6 8

Late Realizations

Article by You Nakai
Share this article
go to article
If a dog swims across a river carrying a piece of meat or anything of that sort in its mouth and sees its reflection, it opens its mouth and in hastening to seize the other piece of meat, it loses the one it was carrying.
Aesop . “The Dog and Its Reflection”


Among the many David Tudor stories that Gordon Mumma told me over the years, there are two that have continued to replay over and over again in the back of my mind. One is a story about a certain ability that Tudor had developed during his early organist days which influenced his later electronic music. The version Mumma recounted on November 11, 2011, went like this:

  • Tudor was an excellent accompanist as a pianist. He had acquired part of that skill by playing the organ at churches where he was accompanying hymn singing. He played with others even as a young organist. And there is one feature of this that is significant, this experience that all organists really have: you’re in a big church and your organ is up somewhere there and the pipes aren’t necessarily in the same place. So Tudor developed early the skill of being able to adjust to the time lag, the latency, if you will, between what he does and what he hears. This is profoundly significant. And in understanding the evolution and development of his electronic music way later . . . profoundly significant! I pound the table! Because he could deal with those time lags—he could adjust to it. (Mumma 2011)

The other is a story about a certain inability that Tudor had developed also during his organist days which likewise influenced his later electronic music. The version Mumma recounted on November 4, 2016, went like this:

  • He would adjust the levels to how he’d heard them [sounds] from where he was performing. Not just that piece, but lots of other things. And there were certain places where even when we were using our own loudspeakers—we had four of them, I think—he made his choices without hearing what was coming out for the audience. And I would go out, I always went out to be sure that I understood what the decibel levels were because when it gets too loud, the audience’s ears get crushed in a way that they can’t really hear what Tudor was hearing. He and I, I don’t want to say disagreed, but we were different worlds about that aspect . . . But it intrigued me because it was a situation in which he couldn’t separate himself from being an organist in large cathedrals, for example, where what you hear in any different place is never the same, but those were different kinds of acoustical spaces which were made to merge all of that stuff, rather than proscenium theatre which pushes to an audience. And he never quite, never really separated his sense of location, I suppose. (Mumma 2016)

These two stories, told to me on different occasions, nonetheless seemed to speak to one another as if they were two sides of the same coin. For one thing, they both concerned the issue of adjustment. In the first story, Tudor appeared as someone who was exceptionally good at adjusting to the latency between his physical input to the system and the output sound from his standpoint at the center of a giant instrument. In the second story, he appeared as someone who was exceptionally bad at adjusting his ears to other listeners who might hear the sounds of his giant instrument differently from a different standpoint. At the heart of the matter was the coordination between what Tudor did and what he heard, a process of adjustment realized through feedback—his physical engagement with his instrument and what his instrument did in response which returned to him in the form of sound to influence his next move. For the sake of this “single performer feedback,” so to speak, other possible acts of listening were either taken for granted or just ignored. Unless, that is, they performed—not only listened but did something in response. Because if they produced an output from the input, these other listeners could also be inserted into the loop and become part of the giant composite instrument that Tudor played so well. This explains why Mumma spoke of Tudor’s ability as an accompanist in the same breath as his ability to deal with latency. Tudor’s virtuosity at adjusting to the time lag in performer feedback caused by the scale of the instrument incorporating a multiplicity of components in the return path may have placed listeners who just listen outside the loop. But, as Tudor himself had written to his partner M.C. Richards in August 1960, “of course one is not just a listener” (Tudor 1960).


“Single performer feedback” was in fact the word that Tudor chose in 1973 to describe his method in one piece of music he had performed seven years before, a piece he had incidentally acknowledged as his own “composition” for the first time (Tudor 1973). The only two performances of Bandoneon ! took place in October 1966 at the massive 69th Regiment Armory in New York City, as part of 9 Evenings: Theatre & Engineering, a festival dedicated to the collaboration between artists and engineers. In both performances, the sound of the titular accordion-like instrument produced by squeezing its bellows in and out—which Tudor had grown fond of for several years by then—was picked up by multiple microphones and processed through a network of electronic instruments before being output as sound from twelve surrounding loudspeakers, or as dancing Lissajous figures on oscilloscopes blown up in scale via three projectors. The same bandoneon sound was also used as a control signal in the processing of sound, as well as in the switching of eight spotlights shining the performer platform and eighteen Lekolites placed in the balcony surrounding the venue.


In other words, the bandoneon sound was not only modified by instruments but also modified how instruments modified. Ron Kuivila gave a nice name to this new ability of sounds, once converted to electronic signals, to control the means of sound production: the instrumentalization of sound (Kuivila 2004, 20). But the instrumentalized sounds Tudor produced in Bandoneon ! not only went into other instruments; they also returned to Tudor’s own ears after going through his giant composite instrument and the enormous space of the Armory. And when it did, it influenced what Tudor did next. So “single performer feedback” was indeed a good description of how the total configuration functioned—“with an extended amount of latency,” one might add for accuracy, since even with all the electronics aside, the long reverberation time in the 31,000-square-foot venue had been measured up to six seconds.1

According to the belated description written seven years after the event, the method of single performer feedback rendered any external compositional means unnecessary, as sounds could now be generated and modified solely through the recycling of signals within a giant feedback loop. So although Tudor had signed his name as the composer of Bandoneon !, it was a composition one step removed—what he composed was a complex network of instruments that once activated composed the music themselves:

  • The situation obtaining when a performer scans two media simultaneously to which I had been introduced through Lowell Cross’s ‘Musica Instrumentalis’, contributed the performance method: a single performer feedback, which also obviated the need for any compositional means. (Tudor 1973)


