Home Search Collections Journals About Contact us My IOPscience

Resonances of a Violin Body Studied by Hologram Interferometry and Acoustical Methods

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1970 Phys. Scr. 2 243

(http://iopscience.iop.org/1402-4896/2/6/002)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 198.101.234.89
This content was downloaded on 31/05/2015 at 09:12

Please note that terms and conditions apply.

 Physica Scripta. Vol. 2, 243-256, 1970

Resonances of a Violin Body Studied by Hologram

Interferometry and Acoustical Methods
E. Jansson, N.-E. Molin and H. Sundin

The Institute of Optical Research and The Department of Speech Communication, Royal Institute of Technology, Stockholm

Receiced Nocember 12, 1970: wcised aersiori receiued November 30, 1970

Abstract Acoustical Society, a society devoted to violin research. The

modern computer has also been used t o analyze and synthesize
Resonances of a violin body studied by hologram interferometry and acoustical
violin tones [lo].
merhods. E. Jansson, N.-E. Molin and H. Sundin (The Institute of Optical
Research and the Department of Speech Communication, Royal Institute of The investigations of the violin are not made merely to satisfy
Technology, Stockholm, Sweden). the curiosity of the scientist. The goal of most investigations has
Physica Scripta (Sweden) 2, 243-256, 1970. been to find correlations between the tonal qualities and physical
properties, and to be able to perform objective tests of a violin’s
Hologram interferometry is applied t o study resonances of a violin body. A
violin is made and after each major step, modes of vibration are studied. The
quality [ll]. With a good understanding of the physics of the
top plate is studied in six steps: without and with f-holes, with f-holes and violin and with the just mentioned correlation known, acoustical
bass bar, with an artificial sound post, in the assembled instrument without test methods can be of great help to the violin maker. Electro-
and with strings. The back plate is studied in four steps: without and with acoustical methods are frequently used by Carleen Hutchins as
artificial sound post, in the assembled instrument without and with strings.
guidance in the making [ 12, 131 and in her opinion we have today
Interferograms of the lowest five t o seven modes ranging from about 400
to 1 300 Hz are shown. The top plate is shown to be divided into plate and knowledge to build good but not excellent fiddles.
body resonances. The back plate is acting more as a unit together with the The whole sound producing system has been studied [14, 151,
ribs thus mainly giving body-resonances. Deformations of the violin are as well as links of the system, such as the function of the strings,
studied when strings are pressed against the fingerboard, by double exposure bridge, air volume, plates and radiation [16]. However, the
hologram interferometry. The lowest air resonance of the sounding box
and frequency responses are acoustically measured. In playing the unvar-
methods of studying plate vibrations, as Chladni patterns for
nished violin the tone is found powerful and even. instance, have not been accurate enough or cumbersome and not
giving pictures of the vibrations in real-time, as scanning over
the vibrating plate by a probe [17, 181. This lack of an accurate
and informative method has seriously limited the possible studies
1. Introduction and the understanding of the violin.
The quality of a violin depends to a great extent on the vibration
The violin is a challenging, complicated, mechanical system [ l , 21 properties of its plates. I n 1968 K. A. Stetson suggested the use
which has inspired many physicists to examine it closely using of hologram interferometry to study the vibrations for
the best means within reach. There are still links of the system, stringed instruments [19].
however, which are not understood and which, today, still attract This technique was first applied a t the Institute of Optical
the interest of physicists, engineers, etc. Research in Stockholm to study plate resonances of a treble viol
F. Savart (1791-1841), in cooperation with the violin maker and to design an improved instrument [20]. Thereafter the
J. Vuillaume, was one of the first to examine the violin by physical plate resonances of a guitar have been measured by hologram
methods [3, 41. Savart built a violin of simplified geometry, flat interferometry and compared to acoustical measurements [21].
plates, and trapezoidal ribs, studied the plate’s vibrations by The resonances were adjusted by altering the bracing of the
means of Chladni patterns and was able to identify some reso- guitar in order to improve the instrument. Reinicke and Cremer
nances of the ordinary violin. H. Helmholtz (1821-1894) studied reported the use of hologram interferometry for the study of
the bowed string and observed the harmonics of musical tones by a violin [22].
means of tuned resonators [5]. Later more accurate and detailed I n this work, hologram interferometry is applied to study reso-
studies of the bowed string were made by C. Raman (1888-1970), nances of a violin, using the experience of earlier investigations.
who also made thorough studies of the complete instrument using The goal has been to gather knowledge about the general be-
hand and automstic bowing [6]. haviour of a violin, to obtain a first quantitative understanding
Electroacoustical methods gave new powerful analysis tools of a violin’s vibrations, and to provide the base for a physical
to the scientists. These were early used by H. Backhaus and H. description of the mechanism of the sound generation in a violin,
Meinel to obtain frequency responses and vibration patterns of which in the future may be correlated to its tonal qualities.
the violin [7, 81. F. Saunders and co-workers started a long and The program was to make a violin and to investigate its vibra-
extensive study of the violin [9]. Saunders also founded the Catgut tion properties after each major step in making by means of im-
Physica Scripta 2
244 E. Jansson et al.

