1874.] Mr. A. J. Ellis on Musical Duodenes.

II. Semidiurnal Tide,

Mean Lunitidal Interval (observed),
II. W. L. W.
-0 h 26 m -7. 6 h l m -l.

III. "On Musical Duodenes, or the Theory of Constructing

Instruments with Fixed Tones in Just or Practically Just
Intonation." By Alexander J. Ellis, F.R.S., E.S.A.,
F.C.P.S., F.C.P. Received October 28, 1874.
This paper is intended to complete and supplement three papers on
Music which I have already read before the Royal Society It contains
1
.

a more complete theory of- temperament, embracing that indicated by

Helmholtz 2 but not worked out by him, and its application to the theory
,

of constructing musical instruments with an intonation practically just,

without change of fingering, and, if there are three or four performers,
without change of mechanism. The name Duodene refers to that collec-
tion of twelve notes, suitable to the present manuals, which is made the
unit of construction. To obtain its precise form, and determine the
number and value of all such duodenes as it is necessary to tune, I have
been obliged to indicate a theory of harmonic scales and modulation,
which I believe to be entirely new, and which has of course other uses.
The great extent of the subject obliges me to confine this part of my
paper to a mere indication.

A. Notation of Pitch.
The letters C, D, E, F, Gr, A, B indicate both musical tones and the
number of vibrations made by the prime or lowest partial tone of each in
a second ; so that, C being known,

D = £C, E=f C, F=-JC, G=f C, A = |C,B = MC.

The marks # t + IT Jt are used
t> for fractional multipliers, having the
followingnames and values ;

» 135 „ * 128
,
sharp, jf = jgg ;
,

flat, Q =
I35'

hi S n
>
+ = 80 ;
low + ' = §F
skhismic, II = 32768 ; hyposkhismic, J^
= ggyjg'
1 "
On the Conditions, Extent, and Realization of a Perfect Musical Scale on Instru-
ments with Fixed Tones," read Jan. 21, 1864, printed at length in Proceedings,
vol. xiii. p. 93 " On the Physical Constitution and Relations of Musical Chords," and,
;

lastly, " On the Temperament of Musical Instruments with Fixed Tones," both read
on June 16, 1864, and printed at length in the Proceedings, vol. xiii. p. 392 and
p. 404.
2
Tonempfindungen, 3rd ed. p. 495.

b2
4 Mr, A. J. Ellis on Musical Duodenes. [Nov. 19,

the first two being written and read after, and the last four being written
and read before, the letter which marks the note. Thus : —
fAb = high A flat =^ *
| •
^ . C= g 0,

£(x# = low G sharp = gj •

g
* 12I
#
C = y|C.

The precise pitch of every tone is therefore indicated by its symbol, when
C is known.
When Italic letters are used, O i9
(7, c, c
(

, c", <fcc. indicate Octaves, and

O is the pitch of the lowest note of the violoncello. When Roman
letters are used, no such relative values are attributed to large and small
letters.

B. Temperament.

All intervals here considered can be made up of Fifths, major Thirds,

and Octaves, taken up or down. In other words, the ratio of the vibra-
n
tion-numbers of any two tones can be represented by (~j • (~Y • 2P ,

where m, n, and p are zero, or some positive or negative integers. In

Table I. the ratios of the vibrational numbers of all the tones in the same
P
horizontal line is (g) * 2 , and in the same vertical column is
(gj
#
%P ,
thep or Octave being left indeterminate. If, then, we proceed from any
note supposed to be the lower, column containingfirst horizontally to the
a second note, supposed to be higher and in the same Octave, and then

vertically to the second note itself, and multiply or divide by | for each
horizontal step, according as it is to the right or left, and multiply or
3
divide by g for each vertical step, according as it is upwards or down-
wards, and finally multiply or divide the result by 2, until the final result
lies between 1 and 2, that fraction will be the ratio of the higher vibra-

tional number to the lower. Thus C to F# gives C to E, 7 ; E to B,

3 2 45
3 5 /3\
= jg*
g; B to FJf, g ; and 4 •
( g] which, being greater than 2, on being
AK
divided by 2 gives the correct ratio of the Tritone C to E# as 05. An
inspection of the Table then shows that £D to jGfl, f Grb to f C, &c. are
the same intervals as C to Fjf, because they have the same relative posi-
tion, and this is sometimes very convenient.
But as 5 are primes, no one tone in Table I., supposing it to be
2, 3,
indefinitely extended, will have the same pitch as any other tone. The
object of Temperament generally is to obviate this inconvenience, by
slightly altering the ratios of the Eifth or major Third, or both. From
doing so, as Helmholtz has proved, more or less dissonance must result.
1874.] Mr. A. J. Ellis on Musical Duodenes. 5

Another object, therefore, is to make that dissonance as little annoying

as possible.
We find immediately by actual multiplication,

*^' the comma of Pythagoras, (1)

GO *
(2) Hii&iaag—
3

G)
-
2 = m = +t ^' the diesis (2)

Multiplying these equations together, and extracting the cube root,

(2)
#
5 *
(2)
= 80
= +' tlie comma ( of Didymus) (3)

Dividing (1) by (3),

/3\ 8

'
5
4 '
/1\
5
32805 ff
= 327G8 ==1[
,,
the u .

skmsma (
,..
4)
(2) (2J
'

On these equations depend all uniform temperaments in which every

Fifth and major Third preserves the same ratio throughout.
Let V, T, 7c, s be any four fractions having relations similar to f-, f-,

t and 1T respectively in (1) and (2) ; then

V la
-7-2
7
=fo, and 2-rT 3 = 7c 2 ^s (5, G)

Subtracting the logarithms of (1) and (2) from the logarithms of (5)
and (6) respectively, we find

log V=log g-jg • (log t-log h + \og % -log s), (7)

log T=log J +? • (log t-log 7,)- J

(log f -log s), (8)

which are the fundamental equations of temperament, and are identi- all

ties, of course, for just intonation. contain 4 unknown expres- As they

sions, two may be assumed and the rest found, giving rise to an endless
variety of temperaments. Without discussing these generally, the fol-
lowing cases should be mentioned :

Commatic System (for which 7j=1). This is the only system discussed —
in my previous paper, where 50 cases were considered.
1. Quintal or Pythagorean Temperament. Assume 1c = 1, and V=!,»
then by (7),
log t + log 1T=log*;
whence by (8),

logT=log| + logt.
This temperament proves to be thoroughly unsuitable for harmony.

2. Tertian, Mesotonic or Mean Temperament. Assume h = 1 and

T=4thenby(8),
2 log t=log H— log*;
6 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

whence by (7),

log V=log ^— i log t.

In my previous paper I have shown this to be the best suited for

harmony of all in this system but it requires 27 tones to the Octave.
;

3. Hemitonic or Equal Temperament. Assume 7c = 1, and also s = l,

then by (7),
O 1

log V = log ^-^ • (log t + log f ),

by (8),

log T=log;? + i .
(2 log t-log f ).

Since log t=0-005 3950 319 (whence -j log t= 0-000 4904 574) and
log 1T = 0-000 4901 071, we may, for all acoustical purposes, assume
log ir=if log t, and hence that

log V=log g_ n . log t=log —log T,

log T=log J + n ^g t=log Jh-7

• log %
This is the only uniform temperament whicli requires no more thai i

12 tones, and hence, although very ill suited to harmony, it has been
theoretically adopted by almost all musicians.

Slchismatic System (for which s=l).

4. Slchismic or Arabic, according to Helmholtz's indication {op. cit.

p. 441). Assume .9=1 and ¥='.,; then by (7),

log &=log t + log f,

and by (8),

log T = log |
-log IT.

The Arabs tune 17 tonesto form perfect Fifths, and thus obtain W
S major Thirds which are only one skhisma too flat. Put they do not
•ise harmony in our sense of the word, although this would give almost

j'lst intonation in certain keys.

5. Skhistic, or Ilelmholtzian, as it may be called, because suggested by

Helmholtz {op. cit. p. 495). Assume 5=1 and T=^ ; then by (8),

2 log lc =2 log f -log %

and by (7),

logV= log ^-g log f.

