Tw elve-Tone M usic II: Invariance, Sym m etry, and C om binatoriality

Conclusions
In this piece we find again two of the characteristic traits of Webern's twelve-tone mu­
sic: use of symmetry and invariance as essential compositional elements. As we listen
to this movement, we experience a sense of sweeping coherence and unity. Both rhythm
and meter contribute to this immediate perception. Rhythmically, we hear a constant
and driving flow of eighth notes. Eighth-note attacks are grouped into two and four.
Some of the two-attack groups include grace notes, and some are chordal rather than
single notes. Groups of attacks are separated by rests. With only two exceptions, the
duration of rests is of one eighth note. The two exceptions are m. 3 (three eighth notes)
and mm. 21-22 (four eighth notes). Rhythm is thus greatly unified, although unpredict­
able. It is unified by the flow of eighth notes, by the groupings into two and four, and by
the eighth-note rests. Unpredictability comes from the irregular order of two and four
groupings, by the two rests longer than an eighth note, and by elements such as the grace
notes and the chords. Even more unpredictable are the various and constantly changing
dynamic markings, the articulations, and the widely disjunct registers of attacks.
From the point of view of pitch, strong coherence is provided by several factors.
The row provides, as is always the case, intervallic invariance between row forms.
Harmonic unity comes from dyadic invariance within row pairs and among row pairs,
along with the use of the (016) Viennese trichord. Finally, and very significantly, sym­
metry around A (emphasized by the A 4-A 4 dyad at four points in the piece) provides a
type of pitch centricity that we don’t usually associate, at least as a norm, with twelve-
tone music. In its conciseness, cohesion, extreme economy of means, and structural
coherence, this movement is arguably one of W ebern’s compositional masterpieces.

HEXACHORDAL COMBINATORIALITY
In our studies of twelve-tone music so far, it has become apparent that W ebern’s rows
often feature derivation and symmetry, and that he often took advantage of invariance
in his structural designs involving complexes of row forms. We also find use of in­
variance, although perhaps to a lesser extent, in Schoenberg’s twelve-tone music. But
Schoenberg was even more interested in a property of some rows which we call com-
binatoriality.4 Combinatorial rows are those that can produce some transposition,
inversion, retrograde, and/or retrograde inversion in which the first six pitch classes
will be completely different from the first six pitch classes in the original row. The
first hexachord of the original row and the first hexachord of the transformed row will
thus be complementary (that is, the combination of both first hexachords will contain

4An essential source for combinatorial theory is Babbitt’s article “Set Structure as a Compositional De­
terminant,” cited in Chapter 7, footnote 1. Two important sources for further study of combinatoriality are
Daniel Starr and Robert Morris, “A General Theory of Combinatoriality and the Aggregate,” Perspectives
o f New Music 16/1 (1977-78): 364-89, and 16/2: 50-84; and Robert Morris, “Combinatoriality without
the Aggregate,” Perspectives o f New Music 21/1-2 (1982-83): 432-86. See also Morris, Class Notes fo r
Atonal Music Theory, Chapters 12 and 13.
196 CH A PTER 8

w # Example 8.11 Combinatoriality in Schoenberg’s row from Klavierstück, op. 33a

the complete collection of twelve pitch classes, the aggregate). Because this type of
combinatoriality involves the segmentation of the row into hexachords, we call it hexa-
chordal combinatoriality.

^ NOTE

Other types o f combinatoriality are possible, as we will study in Chapter 9: tetra-

chordal com binatoriality (in which the aggregate results from combining tetrachords
fro m three different row forms), trichordal combinatoriality (the aggregate results
from combining trichords from fo u r different row forms), dyadic combinatoriality (the
aggregate results fro m combining dyads fro m six different row forms), and unitary
combinatoriality (the aggregate results fro m combining individual pitch classes from
twelve different row forms). The segments combined to form the aggregate may also
contain irregular numbers o f pitch classes.

