490/8 = 186 strokes per second.
422 Solutions

Problems of Chap. 6

6.1 c/4L ≈ 43 Hz.

6.2 Open pipe resonances would occur at about 92 Hz, 184 Hz, 276 Hz, and so on.
√
6.3 (a) The voice frequency is proportional to 1/ M, where M is the effective
molecular weight of the gas in the lungs. Therefore our SCUBA diver wants M to
be about 29 × (120/77.8)2 ≈ 69.
(b) Let X = the fraction of oxygen in the tank and Y = the fraction of krypton
in the tank. Then X + Y = 1 32X + 84Y = 69
Solving these two equations yields X ≈ 0.288 for the fraction of oxygen, and
Y ≈ 0.712 for the fraction of krypton.
6.4 No. He’ll only be able to reach 2725 Hz (about halfway between E and F.)

Problems of Chap. 7

7.1 Answer: The total pressure from a 2-inch difference in the height of water per
unit area in a U-tube manometer is 5.08 g/cm2 . The air-pressure regulator area is
6 ft2 = 864 in.2 = 5574 cm2 . Hence the total weight needed is about 28.3 kg ≈ 62 lbs.
7.2 As a first approximation, the air velocity producing the edge tone in Fig. 7.13 is
simply proportional to the air pressure. Hence the cut-up (L in the figure) should be
reduced by 2/5 = 0.4 on each pipe.
7.3 Up to some point, the pressure in the toe of the pipe would increase
proportionally to the area of the toe-hole after the pipe has been turned on.
7.4 From Appendix D, the fundamental pitch should be 261.6 Hz. Since the Rohr
Schalmei is an open pipe, the overall length should be L = 1100/(2261.6) = 2.10 ft
= 25.2 in. Since the large cavity is tuned to the second harmonic of the pipe, both it
and the copper tube should be 25.2/2 = 12.6 in. long.
7.5 Answer: The orchestral oboe is made from a narrow-scale conical piece of
wood. (Of course, the reed is placed at the vertex where the pressure is a maximum.)
7.6 Use two pipes for each note, chosen to produce successive harmonics of a 16-ft
open pipe. The ear will then interpret the combination as having 16-ft pitch. An 8-ft
closed pipe would provide the fundamental pitch and its odd harmonics. Adding an
open 8-ft pipe would provide the needed even harmonics. If the two are on the same
note channel, they will also tend to lock in phase.
7.7 The total length of the large diameter semi-closed pipe should be L =
1100/(4440) = 0.625 ft = 7.5 in. The length of the short open pipe at the end
should be tuned to the fifth harmonic of 440 Hz = 2200 Hz. Hence, the short length
should be about 1100/(2 × 2200) = 0.25 ft = 3 in.
Solutions 423

7.8 From Eq. (7.3),

5.80 = 12k or k = 0.346/12 = 0.0289 We want N such that loge (2) =
8.20

0.0289N or N = 24 steps.
7.9 Answer: The 32-ft diapason is an open pipe with harmonics at 32-, 16-, and
8-ft pitch (the second, fourth, sixth, and eighth harmonic of the non-existent 64-ft
pipe). The bordon is a closed pipe with a fundamental that is a fifth higher (i.e., third
harmonic) than its nominal 64 foot pitch. Hence, the combination has frequencies
at the second, third, fourth, sixth, and eighth harmonic of the phantom 64-ft pipe
and the human ear will conclude that it actually has 64-ft pitch, even though the
fundamental is completely missing. Assuming the velocity of sound is 1100 ft/sec,
the frequencies would be:
64-ft open 8.59, 17.18, 25.7, 34.4, 42.9, 51.5, 60.1, 68.7, 77.3 Hz 32-ft
Diapason—– 17.18, —–, 34.4, —–, 51.5, —-, 68.7, —- Hz 32-ft Bordon Quint
—–, 25.7, —-, —-, —-, —-, —-, 77.3 Hz Effective harmonic 2 3 4 6 8 9
References

Abele H (1905) The violin and its story. John Leng & Co, New York
Allen EL, Holien H (1973) A laminagraphic study of pulse (vocal fry) register phonationİ. Folia
Phoniatr 25:241–250
Apel W (1972) Harvard dictionary of music. Harvard University Press, Cambridge
Appelman DR (1967) The science of vocal pedagogy. Indiana University Press, Bloomington
Arms WT (1959) History of leyden Massachusetts. The Enterprise and Journal, Orange; Colrain,
MA Library, call number 9744 A73
Arnold GE (1973) Physiology of speech. The new encyclopedia Britannica, vol 17. Encyclopedia
Britannica, London, pp 477–492
Atkins H, Newman A (1978) Beecham stories. Futura Publications, London
Attenborough D (1998) The life of birds. BBC Video Co, London
Atwood W (1997) A bridge collection examined. Catgut Acoust Soc J 3(4):38–43
Audsley GA (1905) The art of organ building. Dodd, Mead, and Co, New York; Reprinted by
Dover, New York in 1965 in two volumes. Emphasizes the history of organ building, vol 1 and
Emphasizes the constructional details, vol 2
Backus J (1969) The acoustical foundations of music. WW Norton, New York
Baken RJ (1998) An overview of laryngeal functions for voice production. In: Sataloff RT (ed)
Vocal health and pedagogy. Singular Publishing Group, San Diego, pp 27–45
Ballot B (1845) Akustische Versuche auf der Niederländischen Eisenbahn, nebst gelentlichen
Bemerkungen zur Theorie des Hrn. Prof. Dopperİ. Poggendorff’s Ann (also known as Ann
Phys Chem, Leipzig) 66(11):321–351
Bartoli C (2002) Interview on “All things considered”. National Public Radio, Washington, D.C.;
Wednesday, September 25
Beare C (1980a) Dominique Peccatte. In: Sadie S (ed) The new groves dictionary of music and
musicians, vol 14. MacMillan, London, p 323
Beare, C (1980b) Eugène Sartory.İ In: Sadie S (ed) The new groves dictionary of music and
musicians, vol 16. MacMillan, London, p 512
Benade AH (1975) The Wolf tone on violin family instruments. Catgut Acoust Soc Newslett
24:21–23; Reprinted in Hutchins and Benade (1997) 1:117–120. This material was also
reproduced in Benade, 1976, pp 567–575
Benade, AH (1976) Fundamentals of musical acoustics. Oxford University Press, New York
Bennett WR (1948) Spectra of quantized signals. Bell Syst Tech J 27:446–472
Bennett WR (1983) Secret telephony as a historical example of spread-spectrum communication.
IEEE Trans Commun COM-31:98–104

