You are on page 1of 33

Rethinking Form: Stravinsky's Eleventh-Hour Revision of the Third Movement of

His Violin Concerto

Lynne Rogers

The Journal of Musicology, Vol. 17, No. 2. (Spring, 1999), pp. 272-303.

Stable URL:
http://links.jstor.org/sici?sici=0277-9269%28199921%2917%3A2%3C272%3ARFSERO%3E2.0.CO%3B2-%23

The Journal of Musicology is currently published by University of California Press.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. 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/journals/ucal.html.

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.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.

http://www.jstor.org
Wed Aug 1 16:32:49 2007
Rethinking Form:
Stravinsky's Eleventh-Hour
Revision of the Third
Movement of His Violin
Concerto*
L Y N N E ROGERS

"T his I havejust completed," wrote Igor Stravin-


sky on 7 June 1931 of Aria II, the third movement of his Violin Con-
cert0.l Soon after writing this letter, however, he sketched a substantial
new passage and inserted it into Aria 11, forever changing the move-
ment and its form. Why did Stravinsky revise his opinion of the original
Aria 11and alter the movement so significantly? Two primary reasons
emerge from study of Aria 11and its compositional history: to incorpo-
rate sharper contrast, and to enhance motivic content and connection.
To explain these circumstances more fully and to explore what such a
late and dramatic revision reveals about his creative process, I will pre-
sent a brief analysis of the published form of the movement, then trace

.
Volume XVII Number 2 Spring 1999
TheJournal of Musicology 0 1ggg by the Regents of the University of California

* This article is a revised and expanded version of a paper read


at the joint meetings of the American Musicological Society and
the Society for Music Theory, New York, 1995. I am grateful to
Professor Warren Darcy for his valuable comments on the confer-
ence paper, and to Professors Severine Neff, William Rothstein,
and Joseph Straus for their excellent advice and many helpful
suggestionsduring the preparation of this article. I wish to thank
the Paul Sacher Foundation (Basel, Switzerland),where the Igor
Stravinsky Collection is preserved, for allowing me access to the
Collection and for granting permission to include in this article
transcriptions and a reproduction of excerpts from sketches and
the first short score of Stravinsky's Violin Concerto.

' Igor Stravinsky, Selected Correspondence, vol. 3, ed. Robert Craft (New York, 1985).
226.
ROGERS

and interpret its history through the composer's sketches and drafts.
The reader will find it helpful to have a score of the Concerto at hand.

The final, published version of Aria 11


The published version of Aria 11 displays ternary
form, as illustrated in Table 1. The opening part, Part A, is divided into
two sections (Ri/Si and R2/S2), each of which consists of a short
statement (R) by the entire instrumental ensemble (flute, clarinet,
trumpet, trombones, solo violin, and strings) and a longer passage (S)
for solo violin with string accompaniment. The first beat of each R fea-
tures the striking sonority (Dq/E.3/A6) in solo violin that launches
each movement of the Concerto. Stravinsky called this simultaneity
the " 'passport' to the concert^."^ The last beat of each R, played only
by flute, clarinet, and trumpet, overlaps the first beat of each S. Within
both Si and S2, the solo's melismatic, cantabile melodies are accompa-
nied only by strings, marked piano and differentiated from the solo by
their lower tessitura, generally longer durations, and steady bass line
in eighth notes. In S2 the solo violin introduces a new motive: a lower-
neighbor figure employing doubly dotted rhythm. On the second beat 273
of #79:2, where this figure is first heard as G4-F#4-G4, the second dot -
occurs as a rest. The neighbor motive later appears singly dotted, as
on beat three of #80:4 Both sections of Part A close with chorale-style
cadences in four parts.
The end of Part A elides on the third beat of #8 1:1 with the begin-
ning of the movement's central portion, Part B, a highly contrasting,
faster passage immediately distinguished by the accompaniment's
rhythmically animated succession of dense chords. The solo violin's line
in Part B, exhibiting overall a lower tessitura than in Part A, is divided
into two sections. The first ( B i ) , beginning at the elided cadence in
#8 1:1 , is played primarily on the G string. The singly dotted version of
the lower-neighbor motive from S2 reappears to embellish C#4/Dbq
with B#4/C4 in #81:2 (b. 2 ) and #81:3 (bs. 2-3). The last eighth in
#8 1:q launches the second section (Bz), distinguished initially by
eighth-note arpeggio-like groupings implying three-voice polyphony.
Throughout Part B, violas provide a distinctive countermelody that
reinforces the solo's structural pitches. Part B ends with a chorale-style
cadence.
The third and last portion, Part A', is a varied return of the first.
Like Part A, Part A' is divided into two sections (R3/S3 and R4/S4),

Samuel Dushkin, "Working with Stravinsky," in Igor Stravinskj (A Merle Armitage


Book), ed. Edwin Corle (New York, 1949)~
1 82.
THE JOURNAL OF MUSICOLOGY

TABLE 1
Stravinsky, Violin Concerto, Aria I.: formal divisions
(published score).
Measures Formal Division
#77:1-#81:1 (b.3) Part A

#77:1-2 (b. 1 ) R1

#77:2-#78:4 S1

#79:1-2 (b. 1 )
#7g:2-#8i: 1 (b. 3)
#81:1 (b. 3)-#83:i (b. 1) Part B

#81:1 (b. 3)-4 (seventh eighth) B1

#8 1:4 (last eighth)-#83: 1 (b. 1 ) B2

#83:1 (b. 2)-#85:3 Part A'

#83:1 (b. 2)-2 (b. 2 ) R3

#83:2 (b. 2)-3 (b. 3) S3

#83:3 (b. 4)-#84:i (b. 3) R4


274
- #84:1 (b. Q ) -85:3
# s4

each beginning with a short statement by the full ensemble (R) and fol-
lowed by a longer passage played by solo violin and strings (S). Sg, only
1 - 1 / ~ bars in length, strongly resembles S i motivically and texturally,
but introduces a different pitch-class collection and new focal pitch-
classes. In Sq, five bars long, a trio of two flutes and solo violin recalls
motivic material from S 1 , accompanied only by a sustained A in double
basses.
In this, the only movement of the Concerto's four that is not
D-centric, both C # and F# receive considerable emphasis, with the latter
emerging ultimately as centric for Aria ZI. Parts A and B are weighted
overall toward C#, although other pitch classes, most prominently F# in
the cadence closing S1, are stressed. In Sg, the first section of Part A', C
unseats C#.When it returns in Sq, C# leads to F#, upon which the move-
ment closes.

Sketch study: History of Aria II


The score's ternary scheme of statement-contrast-
restatement is structurally so elemental that it is difficult to imagine
that it was not part of Stravinsky's original conception of the move-
ment. Yet, the composer's musical manuscripts for Aria ZIshow that the
movement first emerged as a ritornello form containing early versions
ROGERS

of the published movement's outer sections only. The contrasting central


section was added after Stravinsky pronounced the movement complete
in his aforementioned letter of 7 June.
The musical history of Aria 1 1may be traced in four manuscript
sources in the Igor Stravinsky Archives of the Paul Sacher Foundation
in Basel, Switzerland. These n~anuscripts,which contain Stravinsky's
work on the Violin Concerto exclusively, comprise a bound sketchbook,
a folder of unbound sketches, an autograph short score, and an auto-
graph full score. Since Stravinsky typically composed at the piano, it is
inevitable that some steps in his compositional sequence went un-
recorded.3 Nonetheless, significant stages of the movement's creation
are documented in the manuscripts.
Evidence indicates that Stravinsky's earliest written ideas for Aria I1
appear in the sketchbook, a bound volume of 1 2 0 pages, intended for
the writing of text. Stravinsky transformed it into a musical sketchbook

