Locanto - Stravinsky's Late Technique

:~ssI:IiI~No ioc~N1o
Co:iosINo vI1n IN1riv~is: IN1riv~iiIc S.N1~x ~No SriI~i

TrcnNIqir IN L~1r S1i~vINsi.musa_302 221..266
Ti~Nsi~1ro n. Cn~ovIci JrNiINs
The use of intervals as the basic material of musical construction consistently
served as an important and deeply personal characteristic of Stravinskys com-
positional process. In his nal serial compositions, however, this aspect assumed
a more decisive role and underwent signicant changes. Stravinskys late writ-
ings, composed in collaboration with Robert Craft, reect this renewed interest
in intervallic construction. In them, the composer repeatedly describes the rst
stages of the creative process as work with intervals and even projects this
practice backwards to cover the entirety of his musical output. Stravinsky thus
emphasises, perhaps excessively, the continuity of his thinking despite the
evident changes that took place throughout his compositional career.
1
The new and more important role occupied by the intervallic component in
Stravinskys serial compositions is, to a large extent, a consequence of the
gradual abandonment of pitch collections (octatonic, whole-tone, diatonic, and
so on),
2
which had played such a decisive role in his earlier music, in favour of
a growing tendency to make systematic use of twelve-note aggregates a ten-
dency accentuated particularly in the compositions which follow Threni.
3
Although Stravinsky continued to prefer the same pc sets (particularly tetra-
chords) which in his earlier compositions were derived from diatonic or octatonic
collections, the sketches for his last compositions seem to indicate that the
creative process has moved from specic intervals to larger combinations (and
not vice versa), thereby producing harmonic environments which can appear
variously diatonic, octatonic or chromatic.
By freeing the treatment of intervals from a broader system of pitch organi-
sation,
4
Stravinsky seems to have followed a path which presents some similari-
ties to but also some profound differences from the predominantly motivic
pathway followed by Schoenberg and his pupils in their gradual departure from
traditional tonality. Although all of these composers treated the intervallic com-
ponent in different ways, it became for them the foundational aspect of a
motivic technique that is, one based on the use of a restricted number of
intervallic congurations which serve a basic unifying function within a musical
work. Yet the specic way in which Stravinsky used intervallic motives emerges
through a study of his many unpublished versions of pieces, and these must be
evaluated in relation to the peculiarities of his aesthetic.
DOI: 10.1111/j.1468-2249.2011.00302.x
Music Analysis, 28/ii-iii (2009) 221
2011 The Author.
Music Analysis 2011 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK
and 350 Main Street, Malden, MA 02148, USA
All of this raises an interesting issue: although in so-called atonal music the
(presumable) absence of a hierarchy among the sounds allows the composer to
employ all twelve pitch classes as he pleases (considering them exclusively in
terms of their intervallic relationships),
5
with the adoption of serial procedures a
new constructive order is imposed. Now, while in the work of other composers
one could say that this new constructive order encompasses and essentially
identies with motivic-intervallic syntax, in Stravinsky the result of such identi-
cation is instead rather problematic, because these two aspects operate accord-
ing to slightly but signicantly different criteria. In the following pages, I
will attempt to demonstrate how, particularly in the compositions from Agon
onwards, this approach provided Stravinsky with a stimulus, rather than an
obstacle, to composition. I will also attempt to interpret some characteristics of
Stravinskys creative process and serial technique which are by now well known
but whose deeper motivations still require further investigation.
Motivic-Intervallic Syntax: General Characteristics
Even if it is evident that the use of intervals as the basic material of the
compositional process constitutes a central feature of Stravinskys late musical
thought, the specic technical means employed lend themselves to being
described in rather different ways.
6
Stravinsky never specied exactly what he
meant by the expression composing with intervals, nor does the study of his
sketches offer any denitive answers. A great deal of his work with pitch material,
in fact, took place at the keyboard, in a phase prior to that documented in the
earliest sketches, which in actuality record a stage in the creative process that is
already quite advanced.
7
Given this modus operandi one can easily imagine that
Stravinskys interval-based procedures were not codied according to any kind
of systematic approach. This, however, does not exclude the possibility that such
procedures can be described retrospectively in theoretical terms.
To this end, I will use as a brief rst example the original twelve-note row of
A Sermon, a Narrative, and a Prayer (Ex. 1). The row can be subdivided into four
distinct trichords, three of which the rst, second and fourth belong to set
class [014]. Segments 2 and 4 are ordered as <0, 1, 4>, while segment 1 is
ordered as <1, 0, 4>.
8
From an intervallic point of view, if we consider the
intervals apart from their melodic direction (ascending or descending), that is, as
unordered pitch-class intervals (interval classes), all three segments contain a
Ex. 1 Stravinsky, A Sermon, a Narrative, and a Prayer: subset structure of the original
twelve-note row
[014] [014] [014] [015]
1 1 1 3 3 4
3 4 4
222 :assi:iLiaNo LocaN1o
Music Analysis, 28/ii-iii (2009) 2011 The Author.
Music Analysis 2011 Blackwell Publishing Ltd
semitone (ic1), a minor third (ic3) and a major third (ic4).
9
In consequence, we
could describe the three segments as three statements, differently ordered, of the
same group of three interval classes, or equally well as three statements, differ-
ently ordered, of the same set class [014].
Considering the segments of the row as unordered sets corresponds to a
constructive logic which, far from being exclusive to Stravinsky, seems deeply
rooted in most twelve-note and serial music in general and is certainly very
familiar from the published literature on serialism. Many of the basic operations
which concern the subset structure of the twelve-note row are based on this
logic. In Schoenberg, Berg and Webern, the rows are often organised in such a
way as to maximise certain segments which, if considered apart from the order
of the pitches, belong to the same set class. To mention only a couple of
examples, one could cite the row of Schoenbergs String Quartet No. 4, in which
one can identify four segments of three notes as belonging to set class [015]
(Ex. 2a),
10
or the row employed in the twelve-note section of the third movement
of Bergs Lyric Suite, which contains four segments belonging to the class [0126]
(Ex. 2b).
11
Furthermore, in many twelve-note compositions the idea of considering some
segments of the row as unordered sets constitutes an essential premise for
establishing various types of formal relationships. The internal structure of the
row, in fact, allows some of its subsets to preserve the same global pitch content
even after the typical transformational operations (transposition, retrograde and
inversion) are applied.This gives rise to a network of relations among the various
forms of the row employed in a composition.
12
The very notion of hexachordal
combinatoriality,
13
which plays a fundamental role in Schoenbergs twelve-note
music, is based on the possibility of conceiving the hexachords as unordered
collections.
The global intervallic content of these serial segments of course plays an
important role in enabling this type of relation from the moment that each
Ex. 2a Schoenberg, String Quartet No. 4: subset structure of the fundamental
twelve-note row
[015] [015]
[015] [015]
Ex. 2b Berg, Lyric Suite: subset structure of the twelve-note row of the third
movement
[0126] [0126]
[0126] [0126]
Co:iosiNc wi1n IN1cnvaLs 223
segment can be considered either as an unordered set of pitch classes or as an
unordered set of interval classes. Particular attention to intervallic content forms
the basis of Weberns practice of deriving a row from the reiteration of a single
basic cell. To mention only one well-known example: in Cantata No. 1, Op. 29
(Ex. 3),
14
the four discrete trichords of the row are all members of the same set
class [014]. The importance of the global intervallic content of set class [014]
(semitone, minor third and major third) is underlined by the presence of the
semitone between the rst and second segments and between the third and
fourth segments, and by that of the minor third between the second and third
segments.
To summarise: the idea of globally considering the pitches and/or intervals
contained in some serial segments typically constitutes one of the basic con-
structive criteria of twelve-note serialism. Nevertheless, this criterion corre-
sponds only in part to the concept of intervallic motive which I hope to
illuminate in the music of Stravinsky. Generally, the constituent pitches of a
motive can be used in either a harmonic or a melodic sense,
15
and compared in
any order. Nevertheless, Stravinsky, working with the orientation of single inter-
vals, radically modies the physiognomy of his motives, which can thereby
assume forms corresponding to different set classes. From this point of view,
then, an intervallic motive no longer corresponds, in any sense, to a class of
unordered pc sets. Rather, Stravinskys operations act more on the level of single
intervals than on the level of the global congurations within which these single
intervals are included.
The difference becomes clear if we return to the third segment of the row of
A Sermon, a Narrative, and a Prayer (see again Ex. 1). This segment belongs not
to set class [014], but rather to the class [015]. As a consequence, its global
intervallic content is different. However, this segment shares with the other
segments two of its three intervals (ic1 and ic4), which are merely arranged
differently (Ex. 4): in the third segment they are joined in the same direction,
thereby producing an ic5; in the rst, second and fourth segments they are joined
in opposed directions, thereby producing an ic3 (Ex. 4).
16
In short, considered as
unordered pc sets, only three of the four segments of the row of Ex. 1 turn out
to belong to the same class; considered, however, as intervallic motives formed
through the combination in varying directions of two intervals, they turn out
to all be members of the same motive class (Ex. 5): the semitone and the major
third conjoined, expressed symbolically as 14.
17
Ex. 3 Webern, Cantata No. 1, Op. 29: subset and intervallic structure of the original
twelve-note row
[014] [014] [014] [014]
O RI
1
3 1
This simple alteration in intervallic direction constitutes a precious compo-
sitional resource in Stravinskys hands. Joining various forms of a single (or at
most two) motivic class(es), he creates twelve-note rows as in the previous
example as well as smaller or larger successions of pitches to be employed
either melodically or harmonically in a musical passage. For instance, in Ex. 6 we
can observe the succession of pitches which serves as the basis for three episodes
(bars 722) included in the rst section (up to the prima volta) of the rst of the
ve Movements. This passage conceals a closely woven fabric of overlapping
motives of the semitone-tone (12) and semitone-tritone (16) types (indicated
by square brackets). Depending on the orientation assumed by the two intervals,
the rst motive (12) produces sets of three pitches belonging to set classes [012]
and [013]. The second motive (16), on the other hand, produces collections
belonging to set class [016] regardless of the orientation assumed by the two
intervals.
18
It should be apparent that, in dening this type of intervallic syntax as motivic,
the term motive is being used with some degree of latitude. In the Formenlehre
tradition, for example, a motive is typically conceived as a structural nexus of
rhythm and intervals.
19
In Stravinskys music, however, a motive is essentially an
abstract conguration of intervals: pitch components and rhythm are treated as
initially distinct and separate dimensions which can subsequently be related.
20
As
suggested above, in spite of some apparent similarities, this conception of inter-
Ex. 4 Intervallic motive class 14 in the two forms of set class [015] and [014]
5
1
1
0 0
ic4
(ic5)
(ic4)
ic3
ic1
ic1
4
14 motive
Ex. 5 Stravinsky, A Sermon, a Narrative, and a Prayer: motivic-intervallic structure of
the original twelve-note row
[014] [014] [014] [015]
1 1 1 3 3 4
3 4 4 4
5 1
E D B A
C B G
F
F
E D A 4 4
4
4
(3)
(5)
(3) (3)
1
14 motive
1 1 1
vallic motive also differs radically from that familiarly applied to the post-tonal
music of Schoenberg, Berg and Webern. For these composers, the concept of
motivic elaboration which guaranteed coherence in tonal music was gradually
replaced by a constructive principle based on the use of fundamental intervallic
constellations which operate at a more basic level. According to Martina
Sichardt, this reduction of the various Gestalten within a passage to its most
elementary intervallic basis a tendency Schoenberg himself had consciously
put into practice in his own analytical formulations represented a fundamental
premise for the elaboration of the twelve-note method.
21
In this respect, it is
interesting to note that most of the basic intervallic constellations which form the
expressive vocabulary of melodic gestures in Schoenbergs compositions consist
merely of the union of two or three intervals one of which is usually the
semitone disposed in a particular arrangement.
22
An interesting analogy with Stravinskys practice can nonetheless be glimpsed
wherever Schoenberg subjects these basic combinations of intervals to a process
of variation. Jack Boss, for example, has demonstrated that the majority of the
intervallic motives in the rst of Schoenbergs Vier Lieder, Op. 22, could be
derived by applying three types of modication to a motive formed from the
combination of one ic1 and one ic3.
23
Boss considers all of the possible arrange-
ments of these two intervals (that is, <+1, +3>, <+1, -3>, <-1, +3>, <-1, -3>, <+3,
+1>, <+3, -1>, <-3, +1> and <-3, -1>) as variants belonging to the same motivic
category. Moreover, each of these forms can undergo in its turn three funda-
mental types of variation, two of which involve octave complementation and
pitch reordering. All of this corresponds exactly to my denition of motive
class 13. However, despite this similarity, a profound discrepancy remains
regarding the very concept of motive. Schoenbergs procedures, as described
by Boss, effectively identify motive as an entity which may be subjected to a
wide range of transformations while remaining largely recognisable. According
to Boss, for example, the third basic category of variation employed by
Ex. 6 Stravinsky, Movements, i: motivic-intervallic structure of the succession of
pitches contained in each of the three solo episodes of bars 722
[012] [016] [016] [016] [016] [016] [016] [012] [012]
[012] [016] [013] [013] [016] [016] [016] [012] [012]
[016] [016] [016] [016] [012] [012] [013]
12 motive 16 motive
[013] [012] [016] [016]
1 1 1 1
2 6
(3)
[7] = 5
2
6
(5) (1)
Schoenberg involves the expansion of intervals.
24
In this respect, the Schoenber-
gian concept of variation implies a decisively greater quantity and variety
of forms derivable from a single motive than those which occur in Stravinsky.
Still more important is the fact that the Schoenbergian concept of variation
implies a broader process, one which involves the entire plan of the musical form.
Indeed, in Schoenberg, the variation of a motive cannot be dissociated from
the concept, central to the Austro-German tradition, of motivic elaboration,
understood as a means of conferring coherence and organic unity on a compo-
sition.
25
All of this is foreign to Stravinskys musical thought, in which the
manipulation of intervallic motives is understood as a procedure for generating
primary compositional material capable of being employed as a point of creative
departure.
The idea developed in particular through the work of George Perle
26
that
a basic cell or referential sonority can be presented even in a vertical sense
could be considered another point of contact between Stravinskys motivic-
intervallic syntax and the post-tonal harmonic language of Schoenberg, Berg
and Webern. However, unlike the notion of the Stravinskian intervallic motive,
the concept of the basic cell consists of a xed conguration of intervals and
corresponds therefore to a single set class.
27
The most decisive difference,
however, concerns the contrasting aesthetic-musical aspects within which a
motive unit is taken to function: in the music of Schoenberg, Berg and Webern
an intervallic conguration disposed vertically always maintains a motivic char-
acter from which, in fact, the idea of chord as motive arises even in a
dynamic sense. The nature of this element is expressed by the Schoenbergian
concept of unrest:
What is a motive? A motive is something that gives rise to motion. A motion is that
change in a state of rest, which turns it into its opposite. Thus, one can compare a
motive with a driving force ... . What causes motion is a motor. One must distin-
guish between motor and motive ... . A thing is termed a motive if it is already subject
to the effect of a driving force, has already received its impulse, and is on the verge of
reacting to it ... . The smallest musical event can become a motive if it is permitted
to have an effect; even an individual tone can carry consequences. (Schoenberg
1995, p. 386; emphases in original)
In the music of Schoenberg, the simultaneous presentation of pitches produced
by an intervallic conguration can be considered the result of an extreme
concentration in time of an event whose essence is decidedly dynamic tied,
that is, to the movement of time. Therefore, if Schoenbergs motivic conception
is essentially temporal the very idea of a suppression of temporality associated
with the Schoenbergian law of the unity of musical space implies in actuality
the concept of time the Stravinskian approach conversely returns to a con-
ception we may dene as spatial or plastic-visual: the motive is understood as
a conguration of intervals which can be arranged in two dimensions, as in
visual space.
Theoretical Aspects
Before examining some specic examples, a number of purely theoretical con-
siderations are worth reviewing in greater detail. As mentioned earlier, the
possibility of freely orienting any intervallic conguration ensures that a single
motivic class corresponds to more than one set class. With respect to the motives
formed from two different intervals (the model Stravinsky most frequently
employs), each of the fteen possible combinations generally produces two
different set classes, depending on the orientation of the two intervals conjoined.
Only those motives containing the tritone produce a unique set class regardless
of intervallic orientation (see the left-hand side of Table 1).
28
The right-hand side of Table 1, however, shows the motives which are capable
of producing a particular set class of three pitches. For example: the set class
[013] can be obtained by uniting one ic1 and one ic2 in the same direction
(motive 12), or by uniting a single ic2 and a single ic3 in opposite directions
(motive 23). As can be seen, according to the global intervallic content,
29
each
set class can be produced by a variable number of up to three intervallic motives.
Only set class [048], which contains three identical intervals (three ic4s), cannot
be produced by any motive formed from two different intervals.
The motives formed through the union of three different interval classes do
not constitute an analytically relevant object since too many different set classes
are generated. For example, a motive which combines a semitone, tone and
minor third in whatever order and direction would produce ten different set
classes containing either three or four distinct pitch classes:
30
Table 1 Motives formed by two different conjoined intervals
Motivic-
intervallic
class
Set class(es)
produced
Set class Associated
motivic-intervallic
class(es)
12.............................................(012) (013) (012) ............................12
13............................................ (013) (014) (013) ............................12 13 23
14.............................................(014) (015) (014) ............................13 14 34
15.............................................(015) (016) (015) ............................14 15 45
16 .............................................(016) (016) ............................15 16 56
23.............................................(013) (025) (024) ............................24
24.............................................(024) (026) (025) ............................23 25 35
25.............................................(025) (027) (026) ............................24 26 46
26 .............................................(026) (027) ............................25
34.............................................(014) (037) (036) ............................36
35.............................................(025) (037) (037) ............................34 35 45
36............................................ (036) (048) ............................
45.............................................(015) (037)
46 .............................................(026)
56 .............................................(016)
123: [013][023][0123][0124][0125][0134][0135][0136][0146][0236]
Moreover, the majority of these ten set classes can be derived from other motives
covering three different intervals. Thus every set class can be associated with an
excessively broad number of motives and vice versa.
However, motives that may be realised using only two interval classes, one of
which repeats once (for example, 121) to form a set of four pitches, are
relatively common. In Stravinskys case, motives of this type are typically those
which employ the semitone in conjunction with the whole tone or perfect fth
(121, 212, 151 and 515). As can be seen on the left-hand side of
Table 2, these motives produce only two or three different set classes, according
to the orientation assumed by the intervals. The right-hand side of the same
table, on the other hand, show how two different motives of this type can, at
times, produce the same set class. For example, the set class [0123] is obtained
by both 121 and 212.
Intervallic Syntax and Serial Technique
The major discrepancy between the motivic-intervallic syntax described so far
and serial technique as conceived by Stravinsky himself consists in the fact
that while the rst operates predominantly on the level of single intervals, the
second acts essentially on the level of pitch-class sets, understood as the units of
primary structural value.
31
Different orientations of the single intervals of a
motive can produce forms which belong to different set classes; by contrast,
neither the retrograde, nor the inversion, nor the retrograde inversion, nor any
type of permutation of the order of a particular pc set is capable of generating a
different set class. From the point of view of musical perception, one could even
say that motivic-intervallic syntax attributes to the quality of single intervals an
importance superior to the globalising tendency of pc sets. Put simply, intervallic
logic tends towards disintegration, serial technique towards unication.
Table 2 Motives formed by two different intervals, with one of them repeated once
Motivic-
intervallic
class
Set class(es)
produced
Set class Associated
motivic-intervallic
class(es)
121............................(012) (0123) (0134) (012)..............................121
212............................(0123) (0235) (0123)............................121 212
151............................(0145) (0156) (0167) (0134)............................121
515............................(0156) (0167) (0158) (0145)............................151
(0156)............................151 515
(0158)............................515
(0167)............................151 515
(0235)............................212
Despite these discrepancies, in Stravinskys compositional thought the two
aspects seem to aim towards the same end. In order to clarify how this occurs,
consider Ex. 7, which reproduces the three choral statements at the beginning,
in the middle and towards the end of the Dies irae of the Requiem Canticles.
These three fragments constitute an autonomous formal layer which interacts
with the surrounding layers of contrasting musical material (omitted from the
example). The rst choral statement (bars 8283) divides into two parts: in the
rst part the chorus (doubled by the brass) intones the words Dies irae on a forte
chord repeated in a dotted rhythm; in the second, the single word irae (chorus
and horns con sordini) is repeated as an echo to a piano chord which bears a
certain afnity with the preceding harmony. The second statement (bar 86) is
limited instead to a repetition of the forte chord on dies illa, and without the
echo response. The third statement (bar 97 onwards) is essentially a recapitula-
tion of the rst, although it presents a slight harmonic departure.Thus, the entire
layer assumes a type of ABA form.
My analytic symbols placed below the score in Ex. 7 show the motivic-
intervallic construction of this layer. The two chords of the rst statement (bars
8283) correspond to the two forms [015] and [016] of the motivic class 15.
Moreover, notice that these forms share the pitches E
and A
, which together
form ic5. The impression that the rst chord is echoed by the second (come eco)
therefore derives not only from the presence of two common tones, but also from
the intrinsic motivic-intervallic afnity of the two harmonic simultaneities. The
second choral statement (bar 86) opens onto a symmetrical sonority, a member
of set class [0156] containing two ic1s and two ic5s. This sonority is obtained
through the sum of the two 15 motives appearing in the two chords of the rst
statement: E
[015] + E
B [016] = E
B [0156]. This time the