The name sl-histic is derived from dhist, or by abbreviation a, which is

1874] Mr. A. J. Ellis on Musical Duodenes,

my name for %/%. Since log f =88 log o- very nearly, one comma may
be said to contain 88 skhists. Skhistic temperament is indistinguishable
from just intonation, and I shall use it in the theory of constructing
instruments. If in Table I. we suppose the horizontal intervals to be still

5 3 81
V, but the vertical intervals to be s-, and the sign t to stand for oq Vjfc>

while tj=l as before, then this Table represents skhistic relations ; so that
iffrom any note, as E|? (col. 5, line x), we proceed by 8 skhistic Fifths
up to tB and then one major Third to the right, we find a tone D#,
which, when reduced to the same Octave, is identical with E|?. We thus
find a number of skhistic synonyms shown in Tables II. & III.
We may express the errors in the temperaments just discussed in terms
of skhists thus, using $56o- to mean " too sharp by 56 skhists," and so
on, and to mean " no error ":

Just. 1. Quintal. 2. Tertian. 3. Equal. 4. Skhismie. 5. Skhistic.

Minor Third t>88<7 \)22<r t>64<r #8(7 \>i*

Major Third #88^ J56c b8<r
Fourth #22,7 # 8<r .
#1»
Fifth \>22tr ? 8(7 J>l«r
Minor Sixth t>88<7 ?56<7 $8(7
Major Sixth #88<r *22<x #64o- b8* #1*

The error of one skhist is quite inappreciable by the most practised

ears in melody, and can be detected harmonically only by very slow beats
for tones in the highest Octaves used in music.

6. Cyclic Temperaments, The following method is far more general

than that given in my previous paper (Proc. vol. xiii. p. 412). Put
m . log Y=v . log 2, m . log Jc~q . log 2,
m . log T==t . log 2, m . log s =z . log 2,

and, after substituting these values for log V, log T, log 7c, log s in the
logarithms of equations (5) and (6), divide out by log 2, and multiply up
by m. Then
12v-~7m = q+z, m—3t = 2q~z (9,10)

Take any integral values for q and z, and find the integral values which
satisfy one of these indeterminate equations for v, m, or t, m, and substi-
tute in the other, taking the resulting integral values of t or v respec-
tively. The fiye integral values determine a
which the Octave is cycle in
divided into m aliquot parts, which may be termed octs, v of which make
a Fifth, t a major Third, q a comma, and z a skhisma of " the cycle of m."
Most of the results are valueless, but the following present either theo-
retical convenience or historical interest :
8 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

Cycle of Fifth, Major Third, Comma, Skhisma,

No.
m. v. t. ?• z.

1 30103 17609 9691 539 48

2 3010 1761 969 55 7
3 301 176 97 5
4 53 31 17 1
5 53 31 18 1
6 31 18 10 -1
7 12 7 4

Of these, the first three are here, I believe, for the first time shown to
be true cyclic temperaments.
1. The cycle of 30103 is such an excellent representative of just
intonation (giving even 6 octs for the skhist), that it can be used without
sensible error, in place of ordinary logarithms, to reduce the relations of
intervals to addition and subtraction, for general use among musicians
or learners unacquainted with higher arithmetic. By dividing out by
100,000 we obtain almost precisely the five-figure logarithms of the
intervals.
2. The 3010 is almost as correct, with smaller numbers.
cycle of
3. The 301 is almost a perfect representation of skbistie
cycle of
temperament, in which the skhisma is eliminated, and for that reason
becomes perhaps the most practical representation of general musical
intonation.
4. The first cycle of 53 is Nicholas Mercator's representation of just
intonation, but it is more correctly a representation of skhistic tempera-
ment, and not so good as JNo. 3\
5. The second cycle of 53 is a very accurate representation of Py-

thagorean intonation, and has actually been proposed for the violin by
Drobisch.
6. The cycle of 31 is Huyghens's Cyclus Harmonicus, and closely repre-
sents the tertian or mean temperament.
7. The cycle of 12 is the ordinary equal temperament, and its principal
convenience consists in the very small number of its octs, here called
Semitones.
Unequal Temperaments, whether they consist of 12 selected tones from
uniform temperaments, or of 12 tones turned intentionally false (see my
former paper, Proc. vol. xiii. pp. 414-417, for their theory), are now aban-
doned. But the difficulty of tuning equal temperament by estimation of
ear, or even by the monochord, and of retaining the intonation of the
piano or organ unchanged for even an hour, makes all temperaments in
actual use really unequal. The difficulty of original tuning by estimation
* When paper was read, I mentioned that this was the cycle used by Mr.
this
J3osanquet in his paper read before the Eoyal Society on 30th January, 1873.
1874.] Mr. A. J. Ellis on Musical Duodenes. 9

of ear in the case of skhistic temperament, where the Fifths have to be

flattened by an almost inaudible skhist, is so much enhanced as to be
1
insuperable except by Scheibler's method Hence it is necessary to find
.

a practical substitute. This I term

Unequally Just Intonation. —
Suppose that the 48 tones marked off by
a dotted line in Table I. have to be tuned in this substitute for skhistic
intonation. Tune to the fork. Take 4 just Fifths up (or Fourths
down), C to Gr, Gr to D, D to f A, t A to f E, without beats and three just ;

Fifths down (or Fourths up), to F, F to B\>, B\) to E[?. Then tune C
to E as a just major Third up, without beats, and from E proceed to its
just Fifth B, verifying the result by determining that it is a just major
Third above Gr, and so on to Gr$ up and JG down. Then if from Gr# we
proceeded to the just Fifth, DJf, the resulting tone would be exactly one
skhisma sharper than E[}, whereas in skhistic intonation it would be
identical with E|?, as already shown. It is needless to say that no tuner
could effect this exact difference of a skhisma, but he will come practically
near it, and the error is that of the Fifth (the least of the errors) in equal
temperament. If we were to proceed in this way for all the six columns
of 8 tones marked off in Table I., we should have just major Thirds
throughout, and just Fifths also in all but 5 cases— namely, jDJf to JB,
B# to jGr, Gr# to E|?, f E to Op, and f to f A\)\>, each of which would be
too flat by one skhisma. Since JGr (Table I. col. 6, linea?) is a just major
Third below £B, and a just minor Third above JE, but JE is a whole
skhisma flatter than JDJJf (col. 8, line p), which would be played for JE,
it follows that the minor Third, £D$ J to £Gr, would be a skhisma too flat

or close and similarly that the minor Thirds, BJ to Ef>, GJ to C\), and
;

f E to f AJp/p, or 4 minor Thirds on the whole, would be a skhisma too

2
flat Hence this style of tuning gives 5 Fifths and 4 minor Thirds, as
.

1 Calculate the logarithms of the ratios of all the skhistic tones by perpetual addition
or subtraction of 0*1760300 (=log § — log <r) to or from 0, continually adding or sub-
tracting 0*3010300 (=log 2) tomake the results positive and lie between this and 0.
Add the logarithm of the vibrational number of C, and then find the numbers (to three
places of decimals) corresponding to these logarithms. This gives the vibrational
numbers of all the skhistic tones in the Octave, of which 48 will be required. Subtract
4 from each of these values, and procure tuning-forks giving exactly the tones thus
determined to at least the hundredth of a vibration in a second. These may be obtained
of the great manufacturer of acoustic apparatus, Mons. R. Koenig, of Paris but it is ;

necessary to state that the English and German (not the French) system of counting
vibrations is to be used. Then tune each tone roughly to the corresponding fork, and
afterwards sharpen it until it beats 4 times in a second with the fork. By this means,
and by this means only, with great care and attention, the pitch may probably be ob-
tained with sufficient accuracy to distinguish skhistic from just intonation. And simi-
larly for equal and tertian intonation.
2
Taking the cycle of 30103, }G contains 17070, +B 26761, and fDU 9200 octs.
Hence the Fifth, £D## to |B, has only 17561 in place of 17609 octs, and the minor
Third, }D#$ to }G, only 7870, in place of 7918 octs— that is, in each case 48 octs too
little ; that is, these intervals are one skhisma too flat.
10 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

bad as the best intervals (Fifths) in equal temperament, and all the other
intervals absolutely just. Hence the name unequally just. In future I shall
consider that skhistic intonation is practically realized by unequally just
intonation, for which the practical rule is : tune six tones malcing just
major Thirds without beats (as fF]p to t A\> to C to E to JG# to JB$, line t
in Table I.), and from each of them tune seven other tones malcing just Fifths
(as C to G to D to tA to tE, and C to E to B\) to Ej?, in col. 5 of
Table I.).

