Sie sind auf Seite 1von 35

New Approaches to the Linear Analysis of Music

Author(s): Allen Forte

Source: Journal of the American Musicological Society, Vol. 41, No. 2 (Summer, 1988), pp.
Published by: University of California Press on behalf of the American Musicological Society
Stable URL: .
Accessed: 12/06/2014 21:08

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

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

University of California Press and American Musicological Society are collaborating with JSTOR to digitize,
preserve and extend access to Journal of the American Musicological Society.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
New Approachesto the LinearAnalysis
of Music

y "linearanalysis"I referto the broadspectrumof approachesto
the study of music--especiallyrefractoryor unusualmusic, such
as much of the music of the late nineteenth century and early
twentiethcentury--approacheswhich emphasizethe contributionof
large-scalehorizontalconfigurationsto musical form and structure
and which may place local harmonicsuccession,diminutions,and
other musicalcomponentsof smallerscale in a subsidiarycategory.
From the contemporaryperspective, linear analysis can be seen
developingas a majorarea of activity and interest among scholars
today, especiallytheorists. A numberof recent publicationsrepre-
sentingdiversepointsof view and interests,such as those by Baker,
Berry, Forte, Lewin, Meyer, Morgan, Rothstein, and Williamson,
provideamplewitnessto the centralityandimportanceof the topic. It
is this burgeoninginterest which the present article addresses,not
primarilyas a survey of what even now comprisesa rathercomplex
arrayof orientationsand motivations,but ratheras a presentationof
recent originalwork, which I hope will exemplifythe advantagesas
well as shortcomingsof linearapproachesto the study of music.
Of course,the fountainheadof this approachto analysisresidesin
the writingsof Heinrich Schenker,althoughno doubt that eminent
personagehas turned over many times in his final resting place in
responseto some of the uses to which his conceptshave been applied
posthumously (see Beach 1985, especially 288-93). Indeed, it is
difficult to pick up a recent professionaljournalor book without
finding an article which has musical illustrations that contain
Schenkerianbeams, slurs, and other notationalapparatus.
Although the linear graph is the hallmarkof the contemporary
linearapproach,no uniformityeitherof analyticalintentionor result
can yet be detectedin the literature,as we can see from such recent
analyticalapplicationsof graphingproceduresas those in Ayrey 1982
and Wilson 1984. And while new kindsof lineargraphsare servinga
diversityof analyticalends, olderand morestablemethodsare being

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

employedin new ways. Two recentexamplesof the latterare Forte

1985 andPuffett1986.Comparisonof thosestudieswith studiesbased
upon the more traditional approach exemplified in Salzer and
Schachter(1969)is instructive.
It is clear that the Schenkerianinfluence has become a basic
componentin the history of music theory in the twentieth century,
particularlyin the United States. However, the present article is
basedupon the positionthatthe uncriticalapplicationof Schenkerian
paradigmsto certain kinds of music has often led to poor results,
indeed, has misled scholars,in some cases quite seriously, and has
obscured important issues that require confrontation(see Straus
1987).This circumstanceis the skewedobverseof the one which is all
too familiarto battle-scarredSchenkerians,the situationin which one
sees the poorapplicationof Schenkerianmethodto musicfor which it
was actuallyintendedby its founder.
Among the major issues which the present study raises is the
conflictbetweenthe specialcharacteristics of the individualworkand
the generalmodelof linearstructuresofferedby Schenkeriantheory,
hardlya new dish on the menuof music-theoretical controversy.With
respect to the presentationoffered here, I believe not only that
Schenkerianparadigmsof linearmotion-for example,linearprogres-
sions-are often inadequateas componentsof theoriesof individual
works that lie outside the mainstreamof tonal music, but also that
when they areappliedmechanicallythey may actuallyobscurehighly
significantaspectsof those compositions.
At the sametime, I remainconvincedthatthe generalSchenkerian
concept of structurallevels remains valid for and perhaps even
essentialto the developmentof effectivelinearmethods. Indeed, the
examplesfor this article,an odd assortmentof worksdrawnfromlate
nineteenth-centuryand early twentieth-centurymusic, are intended
to representthe applicationof this conceptin ways that, as the reader
will soon recognize, are eclectic, to say the least, drawing upon a
variety of sources, some of which are acknowledgedat appropriate
points in the exposition.
I begin with two short non-tonalpassagesfrom the music of a
prominentcomposerfor the ballet, proceedto excerptsfrom a long
tonalcompositionby a well-knownnineteenth-centuryGermancom-
poserof operas,and end with a workby an exotic and eroticRussian
composer, a work which representsa transitionalstage between
end of the articleI will offersome theoreticalobservationsalongwith
guidelinesto linearanalysisthat I hope will have been convincingly

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

illustrated.Althoughsyntacticrulesthatgovernthe constructionof
the analyticalgraphsmay be inferredfromthe analyses,I will not
attemptto formalizeanalytical
procedure in thisarticle,butpostpone
thatsubjectto a subsequentpublication.

Rite of Spring,Introduction
In orderto supplya historicalcontextfor my linearanalysisof the
openingof Stravinsky'sRiteofSpring,I wouldlike to spenda moment
on an earlierpublishedlinearanalysisof that music by Roy Travis.
The basis of Travis'sanalysis,which he makesquite explicit, is
providedby Schenker-derived linearand harmonicconceptsas trans-
mittedthroughthe writingsof the late Felix Salzer.Thus, for Travis,
the openingmusic of TheRiteof Springrepresentsa "prolongation" of
what he regardsas a "tonic"sonority(fromthe bass up) Ab-Db-C.
Examplei reproducesTravis'smaingraphof the excerptfrom The
Riteof Spring.The basic linearfeaturesof the analysisare extracted
and shown at the lower right of the graph as the succession of
descendingthirdsin descantand bass that prolongwhat Travis calls
a "tonic-sonority" from its initialstatementuntil its restatementand
confirmationat the end of the passage.
In one specific sense, this analysisis difficultto evaluate, since
Travis has alreadyreducedout a good deal of the melodic detail,
beginningwith the "e-minor"triadin the very firstmeasure,without
telling us why. However, the largeoutline of his analysisis clear:a
seriesof descendingthirdsprolongsthe upper-voicec"until it arrives
at d', at which point it proceedsby step to its finaldestination,c'.
In similarfashion, the two lower constituentsof Travis's"tonic-
sonority"areprolongedthrougha seriesof descendingthirds,arriving
at their destinationvia a finalstepwisemotion. These prolongational
motions in their entiretyform a "contrapuntal structure"in Travis's
There arises a basic question about this contrapuntalstructure,
namely:how does the substructureof the bassrelateto the rest of the
music?Put moresimply, why has the analystextractedthis particular
successionof thirdsfromthe overallchromaticdescentand not some
Beforeproceeding,I offera reminderto the effectthatthe workof
Pietervan den Toornhasprovidedbothimportanttheoreticalbasesas
well as convincingmusical evidence from which a new linear ap-
proach to Stravinsky'smusic may be attempted. Indeed, van den
Toorn's own analysesare, in large part, linear, althoughto charac-

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example I
Stravinsky, The Rite of Spring, Linear Graph (Travis)