© Springer Nature Switzerland AG 2018 425

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
426 References

Bennett WR Jr (1962) Gaseous optical masers. In: Heavens OS Applied optics supplement no.1 on
optical masers. Optical Society of America, Washington, D.C.
Bennett WR Jr (1976) Scientific and engineering problem-solving with the computer. Prentice
Hall, Englewood Cliffs
Bennett WR Jr (1977) The physics of gas lasers. Gordon and Breach, New York
Bennett WR Jr (1990) Heart sounds and murmurs. Lecture with digitally recorded illustrations in
compact disc format. MCG International Inc., New Haven; CD No. DIDX 009454 (December)
Bennett WR Jr, Bennett J (1990) Dynamic spectral phonocardiograph. U.S. Patent No. 4,967,760
(November 6)
Berlyn GP, Richardson AD (2001) Wood: its properties in relation to its use in turning. In: Wood
turning in North America since 1930. Amilcare Pizzi, Milan, pp 152–160
Blot E (2001) Liuteria Italiana IV (1800–1950), Eric Bloc editioni, Cremona
Boalch D (1974) Makers of harpsichords and clavichords. Oxford University Press, London
Bouase H (1929) Tuyaux et Resonators - Introduction a l’Etude des Instruments da Vent. Libraire
Delagrave, Paris
Bouasse H, Fouché M (1929) Instruments à vent, par H. Bouasse avec la collaboration expérimen-
tale de M. Fouché. Delagrave, Paris
Boyden DD, Schwarz B (1980) Sadie S (ed) Violin in the new groves dictionary of music and
musicians, vol 20. MacMillan, London, pp 819–855
Braunwald E (1984) Editor of heart disease. W.B. Saunders Company, Philadelphia
Brigham EO, Murrow RE (1967) The fast Fourier transform. IEEE Spectr 4:63–70
Brookes K (2002) Craft masters, Tempo (Oct/Nov/Dec), vol 6. Temple University Radio,
Philadelphia, pp 4–11
Broos B (1995) Un celebre Peijntre nommé Verme[e]rİ. In: Johannes V National Gallery of Art,
Washington, DC, pp 47–65
Browne, MW (1996) Computer recreates call of dinosaur sound organ. The New York times, New
York, p C7; Tuesday, March 12
Bucur V (1988) Wood structural, anistropy estimated by acoustic invariants.İ IAWA Bull 9(1):67–
74; Also, see reproduction in Hutchins and Benade (1997) 2:795–802
Bucur V (2006) Acoustics of wood. Springer, Berlin
Campbell GA (1911) Cisoidal oscillations. Trans Am Inst Electr Eng 30:873–909
Campbell GA, Foster RM (1920) Maximum output networks for telephone substations and repeater
circuits. Trans Am Inst Electr Eng 39:231–280
Cazden J (1993) Vocal acoustics in electronic musician (February) pp 97–100
Cencié ME (1990) Boy soprano, in Schubert, Der Hirt auf dem Felsen, with Roman Ortner (piano),
and Erwin Monschein (clarinet), in Wiener Sängerknabe Philips CD Recording No. 426307-2;
Band 7
Chladni EFF (1809) Traite d’Acoustique. Courcier, Paris
Conan-Doyle SA (1981) The adventure of the cardboard box. In: The original illustrated Sherlock
Holmes; Reprinted from The Strand Magazine of 1892 to 1893. Castle Books, Seacaucus, p
207
Cooke P (1980) Pygmy music. In: Sadie S (ed) The new groves dictionary of music and musicians,
92 vol 15. MacMillan, London, pp 482–483 and “Yodel”, In: Sadie S (ed) The new groves
dictionary of music and musicians, 92, vol 20, MacMillan, London, p 574
Cooley JW, Tukey JW (1965) An algorithm for the machine calculation of complex Fourier series,
vol 19. Math Comput, pp 297–301
Crandall IB (1926) Theory of vibrating systems. Van Nostrand, New York
Cremer L (1981) Physik der Geige. S. Hirzel Verlag, Stuttgart; Translated into English by John S.
Allen as The physics of the violin. MIT Press, Cambridge
Cuesta C, Valette C (1988) Evolution temporelle de la Vibration des Cordes de Clavecin. Acustica
66:37–45
Davis PJ, Fletcher NH (eds) (1996) Vocal fold physiology. Singular Publishing Group, San Diego
Dick Heinrich Company (2003–2005) Quality products for musical instruments. Dick GMBH,
Germany
References 427

Diegert CF, Williamson TE (1998) A digital acoustic model of the lambeosaurine hadrosaur
Parasaurolophus tubicen. J Vertebr Paleontol 18:38A
Dolge A (1911) Pianos and their makers: a comprehensive history of the development of the piano,
new edn, 1972. Dover, New York
Dom Bédos de Celles F (1766–1768) L’art du facteur d’orgues, vol I text, vol II plates. Delatour,
Paris
Donnington R (1980) Vibrato. In: Sadie S (ed) The new groves dictionary of music and musicians,
vol 19. MacMillan, London, pp 697–698
Drower MS (1974) Thebes (Egypt).İ In: The new encyclopedia Britannica, vol 18. Encyclopedia
Britannica, London, p 264
Dudley H (1939) Remaking speech. J Acoust Soc Am 11:169–177
Dudley H (1940) The carrier nature of speech. Bell Syst Tech J 19:495–515
Dudely H (1955) Fundamentals of speech synthesis. J Audio Eng Soc 3:170–185
Dudley H, Tarnoczy TH (1950) The speaking machine of Wolfgang von Kempelen. J Acoust Soc
Am 22:151–166
Dudley H, Riesz RR, Watkins SSA (1939) A synthetic speaker. J Franklin Inst 227:739–764
Elder D (1982) Pianists at play—interviews, master lessons, and technical regimes. The Instru-
mentalist Company Evanston, Illinois
Eulenstein K (1892) Eulensteins musikalische Laufbahn (edited by his daughter, Fanny Rooden-
fels, Stuttgart). Stadtarchiv, Heilbronn
Faber T (2005) Stradivari’s genius. Random House, New York
Farga F (1950) Violins & violinists, translated by Egon Larsen and Bruno Raikin. Barrie &
Rockliff, The Cresset Press, London
Farinelli (1994) Film biography of Carlo Broschi, directed by Gerard Corbiau (collaboratively
produced in France, Belgium and Italy for Sony Pictures Classics)
Farnsworth BW (1940) High speed motion pictures of the human vocal tract. Bell Lab Rec
18(7):203–208
Feynman RP, Robert BL, Matthew LS (1963) The Feynman lectures on physics. Addison-Wesley,
Reading, MA
Firth I (1976/77a) The nature of the tap tone in stringed instruments. Acustica 36(1):36–41
Firth I (1976/77b) A method for adjusting the pitch of top and back plates of the cello. Acustica
36(4):307–312
Flanagan JL (1972) Speech analysis synthesis and perception. Springer, Heidelberg
Forte M (2002) Chopin played by Madeleine Forte on the Erard Piano Paris, 1881—Yale collection
of musical instruments, Roméo Records No. 7214
Fostle DW (1993) Henry Zeigler Steinway, a grand traditionİ, interview in audio magazine
(January), pp 52–58
Fourier BJv (1822) La Theorie analytique de la chaleur (Paris). See Freeman’s translation.
Cambridge University Press, Cambridge, 1878
Friedlander FG (1953) On the oscillations of the bowed string. Proc Camb Philos Soc 49(3):516
Gaillard F (1939) Bach six suites for Violoncello Solo.İ Schirmer’s library of musical classics,
vol 1565; Preface
Gill D (ed) (1981) The book of the piano. Phaidon, Oxford
Giordano N (1998a) Mechanical impedance of a piano soundboard. J Acoust Soc Am 103:2128–
2133
Giordano N (1998b) Sound production by a vibrating piano soundboard: experiment. J Acoust Soc
Am 104:1648–1653
Giordano A, Dilworth J (2004) A fitting conclusion (Paganini’s” Cannon” violin). STRAD 115,
no. 1374: 1046-+
Giordano N, Winans JP II (1999) Plucked strings and the harpsichord. J Sound Vib 224:456–473
Giordano N, Winans II JP (2000) Piano hammers and their force compression characteristics: does
a power law make sense? J Acoust Soc Am 107:2248–2255
Giordano N, Millis JP (2001) Hysteretic behavior of piano hammers. In: Proceedings of the
international symposium on musical acoustics, Perugia
428 References