Musica Instrumentalis, which Tudor claimed in his reminiscence as having inspired the method of single performer feedback, was a piece for bandoneon composed for Tudor and premiered on May 13, 1966. Since the previous year, Lowell Cross had been experimenting in visualizing sounds by connecting two outputs of his sound system to the two inputs (corresponding to the vertical and horizontal axis) of a television set he had modified into a giant oscilloscope. If the two signals were the same except in differences in phase or frequency, it resulted in a pretty visual pattern known as Lissajous figures which appeared to rotate or wiggle as the phase or frequency of sound changed. When Tudor visited him in Toronto in early 1966, Cross showed him what he had been up to. As Anthony Gnazzo, who was assisting Cross at the time recalled, Tudor was fascinated: “David just thought this was the greatest thing” (Gnazzo 1999). He even bought into the pseudo-musicological theory that Cross, then a musicology graduate student, had developed, which held that the visualization of sound enabled by his instrument could be used to tell good music from bad music. “They would just sit there for hours playing sounds and just roaring in front of this thing, deciding who was good and who wasn’t good” (ibid.). But sooner or later, they realized that the bandoneon’s ability to produce two signals at once, in addition to the phase differences created by the movement of the bellows in relation to the microphones made the instrument an ideal input source for the modified TV set: the two signals could be mapped onto the two perpendicular axes of verticality and horizontality defining the monitor screen, and their phase relationships directly affected the shape of Lissajous figures. So Cross proceeded to compose a piece that coordinated his instrument and Tudor’s.

In Musica Instrumentalis, just as in Bandoneon ! performed five months later, Tudor’s bandoneon sound was picked up by multiple microphones and output both as sound from loudspeakers as well as moving Lissajous figures on the modified TV set. Unlike Tudor’s own composition, however, Cross created a visual score with images that the bandoneonist had to reproduce during the performance (Cross 1966?). So what Tudor did was triggered by what he saw, not what he heard. Sound emerged as a by-product of the effort to produce a specific image. In other words, the simultaneous scanning of two media was not on an equal footing.


When he began working on his own composition a few months later, Tudor honored his source of inspiration by asking Cross to be part of Bandoneon !, putting him in charge of visualizing sound via his modified TV set. In September 1966, in the final days leading up to the 9 Evenings, Tudor invited Cross to stay with him to work on the project. As Cross reminisced much later in 2001, one of the things his host was inquisitive about that summer was latency in feedback:

  • “Tell me about phase shift.” He knew that phase (time) relationships between two channels were important considerations in stereophonic sound and the consequent relationships between stereo and x-y displays. He wanted to pursue the use of phase shift in electronic feedback circuits. (Cross 2001, 13)

Cross’s response was that Tudor had already done this in person, albeit unknowingly. He just needed to regard himself as part of the total configuration:

  • I explained to him that he had already experienced himself as a feedback “component” in Musica Instrumentalis, during which his physical movement of the bandoneon between the microphones produced audible stereophonic phase-shift effects over the loudspeakers and visible kinetic phase-shift effects on the TV screens. We talked about feedback in general, which can bring audio, video, or servo systems to conditions of sustained oscillation. (ibid.)

Since sustained oscillation is something that recycles itself over and over again, it obviates the need for any external control. Tudor was to actually stage this principle several weeks after his conversation with Cross when he performed his first composition which was not really a composition but a single performer feedback with a lot of latency involved. According to the reflection of Fred Waldhauer, the Bell Labs engineer who assisted Tudor during the 9 Evenings, the second performance of Bandoneon ! on October 18, 1966, proceeded in the following way before it escaped the capture of language:

  • David began the piece on a sort of drone sound which turned on one of the lights, at first dimly, then brighter as the sound came up. Sound came through one of the twelve balcony loudspeakers. David added other tones and the armory responded with its sound. The piece thus launched, having no choice, moved on. It would be feckless to suppose that it could be described in words. (Waldhauer 1966)

But there were other listeners, mostly constrained in their own seats at the Armory, who felt they were left out of the giant loop. The poet M.C. Richards, who had lived with Tudor for most of the 1950s, wrote in her diary a belated realization she had attained after seeing the same second performance that Waldhauer reflected upon so dramatically:

  • October 18 I attended a program at the festival of the arts and engineering: I went to hear David Tudor’s piece. I fled after the volume of amplified sound became deafening. I suddenly realized the “hard fact” of David’s disregard for simple human qualities, and his lack of respect for his own “ear” as well as mine and others. (Richards 1966)


Although Tudor only recalled Musica Instrumentalis as the source of inspiration in his 1973 reminiscence, there was another piece of music he had performed in the days leading up to the 9 Evenings which appears to have inspired his approach in Bandoneon !. This second music happened to place more emphasis on the multiplicity of listeners involved in the loop than in the fact that it all returned to a single performer-listener at the center. No wonder perhaps, since it was a piece conceived by Gordon Mumma, who would later describe the difference between his own approach and Tudor’s indifference to sounds heard by others as pertaining to “different worlds”: “. . . which was something I was always very conscious of. Part of it is from some of the works that I have done which relied upon me moving around the space. But he didn’t do a piece where he moved around the space, he was always sitting on the thing, right?” (Mumma 2016).2 If mobility on stage was an issue, the difference in question can be mapped onto the difference between instruments: Mumma could walk around because he was a horn player. So similarly to the adjustment to the organ influencing Tudor’s electronic music, the adjustment to the horn appears to have biased Mumma’s particular approach to electronics, focusing on the multiplicity of listening agents dispersed across the complex network of instruments that operated as one big feedback circuit.