JSaunders's "Tilting Action" — Before proceeding to show that 48 skhistic

tones suffice for modern modulational music, it will be useful for future
constructions to remark that a method of realizing all the effects which I
contemplated by my duplex finger-board (Proc. vol. xiii. p. 422) has been
invented by Mr. T. "W. Saunders \ by means of stops, which allow the
manual and fingering to remain unaltered. Two sets of harmonium
vibrators are arranged one behind the other, tuned in tertian intonation
(by means of beats, counted by a pendulum, which gives a fairly accurate
means of approximating to the correct result) as follows, the capital
letters referring to the white or long and the small letters to the
digitals,

black or short digitals, a mode of distinction which I shall constantly

employ :

Bach B# db C## e[> E[> E# g\> E## a^ G## b[> q>

Front ...... C 4 D df E F ff G g# A a# B
2
There are 12 stops one corresponding to each digital in the Octave,
,

which, by a " tilting action," enables one, and one only, out of the two
vibrators in. the same column, as shown above, to be " damped" at plea-
sure throughout all the Octaves of the instrument. When all the stops
are pushed in, the front vibrators only are free, and any one may be
exchanged for a back vibrator by pulling out its stop. Hence 24 out of
27 tones are under the command of the player ; ~B\)\), E[>|>, A \)\) are
omitted.
0. Harmonic Scales.

A series of tones, each of which is consonant with two other tones in

the same series that are themselves consonant with each other, forms
what I here mean by an harmonic scale. This was not the principle on
which scales were originally formed but this is the way in which the pitch
;

of the tones must be determined for the just intonation of modern har-
mony and modulation.
1
As Mr. Saunders has not patented his invention, I am unable to give more than
the indications in the text, and refer to him personally, at E. Lachenal's Concertina

Manufactory, 4 Little James Street, Bedford Row, W.C. His invention offers great
the construction of experimental instruments in any uniform or just
facilities for

intonation. His harmonium was shown when this paper was read.
2
In the specimen shown there are only 9 stops, the C%%, F##, G## having been
omitted but as the principle admits of the construction of 12 stops as easily as 9, the
;

complete form is mentioned in the text.

1874.] Mr. A. J. Ellis on Musical Duodenes. 11

The Harmonic Elements are the Fifth, Cx Gr ( = 39), the major Third,

C +E (
= t) and the minor Third, C — f E^> ( =g)
> , using a notation which

I have found practically very convenient for representing the intervals

between two tones ; the symbols x -f — are not to be employed with
, 5

any other meaning between the names of tones. In these elements it is

supposed that either note may be raised or depressed by any number of
Octaves, or be accompanied by such Octaves of itself.

The Harmonic Unit of Concord, consists of a major triad,

Cell, or
C+E — G,
and a minor triad, C— tE|? + Gr, arranged as in
fE[> G- the margin, and having the same Eirst 0, and hence the
E same Fifth Gr. The Fifth x Gr is placed vertically, the
two major thirds, C+E and fE|? + Gr, are horizontal, and
the two minor Thirds, 0— fE|> and E— Gr, slope obliquely from the
bottom upwards to the left. These positions, then, replace the symbols
X + —
,
, Allowing any one of the tones to be altered by any number
.

of Octaves, or to have Octaves of itself added, and any tone to be taken

as the First, this cell, whence all harmony is developed, contains every
chord recognized by musicians as a concord in Tertian Harmony that is, —
harmony depending on Octaves, Fifths, and Thirds alone, excluding
natural Sevenths (=j)i which form Septimal Harmony. By Table I.

cells can be readily constructed on any tone as a First.

The Harmonic Heptad, or Unit of Chord-relationship, consists of two
cells, the First of one being the Fifth of the other, as
tE|> Gr in the margin. Allowing Octave variations as before, this
tA|> C E contains all the three major and three minor triads which
F A have as one of their constituents, and are thus related
in the first degree. Two of these chords, the minor triad
A — C + E on the right or major side, and the major triad fAj^ + C— fE|?
on the left or minor side, connecting the two
and due to their cells

union, may be called union triads, to distinguish them from the four cell
triads. The heptad also contains all con-dissonant triads (as I term them),
consisting of three tones, two of which are consonant with C but dissonant
with each other. Of these the trine f A\) + C + E, which forms the central
horizontal line, is most important for future work.

The Harmonic Decad, or Unit of Harmony, consists of two heptads

having a common cell, and hence of three cells, the Fifth
tB[> D of the first, lowest, or suhdominant cell, and the Fifth of the
•f~E\) Gr B second, middle, or tonic cell, being the First of the second
f A\) C E cell, and First of the third, highest, or dominant cell,

F A respectively. The decad contains three major and three

minor cell triads, and two major and two minor nnion triads
—that is, ten triads in all, together with all the discords possible without
12 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

modulation. The First of the tonic cell is called the tonic of the decad,
and gives its name to it. The example, therefore, is a C decad.
Harmonic Trichordals consist of three triads, one from each cell in a
decad, and form eight groups. Contracting major triad and minor triad
into ma and ml respectively (with Italian vowels), and naming the three-
cell triads in order from bottom to top, these 8 trichordals are distinguished

as follows in the C decad. The triads are spread out and marked by +
and — and the terminal triads are repeated in part with an interposed
,
|

to indicate the dissonant interval of a Pythagorean minor Third

£Y
•
5=97} so th a t a H t ne harmonies, consonant and dissonant, peculiar to
any trichordal, may be collected at a glance.

i. Mamama B -D |
F+ A -0+ E -G+ B ~D F+ A
ii. Mimama B -D |
F-tAb + C + E -G + B -D F-tAb
iii. Mamima B -D |
F+ A -C-tEb+G-f- B -D F+ A
iv. Mimima B -D |
F-tAb-f- C-tEb+G+ B -D F-tAb
v. Mamami tBb+D | F-f A -C+ E -G-tBb+D F+ A
vi. Mimami tBb+D |
F-tAb + C+ B -G-tBb+D F-tAb
•yii. Mamimi tBb + D |
F+ A -C-tEb+G-tBb+D F+ A
viii. Mimimi tBb+D I F-tAb+C-tEb + G- tBb+D F-tAb
Each and when these are reduced
of these 8 trichordals contains 7 tones,
to one Octave and sounded in order of pitch, they form that particular
scale in which a piece of music is usually written. But in repeating them
each may begin on any tone of the seven, giving 7 modes (in the ancient
Greak sense) to each trichordal. To distinguish these, change the m of
the name of the triad containing the initial tone into _p when it is its Eirst
(p rima), t when it is its Third (t ertia), and qu when it is its Eifth
(<pdnta), which last is dominant
of course required for the highest or
triad only. The final cadence 56 resulting harmonic fully distinguishes the
scales. Of these I append such as are usually acknowledged, making them
all begin with C, and changing the decad accordingly. Between the
16 27
tones I use (.) for the Semitone y^, (:) for the high Semitone hf, (..) for

10 • 9
the minor Tone -g~, (...) for the major Tone g, and (/.) for the augmented
. 75
tone q£

1. mapama, or ordinary scale of C major.

C • • • Cl> . . C • T . . . Q .. a . . . o . c .

2. E mama/pa, one of Helmholtz's modes of the Eourth, or Quartenge-

schlecht.

c . . id . . . e . / . . .
g . . a . h\} . . . c.

3. C mipdma, Helmholtz's minor-major mode, or Moll-Durgeschlecht.

; e . . . d . . e .
f . . . g . fa\) .*. b . c.
1874.] Mr. A. J. Ellis on Musical Duodenes. 13

4. C mapima, Helmholtz's mode of the minor Seventh with the leading

note, or Septimengeschlecht mit clem Leittone, a very usual form
of the modern ascending scale of C minor.
c . . . d . fe\} . .
/ . . . g . . a . . . b . c'.

