03 Lewin 1998 - A Formal Theory of Generalized Tonal Functions
03 Lewin 1998 - A Formal Theory of Generalized Tonal Functions
03 Lewin 1998 - A Formal Theory of Generalized Tonal Functions
Duke University Press and Yale University Department of Music are collaborating with JSTOR to digitize,
preserve and extend access to Journal of Music Theory.
http://www.jstor.org
A FORMALTHEORYOF
GENERALIZED TONAL FUNCTIONS
DavidLewin
Our point of departureis a common method of constructinga system of tonal functions and relationshipsgiven a tonic pitch-classT, a
dominant intervald and a mediantintervalm. The method is portrayed
visuallyby Figure 1.
In the figure we see a C major system constructed given the tonic
pitch class T=C, the dominant interval d=fifth (modulo the octave),
and the mediantintervalm=major third (modulo the octave). From the
tonic note C we construct notes a dominant and a mediant interval
away, that is G=C+d and e=C+m. Togetherwith the tonic note, these
dominant and mediant notes constitute the tonic triad of the system.
Next we construct a dominant triad, comprisingthe dominant note G
together with the notes D=G+d and b=G+m which lie the intervalsd
and m respectivelyfrom that G. Similarly,we construct a subdominant
triad comprisingthe tonic note C, the subdominantnote F from which
C lies the interval d away, and the submediantnote a which lies the
interval m from the subdominantnote F. Since F is constructed to
satisfy the relation F+d=C, we may symbolicallywrite F=C-d; then
a=F+m=C-d+m.
The set-theoretic union of the three triads so far generatedcan be
called the "diatonic set" for the system, that is, the unordered set
23
G
d
FaCe
bD
Figure 1
24
some interval (each modulo the octave), then X+i is the pitch class
lying the intervali from X, and X lies the intervali from the pitch class
X-i. As the context may suggest,i can be imaginedto be measuredin
equally tempered semitones modulo 12, or in cents modulo 1200, or as
the logarithmof a just ratio modulo log2, and so forth.
1. By a Riemann System (RS) we shall understandan
DEFINITION
ordered triple (T,d,m), where T is a pitch class and d and m are
intervals,subject to the restrictionsthat d*O, m40, and m d.
The restrictions are necessary and sufficient to guaranteethat the
pitch classes T, T+m and T+d will be distinct, so that we can talk of a
"triad"without awkwardness.T will be called the "tonic pitch class"of
the system;d and m are its "dominantand mediantintervals."
DEFINITION
2. The tonic triad of the RS (T,d,m) is the unordered
set (T, T+m, T+d). The dominant triad of the system is (T+d, T+d+
m, T+2d). The subdominanttriadof the system is (T-d, T-d+m, T).
These triadsare the primarytriadsof the system.
One sees that the primarytriads are transposedforms, each of any
other.
3. The diatonic set of (T,d,m) is the unordered setDEFINITION
theoretic union of the primarytriads, comprisingthe variouspitch
classes T-d, T-d+m, T, T+m, T+d, T+d+m, and T+2d.
Given the restrictionsof Definition 1, there may be anywherefrom
3 to 7 distinct notes (pitch classes) in the diatonic set. For instance,if
d is exactly one-third of an octave, then T-d and T+2d will be the
same note, here representingtwo distinct functions: subdominantand
dominant-of-the-dominant.
4. The canonical listing for (the diatonic set of) the RS
DEFINITION
(T,d, m) is the orderedseries(T-d, T-d+m, T, T+m, T+d, T+d+m,
T+2d).
This ordered series will alwayshave exactly seven entries,even if the
same note be entered more than once. For example, if d is exactly half
an octave and m is exactly a quarterof an octave, then the canonical
listing for (C,d,m) can be written as F$aCebGbbbbDbb,with the convention that F# and Gb are the same pitch class, and so forth. This
series has seven entries, representingthe seven functions of the system,
but the unordereddiatonic set itself has only four distinct notes.