C.S. C.S. ---

a contrapuntal st
fills the space of an o

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

terizethemasprimarily linearwouldbe inaccurate, sincehe is often

interestedin fixedverticalities andotherfeatures.1
Example incorporates some of van den Toorn'sideas in an
attempt to construct a convincing linearanalysisof theopeningmusic
of TheRiteofSpring.It assumesthatboththelinearstructures of larger
scale as well as the detailsof local motionrelatenot to a single
sonority,as in Travis'sanalysis,but derive from the dynamic
interaction of two underlyingconstructs: the octatonicoctad,pitch-
classset 8-28 andthe diatonicoctad,pitch-class set 8-23. Thesetwo
setsareshownatthetopof Example2 inpitch-class numericnotation,
with the circlednumbersin bothrepresenting the non-intersecting
pitchclasses,thosedistinctwithrespectto eachset.2
Introducingpitch-classset notationin this way providesan
opportunity to recordthefactthata gooddealof recentlinearanalysis
hasinvolvedpitch-class setsandrelationsamongsets. Studiesof this
type represent-for better or worse-a synthesisof Schenkerian and
pitch-classset analysisandwill no doubtcontinueto appearin our
professional publications forsometime.3(SeeBaker1983.)
It is customaryin contemporary theoreticalwritingsto regard
septad 7-35 as the diatonic pitch-classset. However,since the
octatonicoctad8-28 has beenassigneda majorreferential function
here,I haveincreasedthe sizeof 7-35 by one pitch-class,makingit
intothediatonic octad8-23 andthusequivalent in sizeto 8-28.4
As a resultof the approachto the music throughthese two
referentialcollections,the analysisis at once simplerand more
complicated: simplerin thatthebroadoutlinesof descantandbasscan

See van den Toorn 1983, Example27.

2 A pitch-class set is a collection of numbers representing a selection from the 12
notes of the chromatic scale. The number o is assigned to all notated forms of C, I to
all notated forms of C#, and so on. One form of the octatonic octad, 8-28, for
example, consists of all the numbers between o and I I, excluding the numbers o, 3,
6, and 9. Note that the name of the pitch-class set, 8-28, consists of two numbers, the
first of which gives the number of members in the set, the second of which is the
position of the set name on a list of all the octads (see Forte 1973, 179-81). The
pitch-class contents of 8-28 are listed in ascending order (by convention) as:
[o,1,3,4,6,7,9,I0]. See Forte 1973, 1-3-
3 In whatmayhavebeenan authentically propheticstatementJohnRahnhas
written:"Probably, analysesof the preserial
satisfactory worksof Stravinskywill,
whentheyfinallyappear,employtheoriesthatgraftnontonal referential
rulesintoa wildlySchenkerian-derived
of theory of pc set 'prolongation' in various pitch-structural and rhythmic-structural
'levels'." Rahn 1980, 79.
4 Pitch-class set 7-35, one form of which is the diatonic scale, occurs within only
three octads. Of these, 8-23 has the greatest number of fifths. It can be generated by
extending a cycle of fifths to eight terms.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example 2
Stravinsky, TheRite of Spring, Linear Graph
To (8-28): 0 1 3 4 6 7 9 10
Til (8-23): 11 0 1 2 4 6 7 9

(DO To(8-28) 4-23

/ 4-23 4-3 4-10


< To (8-28)& (8-23)/ T,, (8-23)

TI (8-28) complete

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

be described in terms of the octatonic domain; more complicated

because the details of motion reflect various modes of interaction
between diatonic and octatonic elements.
Thus, the initial gesture of the music (Example 2) is circumscribed
by the minor third c"-a'in the descant and by the first note of the bass
c#', both constituents of the octatonic set in its To form. However, set
4-17, beamed and labeled on Example 2, comprising the pitches c",e',
a', and c#', belongs both to 8-28 (octatonic) and to 8-23 (diatonic),
and in the later music of the ballet it performs a more extended linking
function based upon this dual identity. I should say here, in passing,
that all the tetrachords labeled on Example 2 are thematic objects in
The Rite of Spring. The prominent diatonic tetrachord 4-23, for
example, is familiar to everyone as the basis of the poignant four-note
theme of the "Mystic Circle of the Adolescents."
At Rehearsal I in Example 2 we find Travis's tonic surrogate,
comprising A6, Db, and C, as shown in Example i. But in the present
analysis this vertical formation assumes quite a different cast. The
descant c"is now the headnote for the motion that extends downward
through a' to f#', outlining the diminished-triad sonority often so
characteristic of octatonic harmony. At the same moment (at Re-
hearsal I) the bass begins its descent, describing the diminished-
seventh chord beamed on Example 2, which is clearly supported by
meter and rhythm. As indicated, the referential collection here is T1
of the octatonic set 8-28.
Turning now again to the upper voice of Example 2, I note that the
second arrival on f#' is followed by a motion that incorporates two
temporal levels of structure: first, the beamed form of diatonic 4-23
from T1 of 8-23 and within this motion two interlocking forms of
4-1o, the first from 8-28 (To), the second from 8-23 (T11).
Since tetrachord 4-1o enjoys octatonic as well as diatonic status,
the prolongational motion here may be described more simply as the
prolongation of an octatonic pitch-class, F#, by two diatonic
tetrachords, namely, 4-23 and 4-10. There is an obvious similarity
between this analysis and Travis's: both regard directly chromatic
motions as diminutional detail. But the interpretation of the more
basic structures underlying the direct chromatics differs radically
from the one reading to the other.
Other features of the graph in Example 2 can readily be observed,
and I will comment upon only one of those: the lower parts at the very
end. While the bass motion here, to TI of 8-28, and
Ab-B,-AI, belongs
thus represents a continuation of the preceding bass motion, the lower
voices together comprise diatonic set 4-23 and belong to its octadal

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Example 3
Stravinsky, Petrushka,"Chez Petrushka," Analytical Graph
Diatonic(T1of 8-23) 419 (11. 3, 6, 7)
+ A]
Octatonic(8-28) [E WholeTone/Octatonic516 5-16: (0, 1, 3, 4, 7)

5-16 (0, 3, 4, 6, 7) 7-31: (8. 10 11, 12, 4. 5) 4-z29: (0 4, 6, 7)

(11,0. 2. 3, 5, 6, 8) Octatonic

complement 8-23 in a new transposition. A very special pitch-class

motivic feature of the music is associated with this change: the upper
dyad of the configuration, d-eb-d, constitutes the first appearance of
a figure that pervades the work, most obviously as the principal
component of the ostinato pattern of the "Augurs of Spring" move-
ment that follows this introductory music.


Example 3, a graph of the opening music of the Second Tableau of

Petrushka,shows its linear as well as large-scaleharmonic constituents.
In succession, these units are octatonic at the outset, then diatonic,
beginning with the second f#" in the soprano, whole-tone and
octatonic, beginning with d#" in the soprano, and octatonic again,
beginning with g' in the soprano.
Each of these harmonic areas corresponds to a "gesture" in the
music, making segmentation straightforward. Moreover, the whole
passage is closed, in the specific sense that a pentad of the same class
begins and ends it, the octatonic pentad 5-16. Of these harmonic
areas, the most complex is the mixed whole-tone and octatonic.
Notice that the upper voice of that portion, beginning with d#", is
whole-tone, while the accompanying parts are octatonic.
I draw your attention now to the linear structures in the upper
voice and bass of Example 3. The notes comprising the bass are joined
by a beam that begins with c"in the "alto"voice. Beginning with e ',
the first note in the bass register, we hear the motion that drives
toward g at the end of the passage. Here the bass line in its entirety
includes two embellishing notes: d' as a lower neighboring tone
between the two "s, and e#, a passing note that connects e andf# of
eb, set. In this instance and elsewhere in Petrushka,
the ordered octatonic

