Interdisciplinary Science Reviews

ISSN: 0308-0188 (Print) 1743-2790 (Online) Journal homepage: https://www.tandfonline.com/loi/yisr20

The Acoustics of Stringed Musical Instruments

M. E. McIntyre & J. Woodhouse

To cite this article: M. E. McIntyre & J. Woodhouse (1978) The Acoustics of

Stringed Musical Instruments, Interdisciplinary Science Reviews, 3:2, 157-173, DOI:
10.1179/030801878791926128

To link to this article: https://doi.org/10.1179/030801878791926128

Published online: 20 Nov 2013.

Submit your article to this journal

Article views: 95

View related articles

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=yisr20
 The Acoustics of
Stringed Musical Instruments

M. E. McINTYRE AND J. WOODHOUSE

Department of Applied Mathematics and Theoretical Physi ,Cambridge, England

Attempts to understaod the action of musical instruments at a level having some impact on the problem of
mu icians and instrument maker must face formidable un olved problems in psychoaeou tie a welJ a in
classical phy ics. Tbis review outline what is known about the phy ieal properties of tringed in trument , with
emphasis on the violin family, and tries to identify basic outstanding problem where progress may be hoped for
within the next decade. One ueh area is assessment of the widely used simple t idealization of violin behaviour
in whieh the driving at the bridge i as umed to be traosver e, and the body is modelJed a a black box' with
one input and one output. Other major problem include the physical nature and p yehoaeou ticaJ relevance
of comple fine detajJ in the behaviour of the bowed tring.

According to tradition, Pythag ra wa first guided problem to which phy ici ts addres them Ive and
to the noti n that mathematics h ld a key to under- also the scope of their enquiries: along ide an
tanding nature by observation of the relation examination of the behaviour of instrument they
between musical interval and natural numbers. must do the psycboacoustical work nece sary to
From this seminal po ition musical acou tics correlate the ubjective judgement f the mu ician
remained a recognised part of scientific endeavour with phy ical effect amenable to m a urement and
until the early years of this century, with scientists of theoretical inve tigation.2.3 As we hall e, thi
the calibre of Rayleigh and Helmholtz making major work presents challenge that can be ev n more
contribution . However most of the physicists and formidable than tho e of phy ics.
applied mathematicians of the next generation We review here recent work on th tring d
turned away from uch problems to the new and mu ical instruments, especially those of th violin
exciting developments of relativity and quantum family a they pose some of the bigge t questions.
theory. Only in recent years ha there been a general We emphasize problems that are not dealt with in
revival of intere t in classical problems, including the excellent recent book by B nade 3 and we
musical acou tic . emphasize question rather than answer : the r ader
Indeed it has been recently suggested that musical mu tnt expect to find any ecrets of Stradivari .
acou tic may again hold a key to progre in one of Seriou research has not yet reached thi 1 vel f
the frontiers of science, namely the understanding of confident predi~tion. It i unfortunate that the
the human brain. In a recent paper entitled popular image of violin acou tics has been largely
'Research potential in auditory characteristics of moulded by purveyors of pet theorie on this ubject
violin tone' P. . Boom liter and W. Creel I pint ( ee for example chapter 5 of Heron-All n4 ntitl d
out that 'Tbe violin, its vagarie and variegators').
Our understanding of hearing and of the violin can
and hould move forward together... When we
study the variation introduced by the lip- tick'
THE STRING AND THE SOU D BOX
action of the bow, we are al 0 studying the nature of
hort-time organization by the nervous system.
When we study note tunings in violin performance, The phy ical behaviour of a tringed in trument can
we are also studying principle of association in be examined under thre heading: first the
hearing. When we study richne and body in violin behaviour of the tretched tring who e vibration is
tone, we are at 0 studying what make for continu- controlled in one way or another by the play r
ing attention and ati faction in human perception.
second the respon e of the wooden sound b of the
When we study the phrasings of virtuo 0 we are
instrument and the neighbouring air in re pon e to
at 0 tudying the longer organizing systems of the
human mind.
the tring motion, and third the radiation of ound,
almost entirely from the ound box and generally
This growing awareness of the psychological involving a complicated directional dependence.
dimension of the ubject in turn affect tbe type of Tbese three cannot be treated entirely eparately,
CCC-030 -0188/78/0003-0157$08.50/1
© Heyden & Son Ltd, 197 INTERDISCIPLINARYSCIENCEREVIEWS,VOL. 3, NO.2, 1918 157
because there is a back-reaction of the body vibra- fibre composite material t duplicat the ne e ary
tions on the string which in th bow d in trument properties of OTWay Spruc , Picea excelsa the
has an important effect on the 'f el' of the instru- wood used for the oundboard of alma tall strin ed
ment to the player. The body vibrations are in turn instruments.
affected by a radiation reaction. We deal here mainly Under tanding musical in trum nt involv
with the first and econd topic ; a pect f the third making theoretical model f their behaviour.
topic concerning directional radiation patterns are Before can jdering in detail variou models that
covered in an earlier review by remer5 and else- have b en propo ed for the bowed tring in tru-
wh re.6 ment ,we hould ask a basic general qu ti n: what
Though the behaviour of a tretched tring which can titute an adequate model in mu ical ac u tics?
i plucked or struck is well under to d, the
behaviour of a bowed tring i much more compli-
cated. Th motion was understood in outline by MODELS AND RES AR
H lmholtz a far back as 1860 and much additional MU ICAL ACOUSTICS
detail wa elucidated by C. V. Raman early in thi
century. However, many of th mu ically all- From the music of the phere 7 to the Glass Bead
important finer d tail have only recently begun to Game, it ha been c mm n to peak of connecti n
be explored. Of particular interest are the tolerance between mu ic and math mati . H wever mu i-
range in the variou parameter under the player's cians and cienti t hav by no means alway
control for a mu ically acceptable 'steady' note to be cooperat d fruitfully in th study f musical ac u -
produced, and also the length and nature of the tic. Mu ician t nd to think that cienc ha n thing
tring transients involved in vibrato attack' and to tell them, while some phy ici t t nd to a um
'articulation', to whicb the ear i particularly en i- that the study of instruments and can ert halls
tive. The 'steady' ca e ha been inve tigated tome involves no more than traightforward measure-
extent already the tran ient ca e hardly at all. Both ment. In fact, nothing c uld b further from the
present out tanding problem for the future. truth: recent developm nt in p ychoacoustics,2.3 a
The behaviour of the ound box i a c ncern of the well as celebrated di a ter in oncert hall d~sign,
instrument mak r, and much information is avail- have empha jz d th ubtl ty f the human ar-
able from master luthiers who have been prepared to brain system as well as m difying me f th Id r
talk about their ideas and method. Especially e tabLished ideas about the mechani m f bearing.
important contribution have come from Vision involves similar ubtl ti , and Fig. 1 pro-
scientifically minded maker who take extensive vides an excellent illu tration. Th reader will hav
acoustical mea urement on components of instru- no difficulty in eeing a Dalmatian dog with a wealth
ment bodie at variou stage of construction. These of information about what it look like and what it i
re ult ,and theoretical investigations based on tbem, doing. ow one c uld readily measure uch thing a
are apparently beginning to yield some of the essen- tati tical distributions of spot izes and pa ing in
tial characteristics of good in trument , although not the picture, or the chemical comp ition f the
the distinguishing feature of the very be tone . printing ink. and it i cl ar that u h data, however
(Indeed it i que tionable whether 'be t' has any elaborate, would n t r v al v n the existence f th
unique meaning.) dog. Indeed, it doe n t take much thought t di-
One important fact to emerge is that at present we cov r the practical impo sibility of writing a en ral
do not know enough about the properties of wood: purpo e computer program which w uld recogniz
we n ed t know the elastic properties and e pecially the presence of the dog out of th myriad oth r
the internal damping behaviour in more detail than po sibilities.9
was previously reali ed. U eful light i being h d on Similarly, while acknowl dging that standard
this problem by a recent effort to develop a carbon acou tical mea urem nt have their use we mu t

DR MICHAEL E. MciNTYRE is a lecturer in applied mathematics at the University of

Cambridge whose research interests include fluid dynamics and wave propagation as well
as musical acoustics. His interest in acoustics stems from experience as a concert violinist;
he has appeared professionally as concerto and chamber soloist with a number of
distinguished musicians. including lona Brown, Arthur Davison, Antony Hopkins,
Christopher van Kampen and Tamas Vasary. His concern to bring musical acoustics into
contact with artistic experience was instrumental in setting up a project at the Royal College
of Music, London, to explore the uses of the Hutchins 'new violin family'.
DR JAMES WOODHOUSE is a Cambridge mathematics graduate with a spare-time interest
in musical instrument making. At an early stage in his doctoral research he won a Smith's
Prize with an essay on plate vibration theory and its relevance to violin makers' procedures.
He hclds the Denman Baynes Senior Research Studentship at Clare College.
Address: Department of Applied Mathematics and Theoretical Physics, Silver Street,
Cambridge, CB3 9EW, England.