26
(T+m)
(T+d)
(T-d)
(T-d+m)
(T-d)
(T-d+m)
(T+m)
(T+d)
(T+d+m)
(T+2d)
(T+d)
(T+d+m)
(T+2d)
Figure2
Table 1
RS
canonicallisting
diatonic set
(C, 8, 4)
(0, 4, 8)
(C, 6, 3)
FbabCeG~bSt
F$aCebGbbb Dbb
(C,7, 2)
FgCdGaD
(0, 2, 5, 7, 9)
(C, 2, 4)
BbdCeDfgE
(10, 0, 2, 4, 6)
(C, 4, 7)
(11,0, 3, 4, 7, 8)
(C, 7, 10)
AbebCgEbaG
FebCbbGfD
(C, 2, 6)
BbeCfgDg#E
(0, 2, 4, 6, 8, 10)
(C, 4, 2)
AbbbCdEf#G$
(0, 2, 4, 6, 8, 10)
28
(0, 3, 6, 9)
(10, 0, 2, 3, 5, 7)
pitch-class set can be the diatonic set for more than one kind of Riemann System: for example, the whole-tone scale is generatedby two
essentiallydifferent systems on the Table.
A naturaltheoreticalquestion presentsitself: what are necessaryand
sufficient conditions on d and m, in order that the Riemann System
(T,d,m) should be irredundant?Theorem 1 answersthis question.
THEOREM
1. For the RS (T,d,m) to be irredundantit is necessary
and sufficient that conditions (1) and (2) following obtain.
(1) For N=1, 2 or 3, Nd is not zero. (That is, one, two or three
dominant intervalsdo not come out an exact number of octaves.
That is, d is not 0, 6, 4, or 8 equally temperedsemitones.)
(2) For N=O, ?1, ?2, or 3, m is not equal to Nd. (This condition
does not exclude the possibility that m may equal -3d.)
The theorem can be verified by a systematic inspection of various
intervallic relations among the various notes T-d, T-d+m, T, T+m,
T+d, T+d+m, and T+2d. For example, T-d and T+d+m are the same
note if and only if -d=d+m, which is the case if and only if m=-2d.
Table 2 below essentially lists all irredundant Riemann Systems in
twelve-tone equal temperament. The precise meaning of "essentially"
here will become clear duringsubsequentdiscussion.
The first two columns of Table 2 contain all combinationsof values
for d and m that satisfy the conditions of Theorem 1, up throughd=5.
For instance, d= 1 is allowed by condition (1) of the Theorem;for d= 1,
condition (2) disallows m=1, m=-1 (=11), m=2, m=-2 (= 10), and m=
3. The remainingpossible values for m, that is m=4,9,5,8,6 and 7,
are listed in the second column of the Table opposite d= 1 in the first
column. The order in which these values for m are listed will be discussed later. We shall see that it is sufficient to list on the Tablevalues
of d throughd=5 only: irredundantsystems with d=7,9,10 or 11 can
be derivedfrom systems with d=5,3,2, or 1 respectively.
The third column of Table 2 contains the names of the Riemann
Systems with tonic note C and with dominant and mediant intervals
correspondingto the entries in the first and second columns of the
table. The fourth column contains the canonical listings generatedby
the Riemann Systems entered in column three. The usualspellingconventions for enharmonicequivalenceare assumed.The last column contains the Forte-labelsfor the chord types of the diatonic sets at issue.3
Whereno labelappearsin column five of the Table,the last-writtenlabel
is understood to carry on down. For example, the type of the diatonic
set for (C,1,8) is 7-5; the type of the diatonic set for (C,3,10) is 7-31.
With two exceptions, each system on Table2 is pairedwith another
system, which we shall call its "conjugatesystem." The exceptions are
29
Table2
d
system
1
1
4
9
(C, 1,4)
(C, 1,9)
1
1
5
8
(C, 1,5)
(C, 1,8)
1
1
6
7
(C, 1,6)
(C, 1,7)
(C, 2, 1)
2
2
3
11
(C, 2, 3)
(C, 2,11)
2
2
5
9
(C, 2,5)
(C, 2, 9)
3
3
canonical
listing
B eb Ce
B g# Ca
Forte-labelof
diatonic set
C# f D
C$bb D
7-1
7-5
(C, 2, 7)
B e Cf C f#D
B g C g# C#a D
B f Cf$C# g D
D
B f # Cg C
gg
Bbb CcgD eb E
Bb c$ Ceb D f E
Bba Cb D c gE
Bb eb Cf D g E
Bbg Ca D b E
Bb f Cg D a E
1
2
(C, 3, 1)
(C,3,2)
A bb CdbhEbe
A b Cd Ebf
7-31
3
3
4
11
(C, 3,4)
(C, 3,11)
3
3
5
10
(C, 3, 5)
(C, 3, 10)
3
3
7
8
(C, 3, 7)
(C, 3, 8)
A c# Ce Ebg Ft
A g# Cb Ebd F$
A d Cf Ebg# F#
A g Cbb Eb c~ F#
A e Cg EbbbbF$
A f Cab Eb'b F#
5
5
1
4
(C, 5,1)
(C, 5,4)
5
5
6
11
(C, 5,6)
(C,5,11)
5
5
8
9
(C, 5,8)
(C, 5,9)
30
F#
F#
G g# Ccg F f# Bb
G b Ce F a BA
G c$ Cf#$F b Bb
G f Cb F e Bb
G eb Cab F dbBb
G e Ca F d Bb
7-7
7-1
7-2
7-23
7-35
7-14
7-7
7-35
(C,2,1) and (C,2,7): each of these systems is its own conjugate.To see
the salient mathematicalaspect of the conjugate relationship,notice
that (C, 1,4) and (C, 1,9) are paired, and that 4+9= 1. (C, 1,5) is paired
with (C,1,8), and 5 +8=1. (C,2,3) is pairedwith (C,2,11), and 3+11=2.
