Writing for bowed cymbals — a practical guide

Luigi Marino, 2010

A brief historical consideration

Bowing cymbals is a well known technique and is used regularly in both contemporary written music and free improvisation. Perhaps its earliest use in written music is in the 4th piece of Fünf Orchesterstücke (1909) by Arnold Schoenberg, where a tremolo using a cello bow is written in the cymbal part to obtain an effect. By the 1950s, the use of bowed cymbals had become a common practice among Hollywood sound designers, that used it to produce frightening or unusual effects. In 1964 Mario Bertoncini wrote Quodlibet for viola, cello, cb. and percussion, where the sound of the bowed cymbal is not simply an effect, but is put on the same level of importance as any other sound in the composition. Moreover, in this work, high pitches are obtained on the bowed cymbal with a specific technique. When Bertoncini joined Gruppo di Improvvisazione Nuova Consonanza in 1965, he shared his experience about metal percussion with the other members of the group. In these years, at the beginning of what would later become known as the free improvisation scene, we see the bow becoming a standard tool for playing metal percussion. The technique was used increasingly by improvisors; notable is its use by Eddie Prévost of the AMM. Other examples of compositions using bowed cymbals are Tune (1965) by Mario Bertoncini, probably the first composition entirely for cymbals, Asterism (1968) by Toru Takemitsu, A Haunted Landscape (1984) by George Crumb, and many others.

The sounds produced by bowing a cymbal are difficult to control and it is not an accident that their use found favor mostly as orchestral effects, in ambient music, in indeterminate compositional practices, and in free improvisation. In the orchestral "tutti" of the Fünf Orchesterstücke, the cymbal is perceived as an element within a timbral gestalt rather than a single part; in Quodlibet and Tune the accidental partials of the bowed cymbal are points of strength and the same can be said for many improvised settings.

The following is an attempt to define a set of techniques that might be useful to improvisors interested both in the unpredictability of these sounds but also in taking decisions about which sound to play, to composers in the need of a more deterministic approach, or simply to anyone fascinated by the experience of finding sounds, recurrences and relationships on instruments that come as a blank sheet, whose relationship between gesture and sound can be discovered only through practice and attentive listening.

Mapping sounds on a cymbal

The idea of mapping sounds on cymbals came during several improvisation sessions where I noticed some sonic recurrences bowing on them. In those days I also had the possibility to write for a small chamber group that included three percussionists, so, in order to have some written material, differentiation of the parts, and a vocabulary to communicate the ideas and the techniques within the group, I started a more accurate analysis of the instruments. The cymbals I used the most measure 68cm (pic 1 and 2) in diameter and were handmade in Wuhan, China. The fact of being handmade is very important, because the irregularity of the cymbal contributes a lot to the variety of sounds one can find on it1. If they are sufficiently irregular, even if the diameter is the same, every mapping of a new cymbal will result in a completely different set of sounds and relationships. Therefore, all the techniques explained below will not work exactly the same way on other cymbals and the difference could increase as the shape varies.

upside down wing nut

Pic 3 - Upside down flat wing nut

erhu bow

Pic 4 - An erhu bow with a piece of wood to keep the distance between the hair and the bow

Big cymbals usually have an incredibly rich response to the stroke of a bow. When they present irregularities, such as in the case of the large Wuhan cymbals worked with a hammer, with the proper set up, it is also possible to define many sounds in a fairly consistent way. As a set up, standard straight cymbal stands can be used, but they have to be heavy enough not to fall when consistent pressure is applied on the rim of the cymbal laterally with the bow; moreover the cymbals have to be held tight on the stand and they need one or more felt cushions, either below or above the bell, in order to still let the cymbal vibrate. A useful trick is to use a flat wing nut screwed upside down (pic 3). The tightness usually reduces the complexity of the sound, but it offers more control over the behavior: if the cymbal is left completely loose removing the wing nut, the resonances are richer, but the mapping is far less consistent and less recurrent sounds can be found, whereas if the wing nut is too tight or there isn't enough felt the vibrations are dampened too much, resulting in less interesting timbres. With the right tightness the behavior is still very complex and it is possible to map out many recurrences. Concerning the bow, it takes more energy to bow a piece of metal than a string, more rosin, and the hair breaks easier: a cheap solution is an erhu bow with an object inserted near the handle to keep the distance between the hair and the bow (pic 4). To keep the piece of wood in place I wrapped it with rubber bands. However there are many different possible solutions, from cello bows to custom made ones2.