158 INTERDISCIPLINARY SCIENCE REVIEWS. VOL. 3, NO.2. 1978 © Heyden & S n Ltd, 197
Figure 1. Photograph by
Ronald C. James. (Repro-
duced from The Intelligent
Eye by R. l. Gregory, by
permission of Weidenfeld
and Nicolson Ltd.)

guard against the assumption that such measure- their ound. Correct description f int rm diate
ment can capture everything significant about a time scale phenomena i thu an extr m ly
mu ical ound. Again, it is clear that very small important requirement for models in mu ical
changes to Fig. 1 could remove or dra tically change acoustics.
the dog, and this is equally tru of musical ounds: The foregoing remark nece itate a critical atti-
we must not as ume that small physical changes tude to two research approaches which are
nece arily produce small perceptual effects.i.3 So frequently suggested. On the one hand, one could
when we ask the que tion 'what i an adequate take an instrument and try to mea ure all th rele-
model in musical acoustics? , one thing i plain: the vant data about it phy ical behaviour: uch blind
final te t of 'adequacy' mu t come from the ears of measurement is usually terile inc knowledge
the listener, and the finger of the player. Lacking as about what is 'relevant' to the ear i 0 limit d. On
we do a ufficient understanding of the auditory the other hand one could try to imulat n a
sy tern we should aim to listen to, and play, our computer the entire chain of acou tical vent from
modeLs.1o the actions of the player to the ignal at the
The ability t recognize the dog in Fig. 1 is eardrums of the Ii tener: uch large scale numerical
thought to depend on a hierarchy of feature detec- modelling i , however, much more limited in scope
tors' in our neural signal processing system. Among than i ometimes uppo ed. A sufficiently accurat
tho e imple examples for which direct neurophy- imulation of a mu ical instrument in it acou tical
siological evidence is available are vi ual nerve cells environment is well beyond the reach of pre nt day
which respond pecifically to straight line or edges computers; and even if it were not, it would be
at variou inclination. 11 Pa t experience is impos ibly cumber orne t u a a research tool
important: feature detection apparently involve becau e of the volume of information that would b
matching against internal paradigms.1.2 In the audi- produced.
tory system, feature detectors can be roughJy To make progress we must divide the chain of
classified according to the time-scale over which they events into component and tudy th s individuall
operate. and in simple interacti n b fore trying to
The hortest time cale is that evidenced by pitch comprehend everything at once. uch study should
di crimination: vibration period difference of ten combine re tricted model building with mea ure-
12
of microseconds are involved.D. The opposite ment and listening tests. Two example may b cit d
extreme i encountered in the perception of melody as illustrations, each succe ful in it own way. The
and rhythm, involving time scale from fractions of a first is the work of Benade el al.,3 in which a simple
econd upward. In between are what we can term theory wa u ed to interpret mea urements of the
phenomena of 'intermediate time scale', such as propertie of a wind instrument. The theory th n
tarting tran ients and vibrato, where significant ugge ted that certain small changes in the tube bar
changes occupy tens of millisecond . The e la t and hole geometry would improve the playing pro-
are the time scales of con onant in peech, and perties of the instrument, and ub equ nt playing
are aloof prime importance in the recognition and listening te t confirmed this. The second is the
and quality a e sment of musical instruments by work of Risset and Mathew, 10 who analysed sound

© Heyden & on Ltd, 1978 INTERDISCIPLINARY SCIENCE REVIEWS. VOL 3. NO.2, 1978 159
of a trumpet, used available p ychoacoustical know- second the tring can be played at ufficient ampli-
ledge to gue s important features, including tran- tude to introduce nonlinear effects. 15.17
sient tructure, and on this basi re ynthe ized (c) Many of the problems of real mu ical interest
ound which could then be compared with the ori- are concerned not with the relativel imple motion
ginal in a Ii tening te t. of the string during the middle nominally 'steady
portion of the note but with the complicated tran-
ient motion at the start and fini h of the note.
Furthermore even teady' note may contain hart
COMPLICATIONS OF REALISTIC MODELS and intermediate time cale variations of everal
kinds, the rna t obviou of which is vi brat .
on ider now what might be involved in realistic (d) The bridge d not merely tran mit f rce
models of, say, the violin. Figure 2 recalls the basic from the tring to the body passively: it ha vibration
chain of event occurring when a violin is played. modes of its own many of them involving compli-
The player et a tring of the instrument into vibra- cated three-dim n ional motions 1 and these can
tion with hi bow. The string alone doe not radiate strongly influence the eventual driving force on the
ound waves efficiently into the urrounding air, a a b dy.19 In addition ther will be a certain am unt of
wooden resonating tructure i u ed, namely the direct radiation of very high frequency und from
body or ound box of the in trument. The string the faces of the bridge. .
passes over a bridge in contact with the body, and (e) The geometry of a violin bod doe n t I nd
the movement of the string imparts a fluctuating it elf to simple theory: th combination of a c mpli-
force which excites vibration in the wood of the cated hap and a constructi nal material, wood
body. Thi in turn generates sound waves in the having a large number of independ nt la tic and
urrounding air which radiate away to reach the ears viscoelastic constants make detailed modelling
of the listener. Even if we leave a ide the terra cumber orne. Ala, d pite the small amp]itud of
incognita of the biological links in the chain a clo er typical motion in the body, it i possible that om
look at the purely acou tical aspects reveal a long nonlinear effect may have audible con equence .
list of complications: for example, effects associated with th purfling
(a) The bow which excites the tring is itself a inlaid around the edge of the front and back plate of
dynamical system, who e propertie influence the the instrument. 20
tring's behaviour. 13 Thi fact, well known to (f) The tring mu t be regarded a part f th
players, has received relatively little attention from body for the purpose of analysing it m chanical
14
phy ici t . behaviour. As Schelleng ha pointed out the
(b) While the visible motion of the tring i pre- longitudinal compliance of the tring especially in
dominantly transverse, there will inevitably be the case of metal string, will ubstantially change
tor ional14 and ]ongitudinal15 motions as well, the way the body vibrates.
greatly complicating the drivi'1g force at the bridge. (g) The vibration of the body and bridg can
Even for the transverse motion, the linear wave significantly affect the mati n f the tring. An
equation familiar from acoustic textbook i by no extreme example i what mu icians call a wolf nt',
means the whole story for at least two reasons: first, an unplea ant tuttering effect re ulting from the
a real string is anharmonic because of slightly yield- attempt to play a note falling very n ar a trong
ing termination 3 and finite flexural stiffne s 16 and resonance of the body. In a Ie p ctacular way,

Player's Bowed
hands string Auditorium

Figure 2. Schematic indication of the chain of events when a violin is played. The picture fails to
suggest some of the important ingredients, such as slight movements of player or listener in the
auditorium; see text.

160 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2, 1978 © Heyden & Son LId, 197
back reaction of the body and bridge on tbe string reach this, we first put the complexitie of the radia-
influences the ea e with which any note may be tion and room acou tic problems on one side, and
established, and the lengtb and nature of it starting consider the sound of the violin a heard fr m a
. 21 single, fixed microphone position. ext we uppose
tran lent.
(h) The air motion in and around the instrument that the force on the bridge resulting from the
couple to the motion of the body.22.23 Thi coupling torsional and longitudinal motion of the string i
is generally much stronger than tbat between the sufficiently smaller than that an ing from the trans-
body and tbe string: the body is, after all, designed to ver e motion that we can neglect it. If we then
radiate sound. make the relatively minor a umption that the
(i) The radiation properties of a violin body are radiation proces and the behaviour of th b dy are
56 linear to sufficient accuracy, we are left with a linear
very complicated, e pecially at high frequencies. .
There are many po ibilities for important effects 'black box' with a single input and a ingl output.
here, since the ear-brain sy tem is very killed at Now it i well known that uch a y tem can b
u ing directional information in a sound field; for completely defined by its complex frequ ncy
example, thi contribute to the ability of a listener respon e function, that is a singl function whi h
to follow a conversation in a crowded cocktail party. specifies the amplitude and pha e of tbe output sine
The fluctuation of the directional radiation pattern wave when the system i driven with a inusoidal
due to vibrato might for instance be important; if force of unit amplitude and any given frequency. In
0, thi would help explain why conventional experimental work on the violin family a in
recording technique seem unable to capture the specifications for hi-fi equipment, it i customary to
ound quality of a violin played in a concert ball plot only the amplitude although this may well not
24
BenadeJ has demonstrated that the ear is ensitive to be adequate.
the directional radiation pattern of wind in tru- An example of a respon e curve mea ured from a
ments played indoor; changes in these pattern are Stradivarius violin (the Titian of 1715) is shown in
perceived by the Ii tener as changes in subjective Fig. 3. In this case, the instrument wa placed in
tone quality. nearly anechoic surroundings to minimise effect
U) The room in which an instrument i played ha from the behaviour of th room. A tran verse fore ,
many vibration mode in the audible frequency who e amplitude was held con tant as frequ ncy
range: even a domestic living room has ten of varied, was applied electromechanically to th violin
thou ands. Consequently, acou tical measurements bridge. Peak in the respon e curve indicat either
with a single microphone show wild ftuctuations frequencies at which the radiation patt rn of the
when the microphone or any other object is moved instrument happens to fa our the micr ph n
lightly. Thi complexity, far from being a problem direction, or resonance of the violin bod including
for the ear, i put to positive use;3 for example, blind the air in and around it. Some of the resonant vibra-
people learn to use the information to 'hear the tion patterns of in trument bodi at such peaks
po itions of obstacle in a room. have been explored experimentally by various
Thi Ii t is by no means exhau tive. It sugge t that technique .25-28
the first level in a hierarchy of models mu t inevit- The lowe t peak in Fig. 3, ar und 27 Hz, i a
ably involve drastic simplification. typical feature of re ponse curves of r a nably
goo d m . trument.· 2529 I t h as t h e d"Istmctlv. property
30
of disappearing when the f-h Ie are bl cked and
THE RESPONSE-CURVE MODEL hifting to a lower frequency when th instrum nt
31
b dy is filled with carbon dioxide. The air cavity
In the first in tance we take the string motion a within the body together with th f-holes i acting
given and concentrate on the action of the violin somewhat like a Helmholtz re onator, modified by
body for this purpo e regarding the bridge a part of coup \.109 to b 0 d y motion.'. 2232 (I n gUitars
. the
3
the body. Much of the early experimental work on modificati n i drastic.i The higher air-cavity
the ubject took thi approach, since the main inter- modes do not give ri e to such ea ily id ntifiable
e t then was in eeking scientific ways of di tinguish- features in respon e curve, but th ir po ibl role
34
ing between different violins, especially between has been tudied by Jansson.
good one and bad one, on the ba i of the ound Many r sponse curves have be n published,31.35.36
they made. On the grounds that the motion of the but unfortunately ome of them bear no d e rela-
string is much the same on any violin such ti n to the action of the in trument wh n excited b
differences were naturally sougbt in measured its trings because the driving fore was not h Id
characteri tics of the violin body. When we eek the constant a frequency varied. 7 Comparison of
differences betwen playing properties of instruments respon e curves obtained by diff rent e citati n
38
however, it become essential to model tbe string meth d have been given by Bradl y and Stewart
motion and we can ider such problem in detail in and Hutchin .39 Refinements aimed at minimizing
later section . directional radiation effect 0 a t concentrat on
The very imple t idealization i what may for the resonance of the violin indud the u f many
brevity be called the 'respon e curve model'. To microphone 36 and tbe u e of a r verberation