chord is not simply repeated: in the middle of the bar, the lowest voice moves a
semitone from B to B
, thereby giving rise to a sonority containing an ic1 (FF
)
and two conjoined ic5s (A
). The recapitulation (bar 97) is practically

identical to the rst statement; nevertheless, in the rst part (the forte chord on
Dies irae), the bass moves a semitone from F
to G. Finally, notice that the

perfect fth A
is a constant presence throughout the layer in its entirety, thus

forming a kind of operative tonal axis.
The sketches transcribed in Ex. 8a and 8b concern the composition of the
same formal layer. Ex. 8a reproduces a strip of paper containing an early version
of the rst statement (bars 8283), preceded by the pre-emptive instrumental
gesture which introduces it (bar 81).
32
Note that in this version the harmony of
the choral part includes a move of a semitone from F
to G in the bass voice, a

solution which Stravinsky will subsequently adopt for the varied recapitulation
(compare Ex. 8a with Ex. 7).
33
Ex. 8b reproduces a little sheet containing two
different versions in the upper and lower systems respectively of the rst and
the second statements, worked out as an unbroken succession. Here the second
statement adumbrates the echo response, which in the nal version Stravinsky
uses only for the rst statement (and for the recapitulation). In the rst version
E
x
.
7
S
t
r
a
v
i
n
s
k
y
,
R
e
q
u
i
e
m
C
a
n
t
i
c
l
e
s
,
D
i
e
s
i
r
a
e
,
b
a
r
s
8
1
8
3
,
8
6
a
n
d
9
7
9
8
:
m
o
t
i
v
i
c
-
i
n
t
e
r
v
a
l
l
i
c
c
o
n
s
t
r
u
c
t
i
o
n
P
i
a
n
o
T
i
m
p
a
n
i
D
i
e
s
i
r
a
e
,
i
r
a
e
,
{
c o m e e c o
C
O
R
O
T
.
B
.
A
.
S
.
C
o
r
.
c
o
n

s
o
r
d
.
I
I
I
.

I
V
I
I
{
T
r
.

I
.

I
I
T
r
b
n
.

I
.

I
I
t
e
n
.
D
I
E
S

I
R
A
E

=

1
3
6

(

=

6
8
)
8
2
D
i
e
s
i
r
a
e
,
d
i
c
o
m
e

e
c
o
{
9
7
{
8
6
d
i
e
s
i
l
l
a
,
C
O
R
O
S
.
A
.
B
.
T
.
T
r
.

I
.

I
I
T
r
b
n
.

I
.

I
I
t
e
n
.
,,
c
o
m
e

e
c
o
{
T
r
.

I
.

I
I
T
r
b
n
.

I
.

I
I
t
e
n
.
C
o
r
.
c
o
n

s
o
r
d
.
I
I
.

I
V
I
.

I
I
I
C
O
R
O
S
.
A
.
B
.
T
.
[
0
1
5
]
[
0
1
6
]
[
0
1
5
6
]

=

[
0
1
6
]

+

[
0
1
5
]
[
0
1
6
]
[
0
1
5
]
E
F
A
E A B
A BE
F
B
E
E
A
A
F
G
B
5
5
5
1
1
1
5
5 1
1
1
1
3
of the passage (the upper system), the chord of the second statement is a series
of three conjoined perfect fths (BF
). In the second version (the lower

system) the B is modied to a B
. The alteration forms set class [0157], which

contains two conjoined ic5s (F
) along with an ic1 between B
and C
. In
the nal version, Stravinsky preferred a sonority containing two ic1s and two
ic5s, as we have already seen. In the event, all of the variants in the sketches, like
the nal version, can be interpreted from the point of view of a systematic use of
combinations of ic1 and ic5.
The only serial symbols discernible in the sketches are on the page transcribed
in Ex. 8b and refer to the second of the two fundamental twelve-note rows
employed in the Requiem Canticles,
34
or, more precisely, to the two rotational
arrays generated respectively by the rst hexachord of series I (Ia) and the rst
hexachord of series RI (RIa) of Ex. 8c.
35
Without going into detail on the various
properties of this type of table and the ways of using it,
36
I will briey describe
its construction. The pitches of the original hexachord are rst made to rotate
systematically from right to the left: the rst rotation begins with the second
pitch of the original hexachord, through which the rst pitch moves into the nal
position; the second rotation begins with the second pitch of the rst rotation
(the third pitch of the original), and so on for ve iterations (after which it
returns to the original form). The ve rotated forms thus obtained are trans-
posed successively so that they all begin on the same pitch as the original
hexachord (in the specic case of Ex. 8c, F for the forms generated by hexachord
Ia; G for the forms generated by hexachord RIa). Each of the ve rotated-
transposed forms thus obtained contains the same succession of intervals
Ex. 8a Stravinsky, sketches for Requiem Canticles, Dies irae (compare bars 8183 of
the printed score) (Paul Sacher Foundation, Igor Stravinsky Collection)
B
T
Di es i
3
rae
3
i - [illegible]
S
A
3
3
eco
3
3
Tmp
II Inv.
globally each time beginning at a different point within the succession but
different pitch classes.
37
The symbols on the sketch shown in Ex. 8b clearly indicate that the chords of
the choral part result from the combination of dyads freely selected from the two
Ex. 8bd (b) Stravinsky, sketches for Requiem Canticles, Dies irae (cf. bars 8183, 86
and 9798 of the printed score) (Paul Sacher Foundation, Igor Stravinsky Collec-
tion); (c) rotational arrays of the hexachord a of the inversion (left-hand column)
and retrograde inversion (right-hand column) of twelve-note row II, with encircled
serial segments employed in the upper sketch (the circles and connecting arrows are
not part of Stravinskys original autograph); (d) motive 15, in the forms of sets
[016] and [015]
T
B
II
1
2
inv.
{
{
S
A
Di es i rae, (irae) Di es illa
R inv. 1 (5 & 6)
illa
B
T
S
A
Di es i rae (ir rae) di
R
es
inv. 1 (4 & 5)
illa
(ill a[sic])
1
2
3
4
5
6
1
2
3
4
5
6
[016] [015]
B
C
F
A
D
B
5 5
1
1
Row II
hexachord I
Row II
hexachord RI
st
nd
1
1
st
st
1 & 2
1 & 2 3 & 2 [T ]
1 & 3
2 & 3
11
(b)
(c) (d)
rotational arrays (Ex. 8c).
38
In general, the dyads derive from segments of two
consecutive notes within a line of the tables. In one case, they even derive from
two non-consecutive notes (1st, 1 & 3= the rst and third note of the rst line).
In another case, the dyad derived from the third and second note of the second
line (2nd, 3 & 2 = G
) is transposed down a semitone (T