Mumma gave a nice name to what he was doing: Cybersonics. It was a modification of a well-known term that the mathematician Norbert Wiener had coined some ten years before in the late 1940s to call a new interdisciplinary study of semi-automatic behavior of systems based on feedback. Generalizing the lessons derived from his war-time effort to develop anti-aircraft guns that automatically adjusted their aim according to whether it had hit the mark or not, Wiener observed a mechanism of auto-correction in a variety of phenomena whereby changes produced in the environment by a system are read or fed back to the same system to adjust its behavior so that it may better reach a given goal. From this perspective, a person reaching out to drink a glass of water, a thermostat maintaining the room temperature steady, and a gun trying to shoot a plane down could all be seen as engaging in a similar task. To address this circular form of control through feedback, Wiener used the Greek word κυβερνάω (kybernao), meaning “to steer, navigate or govern” and created the name Cybernetics (Wiener 1948).

Mumma’s “little invention of taking the ‘netics’ out and putting the ‘sonic’ in” (Mumma 2011) grafted this general theory of self-correcting systems to the performance of live electronic music. He knew analog circuits well enough to compose instruments that adjusted their behavior in reaction to a sound input, which could cybernetically be their own output. But Mumma’s aim was neither to shoot down fighter planes nor to make instruments play some given music correctly; it was instead to use the principle of feedback to make interesting live music with electronics. For this reason, cybersonics departed from cybernetics in several significant ways.

First of all, the delineation of “system” which determines what counts as input and output and therefore as feedback,3 was considerably broader in Mumma’s case, as he explained in an article written in 1967:

  • I am concerned primarily with “system concepts”— configurations that include sound sources, electronic modification circuitry, control or logic circuitry, playback apparatus (power amplifiers, loudspeakers, and the auditorium), and even social conditions beyond the confines of technology. I suggest that the most important creative aspect of live-performance electronic music technology is not this or that circuit innovation but rather the total configuration itself. (Mumma [1967] 2015, 44)

Since the “total configuration” was a complex network of many instruments, each of which formed a system of its own, feedback became something other than the solitary endeavor of a single system listening to itself talk in order to correctly say whatever it was meant to say. In Mumma’s grafting, feedback was rather a form of interaction across multiple systems, each with a different thing or two to say or do depending on what it heard. As a result, the goal-oriented nature of systems became itself a means to an end. What mattered more than what the total configuration accomplished was what it produced in the attempt. As with the visual score of Musica Instrumentalis, the given goal was a mere pretext for generating good music as a by-product.

So instead of tightly coordinating all the components of the configuration to achieve a global goal, cybersonics focused on the indeterminacy between one component and another, the mismatch between different local goals whose outcome could not be completely foreseen. When he talked about an actual cybersonic instrument, therefore, Mumma often talked about how it was designed not to be obedient to human command, but to deliberate on its own so that the response was always something of a surprise—not a reaction but a “decision.” This led him to think of sound input as a matter of perception, something each instrument “listened to” or “heard”:

  • The circuit had ears, microphones, and it listened, and then it responded. And, of course, when it responded, it added itself to the ensemble, and it was still making an analysis of what was going on, so it changed what it was doing according to what it was doing as well. (Mumma 1982, 101)

That is to say, Mumma admitted a lot of listeners in his total configuration. But the reason he thought these other listeners were listening was because they did something in response. The latent act of listening Mumma could not see was indirectly sensed through the changes in behavior he could see. In other words, the cybersonic system was conceived as a network of multiple listeners each of whom engaged in their own performer feedback—reading the input, making decisions based on that reading, and doing something in response that would go through the configuration before returning to itself.


But Mumma, who had developed a habit of walking around and checking up on how listeners who did not respond were listening, thought of ways to keep such listeners in the loop of the total configuration.4 One way was to turn the table around: if they did not move around the space like he did, then the space could virtually move around to highlight the locality of each listening process. This was indeed what he did in the piece he wrote for Tudor’s bandoneon and premiered on August 6, 1966, just two months before the 9 Evenings. In Mesa, the sustained tones of the bandoneon were modulated through four processing channels using a variety of cybersonic modules, most of which had parameters controlled by the instrumentalized sound of the bandoneon. All in all, the sound of the bandoneon processed the sound of the bandoneon, enhancing the complexity of its overtones, and feeding back the multi-layered result to modify various facets of the same process. The signals that came out of the processing channels were then sent to multiple loudspeakers. Aside from this output returning to Tudor’s ears to influence what he did next, the spatial distribution of inharmonically related and phase-shifted sounds across different loudspeakers created the illusion of sound sources moving when they were “mixed” in the two ears of each audience member (Mumma 1967, 14). Even if the listener did not move, it would sound as if she had, which could then trigger her to move her head around in response and change the sound further. As Mumma explained, the slight latency between signals which made Lissajous figures dance on Cross’s modified TV set made the sound space move in each listener’s mind: “Not only is the ‘place’ of the sound articulated by this means, but the apparent size of the sound space is continually changed” (ibid.). 5

The influence of Mesa may be observed in one particular component of Bandoneon ! that resists its description as a single performer feedback. For although Tudor himself remained stationary throughout the performance other than moving the bellows of the bandoneon, one group of instruments on the receiving end roamed freely around the giant floor of the Armory: objects turned into loudspeakers (by attaching transducers on them) placed on five remote-controlled carts operated by David Behrman, Tony Gnazzo, along with Bell Labs engineers Larry Heilos and Per Biorn. Audience members were seated in fixed chairs so if a certain moving speaker moved closer, the particular sound coming out of it would become louder. Which is to say that what one person heard could be quite different from what another person heard elsewhere, most obviously Tudor engaging in his single performer feedback. 

