Electroacoustic Music of Iannis Xenakis

33Harley

James HarleyMinnesota State University Moorhead1104 7th Avenue SouthMoorhead, Minnesota 56563, [email protected]

The ElectroacousticMusic of Iannis Xenakis

Computer Music Journal, 26:1, pp. 33–57, Spring 2002� 2002 Massachusetts Institute of Technology.

Of the close to 150 compositions that Iannis Xe-nakis created, only a handful involve electroacous-tics (sounds directly produced by electronic, digital,or other studio means). Those works, however, areinfluential beyond their number. Any history ofelectronic music must place Xenakis as a centralfigure, both for his innovations and for the impacthis music has had on successive generations. Hisinvolvement in the creation of multimedia ‘‘specta-cles’’ brought him wide exposure, although his un-compromising aesthetic vision precluded fame andfortune on a popular scale.

Nonetheless, through his work, Xenakis (see Fig-ure 1) presented a bold, charismatic persona: he wasa revolutionary, both in politics and in art. The res-tive students of the 1960s, in particular, were drawnto him, to his peculiar mixture of forward-reachingmodernism and noisy, pounding primitivism. Thereare many parallels between Xenakis’s work and ex-perimental elements of popular music, particularlyin the embracing of technology and high-density/high-amplitude sound set off by disorienting, hallu-cinatory light-shows. The electroacoustic music ofXenakis, Concret PH in particular, provided a linkbetween the ‘‘academic’’ components of the 2000 In-ternational Computer Music Conference in Berlinand the club-oriented ‘‘off-ICMC’’ celebration, heldconcurrently. Additionally, a new Asphodel releaseof his massive electroacoustic work, Persepolis, isaccompanied by ‘‘re-mixes’’ by, among others, Japa-nese techno artists.

It is my aim to present here a brief overview ofXenakis’s electroacoustic music in the manner of atutorial. Certain aspects or selected works fromthis area of his compositional output have been ex-amined in some detail (Di Scipio 1998, 2001; Hoff-mann 2000a; Solomos 1993), and a more detailedstudy of these works within the context of the restof his output is found in Harley (forthcoming).While there is a strong unity of aesthetic and com-positional technique running through his entireoeuvre, five stages in the trajectory of his studio

output can be distinguished (see Table 1). In each,the technical means change somewhat, as do theaesthetic concerns. There are connections tothe instrumental music he was writing aroundthe same time, and there are also connectionsto the electroacoustic music being written by oth-ers (though surprisingly few). Throughout his life,as clearly exemplified in his electroacoustic works,Xenakis sought to ‘‘extend the limits of musicalthought’’ (Robindore 1996).

Stage One: Musique Concrete

Xenakis arrived in Paris as a 25-year-old refugee inNovember 1947. He decided, having barely escapedGreece (as a condemned insurgent) with his life, todevote himself to music, a dream he had guardedwithin himself but which had been sidelined byturmoil, both political and personal (see Matossian1986 for fuller details of Xenakis’s early life). As atrained civil engineer, Xenakis found himself work-ing in the architectural studio of Le Corbusier,where he was able to earn a living while pursuingmusic in his spare time. He would have known vir-tually nothing of contemporary music, but the mi-lieu he worked in was certainly cultured (LeCorbusier was acquainted with Edgard Varese, forexample), and he would no doubt have heard vari-ous broadcasts on Radio-France. He very likely lis-tened to the early broadcasts of Pierre Schaeffer,who, in 1948, presented his first experiments inmusique concrete on the radio and in concert.

Xenakis met Olivier Messiaen in 1951 and beganattending his classes that year, a habit he wouldcontinue more or less regularly for the next twoyears. Messiaen was a central figure in the Parisiannew music world; Pierre Henry, who began work-ing closely with Schaeffer in 1949, had been hisstudent between 1944 and 1948 (Boivin 1995). Dur-ing this period (1951–1952), Messiaen himself vis-ited Schaeffer’s studio, producing a short piece,Timbres-durees. Karlheinz Stockhausen, who satwith Xenakis in Messiaen’s class during 1951–1952, also created a tape study during that time.

34 Computer Music Journal

Table 1. Xenakis’s electroacoustic music

Year Title Medium Duration Studio Recording

Stage I: Musique concrete1957 Diamorphoses Tape 6:53 GRM EMF CD 0031958 Concret PH Tape,

multimedia2:42 Philips/GRM EMF CD 003, Caipirinha

CAI.2027.2, Bvhaast CD06/0701

1959 Analogique B Tape 1:30 Gravesano/GRM Vandenburg VAN 0031960 Orient-Occident Tape

(soundtrack)10:56 GRM EMF CD 003

1960 Vasarely Soundtrack GRM Withdrawn1961 Formes rouges Soundtrack GRM Withdrawn1962 Bohor Tape (8-channel) 21:36 GRM EMF CD 003

Stage II: Mixed1967 Polytope de Montreal Orchestra (pre-

recorded)c. 6 min Radio-France LP

1969 Kraanerg Orchestra andtape (4-channel)

75:00 Radio-France Etcetera KTC 1075,Asphodel 0975

1970 Hibiki-Hana-Ma Tape (12-channel)

17:39 NHK Tokyo EMF CD 003

Stage III: Multimedia1971 Persepolis Tape (8-channel),

multimedia55:06 FractalOX, Asphodel

forthcoming1972 Polytope de Cluny Tape (7-channel),

multimediac. 25 min CEMAMu Mode 98/99

1977 La Legende d’Eer (LeDiatope)

Tape (7-channel),multimedia

46:00 WDR Cologne/CEMAMu Montaigne MO 782058

Stage IV: UPIC1978 Mycenae Alpha

(Polytope deMycene)

Tape (2-channel),multimedia

9:36 CEMAMu Neuma Records 450-74,Mode 98/99

1981 Pour la Paix Voices and tape,radiophonic

c. 27 min CEMAMu/Radio-France none

1987 Taurhiphanie Tape (2-channel),multimedia

10:46 CEMAMu Neuma Records 450-86

1989 Voyage Absolu desUnari versAndromede

Tape (2-channel) 15:25 CEMAMu Perspectives of New MusicPNM 28

1997 Erod Tape (2-channel) c. 5 min Ateliers UPIC Withdrawn

Stage V: Stochastic Synthesis1991 GENDY301 Tape (2-channel) 14:15 CEMAMu Withdrawn1991 GENDY3 Tape (2-channel) 18:45 CEMAMu Neuma Records 450-861994 S.709 Tape (2-channel) 7:03 CEMAMu EMF CD 003

35Harley

It was a few years later, in 1954, that Xenakiswas accepted as one of the first members of theGroupe de recherches de musique concrete (re-named Groupe de recherches musicales in 1958,now commonly called GRM). His assimilation ofthe concerns of contemporary musical culture hadproceeded rapidly. That same year, he completedhis first major orchestral score, Metastaseis, a workthat explores novel sonorities built from complexconfigurations of string glissandi (see Figure 2); itprovoked a major scandal at its premiere in Don-aueschingen the following year. Drawn to the newpossibilities of the studio, Xenakis expressed his in-tention to ‘‘study the evolution of timbres, dynam-ics, and register . . . to make chromosomes ofattacks’’ (Matossian 1986).

Diamorphoses

Xenakis completed his first electroacoustic compo-sition in 1957. Diamorphoses (Greek for ‘‘continu-ity—discontinuity,’’ ‘‘two aspects of being’’) is a

relatively brief work, but sonically intense andquite unlike other music being produced at GRMat that time (such as Orphee, the opera by Schaef-fer and Henry, or even Deserts, by Varese). Therewas a strong concern among many of the compos-ers working in the studio to create sounds fromrelatively discrete instrumental or real-worldsounds. Xenakis, on the other hand, sought to cre-ate dense, noisy textures that bore little direct re-semblance to the world around him. Instead, hissounds derived from the same imagination thatwas creating complex clouds and masses in the or-chestra—an imagination, it should also be pointedout, strongly affected by the chaotic, traumaticsounds of war. (One might also speculate that thehearing loss, particularly in the higher frequencies,Xenakis suffered from his injuries during that timewould have shaped, to some extent, his attractionto ‘‘noisy’’ sounds and to high-decibel playback lev-els.)

Indicative of the dichotomy described by the ti-tle, Diamorphoses combines noisy, primarily low-frequency sounds—derived from an earthquake, ajet engine, and a train—with more sharply defined,high-register bell sounds. The natural glissando ef-fect of the jet engine winding up is combined withglissandi of other sounds, produced in the studio bymeans of tape manipulations. Often, the compositesonority produces something quite new, thesources being submerged.

