RAISING THE BAR

Joseph Curtin reports on a series of experiments at the 2023 Oberlin Acoustics Workshop, which attempted to ascertain the acoustic effects of altering the height and scoop of the violin’s bass-bar

The bass-bar is a gracefully scooped beam of spruce running along the inside of the violin top, extending almost end to end and passing directly beneath the bass foot of the bridge. One of its functions is to prevent the top from collapsing under the sustained downward force of the strings. Another is to provide the structural asymmetry that is crucial to sound production. A stiff bar would best serve the supportive function – and the stiffer the better. What best serves the acoustics of the instrument is more complicated and involves the mass of the bar – typically between 3.5 and 4.5 grams – as well as the distribution of stiffness along its length. The trade-offs between what might be called the bass-bar’s ‘static’ and ‘dynamic’ functions have tacitly governed the bar’s evolution from its diminutive Baroque form to the longer, stiffer bars we use now.

On a sunny afternoon at the beginning of June 2023, violin makers and researchers from as far away as Japan and Australia and as near as Ann Arbor, Michigan, arrived in Oberlin, Ohio, a mid-Western college town that for a few weeks every summer becomes the centre of the violin making world. Last summer marked the first post-pandemic meeting of the Acoustics Workshop. A major part of the week was devoted to two experiments aimed at better understanding the violin bass-bar. The first investigated the acoustical effects of incrementally reducing its overall height, and the second the effects of incrementally increasing its scoop. Each experiment had its own set of instruments and its own team of makers. Dorian Barnes, Alex Currin, Johann Lotter, Wataru Shinozaki and Sibylle Ruppert participated in the height experiment, while Jason Starkie, Gaian Amorim, Todd Mathus and Steve McCann worked on scoop.

Luthiers making preparations for the bass-bar experiments at the 2023 Oberlin Acoustics Workshop

PHOTOS OBERLIN ACOUSTICS WORKSHOP

Each maker brought a violin with a specially prepared bar. In fitting it, they were asked to:

• record the density of the wood used

• maintain a thickness of 5.5mm along the entire length

• position the bar as they normally would

• fit the bar without ‘spring’ (i.e. without forcing a little extra arch into the top, as is often done)

• match the heights to those indicated in Step 1 of the experiment

• keep the side walls of the bar parallel, then round the top to a semicircle of 5.5mm diameter (this being easier to specify and control than the more typical elliptical cross-sections)

• glue the top as lightly as possible (for ease of removal)

• otherwise set up the instrument as they normally would

Brooklyn-based maker M.J. Kwan drew up designs for the modifications each bass-bar would undergo. The five-step height experiment (see Figure 1, page 48) began with an unusually high bar: 14mm at the geometrical centre and 4.5mm at the ends. It finished with an unusually low one: 7.3mm going down to 2.3mm. (None of the measurements given here include the thickness of the top.) The scoop experiment took the centre- and end-heights from Step 2 of the height experiment and kept these constant while the lengths in between were progressively scooped out. For the fifth and final step, the ends of the bar were taken down to zero while leaving the centre height the same. Figure 2 shows the five steps used in the experiment.

A violin top fitted with one of the bass-bars under examination

FIGURE 1 At each of the steps in the experiment, the heights at five positions along the bar were reduced by 15 per cent. The rest of the bar was then trimmed to a smooth longitudinal profile.

FIGURE 2 In the scoop experiment, makers controlled the heights at nine measurement points as they progressively increased the scoop. The last step tests the effects of removing wood at the ends of the bar.

The workshop was six days long, meaning that every day each maker had to remove the top, trim the bar, re-glue the top, and string up the instrument in time for acoustic measurements the following morning. There were, of course, setbacks – an unglued seam here, a measurement malfunction there – and yet heroic efforts by the assembled makers and measurers produced a full data set for eight of the nine instruments.