Take, for instance, the row for Schoenberg’s Klavierstiick, op. 33a, shown in Ex­
ample 8.11. If we look at the pitch-class content of the first hexachord of PQwe can see
that it is totally different from the pitch-class content of the first hexachord for I 5 (hence
the labels Hj and H 2, which denote different hexachordal content). Consequently, the
pitch-class content of the first hexachord of PQ and the second hexachord of I 5 will
be the same, although not necessarily in the same order (both are labeled as Hj), and
the same applies to the second hexachord of P 0 and the first hexachord of I 5 (both are
labeled as H 2). We can say, then, that this row is combinatorial, and that PQ and I 5 (or
any similarly related row forms, such as R q-R I5, P 2 - I 7, R 2 -R I 7, P 5 - I 1 0 ’ anc*so on) are
combinatorially related.
We can now examine a musical application of this property in Example 8.12. In
mm. 23-25, the right hand of the piano is based on the first hexachord of P ]0, while the
left hand presents the first hexachord of I3. P 1 0 and I 3 are combinatorially related, so
there is no pitch-class invariance between both hexachords (or both hands). The same
 Twelve-Tone Music II: Invariance, Symmetry, and Combinatoriality

Ah Example 8.12 Schoenberg, K la vierstü ck, op. 33a, mm. 23-26

applies to m. 26, where we hear the second hexachord of P ] 0 in the right hand and the
first hexachord of I 3 in the left hand.
Looking back at Example 8.11, we see that a result of combinatoriality is that we
hear the twelve pitch classes twice “horizontally” as rows (that is, as ordered sets), and
twice “vertically” as the hexachordal combinations H j-H ^ and H 2 -H j. The latter are
unordered collections of twelve pitch classes (that is, aggregates). The complete combi­
natorial complex in Example 8.11 is called an array, and it contains two rows and two
unordered aggregates. The musical consequences of such twelve-tone complexes are
quite evident: in Example 8.12 we can see that by using an array of two combinatorially
related rows, Schoenberg avoided any kind of invariance between row forms (contrary
to what we saw W ebern was doing in pieces such as the Concerto, op. 24, and the Sym-
phonie, op. 2 1 ), and he presented the complete twelve-tone complex in two dimensions:
we hear it twice (allowing for musical repetitions of some pitch classes) throughout
mm. 23-26 in each of the hands, and twice more as the combined pitch-class content of
both hands together in mm. 23-25 and 26, respectively.

Types of Combinatoriality
The type of combinatoriality most commonly found in Schoenberg’s music is inver-
sional combinatoriality, in which a row is combinatorial with an inverted form of itself.
This is the type we have illustrated in Examples 8.11 and 8.12. There are, however, four
types of combinatoriality: prime, inversional, retrograde, and retrograde-inversional.
Before we examine each of these types, let us recall some properties of hexachordal
complementarity that we studied in Chapter 3. O f the fifty hexachordal set classes,
twenty are self-complementary (that is, the complement of each of these sets is the
same set), and the remaining thirty sets are all Z-sets, related to their complements
by the Z-relation. In all cases, a hexachord and its complement (which will be either
the same hexachord or a Z-related one) have the same interval-class vector (that is, the
198 CH A PTER 8

a. P0:

Pn:
Tr
)

b. Pn: 0 1 5 4 9 8 - 7 E T 3 2 6

P 6: 6 7 E T 3 2 - 1 5 4 9 8 0
Figure 8.4 Prime combinatoriality

same interval content). A ll self-complementary hexachords, moreover, can by defini­

tion map onto their complement under transposition or inversion (whereas this is not the
case for Z-related complementary hexachords). Having reviewed these matters, let us
now define the four types of combinatoriality.