The formal organization is quite clear: the twoouter sections of more-or-less sustained sounds(layered noises, glissandi) are contrasted with a cen-tral, more discontinuous passage, filled with manyshifts of sonority (starting with the bell sounds,then adding others). Xenakis worked in particularto create shifts and continuities of density throughlayered variations of sound objects (Delalande1997). For the discrete sonorities, he used probabili-ties to calculate attack points, layering the tracksin the studio to achieve a perceivable range or scaleof activity.

Concret PH

Xenakis carried these concerns to his next electro-acoustic piece, Concret PH, produced in 1958. (Ac-cording to Delalande, it was revised in 1961 to

Figure 1. Xenakis at themixing console, Le Dia-tope. (Photograph by Mali;used by kind permission ofthe Xenakis family, alongwith all subsequent sketchand score examples.)


create a stereo version, then re-mixed in 1969 ontofour channels.) This 2-min miniature was producedas an introduction to the Poeme electronique of LeCorbusier and Varese for the Philips Pavilion inBrussels. Xenakis had been heavily involved in thedesign and construction of the Pavilion (see Figure3), often standing in for Le Corbusier who was alsoworking on a major project in India (Treib 1996).His music was intended to fill the gap between pre-sentations of the 8-min performances of the mainmultimedia work, which occurred at 10-min inter-vals. In other words, it was intended to fill in thetime as the audience filed into and out of the pavil-ion. Nonetheless, for a young, relatively unknowncomposer, the opportunity to create a work for pre-sentation at a major international exposition that

would be heard by many thousands of peoplewould have been enticing. In addition, the facilitiesthat Philips had developed in order to projectsound, light, and images were among the most ad-vanced in the world.

Unfortunately, Le Corbusier, who was operatingunder enormous pressure and seriously behindschedule, would not grant Xenakis leave to work atthe Philips studio in Eindhoven, Holland. Heneeded him to supervise the construction of the pa-vilion, and seemed to consider the composition ofthe little interlude of secondary importance. Va-rese, on the other hand, was able to take advantageof Philips’s technology, including special equip-ment for the projection of sounds over multipleloudspeakers. (The pavilion ended up housing over

Figure 2. Excerpt fromMetastaseis, beats309–314.

37Harley

400 loudspeakers.) Xenakis, in his spare time, wasinstead forced to work in the rather primitive fa-cilities of the Philips offices in Paris. He produced amonophonic version there, and then completely re-composed the piece later at GRM. (Xenakis pro-duced multiple versions of all his tape pieces, up toLa Legende d’Eer, mixing for different numbers ofchannels, accounting for different playback situa-tions, and so forth.)

The only sound source used in Concret PH is thecrackling and hissing of burning charcoal. Thestudy of density and the application of stochasticfunctions to the calculation of articulation pointsfor each layer of sound were the main concerns.The crackling texture evolves in a continuous fash-ion, much like the central portion of Pithopraktra,his second orchestral score, completed in 1956. Thesense of spatial movement as the music movedalong the trajectories through the loudspeakers ofthe pavilion in Brussels would have been particu-larly effective using these primarily mid- to high-register grains of sound. Certainly, it was utterlyunlike any other electroacoustic music of the time,and remains a gem of the tape music idiom.

Analogique B

Xenakis first met the renowned Swiss conductorand researcher Hermann Scherchen in 1954. Theelder musician became something of a mentor to

the fledgling composer, encouraging him in partic-ular to work out his ideas of combining mathemat-ics and music so that they could be expressedcogently. Scherchen held a ‘‘retreat’’ every year athis estate in Gravesano, to which he invited lead-ing thinkers of the time to discuss a range of top-ics. (Guests included Werner Meyer-Eppler, LejarenHiller, Max Mathews, and others.) The proceedingswere published in the Gravesaner Blatter.Scherchen’s interests ranged to acoustics and newtechnology, and he had actually developed a rudi-mentary studio for carrying out experiments andproducing electroacoustic music.

By 1958, Xenakis had become interested in thetheory of acoustical quanta (Gabor 1947), the pre-cursor to granular synthesis (Roads, 2002). He haddeveloped a proposition for composing on the basisof ‘‘screens,’’ a representation of clouds of quantafor each ‘‘slice of time’’ (Xenakis 1992). The pro-gression from one screen to the next is governed byMarkov chains, whereby the settings for one screenexert an influence on the calculations for the next,building a form of ‘‘memory’’ into the temporalorganization of the music.

That year, Xenakis composed Analogique A, ashort work for strings in which there are nothingbut discrete notes (no held pitches, glissandi, tre-moli, etc.). The music proceeds as a succession ofchanging densities, registers, ranges, and intensities(along with timbral changes from plucked notes tostruck or short bowed ones). In the summer of1959, during his stay with Scherchen in Gravesano,Xenakis produced a complementary composition,Analogique B, for discrete electronic sounds, pro-duced in this case by oscillators. The procedurewas the same, though the particular parametricalconstraints are of course different. The composerworked out an interlocking scheme whereby thetwo pieces can be performed as a unit, the stringsalternating with the pre-recorded electronic sounds,with some moments of simultaneity.

Analogique A�B was premiered in Paris in 1960along with the first French performance of Pitho-prakta, conducted by Scherchen. From that point,Xenakis’s music began to receive more dedicatedattention in his adopted land. A final break with LeCorbusier provided him with the opportunity to

Figure 3. Philips Pavilion,Brussels World Fair, 1958.(Used by kind permissionof the Getty ResearchInstitute. � 2002 ArtistsRights Society (ARS), NewYork/ADAGP, Paris/Fondation Le Corbusier.)


quit his day job (for the most part) and devote him-self at last professionally to music.

Orient-Occident

One of the first commissions Xenakis receivedthrough his association with GRM was for a sound-track to a documentary film by Enrico Fulchignonifor UNESCO. The film, Orient-Occident (1960),tried to express by visual and sonic means the rela-tions and resonances of artistic relics from variousancient cultures. The film was based upon an exhi-bition at the Cernuschi Museum in Paris, wherethe director spent three months shooting. Xenakiswas not brought into the project until that phasewas complete. Mr. Fulchignoni intentionally gaveno instructions or ‘‘interpretation’’ to the com-poser, preferring to allow him the liberty of work-ing from his own reactions and analyses(Fulchignoni 1981). The film links disparate erasand cultures primarily through visual resemblancesrather than any sort of chronological or culturalnarrative. In effect, this more abstract approach en-abled Xenakis to construct his own ‘‘narrative’’ tounderscore the images.

The original 22-min soundtrack is not wellknown. What is available is the concert work ofhalf that length extracted from it. A 4-channel ver-sion was apparently produced at GRM in 1969 (De-lalande 1997), and available recordings show someslight discrepancies of detail. (For example, the ver-sion released on a compact disc from the ElectronicMusic Foundation stops 16 sec before the versionthat had been released on vinyl by Nonesuch.)

Compared to Diamorphoses or Bohor, the soundsof Orient-Occident are much clearer (i.e., lessnoisy). A bow drawn over various objects is thesource of many of the sustained sonorities. Thepercussive sounds are often presented as regularpulsations or perceptible patterns rather than sta-tistical ‘‘clouds.’’ The music is dramatic, with themore reflective sounds being surprised by sharp at-tacks or ‘‘ethnic’’ drumming-like patterns. Thereare some silences, unusual in Xenakis’s electro-acoustic output, but also grating metallic soundsthat resemble sonorities from other pieces. The

granular texture of burning charcoal from ConcretPH makes an appearance in the latter part of thework, this time mixed with water droplets andother sounds.

The formal outline of Orient-Occident is lesssharply defined than is usually the case in Xe-nakis’s music, no doubt due to its original concep-tion as a soundtrack. The music proceeds from onesection to another by shifts of sonority, effected attimes by gradual transition and at other times bysharp divisions. The liner notes of the NonesuchLP state that ‘‘Xenakis looks upon this work as asolution to the problem of finding many differentmeans of transition from one kind of material toanother’’ (Brody 1970). In drawing upon a relativelywide range of sonorities, none recognizably linkedto any particular culture but related to each otherby various means of transition or juxtaposition, Xe-nakis enables the listeners (and presumably theviewers, originally) to create their own associationsbetween the sounds and images from ancient cul-tures.

Vasarely and Formes rouges

In the period 1960–1961, Xenakis produced twomore soundtracks in addition to Orient-Occident.Never again would he work in this domain. (Holly-wood decidedly did not beckon!) Vasarely (1960)was a similar project to the previous, more well-known piece, a short film by Peter Kassovitz and E.Szabo based on an exhibition of Op Art paintingsby Victor Vasarely (see Figure 4). The composerwas quite taken with this artist, and he ended upworking with him a few years later on his ballet,Kraanerg.

