Innovation in Violinmaking
By Joseph Curtin, Proceedings, International Symposium of Musical Acoustics (ISMA), July 1998
Abstract: The violin is a cultural icon as well as a working tool, and departures from its traditional form have been variously regarded as impossible (it would no longer be a violin), unnecessary (the violin is already perfect), and unacceptable (players would not play it). Is it possible to change, or even “improve” the classical violin? Where would one begin? A number of highly qualified violinmakers and researchers are investigating innovative approaches to violinmaking. While the craft has withstood many similar efforts in the past, makers today are able to learn from acousticians, engineers, and material scientists, as well as from the four centuries of violinmaking which precede them. In order to be successful, an innovation – whether in the acoustical, aesthetic, or ergonomic domain – must offer the violinist some tangible advantage over a more traditional instrument. In this paper the author will consider what it is that musicians look for in a violin, then examine the possibilities of changing the traditional violin in order to give them more of what they want.
The violin emerged, in remarkably perfect form, from the workshop of Andrea Amati in the early fifteen hundreds. Since then it has changed only in detail. Stradivari and Guarneri del Gesu developed the models now considered most successful for concert use. Within fifty years of their deaths, the guild system in Italy disintegrated, and with it the legacy of workshop know-how these guilds had harbored. For the next two centuries the thrust of violinmaking was toward recapturing the “lost secrets” of Cremona: the methods and materials, and the understanding of how to use them, that the old masters were presumed to possess. The eighteen hundreds saw the introduction of the “modern setup” – longer necks, fingerboards, string-lengths, and bassbars. And that, more or less, was that. Most of the profession’s ingenuity has since gone into mechanizing production, creating better strings, making new violins look old, and developing the highly sophisticated restoration techniques needed to keep the Old Italian instruments from collapsing in exhaustion.
While the twentieth century has seen a vast increase in the scientific understanding of the violin, we violinmakers have to a large extent resisted contamination by such understanding. This may be because it too often comes couched in language opaque to us, or because partial understanding – by casting doubt on our day-to-day work habits – can be paralyzing, or because our profession has a kind of over-developed immune system which rejects any ideas not coming from within its ranks. Or it may be that we prefer to believe our beloved instrument operates less by cold physics than by an ongoing succession of miracles. For whatever reason, understanding is too often the last tool we bring to the workbench.
Fortunately, this is changing. An ever-increasing number of violinmakers – and I think much credit for this must go to the Catgut Acoustical Society – are listening to some of the answers science is giving to the fundamental questions of the craft. At the same time, a number of highly qualified craftsmen have turned their attention to innovation. Violinmakers Christophe Landon, Roger Lanne, and Guy Rabut have experimented with new aesthetics for their instruments. The synthetic bows of Michael Duff and Benoit Rolland have achieved wide acceptance by professional players. Research scientist and luthier Charles Besnainou has laid the groundwork for building superb stringed instruments using carbon fiber composites. The above are people whose work I am familiar with. There are undoubtedly others.
For my part, after building some 150 instruments on Guarneri and Stradivari models, I feel somewhat like a civil-war re-enactor – one of those people who dress up in period costume and replay battles lost or won a long time ago. And so, to celebrate my twentieth year as a maker, I am devoting as much time as I can to exploring different possible futures for violinmaking – or at least my own violinmaking.
If I were to guess in which direction the violin is trying to evolve, I would say, in the same direction as it always has – toward a bigger sound. The lower arching introduced by Stradivari and Guarneri, along with the modern bridge, bow, and setup, have all tended to maximize power and focus. I believe this evolution will continue, if it can, because the forces which originally fueled it – larger halls, the greater volume of its frequent accompanist, the piano, the competitive instincts of violinists – are still very much with us. Added to this, the recording industry has increased the pressure on players, I think, by raising listener expectation. Recordings allow, in exchange for a certain loss of “realism,” a kind of intensity and presence to the sound that is almost completely lost in large halls. Recordings allow one to hear the violin not as it sounds in a hall, but as it sounds under the ear of the violinist. Purists may well say such “close-miking” of the instrument is unrealistic, even vulgar. Personally, I love it. I love hearing violins in small rooms where the intensity of sound can be almost painful. I love violins that project something of this intensity in large halls. And though I am not an accomplished violinist, I love hearing what a great violin does under my ear, being fully aware that some important portion of this experience is inaccessible to the listener. It is part of a closed loop connecting player, bow, and instrument. But it is within this closed loop, I believe, that innovation in violinmaking will either succeed or fail.
Is Innovation Possible?