Gleason JB (1997) The development of language. Allyn and Bacon, Boston

Gold J (1995) Paganini: virtuoso, collector and dealer. J Violin Soc Am XIV(1):67–87
Gough CE (1981) The theory of string resonances on musical instruments. Acustica 49:124–141
Gray H (1918) Anatomy of the human body. Lea and Febiger, Philadelphia
Gray H, Warwick R, Williams PL (1980) Gray’s anatomy. Saunders, Philadelphia
Grissino-Mayer H, Burckle L (2003) Stradivari violins, tree rings and the maunder minimum: a
hypothesis. Dendrochronologia 21(1):41–45
Grissino-Mayer HD, Sheppard PR, Cleaveland MK (2004) A Dendroarchaeological re-
examination of the ‘Messiah’ violin and other instruments attributed to Antonio Stradivari.
J Archaeol Sci 31:167–174
Gruberova E (1988) Soprano, in Opera collection. Mozart, Die Zauber Flöte, Querschnite,
Nikolaus Harnoncourt, CD 063-13810-9, Teldec Classics
Guettler K (2002) On playing ‘harmonics’ (flageolet tones). Catgut Acoust Soc J 4(5):5–7
Guhr K (1831) Uber Paganinis Kunst die Violin zu Spielen (Main, 1831). Translated from the
original German by Sabilla Novello in 1915, revised by C. Egerton Lowe. Novello & Co.,
London; Also see the translation by Gold J (1982) Paganini’s art of violin playing. Teresa
Parker Associates, San Francisco
Gyütö M (1986) The Gyütö monks: Tibetan tantric choir WD-2001. 360Publishing, ASCAP and
Windham Hill Records, Stanford
Hannings L, Chin Y (2006) J Violin Soc XX:146–162
Hanon CL (1900) The virtuoso pianist in sixty exercises for the piano translated from the French
by Dr. Theodore Baker. G. Schirmer, Milwaukee
Hansell KK (1980) Sadie S (ed) Manzuoli, Giovanniİ in the new groves dictionary of music and
musicians, vol 11. MacMillan, London, p 638
Hanson RJ, Schneider AJ, Hagedahl FW (1994) Anomalous low-pitched tones from a bowed violin
string. Catgut Acoust Soc 2(6):1–7
Harrison JM, Thompson-Allen N (1996) Loudness level survey of the Newberry Memorial Organ,
Yale University. J Acoust Soc Am 100:3909–3916
Hartley RVL (1939) Oscillations in systems with non-linear reactance. Bell Syst Tech J 15:424–
440
Hawkshaw M, Sataloff RT, Bhatia R (2001) Chaos in voice research. In: The professional voice—
the science and art of clinical care. Singular Publishing Group, San Diego, pp 185–190;
Chapter 9
Hayes C (1951) The ape in our house. Harper, New York
Hayes JW (1982) String behavior and scale design. In: Travis JW (ed) A guide to restringing.
JohnTravis, Takoma Park, pp 31–64
Hedley A, Brown M (1980) Chopin, Fryderyk Franciszek. In: Sadie S (ed) The new groves
dictionary of music and musicians, vol 4. Oxford University Press, Oxford, pp 292–312
Helmholtz HLF (1885), Die Lehre von den Tonempfindungen. Longmans & Co, Harlow;
Translated into English by Ellis (1885) and republished as “On the sensations of tone”. Dover,
New York, 1954
Heriot A (1975) The Castrati in Opera. Da Capo Press, Cambridge
Herken G (2002) Brotherhood of the bomb. Henry Holt & Co, New York
Heron-Allen E (1885) Violin-making: as it was and is. Ward Lock & Co., London; Reprinted in
2000 by Algrove, Ottawa
Hill A, Hill A, Hill WH (1902) Antonion Stradivari his life & work. William E. Hill & Sons,
London, pp 1644–1737; Reprinted by Dover Publications, New York in 1963
Hill WH, Arthur F, Ebsworth A (1931) The violin-makers of the Guarneri family. William F. Hill
& Sons, London, pp 1626–1762; Reprinted by Dover Publications, New York in 1989
Hirano M (1970) Regulation of register, pitch and intensity of the voice. Folia Phoniatr 22:1–
20; Also see, “The regulatory mechanisms of the voice in singing”İ, instructional video tape
produced by The Voice Foundation, Philadelphia, p 19103
Hubbard F (1965) Three centuries of harpsichord making. Harvard University Press, Cambridge
References 429

