Professional Documents
Culture Documents
Tibetan Singing Bowls
Tibetan Singing Bowls
Department of Mathematics,
Abstract
We present the results of an experimental investigation of the acoustics and
fluid dynamics of Tibetan singing bowls. Their acoustic behavior is rationalized
in terms of the related dynamics of standing bells and wine glasses. Striking
or rubbing a fluid-filled bowl excites wall vibrations, and concomitant waves at
the fluid surface. Acoustic excitation of the bowls natural vibrational modes
allows for a controlled study in which the evolution of the surface waves with
increasing forcing amplitude is detailed. Particular attention is given to rationalizing the observed criteria for the onset of edge-induced Faraday waves and
droplet generation via surface fracture. Our study indicates that drops may be
levitated on the fluid surface, induced to bounce on or skip across the vibrating
fluid surface.
Introduction
Tibetan singing bowls are thought to have originated from Himalayan fire cults
of the 5th century BC and have since been used in various religious ceremonies,
including shamanic journeying and meditation. The Tibetan singing bowl (see
Fig. 1) is a type of standing bell played by striking or rubbing its rim with a
wooden or leather-wrapped mallet. This excitation causes the sides and rim
of the bowl to vibrate and produces a rich sound. Tibetan bowls are hand
made and their precise composition is unknown, but generally they are made of
a bronze alloy that can include copper, tin, zinc, iron, silver, gold and nickel.
When the bowl is filled with water, excitation can cause ripples on the water
surface. More vigorous forcing generates progressively more complex surface
1
wave patterns and ultimately the creation of droplets via wave breaking. We
here quantify this evolution, and demonstrate the means by which the Tibetan
singing bowl can levitate droplets.
b)
d)
a)
c)
Faraday waves have a wavelength prescribed by the relative importance of surface tension and gravity, being capillary and gravity waves in the short and long
wavelength limits, respectively. As the forcing acceleration is increased, progressively more complex wave patterns arise, and the interfacial dynamics become
chaotic [20, 21, 19]. Ultimately, large amplitude forcing leads to surface fracture
and the ejection of droplets from the fluid bath. A recent study on the breaking
of Faraday waves in a vertically shaken bath has been performed in both the
capillary and gravity wave limits by Puthenveettil et al. [22]. Goodridge et al.
[23] studied the drop ejection threshold of capillary waves in a glycerine-water
solution for frequencies up to 100 Hz.
Faraday [14] reported that such parametric waves can also be emitted by a
vertical plate plunged into a liquid bath and oscillating horizontally: along both
sides of the plate, waves aligned perpendicular to the plate oscillate at half the
forcing frequency. These so-called cross-waves, or edge-induced Faraday waves,
are typically produced by a wave-maker, and have received considerable attention [24, 25, 26, 27, 28]. Hsieh [29] examined theoretically wave generation in
a vibrating circular elastic vessel, specifically the axisymmetric capillary waves
and circumferential crispations that appear in an inviscid fluid subject to radial
wall displacement. These studies have demonstrated that the excitation of the
cross waves is due to a parametric resonance. The complexity of this problem
lies in the nonlinear interactions between the motion of the oscillating rim and
the resulting wave field.
Droplets ejected on the liquid surface by breaking Faraday waves may bounce,
skid and roll before coalescing. A number of recent studies have examined
droplets bouncing on a vertically vibrating liquid bath below the Faraday threshold [30, 31, 32]. The air film between the drop and the liquid surface is squeezed
and regenerated at each successive bounce, its sustenance precluding coalescence
and enabling droplet levitation. A similar effect arises on a soap film, a system
more readily characterized theoretically [35]. The bouncing periodicity depends
on the size of the drop and the vertical forcing acceleration of the bath [33, 34].