If one sets about designing a new kind of violin, one senses immediately many possibilities for aesthetic changes – changes which do not effect function in the least. There also seem to be a number of things one could change to make the instrument more comfortable to play, more resistant to damage, and easier to maintain. Because this is an acoustics convention, I will consider only innovation related to the sound of the violin. Here, the issue becomes somewhat confusing. Are we trying to make a better sounding violin, or a violin with a completely new sound? I believe there is little chance of doing the latter, at least without the aid of electronics. Any acoustic instrument relying on bowed strings and played with normal violin technique will sound, I think, like a violin. I say this after listening to two extreme cases. The first was a solid-body violin connected to an amplifier and speaker. One hears the unfiltered string signal, but the sound is still, unquestionably, a violin’s – albeit an unpleasant one. At the other extreme, when a CD of a good violinist is played through Weinreich’s DTC loudspeaker (1) – which superimposes a frequency response as complex as that of the violin being played – it is surprising how little difference it makes: the difference between listening to the violin in one room rather than another, perhaps. From these and other informal experiments, such as listening through an equalizer set every possible way, I conclude that we are more or less stuck with the sound of a violin. The best we can hope to do, through innovation, is maximize the things about violins that good players and their audiences seem to like, and/or get closer to our own individual tonal ideals.
There is a stale debate as to whether it is, in principle, possible to improve on the best old violins. Can one imagine something better than Perlman’s Stradivari or Paganini’s “Cannon?” From the violinist’s point of view, I think the answer is an easy yes. Make them more powerful and faster responding. Make them more even, less prone to wolf-notes, stable in the face of changing humidity, crack-resistant, and for that matter, less expensive. Any one who has spent time in the violin world knows that, whether or not the violin is in some platonic sense perfect, real violins are not.
At the same time, makers have been trying to improve violins for a very long time, and so succeeding at it is to some extent like shaving a few tenths of a second off the hundred-meter dash. One senses some fundamental limit being approached by ever-narrowing increments. On the other hand, if you want to travel a hundred meters as quickly as possible, running may not be the quickest way. Are there alternative approaches to the craft that can perhaps change the limits set by traditional designs and materials?
What is a Good Violin?
Let us define a good violin as one a good violinist loves to play. There are many other possible definitions, but I believe this to be the most useful to makers and researchers. What then are the characteristics that good players respond to in a violin? Tone quality, projection, response, evenness, sensitivity to vibrato, and dynamic range are undoubtedly important. Projection and response I consider “absolute” qualities. By this I mean the more the better. Of course there will always be players who do not want to stand out in an ensemble – or disturb the family cat. Their needs are well met by existing violins, but I think they would be still better met by violins with more projection. These can be strung with lighter strings, reducing the overall output and giving the bonus of increased playability (2). I have never heard a violinist say, “This instrument responds too quickly.”
Projection might be defined, quasi-scientifically, as how loud an instrument sounds to the listener for a given string signal. This would seem to be related to how loud it sounds to the player, but in fact there is an odd independence. There are violins which sound loud under the ear but do not project well, and those which seem quiet under the ear but carry very well. There are also, of course, instruments that are quiet or loud to player and listener alike. In my experience violinists vary a fair amount on how much sound they want under their ear, but not a lot in their requirement that the instrument project.
The above definition of projection relies on reports of a listener’s experience, and so is not readily measurable. However, projection can presumably be related to both the total amount of sound radiated by an instrument, and its spectral composition. The frequency range between about 2000 hz and 4000 hz has often been singled out as especially important, lying as it does in the region of the ear’s greatest sensitivity. In the remainder of this paper, however, I will not worry too much about the particular recipe required. Having claimed that being heard is of primary importance to most violinists, I will simply explore the possibility of making more sound in all registers.br />
Building a More Powerful Violin
Cremer (3) suggests that about 4% of the energy applied to the violin via the bow is radiated as sound. Before setting out to build a more powerful violin, one should ask whether it is possible to get more sound out of the bowed string than a good traditional violin does. This is an important question; if the violin already radiates as much energy as the string has to offer, there is little point in trying to do better.
Now, the regular vibration of the string depends upon its oscillating energy not being lost too quickly. A violin drains energy from the string mainly via motion of the bridge. When the bridge motion – and thus the drain in energy – becomes excessive, a wolf note occurs. Wolf notes are, in a sense, indicators to the maker that he is arriving at the limit for the amount of power that can be extracted. As most traditional violins are driven at least to the point where a wolf note shows up at the strongest resonance, it can be assumed the bowed string is giving about as much as is feasible. Perhaps more could be taken if one designed an instrument where the strong resonance causing the wolf note is replaced by several smaller resonances. The instrument could then be “pushed” further – until a wolf note began to appear at the next highest resonance.