11
), so as to become
GE
. As can be seen, Stravinsky does not seem to have selected the dyads on the
basis of a pre-established criterion or precise order within the table. Rather, it
seems that his only intention was to ensure the production of numerous ic1 and
ic5 relations. These intervals attain a certain importance within the original form
of hexachord a, where they form two motives of class 15, in the forms [016] and
[015], respectively (also shown in Ex. 8d). Moreover, given the structure of the
tables, these intervallic motives also appear in the rotated(-transposed) forms.
This justies Stravinskys recourse to the rotational arrays, but it does not
explain his reason for extrapolating only dyadic segments, rather than complete
hexachordal units. Nevertheless, it is evident that, by operating in this manner,
Stravinsky hoped to obtain a denser and more cohesive motivic construction
than could be achieved using the hexachords in their entirety. Note, for example,
that the two 15 motives interlaced to form the symmetrical set [0156] in
the second choral statement derive from neither hexachord a nor from its
rotated(-transposed) form. This demonstrates that from Stravinskys point of
view serial technique is not essential per se, but instead functions only as a means
to an end with regard to motivic-intervallic syntax. The use of complete serial
forms does not, as a matter of fact, represent a restriction: if necessary, their use
can pass into the background in favour of a more immediate and direct engage-
ment with single intervals.
From Intervallic Motives to Rows
The problem of the interaction between intervallic-motivic logic and serial
technique becomes central in the compositions following Agon, which are sys-
tematically based on the use of ordered pc sets (tetrachords, hexachords, twelve-
note rows, and so on). This interaction can be observed in two distinct
conceptual stages of the creative process: (1) the initial denition of a row of
pitches and (2) the transformation of the abstract row into concrete musical
contexts. In either of these stages, motivic-intervallic logic can take on a role of
greater or lesser importance. In the rst stage, such logic can determine the
physiognomy of the pc sets to be used as primary series. In the second stage, it
can determine the manner in which the rows are manifested musically. The
implicit motivic-intervallic relations in the abstract formulation of the row, in
fact, can be revealed through a variety of devices. These devices can occasionally
elucidate different motivic-intervallic aspects latent in the row, which is then
subjected to a process of continual motivic interpretation.
In each composition, motivic-intervallic logic may be present in one rather
than the other of these two conceptual stages of the creative process. In Agon, for
example, this logic seems to have characterised the pre-compositional stage (that
is, the initial denition of a tone row). Indeed, the majority of the rows employed
in the work demonstrate a very clear and denite motive-intervallic design.
Elsewhere, I have tried to show how the overwhelming majority of rows (ranging
from four to twelve pitches) which in recent studies of the compositional process
have been identied throughout the ballet,
39
from the Triple Pas-de-Quatre
forwards, can be traced back to motivic-intervallic combinations of ic1 and ic2.
40
Consider, for example, the ordered tetrachord <0, 1, 4, 3>, which appears for the
rst time in the Pas-de-Deux: it can be generated by a 121 motive, with the
intervals oriented in the same direction to form set class [0134], by simply
ordering the sounds according to the succession <0, 1, 4, 3> in which the whole
tone is found between the second and fourth notes and the two semitones on
either side. The central position is then occupied by a (non-structural) major
third (Ex. 9).
The majority of the rows (from six to thirteen pitches) employed in the
succeeding movements of the ballet are obtained by combining different state-
ments of this characteristic tetrachord.
41
Consequently, the 121 motive
becomes the generating nucleus for the remainder of the work. Other rows not
based on the <0, 1, 4, 3> tetrachord can still be traced back to a particular
combination of ic1 and ic2. The rst ve notes of the hexachord stated in canon
at the beginning of the Bransle Simple (Ex. 10a), for instance, are formed by a
succession of alternating tones and semitones, interpretable as two 12 motives
united by a common pitch class. In this case, the particular design of the intervals
the rst motive is in the form [013], with the intervals oriented in the same
direction; the second is in the form [012], with the intervals in opposite direc-
tions guarantees that between the highest pitch, D, and the lowest, G, an ic5 is
formed, the same interval as that produced by the concluding B (the only note
which lies outside the 12 pattern) and the preceding F
. This results in a
symmetrical structure, with two ic5s (DG and BF
) at the distance of a
semitone. The twelve-note row employed in the coda of the rst Pas-de-Trois
(presented for the rst time in bars 185189) is entirely formed from a chain of
12 motives in the two forms [012] and [013] (Ex. 10b).
In Agon, the intervallic motives containing the tone and semitone, aside from
generating the majority of the fundamental rows, perform an important role even
in the movements not based on serial technique.
42
Thus their presence imposes
coherence on the work in its entirety, despite the different compositional tech-
niques employed.
43
The passage for the rst violin (doubled by the cello) at bars
Ex. 9 Motive 121 in [0134] form, reordered as tetrachord <0, 1, 4, 3>
121 motive in <0134> form the same reordered as <0143>
1
2
1
97102 of the Triple Pas-de-Quatre, for example, derives from a dense chain of
overlapping 212 motives (Ex. 11). The intervallic orientation of the motives is
almost always a zigzag, forming the chromatic set class [0123], but at times (see
the circled motives in Ex. 11) the intervals are oriented in the same direction,
thus forming the set class [0235] (see again Table 2).
44
By comparison with the rows used in Agon, the motivic structure of the
twelve-note rows employed in the compositions which succeed it chronologically
appear to be less well dened. Beginning with Movements, Stravinsky seems to
have derived many of his twelve-note rows from a reading of a concrete musical
idea most often a brief polyphonic passage. This procedure guarantees that the
intervallic motives contained in the initial musical idea are less evident in the
related twelve-note row, in which the structural intervals can be found between
non-adjacent pitches. This creates a sort of circularity between the two stages
into which I have conceptually subdivided the creative process: from a concrete
musical idea comes an abstract row of pitches, and on the basis of this row, new
Ex. 10a Stravinsky, Agon, Bransle Simple (opening): motivic construction
2 1
2 1
5
[012]
[013]
5
Ex. 10b Stravinsky, Agon, twelve-note row of the coda of the rst Pas-de-Trois:
motivic construction
2 2
2
2
2 1 1 1 1 1
5
[012]
[012]
[013]
[013] [013]
Ex. 11 Stravinsky, Agon, Triple Pas-de-Quatre: motivic construction of bars 97102
= 212 motive
[0123] [0123] [0123] [0123]
[0123] [0123] [0123] [0123]
[0123] [0123] [0123]
[0123] [0123]
(C. ni)
[0123] [0123]
[0123] [0123] [0123] [0123] [0123] [0123] [0123]
[0123] [0123] [0123] [0123] [0123] [0123] [0235] [0235]
and different concrete musical passages are realised. In effect, Stravinsky appears
to open up a wider eld of compositional possibilities: if indeed the motivic-
intervallic structure of the resultant row is more ambiguous, the row more easily
presents various musical realisations capable of illuminating different motivic-
intervallic aspects that are implicit within it.
In his last works, Stravinsky hinted at different cases in which the formulation
of a twelve-note row could derive from an initial concrete musical idea.
45
Con-
sider for example the following declaration concerning the composition of
Epitaphium:
I began the Epitaphium with ute-clarinet duet (which I had originally thought of
as a duet for two utes, and which can be played by two utes ... ). In the manner
I have described in our previous conversations, I heard and composed a
melodic-harmonic phrase. I certainly did not (and never do) begin with a purely
serial idea, and, in fact, when I began I did not know, or care, whether all twelve
notes would be used. After I had written about half the rst phrase I saw its serial
pattern, however, and ... began to work toward that pattern. The constructive
problem that rst attracted me in the two-part counterpoint of the rst phrase was
the harmonic one of minor seconds. The ute-clarinet responses are mostly
seconds, and so are the harp responses, though the harp part is sometimes
complicated by the addition of third, fourth and fth harmonic voices. (Stravinsky
and Craft 1960, pp. 99100)
Assertions of this sort often nd conrmation in the sketches of the compo-
sitions from Movements onwards, where one encounters some melodic or con-
trapuntal annotations which probably served as the model for the formulation of
the rows.
46
A circumstance of this kind probably explains the origin of the two
twelve-note rows employed in the Requiem Canticles, which Stravinsky explicitly
attributed to some intervallic designs which I expanded into contrapuntal
forms.
47
The intervallic designs to which this quotation alludes can be found
among the sketches for the instrumental Interlude which in actuality was the
rst movement in chronological order of composition.
48
The amount of preparatory material which survives for the Interlude is
uncommonly large, considering the Stravinskian standard: more than twenty
small sheets and strips of paper of different sizes, forms and typologies for a
passage lasting only 67 bars.
49
Some sketches contain only one or two brief
musical phrases, mostly corresponding to the exposition of a single form of row
I or row II (or of one of their constituent hexachords). Other pages assemble
their content from various earlier sketches. The nal form of the passage thus
results from a sort of montage of single ideas which had been elaborated
individually in the rst stages of composition. Comparison of these versions,
together with analysis of the various written materials employed and the auto-
graph dates placed by Stravinsky on some pages, allows for chronological recon-
struction of the sketches for the Interlude with a good degree of certainty.
50
One
of the very rst ideas notated by Stravinsky is reproduced in Ex. 12a. It is formed
from the union of two brief contrapuntal phrases based on the two original rows
employed in the movement, as indicated by the autograph serial symbols.
51
The
two phrases, initially notated separately on two small clippings of paper, were
then pasted onto a piece of cardboard (the continuity between the two phrases
is indicated by Stravinskys autograph arrow).
52
These phrases correspond
respectively to bars 161162 and 173175 of the score, of which the clippings
preserve a very rudimentary version. In the following sketches, Stravinsky added
new musical material between the two phrases, which are at the same time then
gradually reshaped. To this extended musical passage thus obtained (bars 163
172) Stravinsky subsequently added bars 176192, thus creating the entire
episode for four utes (bars 161192), the largest and most important formal
section of the piece. In summary, it seems that the two musical ideas contained
in the sketches transcribed in Ex. 12a were indeed the point of departure in the
composition of the Interlude. If this is so, they may well feature the original
intervallic designs to which Stravinsky alludes in the statement quoted above. In
fact, the contrapuntal relations of the two musical phrases illuminate a very clear
Ex. 12a Stravinsky, sketches for Requiem Canticles, Interlude (Paul Sacher Founda-
tion, Igor Stravinsky Collection)
II
[ ]
0
3
I
0
Ex. 12b Motivic analysis of the musical ideas contained in the sketches of Ex. 12a
3
F
C
B
D
A
A
D
F
G
C
G
E
F
G
E
F
F
D
E
C
B
C
D
C
B
A
G
A
[016] [015] [013] [013] [013] [012] [013] [012]
1
2 2
1
1
2 2
1
1
1
1
2
1
2
1
5
5
5
1
5
5
1
motivic-intervallic construction, based on the 12 and 15 motives (Ex. 12b).
The triplet in the rst crotchet of the second phrase, shown on the right in
Ex. 12b probably composed rst
53
presents within itself a sort of polyphony:
the lower part, delineated by the pitches FGE placed in the same register,
produces a 12 motive; the D
of the upper part forms, however, a relation of a

semitone with the lower E. By holding these two pitches rm and adding the F
,
another 12 motive is obtained in the second crotchet of the phrase, this time
vertically (D
EF
). Furthermore, the two motives (EFG and D
EF
) are
separated by the distance of a semitone. In the remaining part of the phrase,
three overlapping 12 motives, in both [012] and [013] forms, are unfolded
horizontally. In the rst phrase (reproduced on the left in Ex. 12b), the rst two
vertical simultaneities of three pitches form two motives of class 15 respec-
tively in the forms [016] and [015] while the following group of four pitches
delineates a cycle of three ic5s (C
), divided symmetrically into two

