You are on page 1of 99

Varese's 'Density 21.

5': A Study in Semiological Analysis


Author(s): Jean-Jacques Nattiez and Anna Barry
Source: Music Analysis, Vol. 1, No. 3 (Oct., 1982), pp. 243-340
Published by: Blackwell Publishing
Stable URL: http://www.jstor.org/stable/854178
Accessed: 21/05/2010 19:11

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=black.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Blackwell Publishing is collaborating with JSTOR to digitize, preserve and extend access to Music Analysis.

http://www.jstor.org
JEAN-JACQUESNATTIEZ

VARESE'S'DENSITY21.5': A STUDYIN
SEMIOLOGICAL ANALYSIS
Translated
by AnnaBarry

CONTENTS
Introduction 244

I - PART I (bs 1-23)


1. The firstfive bars. 248
2. Digressionon musicsemiologyandthe informationalapproach. 255
3. The progressionto high G (b. 17). 259
4. The zoneof B (bs 18-23). 270

II - PART II (bs 2X40)


1. The percussivesection(bs 2X28). 272
2. Verticalfallsandflights(bs 29-32). 273
3. The flightsof 'Density'. 276
4. Permutationsof B, F# andA (bs 29-32). 280
5. The end of PartII (bs 3g40). 282

III - PART III (bs 41-61)


1. Repriseof the opening(bs 41-43). 284
2. Permutationson B-D (bs 46-50). 285
3. The lastsegment(bs 51-61). 287
IV - RECAPITULATION 289
V- POIETICANALYSIS
1. The poieticproblem. 301
2. Melodicpoietics. 303
3. Harmonicpoietics. 303
VI - ESTHESICANALYSIS 319
VII - COMPARISONOF ANALYSES 329

(OMUSICANALYSIS1: 3, 1982 243


NATTIEZ
JEAN-JACQUES

INTRODUCTION
Music analysis, as we understandit in the semiologicalperspective of
attentionto minutedetailandclarification,does not lend itselfwell to exhaus-
tive presentation.
In my book on music semiology (1975), I was able to give only a few
examplesof my approach,and the small numberof semiologicallyinspired
analysespublishedin periodicalsis confined, for the most part, to parts of
works.l I am, therefore,particularlygratefulto JonathanDunsby for having
offeredto devotemanypagesof MusicAnalysisto the publication,in English,
of this analysisof 'Density 21.5'. It first appearedin a Frencheditionof 300
copiesin 1975, and is long since out of print. I havemadevariouschangesin
orderto correcterrors,to take into accountthe evolutionof my theoretical
ideasin relationto a text now eight yearsold, and to clarifymy positionon
some issues.
The uncomfortablesituationof analysistoday can doubtlessbe explained
by the difficulty experiencedin drawingup and publishingwritten music
analyses.Whenone venturesto reproachanalysesfor not comingto gripswith
the detailof a workand the multipleconstituentvariableswhichgo to makeit
up, one is often told that analysisprofessors,in their classes, can 'go a long
way' into a work. Could music analysisbe an oral genre, or even an oral
tradition?2It must face the followingproblem:no analysisis truly rigorous
unless written down (Granger),an epistemologicalelaborationof the adage
'Verbavolant,ssnptamanent',since the recordof the analysisenablesit to be
checked: once it is written down, it is possible to review, criticise and go
beyondan analysis.Even with a very elaborateoralanalysis,the listenerhas
the physical problem of being unable to retain everything. If the teacher
managesto give the impressionof having penetratedthe work deeply, the
listenerwill be left with a positive 'aura',but a cumulativeadvancementof
knowledgecannotbe developedon the basis of impressions.
The presentstudy thereforeaimsto urgemusicologistsinterestedin analy-
sis to takethe time to recordtheirresearchand offersthe first ratherdetailed
analysisof an entire work from a semiologicalperspective.I am gratefulto
David Lidov for having understoodthis: 'This long study is an important
complementto Fondements. . . It gives a much fullerpicturethan the latter
does of the scope and force of the author'smethods' (Lidov, 1977: 45).
Writtenanalysisenablesus to take in all parameters,not that an oralanalysis
cannotdo this, but it is extremelydifficultto masterthe combinationof all
parametersin the absenceof rules, tablesand diagrams.
This analysisis alsothe first to illustratelinksbetweenthe neutrallevel and
poieticandesthesicdimensions,thoughit in no wayclaimsto offerexhaustive
poieticand esthesicanalyses.It is not proposedto give a new presentationof
the perspectivefromwhich I am working:3it shouldsufficeto rememberthat
a neutrallevel is a descriptivelevel containingthe most exhaustiveinventory
possibleof all types of configurationsconceivablyrecognisablein a score.The
level is neutral because its object is to show neither the processes of

244 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

productionby which the workunfolds(poietics)nor the processesof percep-


tion (esthesics)to which it gives rise. In this sense it provisionallyneutralises
the poietic and esthesic dimensionsof the piece. On the other hand, the
neutrallevelprovidesthe units in relationto whichpoieticandesthesicdatain
SectionsV and VI will be examined.
Yet anotherjustificationof this neutralstatusis the use, fromthe beginning
to the end of the study, of the neutrallevel as an analyticaltool whichis never
called into question, the partitioningof the work into units accordingto
abstractparadigmaticaxes, that is, axes which group togetheridenticalor
equivalentunits from an explicitly stateslpoint of view. This techniqueis
inspiredby methodssuggestedby Ruwet(1972:Ch. 4), a continuationof the
teachingsof Jakobsonand Levi-Strauss,but it is not followed blindly: the
problemsit presentsare discussedelsewhere(Nattiez 1975: 239-356). The
readeris referredto this sametext for a completepresentationof the method-
ology used in the neutraldescriptionof 'Density 21.5'4.
The analysisproceeds'frombottomto top', thatis, fromthe smallestunits
to the largest, since Varese works with the differentiationof short units.
Nevertheless,largersectionsappearin the piece. As these are justifiedonly
lateron, I shallbeginby giving, withoutcomment,a pictureof the hierarchic
structureof 'Density'in so far as it results fromthe completeanalysis,so that
the reader can see how the minutiae which are to be examinedrelate to
broaderphenomena.
Numbersin squarebracketsabovethe stavereferto the smallestunits. Bar
numbersare unbracketed:

MUSIC ANALYSIS1: 3 , 1982 245


units j r X 10 r 1 f -, 1 1[ 5 ^

JEAN-JACQUES NATTIEZ

Parts 1st part


Sectl ons A
Sequences I II

246 MUSICANALYSIS1: 3, 1982


<:e-5
9 fmf >A f > [795
- _>f - S: b
< >Sgpv
1l[80] > X#r p t r3_ > #f
^,
43^
S81]s ' > > l z>
11
r3 . " t, F 6830<
6lf
w PS - -f ;=60
3 CC.moteo
41 Pgw 42 11

VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL


ANALYSIS

( 2 r]d pd r ' )

I u _ m
; [4i L45 169 L47q [480 993 L503 0510
O . .

9 34 4 ) t. 5;i.- t * ; : A

t1 .CLX , r
.,
(2n d pdrt )
D

loco J 60 0530 [540 [c25J 090 [5/ [580 559i


> 3 ;= 72

s Ir 41>1 ;7FitFJE--b-TJ1w;7l4
3rd part
I z

060] [61 ] [6 2, 06 3g

< ' -L;; m - j >1g: Wf' ----I


(3rd part)
B
I u m
8
[64] [65] 066] [670 [68g
-- > - --
L690 _ to co
L7d

4j f1t^1t71t
S5f S2f 5'51 >

JX pcJ?,l

(3 rd part )

Ea
[710 11 [7251 L730
1[740 [75] [76] [770 0780

; $ "t b
(3rd part)
(C)
I[b

ff
(r) ColfrancMusicPublishingCorporation-Canada.
By arrangement.
Agent:G. Ricordi& Co. (London)
Ltd.
MUSICANALYSIS1: 3,1982 247
f =P ' _ - -3 J? p

NATTIEZ
JEAN-JACQUES

I - PARTI (bs 1-23)


bars
1. Thefirstfive
Beforegoing into the work in detail, I shall illustratethe principlesof
segmentation of the musicalsyntagmaccordingto the variousparadigmatic
axesin an analysisof the openingof 'Density':
Ex.1
I [1] A[2ad B[2b]

;; 1 ; 2 $ aS j 2J $
3 , j | [4b], _3

rf , ' , '

>= p 9 ]w 3X, F ] ^ 3 j

First, the left of the example:bs 4 and 5 can be seen, in fact, in two
differentways(A andB), showingstraightawaythatthe neutrallevelis not
restrictedto onemannerof presentation, but on the contrarydisplaysthe
diverseconfigurations possible,thoughwithoutprofessionto unitethemall.
To avoidconfusion,the numbersin squarebracketscorrespond to the parti-
tionedunits;a letterfollowinga square-bracketed numberdesignatesthe
particularparadigmatic alternativewherethereareseveralpartitions.Roman
numeralsdesignateunits regroupedat a higherlevel; these are discussed
below.5
As faras pitchis concerned,the firstthreenotes(unit[1])arerepeatedat
theendof b.3 ([3]).It is temptingto addto thesethe F#-G-F# ([5])of b.4.
Theparadigmatic axisregroupsunitswhichareequivalent froma givenpoint
of view:thisdoesnotmeanthattheyarehomogeneous. For[1], [3]and[5], I
useMolino'sterm block,[bloc].Theevaluation of affinitywhichallowsus
to maketheseassociations, dependsupona mixtureof separatecriteria:
(1) The melodicidentity of [1] and[3] (F-E-F#)
(2) The rhythmicsimilarityof [1], [3] and[5]:
Ex.2 [1XfUlJuJ ,2

t3]n l J

. '
[5g.r:' JS

248 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

Theparadigmatic rhythmictheme,thatis, thetraitcommonto a group


of units,consistshereof two shortvaluesplusone long.6
(3) Butthis similarityis reinforced by twootherfactors:
(a) the two shortvaluesforma kindof mordent(lowerin [1] and[3],
upperin [5]);
(b) the longvaluesarein everycasean F#.
(4) Unit [5] beginson the samenotethatends[1] and[5].
Allthisleadsus to 'neglect'thedifferencebetweenthetwosemiquavers ([1]
and [3]) and the two tripletsemiquavers([5]). The distrtbution of these
elementspermitsthe connectionof the initialF's andF# of the threeunits,
althoughtheyaredifferent.
Thesetwoexamplesof assimilation, to whichcanbe addedthelengthof the
finalF's, illustratewhatmaybe calledequivalence classes.Withinthis class
thereis a wholerangeof relationships: becausetheyarephysicallyclose,it is
easierto assimilatethe semiquaver andtripletsemiquaver thanthe crotchet,
and the compoundvaluequavertied to dottedminimtied to tripletsemi-
quaverunderthe samecategory,'long',sincethe feelingof lengthby oppo-
sitionto the two shortvaluesdoesnot preventperceptionof the durational
differencesbetweenthe three(progressively shorter)finalnotes.
Thereis anotherreasonformakinga paradigmatic associationof Ex.1:from
a widerdistributional pointof view,takingthe broadercontextintoaccount,
[1], [3] and[5] initiatethreelargersegments,I, II andIII.7
Thethreeremaining units[2], [4]and[6]canbe organised in twoways.In
versionA of Ex.1, the criterionis essentiallyrhythmic:

Ex3

[2ag zP XJ) 4 J S
3 1 3 & t

[4ad SJ 3
:S 3
J &
t I |

[64 ;. j J J
3 1 3 } 3

fP

Despitedifferences hasthepatternlong-short-long8
in detail,thisparadigm as
a commontheme.Thereare,however,two anomaliesto be considered: the
inclusionof a semibrevein [2a]andthe two equalvalues(tripletcrotchets)
whichend [6a].
The semibreveis all the more importantbecauseit forms part of a
procedure whichis especiallycommonin Varese:theconstantlengthening of
eachnew note withinone musicalsegment.9Onephenomenon justifiesthe

MUSIC ANALYSIS1: 3, 1982 249


NATTIEZ
JEAN-JACQUES

paradigmatic isolationof the semibreve:the slur fromthe initialF to the


secondC of b. 2. Thedynamics p-f-p alsogivetheC conspicuous autonomy.
Obviously,strictlyspeakingonecannotclaimthesecondtripletcrotchetof
[6a]to be longerthanthe note immediately precedingit, sincethey areof
identicalduration.The totalparadigm,the blockwhichtakesall parameters
intoaccount,willin factneutralise thisanomalywithregardto theshort-long-
shortpatternthroughthe identicalfinalpitchesof [4] and [6]:C#-G. This
meansthat, in makingthe paradigm,anothervariable(pitchidentity)is
dominant
hierarchically in relationto the rhythmicvariable.
ForthisreasonlayoutB of Ex. 1 maybe morepertinent:thisparadigm is
basedon the identityof the finalnotesC#-G in all threeunits,not justtwo.
Thusnewrelationships beginto appear:the C# in particular, initialonlyin
[2b]andcentralin [2b],[4b]and[6b],playstheroleof a pivotnote.Becauseit
alwaysprecedestheG, whichendsnotonlythethreeunitswearedealingwith
here but also segmentsI, II and III, the F# of [2a] is not unconnected,
paradigmatically, with the F# of [1] and delaysthe arrivalon G. These
melodicrelationships areobviousin the paradigm in Ex. 4:

Ex.4
['

< #2>-'J 3>

[2S_Xss

1;gL \ semitone

1 j J k
[3] [4a]

03 [5,0 3

[6] , 3 }

zJ w1

250 MUSICANALYSIS1: 3, 1982


ANALYSIS
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL

It is no exaggerationevento callthe F# a kindof leadingnoteto G whose


syntagmatic positionanddurationmakeit a 'polarnote'.In so farasthistable
opensa separate paradigmatic axisforeachnewnote,its configuration canbe
calledoblique,establishedbetweenlines 1 and 3 by the additionof G, one
semitoneaboveF#.(Thisdetailwill assumea certainimportance lateron.)
Thefinalnotesof [1], [3]and[6]haveprogressively shorterrhythmicvalues
andthe sameis truehere(semibreve,crotchet,tripletcrotchet).The initial
notesof [2b], [4b]and [6b], however,becomeprogressively longer(triplet:
dottedquaver,quavertied to crotchetanddottedcrotchettied to crotchet
respectively).
LayoutB shouldnotsupersede layoutA. Justasit is perfectlylegitimate to
emphasise therepetitionof C#-G whichconcludessegmentsI, II andIII, we
mustalsotakeintoaccountthecaesurabetweenG of b.2 andallthatprecedes
it becauseof the slur.Wehave,therefore,to takeaccountof a contradiction
betweenthe two paradigmatic choices.The neutrallevel shows, by its
positioningin eachparametric framesuccessively, thatthemelodic,thenthe
rhythmicorganisation bothhavetheirownlogic;thiscanbe seenonlywhen
the othervariablesareprovisionally neutralised. By puttingtogetherall the
information tabulated,theworkingsof a principlewhichappearsto becharac-
teristicof thispiecel°is revealed:theprinciple of deception.
Thewholeof thisopeningpassagewillnowbe re-examined withtheaccent
on the syntagmatic progression: one unit [1] of threenoteswitha chromatic
rise(F-F#Sends on a longF#;thearrivalon G is delayedby a secondgroup
of threenotes(C#-F#-C#) whichis connectedto whatprecedesit by a
slur.A restfollows.The initialunit reappears, witha slightlyshorterfinal
note, but this time goes directlyto G which,in [4], framesthe C#; here
though,with the samerhythmictype (long-short-long), it is the C# which
framesthe F# ([2]). In the samebreath,the motivefrom[1] returnsin the
formof [5] describedabove,herefollowedneitherby C# norG, butby an E
([6])whichprecedesthefinalC#-Gmarkeddiminuendo. Thesethreenotes,at
distributionally equivalent points,outlinea diminished fifthchord(C#-E-G,
suggestedby the C#-G fromb.2 onwards,andcompletedby the E of b.5 in
exlremis). Thismeansof delayingthe G in b.2, withan intermediary unitof
threenotes,representsthe principleof deception.The privilegeddistribu-
tionalpositionof C#-Gin [2], [4]and[6]strengthens thelistener'simpression
of havingbeendupedin [2]. Naturally,the listenerdoesnotconsciously and
discursivelyperceivethe workexactlyaccordingto the processdescribed
here,buta detaileddescription of the neutrallevelcansubsequently be used
to describea phenomenon whichisfunctionally pertinentfromanesthesicpoint
of view;whenwe speakof deception,we are,afterall, describingan effect
uponsomeone.
Attentionmustbe drawnto a featureof the methodusedup to thispoint:
the researchprocessconsistsof isolatingunitsaccordingto criteriaof para-
digmaticassociation. But a differentpointof viewhasnowbeenadopted
that of syntagmaticsuccession.Havingbegun with a relativelyabstract

MUSICANALYSIS1: 3, 1982 251


JEAN-JACQUES NATTIEZ

description of unitsregrouped in absentia to use Saussure's expression


theyarethenprojectedontothe realaxiswheretheylinkup.
Note thatthe descriptionof the syntagmatic progression suggestedabove
usesinformation andsegmentation established froma paradigmatic pointof
view.Somemusicanalyseshavealreadyusedparadigmatic presentation,1l but
in introducing to musicology theconceptsof syntagm andparadigm, structural
linguisticsandsemiologyoffera thematisation of thisdistinction.A systematic
searchfor the paradigms of relatedunitseffectsregroupings andclassifica-
tions, thus advancingtowardsa typologicalknowledgeof worksand their
constituentelements;the thematisation of the paradigmatic principlewill
requireus to identifya categoryof upward-surging figures('flights':see,
belowof II-2);the thematisation of syntagmatic organisation bringsus to
studytheirdistribution.In previouslypublishedanalyses,12 we emphasised
paradigmatic decomposition of worksbecausethisaspectwasnotconsidered
systematically in classicalmusicology.In the presentstudy, we wish to
concentrate on syntagmatic progression, butto takeas a basistheinfonnation
yieldedbytheparadigmatic viewpoint.Theorderof discoveryandof exposition
doesnot necessarily coincide:in thisparagraph we havestressedthe method
used(projection of the resultsof paradigmatic decomposition ontothe syn-
tagmaticaxis);fromnowon we shallfollowmorecloselythe orderin which
the unitsappear.It is simplya matterof remembering that paradigmatics
enabledthemto be identified.
Theparadigmatic andsyntagmatic description of thenumbered unitsin the
scoreof 'Density'doesnotcompletetheanalysis.In note7 threereasonswere
givenforregrouping thefirstsix unitsintolargersegments,labelledI, II and
III. Onemightaskwhy, bothinitiallyandthroughout this study,the small
unitsappearto be usedto establishthe largerones,whilstonehearingof the
piecepermitsimmediateidentification of the largestsections.
Thistextcouldhavebegunby justifying, in broadterms,thedivisionof the
pieceintothreeparts,thedivisionof PartI intotwosections-A andB, the
divisionof A into threesegments-I, II andIII, the divisionof I intotwo
units-[1] and[2], etc. Butthe procedure 'frombottomto top'is preferred
becausethehierarchically moreimportant unitsarenotidentifiedaccording to
the criteriain use for classicalandromanticmusic:repetitionof themes,of
longphrasesandof periods.Vareseplayson subtlerhythmicdifferentiations
and avoidsstrictmelodicrepetitions;it is thesedistinctionswith whicha
scrupulous analysisshoulddealfirst.As Lidovrightlysays,'Varesehasleftall
the a prioriimplicational relationsof musicaltonalitybehind.The unique
systemof the workis its onlysystem.Insteadof the tensionbetweenstyle
(abstract)andexample(concrete),the worktakesits life andits energyfrom
the complexityand ambiguityof its internallydevelopedassociations and
contrasts.The taxonomyrendersthese explicit'(Lidov 1977:4445). An
emphasised note, a generalmelodicconfiguration, a rhythmicor intervallic
contrastbetweentwo passageswill definea largesection.This is why it is
easierto understandthe largeunits whenone knowsfromwhichsmaller
phenomena theyareconstructed.

252 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

At the end of this study the divergencesin decision-making among


musicologists regardingformalorganisation will be examined,but what,in
fact, is a form?To speak,for example,of the formABAis to recognisea
certainparadigmatic familiaritybetweentwo segments,separated by a third
whichis considered to constitutea separateparadigm.In analysis,the para-
digmaticprecedesthe syntagmatic. If ABAis inflectedas ABA',it is because
the paradigmatic linksbetweenA andA ' areconsideredto be looserthan
thosebetweenA and A. In classicformalanalysisthe use of two identical
symbolshasneversignifiedstrictidentityof A andA, butrathertheneutralis-
ationof differences considered to be negligible.In a worklike'Density',the
identificationof partsor sectionsgreaterthanthesmallunitsis possible,even
obvious,butonlystartingfroma smallnumberof variables whicharehierar-
chicallydominantin relationto others.If formalanalysesof 'Density'differ
fromoneresearcher to another(seeSectionVII),thisis becausethedominant
variablesarenot necessarily the sameforeveryone:here,theformis not the
resultof commonpracticeamongcomposers in thecourseof a givenperiodof
musicalhistory,butis the consequence of a relativeandmobileconvergence
of paradigmatic equivalencecriteriabetweensections,for listenersand
analystsalike.Thesedivergences areneitherdramaticnorregrettable, pro-
videdthatwe areconsciousof the reasonsbehindthem.
The regrouping of the firstsix unitsinto threesegmentswas justifiedby
threecriteria(seenote7). Othervariablesexplainwhythesethreesegments
eachhavea certainautonomy,but alsoa familyresemblance whichdisting-
uishesthemfromwhatfollowsandenablesus to look,to beginwith,at the
firstsix barsalone:
(1) SegmentI presentsF-E-F#-C#-G in succession.SegmentsII andIII
use the samenotes,butin a differentorder.
(2) Thereis a certainanalogyin the distribution of intervalsin the three
segments:
I [1] ld 2a 5a
[2] 5a 5d 6a (2d)
II [3] ld 2a la
[4] 6d 6a (ld)
III [5] la ld 2d
[6] 3d 6a (6d)
Thenumeralsdesignatethe numberof semitonescontainedin theinterval
(no distinctionis madebetweenan augmented secondanda minorthird):a
and d mean'ascending'and 'descending'; bracketedintervalsmark'joins'
betweensegments.TheintervalbetweenG andF in b.3 is effectivelyneutral-
ised by the rest, but the joiningintervalbetween[4] and [5] has greater
weight,sincetheslurcoversthewholeof segmentsII andIII.Thecriterion of
a resthas not beenusedto identify[4] and [5]. A problemsuchas thatof
joiningintervalsshowsclearlyhow the weightof eachvariabledependson
manydifferentfactorswhich, themselves,changeaccordingto particular

MUSICANALYSIS1: 3, 1982 253


JEAN-JACQUESNATTIEZ

contexts.Nevertheless,a constant canbe drawnfromtheexamination


of these
intervallicsyntagms:everysegmentbeginswitha semitoneandendswitha
tritone,in accordancewitha seriesof continuousaugmentation:
I 1 2 5 6
II 1 2 1 6 6
III 1 1 1 2 3 6
(3) The studyof intervallicdirectionsdisclosesanothersyntagmatic
prin-
ciple:
I 1X,/(\) 2/\/ (X)
II 3\ /(/) 4N/(<)
III 5/<(X) 6h/(X)
Thisabstractionof intervalsjoining[2] to [3], [3] to [4] andthe descending
movementof [5] and [6] (G-Ft-E-Ct), showsa certainpredilection in the
piecefor systematicalternationof ascendinganddescendingmovement.
(4) Evenif segmentsII andIII arelinkedby thelargeslurof bs 3-5, each
one of theseunitsis distinguishedby secondaryslursin [3] and [5]
whichisolate[4] and[6] by virtueof theirdifference.
(5) Rhythmicequivalence classeswerediscussedabove.
Followinganexamination
of thewholepiece,thefollowingrhythmictypology
is proposed:
al : shortpluslong
a2 : two regularshortsplusone long
a3 : one (ortwo)short(s)plusone longplusone short
b : constantaugmentation
c : one longplusone short
d : one longplusone shortplusone long
e : regularrhythm.
Theseseventypescan be groupedinto threefamilieson the basisof initial
short,longor constantrhythm.
Fromthis viewpoint,the homogeneityof the threesegmentsis perfect:
I [1] a2
[2] d + long(ortripletquaver+ b)
II [3] a2
[4] d
III [5] a2
[6] 'd'(byassimilation)

254 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

2. Digressiononmusicsemiology andtheinformational approach'3


The analysisof the openingof 'Density'suggestsseveralconclusions
concerningthe epistemological statusof the neutrallevel. When Ruwet
imagined 'a machine for identifying elementary units'(1972:112),he pro-
videda metaphor whichevokestheexplicitcharacter of themethodproposed,
but which could be said to imply, ratherdangerously, thattheseprocedures
areof analgorithmic character; thefactthatanalysiscanbegin'fromthetop',
and requiresconstantzig-zaggingbetween'small'and 'largeunits',shows
quitewellthatexplicitis not synonymous withalgorithmic or mechanical.
The analysisof the firstfive barscombinedtwo steps:the firstgesture,
intuitivein thatit impliesa confusedcollectionof criteria,consistsof defining
whatMolinocallsblocks.It soonappearsthatin noneof themarethecriteria
of paradigmatic associationhomogeneous.A study of the 'partitionings'
(decoupages), no longercarriedout by block,but parameter by parameter
(pitches,thenrhythms,then intervalsetc.) showsthatno singleparameter
dominatesthe constitution of the blocks,whicharetheresultof a confluxof
variableswhoseweightchangesin eachnewcontext.
Asa comparison, consideranapproach of theinformationaltype.Atfirst,it
wouldnecessarily startat thebottomandproceednoteby note.A scanwould
identifyall identicalunits,takingthe notestwo, thenthree,thenfourat a
time,andso on. At thelevelof pitch,themachinelooksfirstforalltheF- E's
in thepieceandfindsthemin b.3. Then,afterscanningthe sixty-onebarsof
the work,it startsagainwith E-F: whichit findsin bs 3 and54 etc. The
procedurefor unitscomprisingthreenotesis simplersincewe canstatethe
followingrule:if unitA (twonotes)is followedby unitB (twonotes)andthe
last note of A is the first note of B, A + B constitutesa three-noteunit,
providedthatthe samesuccessionof notesis foundelsewhere.Thus, there
appearto be twenty-onethree-noteunitsin 'Density',forexampleF-E-Ft,
bs 1-3,G-Ct-G, bs 4 and 5-6. Eventhen, in the lattercaseit shouldbe
recognibed thatthemachinedoesnottakerestsintoaccountsincetheunitof
bs 5-6 containsa tripletquaverrest.In thisway,we cango up to thelongest
unit in the piece. It containsnine notes:B-Ft-A-F#-B-A-B-F:-A (bs
32-33, repeatedin b.34).
Thisoperationis repeatedon the otherdimensionsof the piece:intervals,
rhythm,dynamics,modesof attack,andslurs.
Note that the graphicrepresentation of unitsdefinedin this way is not
differentfrom the system of paradigmatic notationproposedby Ruwet,
thoughthetablesarenecessarily wider and shorter.Thetableof unitsformed
by notes takentwo by two comprises, syntagmatically, one hundredand
thirty-sixunitsof whichfifty-one are repeated.The intervalinventorybegins:

MUSICANALYSIS1: 3, 1982 255


JEAN-JACQUESNA1TIEZ

(1 = semitone,d = descending,
a = ascending)
ld 2a 5d 5a
5d 6a 2d
ld 2a la
6a
ld la
ld 2d 3d
6a 6d
6a
2a la 3d 3a
3d 3a
2a la
ld la
ld la
etc.