Couder et al. [30] have shown that, through the waves emitted at previous
bounces, some droplets can walk horizontally across the liquid surface. Several
factors are needed to sustain a so-called walker [36]. First, the drop must
bounce at half the forcing frequency, so that it resonates with the resulting
Faraday wave field. Second, the bath must be close to the Faraday instability
threshold so that Faraday waves of large amplitude and spatial extent can be
excited by the drop impacts. The droplet bounces on the slope of the wave
emitted at the previous bounce and so receives an impulsive force in a specific
direction, along which it walks with a constant speed. Such walkers have both
wave and particle components, and have been shown to exhibit quantum-like
dynamical behaviour previously thought to be peculiar to the microscopic realm
[37, 38, 39, 40]. Might such modern physics arise in our ancient bowls?
The manuscript is divided into two main parts. In 2, we examine the
acoustics of the Tibetan singing bowls. We begin in 2.1 by reviewing the
related dynamics and theoretical description of the wine glass [1]. In 2.2, our
bowls are presented and their deformation spectra analyzed. Then, by adapting
3
the theoretical description of the vibrating wine glass, we infer the Youngs
modulus of the alloy comprising our bowls. In 3, we examine the dynamics of
flows generated within liquid-filled vibrating bowls. A review of Faraday waves
and droplet ejection on a vertically shaken bath is presented in 3.1. In 3.2,
our experimental technique is detailed. In 3.3, we analyze the surface waves
generated on the liquid bath, and their relation to Faraday waves. In 3.4, we
examine the limit of large amplitude forcing, in which droplets are ejected by
breaking Faraday waves. Comparisons are made with experiments performed
on a vertically shaken liquid bath. Droplet levitation is considered in 3.5,
where particular attention is given to developing criteria for droplet bouncing
and exploring the possibility of walking droplets. A summary of our results is
presented in 4.
2
2.1
Acoustics
Background
Both the wine glass and the Tibetan bowl can be excited by either tapping or
rubbing its rim. We denote by (n,m) the vibrational mode with n complete
nodal meridians and m nodal parallels. Tapping excites a number of vibrational modes [10], while rubbing excites primarily the (2,0) fundamental mode
(see Fig. 2b). Entirely flexural motion implies radial and tangential displacements proportional to n sin n and cos n, respectively, being the azimuthal
coordinate [41]. For the (2,0) mode the maximum tangential motion is necessarily half the maximum normal motion.
A leather mallet can excite bowl vibrations via a stick-slip process, as does
a finger moving on a wineglass. The moving mallet forces the rim to follow the
mallet during the stick phase; during the slip phase, the bowl rim relaxes to its
equilibrium position. This rubbing results in a sound composed of a fundamental
frequency plus a number of harmonics. While the mallet is in contact with the
bowl, one of the nodes follows the point of contact [11], imparting angular
momentum to the bound liquid.
To simplify the acoustic analysis, one can approximate the glass or bowl by
a cylindrical shell with a rigid base and an open top (Fig. 2a). The system
can then be described in terms of 7 physical variables, the radius R, height
H0 , thickness a, Youngs modulus Y and density s of the cylindrical shell, and
the frequency f and amplitude of its oscillating rim. The system can thus
be described in terms of 4 independent dimensionless groups, which we take
to be R/H, /a, /R and a Cauchy number Ca = s f 2 2 /Y that indicates
the relative magnitudes of the inertial and the elastic forces experienced by the
vibrating rim.
The sound is emitted by bending waves that deform the rim transversely as
they propagate. The speed Vb of bending waves on a two-dimensional plate of
f,
a)
b)
H0
Figure 2: a) Schematic illustration of a vibrating vessel filled with liquid. Relevant parameters are the height H0 and the radius R of the vessel, the thickness
of the rim a, the liquid level H, the frequency f and the amplitude of the
oscillating rim. b) Top view of a vibrating vessel in its fundamental mode (2,0),
characterized by its 4 nodes and 4 antinodes.
thickness a is given by [42]
Vb =
VL f a
1/2
,
(1)
where VL is the longitudinal wave speed in the solid (approximately 5200 m/s
in glass). In order for the bending wave to traverse the perimeter in an integer
multiple of the period, we require
1
2R
.
f
Vb
(2)
3Y
5s
1/2
"
4 #1/2
a
4 R
1+
.