:3 Support for Stravinsky's habit of composing at the piano includes accounts by


Samuel Dushkin, Robert Craft, and the composer himself. Dushkin, the violinist for
whom the Violin Concerto was written, recalled Stratinsky's work habitz during the com-
position of the Violin Concerto: "He often composes at the piano, grunting and strug-
gling to find the notes and chords he seems to be hearing" (Dushkin, "Working with 275
Stravinsky," 184). Craft, Stravinsky's musical assistant beginning in 1948, observed that -
"[Stratinsky] almost always begins with a melodic idea, . . . The piano is not resorted to
in this melody-forging stage, but only when harmonic and contrapuntal ideas begin to a p
pear; . . . (Igor Stratinsky and Robert Craft, Dialogues [Berkeley and Los Angeles, 19821,
1 4 ) .Stravinsky commented in An Autobiopphy (New York, 1962; originally published as
Chroniques de m a vie [Paris, 19351) that "I d o compose at the piano and I d o not regret it"
(p. 5) and "Fingers are not to be despised: they are great inspirers, and, in contact with a
musical instrument, often give birth to subconscious ideas which might otherwise never
come to life" (p. 82). Later, in Igor Strarinsky and Robert Craft, Convrrsations zuith Igor
Stravinsky (Berkeley and 1.0s Angeles, 1959). 15, he stated, "But long before ideas are
born, I begin work by relating intervals rhythmically. This exploration of possibilities is
always conducted at the piano." The strength of Stravinsky's claims must be tempered by
their sources. An Antobzography was ghost-written by Walter Nouvel. Some questions have
arisen regarding the extent of Craft's participation in the cor~versatior~ books, but both
Kathryn Bailey and Richard Taruskin attribute the early books, including Conziersations
ruith Igor Stravinsky, to the composer. For an evaluation of the authorship of An Autobiqqra-
phy, see Kathryn Bailey, "The Craft/Stra\lnsky Conversation Books: Bibliography and
Commentary," Studies i n music from thr C'nivrrsztj of W e s t m Ontario 111 ( I 978), 54; Robert
Craft, Present Prrspectives: Critical Writings (New York, I 984), 26;; and Richard Taruskin,
Stravinsky and the Russian Traditions: A Biography ofthr Works Through Mama, vol. 1 (Berke-
ley and Los Angeles, I 996), 460-6 1 . For disc~~ssion of the authorship of one or more of
the conversation books, see Bailey, ' T h e Craft/Stra\insky <:onversation Books," 49; Vera
Stravinsky and Robert Craft, Stravinsky in Picturrc and Documentc (New York, I 984), 438-
39; Taruskin, Stravinsky and thr Russiccn 7i-nditions I , I 2-1 y, n. 42; and Stephen Walsh, Thr
Music of Stravinsky (London, 1988), 245-46 and 290-91, 11. 4. Some of Stravinsky's state-
ments quoted above also appear in Pieter van den Toorn's 7'hr Music of Igor Strauinsk~
(New Haven, 1983). 204436. Van den Toorn comments on Stravinsky's use of the piano
during composition on p. 2 1 I . In addition, in his Stravznsk~~ and "7%eRite of Spring: The Be-
ginnings of a Musical Language" (Berkeley and Los Angeles, 1987), 34 and 37, van den
Toorn speculates from the degree of development in initial sketches and from the pres-
ence of pianistic characteristics in their final forms, which sections of Thr Ritr of S F ' n g
were likely created through extensive improvisation at the keyboard.
THE JOURNAL OF MUSICOLOGY

by producing staves in ink, as needed, with a stylus. Other notations are


in pencil. Since neither pages nor leaves of the sketchbook are num-
bered, I will refer to specific pages using the page numbers assigned in
my inventory of the sketchbook and transcriptions of its contents. In
addition, I will provide in parentheses after the first mention of each
page its location on the Sacher Foundation's microfilm of the manu-
script (microfilm log).
Dates in the composer's hand, scattered throughout the sketch-
book, indicate that he filled its openings sequentially. After sketching
the first and second movements on pp. 1 through 61 of the sketchbook
(frames 139-6g), Stravinsky used pp. 62-79 (frames 170-78) for work
on Ai-ia II. He frequently wrote directly across two facing pages, treating
most openings as double-spreads.
Stravinsky began the movement on the facing pp. 62 and 63
(frame 170). As noted in Table 2, this opening, dated 24-26 May 1931,
exhibits work on the first ritornello and solo, Ri and the seven-bar Si
respectively. (All sections of the movement named during this discus-
sion of the sketches correspond to those of the published version of the
score, shown in Table 1.) The music of the sketch is very close to that in
276 the final version. A notable exception is the tempo, marked as eighth =
-
60, faster than the 48 of the published score.
Stravinsky went next to the opening formed by pp. 64 and 65
(frame 171 ) , dated 30-3 1 May. Here, he worked on S2. Only the final
gesture of R2 (equivalent to the last three sixteenths of #79: 1 and to
the first beat in flute, clarinet, and trumpet of #79:2) is present, over-
lapping the beginning of S2. The presence of this abbreviated repre-
sentation of the ritornello suggests that Stravinsky had determined by
this point that the movement would be in ritornello form and that the
short opening statement would act as the ritornello. He continued to
compose and revise S2, just over seven bars in length, on the opening
comprising pp. 66 and 67 (frame 172),dated 4 June.
Especially pertinent to this examination of changes in the move-
ment's form are Stravinsky's sketches for the cadence closing S2, that is,
the end of Part A. While working on S2, Stravinsky notated on pp. 64-
67 five very similar versions of this cadence. The version that appears,
by virtue of its location near the bottom of p. 67, to have been the last
composed, is shown as Example la. (In this and all other transcrip-
tions, my additions are enclosed in square brackets.) Example l a is
identical as regards pitch and rhythm to the same cadence in the p u b
lished score (Example i b ) with two exceptions. First, the sketched ca-
dence begins on the second rather than the first beat of the bar. Sec-
ond, in the published score, the final chord of the cadence is elided
with the beginning of Part B on the third beat of #8 1: 1. In Example la,
ROGERS

EXAMPLE 1. Cadence closing Part A ( S 2 ) : (a) transcribed from p. 67 of


the sketchbook; (b) #8 1:1 of the published score.
(a)

Viol 1

Viol 2

I
Vcl

as in the four other versions on pp. 64-67, the cadence is not elided
with the next part, but rather occupies the entire bar.
After creating Part A, Stravinsky filled pp. 68-71 (frames 173-74)
with sketches for S4 (see Table 2 ) . 4 Stravinsky apparently knew by this
4 This section is labelled "Sq" to correspond with its eventual position in the move-
ment, as seen in Table 1, rather than with its place in the compositional chronology.
THE JOURNAL O F MUSICOLOGY

time that the freely imitative trio for two flutes and violin would close
the movement, since each sketch on pp. 68-71 that reaches the final
chord of the work ends with a double bar.3
Pages 68-69 (frame I is), dated 4 June, reveal the earliest version
of Sq. The presence of Rq's final beat over the beginning of the trio
presumably implies that the entire statement is to precede the final
section. The sketches of the movement's five closing bars resemble
overall the corresponding passage in the published score, although
much material present at this stage is subsequently discarded or reap-
pears greatly changed in later versions. On the opening formed by pp.
70-7 1 (frame 1 ~ 4 dated ) ~ 6 J u n e , Stravinsky continued to refine this
passage, pushing it still closer to its ultimate form.
On the next opening, undated pp. 72-73 (frame i 7 5 ) , Stravinsky
sketched S3 for the first time." Lower on the opening, Stravinsky also
entered Rq and another revision of Sq. No notations for R3 are pre-
sent; given the regular appearances of the ritornello in the sketchbook
thus fa^; it is likely that Stravinsky planned to include Rg but neglected
to indicate his intentions.
This first version of S3, shown as Example nb, is texturally and
278 motivically similar to the published one, but differs greatly as regards
-
pitch-class collection and emphasis. By comparing Example 2a, the first
two measures of S i , with Example 2b, the newly composed Sg, it may
be seen that the original S3 is actually a quotation of the first two mea-
sures of S I , metrically realigned and lacking the third and fourth beats.
The compositional chronology of the solo sections of the move-
ment at this point-Si, Sn, Sq, then Sg-is provocative, since it suggests
a ternary conception lurking behind the explicit ritornello form. The
sketches through pp. 70-71 provide evidence that Stravinsky, at least
for a short time, envisioned the movement as follows: Ri S i Rn S2 Rq
Sq. The similarities of Sq to S I and the subtle variety provided by S2 in-
timate ternary form. Sq, while differing from S i in texture, uses its mo-
tives; in fact, large portions of the solo violin's melody in S i , including
most of its first two bars, return intact in Sq. Additionally, both outer
solo sections begin with C#3 in solo violin. In contrast, in the inner solo
section Sn, the solo violin opens F#q-Gq and introduces the new lower-
neighbor motive.
The subsequent creation and insertion of Sg strengthened the im-
plications of ternary form. In sketches on pp. 72-73, S3 quotes the
7 Stravinsky did not close the movement with the traditional final ritornello. It is

