The last secret of Stradivari III

Voicing the violin

After the foregoing it will surprise hardly anyone that I can only come up with the hypothesis that the old masters voiced their violins in the white, by scraping the outside and testing the results by playing. This is a marvellous, and most probably the only way to attain the chaotic results so characteristic of their work. This also has shown to be a reliable way to obtain a consistently high level of sound quality and playing characteristics, and isn’t this considered to be one of the main distinguishing attributes of the Cremonese School?

For several other reasons too I deem it quite probable that this was standard practice in Cremona, and to all probability only in Cremona, until the breaking of the thread. Imagine the early development of the method. You come to know somehow that properly thicknessing the plates is an efficient way to control the sound, but knowledge about where to scrape has to be gathered by trial and error. With only the ear to rely on, gathering some understanding of violin tone and how to influence it, is a purely qualitative task. Inevitably patterns will emerge, but only after oft repeated practising and experimenting.


For the future master of those days, to get to learn the trick and to be able to apply it independently will easily have taken years. And the almost exclusively aural transmission of the trade secrets will surely not have speeded up the process. Besides that it is clearly not for everybody to acquire the knack: a good ear and aural memory, good eyes and hands, string playing ability, understanding of violin tone, fearlessness and the drive to experiment, are some, and I guess not the only, prerequisites to be united in one person. So, if circumstances did not allow for a robust period of training, we may safely assume that something got lost. Imagine also how simply the voicing trick could be kept within the family business, without overtly offending the rules for training apprentices and journeymen set by the guild.


But most of all, imagine the essential difference in sound quality between, and the consistently high level of, violins by Amati c.s. and all other violins. If you, the king, the emperor, the archduke, the pope, were looking for something really outstanding and reliable, Cremona was the place to be. The kind of family monopoly you would be well advised to guard jealously. And now for some practical considerations. There are only a very few principles that I try to adhere to in the course of making an instrument. One of them is choosing wood with a, to my eyes, viable density, with the main aim to avoid extremes1). Though I am going to scrape it, I finish the outside arching meticulously. The initial thicknessing of the plates is done according to simple recipes, that, of course, are being subject to constant refinement2).

Though years of diligently watching trembling plates covered with bouncing tea leaves have not yielded me the key to consistent results, I still use a basic form of this technique when fitting bass-bars3). When the top is finished, without f-holes yet, I tap the plate and listen carefully for the sound quality of the second and fifth modes. I am not interested in their absolute frequency, and only slightly in the interval between them. When it comes to finishing the bass-bar, I work until the original sound quality is largely there again. I do think that in this way you more or less adapt the dimensions of the bass-bar to the quality of the wood and the chosen height and form of the arching. It seems to work and the old ones could have done it this way easily too.

Voicing challenges

The violin is in the white and has been played for the first time. You cannot wait to amend the most obvious imperfections. Where to start scraping? I have to admit that starting somewhere at random, and listening for the results, is a particularly unattractive prospect. You simply do not want to spoil the violin you built with so much loving care4). Not to speak of cello’s or double basses and the sweat involved therein. To be true we should be able to suggest the violin plate that it has been thinned in a certain place, without actually having to thin it, so that we can play, feel and hear the effects of the suggestion5). Why, we are. That is to say, we have at least the rough outline of the voicing routine just sketched out. The theory behind it may be somewhat lacking in scientific refinement, but on workshop level by and large it works. Starting point is the observation that the vibration modes of a wooden membrane, or part of it, will be lowered in frequency by thinning it. Mass reduction is linear, but stiffness reduction is cubic6).

Two side of a coin

It was not until 2002 that the outlines of voicing began to dawn upon me, when Heinrich Dönnwald let out a small bird7): “If we told you, from tomorrow everybody would make good violins”. And a few lines later: “If we say that we glue a coin to the top, and then do something, [.] on p. 257. Why, gluing a coin to a wooden membrane (“the top”) lowers the frequency of that membrane or part of it. There is only mass attached to the membrane, but no stiffness added whatsoever. So by attaching a small weight to a violin plate, you suggest the violin that it has been thinned around that spot. What Greiner and Dönnwald do afterwards, will be a secret forever. We know that they have an extremely well-equipped sound laboratory at their disposal. The simple way is, of course, to glue, play and listen with your own ears. And that is exactly what I began to do. The gluing was done with double sided tape, but that proved to be rather cumbersome, with the soft wood of the top constantly getting torn out in some measure.