Two players, Alex Malaimare and Sean Hardesty, informally evaluated the instruments at every step. There was no time for rigorous subjective testing, however, so we relied on acoustical measurements to track changes in the instruments. Figure 3 shows Lonny Marino using an impulse hammer and laser vibrometer to measure bridge admittance, which shows how much energy from the bowed string gets into the instrument. In Figure 4 M.J. Kwan measures radiativity, or how much sound is radiated across the frequency range for a given force at the bridge. In this article we will consider only the sound radiation data.

A standard method for measuring sound radiation is to tap the bridge with a tiny hammer equipped with a force sensor, and then record the response with a microphone. To get a representative sampling of the violin’s complex sound field, we did this at twelve positions around the instrument, tapping the bass corner of the bridge horizontally and then repeating the cycle tapping the top centre of the bridge vertically. An average of these 24 measurements provides an estimate of the total sound output for a given force. In this article, general trends are highlighted by averaging the measurements of all four violins at each step. The jagged lines in the graphs have been smoothed somewhat for clarity. The horizontal lines represent the average sound levels in each band. Where sound engineers typically work with 1/3 octave bands, the four much-broader bands used here correspond with natural subdivisions in the spectrum of any normally built violin.

Makers investigated the effects of altering bass-bar height and scoop

ALL PHOTOS OBERLIN ACOUSTICS WORKSHOP. FIGURES COURTESY JOSEPH CURTIN

FIGURE 3 Luthier Lonny Marino measures bridge admittance using an impulse hammer and laser vibrometer

FIGURE 4 M.J. Kwan (second left) The measures radiativity, or how much sound is radiated across the frequency range for a given force at the bridge

FIGURE 5 Unlike a recording engineer adjusting the spectral balance with a few sliders, luthiers have a vast number of variables they can adjust, with results that can be unpredictable

To date, there are no well-tested correlations between a violin’s perceived playing qualities and any acoustical measurement, making it reasonable to ask: why bother with more measurements? One answer is that the inner workings of the violin are interesting to makers for the same reason that those of a combustion engine are interesting to automobile manufacturers, regardless of the preferences of any particular drivers or players. The more deeply that makers understand the instrument at a physical level, the better they can cater to the wildly varying tastes of individual players.

Recording engineers can tweak the spectral balance of a sound using the sliders on an equaliser, each of which independently adjusts the sound level in one particular frequency band (figure 5). Makers do something similar when making an instrument, though without the benefit of sliders. Or one could say there are a great many sliders, but they are linked to each other in ways no one really understands, making it difficult to predict the result of shifting any given one. Although makers can easily control the height of the arching or the length of the f-holes or the position of the soundpost, they can at best make intelligent guesses as to the tonal outcome.

One promise of scientific research is to tease out the interconnections, allowing the guesses to become more intelligent. For example, research has shown the strong influence of the mass and tuning of the violin bridge on the instrument’s high-frequency output, allowing a good deal more control than was previously available. There has been no such research on the bass-bar. Common wisdom is that raising the height will increase brilliance and projection, while lowering it will darken the sound. How does this hold up, experimentally?

The graphs in this article show sound output across a frequency range of 200–7,000Hz. The horizontal lines indicate band averages – i.e. the average level for all the peaks and valleys in a particular frequency band. The bands used here are delineated by the following frequencies: 200Hz– 780Hz–1,740Hz–2,930Hz–7,000Hz. All graphs except that in figure 9 represent the average output for all violins in the experiment. The levels of each can be thought of as the relative positions of a slider on an equaliser. Now let’s see how these positions change with each step of the height experiment.

ALTHOUGH MAKERS CAN EASILY CONTROL THE HEIGHT OF THE ARCHING OR LENGTH OF THE F-HOLES, THEY CAN AT BEST MAKE INTELLIGENT GUESSES AS TO THE TONAL OUTCOME

FIGURE 6 The black line shows the average sound levels in Step 1 of the height experiment. The red line shows Step 2. Levels have gone up in all four bands.

FIGURE 7 The red line is Step 2 in the height experiment. The grey lines are the other steps. Step 2 has higher levels in the top three bands, and second highest in the other.