In the end, however, the moving loudspeakers of Bandoneon ! were deployed not to emphasize such differences, but to contribute to the global goal that Tudor had set for his composite instrument: to create a giant white noise generator out of the total configuration.6 The multiplicity of speakers was primarily tasked to increase the density of overall noise, and any difference between various listening points would have sooner or later been drowned in a sea of noise—something quite eloquently documented in the recording of the performance, which is filled for the most part with a majestic storm of feedback noise that makes it easy for the listener to reflect on what M.C. Richards heard that evening, but not so much on what Waldhauer did.7


Tudor continued his effort to compose giant instruments after the 9 Evenings, and the scale of his single performer feedback grew bigger and bigger. In 1970, he designed the entire sound system of the Pepsi Pavilion—a 45-meters-wide and 23-meters-high dome constructed for the World Expo in Osaka, Japan—as a musical instrument (Lindgren 1972, 15).


Although the money came from the famous beverage corporation that named the building after itself, the actual project was led by Experiments in Arts and Technology (E.A.T.), a non-profit group supporting collaboration between artists and engineers that was formed in conjunction with the 9 Evenings. But the conditions this time differed greatly from the Armory. For one thing, the Pepsi Pavilion expected an average of 350,000 visitors per day who would be free to roam around the space. For another, the common philosophy of the E.A.T. artists happened to celebrate the uniqueness of what these numerous visitors each experienced—an attitude they termed “anti-Disney,” in opposition to amusement park rides designed to make everyone experience the same prearranged sequence of events (Martin 2020).8

In tune with this focus, Tudor worked to emphasize the directionality of sounds coming out of the thirty-seven speakers arranged in a rhombic grid pattern across the dome. The eight-channel sound modification console he asked Mumma to build included a high-pass filter (in addition to amplitude and frequency modulation) for this purpose since higher frequencies are more directional. The sounds could also be moved across the speakers in different speeds and patterns, a virtual simulation of mobile loudspeakers Tudor had deployed four years before at the Armory. In spite of all these efforts to highlight the locality of each listener, however, the Pepsi Pavilion still had a “console” in the middle of it all, a central performance interface of the giant composite instrument which not only shared its name and function with the interface of church organs but also its scale. Since each channel had twelve knobs, the total number of controls in all the eight channels added up to ninety-six. Mumma saw a special significance in this number. Ninety-six, he later explained, was “the same order of magnitude of combinations as one has with a large organ” (quoted in Lindgren 1972, 19). Knowing very well about Tudor’s early connection to that instrument, he added, “a person with practice can, in an hour or so, get a good idea of possible configurations and then be able to treat it as a performing instrument” (ibid.).

Pp console

The same familiarity with the organ appears to have also led Tudor to remain in front of the console even in the absence of bandoneon. Out of the six programs that he made, the first five required him to sit at the console and operate the knobs in the style of single performer feedback. In fact, “program 2” Pepscillator was a straightforward electronic feedback piece that routed the output of one modulator channel to the input of another creating what Tudor conceived as a complex oscillator. 

Only the last “program 6,” titled Microphone, put the newly gained ability to walk around the space to good use. Taking a stroll away from the console, Tudor operated two shotgun microphones, one directed at the loudspeakers and the other at some random space, to create discrete bursts of acoustic feedback that were regulated by the slow rotating pattern of activated speakers. Fourteen years later, he would reflect on the experience in which the extreme latency between what he did and what he heard made him almost forget he was performing and become a listener who just listened:

  • I used the slowest clock patterns amongst the speakers. And it worked like a charm. I mean, you could never tell when there was going to be a sound. And of course they were modified through the modification capability, so it was very beautiful . . . Eventually you felt like you were on a barren seashore, listening to birds coming and going occasionally and there’d be a silence for a long period, then all of a sudden . . . (Tudor 1984)


Tudor probably performed Microphone towards the very end of his stay at the Pepsi Pavilion, which was suddenly cut short when Pepsi-Cola came to its corporate senses and realized that it was spending too much money on a futile experiment that promised no return as far as drinking carbonated soda was concerned. Six weeks after the pavilion opened its doors, E.A.T. personnel were forced out of its premises.

But Tudor had plans for his next piece which would extend the scale of his giant instrument even further. Forty years later, Fujiko Nakaya, who collaborated with Tudor on the project, recalled how it was shortly after their expulsion from the Pepsi Pavilion that Billy Klüver, the director of E.A.T., began searching for an island upon Tudor’s request. They had a specific purpose in mind, she reflected on January 7, 2014: “They were looking for an island that could be used as a ‘musical instrument’” (Nakaya 2014). Tudor had very specific ideas about its shape and size: “He was looking for a fairly large and elongated island that would take about twenty minutes to get from one side to another and about an hour to go around” (ibid.). She then went on to describe why her friend wanted an instrument that big: “The particular scale of the place was very important. It was to be an island of a size that matched the maximum scale for feedback to occur” (ibid., emphasis added).

That the physical limit of the instrument should correspond to the limit of feedback is no surprise given how the process of single performer feedback actively delineated the spatial extent of the instrument along with the temporal duration of each performance in Tudor’s music. But just what amounts to feedback in the project which became known as Island Eye Island Ear is quite unclear. The strangeness of this problem is especially highlighted in contrast to the deceptive simplicity of the basic idea:

  • (a) Sound is recorded at different locations on the island over the course of a year;
  • (b) The recordings are played continuously from loudspeaker drivers at the focus of parabolic reflectors and smaller woks, installed across the same island;
  • (c) The reflectors and woks are positioned either in pairs facing each other to create “sound beams,” or on their own as “throws” to direct sounds to “natural reflectors” like rocks, trees, hills and ravines, thus covering the island with scattering beams, reflections, and refractions of sounds displaced in time and space;
  • (d) Other than Tudor’s sound beams, artificial fog by Nakaya and kites by Jackie Matisse Monnier are also installed;
  • (e) Audiences are free to move around the island in any way each person wanted;
  • (f) The overall aim of the project is to reveal the nature of the island through technology—“Decided that you are revealing the island through sound. FN [Fujiko Nakaya] said you are revealing the wind by the fog” (Experiments in Art and Technology, 1974).