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

the octatonic substructureis embellishedby chromaticallyrelated

pitch-classesthat lie outsidethe prevailingform of the octatonicset.
The descant of this passagedescribesa more complex motion.
Whilethe initialdescendingskipfromg"to c"is a specialgesturein the
passageand one that has multipleimplications,the stepwise contin-
uationof the descantreflectsthe integrationof diatonic,octatonic,and
whole-toneelementsalsoevidentin the largeharmonicgroupingsthat
I described above. Although the octatonic successiong"'-f#"-e"-e"
governs the immediateforeground,the pitch e" is understoodas a
passingtone, primarilybecauseof the bassarticulationd'-el' when the
In a similarway, the rhythmiccorrespondence of upperandlower
partsthen articulatesthe upper-voicewhole-tonedescentd#'-c#'-b',
the functionof the latteras the terminalnotemadeclearby the change
in rhythmicconfigurationthat follows. Again, the bass and accom-
panyinglower voices supportthe linearreadingof the descantat the
conclusionof the passage,so that b' over bassf# arriveson g' over
bassg. This kindof temporalcoincidenceis of the utmostimportance
to a linear reading in which the analyst cannot appeal to a priori
operativedeterminants,such as those representedby tonal voice-
The uppermostbeamon Example3 connectsthe primaryconstit-
uents of the descantline:g'-f#"-eb"-b'-g'. Thus, as indicatedon the
example, the line in its entirety projects tetrachord4-19, whose
affiliations are neither diatonic nor octatonic, but whole-tone.5
Counterpointedby the clearoctatonicbass line, this unexpectedand
apparentlyconflictingupper voice structuremay certainly be con-
strued as yet anothermusical representationof Petrushka'sunbal-
ancedpersonality,a depictionreinforcedby the low bassf#, which,
sounding against the "c-major triad" prepares for the famous
Petrushkachord (hexachord6-30) that follows. From the standpoint
of structuralanalysis,we acceptthis as the uppervoice at this level,
possibly subjectto furtherreduction,but certainlynot convertibleto
a Schenkerianlinearprogressionconnectingg" withg'!

5 Neither 8-23, the diatonic octad, nor 8-28, the octatonic octad, contains
tetrachord 4-19. Because the other large-scale harmony, one that has prominent
surface manifestations throughout Petrushka,is the whole-tone octad, 8-21, I under-
stand the linear projection of 4-19 here to refer to it rather than to one of a number
of other theoretically possible but quite "abstract"octads.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Wagner:Preludeto Tristan

I turn now to a famous work, familiar to all readers, a composition

which is perhaps the primary musical experiment of the mid-
nineteenth century, and one from which many a research expedition
has returned in considerable disarray over the years.
At the end of the operatic version of the Prelude to Tristan(mm.
io7-9)--as distinct from the concert version-Wagner has the cellos
and basses play a linear form of the Tristan chord: eb-B-F-AK.This
final reference to the famous chord in a linear projection is but a brief
reminiscence of several highly-charged linear projections that have
already occurred during the course of the prelude-forms of the
Tristan chord that seem to have gone virtually unnoticed by the many
analysts who have studied the Prelude over the years, including such
eminent persons as Paul Hindemith and Alfred Lorenz. I attribute
these oversights not to any unusual obstacles posed by the music, but
rather to the operative analytical approaches, of which the key-
oriented, vertical-harmonic may be the most unproductive. I shall
return soon to the linear role the Tristan chord plays in the Prelude as
a whole. Now, however, as a point of departure for my discussion of
the linear aspects of this music and in order to provide historical
background (with respect to contemporary analytical practice), I
would like to consider briefly the most thorough-going and best
known of the published analyses, the study of the Prelude to Tristan
by WilliamJ. Mitchell.6
To avoid digressing to discuss a number of details and issues that
may be interesting but are irrelevant to the central thrust of this
article, I concentrate on Mitchell's reading of the upper voice,
beginning with the section that extends from bar I through bar 17
(Example 4). He writes (p. 169):

. . the upper voice moves from the opening a to a" of bar 17. The
connectionbetween these two points is formed by the g#' of bar 3
makingan ascentin a stepwisemotionconsistingof four groupsof three
notes each until the terminala' is reached.

6 Mitchell
I967. AfterMitchell'sanalysisof the TristanPrelude,surelythe best
knowncontemporaryanalyticalstudy of that musicappearsin Boretz 1972. For an
exhaustiveand characteristially Gallic chase of the Tristanchord in supportof an
elaboratelyconditionalphilosophyof "relativism," see Nattiez 1985.The combatant
to enter this arenamost recently is Wagnerexpert Robert Bailey. Of particular
interestis his essay, "AnAnalyticalStudyof the Sketchesand Drafts,"Bailey 1985,

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example 4
Wagner, Prelude to Tristan, Linear Graph (Mitchell)

1 5 10

I\ \A T/ N

d2 f$2

Example 5
Wagner, Prelude to Tristan, Linear Graph, continued (Mitchell)
S- to c$, bar - -
17 20

-" f
'-'"r "

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

I have reproduced Mitchell's reading of the sequel to this opening

passage, bar 17 and forward, in Example 5. Mitchell writes (p. 182):

Bars i- 17have, in the structuralactivityof the upperparts,opened up

threeoctaveregisters,froma tog#' and, in bar 17, a' anda".Bars 17-24
carryout a first ascentfroma' to c#" by way of b', as indicated....

Now, for all the masterful technique that Mitchell brings to bear
upon this work and for all the trenchant observations he makes and
the interesting questions he raises, it is still possible to consider viable
alternative readings, readings which bring out certain features indig-
enous to this music that are missing from Mitchell's reading.
I will also venture to be critical of Mitchell's general approach and
point out that it represents an unreflective Schenkerian orthodoxy
insofar as the major constituents of the linear structures (represented
by open noteheads in the conventional way on the graphs, Examples
4 and 5) are directly referable to an assumed tonality of A
major/minor. Thus, he sees the large-scale descant of bars 1-17 as
governed by the octave relation between scale degree i, as a, the
opening pitch of the cello, and its counterpart, a", played by the first
violin in bar 17. In similar fashion, the primary goal of the large-scale
descant motion of bars 17-24 is c#"' in bar 24, the third of the major
tonic of the entire Prelude, in Mitchell's view (Example 5). This
mono-tonal orientation has been questioned by several authors, and I
now add my objection to theirs, for reasons which will become clear
in the sequel.
First, however, a brief comment or two on Benjamin Boretz's
lengthy and very influential study of the Prelude (and other parts of
the opera) would seem to be in order at this point. While it is true that
Boretz's approach is essentially linear, most convincingly evident in
the long linear graph of the entire Prelude which he presents in his
Example Ioa, his approach differs markedly in methodological ap-
proach from Mitchell's, primarily with respect to the structural
"background" of the music, which, in Boretz's view, is a non-
traditional construct (with respect to tonality) that he describes as a
"partitioning" of the total chromatic into mutually exclusive
tetrachords of the "diminished-seventh"type. Despite the apparently
radically different approach, Boretz's analysis shares with Mitchell's a
major flaw: it exhibits an overly rigid adherence to a background
model which determines virtually every aspect of the interpretation of
the foreground and thereby loses contact with certain important and

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Example 6a
Wagner, Prelude to Tristan, Linear Graph, bars i-17
1 6 9 16 17

