The Hexachordal Theorem: A Mathematical Look at Interval Relations in Twelve-Tone

Composition
Author(s): Steven K. Blau
Reviewed work(s):
Source: Mathematics Magazine, Vol. 72, No. 4 (Oct., 1999), pp. 310-313
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2691227 .
Accessed: 28/02/2013 20:50
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

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

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to
Mathematics Magazine.

http://www.jstor.org

This content downloaded on Thu, 28 Feb 2013 20:50:19 PM

All use subject to JSTOR Terms and Conditions

310

MATHEMATICS MAGAZINE

The HexachordalTheorem:
A Mathematical
Lookat Interval
Composition
RelationsinTwelve-Tone
STEVEN K. BLAU

PhysicsDepartment
RiponCollege
Ripon,WI 54971

Introduction Take the numbers1 through12 and dividetheminto two complementarysix-member

subsets A and B. Each subsethas (6)= 15 pairsof members;
betweenthem,
the intervalbetweenanypair of numbersis the (positive)difference
subjectto the equivalencen - 12 - n. So, forthe pair(1, 11) the intervalmaybe said
to be either10 or 2. A convenient
pair of interval
wayto representa complementary
sets,whichmakes visuallyapparentthe equivalencejust described,is to presenta
"clockface" withsixA's and sixB's, as follows:

B
A

A
B

In the case shown, A = 1,4, 5, 8, 11,12} and B = (2, 3,6, 7,9, 10}. The intervalbetweenthe A's at one o'clockand eleveno'clockis either10 (countingclockwise)or 2
(countinganticlockwise).In what followsI will use the equivalence relationto
describeall intervalsas beingbetween1 and 6.
The intervalmultisetassociatedwith a six-memberset is the collectionof 15
intervalsdeterminedby all possiblepairsdrawnfromthe set. For the set A shown,
the intervalmultisetis (1, 1, 1,2, 3,3, 3,4,4,4, 4,5,5, 5, 6}. As one can checkdirectly,
the set B has the same intervalmultiset.This is no accident-the equalityof the
intervalmultisets
is the contentof the hexachordaltheorem.
Let thenumbers1 through12 be partitionedintoany two complementary sets A and B, each with six elements.Then A and B have identicalinterval
multisets.
THEOREM.

Beforeprovingthe theorem,we considerits musicalmeaning.