Another way to maximize the energy available for radiation is to minimize the energy lost in other ways, such as internal losses in the violin’s vibrating components. The use of materials with lower damping will accomplish this, and by doing so yield an increased amplitude of vibration for a given force input. But it will also slow the response time. The transients will take longer to die away, and thus each note will take longer to get started. Rather than trade off on response time, a better strategy might be to reduce the mass of the vibrating components. For a given force input, a lighter body will reach a larger amplitude of vibration, while taking no longer to get to this level.
Now, if we decide to build a violin with lighter plates, the greater amplitude achieved implies greater bridge motion, with it the associated tendency toward wolf notes. There is, however, a way out – make a bridge that, for a given movement at the string notch, provides a greater movement at the feet than does a conventional bridge. In other words, change the effective leverage of the bridge. This can be accomplished by making the bridge lower and/or wider. Beyond a certain point, both these options necessitate changes in the traditional design, but this, after all, is what we set out to do.
Before examining further the possibilities of building lighter violins, I would like to touch on the radiation of the instrument with regards to frequency. It is fair to look at high and low frequency radiation separately, as frequencies above about 1000hz are radiated by the top alone, and those below by whole body modes and the Helmholtz resonance.
It is widely known that a vibrating plate radiates sound of a given frequency most effectively when the length of the bending wave in the plate is equal to or larger than the wavelength of that frequency in the surrounding air. If the speed of sound in air is 340 meters per second, then the wavelength of sound for 1000 hz is 34 cm. This is about the vibrating length of a violin top, so if we were able to arrange for the lowest mode of the top be 1000hz, then it, and all higher modes, would radiate very well. As the lowest mode for the supported top of a traditional violin is in the neighborhood of 600 hz, a considerably stiffer than normal top would be required.
An obvious way to achieve this is to leave the wood thicker. Unfortunately, this stiffer but thicker plate will be more massive, thus decreasing the amplitude of vibration. A more fruitful approach is to use less dense wood. Even left the same thickness, its resonances would be higher in frequency and its impedance lower – a double advantage. How far can one take this? The average density for the European spruce used by violinmakers is about .4 grams per c.c. It is possible to find wood much lighter than this – I have samples as light as .33, and even lighter with some North American species. My guess is that the average density of the top wood of good old instruments is in the range of .36 to .38. As this is easy to measure using a CAT scan, and as such scanning of instruments is now being done, I eagerly await a survey on the subject.
Are there undesirable consequences to using less dense material? Yes, several. The plate becomes more susceptible to dents and abrasions. More importantly, the pressure on the soundpost causes it to dig into and otherwise damage the inside surface of the top. I installed a veneer of maple about .25 mm thick in the soundpost area of an experimental top made with very light spruce. It worked so well than I now do this on all my instruments. It reinforces this otherwise fragile area, seems, if anything, beneficial to the sound, and can be easily removed.
A final consequence to using less dense wood is that, if the plates have been tuned to normal tap tones, they will be thinner than normal plates, and thus less able to withstand the considerable static forces on them. The result is an acceleration of the distortion that tends to occur, over the years, in all violins.
Violinmakers often attempt to stiffen wood with special varnishes or grounds. Although it is fairly easy to somewhat stiffen spruce across the grain this way, spruce is such a stiff material for its weight along the grain that there are almost no natural materials you can apply that will increase the stiffness more than simply leaving the plate thicker. An engineer, looking at the problem, might suggest we find a way to get the stiffness to migrate toward the surfaces of the plate, where it can do the most work. This can be done by putting a veneer of wood or other material on a very light core material, thus achieving a much higher stiffness to weight ratio than is possible with a homogenous material. There is the further advantage that the dense surface will resist damage more effectively. If the surface material resists plastic deformation, then long term distortion will become less of an issue.
Charles Besnainou has developed workshop techniques for using layers of carbon fiber, plastic foam, and wood to build string instruments. His system, presented in a workshop at this conference, is built around carbon fiber cloth that has been impregnated with epoxy resin. This is layered with wood veneers and sheets of foam, then vacuum-formed into a cast and cured in a specially built oven. None of the equipment needed is particularly expensive. More importantly from the maker’s point of view, it is a “friendlier” process than one might expect – a different experience than carving wood, to be sure, but an equally satisfying one, I find. It offers the primal satisfaction of combining ingredients and putting them into the oven, followed by the excitement of seeing how they turned out. Besnainou reports (personal communication) that this approach offers possibilities for building instruments significantly lower in weight than is feasible using traditional methods. Furthermore, such instruments have so far proved stable over time and insensitive to changes in humidity.
To get back to the frequency characteristics of the violin, we can in principal improve the radiation at high frequencies by using stiffer and lighter tops. I do not yet know how this might effect an instrument’s low frequency response, but imagine that changes in the total design would be necessitated. Such questions aside, what can be done to help the radiation at low frequencies?