Prime Combinatoriality
In prime combinatoriality, a row is combinatorial with a transposition of itself, as
shown in Figure 8.4.
In P-combinatoriality, a hexachord (Hj) maps onto its complement (H ,) under Tn.
H 9 is obviously the complement of Hj (because H, and H ? make up the complete row).
But if PQand Pn are related by prime combinatoriality, as shown in Figure 8.4a, Hj also
maps onto H 2 (its complement) under Tn to form an aggregate that combines the first
hexachords of PQand Pn, respectively. The row shown in Figure 8.4b is P-combinatorial
at Pg. Verify the properties of P-combinatoriality as applied to this particular example.

Inversional Combinatoriality
In inversional combinatoriality, a row is combinatorial with an inversion of itself, as
shown in Figure 8.5a.
In I-combinatoriality, a hexachord maps onto its complement under T I. Hj maps
onto H t (its complement) under TnI to form an aggregate that combines the first hexa­
chords of P 0 and In, respectively. Verify the properties of I-combinatoriality as applied
to the I-combinatorial row forms shown in Figure 8.5b.
The fragment from Schoenberg’s String Q uartet no. 4 reproduced in Example 8.1
is also an excellent example of a musical application of inversional hexachordal com­
binatoriality. The fragment is reproduced again in Example 8.13. If you refer back to
Figure 8.2 at the beginning of this chapter, you will find P Qand I 5 for the row in this
 Tw elve-Tone M usic II: Invariance, Sym m etry, and C om binatoriality

ini
a. ' H, h2
P°: /-
i„:
Tnl (
h2 Hi
)

b. P0: 0 1 5 4 9 8 - 7 E T 3 2 6

I3: 3 2 T E 6 7 - 8 4 5 0 1 9
F ig u re 8.5 Inversional com binatoriality

piece. If you examine these two row forms for combinatoriality, you will see this row is
indeed inversionally combinatorial at Iv We already know that the fragment in Exam ­
ple 8.13 is based on P 0 and I7. The two violins first state P 2 and then I7, while the lower
instruments reverse the order of presentation to I 7 - P 2- In other words, we hear P , and
I? simultaneously twice. Not only this, but we also hear each hexachord of P 9 simul­
taneously with the corresponding hexachord of I?. Because P 2 and I 7 are inversionally
combinatorial (they are related by T 5 I, as P 0 and I 5 are), the first hexachords of each row
form create the aggregate, and so do the second hexachords. In our passage, then, we
hear four statements of the row (P 2 and I 7 in the violins, Iy and P 0 in the cello and viola)
and also four statements of the aggregate (because we hear four vertical simultaneities
of Hj and H 9, as shown in Example 8.13.

Retrograde Combinatoriality
In retrograde combinatoriality, a row is combinatorial with a retrograde form of it­
self, as shown in Figure 8 .6 a. All rows are trivially R-combinatorial with their own
retrograde (P 0 is obviously combinatorial with Rq).
In R-combinatoriality a hexachord maps onto itself under Tn. Under T , Hj of PQ
would map onto Hj of Pn and H 2 of P Qwould map onto H 2 of P n (that is, both hexa­
chords would map onto themselves under T ). Because we read Pn backward as R n,
however, the mapping of Hj onto itself becomes the second hexachord of Rn, and thus
P 0 and Rn are combinatorial. Verify these properties in the R-combinatorial row forms
shown in Figure 8 .6 b.

Retrograde-Inversional Combinatoriality
In retrograde-inversional combinatoriality, a row is combinatorial with a retrograde-
inverted form of itself, as shown in Figure 8.7a.
200 CH A PTER 8

TcIEKJIKm Schoenberg, String Quartet no. 4 ,1, mm. 27-28

b. P0: 0 1 5 4 9 8 - 7 E T 3 2 6

R8: 2 T E 6 7 3 - 4 5 0 1 9 8
F ig u re 8.6 Retrograde com binatoriality

In Rl-com binatoriality a hexachord maps onto itself under TnI. Under T I, Hj of P 0

would map onto Hj of In and H 0 of P Qwould map onto H , of In (that is, both hexachords
would map onto themselves under TnI). Because we read In backwards as R In, how­
ever, the mapping of H, onto itself becomes the second hexachord of R In,‘ and thus P Q
and RI are combinatorial. Verify these properties in the Rl-combinatorial row forms
shown in Figure 8.7b.
 Tw elve-Tone M usic II: Invariance, Sym m etry, and C om binatoriality

b. P0: 0 1 5 4 9 8 - 7 E T 3 2 6

RI ^ 7 E T 3 2 6 - 5 4 9 8 0 1
Figure 8.7 Retrograde-inversional combinatoriality