supplied instead bp the immediatelv follo\ving final movement, which opens with the
"passport."
" X date for the opening formed by pp. 72-73 and justification for its inclusion in
the series of sketches begun with p. 62 will be provided during the upcoming discussion
of the first short score, which supplies the Ilecessan. evidence.
ROGERS

EXAMPLE 2. (a) first two measures of S i , transcribed from pp. 62-63


of the sketchbook; (b) Sg, transcribed from pp. 72-73 of
the sketchbook.
(a)

('I

Vlr

v-C

Vno

Vnl'

vlr2

opening two bars of S i and thus functions as a partial return. The solo
sections heard in "piece order" through S3-Si S2 Sg-convey the
scheme ABA'. Sq, which begins with the same solo ~ i o l i nline heard in
Sg, repeats and develops the brief Sg. Although separated bv Rq, Sg
and Sq might be heard as grouped together to form the return: initially
abridged, then repeated and de~eloped.7Sg and Sq total slightly over
seven bars, balancing the seven bars each of S i and Sn.
7 It is intrrrsting that &. the only ritornello other than R I notatrd for its full dura-
tion in the sketchbook. lacks thr initial quartrr-not? chortl of Ri, and thus tlisrupts the
continuity betwrrn solo sections less trongly than had earlier ritornrlli. Stravinsky rr-
rained this version of Rq for the p~~blishrtl score.
THE JOURNAL OF MUSICOLOGY

In summary, on pp. 62-73 of the sketchbook, Stravinsky composed


the music for the outer parts of Aria 1 1and revised much of it exten-
sively (see Table 2 ) . As a result, the latest versions of Parts A and A'
seen in this portion of the sketchbook are generally very close to the
corresponding music in the published score-with two important ex-
ceptions. First, the cadence that completes Part A occupies an entire
measure, not eliding with the beginning of an as yet nonexistent Part B.
Second, Sg uses a pitch organization vastly different from the corre-
sponding section in the published score.
After creating and revising Parts A and A' on pp. 62-73 of the
sketchbook, Stravinsky notated a clean short score of Aria 1 1in ink and
inscribed it with the date 6 June 193 1 (see Table 2 ) . For his composi-
tional procedure, Stravinsky's creation of a short score was an impor-
tant step since typically it signified that he considered the composition
of a movement to be essentially c ~ m p l e t e The . ~ short score was then
used as the basis for notating the clean copies that he sent to publishers
for engraving.
That the short score of Aria 11dated 6 June signified to Stravinsky a
completed movement is verified by the correspondence between Stra-
280 vinsky and Willy Strecker of B. Schott's Sohne, publishers of the Violin
-
Concerto. In the aforementioned letter to Strecker, written 7 June
1931, the day after Stravinsky completed the short score, the composer
thanked Strecker for his compliments on the Concerto's second move-
ment, Aria I, and then wrote: "the second Aria is shorter than the first
and totally different in character; this I have just completed and will send
to you in about eight days . . ." (italics mine) . g According to the pattern

Documentation for the significance of the short score in Stravinsky's composi-


tional routine is found in Stravinsky's correspondence, his dating of manuscripts, and
Robert Craft's accounts of compositional chronologies of various works. For example, the
date at the end of the short score of the first movement of the Violin Concerto is 27
March 1931. O n 30 March 1931, Stravinsky wrote to MTillyStrecker of the publishers
B. Schott's Sbhne: ". . . I have finished the first part of the Violin Concerto, both the mu-
sic and the instrumentation, but the latter is not yet in orchestra score form. Now I am
working on the piano reduction" (Stravinsky, Selected Correspondence 3, 225). The end of
the short score for the second movement of the Concerto i n Eb ("Dumbarton Oaks") is
dated 1-2 J a n u a n 1938. O n 3 January, Stravinsky wrote Strecker that he had finished
the movement (Ibid., 256) and on 4 January, wrote the violinist Samuel Dushkin that he
had already composed the first and second movements of the Concerto (Igor Stravinsky,
Selected Correspondence, vol. 2, ed. Robert Craft [New York, 19841, 310). Robert Craft cites
the completion of the first movement of the Ortet in "abbreviated full score" (Ibid., 460)
and of the first and second movements of the Symphonj i n C in "short-score form" (Igor
Stravinsky, Sekted Correspondence, vol. 1 [New York, 19821, 437-38). Richard Taruskin
characterizes the function of short scores similarly for Stravinsky's earlier music, stating
that the "preparation of such drafts was generally the last stage of composition before or-
chestration" (Richard Taruskin, Strauirisky and the Russian Eaditions: A Biography of the
Works Through Mavra 1, 173).
!I Stravinsky, Select~dCorrespondenc~3, 2 26.
ROGERS

TABLE 2
Stravinsky, Violin Concerto, Aria 11:proposed chronology

for composition. Dates are taken from those in Stra~~insky's

hand unless listed as "n.d."

Portion of
Date (all 193 1 ) Location in manuscript published score
Part A
24-26 May sketchbook, pp. 62-63 Ri , Si
30-31 May sketchbook, pp. 64-65 R2 (partial), S2
4 June sketchbook, pp. 66-67 Se

Part A'
4 June sketchbook, pp. 68-69 Q (partial), S4
6 June sketchbook, pp. 70-7 1 S4
n.d. (probably 6 June) sketchbook, pp. 72-73 S3, R4, S4

6 June first short score Parts A and A' (incl. Rg)


n.d. (probably between first piano-violin score Parts A and A'
7 and l o June) 281
-
n.d. (probably between sketchbook, pp. 74-79 Part B, Sg (pp. 77, 79)
7 and l o June)
i o June second short score Parts A, B, and A'
(complete, with revised Sg)

established with the first two movements, what Stravinskv intended to


send Strecker shortly was the piano-violin score. The orchestral score
would be sent considerably later.
Instead, however, the short score dated 6 June 1931 was soon
followed by another short score of Aria 11, reflecting the dramatic re-
evaluation that required Stravinsky to "complete" the movement more
than once. The score of 6 June (hereafter referred to as the "first short
score") was never sent to Strecker; it was discarded in favor of the later
version. This explains the survival of only three pages of the first short
score and their presence in a folder of unbound sheets labelled by
Stravinsky "rough sketches" (also reproduced on microfilm l o g ) Al-
though two pages-3 I and 32-are missing, the three remaining-the
first, numbered "30," and the last two, numbered "33" and "34"-
reveal much about the movement at this stage.
On p. 30 (frame 99), the first three measures of Aria 11appear in a
form nearly identical to the same measures on pp. 62-63 of the sketch-
book and the published score. The page is headed by the number of
PI.ATF: 1. First short score, p. 33: last 2-3/4 ~neasuresof Part A and first 4-1/4 measures of Part A'.
EXAMPI.E 3. (continued)
~201 1211
FI.
CI