Formes rouges (1961) was a short animated filmby Piotr Kamler. Little is known of the music forthese films, as they were both quickly withdrawnfrom the composer’s catalog. Produced at GRM,which is the reason for noting them here, thesepieces were primarily instrumental rather thanelectroacoustic, apparently. Regardless, the soundswould have been assembled in the studio.

During this same period, however, Xenakis wasbecoming more and more drawn into completing a

39Harley

stochastic algorithm for composing music usingcomputers. These mathematical, abstract concernsconflicted both with the dictates of producing mu-sic for film and with the general aesthetic directionof the GRM studio as directed by Pierre Schaeffer.

Bohor

In 1962, Xenakis completed his most ambitiouselectroacoustic work to date, and subsequentlywithdrew from being an associate of GRM. Bohorwas one of the first 8-channel works created atGRM (John Cage and Earle Brown had created 8-channel works back in 1952/1953). The piece con-

stitutes a radical exposition of the issues of densityand transition that Xenakis had been concernedwith throughout this period. It is conceived as asingle, slowly evolving gesture lasting close to 22min. The composer describes it as being ‘‘monisticwith internal plurality, converging and contractingfinally into the piercing angle of the end’’ (Brody1970).

There are two basic textures in Bohor: one, aconcentrated, teeming sonority, constantly in mo-tion, ranging in timbre from bell-like sounds (fil-tered to muffle the resonance) to metallic rattlings,finer-grained metallic sounds closely resemblingthe charcoal crackles of Concret PH, and noisy,clashing and crashing sounds; the other is a low,

Figure 4. Vega image byVictor Vasarely.


sustained sonority adapted from a Laotian mouthorgan, the characteristic crescendo-decrescendo andbreaking off for breath being preserved. The lowsounds are heard twice, primarily, beginning at the5’30’’ mark, and lasting about 8 min, and reappear-ing after the 17’00’’ mark and lasting 1–2 min. Theother texture continues throughout, adding layersto itself one by one until there are several going onat the same time. On the 8-channel version, thesewould be distributed around the different channels.Each layer builds and subsides independently, pro-ducing an opaque, constantly-evolving texture thatat the same time seems to have no clear linear tra-jectory. When the low sustained sound drops outthe first time, there is a noticeable loss of intensityand tension, but the metallic sounds build upagain, leading at the end to a sublimation of all theother layers by the noisiest, crashing sonority,which sharply boosts its mass and dynamic levelover the final three minutes. (As with Orient-Occident, the version presented on the ElectronicMusic Foundation compact disc stops 20 sec soonerthan the Nonesuch LP recording. This is regret-table, as this last segment consists of a greatlyheightened culmination of the crescendo of dynam-ics and density—the ‘‘piercing angle’’ Xenakismentions.)

Bohor is, perhaps ironically, dedicated to PierreSchaeffer. The elder director, who had in fact beenquite supportive of Xenakis throughout his sevenyears of ‘‘apprenticeship’’ at GRM, was horrified bythe piece. Part of his reaction was to the extremelevel of amplification at the Radio-France premiere,intensified by the distribution of loudspeakers sur-rounding the audience. As he put it, ‘‘this was nolonger tiny embers, each with its own allure, thiswas an enormous burst of explosions, an offensiveaccumulation of lancet jabs to the ear at maximumvolume level’’ (Schaeffer 1981). Xenakis’s ‘‘war’’aesthetic was obviously still present, but also para-mount was the formal concern for sustaining ten-sion for a long period of time through the layeringand variation of density and other sonic parame-ters. Later electroacoustic pieces would carry thisapproach much further.

Xenakis left GRM, but he did not cut himself offfrom the studio definitively, nor did Schaeffer de-

mand it. His works continued to be performedthere and issued on recordings, and he visited anumber of times in subsequent years to producenew mixes of his pieces and to make other uses ofthe facilities (Delalande 1997).

One of the sources of conflict was Xenakis’sgrowing conviction that digital technology was vi-tal to the future of music. His growing stature as acomposer added weight to his argument that GRMshould be the first French institution to equip itselffor carrying out research in computer music. Thisorientation was anathema to Schaeffer, and Xe-nakis realized that he would be better off workingto establish such a facility elsewhere.

A rift had also developed between Xenakis andsome of the other GRM composer-associates. A‘‘collective concert’’ had been planned for 1962 in-volving nine composers, the others being ClaudeBallif, Francois Bayle, Edgardo Canton, Luc Ferrari,Francois-Bernard Mache, Ivo Malec, Bernard Parme-giani, and Michel Philippot. Xenakis took charge oforganizing the event, and put forward the idea thatthe succession of electroacoustic fragments be de-termined according to a probability matrix. Hisproposal was rejected by the others, who preferredto work more intuitively. Xenakis withdrew fromthe project, and there was evidently a certainamount of ill feeling (Schaeffer 1981).

Stage Two: Mixed

Having made the decision to distance himself fromGRM, Xenakis became preoccupied with other con-cerns. He wrote music for the stage (Hiketides, Or-esteıa, Medea), began lecturing, finished writing hisbook Musiques Formelles, composed various in-strumental and vocal pieces, did a residency in Ber-lin, and, in 1967, took up a position at IndianaUniversity.

One of his concerns through this period (andthroughout his career) was the spatialized presenta-tion of his music. He had worked with Philips onthe elaborate sound projection system for the Phil-ips Pavilion in 1958, and had produced Bohor withthe 8-channel ‘‘surround’’ diffusion in mind. Xe-nakis extended this thinking into his instrumental

41Harley

and stage music as well. Eonta (1963), for brassquintet and piano, requires the brass players tomove around the stage, point the bells of their in-struments in different directions, and so forth. InOresteıa (1966), the chorus members are asked tocreate ‘‘clouds’’ of stochastic sounds by playingsmall percussion instruments and sirens whilemoving around the stage. Terretektorh (1966), fororchestra, distributes the individual musiciansamong the audience, so that each listener has aunique experience of the piece.

Polytope de Montreal

In 1966, Xenakis received a prestigious commis-sion: to create a multimedia work for the FrenchPavilion at the 1967 Montreal World Expo. Hewould have preferred to design the pavilion itself,but that job was awarded to architect Jean Faugeron

(who won the Grand Prix de Rome for this work).In fact, the pavilion is still standing, presentlyhousing the Casino de Montreal.

For the large central space of the pavilion, whichrises through several floors and is accessible fromall of them, Xenakis constructed five networks ofintersecting steel cables, each outlining curved geo-metrical shapes (see Figure 5). Onto these cableswere attached 1,200 flash bulbs (800 white, 400colored), which were triggered by a control systeminvolving a perforated tape and photosensitivecells. This Poeme de lumiere comprised a succes-sion of visual configurations, the aim being to ‘‘cre-ate a luminous flow analogous to that of music’’(Fleuret 1988). Xenakis used interconnected tech-niques to compose the music and the ‘‘poem oflights,’’ but for him, ‘‘the link is not between thembut beyond or behind them’’ (Varga 1996).

The music for Polytope de Montreal is scored forfour identical instrumental ensembles, each to be

Figure 5. Sketch forPolytope de Montrealinstallation.


placed along one of the four cardinal directionswith the audience placed in the intervening quad-rants. The score, however, has apparently neverbeen performed live. It was recorded at Radio-France prior to the presentations in Montreal, andthe music was played by means of loudspeakers.The use of a recording, even broadcast from fourspaced loudspeakers, would obviously not be thesame as performing the music live, but Xenakis de-signed the piece very much in terms of layers of so-norities, much as he had composed Bohor. Thus,while not strictly electroacoustic, Polytope deMontreal exhibits similar concerns.

Instrumental sonorities produced by similar in-struments in each ensemble are fused primarilythrough register, and dynamic swelling effects areused throughout to produce the sensation of move-ment as the aural spotlight shifts from one ensem-ble (or channel) to another. The layers are treatedindependently, so that the overall spatialization isquite complex.

The architecture of the music, though, is clear,being constructed from three contrasting sections,each containing a number of distinct sonic ele-ments. While the music is intense, concentratingon registral extremes and being filled with dynamicgestures, it does seem to leave perceptual space forthe visual elements that were intended to occuralongside. There are no programmatic links be-tween the design of the cables, the succession offlash bulbs, and the music. But, as Xenakis stated,‘‘We are capable of speaking two languages at thesame time. One is addressed to the eyes, the otherto the ears. The content of the communication isdifferent but sometimes there’s a link between thetwo’’ (Varga 1996).

Kraanerg

The Polytope de Montreal made quite an impact inMontreal during its run there in 1967. So, it wasperhaps not surprising that Xenakis’s name cameup a year later when officials at the National Balletof Canada were seeking to commission a new,modernist dance work for the gala inauguration ofthe National Arts Centre in Ottawa, the nation’scapital (Harley and Harley 1997).