A68 112(T3) A4690(TI)

t t t t t
AT A'T7 A A' A A&A

immediate musical apprehensions,most notably with the motivic

featuresof the music that are so essentialto both music and drama.
To illustratean alternativeapproach,one of the "newapproaches"
suggestedby the title of the presentessay, Example6 offersanalyses
of a series of five segmentsof music from the Preludeto Tristan.It
does not purportto be a completeanalysis,whateverthat might be,
but highlightslinearfeaturesof the music as they occur at what one
might designate, provisionally,as the middlegroundlevel. As we
proceed,it will be clearthatthe analysisgivesshortshriftto harmonic
progressionand key-matters which, in my view, have provedto be
obstaclesto the developmentof more informativeobservations,and
does not reflectseriousconcernfor Schenkerianparadigms,such as
linearprogressionsthat spanconsonantintervals.
The analyticalgraph, Example6a, representsa readingof the
opening section of the Prelude, but one that stops short of bar 17,
which Mitchellregardsas the goalof the openingmotion. Designated
by the Greekletterdelta (A)on the exampleis the Tristanchord, an
instanceof pitch-classset 4-27, the half-diminishedseventhchordof
traditional tonal harmony. Here and elsewhere in the example
symbols that immediatelyfollow A serve to modify it. Thus, A
followed by To symbolizesthe firstoccurrenceof the Tristanchord,
with To signifyingthe Tristanchordtransposedat level o, its original
pitch-classform. A prime(') afterA meansthatthe Tristanchordhas
been inverted,producinga sonorityone formof which is the familiar

7 Compare Boretz (1972, 162), where the two verticals are described as "exact,
balanced, simple inverses of one another." Although it is evident from the analytical
graph, I should point out that I regard the Tristan chord as a self-standing musical
object, not dependent for its meaning upon a resolution to some other sonority, such

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

I regard the "Tristan Chord" primarily as one possible form of

pitch-class set 4-27 for which any other form may serve as surrogate.
This view was certainly shared by Wagner's contemporaries, includ-
ing Liszt, whose cryptic introduction to the piano transcription of the
Liebestodmusic of Tristan und Isolde features two transpositionally-
related forms of A not only in a pitch-class form that differs from what
is conventionally regarded as that of the "Tristan Chord," but also in
a completely different vertical ordering.
Although the "dominant-seventh"here may display some func-
tional significance with respect to an implied tonality and therefore
demand attention from Roman-numeral addicts, I elect to place that
consideration in a secondary, even tertiary, position compared to the
most dynamic aspect of the opening music, which is clearly the
large-scale ascending motion that develops in the upper voice, in its
entirety a linear projection of the Tristan chord transposed to level 3,
Perhaps even more extraordinaryis the sonority at the conclusion
of the linear projection of A, above bass e. This can be construed as an
amalgamation of A, in the form just negotiated in linear fashion, and
A', the latter being the inversion of A as it occurs immediately after
the initial statement of the Tristan chord-the "E7chord" of bar 3-
to produce an extraordinarysynthesis of small-scale simultaneity and
large-scale linear succession. Although I am aware that this vertical
sonority may be read as a "dominant9th chord," I regard that label as
entirely secondary, perhaps even inconsequential, in relation to the
dynamic correspondence of horizontal and vertical which derives its
significance from features unique to this work, among which the
Tristan chord certainly deserves primary consideration as the most
intensively expressive musical symbol of the entire opera.

as the "French Sixth" which succeeds it. Judging from extensive musical evidence,
this was the view taken by many composers who quoted Wagner's chord, often in
pristine form, a striking instance of which occurs in the Scriabin work discussed later
in this article.
8 See
Bailey (1985, I29 and 290), and Boretz (1972, 172, Example 5), where the
author displays the first linear projection of A as well as the second, F#-A-C-E, but
apparently failed to discover the subseqent occurrences, probably due to his
overriding commitment to a particular "model"of the large-scale structure, namely,
". .. a simple partitioning of the (twelve) pitch-class 'octave' by the (o 3 6 9) construct
and its complementary mutually pitch-class exclusive transpositions (i 4 7 io) and (2
5 8 11). . ." (p. I72).

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

The linearanalysisin Example6b offersa readingof the passage

thatfollowsthe openingportionof the Prelude.9 Here,in contrastto
Mitchell'sreading(Example4) the higha"is viewednot so muchas
the completionof the openingsectionas the beginningof a new
section,representing a kindof dovetailingso characteristic of Wag-
ner'smusic.In general,thecorrespondence betweenthisreadingand
Mitchell'sis minimaland thereforedetailedcomparisonwould be
pointless.I drawattentionto two pitch-class motivicfeaturesof this
sectionapparently overlooked by Mitchell: firstto the flaggedbI'of
the "Glance" leitmotive,whichis directlyassociatedwith Tristan's
death,andsecondto the flaggede"',whichhasa "neapolitan" castto
it, but refers to the tritoneEL-A that is such an importantaxis
throughout opera.Moreover, hereis a pitch-specificreference
to thed#' in bar2 whichcompletes e,'
theopeningmelodicgestureof the
cellos. The musicis so rich in detailof this kindthat I need not
apologizeto thereadersof thisjournalforstayingcloseto thecentral
topicof this article.
Thus,to returnto thelinearanalysisof thispassage(Example 6b),
the eventof largestscaleis againthe projectionof A, the Tristan
chord,comprising thepitchesa"(barI7)-c#'(bar2)-f#'-d#' (bar23)-
This is given above the staves in pitch-class numerical notation as I 3
6 9 and indicatedas a transpositionof the originalA at level Io. This
formof A, Tio, sharesoneandonlyonepitch-class withthe original
form, and thatis by d#',
represented the terminalnote of the linear
of A anda pitchthatis exactlyin theregisterof theoriginal
d# of A, andis thus not only a pitch-classbut alsoa pitch-specific
Two factorsrenderthis secondlinearprojectionof A more
complexthantheinitialone:first,thelongunfoldingfroma"in bar17
to c#' in bar 22 and, second, the slow ascent from a to g' that
coincideswiththe firstappearanceof the "Glance"leitmotive,a line
which Wagnerscoresfor cello. This was a wonderfullyartistic
decision,sincethe headnotea of thatmotionis identicalto the first
note of the Preludebothwith respectto registeras well as timbre.
This motion incorporates the "Glance"leitmotive with its three-note
figure e'-e'-d', on which Tristan sings his final word, "Isolde," in act
3, scene 2 at the momenthe dies. Preciselyhere are two juxtaposed
forms of the Tristansonority,A followedby its inversion,A'.

For some aspects of this graph, in particularthe connection from a to g' in the
descant, symbolized by the Schenkerian dotted-slur notation (coupling), I am
indebted to Professor Stephen Hefling of Case Western Reserve University.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example 6b
Wagner, Prelude to Tristan, Linear Graph, bars I6-23
16 17 19 20 21
A24710(T1) A
A4690(T )


t t
A &A' A A'

Example 6c
Wagner, Prelude to Tristan, Linear Graph, bars 29-35
29 33 34

death A4690(TI)