Hussey LW, Wrathall LR (1936) Oscillations in an electromechanical system. Bell Syst Tech J
15:441–445
Hutchins CM (1962) The physics of violins. Sci Am 207(5):78–93
Hutchins CM (1981) The acoustics of violin plates. Sci Am 245(4):170–187
Hutchins CM (1983) A history of violin research. J Acoust Soc Am 73:1421–1440
Hutchins CM (2004) The catgut acoustical society story. Catgut Acoust Soc J 5:16–30
Hutchins C, Schelleng JC (1967) A new concert violin. J Audio Eng 15(4)
Hutchins C, Benade V (eds) (1997) Research papers in violin acoustics, vols 1, 2. Acoustical
Society of America, Woodbury, pp 1975–1993
Jansson E (1973) On higher air modes in the violin. Catgut Acoust Soc NL 19:13–15
Jansson E, Molin NE, Sundin H (1970) Resonances of a violin body studied by hologram
interferometry and acoustical methods. Phys Scr 2:243–256
Jansson EV, Niewczyk BK, Fryden L (1997) On body resonance C3 and violin construction. Catgut
Acoust Soc J 3(3):9–14
Javan A, Bennett WR Jr, Herriott DR (1961) Population inversion and continuous optical maser
oscillation in a gas discharge containing a He-Ne mixtureİ. Phys Rev Lett 6:106–110
Jeans SJ (1937) Science & music. Cambridge University Press, Cambridge
Jonas F (1968) Translation of selected essays on music by Vladimir Vasilevich Stasov. Praeger,
New York
Kaplan HM (1971) Anatomy and physiology of speech. McGraw Hill, New York
Katz M (2005) Capturing sound: how technology has changed music. University of California
Press, Berkeley
Kempelen WRv (1791) Mechanismus der menschlicken Sprache nebst der Beschreibung seiner
sprechenden Maschine. Available in the Yale University Beinecke Library. Also see, Dudley
and Tarnoczy, 1950
Kent RD (1997) The speech sciences. Singular Publishing Group, San Diego
Kimura M (2001) The world below G—subharmonics extend the violin’s range. STRINGS
(August/ September), pp 24–29
Kindel J (1989) Modal analysis and finite element analysis of a piano soundboardİ. M.S. thesis,
University of Cincinnati
Kindlmann PJ (1966) Measurement of excited state lifetimes. Ph.D. dissertation in Engineering
and Applied Science, Yale University
Kolbert E (2005) The climate of man. The New Yorker, New York; April 25, pp 56–71; May 2, pp
64–73; and May 9, pp 52–63
Kottick EL, Marshall KD, Hendrickson TJ (1991) The acoustics of the harpsichord. Sci Am 110–
115
Kroodsma D (2005) The singing life of birds. Houghton Mifflin, New York
Kuperman WA, Lynch JF (2004) Shallow-water acoustics. Phys Today 57(10) 55–61
Lebrecht N (2001) The maestro myth. Citadel Press, New York
Leipp E (1963a) Un ‘Vocoder’ Mécanique: La Guimbarde.İ Ann Télécommun 18(5–6):82–87
Leipp E (1963b) Étude Acoutsique de la Guimbarde. Acustica 13(6):382–396
Leipp E (1967) La Guimbarde. Bulletin du Groupe d’Acoustique Musicale, No. 25. Laboratoire
d’Acoustique Faculté des Sciences, Paris
Lenehan M (1982) Building steinway grand piano K2521—the quality of the instrument. Atl Mon
1–27
Levin TC, Edgerton ME (1999) The throat singers of Tuva. Sci Am 281:70–77
Litten A (1963) Soprano, in Schubert, Der Hirt auf dem Felsen with Stanley Udy (piano), and
William R. Bennett, Jr. (clarinet), recorded at Yale, December 1963
Loen JS (2001) Thickness graduation mapping: methods and goals. Catgut Acoust Soc J 4(4):5–6
Loen JS (2002) Fingerprints under the varnish: comparing thickness graduations of the ‘Messiah’
violin to golden age strads. Catgut Acoust Soc J 4(6):14–16
Loen J (2003) Thickness graduation mapping: surprises and discoveries. J Violin Soc Am
XIX(2):41–66; Proceedings of the thirty-first annual convention
430 References

Loen JS (2004) Thickness graduation maps of classic violins, violas, and cellos, 1st edn. Jeffery S
Loen, Kenmore
Luke JC (1971) Measurement and analysis of body vibrations of a violin. J Acoust Soc Am
49(4)(part 2):1264–1274
Mackay J (1997) Alexander Graham Bell—a life. Wiley, New York
Marshall KD (1985) Modal analysis of a violin. J Acoust Soc Am 77(2):695–709
Martin WH (1924) The transmission unit and telephone reference systems. Bell Syst Tech J 3:400–
408
Martin DW, Ward WD (1961) Subjective evaluation of musical scale temperament in pianos. J Am
Acoust Soc 33:582–585
McIntyre ME, Schumacher RT, Woodhouse J (1982) Aperiodicity in bowed-string motion.
Acoustica 50:294–295
Mersmann H (1972) Editor of letters of Wolfgang Amadeus Mozart. Dover, New York; Reprinted
from an original edition by J.M.Dent & Sons, London, 1928
Michelson AA (1903) Light waves and their uses. University of Chicago Press, Chicago
Millant R (1972) J. B. Vuillaume. W. E. Hill & Sons, London; This book is in English, French and
German
Milstein N, Solomon V (1990) From Russia to the west—musical reminiscences of Nathan
Milstein, translated by Antonina W. Bouis. Limelight Editions, New York
Moreschi A (1902, 1904) Vatican cylinder recordings reproduced on LP disc by opal records circa
1984; Record no. 823
Morrison M (1983) Video highlights from the voice clinic, vol 1. The Voice Foundation,
Philadelphia, pp 19103
Neumayr A (1994) Music and medicine—Haydn, Mozart, Beethoven, Schubert—notes on their
lives, works, and medical histories, translated by Bruce Cooper Clarke. Medi-Ed Press,
Bloomington
Neuwirth R (1994) From sciarrino to subharmonics. STRINGS, September/October, pp 60–63
Nilsson B (1984) Master class, vol 60.İ The New Yorker, New York, pp 44–45
Nyquist H (1924) Certain factors affecting telegraph speed. Bell Syst Tech J 3:324–346
Odiaga L (1974) Johann Sebastian Bachİ, Editión del Patronato Popular y Porvenir Pro Música
Clásica; CBS Discos del Perù; Recorded in New Haven by William R. Bennett; P.E.002-A and
P.E.002-B
Olson HF (1952) Musical engineering. McGraw-Hill, New York; Republished as music, physics
and engineering. Dover, New York, 1967
Pedersen PO (1934) Subharmonics in forced oscillations in dissipative systemsİ, Part I [Theoreti-
cal]. J Acoust Soc Am VI:227–238; Part II [Experimental], J Acoust Soc Am VII:64–70
Pepperberg IM (1991) Referential communication with an African grey parrot. Harvard Graduate
Society News Letter (Spring), pp. 1–4. Also see, “Brain Parrot Dies. . . ”İ by Benedict Carey,
N.Y. Times, Tuesday, September 11, 2007, p A23 and “Alex Wanted a Cracker. . . ”İ by George
Johnson, N.Y. Times, September 16, 2007, pp 1,4 wk
Perrins CM, Middleton ALA (1985) The encyclopedia of birds. Facts on file, New York
Peschel ER, Peschel RE (1987) Medical insights into the Castrati in opera. Am Sci 75:578–583
Peterson (1999) Skubic M (ed) AutoStrobe 490-ST strobe tuner instruction manual. Peterson
Electro-Musical Products, Alsip
Pickering, NC (1991) The bowed string. Amereon Ltd, Mattituck
Pickering NC (1993) Problems in string making. Catgut Acoust Soc J 2(3):1–4
Pierce JR, alias, Coupling JJ (1983) The science of musical sound. Scientific American Books,
New York
Pleasants H (1981) The great singers from the dawn of opera to Caruso, Callas, and Pavarotti.
Simon and Schuster, New York
Potter RK, George AK, Green HC (1947) Visible speech. D. Van Nostrand Company, New York
Railsback OL (1938) J Acoust Soc Am 9:274
Raman CV (1918) On the mechanical theory of vibrations of bowed strings. Bull Indian Assoc
Cultiv Sci (Calcutta) 15:1–158
References 431