© Heyden & Son LId. 1978 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2. 1978 161
 30

dB

20

10

o
I I I I I I I I
0.2 0.5 2 3 4 5 6 7

FreQuencv in kHz

Figure 3. Response curve of a Stradivarius violin (the 'Titian' of 1715, after Saunders31). Sound
intensity at one microphone position is plotted against frequency. The musical notes shown are
those to which the open strings are tuned. The shaded bars indicate the range of variation of
harmonics of the note 440 A during a vibrato cycle.

chamber in place of an anechoic one.30.40 A related in the proce s giving insight into th phy ical mean-
mea urement on our black box which focuses on the ing of specific feature of th m a ured curves.
a pect of body behaviour relevant to the dynamic Beldie 30 for in tance, u ed a implified mechanical
of the string motion is the 'input admittance curve', a model of the violin body to elucidate it I w
plot of the amplitude of the tran ver e bridge motion frequency behaviour. Another uch m d I i cur-
42
which results from the given driving force. 5.30 rently being studied by C x and F lIgett.
The spikiness' of violin re p n e curves has a Reinicke 19 u ed a similarly impl mod I f the
imple but crucial consequence. When a note is bridge dynamics and convincingly h wed the
played with vibrato on the in trument, the frequency importance of bridge mod t th re p n e above
of each harmonic moves up and down over a range 2kHz.
of up to a semitone, a ugge ted by the haded bars Schelleng,22 in an arlier paper which has become
in Fig. 3. It i plain that the re pan e curve can have a cla ic in the field, used an electric circuit anal gue
con iderable variation within such ranges different model to elucidate a wide range of que tion . Th
for each harmonic. Thus in addition to frequency included the general shape of the re p n curv, the
fluctuation, the output waveform from the black role of the lowest 'air' mode in prom ting fficient
box will exhibit appreciable fluctuations in the pec- radiation above its resonant frequ ncy, and al a the
trum of harmonics with orne increa ing in trength mechanism of the 'wolf note' to which we hall
while others decrea e. This fact was first pointed out return briefly in a lat r ection. In addition Schelleng
by Fletcher and anders,41 who also showed by e tabli hed rules for scaling of different in trument
imple listening te ts that the effect is e sential to the of the violin family to give approximat Iy the same
characteri tic subjective quality of violin ound. The relative response curve with in truments of different
listener does not generally notice the pectral varia- tunings.43 Thi throw light on th difference
tions as uch' rather, the e are 'feature detected' in between the conventi nal vi lin, viola and 'cello,
terms of tone quality - an intermediate time scale which are not caled in thi way.
effect. If vibrato is played in 'slow motion' it i no Attempt have been made to associate many f
longer feature detected in this way and the pectral the ubjective characteri tics of in trum nt with
and frequency fluctuation become clearly audible as recognisable features of re pan e curv - we have
uch. already mentioned tbe 'p ctrum vibrato' effect.
Response curves and input admittance curves can Another point which eems fairly well
be readily calculated for implified theoretical established29.3s.36 i that relatively trang respon e at
models of instrument bodie, and the e calculation the lowe t frequencie i a characteristic of tho e
compared with mea urements uch a Fig. 3. Such violins traditionally most admired by mu ician .
compari on enables the model to be adjusted to Relative strength' is a matter of me d lica y: 'the
bring it into closer agreement with the observation , master maker', it appear, mu t can rve every

162 INTERDISCIPLINARV SCIENCE REVIEWS, VOL. 3. NO.2, 1978 © He den & n Ltd. 197
decibel with miserly care,.44 It is of interest that the ment being played, a viola in thi case. ven th ugh
a-called soundpo t' in ide the body, known in this nece sitate a mall loud peaker, the r ulting
French as the 'soul' of the instrument, play a crucial sound seem rather more reali tic than any that
role in the low frequency respon e. It a ymmetric Mathew and Kohut have yet achiev d. One possible
po ition promotes the 0 cillatory volume changes reason is the more reali tically variable radiati n
which the air in the cavity must undergo a every pattern of the hybrid instrument b th a the
acoustician knows, if there i to be efficient radiation frequency changes and as the play r moves during
at low frequencies.5.45--47 performance. ( ee (i) and U) on p. 161.)
Attempt to discern further subjectively important The more reali tic feedback to the player may also
features of re pan e curve run into two evere be ignificant. Another po ibility is that, although
difficulties. The fir t difficulty is the obviou one, that Gorrill' viola body i trongly trutted internally to
with real in truments it is never po ible to vary one minimi e vibration, there i a mall amount f 'liv
parameter while keeping everything el e constant; ound from the in trument which may be audibl , in
any two real instrument differ in a large number of particular in the tarting tran ient f note to which
ways. The second difficulty is that of communicating the ear i a sensitive. The y t rna tic blindfold
ubjective impressions in word : tl1U Meinel,35 in listening tests required to decid thi last point hav
the cour e of a ummary of many years' work on not yet been made as far as we know.
response curves, states that mall amplitude near Much more work could, and hopefully will be
1500 cycles per second prevent a very nasal charac- done on this key experiment. A very ba ic que tion
ter' and it is clear from the context that thi na al is: what kind of changes in respon e curve ar lea (
quality is can idered undesirable. The word 'nasal noticed by Ii teners? This que tion could in principle
however ha different meaning for different be tudied by the powerful 'objective listening te t'
authors: to Leipp4 it i one attribute of the desirable techniques used in modern p ychoac u tic ,52
old Italian ound. although the magnitude of the task i daunting. A
What i needed i the type of experiment indicated start has been made by Jansson and Gabri I on 5
in the third section, and thi has been very ele- using the technique of long-time-avera e pectra.'40
49 A related question i ugge ted by the violini t'
gantly provided by Mathews and Kohut. They
converted the transver e motion of a et of violin need, especially in chamber music, to command th
trings into an electrical ignal, and then passed thi gTeate t p ible range of perceived ton quality. At
ignal through a et of electronic filters giving a piky lea t two factors contribute to pr dueing uch a
respon e function like that of Fig. 3 before Ii tening range. First, some violin may produe a greater
to it through a loudspeaker. Thi approach opens variety of waveform than th rs, for instance
two very profitable lines of enquiry. Fir t, the elec- through the mechanisms to be di cu don pp. 16
tronic re pon e function can be varied at will, and 169. Second for a given rang of bowed tring
keeping exactly the same recorded player input so input waveforms, some re pan e curves more than
that tatements like Meinel' can be put to others may bring the range of output waveform into
controlled listening te t. econd, the strengths and a part of perceptual space,52 where the ar notices
limitations of the re ponse curve model it elf are the changes most.
revealed by the extent to which the hybrid violin'
can be made to sound in the hand of a competent
player like a monaural recording of a good real
violin.
I TERPRETI G THE
The conclusion of Mathews' and Kohut's pilot TRADITIO
tudy was that a rea on able first approximation to
the sound of a real, monaurally recorded violin could In tead of trying to find out how we want a violin
indeed be produced. The subjective quality was body to behave we now look at the instrument-
fairly en itive to the sharpne of the resonance making tradition to ee how a skilled luthier trie to
peak: this appear attributable to the effects on make it behave. This is a very large field, and there is
5o by no mean perfect agreement between makers on
vibrato and on other tran ients. But none of the
re ponse function u ed by Mathews and Kohut how the main problems hould be handled, or indeed
produced a re ult good enough to fool trained rou i- on what the main problem are. We shall c ncentrate
cians' ears. It i not yet clear whether thi failure on a single example which has shown the rt f
is due primarily to unrealistic features of the work whieh can be done in this area: we examin the
re pan e function, or whether it points to major methods u ed by maker to arrive at th be t dis-
inadequacie in the re pon e curve model itself, tribution of thickne in the back and front plate f
such a the neglect of string motion other than an instrument. Thi proce i apparently regarded
transverse. by rna t maker a a major one in determining th
One exten ion of Mathews' and Kohut's experi- sound and the playing properties of the flni h d
ment i being tried with orne ucces by Gorrill,51 in trument, and has been investigated in am detail
who ha incorporated the loud peaker playing the by makers who use acou tical mea urement tech-
processed string signal into the back of the in tru- nique , notably Meine125.35 and Hutchin .54