This table clearlyshows a patternof units of 2, 3, etc. intervals:for


example,ld-2a (bs 1 and3), 2a-la (bs 6 and8)
Such 'accounting'cannotbe calledanalysis,but is rathera 'physical'
inventory.Giventhatnotallpossibleformsof transformaton areforeseeable,
as soon as relationships
are establishedbetweenunitsthat are not strictly
identicalwe enterthe realmsof analysis,but it must be recognizedthat
culturaland theoreticalknowledge,a priorisand auralimpressionsaffect
* u

c beclslons.
The differencebetweenan inventoryandactualanalysisis thatit doesnot
appearto bepossibletodeducethelatterfromthesumof theinformation provided
bytheformer.In fact,partitioning carriedoutnoteby noteandparameter by
parameter doespresentproblems:l4
(1) A certainnumberof variables is reduced,fromthestart,to thestatusof
hapax,thatis, theyarenot attachable to othervariables,remainingisolated
andunusablein the inventory.Rhythmicvaluesarea casein point:withthe
exceptionof initialnotesof 'phrases',characterisedby twosemiquavers (bs 1,
3, 9, 15,21, 41, 43),thereareveryfewstrictrepetitions in thepiece.Fromthe
thirdnoteof the piece,the valuesof the F: in bs 1-2 andthe F: of bs 3- 4
mustbe madeequivalent in orderto obtainaninteresting result-a departure
fromthe data.The algorithmic procedure lendstoomuchweightto the note
as the minimalpertinentunit. As Molinowrites,it 'is an "amalgam" of
heterogeneous characterisations:
it indicatesoneabsolutepitch,virtualinter-
vals,degreesand function,and virtualdurationswhichcould,potentially,
carryrhythms.Thisis whyanisolatednotecouldneverconstitutea unit:its
mostimportant properties(intervals,degreesandfunctions,rhythms)remain
virtualuntilat leasta secondnote is joinedto it' (1975:55). The example
showsclearlyhowphenomena pertinentfor analysisarepresentat the 'top'

256 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

but not at the 'bottom':the equivalencewhich the proceduredemonstrates


betweenphysicallydifferentvaluesis one suchcase, andit is hardto see how a
computercould automatically establishan equivalencewhich depends on a
judgement of similaritytranscendingconcreteresemblancesand differences.
(2) Even amongrepeatedunits, the inventoryrevealsphysicallyidentical
phenomenawhich do not have the same significance.For example:the unit
G-F of b.3 would be picked up by the machinebecauseof its occurrencein
b.43 (G-Et: we must allowenharmonicnotesto be codedin identicalfashion
which is justifiablein this music). The sameis true of the unit G-F# which
appearstwicein b.4 andreappearsin bs 42 and43. In view of the restsin both
these examples, and also of the distance between and diversity of their
contexts,these 'repetitions'do not appear to be of any use. Transpositionis a
typicalcase. The inventorydealingonly with pitch will pick up the repetition
of F-E-F: in bs l and 3, but will establishno relationbetweenthis and the
E-D:-F fromb. 15. The intervalinventory,on the otherhand, will pick out
ld- 2a in bs 1, 3 and 15, but withoutbeingable to show, sincepitch does not
come into it, that the ld-2a of b. 15 is a transpositionof the othertwo. Then
the rhythmicelementmust be added:the units takenfrombs 1, 3 and 15 have
in commonwith the firstthreenotesof b.9 the rhythmtwo semiquaversplus a
long. The intervallic unit is not identical: Db-C-Db = ld- la, but the
connectionof the units throughthe intermediaryof the rhythmgives la a
presumedequivalencewith 2a. We see, then, thatthe interest of a recurrenceis
not independentof its context:it dependson its insertioninto a block,thatis,
a moreor less homogeneousgroupconstructedby the analyston the basis of
one or severalcriteria,dominant and convergent,thatdo not constituteall the
criteriawhich couldhave been broughtinto the analysis.Until we haveproof
that the criteriaare unsuitable,they justifythe analyticalchoices;this is why
they must be renderedexplicit. For this reason, Molino calls them quasi-
crzterza.
(3) Finally,the note by note inventorydoes not permitthe identificationof
phenomenawhichnormalmusicalcompetenceisolatesat once. We see in the
paradigmatictablesthat [4] introducesa G wherebeforetherewas a Ct, and
laterwe shallhavean E. The importancegivento thesenotespresupposesthat
all precedingmaterial(i.e. the rhythmicfigure,and the analogiesbetween[1],
[3] and [5]) and whatfollows(Ct-G) has been analysed.Does this meanthat
inventoriesof materialareuseless?GillesNaud (1979)has proposeda method
which, on a single table, 'reports'recurrences,parameterby parameter,
variableby variable, together withinformationdrawn from the completed
analysis.There is, therefore,no limit to the numberof possiblecolumns.l5
Therefollowsan exampleof what can be obtainedwith this method, applied
to the beginningof the piece:

MUSIC ANALYSIS1: 3, 1982 257


JEAN-JACQUES
NATTIEZ

x = semiquaver
y- triplet quaver
z- crotchet

Ido notintendto undertake a laborious


inventoryof units:it is possible,in a
specificstudy, to take shortcuts. Froma methodological point of view,
however, thecomparison suggestedby Naudto whichwecanreferforlimited
verification
focusesourattentionon threepoints:
(1) It tells us, by comparison of physicaldataand analyticaldecisions,

258 MUSIC
ANALYSIS1: 3, 1982
ANALYSIS
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL

which variablesare strategic (Molino's term). For example, the unit G-F
which joins the first two segmentsof Ex. 1, does not have the samevalue as
the F-E or the E-F# of the opening.G-F assumesthe statusof a joiningunit.
All the joiningunitsof the piecemightbe studiedfor certaincommontraits-
an interval,perhaps,which is characteristicin relationto all the other inter-
vals of the work. These comparisonsshould permit the identificationof
contexts and situationswhich give certainvariablesmore weightthan others.
(2) Secondly, Naud's inventorytable is an aide-memoire, an instrument
which confrontsus with phenomenathat the 'from the top' procedureof
paradigmaticanalysismight leave out. Once it is discoveredthat F-E-F#,
with its characteristicrhythm,is an important'thematic'elementof the piece,
the intervaltable can be scannedto see if ld-2a is found elsewhere.This
successionappearsin bs 10-11, in the trill of b.20 and in b.39, and makesit
possibleto show how these two characteristicintervalsforma developmental
thread,and at what privilegedpoints(beforethe returnof the real 'theme'at
bs 15, 21 and 41).
(3) Finally, the table in columnsshows up, on one page, conflictingseg-
mentations,the identificationof whichwouldotherwiserequirecomparisonof
paradigmatictables, often spreadover severalpages. Here, a conflict arises
betweenthe melodic segmentationwhich isolatesC#-G and the slur which
isolatesG. The partitioningfinallyadopteddoesnot excludethe other,sinceit
has been shown that it could be pertinentfrom anotherpoint of view.

to highG (b. 17)


3. Theprogression
Concerningthe first five bars in toto.What, in fact, is the beginning of
'Density'? Looking ahead as far as b. 17, we can call it a melody which
remainsfaithfulto romanticgestures,characterisedby a crescendo. The cres-
cendois, of course,dynamic(thefofb.3 andb.8, thettofb.9, thetftofb.11,
the crescendo of bs 13-14 and 16-17), but exists also in a more metaphorical
sense: the progressionof rhythms and intervals, and the melodic ascent
which, at b.5, has only just taken off. A 'romantic'ascent, then, but one
which constantlyfrustratesexpectationsconditionedby tonal dynamics:the
samephenomenonis foundin b.8, as in b.2, then betweenbs 8 and 9, and in
b.16.
The functionof sectionA (bs 1-5) is to introduce,afterthe motivewhich
winds aroundF#, the tritone C$G three times. Momentumis generated
C#-G reappearsin b.6 and the upsurgeof the melodyends only in b. 17 with
the high G. For this new section, B, Ex. 5 shows the organizationof the
melodicprogression(bs S8):

MUSICANALYSIS1: 3, 1982 259


3

NATTIEZ
JEAN-JACQUES

^ [63 3 j
Ex.5 ,

3
v ton e
p sud iho

-J br r ( b*

[6]is reintroduced asa memorandum. InthesamewaythatC$-Gof [7]


Here, [7]. If in b.8 therewere
isretake
a of [6], [8] containsthe G-Bb takenfrom
a clearsluroverthe Bb andC which dividesthemfromthe preceding
not oncemorebeforegoingup to
onemightsaythatVareserepeatsG-Bb C#-G-A-Bb of [7], picked
notes,
But this is a newdeception:the progression
C. breath-mark afterthe longBb, is
by the slur,the precedingrestandthe
out
but onlyin respectto pitch,as G-Bb-C-Db:
repeated
Ex.6

whichconnectsit to C, andtheC,
ButtheBbis isolatedfromtheG bythesluris separatedfromDb by duration
to distinctivedynamicemphasis,
subjected themomentum: themove-
its arrival,andby a breathwhicharrests
delaying / G-Bb-C-Db. There not be
will
mentC#-G / C#-G-A-Bb delaysG-Bb
symmetry.
perfect by a rhythmicdescription.Using
Themelodicanalysismustbe completed
thetypologypresentedabove,we obtain:
[7] b
[8] c or e
al
[9]

quaverof b.8 maybe consideredto


Note thatc or e for [8] indicatesthattheof the precedingtripletcrotchets.In
havea durationperceptivelyequalto that

MUSICANALYSIS1: 3, 1982
260
'\ ^ p1 3] k r 3 ,,

VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

any case, [8] is framedby two units whoserhythmis characterised by


prolongation.It is alreadyapparentthat the rhythmicconfigurations are
organiseddifferentlyfromthoseof sectionA. In b.6, somethingquitenew
begins.
Theparadigmatic profileof Ex.4 wasobliqueonlybecauseGwasaddedon
therightof theF: axis.Thisprofileis primarilyvertical(andlaterin Ex. 8 its
obliquecharacter is moremarked).Fromb.9 to b.11.it returnsto vertical:D
is the onlynoteoutsidethe playof permutations betweenDb andC whichis
spreadovertwo bars:
Ex.7
melody rhythm

ka r E°]g _f

3 t11]

7mf
fi r
subsoo r 5 6
mf
brr

Subito

H ,\, E2] H
t semitonc

¢ Stf t
3 ' 3

: tf r

[10]mustbe paradigmatically with[1], [3]and[5]:it borrowstheir


associated
rhythmictype a2 (two semiquavers plus a dottedminim, as againsttwo
semiquavers plusa triple-dotted minimin [1])andtakesthe formof a lower
ratherthanan uppermordentas in [1]. In so faras [10]doesnotcontainthe
intervallicsuccessionld-2a whichwouldallowthe introduction of Ft, one
semitonehigherthanF, theintervallic patternld-la allowsit, in contrast,to
stayput. Thus, [10]andits prolongations ([11]and [12])constitutea tran-
sitionbefore[13] whichintroducesD, underlinedby a crescendo andtriple
forte,according to a rhythmic (a2)andintervallic configuration
(ld---la) pre-
ciselyas in [1]. The pattern thus
'short-short-long' definesan equivalence
class,justlike the developmental procedurewhichconsistsof upwardsemi-
tonalprogression. Db appearsthusto be a kindof preparation for the D in
b.ll. In a sense, the play of permutations on two notes contributes to the
principleof deception:it delaysthe appearance of a predictable event,the
ascentto D.16

MUSIC ANALYSIS1: 3, 1982


261
JEAN-JACQUESNATTIEZ

A studyof the rhythmicparadigm(Ex. 7 right)showshowVareseintro-


ducessomevarietyintoa melodicsequencewhichcouldhavebeenmonoto-
and the
nous. [10], distinctivebecauseof its long finalnote, its crescendo
semiquaver restwhichseparates it fromwhatfollows,belongsto therhythmic
typea2as we noted;[11]takesthe lasttwonotesof [10],justas [9] didwith
[8]; typea1(short-long) reinforcesthe parallelismbetweenthe Db's of [10]
and[11],a2beingsimplyanextensionleftof a1(ora1a transformation of a2by
elisionof theinitialnote);[12],themelodicinversionof [10],belongsto type
b (tripletquaverplusquaverplusquavertiedto tripletquaver);[13]begins
withthesametwonotesas [10]andhasthesamerhythmictype(a2),butends
on D. The crescendo whichbeginson Db and endsfff contributesto the
emphasisof thisunit.Thereis biasin thesegmentation of [11]and[12].They
areseparated neitherbybreathing norslur.Thefollowingmightbeproposed:

Ex8
[11] 012]

1 F
3
b;

3 3 z3

v tr- r

Thefirstunitof thisexamplewouldbecomethemelodicinversionof [10]and


the secondwouldborrowits firsttwonotes.The lengthof Db in [11]makes
thissegmentation difficultto accept,especiallysinceits repetitionandlength
in-[10]giveit decisiveimportance. Theneutralanalysis,however,mustadmit
the configuration of Ex. 8 if only to foreseecertainperformers' choiceof
phrasing.
Thefollowingtable(Ex. 9) forbs 11and12is bothmelodicandrhythmic.
It is melodicin showingthetreatment of G#-D: G#in theloweroctavein [15]
thenD in the loweroctavein [16]followedby an expansionto the rightof
A-D#, semitonesrespectively aboveG#-D; in theupperoctave,A, BbandE
too willbe a semitoneabovethe notesof [16].Rhythmically, we findtypesa
andb: moreprecisely,a1in [14]and[15],prolongedby b in [16]and[17].
The rhythmicmomentumfollowsthatof the melodicprogression.
In [14]the D waswrittenin brackets.It wouldbe possibleto imaginea
partitioningbasednot on the paradigmatic headingG:-D buton D-GS, as
shownby the right-hand paradigm.I amreluctantto see it in thisway,not
becausetheD mightbelongto twounits(weshallseein b. 36thatthisis nota

262 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

Ex .9

P r5r C : >
s
E5] 3 , 3

Sai r
st s. . st

L<?e' 42$m
. , ,

54 '
9 I tt
.- (st= semitone)

problem),butbecausethef on theG#, validforthewholeof b. 12, separates


clearlyby contrast[14]from[13]wherethetfton D of b. 11is thegoalof the
crescendobeginningat the endof b. 10. In addition,oursegmentationthrows
into reliefa rhythmicprocedurewhichseemsto be usedfrequentlyin this
piece.Butin the samespirit,it wouldbe possibleto partition[15]and[16]in
the followingway:
Ex.10

Rhythmically typea3is followedby typeb, andmelodically a successionof


threeoccurrences of the ascendingtritoneappears,prolonging the join[13]-
[14] and the two descendingtritonesof [14], but emphasising the parallel
movementof theunits.Dependinguponwhetheronechoosesto lendweight
to thevariable'rhythmin augmentation' or on theotherhandto 'importance
of tritones',the boundaries between[14], [15]and[16]mayvary.
Musicanalysisis, like the worksof whichit attemptsto give account,a
symbolicphenomenon, sinceit is the resultof humanactivity(andtherefore
hasits ownpoietics);it leavesa 'trace'(thetextof theanalysis)andis subject
to reading,interpretation anddiscussion(theesthesicpole).Thus,the direc-
tionof attentionon one aspectof the workratherthananothermodifiesthe

MUSICANALYSIS1: 3, 1982 263


JEAN-JACQUES NATTIEZ

organisation of things.Whatis appropriate to the neutrallevelis to makean


inventoryof allanalytical possibilities
andI shallshow,as faras possible,the
poieticor esthesicpertinenceof someof these.
[17]wasconstructed from[16],usingthe familiarprocedure of raisingby
onesemitone.UnlikeEx. 9, the intervalsarenotidentical,eventhoughboth
unitsdo endwitha tritone,butin bothcasestherhythmformsa continuous
progression (typeb: tripletsemiquaver, tripletquaver,quaver,dottedquaver
on theonehand,andquaver,quaver,dottedquaver,semibreve on theother)
andthe successionof directionsis the same( \ / / ).
Concerning the roleof semitonesandimportance of tritones:as forsection
A, it is interestingto lookat intervallic
behaviour betweenunits[7]and[17].
First,we shallgrouptheseunitsintolargersegments:
B I [71 6a 2a la (3d)
II [8] 3a 3d (3a)
[9] 2a (la)
III [10] la la (ld)
[11] la (ld)
[12] la ld (la)
[13] ld 2a (6a)
IV [14] 6d (6d)
[1S] 6a (6a)
[16] Sd + 6a 6a (6a)l7
V[17] 12d l+a 3a (12d)
UnlikesectionA segmentsI andII of B showa tendencyto diminution: 62
1 3 3 2. Thenin the 'permutation'zone(segmentIII)Vareseworksonlywith
semitones.The tone ending[13] is a step towardssegmentIV whichis
dominatedby tritones(four occurrencesbeforeG of [16], then after a
descending leapof a compoundperfectfourththreetritonesleadingto A of
[17]).Comparing [13] with previousintervallicsequences,we can see how
Varesevariesthe pathsfromthe semitoneto the tritone:
A I: 1 2 S 6
II: 1 2 1 6
III: 1 2 3 6
B [13]: 1 2 6 (theinversionof 6 2 1 in [7
Thetritoneis a characteristic interval:it splitsthe temperedscaleinto two
equalpartsandis notwithoutanalogyto thesemitone,whichdividesthescale
intotwelveequalparts.18 Thiscontributes to thetautnessof thepiece.As we
cansee, the intervalsare distributedin privilegedzones:the tritonesend
segments I, II andIII of A, anddominatesectionIV of B, afterthe semitone
zone.SegmentV combinesthese two features:the initialA of [17] is a
semitoneaboveGS, a tritoneaboveD: ; the intervalbetweenA andBb is a

264 MUSICANALYSIS1: 3, 1982


9 , 0 st s

VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

semitoneand the last interval(Bb-E) a tritone.(Thereis one intervalmissing


from the first thirteenbars and almost entirelyabsent from the rest of the
piece, the majorthird.19This is probablybecauseit is a particularlyconsonant
interval,and 'Density 21.5' is articulatedby seconds,tritonesand sevenths.)
In [17] (segment V) there are intervalsnot heard before: a descending
octave, a compound semitone which is followed once again by a tritone
ending.With the compoundfourthof [16]thesearethe largestintervalsof this
opening passageand they widen the ambit (startingfrom the semitone)in
which the melodic curve can flower. But the progression,in evidencesince
[7], is not over. In b. 15, we find the melodico-rhythmicfigureof the opening:
[4] is transposedup a seventh(inversionby semitone),but remindersof the
openingarenot confinedto [18]. [19]takesthe notesof [18]in a gesturewhich
is not withoutanalogyto the passagefrom [1] to [2]: E-D#-F / D#-F here,
andF-E / F# / C#-F# before.[20]and [21]containthe samenotesre-ordered
as the first few bars:E-F-F#-G.
Finally,as regardsintervals,[18] to [21] showthe sametendencynoted for
segment A and between [10] and [16], that is, the continuousbroadening,
after the celebrationof tritonesin [4] to [17], in tones, semitonesand their
compoundrevisions,which does not pass throughthe intermediaryintervals
3, 5 and 6:
[18] 1 2 2
[19] 2 2
[20] 1 2 1+ (1)
[21] 1++

SectionB ends on the same note as sectionA, afterwhich the rise to high G
began(b. 7). The intervalof [21] is a doublecompoundsemitonebetweenF#
and G, the two predominantnotes of [1] and [2]. It wouldbe an exaggeration
to speakof a coda, but the composeris roundingoff andsummingup, andnot
only throughallusionsto the very openingof the piece. Hereis the paradigm
of bs 15-16:
Ex11 i0180 S 4<

E9]e

kOa] 4;

-; A

MUSICANALYSIS1: 3, 1982 265


P *c

NATTIEZ
JEAN-JACQUES

It is not unrelatedto the precedingones. ComparingEx. 14andEx. 10, we see


that [19] takesthe last two notes of [18], just as [11] took those of [10]. Even
rhythmically,the articulationof the two notes from [19] afterthe threenotes
of [18] resembleswhat happenedin [11]. [20a] is relativelycomplex: the
E, D# and F of [18] arepermuted(in the sameway that the Db and C of [10]
werepermutedin [12]);the F from [19] is transformedinto an F#, just as [4]
introduceda G afterF#, [9] a C afterBb, [13]a D afterDb, [16]a D# afterD,
and [17] an A, a Bb and an E after G#, A and D#. This last comparison
enablesus to drawtogether[16]-[17] and [20a]-21a].