The All-Combinatorial and Semicombinatorial Hexachords

The most relevant type of com binatoriality (and the one normally used by Schoenberg
and other composers) is inversional combinatoriality, in which a hexachord and its
complement are inversionally related. All fifty hexachords are R-combinatorial even
if it’s only at R0. Prime, inversional, and retrograde-inversional types of combina­
toriality, on the other hand, do not apply to any of the Z hexachords. O f the twenty
rem aining hexachords (the self-complementary ones), six feature all four types of
combinatoriality. These are the all-com binatorial hexachords, and they are listed
in Figure 8 .8 a (where each of them is identified with a letter designation ) . 5 Thirteen
hexachords are combinatorial only under one or two operations. These are the sem i­
com binatorial hexachords, and they are listed in Figure 8 .8 b. A ll of these hexachords
are I-combinatorial, and one of them, (013679), moreover, is also combinatorial at R6.
None is P- or Rl-com binatorial. Finally, one hexachord, (013458), is combinatorial
only under Tg.
It is im portant to keep in m ind that com binatoriality refers only to pitch-class
content of hexachords, not to pitch-class order. Take, for instance, the row for op. 33a.
The prim e form for H, is (012367), which is listed as a semicombinatorial hexachord
in Figure 8 .8 . Because in an I-com binatorial row, H,, is always an inverted form of
Hj (a TnI transformation), both Hj and are of the (012367) set class. We already
know that this row is combinatorial at I5, which means that H 2 is T5I of H r The first
hexachord of I5, on the other hand, also results from a T5I transformation of the first
hexachord of P0, and although the pitch-class content of both T5I transformations will
be the same (the H 2 hexachord), the pc ordering will not be the same, as you can verify
in Figure 8.9.

5The customary letter designation for all-combinatorial hexachords was originally used by Donald Mar­
tino in “The Source Set and Its Aggregate Formations.”
202 CH A PTER 8

a. All-combinatorial hexachords b. Semicombinatorial hexachords

A (0 1 2 3 4 5) (0 1 2 3 4 6 )
B (0 2 3 4 5 7) (0 1 2 3 5 7)
C (0 2 4 5 7 9) (0 1 2 3 6 7)
D (0 1 2 6 7 8) (0 1 2 4 5 8)
E (0 1 4 5 8 9) (0 1 2 4 6 8)
F (0 2 4 6 8 T) (0 1 2 5 7 8)
(0 1 3 4 6 9)
(0 1 3 5 7 9)
(0 1 3 6 7 9)
(0 1 4 5 6 8)
(0 1 4 5 7 9)
(0 2 3 4 6 8)
(0 2 3 5 7 9)
F igure 8.8 T he all-com binatorial and sem icom binatorial hexachords

TsI

H2
1 E 3 5 9 T 4 6 [both set types: (012367)]
Hi

4 6 9 2 0 8 7 1 E
Figure 8.9 Pitch-class content com binatoriality in the row for Schoenberg’s Klavierstück, op. 33a

Determining the Combinatorial Properties

of a Hexachord or a Row
How can we determ ine if a given row is combinatorial, and how can we find out at what
T, I, R, or RI levels it is combinatorial? Conversely, if we want to build a combinatorial
row, how do we go about it?
Determ ining whether a row is combinatorial is simple enough given the list of
combinatorial hexachords in Figure 8 .8 . All we need to do is figure out the prime form
for the first hexachord and verify whether that set class is listed as a combinatorial hexa­
chord in Figure 8 .8 . This does not tell us, however, at what index number or numbers
the row will be combinatorial. Similarly, to build a combinatorial row we can start from
 Tw elve-Tone M usic II: Invariance, Sym m etry, and C om binatoriality