© Heyden & Son LId, 1978 INTERDISCIPLINARY SCIENCE REVIEWS, VOL 3, NO.2, 1978 163
 We de cribe a commonly u ed method: once the used for studi of violin plate and complete
outside hape of a plate i completed, the violin instruments by Stetson and Agren 26 Jansson, Molin
28
maker h Ilows it out to achieve a thickness dis- and Sundin,27 and Reinicke and Cremer. Figure 4
tribution approximating a standard pattern. shows the Chladni pattern and the hologram of the
However no two piece of wood are identical, so no same mode of a violin t p plate.
one pattern can be perfect for all plate. Thu the The results obtained by the e methods are
luthier make mall adju tments, occasionally hold- ufficiently complex that some the retical in ight i
ing the plate up lightly between finger and thumb needed to interpret them usefully. When n trie to
and tapping it with a finger. He listen carefully to do the theory of plate vibrations it bee me apparent
the note produced, and more particularly to the that, quite apart from the c mplication a dated
quality of the note, and he tries by his thickne with the geometry of the violin, great pr bl ms are
adju tments to achieve what ha been described a a posed by the anisotropic tructure of wood. Wherea
'clear full ring'. He might do thi for several po i- i otropic material have only two independent ela -
tions of holding and tapping, obtaining in thi way tic constants, wood has nine in g neral and at least
more than one note. four are significant for flexural vibrati n of thin
In more scientific terms he make heuri tic obser- plates. While all nine con tant have been measured
vation of the frequency, damping and mode hape for a few specimen (not generally w od of inter t
of one or more vibration modes of the plate, and to in trument maker: ee, e.g., HearmonS7), mo t
adju t the thicknes distribution apparently to large programmes of measurement Ii t only tw ,5
minimise the damping of the e modes. Some makers In addition, since makers feel that damping i of
are al 0 concerned to place the frequencies of crucial importance the viscoela tic damping
certain modes in a particular place or in a particular con tants corresponding to each of these ela tic
relation to one another. A remarkable feature i the con tants should be measured before a relevant
en itivity of the method: sometimes a difference can theory is con tructed. Information on the e damping
be heard a a result of removing just 0.1 mm. of constants i even more limited. We should n t that
wood from a few square centimetres of a plate of varni h and other treatments of the w od will
ome 3 mm. thickness. modify the vi coela tic constant 44 and ome
Hutchin 54 come to the urprising conclusion that maker partially varnish their plate before final
subjectively good re ults in the fini hed instrument tuning.
can be achieved by confining attention to a few of One interesting experimental approach to this
the gravest modes of the plates in their unattached problem i the recent effort by Haine 59 to tailor-
state. Their frequencie cover only a small part of
the audible range, and in any case the plate behave
quite differently under the different boundary
conditions in the a embled in trument. Thu it
would seem that the free plate modes are in ome
way acting a barometer for the behaviour of the
plate under fairly general circumstance, and if thi
could be under tood we might learn much about
desirable behaviour of violin b dies. Some preli-
minary gue es about uch barometer effect have
been made n the basis of modelling simpler y tems
. I'In pates.
t han VIO I 55

The extensive measurement of Hutchin have

revealed much detail about the change in the vibra-
tion properties of plate over the entire audible
frequency range as uch plate tuning operations are
Figure 4. Chladni pattern (left) and holo-
carried ut. Three main methods have been used to graphic interferogram (right) of a violin top
thi end. One i to measure frequency re pon e plate vibrating in one of its natural modes, at
curve of the plat by driving it inu oidally with an about 320 Hz. Hutchins calls this the 'ring
electromagnetic transducer and detecting the resul- mode' and considers it a particularly valuable
ting radiated ound at a tandard microphone posi- guide to the final adjustment of plate thick-
ness. The fringes in the interferogram
tion in a tandard room.56 The other two are ways of indicate contours of constant vibration amp-
visualizing individual vibration mode of the plate, litude. Since the interferogram shows the
to see how their frequencies, shape and amplitudes outside and the Chladni pattern the inside of
vary. the plate, the left hand photograph has been
The fir t method i the classical one of observing reversed to facilitate comparison. (By
permission of C. M. Hutchins and K. A. Stet-
Chladni pattern : the plate i cau ed to vibrate in the son). For a useful comparison of inter-
required mode and powder i sprinkled on it to ferograms for well- and badly-adjusted
make the nodal lines visible. The econd is the much plates, and related acoustical measurements,
more modern technique of hologram interferometry see Ref. 56,

164 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2, 1978 © Heyden & S n Ltd. 197
make a synthetic material which reproduce the haws the four lowe t modes of a constant thickne
relevant properties of pruce, the wood used for the square plate of isotropic material with fre edge.
oundboards of all tringed instruments. The pro- The pictures show the plate in per pective with its
perties of man-made materials can of course be vibrations frozen a by a troboscope. Th amplitude
manipulated over a wider range and in a more is exaggerated for clarity. olumns two and three
controlled way than can be achieved with wood. indicate what happens to the frequency and damping
Nevertheless it i remarkably difficult to imitate even of these mode when a small perturbation of thick-
a few of the properties of spruce: the reason for the ne i introduced. If a mall amount i removed
traditional role of that wood in instrument making i from the plate in a region where the column two
it almost unique combination of extreme pro- function is positive, the frequency of that mod will
pertie .22 drop, while if the column three functi n i positive
The only material currently available which there, the damping will rise. Thi applie to each
combines the low den ity high along-grain stiffness mode separately, so that a large enough t of the e
and low damping of spruce is a carbon fibre picture would show how ev~ry mode of intere t i
composite used in a andwich construction with a affected by the thicknes perturbation. Analogou
light core material such as cardboard or balsa wood. pictures can readily be computed for flat wooden
A violin and a guitar have been constructed by plates, but arched plates pre ent computational
Haines and his collaborators with soundboard of problems not yet tackJed.
such a material and both have been surprisingly
s9
successfu1. We can hope for much more progress in
the near future on thi approach not lea t becau e it
ha commercial possibilities until recently a rare THE BOWED STRING: TOLERANCE
occurrence in musicaJ acou tics. RANGES AND WOLF NOTES
As an example of the sort of theory which can be
done when the vi coelastic constants are known, the We now return to our general con id ration of levels
sS
authors have explored the simpler problem of the of idealization in modelling the violin, and examine
effect of thickness perturbation on the damping what is known about the motion of th bowed tring.
of vibrations in isotropic materials. This has given It is at this point that we first confin our int rest
some clues about the relevance to the finished trictly to the bowed in truments: our discu ion of
instrument of the behaviour of the plates before b dy behaviour is applicable in general term to
assembly, and about ome of Hutchin' other p Iuc ke d·Instruments.' 33 60 The Iarge t stnng
. motIOn
.. I
ob ervation . transver e, and most is known about thi ; the longi-
The results can be represented in pictorial form tudinal and torsional motions, as well as the effect of
and we illustrate this in Fig. 5. The first column bow hair motion, are less well explored.

Figure 5. Computer cal-

culations for vibrations of a
square, isotropic plate of
constant thickness. The first
column shows the lowest
four vibration modes, with
the nodal lines marked. If a
small perturbation of thick-
ness is introduced, the
change in frequency of each
mode is given by the
integral of the perturbation
with the function shown in
column two, while the
change in internal damping
is given by the integral of
the perturbation with the
function in column three.