0) Vn.I WW
Hn glance
L .
. +o,

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Together, these two sonorities have a motivic significance that

completely transcendstheir textbook harmonic role as secondary
Example6c is an analysisof an intricatesectionof the Prelude, a
sectionin which the descantjoinstwo large-scalelinearformsof A in
a singularway. As can be readfromthe example,A firstascendsfrom
f#' in first violin, with the "Death"leitmotive in the foreground,
then, with the "Glance"leitmotive in the foreground, continues
upwardfroma' throughc"to its completionon e"at bar 34--the latter
partof the motion carriedby woodwindsand horns.
Preciselyat bar 34 and coincidingexactlywith the completionof
the bass motionthat has ascendedfromD# to A, the secondform of
A begins on e". Obviously, this music repeatsthe earliermusic in
Example6b. But the contextdiffersconsiderably,becausethe formof
A thatbeginson e"at bar 34 hasno commitmentto another,overriding
formas it did earlier.The climacticfinalpitch of A hereis AI",which,
as I remarkedearlier,is symbolicallyassociatedwith Tristan'sdeath,
an associationthat provides a furthermotivic enhancementof this
To sum up, the descant representedon Example 6c joins two
forms of A, the first a transpositionof the original at level I, the
second a transpositionof the originalat level I I. At the risk of being
overly systematic, I point out that inverse-relatedvalues of the
transpositionoperatoralwaysproducethe samenumberof pitch-class
representativesin commonwith the transposedset--in this case none.
Therefore, of more immediate analytical interest is the relation
between the two forms of A which join at bar 34. The second is an
unorderedtranspositionof the first at level io, with the result that
they shareonly one pitch class, e",the last note of the first form and
the first note of the second. It is difficult to believe that this
correspondencebetween abstracttransformation(the transposition)
and the actualmusicallocationof the commonpitch class came about
merely by accident.
Example6d displaysan uncomplicatedlinearprojectionof A in its
level 3 transpositionwith respectto the originalform--that is to say,
in the same"unordered" pitch-classformas the firstlinearprojection,
shown in Example 6a. Here, however, the foregroundreflects the
leitmotivicdevelopmentthat has occurredin the interveningmusic of
the Prelude, for we find that the leitmotive "Desire"(the original
chromaticascentfromg#', labelleda on the examples)againconnects
the componentsof A as it did at the beginning,but, in addition,now
is embeddedin the descendingarpeggiationof the "Deliveranceby

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Wagner,Preludeto Tristan,LinearGraph,bars62-66
A 6 8 112(TO)
62 64 66


, " -L .
, - i~ -- - - M-- T. -

Death" leitmotive. The single stave example below Example 6d

providesa simplification.
Finally, in Example 6e we find perhaps the most remarkable
large-scalelinear manifestationof the Tristan chord in the entire
Prelude. Beginningfrom the verticalform of A in bar 79 (T7), the
descantgraduallyprojectsA in its originalform, arrivingin bar 81 on
the climacticpitch d,'"played by violin I. The second occurrenceof
this climacticpitchis then followedby its enharmonicequivalentg#',
two octaveslower, and a briefreturnto the openingmusic. This time
the descantdoes not projecta formof A, as shownin Example6a, but
rathera form of A'! In fact, with respectto pitch-classcontent this
form replicatesthe verticalformof A' in bar 3. In this extraordinary
way, the linearprojectionsof the Tristanchordin the finalsectionof
the Preludecorrespondexactlyto the initialverticalstatementsof that
harmony,a dramaticexpressionof the relationbetween verticaland
horizontaldimensionsanda specificinstanceof the correspondenceof
musical events of differenttemporalscale so characteristicof this
We have encounteredthe Tristanchordin variouslinearformsin
the excerptsdiscussedabove.Amongthese, the culminationis surely
the interlockingA and A' at the end of the Prelude displayed in
Example7. Althoughthe symmetricalarrangementproducedby the
interlockingof these two forms of the sonority can be viewed in a
numberof differentways, for example,to stressthe centrallocationof
set 4-28, the diminishedseventhchord, I draw attentionto the fact
that A' presents a successive intervalpattern that is precisely the
reverseof the patternfor A: the patternconsists of intervalclasses

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example 6e
Wagner, Prelude to Tristan, Linear Graph, bars 79-89
79 79 A3 5 8 11(To) 82
A'8 112 4(TJ)

t T t To To tTt
AtT7 tA' ATo At At AtT0 tA'T7 At

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

InterlockingFormsof the TristanChord
2-3-3 428

A' 3-3-2

Formsof the TristanChordand their IntervallicPatterns

19 25
33 2-3

f 3

A' 2

3-3-2for A' and intervalclasses2-3-3for A. This suggeststhata more

systematicconsiderationof orderrelationsmightshed additionallight
upon the large-scalelinearmanifestationsof A (and A') that pervade
the music, as well as upon its many poignantlyexpressivevertical
manifestations.Becauseanythingapproachinga complete survey of
this dimension of structurewould requirean extensive theoretical
expositionand would entaila digressionthat would consumeconsid-
erablymore spacethan is appropriateto the modestintentionsof the
presentarticle(the study would perhapsbe called "A Permutational
View of the TristanChord"),I restrictmyself to a few observations
which I hope will lend furthersupportto the analysisand which may
also be of interestto readerswho wish to pursuethe generaltopic of
orderrelationsin unusual19th-century music as they relateto linear
configurations(Examples8 throughio).
Example8 showsone of the manycorrespondences betweenforms
of A. At bar 19 the second part of the "Glance"leitmotivecarriesa
slightly concealedlinearversionof A, with linearintervallicsucces-
sion 2-5-3, as indicatedby the numbersbelow the staffwhich referto
the beamedopen noteheads.This portionof the motivealso contains
A' in the "silhouette"form representedby beamedclosed noteheads.
The two intertwinedforms are relatednot only as membersof the
same set-classtype, but also with respectto order, since A' as 3-5-2
presentsthe reverseof the linearintervallicsuccessionof A: 2-5-3.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Example 9
Wagner,Preludeto Tristan, of OpeningMusic

5 4 3
4 4
6- 6
6 6 4 4
4-27 4-25 4-25 4-27

At bar 25 (Example8) the leitmotive"The Love Philtre"begins,

and it also carriesa slightly embellishedform of A. Again, its linear
intervallicsuccessionis far from arbitrary;indeed, it is preciselythe
reverseof the pattern3-3-2associatedwith A in its finaloccurrencein
the Prelude,as shown in Example7. Further,the secondnote of the
leitmotive,d#', which is a pitch-specificreferenceto A in its original
form in bar 2, is set by a verticalform of A, as indicated,and the
arrangementof intervalsis such thatthe ascendingsuccession3-5-2is
created,thus effectingan intervallicassociationbetweenthis passage
and the one shown in Example8, in which the 3-5-2patternand its
retrogradeimagewerecombined,andprovidinga clearinstanceof the
transferenceof a horizontalstatement(bar i9) to a verticalone (bar
The appearanceof motivecafromthe openingmusic(Example6a)
as a distinct componentof the melodic leitmotivehere (Example8)
intensifiesthe associationof this musicwith the horizontalandvertical
formationsthat initiatethe Prelude.Indeed,at this junctureit will be
instructiveto returnto thatopeningmusicandto considersomeof the
orderrelationsthat it bringsinto play.
Example9 shows the verticalorderingof the four "slices"that
comprisethe openingphrasein its "chordal"aspect.A is connectedto
its inverse,A', by two passingchords,markedrr,the secondof which
is a reorderedtranspositionof the first. Specifically,two of the pitches
of the second, d' and e, are "ordered,"in the sense that they follow
registrallythe pitches from which they derive, while the other two
(g# and a#') exchange positions, producing a traditional voice