The choreographer was Roland Petit, fromFrance. Xenakis was given complete freedom bythe choreographer and the commissioners to createhis music, aside from the proviso that it be at least75 min in length, and that it incorporate, if possi-ble, an electronic element, along with the chamberorchestra, to highlight the sophisticated playbacksystem installed in the new facility (with loud-speakers all around the hall). It was the composerwho recommended that Victor Vasarely be engagedto create the set (see Figure 6).

Kraanerg (Greek for ‘‘to perfect, accomplish; [ce-rebral] energy’’) is concerned, according to the com-poser, with ‘‘the overwhelming fight of man’s brainand the [un]ending obstacles that exist or that hehimself creates’’ (Xenakis 1969). Xenakis was alsothinking, apparently, of the student demonstrationsgoing on in Paris and elsewhere through the periodhe was working on this piece, extending those so-cial conflicts to global concerns:

In barely three generations, the population ofthe globe will have passed 24 billion. 80% willbe aged under 25. The result will be fantastictransformations in every domain. A biologicalstruggle between generations unfurling all overthe planet, destroying existing political, social,urban, scientific, artistic and ideological frame-works on a scale never before attempted by hu-manity and unforeseeable. This extraordinarymultiplication of conflict is prefigured by thecurrent youth movements throughout theworld. (quoted in Toop 1989)

The music of Kraanerg is primarily instrumentalin conception, similar in its textures to the orches-tral score, Nomos gamma, completed that sameyear. The tape part, originally conceived in fourchannels, is built from processed instrumentalsounds, apparently based on recordings done withthe same 23-member chamber orchestra as the liveparts. The music needed to be written, produced,and recorded within about six months, in time tobe ready for the rehearsals with the dancers. Itseems likely that the orchestra parts were recordedand then sections of that used to create the electro-acoustic part.

The piece is built from blocks of material alter-

43Harley

nating between tape and live ensemble. The obvi-ous reference is Edgard Varese’s Deserts (1954), butKraanerg is much more elaborate in the structuringof the mosaic-like form. The tape often overlapsthe live parts, to create more of a sense of continu-ity than is found in Deserts. The overlapping maybe slight or may be more extended. The continuityis enhanced by the similarity of material betweenthe recorded sounds and the instrumental ensem-ble. There is never any confusing them, though, asthe produced sounds are always processed, some-times quite heavily (primarily through filtering, of-ten to obtain degrees of distortion). Silence is alsotreated as an important structural element, and

there are 21 such passages of stillness, ranging induration from 2 sec to 28 sec. These occur primar-ily in the middle portion of the work, so are not in-tended to delineate large-scale formal divisions.The proportional relationship between tape and en-semble does shift at certain points, and thesechanges create a tripartite form. The first sectionpresents the two entities with more or less equalattention (eight blocks each); the second, beginningat 23’00’’, shifts the focus to the live ensemble; thethird, beginning at 52’00’’, settles on the tape, pri-marily. The final 6 min, for example, are entirelyfor tape.

Within these blocks of sound, Xenakis plays off

Figure 6. Photo of stage de-sign by Vasarely and per-formance of Kraanerg.(Photo by Anthony Crick-

man. Used by kind per-mission of the NationalBallet of CanadaArchives.)


the orchestral groups (woodwinds, brass, andstrings) one against another, sometimes layering,sometimes juxtaposing them. The music is oftenbuilt from global sonorities, or textures, but thereare also moments of solo melodies and exposedchamber-like combinations of instruments.

The ballet as choreographed by Mr. Petit was lessthan successful, though the National Ballet of Can-ada did present the work numerous times on tourin Europe and North America over the next fewyears. Although there were a few concert perfor-mances, no other company took up Kraanerg untilGraeme Murphy, of the Sydney Dance Company,created a new choreography in 1988. The music,though, was praised from the beginning, and be-came known through the LP recording that was is-sued from the original recording produced atRadio-France for the rehearsals in Ottawa.

Hibiki-Hana-Ma

By 1970, 10 years into his career as a professionalcomposer, Xenakis was known around the world.For the Osaka World Exposition that year, he wasinvited to the studio facilities of NHK (NipponHoso Kyokai) in Tokyo to compose a tape work tobe presented in the Japanese Steelworkers Federa-tion Pavilion. No doubt one of the great attractionsof this commission was that the music was to beprojected through a sound system of 800 loud-speakers grouped in 250 locations.

Hibiki-Hana-Ma (‘‘reverberation-flower-interval’’), originally conceived for 12 channels,was later mixed down to eight for concert diffusion(and then to two for the issuance of the recording).At the radio studio in Japan, Xenakis had access toan orchestra, and, like Kraanerg, much of the mate-rial comes from orchestral sonorities. In this case,the sources were ‘‘typical’’ textures from existingscores with emphasis on the strings, particularlyfeaturing glissandi and natural harmonics. To thisthe composer added the biwa, a Japanese lute-likeinstrument, and some additional percussionsounds.

There are varying degrees of studio manipulationof these instrumental sounds, from virtually none

to so much that the original sources are unrecog-nizable. There is a much wider range of sounds pre-sented in Hibiki-Hana-Ma than in Kraanerg, whichis understandable considering that the electro-acoustic element must stand on its own. The possi-bility of deploying up to 12 channels enabledXenakis to build up layers and complex superposi-tions of sonorities. The music is put together fromblocks of material spliced into the different chan-nels, in a similar process to the ballet piece, thoughextended from three or four layers to 12. There aremany sudden shifts of sonority, density, and inten-sity, and various layers are brusquely cut in or out.

Major articulation points serve to loosely dividethe piece into four sections. The first section, last-ing to 3’00’’, is built from a low, booming, undula-tory sonority over which orchestral glissandi stringsounds are layered. A sweep up to a sustained high-register cluster signals the start of the second sec-tion, which introduces a layer of tinkling bells,stochastic clouds of whips and pizzicati, and muchelse. A sudden drop in dynamic level and in thenumber of layers at 6’32’’ signals the third section,resembling the previous one in featuring succes-sions of a wide range of sonorities, most of whichhave been heard already. Here, though, the orches-tral winds are introduced, playing both sustainedsonorities and glissando textures reminiscent ofsimilar passsages in Nomos gamma and Kraanerg.At 11’07’’, another sudden drop in intensity/densitysignals the shift to the final, most substantial, sec-tion. Through this passage, various noise-based so-norities are introduced, of both sliding andfixed-band types. This material carries on to theend, layered with previously introduced sounds.

The impact of hearing such a wide range of pro-duced sonorities based on acoustic sources, bothsustained and percussive, as they were projectedthrough the pavilion’s large sound system musthave been powerful. What made a strong impres-sion on the composer, though, were the dynamiclight sculptures and laser projections presented inthat same pavilion by Japanese artist Keiji Usami.Xenakis was particularly interested in the newtechnology used to control the lasers and to syn-chronize the lights with the sound. In this, he waspreparing to take on a series of large multimedia

45Harley

projects for which he alone, finally, would be re-sponsible.

Stage Three: Multimedia

In 1969, Xenakis had traveled to the ancient site ofPersepolis, Iran, for the premiere of his ambitiouswork for six percussionists, Persephassa. The Shi-raz Festival was organized under the patronage ofthe Empress of Iran, and Xenakis and his musiccreated quite a sensation (Matossian 1986). He wasapparently engaged on the spot to create an evenmore ambitious work for the 1971 festival. Thistime, the composer was to create a multimediaevent, with all the resources of Iranian Radio andTelevision placed at his disposal.

Persepolis

The one limitation Xenakis was faced with in plan-ning his first real ‘‘polytope’’ was that the locationwithin the ruins of such an important historical

site necessitated a ban on construction or renova-tion (see Figure 7). All loudspeakers (59), lights (92spotlights), and lasers (2) needed to be installed inas non-invasive a way as possible (see Figure 8).

The lasers and spotlights were ranged throughoutthe site and projected to create ‘‘luminous patternsevoking the Zoroastrian symbolism of light as eter-nal life’’ (M. A. Harley 1998). From the central por-tion of the site, where the 59 loudspeakersprojected the eight channels of sound throughoutthe audience area, the lights swept upward and outtoward the hillside tombs of Darius and Artaxer-xes. There, in the distance, bonfires were burning,and parades of children carrying lighted torcheswended their way up the hillsides, creating ever-changing linear patterns.