© Heyden & S n LId, 1978 INTERDISCIPLINARY SCIENCE REVIEWS. VOL. 3. NO.2. 1978 165
 Frictional visible to the naked eye. ate that the form of thi
force
Helmholtz motion' is to a large extent independent
of the po ition of the bow, in contra t to the varia-
tion with plucking point of the motion of a plucked
Relative string.
velocity
For most of the cycle the tring tick t the bow,
until the 'corner' dislodge it, at which the string flies
back rapidly to be recaptured by the bow when the
corner arrives back from the bridge. Thi e ntial
timekeeping role of the Helmholtz corner i what i
missing from the usual layman's explanation of the
bowing mechani m. A imple but very effective
Figure 6. Frictional force between bow and
string idealized as a function of their relative
theoretical de cription of the mati n wa given by
velocity, for constant normal bow force. Raman 2\ who worked on violin ac u ti in his early
year before turning to the work in pectro copy for
which he won a abel prize. Raman al a tudied
The tran ver e vibration of a stretched tring ha the hierarchy of pas ible mati n having m r than
been tudied ince the earliest days of mathematical one corner travelling on th tring. uch higher
phy ics, and in most textbooks the string is used a types of motion are readily produced when the
the first and simplest example of a continuou vibra- bow is not too cia e to the bridge, and they are
tory system, that.i a y tern having an infinite series u ed by musician for colouri tic effects in sui tasto
of normal mode of vibration and corresponding playing.
natural frequencies. Many scienti tare thu The u ual concern of the player, however i to et
urpri ed to learn that the tran ver e motion of a up a Helmholtz motion in the tring. He mu t
bowed tring presents many un olved problem of control three parameters: bow peed, po ition of the
both practical and mathematical intere t. The rea on bow on the string, and force between the b wand
is that the vibration are excited by the frictional the string, known perver ely by mu ician a bow
force between bow and string, and this force is a pressure. He must keep the thr e quantitie within
severely nonlinear function of the relative velocity of certain range in order that the teady vibration can
the bow and string 5 as illu trated in Fig. 6. This exi t: the e tolerance ranges vary among different
renders the usual mathematical techniques inap- violin and are evidently extremely important pro-
plicable. Some of the complexity of the problem is perties of an instrument. A second and more ubtl
perhaps intuitively apparent if one considers the problem is that he must control th nature and
large variety of ound elicited from a violin in the duration of the tran ient motion f the tring
hands of a novice, compared with the effort of a especially the starting tran ient f the note. The ea e
beginner plucking a guitar string. with which thi can be done may b connected with
The motion of a string during the bowing of a the teady state tolerance of the instrument, but
steady, mezzoforte, mu ically acceptable note with there i a yet little definite knowledge
the bow fairly near the bridge was first observed by aval'1a bl e. \53162
.. W e d'ISCUSS maIO . Iy t he ca e a f a
Helmholtz.61 The form of this vibration is somewhat teadily bowed note.
surpri ing: a iJlustrated in Fig. 7, at any in tant the For definiteness, suppose that the p ition and
string i in two approximately straight portion speed of the bow are kept can tant whil the player
epa rated by a corner which travels back and forth, gradually increa e the force from zero. At first he
tracing out the curved 'envelope' of the vibration elicit a urface ound in which the fundamental i
weak, as in sui ponticello playing. He will th n find a
rather sharply defined minimum bow force at which
the Helmholtz motion tart - rou ician call thi
'getting into the tring' - and a less well defined
Bridge Nut maximum force where the note either g e unac-
ceptably out of tune or give way to a raucou ound,
~~~ depending on detailed circum tance .
Bow ----~
---
---------- Minimum bow force was fir t tudied theoretically
and experimentaUy by Raman.6 Further p ri-
Agure 7. The simplest motion of a bowed
mental confirmation of aspect of hi theory ha
string, first observed by Helmholtz. The
transverse scale is exaggerated. The visible been obtained by Saunder 31 and Lazarus.5 A
envelope of the vibration is shown dashed, synthesi and exten ion of Raman' early work wa
together with two 'snapshots' of the moving recently given by ScheJleng 14 in whose penetrating
string at different instants. The position and discu ion the first explicit theoretical formulae for
direction of motion of the bow are indicated,
and the directions of travel of the Helmholtz
both bow force limits were given a well a a good
corner. The string is sticking to the bow at deal of physical in ight into the b wed tring in
both instants shown. general.

166 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2, 1978 © Heyden & n Ltd. 197
 1.000 variation of tolerance am ng in trument , and from
.,
<J ....
note to note on one instrument.b3
0.100 The 10 se tend to be greate t, and hence the
~
.,
.~
bowing tolerance least, at fundamental tring
iii frequencies corresponding to re onances of the
"ii 0.010
a: ba dy. 31.64 AI 0, at uc h f'requencles unu ually large
.
amount of energy may be tared in body vibration;
0.001
these are particularly noticeable to the player in the
ca e of the large peak near the top of the fir t ctave
in Fig. 3. This ability to tore energy at re nance
Figure 8. Graph showing the trends of
maximum and minimum bow force as the implies a time lag in the re ponse of the body which
bow is moved away from the bridge, for a can evidently affect the instrument' tran ient
50
given bow speed and a given note on the behaviour. Such a time lag is known to play a
instrument. The slope of the minimum bow crucial role, moreover in that m re bizarre form of
force line is twice that of the maximum bow
intolerance the 'wolf note'.
force line. The graph shows immediately why
bowing near the bridge demands greater
control On the part of the player. (By permis-
sion of J. C. Schelleng and J. Acoust. Soc.
Amer.)
THE WOLF OT

Suppose that we bow a note who fundamental

He summarized the re ults in a graph of typical frequency coincides with a trong body resonance
force limits again t the bowing point a a proportion and that a Helmholtz motion i set up at the tart.
of the string length, reproduced in Fig. 8. Schell eng s The energy stored by the body takes a number of
approximate formulae, which in practice should be cycles to build up, and during thi time there i a
viewed a giving order of magnitude only, express continual increase in the rate of energy 10 from th
the fact tj1at above maximum bow force the arrival string, and hence in minimum bow force, which is
of the Helmholtz corner from the nut i insufficient dominated by 10 es at the fundamental, a
to cause the string to lip, while below minimum mentioned before. If the minimum bow force
force the bow fails to keep hold of the string while needed exceed the actual bow force b fore a teady
the Helmholtz corner is travelling between bow and tate i reached, the Helmholtz motion gives way to
nut. The urface ound elicited below minimum bow a 'double lip regime, who e on et is indicated by
force involves two or more slips per cycle. the arrow in the top curve of Fig. 9. The energy
We should note that the picture of tolerance tared in the body is in just the right pha e to pro-
range given by Fig. 8 become over imple when the mote the growth of this second lip which tak over
bowing point moves too far from the bridge. Ca ual as a new Helmholtz motion, out of pha e with the
experimentation confirm what Raman's theory old one. Then the whole cycle repeat it elf and the
predict , namely that as the bowing point is moved result i the characteristic tuttering ound which i a
toward a imple fraction such as one-fifth or one- particularly common problem with certain notes on
quarter of the string length the tolerance range cello.
fluctuates wiJdly. The main reason j the appearance
of the higher type of vibration already mentioned,
which are more complicated than the Helmholtz
motion.
It is in tructive to look more closely at the
mechanism behind minimum bow force. For an ideal
textbook string with rigidly fixed ends, minimum
bow force would become zero. The Helmholtz t 20 ms
motion i a free motion of such a tring, a that the
bow would not be required at all to maintain it once
it had started. In the real situation the bow i needed
to su tain the motion against small los e principally
from the ends of the string5•19 and ultimately attri-
butable in part to radiation los es. As Schelleng
made clear it is the rate of loss of energy from the Frequency --_
fundamental that largely determine the position,
though not the slope, of the minimum bow force line Figure 9. Simple wolf note (played on the G
in Fig. 8, and hence the bowing tolerance range for a string of a viOlin). Top: waveform of trans-
given bow po ition and peed. The 10 se in tum are verse force exerted by string on bridge
(measured by the authors USing a piezoelec-
largely determined by the behaviour of the instru-
tric transducer developed by W. Reinicke).
ment body. This appears to be a major rea on for Bottom: corresponding frequency spectrum.

© Heyden & Son Ltd, 197

INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3. NO.2, 1978 167
 This qualitative explanation of the simplest kind inve tigate their stability. Predece or to thi line of
of wolf note i ub tantially due to Raman,6 work were the tudies by Friedlander69 and Keller,70
although he did not make clear the importance of although they used the imple t textbook idealisa-
the phase reversal between alternate cycles of the tion of the string. Their models produced orne
wolf. This latter point was first brought out by interesting mathematics - indeed this i the only area
Schelleng,22 who confirmed Raman s picture using of our subject which ha so far yielded problems of
an alternative approach to the imple t wolf note great formal interest - and Friedlander found the
starting from the idea that 'its behaviour immedi- ignificant re ult that di ipation i e ential for
ately suggest beating and coupled circuits'. table, periodic motion.
In technical language Schelleng's is a frequency- In the model tudied by Raman, Fri dlander and
domain viewpoint complementing the time-domain Keller there is no change in the tring motion a b w