(F
in the bass; G
in the upper parts). The last vertical sonority of three

pitches (F
EG) forms a 12 motive which creates a strong link with the

following phrase, beginning with the motive FGE, another member of the 12
motivic class. The link illustrated also by Stravinskys cue in the upper right-
hand corner of the rst sheet of Ex. 12a is reinforced by the presence of pitch
classes E and G in both of the motives.
The intervallic designs contained in the two ideas thus become relatively
clear. We might ask at this point which came rst, these two musical ideas or the
two twelve-note rows whether, in other words, the rows were obtained from the
musical ideas or could instead have been xed in advance as an abstract
sequence of pitches on the basis of which the musical ideas were subsequently
elaborated. The fact that the two musical ideas contain all twelve notes without
repetition does not mean we must prefer the second solution: generally speaking,
in fact, we may suppose that Stravinsky initially elaborated his musical ideas
following a predominantly motivic-intervallic logic, and although even at this
stage of which, however, hardly any written traces remain he tended to exploit
all twelve notes of the chromatic gamut, that did not prevent him from using
some pitches more than once. Only in the nal formulation of the idea were the
repetitions eliminated until a fundamental twelve-note row was obtained. This is
clearly demonstrated by an important document to which Joseph Straus has
drawn attention:
54
the photographs taken in 1967 by Arnold Newman in Stravin-
skys Hollywood studio.
55
Like the stills of a lm, Newmans photographs record
step by step the creation of a musical idea a brief instrumental passage and
its successive transformation into a twelve-note row.
56
According to Straus, the
musical passage reveals, above all else, some semitone-tone motives belonging to
set class [013].
57
From my point of view, conversely, the rethinking which took place in the
course of elaboration was determined by a motivic idea of the semitonemajor
third type (14) in its two possible forms, [014] and [015]. In the very rst bar,
Stravinsky notated a portion of an Allintervallreihe (all-interval row; Ex. 13a) as
Straus also observes. In the next stage of composition (Ex. 13b), he abandoned
this initial idea to compose a brief counterpoint between two voices, without
attempting to avoid the repetition of pitches (note the initial pitches B and C,
repeated at the close of the passage). The third stage reveals the rst signicant
stage of rethinking: comparing Ex. 13b with Ex. 13c, observe that Stravinsky
replaces the rst of the two Cs with an E
, thus avoiding the presence of

repetition. The choice of E
throws light on the motivic-intervallic logic which

guides the composition of the passage: the rst three notes of the viola (BE
)
now form a motive of the semitonemajor third type (14). This motive in its
two forms, [014] and [015] appears at numerous other points within the
passage: in the rst ve notes of the viola part (twice: BE
and B
A, with
B
in common); in the last three notes of the cello (B
F), grouped together

as a triplet; between the rst two notes of the viola (BE
) and the D of the cello

which follows immediately afterwards (a semitone lower); and nally in the
contrapuntal relationship between the B
of the viola and the D of the cello

(Ex. 13f).
There is a second signicant redrafting at the following stage (Ex. 13d), where
Stravinsky replaces the rst and the third notes of the viola (B and B
), both
repeated, with G and A
respectively. By making this adjustment, Stravinsky not

only obtains all twelve notes without repetition, but also preserves intact the
Ex. 13ag See Arnold Newmans photos, reproduced in Craft (1967), pp. 1415 and
1617. The sequence of photos goes across the volumes two-page spreads of the
sketches; photos 1 and 2 are on p. 14, photos 3 and 4 on p. 15, and so on
(a) stage 1
(cf. photos 110)
(b) stage 2
(cf. photos 1113)
(c) stage 3
(cf. photos 1417)
(d) stage 4
(cf. photos 1820)
(e) analysis of stage 5
(cf. photos 1825)
(f) analysis of stage 3
(g) analysis of stage 4
3
3
[3]
3
[3]
inverted
[025] [027]
D
G
E
A
D
E
5
5
2
2
3
[014] [015] [014] [014]
B B B A
D D E
E
B D D B
4
1
1
4
4 4
1
1
[3]
[015]
F
G
B
4
1
3
[015] [015] [015]
D
E
G A
G
E A
D
D
1 1
4
1
4 4
[3]
[015]
F
G
B
4
1
motivic construction of the passage, which remains based on the conspicuous
presence of a 14 motive now in the form [015] (Ex. 13g).
The third and nal phase of development occurs in the nal stage (Ex. 13e),
at the end of which Stravinsky obtains a twelve-note row, arranging in succession
the notes of the passage just composed.
58
Here a small adjustment in the order
of the pitches sufces to mask the original motivic-intervallic aspect of the
passage while at the same time illuminating a new one. The second and sixth
pitches (E
and E
) are inverted (see the circled notes in Ex. 13e). Thus, the rst
two segments of three pitches (GED and A
) become two motives of the

class 25. Moreover, the four segments of three pitches which form the row
delineate a symmetrical structure: the combination of the even-numbered seg-
ments forms a partial circle of fths from G
to F, while the combination of the

odd-numbered segments forms the remaining part of the circle (from C to B):
59
1. E
(segment 2) + B
F (segment 4) = G
F;
2. GDE (segment 1) + ABC (segment 3) = CGDAEB.
In the end, the simple exchange of E with E
in the nal row creates, in this case,

a marked estrangement of the motivic-intervallic construction of the original
musical idea.
60
From the Row to Intervallic Motives
At this point it is worth reecting on the ways in which motivic-intervallic logic
inuences the musical concretisation of the row, once it has been denitively
established. I will rst consider a brief musical fragment drawn fromthe beginning
(bars 4648) of the second of the ve Movements (Ex. 14a and b).
61
The passage
is based on the two discrete hexachords (labelled a and b) of the fundamental row
Ex. 14a Stravinsky, Movements, ii: hexachordal forms employed in the sketch shown
in Ex. 14b
1
1
1
1
2
2
2
2 3
3
3
3 4
4
4
4 5
5
5
5 6
6
6
6
[016]
[016]
[012]
[012] [012]
[012]
[016]
[016]
Hexachord
Hexachord
RI (T )
R
6
used throughout the entire composition (Ex. 14a, rst and third lines). In the
upper part of the sketch transcribed in Ex. 14b, Stravinsky wrote the RI (T
6
) form
of hexachord a and the R form of hexachord b (Ex. 14a, second and fourth lines)
as indicated also by the autograph labels Riv-Inv a and Riv b.
62
As my analysis in Ex. 14a shows, hexachord b can be divided into two
trichords belonging to set class [012], which can in turn be related to motive
class 12 (see again Table 1). Hexachord a is formed by two trichords of set class
[016], which can be related to three different intervallic motives: 16, 15 or 56.
In this case, Stravinsky clearly placed the intervals of a semitone and perfect fth
(motive 15) in relief. To this end, a particular permutation of the order of the
pitches of hexachord a is carried out:
63
besides placing the pitches in reverse
order (from the sixth to the rst), he also reversed the order of the rst two
pitches of each trichord. In this manner the pitches which form ic5 (represented
in bold in the schema below) are always adjacent:
654/321 becomes 564/312
(GF
C/DD
becomes F
GC/DA
).
The purpose behind this particular reordering can be appreciated in the
musical passage outlined in the sketch transcribed in Ex. 14b, immediately
below the two hexachords: the pitches which form ic5 are arranged vertically as
a perfect fth; the pitches which form ic1 precede the fth in a register at the
distance of an octave. The two fths (CG/D
) are separated by a semitone,

thereby producing a symmetrical conguration. All these choices are clearly
intended to throw the ic1 and ic5 into relief.
Ex. 14b Stravinsky, sketches for Movements, ii (cf. bars 4648 of the printed score)
(Paul Sacher Foundation, Igor Stravinsky Collection)
5 6 4 3 1 2
6 5 4 3 2 1
6
4
3
1
2
5
2 3 4
6
5
1
Riv.
Riv
Riv- Inv
Riv-Inv
This brief example shows how some operations which alter the physiognomy
of the row were intended by Stravinsky to facilitate the transformation of an
abstract row in a specic musical passage which highlights some particular
motivic-intervallic characteristics. One such operation is order permutation, as
we have just seen; another consists of extrapolating small segments, usually of
two to four notes in length, and successively reconguring them. This technique,
which I call serial fragmentation-recombination, was probably adopted for the
rst time by Stravinsky in Threni and was subsequently used in an increasingly
sophisticated fashion.
64
We have already observed one application, albeit a rather
limited one, in the Dies irae of the Requiem Canticles. To further illustrate its
function in relation to the motivic-intervallic syntax, I will now consider its use
in the rst part of the Rex tremendae of the Requiem Canticles.
The serial construction of the passage is clearly illustrated on the rst page of
the autograph short score (containing bars 203208), transcribed in Ex. 15a.
65
The symbol I Ra stands for row number I, retrograde form, hexachord a. The
Ex. 15a Stravinsky, rst page of the short score for Requiem Canticles, Rex tre-
mendae (compare bars 203207 of the printed score) (Paul Sacher Foundation, Igor
Stravinsky Collection)
B
6
T
8
Rex
4
vla
vc
cb
A
Rex
1
3 Fl
4 5 6
S
I R
Rex
3
Rex tre
men
Verticals
dae ma je sta tis
1 2 3
I
4
inv
5
3
6
rd
3 Fl
[ ]
th
R 5
[ ]
I inv 1 line
st
vc
I inv 4 line
th
3 line
rd
I
3 4 5
trmb
6 line
th
5
th
5 4 6
4 5 6
R 1
st
Strings
vlc
3 line
rd
circled numbers at the beginning of each choral part in the short score indicate
the lines of the rotational arrays of the rst hexachord. Only limited portions
(three to ve notes) of each line are used (shown circled in Ex. 15b). The zigzag
line traced across the choral parts
66
corresponds to the brass part (trumpet and
trombone) elaborated on the lower portion of the page. The autograph symbols
indicate that even this line was obtained by the combination of three serial
segments drawn from the two rotational arrays of the R hexachords. For
example, the symbol Ra 5th 3 4 5 stands for retrograde, hexachord a, fth line
of the rotational array, notes 345 (see also Ex. 15b, where the segments used
in the brass part are indicated within boxes). In sum: the rst three bars
67
of the
choral part and the brass parts are obtained exclusively through a combination
of serial segments extrapolated by two rotational arrays produced by hexachords
a and b of the retrograde form of row I. As in the case of the Dies irae, the
segments are selected in an apparently arbitrary manner. Nevertheless, they
demonstrate a signicant presence of motives of class 12: four of the ve
segments of three-note segments (Ra 1st 46, Ra 4th 46, Ra 5th 35, Ra 6th
46 and Rb 5th 46) directly correspond to this motive class; one of the two
segments of ve notes (Ra 3rd 26) contains two overlapped 12 motives
(C
C + CBA); and the other (Ra 1st 15) begins with a 12 motive. The
reason for this arrangement was Stravinskys desire to create an imitative texture
based on the motivic-intervallic element: at regular intervals of a minim the
contralto, the trombone, the sopranos and the tenors display motive 12; but
because this motive can assume two different forms [012] and [013] and
given that the three pitches can be combined in any order, the imitative responses
repeat neither the same melodic prole (as in traditional imitative style) nor the
Ex. 15b Stravinsky, Requiem Canticles, twelve-note row I: rotational arrays of the
hexachords a and b of the retrograde form, with encircled serial segments employed
in the sketch of Ex. 15a
1
2
3
4
5
6
Row I Retrograde

same set class. Therefore, motive 12 is repeated three times in the brass part
which runs throughout the choral passage. Note also that the initial pitches of
each imitative part (A
) gradually take the form of a sequence of

perfect fths (F
), which is completed in bar 3 with the addition of

the pitches G
and C
in the bass part.

This homogenous motivic design is due largely to the structure of the row
itself. Indeed, it should be evident that numerous 12 motives are already
contained in row I of the Requiem Canticles (see again the right-hand section of
Ex. 12b). Nevertheless, by segmenting the hexachords selectively, Stravinsky
created an imitative texture which was more coherent than would have been
possible had he used complete hexachords. One might ask in what sense and to
what degree such a procedure could be dened as authentically serial. However,
the fact that the composer had indicated the serial origin of the various segments
in the short score demonstrates that he conceived of the row as a reservoir of
motivic-intervallic material, and that he understood serial technique as a means
of managing this material systematically.
The beginning section of the rst of the ve Movements (see again Ex. 6)
is likewise based on a systematic application of serial fragmentation-
recombination. The passage divides into an initial introduction (bars 16) and
three solo episodes for, respectively, piano, rst ute (accompanied by piano and
clarinet) and piano again (accompanied by strings). The central ute solo is very
familiar to Stravinsky scholars; indeed, the composer himself drew attention to
its complex serial construction, thereby instigating a long series of attempts at
analysis.
68
As I will try to show, a motivic-intervallic approach can provide a new
and logical key to reading the passage.
To this end, it is useful to begin with the subset structure of the two hexa-
chords of the fundamental row (Ex. 16). The rst hexachord is formed by two
disjunct trichords belonging to set class [016]. In the middle, starting with the
third note, one nds a trichord of set class [012].The second hexachord contains
two disjunct trichords of set class [012] and one [016] trichord at its centre, thus
complementing the rst hexachord. Other interesting properties of the two
hexachords emerge when one takes into consideration their rotated forms.
Ex. 17 cites one of the rotational arrays employed by Stravinsky for the compo-
sition of Movements: columns a and b contain the rotated forms of the two
hexachords of the original form of the row; columns g and d display the rotated-
transposed forms. Ex. 18 reveals the subset structure of all these rotated
(-transposed) forms. Eight of the twelve hexachords contain within them three
Ex. 16 Stravinsky, Movements, i: subset structure of the original twelve-note row
Orig.
[016] [016] [012] [012]
[016] [012]