Ex.12 5601 +

Xg 4tS J) #r

[,5 f f
520ag 3 § t21b]H

7 $N, r LS

directions
Justas in [16] a D# was introducedaftera D with the characteristic
/ , so in [20a] Vareseintroducesan E# after E, with the same directions
/ X . Even if these intervalsare not the same (ascendingtritone and major
second, decendingcompoundperfectfourthand compoundminor second),
the leapwhichleadsrespectivelyto D# in [16]and to E# beforethe returning
upward sweep (D#-A-D# / E#-F#-G) in [20a] is some justificationfor
drawingthese musicalsegmentstogether.Finally,the culminatingG in b. 17
has the samesemitonerelation(evenin octavedisplacement)to F#, as the E#
has to E, the Bb to A, and the A to G#.
It will be notedthatthe implicitpartitioningin the threesegmentsof Ex. 12
does not correspondto [16] and [20a]-[21a]. Recall that whilst in Ex. 9
rhythmicfactors,indispensablefor the analysisof bs 11-14, were taken into
account,in Ex. 11 the accentwas placedon melodicrelationships.The break
between[20a]and [2la] cannotbe retainedbecauseof the phrasing,dynamics
and breathing.It should thereforebe written:

266 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

Ex.13

D 20]<2= '
[21] -
L
,y C

Of course,this segmentationoffersnothing,paradigmatically, in respectof


[18]and[19],preciselybecausethe F# andthe G arenew.It is thissegmen-
tation,however,whichis retainedas the slursandthe breathingimposea
partitioningwhereit is only the melodicrelationships that are no longer
readable.Thismeansthat,onceagairl,thejoiningof Ex. 11 andEx. 13on the
neutrallevel throwsinto reliefa new formof the principleof deception.
Followingtheparadigmatic analogyof Ex. 15the E# of [20]shouldgo to the
F# of [21]in thesamemotionwhichtakesD# to A ([18])orA to Bb([17]).In
fact,thisnaturalmomentum is arrestedbytheexpansion of thephrasingafter
thefirstF# andthe breathbetweenE# andF#, in themiddleof a crescendo.
SectionB is autonomousbecauseit displays,in its ownright,traitswhichA
doesnot possess.The rhythmictypesin the firstthreeunitsof B havebeen
shownto be differentfromthoseof A. Let us look at B fromthis pointof
view:

B I [7] b
II [8] c or e
[9] al
III [10] a2
[1 1] al
[12] b
[13] a2
IV [14] a
[15] a
[16] b
V [17] b
VI [18] a2
[19] c
[20] al a
[21] a1
Exceptin [8]and[19],typesa andb, thosewhichgo fromtheshortestto the
longestduration,predominate.Type d, presentin A, is peculiarto the
beginningsinceit doesnotappearhereat all. Its returnfrom[22]to [26]will
be allthemoresignificant.Theomnipresence of a andb impliesthatallthese

MUSIC ANALYSIS1: 3, 1982 267


E. P L

JEAN-JACQUES NATTIEZ

units([7]-[21])end witha longervaluethanthosealreadyheardwithinthe


sameunit.Thisis alsothecasewithsegments.Onecannothelpbutestablisha
correlationbetweenthis tendencyto lengthendurational
valuesin eachunit
and each segment,and the melodicprogression whichis the objectof B.
ConsiderEx. 14:
Ex.14a

I + J >
nf f mff w f

3- , 3 '

m dF3
# w-
1)1
##
tWu L

(I)74/-

gJ b
p s6izo f
I r

m UfT
*g--

mfs>Eto 3 f*g
Ex.14b

EH-e
f cL"-'

-
^

fP w - _,ff9c -

268 MUSIC ANALYSIS 1: 3, 1982


bo } bo I

VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

Thenewnotesfromeachunitactuallyforman ascendingstep-wiseprogres-
sionin relationto thosepreceding:
Ex.1 5

;> o .. bo w bo zqoWo ^s o ct,,)#O o

Canwe speakof tonalityhere?A scale,firstminor(G, A, Bb, C, D) then


chromatic(Dt, E, F, F,t, G) unfolds,but not everyaddednote is felt as
beingin thekeyof G:on the contrary,at everytransient pointthetonalityis
uncertain sincethenotesin a unitor segment lean towards a polarnotewhich
is differenteachtimeandwhich,itself,leans towards another notethatwill
predominate subsequently.Eachtransient point of thesefirstseventeen bars
delineatesa moment whichestablishes itselfin a preciseambit and is exceeded
onlyby successivetonesor semitones:

Ex.16
[1g-t2 [2,

@' o bo t.. o I
[1] [6] [72
o | U bo
t80 t9]
O bo l s
[lo]-[120 [13]

014]-E5] E60 [17]


n bt -
o I to I I

[18]-[20] t2og
[2<

o t° $° 14" 1$°' I

Thereis onenotemissingfromtheopening,B20,left,soto speak,in reserve


for bs 1W23(sectionC) whereit evidentlyassumesthe role of a pole of
attraction.

MUSICANALYSIS1 : 3, 1982
269
sJ -

JEAN-JACQUES
NATTIEZ

4. Thezoneof B (bs18-23, [227-[287)


The zoneof B maintainsa numberof linkswithwhatgoesbefore:
Ex.17

v l /

[2i

p tubito
>02603

I
r27g
[24Xff ft
--L*S-
p wubito
pt
Lookingfirstat the left-handparadigmatic axis, we see thatmelodically
[22]is a lowermordentlike[10],butrhythmically it belongsto typed of [2].
[23]transforms[22]byloweringA: onesemitone,thefamiliar procedure, but
operatingdownwards this time. G: is added,on the right,alsoa semitone
lower.Contraryto whathas happenedup to b. 17, herethe tendencyis to
descent.
In [24]thereis the samemelodiccontouras in bs 1S17:
Ex.18
,Q sl to 4;.w ("v) n

[24,

o ts. to °

270 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

It shouldalsobe notedthat,whereasin [12]-[13]thereis a slowtrillendinga


semitonehigher(D), in [24] thereis a rapidtrill whichfollowsthe same
intervallicpattern(descending semitone,ascendingtone),the patternof [1]
whichthe phrasingof bs 19-20(anotheruseof deception)is not supposedto
hide.[24]is a kindof responseto [22]-[23]:theselastdescendchromatically
towardsG: withfournotes(B-A-B-G:); [24]climbsup to D in inversion
(B-Ct-B:-D): only B: altersthe symmetry.But this note is important,
heraldingthe B$tof [28].[24],whoseanalogyto [1]hasbeenmentioned,is a
sortof transitionto [25]whichclearlyborrowsits rhythm(or thatof [10]).
SectionC is thus attachedto sectionA by two procedures: in [22] by the
rhythmof [2], in [25] by the rhythmof [1], andin [24]by the intervallic
patternof [1]. [26] plays on the A alreadyintroducedwhich occupies,
distributionally,
theplaceof B andtakestherhythmof [2]and[22].[27]uses
onlythefirsttwonotesof [26].Since[25]wehavenotlefttheambitA-B, just
as [22]-[23]establishthemselvesbetweenG: andB. [28]willplay,anoctave
higher,on the C: andB: introducedby [24].The appoggiatura is a sortof
inversionof thelowermordentof [22]and[25],andsectionC endsa semitone
abovethe polarB: a transitionis assured.
Thissectionis divisibleintotwo'moments',eachof whichmayin turnbe
dividedinto two:
C Ia [22] 23
Ib [24]
IIa [25] 26 27
IIb [28]
Ia-Ib andIIa-IIbarein factsymmetrical: Ia andIIa showa tendencyto
descentor to stasisand botharefollowedby an abruptchangeof register.
Theyarecharacterised by tonesandsemitones([22]:1 1; [23]:2 2 3; [25]:1 1
2; [26]:1 1 2; [27]:1). Thetwosegmentsendby leapsof a tone,displacedup
anoctave,in thesamewaythattherisesof [17]and[21]closesegmentsV and
VI of B.
Mostmusicologists21 agreethatb. 23is theendof thefirstpartof thepiece.
Actually,fromb. 24 onwardsVareseusescompletely differentcompositional
procedures. WeseesectionC moreasa transition betweenwhatprecedesand
whatfollows:becausein b.17 the climaxon thehighG completeda progres-
sioninstigatedat the beginning;becausesectionC is theprivilegedzoneof a
notewhichwe hadnot heard;becausethe alternation fall/rise-stagnation/rise
doesnotshowa clearpictureasin thefirsttwosections theimportant notes
of thefoursegmentsfirstoutlinethechordB-G:-D, thenthegroupB-At-
B: whichstayssuspendedbut is identicalto the intervallic patternof [1].

MUSIC ANALYSIS1: 3, 1982 271


JEAN-JACQUESNATTIEZ

II PARTII (bs 24-40)


PartII comprisessectionA (bs24-28)whichis characterisedby percussive
useof theflutekeys,sectionB (bs29-32), sectionC (bs32-36)andsectionD
(bs 36-40). The thirdpart(III)beginswitha returnto the motiffrom[1].

section(bs24-28)
1. Thepercussive
Thesegmentation rests:
of thissectionis obviousbecauseof thenumerous
Ex.19
[29], 3 ,

[3] + + X

.
[312 + [3i

#S
, p
77S7'?P

t3i3 + +

[34] + +

nn [3i +3 +'

03i +

,b:

Varesedwellsfirston E-Ct. Can[29]be drawnfromtheE-C: of [6]?It is


temptingto notethatbs 2X28 arecastin theambitCt-(E)-D whichis that
of the openingof the piece (bs 1-11) until the introduction by a play of
tritones(b. 12) of a new modeof progression.In [31], Varese invertsthe
E-Ct. This last note, prolongedby the D of [32], becomes the pivotnote
([33]- [34])followedby D ([35]- [37]). The passage ends on an Eb, heard
hereforthe firsttime,andstands above D as a compound minor second.
This whole sectionis, then, autonomous. The first unit ([29]) moves
withouta break to the last ([37]) by a play of successive additionsand
suppressions.In contrast to the firsttwenty-three bars,accented and dotted

272 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

notesareusedsystematically. The regularrhythmtypeis predominant and


appearsfivetimes.Therestsarelongerandmorefrequentthanin preceding
sections.In PartI, sub-sections A andB areseparated by a rest:thereis a
tripletcrotchetrest between[6] and [7], [19] and [20]; semiquaverrest
between[10]and [11];quaverrestbetween[23]and [24], [24]and [25]. A
dottedcrotchetrest divides[17] and [18], beforethe returnof the initial
motive,andPartI endson a minimrest.In PartI, theselongrestsenjoya
privilegeddistributionalposition:theyprecedethereturnof theinitialmotive
([3] and [18]in the two versionsclosestto [1] rhythmshort-short-long;
descendingsemitone,ascendingtone).
In this we find:a minimrestbetween[29]and [30]and(approximately)
between[32]and[33],a dottedcrotchetrestbetween[33],[34],[35]and[36]
(to thenearesttripletsemiquaver), a crotchetrestbetween[30]and[31],[31]
and[32],[35]and[36],[37]and[38],anda quaverrestbetween[36]and[37]:
longrestspredominate.
Finally,thisis theonlymomentof thepiecewhereVareseusesthepercus-
siveeffectof the fingerson the flutekeys.Becausethiswasa newuseof the
instrument,these five barshave attractedmost commentand sufficedto
inscribethe piecein musichistory:it is reallysincethenthatcertainpurely
technicalpropertiesof instruments havebeenusedto musicalends.

2. Verticalfalls
andflights
(bs29-32)
With the tempochangein b. 29 and the descendingleapswhichthe
composerusessystematically
forthefirsttimehere,something
differentagain
begins:
Ex.20

t38,

[39]29

_[40Xt $+

MUSICANALYSIS1: 3, 1982 273


NATTIEZ
JEAN-JACQUES

[420 3 9
[43> [43in ;-/st t

--43 A; J4T'lQl

[432 t br^t

Theseleapsarein fact similarto thoseof [16](G#-D#), [17](A-A) and


[20](F#-E#), butthisfigureis combinedforthefirsttimewiththecharacter-
isticrhythmictypea2from[1](short-short-long). Thedifference betweenthe
longF#'s of [38]and[40]is onlya semiquaver, andthis allowsus to group
themin an equivalence class.
In [38] to [40], we are struckby the playof minorand majorseconds,
simple,compoundor invertedG-F#, F#-E#, E#-G (at the join),G-F#,
F#-G,G-F#. Throughthissystematic preservationof smallintervals,Varese
makesa specialuse of clashesof seconds.
[42]and[43a]presenta patternreminiscent of the firstbarsof the work:

Ex.21
bs 1-2

Ws' O tsZ t (# " ) O


[42] [43ai

o ("-) ^ O bo

Here,theintervallic sequenceis identical:anascendingtone,a descending


perfectfourth,an ascendingdiminished fifth.The combination of intervallic
directionswith theircontinuousbroadening makeseachnew note, on the
ascending or descendingslopes,higherorlowerthanthepreceding Thisone.
principleof motivicdevelopment is foundelsewhere,evenif theintervalsare
notidentical.In Ex. 22 the twoinstancesfromEx. 18arereinserted to show
the link betweenthesefourfragmentswhichreliesupon the alternationof
directionsandthe playof simpleor compoundsemitones, even though there
is no intervalidentity:

274 MUSICANALYSIS1: 3, 1982


nb I

VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL


ANALYSIS

EX.22 4b 12 S,} (o) - ZX

n to ) StZ

bs 16-17

.s go

(°} $0-----,<.

bs 20-21

O ta. t0 0

b.30

i,,, to n
o - - - -_ _ _ _X..

The thirdsectionof PartII is dividedintotwo'moments':first,a seriesof


descents,then, after[41]whichrisesandfalls,the progression towardsthe
highA of b. 32. Ex 23 showshowthe ascentto A is effected:

Ex.23 4 /

. 3 [42] 3
.

[4i, ''s

[43io 4e

[43@ f

This is not as 'slow'or regularas the approachto high G: everything


happensin a baranda half.Buttheendof [42]introduces a G#andanA not
containedin [41],the secondnoteof [43a],Bb, is a semitoneabovethe A of
[42],thesecondnoteof [43b]is a semitoneabovetheBbof [43a],andthefinal
A is a semitoneabovethe G# of [43b].The semitoneis, once again,fun-
damental to thissection:from[38]to [40]in thefirstsegmentof C, it frames
theoctaveleaps;in [41]and[42]it slipsin betweenthewideintervals (seventh
andthirteenth)andfollowsthem,but betweenall theseintervals(10 11 12
1+) thedifference is alwaysa semitone.Ontherhythmiclevel[43a]and[43b]
areof typea1and[43]as a wholeis of typeb. All of sectionC, then,is filled

MUSICANALYSIS1: 3, 1982 275


JEAN-JACQUES
NATTIEZ

withprogressive
rhythmictypes:typea from[38]to [42]andtypeb in [43].

3. TheflightsofDensity
I shallnowbreakaway,exceptionally, fromtheprinciplefollowedfromthe
beginningof thisanalysisby depariingfromthelinearunfoldingof thepiece.
[43]is compared withanalogous phenomena alreadyencountered orto come.
I havegiventhemthe genericname'flight':

Ex.24
bS 12-13

Theyarenot all of one type:therearebreathsin bs 13 and 16 withslurs


straightafterwards in bs 13, 31, 58. But one couldsay thatthey articulate
threetypesof 'crescendo':a non-metaphorical crescendo (< fff), augmentation
of rhythmicvalues(bs 12-13:quaver,dottedcrotchet,thenquaver,dotted
crotchet,semibreve;b. 16: crotchet,crotchet,semibreve;bs 31-32: semi-
quaver,dotted quaver,crotchet,dotted crotchet,dottedcrotchet;b.44:
quaver,dottedcrotchet,dottedminim;bs 5W61:tripletcrotchet,triplet
crotchet,crotchet,then quaver,crotchet)dotted crotchetand crotchet,
minim,semibreve),andfinallyan unrelenting risein pitch(towardsE in b.
13,Gin b. 21, A in b. 32, C: in b. 44, B in b. 60).TheseflightsaredifElcult
to
describefroma melodicpointof view:in thefivesegmentsthereis a feelingof
functionalsimilarity,buthowis thisto be madeexplicit?Thereis no obvious
regularity
in the intervalseries.22 All thatcanbe observedis a certainprefer-
encefor'taut'anddissonantintervals,butif welookat thewholepiece,there
is nothingspecialin that.

MUSICANALYSIS1: 3, 1982
276
b.44 s _ 1+R
bs 58-61 \ 1+R;

VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL


ANALYSIS

Theconstructional
principleof theflightsmustbe soughtelsewhere,in the
roleplayedby semitonesin the appearance of newnotes:

Ex.25
bS t2-13 bs 31-32

t b. 1 6
O b. R2

, .

R2, ' | { _
! bo .
* 7, R

b- bo
{b o \
. tone \ \ \t+22 R

ts' ' . '


0 jR2
'R R2 _ '
80 #s'. q

(R signifiesoctavedisplacement
andR2 double-octave
displacement)

But the flightsdo not haveonlyparadigmatic analogies.Syntagmatically,


theyhavea characteristic distribution.
First,andmostobviously,theyendan
important section(in b. 13, segmentB = IV of PartI; in b. 17, PartI; in b.
32, sectionB of PartII; in b. 45, sectionA of PartIII; in b. 60, the entire
piece).Equally,we see thatthepositioning of theechappees canbe studiedin
relationto anothermelodictypealreadyencountered - permutations. There
followsa list of the two typesconcerned:
Flights Permutations

[10] - [13]
[14] - [16]
[17] [18]
[21] - [20]
[25] - [27]
[43] [44] - [51]
[54] - [56]
[59] - [62]
[63] [64] _ [70]
[71] _ [74]
[79] - [81]
[82] [83]

MUSICANALYSIS1: 3, 1982 277


NATTIEZ
JEAN-JACQUES

Infourcasesoutof five([17],[21],[63],[82]-[83])theflightsareprecededbya
permutation: in threecasesoutof five([7], [43],[63]) exceptthatsincethe
lastechappeeendsthe pieceit is literallythreecasesout of four theyare
followedby a permutation (thus,two 'flights'areframedby permutations,
[17]and[63]).Evidently,permutations constituteoneaspectof thecontextof
one
flights,justasflightsconstitute aspectof thecontextof permutations: this
is an important melocticsyntagmaticelement of 'Density'.
Finally)flightshavea thirddistributional on the left this
characteristic,
time:in fourcasesout of five, one of the two unitscontains a descending
compoundinterval:
(1) beforetheflightof [17],descending compound fourthat the beginning
of [16];
(2) beforethe flightof [21],descendingcompoundsemitoneat the endof
[20];
(3) beforethe flightof [43],descendingcompoundsemitoneat the endof
[41];
(4) beforethe flightof [82]and[83],descendingcompoundmajorsecond
at the beginningof [81].
Returning fora momentto thecontentof theflights,werecallthatthefirst
two ([17] and [21]) containan ascendingcompoundsemitonein one- and
two-octave displacement andwe aretemptedto see if intervals
respectively,
greaterthan an octavehave someprivilegedrelationship to flights.Every
instanceof descending compoundintervalsnotin oneof thetwounitspreced-
ing an echappeeis a join([24]-[25]:compoundminorthird;[28]-[29]:com-
poundaugmented fourth;[32]-[33]and[33]-[34]:compound semitone;[69]-
[70]: majorthirdwith doubleoctavedisplacement; [74]-[75]:compound
minorsixth).Withascendingcompoundintervalsoutsidethe flights,seven
casesoutof tenarealsooverjoins([27]-[28], [38]-[39],[80]-[81]:compound
majorsecond;[36]-[37],[39]-[40],[63]-[64]:compoundsemitone;[70]-[71]:
compoundperfectfourth).[24]and [77]containa compoundmajorsecond
and [33] has a compoundsemitone.The inventorywill be affectedby the
followingobservations:
(1) Everyinstanceof a compoundsemitone,ascendingand descending,
from[33]to [37],is foundin a sectionof PartII wherebecauseof the
of unitsanabsolute
restsit is difficultto givetheproposedpartitioning
value.Should[31]and[32](b. 25)constituteoneortwounits,and[35],
[36]and[37](b. 28) one, twoor threeunits? We mighthesitate.It is
thereforeinteresting,whetheror notjoinsareinvolved,to examinethe
importance of compoundsemitonesin sectionB.
(2) Thesametypeof problemis posedby thejoin[27]-[28],since[28]is to
[27]as the C::-B: trillin b. 20 wasto theprecedingB. Moreover,the
intervalis thesame.Thisassimilation is allthemoreappropriatein that
[24] and [27]-[28]are not withoutanalogyto the flights:leap to a

MUSICANALYSIS1: 3, 1982
278
a

VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL


ANALYSIS

muchhighernote, crescendo, contextof permutation. It will laterbe


shownthatunit[77]couldhavebeenconsidered anechappee wereit not
'afflicted'with a diminuendo. If, therefore,'assimilatedflights'seems
appropriate, it is interestingto note that [24], [27]-[28]and [77] all
containa compoundmajorsecond.Allothercases- exceptthedebat-
able exampleof [36]-[37]from sectionA of PartII - are clearly
positionedat joins ([38]-[39],[39]-[40],[63]-[64],[70]-[71],[80]-
[81])
(3) Oncetwoof thesixdescending intervals
havebeenlocatedin sectionA,
the otherfour([24]-[25],[28]-[29],[69]-[70],[74]-[75])areclearlyat
olns.

On the basis of these observationsand distinctions,a table of the


distribution
of compoundintervalsin the piececanbe formulated:
Ascendingintervals Descendingintervals

Beforea Tritone [17]


flight Semitone [20]
Semitone [41]
Majorsecond [81]

Withina Semitone [17]


flight Semitonein
double-octave
displacement [21]

In an Tone [20]
assimilated Tone [27]- [28]
flight Tone [77]

PartII, Semitone [33] Semitone [32]- [33]


SectionA Semitone [36]- [37] Semitone [33]- [34]

Joins Semitone [39]- [40] Minorthird [24]- [25]


[63]- [64] Tritone
Augmentedfifth [69]- [70]
Tone [38]- [39] [28]- [29]
[80]- [81] Majorsixth
[74]- [75]
Perfect [70]- [71]
Fourth

MUSIC ANALYSIS1: 3, 1982 279


t
JEAN-JACQUES
NATTIEZ

It was[43]whichmotivated thisdigression.It is thethirdof thefiveflights


and,in a way,it is central.The firstflightendson a highE, the secondon a
highG (b. 17),thehighestnoteup to thatpoint.G thendisappears foreleven
bars,andreappears as highG in b. 29. It is usurpedin b. 31 by A. The last
flightconcludesthe pieceon a highB, the privilegednotefromsectionC of
PartI, absentfromthefirsttwosectionsof PartII. G, A andB maintaintheir
privilegedrelationshipin termsof immediatecontext,of registerand of
distribution.
Thethirdsectionof PartII is in fact,a permutation zonewhere
the intermediaryF# insertsitselfbetweenB andA.