view of Raman. As might be expected in a trongly force is varied between maximum and minimum. An
nonlinear problem, ome feature are ea ier to ideally flexible tring with a frequency independent
under tand from one viewpoint and vice versa but loss mechanism gives no corner rounding, and a
the two approaches are consistent. The frequency- Cremer and Lazarus71.72 first pointed out corner
domain viewpoint is suggested by the econd curve rounding is necessary for the b erved variati n of
of Fig. 9, which show the frequency spectrum of the string motion within the toleranc rang. W thu
wolf note in the top curve. The fundamental is plit require rounded corner model t tudy uch varia-
into two peaks of comparable amplitude a predicted tion. (It is in constructing uch mod I that th input
by Schelleng, and the wolf phenomenon can be admittance curve mention don p. 162 are relevant.)
de cribed in term of beating between these. Such model also make po ible a clo er tudy of
Raman's viewpoint readily explains why pressing maximum bow force. As chelleng realized,14
harder with the bow can suppress the wolf at least in maximum bow force in practice is generally less than
some ca es, while Schelleng' led him to a imple the force causing cata trophic breakdown of the
quantitative criterion for susceptibility to wolve mu ical ound to a raucou ound and it i ignalled
which haws why cellos are more prone to them than by one of everal Ie dra tic phenom na. Of the e
violin, although Fig. 9, a it happen, come from a phenomena, two matter a great d al to the mu ician.
violin. Both approache explain why wolves can be The first i that the noi content f an oth rwi e
alleviated by fitting a correctly tuned wolf elimina- musical note can reach an unacceptable level; we
tor66 or a lighter string either of which reduces the shall return to this subject in the next section.
coupling of the string to the body and hence the The second is the very slight deviation of pitch -
maximum stored energy. More complicated wolves, almost always on the flat side of the string tuning - a
orne of which are analogou to a called multi- bow force increa e . Thi flatt ning i ea ily d mon-
phonics in wind in trument have been ob erved by trated with the bow a moderat di tanc fr m the
Raman,65 Firth and Buchanan67 and others, and bridge at a low bow peed.73 When flatt ning i
have been discussed by Benade.3 A quantitative audible, pitch is sen itive to bow force that control
theory covering aU known cases has yet to be con- of intonation become difficult. Player av id thi
74
tructed and verified. regime. It ha recently become apparent .75 that the
flattening effect is strongly connected with th
mechanism of waveform change through corner
rounding, and so we discuss both phenomena
MORE REALISTIC MODEL ROUNDED together.
COR ERS When a rounded Helmholtz corner pa e th
bow, the effect of the friction curv n nlin arity i t
Real strings on real instruments have slightly harpen the corner to an extent which depend very
anharmonic overtone when allowed to vibrate much on bow force. Thu a periodic motion of the
freely, and 10 e occur at different rate for different string represent an equilibrium b tween c mer
overtones.3.5.19.6 One re ult of thi i that the rounding by the tring and it t rminati n , and
Helmholtz corner i not perfectly harp, but i corner sharpening by the b w. A b w fore i
somewhat rounded. Such corner rounding influences increased corner sharpening at the bow become
the string's response to the bow in a musically more pronounced, so that the periodic solution has
ignificant way, and model taking account of it are more high frequency content.71.72
needed. Such model are analogou to tho e u ed by A further effect come int play when b w force
Benade and others3 in work on wind in truments. exceeds a certain value: the am unt f c mer har-
However, the friction curve nonlinearity i much pening during the tran iti n from ticking t lip-
more ill behaved than the nonlinearities encoun- ping, release, i greater than that during the tran-
tered in wind instruments so Benade's approach ition from lipping to ticking, capture.75 Th re ult
po e greater difficultie in the case of the bowed of thi a ymmetry i not only a furth r chang in th
tring. shape of the periodic waveform, but al a a delay in
The mathematical technique most promising at the round trip time of the Helmholtz corner. In other
first sight is to seek exactly periodic motions, and words, as bow force is increased beyond a certain

168 INTERDISCIPLINARY SCIENCE REVIEWS, VOL 3, NO.2, 1978 © He den & Son Ltd. 197
limit there is a tendency for the note to play flat, and far. Such departures have been observed: remer72
the degree of flattening increases with bow force as and other have reported measurement of the
observed.7 76 period length of many cycle of a bowed string
There is yet another can equence of corner shar- signal in which variation of up to 30 cent between
pening by the bow, first pointed out by Cremer and the shortest and longest cycles were found (a cent i
Lazaru 71.72 and further elucidated by SchelJeng.14 a hundredth part of a semitone). The observation
When the hape of the corner transmitted past the rai e two que tion . Fir t is this the source of the
bow is changed, <l.reflected wave i also generated, audible noi e? Psychoacou tical data 78 taken with
travelling in the oppo ite direction. These secondary headphones suggest that uch jitter r random
waves can then reverberate in ODe ection of the variation of period, need nly attain 20 cent r to
string, reflecting from the ticking bow, until they be audible. Second, what i th ource of th varia-
happen to arrive at the bow during a lipping phase tion? It could be simply a reflection f e t mal
in the Helmholtz cycle. The reverberating econdary factors such as the unequal distribution of ro in on
wave give ri e to a characte 'stic pattern of ripple the bow, but on the other hand it could in orne way
in the tring motion away from the bow which is be intrin ic to the bowed tring mechani m as
almo t certainly an important ingredient of tone Cremer suggested. This would eriou Iy limit th
quality,77 although the requi ite Ii tening tests have usefulness of theoretical studie of trictly p riodi
yet to be made. motion.
[n real strings ripples are also visible in the wave- The present authors have recently made ome
form of string velocity at the bow: during the ticking careful mea urement of jitter. The extremely teep
part of the Helmholtz cycle the observed velocity i flyback in the awto th wave felt by th bridg when
not exactly equal to the bow velocity.5.14.71 This is a Helmholtz motion is imparted to th tTing make
not urprising, since the bow hair can yield lightly; po sible an accurate determination of period length.
moreover the string can roll on the bow hair. In fact One example from our re ult is hown in Fig. 10: it
the latter effect is the larger of the two,5.13.14 and appear that maximum f1yback jitter' can be as little
indeed tor ional motion are an important aspect of as 3 cents for open string on a real in trument. Thi
bowed string behaviour altogether. They affect the observation sugge t that the bowed tring i indeed
detail of the flattening and corner- harpening effect capable of precisely periodic motion under at lea t
already described and they result in extra 10 e, tbu orne playing conditions. On the other hand
changing the bowing tolerance limits. Since torsional different conditions, uch as when playing a note
yielding of string i affected by tring diameter and high on the violin G tring, can pr duce much mar
construction while it is not much affected by tring jitter; and in fact the amount f jitter i fund to be
tension, it i clear that bowing tolerance will vary roughly proportional to the amount of corn r
with different types of strings, as player are well rounding, in the sense of the previou ction. Thi i
aware. consistent with the idea that unevenness of bow hair
A theoretical inve tigation of the effects of torsion and rosin is primarily responsible for jitter.
pre ent no great difficulty in principle: any method We hould not jump to the conclusion that jitter of
capable of analysing transver e motion can rather Ie than 20 cent can have n mu ical effect.
imply be extended to include tor ion.14.76 The Benade3 (econd reference) ha ugge ted that in
tor ional a cillations are known to have a con equence of the extraordinary en itivity of the
fundamental frequency several times higher than acoustical behaviour of a large c ncert hall t any
tho e of the tran verse motion,5.14 but their rate of light unsteadiness in the ound urce jitter may
damping ha not yet been ati factorily measured. reduce the carrying power of an instrument that i
More detailed observations and theory will no doubt to ay it audibility in a concert hall against a back-
become available soon. ground of other ound. Th amount f jitter needed
to affect carrying power might be mailer than the
threshold for direct audibility in headphone.
Our jitter mea urements do however ugge t that
POWER, AND OTHER the audible noi e i not connected with flyback jitter.
Even during loud playing near the bridge, measured
flyback jitter often remains well below 20 c nt .
The ound of a real bowed tring i more or less The ource of the audible nib com apparent,
noi y, and in the previous ection we mentioned that however in Fig. 11. This shows transver e force at
maximum bow force can in some circumstances be the bridge during a noisy note, which for an ideal
governed by the buildup of noise. This is especially Helmholtz motion would be a sawtooth wave. We
apt to happen when trying to play more and more note the aperiodic pike uperimposed on this
loudly near the bridge. This noisy regime i often sawtooth: many ob ervation have hown that these
used to deliberate mu ical effect but the noi e can always appear when the audible noi e build up.
reach an unacceptable level, depending on context. The main clue about the ource of pike came
The presence of noise indicate orne kind of from experiments in which tring wer bowed with
departure from the periodic motion di cus ed so a smooth, round, rosined stick in place of a bow. It

© Heyden & Son Lid, 197 INTERDISCIPLINARY SCIENCE REVIEWS, VOl. 3. NO.2. 1978 169
 5 ms

Figure 11. Waveform of transverse force

exerted by string on bridge for a loudly
bowed open violin G. The main 'flybacks'
occur when the Helmholtz corner reflects
from the bridge, and their timing is extremely
regular. By contrast, the 'spikes' in between
are aperiodic, and cause audible noise.