Ex. 17 Stravinsky, Movements, sketch showing the rotated (columns a and b) and
rotated-transposed (columns g and d) forms of the two hexachords of the original
twelve-note row (Paul Sacher Foundation, Igor Stravinsky Collection)
I
II
III
IV
V

Ex. 18 Stravinsky, Movements: subset structure of the rotated-transposed forms of
the two hexachords of the fundamental row
Orig.
I
II
III
IV
V
[016] [016] [012] [012]
[016] [012]
[016] [012]
[012] [016]
[016]
[012]
[012]
[016] [016]
[016] [016]
[012]
[012]
[016]
[012]
[016]
[012] [016]
[012]
[012] [012] [016] [016]
[016]
[012]
[012]
[016]
[016]
[012]
[016]
[016]

trichords belonging to set class [016] or [012]; two hexachords (IIg and IIIg)
contain four which closely overlap with one another. The remaining two hexa-
chords (Id and IVd) contain two each.
It is evident, therefore, that the internal structure of the two hexachords is
such that the rotated forms generate a large number of trichords belonging to set
classes [012] and [016]. Now, set [012] can be understood solely as a 12 motive
with the intervals arranged in opposite directions,
69
while set [016] could be
associated with three different motives: 15, 16 or 56 (see again Table 1). A
deeper analysis will clarify which of these intervallic motives was the object of
Stravinskys interest.
With the aid of the sketches, I have reconstructed the chronology and serial
origin of the three episodes which form the entire section, summarised in
Table 3.
70
As can be seen, all three episodes are based on a combination of serial
segments chosen from columns g and d of the rotational array. The serial
segments employed for the second episode, the ute solo (bars 1317), are
indicated in the sketch transcribed in Ex. 19a; the symbols here refer to the
sketch with the rotational array transcribed in Ex. 17 (the Greek letters refer to
columns g and d of the array and the roman numerals to the lines; the arabic
numeral indicate the selected segments). The 34 notes of the ute melody
(Ex. 19a) are obtained by combining ten serial segments, freely chosen from the
array (Ex. 19b). In the nal version (Ex. 19c), the melody thus obtained was
completely transposed a major third higher (T
4
) to start on G instead of E
. In the
third episode (bars 1822; see Ex. 20) the piano and string parts employ the
same serial segments as the ute solo, as is suggested by the serial symbols and
the indication follow the ute solo before (same series) in the short score;
however, this time it is not transposed (compare Ex. 20 with Ex. 19b).
The rst episode (bars 712; Exs. 21 and 22) is essentially based upon the
same succession of serial segments, even if in the nal version the resulting
correspondence is obscured owing to some errors Stravinsky committed in the
Table 3 Chronology and serial origin of the three episodes in Stravinskys Movements,
i, bars 722
Episodes (in chronological order) Twelve-note material employed
bb. 1317: second episode
(solo for ute I)
Serial segments drawn from columns g and d of the
rotational array of the hexachords. The entire
succession is transposed T
4
to G. The accompaniment
uses complete hexachords.
bb. 712: rst episode
(rst solo episode for piano)
The same serial segments from the ute solo but
transposed T
2
(on F).
bb. 1822: third episode
(second solo episode for piano)
The same serial segments from the ute solo, not
transposed (on E
).
preparation phase. On the rst two lines on the page of sketches transcribed in
Ex. 21, up to the pitches BCD
before the notated clefs, the composer had

initially outlined a rst version of the passage transposed to begin on G, in a
manner identical to the nal version of the episode for ute. Ex. 22 presents this
rst version together with the related serial symbols derived from his autograph
short score. As is evident by comparing Ex. 22 with Ex. 19c, Stravinsky obtained
exactly the same pitches as the nal version of the ute episode by utilising only
slightly different serial fragments.
71
Up until this point, therefore, the rst piano
episode corresponds exactly with the ute solo, with respect to pitch content. At
a later point, represented in lines 69 of the sketch in Ex. 21, Stravinsky elabo-
rated a second version different from the rst in both rhythm and the octave
registration of some pitches transposed to begin on F instead of G. Later, after
Ex. 19a Stravinsky, sketches for Movements, i, bars 1317 (Paul Sacher Foundation,
Igor Stravinsky Collection)
1 2 3 4 5 6 1 1 2 2 2 3 3 3 3 3 4 4 4 4 [4] [4] 5 5 5 5 5 5 6 6 6 6 6 6
[ ] ] [
I I I II II III III III V V
Ex. 19b Columns g and d of the array reproduced in Ex. 17 (hexachordal rotated-
transposed forms), with encircled serial segments employed in the sketch shown in
Ex. 19a
Orig.
I
II
III
IV
V

1
1
1
1
1
1 2
2
2
2
2
2 3
3
3
3
3
3 4
4
4
4
4
4 5
5
5
5
5
5 6
6
6
6
6
6 1
1
1
1
1
1 2
2
2
2
2
2 3
3
3
3
3
3 4
4
4
4
4
4
5
5
5
5
5 6
6
6
6
6
6 5
Ex. 19c Final version of the ute solo in bars 1317 showing the serial segments
employed in it
Fl. I
I 13 (T )
4
I 46 (T )
4
I 46 (T ) I 36 (T )
4 4
V 13 (T )
4
V 13 (T )
4
III 36 (T )
4
II 46 (T )
4
III 46 (T )
4
II 26 (T )
4
the clefs (where the elaboration of the rst version was interrupted) on lines 12
of the same page, Stravinsky completed the episode with the nal missing part.
Although neither the sketch nor the short score presents serial symbols at this
point, it is obvious that the segments employed are the same as those used for the
last part of the ute episode (transposed by T
2
).
72
To summarise, the second version of the piano passage stretches in part across
lines 69, up to the dyad AB
circled in the middle of the ninth line which

corresponds to the pitches B and C of the preceding version on G, as demon-
strated by the vertical line drawn across the page and in part (following
Stravinskys arrow) on lines 12, to the right of the treble and bass clef.
73
In
setting the entire passage in the short score, however, Stravinsky failed to recopy
pitches B and C, placed on the rst line immediately after the treble clef, perhaps
mistaking them for a repetition of the B and C immediately preceding it on the
second line which, however, belonged to the rst version on G. Thus these two
notes do not appear in the nal score. By restoring them, one will easily notice
that the entire sequence of pitches in the pianos rst episode corresponds
exactly to that of the utes solo episode, transposed a tone lower.
74
According to Stravinskys initial intent, then, all three episodes were to be
based on the same succession of pitches, transposed onto three different levels:
G, F and E
. It was most likely in this manner that Stravinsky sought to obtain

something similar to tonal organisation on a broader formal plane.
75
However,
considerations of this type are beyond the scope of this study. What is important
from my point of view is to observe how the intricate combination of serial
segments on which the three episodes are based produces a homogeneous and
cohesive motivic-intervallic texture.
Ex. 20 Stravinsky, Movements, i, bars 1822
r.h.
2
3
,
3
l.h.
18 19 20 21
3 2
1
3
Piano
I 13 I 46 V 13 V 13 III 36 I 46 II 46 III 46 I 36 II 26
22
3 pizz.
{ {sim.
3
(non div.)
3
pizz.
Vle.
Vc.
C. B.
Ex. 21 Stravinsky, page of sketches for Movements, i, bars 712 (Paul Sacher Foun-
dation, Igor Stravinsky Collection)
[ ]
[ ]
[ ]
5
8
[ ]
] [
[ ]
[ ]
5
va
vb
8
va
[ ]
[
[ ]
]
[ ]
[ ]
[ ]
8
8
vb
[ ]
To this end, the question which arises is: according to what criterion did
Stravinsky select and combine these serial segments? Typically, if one takes into
consideration the original hexachords from which the segments were extrapo-
lated, it becomes apparent that a procedure of utilising pivotal pitches is ren-
dered central. For example, the rst segment of three pitches (gI 13) is followed
by the pitch A
from the middle of its original hexachord gI (see again Ex. 19b).
The second segment must therefore begin with this pitch. In its turn, the second
segment (dI 46), also containing three pitches, is followed,
76
in its original
hexachord, by the pitch C; thus the third segment must begin with that pitch.
The passage consequently unfolds as follows:
I E A G A 1 3 4 ( ) = ( )

I A G D C 4 6 1 ( ) = ( )

V C G G E 1 3 4 ( ) = ( )