(C,3,7) is paired with (C,3,8), and 7+8= 3. (C,2,1) is self-conjugate,
and 1+1=2. In general, (T,d, m) and (T,d, m') are conjugatesystems if
m+m'=d. Some formaldefinitions are in order.
7. The conjugate system of the RS (T,d,m) is the RieDEFINITION
mann System (T,d,d-m). The operation that transformsa given RS
into its conjugate will be called "CONJ".Weshall write, symbolically, CONJ(T,d, m)=(T, d, d-m).
The conjugateoperation can be applied to any RiemannSystem, irredundantor redundant.Formally,it is necessaryto verify that (T,d,dm) is indeed always a "RiemannSystem" whenever(T,d,m) is. That is,
one must verify that dO, d-m/O and dkd-m, supposingthat dO,
m 0Oand d:m. This is easily done. In discussingsubsequenttransformations of Riemann Systems, we shall always understandsuch methodologicalniceties to have been carriedout.
Givena RiemannSystem(T,d, m), let us set m'=d-m. The conjugate
system articulatesthe dominantintervald, within each primarytriad, as
d=m' +m, ratherthan d=m+m'. See Figure3.
The relation between original and conjugate subdivisions of the
intervald thus generalizessuch traditionalnotions as arithmetic-versusharmonic division of d, and such traditionalrelationsas that of CeGto
CebG.4
Extending the format of Figure 3 to cover the complete canonical
listings of the systems underexamination,we obtain a relationshipportrayed visuallyby the uppertwo-thirdsof Figure4.
Let us now read the conjugate listing backwards.This produces a
listing for a "retrogradeconjugate" system. As the bottom part of
Figure 4 indicates, the latter system has dominant interval -d and
mediant interval-m. In this sense, it is an inversionof the given system.
In fact, the canonical listing for the RET CONJsystem is a serialinversion of the canonicallisting for the originalsystem. Figure4 makesthis
relation clear: the seven successivenotes of the originallisting spanthe
six successive intervals m, m ', m, m ', m, m ', while the seven successive
d
m
m%
ORIGINAL
CONJUGATE
*
e
Figure3
d
m
ORIG 0
CONJ e
,*
m
d
RET
CONJ-d
=TDINV
-d
Figure4
32
Let us now return to Table 2. As we readthe variouscanonicallistings in the fourth column of that table backwards,we shallthereby be
reading, en masse, the canonical listings for the TD-inversionsof the
variousSystems appearingin column 3 of the Table. For instance,the
F major Riemann System (F,7,4) has dominant note C; hence it is the
TD-inversionof the System (C,-7,-4)=(C,5,8). As the TD-inversion
of (C,5,8), (F,7,4) is also the retrograde-conjugate
of (C,5,8). So the
listing for (F,7,4) can be read, on Table 2, as the retrogradeof the listing for the conjugate system of (C,5,8), that is as the retrogradeof the
listing for (C,5,9).
Generalizingthis example, we can now see in what sense Table 2 lists
"essentially"all types of irredundantRiemann Systems in twelve-tone
equal temperament. If d=1,2,3 or 5 and the pair (d,m) satisfies the
conditions of Theorem 1, then the System (C,d,m) appearsdirectly on
Table 2. If d=7,9,10 or 11 and the pair (d,m) satisfies the conditions
of Theorem 1, then the pair (-d,-m) will also satisfy the conditions of
the Theorem, so that the System (C,-d,-m) will appearon the Table.