(Measured by the authors using a piezoelec-
tric transducer developed by W. Reinicke.)

76
motion. The mathematical problem is clo ely
analogous to that of incorporating t r ional moti n
of the. tring, ince again the extra equation involved
-2 -1 0 2 is linear, and couples to the others only through th
Cents friction force. hi i only th ca e how v r as long
as the bow is idealized a ha jng infinite imal width.
F:~ure 10. Statistical distribution of period To allow fully for finite width of the bow and thu
lengths (expressed as departures from a
imulate spike production would be a much more
nominal constant value) for 560 cycles of a
Helmholtz motion bowed by hand on the complicated undertaking.
open E string of a violin. Special care was The final complication is longitudinal tring
taken to minimize spurious effects such as motion.3.15.48 Since the length of a tring i increa d
drift in string temperature. The total spread is somewhat when it vibrate tran ver Iy, its ten i n
only three cents or 0.2%.
will also increase, 0 that at the bridge th fluctuating
force will have a component along the tring as well
a transver e to it. A compon nt of thi will be
should be noted incidentally, that such a stick i exerted downward on the bridge and hence on the
much closer tban a real bow to rna t of the theoreti- in trument. The waveform of this 'indirect excitati n
cal ideaJization used 0 far, and a such may prove will depend on the behaviour of th longitudinal
an important tool for linking experiments, listening waves in the string; their fundamental frequency i
test and theory in a wider context than the present typically about three octave above the fundamental
investigation of spikes. With the stick spikes are tran ver e frequency, as is well known to player who
conspicuou Iy absent from the bridge-force wave- inadvertently excite them when cleaning ro in from
form; and further experimental and theoretical evi- their strings.
dence has recently made it clear that the finite width The coupling of these waves to the transverse
of the ribbon of bow hair in contact with the string is motion i nonlinear; 17 moreover one fiT]d that
the essential ingredient in spike production.74 transver e motion with sharp corners in it, like the
This discovery reinforces a point made by Schu- Helmholtz motion, i particularly effectiv for exci-
macher, l3 that we should examine more clo ely the ting longitudinal motion. Also, th nonlinear c up-
influence of the bow it elf. The bow hair and the ling implies that a feature of the tran verse tring
stick to which it i attached are by no mean rigid, displacement which we omitted for clarity from Fig.
and will yield to some extent under the fluctuating 7 become important. There i a tationary comer on
friction force. Indeed we would expect to find some the tring at the bow caused by the friction force
significant effect of bow dynamics on string motion there; and thi force ha a sub tantial teady
since player can distinguish between different bow component. The re ult i a ignificant indirect exci-
by their playing propertie . tation at the fundamental a well a the octave and
Investigation into bow behaviour has been rela- higher harmonic, a point which distingui hes the
tively neglected in the past. Measurements on violin from the guitar and which ha sometime been
variou bow have been made by Schumacher, 13who overlooked.
has also taken the first step toward incorporating It appears certain that longitudinal motion i
bow hair motion in a theoretical treatment of string excited in a bowed string, and may have audibl

170 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2, 1978 © He den & on Ltd. 197
 consequences IS - especially in starting transient, eventually be augmented when a better under tand-
and especially with metal strings which have a higher ing of what makes a top cia in trument, aM. top
Young's modulu than gut string. The strength of class bows and trings, enable them to be produced
the indirect excitation i obviously sen itive to the lightly more reliably .• Even more important uch
angle the strings make with the bridge, a parameter knowledge might rai the standard of mass pro-
long regarded by maker a critical. A promjsing first duced tudents' instruments from the pre nt very
step toward a essing the importance of indirect low level to something at least decent, if not
excitation would be to combine Mathews' and excellent. If in the process our under tanding f
Kohut's technique with a simple computer simula- auditory perception mechanism al develop, a
tion of the 10ngitudinaJ string motion, to give the was suggested in the introduction, that will be a
po sibility of a Ii tening test in which the indirect valuable bonus.
excitation could be witched in and out.
In another ten year, a review similar to this
hould be able to report very substantial progres
toward understanding a number of feature of Acknowledgements
stringed instrument behaviour which are of real
concern to mu icians. As has been made clear here, We thank C. Blakemore, C. M. Hutchins, ambridge
University Press and Weidenfeld and Nicol on td. for
many lines of investigation are now being pursued
help in obtaining material for figures' and . H. Benade,
which promise to move toward this goal many of L. Cremer, P. B. Fellgett, R. Hecht, C. M. Hutchin , A.
which have only recently become available with the Ki s, J. Meyer, A. Pay, J. C. Schelleng and R. T. Schu-
development of computer technology to its pre ent macher for helpful comment during preparation of the
level. manuscript. Our own research ha been supported by the
Science Research Council and by lare ollege,
either violin makers nor violin players will be
Cambridge, and experimental facilities were kindly made
displaced by this knowledge, but possibly the world's available by J. E. Ffowes William and the Cambridge
very limited supply of top class in truments will University Engineering Department.

LITERAT CITED

Note that some of the papers referred to below, as well as 8. Hermann Hesse's novel is set in a society whose culture
many others relating to violin acoustics, are reprinted in is founded on a supposed union of mathematics and
Musical Acoustics, Parts I and II, ed. by C. M. Hutchins music.
(Benchmark Papers in Acoustics 5, 6), Dowden, Hutchinson 9. M. J. Lighthill,Artificialintelligence: a paper symposium.
and Ross, distributed by Wiley-Halsted, New York and SRC report, Science Research Council, London
London (1975/1976). The relevant papers are marked'" (Part (1973).
I)and t (Part II)below. 10. J. C. Risset and M. V. Mathews, Analysis of musical
Much of the most recent work is as yet reported only in the instrument tones. Physics Today 22, 2, 23 (1969). An
Catgut Acoustical Society Newsletter. Readers having no earlier example of analysis-synthesis work is H. Flet-
access to this may write to: 112 Essex Avenue, Montclair, N.J. cher,t E. D. Blackham and O. N. Geertsen, Quality of
07042,U.S.A. violin, viola, cello and bass viol tones I. J. Acousr. Soc.
Amer. 37, 851-863 (1965).
1. P. C. Boomslitert and W. Creel, Research potentials in 11. H. B. Barlow, R. Narasimhan and A. Rosenfeld, Visual
auditory characteristics of violin tone. J. Acousr. Soc. pattern analysis in machines and animals. Science 177,
Amer. 51,1984-1993 (1972). 567-575 (1972j.
2. J. C. Roederer, Introduction to the Physics and Psy- 12. B. C. J. Moore, Frequency difference Iimens for
chophysics of Music, 2nd Edn., Springer-Verlag, New short-duration tones. J. Acousr. Soc. Amer. 54, 61~19
York(1975). (1973); see also p. 210 of the first reference of 3.
3. A. H. Benade, Fundamentals of Musical Acoustics, 13. R. T. Schumacher, Some aspects of the bow. Catgut
Oxford University Press, New York (1976).See also A. H. Acoustical Society Newsletter 24,5-8 (1975).
8enade, Musical acoustics today: a scientific crossroads 14. J. C. Sehel/eng •.•.The bowed string and the player. J.
in Proc. 9th Int. Congr. on Acoustics, Madrid (1977); and Acoust. Soc. Amer. 53, 26--41(1973).
Musical Instruments in Encyclopaedia of Physics, ed. by 15. F. A. Saunders, A scientific search for the secret of
R. Lerner and G. Trigg, Dowden, Hutchinson and Ross, Stradivari. J. Franklin Inst. 229, 1-20 (1940j.
distributed by Wiley-Halsted, New York and London 16. J. W. S. Rayleigh, The Theory of Sound, § 137. Dover,
(1978). New York (1945).
4. E. Heron-Allen, Violin Making as it Was and Is, Ward 17. P. M. Morse and K. U. Ingard, Theoretical Acoustics.
Lock,London (reprinted 1976). McGraw-Hili,New York (1968).
5. L.Cremer,· Die Geige aus der Sicht des Physikers. Nachr. 18. M. Minnaert* and C. C. Vlam, Vibrations of the violin
Akad. Wiss. G6ttingen: /I Math, Physik, KI. 12, 223-259 bridge. Physica 4. 361-372 (1937); also E. Barton and T.
(1971). Ebblewhite, Vibration curves of violin bridge and strings.
6. L. Cremer and F. Lehringer, Zur Abstrahlung ge- Philos. Mag. 20, 456 (1910).
schlossener Korperoberflachen. Acustica 29. 137-147 19. W. Reinicke..•. Ubertragungseigenschaften des Streich-
(1973); J. Meyer, Akustik und musikalische Auffiihrungs- instrumentenstegs. Catgut Acoustical Society
praxis, Verlag das Musikinstrument, Frankfurt (1972). Newsletter 19, 26-34 (1973). See also Ref. 3, p. 537, and
,. C.Ronan, The music of the spheres. Interdiscip. Sci. Rev. Ref. 5, p. 238.
1.149 (1976). 20. C. M. Hutchins, Personal communication.

© Heyden on Ltd, 1978 INTERDISCIPLINARY SCIENCE REVIEWS, VOL. 3, NO.2, 1978 171
21. C. V. Raman, On the mechanical theory of vibrations of 2-8 (1975). (Abridged English translation in Catgut
bowed strings. Indian Assoc. Cult. Sci. Bull. 15, 1-158 Acoustical Society Newsletter 23, 2-6.)
(1918). The beginning of this monumEllltal work is 37. A standard electroacoustic technique for driving an
reprinted in the first Benchmark volume·. instrument has yet to be universally adopted. An
22. J. C. Schelleng,· The violin as a circuit. J. Acoust. Soc. inexpensive method of exerting an approximately
Amer. 35, 326-338 (1963). For a further explanation of constant force is that advocated by Hutchins,56 using a
certain crucial points see also J. C. Schelleng,t On the very light moving coil placed in a fixed magnetic field.
polarity of resonance. Catgut Acoustical Society The coil is driven at constant current (not at constant
Newsletter 10, 14-18 (1968). voltage as for a loudspeaker) as frequency is varied.
23. E. Jansson and H. Sundin, A pilot study on coupling 38. J. S. Bradley and T. W. W. Stewart, Comparison of violin
between top plate and air volume vibrations. Phys. Scr. response curves produced by hand bowing, machine
