The Soundboard of a Harpsichord

If you look at solely the physical dimensions of harpsichord soundboards and the variety of properties of the wood of which they are made, you will see so many variables that it is easy to give up trying to understand them. But there is another way of looking at soundboards - from the view of a harpsichord string.

A string is a chunk of isotropic gunk (isotropic: a material with the same properties in all three dimensions). Nothing could be further from living things like wood than metal is! Any view of the universe a metal string has is going to be childishly simple by comparison with ours. And so, the smart question is to ask, how does a string view a soundboard? Well, it sees a mass (weight, sort of) and compliance (give). That is what affects the frequency at which the string will vibrate and how much of its energy is coupled to (in common with) the soundboard. And, it sees an acoustic damping, which determines how fast its energy is dissipated into the surrounding world after it is plucked.

All instrument makers know this at heart. Get any of us in a workshop, and the first thing we do with any soundboard - harpsichord, guitar, violin, whatever - is to tap it to see if it sounds 'alive', to ensure that the sounds of a string will be sent to the surrounding air rather than all soaked up in dead wood. Then we heft it to judge its weight and bend it to judge its flexibility.

Maybe, some day, I'll figure out how to communicate the aliveness we look for in words, but the truth is that I spent an afternoon with Sydney Evans, a luthier supplier in England when I was there. At the end of the afternoon, after picking up a few hundred pieces of various qualities of violin wood, I knew the feel and have never forgotten.

However, mass and compliance are things on which one can put numbers. And, one can relate those numbers to several important things about soundboards. In particular, they are related to the direct-sound sustain of the instrument, to the coupling between plucked and unplucked strings, and to the frequency response (warmth) of the sound. The first study I am aware of in this vein was done by Fletcher (Acustica 37:139 1977). One of Fletcher's conclusions was that the traditional barring design of northern harpsichords should result in a very even frequency response from the fundamental resonance up to about 4 kHz. This was confirmed by Kottick (Scientific American 264(2):110-115 1991) who observed an average resonance spacing of 17 Hz in a new instrument, closer than the width of the individual resonances. However, when Kottick measured old instruments (GSJ 38:55-77 1985) he found much more variable response - most had at least one very strong peak in the octave below c', and response dropped off rapidly above 1 kHz. I suspect that when these old instruments were new, their makers were aiming for a more even and brilliant response than we hear from them today.

There are two basic designs in old harpsichord soundboards. The model used by most Italian instruments was to put bars across the underside of the soundboard (dashed in the diagram at right), so that the string is connected to as large a sounding area as possible (green at right - the light green shows area that is only coupled to the soundboard by the internal air resonance). The ends of the soundbars are tapered, and the sides of Italian instruments are of the same construction as the soundboard, flexible between the 'barring' provided by the support knees. The soundbars usually seem to be placed to meet the bentside between knees, but to come close to them on the spine. Older instruments tend to have fewer and lighter soundbars than later ones. The entire soundboard of such an instrument moves up and down, and the bentside between the knees in and out, in response to a string vibration. One essential function of Italian soundbars is to establish a slight curvature (crown) to the soundboard. If the soundboard were flat, it would pull in the sides both when moving above center and below center - the sides would vibrate at twice the frequency of the soundboard. With a crown, the sides always move in when the soundboard moves up, and move out when the soundboard moves down, with no frequency doubling.

The lower the frequency, the larger the area is required to couple a soundboard to air. Being almost entirely soundboard, the sound of Italian instruments has lots of fundamental, especially the low frequency drum-like pulse of the initial pluck. In isolation, the sound of an Italian string would be quickly coupled to the surrounding air and thus die away quickly, but another factor comes into play. The sections of string between bridge and hitchpin were not wrapped in felt as with modern pianos, so they are set into resonance by the pluck of the sounding string. Due to the light soundboard and flexible sides, these string ends can radiate sound very effectively. In fact, in many early Italians the nut was placed on a soundboard of its own as in this diagram - the string ends between nut and tuning pin could radiate effectively too. So, a good Italian instrument has a powerful fat direct sound (ictus) that is rapidly transformed into a cathedral-like 'echo' from the sympathetic strings.

The model used by northern instruments was fundamentally different. The bottom and sides are rigid and heavy. Frank Hubbard once made a series of Flemish instruments with cases that included ones made of Douglas fir plywood, and insisted that the only difference in the sound was that the plywood was better in the sense that, being more stable than solid wood, it permitted the soundboard to be made more responsive. The northern soundboard is unbarred under the bridge so it sounds in the manner of a harp, as a thin membrane. No crown is required acoustically as with the Italian design. (They are, however, put in under compression to avoid splitting with weather changes under the constraint of the rigid sides.) The primary width of this harp-like soundboard (green at right) is set by a 'cutoff' soundbar that is roughly parallel to the bridge. In instruments with a 4' stop, this first bar (dotted at right, between the yellow and green areas) is the hitch rail for the 4' strings, then the 4' soundboard (yellow) has its own cutoff bar. The soundboard in the triangular area defined by this bar (in magenta) has a few more soundbars to divide it up a bit irregularly so that it has lots of the plate resonances shown in Kottick. The nuts are on a rigid wrestplank, so the string ends at the tuning pins are acoustically inactive. And, the string ends between 8' bridge and hitchpin can only radiate from one end, not from both as with the Italian design. In effect, their place is taken by the triangular section of the northern soundboard. The northern soundboard can be made more acoustically rigid than the Italian model, so the sound of a northern string can have less ictus and sustain longer directly, that is by itself rather than via the sympathetic vibration of the string ends. However, the cathedral-like echo of the Italian was still valued in the north - several northern makers used side-string (mouse-ear) dampers that did not touch the string when the choir was not being plucked, which gave a similar effect as the Italian end strings when at least one choir was off.

For music of rapidly changing tonality, especially to remote keys, the northern sound adapts faster, it's 'cleaner', more precise tonally than the Italian. But the richness of a cathedral sound is magnificent, when used for music whose tonal character matches it.

John Sankey
other notes on instrument making