5. C minima, the theoretical modern ascending scale of minor.

c . . . d . fe\p . .
f . . . g . ta|? •'• & • c '«

6. F mamapi, considered by Helmholtz (op. c^. p. 434. no. 6) as a

variant of the mode of the minor Seventh.

c . . %d : fe\) . . / . . . g . . a . b\} . . . c'.

7. C mapimi, Helmholtz's mode of the minor Seventh without the

leading note.

c . . d . 1y|? . .
/ . . . g . . a : f &|? . . c'.

8. C mipimi, Helmholtz's mode of the minor Third, or Terzenge-

schlecht, the ordinary form of the modern descending scale of

C minor.
c . . . d . f dp . .
f . . . g . fa\) . . . fb\} . . c'.

9. F mimipi, Helmholtz's mode of the minor Sixth, or Sextenge-

schlecht.

c.cdp . . . felp . .
/ . . .
q ' o t«[> . .
&t>
. . . c\

These 56 harmonic scales are all that can be produced without modu-
lation.
To retain old names as much as possible, mwpdma will be called

C major, and all three, C mipima, mapima, will be considered as

mijpimi,
making up C minor, whilst other forms will be termed unusucd minor
scales. All these, however, and more of the 56 scales mentioned above,
actually occur in modern music, at least for short phrases, although the
usual major and minor alone characterize whole compositions.

D. Modulation and Duodenation.

Although a decad consists of complete triads and cells, yet it is evident

that one or two of the cells may be made parts of other decads, and that
the union triads may be regarded as parts of cells left incomplete. The
tones forming these cells and unions are therefore ambiguous, and there
is alwavs a tendency to complete them
in a different way from that in

the original decad, or, in other words, to proceed to the other decads of
which they form a part. By an extension of the term modulation,
which originally referred to a mere change of mode, this change of
decad might still be called modulation, although decadation might be more
appropriate.
14 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

The Harmonic Heptadecad f

or Unit of Modulation (or Decadation),

n\) tf fa ru f° le

tgt> tBt> D f# su To Me fi
fct> fE|? a B *d# du Mo So Ti ri
or
fffc> fAt> E ig# A Lo Do Mi se

dt> F A ro Fa La de
bb 4.C1 ta ra fe
consists of seven interwoven decads, which are constructed on the
seven tones of a heptad as tonics, and contains 24 tones. On the left is
the heptadecad of 0, in which the decad of is printed in capitals, and

the added tones necessary to complete the heptadecad in small letters.

On the right solfeggio names are proposed as substitutes, to be pronounced
with Italian vowels. These names are founded on those used by the
Tonic Solfaists, and are suitable to any original tonic Bo and they are ;

introduced because singers in just intonation should become accustomed

to the " mental effect " of each of these tones in relation to the Bo selected.
The decad s are named from the names of the tones in the original decad.
The G decad is the dominant or So decad the F the subdominant or Fa
;

decad ; the A the right relative, or major Sixth, or La decad ; the tEf> is

the left relative or minor Third or Mo decad ; the E is the right cor-
relative or major Third, or Mi decad; and the t-A|? is the left correlative
or minor Sixth or Lo decad — all with reference to the C or tonic or Do
decad. These six decads are related to the original decad in the first
degree. The dominant and subdominant decads have each seven tones,
the relative and correlative decads have each siw tones, in common with
the original decad. The dominant decad raises two tones, F and A, by a
81 135
comma, ^, to tf and ta, and one, F, by a sharp, j^g, to
f# . The subdo-
minant depresses two tones, tB|? and D, by a comma, to bf> and Jd, and
one, D, by a sharp, to d^. These two decads are therefore equally related
to the original. The right relative decad depresses one tone, D, by a
comma, to id, and raises three tones, F, 0, Gr, by a low sharp, ^r, to

Jf#, Jcjf, Jgjf. The right correlative decad raises one tone, F, by a sharp,
to f #,and three tones, 0, Gr, D, by a low sharp, to Jc#, Jgj, JdJ. Hence
the right relative is more nearly related to the original than the right
correlative. Similarly the left relative, changing F in f f, and 0, Gr, D
into tcj?, fgj>, tdf>, is more nearly related than the left correlative, which
changes D into d^, and F, 0, Gr into ffp, tef>, tg£>. In commatic tem-
peraments, where Jd, f f are not distinguished from D, F, the relative
decads seem, like the just dominant and subdominant, to have 7 tones in
common with the original, and similarly the dominant and subdominant
decads appear to have 9 tones in common with the original. Hence
various important confusions have arisen, of which it must suffice to have
indicated the source. Since the most natural and easv harmonic tri-
1874.] Mr. A. J. Ellis on Musical Duodenes* 15

chordal, with the fullest and best harmonies, is undoubtedly the mamama
or major, consisting of the central and right columns of the decad con-
taining it, modulation (or decadation) to the right is more common than
modulation to the left ; and, owing to the closer relationship, modulation

to the right relative is more common than into the right correlative, which
generally occurs as a vertical (dominant) modulation from the latter.
Vertical (dominant or subdominant) modulations are, however, the most
common of all, unconsciously (owing to commatic temperament) into
the subdominant (when the minor chord, +d— f+a, is used for the chord
of the added Sixth, f + a d), and consciously into the dominant (in
|

which, however, only f J, and not fa, is commonly recognized).

The vertical modulation is so common that it influences scales, producing
actual tetrachordals, which are disguised in melody by being occasionally
deprived of their extreme tones, so as to reduce their apparent number at
any time to 7. The fourth chord may be added on to the name by a
hyphen. Thus we have

C ma-mapdma. ..... b^ + id— f -fa — c-f-e-— g-f b — d,

in which the b\} is seldom touched except in the chord of the dominant
Seventh, c+e— g |
Wp, and then not in melody, but Jd often comes into
melody. Similarly we have
mapama-ma f + a — c+e— g+b — d + fjf — fa,
where fa is not touched in the melody. But in minor scales this is more
marked, as

A mimvpa-ma Jd— f+a-— c-f-e + Jg#— b+id#— f#,

where Jd and f J are not touched in the melody ; so that the scale reads

e ./ .*.
%g$ .a . . . b . d .*.
Wjf . e\

with 4 semitones and 2 augmented tones, which has an extremely strange

effect \ Another scale of this kind is
A mimijpi-ma Jd — f + a — c+e—g+b-f+dj — fjf,
which occurs in the modern treatment of Helmholtz's mode of the minor
Sixth (No. 9, above). The apparent scale is

e . / . . . g . . a . . . b . c .*. JcZ'J . e' 9

which has 3 Semitones. These are, in fact, all cases of vertical modula-
tion (or decadation) and it is only by recognizing this fact that we are
;

able to reduce them to just intonation. They have not been, however,
hitherto so conceived, and hence it became necessary, for the purposes of

1 This scale and its harmonies are taken from C. Child Spenser's Rudimentary and '

Practical Treatise on Music,' vol. ii. p. 42. He does not acknowledge either Jd or f# but ;

he really uses {d in his second chord, \d a \$ /', and he only avoids f % by using
f-j-a. .. b + +d$ for the usual chord of the dominant Seventh, fo-j-'JAfl — ff a. |
 16 Mr. A. J, Ellis on Musical Duodenes. [Nov. 19,

this paper, to explain A means of putting the strange-looking

them.
chords
1
, and f -f a— c with $d#, or f -fa with +d#, which
f +a ... b + +d#,
they contain, under the hands of the performer on justly intoned instru-
ments, is absolutely necessary.