R2
3 H0
5
(3)
When the bowl is partially filled with liquid to a depth H (Fig. 2a), the frequency
decreases. French [1] captured this effect by considering the kinetic energy of
the liquid near the rim, and so deduced the frequency of the fundamental mode:
f0
fH
2
l R
1+
5 s a
H
H0
4
,
(4)
where l is the liquid density and 1.25 is a constant indicating the coupling
efficiency between the rim and fluid displacements. Similarly, frequencies of
higher modes can be calculated by considering a radial position proportional to
cos n and with m nodal parallels [1]:
f(n,m)
2.2
1
=
12
3Y
s
1/2
1/2
a (n2 1)2 + (mR/H0 )4
R2
1 + 1/n2
(5)
Tibetan bowls
Four different antique bowls of different sizes have been studied (Fig. 1). They
are referred to as Tibet 1, 2, 3 and 4 and their physical characteristics are
reported in Table 1.
Bowl name
Symbol
f(2,0) (empty) (Hz)
Radius R (cm)
Thickness a (cm)
Mass (kg)
Volume (cm3 )
Density (kg/m3 )
Tibet 1
236
7.5
0.34
0.690
76
9079
Tibet 2
187
8.9
0.38
0.814
97
8366
Tibet 3
347
6
0.31
0.306
35
8754
Tibet 4
428
5.9
0.37
0.312
33
9372
Table 1: Physical properties of the four Tibetan bowls used in our study.
When a bowl is struck or rubbed, the sound emitted by the resulting bowl
vibrations is recorded with a microphone and a fast Fourier transform performed on the signal. Different peaks clearly appear in the frequency spectrum,
corresponding to the bowls different vibrational modes. Fig. 3a indicates the
frequency spectrum generated by striking the empty bowl Tibet 4. When the
bowl is rubbed with a leather wrapped mallet, the lowest mode is excited along
with its harmonics, an effect known as a mode lock in [43]. The frequency
spectrum of Tibet 4 when rubbed by a leather mallet is presented in Fig. 3b.
Due to the bowl asymmetry, two peaks separated by several Hz arise and a
beating mode is heard. This split is highlighted in a magnification of the first
peak f(2,0) in the inset of the figure. The deformation shapes are the same with
both frequencies but there is horizontal angular shift observed to be /4 between them for the fundamental modes (2, 0) and /2n for other (n, 0) modes.
Finally, we note that, owing to the relative squatness of the bowls and the associated high energetic penalty of modes with m 6= 0, only modes (n, 0) were
excited; henceforth, such modes are denoted simply by n.
a)
0.14
b)
(3,0)
0.12
0.07
(2,0)
0.07
0.06
0.06
(2,0)
0.1
(4,0)
0.08
relative magnitude
relative magnitude
0.05
(6,0)
0.06
(2,0)
0.04
(5,0)
0.02
0
0.04
0.05
0.03
0.02
0.04
0.01
0
410 420 430 440 450 460
0.03
0.02
0.01
1000
2000
3000
4000
5000
6000
f(Hz)
1000
2000
3000
4000
5000
6000
f(Hz)
Figure 3: a) Frequencies excited in the bowl Tibet 4 when struck with a wooden
mallet. The different peaks correspond to the natural resonant frequencies of the
bowl, and the associated deformation modes (n, m) are indicated. b) Excited
frequencies of the bowl Tibet 4 when rubbed with a leather mallet. The first
peak corresponds to the deformation mode (2,0) and subsequent peaks to the
harmonics induced by mode lock-in. In the inset, a magnification of the first
peak provides evidence of its splitting due to the asymmetry of the bowl.