up from the sounding box by the bridge. The tailpiece is held

by a string around the end button.
The sounding box consists of the top plate, the ribs (sidewalls)
and the back plate. The top and back plate are usually made of
two bilaterally symmetrical pieces of spruce and maple, respec-
tively. The plates are arched and vary in thickness, the top plate
from about 2 mm close to the ribs to about 3.5 mm in the middle
of the plate, and the back plate from 2.5 to 4 mm, see photo-
graphs of the plates in translucent light (Fig. 2). In the top plate
the f-holes are cut. The plate is strengthened by the bass bar, which
is glued underneath the plate under the left foot of the bridge
usually almost in parallel with the glue-joint of the plate. The
plates are glued at the ribs, usually of maple, all way round. The
top plate is supported by the sound post, which is squeezed in
between the top and back plate closely behind the right foot of
the bridge. The violin is usually put together, varnished, tested
by playing, taken apart and adjusted several times before con-
sidered finished.
The tone of a violin is produced mainly according to the follow-
ing model. The string is set into vibration and the vibrations are
transmitted by the bridge, neck and tailpiece to the sounding
box. The walls and the enclosed air volume of the sounding box
are set into vibration and sound is radiated from the walls and the
Fig. I . Front view and section of a violin showing neck, finger board,
f-holes. Little sound is radiated directly from the strings because
tailpiece ( T ) , and four strings, top plate ( T P ) , back plate ( B P ) , f-holes (F), of their small area.
bridge ( B ) , sound post (SP),bass bar (BB) and lower end block (LB).

3. Hologram interferometry
proved holographic equipment. We studied the shapes and the
frequencies of the lowest five to seven vibration modes and how In two papers, in 1965, Powell and Stetson presented the prin-
they were changed by the f-holes, by the bass bar etc. The opti- ciples of hologram interferometry [23, 241. Since then, hologram
cal measurements were supplemented by acoustical measure- interferometry has grown into one of the most important fields
ments. of holography [25, 26, 271. By means of hologram interferometry,
In the following a short description is given of the violin. There- it is possible to measure with interferometric accuracy, deforma-
after the hologram interferometry technique is described. Finally tions of an object of any shape and surface roughness, even of
the experiments are presented and discussed. objects such as wood. The deformation of the whole surface is
obtained in one record without disturbing it by polishing or by
a probe.
2. The violin We will start with a description of the equipment and the holo-
graphic technique used. The reader interested in more detailed
A violin mainly consists of a sounding box, neck with finger general information is referred to [25, 26, 271, in combination of
board and four strings (Fig. 1). The strings are stretched from modes to [28], in the use of the normal mode theory [20] and in
the upper end of the finger board to the tailpiece and are held a method to measure bending moments to [29].
A schematic diagram of the system used is shown in Fig. 3.
The thin light beam from a laser, L, is reflected by a mirror, M,
through a shutter, S, to a beamsplitter, BS, where the beam is
split up into two beams. The transmitted beam, the object beam,

Fig. 3. The holographic system used. Abbreviations: laser (L), first surface
Fig. 2. Photographs in translucent light of (0) thc top plate, (b) the back mirrors ( M ) , glass wedge beamsplitter (BS), beam expander lens-pinhole
plate. The density in the pictures gives an estimate of the thickness distribu- system ( L P ) , shutter (S), liquid-gate hologram plate holder and holder for
tion in the plates. The left dark line is the bass bar and the central one the holograms o n 35 mm film ( H ) and object violin (0)clamped to heavy wooden
glue-joint. Some driving points also show up. jig ( J ) .
Physica Scripta 2
 Resonances of a violin body studied by hologram interferometry 245

is reflected by a mirror, M ythrough a beam expander, LP, ex-

panding the beam to cover the whole object, and is thereafter
reflected by the object, 0, held in a heavy jig, J, to a liquid-gate
hologram plate holder and a camera back for 35 mm film, H.
The beam reflected at the beam splitter, BS, the reference beam,
is reflected by a mirror, M y passed through a beam expander,
LP,and is finally reflected by mirrors, M yto the hologram holders,
H.
The laser used is a He-Ne Spectra-Physics, Inc., mod. 124
with an output of 15 mW at a wavelength of 6 328 A. An absorb-
ing filter was used in either the object or the reference beam to
obtain correct beamratios in every case. The beam splitter is
arranged to direct approx. 85% of the laser flux to the object
beam. The beam splitter is a weak prism which changes the cross-
section of the transmitted beam from circular to elliptical in order
to provide uniform illumination over the whole violin plate. A
beam splitter with continuously variable reflectance and trans-
mittance ratio would have been useful for adjusting the beam
intensity ratios necessary for the different techniques of holo-
gram interferometry.
The beam expander for the object beam consists of a micro-
scope objective ( x 25) with a pinhole (20 pmj. The corresponding
data for the reference beam are ( x 10) and (10 pm).
All equipment except the shutter was placed on a granite slab,
1.1 m x 2.1 m x 0.1 m, resting on a damped spring mattress. The
shutter was mounted on a separate stand on the floor.
The violin is an element of an interferometer and therefore
humidity and temperature should be kept constant since un-
varnished woodplates are extremely sensitive to changes in humi-
dity. The noise-level in the room should be low as the plates are
easily excited by sound.
The experimental set up enables simple ahd rapid use of both
real-time and time-average hologram interferometry [23, 241.
Time-average holograms cannot be evaluated until the hologram Fig. 4 . Real-time hologram interferograms of the back plate obtained by a
is processed. In vibration analysis, it would be necessary there- camera looking through the liquid gate. ( U ) the plate at rest 3 min after holo-
gram exposure; (6) the plate excited at the lower right at 960 Hz (compare
fore, to record extensive series of holograms to achieve the right
time-average interferogram in Fig. 12c); (c) the plate at rest 4 min after
excitation frequency and amplitude for relevant vibration modes. hologram exposure (after that of b); (d)the plate at rest 20 min after holo-
Here real-time hologram interferometry offers a valuable guidance gram exposure and rough experimenting. The exciters are placed in upper
as the vibration is studied at the event. The draw-backs, however, middle and lower right.
are again the time interval between hologram recording and
evaluation, now in real-time hologram interferometry, due to the in rest reconstructed by the hologram in the liquid gate and the
photographic process and the time-consuming difficulties to object field transmitted through the hologram at the moment
relocate the hologram into its initial position. During this time the photograph was taken. An interferogram obtained 3 min
the wooden plates will undergo deformations due to changes in after exposure of the hologram is shown in Fig. 4a. Two black
humidity and temperature. fringes can be seen inside the plate. No fringes are cutting the
Therefore an integrated real-time hologram interferometer edges of the plate but the plate itself is slightly deformed after
around a rapid photographic process, has been designed for the hologram exposure. The deformation represents the plate’s
these experiments [30]. The photographic plate is fixed in posi- “memory” of a previous experiment. The plate had been driven
tion in the liquid-gate hologram plate-holder, exposed presoaked in the center of the inner black fringe on the opposite side of the
in rapid developer and evaluated in the fixing bath. The time plate and is slowly creeping into an equilibrium state after former
needed between exposure and evaluation is less than 3 min. Expos- deformation. In Fig. 4b, an interferogram obtained shortly after-
ure time is shortened to about 113 compared with exposure of wards is shown when the plate is excited at 960 Hz using the earlier
photographic plates in the usual dry state; with Agfa-Gevaert mentioned driving point. In the lower part of the plate two anti-
Scientia 10E70 plates the exposure time is 1 sec. A beam ratio of nodes on each side of a vertical nodal line appear. The interference
about 1.5 is used. In the evaluation of the real-time fringes patterns in this interferogram are quite complicated and consist
the intensity of the reference beam, which now acts as the recon- of two sets of patterns multiplied [30]-a sinusoidal fringe system
struction beam, is increased about 5 times by removing an absorb- due to static deformation as seen in Fig. 4 a and c, and a zero-
ing filter in order to match the intensities of the reconstructed order Bessel function distribution of fringe patterns due to
field and the object field. vibration. Fig. 4c, obtained 4 min after the hologram exposure,
As an example of the real-time method used, four real-time shows a fringe system similar to Fig. 4a, but the deformation is
hologram interferograms are shown in Fig. 4, obtained with a increased especially around the driving point. Fig. 4d, finally,
camera, placed behind the liquid gate thus replacing the observer. shows the plate at rest after 20 min rough experimenting after
This is the interference between the replica of the field of the object the hologram exposure. The deformation is maximum at the
Physica Scripta 2
246 E. Jansson et al.