Raman CV (1988) Ramaseshan S (ed) Scientific papers of C. V. Raman volume II acoustics. Indian
Academy of Sciences, Bangalore
Ramanujan S (1914) Modular equations and approximations to p. Q J Pure Appl Math 45:350–373
Rayleigh L, alias, Strutt JW (1877) Theory of sound, vols I and II. Macmillan Co. Reprinted by
Dover, New York, 1945
Riesz RR (1932) The relationship between loudness and the minimum perceptible increment of
intensity. J Am Acoust Soc 4:211–216
Ripin EM (1980) Groves dictionary of music, vol 14. Macmillan, London, pp 682–714
Ripin EM, Schott H, Barnes J, O’brien G (1980) Groves dictionary of music, vol 8. Macmillan,
London, pp 216–246
Rioul O, Vetterli M (1991) Wavelets and signal processing. IEEE Signal Process Mag 8:14–38
Robbins WW, Weier TE (1950) Botany: an introduction to plant science. Wiley, New York
Rosen C (2002) Piano notes—the world of the pianist. The Free Press, New York
Rossing T (1990) The science of sound. Addison-Wesley, Reading
Rossing T, Fletcher N (1991) The physics of musical instruments. Springer, New York
Rothstein E (1994) A violinist tests music of her time. New York Times, New York, p C21; April
21
Rothstein E (1995) Emblems of mind—the inner life of music and mathematics. Avon Books, New
York
Runge C (1903) Z Math Phys 48:443
Rymer R (2004) Saving the music tree. Smithsonian 35(1):52–63
Sacconi SF (2000) The “secrets”İ of Stradivari, translated by Andrew Dipper and Cristina Rivaroli
with corrections to the first English edition of 1975 by Andrew Dipper. Eric Blot Edizioni,
Cremona
Sadie S (1980) Mozart (3), Wolfgang Amadeus. In: Sadie S (ed) New groves dictionary of music
and musicians, vol 12. MacMillan, London, pp 680–752
Salgo S (2000) Thomas Jefferson, musician and violinist. Thomas Jefferson Foundation, Moticello
Sataloff RT (1992) The human voice. Sci Am 267:108–115
Sataloff RT, Hawkshaw M, Bhatia R (1998) Medical applications of chaos theory,İ in: Davis PJ,
Fletcher NH (eds) Vocal fold physiology. Singular Publishing Group, San Diego, pp 369–387
Saunders FA (1962) Violins old and new – an experimental study. Sound: Its Uses Control 1(4):7–
15 (1962)
Savart F (1819) Rapport sur un mémoire relatif a la construction des instruments àă cordes et àă
archet. Ann Chim Phys 2:225
Schawlow AL (1996) Charles townes as i have known him. In: Chiao RY (ed) Amazing light, a
volume dedicated to Charles Hard Townes on his 80th birthday. Springer, New York, pp 1–6
Schelleng JC (1963) The violin as a circuit. J Accoust Soc Am 35:326–338; Especially, see Section
VI
Schleske M (1998) On the acoustical properties of violin varnish. Catgut Acoust Soc J 3(6):27–43
Schneider D (2004) Living in sunny times. Am Sci 93:22, 23
Schonberg HC (1984) History’s last castrato is heard again. The New York Times, New York, p
H25; September 16
Schonberg HC (1985) The glorious ones: classical music’s legendary perfomers. Times Books,
New York, p 3
Schroeder MR (1999) Computer speech. Springer, Berlin
Schuck OH, Young RW (1943) Observations on the vibrations of piano strings. J Am Acoust Soc
15:1–11
Schwartz D (1996) Snatching scientific secrets from the Hippo’s gaping jaws. Smithsonian
26(12):90–102
Segré G (2002) A matter of degrees—what temperature reveals about the past and future of our
species. In: Planet and universe. Viking, New York
Serber R, Crease RP (1998) Peace & war. Columbia University Press, New York
Shakespeare W, Craig WJ (1966) Shakespeare: complete works. Oxford University Press, London.
Print
432 References