The system and its global goal seem quite clear. The latent nature of the island that one could not see, the prime example being wind, is indirectly sensed through the changes in the behavior of sound beams, fog, and kites that one could see. And in the case of Tudor’s sound beams that reflect and are reflected by the wind and stones and other natural resources of the island, these were themselves indigenous sounds of the island, stored, delayed, modified, transported, and output at other locations and at other times. Technology thus allowed the island to become both the modifier and the modified.

Nevertheless, there is something very puzzling about Tudor’s overall conception, for nothing appears to make the process repeat itself: there is no return path anywhere. In a project that claimed to pursue the “maximum scale for feedback to occur,” no feedback as such appears in his plan, neither acoustic feedback (à la Microphone) nor electronic feedback (à la Pepscillator). In other words, if Tudor was indeed exploring the limit of feedback in Island Eye Island Ear, the path that returned the output to the input must have been neither acoustic nor electronic.

There is in fact only one component in the system even remotely capable of establishing feedback in spite of the separation in time and space between input and output: the individual visitors who walked inside the giant instrument, listening to sounds and committing some of them to memory so that they might be identified with some other sound heard later in another part of the island. A human could connect space by moving from one place to another, and connect time through reflection. As Cross had told him in the fall of 1966, Tudor had already experienced himself as a feedback “component” in Musica Instrumentalis. The estimate for “the maximum scale for feedback to occur,” which Tudor specified as the particular size and form of the island—“twenty minutes to get from one side to another, and about an hour to go around”—may not have been derived from any technical considerations of non-human equipment, but rather from the psychological and physiological conditions of human memory.

The individual experience of each audience indeed appears to have been the central focus of the whole project reflected in the very use of extremely localized sound beams. When Klüver interviewed him about the island project on November 24, 1978, Tudor described what kind of experience he wanted to create: “First of all, intimacy. So that when you come upon a feature that I’ve arranged within a given space, you feel that you are alone with it. And it won’t matter whether you’re surrounded by people or not” (Tudor 1978, 3-4). The insular nature of Tudor’s giant instrument in this way would have reflected the insular nature of each visitor’s experience. To put it differently, the physical scale of the instrument in the island project had precluded Tudor’s tendency for single performer feedback centering around a single listener. What was brought to the center in its stead was the multiplicity of listeners who each listened differently, leading to a multiplication of performer feedback.


Although Tudor long regarded it as his most important project, Island Eye Island Ear was never realized during his lifetime.9 In the meantime, he was often performing the large-scale version of Rainforest—an outgrowth of those mobile loudspeakers of Bandoneon ! in which objects were converted into loudspeakers by attaching transducers on them—with younger musicians who had participated in a workshop he taught at the New Music in New Hampshire summer festival in June 1973.


Soon, Tudor began describing the forest of instrumental loudspeakers as if it were a miniature analog of the instrumentalized island—a virtual Island Eye Island Ear, so to speak. Now that the audience walked around a room full of giant loudspeakers, what became central in Tudor’s mind was the relationship between the sound of one instrument heard directly and its transformed reflection, the same sounds picked up by contact microphones, equalized and output from commercial loudspeakers set around the space. On September 3, 1984, Tudor gave a detailed description of Rainforest explaining all this:

  • [T]he purpose of the contact mike is to take the resonant frequencies which you hear at best very close to the sounding object; to take those into an ordinary loudspeaker which you can consider not as auxiliary but as enhancement. What that does when you establish the proper tonal balance is that you’ve got a reflection of the sound which you can distance in space. [...] [The piece] becomes larger and there’s a coherence there that’s deceptive which I like very much, because as you move through the space there are sounds that you hear, that you’ve heard before or have heard somewhere else. (Tudor 1984)

Sonic reflection thus triggered mental reflection—it composed a “deceptive coherence” in the mind of each moving listener. And according to a narration inserted in the video documentation of Rainforest from November 24, 1975, the coherence that only exists privately in the mind of each wandering listener was ultimately the “composition” itself. In a matter of less than a decade, composition for Tudor had come a long way from the single performer feedback of Bandoneon !.

  • What is left then is for the listener to roam the space integrating the sound of each object into the total sound environment. Listening to each object emphasizes its sound over all the others. This process shapes the sounds of Rainforest. This is not passive listening. It is the active role of composing. David Tudor brings to Rainforest a selection of sounds and defers the final act of composing to the audience. Because each person experiences Rainforestdifferently, there are as many compositions for each performance as there are members in the audience. (quoted in Bolling 1975?10)

Or, as many performer feedback as there are visitors. The particular trajectory taken by each performer-listener mattered, as well as the memory accumulated along the way, since each path formed a different adjustment to the instrumentalized island or forest. This view explains Tudor’s peculiar ideas for how to document Island Eye Island Ear: he thought of recording the entire soundscape from up in the air, or to make a recording for four hours per day for seven days and then compress the entire recording into four hours (Tudor 1978, 5-6). It was as if the island had its own sensory organs to perceive its own sounds through the reflection of each human visitor, as if the sum of all the multiple performer feedback amounted to the island listening to itself talk across very long latency—but then again, this was what Tudor’s title had been suggesting all along: Island Eye Island Ear.


The opening up of single performer feedback to multiple performer feedback was also reflected in the deeply collaborative nature of the projects Tudor engaged around this period. After “giving away” (Tudor 1995, quoted in Rogalsky 2005, 190) Rainforest to the students of his summer festival workshop in 1973, the same participants formed a group to perform the piece which subsequently became known as the "Composers Inside Electronics." Following Tudor’s death in August 1996, the surviving members of CIE have continued to perform Rainforest to this day, which has now evolved into its fifth version, an installation that can be set running in museums without performers present.