only normal modes. This was done by observing, in real-time,

that the nodal lines did not move while sweeping through a
resonance and that no closed fringes contained a nodal line or
point. If not, the driving points were moved. For some modes,
however, especially when the whole instrument was vibrating, i.e.
a body resonance, or when the modes are close in frequency, it
was not always possible to record them distinctly and separately.

4. Preparation of the violin for testing

During the experiments the instrument should stand on the

hologram table with the highest possible stability to get hologram
recordings not disturbEd by displacements and yet still be an
accurate simulation of a violin in action. Therefore a heavy jig
of wood, 0.75 m ~0.35m ~ 0 . 0 8m, was constructed. A hole
slightly larger and of the same shape as a violin plate (as seen in
the front view of Fig. 1) was cut in the middle of the jig and
special clamps were made for the clamping of the instrument.
This enabled driving at any point on the plates and clamping at
any point along the edges of the assembled violin’s plates. The
jig was later modified to allow clamping of the neck.
The plates were excited by an electro-magnetic system. The
signal from a sine wave oscillator was fed to one of the two input
terminals of a stereo amplifier and via a phase shifting circuit
to the other. Each amplifier output was connected to an electro-
magnet. Small iron strips were fastened at the plate by sealing
wax or double-sided adhesive tape. The electro-magnets excited
Fig. 5. Back view of the back plate mounted to the jig. The exciters are the object through a 1 mm air gap separately or together with
mounted and the artificial sound post is visible. suitable phase and amplitude between the driving signals, which
often is useful to separate modes in mode combinations [28, 201.
The static force and the weight of the iron strips, about 0.1 g
driving points (on the opposite side of the plate, unseen)
each, did not seem to affect the resonance frequencies within the
in the lower right and upper middle part of the plate. The defor-
measuring accuracy.
mations are caused by changes in air humidity and the mechanical
When the violin plates were tested alone, they were glucd to
forces applied. the ribs, which in turn were glued to a thick wood plate firmly
By sweeping the frequency of excitation a maximum response bolted to the jig. A large hole was cut in the wood plate to allow
of the plate is easily found by observations in real-time (cf. Fig.
driving at the inner side of the violin plate (Fig. 5).
4b). The amplitude distribution is obtained, but, due to the When testing the assembled instrument it was found that a
static deformations the analysis of the real-time hologram inter-
clamping at the lower end block and the neck was not stable
ferograms are often complicated, especially when large static enough. Therefore the assembled instrument was usually clamped
deformations and vibrations are involved. A compound recording
at three places, one at the lower end block, two at opposite pairs
technique is used, therefore. When the resonance frequency is
of corner blocks. The clamping at blocks was chosen in order
found, a suitable vibration amplitude is set, and a time-average
to minimize the influence on the resonances. This was fulfilled
hologram is exposed on 35 mm Agfa-Gevaert Scientia 10E70
in the case of plate resonances but not in the case of body
film in a camera-back. As the unwanted deformations during the
resonances, where, especially the clamping at the corner blocks
exposure are small compared with the wavelength of the light
they will not show up in the time-average hologram interfero-
grams. In the reconstruction of the fields from this hologram, the
time-average of the field during the exposure is obtained. If
sinusoidally vibrating, the object will be covered with zero-order
Bessel function interference fringes. The time-averagz interfero-
gram corresponding to the real-time interferogram of Fig. 4b
is shown in Fig. 12c. The plate is divided by the bright vertical
nodal line into two vibrating regions. The antinodes are sur-
rounded by closed zero-order Bessel function fringes, each
fringe connecting points with almost equal magnitude of vibra-
tion. The magnitude is increasing with about 0.3 A, from one
fringe to the next starting from the nodal line.
In the following time-average hologram interferograms are
shown. The interference fringes can be regarded as equal magni-
tude lines. However, these lines do not necessarily represent lines
Fig. 6. Diagram relating frequency to the equally tempered scale. The
of equal phase. Some vibration patterns shown here may represent strings of the violin are tuned G (196 Hz), D (294 Hz), A (440 Hz), and E
combinations of modes [28]. The attempt was made to record (659 Hz).
Physica Scripta 2
 Resonances of a violin body studied by hologram interferometry 247