We can also find a relation between the different mode frequencies. Since the
speedof the bending wave is proportional to the
square root of the frequency,
v f and since = v/f , we expect 1/ f a. For the mode n, we thus
have 2R = nn . The frequency of this mode n should thus be proportional to
f
n2 a
.
R2
(6)
In Fig. 4a, resonant frequencies of the 4 bowls are plotted as a function of their
corresponding mode n and a power 2 curve is fit onto each curve. In Fig. 4b, we
collapse all these curves onto a line by plotting the characteristic speed f R2 /a
as a function of the mode n, thus validating the proposed scaling (6).
It is readily observed that the resonant frequencies decrease when liquid is
poured into a vessel. In the inset in Fig. 5a, we report the measurements of the
fundamental frequency of the bowl Tibet 1 as a function of the dimensionless
2
liquid height H/H0 . In Fig. 5a, we report (f0 /fH ) as a function of (H/H0 )4 .
According to (4), the slope of this curve gives the ratio 5 lsRa . In Fig. 5b, we
present the dependence of (1 + 1/n2 )(R2 /a)2 f 2 on (n2 1)2 for the deformation
modes n =2 through 6 of the bowls Tibet 1, 2, 3 and 4. According to (5),
1 Y
the slope should be equal to 48
2 . For each value of the abscissa there are
s
4 measurements corresponding to the 4 Tibetan bowls. Data points from the
7
a)
7000
104
b)
Tibet 2
Tibet 1
Tibet 3
Tibet 4
all
Tibet 2
Tibet 1
Tibet 3
Tibet 4
6000
f R /a (m/s)
4000
3000
1000
f(Hz)
5000
2000
1000
0
10
100
12
8 9 10
Figure 4: a) Measurements of the resonant frequencies of the different deformation modes n for the 4 different Tibetan bowls. b) Characteristic speed f R2 /a
as a function of deformation mode n for the 4 different bowls. Data are fit by
a power 2 curve indicating that fn n2 a/R2 , consistent with Eq. 6.
different Tibetan bowls overlie each other, especially at low n, indicating that
all bowls have the same value of ratio Y /s , and so are likely made of the same
material. A linear fit gives Y /s = 8.65 106 Pa.m3 /kg.
a)
7
b) 2.5 10
1.6
1.5
2 10
2
(1+1/n ) R4/a2 f
1.3
240
230
220
1 107
1.1
210
5 106
200
1
0.9
1.5 107
250
1.2
f (Hz)
(f0/fH)2
1.4
Tibet 1
Tibet 2
Tibet 3
Tibet 4
B
190
0
H/H
0.2
0.4
0.6
(H/H )
0.8
200
400
600
(n2-1)2
Fluid dynamics
3.1
Faraday waves
with am (t) being the oscillating amplitude of the eigenmode m. Benjamin and
Ursell [16] demonstrate that the coefficients am (t) satisfy:
2
a
m + m
(1 2 cos 0 t)am = 0,
where
2
m
=
gkm +
3
km tanh (km h)
(8)
(9)
2(1 + Bo1 )
(10)
g
is the dimensionless forcing acceleration, and Bo = k
2 is the Bond number.
km = 2/m is the wave number of the mode m, and m is the corresponding
wavelength.
In the absence of forcing, = 0, (8) describes a simple harmonic oscillation
with frequency prescribed by (9) corresponding to the free surface vibrations
of the liquid. When > 0, (8) describes a parametric oscillator as the forcing
term depends on time. The resulting Faraday waves can be either capillary
or gravity waves according to the magnitude of Bo; specifically, the former and
latter arise in the respective limits Bo 1 and Bo 1. Equation 8 is known as
the Mathieu equation and cannot be solved analytically since one of the terms
is time dependent. However, as the forcing is periodic, Floquet theory can be
applied to show that an inviscid fluid layer is always unstable to Faraday waves
with a frequency F that is half the forcing frequency 0 = 2F [16]. In the
deep water (kh 1), capillary wave (Bo 1) limit, we can infer from (9) a
Faraday wavelength :
F = (2)1/3 (/)1/3 (f0 /2)2/3 .