Set: (014589) Summation square

Vector: [3 0 3 6 3 0] 0 1 4 5 8
0 0 1 4 5 8 9
1 1 2 5 6 9 T
4 4 5 8 9 0 1
5 5 6 9 T 1 2
8 8 9 0 1 4 5
9 9 T 1 2 5 6
F ig u re 8.10 Interval-class vector and sum m ation square for (014589)

any hexachord listed in Figure 8 .8 . But then we still don’t know what exact operation we
need to apply to the chosen hexachord to achieve a combinatorial form of the row.

General Procedure
We know that in P-combinatoriality, Hj maps onto its complement (H2) under Tn. In
other words, the pitch classes in common between Hj and such a Tn of Hj (that is, H ?)
will be zero (because the complement of Hj contains “the other” six pitch classes). We
need to find a Tn of with zero invariants with H j.
In I-combinatoriality, Hj maps onto its complement (H9) under TnI. Hj and such
a TnI of Hj (that is, H 2) will have zero pcs in common (because the complement of H :
contains “the other” six pitch classes). We need to find a TnI of Hj with zero invariants
with H r
In R-combinatoriality, Hj maps onto itself under Tn. Hj and such a Tn of H x will
thus have six pcs in common. We need to find a Tn of Hj with six invariants with H r
In Rl-combinatoriality, H ; maps onto itself under TnI. Hj and such a TnI of H : will
have six pcs in common. We need to find a TnI of H x with six invariants with H r
We can do all this by applying the procedures to determine invariant tones under
transposition or under inversion, which we studied in Chapter 3. Recall that the number
of invariant pcs under transposition by interval n is equal to the number of times interval
n occurs in the set (as shown in the set’s interval-class vector), with the only exception of
interval-class 6 , for which the number of invariant pcs is twice the number of occur­
rences of the interval. The number of invariants under TnI, on the other hand, will be
equal to the number of times integer n appears in the summation square for that set.
Let's apply these procedures to a specific hexachord. We will choose an all-
combinatorial hexachord, (014589), so we can demonstrate all four types of combinato­
riality. First, we list the interval-class vector for this set and build a summation square,
both shown in Figure 8.10.
For P-combinatoriality we need zero invariants under T . The second entry in the
vector (for T 0 and T 10) and the last entry (for Tg) show zero entries. The hexachord is
then P-combinatorial under t 2, t 10, and Tg.
204 CH A PTER 8

Po: 0 1 5 4 9 8 — 7 E T 3 2 6
P2: 2 3 7 6 E T — 9 1 0 5 4 8
In: E T 6 7 2 3 — 4 0 1 8 9 5
R4 : T 6 7 2 3 E - 0 1 8 9 5 4
Rig: 3 7 6 E T 2 — 1 0 5 4 8 9
F ig u re 8.11 A com binatorial row built from hexachord (014589)

For I-combinatoriality we need zero invariants under T I. We need to find what

integers do not appear at all in the summation square. Integers 3, 7, and 11 are absent,
so the hexachord is I-combinatorial under T 3 I, T ?I, and T n I.
For R-combinatoriality we need six invariants under T . These are provided by the
fourth entry in the vector (for T4 and Tg). Moreover, all hexachords are R-combinatorial
under TQ.
For Rl-com binatoriality we need six invariants under T I. Three integers appear
six times in the summation square, 1, 5, and 9. The hexachord is Rl-combinatorial un­
der T jl, T 5 I, and T 9 I.