caused noticeable effects. In the body-resonances the whole

instrument, plates, ribs, and sometimes the neck, were vibrating.
The experiments were performed in the following way. First
driving and clamping places were chosen by looking at the speckle
patterns with the instrument illuminated by laser light [20],
from interferograms of earlier experiments, or in few cases by
trial and error. In the last experiments a speckle interferometer
was successfully used [31, 321, which was especially hdpful since
nodal areas and even iso-amplitude lines can rapidly be found
almost directly after driving points or clamping positions have
been moved. The speckle interferometer is used as a real-time
instrument and has proved to be a good complement to the
real-time hologram interferometer. The next step was to mount
the object, jig and drivers as stably as possible on the hologram
table and leave the equipment, especially the violin, to stabilize
for an hour or two. Thereafter, all evaluations were done in the
real-time interferometer. A resonance frequency and corresponding
Q-value were obtained by determining the relative amplitude of
vibration for constant excitation by counting number of fringes,
as a function of frequency, at three points around the resonance
peak where the maximum number of fringes occur. A resonance
curve was fitted through the points thus giving center frequencies
and Q-values. The resonance frequency, however, can usually be
read accurately enough directly on the oscillator at the point of
maximum amplitude. A frequency range from 70 to 1 300 Hz
was investigated with both drivers used separately and if necessary
together.
The five lowest modes were sought especially. At each reso-
nance one or several time-average holograms were recorded. The
time needed for the evaluation and recording was usually about
one hour and the same real-time hologram could usually be used
all the time. Finally the 35 mm film containing the time-average
holograms was processed and interferograms of the reconstructed
virtual image fields were photographed by a mirror-reflex camera
on Agfapan 400 Professional film.

5. Experimental results using hologram interferometry

Ribs, neck with fitted tuning pegs, and finger board were bought
ready-made. Roughly worked plates of sufficient thickness to
enable the wanted arching and thickness were selected and bought.
The arching was first worked out and thereafter the plates were
worked close to the appropriate thickness. The f-holes were cut
and a prestressed bass bar was glued in. In the last step the neck
and plates were glued to the ribs and the instrument was strung.
The plates were not varnished, and in order to obtain visible
interference fringes at the black finger board and tailpiece, these
were covered with white tape. The effect of the drivers was set
individually in order to make different modes to appear with a
suitable number of fringes. Hence, the vibration amplitudes Fig. 7. Time-average hologram interferograms of series (a), the top plate
without f-holes and bass bar. (a) 550 Hz (lo), (b) 800 Hz (lo), (c) 810 Hz
marked in different interferograms are not simply related to one (30), ( d ) 1000 Hz (30), (e) 1 110 Hz, (f) 1220 Hz.
another.
The plates were tested after major steps in the making, viz.
(h) the back plate in the assembled instrument without finger
(a) the top plate without f-holes and bass bar, board, strings and devices for stringing,
(b) the top plate with f-holes but without bass bar, (i) the top plate in the complete instrument,
(c) the top plate with f-holes and bass bar, ( j ) the back plate in the complete instrument, and
( d ) the top plate with f-holes, bass bar and artificial, immovable ( k ) deformation of the instrument when strings are pressed
sound post, against the finger board.
(e) the back plate,
( f ) the back plate with artificial, immovable sound post, Frequencies and Q-values of the modes are in the following given
(g) the top plate in the assembled instrument without finger in the figure captions. The Q-values, measured only for some
board, strings and devices for stringing, modes, are put in parentheses.
Physica Scripta 2
248 E. Jansson et al.

Fig. 8. Interferograms of series

(b), the top plate with f-holes but
without bass bar. (a) 420 H z (15),
(6) 600 H z (30), (c) 640 H z (IO),
(d) 880 H z (25), (e) 920 H z (35),
(f) 1050 H z (30), (g) 1090 Hz.

