The guitar is a very complex non-linear acoustic device. Anyone who claims to understand it is fooling you and himself as well. Of course principles from the science of acoustics do apply, but ascertaining exactly how they apply is no simple matter. Acousticians who have looked at guitar function have mostly confined themselves to generalized studies that don’t have much relevance when you want to know the effect of changing the position of a brace or two, or sizing and shaping a bar in some particular way.
Most luthiers work by way of trial-and-error and intuition. I do too, but I have always wished for a more deliberate and rigorous approach that would help enable a deeper understanding of how our instrument works. I conceived of a program of research that would rely on a set of four experimental guitars that could be disassembled and reassembled in different combinations and then compared.
In 2009 I had the opportunity to build just such a set of instruments when Finnish luthier Keijo Korelin visited for several months and most generously applied his skills to making this a reality. We made 4 removable necks, 4 side sets, 4 backs, and 6 soundbaords. They can be assembled interchangeably with screws into and through the linings. The soundboards were made in pairs thicknessed to three different dimensions (1.5 mm, 2.25mm, 3 mm). The braces are glued on with dilute hide glue for easy removal. The backs were also thicknessed and braced in pairs (to roughly 2 mm and 3 mm). Likewise, the sides were treated in pairs, with one pair thin (1.6 mm) and one laminated to a thickness of about 2.6 mm, with, additionally, massive ebony bars distributed around the lower bout. Because of the way everything is screwed together, the linings are quite substantial and are uniform throughout all sets. Thus, we are limited to massive linings in all our experiments.
Overall, however, this creates a very flexible system for experimentation. Now it is possible to do paired comparisons for any treatment on which one focuses. Pairs of guitars can be assembled with everything the same except, for instance, the mass and stiffness of the sides. Or they can be assembled identically except for the length or stiffness of the braces in the lower bout of the soundboard. The number of possible tests is only limited by one’s imagination. Also, the interaction between different design elements is possible to investigate by assembling all four sets. Because one is making paired comparisons there is a basic level of verifiability with every test. This is what is missing when we luthiers wander from guitar to guitar through trial and error. Here it is possible to quickly hone in on the particulars of a specific design. One can also ask more basic questions, such as, what is the difference between a guitar with a thin top and heavy braces vs. one with a thick top and light braces?
For a few weeks in the Fall of 2011 lutherie student Veit Lätsch visited my shop. He did some of the preliminary experimental work while we worked some of the bugs out of the system.
In the Fall of 2014 I was very fortunate to have Swedish engineer and budding luthier Johannes Larsson work in my shop for a few months. He ran many of the experiments I had envisioned and did so in a rigorous way. He also developed a way to measure vibration patterns of resonant frequencies on the top, sides, and back. Without his electronics expertise this would not have been possible.
At this time the results of his work are still tentative, but the data is recorded and a number of interesting results can eventually be extracted. Now we are mostly limited to impressions elicited as answers to a series of questions asked as we tested the instruments. The first thing that struck us both is how subtle the differences in sound are between treatments, until we focus very deeply. With more and more familiarity the differences became clear. This should not surprise; the same qualification applies when listening to different guitars in concert or on record. I intend to spell out some of these results soon in a Blog post. In the meantime, the results may not be earth-shaking but they confirm many impressions gained over a lifetime in lutherie and add a new layer of understanding.