But from the onset it was clear that if the violin was happy with the coin in a certain position, you could repeat and consolidate the result by scraping the outside of the plate at or around that spot. I had to wait for a phone call from my former pupil, the Israel based maker and player of viols Amit Tiefenbrunn, to have pointed out to me the possibility of using a pair of magnets, one inside, one outside8). Yes! Now we had at our disposal an elegant and fast method of searching the plates for where to scrape. We now could put some additional warmth and strength in the G-string, highlight the e-string, appease the odd unruly note on the a-string, get the f sharp on the e-string in line etc. And so I realized around 2004 that I could in a certain way guarantee prospective violin makers, amateur or professional, that they would be able to make violins well above the average level.

That bolstered my self-confidence enough to start teaching violin making. And, upon my word, teaching, and making, this way is great fun. Most pupils tend to be bewildered at first by the idea that almost all measurements can be taken with a proportionate amount of salt. Once accustomed to the idea, they work happily to their abilities, without having to worry with every small deviation or adaptation about the possibly ruinous effect on the sound of their instrument9). This way of working allows you to free yourself of the accretions of 19th century superstition about bass-bar, soundpost, tailpiece, fingerboard, ground, varnish, string height, strings etc., as well as of more recent ones about wood treatment, plate tuning, bench copies and the computer.

What if

The endless reservoir of “what if” questions, typically at the disposal of the amateur maker, can be disposed of elegantly by remembering that in almost all cases the influence of voicing is a magnitude bigger than the supposed effect of the original issue. This is not to say that we can do away with the subtleties of the craft. The difference between a good violin and an exceptional one is subtle, but not to be overlooked. But there is no need any more to accidentally end up below a certain level, and that level is a comfortable way above average. And is not the high general level of production one of the hallmarks of classical Italian violin making? As a bonus I promised to say something about testing this hypothesis. I now have to resort to definitely 21st century methods of violin research10).

Fingerprints of the master

A resolution of 50 microns opens the possibility to scan inner and outer surfaces of plates to the magnitude of a single stroke of the scraper. From these scans we can not only produce an idealized curve, or surface, but also the measure of deviation from that curve, the “roughness” of the surface. If an instrument has been finished by scraping the outside, this side of the plate should consistently be found to be rougher than the inside. If outside roughness appears to be a characteristic of classical Cremonese instruments only, they may well have been finished that way. If Stainer’s instruments also have the characteristic outside roughness, that may be another indication of his still enigmatic connection with the Amati’s. We might even end up with checking the “Messiah” this way. As an afterthought, having to play the violin in the white sheds at least some light on the alleged presence of fingerboards before varnishing. Having to remove a fingerboard with veneered sides, and to refit it exactly on the finished violin, wouldn’t be my first choice either.

1) Borman & Stoel, Material Facts,The Strad, January 2013, pp. 47-52
2) The dimensions given by Sacconi are in my opinion still a sound starting point.
3) Vibration modes of free violin plates, advocated widely by Carleen Maley Hutchins in i.a. the Journal of the Catgut Acoustical Society.
4) Robert F. McGowan, My Method of Tuning Plates, Journal of the Australian Association of Musical Instrument Makers, Vol. IV, No 1, February 1985. Reprint from the Violin Makers Journal of the Violin Makers Association of British Columbia, 1963. To me this publication was absolutely seminal. It was quite sobering to re-read, for most of the reasoning to me still makes sense.
5) From Sverre Kolberg (Norwegian physicist and recorder maker) I learned, somewhere in the 80-ies, how to suggest a recorder that its bore was locally narrowed, and how to shape the bore accordingly for good tuning and sound, reaming, playing and listening. I since kept pondering about how to voice violins that way.
6) Without much understanding I read this in a lucid article by a guitar maker, almost certainly Graham Caldersmith, the name and source of which I am, alas, unable to dig up anymore, way back in the 80-ies too. Don’t know why it stayed with me, but it obviously did.
7) Masters of the Bel Canto, The Strad, March 2002, pp. 252-257.
8) A good source of small and strong magnets is for instance
9) This is basically what is taught at the Dutch School of Violin Making.
10) Franco Zanini, Learning the Finer Points, The Strad, jan. 2012, pp. 36-41. More recently: Hopfner & Weber, A better Resolution?, The Strad. May 2014, p. 8.