(11)
To incorporate the influence of the fluid viscosity, one can add to (8) a
phenomenological dissipation term proportional to the velocity [18]:
2
a
m + 2 a m + m
(1 2 cos 0 t)am = 0,
(12)
dm F (/)1/3 f0
(14)
in the capillary wave limit. Droplet ejection arises when the destabilizing inertial
driving force mg (with m 3F ) exceeds the stabilizing surface tension force
F . This implies, via (14), a threshold acceleration that scales as:
d
1
4/3
(/)1/3 f0
g
(15)
3.2
Experimental technique
The experimental setup used for studying the surface waves generated within
the Tibetan bowls is presented in Fig. 6. A loudspeaker is set in front of the
bowl, its signal received from a signal function generator then amplified. When
the applied signal frequency is close to one of the bowls resonant frequencies,
it oscillates in the corresponding deformation mode. Conveniently, with this
method, we can select a single deformation mode. Recall that by striking or
rubbing the bowls, we excited several modes (Fig. 3): moreover, rubbing induced
a rotational motion that followed the mallet. We can now examine the vibrationinduced flows in a controlled fashion.
In order to extend the range of natural frequencies, we consider Tibetan
bowls, wineglasses and soda cans (with their tops removed), whose resonant
10
Loudspeaker
Accelerometer
11
the initial segment length. Since the deformation is small, we can approximate:
1
1
3x
rc R + R2 where x is the radial displacement of the wall [1]. The strain gauge
gives the measurement of the convex longitudinal extension of the rim = l/l0 .
We can thus deduce the maximal radial amplitude of the rim deformation [1]:
2
xm = 23 Ra .
3.3
Surface waves
Two different liquids were used in our experiment, pure water with density
w = 1000 kg/m3 , viscosity = 1 cSt and surface tension = 72 mN/m, and
Dow Corning silicone oil with o = 820 kg/m3 , = 1 cSt and = 17.4 mN/m.
The fluid depth and resulting natural deformation frequencies of the bowl were
varied. Specific deformation modes of the bowls were excited acoustically. We
proceed by reporting the form of the flow induced, specifically, the evolution of
the free surface with increasing rim forcing.
In Fig. 7, we present snapshots of the bowl Tibet 1 resonating in its fundamental deformation mode with different when it is completely filled with
water. The loudspeaker produces a sinusoidal sound at a frequency f0 = 188 Hz
that corresponds to the mode (2,0) with four associated nodes and antinodes.
The vibration of the water surface is forced by the horizontal oscillation of the
rim. When the amplitude of the rim oscillation is small, axisymmetric progressive capillary waves with frequency commensurate with the excitation frequency
appear on the liquid surface. Though almost invisible to the naked eye, they can
be readily detected by appropriate lighting of the liquid surface (Fig. 7a). When
is further increased, relatively large amplitude circumferential standing waves
appear at the waters edge (Fig. 7b). These standing ripples, aligned perpendicular to the bowls edge, are spaced by approximately a Faraday wavelength
F , as defined in (11). Moreover, their frequency is half that of the axial waves,
indicating that these waves correspond to classic cross waves or, equivalently,
edge-induced Faraday waves [14]. More complicated wave modes appear at
higher excitation amplitude (Fig. 7c). At sufficiently high , the Faraday waves
break, and water droplets are ejected from the edge of the vessel (Fig. 7d),
specifically from the anti-nodes of the oscillating wall. The ejected droplets
may bounce, slide, and roll on the water surface before eventually coalescing.
One of our bowls (Tibet 2) resonates readily in both modes (2,0) and (3,0).