The Harmonic Duodene or Element of Modulation, as distinguished from

the heptadecad or unit of modulation, contains the 12 tones inclosed
within an oblong in the figure of the heptadecad. It is seen to contain
a complete decad of and two additional tones, 1% and dj?, which I term
mutators, as each of them is part of two cells, and hence lead the old
decad to change into the new decads containing them. Thus f# is
part of

the vertical cell f f fa and of the lateral cell D f

I) f# B Jd#,

and hence leads both to the dominant decad and to the right correlative
decad. Again, d\? is a part of

the vertical cell d\) E and of the lateral cell ff[? f A[>
b\> id d[? E,

and hence leads to the subdominant decad and to the left correlative
decad. But these mutators, fJ and d|>, also complete two scales left
incomplete in the decad because they required vertical modulation (or
decadation), namely,

f A\> mapdma .... dp + f - ta|? + c — f e|? -f g — fb[>, and

E mipimi .... a —c+ e ~g-\~ b — d-f f#,

and d(? + E ... Gr-fB, fA[>-f-C ...

also complete the peculiar chords,
D+f#, D|?+E-tAf? with B, f A|? C-fEk> with f#, which occur in the +
tetrachordals of minor scales already mentioned.
A
duodene, then, consists of 12 tones, forming four trims of major
Thirds arranged in three quaternions of Eifths. Hence the duodene con-
structed on the second tone of any trine, fA^-j-C + E, contains the
mapdma or major of the first tone f Afc>, the complete decad of the second
tone C, and the mipimi or common descending minor of the third tone E.
It has therefore three tonics, f A|?, C, and E ; but the tonic of the decad
being most characteristic, this is called the root of the duodene, and the
duodene is named after it.
Any duodene is clearly and sharply separated from its adjacent trines
and quaternions, as shown in Table L, where the small innermost oblong
marks off the duodene of 0. Eor in the duodene the smallest intervals
between two adjacent tones are the Semitone,
jg,
(f# to G, B to C, E to E,
• The justification of these chords is that the interval fio M'$-2 . &.1 , -| .
41Ajl
'
8' :
i -
fff-*
anc^ 1S bence very nearly the interval of the natural Seventh =-£•.
1874.] Mr. A. J, Ellis on Musical Duodenes. 17

D to W\), G- to f A|>, C to d|?) ; the Sharp, ^, (E to f J, d[> to D) ; and

25
the low Sharp, ^4, (f Bp to B, tEj? to E, f A\> to A). But if we take the

trine above, f f -f ta -f c #, we nave two intervals of a comma, t, (E to f f

A to fa), and one
of a diaskhisma, f J^, (c( to &\>). If we take the trine
below, g|?'+b|?+ id, we have the same intervals of a comma (b[> to fBj>,
td to D), and a diaskhisma (f J to g[?). If we take the quaternion to the
right, as JcJ x *g# x Jd$ x a(, we have three intervals of a diesis, f tj^,
(a# to f Bfc>, JdJ to tEf>, tgjf to tA|?, and JeJ to d|?) ; and similarly if
we proceed to the left. Hence the intervals introduced by adjacent
trines and quaternions are all less than two commas. In equal tempera-
ment no new intervals would be thus introduced for all the Fifths are ;

there so altered that the new upper trine, tempered ff +fa-j-cj, would
become identical with the original bottom trine, tempered dj^+E + A,
except in order of terms and the new quaternion to the right, tempered
;

£c# X Jg# x idj x a#, would be identical both in value and order of terms
with the old quaternion to the left, tempered d\) x f A\} x f E[? x 'YBp. The
consequence is that only one duodene exists for equal temperament, and
the real nature of modulation is thoroughly disguised. In tertian tem-
perament this would not be the case the quaternions would be distin- ;

guished, but the trines would partly coincide, and hence some, but not all,
1
of the meaning of modulation would be lost .

1
If in Table I. the signs t J be omitted, and the letters and the signs % b be taken
to have their values in Tertian or any uniform commatic temperament (except the
Equal, which is also skhismatic), the Table will represent the corresponding duodenes.
But if the letters and signs % b are taken to have their value in the Equal tempera-
ment, so that
c D E F G x*» B
Dbb Ebb Fb Gbb Abb Bbb Ob
and B# on r>8# E| •*- ## G## A$$
and
c* m E#
F
F$
Gb
G#
Ab
A#
Bb
M
Db Eb

(showing the utterly absurd relations between symbolization and signification), then the
same Table will reduce to the one central duodene with its tones differently distributed.
This will be still better shown by using

C cd D de E F fg G- ga A ab B
for the 12 digitals on a piano, so that the central duodene and its adjacent trines and
quaternions reduce to

cd F A cd F

voi*. xxiii. c
18 Mr. A. J. Ellis on Musical Duo denes. [Nov. 19,,

In skhistic intonation, the modification of the Eifth leads to a modi-

fication of the comma and obliteration of the skhisma ; so that the two
first tones, skhistic ff, fa, of the new upper trine, f f -f- f a 4- cf are one
,

skhistic comma higher^ and the third, skhistic c#, is one skhistic comma
lower than the two last tones, E, A, and the first tone D\) of the old
trine, D^ + E-j-A. And the tones of the new right quaternion will be
in the same order, exactly two skhistic commas flatter than the old left-
hand quaternion 1
.

The consequence is that if we took 4 independent duodenes (that is,

such that no tone of one is common to any tone of the other) as the duo-
denes of f B|?, A J, G[>, and JE J, the tones of which are contained within
the dotted lines and right side of the inner oblong of Table I., the tones
of the duodenes Af and JEjf will be two commas flatter than those of
f B[> and Gr|? and the tones of the two first quaternions of the f B|? and
;

A$ duodenes will be one comma sharper than those of the two last
quaternions of Gr[> and JEJ, while the tones of the third quaternions of
f B|? and A J will be one comma flatter than those of the duodenes of G\}
and JEjf respectively.
The 48 tones will consist of four corresponding
result, then, is that the
sets of 12 tones each appearing in 4 forms, differing in pitch by one
skhistic comma. This will appear more clearly by the following Table,
in which the value in octs of the cycle of 301 is given for 73 tones, being
those in cols. i. to vi. of Table I., less those in col. I., lines I, m, n, and
col. vi., lines y, z. The 48 of those tones contained in 4 independent duo-

denes are in Roman

capitals, the other of the 73 tones, which are some
of their skhisticsynonyms, are in Eoman small letters, and other syno-
nyms are added in Italics the. whole are divided into groups of 4, the
;

constituents of which differ from one another by 5 octs, or one skhistic

comma.

1 This is readily seen by expressing the tones1 in terms of the octs of the cycle of
-
301,
by continually adding and subtracting 176 for the Fifths and 97 for the major Thirds,
adding or subtracting 301 as often as is necessary to reduce to the same Octave. A
skhistic comma is represented by 5 octs. This gives

Tones. Octs.

fdb tf ta c# e# 33 130 227 23 120

tgb
fcb
tBb
fEb
D
G
n
B
a# 158
283
255
79
51
176
148
273
245
69
Jdtf
tfb fAb C E teflL 107 204 97 194
fbbb Db F A Jc# 232 28 125 222 18

tebb gb bb Jd p# 56 153 250 46 143

 1874.] Mr. A. J. Ellis on Musical Duodenes. 19

Tones. Octs. Tones. Octs. Tones. Octs.

JCg \\db bgg 18 $m Jtt^bb 115 JGgg Ja fl&bb 217

Cg Jdb. t&gg 23 Eg +fi
+ tt^bb 120 ggg A J6bb 222
teg Db nbn 28 teg •
P fe'bb. 125 tggg tA bbb 227
ttcg tDb ttt6#* 33 tt4 tE gbb 130 ttggg tta tBbb 232

JCgg
egg
tfc*
JD
JJebb
jebb
41
46
mEg
J:^b
fcb
egg
un
143
148
JAg
Ag
\\bb
\bb
240
245
tegg D ebb 51 tfg Gb ttegg 153 tag Bb 250
ttegg tD tebb 56 tt/g tGbtttegg 158 tttfg tBb 255

IPS tJeb 64 .{Fgg tiff H«bb 166 agg JB IJcb 268

JDg Jeb 69 fgg JG t«bb 171 tagg B \cb 273
dg Bb 74 tm G abb 176 ttagg tb Cb 278
tdg tEb 79 tt/gg tg tAbb 181 tttagg tti 1hCb 283

«
JDgg {e
E
JJ/b
t/b
92
97
mGg
Jab
ab
194
199
JBg
Bg
ft*
tc
ffi\xlbb
+ : :<^bb
291
296
tdgg tE fb 102 t^g tAb 204 -\b% : \dbb
ttdgg tte tEb 107 tt^-g ftAb 209 tt&fl tC dbb 5
i