Following contemporary practice, the numbers in the interval successions
represent interval classes. Thus, in bar 25 of Example 8, the first interval above the
bass, formed by A andf#, spans 9 semitones. This number is by convention reduced
to its numerical inverse, modulo 12, which is 3. This corresponds to the traditional
equivalence of major sixth and minor third.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

exchange.As a result,the intervalsuccessionsof the two passing

chordsurlackany intervalsof class 2. In fact, that intervalclass,
togetherwithintervalclassi, is totallyunrepresented in the"vertical"
dimensionof this openingmusic,if directinterval-succession (adja-
cency) is taken to be the measure of relations. Thus, not only is the
progression from A to A' unified by virtue of the factthat thosetwo
sonoritiesareinversionally related,but alsothe uniform occurrence of
setsof class4-25 asvoice-leading productions bondsthetwotogether.
The basicquestionremainsconcerning the relationbetweenthe
verticalorderings of A andA' in theirinitialthematicmanifestations.
Table i summarizes the situation,showingthatthe circularpermu-
tationsof A (comprising fourof the possibletotalof 24)producefour
distinctpatternsof intervalsuccession,reduced,in the rightmost
column,to a normalized form,calledthe basicintervalpattern,in
whichthe numbersthatrepresentthe intervalsof the successionare
placedin ascendingorderwithoutintervening hyphens.WhenA' is
subjected to the same each
rotations, of the resultingintervalsucces-
sionsis a retrograde image of an interval successionof A. Thus, the
fourthrotationof A (takingthe topmostorderingof A as the first
rotation)will producea linearintervallicsuccessionthatis precisely
the reverseof thatof A', its inverseimage,andbothA andA' then
haveidenticalbasicintervalpatterns.This is not trueforanypairof
inversionally relatedformsof 4-27, butit is trueforanypairof such
formsif the intervallic orderingis correctlydeployed.Forexample,
undercircularpermutationthe followingpitch-classform of 4-27, 7
11ii5 2 (IT1o),will producethe same set of basic intervalpatternsas
The purposeof the finalexampledevotedto the TristanPrelude,
Exampleio, is to set out in termsas preciseas possiblethe relation
betweenthe verticalorderingof the TristansonorityA and linear
expressionsof that sonorityat two locationsin the music. This
exampleis intendedto beillustrative
andsuggestive onlyandis by no
meansofferedas a completeandsystematicstudyof suchrelationsin
the Prelude(not to mentionthe entireopera!).The first part of
Exampleio showsa two-stage"transformation" of A, beginningwith
the sonorityin its originalform.The firststage,shownin ordinary
music notationas well as in numericalnotationbelow the staff,
consistsof therotationof A (seeTableI) to createtheverticalinterval

I"The number of basic interval patterns varies according to the interval construc-
tion of the pitch-class set. Pitch-class set 4-27 has associated with it ten distinct basic
interval patterns, four of which are represented on Table I.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

of A andA', IntervalSuccessions,

A A'
Circular Intervals Bip Circular Intervals Bip
Permutation Permutation

5 II 3 8 6-4-5 456 4 8 2 II 4-6-3 346

II 3 8 5 4-5-3 345 8 2 11 4 6-3-5 356
3 8 5 II 5-3-6 356 2 II 4 8 3-5-4 345
8 5 II 3 3-6-4 346 II 4 8 2 5-4-6 456

succession4-5-3, symbolizedby the Greek letter rho (p) with sub-

script i designatingthe first rotation.In the second stage the new
orderingof A is transposeddown two semitonesto producea new
pitch-classform of the harmonywhile retainingits new ordering
4-5-3. It is this vertical ordering, 4-5-3, which is then projected
horizontallybeginningat bar 17, a configurationnotatedas beamed
open noteheadsaccompaniedby numbersthat representthe linear
intervalsuccession.The transformationis representedby the Greek
letter sigma(u).
The next transformation,representedby the Greek letter kappa
(K), is both more novel and more problematic.It compressesthe
previouslinearorderingcr registrally;that is, the horizontalconfig-
urationis verticalizedin register.The resultis startling,for now we
have the 2-3-3 orderingthat occursso often in the Prelude(Example
6d), most tellingly at the very end of the Prelude from bar 79
(Example7), symbolizedin Example io as the compoundtransfor-

Example io
The Tristan Chord in Vertical and Horizontal Configurations

2 17 22 23 79

A P1(A) To(A) o(A) K(A) o (T2(A))

6 44

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

mationcrof T2 of A.12


Scriabin'sFourthSonata,composedin 1903,is a workthatstands

at the veryedgeof the atonalprecipice,as we shallsee. Becausethe
musicis apt to be unfamiliar to manyreaders,I haveincludedthe
openingportion of the scoreas Exampleii. My analyticalgraphis
distributed overExamplesI2a throughI2d, whileTables2 through5
provide information intendedto clarifyspecificaspectsof thisunusual
music, which marks the endof Scriabin'stransitionalperiod.'3
The faint-of-heart maytakecomfort:I do not intendto offera
narrative accountof thegraph,butwillmerelypointoutcertainof its
salientfeatures,featuresgermaneto the topicat hand.Forexample,
I will not discussthe detailsof verticalorganization, althoughthey
begto beexamined.ButI willpointoutthatalthoughthepieceatfirst
seemsto havea tonalcastto it, furtherstudyrevealsthatthe vertical
sonoritiesaswellasotherdimensions belongprimarily to othermodes
of organization.Forexample,thefirstsection,upto andincludingbar
7, contains onlya singleandunadorned consonant triadas a vertical,
the F# triadoverbassa# at the endof bar4.
But illusorytonal harmoniesabound.For example,the chord
abovebassa in bar3, whichmayat firstappearto be a haplessF#
minorchord(withdoublesuspension),is actuallya formof atonal
pitch-classset4-19, asis theentiresonorityof bar5 abovebassd. The
presence 4-19 here,thehallmark
of of experimental non-tonalmusic,
is but one of manystylisticfeaturesthat determinethe historical
placementof the composition. At certainmomentsit appearsto look
backto the arcadiaof tonality,yet againand againits essentially
non-tonaldesignstronglypropelsit forwardtowardthe atonalworld
of the FifthSonata.

For an extensive, complex, and suggestive treatment of transformationsapplied
to pitch relations see Lewin 1982-83.
13 See Baker 1986, 195-202 for a discussion of the relation between this sonata and
the Fifth Sonata. As will be apparent from the graph, Example 8, my reading of the
form of the first movement differs from Baker's, but does not contradict his, which
takes into account the second movement (see Baker 1986, 196-97). In particular, I
regard the first 35 bars of the first movement as a miniature, self-contained sonata
form, the parts of which are clearly demarcated by differential forms of the octatonic
set, to be explained below.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Examplei i
Andante 63

~con voglia
ru to
A 4 4f. ,"I L I I 9 C

O?,p-" . , ,
'" -
fi' 91.I' /7
Y-9 T

My Examples I2a through I2d show that the linear analysis of the
descant is straightforward,with reduction determined by register and
voice membership as the principal analytical strategy. To illustrate,
the motion d#"-g#"-c#"' over the first three bars appears as the
uppermost linear strand on Example I2a, while the return to the inner
voice d#" and e#"in bar 3 is a momentary digression that prepares the
f#" of bar 4. As a further illustration of this procedure, consider bars
7 and 8 (Example I2a), where the descending leap from d#" to fx is
construed as a motion to the alto voice, with the continuity of the
uppermost voice then restored by the stepwise ascent to cx" on the
downbeat of bar 8. The graph interprets this chromatic note as a
lower neighboring tone to d#" of bar 7.
In order to simplify the discussion of this rathercomplicated series
of graphs, I will comment first upon the broad outline of the descant,
then say something about the melodic inner voice, and finally offer
general remarks on the bass line over the span of the excerpt.
The descant throughout consists of successive and interlocking
forms of tetrachord 4-2 3. In this context I regard 4-2 3 as essentially
non-tonal, since it does not operate within a traditional tonic-
dominant tonality, but instead interacts with another structure indig-
enous to this work which I will explain in a few moments. Still,
because of its luxuriant fifths 4-2 3 must be regarded as the archetypi-
cal diatonic tetrachord, and I will designate it as the representative of
that harmonic area without assigning it tonal significance in the
traditional sense.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Example 2
Scriabin, Fourth Sonata, First Movement, Linear Graph
a) bars 1-8
Exp 1 1368 81013 5 7 8