The sound changes according to the speed of the stroke, the pressure applied, and the position of the bow on the circumference. The position where one bows on the circumference is the most important element. All the other parameters can be found only when one has a consistent reference to a particular spot, and as little as one centimeter can completely change the sound. In this sense, the mapping of these spots is a physical measurement of the space or, better still, a measurement of the space led by the sound because, having almost no visual marks, the only way to discover the spots is an attentive listening and a careful detection of the acoustic recurrences. The behavior not only varies according to the parameters specified, but it can also have a development in time in response to the same sustained action. This development is often hard to control, but a few spots present a fairly consistent recurrence, at least in the general shape of the spectrum.

The amount of predictability of the mapped sound varies according to the spot. Changing spots before the cymbal stops vibrating almost always changes the response of the spots because the energy is added to the previous resonance, and the behavior becomes increasingly unpredictable. I think that, under ideal conditions, it would be possible to map all the acoustic behaviors of the cymbal with certainty, or at least a very close approximation. But these conditions are usually far from the ones in which experimental musicians work; moreover taking into account that some spots behave unpredictably can often be a point of strength for the performance. For the analysis I used mostly my ears, running a computer analysis of the spectrum when I needed more precision. For some spots the response is totally predictable, so they do not require any particular attention with the stroke used. Other spots have a predictable response only according to a very specific stroke (e.g. high pressure, fast stroke, etc.). Most of the problems that arise in performance are related to having the desired sound at the attack, and then being able to sustain it or control its development: therefore the attack phase is of fundamental importance. It is common that, once the resonance is established with the right stroke, the sound can be sustained regardless of the stroke type: in other words, when the right sound comes out, it is usually easy to sustain it.

To have a good sustain with only one bow it is necessary to slightly relieve the pressure at the end of each stroke so that the cymbal continues to ring, and start the new stroke without damping the previous resonance. The duration of the pressure relief should be minimized until it is impossible to hear the break between strokes. Hiding the break is easy with sounds having a long resonance, but almost impossible with more dry ones, such as some pitches in the high register: in this case the break will be noticeable, not unlike the bow change on a string instrument. A good sustain is especially useful in creating continuous dynamics such as a continuous crescendo, although, for the crescendo, the spot where one bows is crucial because the loudness range of the sounds varies from one spot to another in a very sensible way.

A classification of the sounds

In order to write for cymbals, I found it useful to have a classification of the sounds to be able to communicate with other performers, so I grouped the sounds in 4 categories: pitches, multiphonics, high register and processes. However, classifying the sounds of a bowed cymbal is not a simple task, since there is a continuity between these groups. For instance, the distinction between a pitch and a multiphonic can be loose because very few pitches can be perceived as pure, and in a multiphonic a partial can be prominent and be perceived as fundamental pitch. Moreover, some pitches obtained by regularly bowing one spot, especially with a fast stroke, can be higher than the pitches obtained on another spot with the technique used to obtain the high register. The categories below were decided upon in this way: the distinction between pitch and multiphonic is based on the acoustic quality of the sound, though the decision about when sounds other than the fundamental are consistent enough to constitute a multiphonic is arbitrary; the high register is defined exclusively according to the technique used; the definition of process is based solely on the acoustic result, as it was for the pitch and the multiphonic, and once a recurrence is established, the performer may need to adjust the technique case by case to obtain the desired development.

• Pitches
Establishing the pitches is a good starting point to map the spots on a cymbal because they are the easiest to define. Bowing in different spots of an irregular cymbal will result in the discovery of different pitches (example 1). Every cymbal has a range where the most of the stable pitches are, but exceptions are frequent, so it is common to come across isolated pitches that differ from the average of the others. There is a relationship between this range and the diameter. Often it is possible to map pitches in this range with very small intervals. The quality of the pitches can vary significantly. In my experience, a few spots resulted in the clear sound of a doubled octave which was possible to control (example 2).

• Multiphonics
The multiphonic is hard to define because of the complexity of the spectrum. Multiphonics vary according to the spot, they are very sensitive to the stroke, and have the most irregular behavior in time. As with wind instruments, it is possible to recognize a difference between standard multiphonics (example 3) and beating multiphonics (example 4), but the instability of the cymbal multiphonic makes this distinction arbitrary. Generally it is possible to define several spots on the same cymbal, associable to a different multiphonic, with their own character and movement.