The music, with its noisy sonorities and overlap-ping waves of intensity, looks back to Bohor. Per-sepolis, though, is 55 minutes in length, a verylong span for a continuously evolving form. Ac-cording to his sketches, Xenakis constructed thetape from 11 sonic entities distributed among theeight channels (see Figure 9). The sketch showngives precise timings for the first reel of the origi-

Figure 7. Archeologicalsite of Persepolis, Iran.


nal tape, which ends at 31’30’’. The performancewould have required two 8-track machines so thatthe piece could be performed with no breaks. Mul-tiple layers of similar material create overall tex-tural ‘‘zones’’ which serve to delineate the form,though the shifts from one to another are rarelyeasy to distinguish.

The sonic entities range from clarinet multi-phonics (3, as numbered in the sketch), to complex,high sustained sounds derived from string harmon-ics (2), to low, sliding distortions of timpani rolls(9), to gongs (7) and ceramic wind-chimes (11). Oth-ers are harder to identify, but one is derived fromrecordings of cardboard being handled (6), and an-other sounds as if a strong, buffeting wind had beenfed through a distortion module (8). The remainingentities can be identified as metallic, noisy sonori-ties. Entities 1 and 4 are not used at all in the firstpart, occurring only in the final moments of thepiece.

None of these entities are simple or ‘‘pure’’ so-norities, by any means, and the sonic intensity of-ten seems overwhelming. All of the material isdeveloped rather than just being repeated, so thatthe music evolves, while remaining unified, overthe course of its journey through this violent,shrouded soundscape. Hearing the music in the set-ting for which it was intended, the dark ruins of

Persepolis within the desolate beauty of the Iraniandesert, with spotlights sweeping the sky and firesburning in the distance, would have been an awe-some experience.

Polytope de Cluny

Polytope de Cluny (1972), Xenakis’s next multime-dia creation, followed fast on the heels of Persepo-lis. It was commissioned for the Festivald’Automne in Paris, and was set in the historic Ro-man Baths of Cluny, just off the Boulevard Saint-Germain-des-Pres. It premiered in October 1972,and ran for 16 months, four times daily, with thecumulative audience reaching over 200,000. Theexperience was popular among young people, who,in their radical political and social orientation,sought ‘‘music that transcended the limits of tradi-tion and nationalism’’ and rejected the ‘‘formal ap-parel and conventions’’ of the concert ritual.Underground in Cluny, they ‘‘sat on the floor, sur-rounded by strange sonorities and subjecting them-selves to perceptual and aestheticexperimentation’’ (M. A. Harley 1998).

The T-shaped chambers of Cluny were, like thehistoric site in Iran, not to be altered, so Xenakis’sinstallation was erected within the walls by means

Figure 8. Composer’ssketch for distribution ofloudspeakers and lights forPersepolis.

47Harley

of scaffolding and cables. Along with the flashbulbs reminiscent of Polytope de Montreal (600 innumber) were three lasers (colored red, green, andblue) projected along pathways determined by 400adjustable mirrors. Xenakis was able to control theoperations of the lights, lasers, and mirrors throughdigital means. All of the calculations were pro-grammed on a computer, and the output data wasthen converted to electromagnetic signals.

This implementation of digital technologybrought to fruition one of Xenakis’s long-helddreams. In 1966, he had established an organiza-tion, EMAMu (l’Equipe de mathematique et auto-matique musicales), which in 1972 became

CEMAMu (Centre d’Etudes de mathematique etautomatique musicales), with the goal of pursuingcomputer music and related research. He had ap-parently attempted to set up a parallel operation atIndiana University during his tenure there, but hewas unsuccessful, and he resigned his position in1972, the same year of Polytope de Cluny and theconsolidation of CEMAMu. One of his interestswas the refutation of Fourier analysis as the basisfor sound synthesis. As he put it in his chapter on‘‘New Proposals in Microsound Structure:’’

[S]pectral analysis of even the simplest orches-tral sounds . . . presents variations of spectral

Figure 9. Composer’ssketch for the 8-channeltape piece, Persepolis, upto 31’30’’.


lines in frequency as well as in amplitude . . .These tiny variations . . . would certainly re-quire new theories of approach, using anotherfunctional basis and a harmonic analysis on ahigher level, e.g., stochastic processes, Markovchains, correleated or autocorrelated relations,or theses of pattern and form recognition. (Xe-nakis 1992)

The electroacoustic music for the Cluny instal-lation was produced on seven channels of the 8-channel tape, with the last track reserved forelectromagnetic control signals for the lights, la-sers, and mirrors. Timing between the various ele-ments, including the music, could thus be preciselycontrolled—although, as before, Xenakis soughtparallels rather than unions: ‘‘I wanted to establisha contrast: the lights are a multitude of points,with stops, starts, etc., and the music is continu-ous, for although the sound changes it does notstop’’ (Fleuret 1972).

The music was presented through 12 loudspeak-ers distributed around the site. The event, whichran some 25 min, falls in between the 6-min dura-tion of the Polytope de Montreal and the hour-longlength of Persepolis. The music bears some resem-blance to the Iranian piece, even borrowing sonicmaterial. There is much that is new, as well, in-cluding a wild, brassy sound that is treated exten-sively throughout. There is a greater prominencegiven to textures of percussive sounds, such as ce-ramic wind-chimes from the previous piece, butothers as well, including a plucked African thumb-piano. There is a striking moment at the end, whenthe focus suddenly turns to the thumb-piano, anexposed single note being plucked slowly andevenly, the ear zooming in on the rich, rattling, res-onances of this distinctive sound.

As one would expect, the various sounds are gen-erally layered and distributed across the sevenchannels, though the density is not as high as theearlier pieces, possibly in deference to the vaulted,reverberant performance space. Particularly notablein Polytope de Cluny are the synthesized sounds,created using stochastic algorithms. Xenakis wasproud to have been the first in France to producecomputer-generated sounds, even if he lagged be-

hind Bell Labs, where digital synthesis had beenlaunched by Max Mathews in 1957.

Le Diatope; La Legende d’Eer

At first glance, Le Diatope (the title of the installa-tion) would appear to be a variation of Polytope deCluny. The electroacoustic music is recorded onseven channels, with the final channel reserved forcontrol signals. The multimedia performance in-volves 1,680 flash bulb lights, 4 colored lasers, and400 programmable mirrors. This was a far moreambitious project, however. It was commissionedfor the inauguration of the acclaimed CentreGeorges Pompidou in Paris, and it also included anoriginal architectural component. At long last, Xe-nakis was able to design the free-standing shellwithin which the spectacle would take place (seeFigure 10). It was also, at 46 min, about twice thelength, an ambitious duration for lighting effectsthat required updating 25 times per second.

In addition, the music was commissioned sepa-rately, and was clearly conceived to stand as an in-dependent piece. WestDeutscher Rundfunk (WDR)invited Xenakis to produce the piece using the fa-cilities in Cologne, and it was in fact premieredthere in February 1978, months prior to the open-ing of Le Diatope outside of the Centre Pompidouin Paris. Its North American premiere took place atthe 1978 International Computer Music Confer-ence in Evanston, Illinois (to much acclaim, andsome controversy).

Xenakis, almost uniquely in his entire output,expended much effort on the program notes for LeDiatope, presenting his own thoughts and gather-ing inspirational texts which, while not forming anarrative, resonate in multiple ways with the cos-mic, apocalyptic scope of the sounds and lights.The title of the music, La Legende d’Eer, is takenfrom the concluding passage of Plato’s Republic, inwhich a soldier is killed in battle then broughtback to life full of images of the afterlife, includingthe famous ‘‘music of the spheres.’’ The medievalera is represented by Hermes Trismegistus, an al-chemist, who, in a similar way, is given a vision ofthe boundless darkness and light of immortality.

49Harley

The passage from Blaise Pascal’s Pensees contem-plates the insignificant place of humanity withinthe infinity of nature, and Jean-Paul Richter carriesthe vision further, writing of the terror of beingalone in the universe. The final text is a scientificdescription of a supernova, presenting its awesomesize and energy with detached precision. It is al-ways easier to interpret words than music, andclearly, these texts all share a vision of the vastnessof the universe, with different images of light andlife within that infinity. The composer, in discuss-ing Le Diatope and his other polytopes, confessed:‘‘I want to bring the stars down and move themaround. Don’t you have this kind of dream?’’ (Ma-tossian 1981).

Even if less protracted in duration than Persepo-lis, La Legende d’Eer is vast in scope, with a more

concentrated, directed formal outline. The overalltrajectory, which proceeds without break, is one ofa gradual descent, returning at the end to the highwhistling sounds of the opening. The multi-channel projection of the sounds aids perception ofthe different, often overlapping, entities, and alsoallows rather subtle shifts in emphasis to be ef-fected.