3 4-23:

T1(7-31): 11 10 2 8 1

b) bars 9-19
9 13 14 15 16
1368 81013 57100


11 10 2 4 5 10

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
c) bars 2 2-29

9 1124
Dv 22 23 26 27 28

1 3 4 6 7 9 10 (1 3 4 6
4 6 7 9100

To(7-31): 0 10 1 9 7 9

d) bars 29-35

29 32 35
8 1013


* 10) 0 2 3 5 6 8

0 8 6

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Forms of 4-2 3

o 2 5 7 4 6 9 II 8 10 I 3

3 6 8 5 7 1O o 9 2 4
2 4 7 9 6 8 I 3 5
II Oo0
3 5 8 Io 7 9 0 2 II I 4 6

Table 2 lists the twelve forms of 4-23, some of the special

properties of which Scriabin seems to have taken into account as he
composed this unusual movement. Observe that each pitch-class
integer appears in four forms of the set, with this pattern of
transposition levels, designating the first form in which it occurs as
level o: 0 2 5 2, a representation of trichord 3-7-
Even more interesting and implicative is the fact that pitch-class
set 4-23 contains trichords of classes 3-7 and 3-9 only. But, while any
single pitch-class representationof 3-7 occurs only within one form of
4-23, any representation of 3-9 occurs within two forms of 4-23. For
example, o 2 5 (3-7) occurs only within o 2 5 7, while o 5 7 (3-9) occurs
both within o 2 5 7 and within 5 7 10oo. Thus, 3-9 is a potential 'pivot'
between two forms of 4-23, a feature Scriabin utilizes to organize
linear progressions in the descant of this work. It is this pivotal
capability of trichord 3-9 that permits two forms of 4-23 to be nested,
as at the very beginning of the movement, where the initial trichord
d#"-g#"-c#"' is completed first by f#" at stratum 3 and then at bar 5
by a#" at stratum 2, creating the nested structure that sums to pc set
I introduced the terms stratum 2 and stratum 3 in the course of the
foregoing discussion. The graphs of Example 12 show those two
structural strataover the descant in its entirety and give the pitch-class
content of their constituents in numerical notation on small platforms
placed above the beams of the upper staves.
From Table 3, which summarizes the progression of the forms of
4-23, we can see that, for the most part, the forms of 4-23 are
combined at transpositions of fourths and fifths, given as Ts or T7 on
Table 3. The exception occurs at the beginning of what I have called
the Development (Dv), in bar 22 (Example 12c), where transposition
T9 effects a radical change in the pitch-class content of the descant, as
reflected by the many accidentals introduced in the notation.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Formsof 4-23 in Descantof Scriabin's

4-23 T Invar.

Theme I
Level 2: 8 Io I 3
Level 3: I 3 6 8 T517 [1,3,8]
Theme 2
Level 2: 5 7 o
Level 3: 0 2
5 7 T715 [5,7,0]

Development(m. 22) [2,7] w/r Theme 2

Part I T9
Level 2: 9 II 2 4
Level 3: 2 4 7 9 T517 [2,4,9]
Part 2
Level 2: 3 o0 (i 8)
Level 3: 3 5 8 io (T715) [3,8, Io]

Level 2: 8 1o I 3
Level 3: I 3 6 8 T517 [1,3,8]

The reader may have noticed that I have avoided specifying the
content of stratum i, the "Urlinie"of the excerpt. One very strong
candidate for this honor would be the single pitch a#", which is very
prominent in the upper voice throughout the excerpt and even more
so in the reprise of the short movement, which begins just where
Example I2d leaves off. I would then select d#" as the other
component of the Urlinie, for both pitches perform special roles in the
music. They serve as boundary elements, for example, in the first
section, and they link themes one and two, when a#" becomes b6"in
bar 14. In the return to the opening music at bar 35, both a#" and d#"
are fundamental, as can be seen from the graph, Example I2d.
The inner voice, the alto, if you will, makes a very subtle and
important contribution to the progression of the music. From the
standpoint of structural analysis, it provides a major clue to the other
fundamental constituent of this music, the octatonic set 8-28. From
the music-structural standpoint it provides a measure of connection
between the outer-voice linear strands in descant and bass. By "inner
voice" I mean specifically the ascending figures which appear at
regular temporal intervals throughout the excerpt: in bars 7-8 (Exam-
ple 12a), I5-I6 (Example I2b), I8-19 (Example 12b), 22-23 (Example
i2c), 26-27 (Example I2c), 28-29 (Example I2c), and 32 (Example

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Each of these figures is an ordered segment of one of the three

octatonic sets. For example, the first of these, beginning at bar 7 on
the graph in Example I2a, is an ordered segment of the octatonic set
that begins on pitch-class o. Here the figure begins with e#', the
upper note of the left-hand part of the piano, and extends upward to
cx". Again, in bar 18, the inner-voice figure begins on d' and ascends
through the octatonic scale to b', representing pitch-class i i. For
convenient reference, the pitch-class numbers of these octatonic
gestures are given between the staves on the graphs of Example 12.
That these octatonic figures in the inner voice are not merely
"fillers"or arbitrary gestures of some kind becomes evident when we
analyze the bass. We then understand that the inner-voice motions are
miniature representations of the large-scale underlying set that un-
folds in the bass register. Thus, the bass presents five pitches of the
octatonic set-unordered with respect to the conventional scalar ar-
ray-ending on c# at bar 7 (Example 2a). Precisely at that moment,
the inner voice begins its octatonic segment with e#' (pc 5) and fx'
and continues until it has brought in pitch-classes 8, io, i i, i, and 2
that have just appeared-in a different order, of course-in the bass.
In this way, the octatonic inner-voice figure, which "prolongs" the
descant pitch d#" amalgamatesthe large-scale bass from the octatonic
collection and the large-scale upper voice from the diatonic collection.
This process repeats throughout the excerpt, so that the seemingly
disparate fundamental harmonic components, the diatonic and the
octatonic, are literally joined.
At bar 22 (Example I2c) a new form of the octatonic set begins on
bass C, designated To. And finally, at bar 29 (Example I2d), the third
form of the octatonic set, T2, begins on bass D, coinciding exactly
with the onset of the final form of 4-2 3 in the descant, beginning with
the pivotal a#", just as the initial bass note of To in bar 22 coincided
exactly with the onset of a new form of 4-23.
Table 4 provides a summary of the octatonic set as manifested in
this intriguing and highly innovative composition. Here, instead of a
chronological arrangement, I have shown how the ordered forms in
the piece correspond to the referential, "background"forms, begin-
ning with the three statements of octatonic segments in the bass.
Statements of octatonic segments in the inner voice are shown in the
lower part of the example.
Table 4 shows that all three forms of the octatonic scale are
represented by linear configurations that span the entire excerpt. It
also shows that not all transpositions of the octatonic set appear to be

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions
Formsof 8-28 in Scriabin's

Reference Forms Ordered Forms Bar

Bas :
To: o 1 3 4 6 7 9 10 B
o :O 9 7 9 22
TI: I 2 4 5 7 8 iO Ii II O 2 8 I II 2 4 5 10 I
T2: 2 3 5 6 8 9 II o 2 o 6 8 29