Nevertheless, this criterion establishes only the rst pitch of the following
segment and not the internal characteristics of that segment (as the serial table
shows, there are different segments which also begin with the same pitch).
Moreover, the mechanism of pivotal pitches is employed only in a limited
number of cases. A fuller rationale nevertheless emerges if the ten serial segments
employed across the three episodes are re-examined in relation to the concept of
intervallic motive. Given the structure of the hexachords (see again Exs. 16 and
18), these segments are almost all members of set classes [012] and [016], as
Ex. 23 demonstrates (the only exceptions are the two segments indicated by the
exclamation marks). All of the four-note segments (nos. 5, 9 and 10) contain
overlapping sets [012] and [016], as Ex. 23 indicates.
What is intriguing, however, is the global result obtained from the combina-
tion of the segments. Ex. 24 examines all of the consecutive three-note groups
(bracketed above and below the musical stave) starting from each note of the
Ex. 22 Stravinsky, Movements, i (compare bars 710 of the printed score). The
symbols above the staves represent serial segments drawn from the short score (Paul
Sacher Foundation, Igor Stravinsky Collection). Below them is an excerpt (rst two
staves) from the sketch transcribed in Ex. 21
[ ]
V 36 (T ) I 56 (T ) V 13 (T )
5
III 46 (T ) I 46 (T ) II 46 (T )
2 4 4 4 4 4 4
V 14 (T )
complete succession of pitches:
77
with the sole exception of seven groups (indi-
cated by the exclamation marks), all of the trichords belong to set class [012],
[013] or [016]. These set classes, in turn, can be obtained from only two
motivic-intervallic classes; set classes [012] and [013] are, in fact, the two forms
assumed, according to the orientation of the two intervals, by the characteristic
tone-semitone motive (12). It is important to note that the [013] form of this
motive is not included in the serial hexachord, which contains only the [012]
form (see again Exs. 17 and 18). The presence of form [013] in the succession
of pitches on which the three episodes are based is therefore the result of a
collection of serial segments created ad hoc, which clearly demonstrates how the
technique of serial fragmentation-recombination permitted the composer to
generate different motivic forms without limiting himself to those contained in
the row. Set class [016] could have been generated by three different motivic-
intervallic classes: 15, 16 or 56 (see again Table 1). Nevertheless, the form
[015] of motive 15 obtained by orienting the two conjoined intervals in
opposing directions never appears in the entire succession of pitches, which
sufces to exclude it as a possibility. Of the remaining two motives (56 and
16), it is more logical and economical to think primarily in terms of motive 16
since the entire passage can then be read from the point of view of only three
interval classes ic1, ic2 and ic6 contained in motives 12 and 16.
It is evident, therefore, that the selection and arrangement of the serial
segments was done in such a way as to obtain a continuous interlocking of two
Ex. 23 Stravinsky, Movements, i: serial segments employed in the succession of
pitches contained in each of the three solo episodes of bars 722
[016] [016] [016] [016] [016]
[016] [016]
[012]
[012]
[012] [012]
! !
I 13 I 46 V 13 V 13 III 36 I 46 II 46 III 46 I 36 II 26
1 2 3 4 5 6 7 8 9 10
Ex. 24 Stravinsky, Movements, i: intervallic motives in the succession of pitches
contained in each of the three solo episodes of bars 722
[012] [016] [016] [016] [016] [016] [016] [012] [012]
[012] [016] [013] [013] [016] [016] [016] [012] [012]
[016] [016] [016] [016] [012] [012] [013]
12 motive 16 motive
[013] [012] [016] [016]
1 1 1 1
2 6
(3)
[7] = 5
2
6
(5) (1)
! ! ! !
! !
!
motivic classes (12 and 16),
78
generated by just three interval classes (ic1, ic2
and ic6). The use of complete serial forms (twelve-note rows or hexachords)
would not produce so dense and coherent a fabric: hence the exigency of
fragmenting the hexachords into smaller units, which can then be recombined at
will.
Conclusions
In the evolution of compositional technique which runs from the melodic chains
of Agon to the complex combinations of serial segments in Movements or the
Requiem Canticles, it is possible to discern a continuity of thought centred on the
problem of interaction between motivic-intervallic syntax and twelve-note tech-
nique. In Agon, the motivic-intervallic logic determines, above all, the construc-
tion of ordered sets, the combination of which gradually forms larger
agglomerations, employed in their turn as fundamental rows. In later composi-
tions, the row confronts a continuous motivic rereading, whether through dif-
ferent means of musical realisation or through fragmentation into small groups
of pitches, which are then recombined into new congurations.
Examination of the creative process reveals that the tendency towards dis-
integration resulting from the direct manipulation of single intervals is not
incompatible with serial technique, providing the latter is understood in a cre-
ative rather than a strictly procedural sense. Stravinskys adoption of serialism
relies upon an aesthetic vision which does not attribute to the row the value
of a fundamental Gestalt for the composition. The same can be said of
Stravinskian motivic-intervallic syntax, at the base of which lies an aesthetic
conception foreign to the ideal of organic coherence which characterises the
Austro-German tradition. The idea that intervallic congurations should
provide a unifying function for the general internal relations which govern an
entire composition a function comparable to that of traditional tonality is
largely alien to Stravinskys approach, which conceives of the intervallic
motives simply as starting materials for the act of musical construction. This
creative process consequently proceeds from the particular to the general,
following an itinerary open to deviations and metamorphoses which are
realised through continuous motivic-intervallic rereading of both the pre-
compositional row and its concrete musical correlates. The elementary mate-
rials (the intervallic motives) from which this mode of construction primarily
derives evidently leave a mark, a recognisable impression on the nal physi-
ognomy of the musical edice, without, however, assuming a determining role
with regard to structural connection. To what degree and at what level
this impression is discernible and perceptible is a question which evidently
remains open. Whatever the answer, we must not lose sight of the essential
signicance of working with intervals that characterised Stravinskys musical
thought.
NOTES
This article derives, in large part, from my doctoral thesis (Locanto 2002). All
transcriptions and facsimiles are published with the kind permission of the Paul
Sacher Foundation in Basel. The sketch transcriptions aim to reproduce the
originals as faithfully as possible: all of the authors annotations are either enclosed
within square brackets or signalled in the accompanying captions. Since the
numbering of the microlms in the Paul Sacher Foundation may change from
time to time, items are identied here not by microlm number, but by the
collection to which they belong and the type of material (sketch, short score, full
score, etc.). In the case of the sketches for the RequiemCanticles, for which a micro-
lm copy has not yet been made, I refer only to the collection to which it belongs.
Copyright clearance for musical examples was secured fromthe following sources:
Stravinsky, Requiem Canticles copyright 1967 by Boosey & Hawkes Music
Publishers Ltd., reproduced by permission of Boosey & Hawkes Music Pub-
lishers Ltd.; Stravinsky, Movements copyright 1960 by Hawkes &Son (London)
Ltd., reproduced by permission of Boosey & Hawkes Music Publishers Ltd.
1. See for example Stravinsky and Craft (1959), p. 11: I begin work by relating
intervals rhythmically. (See also Stravinsky and Craft 1966, pp. 601, on the
Variations, and Craft 1972, p. 98, on the Requiem Canticles.) In an interview with Jay
S. Harrison in the NewYork Herald Tribune, 21 December 1952 (cited in Tucker
1992, vol. 2, p. 187), Stravinsky emphasised: Always I have been interested in
intervals. Not only horizontally in terms of melody, but also the vertical results that
arise from the combinations of intervals. This point has been strongly endorsed by
Milton Babbitt: One of the remarkable things that Stravinsky said, when people felt
that he committed a treasonable act by starting to write pieces where you could nd
a succession of twelve [notes] at the beginning, was Theres nothing to it; Ive
always composed with intervals. Basically, of course, it was something of a witti-
cism, but what it did show, much more than a witticism, was how profoundly this
is an interval kind of syntax and not just a pitch-class syntax fundamentally and
centrally an interval syntax (Babbitt 1987b, p. 20; see also Babbitt 1968, p. 167).
2. On these topics, see especially Berger (1963), van den Toorn (1983) and Taruskin
(1996), pp. 255307. On the use of other non-diatonic collections in Stravinskys
music, see also Johnson (1987), Tymoczko (2002) and van den Toorn and
Tymoczko (2003).
3. After Threni, the systematic use of diatonic collections, which characterised even
Stravinskys earlier serial compositions, is noticeably reduced in favour of more
markedly chromatic situations, even if the latter contain a diatonic core (see
Locanto 2002, pp. 177212). On the diatonic component in serial compositions up
to Threni, see Neidhfer (1999). Taruskin (1993) and (1996), pp. 164873, has
hypothesised the persistence, up to the nal serial compositions, of a similar routine
based on the use of octatonic collections. More recently this idea has been placed
in a markedly different perspective, especially by Straus (2001), p. 39 and n. 79.
4. On the use of intervallic motives in Stravinskys pre-serial music, see for example
Straus (1991).
5. The underlying theoretical perspective is neatly summarised by Wason (1996), p.
111: In tonal music, motivic events are generally regarded as intervallic relation-
ships, whose actual pitch-class representations change with reference to a xed-
pitch background the tonal centre itself, or a related tonal region which
temporarily comes to the fore. In so-called atonal music, that background is
presumably absent, leaving us with only the intervallic relationships the immediate
object of most analytical investigations of this repertoire.
6. For example, Straus, in attributing a basic role to the manipulation of single intervals,
denes the concept of motive in a way which differs from the denition offered here.
For him, a motive results from the combination of a limited number (usually two) of
atomic intervals (one of which is generally a tone or semitone).Through the choice
of an appropriate transpositional level, these intervals produce particular sets (see
Straus 2001, pp. 8292), which are then used as motives in melodic construction
(Straus 2001, pp. 92103). In this case, a motive, although initially obtained by a
particular combination of elementary intervals, corresponds to a set of pitch classes
which could be represented in retrograde, inverted or reordered, but would continue
nonetheless to belong to the same class. From my point of view, on the other hand,
a motive is created by a variable conguration of intervals which can produce sets
belonging to different classes. Furthermore, Straus conceives of motives in an
exclusively melodic sense: his analyses demonstrate the use of motives only in the
horizontal dimension. To my mind, however, the pitches produced by a particular
arrangement of basic intervals can be situated either horizontally in any order or
vertically. As an example of the different results to which these two approaches can
lead, see n. 40.Yet another approach is adopted by Smyth (1997), pp. 213, which
considers interval types (not interval classes). See also Smyth (1999) and (2000).
7. As other documents also demonstrate: see for example Stravinsky and Craft (1959),
pp. 1112: This exploration of possibilities is always conducted at the piano. Only
after I have established my melodic or harmonic relationships do I pass to compo-
sition. Composition is a later expansion and organisation of material ... . I start to
look for this material, sometimes playing old masters (to put myself in motion),
sometimes starting directly to improvise rhythmic units on a provisional row of
notes (which can become a nal row). Other sentiments of this type can be found
in Stravinsky and Craft (1962), p. 52, where the keyboard is described by Stravinsky
as the center of my life and the fulcrum of all my musical discoveries; in Stravinsky
and Craft (1966), pp. 234 and n. 8; and in Craft (1972), p. 131. See also the
testimony of Nicolas Nabokov (1949), p. 146, along with that of Stravinsky himself
in the documentary A Conversation with Igor Stravinsky, directed and produced by
Robert D. Graff for the National Broadcasting Company in 1957 (the passage in
question is transcribed with commentary in Tucker 1992, vol. 1, p. 23).
8. In this study, the prime forms (see Straus 1990, pp. 412) of unordered sets are
represented by a sequence of numbers each of which stands for a pitch class
between square brackets, arranged according to the conventional criterion dis-
cussed in Straus (1990), pp. 412. Ordered sets are, however, represented by a
numerical sequence in angle brackets which follows the actual order of the notes.
For example: the succession of notes B
ACB belongs to set class [0123]; con-

sidered as an ordered set, however, it would be represented as <1, 0, 3, 2>.
9. For a denition of interval class see Straus (1990), pp. 68. In this study, interval
classes are indicated in the orthodox manner by ic followed by an arabic numeral
indicating the interval class measured in semitones. The terms of traditional tonal
theory, when used, refer unless otherwise indicated to interval classes. For
reasons of space, in most of the musical examples the ic is omitted and the intervals
are indicated by arabic numerals alone.
10. See Straus (1990), pp. 12830.
11. This allows different row forms to project set class [0126] using the same global
pitch-class content, a property which Berg exploits to create a network of formal
relations.
12. This concerns, in essence, the phenomenon of pitch-class invariance, to which
Milton Babbitt drew attention in a fundamental collection of essays; see in particu-
lar Babbitt (1960) and (1961). On invariance see also, for example, Lewin (1962)
and Beach (1976).
13. On which there is an extensive body of literature; for an overview, see Perle (1991),
pp. 989.
14. The example is based on Weberns autograph row chart transcribed in Bailey
(1996), p. 196.
15. It is instructive to compare this denition of intervallic motive to the following
afrmation of Stravinskys: Always I have been interested in intervals. Not only
horizontally, in terms of melody, but also the vertical results which arise from the
combinations of intervals (see again n. 1).
16. Since I am considering unordered pitch-class intervals, it may seem senseless to
speak of their direction or orientation. However, the intervallic motives which are
the focus of my interest here always result from the union of two (or more)
conjoined intervals, for which these terms refer simply to the orientation assumed
by the intervals relative to each other. In fact, two conjoined intervals united in the
same direction produce a third interval corresponding to their sum (such as ic1 and
ic4, which together produce ic5 in Ex. 4) and, conversely, two conjoined intervals
united in opposite directions produce an interval corresponding to their difference
(such as ic1 and ic4, which together realise ic3 in Ex. 4). This is represented
graphically by the arrows in my examples, in which the pc sets are conventionally
arranged in their normal form from lowest to highest (for a denition of normal
form see Straus 1990, p. 27).
17. From here on, the intervallic motives will be represented by two or more arabic
numerals (corresponding to the interval classes) separated by a dash () and
ordered, only as a convention, from the smallest to the largest. For example: 12
indicates a motivic class which includes the following possible congurations:
<+1, +2>, <+1, -2>, <-1, +2>, <-1, -2>, <+2, +1>, <+2, -1>, <-2, +1>, <-2, -1>. In
what follows, I will often use, for the sake of simplicity, the termmotive in the more
precise sense of motive class.
18. See in particular Table 1.
19. See for example the denition given in Schoenberg (1967), p. 8. According to
Dahlhaus (1986), p. 283, the concept of motive which correlates with the Schoe-
nbergian idea of developing variation essentially concerns only the intervallic
aspect. Nevertheless, in Schoenbergs pedagogical writings, the motive is conceived
always as a complex of different interconnected aspects (intervals, rhythm, metrical
position, dynamic level, and so on). See for example Schoenberg (1995), pp.
16871.
20. In Stravinskys own words: I begin work by relating intervals rhythmically (see
n. 1).
21. See Sichardt (1990), pp. 3052.
22. See for example the motives identied by Sichardt (1990), pp. 502, in the com-
positions and fragments dating from the years immediately preceding 1919.
23. See Boss (1992), pp. 12550.
24. See Boss (1994), pp. 1946.
25. It is more difcult, especially in twelve-note music, to establish whether (and, if so,
to what extent) motivic elaboration, applied in the sense of the Schoenbergian
concept of developing variation, also confers a teleological orientation on the
musical discourse, as maintained, for example, in Haimo (1997).
26. See for example Perle (1991), pp. 938, in particular his analyses of Schoenbergs
Op. 23 No. 1, based on a minor thirdsemitone cell. A cell is dened by Perle (p.
9) as a group of pitches which may operate as a kind of microcosmic set of xed
intervallic content, statable either as a chord or as a melodic gure or as a combi-
nation of both.The difference between this and my denition of intervallic motive
is obvious.
27. The same thing can be said of the cells which, according to different authors (see
especially Perle 1955, Treitler 1959 and Antokoletz 1984, pp. 78137), play a
determining role in the music of Bla Bartk. Nonetheless, Bartks use of inter-
vallic cells presents some analogies with Stravinskys practice, especially in the
preference for symmetrical aggregates (see for example the cells labelled X, Y and
Z in Treitlers analysis). Different theoretical aspects concerning the use of inter-
vallic cells in the music of Bartk are addressed in Antokoletz (1984), p. 16 n. 27
and pp. 78137.
28. This is because the tritone subdivides the octave into two equal parts.
29. The global intervallic content of a set class is represented synthetically by the
interval vector dened in Forte (1972), p. 179.
30. In order to obtain all of the set classes produced by the union of three different
intervals, one can proceed as follows. For each of the six permutations of the three
intervals (123, 132, 213, 231, 312, 321) all of the possible arrange-
ments of their directions (ascending or descending) are formed (four: <+ + +>, <+ +
->, <+ - +>, <+ - ->). (Note: the four arrangements with initial descending intervals
produce sets of the same classes as the four preceding arrangements, of which they are
simply inversions.) The number of set classes thus obtained will not necessarily be
6 4 = 24, because some arrangements will generate the same set class.
31. This latent discrepancy is also recognised by Straus (2001), p. 92, who notes the
basic formal paradox of [Stravinskys] music, namely the centrifugal tendency of
the musical units [intervals] toward isolation and the centripetal tendency of the
transpositions and inversions to link them together into larger wholes. Neverthe-
less, in Strauss vision the discrepancy concerns the very immediate level of
structure, that is, the structuring of the motives and ordered sets used in the
composition. From my point of view, conversely, the necessity of nding a com-
promise between the two different constructive logics conditions the compositional
process across multiple levels.
32. The serial symbol II inv in the sketch transcribed in Ex. 8a indicates that the
passage is based on the inverse form of the second fundamental twelve-note row of
the Requiem Canticles.
33. The version featuring the move of a semitone from F
to G is found also in the lower

system of the sketch reproduced in Ex. 8b. In the upper system of this last sketch
one nds a third version, with the F
which leaps directly to B, thus anticipating the