Since, then, the duodene of C is precisely adapted for placing on our

ordinary manuals, and no corresponding tones which have to be intro-
duced within these limits will be more than two or three commas sharper
or natter than these, such corresponding tones (owing to our habits of
reading musical notes into directions for using digitals) will be all fitted
for being played on the same digitals. This is the most important point
in the practical construction of instruments, and is for the first time
pointed out in this paper.
Another important result is, that if we take any 12 consecutive skhistic
tones in Order of Fifths, or 8 consecutive tones in that order, and 4
others separated from them by 24 or 48 Fifths, although such tones will
not form a duodene, they will be 12 tones suitable for our manuals,
and will therefore afford the means of temporarily supplementing other
arrangements.
In skhisfcic intonation, then, the modulational peculiarities of just
intonation are preserved and it will be convenient in future to consider
;

modulation as taking place by duodenes, and hence consisting of duodena-

tion. We shall therefore have in just intonation both vertical and
lateral duodenation to consider ; but in skhistic intonation it will be seen
by Table II. that one right lettered duodenation, as from root f Bj? to root
D, is the same as eight descending vertical duodenations, for these would
in just intonation lead to the root ~E\)\), which, as shown in the last Table,
is skhistically identical with D. Hence in skhistic intonation we have, so
far as instruments are concerned, only to render vertical duodenation
possible and easy.
c2
 20 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

And in writing music, if we note at the top of any bar the name of the
duodene to which the notes to be played belong, and suppose this duodenal
(as the mark may be called, in contradistinction to the signature, which
will remain as before) to hold till a new one is written (according to the
custom of musical signatures), we shall be able precisely to mark the pitch
of every tone in just or skhistic intonation, without introducing any change
or any additional sign into the staff-notation of music 1 This is again an ,

entirely new from the present theory. The

practical principle resulting
duodenal will direct the player to the mode of arranging the manual he
has to use. It should be the duty of the composer to insert the duodenals
himself; but in respect of existing compositions, which were composed for
some commatic system of temperament, it will be often difficult to deter-
mine which of two adjacent vertical duodenes it would be best to use
and it will probably be necessary to introduce commatic changes, when
they can be made within the limits of a single heptadecad. Also in the
case of compositions in a major scale, which do not change into the minors
of the same decad, and hence use only two quaternions of a duodene, but
will necessarilyand frequently modulate to the right, it is more convenient
to consider the music as performed in the first and second quaternions of
a duodene having for its root the Third of the major scale, because the
third quaternion of that duodene contains the tones required for right
lateral modulation. Thus major will be assigned to the duodene of E,
No. 19 of Table II., and E major to the duodene of A, No. 20 of Table II.,
&c. This makes the modulation from the major into the relative minor
as simple and direct as vertical modulation, for C major passes into any
form of A minor or major by descending vertically from the duodene of
E to that of A. All pieces in any minor scale pass into each of the three
quaternions of a duodene, and hence their duodenal will be the tonic of
their decad, which gives its name to the duodene. The duodene is then
prepared for playing the synonymous major of that minor scale. Such
duodenals might be distinguished by an added star.
It often happens that passing tones, changing notes and appoggiature,
are introduced which do not belong to the harmony. They are written
usually after the laws of Pythagorean temperament, but their pitch is

really indeterminate. Eor these there no occasion to change the duo-

is

denal at all. They will then be played in the duodene of the other

1 For theoretical and experimental purposes it may be sometimes convenient to use

signs equivalent to t J in the staff-notation itself. The signs L p for t, ft, and

I
£ for J, {{,
being the tails of quavers and semiquavers, are well adapted for this pur-

pose. The direction of the angles show ascent and descent, and the forms exist as
types for every required position on the staff; thus fa, td, tbb, and Jgjjl, would be

$P=q«£
35^E=S~£EEl^j
1874.] Mr. A. J. Ellis on Musical Duodenes. 21

harmonies by a tone of not more than two commas different, which must
be considered as their proper representative in just intonation.

E. Number of Tones required.

The next important point is to determine how many duodenes must be

provided. In Table I. the large inner oblong contains all the duodenes
which have at least one tone in common with the original duodene of C.
Thus the duodenes tD[? and JB have respectively the tones f B[? and A
in common with the original duodene, and no others. If we proceeded
further, as vertically from the f D\) to the ttA[> duodene, we should no
longer have any connexion with the decad of the original tonic C. The
confusions of modern equal temperament might lead much further but ;

in that case, we must restore the commatic changes which equal tempera-
ment ignored, considering, for example, that when the composer modu-
lated into tempered A[> from tempered D\} he really meant to make a
modulation from just D|? into just f A|?, and not from, f D\) to just ttA|?.
It would probably have never been the composer's intention to proceed to
such unrelated duodenes as these two last.
The limits of the original roots of duodenes may be taken to be the tones
of the duodene of C. Practically, composers had no others in their minds.
Any smaller changes of pitch were relegated to differences in the pitch
of 0, whence all the others were derived. If, then, we construct the

limiting duodenes to the extreme tones of the duodene of as original

roots, we shall obtain all the tones in Table I., being 9x13=117 in
number. This is the number of tones required, therefore, in just
intonation.
Skhistic intonation would introduce identifications which would reduce
this number to the 9 x 8=72 tones in the lines p to x in Table I., toge-
ther with three in col. 1, lines Z, m, n, and two in col. 9, lines y and z, that
is to 77 skhistic tones in all. The last 5 are so extremely unlikely to
occur, however, that we may
consider these 72 skhistic tones as suffi-
ciently representing the whole 117 of the Table. These 72 tones form
6 independent duodenes, those of ftE^, D and JCJJf, and of fC^, Bj?
and JAjf. It will be shown that there is really no difficulty in playing
them all, with existing means, if required; but they would not be
required. The tendency of musicians is not to modulate to both right
and left equally in the same piece. It has been already noted that on
account of the prevalence of major scales duodenation is generally to the
right. The fingering for the duodenes of fB|? and AJ would be the same
on manuals constructed on the duodenary theory, although the tones in
skhistic intonation would differ by two skhistic commas. If, then, a piece

in tB[? duodenated much to the left, that is (for skhistic intonation),

ascended vertically, we could play it as A#. would simply be neces-
It
sary to write tB\} as its duodenal, as that is shown to be identical with
A# in the last Table. We should then be able to use ascending duodena-
22 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

tion with great ease, asshown in Table II., even if columns 1 and 2 in
Table were omitted. Hence we may begin by cancelling columns 1 and
I.

2 of Table II. In view of the greater frequency of right lateral or

descending duodenation, we need only reject column 9 to the right 1 We .

have thus reduced the just tones to the 6 x 13 = 78 in columns I. to VI.

of Table I. The skhistic identifications reduce these further to the
6 x 8=48 tones in lines p tox of the same columns, together with the
three tones in col. i., I, m, n, and the two in col. vi., y, z. As these tones
may, I think, be always avoided by properly choosing the original root,
motives of convenience induce me to reduce the number of skhistic tones
necessary to the 48 included by the dotted lines in cols. i. to vi. of
Table I.

In my
former paper (' Proceedings/ vol. xiii. p. 98), not having taken a
comprehensive view of the nature of modulation, I fixed the
sufficiently
number of just tones required at 72 instead of 117, and showed that they
would reduce by skhismatic substitution (for I had not then worked out
the theory of skhistic temperament) to 45 ; and on examination it will be
found that these 45 include the 48 which I have just named, with the
exception of those in col. i., lines p and q, and col. vi., line x of Table I.
The tones used in Mr. Liston's organ (according to the statement I was
able to give in Proceedings,' vol. xiii. p. 417, note §), on being treated
'

skhistically, include 44 of these 48 tones, omitting the 4 tones in col. i.,

lines w, x, col. v., line x, and col. vi., line x, and introduces two others
found in col. 2, lines p and x, and probably only due to his system of
tuning. He has thus 46 tones in all. G-en. T. Perronet Thompson's organ
(see my paper in Proceedings,' vol. xiii. p. 102), when similarly reduced,
'

has 38 of my 48 tones, omitting all col. i., col. n., line x, and col. vi.,
line r in Table I., and retaining the two tones of col. vi., lines y, 3, which
I do not find necessary. Mr. Poole's latest organ (Silliman's Journal
for 1867, pp. 1 to 45), after rejecting his 39 natural Sevenths, which I
expressly exclude, has 61 just tones, which reduce to the 36 skhistic
tones in col. n., lines q to x, cols, in., rv., and v., lines^ to x, and col. vi.,
lines p to t. These are the principal attempts at limiting the scale actually
made up to this time and hence I conclude that my reduction to 48
;

skhistic tones (that is, practically, unequally just tones) would embrace
almost every case, though it is conceivable that some extraordinary music
might make it advisable to introduce 12 more, namely, col. 2, lines p to
x, and col. 9, lines p to s, making 60 tones in all, for adding which pro-
vision should be made.