In Fig. 6 a diagram is given relating frequency to the equally (6) Top plate with f-holes but without bass bar
tempered scale. The strings of the violin are tuned (starting from Interferograms from series b are shown in Fig. Sa-g. NOWthe
the left one in Fig. 1) at G 196 Hz, D 294 Hz, A 440 Hz, and second and third modes have exchanged their appearances. The
E 659 Hz. mode with one vertical nodal line is obtained at lower frequency
than the one with a horizontal. A new mode is found (Fig. S f ) ,
( a ) Top plate without f-holes and bass bar one vertical and two horizontal nodal lines, at lower frequency
In Fig. 7a-f interferograms from series a are shown. The dark than the mode with three vibrating areas in the lower part of
fringes (rings) can be regarded as equal vibration amplitude the plate. The plate seems not to be firmly glued to the lower
lines. The amplitude difference between neighboring fringes is end block, which only seems to influence on the seventh mode
approximately 2 x lO-5 cm. The interferograms give the ampli- (Fig. Sg). The amplitude is maximum in most cases at the f-
tude distribution over the whole plate with high resolution. holes, which implies that the place and shape of the f-holes may
The first three modes are simple; the first mode consists of be crucial. The amplitude and area of vibration tend to increase
one vibrating area, the second, and third modes of two vibrating in the lower parts of the plate. The fourth mode, for instance,
areas divided by a horizontal and a vertical nodal line, respectively. is now clearly divided by a cross node.
However, in the third mode the complexity of the plate vibrations The plate divides more easily in vertical than in horizontal
already starts to appear. Only the lower part of the plate is vibrat- nodal lines. All mode frequencies are decreased (Fig. 19). This
ing. In the fourth mode the upper part of the plate is divided by a means that the effective stiffness is decreased, which seems logical,
vertical nodal line into two vibrating areas. But two vibrating as the top plate is partly cut free from the ribs.
areas also show up in the lower part. These may be residual in-
fluence of the third mode or, more probably, the fourth mode has
a cross node. The fifth mode consists of three vibrating areas (c) Top plate with f-holes and bass bar
divided by two horizontal nodal lines. The sixth mode has three Fig. 9a-f show interferograms from series c. The major appear-
vibrating areas divided by two vertical nodal lines in the lower ance of the modes compared with series b is not changed. However,
part and a vibrating area in the upper part, which may be a Fig. 9e may not represent a simple mode, but rather a combina-
residual of the fifth mode. tion of modes. The pattern is easily excited and the influence of
The waist of the plate seems to divide the plate into two parts, the bass bar is pronounced (cf. Figs. S f and 9 4 . The bass bar
the upper and the lower bouts as regards to vibration. increases the amplitude and the area of vibration in the upper
Physica Scripta 2
 Resonances of a violin body studied by hologram interferometry 249

Fi3. 10. Interferograms of series

(d), the top plate with f-holes, bass
bar and artificial, immovable
sound post. (a) 540 Hz (ZO), (b)
775 Hz (15), (c) 800 Hz (30), (d)
980 Hz (30), (e) 1 110 Hz (45).

Fig. 9. Interferograms of series (c), the top plate with f-holes and bass bar.
(a) 465 Hz, (b) 600 Hz, (c) 820 Hz, ( d ) 910 Hz, (e) 1040,Hz, (f)1090 Hz.

left part of the plate, but shows little influence on the lower parts. frequency differences occur in the first and the third mode, in
All mode frequencies are increased (cf. Fig. 19), which means which the plate is bending most at the place of the bass bar, as
the effective stiffness of the plate is increased more than the seen in series b. The plate is not splitting in horizontal nodal
effective mass by the bass bar. Note that the most pronounced lines as easy as earlier.
Physica Scripta 2
250 E. Jansso:z et al.

Frg. 11. Interferograms of series ( e ) , the back plate. ( a ) 490 H z (lo), (6) F ~ E .12. Interferograms of series (f), the back plate with artificial im-
660 Hz (lo), (c) 840 Hz (35), (d) 910 Hz (30), (e) 1 030 Hz (20), (f)1 120 Hz movable sound post. (a) 740 Hz (15), (6) 820 Hz (15), (c) 960 Hz (30), (d)
(20). 1 110 Hz, (e) 1 200 Hz (ZO), (fl 1 300 Hz.

( d ) Top plate with f-holes, bass bar, and artificial, immovable the shape of the fundamental mode is changed, it has a nodal line
sound post at the sound post. The upper part is vibrating in one additional
I n series d (Fig. loa-e) the appearance of the second and the area in the third mode, which probably is a residue to the second
third mode is changed back to mainly that of series a. However, mode, the Q of the third resonance is considerably higher than
Physica Scripta 2
 Resonances of a violin body studied by hologram interferometry 251

The modes of the back plate are more difficult to separate

than those of the top plate. The back plate is not as sensitive to
changes in humidity and temperature. The back plate is made
of harder wood than the top plate and is thicker.

(f) Back plate with artificial, immovable sound post

The order of main appearance is kept from series e to series f
(Fig. 12a-f). In all cases a nodal line or a nodal area is situated
at the sound post. The vibrating area of the upper part (Fig. 1 2 4
is probably a residue of the second mode (cf. Fig. 1Oc). In Fig.
12e one of the earlier four vibrating areas is suppressed (cf.
Fig. 10d). All resonance frequencies are increased in good agree-
ment with changes from series c to series d (Fig. 19).

( g ) Top plate in the assembled instrument, without finger

board, strings, and devices for stringing
Plate resonances from series g are shown in Fig. 13u-d. The
second mode of series d was difficult to find and no successful
recording of this mode was obtained. The mode (Fig.: 13b) is
believed to be a distorted version of the third mode, two vibrating
areas in the lower part divided by a vertical nodal line.
The main appearance of the other modes resembles closely of
those in series d (Fig. 10). A nodal line is at the sound post in
three interferograms. The sound post is vibrating in the distorted
mode (Fig. 13b). The sound post is now placed at a springy
support, the back plate. Series c and d represent the extremes of a
sound post influence, no sound post, and immovable sound post.
The mode frequencies of series g are likely to be in-between those
of series c and d. This assumption is fulfilled in the experiments
in three cases out of four.

( i ) Top plate in the complete instrument

Plate resonances from series i are shown in Fig. 14a-e. No pro-
nounced changes of the modes’ appearance occur compared with
series d and g. The sound post is not visibly vibrating in any of
Fig. 13. Interferograms of series (g), the top plate in the assembled the interferograms. The frequencies are increased for two modes
instrument, without fingerboard, strings and devices for stringing. (a) 495 Hz but decreased for two compared with series g. Most changes are
(25), (b) 740 Hz, (c) 930 Hz, ( d ) 1 150 Hz. however small. The resonance frequencies in series i are in-between
those of series c and d in three cases out of four. The bridge and
strings seem not to give drastic or one-way changes in frequency.
the Q of the second. In the fourth mode the vibrating area of the However, already in series g, new vibration patterns appear
lower right part of the plate is suppressed by the sound post. with the whole instrument vibrating. Two examples are shown in
The immovable sound post gives an additional fixed point of Fig. 15a and b. In Fig. 15a, the tailpiece, finger board, and the
the plate besides the ribs, which is likely to increase the reso- upper part of the plate are vigorously vibrating. This type of
nance frequencies. This is in agreement with the experiments with resonance in the whole instrument will be called body resonance
only one exception (Fig. 19), the second mode with a horizontal in this work. The body resonances are much influenced by the
nodal line at the sound post. The position of the sound post can clamping. In Fig. 15b the edge of the lower right part is vibrating.
be traced by unusual bends of the interference fringes (Fig. The tailpiece seems to be rocking around an almost hori-
10a-d). zontal axis.