Tr-ba
'Tr-nc

V-no
ROGERS

the movement ("111") and its title ("ARIA 11"). Stravinsky retained the
tempo of eighth = 6 0 from pp. 62-63. The real find is on p. 33 (frame
l o o ) , seen as Plate 1 and transcribed as Example 3."'
Here, the end of SY,derived from pp. 66-67 of the sketchbook and
equivalent to #80:3-#8 1:1 of the published score, is followed immedi-
ately by Rg (equivalent to #83: I , b. 2-#83:2, b. 2 ) , that is, Part A is fol-
lowed directly by Part A'. The cadence closing SYin m. I 7 on p. 33 is vir-
tually the same as that shown as Example l a except that the cadence in
m. 17 begins on the downbeat rather than the second beat of the bar.
(For measure numbers, please refer to Example 3.) On the fourth beat
of m. 17, Stravinsky notated the initial chord of Rg, the remainder of
which appears in the next bar. Presumably transferred from pp. 72-73
of the sketchbook, Sg, Rq,and Sq follow on the lower system, with Sq
and the movement itself closing on p. 34 (frame I o I ) . l
After notating this short score, Stravinsky wrote the piano-violin
score, to be sent to Strecker. It is impossible to determine, based on the
surviving pages, if Stravinsky completed the piano-violin score. Al-
though he did not date the score, the fact that in it Part A' immediately
follows Part A indicates that he notated the manuscript after writing to
Strecker on 7 June but before completing the revised short score of the 285
-
movement, dated l o June (see Table 2 ) . Like the first short score, the
first piano-violin score was rejected after Stravinsky revised the form of
the movement, and never sent to Strecker. Extant pages of the score
(frames 103-06) are now f6und within the folder of unbound sketches.
The dated first short score and the first piano-violin score of Aria II,
along with Stravinsky's correspondence with Strecker, demonstrate that
as of 6 June 1931, Stravinsky assumed that he had finished composing
Aria 11, a movement then exhibiting ritornello form. The contrasting
central passage that would group the alternating ritornelli and soli into
Parts A and A' was evidently not even a twinkle in the composer's eye.
Yet, at some time after the first week of June 193 1 , he reconsidered the
movement as it then stood, and embarked on a major revision.

Ii1 The missing pp. 31-32 must have contained the bars filling the gap between the
last measure on p. 30 (equivalent to #77:3) and the first on p. 33 (equivalent to #80:3),
Stravinsky presumably transferred these bars (equivalent to #77:4-#go:') from selected
versions on pp. 62-67 of the sketchbook. If the scale of Stra~insky'snotation on pp.
31-32 was consistent with that on p. 33. pp. : ~ I - Y Z would have comfortably accommo-
dated these eleven bars.
'I A date of 6 June 1931 can now be suggested for the undated pp. 72-73 of the
sketchbook (see Table 2). Since Stravinsky seems to have transferred material from pp.
62-71 of the sketchbook to the short score, it is likely too, that the music on pp. 72-73
was composed in the sketchbook, then copied onto the first short score. Thus, since the
music on pp. 72-73 was probably composed after that on the opening formed by pp.
70-7 1, dated 6 June i 93 1, and before the first short score, also dated 6 June 193 1, the
same date must be assigned to p. 72-73,
THE JOURNAL OF MUSICOLOGY

The Creation of Part B


Once Stravirlsky decided to add a contrasting cen-
tral passage to the movement, he returned to his sketchbook (see Table
2 ) . O n pp. 74-79 (frames I 76-78), he created Part B and thoroughly
revised Sg, which would follow it directly. Although pp. 74-79 are un-
dated, we may assume that they were filled after Stravinsky wrote to
Strecker on 7 June and before he completed the second short score on
I o June.
The opening spanning pp. 74 and 75 (frame i 76) contains Stravin-
sky's earliest sketches of Part B, which already feature the accompani-
ment's new rhythmic pattern and new nlelodic ideas for the solo violin.
In these initial sketches, the solo violin also develops the lower-
neighbor motive introduced in S2. In addition, Stravinsky had appar-
ently decided at the earliest stage of composing Part B to elide its open-
ing with the end of Part A. Example 4, transcribed from pp. 74-75,
shows the beginning of Stravinsky's first sketch of the opening of Part
B. The sketch and Part B begin on the third beat of the first bar shown
with a rhythmically active A-major triad in second inversion. A sustained
286 but identically voiced presentation of this same triad closed Part A
- on the same beat of m. 17 on p. 33 of the first short score (see Plate 1
and Exampie 3 ) . Stravinsky retained this elision for all later versions
(see Example I b) .
Also present on pp. 74 and 75, and already in its final form, is the
cadence closing Part B. It is significant that the last chord of this ca-
dence, directly preceding Part A', is a root-position A-major triad. In
the first short score, Part A' immediately followed the A major triad in
second inversion that closed Part A (Example 3, m. 17). Thus, Stravin-
sky composed Part B so that it ended on a repositioned form of the
same harmony that had previously preceded Part A'. Stravinsky refined
Part B on pp. 76-77 (frame 177), bringing this version of the contrast-
ing passage very close to that in the published score.

Reasons for the Creation of Part B


Why did Stravinsky decide to compose Part B and
transform Aria 11into a ternary form? Why is the ternary form, with its
new passage, essential to the movement? No known writings by Stravin-
sky explain his changes to Aria II. As there is no evidence to indicate
that extramusical considerations were involved, we must conclude that
Stravinsky's reasons were entirely musical and thus reflected by the
structures and relationships produced by the inclusion of Part B that
would be absent otherwise.
ROGERS

+
4 2
C

5
u

5
Lk

0
LC,

r.

I
Tt'
r.

Q
a

s
C

T
eL

zE:
2:

-
z
a-

Lk

0
p
-
.
8

-4
W

c)

Y
T H E JOURNAL OF MUSICOLOGY

With its many striking new features, Part B provides Aria II most
evidently with strong contrast. (The movement at this compositional
stage is close enough to the final version that the published score will
be the reference for the discussions that follow.) Particularly notable
features include the presentation of the accompaniment's harmonies
as repeated thirty-second-note chordal segments, and the solo violin's
lower range and new, arpeggio-based constructions. Incorporation of
the contrasting central section altered the formal contour of the move-
ment, grouping Rl/Si/R2/S2 as A and R3/S3/Rq/Sq as A', articulat-
ing a ternary form. Heard after Part B, R3 and S3 are more clearly es-
tablished as a return of the opening material, thereby increasing their
poignancy and expressive power.
It is significant that the seeds for the movement's new form lie in its
past. As discussed above, earlier versions (through p. 73 of the sketch-
book) are explicitly in ritornello form, but suggest the possibility of
ternary form through characteristics differentiating S2 from the other
solo sections. Stravinsky ma) have recognized the latent capacity of
such differentiation, inspiring him to create Part B.
Although the most obvious function of Part B is that of providing
288 contrast, its presence is equally important in creating connection. First,
- the solo violin's framework of emphasized pitch classes in Part A,
C#-F#-G-C#, reemerges, reclothed, in Part B. In Part A, the solo violin
begins on a long C#5 in #77:2 and returns to rest on it several times be-
fore ending S I in #78:4 on F#4, notable as the first strong arrival on
the movernent's centric pitch class. The solo begins S2 in #79:2 with
stress on Gq, reemphasized in the following bars, and cadences on C#q
in #81:1. The same C#q opens Part B. The solo violin continues
to stress C# (also spelled Db) in #8i:n-q (b. 1) as the main note of the
lower-neighbor motive. F#, spelled as Gbq, closes Section BI on beat q
of #81:4 and opens Section B2 an octave higher on the next eighth,
the apex of the solo's line in Part B. The solo's series of descending
arpeggio-like figures trickles by #82:3 to a close on open G3, the nadir
of the line. On beat 4, G is transferred up an octave, initiating a scalar
descent that is completed on C#q by second violins, playing the highest
moving line in the cadence.
The tonal framework C#-F#-G-C# is not only expressed by the solo
violin (assisted by second violins) in Part B, but also by the violas' coun-
terpoint. After sustaining C# along with the solo violin in #8i:i-2, the
violas spin a melody based on a sequence of ascending perfect-fourth
skips (E3-A3, F3-Bb3, and Gbg-Cbq), each followed by descending
steps to F#3/Gb3. The second F#3/Gb3, in #81:4, anticipates then dou-
bles the solo. The violas move in #82:3 to G3, with which they prepare
then reinforce the same pitch in the solo. In sum, solo violin and violas
ROGERS