FIGURE 8 The black and red lines show average sound levels for Steps 1 and 5 of the height experiment, respectively. In Step 5, the spectral balance is shifted towards the bass.

FIGURE 9 The red line depicts a single violin with the bar removed altogether. The grey lines are the five steps for the height experiment for that same violin. Removing the bar dramatically increased the bass and decreased the treble. It also significantly lowered the frequencies of the so-called B1 modes, which are hardly affected by the other steps.

FIGURE 10 The red line is the average of all violins for Step 4 of the scoop experiment. It shows slightly more sound output in all bands than the four other steps (gray lines)

FIGURE 11 Taking the ends of the bar down to zero (black line) while maintaining full height at the centre reduced output in Bands 1, 2 and 4. The band level was unchanged in Band 3, and yet the height of two peaks there actually went up.

Because there were countless uncontrolled variables and opportunities for experimental error, results should be considered suggestive rather than definitive. Still, it can be seen in figure 6 that the average sound output of all four violins in the height experiment went up in all four frequency bands. This was true of each violin individually: sometimes subtly, sometimes dramatically.

Figure 7 overlays Step 2 (in red) over all other steps (in grey). Step 2 produces the highest output in the three top bands, and very nearly the highest in the lowest band. If one wants to maximise overall sound output, Step 2 seems a clear winner.

Figure 8 compares all violins at Step 1 and Step 5. The spectral balance has clearly shifted towards the bass, confirming the common wisdom that lowering the bar tends to reduce brightness. One maker, Johann Lotter, went a step further and removed the bar altogether. Figure 9 shows that this moved the balance far more dramatically. It also significantly lowered the frequencies of the so-called B1 modes, which were hardly affected by the other steps. The term ‘bass-bar’ may be misleading in that its very existence inhibits bass output while increasing treble.

The results from the scoop experiment are less obvious, as there was a great deal of variation among instruments from step to step. And yet Figure 10 shows that on average (and only on average) Step 4 provided the highest sound output in all bands.

Progressively altering the height of the bass-bar

FIGURE 12 Step 2 from the height experiment and Step 4 from the scoop. Each produced, on average, the highest levels in all frequency bands, albeit in different sets of instruments

FIGURES COURTESY JOSEPH CURTIN. PHOTO COURTESY OBERLIN ACOUSTICS WORKSHOP w m s s g t

Step 5 of the scoop experiment changed the question to what happens if you take the end of the bar down to zero while maintaining the height at the centre. Figure 11 shows that sound levels go down in Bands 1, 2 and 4, while Band 3 holds steady. Moreover, the height of two prominent peaks in Band 3 goes up – a tantalising finding given the tonal importance of this band, which covers the so-called Bridge Hill region.

What should makers take home from all this? Figure 12 shows Steps 2 and 4 for the height and scoop experiments respectively. When averaged over the four instruments in each experiment, these steps produced the highest sound levels in all frequency bands. While Step 2 of the height experiment will look like a typical bass-bar to many makers, Step 4 may seem more scooped than usual. Though the data for this experiment is relatively weak, statistically it does suggest a direction makers may want to explore. Should we conclude that relatively high bars with a fair bit of scoop are optimal? Well, no – at least not without a well-grounded definition of optimal, and this surely varies from one instrument (and one player) to the next. A previous article on the bridge’s evolution (‘Views on the bridge’, The Strad, November 2019) suggested that the modern bridge was adopted because it increased sound output, especially at frequencies associated with brilliance and projection. This experiment implies that the modern bar, which is higher and longer than its Baroque counterpart, was adopted for the same reason. Players want to be heard.

More research would be helpful. One of the most promising ways forward is to use computer models of instruments.

Although modelling the violin’s geometry and material properties is a complex and challenging project, once done, numerous experiments can be undertaken without the punishing time investment and uncontrollable variables associated with modifying real instruments. Qualified volunteers please apply!