B of the following chord.
34. The serial topography of this passage would be difcult to reconstruct without the
assistance of the sketches: see for example the serial analyses of Powles (1995), p.
288, which are theoretically plausible but in total contradiction of the documentary
evidence.
35. A clarication is needed with respect to the serial symbols O, R, I and RI employed
in the sketches. Normally, Stravinsky does not obtain the fourth basic ordering of
the row (RI) by reading the inverse form (I) backwards (as is typical of Schoenberg
and his students), but by inverting the retrograde form (R) (on the respective
consequences, see for example Krenek 1940, p. 11). The two forms differ with
respect to the level of transposition: the inversion of the retrograde begins on the
same pitch as the retrograde; the retrograde of the inversion begins on the last pitch
of the inversion. In order to avoid a disparity with the sketches which I have
transcribed, I will likewise employ the symbol RI to indicate the inversion of the
retrograde. The factor of transposition (T
n
) is computed in ascending semitones
from 0 to 11 (taking octave equivalence into account).
36. On which there already exists copious literature: see especially Spies (1965a),
(1965b) and (1967); Rogers (1968); Hogan (1982); Van den Toorn (1983), pp.
4424; Babbitt (1986) and (1987a); Morris (1988); and Straus (2001), pp. 2633.
Stravinskys rst composition to make use of rotational arrays is Movements (1958
9). In Locanto (2002), pp. 5961, I interpret some characteristics of these tables in
relation to the serial procedures employed by Stravinsky in earlier serial composi-
tion of the 1950s.
37. The two steps in which the procedure (rotation and transposition) is articulated are
clearly illustrated in the autograph table transcribed in Ex. 17: columns a and b
contain the rotated forms; columns g and d contain the rotated-transposed forms.
38. The serial symbols of the sketch can be deciphered as follows: II = second funda-
mental row; Inv = inversion (I); R inv = retrograde inversion (RI); a = rst
hexachord; 1
st
= rst line of the rotational array; 2
nd
= second line of the rota-
tional array (etc.); 1
st
, 1 & 2 = rst and second notes of the rst line of the array;
2
nd
, 1 & 2 = rst and second notes of the second line of the array; and so on.
39. See in particular Tucker (1992), vol. 2, pp. 6092; see also Pousseur (1971/i), pp.
2730, and Van den Toorn (1983), pp. 390413.
40. See Locanto (2002), pp. 3049. The use of motives formed by the combination of
tones and semitones is a typical trait of many of Stravinskys serial compositions.
Particularly common are the motives dened by Straus as twist motives, formed by
the tone and semitone conjoined in opposite directions to form set class [012] (see
for example Straus 2001, p. 91). These motives can be identied in many twelve-
note rows, from Surge, Aquilo (Canticum sacrum) to The Owl and the Pussy-Cat (see
the statistic in Straus 2001, p. 90 n. 13, based on Jers 1986, pp. 335) and in several
pre-serial compositions (see for example Carter 1997). Although Straus (p. 90)
observes that the semitone and tone often appear united in the same direction to
form a set of class [013], in his analyses the two forms are considered to be distinct
motives. For example, his analysis of the serial melody of Fanfare for a NewTheater
(p. 91) notes only the presence of the motive [012] in which the two intervals are
oriented in opposing directions (see Schema 1). However, considering the set [013]
as another form of the same motivic class 12, one will see that the row contains two
more motives between B and C
D and between DC and D
, thus appearing as a
continuous chain of seven motives (see Schema 2).
Schema 1
A A B C D C D F E F G G
<1, +2> <+1, 2> <+2, 1> <+2, 1>
<1, +2>
Schema 2
41. As was demonstrated by Tucker (1992), vol. 2, p. 186, on the basis of a thorough
examination of the sketches. For the sake of further clarity and completeness, I will
briey summarise the way in which the principal rows employed in this part of the
ballet are obtained from the initial tetrachord (for further details I refer the reader
to Tucker 1992, vol. 2, pp. 182242; on the technique of tetrachordal linkage, see
also Van den Toorn 1983, pp. 40914, and Locanto 2002, pp. 3447). The serial
heptachord GA
ACD
that appears in the coda of the Pas-de-Deux (bars

495496) results from the union through a common tone of a <0, 1, 4, 3> tetra-
chord (GA
) and its RI form (B
ACD
). Then, by combining through a

common tone two forms of this heptachord placed a tritone apart (DE
FE
GA
+ A
ACBB
D), Stravinsky forms the thirteen-note row employed in

the second section (Adagio) of the Pas-de-Deux a row containing all twelve pitch
classes with a single repetition. The twelve-note row used in the two nal move-
ments of the ballet (Four Duos and Four Trios) is obtained by uniting, through two
common tones, the heptachord of the coda of the Pas-de-Deux with another row of
seven notes, arrived at through the union (again via a common tone) of the
fundamental tetrachord <0, 1, 4, 3> and a slightly modied form of that tetrachord,
ordered as <1, 4, 3, 0> instead of <0, 1, 4, 3>.
42. That is, from the Pas-de-Quatre to the Triple Pas-de-Quatre.
43. As Tucker (1992), vol. 2, pp. 7680 (on the basis of the transcribed sketches, p. 13),
demonstrates, a point of contact between the use of intervallic motives in the strictly
melodic sense, which characterises the non-serial movements of the ballet, and the
serial technique employed in the following movements can be glimpsed in bars
104107 of theTriple Pas-de-Quatre, where the 212 motive, which rst appeared
in various congurations (see for example its use in the melody of Ex. 11), now
becomes xed in the form of the ordered tetrachord <1, 3, 2, 0>, which from that
moment comes to be used as the referential row.
44. See also the clarinet melody in bars 6768 of the Double Pas-de-Quatre, analysed
in Pousseur (1971/i), p. 42.
45. See for example Stravinsky and Craft (1966), p. 60, where the twelve-note row of
the Variations is described as a succession of notes that came to my mind as a
melody.
46. The genesis of the row for Epitaphium nds strong conrmation in a page of
sketches which was addressed recently by Straus (2001), pp. 613. Strauss analyses
nd within the rows structure a certain number of segments belonging to set class
[016], separated by the distance of T
4
(see Straus 2001, pp. 613, 99102 and
1301). The importance of the semitonal relation, which Stravinsky pointed out, is
set in relief in Locanto (2002), pp. 12831.
47. See Craft (1972), p. 98. As I have already mentioned, in the Requiem Canticles two
different fundamental rows are employed, indicated in the sketches by roman
numerals I and II. Nevertheless, the two rows are employed simultaneously only in
the Interlude and the Postlude, while the remaining movements employ them
alternatively.
48. According to the date of the sketches and the testimony of Robert Craft (see
Stravinsky 1984, pp. 46771), the Interlude was composed between the middle of
March and 17 October 1965.
49. For a complete inventory and a summary description of these sketches, see Locanto
(2002), pp. 11517.
50. See Locanto (2002), pp. 11617.
51. Stravinsky (1984), p. 467, maintains that this sketch was preceded by a little sketch,
dated March 1965, containing three annotations of row II, in each of which various
metrical indications are visible, followed by a brief musical passage (also based on
row II) of which there is no trace in the nal score. However, there is no proof that
this sketch denitely preceded the sketch transcribed in my Ex. 12a, and in any case
it does not contain annotations of row I.
52. Currently the second clipping is detached from the piece of cardboard, on which
remain the traces of the adhesive tape which originally held them together.
53. Indeed, the row extracted from this phrase is numbered I.
54. See Straus (2001), pp. 4952.
55. Reproduced in Craft (1967), pp. 1317.
56. The row was then to have been employed in a symphonic composition, which was
never brought to completion.
57. See Straus (2001), pp. 4952. The divergence between my analysis and that of
Straus depends, even in this case, on our different conceptions of intervallic
motive, with respect to which see nn. 6 and 40. In this case, moreover, the
difference is accentuated by the fact that in my analysis, as opposed to Strauss, the
presence of the motive also arises in the vertical dimension.
58. The row is clearly visible in Newmans nal photograph.
59. Note also that the enharmonic spelling of the row claries its basis in the circle of
fths.
60. One could perhaps glimpse an allusion to this substantial modication in Stravin-
skys story, reported by Craft, commenting on Newmans photographs: During the
morning of December 13, Stravinsky mentioned the need to put an idea in order,
but when the sketch was completed, in about thirty-ve minutes, he said that the
music he had actually written was something different and that it had not been in
his mind as long as an hour before. And he always seems to know exactly when his
imagination is at the starting line. Shortly after he had nished, Mrs. Stravinsky
clapped her hands together in the hall below his studio, her signal that lunch is
ready, and Stravinsky applauded back, his signal that he is, too. At table, to show
what he meant by putting in order, the composer placed three wine glasses in
parade formation, then interchanged the rst and third, saying, It is a matter of
knowing that the notes must be this way and not the other (Craft 1967, p. 13, italics
added; see also Craft 1972, pp. 303 and 311).
61. The dotted minim in the middle of the third line of the sketch shown in Ex. 14b
clearly appears to be a B3 in the sketch. However, this is obviously an error deriving
from the inconsistent inscription. The understood note is doubtless C4, as is seen
either in the serial hexachord employed in the passage (rst line above in the
sketch), or in the nal score (bar 47, trombone I).
62. Which stand for RIa and Rb, respectively.
63. This demonstrates, once again, that from Stravinskys point of view the order of
succession of the notes was less important than their global intervallic content.
64. On the use of this technique in De elegia tertia (Sensus spei) of Threni, see Tucker
(1992), vol. 2, p. 251, where the author establishes, with the assistance of the
sketches, the serial segments employed for the construction of the tenor melody of
Eradicationem (bars 252259). However, according toTucker, Stravinskys choice
of various segments was determined by no compositional system (p. 251). On
the contrary, in Locanto (2002), pp. 1347, I attempt to demonstrate that in all of
the preliminary versions of the passage, as in the nal version, the choice is
determined by the desire to create a dense chain of overlapping 15 motives.
65. The complete autograph short score comprises four separate pages. Only the page
transcribed in Ex. 15a which, unlike the others, is merely drafted contains serial
symbols. Its content corresponds to the printed score, with the sole exception of the
E in place of D in the third bar of the tenor.The reading in the score (D) is probably
incorrect, inasmuch as the E of the autograph short score nds conrmation in the
serial tables. However, even if this reading is treated as an intentional variant
introduced by Stravinsky (it appears for the rst time in the clean copy of the score),
that does not challenge the validity of my analysis, which concerns the creative
process from its earliest stages.
66. Note also the effect created by the occasional doubling of pitches in the vocal parts.
67. Starting in bar 4, the pitch organisation of the choral part is based on a different
serial technique, which involves reading the rotational arrays vertically. On this
technique see especially Straus (2001), pp. 15264.
68. See Stravinsky and Craft (1960), pp. 1001. Analyses of the ute solo can be found
in White (1966), p. 612; Mller (1984); Babbitt (1986), p. 255; Tucker (1992), vol.
2, p. 258; Rust (1994), pp. 6471; and Straus (2001), pp. 658 (based on Rust) and
12530.
69. Set class [012] could also be formed by uniting two semitones, but the motives
formed by two different intervals are far more typical of Stravinskys music.
70. The sketches which I consider (all in the Paul Sacher Foundation, Igor Stravinsky
Collection) include (1) the sketch for bars 1317 transcribed in Ex. 19a; (2) the
page of sketches with two versions of bars 712 transcribed in Ex. 21; (3) a clipping
with a preliminary version of bars 1821, without serial symbols (not reproduced
here); (4) a clipping with the denitive version of bars 1821 and (on the other side)
bar 22, without serial symbols (not reproduced here); and (5) the autograph short
score, some of the serial symbols from which are transcribed in Ex. 22. If one
excludes the serial analysis of the accompaniment in bars 1317 in Rust (1994), p.
70, based on the score (no. 5), sketches nos 25 of the list have never been
considered until now. In contrast, sketch no. 1 has been cited and discussed
numerous times in the musicological literature. One of the rst transcriptions
appeared in Neidhfer (1991) (in the same year Joseph Straus presented this
document in a paper at the Annual Meeting of the Society for Music Theory in
Cincinnati, OH). Later the document was analysed by Tucker (1992), vol. 1, p. 66
(transcription) and (1992), vol. 2, p. 258 (commentary); Rust (1994), pp. 634; and
Straus (2001), p. 67, which nally established the serial construction of the ute
solo, without, however, comparing it with either the construction (actually quite
similar) of the other two episodes, or with the motivic-intervallic construction to
which it is subjected.
71. The major discrepancy concerns the rst two segments: in the short score of the
piano episode they are labelled segment 36 of hexachord g5 and segment 56 of
hexachord d1. In the sketch for the ute episode, however, they are labelled segment
13 of hexachord g1 and segment 46 of hexachord d1. However, note that when
the two piano segments are transposed by T
2
(rather than T
4
, as in the ute), the
resulting pitches are the same as in the ute episode.
72. In the rst version, the serial segments were transposed a major third higher (T
4
);
the second version, on the other hand, was obtained by lowering the rst version by
a tone (T
2
). Therefore, in the second version, the serial segments prove to be
transposed a tone higher (T
4
- T
2
= T
2
). The short score does not contain serial
symbols for the nal part of the passage, formed by the succession of thirteen
pitches notated in the nal part of the third line of Ex. 21. The same succession of
pitches this time without the initial C
is transcribed again lower on the page