4. Permutations of B, F# andA (bs32-36)


A does not play a role comparable to thatof G. Hereit closessegment
[44]-[45]witha longnote,butin [51]it is B thatclosesthe secondsegment
withthesamevalue(adottedminim).Thezoneof A is, in fact,intermediary:
the restof the piecepreparesthe climaxof the workon highB.
Aftera briefpassagein a slowertempo(crotchet= 60, bs 29-32)theinitial
temporeturnsin [44].TheB-F#-A adopts,in thesameregister,thelasttwo
notesof the flight, fillingthem out, and, as GillesNaud pointsout, the
rhythmof the beginningof [44]is relatedto thatof [43]:
[43] = semiquaver,dottedquaver,crotchet,
[44] = tripletsemiquaver,
tripletdottedquaver,tripletcrotchet
In [44]thereis a tripletandin [43]the G# is a crotchet,butthe articulation
pointof eachnoteis, in bothcases,proportionally comparable in relationto
the precedingone. Thus, this openingsectionborrowsits rhythmicand
melodiccomponents fromtheprecedingechappee, andthewholeof sectionC
will be castin the ambitof its lasttwonotes.But considerthe development
from[44]:
Ex.26

s . ,

gr j

s . . . n

8 - n

280 MUSICANALYSIS1 : 3, 1982


r
VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

This paradigm is melodic,but the onlyrationalpartitioning


musttakeinto
accountthe slurwhichlinksthe lastB of b. 32 to b. 33:
Ex.27
[440 i4
& . . 1

t460 3 9 g 3
& . ,

_#t T;
[48S
8 - 1

3 3

Thisis certainly theonlyplacein thepiecewherea segmentis repeated.There


is, of course,a transformation(elisionof the A) at the endof [47], but [48],
whichprolongsit, adoptsmelodicallythe last two notesof [44] and [46].
Hencethe A from[47] in Ex. 27 iS in brackets: Varesecannotaccepta strict
repetition.The similaritybetween[45] and[46] iS SO greatthat,according to
the principleof deception,it arreststhe momentum tovvards
A the second
slurendson theFt. How,then,is thesubsequent musicto be analysed? The
problemis identicalto thatencountered in bs 10-13. Rhythmictransforma-
tionswill nowbe shownin relationto the melodicinvariant:
E>c.28
X [Q]t

E i O------ n

L.ME
[502 8 ---- - -- -n

r#S
3 3
; 60
251] ,< ,

3ar^: G7'SS''
_ B

MUSICANALYSIS1: 3, 1982 281


JEAN-JACQUESNATTIEZ

It is in extremis
that [51] reintroduces
the B. Note that, melodically,[51]
adoptsthe descendingsuccessionA-F#-B whichconcludesb. 32, but as
alwaystheunitwhichcreatesthisequivalence is imaginary,
becauseof theslur
whichseparates[44]from[45].23
Rhythmically and intervallically,
this sectionis distinguishedfrom the
preceding andensuingsectionsbytheabsenceof semitones,thelargenumber
of minorthirds,perfectfifthsandminorsevenths,andby typesa1,b ande:
I [44] a1 a1 7a 3a 3d (7d)
[45] b lOa lOd 7a 3a (lOa)
II [46] a1 a1 7a 3a 3d (7d)
[47] b lOa lOd 7a (3a)
III [48] a1 3d (3a)
[49] e 3d (3a)
[50] e 3d (3a)
[51] b 3d 7d
The breathbetween[45] and [46], like the repeatof the same units in
[46]-[47],justifiesthe demarcation of segmentsI andII. SegmentIII, up to
the fallingfifth(F#-B), is dominatedby minorthirds.The rhythmictypes,
then, havea characteristic distribution:a1 is alwaysat the beginningof a
section,b at the endande in an intermediate position.
5. Theendof PartII (bs36-40)
SectionC hasdevelopedon the samethreenotes:C, a semitoneabovethe
precedingB, seemsto be an intruder.Armedwith a tripleforte,it is the
culmination of the crescendobegunat the end of b. 35, but becauseof unit
[53],it willbe linkedto E. ThisC caneasilybeacceptedasmarkingtheendof
sectionC andthe beginningof sectionD:

Ex.29

X ''7^t'lr
.
[5! t9i

055] , 3 t >y

P SU6fb fp

282 MUSICANALYSIS1: 3, 1982


- D

VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

, Lt , 7
L572--
-I br

[56 SP
vr 2^"
3 '2 '
fP

In the sameway, it does not matterwhetherthe Eb of [53]is the end of [53]or


the beginningof [54]. We attachit to [53] since the diminuendo makes it a
prolongationof C andof the firstEb. In contrast,thetton D isolates[54]with
its own slur. The most importantthing is thatthe rest of the paradigmshows
how [54]-[58] is constructedin relationto Eb, D and Db by a play of
permutationswhich can be read from the table above. Rhythmicallyand
dynamically,[55]-[58]is characterised by the sameprocedure:24 the finalnote
of the four units is alwaysshortened;it is eitherpianoand staccato([55], [57],
[58])or of a shorterwrittenduration([56])andfollowsan accentedfortenote.
In the sectiontaken as a whole, the distributionof rhythmictypes is rather
scattered:al three times, b once, c twice, e once, and [56] which is the only
exampleof b inverted.25There are, therefore,roughlyas many short-long
types as long-short.The intervallicdistribution,combinedwith the zones
among which the differentnotes are divided (C-Eb / D-Eb / Eb-Db-B)
delineatesthree segmentsin this section, wherethe semitonealternateswith
majorsixths, where the semitoneclasheswith the tone and where the final
intervalis a descendingdiminishedfourth.The wholeis, however,dominated
by a tendency to descent, felt from [51] of the previoussection. The first
segmentends on Eb, the secondon D (a semitonelower)and the thirdon B
the lowest note of section D: the B dominatedsection C but had not been
heardagain.
PartII can be describedas full of contrastsand hesitations.SectionA: two
permutationzones (E-Ct-D and Ct-D-G:). Section B: three rapid falls,
then flight to high A. Section C: permutationzone (B-F:-A). Section D:
descentto B with permutationzone on Db-D-Eb. WhilePartI wascharacte-
rised by the rise to high G, Part II, with its varieduse of rhythmictypes,
dynamicsand melodicdirections,seems to be intermediaryand this is partly
due to the largenumberof 'permutationzones'. Evidently,three segmental
types will have three functions in this piece: the permutationis stagnant,
delayingthe appearanceof a new note whichis generallya semitonehigher;or
oblique paradigmsallow the piece to progress;or rapid flights lead to a
climax.Betweenthem, these typesset up a dialectic:the permutationactsas a
brakeon development in relationto the obliqueparadigmsandthe flightsit
favoursa periodof rest ratherthan momentsof tension. Vareserestores,on

MUSICANALYSIS1: 3, 1982
283
JEAN-JACQUESNATTIEZ

anotherlevel,whatthetonalsystemis nolongerableto offer,byalternation


of
distinctfunctionaltypes.

III PARTIII(bs41-61)

1. Repriseoftheopening
(bs4143)
Fromb.41 thereis a reprise.The initialtemporeturns(crotchet= 72)while
[59]is an exacttransposition,
a semitoneabove[1], of the firstthreenotesof
thepiece,whichpermitsa directarrivalon G withoutthe 'suspensions'of [2].
This is probablywhy it is repeated(withdurationalshorteningof the G)
withoutintermediary development in unit [60]:
Ex.30
01 ]

4 F, _J_,>
mf w
J J) S W4
- - w

r - - e

p
26i 3

p
[610 3 3

1e1
P=r y
2
26i 3

i2xJ :
1 06$

2sJ q;t
Lg

I
i b^t?t-

Thisparadigm is simple:thenotesin unit[61]arepermuted,thentheorder


at theheadof theparadigm returnsin [62].It is interesting
to tracetheorigin
of the D andthe Ab of [62]:
Ex.31
[2] 3
[lg

g ° t$s t(F"_,° r 1 °

[62] 3

- to ° v b. v j 4

284 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

[62]is anexactmelodictransposition of themelodiccontourin thefirsttwo


unitsof thepiece.Therhythmsof [59]to [62]arenotidenticalto theopening.
However,in [S9]and[60],typea2of [1] occurs.
Interestingly,beforethe flight of [63], alreadyanalysed,the intervallic
syntagmbearssomeanalogyto sectionA of PartI, in theprogressive growth
of the sameintervals,fromthe semitoneto the tritone:
A I [S9] a2 ld 2a (ld)
II [60] a2 ld 2a (2d)
[61] a2 la la (ld)
[62] b ld 2a Sd 6a (la)
This intervallicprogression,togetherwith the tendencyto rise whichis
initiatedat theendof unit[62]withtheintroduction of anAb(notyetheard),
is reinforced rhythmictypes,a2andb: it is easyto see, here,
by progressive
howb is an expansionof a2
Theanalysiswasbegun,in respectof theflightsin Ex. 24, on theD of [62].
Buttheslurandcrescendo clearlyisolate[63].ThecaesurabetweenAb andA
is another'deception'in the risebeginningwithD.

2. PermutationsonB-D (bs4S50)
Afterthe flight, Varesereturnsto his predilectionfor playingon two
alternating notes:thisrecallsthe wayDb andC areexploitedin bs 9 and10.
TheB andD at issuearepositioned,respectively, a semitoneabovethenotes
of the flight(Bb - Ct). The D, is
moreover, brought in by Ct, a compound
semitoneabove it. This time, the D appearsto predominate:

Ex.32
[6 i 8 ----- - n
, 3 ,

t) , 3

L68-.---------------------,

3 3

t6 6] ; 3 - -- - - - - - - - --- -- - '

f f ft'
Y ,

MUSICANALYSIS1: 3, 1982 285


3 3 3

JEAN-JACQUESNATTIEZ

26703 '

fS f
[ , s ,

06! - ,

Thisis doubtlessbecauseof rhythmictypesa andb whichconfera longvalue


on the finalD's:
Ex.33

[ ] G 5 G r r, ;
3 3
r65] G G r (a2,

3 3 3

0660 : : r r (a2
)
3 >s3
067] : r 5 (a1
)
3 3

0680 r r G S (a1 ,

[69] G C f ( b)
Lessimportant thanG andB, D nevertheless playsa decisiverolein certain
intermediarymoments:it closestheplayaroundB in b. 21;wefindit againin
a privileged
paradigmatic positionin bs 25, 26and28;it willreturnbeforethe
finalflightas the uppernote of the permutations of bs 5S58. It owesits
particularimportanceto the registralleapafterC# of [63]whichmakesit the
highestnoteof the piece.

286 MUSICANALYSIS1: 3, 1982


4 ff

VARESE'S
'DENSITY21.5: A STUDYIN SEMIOLOGICAL
ANALYSIS

Thiswholesectionof alternating B'sandD's endson a littlemelodicfigure


whichis the inversionof the initialF-E-F# andoutlinesa finaldescent:

Ex.34
st to ne

,2

X {w O #.
3
st tone

1
I _
j _
r _
11-1
5 . #

B P f P

[70]is treatedas anappendixto sectionB forseveralreasons:first,because


theopeningfffistheendof thecrescendo from[69];second,becausethereis a
breathonly in the middleof the bar;finallybecausethe characterof the
sectionbeginningwith[71]is quitedifferent.

3. Thelastsegment(bsSI-61)
Thisdoesnot mean,as in previouscases,thatthereareno linksbetween
thisfinal'phrase'andwhatprecedesit. BetweenC andF# of [71]wehavethe
samedescendingtritoneas betweenD and Ab of b. 50 in double-octave
displacement. But,aboveall, [71]bearssomerelationto thewidedescending
intervalsin rapidrhythmof [38],[39]and[40],and,to a lesserextent,[53]:

Ex.35,

7
t539
^ > '_ [54]
-

r - ff
070 s

[72]v. <:+

SfPrP
r A: SL,4Jb; ()
3 3 ,_

MUSICANALYSIS1: 3, 1982 287


-

JEAN-JACQUES
NATTIEZ

tW*-*s s [SS WS

t1 >
, (_) 077

. [d P
' JR w
3

br F:
crcsc. rnolto
[29I'/ fE >
-ff4; : F
[802

$ - 3 > I t82]e, te

Eventhough[71]beginswitha C and[38]witha G, thefallto F#- E# on a


practicallyidenticalrhythmandthedemisemiquaver Cmakecomparison with
[38]legitimate.Thisparadigmatic tableshowsthe rhythmictransformations
in relationto melodicdata,but the implicitunitswhichemergecannotbe
retainedbecausethe slurshavetheupperhand.Numbers[74]and[75]mark
theboundaries of morerealunits.Ex. 35 thusbringstwocontradictory levels
of segmentation together.It willbe noticedthat[72]and[53]areimmediately
followedby a unitof thesamerhythmictype([73]and[54]),e, whichalsohas
the samenumberof notes.
From[75]onwards,the composeris slidingto anotherparadigmatic axis.
Ex. 35 demonstrates thatthisentireconcluding segment(fromb. 51)formsa
whole, sincewe progresstowardsthe finalhigh B by successivechanges.
These changesare of a particularly characteristicnature:in [75], E# (F)
becomesE, a semitonebelow.TheC on theleftof theparadigm becomesC#
(a semitonehigher).[76] marksthe firstattemptat melodicflight:for the
momentwe stayon G, a semitoneabovethenewF#, itselfa toneaboveE. In
[77],as in bs 5 and6, Vareseadoptstwoprecedingnotesandrisesto A26(a
compound toneabovetheG). [78]introduces a Bb, thepitch-class
a semitone
aboveA of [77]. In [79], D, a seventh(inversionof the second)fromE, is
added.[80]and[81]playonC, E, D, [82]takestheBbof [78]whichthusfills
in theintervalE-D of [79].Therefollowsthefinalflight,whoseconstruction

288 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

on playsof tonesandsemitones,is shownin Ex. 24, togetherwiththewayin


whichit endson B.
Thisendingis thusa resume of all thatprecedes:the fallsof bs 51 and52,
the playof permutations, the progression on G thenA andB andthe final
flightare all proceduresencountered previously.The tendencyto returnto
eventsfromthefirsttwoparts,whichwenoted forthe twopreceding sections,
is confirmed here:afterhavingrecalledwhatgoesbefore, Varese undertakesa
riseto the lastclimaticpointwhichG andA prepare.

IV RECAPITULATION
The tableon the followingpagesattemptsto give a globalpictureof the
piece by assembling,in synopticform, the essentialelementswhichhave
underpinned atfourlevels:parts(Part
thisanalysis.Ontheleft, segmentation
I etc.), sections,segmentsandunits.Thenthethreeprincipledata:rhythmic
types, intervallicsequencesand melodicpattern.On the right, a general
characterisation of the syntagmsin units:zonesof permutation,of progres-
sion,of flightandof descent.27
Thistablecould,in addition,havecarriedotherdescriptive variables: for
example,slursanddynamics,bothof whichareessentialto the piece.But
theyaroseaboveall, in the precedingpages,to delineatethe unitsso thatI
considerit unnecessary to includethemhere.

MUSICANALYSIS1: 3, 1982 289


o

PartI
Segmentation Rhythmictypes Intervallicsequences Melodic Patt
a2 ld 2a (Sa)
d Sa Sd 6a (2d)
A I [1]
[2] a2 ld 2a (la)
II [3] d 6d 6a (ld)
[4]
a2 la ld (2d)
III [s] "d" 3d 6a (6d)
[6]
B I [7] b 6a 2a la (3d) C$G/A-Bb
II [8] c or e 3a 3d (3a) G - Bb
[9] al 2a (la) /C
III [10] a2 ld la (ld) Db -C
[11] al la (ld)
[12] b la ld (la)
[13] a2 ld 2a (6a) /D
IV [14] al 6d (6d) G#- D
[15] al 6a (6a)
[16] b Sd + 6a 6a (6a) /A- D:
V [17] b 12d l+a (l+d) A-Bb - E
[31] al 3a (lOa) /D , .

Segment ation Rhythmictypes Intervallicsequences Melodicpat


VI [18] a2 ld 2a (2d) E - D# - F
[19] c 2a (2d)
[20] al al l+d (la) /F
[21] la++ (8d)+

c :Ia[22] d ld la (O) B-A#


[23] d 2d 2a 3d (3a) /A
[b [24] d 2a+ ld 2a (3+d) B+C# B# D
Ia [25] a2 ld la (2d) B-A#
[26] d la la (2d)
[273 al la (2a ) /B

Part II
A I [29] e 3d (3a) E - C#
[30] e 3d 0 (O)

to semiquaver C#-D-G#
[33] e l+a (l+d)
Segmentation Rhythmictypes Intervallicsequences Melodicpat

/Eb

a3 ld 12d ld (2+a) G-F$E:


B I [38] a2 ld 12d(l+a)
[39] a3 ld 12d ld (O)
[40] a2 lOdlla la l+d(la)
II [41] a2 la la (Sa) IG-G$A
[42] b 6a lOa3a lOa(lOd) E-Bb-G$B-
[43]
C I [44] al al
7a 3a 3d (7d) KFFA
[45] b lOa lOd 7a 3a (lOa)
II [46] al al 7a3a3d(7d)
[47] b lOa lOd 7a (3a)
III [48] al
3d (3a)
[49] e 3d (3a)
[50] e 3d (3a)
[51] b 3d7d(1a)
oQ -_ v o ¢
I-t-tt o ffi N +m -t t v)
I -t o £ffi t -t -

VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL


ANALYSIS

'e

t: - 7
. -

t # n m A=t ;

u)

Ct

-v CC

- r N 'e cC: r r t r CC:


v v ,/ m m -v s} -v
> < E-vt C -ooN t Ct Ct Ce Y Ct
Q - Lo N m o s} m
= r t 'e t 'e r t r t Ct r Ct t 'e 'e 't

o
u) =
Q u)

.- =
= -o

$ ce D X ce = t) t ce ce ct D D D ce ce ce

N m t V £ S X
=
n
- , I
m m m m
. .
m
. v)££££ ££££

- - -

v
*

ct

bt

ct

293
Segmentation Rhythmictypes Intervallicsequences Melodicpat

[68] al
[69] b
III [70] e

A I [71] a3
6d ld (7a) SF$E:
[72] al
6d (ld)
[73] e 7a 6d (la)
[74] c
7a (8+d)
[75] c+ d 2d ld 3a 3d (3a) /E
IIa [76] a2
2d la (3d) E-F$G
[77] b 3a 2+a (lld) IA
[78] d 6d 6a (lOd) /B
IIb [79] b 4a lOa(lOd) CE-D
[80] al
4d (2+a)
[81] e 2+d 4a (6a)
[82] b 4a 4a 7a (2a) Bb-D-F$C
[83] b 2a 2a 6a D$EtB
oo}^,4... ...* 4O ho I I I o l ., bo

VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL


ANALYSIS

The 'melodic pattern'column summarizesa phenomenonwhich has been


discussedconstantlyhere the progressionof the piece by a successionof
zones of privilegednotes which are overshotby new addednotes. This gives
the followingoverallpicture:

Ex.36,partone
[1]-L63

g '>P °
[70

t82-t9]
, b.

E°d-[120
bo .,

L1
30
,.,

E40-05]
o

016,
o

[170

[180-[19]
0 >, o

[ ] 21]

o >

[2i-S2@ " g o

L2- [280 " t

MUSICANALYSISl: 3 , 1982 295


[38}[42] Xo I [43] -

JEAN-JACQUES
NATTIEZ

E>c.36, part two

[29]-032] [33]_[370

X 4° 14 #°

°-$° o I o b l
[44]-05 1]

[52]-[53] 0540-[570 058]

- b Xo 1x^
bo | b*.<

Ex.36,part three

0590-2620 ,, bn"-
vo ° ' g bsw14vw
264]-0690 [700

° 40 1S4nW
[ ]eLL [7SS [76, L7d29][8t] i2]-g83] #E"+Z_

Theprogression to highG in b. 17is followedby a zoneof hesitationwhich


alsobelongsto B (bs 18-23).PartII, dividedintofour'moments',comprises:
playaroundE, C#, D, thenC#, D, G#;successivefallsthena climaxon A;
playaroundB, F#, A; playaroundC, Eb, D, Db. PartIII contains:reprise,
permutations on B;D anda finalphrase.
Four syntagmatic familiesmay be extractedfromthe melodicpattern:
permutation, progression,flight,descent.Usinga horizontalline ( ) for
permutation, anobliqueline( / ) forprogression,
anarrowed obliqueline( / )
for flights,and a descendingobliqueline ( \ ) for descent,we obtainthe
followinggeneralpictureof the piece:

PartI _/_/ /t /\/_/


PartII _/_/\//_
PartIII _// / /

296 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL
ANALYSIS

Thisplanshowsclearlythe tendenciesof eachpart:PartI-ascent,PartII -


hesitation,PartIII ascent-descent-ascent.
Throughout the analysis,the sectionsandsegmentshavebeendifferent-
iatedmostoftenby contrast:A of PartI groupstogetherrhythmictypesa2
andd, whereasIII of B in the samepartis characterised by successionsof
semitones.Duringthe courseof the analysisdominant variables
havebeen
emphasised momentby xnoment.The completetableenablesan assemblyto
be madeof all the components of the musicalmaterialandconsequently to
replacedominantvariablesby thosewhicharenot dominant.
Thiskindof tablealsomakesit possibleto verifywhetheror notthereis a
correlationbetweentwophenomena. By definition,thefoursegmental types
ontherightarecorrelated withthemelodicpatternsandintervallicsequences.
Do they correspondto the rhythmictypes?A hypothesisis that regular
typeswill correspond to stagnantzonesof permutation, thattypesa, b, c
(augmentation) willcorrespond to zonesof progression,
etc. Thepartitioning
of the fourfunctionaltypesis nowprojected ontothe rhythmictypes(noting
thattheydo not necessarily correspond to segmentsandsections):
Permutation Progression Flight Descent
a2d a2d a2d b c or e al
a2al b a2
al al b b
a2c al al al dd
d
a2d al a2
e e al
eee a3a2a3a2
a2 b
al al b al al b
ae e b al b
e al al c c
a2a2a2 b b
b a2a2al al b e
a3al e c cd
a2b d
b al e bb

Tryaswemight,noparticular correlation
canbediscerned betweenrhythmic
andfunctionaltypes.This negativeresultis still progress:unlessthereare
gaps in our inventoryof variables whichcouldalwaysbe filled in by
anotherresearcher it showsthattheimpression of stagnation
orprogression
is dueonlyto modalitiesof the melodicline.
The valueof the procedurefollowedis, in any case, obvious:without
relyingon the completetable,particularlylookingat functionsin termsof

MUSICANALYSIS1: 3, 1982 297


JEAN-JACQUESNATTIEZ

variables,it could have been statedthat rhythmictype b was absolutely


characteristicof theflight(b fivetimes,asopposedto a1andd1once),whichis
truein a way,but b is foundsix timesin thepermutations andfourtimesin
the progressions whereit is as frequentas a2. It is impossibleto claim,
therefore,thatb is therhythmicvariablewhichcorresponds to theflightsince
it is foundelsewhere,in combination withothervariables.The sameis true
fore, mostlysitedin permutation zones(exceptonecaseof descent)wherewe
findalltheothervariables too.A variableis notpertinentin relationto a given
function,28 in anyabsolutesense,and,exceptin the caseof b, theredo not
appearto be any correlations strongenoughto extracttendenciesof any
interest.
Here,the flightis, in fact, a special,distributionally
privilegedcase(five
occurences) in the progression: if we lookat the kindsof unitsin whichb is
found,we canstatethatin twelvecasesoutof seventeen,theseunitsendwith
oneorseveralascending intervals(insevencasesoutof twelve,alltheintervals
ascend).It appearsto be possibleto establish,on anot)zer level,correlations
betweentwo or moreparticular variablesenteredin the table.
Thereis no recipeforfindinginteresting combinations: the analysisfunc-
tionson the basisof hypotheseswhichcanbe neitherconfirmednorrefuted
unlessa taxonomic description is availablewhichis asresponsible andexhaus-
tive as possible.This is doneby 'churning'the data,thatis, projectingthe
characteristicsof one datumontoanother,startingwiththoseintuitivelyfelt
to be the most promising.29 It was shown,for example,that flightshad
particular internaland distributional characteristics
(sectionII/3). Studyof
the table revealsthat, withinone segment,thereare manypermutations
followedimmediately by a progression. Theseare:
PartI B III [10]-[12]/[13]
IV [14]-[15]/[16]
VI [16]-[20]/[21]
PartIII A I [59]-[61]/[62]
C IIa [79]-[81]/]82]-[83]
Withthe exceptionof [25]-[27]/[28],theseparadigmsshowan interesting
rhythmicanalogy:

298 MUSICANALYSIS1: 3, 1982


JJ$Sy
1F ; r > v -
% 5 - - o - ^l - Ea - >>
g

VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL


ANALYSIS

E>t.37

E00-E3q E4]-016]
E8]-t2lt

tS w r p
6 Ll; E 43 5
-z tf r 4 2tG 7

[59]-[620 [794-t820 [440_[45]

3 3
..1d1 t
#zi e-

A*m
r C- 5, bJ r ]3.1 b9 | / s 3

Beforethenotewhichcreatestheprogression, at thebottomof theparadigm,


thereis rhythmicacceleration throughthe use of the tripletafterbinary
rhythms andanincreasein thenumberof notesin thelastunit(exceptin the
firstcase).Thesameprocedure willhavebeennotedin [44]-[45],butthereit
doesnotleadto a progression.It is rareforoneprocedure to be encountered
in 100%of cases:musicalstyleis not a system.
Let us lookat the permutations on two notes:
Db- C [10]-[13]
B - A: [25]-[28] semitone
E - C: [29]-[31]
F: [49]-[511
L J | - J minorthird
D - B [64]-[69]

MUSIC
ANALYSIS1: 3, 1982 299
JEAN-JACQUES
NATTIEZ

Theintervallic constantis in itselfremarkable.