state, with significant overlap, the first three pitch classes of the frame-
work before it is completed by second violins.
The entire string accompaniment (including violas) frequently sup-
ports the solo's emphasized pitch classes in Aria 11through unison and
octave doublings and harmonization by thirds (or sixths) and triads.
During the movement, solo and accompaniment combine to create si-
multaneities exhibiting variety in number of members and in intervallic
structure; within this context, the consistent appearance of thirds and
triads at formal boundaries and harmonizing emphasized pitch classes
is conspicuous. The solo's focal C#s that occur at important structural
points in Parts A and B are doubled and harmonized with A (e.g.,
#77:2, b. 1; #78:3, b. 1 ) or A and E (e.g., #81:1, b. 3; #83:1, b. 1 ) .
Stressed F#/Gbs are likewise doubled and either belong to a third (e.g.,
with A in #78:4, b. 4) or triad (e.g., with Eb and Bb in #81:4, b. 4). The
solo's emphasized G's are treated similarly (doubled and with E in
#7g:2, bs. 2-3; with D and Bb, then D and B in #82:3, bs. 2-4). When
C#, F # , and G are stressed at less significant formal points, they are
frequently doubled and/or harmonized by sonorities that incorporate
thirds or triads into larger, more complex structures (e.g., the solo's
C # 5 in #77:3 (b. 4) is doubled and harmonized with F#,A#, and G # ) . 289
-
At important structural boundaries throughout Aria 11(including
within Part A'), the solo violin's emphasized pitch classes are harmo-
nized most often with A and E.l2These two pitch classes first attracted
attention in the concerto's opening bar as the upper two-thirds of the
solo's passport sonority. It is noteworthy that in Aria II, the passport's A
and E are doubled by the orchestra for the first time in a passport "set-
ting" (see Example 5). A passport "setting," which, headed by the pass-
port, begins each movement, is identical in Aria 11to the ritornello. Al-
though different for each movement, each passport setting comprises
the passport, the solo violin's passport "extension," that is, the material
it plays immediately following the initial sounding of the passport, and
the orchestra's accompaniment during passport and extension. In the
passport settings beginning the two preceding movements, the orches-
tra reinforces only the passport's D. By the opening of the last move-
ment, all three of the passport's pitches are doubled by the orchestra.
Part B also creates motivic connection within Aria 11by outlining
a descending line that answers those articulated by other sections.
l2 Stravinsky even harmonizes the movement's final F#6 with A and E, as well as
doubling it with FU4. Stravinsky deviates here from one norm established during Parts A
and B for focal pitch classes at important structural points-harmonization as members
of thirds or triads-but adheres to another in his use of A and E. The motivation for do-
ing so in the movement's final sonoritv might stem from the link F#, A, and E provide to
the next movement; these three pitch classes, along with centric D, form the opening si-
multaneitv of Capn'rrio.
THE JOURNAL OF MUSICOLOGY

EXAMPLE 5. Published score, #77:i-2: passport setting (circled) over-


lapping the beginning of S i .
Aria I1

Flnuto gr I

Tromhoni

V~olinoSolo

290

Although descending stepwise lines at various structural levels occur


throughout the Concerto, such a line is highlighted for the first time in
Aria IIas part of the passport setting. The passport's A6 initiates the de-
scending line A-G-F-E-D-B, doubled and completed in flute and
trumpet (see Example 5 ) . While rhythm and instrumentation make dif-
ferent segmentations of this line possible, it is the initial diatonic, scalar
descent through a perfect fifth that recurs within Aria IIas an indepen-
dent structure.I:I

':3 Capnrczo's passport setting transforms this scalar descent into major in #86:4-6:
rZ5-Gg-Ft j-Ej-Dg.
ROGERS

As an immediate reaction to this descent from A to D, the solo vio-


lin plays C#5-Bq-Aq-G#q-F#q in #77:2-3. Two other descents from
C # 5 to F#q in S i unfold within a more complex polyphonic line: that in
#77:3 (b. 4)-4, which substitutes Gq for G#4, and that in #78:3-4. In
S2, the solo's descent does not occur until nearly the end of the section
in #80:4 (b. 2)-#8i: 1 (b. 3 ) , and, as befits the subtly contrasting nature
of S2, moves to, not from, C#: Fq-Eq-Dq-C#q Compared to earlier
five-note scalar descents spanning perfect fifths, S2's four-note slide
through a diminished fourth seems insufficient. Indeed, after the
prominence granted Gq during the first half of S2, a five-note descent
from Gq to C#q-only one semitone shy of a perfect fifth-would not
be unexpected. Even the substitution of G for G# in #77:4 of S1, which
links G to C# for the first time, prepares the ear to be revisited by the
pairing. Nonetheless, the emphasized Gq of #79:2-#8o:1 moves up
rather than down and is not convincingly connected to the Fq of #80:4
except by register.
The addition of Part B solved this problem with a closing descent
leading unequivocally from Gq to C#4. G is heard first as G3 in #82:3,
emerging from the gradually slipping three-part texture implied by the
solo's polyphonic line in #81:4 (b. 4)-#82:2. After an octave leap, the 291
-
solo begins the awaited descent with Gq-Fq-Eq in #82:3-4, leaving sec-
ond violins to finish the line with Dq-C#q in #82:4-#83: 1 .
This second and finally successful attempt to move from Gq to C # 4
is preceded in Part B by the violas' three short descents to F#/Gb, dis-
cussed earlier, which may be heard as incomplete efforts to reproduce
the original descent from S i . In the same way that the too-short a p
proach to C# in S2 called for a complete descent, eventually provided
in Part B, the violas' moves to F# in Part B necessitate a return of the
entire perfect-fifth span. In fact, the highly contrapuntal Sq, which
quotes fragments from S i , brings back C#-B-A-G#-F# numerous times
in various registers in solo violin and flutes. The original S3 (Example
3, mm. 18 (b. 4)-20 (b. 1 ) ) offered an even earlier return of this fig-
ure, which disappears from the revised version.
Its authority in outer sections S i and Sq and its altered presence in
Part B establish the C#-F# descent as an essential pitch structure in Ama
ZI, tied intimately to the movement's F# centricity. With their descents
to C # , Part B and S2 subject the scalar descent to a transformation that
attempts to balance the ultimately primary descent to F#. Part B pro-
vides connection to Parts A and A' when it completes the G-C# descent
attempted in S2, and when it offers too-short descents to F#/GC that
elicit the reappearance in Sq of the original structure.
Part B may also have been added to provide temporal balance
within and without the movement. As it stood without Part B. the
ROGERS

already present in the original ritornello form; the creation of a


strongly contrasting and self-contained Part B affirms this tendency.
The late composition of Part B as a self-contained and compara-
tively insulated unit, inserted into a movement containing other such
passages, is typical of what Craft refers to as "Stravinsky's jigsaw compos-
ing process,"ly which involved the composition of various segments of a
work, not in their ultimate piece order, with their assembly coming
later.P<>In this light, it is interesting that Stravinsky, several months be-
fore beginning the Violin Concerto, characterized the form of his
music, and, in fact, that of all contemporary art, as c o n s t r ~ c t i v e The
.~~
insertion of the highly contrasting, self-contained Part B, while dramati-
cally altering Am'a I1 at a late compositional stage, is nonetheless consis-
tent with known aspects of Stravinsky's compositional history.