One can then read the canonical listing for a System with dominant
interval d and mediant intervalm by readingbackwardsthe canonical
listing for the conjugatesystem of (C,-d, -m) off the Table.
Since the retrogradeof the conjugate listing is a serial inversionof
a given listing, it follows that the conjugatelisting itself is a retrogradeinversionof the given listing. This property is easily seen and heard as
manifested among the listings of Table 2; it explains why the listings
pairedtogether on the Table sound so closely related.
Given a Riemann System and its conjugate, since the canonicallisteach of the other, it
ings of their diatonic sets are retrograde-inversions,
follows that the unordereddiatonic sets themselvesmust be pitch-class
inversions,each of the other. Hence, if the Systems are irredundant,
they will have the same Forte-numberentered in column 5 of Table 2.
One can also see why every irredundantRiemannSystem appearing
on the Table with dominantintervald=3 must have 7-31 as the Fortenumber of its diatonic set. In such a System, the notes T-d, T, T+d
and T+2d will be T-3, T, T+3 and T+6; these four notes will constitute
33
d
mn
-m
-m
ORIG
RET
-d
Figure5
CONJP
ORIG
d
mRET
-m
m
-d
-d
RET
TDINV
Figure6
35
Table 3
IDENT
TDINV
RET
CONJ
IDENT
IDENT
TDINV
RET
CONJ
TDINV
TDINV
IDENT
CONJ
RET
RET
RET
CONJ
IDENT
TDINV
CONJ
CONJ
RET
TDINV
IDENT
38
are concerned, any of the four Systems can be derivedfrom any other
one in absolutelyegalitarianfashion.
It is useful to have made this observationbefore going on to investigate specifically the four Riemann Systems in twelve-tone equal temperament that can be derived from the "C major" System (C,7,4) by
the operations of GSER: C major is equally a derived form of any of
the other three Systems,which are equally derivedforms of each other.
The preceding discussion makes it clear that, if we want to attribute
priority to C major in preference to the other three Systems, that is
only because we wish to attribute priority to the "fifth" and the
"majorthird"as structuralintervalsin this connection.s
The System (C,7,3) is the conjugate of C major. We shall call it "C
minor." As already noted, the conjugate relationship puts into the
terminology of the present paper such notions as the harmonic and
arithmeticdivisionof the fifth, or the inversion(renversement)of major
and minor thirdswithin the fifth.
The System (G,5,8) is the TD-inversionof C major. We shall call it
"dual G minor." Its tonic triad projects, from the tonic note G, the
dominant and mediant notes C=G+5 and eb=G+8, that is, the notes
respectivelya fifth and a majorthird "down"from the tonic. The structure will be familiarto studentsof Riemann'stheories;Riemannrecommended this triad as the "natural"form for minor tonality, though he
shrank rather illogically from calling CabF the "dominant triad" of
GebC in this context.6
The System (G,5,9) is the retrogradeof C major. We shall call it
"dual G major." Its tonic triad projects, from the tonic note G, the
dominant and mediant notes C=G+5 and e=G+9, that is, the notes
respectivelya fifth and a minor third "down"from the tonic. As far as
I know, no aspect of this system has hitherto been investigated,much
less recommended,by any theorist of tonality. The formalitiesof the
present context make that observationquite striking.If one admitsthe
musical cogency of both the relation "Cmin=CONJ(Cmaj)"and the
relation "dual G min= TDINV(Cmaj),"and if in addition one considers
minor tonality to be as basic as major, ratherthan strictly subordinate
to major,then one ought to admit as equally cogent relationslike "dual
Gmaj=CONJ(dualGmin)" or "dual Gmaj=TDINV(Cmin)."Dual G
major would thus naturally come into considerationand, having admitted two species of minor which retrogradeeach into the other, it
would be only logical to admit two species of major which retrograde
each into the other,always assumingthat one puts minor on a structural
parwith major.
Thus one can analyze the absence of dual G major from the literature of tonal theory in the following way. First, some theorists consider
ordinary major tonality, with d=7 and m=4, to have strong priority
40
Lentp
Example 1
Sehr IQngsQm
-O;
Ido
a
Example2
42
Z~
and G appear as fifth and major third below B. Certainly one does
violence to the sense of the passageif one tries to hear an E root for
the triad.