• High register
To produce a random pitch in the high register, it is enough to apply pressure on the cymbal with a finger at about 2.5cm or less from the rim, in the same position as the bow or close. The closer the pressure point is to the rim, the higher the pitch. However, this is only a general principle, and there are plenty of exceptions. There are spots where the high pitch is louder, others where it never gets too high, and others where it simply cannot be played. Generally, the multiphonic spots favor high pitches, and sometimes it is possible to map them exactly, but even a minute change of position, or a change in the object used to apply pressure, can render this distinction useless. Moving the object on a line that goes from the bow to the center of the cymbal, as one is bowing, can generate very interesting shifts in the high register (example 5); on the cymbals in the pictures the most sensitive area is approximately between 0.5cm and 2.5cm from the rim. To obtain the exact pitch or the desired shift, I found it easier and sometimes necessary using a hard and small object like the back of a small soft mallet, which can be used for soft roll and soft hits too. Although, the use of the fingers allows more sensitivity for nuances.

• Processes
Changing the stroke type, such as applying more pressure or increasing the bow speed, may result in a clear change in the timbre quality. When this change involves a multiphonic, the result is particularly interesting, so I defined "process" as a change involving a multiphonic that can be reproduced at least with a similar development to the one desired. It is possible to recognize subtle processes inside a mutiphonic, but they are extremely hard to control and their definition is very arbitrary, while the transition from a pitch to a multiphonic is more apparent, presents a surprisingly rich acoustic development and, in some cases, it is not too hard to control (example 6). Generally all processes are related to volume or/and bow pressure, but I was only able to find a relationship between gesture and sound in the ones from pitch to multiphonic. For these processes, in some spots, the evolution responds primarily to volume, while in other areas bow pressure plays the primary role. In the first case, there is an intensity point where the pitch starts to break and other partials appear; in the second case the partials arise as the bow pressure increases, without any noticeable change in volume. Usually volume is a result of bow speed, but sometimes, according to the spot, it is a result of pressure, so the distinction between volume and pressure can be tricky: sometimes a performer needs more pressure to get to the volume point where the multiphonic happens, other times he or she uses the pressure to generate the multiphonic without involving the volume. Certain processes are reversible, but most are not. Usually it is possible to recognize a certain trend of the sound to collapse toward the low frequencies, and at this point it is hard to change the state of the resonance without stopping the cymbal completely. Sometimes it is possible to reverse this process by damping the cymbal like for the high register, but a little further from the rim and with less force. Very few spots present a trend to go from multiphonic to pitch (example 7). Processes involving more than one spot on the same cymbal are possible, but very hard to define, and they require a perfect knowledge of the instrument (example 8).

Some details

cymbal3 detail

Pic 5 - A detail of the mapping

Finding the spots where the exact sound can be consistently reproduced takes some time. When I started the mapping, I initially wrote preliminary marks with a pencil, then put paper labels on the spots I thought to be more consistent, and started to play regularly. When the practice confirmed the labeling, I removed the paper and marked the safe spots with a marker, whereas I left paper labels in the spots where there were some interesting sounds, but they couldn't be defined consistently.
On the left (pic 5) there is a portion of the rim with 5 spots marked with a marker. M3 and M4 are two different multiphonics. I numbered incrementally the multiphonics that I can recognize as different as I find them proceeding around the rim. B3 half flat and C half sharp are normal pitches. The arrow from M3 to B indicates that while I'm playing M3 I can move to B without stopping the cymbal. Sometimes very interesting things can happen without stopping the cymbal before bowing another spot, so I draw a line everytime there is a particular behavior, even if the second bowed spot doesn't sound like its label. The spot with three marker labels has different behaviors according to: a) The speed of the bow (fb = fast bow), so when the stroke is fast I get a A5 diesis; b) The pressure of the bow (hp = high pressure), so when I bow applying more pressure I get multiphonic 5; c) It presents characteristics of a process, and in this case, like in most of the cases I labeled as processes, the pitch turns into a multiohonic applying more pressure.
Once (luckily only once) I dropped one cymbal that fell from my arms to the ground exactly on the rim, bent like an elastic band and jumped back. Even if the shape in the end seemed unchanged, some of the spots I marked were irremediably lost, and with them some of the details of the parts I had written: if you decide to go so specific with a unique object, it is good to accept some ephemerality, b plans, or being really really careful...

Audio examples

(1) See how Gino Robair cuts standard cymbals to make them irregular.
(2) See the custom bows made by Tatsuya Nakatani.