Shanks D, Wrench JW (1962) Calculation of PTO 100,000 decimals. Math Comput 16:76;
poccupies a 20-page table
Shorto R (2004) The island at the center of the world. Doubleday, New York
Silverman WA (1957) The violin hunter. Paganiniana, Neptune City
Sitwell S (1967) Liszt. Dover Publications, New York
Solomon M (1995) Mozart—a life. Harper Collins, New York
Sonnaillon B (1985) King of the instruments—a history of the organ. Rizzoli, New York
Stasov, V. V. Selected essays on music. London: Cresset P, 1968. Print
Steadman P (2001) Vermeer’s camera. Oxford University Press, Oxford
Steinhardt A (1998) Indivisible by four—a string quartet in pursuit of harmony. Farrar Straus
Giroux, New York
Steinhardt A (2006) Violin dreams. Houghton Mifflin, New York
Stetson KA, Powell RI (1965) Interferometric vibration analysis by wave front reconstruction. J
Opt Soc Am 55:1593–1598
Stetson KA, Powell RI (1966) Hologram interferometry.İ J Opt Soc Am 56:1161–1166
Stevens KN (1998) Acoustic phonemes. The MIT Press, Cambridge
Sundberg J (1987) The science of the singing voice. Northern Illinois University Press, DeKalb
Synge JL, Griffith BA (1949) Principles of mechanics. McGraw-Hill Book Co., New York
Tammet D (2007) Born on a blue day. Free Press, New York
Taub R (2002) Playing the Beethoven piano sonatas. Amadeus Press, Portland
Taylor DA (1965) Paderewski’s piano, Smithsonian, Maryland, pp 30–34
Thayer AW, Forbes E (eds) (1973) Thayer’s life of Beethoven. Princeton University Press,
Princeton
Travis JW (1982) A guide to restringing. John W. Travis, Takoma Park
Trent HM, Stone DE (1957) Elastic constants, hardness, strength, and elastic limits of solids.İ In:
American institute of physics handbook. McGraw Hill, New York; Chapter 2f
TUVA (1990), Voices from the center of Asia, Smithsonian Fokways CD 40017
Valente B (1960) Schubert, Der Hirt auf dem Felsen, in Recital No. 2 by Harold Wright (clarinet),
Rudolf Serkin (piano), Boston Records BR1024CD; Band 10
Van Dyke M (1982) An album of fluid motion. The Parabolic Press, Stanford
Walker T (1980) Castratoİ. In: Sadie S (ed) The new grove dictionary of music and musicians,
vol 3. Macmillan, London, pp 875–876
Wang LM (1999) Radiation mechanisms from bowed violinsİ. Doctoral Thesis in Acoustics,
Pennsylvania State University, State College, Pennsylvania
Wang LM, Burroughs CB (2001) Acoustic radiation from bowed violins. J Acoust Soc Am
110:543–555
Weatherburn CE (1951) Advanced vector analysis. Bell and Sons, London
Weinreich G (1977) Coupled piano strings. J Acoust Soc Am 62:1474–1484
Weinreich G (1997) Directional tone color. J Acoust Soc Am 101:2338–2346
Weinreich G (2002) Sound radiation from the violin—as we know it today. J Acoust Soc Am
5:37–42
West MJ, King AP (1990) Mozart’s starling. Am Sci 78:106–114
Wheatstone C (1828) Erkläring der vermittels des Brummeisens hervorgebrachten Töne. Allg
Musik Z (Leipzig) 30:626
Wheatstone C (1879) The scientific papers of Sir Charles wheatstone. Physical Society of London,
London, pp 348–367
Whittaker E, Watson G (1902) Modern analysis, Chapter 12. Cambridge University Press,
Cambridge
Whittaker ET, Watson GN (1920) A course of modern analysis, 3rd edn. Cambridge University
Press, Cambridge
Winchester S (2003) Krakatoa—the day the world exploded: August 27, 1883. HarperCollins, New
York
Wood AB (1955) A textbook of sound. Bell, London
References 433

Woodhouse J (2002) Body vibrations of the violin—what can a maker expect to control? J Acoust
Soc Am 5:43–49
Wright J (1980) Jew’s Harp.İ In: Sadie S (ed) The new grove dictionary of music and musicians,
vol 9. MacMillan Publishers, London, pp 645–646
Yoshikawa S (1997, in Japanese) Bowing positions to play cello tones on the violin. J Acoust Soc
Jpn (Nihon Onk yo Gak kai-shi) 53(12):955–963
Young RW (1957) Frequencies of simple vibrators. Musical scales. American institute of physics
handbook. McGraw-Hill, New York, pp 3-100–3-107
Yung BN (1980) Chin23 part IV. Theoryİ. In: Sadie S (ed) The new grove dictionary of music and
musicians, vol 4. MacMillan Publishers, London, pp 260–283
Zemansky MW (1951) Heat and thermodynamics. McGraw-Hill, New York
Index

A Blake, Eubie, 124

Aki, Syoko, 150, 226, 383, 400 Bloom, Robert, 87, 309
Alard, Delphin, 156–158 Bolet, George, 118
Amati family Broadwood, James Shudi, 102–105, 111
Andrea, 150
Hieronymous, 150
Nicolo, 151 C
Angular velocity, 7 Campbell, William C., 44, 45
Arnold, Paul, 171, 173, 176, 201, 212, 215, Chickering, Jonas, 89, 114
223, 224 Cicadas’ noise spectrum, 56
Ashmolean Museum, 156 Circle of fifths, 77–79
Clavichord
action, 90, 101
B Clavichord:, 82, 90–92, 101, 110, 123, 125
Babcock, Alpheus, 110, 114 Closed-pipe waveform, 45
Bach, C. P. E., 124, 146 Cohen, Dr. Lawrence, 53
piano fingering, 124 Commins, Eugene, 247
Bach, J. C., 104 Convergent series, 38, 39
Badura-Skoda, Paul, 208 Couchet, Jean, 82
Balkan maple, 148, 175, 178–181 Cremonese makers, 147–149, 151, 153, 178,
Ballot, Buijs, 26 183
Beckerath, Willie von, 119 Cristofori, Bartholomeo, 95–100, 111, 124,
Beecham, Sir Thomas, 87, 212 138
Beethoven, 63, 88, 97, 100–105, 111, 119, 126, Cuesta, C., 84
129, 149, 151, 210, 211, 225, 261
Broadwood piano, 103, 105
Behrent, John, 110 D
Bellot, Louis, 80 d’Alembert’s solution for the plucked string,
Bergonzi, Carlo, 153, 154 37, 60, 61, 225
Berlioz, Hector, 207, 210 Davidov, Karl, 153
Bernheimer, Martin, 126 decibel (dB), 34–36, 44, 223, 271
Bernoulli, Daniel (string solution), 37, 62, de’ Medici, Caterina, 148
251–253, 275, 342, 360 Dolge, Alfred, 95, 110–112
effect, 251, 253, 275, 342 Dolmetsch, Arnold, 89
Black hole pressure waves, 28 dolomite spruce, 148, 175, 178, 180