2,243-256 (1970). bowing and an electromagnetic driver. J. Acouse. Soc.
24. Examples of black boxes with response functions which Amer. 48, 575 (1970).
have complex structure producing drastic audible effects 39. C. M. Hutchins,t Instrumentation and methods for violin
but whose amplitudes, and hence response curves in the testing. J. Audio Eng. Soc. 21, 563-570 (1972). The reader
sense usually defined, are constant, may be found in J. is cautioned that the curve in Fig. 8 labelled 'Hutchins
R. Stuart, An approach to audio amplifier design. Wire- system,J7 was in fact taken by another technique, invali-
less World 79,387-391 (19731. and M. R. Schroeder and dating the comparison. (This misunderstanding had not
B. F. Logan, 'Colourless' artificial reverberation. IRE come to light when the paper was published.) See also C.
transactions on audio 9, 209-214 (1961). M. Hutchins, Variations in violin testing. Catgut Acousti-
25. H. Meinel.t Ober die Beziehungen zwischen Holzdicke, cal Society Newslett~r 15,15-21 (1971).
Schwingungsform, Korperamplitude und Klang eines 40. E. Jansson and J. Sundberg, Long time average spectra
Geigenkorpers. Elek. Nachri. Tech. 14, 119-134 (1937); ... part I. Acustica 34, 16 (1975).
F. Eggers,t Untersuchung von Corpus-Schwingungen 41. H. Fletchert and L. C. Sanders, Quality of violin vibrato
am Violoncello. Acustica 9, 453-465 (1959); J. C. Luke, tones. J. Acoust. Soc. Amer. 41,1534 (1967).
Measurement and analysis of body vibrations of a violin. 42. S. Cox and P. B. Fellgen, Equivalent-circuit investigation
J. Acoust. Soc. Amer. 49,1264 (1971). of the low frequency behaviour of stringed instruments.
26. C. H. Agren and K. A. Stetson,t Measuring the Catgut Acoustical Society Newsletter, in press.
resonances of treble viol plates by hologram inter- 43. C. M. Hutchins.t Founding a family of fiddles. Phys.
ferometry and designing an improved instrument. J. Today 20, 2. 23 (1967). This article describes a pioneering
Acoust. Soc. Amer. 51,1971 (1972). effort to build a full consort of instru ments scaled from the
27. E. Jansson,t N. E. Molin and H. Sundin, Resonances of a violin. The musical uses of this 'new violin family' are now
violin body studied by hologram interferometry and being explored at the Royal College of Music, London,
acoustical methods. Phys. Scr. 2, 243 (1970). and elsewhere.
28. W. Reinicket and L. Cremer, Application of hologram 44. J. C. Schelleng,* Acoustical effects of violin varnish. J.
interferometry to the vibration of bodies of stringed Acoust. Soc. Amer. 44,1175-1183 (1968). See also P. B.
instruments. J, Acoust. Soc. Amer. 48, 988-992 (1970). Fellgett. Working speculations about 'magic varnish'.
29. C. M, Hutchins,· The physics of violins. Sci. Amer. 207, 5, Catgut Acoustical Society Newsletter 28, 24-25 (1977).
78 (1962). 45. J. C. Schelleng,· The action of the soundpost. Catgut
30. I. P. Beldie, reported in L. Cremer, Darstellung des Acoustical Society Newsletter 16, 11-15 (1971).
Geigenkorpers als ein Schwingungssystem mit vier 46. F. Savart,* The violin. Translated from L 'Institut 8, 54-56
Freiheitsgraden im tiefen Frequenzbereich. Nachr. Akad. (1840).
Wiss. G6ttingen: /I Math. Phys. KI. 1, 1-11 (1976). 47. See pp. 29-30 of H. Meinel,t Ober Frequenzkurven von
Partially summarised in I. P. Beldie, Vibration and sound Geigen. Akust. Z. 2, 22-23 (1937).
radiation of the violin at low frequencies. Catgut Acous- 48. E. Leipp, Le Violon, Hermann, Paris (1965).
tical Society Newsletter 22, 13-14 (1974). 49. M. V. Mathewst and J. Kohut, Electronic simulation of
31. F. A. Saunders,· The mechanical action of violins. J. violin resonances, J. Acoust. Soc. Amer. 53, 1620-1626
Acoust. Soc. Amer. 9, 81-98 (1937); tThe mechanical (1973).
action of instruments of the violin family. J. Acoust. Soc. 50. M. E. Mcintyre and J. Woodhouse, Toward a psy-
Amer. 17, 169-186 (1946); tRecent work on violins. J. choacoustically realistic violin physics. Catgut Acoustical
Acoust. Soc. Amer. 25,491-498 (1953). Society Newsletter 22,18-19 (1974).
32. C. M. Hutchins, A note on the function ofthe soundpost. 51. S. Gorrill, A viola with electronically synthesised
Catgut Acoustical Society Newsletter 21,27-28 (1974); resonances. Catgut Acoustical Society Newsletter 24,
The Helmholtz air resonance of the cello and baritone. 11-13 (1975).
Catgut Acoustical Society Newsletter 26, 5-6 (1976); and 52. L. C. W. Pols, L. J. Th. van der Kamp and R. Plomp,
C. H. Agren, Experimental findings concerning Perceptual and physical space of vowel sounds. J.
cavity-assisted resonances in stringed instruments with Acoust. Soc. Amer. 46, 458-467 (1969).
a soundpost. Physica Scripta 14,179-186 (1976). 53. A. Gabrielsson and E. V. Jansson, An analysis of
33. G. Caldersmith, Low range guitar function and design. long-time-average spectra of twenty-two quality rated
Catgut Acoustical Society Newsletter 27, 19-25 (1977); violins. Catgut Acoustical Society Newsletter 27, 13-19
The guitar as a reflex enclosure (In press) J. Acoust. Soc. (1977).
Amer. (1978). 54. C. M. Hutchins. A sequence of articles in Catgut Acous-
34. E. V. Jansson,t On higher air modes in the violin. Catgut tical Society Newsletters 5, 6, 8, 14, 15, 16 and 19.
Acoustical Society Newsletter 19, 13-15 (1973) and 55. M. E. Mcintyre and J. Woodhouse, The influence of
Acoustical properties of complex cavities. Acustica 37. geometry on linear damping. Acustica 39, 209-224
211-221 (1977). (1978).
35. H. Meinel,· Regarding the sound quality of violins and a 56. C. M. Hutchins,t K. A. Stetson and P. A. Taylor,
scientific basis for violin construction. J. Acoust. Soc. Clarification of 'free-plate tap tones' by hologram inter-
Amer. 29, 817-822 (1957). Also, many more response ferometry. Catgut Acoustical Society Newsletter 16,
curves may be found in previous papers by Meinelt 15-23 (1971).
in German, cited in this reference, 57. R. F. S. Hearmon, Elasticity of wood and plywood.
36. J. Meyer, Akustische Untersuchung zur Klangqualitat H.M.S.a. Forest products research special report no. 7
von Geigen. Instrumentenbau (Musik International) 29, (1948).

172 INTERDISCIPLINARV SCIENCE REVIEWS, VOL. 3, NO.2, 1978 © Heyden Son Ltd. 1978
58. For example I. Barducci* and G. Pasqualini, Misura dell 68. M. Hancock, The mechanical impedance of violin strings,
'attrito interno e delle costanti elastiche del leg no. II I. Catgut Acoustical Society Newsletter 23, 17-26 (1975),
Nuovo Cimento 5, 5, 416-466 (1948). Part II is to appear in the Catgut Acoustical Society
59. D. W. Haines, C. M. Hutchins, M. Hutchins and D. A. Newsletter in 1977.
Thompson, A violin and a guitar with graphite-epoxy 69. F. G. Friedlander, On the oscillations of the bowed string.
composite soundboards. Catgut Acoustical Society Proc. Cambridge Phi/os. Soc. 49, 3.516 (1953).
Newsletter 24, 25-28 (1975). 70. J. B. Keller, Bowing of violin strings. Commun. Pure and
60. For example response curves for guitar bodies are Appl. Math. 6, 483 (1953).
shown in D. W. Haines and M. A. Haltiwanger, A study of 71. L. Cremer and H. Lazarus, 6th ICA Congress, Tokyo,
bOdy resonances and their influence on the performance N.2-3 (1968).
of a folk guitar. Catgut Acoustical Society Newsletter 24, 72. L Cremer, Der Einfluss des 'Bogendrucks' auf die
29-34 (1975). Other studies of guitar acoustics include F. selbsterregten SChwingungen der gestrichenen Saite.
T. Dickens, Tuning of guitar plates. Catgut Acoustical Acustica 30, 119-136 (1974). ·Engllsh translation of an
Society Newsletter 26, 19-20 (1976); and H. L. Schwab earlier version in Catgut Acoustical Society Newsletter 18
and K. C. Chen, Finite element analysis of a guitar and 19, pp. 13-19 and 21-25.
soundboard. Catgut Acoustical Society Newsletter 25, 73. Then there is no significant amplitude dependent shar-
(1976). pening, due simply to the increased mean tension of the
a1. H. Helmholtz, On the Sensations of Tone. Dover, New string, to confuse the issue.
York (English translation of the German edition of 1877) 74. M. E. Mcintyre, R. T. Schumacher and J. Woodhouse,
pp. 8Q-88 (1954). New results on the bowed string. Catgut Acoustical
62. E. Rohloff, t Ansprache der Geigenkliinge. Z. Angew. Society Newsletter 28 (1977).
Phys. 17, 62-63 (1964). This paper reports experiments 75. M. E. Mcintyre and J. Woodhouse, On the fundamentals
suggesting that the subjective 'ease of speaking' of a of bowed string dynamics. Submitted to Acustica (1977).
violin is not correlated with the steady-state tolerance at 76. R. T. Schumacher, Self-sustained oscillations of the
all, but rather with the player hearing enough extreme bowed string. Submitted to Acustica (1977).
high frequency (4-8 kHz) in the starting transient. 77. Ripple waveforms are similar to the waveforms
Experiments like these, attempting controlled generated by the jew's harp, which, together with their
modification of the feedback to the player's ears, deserve perceptual correlates, are discussed in C. J. Adkins,
to be pursued further. Investigation of the sound-producing mechanism of the
63. C. V. Raman,· Experiments with mechanically played jew's harp. J. Acoust. Soc. A mer. 55, 667·670 (1974).
violins. Proc. Indian Assoc. Cult/v. Sci. 6, 19-36 (1920). Ripples are not to be confused with Helmholtz's 'crum-
64. A technicality which we gloss over here is that this is not ples',B' which are described by Raman's model2' and are
true of the 'air resonance', the reason being that its effect due to weakness of certain harmonics when the bow
on bridge motion is more like that of a 'parallel' than a divides the str.ing in a simple ratio.'4.7B
'series' resonance, in the electrical circuit analogy.22 78. E.g. B. C. Cardozo, The perception of jittered pulse trains.
65. C. V. Raman,· On the 'wolf-note' in bowed stringed In Frequency analysis and periodicity de~ection in hear-
instruments. Phi/os. Mag. Ser. 6, 32,391-395 (1916). ing, ed. R. Plomp and G. F. Smoorenburg, A. W. Sijthoff.
66. Descriptions may befound in A. H. Benade3 (p. 574), and Leiden (1970).
J. C. Schelleng, Adjusting the wolftone suppressor.
American String Teacher p. 9, Winter 1967.
67. I. M. Firth· and J. M. Buchanan, The wolf in the cello. J.
Acoust. Soc. Amer. 53,457-463 (1973); also I. M. Firth,
The wolf tone in the cello: acoustic and holographic
studies. Royal Institute of Technology (KTH), Stockholm, The manuscript was received 3 November 1977
Speech Transmission Laboratory, Quarterly Progress
and Status Report. pp. 42-56, Jan. 1975. © Heyden and Son Ltd, 1978

© Heyden & Son LId, 1978 INTERDISCIPLINARY SCIENCE REVIEWS, VOL 3. NO, 2, 1978 173