The island project was also conceived as a group work from the beginning, with Nakaya’s fog, Monnier’s kites, as well as Swedish choreographer Margaretha Åsberg's performers forming the four basic components along with Tudor’s sound beams. Like all the latent visitors who could have walked around the island, these collaborators would have certainly experienced the instrument with different eyes and ears and reflect on different matters; each person acting as an island engaged in a performer feedback that Tudor would not have been able to foresee.

One of the collaborators who participated in the 1974 exploration of Knavelskär, an island near Stockholm that perfectly matched Tudor’s conditions, was Julie Martin, who assisted Klüver at the time and later became his wife. In early September 2019, Julie emailed me out of the blue asking if I wanted to fly over to Stockholm to revisit Knavelskär with her and several younger artist friends (Anna Lundt, Jacob Kirkegaard, and Tobias Kirstein). The trip was happening within a few days so I didn’t have much time to think. I decided to do rock-paper-scissors with my ten-year-old son Aevi and go if I win and stay if I lose. I won and hopped on the plane.

IMG 0193
IMG 0150

As we approached the island—which happened to be quite beautifully covered in fog—Julie took out the map they had made during the first expedition in which Tudor had inscribed the possible placement of his sound beams and Fujiko had marked the possible areas to generate her fog. Looking at it, she tried to adjust her memory from almost half a century ago with the scenery spreading before her eyes in the present. This act of reflection across a great time lag continued throughout the several hours that we stayed on Knavelskär, as we walked by the trees and rocks and ravines, around the beaches and inlets, with Julie constantly replaying what had stayed in her mind after all these years and projecting it over the present island. It was a feedback of the mind that may have been on Tudor’s mind when he decided to explore the “maximum scale for feedback to occur,” though the scale on which it was being realized most likely exceeded his rather modest expectations.

One unexpected by-product of this trip was Nakaya’s excitement upon receiving my report of what we had done, which led her to invite Julie to Japan in January 2020 to discuss the possibility of realizing Island Eye Island Ear near the province of Kaga in southwestern Ishikawa prefecture.11 During her stay, we took a weekend trip to find an island in that area. A daylong excursion to a beautiful lake with a little nice island offered much to reflect upon, yet ended up being futile as far as composing a giant instrument was concerned. Somewhere along the way, Julie, who worked for so many years with Tudor and Klüver to find the right island, had a moment of reflection, attaining a belated realization about the nature of this project. Back at the hotel, she turned to me and said, “Maybe this is what the piece has become! Maybe Island Eye Island Ear is now this whole process of trying to find an island!” (personal communication). I laughed and agreed wholeheartedly.12

6 1


It seems best to spin the narrative out of David Tudor’s loop and end with the story of yet another listener who listened to what Tudor did quite belatedly and responded in a very different manner. It was only recently that I realized how in my own work with No Collective I have been using a lot of long-time delay systems. I had never cared to develop a coherent theory about why this is the case until I started writing this paper. But maybe this is a good occasion to bring things to a pseudo-full circle.

2012 11 17 21 34 41
U8 C5559

In 2013, I was forced to make an installation piece instead of a performance because I was single-parenting Aevi who was two years old back then. It was a commission from the art festival Mikke Konohana, taking place in the neighborhood in Osaka where I was living temporarily at the time. What I ended up making was a musical haunted mansion based on the simple idea that the most frightening thing in the world is one’s own reflection that one doesn’t recognize as such. Out of the several locations presented to me by the festival organizers, I chose an abandoned tiny two-stories warehouse, whose second floor was completely dark and filled with many shelves because they used to grow silkworms there. I set a microphone in the middle of the otherwise empty first floor and hung four portable amplifiers around. The microphone was connected to the loudspeakers but with a long-time delay of five minutes and gated by a movement sensor located on the second floor. On that pitch dark second floor, I had placed many mirrors around to create the sense of another being in the space. I had also placed several hundred sound-triggered chirping bird toys on the silkworm shelves which, like the mirrors, were activated by a person’s presence in that space. The movement sensor opened the gate for five minutes only so that if nobody re-triggered it for more than that amount of time, the gate would be closed again and the input from five minutes before be never heard.

So audiences would come in, talk into the microphone, not hearing anything back, go up the stairs to see what’s going on, face their own reflection without knowing so, and if they managed to trigger the sensor while they move about, hear their own voices from five minutes past coming out of the speakers below. The acts of each particular visitor thus completed the feedback circuit—if no one came, nothing would have happened. But they did, and more towards the end of the second day as the festival neared its close, when people after people kept pouring in non-stop so that the gate was left constantly open and feedback constantly active. The end result of this was a gradual increase in the density of overall feedback noise which finally triggered one of the neighbors listening outside the loop to call the police who came and shut down the installation just five minutes before closing time. I named the piece House Music: And the Rest of You.