The Revision of S3
In addition to the creation of Part B, the only major
change in the movement made after the completion of the first short
score involved Sg. For the S3 of the first short score (Example 3: m. 18
(b. 4)-m. 2 0 (b. I ) ) , Stravinsky substituted new, but related music, 293
shown as Example 6b. The first three beats of the new Sg (Example 6b, -
m. 1 ) occupy the bottom right-hand corner of p. 77 of the sketchbook
(see Table 2 ) . Most of the opening formed by pp. 76-77 (frame 177) is
filled by a continuous draft of Part B, which ends on the lowest system
of the opening and is followed immediately by Rg (represented by its
highest line) and the first three beats of Sg. Stravinsky sketched three
versions of the fourth through sixth beats of Sg on p. 79 (frame 178; p.
78 was left blank), selecting for inclusion in the work the version most
likely composed last. This version appears as m. 2 of Example 6b.
The revised Sg differs from the original primarily in regard to pitch
organization. The earlier version (Example nb) is a nearly exact quo-
tation of the first two bars of S i . The original Sg uses the pitch-class

',IStravinsky,Selected Correspondence 2 , 4 6 5
z'' In his appendices to the first two volumes of Stravinsky's Selected Correspondence,
Craft provides numerous accounts of the "jigsaw process." See especially vol. 2 , Appendix
F: "On the Chronologies of the Composition of the Octet, Serenade, and Concerto per
due pianoforte soli," 458-67. In Strauinsky and "TheRite of Sping, " 23-38, Pieter van den
Toorn similarly describes the composition of passages out of sequence for that work.
Joseph Straus has observed this compositional process in Stravinsky's sketches for The
Rake's Progress (see Joseph Straus, "The Progress of a Motive in Stravinsky's The Rake's
Progress," TheJournal of Musicologj IX/z ( 199 1), 17 1-74),
"La forme est prkdominante, . . . ; la forme constructive. Des les debuts, ma
musique est constructive. . . . I1 me semble que tout l'art d'aujourd'hui tend i la concen-
tration constructive." Stravinsky, as quoted in an interview in Le V i n g t i k Siicle, 27 mai
1930. Cited in translation in V. Stravinsky and R. Craft, Strauinsky in Pictures and Docu-
ments, 256.
T H E JOURNAL OF MUSICOLOGY

EXAMPLE 6. S3: (a) transcribed from the first short score, p. 33 (solo
violin only); (b) first measure transcribed from p. 77 of
the sketchbook, second measure transcribed from p. 79.
Arrows and brackets added by this author.
(a)

Vno

(b)
["no

294 collection {C#,D, E#, F#, G#, A, A#, B]. In the solo violin, C# receives
- primary emphasis, while F# is highlighted toward the end of the pas-
sage before C# returns. The accompaniment incorporates limited
melodic movement in each part; nonetheless, violins and violas support
the solo violin's pitches through octave doubling or incorporation into
thirds, fifths, and triads. The bass line, in cellos and basses, is partially
independent of the upper parts~andsupports them inconsistently.
The new version of Sg presents both a new pitch-class collection
and new centric pitch classes (see Example 6b). It uses the collection
{C,D, E, F, G#, A, B), which includes neither of the previously empha-
sized pitch classes. In place of the solo's original emphasis on C # , the
new version stresses C. Where the original version created a secondary
emphasis on F#, the solo violin in the new version asserts A. Although
beginning melodically much as it did in the original, the solo in revised
S3 becomes much more complex. In its last two beats it adds a line and
ends on pitches different from its starting point. The revised accompa-
niment, on the other hand, has become more static and generally less
responsive to the now more adventurous solo. The entire texture is less
well integrated harmonically and more unstable overall than in the
original. This situation renders the return of familiar pitch structures in
Sq that much more powerful.
Despite these significant changes in its pitch organization, the re-
vised Sg remains indebted to the original for its texture and motivic
material. Some aspects of the motivic relationship between the two ver-
ROGERS

sions are noted on Example 6. The arrows and brackets indicate some
of the changes Stravinsky made to transform the solo line of the origi-
nal Sg into that of the new version. To create the solo line in Example
6b, m. 1, Stravinsky merely changed C# to C-natural. To create m. 2, bs.
1 and 3 of Example 6b, he transposed Aq-G#q-F#4 up a minor third to
become C;-Bq-Aq (upper line, b. I ) , and up a perfect fourth to create
Dg-Cg-Bq (lower line, b. 3). E#;-F#5-C#5 of Example 6a is trans-
posed up a minor third to become G#j-Aj-Eg (upper line, b. 3 ) , trans-
forming the original secondary emphasis on F# to one on A. The lower
line's ascending G#q-Aq-Bq in m. 2, b. 2 is an inversion, transposed up
a major second, of the Aq-G#q-F#q of Example 6a.
Although Stravinsky simplified the accompaniment when he re-
vised Sg, he retained the bass line's rebounding eighth notes and the vi-
olins' and violas' long durations and limited motion. More specifically,
the portion of the bass line that moves by third, originally A-C#-C#-A
(Example nb, bs. 3-q), is transposed to E-G#-G#-E for violas and al-
tered rhythmically (Example 6b, bs. 1-4). In addition, the violins' and
violas' G#3 and Dq on beats 3-4 of the original are preserved on the
same beats of the new Sg. In both versions, G#3 and Dq harmonize
beats on which the solo begins with Bq. Stravinsky retained the new Sg 295
-
for all later versions, including the published score.

Reasons for the Revision of Sg


MThy did Stravinsky revise Sg as he did? It may be in-
ferred from his retention of the motives and texture of the first version,
itself a slightly altered recapitulation of the first two bars of S i , that he
wanted Part A' to begin similarly to Part A and thus function as an audi-
ble return of opening material. The compositional history of the pas-
sage strongly suggests that the distinctive revisions in the pitch organi-
zation of Sg were prompted by the addition of Part B.
Most likely, the prominence and frequent recurrence of the pitch
class C#/DC in Part B provoked changes in Sg. After composing Part B,
Stravinsky probably reviewed his original version of Sg (see Example 3,
mm. 18 (b. 4)-20 (b. I ) ) ,with its marked emphasis on C#,and decided
that if this passage were to follow the new middle part, C# would re-
ceive excessive attention, especially if he intended to preserve the C#
initiating Sq. The revised Sg, with its new collection and emphasized
pitch classes C and A, provides relief from C# and thereby revitalizes its
reappearance in Sq. The new tonal juxtaposition between Sg and Part
B also heightens, in retrospect, the contrast presented by Part B.
The change from C# to C in Sg also gives Am'a I1 a newly significant
role in a drama in which it was originally only a minor player. This
THE JOURNAL OF MUSICOLOGY

EXAMPLE 7. Published score, mm. 1-2: Toccata's passport setting.

Corni-Fa l

Trombe-Do

296 I Tempo J = 96

Violoncelli

Contrabassi

important drama, based upon the conflict between C and C # , unfolds


through all four movements of the Concerto. The C/C# conflict is in-
troduced forte in m. 2 of the first movement's passport setting, where
the solo's C#q forms a cross relation with the cellos' and basses' C2 (see
Example 7 ) . While the passport settings of all four movements feature
the solo's C # , only the settings of movements I, 11, and IV also contain
the orchestra's C. In each case, the C#-C cross relation presages the de-
velopment of the same pitch-class rivalry in the body of the movement.
By extension, the absence of C and thus the symbolic cross relation
from the ritornello of Aria I1 should indicate the existence of few or
no signs of the C/C# conflict in the movement. The accuracy of this
assumption depends upon the stage in the compositional history of
ROGERS