(e) Back plate (h) Back plate in the assembled instrument without finger board,
strings, and devices for stringing, and
Interferograms from series e are shown in Fig. 11a-f. Coarsely
the appearance and the frequencies of the modes are the same as ( j ) Back plate in the complete instrument
those of series a, the top plate without f-holes and bass bar. Few normal modes of vibration were found in the back plate.
However, the appearance of the fourth and fifth modes are Nearly independent of driving points the nodal lines were moving,
changed compared with series a, The frequencies tend to be when sweeping through a resonance. Especially at low frequencies,
lower than those of series a. The modes are not perfectly symmetric- the plate is vibrating at its edge, i.e. the resonances are pronounced
al (e.g. Fig. 11a, b, and e). Imperfections in the glueing seem to body resonances. Therefore, only a combined series of interfero-
have caused the vibrating edges in Fig. 1 1 d, e, and f. The mode grams from series h (Fig. 16a andj) and from series j (Fig. 16b,
patterns are, however, about the same as corresponding patterns c, d, e, f,g , h and i ) are shown. All modes are either driven at
of series a. the sound post or the bridge. The sound post is visibly vibrating
Physicu Scripfa 2
252 E. Jansson et al.

however, that because the back plate is harder to excite it is

driven harder in order to obtain the interferograms. This may also
cause the difference in behaviour between the plates. It should be
noted none-the-less, that many experiments were done with
different driving points at the back plate. The results w-d-2 not
consistent because the plates always vibrated together with ths
rest of the instrument, and thus dependent upon the clamping.

( k ) Deformations of the complete instrument

Double exposure holograms were made of the complete instru-
ment with a first exposure when one or several strings were pressed
against the finger board at about 8 cm from the nut and a second
exposure after releasing the strings. Double exposures of a holo-
gram cause double images upon reconstruction of the fields
and, because the reconstructions are mutually coherent, they
interfere with one another and cause cosine interference fringes.
Interference fringes thus obtained give a measure of the object’s
deformation between the exposures. Six double exposure inter-
ferograms of the violin are shown, in Fig. 17a-c of the top side,
and in 17d-f of the back plate. In Fig. 17a and d the G-string is
pressed and released, in 17b and e the E-string and in Fig. 17c
and f all strings.
The interferograms show that the instrument is bent around
the upper corner block in such a way that the free part of the
finger board is deflected towards the sounding box. The top
plate is pushed inwards and the back plate outwards. It is also
shown that the tailpiece is moved when a string is pressed.
A closer study of the top plate deformation shows, that the
deformation is small at the right foot of the bridge, the sound
post side, but close to maximum at the left foot of the bridge,
the bass bar aide, when the G-string is pressed. The static defor-
mation reminds of the first vibration mode of the plate (Fig.
loa). When the E-string is pressed against the finger board, the
deformation at the sound post is bigger as when the G-string is
pressed.
The back plate deformation is maximum at the sound post,
and consequently bigger when the E-string is pressed as when the
G-string is.
The top plate deformations are bigger than the back plate
deformations and the back plate is acting more as a unit with
the ribs than the top plate does.

Fig. 14. Interferograms of series

(i), the top plate in the complete
instrument. (a) 480 Hz, (6) 780
Hz, (c) 830 Hz, (4950 Hz, (e)
1100 Hz.

in 6 of the 10 interferograms. A try is made to identify the

appearance of the modes. The frequency changes from series f to
series h and j resemble of those from series d to series i (Fig. 19).
The back plate seems to vibrate more as a part of the whole
instrument than the top plate. This seems to be in agreement
with the construction of the instrument. The back plate is heavier, Fig. 15. Interferograms of series (i), the top plate in the complete instrument,
stiffer, and glued more firmly to the ribs. It must be stressed, showing two body resonances. (a) 100 Hz, (6) 600 Hz.
Physica Scripta 2
 Resonances of a violin body studied by hologram interferometry 253

Fig. 16. Interferograms of series (h), the back plate in the assembled
instrument without fingerboard, strings and devices for stringing, and series
( j ) , the back plate in the complete instrument. ( U ) 100 Hz, (b) 340 Hz,(c)
470 Hz, (d) 505 Hz, (e) 580 Hz, (f)770 Hz, (g) 870 Hz, (h) 970 Hz,(i) 1 130 Hz,
0') 1 280 Hz.

6. Air resonance and radiated sound

After series k the violin was taken apart, finally adjusted and
put into order for testplaying, still unvarnished. The strings were
damped and the instrument was clamped around the neck to a
heavy stand. The electromagnetic driving system was mounted
to the stand in such a way, that the instrument could be excited
at the bridge between the G- and D-string perpendicular to the
top plate.
First the lowest air resonance, the only one of practical im-
portance [15], was studied. A microphone (BriiA and Kjaer
4135) was placed about 1 cm outside the middle of the right
f-hole. Two frequency responses were obtained, one with the
left f-hole covered by tape and a second with the f-hole open.
A double peak at a frequency about 285 Hz wasrecognizedas
During all these experiments, the violin was clamped close connected with the air resonance. The same double peak was
to the tailpiece at the end block and at the upper corner blocks, found at a lower frequency with the left f-hole covered. The
as seen in the figures. This clamping probably does not give experiment was repeated with the enclosed air volume excited
exactly the same deformation of the instrument as when held for by a small earphone furnished with a 12 cm long and acoustically
playing. damped tube, which was inserted through the left f-hole. The
17 - 71 1971 Physicu Scripfu2
254 E. Jansson et al.