Three years later, I curated a series of semi-private performances with Vadim Pevzner in his almost-one-room apartment in Chelsea, New York. There was quite a lot one could do in that situation that was not possible in a fully public presentation. The audiences were assembled by invite only, and we called the series Public Service (2016). In No Collective’s own contribution, I did nothing other than set up a long-time delay system of thirty minutes in the apartment without telling the visitors. When people arrived we offered them food and drink and other substances and proceeded to engage in conversation with them. As the six or so visitors ate, drank, and smoked, waiting for something to happen, they naturally became more and more intoxicated with whatever they happened to be consuming, until suddenly they began to hear their own voices from thirty minutes before. They listened to it, trying to figure out what was happening, many talking over it as they tended to think that it was a playback of a one-time recording and not an ongoing delay. Whatever they uttered over the first loop would be included in the next one, which is also to say that the moments people did not talk over the playback but silently lent their ears to the sounds of thirty minutes past were registered in the loop as sonic windows to an earlier loop which remained intelligible even after many cycles. The act of listening thus made an audible difference in what was being listened to. We had also sent out invitations with different descriptions of the event to each person, which triggered each attendant to develop a different theory about the meaning of this experience based on the particular description they had received. People generally got more intoxicated with whatever they were drinking and/or smoking as the evening passed, and stayed for nine to ten hours on average just talking among, and over, themselves about what it was that they were experiencing.

All in all, if listening to yourself talk takes too long, you are bound to forget what you said by the time you hear it. The output slips out of your present. And you might even perceive the returning voice as coming from someone else. Beyond certain thresholds of latency, single performer feedback thus becomes multiple performer feedback unbeknownst to the performer.



Bolling, George. 1975? Rainforest, 1981. David Tudor Papers, Box 37A, Item V10, Getty Research Institute.

Cross, Lowell. 1966? Performance Instructions for Musica Instrumentalis. David Tudor Papers, Box 36 Folder 2, Getty Research Institute.

Cross, Lowell. 2001. “Remembering David Tudor: A 75th Anniversary Memoir [Online-Einrichtung: Clemens Gresser, Wolfgang Krebs].” European Journal of Musicology 4.

Experiments in Art and Technology. 1974. “Report on Experiments for ‘Knavelskär’ Project (8 August).” David Tudor Papers, Box 21 Folder 1, Getty Research Institute.

Fillion, Michelle. 2005. Liner Notes. In Mumma, Gordon. Electronic Music of Theater and Public Activity. New World Records 80632, compact disc.

von Foerster, Heinz, ed. 1974. Cybernetics of Cybernetics: or the Control of Control and the Communication of Communication. Urbana, IL: University of Illinois, Biological Computer Laboratory.

Gnazzo, Anthony. 1999. Interview by Julie Martin [video recording]. Oakland, CA, 17 March. Klüver/ Martin Archive.

Goldman, Jonathan. 2012. “The Buttons on Pandora's Box: David Tudor and the Bandoneon.” American Music, Vol. 30, No. 1: 30-60.

Kuivila, Ron. 2004. “Open Sources: Words, Circuits and the Notation-Realization Relation in the Music of David Tudor.” Leonardo Music Journal 14: 17–23.

Lindgren, Nilo. 1972. “Into the Collaboration.” In Klüver, Billy, Martin, Julie, Rose, Barbara, eds. The Pavilion: Experiments in Art and Technology. New York, NY: E. P. Dutton.

Martin, Julie. 2020. “Art, Technology and Collaboration.” Presentation at the conference “Weathering Ear, Breathing Eye," Kaga, Japan, 25 January.

Mumma, Gordon. 1967. “Four Sound Environments for Modern Dance.” Impulse: Annual of Contemporary Dance.

Mumma, Gordon. 1967. “Creative Aspects of Live- Performance Electronic Music Technology.” In Mumma, Gordon. 2015. Cybersonic Arts: Adventures in American New Music. Chicago: University of Illinois Press. First published 1967. Audio Engineering Society, Papers of the 33rd National Convention.

Mumma, Gordon. 1982. Interview with Vincent Plush. 17 May. Oral History of American Music, Yale University Library.

Mumma, Gordon. 2011. Interview by You Nakai. San Francisco, CA, 11 November.

Mumma, Gordon. 2016. Interview by You Nakai. Vancouver, BC, 4 November.

Nakai, You. 2014. “Hear After: Matters of Life and Death in David Tudor’s Electronic Music.” communication +1 Vol. 3.

Nakai, You. 2021. Reminded by the Instruments: David Tudor’s Music. New York, NY: Oxford University Press.

Nakaya, Fujiko. 2014. Interview by You Nakai. Tokyo, Japan, 7 January.

Richards, M.C. 1966. Diary, 18-19 October. Mary Caroline Richards Paper, Box 56 Folder 6, Getty Research Institute.

Tudor, David. 1960. Letter from David Tudor to M.C. Richards, August 26-29. Mary Caroline Richards Papers, Box 26 Folder 2, Getty Research Institute.

Tudor, David. 1973. Bandoneon! Pre & Post-operative Note. David Tudor Papers, Box 16 Folder 6, Getty Research Institute.

Tudor, David. 1975. Letter from David Tudor to George Bolling, 24 October 24. David Tudor Papers, Box 19 Folder 7, Getty Research Institute.

Tudor, David. 1978. Interview by Billy Klüver. 24 November. In Experiments in Art and Technology. 1979. “Island Eye Island Ear: A Sound and Cloud Environment on Boulder Island (January 11, 1979), Appendix I.” David Tudor Papers, Box 21 Folder 8, Getty Research Institute.

Tudor, David. 1984. “‘. . . performing is very much like cooking: putting it all together, raising the temperature,’ Interview by John David Fullemann.” Stockholm, Sweden, 3 September. davidtudor.org. Accessed December 15, 2021. http://davidtudor.org/Articles/fullemann.html

Tudor, David. 1995. Interview with Matt Rogalsky. Tomkins Cove, New York, 28 March. Quoted in Rogalsky, Matthew. 2006. Idea and Community: The Growth of David Tudor’s Rainforest, 1965-2006. Ph.D. dissertation, City University London.

Tudor, David, and Experiments in Art And Technology. 2009. Bandoneon ! (A Combine). ARTPIX, DVD.

Tudor, David. 2013. The Art of David Tudor (1963-1992) Vol. 1. New World Records, 80737, compact disc.