This content downloaded on Thu, 28 Feb 2013 20:50:19 PM

All use subject to JSTOR Terms and Conditions

VOL. 72, NO. 4, OCTOBER 1999

311

The musical context The hexachordaltheoremwas empiricallydiscoveredby

composersworkingwith Arnold Sch6nberg'stwelve-tonemethod. In the musical
context,the numbers1 through12 representthe twelvenotesof the chromaticscale.
Twelve-tonecomposersuse a twelvenote "row" containingall the membersof the
chromaticscale, arrangedas theysee fit,as a basis fortheirmelodiccomposition.
They oftenthinkof the firstand second halves of the row as two complementary
hexachords.In the musicalcontext,"complementai-y"
meansthata note containedin
setting,compleone hexachordis not presentin its companion.In the mathematical
subsetsof (1, 2,,..., 12}.
six-member
hexachordsare complementary
mentary
What I called an intervalin the theoremwould also be called an intervalby
composers:it representsthe numberof semitones(notes on a keyboard)needed to
connectone note to another.Thus the hexachordaltheoremis a statementaboutthe
row.
halvesof a twelve-tone
of the two complementary
intervalstructures
The hexachordaltheoremwas firstprovedby MiltonBabbitt,a celebratedcomand David Lewin, then a graduatestudentin
poser witha degree in mathematics,
In describing
his proofwithLewin,Babbittwrote,"We used topological
mathematics.
methods.We hitthislittleproblemwithall kindsofheavyhammers,and we solvedit"
proofs
[1]. Later,Lewin (workingon his own) and Ralph Fox constructeddifferent
methods.Lewin describedhis workon the hexachordaland
using group-theoretic
relatedtheoremsin the JournalofMusic Theory(see [2] and [3]). He did notinclude
a proofof the hexachordaltheorem,but sketchedhis work on relatedtheorems,
reasoningbywhichI arrived
observingafterone of his proofs[2]: "The mathematical
at this resultis not communicableto a reader who does not have considerable
mathematical
training."The proofof the hexachordaltheoremthatfollowsrequires
no advancedmathematics;
it can be followedby musicianswithlittlemathematical
and it mayinterestmathematicians.
training,
ofwesternmusicup to 1900, I wouldoffer:
history
Challengedto givea four-word
"Modulationsbecame morefrequent."By the late 1800s in theworksofWagner,for
example,thereare sectionswherethe modulationscome so quicklythattonalitiesare
establishedonly for seconds beforetheychange. Beethoven,by contrast,typically
allowedtonalitiesto be establishedfora muchgreatertimeperiodbeforemodulating
to a new tonality.Earlyin his career,ArnoldSchonbergwrotemusicfollowingthe
Wagnerianline,but laterdecided thatsuch musiccontaineda sortof inconsistency.
The momentumof musiccreatedsince Bach was pointedtowardan equalityof the
composersstillfelttied to
notes of the chromaticscale, but earlytwentieth-century
of earliercenturies.Schonbergadvocatedan "emancipationof
the tonalconventions
dissonance," abandoningthe notion that music must be conceived in terms of
tonalities.A corollaiywas the freedomto use all twelvenotesof the chromaticscale
equallyin composing,thoughSch6nbergdid occasionallywritetonalmusicthroughout his career.
The earlypieces of Schonbergand his "Second VienneseSchool" were almostall
so greatthat
The freedomenjoyedby composerswas apparently
shortand aphoristic.
to writeon a grandscale music that maintainedits internallogic.
it was difficult
Sch6nbergsoughta structurethatwould allow atonalcomposersto create coherent
workson a largescale. His solutionwas the "methodof composingwithtwelvetones
the twelve
relatedonlyto one another."One beginswitha scale, or row,containing
notesof the chromaticscale arrangedin some fixedorder.Then one manipulatesthe
rowin prescribedwaysin composing.Partof the composer'sskillis knowingwhento
break Sch6nberg'srules,but these rules do offercohesionto atonal compositions.
Some twelve-tonecomposersfeltthat theirworkwould be furtherunifiedif the
rowswere identical.
intervalmultisetsimpliedby varioussubsetsof theirtwelve-tone

This content downloaded on Thu, 28 Feb 2013 20:50:19 PM

All use subject to JSTOR Terms and Conditions

312

MATHEMATICS MAGAZINE

Consequently,
theyempirically
discoveredthe hexachordaltheorem,as an offshoot
of
theirsearchforunification.
Babbittand Lewin,and thenothers,provedthe empirical
resultmathematically.
Proof of the theorem Given any pair of complementary
hexachordsA and B,
an
displayedas a clockface,one can generatea new complementary
pairbyswitching
A with a neighboringB. Indeed, any complementary
pair of hexachordscan be
generated,througha sequence of switches,fromthe pairingwithA's in positions1-6
and B's in positions7-12. This special pairingclearlysatisfiesthe conclusionof the
hexachordaltheorem.Therefore,the hexachordaltheoremis a consequence of the
lemma.
following
LEMMA. Suppose that two complementary
hexachordsA and B have identical
A
intervalmultisets.
Thenthehexachordsobtainedby switching
a pair ofneighboring
and B elementsalso have idlentical
intervalmultisets.

Proof.FIGURE 1 shows,at left,the clock face introducedearlierto representa

the underlinedentries(an A at eighto'clock
particularpair of hexachords.Switching
and a B at seveno'clock)producesthe the clockface on the right:

B
A

B
A

Afterthe swvitch

Beforethe swvitch
FIGURE 1

Switching
an adjacentpair.