Fig. 18. Frequency responses obtained at a distance of 35 cm perpendicular

t o the plates through the bridge. Electromagnetic driving at the bridge be-
tween the G- and D-string perpendicular to the top plate. (a) The microphone
facing the top plate; (b) the microphone facing the back plate.

a microphone (Bruel and Kjaer 4131) at eight different points

at a plate length distance from the central axis of the instrument
in a plane perpendicular to the axis. Two frequency responses
are shown in Fig. 18a and b obtained perpendicular to the
plates, compare [13]. Fig. 18a shows two prominent peaks at
490 Hz (Q =30) and 1 190 Hz (Q =35) and Fig. 18b two at 490
Hz (Q = 30) and 690 Hz (Q =45). A closer look reveals the double
air resonance peak at about 300 Hz.
Glissandos ranging from open string to octave were played and
recorded with the microphone approximately perpendicular to
the top plate at a distance of 50 cm. Afterwards the strength of
the fundamental as function of frequency was drawn by means
of a narrow band-width spectrograph, Rohde and Schwartz FNA
48301. The air resonance and the peak at 490 Hz were clearly
obtained [cf. 14, 331.
Finally the violin was played by an expert professional. The
player found the tone powerful, even but not as noble as in old
excellent violins. The violin was strung with steel strings, how-
ever, to stay in tune during the experiments, although it was
adjusted for gut strings. Furthermore, the player was not used
to steel strings.

7. Discussion of experiments

This study of a violin can be divided into three major parts: the
investigation of ( U ) the plates alone, (b) the plates in the assembled
instrument, and (c) the radiation properties.
Fig. 19 summarizes the results regarding resonance frequencies.
The upper six lines represent the top plate experiments and the
lower three those of the back plate. Each circle marks the fre-
quency of a mode and the thin lines connect modes of similar
shape. The plates alone are studied after major steps in making

Fig, 17. D o ~ ~ h cxpoyurc

le interferograms of ~ristrunicntdeformntions when
strings are pressed against the fingerboard. (a) top view, G-string pressed;
(6) top view, E-string pressed; (c) top view, all strings pressed; ( d ) back
view, G-string pressed; (e) back view, E-string pressed; (f) back view, all
strings pressed.

frequency responses now showed a single peak at 280 Hz with a

Fig. 19. Resonance frequencies obtained in the experimental series (a)-(j).
Q of 20. Only traces were found of the higher peak.
Each circle represents a mode in a specific series. Modes of mainly the same
Thereafter instrument and driver were moved to an anechoic shape are connected by a thin line. The upper six horizontal lines cor-
chamber. Frequency responses were measured and recorded by respond to the top plate experiments, the lower three to the back plate.
Physica Scripta 2
 Resonances of n ciolin body studied by hologram interferometry 255