It becomesevenmoreremark-
ablewhenit is noticedthatthe firsttwo followpassageswhichcontainno
intervalsgreaterthana minorthird.As fortheotherthreepermutations, they
appearaftersegmentswherethe perfectfifth is present,one of the rarest
intervalsin thepiece,foundonlyin [51],[73]-[74]and[82].It is therefore no
exaggeration to postulatea correlation betweenpermutations on two notes,
theintervalof a minorthirdanda contextin whichtheperfectfifthappears.
Still more little correlationscould probablybe found in this piece.
Moreover, theglobalperspective of ourneutralanalysisin musicalsemiology
is, we know,stylistic:thismeansthatif otherworksby Vareseareanalysed,
it shouldbe possibleto pick out fromour tablecorrelations or phenomena
whose identificationis not permittedby 'Density'alone. Returningto
'Density'on the basisof a widerfield of workswouldbe thatmuchmore
efficient,giventhatwe wouldalreadyhavedrawnup a prettydetailedinven-
toryof variables.Theneutrallevel,it mustbe stressed,is onlya 'moment'of
analysis.
A few generalobservations on the methodology usedhere:
(1) It is possibleto partitionparameterby parameter,but the global
segmentations performedon the syntagmatic chainare the resultof
intuitionabouttheconvergence of differentvariables
ata givenmoment
andthe hierarchic dominance of someof these.The collationof infor-
mation, parameterby parameter,variableby variable,shows the
constitutionof theglobalsegment.Thisglobalsegmentcanbe deduced
fromthe sum of the individualsegmentations, but only to describe
texture.Thereis, therefore,no logicalor necessary orderaccording to
whichan analysisshouldbe conductedand presented:here,as else-
where,the orderof reasoningdoes not coincidewith the orderof
discovery.
(2) One particularconsequenceof this principleis manifestin the
constructionof equivalenceclasses:when the assimilationthe of
rhythmsof [1], [3] and [5] was proposed,in spite of theirphysical
differences,it wasnotonlybecausethesedifferences wereminimal,but
because,onanother level,melodicidentity(F-E-F#) inducedanassem-
bly of theserhythmicunits.It is impossibleto say,therefore,thatthe
rhythmicanalysisis purelyrhythmic:in fact) otherparameters are
takeninto account.The resultof all thisis a mixture.
(3) I havespokenseveraltimesofprojectzon. Thishasa doublerole.First,it
permitstheevaluation of theroleof onevariablein relationto another,
as in the previouscase. But it also has an heurtsticvalue:it enables
interestingcorrelationsto be discovered betweenvariablesof different
types.Consequently, thereis no limitto thenumberof projections it is
possibleto performfromone variableontoanother:thisis essentially
why analysisis endless,andevenif one hopesto put one'sfingeron
somecharacteristic correlations,
thereis stillthe convictionthatother

300 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

equallyvalid relationshipsmay have escapedour attention.There are,


therefore, no rigorous rules for determiningwhich projectionswill
prove viable. We can procedeonly by trialand error.
(4) Thus, correlationsare establishedbetweendisparatecharacterisations
of differenttypes and levels: one exampleof this is the distributionof
flights. This is identified, on the left, by a descendingcompound
interval,a moreconcretedatumthanrhythmictypesin general,or than
type b which, positionedat the end of a segmentand with its play of
semitonesin the pitch structure,characterisesits content.
(5) A great deal of intuition enters into the research,but althoughthe
objective of taxonomic clarificationis confirmation,it may equally
resultin invalidation.This wasthe case,in the researchphase,for rests:
it was assumedthat they could be enteredin differentclassesaccording
to the type of segmentthey delineated.A study of the taxonomicdata
revealsthatthe only interestingconstantis the unusualpresenceof rests
in the 'percussive'sectionof Part II one of its characteristics.

V POIETICANALYSIS
Severaltimes, here and elsewhere,there has been occasionto recallthe fact
that neutralanalysisis an essentialbut intermediarystagein the semiological
approachto musicalworks. In this and the ensuingsectionthe intentionis to
show how the data of neutralanalysisrelateto those of poietic and esthesic
analysis.

1. Thepoieticproblem
There are two ways in which the phenomenacataloguedby the neutral
analysiscan be consideredpoieticallypertinent.To the extent that analysis
deals with the score, it is directed at the only traceleft by the composer.
Thereforeit is possible to consider that recurrenttraits demonstratethe
preferencesof the composerfor certaincompositionalprocedures;they enjoy
the presumption of poietic pertinence. This presumption is confirmed
particularlyif otherworksby the samecomposercontainthesetraits(don'twe
say 'he likes to do this, he likes to do that'?),or if our historicalknowledgeof
the evolutionof musicallanguageestablishesthat, on the basisof the heritage
receivedandexperiencedby Varesewhen he beganto compose,he decidedto
orienthis compositionalpracticein this or that direction.30
This poietic proceduremay be qualifiedas inductive:

poietics; neutrallevel

But it is just as feasibleto startwith an externalpoieticelement-a sketch, a


roughdraft,a commentary- and projectit onto the work, eitherto directthe

MUSICANALYSIS1: 3, 1982 301


JEAN-JACQUESNATTIEZ

analysisaccordingly,
or to reorganise
the neutrallevel,constructed
indepen-
dentlyof the externaldatum:

; work
poieticdata > or
neutrallevel

In the first, inductiveperspective,whichtraitsidentifiedby the neutral


analysishavea poieticpresumption?
(a) Theprincipleof deception:I haveshownhowit governeddevelopmen-
tal procedures,at the sametime frustratingthe expectations of the
Westernlistener.Thus, the procedures describedstandout fromthe
wide rangeof habitshandeddownby the dynamicsof the romantic
musicalphrase,whoseprinciplesof progression andelevationVarese
preservesbut constantlycontradicts.
(b) Theprincipleof maximaldifferentiation: on theneutrallevelrhythmic
equivalenceclasseswere constructed(typesa to e), but it must be
remembered thatVareseis carefulto writevalueswhichareas distinct
as possiblefromone another.Oneof his favouriteprocedures consists
of multiplying thenumberof dotsfromthebasisof a givenvalue(asin
Integrales).Slurs,tripletquaversor semiquavers introducesubtledif-
ferentiationsfrombinaryvalues.Equivalence groupingshave,there-
fore, a presumption of esthesicpertinenceaboveall in oneparticular
case.Thisdoesnotmeanthatthedifference betweena crotchettiedto a
semiquaver anda crotchettiedto a tripletsemiquaver is imperceptible,
but it cannotbe perceivedon the samelevel. The discernment of a
rhythmicprogression (typesa andb) andthediscernment of a minimal
differencebetweentwo valuesarenot incompatible. Detailedexperi-
mentswouldin anycasebe requiredto establishthe thresholdbelow
whichtwo unitsareconfusedandabovewhichtheyaredistinguished
fromone another.
(c) Thepoeiticcounterpart to thealternationtension/relaxation,thatis, the
methodschosenby Vareseto createthis esthesiceffect:playof semi-
tones,dissonantintervals,crescendi,the configuration of flightson the
one hand,andthe permutations of stagnantzoneson the other.
Butin the secondperspective,thereis a commentby Vareseon 'Density',
published,withoutreferenceto its source,by HildaJolivet,givenherein its
entirety:
Despite the monodiccharacterof 'Density21.5', the rigidityof its struc-
ture is definedovertlyby the harmonicschemecarefullydescribedin the

302 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

unfoldingof the melody. 'Density 21.5' is based on two short melodic


ideas, the first, modaland in binaryrhythm,which beginsand ends the
piece, and the second, atonal and in ternary rhythm, which lends
elasticityto the shortdevelopmentsseparatingreiterationsof the firstidea
(Jolivet1973: 110).
I shall use this externalpoieticdocumentin a preciseperspective:after
makirlgan appraisal, its contentswill be appliedto the configurationsof the
neutralanalysisin orderto reorganise it. The objectiveis obviouslyto com-
parethe two picturesof the workthusobtainedandto definethe statusof
neutralanalysisin relationto this kindof poieticanalysis.

2. Melodicpoietics
We begin by examiningthe secondpart of this text whichdealswith
melodicideas.Thequalifications 'modal'and'atonal'willbe setasidesince,if
the two ideascorrespond as I believeto [1] and [2], it is not clearthatthe
secondshouldbe moreatonalthanthefirst,or thefirstmoremodalthanthe
second.One couldobservesoundphilologicalprinciplesand turnto other
textsby Varese,in an attemptto shedlighton the meaningof thesewords.
ButVaresewasno theoretician, andnothingin hiswritings(1983)helpsus to
understand whathe means,here,by 'modal'.Withregardto 'atonality', these
twoquotationscanbe set off againsteachother:
[In contemporarymusic],whetherwe deny its presenceor not, we sensea
tonality.Thereis no need to havea tonic, with its thirdandfifth, in order
to establisha tonality(1934).
My languageis naturallyatonal, althoughcertainthemes, certainnotes
repeatedin the mannerof tonics, constituteaxes aroundwhich sound
massesappearto agglomerate.In this way, musicaldevelopmentgrows,
little by little, throughthe repetitionof certainelementswhich are pre-
sentedneachtime, in a differentaspect,andinterestincreasesthroughthe
oppositionof planesand the movementof perspectives.If themesreap-
pear, they have a differentfunctionin a new medium:dynamics(1930).
All of thislasttextcouldbe appliedto 'Density21.5',butit doesnothelpto
pinpointthe meaningof 'atonal'as opposedto 'modal':on the basisof these
twoquotationsthe piececouldbe qualified'tonal'justas muchas 'atonal'.
Thebinary/ternary oppositionis moretelling.I havetherefore
playedalong
andconstructed a paradigmatic
tablewhereaxesof equivalence aredefinedby
the rhythmiccharacter of units,binaryor ternary(Ex. 38 below).The first
axisthereforecontains[1] andits transformations. Noticethat[5]hashadto
be placedin the secondaxisbecauseits rhythmis ternary.[71]and[72]have
beenplacedin the firstaxisbecauseof theiranalogous distributional
position
andparadigmatic linkwith[1]throughtheintermediary [38].Thesecondaxis
containsall theunitswithat leastone 'irregular'value.Therefore a thirdaxis
must be opened,one not mentionedby Varese,whichgroupstogetherall
unitsthatareneither'areturnto thefirstidea',norin ternaryrhythm,thatis,

MUSICANALYSIS1: 3, 1982 303


_ § l I [ u l Lf .

JEAN-JACQUES
NA1YIEZ

Es.38
t22
tr] 3 ,

LI t S 1D 1 o-
ll - ww

n X 04 ]
S JtL r --3

[5 ,l; t

_ 1 . _
vr _

t6 3 ,,

[100
o a- 5 [93 3 l ,

=FtU S : r f >
L] v E g - s3 ossia 3

tf r Af r
> 3 3

0 50 3 3
01 70
#2
[1 8]

L ] j -

[2 3]
ll

44r ' [27] ag

-
L290 r 3 1

H;; [
304 MUSICANALYSIS1: 3, 1982
:bS' 3 > [532 r

VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL


ANALYSIS

Ex.38 cont.
[312 [3 2]
2300 3

-
StJ77N,$ r w

0340 +
t]7 7P2:
3J,
L6]flC.e
S
\ =
I V rT,< r \

5L4528-r
t Lp
" d: ,r n
,
L4 7 s- - - - - -

L
0540

L553 L3 '

r560

[53 3 3

=+G 7
[580 3

305
1

NATTIEZ
JEAN-JACQUES

Ex. 38 end

t593

L6 1 l

>qi -X
tt- 3 n 0630 bss"e
,621

t20g-J
fi ..
- -44 r'lF
Ve 3 _ . 3
[65] 8 t f

3
L3 ° r- -
& 3;rCC
8 --- n
t6 71 - t_^
J tS 3-
_ Sr
068]ot [69]:5't'

s_3 T

j_
[7O]JCo)z
t7l;
L _ 5
,
[72k# [740
[73] 3- t

-- S Ser 0750
Mu
3

t760 3 3
[770 ; $ 5 j
- nz J poJ.If
078] _ 079]

l -
w
l
tJ
- - f |

0800

[82] X tt

MUSICANALYSIS1: 3, 1982
306
VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

thosethatfollowa ternaryunitandprecedethereturnof anotherternaryunit


or of the first'idea'.
Whatmay be concludedfrom this example?First, it must incite us to
approachpoieticinformation with a measureof caution:this information
definesa fieldof equivalence whichis not systematically precise.31We have
alreadyhadto discardthe designations 'atonal'and'modal'.Varesesaysthat
thefirstideaendsthepiece:in fact,it lastappears in strictrepetitionatb. 42of
a sixty-onebarwork.Moreover,he talksof 'shortdevelopments' between
reiterationsof twomelodicideas,butthesecondparadigm is by farthefullest
(forty-fiveunitsasopposedto twelveforthefirst).If thehypothesis is adopted
thatthe firstbasicmelodicideacorresponds notto [1], butto segmentI, the
development is defactoshortened,but the first'idea'is no longershortor
binary.Finally,thethirdparadigmatic axis(twenty-four units) therest
is far from negligible:Varesedoes not mentionit. In fact, it wouldbe
interestingto knowthe exactsourceof the text by Varesequotedby Hilda
Jolivet:the composer'sspokenwords,programme note, personalmessage?
Hasit beencorrectlytranscribed?
This poieticlaxnessallowsa stretchingof the principlesset out abovefor
the construction of axes:[1] and[60]containan irregular value,butas [1]is
preciselythe one Varesequalifiesas binary,it may be supposedthat the
ternaryaspect(tripletsemiquaver placedat theend)is negligible,sinceit has
no influenceoverthe momentwhenF: ([1])is articulated. The sameis true
for the finalD of [13], but hesitation,in this case,alsocomesfromthe fact
that,as in [7], the longvaluesbelongto a unitdefinedas an entity.
Theexamination of thethirdaxisis of greatinterestto ourstudy.Indeedit
showsthat the projectionof a poieticdatumonto the neutrallevel can
reorganise it, without theneutraldessnption losingitsvalidity.Thatis notall:the
poieticdescription wouldnot be whatit is hadthe neutralanalysisnot been
performed first.
Whatgoesto makeup thisthirdaxis?Theunitsit containsmaybe entered
n tourcategorles:
* ,* .

(1) Descendingunits:[23],[49],[52],[53],[54],[75]and[80]wherethere
is noprivilegedrhythmictype.Theonlydescending unitsnotin thisaxis
are[38]to [40], [58]and[70]:the firstthreebecauseI havedecidedto
assimilatethem,here,withthe 'initialidea';[58]and[70]arethe only
descendingunitsin the secondaxis.
(2)All the unitsqualifiedas flightsarein the thirdaxis.Theseunitswere
by typeb (rhythmin augmentation).
seento be characterised Theflights
maythusbe drawntogether.
(3) Threeascendingunits,[9], [27]and[69],havea rhythmwhichis also
progressive (a1twice,b once).
(4) The lastcategorygroupstogethertheotherascending unitsin the axis:
fiveof thesebelongto thepercussive section([31],[32],[34],[36],[37]),
the other,[74],is a hapax.

MUSICANALYSIS1: 3, 1982 307


JEAN-JACQUES
NATTIEZ

Whatmightbe concludedfromthisdivision?It confirmsthefinalposition


of unitsof progressivetypes(a1andb) andascending direction,sincematerial
separating twoternaryunitsora ternaryunitanda returnto theinitialideais
placedin the thirdaxis. In as muchas Varesedoes not speakof flightsas
specificprocedures, one mightwonderwhetherhe waspoieticallyconscious
of them,but this distributionaccordingto the threeaxes(of whichone was
obtainedbydifferences) showsthatthethreeprinciples of constructionin this
pieceare:the initialbinaryidea, the ternarytransition,and the ascending
unitswithprogressive rhythmwhosefinaldistribution constitutesa character-
isticstylisticelementof Varese'smusicalrhetoric.
By numbering unitsaccording to the paradigmsin whichtheybelong,the
syntagmatic poieticorganisation
of the piecemaybe presentedthus:
I II I (II)x9 III
I II (II III)x2 III II III II II (II III)x3
I I I II II III (II)x5 (II III)x2 III III (II)x4
I I II II III (II)xS III II
I I II (III)x2 (II III)x2 III II (III)x2
The followingprinciplesmaybe established:
(a) It is not possibleto go fromI to III withoutpassingthroughII.
(b) I mayfollowitselfbetweentwo andninetimes.
(c) The syntagmII III maybe reiteratedtwo or threetimesbeforethe
returnto I or II.

3. Harmonic poietics
Thefirstpartof Varese'stextstatesclearlythatperforminga monodicanalysis
is not enoughto givean accountof thispiece.An harmonicanalysishasnot
beenincludedin the 'neutral'partof outstudybecausethereis alreadyonein
existence:thiswillbeexaminedbelow.Theharmonic description of 'Density'
belongsas muchto the neutrallevelas doesmelodico-rhythmic description.
'Neutral'implies,as I havesaidelsewhere(1975),'neutralisation',sinceit is
not trueto say thatmelodyandharmonyshouldbe analysedtogether:this
leadsto neglectingaspectswhicharestrictlymelodicor strictlyharmonic.In
sayingthatthe unfoldingof melodyfollowsthe harmonicplanandmakesit
explicit,Vareseindicatesclearlythatthepoieticanalysisconsistsof showing,
on the basis of two neutralanalyses melodicand harmonic,how the
monodicunfoldinggivesthe harmonicstructureits form.
The completetextof MarcWilkinson's harmonicanalysis(1957:17-18)is
set outbelow,interspersed withcommentsandcriticalremarks.I wouldlike
to say straightawaythat I considerthis text quite remarkable, and the
pernicketynatureof my commentsis due onlyto the explicitperspectiveI
havechosenhere.Wilkinsonwrites:

308 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

I shall concentrateon the harmonicand melodic structure,for the ele-


ments of rhythm and dynamicsserve mainly to define harmonicareas,
phrases,and motivicelements,and they arenot particularlyproblematic.
Thisis a goodexampleof whathappenswhena specificaspectof a workis
given privilegedstatus:otherelementsare takenas reador presentedas
secondary.I believethat I have shown,on the contrary,that interestin
of other
rhythmfor its own sake (that is, the provisionalneutralisation
showsup characteristics
variables) whicharepeculiarto rhythmandcontri-
of the piece.
buteto the globalorganisation
The first phrase(bars 1-17) is composedmelodicallyand harmonically
aroundthe notes E, C#, G. This relationof minorthirdsto a centralnote
(E-C#, E-G) is the harmonicelementon which the whole piece is built.
Bars1-5 arethe firstand simpleststatementof this relationship(withE as
centre);F and F# are ornamentalto the harmony,thoughby choosingto
fill in chromaticallythe higherof the two structuralminorthirdsVarese
prefiguresthe upwardsweep of the melodic line throughoutthe piece.
The little cadenceE-C#-G in bar S merelyrecapitulatesand clarifiesthe
harmonicrelations.
The author'sdescriptionshowsverywell how the ambitC#-G is filled
progressively,by F andFt, thenby the E whichintervenesonlyin b. 5. It
mightbe emphasisedthat the lengthof the E is partlyresponsiblefor its
harmonicimportance.It is not possibleto speakof an ascendingcurve
'throughout anddescent,butit
thepiece',sincetherearezonesof stagnation
is trueto saythatF andF: leantowardsG. Myanalysisof themonodyof the
firstfive barsclearlyshowshowthe rhythmicandmelodicprocessesrealise
the harmonicplan, accordingto Varese'stext, and thus how the poietic
analysisis the resulthere of the combinationof neutralanalysesfor each
separateparameter. Wilkinsondelayscommenton bs S8 andcontinues:
Bars9-17 aremorecomplex,for two statementsof the harmonic'area'are
telescopedinto one, thoughin bars12-13 the ambivalentnotesaredefined
by octavetransference.The first harmonicsphereis centredaroundC#;
the minorthirdrelationsare tracedin the notes C#< (bars9-10) to Bb
(bar 13), and in the notes C#-D-D# (bars 11-12) to EW(bar 13). The
secondsphere,centredon G, is presentedin a melodicandharmonicform
of retrogradation,for the G appearsonly at the end of the phrase(bar17).
The two minor thirds are built with the (G)-G#-A-Bb in bars 12-13
(noticethe importantroleof octavedoublingin makingthismotionclear),
and with the notes E - F- F#- (G) in bars 15-16. The D# in these last
two barsis held overfromthe othersphere(bar12)andresolvesto the EQ
in bar 16. However, by the subtle use of phrasemarkings(such as the
breathbeforethe F# in bar 16), Varesesuggestswith this D# a new minor
third relation(DtF#) as a secondaryharmonicEleld.Meanwhilethe
octavetransferenceof E# and F# (bar 16), and the long durationvalues,
define a cadenceto G in bar 17.
BarsS8 link the two mainsectionsof the phrase.The musicalmaterial

MUSICANALYSIS1: 3, 1982 309


JEAN-JACQUES
NATTIEZ

emergesby elision from the cadencein bar S; the uppernote G becomes


the lower note of a new minor third relation(G-A-Bb). This process
repeatsitself by a simplesequenceupward(Bb-Ceelisionto C# in bar9).
The rhythmand the dynamicsstrengthenthis feelingof motionby minor
thirds.
All of this analysisdeservesattentivestudy.I haveno objectionto the
accountof bs S9. Wilkinsonis explaining,in harmonicterms,theprinciple
of deceptionwhich delaysthe arrivalof Db, which I have emphasised
throughout the piece.Onewonderssimplywhetherit is possibleto continue
fromb. 6, to base harmonicexplanationon a centralnote framedby an
ascending anda descending minorthird,asthepositionof E in b. 5 appearsto
imply.Forbs 6-, Wilkinsondoesnot refuteit andthisis doubtlesswhyhe
makesthis passagea linkingzone betweenthe two main sectionsof the
phrase;but the explanatory principlereturnsin b. 9.
The author'sideais as follows:betweenbs 9 and 17 therearetwo inter-
twinedharmoniccentres.The firstis Db, the second,delayeduntilb. 17, is
G. The descendingminorthirdfromDb explainsthe Bb of b. 13 and the
ascendingminorthird accountsfor E of the same bar. This Bb can be
explainedevenby G of b. 17:thisascendingminorthirdis filledout by G:
andA of bs 11-13;the E of b. 13is a minorthirdbelowG, filledout by the
F#'sof bs 15-16.Thereremainstheexplanation of D: in bs 12and15:it is an
appoggiatura to E, but its privilegedpositionin b. 12 defines,with F#
(doubtlessthe F# of bs 16 and 17),a secondary harmonicfield.
I find this explanation a littlecomplex,particularlybecauseit establishes
linksbetweennoteswhichareprettyfaraparton thesyntagm.Is it notrather
contrivedto go to b. 13to finda Bb to forma minorthirdwithDb when,as
Wilkinsonlatersays,the Db of b. 9 is a minorthirdabovethe Bb of b. 8?
Thesedifficulties canbe avoidedbyexplaining bs S17 astheexpansion of the
twothirdsfromthe firstfive bars:
E>c.39
bS 1-5 , bS 6-9 , bS 9-17

Comingbackto theprincipleof slidingby semitones(usedforthemonodic


analysis),the rhythmicandmelodicmovementemphasises notesof harmonic
importance: thiswasseenin thefirstfivebarsandis confirmedby thelength
of Bb in bs S7, theplayon Db in bs 9-10 (Cbeingits appoggiatura)
andthe
positionanddurationof E (bs 13-14)andG (b. 17).Giventhisframework,
VareseeffectssuccessivesemitonalslidesfromG-Db untilhe arrivesonce
againat Bb-E:

310 MUSICANALYSIS1: 3, 1982


o

VARESE'S'DENSITY21.5':A STUDY IN SEMIOLOGICAL


ANALYSIS

Ex.40
bs 6-10

rP w bo
bs 11-12
t bo ,

do v.
.

b. 12 ,
. ,

.. #°
7

L @

W° a.

t J

This explanationhas the advantage,in my opinion,of showinghow the


harmonic fieldfits,in accordarsce
witha chromatic displacement shownup by
mymonodicanalysis,intothetritoneambitprepared fromb. 2, confirmed in
b. 5 andreaffirmed in bs 11-13.Thetritoneculminatesatthesametimeasthe
firststageof the melodicprogression.
Bs 15-17 are cast in the intervalof a minorthird, E-G. If harmonic
explanationis requiredhere,the D# mustbe madean appoggiatura to E (b.
16)andF andF: considered elementswhichfill outthethird;in as muchas
this 'phrase'(bs 15-17) ends on G like the first section(bs 1-5), it is
interesting
thatthesetwonoteshavethesamefunction.Resumingtheideaof
a secondary harmonic field(D:-Ft) the F wouldbe interpreted, takinginto
accountthemelodico-rhythmic unfolding,asa suspensionof F# whichwould
itselfappear,assoonasit is heard,to bea leading-note,
orappoggiatura, to G.
The intervalof a tone betweenD: and F seemsaboveall to preparethat
betweenE: (thrownintoreliefby a breath)andG.
My explanatory hypothesisof slidingsemitonesis all the moreacceptable
sinceWilkinsonhasrecourseto it to explainwhatfollows:
Varesehas now establishedthe notes E C$G as centresfor bothmelodic
and harmonicactivity, and has consistentlyenlargedthe overallsound
areawhich this firstphrasecovers(the octavetransferencealso servesthis
purpose,of course).But in doingthis, he has automaticallytouchedupon
Bb as a centre(sinceBb is a minorthirddistantfromboth C# and G). He
does not use this note as a centre in the way we have alreadyseen,
however,but instead builds harmonicspheresfrom note centresplaced
symmetricallya minor second above or below Bb (on the notes BWand
AQ),againusing all the intervalsof the minorthirdwhichcan be derived
(B-D, B-G# and A-C, A-F#).