There are eight basic sonic entities used in LaLegende d’Eer, present or dominant at differenttimes (see Figure 11). Each is treated to a great dealof studio manipulation, including transposition, fil-tering, and reverberation. Each is also varied interms of density. These techniques serve, on onelevel, to create links between the different entities.For example, the high whistling tones of the open-ing, smooth at first, are varied with tremolo/

Figure 10. Design ofLe Diatope.


amplitude-modulation effects to create a morestriated variant. This developed sonority resembleshigh transpositions of the rattling ceramic sonority,itself narrowly filtered to produce a recognizableidentity of pitch. These two entities in turn resem-ble the high, narrow-band version of the granulatedundulating noise, and even the drum (tsuzumi) ele-ment, transposed and mixed to create a texture ofhigh density. Thus, while the entities are relativelydistinct, they can also be related along differentparametrical continua.

La Legende d’Eer can be divided into eight sec-tions. The first is the clearest, containing only thehigh, whistling sounds that Xenakis calls ‘‘sonicshooting stars.’’ Over 6 min, this layer alternatesbetween the smooth sound and the striated variant.The second section begins when other sounds be-gin to enter and the high sonority starts to fade out.This is a transitional passage, as five different enti-ties are introduced one after another, none of themdominating the sound-field. A rustling noise alter-nates with a brassy synthetic sonority, until, at17’13’’, this entity begins to layer a number oftracks upon itself, commanding most of the atten-tion. With the abrupt arrest of the brass voices, thefourth section begins with percussive sounds alone,

a mixture of the plucked mbira (African thumb pi-ano), the rattling ceramics, and the tsuzumi. At25’00’’, a rich electronic entity enters and graduallysaturates the texture. The pulsating, wave-like con-tours of this sonority strongly resemble the pound-ing undulations of the second section of Jonchaies,the orchestral score Xenakis completed around thesame time (Xenakis 1977). After close to 8 min,this entity begins to fade out, and the brassy sonor-ity takes over again. A number of other sounds en-ter too, and the brass is less dominant. A wildmetallic sonority, sounding like a cross betweenthe amplified and distorted braying of a donkey andan electric guitar, is prominent throughout thispassage. Up to that point, the general range of thesounds had been descending. The lowest transposi-tions of the brass and metallic entities carry themusic through to the final section, where they fadeout over some 3 min as the high tones of the open-ing enter and carry on to the end.

The pavilion that was designed to house thispiece, together with the light show that went alongwith it, was constructed from red vinyl stretchedover a metal frame. The design somewhat resem-bles the Philips Pavilion, although the point of de-parture was to create a form that would have the

Figure 11. Chart showingprogression of sonic out-lines and sectional divi-sions of La Legende d’Eer.

51Harley

maximum volume for the minimum surface ofouter shell. Rejecting the obvious solution—thesphere—as being acoustically and visually poor,Xenakis created a more complex form. As he put it,‘‘the effect of the architectural form has a quasi-tactile influence on the quality of the music orspectacle presented within it. This is beyond anyconsiderations of optimal acoustics or proportions’’(Xenakis 1978). There is an aesthetic character thatthe structure imposed on the performances withinit. In the case of Le Diatope, the point of departurewas the sphere, but with the design’s double curva-ture, its ‘‘flights,’’ it is an architecture ‘‘open to theworld.’’

Stage Four: UPIC

Throughout 1972–1978, during which Xenakis washeavily involved in overseeing the multifacetedproduction of Polytope de Cluny and Le Diatope,he was also working with his engineers atCEMAMu on the development of a novel approachto digital sound synthesis. The UPIC (Unite Poly-gogique Informatique de CEMAMu) is a computermusic system enabling the user to create sounds bymeans of a graphic interface (Lohner 1986). All theelements of the sound are designed with an electro-magnetic pen on a large electromagnetic drawingboard. (Recent versions have replaced the pen anddrawing board with a mouse.) These elements in-clude the waveforms, the dynamic envelopes of thesounds, and the ‘‘arcs,’’ or notes. Interesting soundscan be obtained by designing noisy waveforms andcomplex envelopes, and also by layering as many asseveral hundred notes (Marino et al. 1993). Xenakishad always sketched much of his music on graphpaper, so such an approach to sonic compositionwas perfectly natural. It has also proven to be anengaging medium for young children (and manyothers, of course) to approach creating sounds onthe computer.

Polytope de Mycenes; Mycenae Alpha

Mycenae Alpha, Xenakis’s first piece created onthe UPIC, was completed in the summer of 1978,not long after the opening of Le Diatope. It was,

like La Legende d’Eer, produced for a multimediaperformance, though also conceived to be an inde-pendent concert piece. The Polytope de Myceneswas an outdoor spectacle mounted at the historicsite of Mycenae in Greece. Similar in style to Per-sepolis, this event included torches on the moun-tainsides, searchlights crossing the sky, and alsoincorporated several of Xenakis’s ‘‘Greek’’ instru-mental and choral works. Mycenae Alpha, under10 min in length, served as an electronic interludebetween the other pieces.

The music is noisy and dense, made up primarilyof massive clusters designed so that the scorewould also be visually compelling (see Figure 12).The basic impetus is to move from complex tex-tures to more stable ones and back again (or to anew complex sound). Interspersed are moments ofmore focused, simpler sonorities. The main limita-tion of that first version of the UPIC was that itwas not possible to mix the different ‘‘pages’’ of thegraphic score. Xenakis’s piece, then, is a successionof different gestures, one following on the other. Inaddition, each of these ‘‘pages’’ could be no longerthan one minute in duration. Mycenae Alpha isbuilt from 12 such graphic/sonic entities, two be-ing repeated, to make a total of 14.

In this piece, a structural distinction can bedrawn between complex sonorities created bymeans of masses of relatively stable note-segments,and others created from dynamic arcs. It is easy todesign intricate glissandi on the UPIC merely bypicking up the electromagnetic pen and tracingthem onto the design board—held notes, in fact, aretreated as a sub-class of the more generalized ‘‘arc.’’The design of the waveforms (a hand-drawn wave-form is written into a look-up table) is also criticalto the sonorities created. The timbres tend to benoisy, but also static. Variation is achieved throughthe layering of the notes and the dynamic enve-lopes.

Mycenae Alpha is remarkable for demonstratingthe innovations of the UPIC system. Clearly,though, this piece is far more limited in its soniccontent than La Legende d’Eer. But, for the poly-tope at Mycenae, the harsh sonorities would havematched the savage magic of the landscape, so per-meated with ancient myths.


Pour la Paix

In 1981, Xenakis was commissioned by Radio-France to produce a radiophonic work for the PrixItalia (although apparently it was never entered inthe competition). In response, he put together arather strange, collage-like work for reciters, choir,and electronic sounds created on the UPIC. Thetexts are taken from two works by the composer’swife (a celebrated author in her own right), Franco-ise Xenakis: Ecoute and Les morts pleureront. Partof this same selection of texts was used in thework composed just prior to this, Nekuıa, for choirand orchestra.

Pour la Paix exists in four forms. The full versionlasts over 26 min, and includes all three elements:the spoken text, the sung parts, and the electronicfragments. There is an abridged version comprisingthe ten choral sequences on their own. The othertwo versions present the work in pre-recordedform, either completely on tape or with just the re-citers speaking their texts live. The spoken textsconstitute roughly half the work, there are ten sungpassages, and there are numerous sequences ofUPIC material.

The electronic sounds heard in Pour la Paixcover a much wider timbral range than those inMycenae Alpha. Some are cinematic, evoking thesounds of war, for example, in response to a pas-

sage of text. Others are more abstract, and the re-maining fragments are supportive, to accompanyeither the choir or the recited parts. The choir partsrange from straightforward chanting on a restrictedset of notes, as in the first sequence, to more com-plex sonorities—a compendium, really, of elementsfrom Serment, an a cappella choral work from ear-lier that same year.

In spite of the intensity of the texts, poetic rail-ings on war and death, Pour la Paix is rather disap-pointing as a radiophonic creation. The sequencesof material mostly succeed each other with littleoverlap, with the exception of the electronicsounds which sometimes appear in conjunctionboth with the spoken and sung parts. The produc-tion lacks sonic depth and spatial organization, apuzzling weakness considering the level of sophis-tication common in all kinds of broadcasts, not tomention Xenakis’s other electroacoustic works.This would be his only foray into the medium ofradio art, in any case.

Taurhiphanie

In 1987, Xenakis was given his last opportunity toproduce a multimedia event. The Festival de Mont-pellier commissioned him to create a work for theancient Roman arena in the Provencal town of

Figure 12. Page of graphicUPIC score from MycenaeAlpha.