Example 2, taken from the opening of Wagner'sParsifal, shows a
similar state of affairs as regardsthe structureoutlined by the succession G4-C4-(D4)-Eb 4. Our attention here is on G4, not on C4. G4
was just heard as leading tone to Ab 4; now Ab 4 becomes an upper
neighborto G4 and in fact resolvesto G4,reversingthe earliertendencyrelationship.C4 andEb 4, recentlyheardas 3rd and 5th of an Ab major
triad, now are heard as 5th and 3rd underneathG4, in a dual G minor
triad. All of this is a textbook example of what RiemanncalledLeittonwechsel.
Example 3 is from the opening of Brahms'sIntermezzo Op. 119, no.
1. It almost seems that the composer has constructed an academic
exercise in the compositional projection not just of dual minor triads,
but further, of large serial segmentsfrom the entire canonicallisting of
the dual F# minor RiemannSystem. One can read these segmentsfrom
left to right on Figure7, progressingfrom top to bottom.
The analysismakes good sense because one orients oneself harmonically, in this context, by the tops and beginningsof the 3rd-chains,
more than by their (anticipated) bottoms and ends. The passagesuggests modulation to a related major System, and the piece as a whole
exploits such modulations. Given the analysis of Figure 7, the related
System is surelynot D major;ratherit is dualA major:the majortriads
at issue are presentedand developedin exactly the sameway as are the
minor triads. Figure 8 tries to analyze the whole passageas projecting
segments of the canonical listing for dual A major. It is this structure,
not ordinary D major, which is in contention with the structure of
Figure7.
The dual A major structure is not very powerful in the first four
lines of Figure 8; it becomes strong however in the fifth line of Figure
8, strongerthan the dual F? minor assertedby the fourth line of Figure 7 for the same music. The sixth line of Figure 8 and the fifth line
of Figure 7 are of equal power in contending to assert their respective
structures.The contention here is not only between tonic notes but
also, perhapsmore, between upper case and lower case for the namesof
all the notes at issue: C# or c ? a or A?, and so forth. That is, is a given
note hereaboutsa root-or-fifthof something(uppercase) or is it a third
of something (lower case)? This question, which would have delighted
Hauptmann,highlightsthe unique status of the note E: it is uppercase
on both Figures.
The E is thus uniquely releasedfrom the uppercase/lowercase tension, and in fact we find it usedto initiatecadentialmaterialimmediately
following the passageat issue, now that complete canonicallistings of
43
Acogio
Example3
F#
C#a
C#
F$a
a
g
E
Figure7
f#
E
A
G
c#
ff
cE
ff
E
44
Figure8
dualF# minor
...G
.. g
e C a F? d BgE c..
E cA ff Db G e C...
dualA major
Figure9
(subdom.(!))
dual G$ maj:
C$tmaj.
(subdom.)
Example4
AP
mai.
(subdom.)
Alla Danzo Tedesco
dual D maj.
AP
(subdom.)
(!)
Example5
46
A min: dom -*
tonic
Allegro vivace
=--
f rubato
m4frubto
A minor
etc.
dual Ab major
Example6
47
F# minor
C major
.fAb
c Eb g Bb d Fa
C e G b D f# A c# E g# B d$...
II
E minor
Bb major
Figure 10
49
not shifted forms of just C major,namely all just majorand minor Systems whose tonics cannot be derivedfromthe note C by addingand subtractingjust fifths orjust majorthirdsmodulothe octave.Whilesuch Systems are indeed transpositionsof just C major, or of just A minor, the
transpositionalrelationshipshave no functional significancein the context of RiemannSystematics.Note in particularthat the "just C minor"
System we can read off Figure 10 as just C majorshifted by -7 places is
not the conjugateSystem of just C major,since the lower case c and the
uppercaseCon Figure10 differ,injust intonation,by a syntonic comma.
On the other hand, if we considerFigure 10 in equal temperamentit
is clear that there will be only 24 distinct Riemann Systems whose
canonical listing are embedded in the extended listing; that listing will
repeat itself indefinitely at every 24th entry. In this case any minor
System involved will indeed be the conjugateof the major System involved that has the same tonic note, and vice-versa.However,even on
this small family of 24 RiemannSystems,SHIFT(-7) and CONJoperate
with different effect. That is, while C minor in this context is equally
SHIFT(-7)(Cmajor) and CONJ(Cmajor),it is not true that C majoris
both SHIFT(-7)(Cminor) and CONJ(Cminor). Rather, C major is
SHIFT(7) (Cminor);SHIFT(-7)(Cminor) is not Cmajorbut Cbmajor=
Bmajor. So the operationsSHIFT(-7) and CONJdo not have the same
effect on all 24 RiemannSystems at issue.