© Springer Nature Switzerland AG 2018 435

W. R. Bennett, Jr., The Science of Musical Sound,
https://doi.org/10.1007/978-3-319-92796-1
436 Index

Doppler Shift Kreisler’s violin, 154, 210

demonstration at Maarsen, 26 Paganini’s “Cannon,” 156, 169
radar, 29 Joseph (del Gesu), 155, 156
rotating fan, 29 Guettler, 219, 227
Double-bow technique, 218
Dowd, William, 89
Duiffoprugghar, Gasparo (alias Kaspar H
Tieffenbrucker or Dieffenbrucker), Haas, Julie, subtones, 238, 239
146 Haberbosch, Christopher, 137, 141
du Prés, Jacqueline, 153 Hahn, Erwin, 255
Hanon, C. L., 125
Hansen, Ray, 124
E Harmonics, 10, 13–15, 28, 31, 33, 42, 44–46,
Einstein, Albert, 158 48–49, 55, 56, 59, 63, 66–69, 76,
Eisenstadt, Frone, 183 77, 83, 84, 86, 88–91, 94, 98, 108,
Eisenstadt, Maurice, 247 113, 115, 127, 128, 130, 164, 165,
Electric guitar, 65–66 169, 197, 205, 206, 216, 224–228,
Erard, Sebastian, 89, 102–107, 114–116, 122, 231, 234, 235, 241–243, 245–247,
125 252–257, 262, 264–266, 272, 274,
Erhu, Chinese, instrument, 145 275, 279–281, 285, 289–296, 315,
Erle, Broadus, 211, 212, 230 316, 318, 341, 417, 422, 423
Ernst, H. W., 226, 233 Harpsichord, 38, 49, 57, 69–76, 79–82, 84,
85, 87–91, 94, 96, 97, 99, 102–104,
118–120, 124, 126, 129, 130, 132,
F 143, 160, 164, 182, 225
Faraday’s Law, 66, 334 Hayes, James B., 120, 139, 257
Fazioli, Paolo, 111, 126, 127 Heart murmur spectra, 52, 53
Fechtner’s law, 224, 230 Helmholtz, 32, 80, 160–163, 166, 170, 183,
Fender company, 127 190–192, 195, 207, 227, 317,
Flaming maple, 156, 179 354–355, 364, 366
Focal length of spherical mirror, 19, 20 Hemsch, Henri, 87
Forte, Madeleine, 116 Hertz (Hz), 2
Fourier Analysis, 36, 44–51, 63, 160, 258, 264, Hirl, William, 89
279, 348, 352, 364, 377–414 Hoffman, Joseph, 124
FFT, 52–54, 89, 279 Holland, Bernard, 95
Fourier transform, 49–56, 138, 279, 332 Hooke’s Law, 112, 237, 241, 329
Frederick, E.M., and Patricia, 95, 116, 117 Horowitz, Vladimir, 107, 124, 126
piano museum, 114, 116 Horsehair photomicrographs, 221, 222
Horsetail or scouring rush, 183, 184
Hubbard, Frank, 63, 70, 71, 75, 76, 79, 81–83,
G 85–87, 89
Galilei, Galileo, 151 Hutchins, Carleen, 179, 189, 191, 192, 197,
Garden hose waveform, 46 198, 201, 204, 205, 207–210, 247
Genualdi, Joseph, 154, 155, 206, 212 Hutchins octet, 207–210
Gibbs phenomenon, 41 Huygens, Christiaan, 16–19, 22
Glissando, 28 Huygens, Constantijn and Vermeer-Ruckers
Gold, Joseph, 156, 226, 233 connection, 82
Graf, Conrad, 101, 102
Guadagnini, J. B., 153
Guarneri family I
Andrea, 150 Iizuka, Hiroshi, 170, 173, 175–177, 180, 184,
del Gesu instruments 191, 192, 215, 223
Alard, 156 viola, 175, 176
Guarneri label, 156 Inverse-square law and diffraction, 22–25
Index 437

J Mozart, Leopold, 164

Jansson, Erik, 168, 193–197, 199, 200 Munchkin effect, 28
Jefferson, Thomas, 110, 151 Museum of Fine Arts, Boston, 92
Mussorgsky, Modeste, 77, 78

K
Kanada, Yasumasa, 5 N
Kantuscher, Josef, 175, 176 Naperian logarithms, 39
Kennedy, Paul, 70–72, 75, 99 Nash, Ogden, 95
instruments, 70–72, 99 Neumayr, Anton, 102, 283
Kimura, Mari, 228, 229, 232–243 Nyquist, Harry, 48
Kindel, J., 132–134 Nyquist sampling theorem, 48
Kindlmann, Peter, 211, 212, 243
Kirkpatrick, Ralph, 89, 90
Klotz, Matthias, 175 O
Kolisch, Rudolf, 169 Obikho, 78
Koto, Japanese, 57, 59 Odiaga, Lola, 87
Kottick, Edward, 73, 80 Open and closed pipe modes, 12, 13
Krakatoa explosion, 55 Oppenheimer, J. Robert, 55
Kreisler, Fritz, 154, 210 Oscillators: A1–A25, 14, 266, 305, 322, 330,
Kreter, Rudolf, 111 351, 352

L P
Landowska, Wanda, 89 Paderewski’s Steinway, 115, 120, 124, 125
Langbell, Kenneth, 126 Paganini, Nicolo, 226
Law of cosines, 20 Pecatte, Dominique, 222, 223
Leyden, Massachusetts, 55 Pedal mania, 101
Lin, Cho-liang 151, 154 Perfect pitch, 29, 256, 258
Liszt, Franz, 105–107, 110, 114, 118, 226 Periodic waveform, 36, 41, 48, 49, 54, 225,
Loen, Jeffrey, 179, 185–187, 193 233, 237, 242, 252, 279, 290, 291,
Lorentzian shape, 50, 336 293, 294, 296, 314, 324, 380
Lupu, Radu, 124 Pernambuco, 218, 220, 222, 223
pi (π ), 5
Piano
M agraffe, 103, 104, 121, 125, 137, 138
Maarsen, 26, 29 aliquot stringing, 126
Maguire, Katie H., 270 Babcock of Boston, 110
Mammoth tusks for frogs, 221 back check, 96, 99, 104, 105, 123
Manikara kauki wood for bows, 220 Beethoven piano, 102–104
Martin, Willard, 76 Blüthner pianos, 111, 126
Masaoko, Maya, 59 Bösendorfer imperial, 127
Maunder, E. W., 178 Brahms’ piano, 77, 99, 116–119
Maunder minimum (Little Ice Age), 178 Bridge, 120
Ma, Yo-Yo, 153 Broadwood action, 104, 142
Medici family in Florence, 95 capo d’astro bar, 121, 125
Mekeel, Joyce, 74 capstan, 123
Michelson’s spectrum analyzer, 33, 55 Chickering company, 89, 114
MIDI, 129, 267, 312 comma of Pythagoras, 78
Mirrors, 195 crashing piano sound, 118, 142
plane, 18, 20–21 double escapement, 104–107
plane mirror moving, 20–21 electronic piano, 127, 129
spherical, 19–20 Erard piano, 105, 116
Mitral prolapse, 52, 53 Fazioli pianos, 111, 126, 127
438 Index