MUSICANALYSIS1: 3, 1982 311


O qs

NATTIEZ
JEAN-JACQUES

I shouldprefernot to invokethe distanceof a minorthirdfromCt, for


reasonsgivenabove,but ratherfromG (expansion by minorthirds):
Ex.41

X 1, bo t.. a (t,
bS 1B-23

"> _ "_ Q "t t_ "

Bars 18-23 establishthe first of these secondaryharmonicareas,and one


noticesthatVaresehas carefullyavoidedsoundingthe note BWin the first
phrase,to preparethis entry.
criteriawhichexplainthe
Note herethatit is the melodicand syntagmatic
harmoniccentreandsemitonalslide,andnot the reverse.
Bars18-19 outlineB-G$; bars2(}21 outlineB-D, andbars22-23 resolve
to Bb (or A#) whichconnects,by a minorthirdrelationship,with CtE in
the next bar. The transferenceof the BWin bar 23 an octave higher
separatesit from the body of the precedingpassage,or ratherrelatesit
only with the B$C$DW in the sameregister(bars2(}21).
Herethereis a problemin interpreting the score.In as muchas Vareseis
carefulto makethe accidentals explicitwheretheycouldbe ambiguous (for
examplethe Au of b. 19),it seemsdifficultto imaginethathe is following
of notationandthattheaccidental
classicalconventions beforethefirstB does
notalsoapplyto thetripletB in b. 23. If B# is retained,theanalogywithnotes
of the sameregisterin b. 20 worksevenbetter,andVaresedoes, as I have
shown,liketo enda melodicsegmentby a notea semitoneaboveits predeces-
sors.Onthishypothesis, Wilkinson's explanation of theG#andD of bs 25-28
remains valid,becauseof its posiiionin relaiionto thedoIxiinantB of bs 18-22:
Startingfrom this B, Varesejuxtaposesin bars 2F28 a new (and now
inverted)areaB-D-G: with a returnof the Ct-E relation,derivedfrom
the first five bars.
explanation
Wilkinson's as follows:
is represented
Ex.42
bs 18-23 b. 23

* bS 24-25 , bS 25-28

($A" A o t0

312 MUSICANALYSIS1: 3, 1982


ANALYSIS
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL

Notethattheauthorplaysaroundwiththethirdsratherashe pleasessince,if
B reallyis thecentreof theambitD-G# (bs25-28)andin thiswayconforms
to the principlesustainedforbs 1-5, 9-17 and 18-21,the thirdC#-E is, on
theotherhand,abovetheA#. Thisshowshowthecriterionof thirdsarounda
pivot-notealternateswiththatof theexpansion of superposed thirds(already
usedin bs S9).
It is interestingto noticethatthe firstD (bar25) is obviouslyrelatedto the
B in 23, that the D in bar 26 is relatedboth to this B and to the C: (the
rhythmicexpressionhere is important),and the D in 27 to C: only
(becauseof octave transference);while the G: completesthe B-D-G:
relation,andfinally,in bar28 both D's appearandthe chromaticfillingin
of CtE is completedwith an Eb. This Eb also links the passagewith the
next, for it is relatedto the F: at the end of the firstphrase(bars15-16).
Bothresolveto G as harmoniccentre.The Eb serves,too, as a leadingnote
to the E of the G-SBb harmonicareawhich follows. Moreover,it has
been transferredan octaveup, and out of its normalregister,becauseit is
to influencethe harmonicmotiona little lateron (see bars3640).
If theideaof Ebfillingin C#-E (b. 28)is consistentwithothermomentsof
the analysis(bs 1-5, 15-17), it does not, however,have quite the same
meaning,since Eb is heardthreebarsafterthe last C#-E and one octave
higher Wilkinsonfeels,moreover, theneedto justifythis.Finally,it seems
difficultto establisha linkwiththe F# of b. 16on thepretextthatthesetwo
noteswerelinkedtogetheratthatpoint(eventhen,onlybythehypothesis of a
secondary harmonicfield).
Onthebasisof theideathatin b. 24 VaresetakestheE-C# of theopening,
thefollowinghypothesis maybe advanced: aftertheexpansion of minorthirds
whichcharacterised the first seventeenbars,then the slidingto B which
introduces,we note, a majorthirdbetweenG of b. 17 andB of b. 18, the
composer takesa newchainof thirds,startingfromthefirstthirdof thepiece,
butintroducing a majorthird.TheEbis thusa semitoneabovethelastnoteof
the passage,as areB# aboveB, G aboveF#, E aboveD# andDb aboveC:
Ei.43
bS1-17 sm b
j- bo bo
bS 18-23 \ $

10 X- °

b - o to

O1 o
MUSICANALYSIS1: 3, 1982
'S_ 4 \

JEAN-JACQUES
NATTIEZ

Notethatif oneallowsa B# atb. 23, it is possible,in Wilkinson's


perspective,
to seea relationship
betweenthisnoteandtheEbof b. 28, bothfinalnotesof a
phrase.
Bars 29-31 are particularlyinteresting,for the harmonicareasfluctuate
and resolve in a most complex manner. The basic area is G-LBb,
suggestedin the high and middleregistersof bars29-30. They are stated
completelyin the low registerafterthe octavetransferenceof the Gb (bars
3(}31).
I preferto takethe morestraightforwardbasisof the E andBb of b. 31 to
establishthezoneE-G-Bb, sinceit is notclearwhatmightsuggestthesetwo
notesin bs 29-30.
The middleregister,whilehelpingto bringaboutthis octavetransference,
alsooutlinesan harmonicmotionby sequenceof minorthirds(see also
barsS8), fromE: by wayof F: (bars29-30) to G: (bar31), andfromG:
to B4 with the A4 transferredan octavehigherto completethe minorthird
relation(FtA) begunwith the firsttwonotesof bar29. It is worthnoting
how this is impliedin the low register,at the beginningof bar31, by the
melodicarrangement andphrasemarkings.It is the firstentryof this
secondaryharmonicarea(A-Ft, A<).
Theobjectiveof thisdevelopment is to explainGSB andA of bs 31-32which
areoutsidethe zoneG-E-Bb. In orderto do this,a certainimportance must
be givento E# andF# of bs 29 and30. Obviously,it is theinsistenceof F#-A
of bs 32-35whichjustifiesthe privilegedextraction of thesetwo notesin b.
31. The overallexplanationis as follows:
Ex.44
bS 29-31

i e o t - bo
' b.29' \ b. 31

b. 29 , b. 31 4 \

> - \
bS 32-364

40

It mightbe noticedthatthezoneG-E-Bb fitsintotheprolongation of the


stackedthirdsfromtheopening,if thelatterhadnotdeviatedtowardsB in b.
18. As to G#-B-A which ends this passage,the explanationby play of
semitonesis preferredto the ratherfar-fetched
relationship
withE: andFt,

314 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

as the developmentalprocess of this flight is entirely comparableto the


tritonalslides in bs 11-13, unless this passageis seen as sketchingan alter-
nationbetweenE-G-Bb and G:-BD as possibleprolongationsbeyondE of
the stackedthirdsof the first seventeenbars:
Ex. 45
bS 1-5 bS 6-9 bS 9-17

j ' bo bo 44° n \
\ bS 17-23

=1$o o

bS 23-28

# (,3 o b-
o
bS 29-31

o
O bo
\

o ho t0 t o
l

b.32

4"
In this lasthypothesis,the permutationzone of bs 32-36, then the (; Eb of
bs 3S38, and the D of b. 38 to the returnof B in b. 40, are explainedthus:

Ex.46
bS 29-31

t0 0 bo

S 00 q ° (

bs 32-36

102v "
"s bS 36-3B

"s
"s S{w

b5 3 B -40

o b. b to
v

MUSICANALYSIS1: 3, 1982 315


NATTIEZ
JEAN-JACQUES

for thiswholepassageis somewhatdifferent:


explanation
Wilkinson's
Bars 32-35 offer a rhythmicelaborationof the new A-F: polarity, in
relationto the alreadyestablishedcentre,B. In bar36, C completesthe
new area, and the following passage (bars 3640) relates the interval
polaritySEb, in a similarway, to B. This Eb is derivedfromthe Eb as
leading-notein bar 28; the weakeningdynamics, the unusual octave
transferencewhich makesa majorsixth of a minorthird, and the resolu-
tion, in bar40, on to a B unrelatedto its own sphereat this moment,are
all intendedto lessen the impulseand to keep the harmonyfloating.
OnewonderswhyWilkinsonreturnsto the Ebof b. 28 to justifythe Eb of
b. 37, whenhe couldexplainit in termsof a literalminorthirdwithC, in the
samewaythathe explainsC in relationto A, ignoringthefactthatthe actual
intervalis a descendingmajorsixth.Note thatthe D of bs 38-40 is inter-
spersedwithrests.
The G-E-Bb relationsin bars4145 areclear,but the low DW(bar43) and
the rhythmicallyimportantAb (bar44) weakenthis sphere,while being
connectedbelatedlywith the B4 of bar40. C: actsas a leadingnote to the
next passage(bars4649), whichdevelopsrhythmicallythe B-D relation.
D maybe the resultof a broadening towardsthe low registerof the pattern
E#-G-E of bs 3s3 1. Butaboveall, Ex. 35 showstheintervallic patternto be
the sameas in bs 3 and4, whichintegratesAb intothe analysis.Thisis one
point amongothers whereit is necessaryto resortto an autonomous
melodicanalysisto explainone note. If C# reallyis a leadingnote to the
followingD, nothingpreventsus fromexplaining it bya minorthirdrelation-
shipwithBb accordingto the principleof expansion:the melodicdevelop-
mentof bs 4145, whoseanalogieswiththe beginningof thepieceareshown
above,is thusreinforced by a stackof minorthirds,startingfromthelowest,
whichis identicalto thatseenin bs 1-9. Notethatthesemitonal slide(Cl D)
takesus intothe secondzone,G#-B-D:
Ex.47
bS 1-9

t O O bo b
bs 41-45

o b0 b°

bS 46-SO

ro 40<
B-D:
of the relationship
Afterthe rhythmicdevelopment
The Ab in bar 50 is connectedwith this intervalrelation(=D, KG:), as

MUSICANALYSIS1: 3, 1982
316
VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

was the C with A4-F# in bar 36; but the Ab resolvesto AWand GW(bar
51). A becomespartof the A-F$ A-C relationin the high register(bars
51-55) andG is relatedto the E-Ct, E-G spherein the low register.The
transferenceof AWan octave higher in bar 55 underlinesthese rela-
tionships.The E: servesas a link betweenthe twoharmonicareas,sinceit
belongsby intervalpolarityto the lowerregister,but is in fact transferred
an octavehigher. It is worth noticingthe use Varesemakeshere of the
sequentialpattern(wholetone, semitone)of barsS8 for he now employs
it to establishstronglyand for the last time the E<$G sphere (bars
53-55), instead of creatinga link between two areasby a sequenceof
minorthirdpolarities.
This commentarymay be representedthus:
Ex.48
bs 1-9

@ O O bo bo
, b.51 bS 51-53
bS 4 6 5 0 , q +

bS 53- 56

F w °F
'n° bo

The E# has been explained, melodically,by paradigmaticanalogywith


b. 29. It seems possible, once again,to join the two harmoniczones to those
alreadyidentified:

Ex.4s
O
n °$ot s

X ' , ts F
/
/

) ° w t

The finalcadentialphraseneutralizesthe relationsestablishedduringthe


piece and coversthe full rangeof registers.It becomesa resolutionof all
centresand all polarities,and dissolvesthe tendenciestowardharmonic
motion.
This remarkis particularlyinterestingand goes well with the spirit of my

MUSICANALYSIS1: 3, 1982 317


NATTIEZ
JEAN-JACQUES

comments:apartfromthefactthatit is thefirsttimein thepiecethatVarese


useslowC, thisnoteis a majorthirdfromE in theprolongation'totheleft'of
my secondharmoniczone. Bb belongsto the firstzone, D and F# to the
second,C# to the firstandD#, E# andthe finalB to the second:
Es.50

j Obtobo4ffi}
'>, tl <, b_

, 8 \ I

, , AK . . . .

,'' ,' X , '

' 1, ' ° j

Thereis a conspicuous neutralisationof harmonicdivisionsintotwodistinct


zones.The firstzoneis madeup entirelyof minorthirds;the secondmixes
majorandminorthirds,whichexplainswhyit containsbothE andEb and
whyto go beyondEbbya majorthirdwouldeffecta returnto thefirstzone.If
zone2 wereprolongedby a minorthirdit wouldleadto F, thento the G#
alreadyheard.WithC-E andD-F#, whatI defineas a secondharmonic axis
is not 'pure'.Whythis mixtureof majorand minor? The octaveis divisible
intothreemajorthirdsandintofourminorthirds.By introducing semitonal
slidesin the successionof thirds,mixingmajor and minor thirds,Varese
createsa developmental processwhich avoidstouching on notes already
heard.Perhapshe is applyingin thispieceoneof his poieticprincipleswhich
Xenakisis saidto haveheardfromOdileVivier:'Varese'swishfor a spiral
scale,thatis, a cycleof fifthswhichwouldnotleadto a perfectoctave'(1971:
266). The explanationof the move to B as a polarnote in b. 18 should
doubtlessbe maintainedhere. Once the cycle of G is completed,Varese
inaugurates a newone,a semitonehigher,fromG#. Andoncethezonesof G,
thenB, areestablished, he constructsthedevelopment withrecoursealterna-
tivelyto one or the other.
It willhavebeennoticedthatmyattemptat harmonic explanationrelieson
theprincipleof development by thirdsproposedby Wilkinsonforbs S8 and
29-31;forall that,the principleof development by descendingor ascending
minorthirdsarounda pivot-notehasnot beenexcluded.Whythisexchange?
It mustbe quiteclearthatWilkinson's explanation,basedessentiallyon this
lastprinciple,hasthesamestatus as mine:thatis, an attemptto assemblethe
maximumnumber of phenomena around a common principle.My own

318 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY IN SEMIOLOGICAL

attemptsto joinupwithanotherpoieticdatum:thespiral.Re-reading
exegesis
thislastsectionin detail,it willbe noticedthatcertainnotesareincompletely,
or badly,integrated intothisexplanation andthatthe dooris openforother
hypotheses.
Throughout thissectionI hopeit is clear,in anycase,howthepoieticpoint
of view reorganises the neutralanalysis,but alsohow someof its elements
cannotbe integrated in the poieticexplanation andthusretaintheirrightto
autonomous existence.

ANALYSIS
VI- ESTHESIC

problem
1. Theesthesic
In the samewaythattherearetwomainapproaches to poietics,depending
uponwhetherone proceedsfromthe workto the poieticsor froma poietic
datumto thework,attemptsatperceptual analysismayequallybeclassifiedin
two categories.The first, inductiveas before,consistsof startingfromthe
musicaltextandpickingoutconfigurations to whicha perceptual is
pertinence
accordedon the basisof esthesichypotheses:
| work | g esthesics|

It is supposedthat a given pictureof the workcorresponds to what we


perceive.Thisis themethodfollowedby LeonardMeyerin thesecondpartof
Explaining Music(1973)andin a stimulating articleby ThomasFay(1971).
The othermethodwouldstartfroman externalesthesicdatum,thatis, a
documentwhichbearswitnessto perception-fromProust'spageson the
strategies
perceptual
sonataby Vinteuil,severalaspectsof his privileged may
be deduced- or an experimentperformedon listeners.As in the case of
poietics,esthesicanalysispracticedin this waywouldprobablyleadto the
modificationof someaspectsof a neutralanalysisundertaken first,or would
be projectedontothe work:
| work| l
or | < j esthesicdata|
| neutrallevel | )

In oneway,experimental esthesicanalysiscouldbe considered of


a validation
inductiveesthesicanalysis.
All this is conditionalbecausetherearefew, evenno convincingesthesic
analyseswhichtakethis direction(withthe possibleexceptionof workby
Frances(1958)andImberty(1979,1981),buttheseconcernthemorespecial-

MUSICANALYSIS1: 3, 1982 319


mf f mf p - S

JEAN-JACQUES
NATTIEZ

ised sectorof musicalsemantics).It is doubtlessbecausethe experimental


approach to worksfroma perceptual pointof viewis no morethanembryonic
that,mostoften,musicanalyseswhichareadvanced as perceptualoperatein
an lncuctlvemanner.
* @

Here,bothapproaches willbe considered.I feelthattheinterpretation


of the
workby a flautistconstitutesa testimonyto his or herperception of it. This
standpointmay be contested.I haveoften been reproached for proposing
'scoreanalysis',but the statuswe give to the scorein its musicalrealisation
doesnotappearto havebeenclearlyunderstood. In Westernmusic,it seems
absolutely obviousthatthescoreis thecomposer's meansof pinninghiswork
down;it alsoguarantees the identityof the workfromone performance to
another.Thescoreis, therefore,a symbolic factwhichis absolutelyessentialto
its transmission.
The performer maythusbe seenas playinganintermediary
but decisiverole betweenthe writtenscore the traceof the composer's
intentions andthelistener,andin thissenseis thefirstto perceivethework,
that is, to makea seriesof choiceson the basisof the composer'stext.32
Followingan enquiryof thatkind,I shallexaminethe analysisproposedby
JamesTenney,inductivethistime,whichhasalreadybeenreferredto in this
study.

2. Interpretation
asanesthesicdocument
Naturally,musicalsemiologydoesnot pretendto tell a performer howhe
mustplaya work.If it tookon a normative statusit wouldbe turningits back
on its scientificintentions.Semiologycancomeonlypostfestumto describe
whathashappenedandnotjustifyit. Sincethetechniques of theneutrallevel,
whenappliedto a monody,defineunits,theirresultis notwithoutinterestfor
thedefinitionof phrasing, aboveallwhenthereareno suchindications in the
score.Fromthisperspective, I havecomparedfourinterpretations
of 'Density
21.5'whichwillbe calledtheZoller,Gazzelloni, DebostandCraftversions.33
I shalldrawattentionto a fewpassageswheredifferences
betweenscoreand
performer,or betweenperformers,are most apparent.Lettersdesignate
precisepointsin the score.
(1) [1] and[2]:
Ex.51
[1] @ 02 >

@eg J 19&1 J l

(a) Gazzelloniignoresthe secondaryphrasingwhichdelineates[1] and


joinsall the notesup to the lastC: in a singlelegato phrase.

MUSICANALYSIS1: 3, 1982
320
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL
ANALYSIS

(b) ZollerandDebostmakeF# of [2] seemlongerthanit shouldbe.


(c) Debostmakesonlya lightbreakbetweenthe secondC# andthe G.
(2) [5] and[6]:
Ex.52
3_ @ 3_

XpZ J IS;. ;011


f fP

(d) DebostconnectsE of [6] to F# of [5] in the samephrase


(e) andinterpretsthef on C# of [6] as the signof a breakwhichisolates
C#-G fromwhatprecedes.Gazzelloni, on theotherhand,breaksthe
phrasebetweenF# andE, justasindicatedby mycaesurabetween[5]
and[6], but C#-G is linkedto the precedingE.
(3) [8]-[9]:
Ex.53
_3 (!) @ t]

6 j b<45>
P subito Cf

(f) in the Craftversion,[8] and [9] arelinkedlike [7], whichseemsto


underplay the differencebetweenthephrasingof [7]on the onehand
and[8]-[9] on the other.
(g) Gazzelloni slursG andBb of [8] and[9], probablyinfluencedby the
repetoon of G-Bb. The interestof the principleof deceptionlies
preciselyin showinghow Varesefrustratesthe melodicdynamism
withwhichwe arenowfamiliar:if one doesnot abideby the sluron
Bb-C ([9]),is thisnot a contradiction?
(4) [11]-[16]:
Ex.s4
[110 b 10 [120 Li b.11 0140 b 12 01i L160

e Zr r tS ,3,r t s 3 t < it

Gazzellonl +6 | r r tC: 32 . i::


MUSICANALYSIS1 : 3, 1982 321
JEAN-JACQUESNATTIEZ

Zo11er
@r tffltC1: tr?;3*=

Debost
-Xa
9 V
tb- -r r? rrr^ltf: t 1Ce
t 3t 1, j . . j 3 i 3J <

Here,the samemusicalfragmenthasbeencopiedfourtimeswiththe
phrasingof eachperformer. Althoughall 'respect'[11],theirchoices
thendiverge:DebostandCraftsplitthe musicup accordingto [12]:
Zoller,on the otherhand,isolatesthesecondC of b. 10. As forbs 11
and 12, we may say that the four versionsillustrateall possible
segmentations: Craftand Deboston D of b. 10, Gazzellonion G#,
Zolleron D of b. 11. This is a perfectillustration of the different
weightingof variablessinceeachof thesechoicesmeansattributing a
specificvalueto certainnotes:in CraftandDebostthelengthof D in
b. 10;in Gazzelloni,thisD is treatedasa pivot-notearoundwhichthe
lowandhighG#'sturn;in Zollerit is, onthecontrary, theimportance
givento the repetitionof G#-D whichseemsto determine the phras-
ing. This particularsectionof researchshowsclearlyhow different
partitionings of the neutrallevel describepotentialitiesof esthesic
pertinence.
(5) [24]:
Withthe exceptionof Debost,whomakesa slightbreakbetweenthe
loudB# of b. 20 andthe D of b. 21, thethreeotherflautistsjoinD to
B# as if the slurfrombs 19-20wereprolongeduntilb. 21. In making
this passage one single segment, I have opted for the same
interpretation. This is withoutdoubtbecauseof the trill (C#-B#)
whichdemandsto be followedby something,andalsobecauseof the
movefromf to ff whichprolongsthe crescendo begunon B.
(6) [25]-[27]:
Gazzelloniignoresthe threeslurswhichled us to distinguishthree
units.Note that,on a higherlevel,as we haveseen,thesethreeunits
forma whole,separatedfromthe previouscontextby a quaverrest
andfromtheensuingcontextby changeof register.Thehomogeneity
of the passageis guaranteed alsoby the playon the samethreenotes:
A, A#, B. We see, therefore,thateventhoughGazzellonidoes not
followthescoreto theletter(thefirsttwobarsshowclearlythatVarese
is capableof indicatingtwo levelsof phrasingif he wishes),he has
optedto outlinelargesections(cf. his phrasingof bs 1n11 and 12)
ratherthansmallsegments.
(7) [43]- [47]:
Gazzelloni proceedsin the samewaywithbs 31-34.WhileCraftjoins
E of [43] to the legatoof [42], Gazzelloni,allowinghimselfto be