For another example, let us study the System (C,3,7) in equal temperament and inspect its shifted forms. One sees that the canonical
listing AeCgEbbbGb will extend on to the right as.. . dbAeCgEb....
The extended listing is thus a serial orderingof an octotonic scale-set
which repeats at every eighth entry ad infinitum. Hence there will be
only eight shifted forms of (C,3,7). (C,3, 7) itself will reappearas
SHIFT(8)(C,3,7). However its conjugate System (C,3,8) will not appear among its shifted forms. Odd-numberedshifts of (C,3,7) will indeed produce Systems with dominant and mediant intervals 3 and 8
respectively, but the availabletonic notes for those Systems will be
only E, G,Bb and Db, never C. The extended canonicallisting imposes
a certain symmetricalorderingon the octotonic scale, a feature which
might have interestinganalyticalor compositionalimplications.
Let us now returnto the basicformulaSHIFT(M)SHIFT(N)=SHIFT
(M+N), and to the special case SHIFT(N)SHIFT(-N)=SHIFT(O)=
IDENT. If M and N are both even, then M+N and -N will also be even.
We can conclude that the even shifts, combining among themselves,
form a group of operations. Mathematicianswould call this a "subgroup"of GSHIFT.
DEFINITION
14. The family of all operations SHIFT(N) such that
Nis even will be called the even shift group and denoted GEVSHIFT.
51
cmajor
'
~.
---'
- dualE minor
*-
(adjusting1.c.and u.c.)
"
-d
Figure 11
TDINV
0+)
eb
TMINV
MDINV
er
((p
(</>)
Figure 12
53
Thus C major and F$ major are of the same type, supposing the
sameintonation. In twelve-toneequal temperament,(C,2,5) and (Bb ,2,
5) are of the same type, namely the type (2,5). Also of type (2,5) is
(C+j,2,5), where j representsa just fifth. The concept of type enables
us to relate the intervallicstructureof (C+j,2,5) to the intervallicstructure of (C,2,5), without havingto assertany functional significancefor
the transpositionalrelationshipbetween them.
The concept of type is also useful to make generalizationsabout the
intervallic structuresof Systems that are functionally related, without
having to worry about their tonic notes. For example, given any RS
(T,d,m), its conjugate System will be of type (d,d-m), and so will its
odd-shifted Systems. In that sense, we can say that the odd-shiftedSystems are all "of conjugate type," even though the conjugate System
itself may not appearas a shifted System.
In similar vein, we can say that the TD-inverted,TM-invertedand
MD-invertedforms of (T,d,m) are all of one type, namely (-d,-m), the
"invertedtype" of (d,m). And we can say that any minor System is "of
retrogradetype" to any dual minor System, in the same spirit. To make
such discourse formally precise, we need only define the operationsof
conjugation,inversionand retrogressionon types.
19. Given a type (d, m), the conjugate type is conj
DEFINITION
(d,m)=(d,d-m). The inverted type is inv(d,m)=(-d,-m). The
retrograde type is ret(d,m)=(-d,m-d). The group comprising the
three operationsconj, inv and ret, along with the identity operation
ident, will be called the "serialtype-group"of operationson Systemtypes.
56
analyze the minor triad as comprisinga fifth "up" and a major third
"up" to a commongeneratee(for example, C to G and eb to G).
Systems involvingcommon-generateerelationships,whether in connection with minor triads or in other ways, can be called "phonic"as
opposed to "sonic."8 A formalismmore general yet than that of the
present paper could perhapsbe developed to generalizephonic, as well
as sonic, intervallic systems, and to interrelate all systems, phonic as
well as sonic, amongthemselvesandeachother. However,the formalities
of the present paper are, I imagine, amply general to satisfy most
readersfor the nonce.
4b
58
NOTES
1. A bibliography and critique of writings by and pertinent to Riemann, convenient for American readers, is to be found in William C. Mickelsen, Hugo Riemann's Theory of Harmony (Lincoln and London: University of Nebraska
Press, 1977). Among the variety of works providing foundations for the approach
taken by Riemann and continued in the formalities of the present paper, one
should particularly draw attention to the work of Jean-Philippe Rameau in his
59
60