and with boundary conditions simulating those of the plates in lowest air resonance and a resonance in the neck, and the peak at
the assemble instrument. After the first step, series a and e, 490 Hz to the fundamental resonance of the top plate (Fig,
the plates are bilaterally symmetrical and split up into vibrating 14a and 19 series i). At higher frequencies, the frequency responses
regions divided by vertical and horizontal nodal lines. The upper are harder to interpret, because the excitation of mode combina-
and lower bouts tend to split up into the same vibration patterns tions and the complicated radiation properties of the instrument
but at different frequencies. The cutting of the f-holes lowers the [21]. The two mentioned peaks also show up in the played glis-
resonance frequencies and alters the order of the modes’ ap- sandos and the peaks appear to be important in the playingof
pearance (series b). The bass bar increases the resonance fre- the instrument.
quencies and the vibrations in the upper left part of the plate The difference made between body and plate resonances is in
(series c ) . The frequencies and shapes of the modes are changed a way fictitious. When a plate mode is excited hard enough, the
the most when the bass bar is bent the most. In series d and f, whole instrument is set into vibration, i.e. a body resonance. The
the plate vibrations are studied under conditions, simulating those main vibration, however, is still in the original mode.
of the plate in the assembled instrument by an immovable sound The interferograms presented show the vibration patterns at
post. This sound post introduces an additional point of zero single frequencies, resonance frequencies. When a violin is played,
tranverse motion. Some modes are distorted and the resonance the plates are simultaneously excited at several harmonic fre-
frequencies are increased. quencies, therefore they vibrate in combinations of modes.
The plates in the assembled instrument are studied in two
series, before and after stringing. In the top plate, series g and i,
the modes of series d appear slightly distorted, i.e. the sound post
8. Conclusions
acts as a support to the top plate and the resonances are-or
are at least close to-plate resonances. The resonance frequencies In this study, vibrations of a violin body are investigated by
are lowered in series g compared to series d, however, which hologram interferometry, and acoustical measurements. The
indicates that the sound post is not a perfect immovable support. results provide a first quantitative understanding of these vibra-
The overall influence of the bridge and strings is neither great tions and how they are influenced by the different parts of the
nor simple regarding mode frequencies. All modes recorded can violin body. Hologram interferometry proves to be an excellent
be driven at the bridge at least in combination of modes. tool to study the vibrations and to open ways for a more complete
The back plate behaves differently, series h andj. Its vibrations understanding of the violin.
are visibly coupled to the rest of the instrument. The vibration
modes, body resonances, are effected by the clamping of the
instrument. All modes shown are driven at the bridge.
Acknowledgements
In series i resonances in other parts of the instrument were
found as bending of the neck and in the tailpiece. Grateful acknowledgement must be gi\en to K. A. Stetson at the National
One possible use of the interferograms should be mentioned. Physical Laboratories, Teddington, England, Klaus Biedermann at the
An interferogram of a mode shows the magnitude of displacement Institute of Optical Research, KTH, Stockholm and J. Sundberg at the
Department of Speech Communication, KTH, Stockholm, for valuable
of the standing wave. Thus the velocity and the bending can be discussions and criticisnl of the manuscript. We thank Profs. E. Ingelstam
obtained in every point of the plate. With these parameters known at the Institute of Optical Research and G. Fant at the Department of
it can be estimated, how changes of the plate will influence its Speech Communication, KTH, Stockholm, for their interest and support.
vibration modes by perturbation theory, such as is used in This work was supported by The Swedish Board for Technical Devel-
calculating the resonance-frequency shifts of the human vocal opment.
tract at small changes in articulation [34]. The distribution of
mass and stiffness determines the vibration modes of the plates.
A mode’s frequency will be raised if the plate is thinned at a References
place of mainly tranverse motion (a decrease of mass) and
1. Leipp, E., Le Violon. Paris, 1965.
lowered if the plate thinnsd at a place of mainly bending (a 2. Mockel, 0. and Winkel, F., Die Kunst des Geigenbaues. Hamburg,
decrease of stiffness). 1967.
The interferograms obtained when pressing strings against 3. Hutchins, C. M., Catgut Acoust. Soc. Newsletter, No. 12 (1969).
the fingerboard, series k , show changes in the tailpiece and 4. McKusick, V. A. and Wiskind, H. K., J. Hist. Medicine and Allied
Science 14, 41 1 (1959).
finger board position, and maximum deformation at the left 5. Helmholtz, H., On the Sensations of Tone (transl. A. J. Ellis). New
f-hole in the top plate and at the soundpost in the back plate. York, 1954.
The deformations look like the fundamental mode of the top 6. See Catgut Acoust. Soc. Newsletter, No. 13, p. 7 (1970).
plate, and the low frequency vibrations of finger board and 7. Backhaus, H., Zeits. f . Physik 62, 143 (1930).
8. Meinel, H., Elek. Nachr. Tech. 14, 119 (1937).
tailpiece.
9. Saunders, F., Sound 1, No. 4 (1962).
The transfer function of any linear system can mathematically 10. Risset, J.-C. and Mathews, M. V., Physics Today 22, No. 2 (1969).
be described by a ratio of two polynomials. A factorization of the 11. Yankovskii, B. A., Soviet Phys. Acoustics 11, 231 (1966).
numerator brings out the zeros and a factorization of the deno- 12. Hutchins, C. M., Scientific Amer. 207, No. 5 (1962).
13. Hutchins, C. M. and Fielding, F. L., Physics Today 21, No. 7 (1968).
minator the poles of the transfer function. This theory has SUC-
14. Jansson, E., Speech Transmission Lab. Quart. Progr. and Status Rep.,
cessfully been applied to the analyses of complicated acoustical No. 3 (1966).
filters, such as the human vocal tract [35]. By applying the theory 15. Schelleng, J. C., J. Acoust. Soc. Amer. 35, 326 (1963).
to the acoustically measured frequency responses of the violin, 16. See the detailed list of references in [131.
it can be said that the poles are determined by the instrument, 17. Eggers, F., Acustica 9, 453 (1959).
18. Menzel, R. E. and Hutchins, C. M., Catgut Acoust. Soc. Newsletter,
and the zeros by the choice of the place of excitation and the No. 13 (1970).
place and choice of the response function. Thus the double peak 19. Stetson. K . A.., Catgut
- Acoust. Soc. Newsletter, No. 9 (1968).
at 280 Hz, made up by two poles and a zero, can be related to the 2 0 . Agren, C.-H. and Stetson, K. A., J. Acoust. Soc. Amer., in print.
Physica Scripta 2
256 E. Jansson et al.

21. Jansson, E., Speech Transmission Lab. Quart. Progr. and Status Rep.
NOS.2-3 (1969).
22. Reinicke, W. and Cremer, L., J. Acoust. Soc. Amer. 47, 131 (1970)
abstract only.
23. Powell, R. L. and Stetson, K. A., J. Opt. Soc. Amer. 55, 1593 (1965).
24. Stetson, K. A. and Powell, R. L., J. Opt. Soc. Amer. 55, 1694 (1965).
25. Goodman, J. W., Introduction to Fourier Optics. New York and
London, 1968.
26. Smith, H. M., Principles of Holography. New York, 1969.
27. The Engineering Uses of Holography (ed. E. R. Robertsson and J. M.
Harvey). Cambridge University Press, London, 1970.
28. Molin, N.-E. and Stetson, K. A., J. Phys. E: Sci. Instrum. 2, 609 (1969).
29. Stetson, K. A., Optics Technology 2, 80 (1970).
30. Biedermann, K. and Molin, N.-E., J. Phys. E: Sci. Instrum. 3, 669
(1970).
31. Archbold, E., Burch, J. M., Ennos, A. E. and Taylor, P. A., Nature 222
(1969).
32. Ek, L. and Molin, N.-E., Submitted to Optics Communications, 1970.
33. Bradley, J. S. and Stewart, T. W. W., J. Acoust. Soc. Amer. 48,575 (1970).
34. Schroeder, M. R., J. Acoust. Soc. Amer. 41, 1002, (1967).
35. Fant, G., Acoustic Theory of Speech Production. The Hague, 1960.

Royal Institute of Technology

S-100 44 Stockholm 70, Sweden

Physica Scripta 2