Aria 11 to which it is applied. This history-specifically, that portion rel-


evant to the C/C# conflict-is best understood as a counterpart to the
drama itself, in which the pitch-class opposition is played out over the
course of the Concerto. A summary of this drama follows:
In the D-centric first movement (Toccata), C and C# are presented
as rivals of equal standing. They compete most frequently for the posi-
tion of seventh scale degree in a primary collection containing D, E, F#,
G, A, and B as its first six members, but also function in other capacities
in passages featuring centric pitch-classes besides D and collections
different from the primary one. Generally, a passage or instrumental
group within a passage uses either C or C# rather than mixing them
freely, causing either temporal or spatial juxtaposition of the two pitch
classes. The C/C# conflict, although not always the tonal focus, is active
throughout Toccata and left unresolved at its end, where the C# of the
theme's last soundings vies with the distinctive Mixolydian tint of the
final scalar descent to claim the listener's ear.
The second movement, Aria I, is also D-centric but uses primarily
D-minor scale collections. C and C# often appear in close temporal
proximity but in separate instrumental lines. Both pitch classes also
sound in passages exhibiting centricities other than D and collections 297
-
other than variants of the D-minor scale. Since C# is more prominent
than C overall in major cadences and at the ends of sections, C# gains
an edge in the rivalry by the end of the movement.
Given the importance and ubiquity throughout the Concerto of
the perfect fifth as melodic interval, harmonic component, and struc-
tural span, it is logical to conclude that Stravinsky's choice of F# centric-
ity for Aria 11 influenced the manifestation of the C/C# opposition
within the movement. In the original version of Aria 11, which lacked
Part B and featured the original Sg, C# prevailed decisively over C.
Only S2, also presenting dissimilarity in other areas, distinguished itself
with a nearly balanced presentation of C and C#.
The situation did not change initially when Stravinsky revised Aria
11, since the added Part B gives more weight to C# than to C. Not only
does C# belong to the contextually stable triads that open and close
Part B, it is emphasized in the solo violin for nearly the first three bars
of the passage. The lower-neighbor motive as it appears in Part B-
C#/DC ornamented by C/B#-symbolizes C#'s virtually unchallenged
position in the movement at this stage.
In contrast, the new Sg revitalized the conflict in an almost theatri-
cal manner. Stravinsky transformed the original C#-laden and -governed
Sg, in which C did not even appear, into a new passage in which C
usurped the roles played by C#, now vanished from the section. This
chapter of compositional history is literally audible: at the opening of
THE JOURNAL OF MUSICOLOGY

Sg, one expects C#-prepared by the close of Part B and suggested by


the function of Sg as a return of Si-and is confronted with C instead.
The shock of the substituted C juxtaposes the tones of expectation and
reality, producing dissonance in the mind's ear.
C# returns to begin Sq and is in effect for nearly the entire section.
With its reappearance, it completes a large-scale neighbor motion
that links focal pitch classes opening Part B ( C # ) ,Sg (C), and Sq (C#).
This neighbor motion connecting Part B to Part A' (and even backward
to Part A, which opens and closes with emphasis on C#) is foreshad-
owed in miniature by the aforementioned lower-neighbor motive
within Part B.
With his transformation of Sg, Stravinsky revived the C/C# conflict,
present only discreetly in the original version of the movement. In spite
of the impact of C's appearance in the revised Sg, C# proves to be the
more powerful of the two overall in Aria 1%After witnessing the fairly
equal presentation of the two pitch classes in Toccata, and the edge
granted C# in Aria I, a listener might conclude, upon hearing the C#-
heavy Sq complete Aria 11,that the rivalry had been resolved-but that
judgment reckons without the single C in the movement's last measure,
298 which casts enough doubt upon the preeminence of C# that the final
- word on the subject is left for Capriccio.
Like Toccata, Capriccio is D-centric and uses the same primary collec-
tion, in which C and C# vie for the seventh The two pitch classes
function in other capacities in passages featuring other centricities and
collections. Unlike Toccata, Capriccio stages a blatantly unbalanced ver-
sion of the conflict. The movement begins by choosing C#, in force for
most of the movement. Dissension resurfaces when C returns as the
movement nears its close in # i 19:4, and intensifies as both pitch classes
compete directly during the long cadential succession beginning in
#129:1. C#, highlighted in the solo violin's closing D-major scales, and
powered by its accumulated exposure and increased importance over
the course of the Concerto, emerges the victor.
Although in revising Sg, Stravinsky made the C/C# conflict a vivid
and essential aspect of Aria 11, the movement's passport setting remains
the only one in the Concerto without the representative C#-C cross re-
lation. Rather, where C sounds in the passport settings for the other
movements, G# appears in the ritornello of Aria I1 (see Example 5 ) .
G#2 is heard simultaneously with the solo's Gg, announcing a pitch-

ss Stravinsky did not compose Capicczo until after he revised Aria II. Still, a brief de-
scription of the C/C# conflict in the final movement is appropriate here since the revi-
sion of Aria II may have affected the manifestation of the rivalry in Caplccio, or, con-
versely, a precompositional decision to resolve the conflict within the fourth movement
may have affected the revision of the third.
ROGERS

class opposition prominent only in Aria II: G versus G#. The substitu-
tion of one pitch-class rivalry for another in the passport setting of only
Aria II likely corresponds to the movement's unique F# centricity.
Unlike the extended drama of the C/C# conflict, the tale of the
G/G# rivalry is told completely within Aria II. G# is featured initially
in S i , noticeable in both the solo's first descent-C#-B-A-G#-F# in
#77:2-3-and its accompaniment. Conflict is ignited when the second
version of this descent substitutes G for G# in the solo on beat 3 of
#77:4 and doubles it in second violins and violas while G# sounds in cel-
los. The two pitch classes vie for attention through the remainder of
Part A. The dissension is prolonged by Part B, suggesting yet another
reason for its creation. Part A' resolves the issue decisively: S3 and Sq
use only G#<: :
Perhaps when he omitted the C#-C cross-relation from the pass-
port setting of Am'a II, Stravinsky intended to let the issue take a back
seat to the G/G# particular to this movement. The first version of Aria
II supports this conjecture: the C/C# rivalry is present, especially in S2
and briefly at the end of Sq, but far from significant. Only with the revi-
sion of S3 did Stravinsky renew the C/C# conflict, which erupts just as
that involving G and G# resolves. Despite the magnitude of the changes 299
made to S3, Stravinsky did not revise the ritornello to reflect them. Nei- -
ther did he alter the ritornello's returns during the movement; virtually
identical to the opening two bars, they remain unaffected by events un-
folding within the movement.
The revision of Sg also creates within Part A' a new, large-scale
tonal framework linked to that heard in Parts A and B. In the original
Part A', consisting of the first Sg and Sq, emphasis in the solo moved
from C# (receiving primary stress in S3 and at the beginning of Sq) to
F# (goal of Sq), recalling the same shift of emphasis in S i . The revised
Sg replaced C# with centric C, allowing C-C#-F# to emerge as struc-
turally prominent in Part A'. C-C#-F# is similar to the solo's tonal scaf-
folding in Parts A and B, C#-F#-GC#: both reduce to [o, i,6] trichords
and contain C# and F#. A significant difference is in t6e positions of C#
and F#. Specifically, relocation of the C#-F# leg from the beginning of
the pitch-class framework articulated by Parts A and B to the end of
that heard in Part A' confirms F# as centric in Aria II. It is noteworthy
that Si and Sq, which both move from focal C#to focal F#, contain the
numerous local, scalar descents discussed above that connect these two
pitch classes. The cumulative effect of these embedded descents from
C# to F# is to make the close of both sections on F# seem inevitable.

z:$ The revision of S g had no effect on the G/G# conflict. Both the original and re-

vised S g use only G#.