Piano (Cont.) Reichardt, Karl, 102

hammers, 110–112, 217 Rephann, Richard, 70, 75, 87, 91, 383, 400
voicing, 113–114 Resonant frequencies, 10, 13, 15, 28, 51, 115,
inharmonicity of strings, 136–140 134, 184, 185, 196, 198, 205, 319,
effect on tuning, 140–142 359, 360, 369
iron frame, 103, 110, 114, 119, 122, 421 Rhapsody in Blue, 268
janizary stop, 101 Rhine Railroad, 26
Knabe hammers, 112 Rhodes, Harold, 127
knuckle, 105, 123 Richard, E. Rouen luthier, 93
Mason and Hamlin action, 122 Richter, Hans, 89
modern action, 124 Robinson, Christopher, 95, 112, 113, 137, 138,
Mozart piano, 99, 101 141
Paderewski’s Steinway, 115, 120, 124, 125 Rocca, Giuseppe
Pedal mania, 101 cello, 158
prellmechanik action, 100, 101 Rosen, C., 102, 103, 125, 127
real strings, 58, 135–138 Rubinstein, Artur, 126
Renner Company Stuttgart, 111, 112 Ruckers, Andreas, 81, 82, 84
Revenaugh modification, 134, 135 Ruckers, Hans, 81
scaling, 121, 140 Ruggieri, Francesco, 150, 151
scaling & bridge shape, 120–122 Russian bowing, 232
single escapement action, 118 Ruth, Babe, 142
sitka spruce, 127
sostenuto pedal, 119–120
sound board modes, 115, 132–136 S
sound pressure levels, 97 Sacconi, Simone, 146, 151, 152, 173, 175, 183,
Steinway D 185, 190, 191, 200, 201, 207, 214
inharmonicity, 136–139 Salgo, S., 210
spectra, 128 San Domenico, 147
stretching (in tuning), 141 San Marco monastery, 28
Streicher action, 119, 142, 420 Sartory, Eugène, 223
Streicher Company, 119 Saunders, Martin, 88
teflon®lubricant, 126 Savart, F., 155, 175, 178, 184, 189
transient build-up & decay, 129–130 Savonarola, Girolamo, 28
Una Corda mode, 97–102 Scelsi, Giacinto, 218
wippen, 123 Schumann, Robert, 124
Yamaha pianos, 127 Serber, R., 55, 258
Piatigorsky, Gregor, 218 Serkin, Rudolph, 102, 285
Pickering, N.C., 163–165, 232, 240, 366 Servais, Adrien-François, 169
Plucked string motion, 60–65, 225 Sikora, Greg, 137, 141
Poll, Hermann, 75 Silberman, Andreas, 303, 307
Propagation constant (k), 8–10 Silberman, Gottfried, 99
Sound pressure levels (SPLs), 35, 216, 309
Specific heats of common gases, 15, 354
Q Spectrum analyzer, 32, 33, 36, 48, 51, 55, 212,
Quintadena organ pipe waveform, 42 230, 235, 236, 264, 271
Spherical mirror, 19–20
Spohr, Ludwig, 151
R Standing waves, 9–13, 28, 194, 196, 240, 254,
Radiation of a spherical wave 255, 293, 314, 317, 333
from a dipole, 23 Stasov, V.V., 106
from a linear multipole, 24 Steam engine whistle, 28, 418
from quadrupoles, 24 Steinberger bass, 66–69, 94
Ramanujan, S., 39 Steiner, Jacob, 175
rebec, Arabian, 145 Steinhardt, A., 222
Index 439

Stokes, George, 76 bowed strings

Stradivari Family Helmholtz model, 160–163
Antonio, 150–151, 154, 178 real bowed string, 163–165
Francesca (grand daughter), 154, 156 stick-slip process, 162, 364
Paolo, 147 tastiera bowing, 228–229
Stradivari instruments bows
Betts violin, 154 Bach bow, 217
Castelbarco violin, 154 bow hair, 145, 159, 163, 164, 218, 220,
Christian Hammer, 152 235, 240
Davidov cello, 153 Chinese bow, 223
Herberlein-Taylor violin, 154, 155, 212 double bow, 218
Lifetime output, 152 Evolution, 217–224
“Messiah” violin, 155–158, 187 octave subtones, 232–239, 241, 242
sound levels, 154, 197, 201, 208 pernambuco, 216, 220, 222, 223
Stradivari label, 151–154 spectral comparisons, 150–158
tools, 147, 153, 183 bridge, 169–173
bridge pulses & spectra, 150–158
cigar box modes, 194, 196
T f -holes, 146, 156, 166, 178, 179, 185,
Tarisio, Luigi, 154 190–194, 196, 199, 201–203
Tartini, Giuseppe, 230 harmonics, 226–227
Tartini tones tables, 229–232 Helmholtz model, 160–163
Taskin, Pascal, 80, 85–89 higher-order modes, 193
Tastiera bowing, 228–229 holographic studies, 199–200
Taub, R., 99, 101–103, 130 Jansson’s model, 196
Telmányi, Emil, 217 loudness curves, 197, 198
Terpenoid resin rejuvenator, 153 Mirecourt, France, 157
Tourte, François-Xavier, 218 Mittenwald School, 175
Tree growth patterns, 181 plate modes, 169, 190, 197
Tree, Michael, 175–177, 232 purfling, 156, 166
Très Grande Vitesse (TGV), 29 radiation patterns, 127, 189, 194, 198,
Tuva throat singer, 293–297, 349 200–205
sound post, 153, 166, 168, 169, 171–173,
175, 183, 188, 190, 192, 193, 199,
U 202
Uncertainty Principle, 50, 51 spectrum (Amati), 150–158
Utrecht, 26 subtone excitation, 238, 240
Tartini tones, 229–232
torsional waves, 239–242
V varnish, 153, 156, 213–217, 223
Valette, C., 84 vibrating strings & membranes
Varnish, 153, 156, 213–217, 223 bowed string, 364–366
Vega Bach bow, 217 circular membranes, 372–375
Velocity of sound, 9, 11, 15–18, 25, 28, 29, 55, general solution, 359–360
182, 188, 191, 194, 307, 309, 327, kettledrum, 374, 375
355, 420, 423 plucked string, 360–361
Verizon Hall, Philadelphia, 204, 309 struck string, 361–364
Vermeer, Johannes, 82, 103 torsional waves, 366–367
Vestergaard, Knute, 217, 218 wave equation for string,
Viola pomposa 357–359
air resonances, 146 vibrational modes
Violin thin membrane, 187–189
bass bar, 173–175 top plate, 182–185
bouts, 195 vibrato, 210–212
440 Index

Violin (Cont.) reflections, 9–11

virginal, 63, 70, 82, 94 running waves, 4–5
wolf tones, 205–206 sinewave generation, 3–4
Vuillaume, J.-B., 155, 223 sound waves, 11–12
Octobass, 207, 208 standing waves, 9–11
wavelength, 2–3
Weinrech, Gabriel, 97–99, 202, 204, 205
W Well-tempered scale, 29, 76, 79, 141, 246, 290,
Walter, Anton, 99, 101 415
Wang, Lily, 204 Wilbraham effect, 41
wave motion Witten, Lawrence, 148
angular frequency (ω), 7
cos, tan, 6
frequency, 2–3
Y
nomenclature table, 3
Yoshikawa, S., 235, 239, 244
period, 2–3
Ysaÿe, Eugène, 210
periodic wave, 2, 49
phase angle
pi (π ), 5
propagation, 1 Z
propagation constant (k), 8–9 Zigetti, Joseph, 212