322 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL
ANALYSIS

carriedby melodicdynamics,beginsthe legatoof b. 43 on the E. In


thesameway,[44]and[45]areincludedin a singlephraseaswith[46]
and[47],withoutundermining Varese'sphrase-marks whicharespe-
cificonlyin [45]and[47].Gazzelloni continuesin thisveinin b. 54by
joiningthe lastC of [75]to the slurof [76].Buton the otherhandC
andE of [81]willbe separated fromthe precedingD andfrom[82].
Debostusesthe samegesturefor [44]-[45]and[46]-[47].
(8) [38]:
Alltheperformers play[38]like[40],thatis witha slurupto E$t.But
whythenhasVaresetakenthetroubleto establish,fromthispointof
view,a differencebetweenbs 29 and30?
(9) [48]-[51]:
Herethereis the sameproblemas in [11]-[16]:the absenceof phras-
ing givesthe performers a certainfreedom.Zollerincludes[48]-[50]
in a singlephrase.Debost,justlikeCraft,anticipates, from[50],the
phrasingof [51].
Unlikethesetwo'ambiguous' passages,[64]-[69]areinterpreted in accord-
ancewiththe partitioning proposed.The importance of D, restsandbreath-
ing doubtlesscombineto suggestto the differentperformersa common
solution.
Withinthislist,twotypesof commentmaybediscerned: mistakespureand
simple,and divergences in interpretation on the basisof a matrixof possi-
bilitiesopenedup by thescore.It is difficult anddangerous to establish
the boundarybetweenthe two: Gazzellonihas his reasonsfor grouping
[25]-[26]together,forexample,andhis interpretation in generalrevealsthat
sucha choicereflectsan overallbias.
In expressingmy surpriseconcerningcertainchoices,for examplethe
identification of [38]and [40], I havealmostslippedfromdescriptivesemi-
ologyinto whatmightrightlybe calledmusiccriticism.But whichmusic-
ologistcanreallyclaimto capturethe 'spirit',the 'essence'of the work,for
examplethe so-calledprincipleof deceptionthatappearsto me to be fun-
damental? Thenagain,the followingpresupposition wouldhaveto allowed:
anyself-respecting interpretationmustreflectthe composer's intentions.
Influencedby the combinationof New Criticismin literatureand the
creativerenewalof theartisticavant-gardes, therehasappeared a newattitude
whichallowsthatthe musicalinterpreter, like the critic,has a rightto free
interpretationin everysenseof the word of the workshe tackles.But
whereis the truth,if it doesin factexist?Arenot the choicesin thismatter
orientedforthe mostpartby the spiritandthe tastesof the time?Therewas
littlenegativecriticismwhenBoulez'renewed' theinterpretation of theSacre,
turningit intosomething'lessRussian'thanwouldanAnsermetor a Marke-
vitch.On the otherhand,not everyoneagreeswiththe anti-romantic biasof
Boulez'sParsifal.If semiologymustabstainfrommakingjudgements, it is
not in the nameof a desirefor objectivitywhichone mightdenegrateas
'positivist'.Semiologicaldiscourse,likethatof criticismandallhumanactiv-

MUSICANALYSIS1: 3, 1982 323


JEAN-JACQUES
NATTIEZ

ity, is itselfa symbolicfactaccountable to a semiology.The neutrallevelof


analysisis thereonlyto facilitatethe comprehension of musicalphenomena
andto providea basisfor comparison.
Criticaljudgements whichone mightbe temptedto makeon the basisof
our analysisaboutone interpretation or anothermust also be tempered.
Phrasing,the issuehere,is onlyoneaspectof thevariables whichplaya part
in the performers' choices.Craft'sversionis perhapsthe most faithfulto
Varese'stext, andGazzellonisometimestakessurprising liberties,34
but the
latterinterpretationis perhapsthemostlivelyandlyricalby comparison with
Craft'sratherdullversion this saidsubjectively.
Theseremarkscallnot onlyfor a semiologyof musiccriticism,but for a
semiological studyof the variablesat workin an interpretation; this study
wouldconstituteanimportant linkin thechainof a methodology foresthesics
whichhasstillto be elaborated.

3. Aninductive esthesic
of'Density'
It is fortunatethat the composerand theoretician JamesTenneychose
'Density21.5' as one of the worksanalysedin his article'Temporal Gestalt
Perception in Music'(1980).This title speaksfor itselfregarding the orien-
tationof the article.In addition,the authorsystematically compareshis
analysiswithmy own.Therearethereforetworeasonsforexamining it here.
Tenney'sperceptual approach canbe summarised thus:Whenwe perceive
a pieceof music,its temporal'continuum' is dividedinto 'a hieratchically
orderednetworkof sounds,motives,phrases,passages,sections,movements,
etc.'. Tenneycallstheseperceptualunits 'temporalGestaltunits'or 'TGs'
(1980:205).A detailedstudyof allthetheoretical andmethodological implica-
tionsof Tenney'sdiversepropositions wouldrequireanentirearticle.Forour
purposes,the followingaspectswillbe retained:
(1) His objectiveis to 'predict [myemphasis]wherethe TG boundaries
willbeperceived' (:206).Thisreallyis, then,aninductive step,starting
fromhypothesesaboutmusicalperception.
(2) Tenneymainlytakesthe followingperceptual datumas a basis:'The
perceptual formation of TGsat anyhierarchical levelis determined by
a numberof factorsof cohesionandsegregation, themostimportant of
whichareproximityandsimilarity' (:208).
(3) Moreconcretely,andcallingon the ordinary experience of a listener,
he considersthatin a monodicpiecetemporal andpitch-class intervals
whicharegreaterthanthoseimmediately preceding orfollowingthem
createtheTGboundaries (:208-9).To thesetwocriteriaTenneyadds
the roleof dynamics.
(4) The modeltakesinto accountneither'harmonicrelationsbetween
pitchesor pitchclasses'nor'motivic/thematic relations'(:217).
(5) The authoris consciousof the factthateachvariablehas a specific
weight,but notesthe impossibility of providingan adequatemeasure

324 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL
ANALYSIS

of it: 'Wehaveno wayof knowing,a priori, therelativeimportanceof


oneparameter verSus another,
in its effectson TG-formation.As yet,
no clear principlehas been discoveredfor determiningwhat the
weightsshouldbe'(:211-12).Andherefers,notwithoutreason,to the
need for acousticpsycho-research. But his modelis dealtwith by
computer.Explicitly,his methodbecomesalgorhythmic. He hasthus
hadto neutralisethe problemof weightby simplytakingthe sumof
temporal,pitch-classanddynamicdistances.
The commentsTenneydevotesto his analysisandmine,andalso
thetablewherehe compares ourtwosegmentations, arequotedbelow.
Notethatcirclednumbersin hisanalysisdesignate thenumberof each
event(notesandrestsaredividedup individually):

The segmentationgiven by Nattiezfor this piece is shownin the lower


portion of Ex. 55, so that a direct, point-by-pointcomparisoncan be
made. Here the correlationsbetweenthe two partitioningsarequite close
especiallyat the clang- and sequence-levels althoughthe two are not
identical,of course,and the similaritiesdiminishat higherlevels. In fact,
some 81%of the clang-initiations in our results,and 85%of the sequence-
initiations(but only 44% of the segment-initiations)coincide with the
correspondingboundariesin Nattiez'ssegmentation.Thereareno coinci-
dences at any higher level. Some of the discrepanciesbetweenthe two
segmentationsare fairly trivial, as where one of two 'models' simply
interpolatesan extra clang-breakbetween two otherwise coincident
boundaries(as at elements8, 25, 54, 109, 117, 118, 140, 179, 224, 226,
233, and 241). A few differencesresultfromthe fact thatNattiezdoes not
prohibit one-componentTGs, as our model does. These occur in his
segmentationin the form of 'one-elementclangs'beginningat elements
109, 117, and 118, and as sequences containingonly a single clang,
beginningat elements22, 52, 74, and 97.
Even if we disregardsuch discrepanciesas these, however,there will
still remaina numberof placeswherethe two segmentationsdiffer. Some
of these probablyhave to do with the fact that neither harmonicnor
motivicfactorsare consideredby our algorithm.For example,the high-
level TG-initiationwhich Nattiez locatesat element 188 is clearlydeter-
minedby the factthatthe initialmotivicideaof the piecesuddenlyreturns
at this point, and a model which includedsome considerationof motivic
relationsmight well yield a resulthere morelike Nattiez's.On the other
hand, the strongelementof surprisethat this returnof the initialmotive
evokes in my perceptionof the piece suggeststhat this motivicfactoris
hereworkingverymuch 'againstthe grain'of most of the otherfactorsof
TG-organization, and that an importantpartof the musicaleffect of this
even in the piece dependson the fact that motiverecursat a pointwhich
would not otherwisebe a high-levelboundary.
After all of the foregoingreasonsfor the differencesbetweenthe two
segmentationshave been accountedfor, a few discrepancieswill remain
whichsuggestthatour weightingsmaynot be quite 'optimum'afterall, or
that they are simplydifferentfrom those unconsciouslyassumedby Nat-

MUSIC ANALYSIS1: 3, 1982 325


tft < P f ;J tJ ti Xt p izte

JEAN-JACQUES
NATTIEZ

tiez, or even thatsomeaspectof ouralgorithmmayneedrefining.Finally,


however,I must say that I think our segmentationrepresentsthe percep-
tual 'facts' here more accuratelythan Nattiez's at certainpoints. These
would include the clang-initiationsat elements 13, 20, and 75, and the
sequence-initiations(and perhapseven the segment-breaks)at 177 and
238 (Tenney1980:221).*

Ex.55
SECTIONS-
S EGMENTS:

SEQUENCES: 11 13 16
-
l ls 1, -7 1@ 1S 1t §7
Cl.ANGS
r r

;:79" ; A I ,s, X 1, I X

HIERARCHICAL j |R 13 4 5 6 7 |!
STRUCTURE I In m X ar

FROM NATTIEZ A B
(1975) lSt¢t


N
0}s 1 123 p>+
6
1s 1}& 1Sz
7
1j! 1{9
16
1nO t; te

S Jl'r-r pElLr rl'rr}, r^rrrillrStSl r;C;S$8!rtely lhltlf 5 ,¢ lirl Cir*v-1


qD 1,, '; 1a3 1Xs 1+5 :a; 1e7 j1t Iss lao
Sm 1s T ar

q to

6 $ , _ = i 4 G G J 1> | f9; 8 g 1'4 X5ito


, >; 1a3 1ns 125 12^ 127 ie
Lx 1IC liLx i
C

z
4 12 53

38 i3; 133 13+ 135 ¢ ; <v 138 139 #°

$ l t$ bS t tplptl ttt}p;7 litaJ4ts li tZrl -7Xmp¢ llFF4p tlfofr ¢[ t twfora

|Rq 30 13 |tR 153 13s t3S 13* 137 j3tt 139 40 |#

|R
1S 9x.t
8

* The Editoris gratefulto ProfessorMarthaHydeforpermissionto quotetext and


music (Ex. 55) extensivelyfrom the 70urnalof MusicTheoty,Vol. 24, No. 2.

326 MUSICANALYSIS1: 3, 1982


9g
6 LJ
,F,;
;7;*;
Pc,}il >;W=n:
' wi^3:#J":<^
| 'rX 751'4e@^:
* W f f f-Ic rf
t r<:
'rr-;e :>fl f,1l;
t
p ^ _ I .,,&p r-; , afS vf1g2
X | rC
;nit ir f8 | rf ;$
r
> ' XSg<
¢;r5E ,s
P PP t> M0. > f 0 , o

VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

, ;+ 153 v ¢5 ^ 1+7 1 1+? 05°

n 1 },
{51 15i 53 r {ss e 147 1s! 59 t ¢,

gr acl: ir: + r'r1s w_,,>,< 8 rlfl-fJ28t2->J t $8


s3 0s tss 156 15, Iss 59 1^4 1w? tha

x 7n 1

34ik

t q

2A A3 ;#

- 63 tS 165 g 167 1 toeo 6,9 1z°

16. ii4 i 166 67 161 ! --- -


t

10

Z5 26 A7

70 17> 173 17+ 75 17; 177 17g 279 g

71 172 173 17# 1.75 76 177 17g n


I 1KD ¢b
C

1sj 18a 0t It

180 1t, Jsa 1X3

Agent:G. Ricordi& Co. (London)


By arrangement.
g) ColfrancMusicPublishingCorporation-Canada.
Ltd.

MUSIC ANALYSIS1: 3, 1982 327


NATTIEZ
JEAN-JACQUES

Attheendof hiscommentary Tenneysays:'Finally,I mustsaythatI think


oursegmentation represents 'facts'heremoreaccurately
the perceptual than
Nattiez'sat certainpoints'(:221). It is the five contentiouspoints then
enumerated by Tenneywhichwill now be examinedin detail.It must be
emphasised firstof all thatthe authordoesnotappearto havetakenaccount,
in his criticism,of my methodological the distinction
premises,particularly
betweenthe neutraland the esthesic:my segmentation is not necessarily
supposedto be esthesically pertinent.

13and20 in Tenney:
(1and2) Elements [4] and[6] inNattiez
Onhis graph,Tenneyretainsonlyoneof thetwoalternative segmentationsI
haveproposed(cf. Ex. 1, B above).Nevertheless to notethat
it is interesting
Tenney'sdifferentcriteriafor the segmentationof Cit-Gareno less legiti-
mate.Ontheotherhand,I shouldfindit difficultnotto registerthereturnof
the melodico-rhythmicfigureof the openingin b. 4, in [5].

(3)Element 75in Tenney:[24] inNattiez


Fromanimmediate perceptual pointof view,theascending compound major
thirdleapandthef obviouslycreatea break.Nevertheless,the slurwhich
linksthetrillto B doesconstituteanargument fortheexistence,atthispoint,
of a unit. Ex. 17 shows,iough, howthis trillis outsidethe paradigm. The
analogyI haveindicatedwith[28]couldalsobe invokedto makeCit-B:-D
an independent unit. The confrontation hereof thesetwo analysesdemons-
tratesunequivocally howanalysisas symbolicfactdependson the respective
weightgivento differentvariables.

(4)Element 177in Tenney: [55] inNattiez


The divergencewith regardto [55] bringsto light the differencein our
approaches. [55]belongs,with[54]and[56],to a largersequence II. For
Tenney,onthecontrary, thisis thebeginning of a segment, a unithigherin the
author'shierarchy.What justifiesthis strongpartitioning? The contrast
betweenff andp and the lengthof the rest (crotchet)have'increased' the
weightof thebreakin thequantification of variables.It is herethatwe realise
thereis doubtlessno directrelationbetweenthe quantification of weightof
parameters andrealperception. Throughout thispassage,despite boththerest
and the dynamicchange,is not Varese'splayon permutation of the same
notes(D- Eb-Db)the dominantcriterionforthehomogeneity of [54]-[56]?
Onthecontrary, howis it possibleto missperceptively thereturnto theinitial
unitin [59]? Howcanthetworestsof thisunitbe strongerthantherepetition
of [60], to the extentthatTenney'selement59 is morecut off fromwhat
precedesthanhis element58?'It nowappearsthatsuchoptimumweightings
areslightlydifferentforeachpieceanalysed'(:212).Thereis everyreasonto
thinktheyvaryfromonepointto anotherin the samework.

328 MUSICANALYSIS1: 3, 1982


ANALYSIS
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL

(S)Element238 in Tenney:[75] in Nattiez


Here, his criterialead him to establisha huge breakon the fortissimoE.
Certainly,[75]is a unitof transition,butforthe samereasonsas in point(4)
we areunableto separateit fromwhatprecedes([71]-[74]) hencesegment
I at a higherlevel since[75]endsthe fallfromhighC to lowCi$: begunin
[72],delayedbut alsoreinforced by the tensionof the E#'sin [71],[73]and
[74](F enharmonically). The paradigm reveals,besides,the roleof the axis
EtE (a semitone lower),then the function of C# in relationto C a semitone
higher ([71],[72],[73]and [74]) which thus attaches[75]to thefourpreceding
units.
Whatmaybe learnedfromthis confrontation of approaches? Thereis no
doubtthatTenney'sbasiccriteriaareperceptual. Buttheneutralisation of the
relativeweightof variables andthenon-consideration of motivicrelationships
do not permitacceptance of his segmentation as moreesthesically pertinent
thanmine.Ourtwoapproaches havea crucialpointin common:clarification,
Buttheydivergeon a no lessdecisiveepistemological question.Tenneytakes
the 'objectivity' of the informational approachas a basis,whereasI feel it
necessaryto adopta semiologicalapproach:perceivingis a symbolicact,
exactlylike musicanalysis.The idea of a blockwith whichI began,less
rigorousbut moresensible,turnsout to accountmoreadequatelyfor the
changeable characterof the hierarchyof variablesin the courseof a piece.
Paradigmatisation takeson theflexibilityof thesymbolic.Fromnowon, themost
important thingis not to carryout ananalysisby computer,butto makethe
analytical criteriaexplicit notexcludingotherswhichmightbe possible
which account forthe symboliccharacter bothof perception andof analysis.
Whatof aninductiveesthesic analysis? Stillpossible,startingfroma neutral
analysis,but only once general perceptual strategies are better known
(throughexperimentation): in this way we shall be able to interpretthe data
fromthe neutralanalysis more adequately.

VII- COMPARISON OFANALYSES


Theneutrallevelis notsimplya fundamental datumforpoieticandesthesic
approaches. In as muchas musicanalysisis, itself,a symbolicactivity,it is
epistemologically crucialto comparedifferentanalysesof the samework.35
Whatrighthasthe neutrallevelto serveas a basisforthiscomparison? It is
distinguished frommostcommonprocedures on the one handby its goalof
exhaustivity or, moreexactly,its refusalto look at thingsfroman a przorz
privileged pointof view,andon theotherhandbytheintegration intoits text
of the mostsearchingclarification possibleof the methodused;this hasthe
effectof combining theresultsof theanalysisitselfwitha levelof methodolog-
icalmetalanguage thatmaybe projectedontoit.
Froma fairlygeneralpointof view, I believeit is possibleto classifythe
analysesof 'Density'in two categories:

MUSIC ANALYSIS1: 3, 1982 329


JEAN-JACQUES
NATTIEZ

(1) If the circumstances of composition36andrealisation areset aside


allthisis moretheconcernof musichistorythanof analysis commentaries
aremostly'synthetic' andoftenquiteshort.Onrecordsleeves,forexample,it
is a questionof graspinga fewcharacteristic traitsof theworkwithoutbeing
able to go into detail:this is no reproach properlydone, such notes
presuppose a greatdealof insight.In additionto thetextswhichaccompanied
the performances examinedin the previoussection,I shallreviewthe pages
devotedto 'Density'by OdileVivierin herarticle'Innovations instrumentales
d'EdgardVarese'(1955:193)andin herbookon the composer(1973:114-
17),Halbreich's briefanalysisin theconversations of VareseandCharbonnier
(1970:150-51), and the remarksby MiltonBabbittin his article'Edgard
Varese:A few Observations on his Music'(1966:18). In the contextof an
'analysis of analyses',thesetextsareof greatinterestbecausetheyshowwhich
particular traitshavebeenprivileged in orderto capturethedominantcharac-
teristicsof 'Density'.
(2) The othertypeof analysisis the kindof detailedstudywhichfollows
theworkstepby step,justas I havedonehere.Thistypeofpublished analysis
is relativelyrareforreasonsgivenat the beginningof thisstudy.Apartfrom
Wilkinson's workalreadyquotedanddiscussed,thereis onlyoneotherarticle
on the wholepiece:'Versuchan VareseDensity21.5' by MartinGumbel
(1970).
Thestudyof diversecommentaries providesananswerto a questionwhich
is oftenasked:whatdoesananalysisin thesemiological perspective butone
mightjustas wellsayanymusicanalysiswhichis fairlyprecise tellus that
we did not alreadyknow?This questionthusconcernsthe cognitive value of
thistypeof approach.The analytical elementswillbe classifiedaccording to
somepointstackledby the differentauthors.

(1)Melodic progression

Vivier(1955:193and 1973:114):'It is a puremelodywhichgrowsaround


certainpivot-notes,castalternately in mirrorchromaticisms andverydisfunct
intervals'.
ForHalbreich,the pivot-noteof the openingis Ft: 'Everything is bornof
the firstbars:theintervalsbroadenprogressively eithersideof thepivotFi$:,
soundmovingoff in questof its rightfulpitch,andthenits dynamiclevel'
(1970:151).
When Viviertalks aboutmirrorchromaticisms, she is alludingto the
inversionsF-E-F: / Ft-G-F or B-At B / B:n:-Ci$:-B:n:. The words'very
disfunctintervals'referto all that is not conjunctchromaticism. Thereis
certainlya broadening of intervals,as Halbreich
says,anda conquestof wider
andwiderspaces(highnotes;maximum rangewiththeintroduction of Cin b.
56). Thereis thereforenothingwrongin all that.The attributeof a precise
taxonomicmethod,or, in Gumbel'scase, of a statisticalanalysis this
presentsdifficulties,but I shallreturnto the problemlater is to showthe

330 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

wherefore of thesegeneralstatements.In my analysisthe paradigms showall


thetransformations basedon [1],thatis, showtheprocessof melodicprogres-
sion;in Gumbel(1970:37) a diagramshowshow the progression in use of
dynamicsis parallelto the evolutionof the rangesused.
Canit be said,however,that'everything is bornof the firstbars'?Thisis
true only on the basisof the developments of the initialmotive,but the
techniqueof systematic of thedevelopment
clarification showsthatnotallthe
relationships betweenunitsaretransitive, whichHalbreich's phrase,if taken
literally,mightimply:he expressesan overallfeelingof expansionandprog-
ressionfromthebasisof theinitialcell. Finally,onemightaskwhytheinitial
pivot-noteshouldbe Ft: the paradigmatic tablesshow,on the contrary,the
importance firstof Ct, thenG.
In her book,Viviershowsspecificallyhow the progression is organised:
'The fundamentalstructure,the skeleton,is composedof a chromatic
ascent,alwaysunderlined by a longvalue,butwitha freedomgreaterthanin
the chromaticdescentof Octandre' (1973:114).Withthe aid of theselong
'prop'notes,the authorreconstructs the followingchromatic scale:
Ex.56

'G: is outof orderin thescale',shecontinues,'butappearsin a highregister


andis repeated in anarpeggiomovement aftertheheldD; it alsoprepares Dt,
whichis presentedonlyas a dottedcrotchet,but emphasised by a sforzando
and followedby a breath it thus reverberates longerin the ear'. The
missingB appearsonlyin b. 18. 'Alreadyin Octandre', she adds,'thenote
missingfromthetwelvesoundsmakesits absencefelt'.Notethatthisprogres-
sionis describedonlyup to b. 23. Shortanalysessometimesgivetheillusion
this generalcomplaintis not directedparticularly at Vivier of having
capturedthe 'essence'of thepiecein describing animportant moment,while
the worktakenas a wholepresentsothermodesof progression.If, in a
two-pagetext whichdoes not claimto be exhaustive,Vivier'sanalysisis
adequate,it is becausethe passagedescribedis symptomatic of a general
tendencyto ascent.Note also thatthe 'scale'she proposesis not quitethe
sameas mine. Perhapsinfluencedby serialism,she attemptsto showthe
presenceof twelvenotes, whereasI haveproduceda kind of mixedscale:
diatonicforbs 1-8 (G-A-Bb-C), thenchromatic up to b. 17. Ft, therefore,
didnotfigurein thebeginning: hadI inscribedit, it wouldhavehadthestatus
of an appoggiatura, or a leadingnoteto G.

(2)Rhythmicdiversity
Vivier:'Rhythmicvalues,of greatvariety,setshortnotesandlongheldnotes

MUSICANALYSIS1: 3, 1982 331


JEAN-JACQUES
NATTIEZ

in oppositionwitheachother,makingsubtleuse of the tripletswhichhave


beenfalselycalled"irrational values"' (1955:193;1973:115).
Halbreich: 'Whatsuppleness,whatvarietyin rhythmicinvention,brought
to life by the verybreathof the life-pulse'(1970:151).
Certainly, buttheimportant thingis to showhow.TherhythmictypologyI
haveproposedcomesin to completeor to fill outVivier'scommenton shorts
andlongs.It hasbeenseenhowthepoieticpointof viewcouldassigna place
to the triplets.