T H E JOURNAL OF MUSICOLOGY

The Second Short Score


After completing his sketches of Part B and the re-
vised S3, Stravinsky notated a new short score, dated l o June 1931
(film 14,' 14).AS noted in Table 2 , he incorporated in this second short
score the new Part B and the revised Sg. The second short score differs
from the published version primarily in minor details. The most no-
table difference involves tempo. The revised short score retains the
original tempo of eighth = 60 for the entire movement. Later versions,
including the published score, assign eighth = 48 to the outer parts
only; as a result, the contrast provided by Part B is strengthened fur-
ther.24
After finishing the second short score and thus "completing" Aria II
for the second time, Stravinsky set to work on a new piano-violin score,
this one actually sent to Strecker, probably on 1; J ~ n e . ~He
5 sent the
full score of the movement to the publisher in mid-September.26

Conclusions
The foregoing study of the history of Aria II reveals
300 that, after he thought that he had completed the movement, Stravinsky
-
composed and incorporated a new, central passage that produced a sig-
nificant change in form. Rather than describing this change as simply
from "ritornello to ternary," it would be more accurate to state that the
movement exhibited elements of both forms for much of its composi-
tional history and that the relative weighting of the forms was what
changed. In the first short score, ritornello form was predominant, and
ternary form only a suggestion. The ritornello and solo sections were
retained by the second short score, but, with the insertion of Part B, the
balance of power shifted to the ternary structure.
Analysis of versions of the movement indicated that Stravinsky
probably composed the new passage, Part B, to respond to what he per-
ceived as not adequately expressed or even as lacking in the original
version, comprising Parts A and A' only. The characteristics of these
lacks are implied by the primary functions of Part B: to provide greater
contrast and juxtaposition, hallmarks of Stravinsky's style, and to enrich

24 The new tempo for Parts A and A' adds approximately 32 seconds to Part A, 18
seconds to Part A', and thus approximately 50 seconds to the entire movement.
z s ". . . Tomorrow morning I will send the manuscript (piano and violin) of the
third part (Aria 11) ." Letter of 1 4June 193 I from Stravinsky to Strecker in Stravinsky, Se-
lected Correspondence 3 , 226. Strecker acknowledged receipt of the manuscript in his letter
to Stravinsky of 24 June (letter in the collection of the Paul Sacher Foundation).
z" In his letter to Strecker of Saturday, 1 2 September 1931, Stravinsky expressed his
intention to send, o n the following Monday the orchestral score of Ana I1 (Stravinsky, Se-
lected Correspondencr 3, 2 2 7).
ROGERS

connection through deeper exploration of significant motives and


pitch structures.
Although the contrast and connection created by Part B are new,
their origins are found in earlier versions of the movement. The seeds
for ternary form lie in the first short score in the subtle contrast pre-
sented by S2 and the return conveyed by Sg and Sq. The solo violin's
framework of emphasized pitch classes in Part B originated in Part A.
The stepwise descent from G to C# that closes Part B may be inter-
preted as an answer to descents from C# to F# in Si and Sq and to the
too-short descent to C# in S2. Thus, the late composition of Part B,
and, thereby, the movement's ternary form, may be seen as Stravinsky's
response to the content of the original outer sections.
The content of Part B, in turn, triggered a response in the form of
the revised Sg. Presumably a reaction to the additional emphasis on C#
in Part B, the new Sg, with its focal C, makes possible a long-range
neighbor motion connecting Parts B and A', and, of even greater im-
port, a role for Am'a IIin the unfolding of the C/C# conflict already de-
veloped extensively in the first two movements.
The idea that the content of Parts A and A' influenced the ultimate
form of the work is consistent with many statements, attributed to 301
-
Stravinsky, regarding decisions about form during the compositional
process. At different points in his career, Stravinsky claimed that, typi-
cally, he did not know the form of a work from the outset of composi-
tion, but rather that the form made itself known as the musical materi-
als were developed. The following two comments, from Stravinsky's The
Poetics of Music, written eight years after the completion of the Violin
Concerto, are particularly telling:p7 "Step by step, link by link, it will be
granted him [the creator of the work] to discover the work.'jZ8"Usually
when I set to work my goal is not definite. If I were asked what I wanted
at this stage of the creative process, I should be hard pressed to say."Zg
Later, in Conuersations with Igor Stravinsky, Stravinsky comments that
'You should not suppose that once the musical idea is in your mind you
will see more or less distinctly the form your composition may evolve.":cO

* i It is difficult to know how much credence to grant these words since The Poetics of
Music was written not by Stravinsky, but by Roland-Manuel, with the assistance of Pierre
Suvchinsky. Still, the text of these Harvard lectures was based o n Stravinsky's notes and
discussion with Roland-Manuel, and apparently approved by the composer, who viewed
proofs of the lectures and read them himself at Harvard. For commentary o n and a
chronology of the writing of The Poetics of Music, as well as a transcription of Stravinsky's
notes for the lectures, see Stravinsky, Sekted Cmespondence 2 , 503-17,
z 8 Igor Stravinsky, The Poetics of Music in the Form of Six Lessons, Arthur Knodel and
Ingolf Dahl, trans. (Cambridge, MA, 1970; first published as Poitique musicale [Cam-
bridge, MA, 1 9 4 2 ] ) ,50.
z t ~ Ibid., 69.

:$'I Stravinsky and Craft, Conversations u~ithIgor Stravinsky, i 5 .


THE JOURNAL OF MUSICOLOGY

He even defines musical form as ". . . the result of the 'logical discus-
sion' of musical material^,":<^ evoking an image of its creation as
content-based, rather than predetermined.~~
Stravinsky's sketches for Aria 11 do not document definitively
whether he began with a predetermined design or, alternatively, em-
ployed a content-based approach during early stages of composition.
Given Stravinsky's habit of composing at the piano, it is unlikely that
complete written records of these beginning stages were made. The
presence of the ritornello in the first sketches, however, suggests that
Stravinsky conceived of Aria 1 1as a ritornello form fairly early in the
creative process. Still, the ritornello form may very well have been
inspired by content, specifically by the "passport," with which both
the completed Toccata and Aria I had begun. Given the dearth of docu-
mentation, theories of the movement's formal origins must remain
conjectural.
On the other hand, evidence provided by the sketches and analysis
of the new relationships spawned by the creation of a ternary form
demonstrate that, ultimately, content did indeed influence form signifi-
cantly. After all, Stravinsky could easily have planned the basic construc-
302 tion of a ternary form precompositionally. Instead, this design arose
- only during the process of composition, as necessitated by the content
of the movement's outer parts. In this light, reading the published
score of Aria 11is a way of looking back into time: most of the music of
the original ritornello form is present, but reconfigured by Stravinsky's
late adoption of ternary form.
Seeing content as determining or influencing form is meaningful
not just for Aria II, but in general for Stravinsky's non-programmatic
neoclassical works. Analysis of these works should seek to discover, as
much as possible, how the form-whether new or traditional-is appro-
priate to, and even required by, the content.33 The forms of Stravinsky's
works might be most effectively understood as responses to, rather than

:<'Ibid., p. 19.
3 . The term "content-based," as applied to form, is the innovation of James H e p
okoski in his work on the music of Sibelius. For Sibelius, the goal of content-based com-
position was the invention of new formal structures that would serve as alternatives to
those of the Fonenlehre tradition. See James Hepokoski, Sibelius: Symphony No. j (Cam-
bridge, 1993), =.
Carl Schachter takes this analytical stance in his discussion of the first movement
of Brahms's Symphony No. 2 (see Carl Schachter, "The First Movement of Brahms's Sec-
ond Symphony: The Opening Theme and Its Consequences," Music Analysis 11/ I [ 19831,
.55-68).
. In the article's Introduction, Schachter quotes statements, attributed to Brahms,
that characterize sonata form as a consequence of its themes (p. 55).
ROGERS

containers for, their musical materials.34 Viewing musical content in a


new role-as agent in the creation of form-might stimulate new dis-
coveries about the relationship between musical architecture and musi-
cal materials in Stravinsky's works.
Oberlin College,
Conservatory of Music

.M For example, consider the first movement of the Symphony in C, well-known as


an example of Stravinskian sonata form. The tonal ambivalence of the opening figure,
B-C-G, soon evolves into the polarity behveen C and E that forms the basis of the move-
ment's tonal conflict. Stravinsky's first notation for this symphony, which, according to
Robert Craft, was dated over a year before Stravinsky began composing the work in
earnest, shows only a melody vaguely resembling the symphony's first theme, but includes
the famous three-note motive, then beginning on C#. (For Craft's comments and tran-
scription of this sketch, see V. Stravinsky and Craft, Stravinsky in Picturer and L)ocumats,
334.) Isn't it possible that the opening motive's potential for tonal ambivalence might
have inspired the choice of sonata form?

You might also like