Afteridentifying
an adjacent pair of A and B elementsto be switched,one may
partition
the remaining10 elementsinto5 pairs:foreach integern from1 to 5, we
considerthe pair of elementslyingn hoursto eitherside of the switchedA and B.
ofeach suchpairs:twoA's,twoB's, or
Thereare fourpossibilities
forthemembership
one of each (in eitherorder). All fourpossibilitiesare illustratedin FIGURE 2 (the
valuesof n correspondto the exampleillustrated
above):
particular
B

B
fl=3

B
A

2=4

=1
FIGURE 2

sideoftheswitched
A andB.
Pairsofelements
lyingn hourstoeither

This content downloaded on Thu, 28 Feb 2013 20:50:19 PM

All use subject to JSTOR Terms and Conditions

VOL. 72, NO. 4, OCTOBER 1999

3 13

By considering
the fivepairsdefinedas above,one considersall the intervals
possibly
affectedby the switch.
We considerfourcases. If a paircontainstwoA's (as illustrated
forn = 3), thenthe
switchchangestwointervals
in the A hexachord:twooriginalintervals,
withlengthsn
and n + 1, become twonew intervals,
withlengthsn + 1 and n, respectively.
Hence
the switchhas no neteffecton the A hexachordmultiset(or on the B multiset,
since
the pair involvesno B's). A similarargumentshowsthatthe switchhas no effectif
bothelementsin a pair are B's (as illustrated
forn = 1).
If theelementofthepairnearestto theswitchedA is an A, and theotherelementa
B (as shownin FIGURE 2 forn = 4), thentheswitchdoes altertheintervalmultisets
for
the A and B hexachords,
but in identicalways-in each multiset,
an originalinterval
of size n becomes a new intervalof size n + 1. The remainingcase (illustratedfor
n = 2) is similar.
Thus, forall n, the switching
operationhas the same effect,if any,on boththe A
and the B intervalmultisets,
and the proofis complete.
A generalization Nothingin the proofabove reliedon the (musical)factsthatthe
chromaticscale has 12 notes and thatcomposersconsidereddividingthe chromatic
scale intotwo hexachordsof equal length.The following
theoremcan be provedby
formby Lewin [3] forthe case
the same method,and was expressedin a different
N= 12.
THEOREM. Let the set (1, 2, .., N} be partitioned
intodisjointsetsA and B, with
size a and N - a, respectively.
Defineintervalmiultisets
for the A chordand the B
chord as was donefor th>ehexachordaltheorem.Let A(i) be the numberof i's in
the A intervalmultiset,and similarly
for B(i). Considerthe special partitionAo =
Then, for all i, A(i)-B(i)=
(1,2,3...,a},
Bo={a+1,a+2,a+3,...,N}.

AO(i)

BO(i).

Acknowledgment. It is a pleasure to acknowledgeKarl Beres, Kurt Dietrich,Norm Loomer, Donald

read thispaper and offeredmanyhelpfulsuggestions.
Passman,and RaymondStahura,wlhocarefully

REFERENCES
1. MiltonBabbitt,in WordsAbo2lt Mutsic,ed. by Dembski,Stephenand Straus,JosephN., University
of
WisconsinPress,Madison,WI, 1986,p. 105.
2. David Lewrin,
Inteivallicrelationsbetweentwvo
collectionsof notes,Journalof MlisicTheornJ3 (1959),
pp. 298-301.
3. David Lewin,The intervallic
contentof a collectionofnotes,inteivallicrelationsbetweena collectionof
notesand itscomplement:an applicationto Schoenberg'shexachordal
pieces,Jourtial
ofMusic Theory4
(1960), pp. 98-101.

This content downloaded on Thu, 28 Feb 2013 20:50:19 PM

All use subject to JSTOR Terms and Conditions