53Harley

Arles. Bull fights were still held in that same arena,so the focus of the event became the incorporationof bulls into the proceedings (along with the fa-mous white stallions of the nearby Camargue re-gion). The plan was to wire the bulls with radiomicrophones to capture their snorting breaths.These sounds would then be processed in real-timeusing the new version of the UPIC system, whichwould be installed in a tower in the middle of thering. Spotlights were to project patterns of lightonto the floor of the arena and the animals wouldcreate dynamic stochastic patterns as they movedaround the ring. To top it all off, twelve percus-sionists were stationed around the upper seats ofthe round arena, and they performed selectionsfrom Xenakis’s percussion music (Idmen B, Pleıa-des, and Psappha).

In the end, a lack of time and resources forcedXenakis to drop the amplification of the bulls. In-stead, he incorporated brief samples from this un-usual source into a tape work created at CEMAMuon the UPIC. This piece, Taurhiphanie, was pro-jected in the arena along with some live manipula-tion of fragments of the sonic material using theUPIC. The animals did not cooperate in creatinginteresting stochastic patterns; they tended to hud-dle at one end of the ring or another, apparentlytraumatized by the pounding of the amplified per-cussion and the extreme dynamic range of the elec-tronic sounds. The performance was not anunqualified success.

Taurhiphanie, though, remains as perhaps Xe-nakis’s most successful piece created using theUPIC. At just under 11 min in duration, it isslightly more substantial than Mycenae Alpha,and, given the technological improvements of theUPIC, it is clearly conceived as an organic wholerather than a succession of parts, as in the earliertwo pieces.

The first 5 min form a continuous gesture, thelayers of sustained sounds gradually rising until ahigh point is reached at 4’22’’. At that point, thesounds begin to descend again, but are interruptedat 5’28’’. After a series of shorter, disruptive ges-tures, a sustained sonority takes over at 5’57’’ fea-turing some stable intervals—a ‘‘still point.’’ By6’45’’, fractures appear in the texture, with a vari-

ety of sonorities making brief entrances. The rangeof timbres is relatively wide for a synthesizedscore, though still not as rich as pre-UPIC workssuch as La Legende d’Eer. A break at 8’00’’ preparesfor a return to the bands of sliding sounds from theopening section, this time interrupted by dramaticinterjections of more concentrated, sweepingsounds that heighten the intensity of the passage.As the music nears its conclusion, the sustainedsounds begin to diverge, some migrating higher,some lower. The piece ends abruptly, as if cut off.Most of Xenakis’s electroacoustic works end insimilar fashion.

While there is a certain roughness to Taurhi-phanie that might indicate a lack of finesse in theshaping of certain details, there is also a strongsense of assurance and cohesion. As with the otherpieces that had their sources in multimedia crea-tions, this music stands on its own and carries on avitality that long-past spectacles cannot retainthrough mere documentation.

Voyage absolu des Unari vers Andromede

In 1989, Xenakis produced his most ambitiousUPIC work, a commission for the inauguration ofan International Exposition of Paper Kites in Japan.Voyage absolu des Unari vers Andromede extendsthe image of a kite soaring through the air to a cos-mic voyage through space in the direction of theAndromeda galaxy. While the music is not exces-sively programmatic, one may quite easily imaginetraveling through space while listening, with vari-ous ‘‘episodes’’ occurring along the way.

The formal outline of Voyage absolu is con-ceived on a broad scale, though the 15-min, 25-secduration is still nowhere near the breadth of Per-sepolis or La Legende d’Eer. The piece is structuredin two parts. The first, lasting until 8’40’’, is themore turbulent, with numerous sonic scrabblingsand short dramatic gestures surrounding more on-going layers of arching contours. One striking tex-ture occurs at 4’48’’, where a percussive noisepulsation briefly sets up a regular beat that mightgo over well at a techno club. The second part ismuch more serene, as the sustained sounds undu-


late more slowly. In this section, the added sonori-ties are less prominent in the mix, though there aremany discrete gestures heard in the background.The sonic arcs evolve both higher and lower, andthere are well-defined episodes where timbres anddensities clearly shift. The music at 10’38’’, for ex-ample, is quite sparse, though by 11’30’’ it starts tobuild up again. The final few minutes encompass agradual ascent into the high register, decoratedwith aliasing effects that produce low buzzingsounds.

Erod

By the time he reached the 1990s, Xenakis wasclearly not well. In addition, he was preoccupiedwith developing his stochastic synthesis algo-rithms. He did, however, return to the studio toproduce one last UPIC piece. Erod was created atLes Ateliers UPIC in 1997 for a commission fromthe Bath Festival in England. Xenakis workedclosely with Brigitte Robindore, and they evidentlyexperimented with deriving new sounds from re-cordings of earlier scores, both electroacoustic andinstrumental. The piece lasts less than 5 min, andit was withdrawn soon after its premiere. (The con-tribution of Ms. Robindore to the production of thepiece was so integral that Xenakis could not ingood faith put his name on it.)

Stage Five: Stochastic Synthesis

As early as the late 1950s and early 1960s, Xenakisspeculated about the possibility of synthesizingnew sonic waveforms on the same basis as his STalgorithm for stochastic composition. He carriedout some experiments to that end at Indiana Uni-versity, and continued at CEMAMu in Paris. Hisfirst results were applied to electronic sounds pro-duced for Polytope de Cluny and La Legende d’Eer.Until the mid 1980s, though, the main preoccupa-tion at his research center was the developmentand promotion of the UPIC system. In 1986, withthe loan of a minicomputer from Hewlett Packard,Xenakis was at last able to return to this project.

By 1991, his team had succeeded in developing acomputer program that would not only synthesizesounds stochastically, but would also implementmacro-compositional procedures similar to thosedefined for the algorithm implemented in 1962(Serra 1993).

On the synthesis level, the aim was to generate awaveform that could be varied continuously ac-cording to some stochastic function. In searchingfor the most efficient way to do this, Xenakis andhis assistants eventually adopted a waveform cycleas the basic unit (rather than, say, a sample), sim-plified as a series of line segments rather than acurve (see Figure 13). The degree of change, bothvertical and horizontal, from one point in a cycle tothe corresponding one in the next, is calculated onthe basis of a probability formula. The cycle is var-ied both in its amplitude (vertical extent) and fre-quency (horizontal extent). If each cycle of a sonicwaveform varies radically, the resulting sound willbe very unstable, or noisy. In order to obtain arange of timbres, a second-order process needed tobe set in place to control the boundaries, or degreeof possible change, for each operation. This processcould of course be dynamic, effecting gradual shiftsin the rates of change. On the higher level, pro-cesses were implemented in order to select thenumber of ‘‘voices’’ (waveforms) activated at anyone time, their points of entry in the time-line ofthe piece, and the duration of each of these seg-ments for each voice. With these basic tools it be-came possible to input control data so as togenerate an entire piece digitally on the basis ofstochastic functions.

GENDY301; GENDY3

Xenakis’s first compositional product from thisGENDYN (GENeral DYNamic stochastic synthe-sis) algorithm was GENDY301, presented at the1991 International Computer Music Conference inMontreal, Canada. A newly-generated work derivedfrom similar control data was produced later thatsame year for the Journees de Musique Contempo-raine in Metz, France. This work, titled GENDY3,is the version that has become known through sub-

55Harley

sequent performances and release on compact disc.In spite of the close relation of the two pieces interms of their genesis, they in fact sound com-pletely different.

GENDY301 is the shorter of the two, at 14 min15 sec in duration. It also exhibits a wider dynamicrange than the other better-known piece, with ex-tremely loud textures entering suddenly on top ofnarrower-range sonorities. In addition, it containsmore breaks of silence. Xenakis never stated hisdissatisfaction with the piece, but he may have de-cided not to release a ‘‘family’’ of works as he hadproduced in 1962 with the data from his ST pro-gram. (Those pieces include ST10 for mixed ensem-ble, ST/4 for string quartet, ST/48 for orchestra,Amorsima-Morsima for mixed quartet, Morsima-Amorsima for mixed ensemble [withdrawn], andAtrees, for mixed ensemble.)

At 18 min 45 sec in duration, GENDY3 is clearlymore substantial than the earlier version. Upon lis-tening, it is apparent that, as in the ST scores,clearly defined sections are integral to the algo-rithm. The piece is structured, from the listener’sperspective, as a succession of eleven short sec-tions, each lasting from one to two minutes, theoverall progression expanding these durationsslightly. The sections are distinguished by registralplacement and scope, and degree of instability ineither frequency or timbre. Each sound is relativelyconsistent in its settings throughout a section,though there are always mixtures of types, and theindividual layers are often laced with a great deal ofsilence, to avoid oversaturation.