1
movements of a long piece modulation took place first
If in changes of key in the
much and then much to the right, we might perhaps make commatic changes
to the left
between the movements, without disturbing the connexion. And when changes are
introduced by successions of discords, such commatic changes could not be observed at
all by the listener. By the use of the duodenal, however, they will be rendered per-
fectly simple to the performer.
1874.] Mr. A. J. Ellis on Musical Duodenes. 23

The 48 tones thus pointed out form the 32 trines, which, with their
synonyms, are shown on Table II. By taking these in quater-
skhistic
nions we In Table II. the root of the duodene is
obtain 29 duodenes.
written against its and hence the root itself is found in
uppermost trine,
the middle of the second trine below, and the whole duodene extends to
the third trine below. Trines on the same line are skhistically identical,
the capitals indicating the names of the 48 selected tones of cols. I. to vi.,
lines p to at, in Table I. In Table III. the roots of these duodenes are
arranged in 4 columns, of which each tone in the same line is skhis-
tically identical; but, proceeding from left to right, each tone is, in
just intonation, one skhisma natter than the next adjacent tone on
the right.
In Table III., also, the tones in each of the duodenes are written down
in the order in which they would stand on a manual ; but the skhistic
identities of the central column of tones in the preceding Table of Octs
(p. 19) have been used to give the same names to
all the tones in one

column, exclusive of the prefixed f and t. thus see clearly that We

three new tones are introduced by each new successive duodene, two
fallingand one rising by a skhistic comma, as respects the tones they
replace. "We also see that each tone prevails through 4 consecutive
duodenes, and that there are 4, and only 4, varieties of f, J in each
column.
This completes the theory of the construction of instruments, because
the rest is properly the work of the mechanician, and consists simply in
devising a method for bringing each of these 29 duodenes under the hand
of the performer, when indicated by the duodenal. It will be sufficient
here to point out a few experimental instruments, to suggest some prac-
tical forms, and to show that means of playing in just intonation with
fixed tones already exist.

I\ Justly or Skhistically Intoned Instruments,

1. Just Concertina (exhibited when this paper was read). The C con-
certina described in my former paper (Proceedings,
vol. xiii. p. 104) contains the portion of a hepta-
D E# —
decad shown in the margin that is, the duodene of
G B tm E with the exception of A#, and the duodene of A
C E JG# with the exception of JE$. It has the whole decad
E A ic# of E, and the major scales of E, C, G, E. I have
B b JD found it a most useful instrument in all my ex-
periments. Using capitals for white and small
letters for black studs, its 14 notes are tuned thus :

Jcjf, D id, E Jd#, E f#, G tg% tA a, B b[>

24 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

2. e/kstf Harmonium (exhibited when this paper was read). For

ordinary lecture and illustrative purposes, this is

tE the cheapest and best instrument. It contains the

fBb D portion of a heptadecad shown in the margin, and
tEfc> a B hence contains the duodene of C with the exception
fAt> c E of E$, the whole decad of C, and the major scales

^b E A of fEJ7 and tA|?, The two last show vertical

modulation. Again, f E|? major or tA[> major to C
decad shows right lateral, and, inversely, C decad to tE|? major or tA[>
major shows left lateral modulation. The E and tE show the influence
of a comma, and the difference between the just triad, D|?-j-E — f A|?, and
the Pythagorean triad, D\> with f E and f A\). Also in the key of f E|?
major, the difference can be shown between the minor chord, E — f A|? -f 0,
and the chord of the added Sixth, tAj^ + O tF. It also contains the ]

German Sixth, Db + E — f A\) with B (which is a close imitation of the

chord of the natural Seventh), the Italian Sixth, P[?-f-E with B, and the
Erench Sixth, Db + E .. G + B. Using capitals and small letters for
.

the black and white keys, they are arranged thus :

C d[? D te> E E ff G tab A tb|? B.

The tE is placed on the E# digital, and the fingering is normal in other

respects. The five tones, d|? x tab x te[? x tb|? x tf, will then enable us
to play all Scotch and other music containing only fire tones, in perfectly
just intonation,on the black digitals, and to show that such music can-
not be harmonized. The practical direction for tuning is, "tune the
following 7 major chords without beats, putting tE on the E# digital,
A D
E C, C E G, G B ; Db E f Afc>, tA[> fE\>, fE\> G tB[>, tB|? tF." D
A four-Octave instrument was thus tuned in two hours.
8. Hephtharmonium. This requires bwo rows of vibrators and Mr.
Saunders's " tilting action/' already described. The vibrators are dis-
posed thus :

Bach . . . . tDb ic# iT> id# tE|? fF Jf# fG\> Jg# JA bb tty
Front .... db B teb E E f# G tab A tb[> B.

This harmonium contains all the tones in a heptadecad (whence its

name), and consequently illustrates every kind of modulation in the first
degree,and becomes a most valuable instrument for the lecturer and
teacher of singing. The front vibrators, to which correspond the digitals
when the stops are all pushed in, contain the whole duodene of C, and
the stops enable the player to exchange their tones for those in the six
other decads. All lie on the usual digitals except tDb and tGb, which
are placed on the long white digitals of and G, in place of the short
black digitals next to their right, as these were wanted for Jcjf and Jgf
This makes a slight difference in the fingering of tEb minor and
similar scales.
1874.] Mr. A. J. Ellis on Musical Duodenes, 25

4. Helmholtz's Harmonium (Tonempfindungen, p. 496). Two sets of

vibrators are tuned, the back set to the duodene of D$, No. 14,
~E$p or
Tables II. and III., and the front set to duodene of tOp or B, No. 18, of
the same. This instrument contains the eight trines,Nos. 14 to 21, and
the five duodenes, Nos. 14 to 18, Table II. The " tilting action " pro-
duces a most useful experimental instrument, which is far easier to use
than Helmholtz's own double manual instrument, because it has only-
one manual, and requires no alteration in ordinary fingering. For this
purpose the stops may be reduced to four, each changing a trine instead
of a single note.
GuerouWs Harmonium (Comptes Bendus, 1872, p. 1188). This,
5.
again, may be treated as the last by means of Mr. Saunders's " tilting
action." Two sets of vibrators must be used, the back set tuned to duo-
dene tD\> or Cft No. 16, and the front to iBftj or A, No. 20, of
Tables II. and III. M. G-ueroult tuned the BJJ of duodene No. 16 as

B # = j^T^BJf,
v
so as to make the combinational tone of GTjf and B'J the

same and E|?, the other tones being tuned in just intona-
as that of B'J
tion from C. Omitting this as unnecessary, the instrument contains the
eight trines, Nos. 16 to 23 of Table II., and the five duodenes, Nos. 16
to 20. M. Grueroult arranged the tones somewhat differently for two
manuals.
6. Buoni, Trioni, Quartoni, Quintoni, Sestoni. The Bussian horn-band
which visited London some years ago, and produed great effects by each
performer's playing a single tone only (and hence, probably, in just
intonation), and the customs of hand-bell and church-bell ringers, who
each play a single note in a melody, have suggested to me the use of two,
three, four, five, or sioo harmoniums or pianofortes, indicated by the above

names, for the purpose of playing in skhistic or unequally just intonation,

by means of two, three, four, five, or six performers, among whom the
tones are distributed. The Buoni are intended for two independent duo-
denes, as in the two last cases, the Quartoni for four such, playing the
whole 48 tones, the Sestoni for six, in the almost impossible case of 72
tones being required. The Trioni supplement the Duoni by using 12
additional tones, forming consecutive Fifths, and hence not constituting
a duodene, by which means the 36 tones of Mr. Poole's compass can be
played. The Quintoni supplement the Quartoni in a similar manner ; but
the first 8 tones are those in col. 2, lines p to x,and the last 4 those
in col. 9, lines j) to s of Table II. —giving 60 tones on the whole,
chosen so as to supplement without changing the arrangement of the
Quartoni.
In each case separate harmoniums or pianos are used, with no change
in existingmechanism or fingering, but only in intonation ; so that the
instruments could be obtained and tuned in unequally just intonation, as
26 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

already described, without difficulty, at a day's notice. The music is to

be marked with the proper duodenals, and the duodenes thus indicated
are to be transcribed separately, and divided into parts by transverse
lines, corresponding to the tones existing on the different instruments.
The copyist writes out a separate part for each performer (which had
better have an indication of the complete harmony annexed), in which
only those notes that belong to his own instrument are written. Thus,
suppose that the duodene E$, and the lines show what tones lie on
is