Inner voice:
To: o 1 3 4 6 7 9 10 7 9 10o 15
4 6 7 9 10 0 22
1 3 4 6 7 9 10 26
4 5 7 8 io II 5 7 8 I
TI: i 2
1o II 2 7
2 4 5 7 8 io 8
T2: 2 3 5 6 8 9 II o o 2 3 5 6 8 32

equally represented. In particular, the form designated T2 seems

especially deprived with respect to its linear representations.
But this situation is remedied at the surface of the music in one of
its most startling aspects-no doubt already perceived by many
readers: the multiple occurrences of the Tristan chord.
Table 5 provides a list of the occurrences of that supercharged
sonority in this work. There are exactly six of these, five of which are
ordered vertically in exactly the same way as Wagner's chord. On the
graphs, Example 12, an eighth-note flag distinguishes them in their
various contexts-with the exception of the first, somewhat concealed
instance in bar 2, which has the same pitch-class content as the first
chord in the Tristan Prelude.
Interesting though these allusions may be, reflecting, no doubt,
Scriabin's interest in erotic symbols, their structural significance lies
in their relation to the fundamental octatonic domain, which is
expressed most completely in the bass. As Table 5 shows, three of the
six occurrences of the Tristan chord are derived from T2 of the
octatonic collection, the form I said was under-represented earlier in
the music. Most important, however, is the fact that the recurring
chord points up the integration of linear and vertical formations in this
work, a general consideration of the utmost significance to linearly
oriented analytical studies.
There are a few prominent formations in the foreground of the
music that do not fit altogether neatly into the structuralpicture I have
presented so far. I refer, in particular, to the strong statements of the
vertical tetrachord4-19, notably at bar 5 above bass d. To resolve this
apparent anomaly I suggest that the augmented-triad "core" of 4-19

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

The Tristan Chord in Scriabin's Fourth Sonata, first movement

Pitch-class form Octatonic location

5 3 8 T2
5 9 2 T2
8 2 6 Ii T2
o10 4 8 I T1
7 I 5 10 T,
6 o 4 9 To

serves as a unifying component with respect to the three forms of the

octatonic set, no single form of which can contain 4-19, and points to
certain key junctures in the music where the augmented triad
occurs-for example, the crux of the development at bar 29.


Three proposed guidelines for linear analysis are exemplified in

the analyses I have just presented.
First, an effective reading of the large-scale horizontal dimension
should relate in specific ways to the motivic structure of the music.
Perhaps the most telling illustration of this is the analysis of the
Tristan Prelude, in which what surely must be regarded as the
primary leitmotive of the opera, the "Tristan chord," extends hori-
zontally over relatively long spans of music. The Scriabin analysis
provides a more elusive instance, one in which an emphasized vertical
feature of the music, the A chord of Tristan, reflects the underlying
operation of a referential set that governs one of the primary linear
Second, where specific non-tonal referential collections are in
operation, as in the Stravinsky and Scriabin examples, the reading
should discover precisely how these are expressed in the music,
without violating such important musical considerations as phrase
groupings, rhythmically determined units, registral and timbral asso-
ciations, and so on.
Third, the reading of linear structures should take into account
onset and closure within the individual linear configuration as well as
the relation between linear configurations in combination-"coinci-
dences," as they might be termed. This is perhaps best illustrated in
the Scriabin analysis, where the closure of an octatonic segment in the

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

bass coincides precisely with the closure of a diatonic segment in the


Yale University


Ayrey, Craig. "Berg's'Scheideweg':AnalyticalIssues in Op. 2/ii." Music

Analysis I (1982): I89-202.
from Tristan
Bailey, Robert, ed. RichardWagner,Preludeand Transfiguration
and Isolde. New York, 1985.
Baker,JamesM. "SchenkerianAnalysisand Post-TonalMusic."In David
Theory.New Haven and London, 1983,
Beach, ed., Aspectsof Schenkerian
. TheMusicof AlexanderScriabin.New Haven and London, 1986.
Beach, David. "The Current State of Schenkerian Research." Acta
Musicologica57 (1985): 275-307.
Berry, Wallace. StructuralFunctionsin Music. Englewood Cliffs, N.J., 1976.
Boretz, Benjamin."Meta-Variations, PartIV: AnalyticFallout(I)."Perspec-
tives of New Music i i (1972): 159-217.
Forte, Allen. TheStructureof Atonal Music. New Haven and London, 1973.
"Motive and Rhythmic Contour in the Alto Rhapsody." Journal of
Music Theory27 (1983):
"Tonality, Symbol, and StructuralLevels in Berg's Wozzeck."
MusicalQuarterly71 (1985): 474-99-
Hindemith, Paul. The Craft of MusicalComposition.New York, 1945.
Balanceas an OrganizingForcein Schoenberg's
Lewin, David. "Inversional
Music and Thought." Perspectives of New Music6/2 (1968): 1-21.
. "TransformationalTechniques in Atonal and Other Music Theo-
ries." Perspectivesof New Music 21 (1982-83): 3 12-7I.
Lorenz, Alfred. Das Geheimnisder Form bei RichardWagner, Vol. II: Der
MusikalischeAufbauvon RichardWagners"TristanundIsolde."Berlin, 1926.
Englishtranslationof Lorenz'sanalysisof the TristanPreludein Bailey
1985, 204-23-
Meyer, LeonardB. "Grammatical
Simplicityand RelationalRichness:The
Trio of Mozart'sG minor Symphony."CriticalInquiry2 (1975-76):
Mitchell,WilliamJ. "TheTristanPrelude:Techniquesand Structure."The
MusicForum I (1967): 162-203.
Morgan,Robert. "DissonantProlongation:Theoreticaland Compositional
Precedents."Journal of Music Theory20 (1976): 49-91.
Nattiez, JeanJacques."The Conceptsof Plot in SeriationProcessin Music
Analysis." Trans. Catherine Dale. MusicAnalysis4 (1985):
Puffett,Derrick."The FuguefromTippett'sSecondStringQuartet."Music
Analysis5 (1986): 233-64.
Rahn, John. BasicAtonal Theory.New York, 1980.
Rothstein, William. "LinearStructurein the Twelve-Tone System: An
Analysis of Donald Martino's Pianississimo."
Journal of Music Theory24
(1980): 129-65.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions

Salzer, Felix and Carl Schachter. Counterpointin Composition,Part 2: "The

Techniques of Prolonged Counterpoint." New York, 1969.
Straus, Joseph. "The Problem of Prolongation in Post-Tonal Music."Journal
of MusicTheory31 (1987): 1-25.
Travis, Roy. "Towards a New Concept of Tonality?"Journalof MusicTheory
3 (1959): 257-84-
Van den Toorn, Pieter. TheMusicof Igor Stravinsky.New Haven, 1983.
Williamson, John. "The Structural Premises of Mahler's Introductions:
Prolegomena to an Analysis of the First Movement of the Seventh
Symphony." MusicAnalysis5 (1986): 29-57-
Wilson, Paul. "Concepts of Prolongation and Bartok's Opus 20." Music
TheorySpectrum6 (1984): 79-89.


A numberof recentstudiesincorporate linearanalysisof non-standard

tonalmusicandnon-tonalmusic.Althoughthe authortakesissuewiththe
application of strictSchenkerianparadigmsto suchmusiche acknowledges
the historicalimportance of the Schenkerian canon.Invokingnew proce-
dures,amongthempitch-class set-analytical
a seriesof analyseswhichreveallinear-motivic
featuresheldto beessentialto
the musicunderconsideration overtemporalspansof varyinglength.The
study,whichincludesworksby Stravinsky, Wagner,andScriabin,endsby
suggestingthree guidelinesfor subsequenteffortsin this new area of
analytical research.

This content downloaded from on Thu, 12 Jun 2014 21:08:00 PM

All use subject to JSTOR Terms and Conditions