When completely filled with water, the resonant frequencies of its (2,0) and
(3,0) modes are f = 144 Hz and f = 524 Hz, respectively. In Fig. 8, we observe
the progression of the surface waves with increasing amplitude for each of these
modes. Note that for the mode (3,0), since the frequency is higher, the wavelengths are shorter. Moreover, the sound amplitude needed to produce surface
waves is necessarily higher for mode (3,0) than (2,0).
The transition from axisymmetric capillary waves to Faraday waves arises at
a critical acceleration F readily measured by the accelerometer. This threshold
was measured as a function of the forcing frequency, the latter having been
tuned to excite the fundamental deformation modes of the bowls with different
liquid levels. We also measured this acceleration threshold for the deformation
12
a)
b)
c)
d)
Figure 7: Evolution of the surface waves in Tibet 1 bowl filled with water and
excited with a frequency f = 188 Hz corresponding to its fundamental mode
(2,0). The amplitude of deformation is increasing from left to right: a) = 1.8,
b) = 2.8, c) = 6.2, d) = 16.2.
a)
b)
c)
d)
e)
f
)
g)
h)
Figure 8: Evolution of the surface waves in Tibet 2 bowl filled with water. a-d)
Mode (2,0) is excited with a frequency f = 144 Hz. e-h) Mode (3,0) is excited
with f = 524 Hz. The amplitude of deformation is increasing from left to right.
For the deformation mode n = 2, a) = 1.0, b) = 1.7, c) = 5.0, d)
= 12.7. For the deformation mode n = 3, e) = 5.7, f) = 7.7, g) = 15.1,
h) = 33.8.
mode (3,0) of the bowl Tibet 2. Higher frequencies were explored with three
different wineglasses filled to different levels using the strain gauge system. All
the measurements with silicone oil of viscosity 1 cSt are presented in Fig. 9
(lower curve). Consistent with (13), the data suggest a dependence F f 5/3 .
In Fig. 10a, we report our measurements of F as a function of frequency for
both silicone oil and distilled water. Each data set is fit by a 5/3 power law.
Prefactors of 3.5 104 for water and 1.7 104 for 1 cSt silicone oil were
inferred.
13
3.4
Surface fracture
When the Faraday waves become sufficiently large, they break, leading to droplet
ejection. A second critical acceleration can thus be measured, d , above which
droplets are ejected from the surface. The droplet ejection starts with very
few droplets ejected, then the ejection rate increases with forcing amplitude.
Our criterion for onset is that at least two droplets are ejected in a 15 second
time interval. The dependence of d on f is presented in Fig. 9 (upper curve)
for bowls and wine glasses filled with different levels of 1 cSt silicone oil. The
drop ejection threshold scales as d f 4/3 , which is consistent with the scaling
presented in (15).
102
Faraday waves
+
jumping droplets
101
Faraday
waves
100
Capillary waves
10-1
102
103
f(Hz)
Figure 9: Phase diagram indicating the critical acceleration F = 2 /g above
which axial capillary waves give way to Faraday waves, and d above which the
Faraday waves break. The , , and symbols correspond, respectively, to
the measurements made with the bowls Tibet 1, 2, 3 4 and the to wineglasses
filled with different levels of 1 cSt silicone oil. The first acceleration threshold
(lower curve) scales as F f 5/3 and the second (upper curve) as d f 4/3 .
In Fig. 10b, our measurements of the dependence of d on f are reported for
bowls filled with different levels of either 1 cSt silicone oil or water. The two d
curves collapse when we use the scaling law (15) expected to apply for vertical
forcing. Specifically, we find
d 0.23(/)1/3 f 4/3 .