It is widely known that the Helmholz resonance is the lowest radiating resonance on a violin. Weinreich (3) has pointed out that its strength in relation to the other modes can be changed only with respect to its damping. A decrease in damping, allowing larger amplitude, might be effected by rounding the edges of the f-holes. Interestingly, the edges of the f-holes of most old instruments have been rounded with time, while violinmakers often pride themselves on the crispness of their cut. Some simple experimentation would determine how significant such differences are. More radically, one might try changing the f-hole’s shape. The damping of the Helmholtz resonance is largely determined, for a hole of given area, by the total length of the edges of the hole. An f-hole has rather long edges for its area. The least-damped hole would be a circular. Given that the length and placement of the f-holes are important in lending flexibility to the bridge-carrying part of the top, circular f-holes don’t seem feasible. However, a simplified f-hole design might reduce damping somewhat.
The rest of the instrument’s low frequency radiation is achieved by modes involving various contortions to the instrument’s body. Because the body is small in relation to the wavelength of sound at these frequencies, the extent that these modes radiate sound is directly proportional to the net changes in the volume of the body they engender. Such volumetric changes depend on the asymmetry of the modes. Martin Schleske has developed practical ways to optimize the shapes of these modes, and will present his ideas in a workshop at this conference. I think he has made an important contribution to violinmaking methodology.
But standing back a little, I believe that one of the weaknesses (perhaps “characteristics” is a less provocative word) of violin-family instruments is the relative paucity of modes at low frequencies. This paucity should come as no surprise. The violin can be looked at as a “flexing shell,” and the modes of such structures are on average evenly spaced with regard to frequency. Musical pitch, on the other hand, goes up in proportion to an exponential increase in frequency. Thus the low modes of a violin are relatively widely spaced with respect to the violin’s low notes. As not all of these modes radiate, the fundamentals and lower partials of low notes are often poorly supported. The opposite is true at high frequencies, where each note has relatively few partials in the audible range, and a great many closely spaced modes to support them. I think this explains why, at least to my ears, violins differ in their highest registers more in their volume than in their tone color. At any rate, given the native variability of wood, the complexity of the violin’s structure, and the inevitable differences among hand-made instruments, it is difficult to predict and control the radiativity of the lower modes. Because they are scarce, the stakes are high – each mode becomes crucial to the success of the instrument in a way that most of the higher modes are not.
I have for some time believed that violins would, on average, sound better if there were significantly more radiating modes in the low range. This belief was reinforced when I played the string signal from a solid body violin through Weinreich’s aforementioned DTC speaker. The speaker has a great many modes in the violin’s low range, and though it introduces several not-violin-like qualities to the sound, the richness and evenness lent to the low and middle range is remarkable.
I don’t know how one might get a significantly greater number of radiating modes in the low range of a traditional violin. Perhaps additional resonant elements could be incorporated into some novel design. I have several ideas in this direction, and will report on them after further experimentation.
Summary and Final Thoughts
If we hope to create new instruments that in some sense work better than traditional ones, it is important to identify just what, in physical terms, we want them to do. We otherwise risk our work ending up on an already crowded shelf marked “Irrelevant Innovations.” I believe that better projection will always be welcomed by musicians, and think that this, and other benefits, will be gained by reducing the mass of the vibrating components of the instrument, while doing everything possible to ensure efficient radiation. I suspect that an instrument with a greater than normal density of radiating modes under 1000 hz will have desirable musical characteristics.
It must be remembered that the violin is a cultural icon as well as a working tool, and iconoclasts – those who smash of icons – cannot expect a warm welcome at the temple gates. At the same time, a growing number of players, frustrated by the sheer amounts of money they are being asked to pay for often-mediocre instruments, are becoming increasingly open to innovation. The success of such innovation should be judged not by looking back over our shoulders to Cremona, but in terms of how well they both meet the needs of musicians and satisfy our own evolving sense of tonal beauty.
My thanks go to Charles Besnainou, for sharing his work with composites, to Xavier Boutillon, for our many conversations about violin acoustics, and especially to Gabriel Weinreich, for reviewing this paper and for his many years of intellectual generosity.
1. Weinreich, G., “Radiativity revisited: theory and experiment ten years later,” Proceedings of the Stockholm Musical Acoustics Conference, p. 434, 1993.
2. Pickering, N., The Bowed String, Mattituck, N.Y., Amereon, 1991, ch. 1, p. 6.
3. Cremer, L., Physics of the Violin, Cambridge, MA, MIT Press, 1984, ch. 9, p. 203.
4. Weinreich, G., “The Directional Tone Color Loudspeaker,” Journal of the Acoustical Society of America, 101, p. 3071, May 1997