Waldhauer, Fred. 1966. Use of the Proportional Control System at the Armory (unpublished manuscript). Klüver/ Martin Archive.

Wiener, Norbert. 1948. Cybernetics: or Control and Communication in the Animal and the Machine. Boston, MA: MIT Press.


31 January 2022
Review status
Double-blind peer review
Cite as
Nakai, You. 2022. "Late Realizations." ECHO, a journal of music, thought and technology 3. doi: 10.47041/WXPP3029


  • 1 ”The Armory was known for its long echo, which we measured up to six seconds, which before we moved in, we knew about and which horrified us. When David discovered it and John [Cage] discovered it, they realized that they had another dimension to play with. I found David one day on the balcony with a signal generator and a microphone. And he turned on the signal generator and twisted the microphone 180 degrees back and forth, picking up the signal when it had bounced off one wall of the Armory and coming back again. I usually think of it as if David were playing the whole armory as an instrument” (Klüver, quoted in Tudor and Experiments in Art and Technology (2009)).
  • 2 This is not entirely right, as Tudor did engage in performances which involved him moving around the space, most of which were Cage’s theater pieces including Theatre Piece (1960), a simultaneous performance of Variations II (1961) and Fontana Mix (1958) on two amplified pianos during the television broadcast of Musik im technischen Zeitalter (1963), or Variations VI (1966). For a detailed analysis of Tudor’s performance at Musik im technischen Zeitalter, see Nakai (2021, 123-137). Also, a notable piece in this regard was Pauline Oliveros’s Duo For Accordion And Bandoneon And Possible Mynah Bird Obbligato (1964) which she composed for Tudor and herself to perform together while sitting and moving around on a rotating see-saw created by the choreographer and designer Elizabeth Harris. For details of Duo, see Nakai (2021, 142-151) and Goldman (2012, 43-35).
  • 3 The problem of the observer who stands outside the system and delineates its limits has been long discussed in the literature of cybernetics. Towards the end of 1960s, Heinz von Foerster and others began to contemplate the role of the observer who inevitably influences the operation of the system itself, formulating the cybernetics of ‘observing systems’ as opposed to that of ‘observed systems.’ See, for instance, von Foerster (1974). I have conducted a more detailed comparison between cybernetics and cybersonics in Nakai (2014).
  • 4 In Cybersonic Cantilevers (1973), a day-long installation on May 19, 1973, at the Everson Museum in Syracuse, New York, Mumma extended the total configuration to include the active participation of audience members, many of them children and teenagers. Michelle Fillion provides a wonderfully detailed description in her liner notes to a recording of the piece: “The primary sound materials for the installation were contributed by museum visitors, who were invited to supply and input sounds of their own choice into microphones and cassette players at four stations. Rock and pop music, the tapping of teletypes, spoken words, and vintage radio crime dramas were among the randomly contributed sounds. These acoustical materials were relayed to cybersonic processors designed by Mumma, which periodically recycled and transformed them by processes of analog synthesis. The resulting delayed-action sound collage was then returned by loudspeakers into the museum space. Besides adding new materials to the collage, the participants were also capable of influencing the system by activating controls at their stations or by interacting with other participants. As the day progressed, the initial “sound playground” developed into an increasingly subtle sonic landscape in which the cumulative layers of reprocessed sounds became less and less identifiable. In the final hour, the participants spontaneously focused on making fine adjustments to the existing material. At this, Mumma abandoned his control console and walked through the museum space with a portable tape recorder, capturing the stunning results as the participants explored ‘a whole new realm of sound, floating it to see where it would go’” (Fillion 2005).
  • 5 Mesa was commissioned as music for Merce Cunningham’s dance titled Place.
  • 6 Tudor had initially planned to do a realization of Mauricio Kagel’s piece Alle Rechte vorbehalten using white noise as input, modified by the sounds of his bandoneon. This idea “abandoned itself thru the process of my projecting my thoughts into the about-to-become available technology, & its potential for the creation of ‘white noise’ from scratch” (Tudor 1973). For details of this inversion where white noise switched from a singular source material to be modified to something to be “(re)generated” through the entirety of Tudor’s giant instrument, see Nakai (2021, 234-250).
  • 7 Part of this recording can be heard in Tudor and Experiments in Art and Technology (2009) and Tudor (2013).
  • 8 “The philosophy of the artists for the pavilion was what they said ‘anti-Disney.’ It was a space where people could explore the environment and the imagery as they wanted. They were free to move around the space” (Martin 2020).
  • 9 For a detailed examination of how the Island Eye Island Ear project developed, see Nakai (2021, 306-324).
  • 10 Despite the year in the title of George Bolling’s video (“Rainforest, 1981”), this footage was very likely filmed at the De Saisset Art Gallery in Santa Clara, CA, on November 24, 1975. Bolling worked as a curator there and was responsible for organizing the performance of Rainforest that day (Tudor 1975). He also attended the show, as attested in other photographs of Tudor and Bill Viola taken by him.
  • 11 Another by-product was the double LP Monobirds: From Ahmedabad to Xenon (Topos, 2021) which compiles unreleased recordings by Tudor that were discovered in Julie Martin’s archive, edited by Jacob Kirkegaard, and accompanied by a long essay I wrote. It was on the boat back from Knavelskär that Jacob asked me if I would write something for the release he and Julie were planning.
  • 12 Starting from February 2022, I will embark on a three-year endeavor called Side Project in Hokkaido, the northern island of Japan, to explore the possibility of realizing Island Eye Island Ear as well as any by-products such an attempt might produce along the way.

Leave reply

Your browser does not meet the minimum requirements to view this website. The browsers listed below are compatible. If you do not have any of these browsers, click on the icon to download the desired browser.