14
(16)
102
104
Faraday waves
0.23
103
Faraday waves
+
jumping droplets
(/)
1/3
101
b)
100
0.3
Capillary waves
10-1
102
0.2
102
102
103
f(Hz)
Faraday waves
103
f(Hz)
a)
10-1
b)
Pure water
1cSt Silicone oil
B
B
185 Hz
281 Hz
354 Hz
0.8
1/3
185 Hz
1.5
1
0.5
d (/)
0.87
0.5
1.5
d (mm)
CDF
0.4
281 Hz
0.6
354 Hz
2.5
0.2
10-2
0
0
0.5
1.5
102
d (mm)
103
f(Hz)
3.5
Bouncing droplets
With a more viscous fluid (e.g. 10 cSt silicone oil), the waves are less pronounced, and the fluid is more strongly coupled to the vibrating sidewalls;
specifically, more of the surface oscillates up and down near the walls antinode. When a droplet of the same liquid is deposited on the surface, it may
bounce, levitated by the underlying wave field. Such sustained levitation was
not observed in the Tibetan singing bowl with liquid viscosities lower than 10
cSt.
In Fig. 12a, we present a still image of a drop of diameter 0.5 mm bouncing
on the liquid surface inside the bowl (Tibet 1) resonating at a frequency of
f0 = 188 Hz. The drop has been made by dipping then extracting a syringe
needle, on the tip of which a capillary bridge forms and breaks, leaving a drop
that bounces near the oscillating rim. A movie of the bouncing was recorded,
then vertical slices of each image through the droplet centerline juxtaposed. We
can thus construct an image illustrating the dynamics of the droplet (Fig. 12b).
16
In this case, the drop experience two bounces of slightly different amplitude
while the liquid surface (and the rim) oscillates twice. When a smaller droplet
(of diameter 0.35 mm) is placed on the oscillating surface, the bouncing motion
can be more complex. In Fig. 12c, we see that the droplet bounces only once
during three oscillations of the surface. Fig. 12d illustrates the corresponding
trajectory. We note that the bouncing motion of sufficiently small drops can
become chaotic.
We sought to sustain walking droplets in our system. Once the horizontal
forcing amplitude is just above the Faraday wave threshold F , circumferential
Faraday waves are sustained at the edge of the vessel. Then, when is increased,
Faraday waves propagate progressively towards the center of the vessel, their
amplitude damped by viscosity. Beyond these waves, the liquid surface was
quiescent unless perturbed by a bouncing droplet, in which case it could sustain
a field of Faraday waves. This phenomenon was observed in the bowl Tibet
2 almost fully filled with 10 cSt silicone oil (f0 =140 Hz). Just beyond the
Faraday waves, a bouncing droplet of diameter 500 m was made such that
its bouncing frequency corresponded to the Faraday wave frequency, that is,
half the frequency of the vibrating rim. Such droplets were unable to excite
sufficiently large Faraday waves to enable them to walk. We note that the usual
range of walker diameters is between 650 m and 850 m [33, 34]: their mass
is thus 4 times larger than that of our drops.
Conclusion
17
18
threshold.
Acknowledgements
The authors thank Rosie Warburton for bringing this problem to our attention, and for supplying the bowls for our study. We are grateful to Jim Bales (of
MITs Edgerton Center), Barbara Hughey and Stephane Dorbolo for assistance
and fruitful discussions. Denis Terwagne thanks the University of Li`ege and
the GRASP for financial support. This research was supported by the National
Science Foundation through grant CBET-0966452.
References
[1] A. French, In vino veritas - a study of wineglass acoustics, Am. J. Phys.
51, 688 (1983).
[2] K. Chen, C. Wang, C. Lu and Y. Chen, Variations on a theme by a singing
wineglass, Europhys. Lett. 70, 334 (2005).
[3] Y.-Y. Chen, Why does water change the pitch of a singing wineglass the
way it does?, Am. J. Phys. 73, 1045 (2005).
[4] G. Jundt, A. Radu, E. Fort, J. Duda, H. Vach and N. Fletcher, Vibrational
modes of partly filled wine glasses, J. Acoust. Soc. Am. 119, 3793 (2006).