Building a Combinatorial Row

Now let’s build P 0 of a combinatorial row using this hexachord. We can build the first
hexachord from the prime form (014589), which, of course, we don't need to keep in
the same order. L et’s say 0 1 5 4 9 8 . Now it will suffice to build H 2 as the comple­
ment of the first hexachord, and again we can choose any ordering, such as 7 E T 3 2 6 .
We thus have the row PQ: 0 1 5 4 9 8-7 E T 3 2 6 . Both of these hexachords are of
set class (014589), so we know that the preceding combinatorial relationships we have
listed should apply. Let us verify one of the relationships for each of the combinatorial
types. In Figure 8.11 we list PQ along with P 2 (which features P-combinatoriality), In
(I-combinatoriality), R4 (R-combinatoriality), and R I 9 (Rl-combinatoriality), and we
see that indeed all these row forms are combinatorial with PQ.

Determ ining the Combinatorial Properties of a Given Row

We have thus learned how to build a combinatorial row. Let us determine the procedure
to figure out the combinatorial levels of a given combinatorial row. Suppose that we try
to find out whether and how the row from Schoenberg’s op. 33a is combinatorial. P 0 is
0 7 3 1 E 8-3 5 9 T 4 6 . We find out the prime form for Hj, which is (012367), and verify
in Figure 8 . 8 that it is a combinatorial hexachord. Then we check its interval-class vec­
tor: [422232], Because neither zero nor six (nor three in the case of ic 6 ) are entries in
the vector, the row is not P-combinatorial nor R-combinatorial (except for the trivial
R-combinatoriality with Rq).
 Tw elve-Tone M usic II: Invariance, Sym m etry, and C om binatoriality 205

0 7 Jo 1 E 8
0 0 7 3 1 E 8
7 7 2 T 8 6 3
oJ 3 T 6 4 2 E
1 1 8 4 2 0 9
E E 6 2 0 T 7
8 8 3 E 9 7 4
F ig u re 8.12 The sum m ation square for the first hexachord of Schoenberg's Klavierstück,
op. 33a row

Now we need to build a summation square for the first hexachord. It is important
to keep in mind that to examine a specific row for I and RI combinatoriality, we need
to realize the summation square fro m the actual pitch classes in the row, and not from
a prim e form . So our summation square in this case will be for 0 7 3 1 E 8 , and not for
(012367), as shown in Figure 8.12.
One integer, 5, is missing from the square. So the row is I-combinatorial at I5.
No integer, on the other hand, appears six times in the square, so the row is not RI-
combinatorial. These procedures verify what we already know: the row for op. 33a is
combinatorial only at I 5 and Rq.

ANALYSIS 8.2: SCHOENBERG, KLAVIERSTÜCK,

O R 33A (ANTHOLOGY NO. 23)

We will conclude our analysis of “classical” twelve-tone music with a study of Schoen­
berg’s Klavierstück, op. 33a (1929), a composition more complex and of larger propor­
tions than the twelve-tone pieces we have analyzed so far . 6 First listen to the whole
piece, and as you listen, try to identify and m ark any distinctive and characteristic mu­
sical event (particular textures, themes and their character, sections and other formal
issues). Then let’s examine the two basic forms of the row on which much of the piece
is built. We already know that P 0 for this row is 0 7 3 1 E 8-3 5 9 T 4 6 (or C - G - D -
D t-B -A t-E t-F -A -B t-E -F It), and that this row is semicombinatorial at I5.

6Recommended analyses of Klavierstück, op. 33a are Eric Graebner. “An Analysis of Schoenberg's Kla­
vierstück, Op. 33a," Perspectives o f New Music 12 (1973-74): 128-140: George Perle, Serial Composition
and Atonality, 6th ed. (Berkeley and Los Angeles: University of California Press. 1991), pp. 111-116:
Joseph Straus, Introduction to Post-Tonal Theory, pp. 253-259; and Robert Morgan, Anthology o f
Twentieth-Century Music (New York: Norton, 1992). pp. 68-70. See also Milton Babbitt: Words about
Music, pp. 75-79. For an analysis of op. 33a's “sister piece,” Klavierstück, op. 33b, see Brian Alegant,
“Unveiling Schoenberg's op. 33b,” Music Theory Spectrum 18/2 (1996): 143-166.