(lines 56) with a different choice of octave register for the single notes and using
a rhythm outlined for the nal cadence.
73. The B that appears as the rst note after the treble clef corresponds to the D
of the
preceding version, placed immediately before the treble clef. See also the line drawn
by Stravinsky. A different version of the nal part of the phrase alone (starting with
the dyad B
A on line 13) is also notated by hand in the lower part of the page (lines
1214).
74. The error in the printed score has created some difculties in the analytical
literature. See for example Rust (1994), p. 68, where the omission of the notes B
and C prevented the author from recognising the T
2
relation between the last ten
notes of the rst piano episode and the corresponding notes of the ute solo.
Moreover, Rusts analysis omitted the last bar (bar 22) of the second episode for the
piano, so that he failed to notice the T
8
relation with the ute solo in the last six
notes.
75. See Boykan (1963), p. 158; Walden (1979); Rust (1994), pp. 6276; and Straus
(2001), pp. 1248.
76. On the basis of the principle of rotation on which the serial tables are based, one
could say that the pitch following the nal pitch of a hexachord is the rst.
77. For the sake of convenience, in Ex. 24 the entire succession of pitches is transposed
to F (as in the rst piano episode).
78. Note that, in the denitive score, one of the very rare points at which this continu-
ous motivic chain is interrupted corresponds to the juncture at which Stravinsky
omitted the notes B and C in the rst episode.
REFERENCES
Antokoletz, Elliott, 1984: The Music of Bla Bartk: a Study of Tonality and
Progression in Twentieth-Century Music (Berkeley and Los Angeles, CA: Uni-
versity of California Press).
Babbitt, Milton, 1960: Twelve-Tone Invariants as Compositional Determi-
nants, Musical Quarterly, 46/ii, pp. 24660.
______, 1961: Set Structure as a Compositional Determinant, Journal of Music
Theory, 5/ii, pp. 7294.
______, 1968: Remarks on the Recent Stravinsky, in Benjamin Boretz and
EdwardT. Cone (eds.), Perspectives on Schoenberg and Stravinsky (Princeton,
NJ: Princeton University Press), pp. 16585.
______, 1986: Order, Symmetry, and Centricity in Late Stravinsky, in Jann
Pasler (ed.), Confronting Stravinsky (Berkeley, CA: University of California
Press), pp. 24761.
______, 1987a: Stravinskys Verticals and Schoenbergs Diagonals, in Ethan
Haimo and Paul Johnson (eds.), Stravinsky Retrospectives (Lincoln, NE:
University of Nebraska Press), pp. 1535.
______, 1987b: Words about Music, ed. Stephen Dembski and Joseph N. Straus
(Madison, WI: University of Wisconsin).
Bailey, Kathryn, 1996: Weberns RowTables, in Kathryn Bailey (ed.), Webern
Studies (Cambridge: Cambridge University Press), pp. 170228.
Beach, David, 1976: Segmental Invariance and theTwelveTone System, Journal
of Music Theory, 20, pp. 15784.
Berger, Arthur, 1963: Problems of Pitch Organization in Stravinskys Diatonic
Music, in Benjamin Boretz and Edward T. Cone (eds.), Perspectives on
Schoenberg and Stravinsky (Princeton, NJ: Princeton University Press), pp.
12355.
Boss, Jack, 1992: Schoenbergs op. 22 Radio Talk and Developing Variation in
Atonal Music, Music Theory Spectrum, 14/ii, pp. 12550.
______, 1994: Schoenberg on Ornamentation and Structural Levels, Journal of
Music Theory, 38/ii, pp. 187216.
Boykan, Martin, 1963, Neoclassicism in Late Stravinsky, Perspectives of New
Music, 1/ii, pp. 15569.
Carter, Chandler, 1997: Stravinskys Special Sense: the Rhetorical Use of
Tonality in The Rakes Progresss, Music Theory Spectrum, 19/i, pp. 5580.
Craft, Robert, 1967: Bravo Stravinsky. Photographs by Arnold Newman (Cleve-
land, OH: World).
______, 1972: Stravinsky: Chronicle of a Friendship, 19481971 (NewYork: Knopf).
Dahlhaus, Carl, 1986: Was heisst entwickelnde Variation? in Rudolph
Stephan and Sigrid Wiesmann (eds.), Bericht ber den zweiten Kongress der
Internationalen Schnberg-Gesellschaft (Wien: Universal), pp. 2805.
Forte, Allen, 1972: The Structure of Atonal Music (New Haven, CT: Yale Uni-
versity Press).
Haimo, Ethan, 1997: Developing Variation and Schoenbergs Serial Music,
Music Analysis, 16/iii, pp. 34965.
Hogan, Catherine, 1982: Treni: Stravinskys Debt to Krenek, Tempo, 141, pp.
229.
Jers, Norbert, 1986: Igor Stravinskys spate zwlf-tonWerke (19581966) (Regens-
burg: Gustav Bosse).
Johnson, Paul, 1987: Cross-Collectional Techniques of Structure in Stravinskys
Centric Music, in Ethan Haimo and Paul Johnson (eds.), Stravinsky Retro-
spectives (Lincoln, NE: University of Nebraska Press), pp. 5575.
Krenek, Ernst, 1940: Studies in Counterpoint Based on the Twelve-Tone Technique
(NewYork: G. Schirmer).
Lewin, David, 1962: A Theory of Segmental Association in Twelve-Tone
Music, Perspectives of New Music, 1/i, pp. 89116.
Locanto, Massimiliano, 2002: Pensiero musicale e procedimenti costruttivi
nellultimo Stravinsky (PhD diss., University of Pavia).
Morris, Robert, 1988: Generalizing Rotational Arrays, Journal of Music Theory,
32/i, pp. 75132.
Mller, Alfred, 1984: Igor Strawinsky: Movements for Piano and Orchestra,
Melos, 46/ii, pp. 11239.
Nabokov, Nicolas, 1949: Christmas with Igor Stravinsky, in Edwin Corle (ed.),
Igor Stravinsky (NewYork: Sloan & Pearce), pp. 12368.
Neidhfer, Christoph, 1999: An Approach to Interrelating Counterpoint and
Serialism in the Music of Igor Stravinsky, Focusing on the Principal Diatonic
Works of His Transitional Period (PhD diss., Harvard University).
______, 1991: Analysearbeit im Fach Komposition/Musiktheorie ber die
Movements for Piano and Orchestra von Igor Strawinsky (masters thesis,
Musik Akademie der Stadt Basel).
Perle, George, 1955: Symmetrical Formations in the String Quartets of Bla
Bartk, Music Review, 16, pp. 30012.
______, 1991: Serial Composition and Atonality: an Introduction to the Music of
Schoenberg, Berg, andWebern, 6th revised edn (Berkeley and Los Angeles, CA:
University of California Press).
Pousseur, Henry, 1971: Stravinsky selon Webern selon Stravinsky, Musique en
jeu, 3/iii, pp. 2147 and 10726.
Powles, Jonathan C., 1995: Continuity and Discontinuity in the Music of
Stravinsky: Analysis, Theory and Meta-Theory (PhD diss., Oxford
University).
Rogers, John, 1968: Some Properties of Non-duplicating Rotational Arrays,
Perspectives of New Music, 7/i, pp. 80102.
Rust, Douglas, 1994: Stravinskys Twelve-Note Loom: Composition and Pre-
composition in Movements, Music Theory Spectrum, 16/i, pp. 6276.
Schoenberg, Arnold, 1967: Fundamentals of Musical Composition, ed. Gerald
Strang and Leonard Stein (Boston and London: Faber and Faber).
______, 1995: The Musical Idea, and the Logic,Technique, and Art of Its Presentation,
ed. and trans. Patricia Carpenter and Severine Neff (NewYork: Columbia
University Press).
Sichardt, Martina, 1990: Die Entstehung der Zwlftonmethode Arnold Schnbergs
(Mainz: Schott).
Smyth, David H., 1997: Stravinsky at the Threshold: a Sketch Leaf for Canti-
cum sacrum, Mitteilungen der Paul Sacher Stiftung, 10, pp. 216.
______, 1999: Stravinskys Second Crisis: Reading the Early Serial Sketches,
Perspectives of New Music, 37/ii, pp. 11746.
______, 2000: Stravinsky as Serialist: the Sketches for Threni, Music Theory
Spectrum, 22/ii, pp. 20524.
Spies, Claudio, 1965a: Some Notes on Stravinskys Abraham and Isaac, Per-
spectives of New Music, 3/ii, pp. 186209.
______, 1965b: Some Notes on Stravinskys Variations, Perspectives of New
Music, 4/i, pp. 6274.
______, 1967: Some Notes on Stravinskys Requiem Settings, Perspectives of New
Music, 5/ii, pp. 98123.
Straus, Joseph N., 1990: Introduction to Post-Tonal Theory (Englewood Cliffs, NJ:
Prentice-Hall).
______, 1991: The Progress of a Motive in Stravinskys The Rakes Progress,
Journal of Musicology, 9/ii, pp. 16585.
______, 2001: Stravinskys Late Music (Cambridge: Cambridge University
Press).
Stravinsky, Igor, 1984: Selected Correspondence, ed. Robert Craft, vol. 2 (Boston
and London: Faber and Faber).
Stravinsky, Igor and Craft, Robert, 1959: Conversations with Igor Stravinsky
(Garden City, NY: Doubleday).
______, 1960: Memories and Commentaries (Garden City, NY: Doubleday).
______, 1962: Expositions and Developments (Garden City, NY: Doubleday).
______, 1966: Themes and Episodes (NewYork: Knopf).
______, 1969: Retrospectives and Conclusions (NewYork: Knopf).
Taruskin, Richard, 1993: TheTradition Revisited: Stravinskys Requiem Canticles
as Russian Music, in Christopher Hatch and David W. Bernstein (eds.),
Music Theory and the Exploration of the Past (Chicago: University of Chicago
Press), pp. 52550.
______, 1996: Stravinsky and the RussianTradition: a Biography of theWorks through
Mavra (Berkeley and Los Angeles, CA: University of California Press).
Treitler, Leo, 1959: Harmonic Procedure in the Fourth Quartet of Bla Bartk,
Journal of Music Theory, 3/ii, pp. 29298.
Tucker, Susannah, 1992: Stravinsky and His Sketches: the Composition of Agon
and Other Serial Works of the 1950s, 2 vols. (PhD diss., Oxford University).
Tymoczko, Dmitri, 2002: Stravinsky and the Octatonic: a Reconsideration,
Music Theory Spectrum, 24/i, pp. 68102.
Van den Toorn, Pieter C., 1983: The Music of Igor Stravinsky (New Haven, CT:
Yale University Press).
Van denToorn, Pieter and Tymoczko, Dmitri, 2003: Colloquy about Stravinsky
and the Octatonic: a Reconsideration, Music Theory Spectrum, 25/i, pp.
167203.
Walden, William, 1979: Stravinskys Movements for Piano and Orchestra: the
Relationship of Formal Structure, Serial Technique, and Orchestration,
Journal of the Canadian Association of University Schools of Music, 9/i, pp.
7395.
Wason, Robert W., 1996: A Pitch-Class Motive in Weberns George Lieder, Op.
3, in Kathryn Bailey (ed.), Webern Studies (Cambridge: Cambridge Univer-
sity Press), pp. 11134.
White, Eric Walter, 1966: Stravinsky: the Composer and His Work (Berkeley, CA:
University of California Press).
ABSTRACT
In Stravinskys nal serial works, the intervallic component assumed a more
decisive role than in his in earlier compositions, becoming the foundational
aspect of a motivic technique, the specic aspects of which are illustrated here
through several sketch-based analyses. In Stravinskys case, motivic-intervallic
syntax and serial procedures operate according to slightly but signicantly dif-
ferent criteria: the rst on the level of single intervals, the second on the level of
pitch-class sets. Using several specic examples drawn from compositions
ranging from Agon to the Requiem Canticles, this article demonstrates that the
discrepancy provided Stravinsky with a stimulus, rather than an obstacle, to
composition, and provides a guide to the interpretation of certain well-known
characteristics both of his creative process and of his serial technique. The
music-theoretical aspects of Stravinskys intervallic syntax are illustrated, and its
interaction with serial technique is observed from two conceptually different
directions: from intervallic motives to rows (the initial denition of a row of
pitches), and from the row to intervallic motives (the transformation of the
abstract row into concrete musical contexts).

Locanto - Stravinsky&#39;s Late Technique

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Locanto - Stravinsky&#39;s Late Technique

Enviado por

Direitos autorais:

Formatos disponíveis

:~ssI:IiI~No ioc~N1o

Co:iosINo vI1n IN1riv~is: IN1riv~iiIc S.N1~x ~No SriI~i

B [0156]. This time the

, thereby giving rise to a sonority containing an ic1 (FF

). The recapitulation (bar 97) is practically

to G. Finally, notice that the

is a constant presence throughout the layer in its entirety, thus

to G in the bass voice, a

). In the second version (the lower

. The alteration forms set class [0157], which

) along with an ic1 between B

) is transposed down a semitone (T

of the upper part forms, however, a relation of a

). Furthermore, the two motives (EFG and D

), divided symmetrically into two

in the upper parts). The last vertical sonority of three

EG) forms a 12 motive which creates a strong link with the

, thus avoiding the presence of

throws light on the motivic-intervallic logic which

in common); in the last three notes of the cello (B

F), grouped together

) and the D of the cello

of the viola and the D of the cello

respectively. By making this adjustment, Stravinsky not

) become two motives of the

to F, while the combination of the

in the nal row creates, in this case,

) are separated by a semitone,

) gradually take the form of a sequence of

), which is completed in bar 3 with the addition of

in the bass part.

before the notated clefs, the composer had

circled in the middle of the ninth line which

. It was most likely in this manner that Stravinsky sought to obtain

ACB belongs to set class [0123]; con-

to G is found also in the lower

which leaps directly to B, thus anticipating the

D and between DC and D

that appears in the coda of the Pas-de-Deux (bars

) and its RI form (B

). Then, by combining through a

D), Stravinsky forms the thirteen-note row employed in

is transcribed again lower on the page

Você também pode gostar

Locanto - Stravinsky's Late Technique

Locanto - Stravinsky's Late Technique