[5] M. Courtois, B. Guirao and E. Fort, Tuning the pitch of a wine glass by
playing with the liquid inside, Eur. J. Phys. 29, 303 (2008).
[6] T. Arane, A. Musalem and M. Fridman, Coupling between two singing
wineglasses, Am. J. Phys. 77, 1066 (2009).
[7] R. Apfel, Whispering waves in a wineglass, Am. J. Phys. 53, 1070 (1985).
[8] L. Rayleigh, Theory of sound (Cambridge University Press, 1945).
[9] T. Rossing, Wine glasses, bell modes and lord rayleigh, Phys. Teach. 28,
582 (1990).
[10] T. Rossing, Acoustics of the glass harmonica, J. Acoust. Soc. Am. 95, 1106
(1993).
[11] S. V. Joubert, T. H. Fay and E. L. Voges, A storm in a wineglass, Am. J.
Phys. 75, 647 (2007).
[12] O. Inacio, L. Henrique and J. Antunes, The dynamics of tibetan singing
bowls, Acta Acust United Ac 92, 637 (2006).
[13] O. Inacio, L. Henrique and J. Antunes, The physics of tibetan singing bowls
(2004), Revista de Ac
ustica. 35, 33 (2004).
19
[31] T. Gilet, D. Terwagne, N. Vandewalle and S. Dorbolo, Dynamics of a bouncing droplet onto a vertically vibrated interface, Phys. Rev. Lett. 100, 167802
(2008).
[32] S. Dorbolo, D. Terwagne, N. Vandewalle and T. Gilet, Resonant and rolling
droplet, New J. Phys. 10, 113021 (2008).
[33] S. Protiere, Y. Couder, E. Fort and A. Boudaoud, The self-organization
of capillary wave sources, Journal of Physics-Condensed Matter 17, S3529
(2005).
[34] A. Eddi, D. Terwagne, E. Fort and Y. Couder, Wave propelled ratchets and
drifting rafts, Eur. Phys. Lett. 82, 44001 (2008).
[35] T. Gilet and J. W. M. Bush, The fluid trampoline: droplets bouncing on a
soap film, J. Fluid Mech. 625, 167 (2009).
[36] A. Eddi, E. Sultan, J. Moukhtar, E. Fort, M. Rossi and Y. Couder, Information stored in faraday waves: the origin of a path memory, J. Fluid
Mech. 674, 433 (2011).
[37] Y. Couder and E. Fort, Single-particle diffraction and interference at a
macroscopic scale, Physical Review Letters 97, 154101 (2006).
[38] A. Eddi, E. Fort, F. Moisy and Y. Couder, Unpredictable tunneling of a
classical wave-particle association, Physical Review Letters 102, 240401
(2009).
[39] E. Fort, A. Eddi, A. Boudaoud, J. Moukhtar and Y. Couder, Path-memory
induced quantization of classical orbits, Proc. Natl. Acad. Sci. 107, 17515
(2010).
[40] J. W. M. Bush, Quantum mechanics writ large, Proc. Natl. Acad. Sci. 107,
17455 (2010).
[41] R. Perrin, T. Charcley, H. Banut and T. D. Rossing, Chladnis law and the
modern english church bell, Journal of Sound and Vibration 102, 11 (1985).
[42] L. Kinsler, A. Frey, A. Coppens and J. Sanders, Fundamentals of acoustics
(3rd edition) (Wiley, 1982).
[43] A. Akay, Acoustics of friction, J. Acoust. Soc. Am. 111, 1525 (2002).
[44] S. Douady, Experimental study of the faraday instability, J. Fluid Mech.
221, 383 (1990).
[45] T. Donnelly, J. Hogan, A. Mugler, N. Schommer, M. Schubmehl, A. Bernoff
and B. Forrest, An experimental study of micron-scale droplet aerosols produced via ultrasonic atomization, Physics of Fluids 16, 2843 (2004).
21