(3)Processes
ofvariation
Babbitt:
Thereare, I believe,no two identicalmeasuresin Density.The durational
successionassociatedwith the attackpoints of the initial three pitches
occurs,in the samemetricalorientation,only at two furtherplacesin the
work,andat thoseplacesis associatedwith the openingintervalsuccession
also, but the pitch successionis altered in each case by transposition
(1966:18).
A preciseremark,anda correctone, whichtaxonomic description
completes
fortherestof thepiece.In his article,Babbittis concerned
simplyto capture
Varese'sstylistictendencyto diversity.He adds:
Varese is one of those composers . . . whose music has necessarily
directedour attentionto the inadequaciesof our analyticalconceptswith
regardto rhythm,by decreasingcompositionalrhythmicredundancy,by
increasingthe numberof rhythmicconfiguratons, and the dimensionsin
which these configurationsare made to appear(1966:19).
Hopefully,typological
classification
of rhythmshasaccountedessentially
for
thisrhythmicspecificity,whicheludesthe usualanalytical
techniques.

(4)Registerandthe'polyphonic'
aspectofthepiece
Vivier:'Theuseof thedifferentregistersof thefluteis remarkable because
theyarecombinedwithdifferentmodesof intensityanddynamiclevels.In
certainpassagesthe modeof attackandthe dynamicchangeon everynote.
An echoeffect,or, moreprecisely,a feelingof expansionandreliefbetween
distancedplanes is createdby changesof registerlinked to opposing
dynamics:a highregisterfortissimois succeededby a mediumregisterpiano
subito,or againa medium-lowregister"enfle"forteis followedbya highnote,
pianosubito,repeatedthreetimesafterornamentation. Severalinstruments
seemto be answeringone another,severalinstruments,not severalflutes,
sincecertainpercussiveeffectsgo beyondthe soundworldwhichwe might
customarily haveexpectedfroma flute'(1955:193;1973:115).
Halbreich: 'Withtheaidof thisuniqueinstrument, Vareseconquersa new
soundspace,combining oppositionsof register,dynamicsandagogicsin such

332 MUSICANALYSIS1: 3, 1982


ANALYSIS
VARESE'S'DENSITY21.5': A STUDY IN SEMIOLOGICAL

answering
a wayasto givetheillusionof severalinstruments oneanother.The
percussiveeffectsof bs 2F28, often imitatedsince, open the way for a
genuinefluteequivalentto violinpizzicato'(1970:151).
ThelinkVivierestablishes by
betweenregistersanddynamicsis illustrated
two examples.It thereforeconstitutesboth the expressionof a dominant
impressionon the partof the analystand a hypothesisfor study:it would
sufficeto comparetwoseparatecolumns(forthedefinitionof registerandthe
enumeration of dynamiclevels)whichwe wouldaddto ourgeneraltable,in
orderto verifyup to whatpointthis correlationis correct.

(S)Formalorganisation
In Vivier'sbookthepieceis dividedintothreeparts:thefirstgoesto b. 23,
that is until the appearance of B has completedthe first elevensounds
introducedby long values.Bs 2F28 (percussiveuse of keys)constitutea
'centralinterlude'.Then 'threeveryfast arpeggiosannouncethe returnto
normalbreathing in a melodywhichbecomesmoreandmoreasceticwithits
wideintervals.A modifiedrepriseof the initialelement(b. 41) bringsin the
finalpart, particularly disjunct,whichends its development in ascending
movementovernearlythreeoctavesin a crescendo tofortissimo'(1973:115).
ForHalbreich,therearethreeperiods,'thefirsttwoof whichareseparated
by the strange"percussive" interludeof bs 2X28. The second(bs 2940),
intenselyexultant,throwsout an unutterable summonsto night,flamesof
light rising againstthe temptationsof despair.The third, a cumulative
synthesis,typicalof Varese,takes the quintessenceof the first two and
culminates,onceagain,in a desperate fortissimo in the upperextreme of the
register'(1970:151).
The authorsagreeon the essenfialdivisioninto threepartsand on the
intermediary roleof the percussivesection.My ownanalysiscoincideswith
theseviews.
MartinGumbelpresentsa quitedifferentbut subtlepointof view. His
objectiveis to showthattraditional analysiscannotaccountforthispieceand
that the statisticalapproachpicks out a more fundamental aspect. 'The
questionwill not be addressed herewhetherthe use of inadequate analytical
metilods can determine, iniluenceor even quite falsifythe of
result ananalysis
in an unreliable way' (1970:31). One might ask alsowhether the traditional
formalanalysisGumbelproposesis nota littlecontrived (1970:31-32).Thisis
the overallsegmentation he proposes:

MUSICANALYSIS1: 3, 1982
JEAN-JACQUESNATTIEZ

Parts Formal Bars Function Overallform


sections

Part I
1 a 1 Exposition 0 A
2 a1 2, 337 Evolution, Exposition
variation
3 b 6, 1 Contrast 5
transformation I
4 a1 9, 1 Exposition A
5 b1 15, 1 Evolution, Evolution
* .

varlatlon varlatlon
6 b1 19, 4 Contrast
transformation
7 24, 1 ) B
Contrast
Transformation

Part II

8 a2 29, 1 Exposition(?) i A
9 a2' 32, 3 Variation(?)
Evolution(?)
10 b2 36, 2 Transformation(?)
11 a3 41, 4 Exposition ) A1'
12 a3' 46, 1 Variation(?) 5 Evolution
13 b3 50, 3 Transformation(?)) Variation(?)
14 53, 1 j B1
5 Contrastand
(?)
8 transformation

Gumbeldiscoversan expositionclassicallydividedinto three,accordingto the


pattern a-a'-b. The first large section of his Part 1 ends on C of b. 8:
everythingdependson the principleadoptedfor segmentationandthe chosen
privilegedvariable.For me, the first five barsformeda whole becauseof the
stagnationon G. Gumbelprivilegesthe returnof the initial motif in b. 9.
What appears more difficult to sustain is the parallelismhe establishes
between Parts 1-3 and F6 and for the entire piece between A-A-B and
A'-A1'-Bl. The author'sbasis is essentiallythe returnsof the initialmotive.
As this is not supportedby a detailedanalysisof the relationshipsbetween
units, it is obvious that 'contrastand transformation'cannothave the same

334 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: A STUDY IN SEMIOLOGICAL
ANALYSIS

meaningfor b1andB or B1of sections6,7 and 14, etc. But he himselfdoubts


the pertinenceof this segmentation(see the numerousquestionmarks).He
considersthe pattern a-a'-b too simple and is not sure that the ostinato
sequencesin bs 32-35 or bs 4g50 aretransformations of the openingmotivic
material(1970: 32). Perhaps, in graftinga historicallydated patternonto
'Density', the authorhas taken a ratherfacile exampleto demonstratethe
legitimacyof the stylisticapproach.This does not meanthat the latteris not
pertinent,but the demonstrationwould have had more force if the segmen-
tationhad respectedthe 'naturalarticulations'of the musicaltext. Moreover,
Gumbel'sposition remainsambiguous:despite his misgivingsabout formal
partitioning,he considersthe divisioninto fourteenpartsas 'quitepracticable'
and it is in relation to theunitsthusdefinedthat he proceedsto makea certain
numberof statisticalcalculations.
If the functionalscope of his partitioningand the more or less precise
criteriawhich motivatedit are ignored, there remainsa frameworkwhose
legitimacymight be questioned:why, for example, does Part 5 cover bs
15-19?The rise to high G and the first appearanceof B aretotallyneglected.
It is also difficult to understandwhy Part 13 begins in the middle of b. 50.
This is relativelyserious, since a statisticalcalculation,like every type of
description,is alwaysrelativeto its given field. The partitioningin units is
fundamentalin thatit conditionsthe validityof all thatis saidaboutthe piece.
The principle of Gumbel'sanalysisis, in itself, perfectlylegitimate:he
draws up diagramsof the evolution of pitches, intervals, durationsand
dynamicsto show the progressionof these parametersand the correlation
betweenpitches, rangeand dynamics.His analysisshows 'a formof develop-
ment . . . accordingto a differentiatedand complexprocesswhichcannotbe
reducedto letters or verbaldenominations'(1970: 38). In pointingout the
differencebetweenconventionalsystems (like serialism,whichVareserejected)
and principles, Gumbel's analysis ties up with the words of Varese at a
conferencein Princetonin 1959: 'Form is a result, the result of a process.
Eachof my worksdeterminesits own form . . . My musiccannotbe put into
any traditionalmusical-box'(Charbonnier1970: 85).
We are now in a better position to evaluatethe scope of these different
analyses.They are rarelywrongin any literalsense;they simplydo not have
scientificstatus-nor do they claimthis becauseit wouldbe impossibleto
reconstructthe work from their proposedcharacterisations.38 If the ideas of
'mirrorchromaticisms'or 'verydisjunctintervals'areto be meaningfulto the
listener,the piece must alreadybe known. Whattaxonomicanalysisprovides
is not the overallconclusionswhich an intuitiveapproachmight often enable
us to pinpoint, but the wherefore of these conclusions.
It is possiblethatwe havehereone of the semiologicalcharacteristics which
distinguishesthe differentmetalanguagesof musicology:withoutdescription,
conclusionsare like words deprivedof their referents.Becausethey are too
general, these appraisalsbecome empty. This is why they are on record
sleeves, or directed on a relativelysimple level, at music lovers who have

MUSICANALYSIS1: 3, 1982 335


JEAN-JACQUES
NATTIEZ

personalknowledgeof a piece. The 'known'work is the objectof such


appraisals, whereasin anyscientificanalysisthereferentis nottheworkasit is
experienced,in amorphousform,but the workas it is organised,already
distinctfromits immediateperceptionthroughcharacterisaiions whichare
explicitat differentlevels.
Mycriticalobservations do notseekto discredittheworkof anyone,butI
am opposedto the mixtureof genres,and, whileit maybe obviousthata
studyis not addressedto record-buyers or interestedmusic-loversbuyinga
bookof information or to professional
musicians,epistemology nevertheless
has the right to a definitionof the differencesbetween'musico-graphic'
commentary andrigorousanalysis.Wehaveinsistentlyevaluated the levelof
truthof thegeneralcharacterisations examined,andsincetruththereis, letus
insiston the fact that they may very well mentionimportantfactswhich
escapedthe taxonomicdecomposition. In this sense,they mayofferhypo-
theseswhichare verifiableby systematicconfrontation with the detailed
analysis.
The epistemological point of view adoptedhere, and the semiological
perspectiveI have attemptedto illustrate,do not thereforeseek to make
watertight divisionsbetweendifferentmodesof analysis,butratherto suggest
a framework whichspecifiestheirrespectivescopesandmerits.Whilethere
maybe no uniqueandgloriousroadto theattainment of musicalknowledge,
it is stilltrueto saythatthe description, classification
anddistinctionof the
phenomena studied,likethetechniquesusedto accountforthem,allowsthe
introduction of orderandclaritywhereconfusionmayreign.This,in theend,
is the objectiveI am pursuing.I believethatits pedagogical valueis self-
evident.

REFERENCES
Babbitt,M., 1966:'EdgardVarese:a Few Observationson his Music',Perspectives of
New Music,Vol. 4, No. 2, pp. 14-22.
Charbonnier,G., 1970:Entretiens avecEdgardVarese,Paris, Belfond.
Deliege, C., 1975: 'Webern:Op. 10, No. 4; un theme d'analyseet de reflexion',
Revuede musicologie, Vol. 61, No. 1, pp. 91-112.
Fay, T., 1971: 'PerceivedHierarchicStructurein Languageand Music',3rournal of
MusicTheory,Vol. 15, Nos 1-2, pp. 112-37.
FrancesR., 1958:La perception de la musique,Paris, Vrin.
Gardin,J.C., 1974:Les analysesde discours,Neuchatel,Delachauxet Niestle.
Guertin, M., 1981: 'Differenceset similitudesdans les Preludes pour piano de
Debussy',Revuede musiquedesuniversites canadiennes, No. 2, pp. 5S83.
Gumbel, M., 1970: 'Versuchan VareseDensity 21.5', Zeitschrift fur MusiEtheorie,
Vol. 1, No. 1, pp. 31-38.
Halbreich,H., 1970: 'Etude de l'oeuvred'EdgardVarese',in Charbonnier,1970,
pp. 12147.
Herndon,M., 1974:'Analysis:Herdingof SacredCows?',Ethnomusicology, Vol. 18,
No. 2, pp. 21942.

336 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5': ANALYSIS
A STUDY IN SEMIOLOGICAL

Imberty,M., 1970:'Polysemieet coherencedu languagemusical- I: La polysemie


dansles reponsesverbalesassocieesa la musiqueet la constructiond'uneechelle
circulairedes expressivitesmusicales',Sciencesde l'Art, Vol. 7, Nos 1-2, pp.
75-90.
1979:Entendrela musique,Paris, Dunod.
du temps,Paris, Dunod.
1981:Les ecritures
Jolivet, H., 1973:Varese,Paris, Hachette.
Lidov, D., 1977:'Nattiez'sSemioticsof Music', TheCanadian3rournal of Researchin
Semiotics,Vol. 5, No. 2, pp. 13-54.
Meyer,L.B., 1973:ExplainingMusic,Chicago,UniversityofChicago.
Molino,J., 1975:'Faitmusicalet semiologiede la musique',Musiqueen3reu,No. 17,
pp. 3742.
1982:'Un discoursn'est pas vrai ou faux, c'est une constructionsymbolique',
L'Opinion,January8 & 15, 1982.
Nattiez, J.-J., 1973:'Quelquesproblemesde la semiologiefonctionnelle',Semiotica,
Vol.9.No.2,pp. 157-90.
1974a:'Sur les relationsentre sociologieet semiologiemusicales',International
Reviewof theAestheticsandSociologyof Music,Vol. 5, No. 1, pp. 61-75.
1974b:'Pourune definitionde la semiologie',Languages,No. 35, September,
pp. 3-14.
1975:Fondements. \
de la musique,Paris, 10/18(Englishtranslation
d'unese'miologie
.

n preparatlon).
1979: 'A propos de Schoenberg:les problemesde la constructiondu modele
poietiqueen semiologiemusicale',2nd Congressof the AssociationInternationale
de Se'miotique(Vienna,July 1-6), 'Actesdu Congres',to appear.
1982: 'Problemesde la poietiqueen semiologiemusicale:quelquesreflexionsa
proposdu "De NaturaSonorum"de BernardParmegiani',in L'atelierdu sonore,
Paris, Buchet-Chastel.
Nattiez,J.-J. andHirbour-Paquette,L., 1973: 'Analysemusicaleetsemiologie: a
proposdu Preludede Pelleas',Musiqueen3reu,No. 10, pp. 4249.
Naud, G., 1979:'Pourune methoded'analysedes analyses',Addressto 1st Congress
Internationale
of the Association (1974),in Chatman,Eco, Klinken-
de Semiotique
berg, eds: Panoramasemiotique, The Hague, Mouton,pp. 1015-18.
1975: 'Apercousd'une analysesemiologiquede NomosAlpha',Musiqueen 3reu,
No. 17, pp. 63-72.
Ouelette,F., 1966:EdgardVarese,Paris, Seghers.
Ruwet, N., 1972:Langage,musique,poe'sie,Paris, Seuil.
Stefani,G., 1974:'Surl'approchefonctionelledes pratiquesmusicales',International
Reviewof AestheticsandSociologyof Music,Vol. 5, No. 1, pp. 77-82.
1976:Introdazionealla SemioticadellaMusica,Palermo,Sellerio.
Tenney, J., 1980:'TemporalGestaltPerceptionin Music', Zournalof MusicTheory,
Vol. 24, No. 2, pp. 20541.
Varese,E., 1930:'E. Varesey la musicade vanguardia',interviewby JoseAndre,La
Nacion, BuenosAires, April 20.
1934: 'Vareseand ContemporaryMusic', Trend,Vol. 2, No. 3, May-June,
pp. 125 ff.
1959:Conferenceat Princeton,in Charbonnier,1970,pp. 83-86. Fullerversion
59, Vol. 1, No. 5, September-October,pp. 276-83.
in Liberte'
1983:Ecrzts,Paris, Bourgois(in press).

MUSICANALYSIS1: 3, 1982 337


JEAN-JACQUESNATTIEZ

Vivier, O., 1955: 'Innovationsinstrumentalesd'EdgardVarese', Revue musicale,


No. 226, pp. 188-96.
1973:Varese,Paris, Seuil.
Wilkinson,M., 1957: 'An Introductionto the Music of EdgardVarese',TheScore,
No. 19, pp. 5-18.
Xenakis, I., 1971:FormalizedMusic,Bloomington,IndianaUniversity.

NOTES
1. See, for example, Gilles Naud's article on NomosAlpha (1975) or Marcelle
Guertin'sstudy of Debussy'sPre'ludes (1981).
2. Perhapslike the passing on of the teachingsof Oliver Messiaenand Nadia
Boulanger,hardlyrecordedat all exceptin the memoryof thosewho heardthem.
3. On semiologicaltripartitioncf. Nattiez (1974a, 1974b, 1975), Naud (1975),
Molino(1975, 1982).
4. This analysiswas the subjectof severalseminarsin the M.Mus. Semiologycourse
at MontrealUniversityin 1974. I thank all the students, the membersof the
Groupede Recherches en Se'miologieMusicale,especiallyGillesNaud who gaveme
concretehelp at the beginningof this study, and colleaguesLouise Hirbour-
Paquetteand JeanMolinofor criticalcommentswhich contributedto modifying
the contentof this analysis.JamesTenney'sanalysispublishedin 1980, and the
fruitfulconversationswith him in June 1982, led me to extend SectionVI
devoted to esthesic analysis which now containshis analysis. Following a
suggestionby David Lidov(1977:44), I haveremovedfromthe text all reference
to Pike'sdistinctionbetween'etic'and 'emic'units. It is not thatthesetermshave
no place in musical semiology, simply that their presenceis, in this context,
superfluous.These two words deserve, furthermore,a profound conceptual
studywhichwill be undertakenelsewhere.The firsteditionof the presentstudy,
in French,was partof a projectin semiologicalmusicanalysissponsoredby the
ConseildesArtsdu Canada(No. S73-1826).
5. The scoreabovewill give an initialoverallview of the differentlevelsof segmen-
tation.
6. The numeral3 below the last semiquaverof [1] indicatesthat it belongs to a
triplet.
7. This is an exampleof the necessityunderlinedby Ruwet(1972: 114)for perform-
ing an analysisbothfrombottomto top andfromtop to bottom.The presentation
of the analysiscouldhavebegunby delineatingsegmentsI, II andIII on the basis
of three criteria:
(a) the similarityof the initialnotes of [1], [3] and [5],
(b) the identicalfinal notes of [2], [4] and [6]: C#-G, and
(c) the rest between[2] and [3].
8. The 'short'and 'long'valuesin this paradigmcan obviouslynot be put onto the
same footingas the shortsand longs of Ex. 2.
9. This can be seen in characteristicfashionin the openingof Integrales(cf. Nattiez
1975: 285-97). One additionalcomment:in FondementsI advocateseriation,
while this monographis devotedto a singlework. The exampleof rhythmictype
whichconcernsus hereclearlyshowshow one traitcannotbe consideredpeculiar
to a single work unless the field of worksstudiedis widened.It is obviousthat
here we have touchedon one of Varese'sstylistictraits.

338 MUSICANALYSIS1: 3, 1982


VARESE'S'DENSITY21.5: ANALYSIS
A STUDY 1N SEMIOLOGICAL

10. It is up to futureresearchto show whetherit is found elsewherein Varese.


11. Analysesof Le Sacredu Przntemps by Boulez, of the Preludea l'apres-midid'un
fauneby Austinandof Hungarianfolkmonodiesby Erdely,andsee Nattiez 1975.
12. Discussion of analyses of the Prelude to Pelleas (Nattiez-Paquette1973),
Debussy's Syrinx, beginning of Varese'sIntegralesand Brahms'sIntermezzo,
Op. 119, No. 3 (Nattiez 1975).
13. It is not my intentionto makea systematiccomparisonof musicalsemiologyand
the use of the computerin analysis.We aretalkinghereof a hypotheticalattempt
at automaticanalysisof monodiesalone. All of this section II is found in the
originaltext of my analysis(1975). Tenney'scomputeranalysisshowsevidently
that it is possible to ask the machineto segment other than on the basis of
paradigmaticassociation.But I believethat everythingsaidhere explainswhy it
was impossible(or at least difficult)to integratethe recognitionof identicalor
analagousmotivic units into his model. See particularly,below, the ideas of
block, amalgam,quasi-criteriaand mixture.
14. The rest of this paragraphis a summaryof verbalobservationsmade by Jean
Molino(May 1974).
15. For example,dynamicscould be addedhere.
16. On the otherhand, the paradigmatic connectionof [13]and [1] isolates[13]from
[12].
17. The sign + indicatesthat the intervalis compound.
18. This is perhapsan argumentin favourof Deliege (1975:93), who considersthe
tritoneto be a chromaticintervalbut does not explainwhy.
19. Two exceptions:in bs 57-59 this final'localisation'is, in fact, characteristic
and at the join [37]-[38] in compoundform.
20. This fact has been noted by musicologists.Cf. in particularVivier 1973: 114.
21. Cf. the extractsfromVivierandHalbreichquotedherein SectionVII. This is not
the case for Tenney, whose model does not pick up motivicanalogies.
22. In numbersof semitones:bs 12-13: 6 6 6 12 1 3; b. 16: 1 1++; bs 31-32: 6 10 3
10;b.44:733;bs58-61:46447226.
23. More simply, one may say that sectionC ends with the descendingmovement
A-F#-B while it had begun ([44]) by the ascendingmovementB - Ft A.
24. This processexplainswhy [55]is not dividedinto two unitsof two noteslike [57]
and [58].
25. Thatis, a unit of morethantwo noteswhichgoes fromthe longestto the shortest
value.
26. This unit has not been describedas a flight because,unlikethe casesmentioned
above, it is characterisedby a diminuendo.
27. Note that this table conformsto paradigmaticprinciples,since the syntagmatic
successionof all the types may be foundby readingeverythingfromrightto left
and from top to bottom.
28. Gino Stefani(1974:82) statedthathe took fromNattiez 1973the ideaof it being
necessaryto go from the materialstudiedto the functionand not the contrary.
LaterStefaniinsistedon the functionalaspectof analysis(1976).
29. We cometo the sameconclusionhereas does Imbertywhenhe projectssemantic
characteristics,obtained experimentally,onto sound material:'The traits of
musicalstructuresare,in the end, onlypertinentfor a givenfactor,althoughthey
may be presentin others'(1970:92).
30. Althoughcouchedin differentterms,this is the ideaof the 'conventionalmatrix'

MUSICANALYSIS1: 3, 1982
339
JEAN-JACQUES
NATTIEZ

expoundedby JeanMolinoas one of the elementsof poieticknowledge.


31. Varesewould doubtlessagree:'The role of creation,in everyart, is to reveala
new world, but the creativeact itself escapesanalysis.The composeris no wiser
thananyoneelse as to wherethe substanceof his workis comingfrom and it is
only as a craftsmanthathe can speakcoherentlyaboutit' (1959:283). But this is
no reasonto ignoreor to dismissthe poieticdimensionof worksas semiological
fact. It should simply be expectedthat poietic analysis(which, unlike neutral
analysis,dealswith processesand not structures)will encounterspecificdifficul-
ties andaboveall will not resembleneutralanalysis,if only becauseof the gapsin
its data. On this subjectsee Nattiez (1979) and (1982).
32. For more details on these problems see Nattiez (1975: Pt. 1, Ch. 5, and
pp. 109-17).
33. KarlheinzZoller,HMV, C 061-28950;SeverinoGazzelloni,Virgo,89836;Michel
Debost,Angel,S-36786;RobertCraft(thenameof the flautistis not mentioned),
Columbia,MG 31078.
34. For example, the two quaversof b. 13 are played as double-dottedquavers,
influenced,it would seem, by the long-shortrhythmsof the previousbar.
35. I have alreadydevoteda systematicstudy to this type in the Preludeto Pelleas
(Nattiez-Paquette1973).An identicaltendencyis comingto light in othersectors
of musicology(cf. Herndon1974).
36. For 'Density', see Ouelette(1966: 147-148) and Hilda Jolivet(1973: 109-11).
37. A figure after a commain Gumbel'stable indicatesthe division of the bar in
crotchets.
38. On this epistemologicalcriterionof analysis,cf. Gardin(1974: 107-14).

340 MUSICANALYSIS1: 3, 1982

You might also like