What is surprising about GENDY3 is the degreeof harmonicity present. Every section containssome number of sustained voices, cutting in andout in stochastic fashion, and there are most oftenconsonant relationships between the stable pitches.In some sections, there are also less stable voices,either in terms of frequency (creating glissandi) ortimbre, resulting in noisy or rapidly fluctuating so-norities. There are some surprisingly lifelike tim-bres at certain points, including vocal-sounding andbrassy (though more like organ stops than realbrass instruments) sounds. In fact, there is a greatdeal of organ-like tone to the held sonorities. It isalso true that the unstable sonorities, because oftheir novelty, stand out from the other sounds inthe sections where they occur.

S.709

After producing the two GENDY works in 1991,Xenakis continued tweaking the GENDYN algo-rithm (Hoffmann 2000b). For a monograph concertat Radio-France in December 1994, a new piecewas generated. S.709 focuses more on unstable, dy-namic sonorities. These are produced by allowingthe vertical and horizontal points of the waveformsegments to vary more widely and rapidly. With aduration of just 7 min, this is a much shorter piecethan GENDY3, but the degree of sonic activity—the acoustical intensity—justifies the truncated ar-chitecture.

There are no clear sections in S.709 by which a

Figure 13. Segmentedwaveform cycle subjectto time and amplitudevariation.


formal outline can be perceived. Instead, the musicis highly fragmented. The harmonic, sustainedtones of GENDY3 are entirely absent. By contrast,each of what seem to be no more than three or four‘‘voices’’ are in constant fluctuation in terms ofpitch, timbre, or both. Amplitudes, too, are highlymodulated. While there are some consistencies—certain characteristic sonorities—each voice under-goes considerable transformation, both graduallyand in sudden shifts. The density of sonic informa-tion within each voice surely dictated the reduc-tion in the number of layers. In fact, for much ofthe time, there is only one layer sounding at atime.

Taken together, the GENDY pieces and S.709demonstrate the wide range of possibilities inher-ent in the General Dynamic Stochastic Synthesisalgorithm.

Conclusion

Once he developed a workable compositional algo-rithm in 1962, Xenakis integrated his creative pro-cesses and experimental approach with computertechnology throughout the remainder of his career.It is worth noting, though, that some of the proce-dures he developed prior to 1962 could also havebeen programmed, had Xenakis had access to com-puting facilities. He applied his stochastic (andother) functions to his instrumental and vocal mu-sic, and worked at CEMAMu to generate newsounds digitally. Toward the end of his career, heat last found a way to reconcile his higher-ordercompositional concerns with the generation of thesounds (waveforms) themselves. Even so, Xenakisdidn’t drop everything else in order to work withthese long-awaited tools. The acoustical richness oflive instruments and voices remained compelling,as the vitality of his late orchestral scores such asDammerschein (1994) or Ioolkos (1996) attests.

In the domain of electroacoustics, Xenakis’s mu-sic evolved a great deal, from origins in the mu-sique concrete style through more abstractnoise-based sonorities that were fashioned intocontinuously evolving waves of intensity, to the in-corporation of transformed instrumental sounds,

to, finally, digital synthesis. A detailed assessmentof his achievements in this domain in relation toother work in the field remains to be undertaken. Itis safe to say, however, that Iannis Xenakis fol-lowed a singular path, guided by his unique charac-ter as a Greek revolutionary, trained inengineering, influenced by architecture, self-taughtin music. The seeds he sowed, his works, and histechnical or conceptual innovations, have proven,and will continue to be, fruitful for succeeding gen-erations.

Acknowledgments

I am grateful to Radu Stan of Editions Salabert forsupplying me with archival recordings of the elec-troacoustic works of Xenakis otherwise unavailableat the time of my research. I am also grateful toBrigitte Robindore for detailing the production ofErod, Xenakis’s last computer piece, and for allow-ing me to hear it. Others who have been helpful invarious ways in my research include: Cornelia Co-lyer, Cort Lippe, Gerard Pape, Jean-Michel Raczin-ski, Curtis Roads, and Malcolm Smith. Thanks alsogo to Doug Keislar for his help in editing the arti-cle. I would also like to acknowledge the support ofIannis Xenakis himself, who permitted me to workwith the UPIC system at CEMAMu during 1985–1987, and who was generous in many other ways.

References

Boivin, J. 1995. La classe de Messiaen. Paris: ChristianBourgois.

Brody, J. M. 1970. ‘‘Iannis Xenakis Electro-acoustic Mu-sic.’’ Liner notes. Nonesuch Records H-71246.

Delalande, F. 1997. ‘‘Il faut etre constamment un immi-gre.’’ Entretiens avec Xenakis. Paris: Buchet/Chastel &INA/GRM.

Di Scipio, A. 1998. ‘‘Compositional Models in Xenakis’Electroacoustic Music.’’ Perspectives of New Music36(2):201–243.

Di Scipio, A. 2001. ‘‘Clarification of Xenakis: TheCybernetics of Stochastic Music.’’ In M. Solomos,ed. Presences of/Presences de Iannis Xenakis. Paris:Centre de documentation de la musique contempo-raine, pp. 71–84.

57Harley

Fleuret, M. 1972. ‘‘Une musique a voir.’’ L’Arc 51:32–35.

Fleuret, M. 1988. ‘‘Il teatro di Xenakis.’’ In E. Restagno,ed. Xenakis. Torino: Edizioni di Torino, pp. 159–187.

Fulchignoni, E. 1981. ‘‘Sur Orient-Occident.’’ In H. Ger-hards, ed. Regards sur Iannis Xenakis. Paris: Stock, pp.257–263.

Gabor, D. 1947. ‘‘Acoustical Quanta and the Theory ofHearing.’’ Nature 159(4044):591–594.

Harley, J. Forthcoming. The Music of Iannis Xenakis.London: Harwood Academic Publishers.

Harley, J. and M. A. Harley. 1997. ‘‘Triumphs of Moder-nity: Xenakis’ Kraanerg at Canada’s National ArtsCentre.’’ Musicworks 67(Winter):24–31.

Harley, M. A. 1998. ‘‘Music of Sound and Light: Xenakis’Polytopes.’’ Leonardo 31(1):55–65.

Hoffmann, P. 2000a. ‘‘Analysis through Resynthesis:Gendy3 by Iannis Xenakis.’’ In M. Solomos, ed. Pres-ences of/Presences de Iannis Xenakis. Paris: Centre dedocumentation de la musique contemporaine, pp. 185–194.

Hoffmann, P. 2000b. ‘‘The New GENDYN Program.’’Computer Music Journal 24(2):31–38.

Lohner, H. 1986. ‘‘Xenakis and the UPIC.’’ ComputerMusic Journal 10(4):42–47.

Marino, G., M.-H. Serra, and J.-M. Raczinski. 1993. ‘‘TheUPIC System: Origins and Innovations.’’ Perspectivesof New Music 31(1):258–269.

Matossian, N. 1981. ‘‘L’artisan de la nature.’’ In H. Ger-hards, ed. Regards sur Iannis Xenakis. Paris: Stock, pp.44–51.

Matossian, N. 1986. Xenakis. London: Kahn & Averill.

Roads, C. 2002. Microsound. Cambridge, Massachusetts:The MIT Press.

Robindore, B. 1996. ‘‘Eskhate Ereuna: Extending the Lim-its of Musical Thought—Comments On and By IannisXenakis.’’ Computer Music Journal 20(4):11–16.

Schaeffer, P. 1981. ‘‘Chroniques xenakiennes.’’ In H. Ger-hards, ed. Regards sur Iannis Xenakis.Paris: Stock, pp.79–88.

Serra, M.-H. 1993. ‘‘Stochastic Composition and Stochas-tic Timbre: GENDY3 by Iannis Xenakis.’’ Perspectivesof New Music 31(1):236–257.

Solomos, M. 1993. A propos des premieres oeuvres(1953–69) de I. Xenakis. Pour une approche historiquede l’emergence du phenomene du son. Doctoral disser-tation. Paris: Universite de Paris IV.

Toop, R. 1989. ‘‘The Music of Iannis Xenakis.’’ Linernotes. Etcetera Records KTC 1075.

Treib, M. 1996. Space Calculated in Seconds: The Phil-ips Pavilion, Le Corbusier, Edgard Varese. Princeton:Princeton University Press.

Varga, B. A. 1996. Conversations with Iannis Xenakis.London: Faber and Faber.

Xenakis, I. 1969. Letter to Wallace Russell, Administra-tor of the National Ballet of Canada, 6 March 1969.Toronto: Archives, National Ballet of Canada.

Xenakis, I. 1977. Foreword to Jonchaies. Paris: EditionsSalabert.

Xenakis, I. 1978. Le diatope: geste de lumiere et de son[Program Book]. Paris: Centre Georges Pompidou.

Xenakis, I. 1992. Formalized Music: Thought and Mathe-matics in Music, revised edition. Stuyvesant, NewYork: Pendragon Press.

Documents

Electroacoustic Music of Iannis Xenakis