^;ne instrument (X, II., III.), as in the mar-

(II.) fE Grit Bit
gin. Suppose that the succession of chords
tA C# E# 4 g% V c'% /# e'# a% c'% and B tcl'% f% V
(I.) D E# A# (III.) has to be played. The tones will be distri-
a b ^ff
-f-
buted as in the margin. Considerable practice
would be necessary to take up the notes truly at the right moment, but
c y there is no longer any instrumental or digitational
V c'$ { difficulty in playing in just intonation.
a-) f'S
Leaving Duoni aside as sufficiently indicated
in the two last cases, and of only experimental
(II.) g%
interest, and Sestoni as practically not required,
(in.) 4 «'# %d% it will be enough to explain the tuning of

Trloni, Quartoni, and Quintoni.

Trioni. Tune the three instruments thus :-

(I.) C Jdfc> D te\> E tE igfc> G ta^ B

(II.) iO d|> fD e\> tE E gf> i'G- a|? A bt» tB
(in.) iB# ic# tc## M# JE -"4f ***
<rf
"+"^-
"fffl* ">~o«H" JB
Then (I.) is in the duodene
No. 10, and (II.) in that of E|?,
of Gr,
No. 14 of Tables IT. and III.
and this readily gives the method of
;

tuning them. (III.) consists of 12 tones forming consecutive Eifths from

E# to JE, col. 7, lines q to y, and from JGr$# to JBJ, col. 8, lines q to £ of
Table I. Eor (III.) begin by tuning JG-jf a major Third without beats
to E in (I.), and then work up to EJ and down to JE by Eifths, verifying
with the corresponding major Thirds below in (I.) and (II.). Then tune
JB# a major Third above £Gr#, and tune up by Eifths to JG-JfJ, verifying
by the major Thirds below, which lie all in (III.). By this the three
instruments are completely in tune, and give the 17 duodenes, Nos. 10 to
26, Table II., containing Mr. Poole's scale of 36 tones.
Quartoni are much simpler, because they contain the four independent
duodenes,

(I.) of tBb, No. 1, (III.) of A#, No. 25,

(II.) of G|?, No. 5, (IV.) of JF#, No. 29

in Table III., where the corresponding lines give the tuning of each
1874.] Mr. A. J. Ellis on Musical Duodenes. 27

instrument. The arrangement of tones is managed as before. Suppose,

for example, we have to play the succession of chords egc' g', ffc' a' 9

d gf f
and c g e' c in the duodene of 0, this duoclene would be
&',

written and the tones would be distributed among the four instru-
ments as follows :

^- }
tEb &
n (III.) (I.) 9 — 9'
9d 9
(II.) c o'f f cc'

tAt> E (III.) — — &'. —

(U) A (IV.) (IV.) e a' e'

Quintoni have Rye instruments, (I.) to (IV.) being tuned as in Quar-

and (V.) added when by some extraordinary vagaries of modulation
toni,

more than 48 tones are needed. (V.) is tuned thus, where the synonyms
show the meaning of the arrangement :

(V.)tDbb Jb#*ttEbb tfbb ttFb tGbb JejMf -ft Abb \m% ttBbb tcbb %k%%
=ttO |Jc# ttD tteb tttB ttF \\i% TO \\g% tttA ttbb ftE

The tuning of (V.) is effected thus : —Tune ttA[?|? as a major Third

below tC|? on and then work up by Fifths to ttF[> and down to
(I.),

tF[?J?, verifying by the major Thirds above in (I.) and (II.). Then tune
tA## as a major Third above iF## in (III.), and work up by Fifths to
JFJfJfJf, verifying by the major Thirds below in (III.). The notation
of the tones, though inevitable, is frightful ; but the tuning is very
simple, and the use of the duodenal leaves the old staff-notation un-
changed. It is most probable that the fifth instrument would never be
wanted.
7. Great and Small Duodenary Harmonium. Although the mode just
explained places just intonation at the immediate command of three or
four performers, yet it seems necessary to suggest a mode of putting all

the 17 or 29 duodenes at the command of a single performer. I suggest

the following for consideration. It seems practicable, but would doubt-
less require much mechanical treatment from harmonium-builders before
it would act properly. It will be enough to indicate the form of the
great duodenary.
Take four sets of vibrators, tuned as for Quartoni, and placed one
behind the other, each opening with a separate valve connected with
a digital. Sometimes two digitals will have to be connected with
the same valve. Conceive the manual as a set of 29 " steps," with
|-inch " tread " and |-inch " rise," the lowest step next the per-
former. Each step for the length of an Octave is divided into 12
digitals corresponding to the columns in Table III. The width of
28 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

the digitals to be as follows for No. 11 of Table III., in eighths of

an inch: —
D E F A
535355353535
C dt> teb tg\> G- fa[? fb[? B

The digitals corresponding to the small letters are to rise | inch above
the others and to be bevelled, so that they are § inch wide at bottom, and
| inch wide at top. Each step is then a miniature finger-board in
the ordinary arrangement. "Whenever any note occurs in 4 consecutive
steps, as shown by the
cross lines in Table III., its 4 digitals are to be
consolidated into one, so that, except in " steps " 1 to 3 and 26 to 29, the
digitals will be practically 3 inches long. To show which digitals are
consolidated, colour the low wide digitals alternately white and light
red, and the high narrow digitals alternately light blue and light
brown, distinctions of colour easily seen. To mark the duodene,
draw a black | inch broad, across the digital bearing the name
line,

of the duodene, and put a black circle of | inch in diameter on the

tonic of the major scale which it contains. The lines thus marked,
together with the alternation of colour, will clearly distinguish each
duodene.
The depth of this manual from front to back would be 21 1 inches, and
the rise 7% inches; the width of an Octave from O to B is 6 J inches,
and from C to c is 6-J inches. This last width 7f inches on the is

piano; but as the hand would on the duodenary always have to dip
between high digitals to strike Octaves of low digitals, it must be
held more upright, and hence its span will be less. A manual of
five Octaves and one note, C to c"" will be 31-| inches long.
9
The
number of movable digitals in each column of Table III. is 8, which
open only 4 valves this will necessitate coupling the details result-
; —
ing from Table III., which may be considered as a ground-plan of this
manual 1 .

1 When this paper was read I mentioned that the 48 tones, making 29 duodenes, of
Tables II. and III. could be played on Mr. Bosanquet's " generalized key-board," as
exhibited to the Royal Society when his paper was read on January 30, 1873, with less
difficulty in mechanism than by the plan I proposed (of which a model was exhibited),
but that slightly new fingering would then be necessary and also that the 72 tones of
;

Table I., lines p to x, making 53 duodenes, might be played by the same arrangement
on a manual not larger than that which I proposed for the 48 tones or 29 duodenes
and hence that the sole advantage of my scheme for a manual was its preservation of
the present fingering, against which had to be set off the advantage that the new finger-
ing of Mr. Bosanquet would be the same in all keys or duodenes. The intonation,'
however, would remain different from Mr. Bosanquet's.
1874.] Mr. A. J. Ellis on Musical Duodenes.

Table I.

Limits of Duodenation and Number of Tones.

I. II. III. IV. V. VI.

2. 3. 4. 5. 6. 7. 8. 9.

I
771
tttBbb
tttEbb
ttDb
ttGb
ttF
ttBb
ttA
tD
tc# tE#
tA#
G##
on
B##
em
mn
mn
l

n tttAbb ttCb ttEb tG tB m Mi A## torn n

P ttDbb ttFb ttAb to tE G# B#

<l
ttGbb ttBbb tDb tF tA E# tB## 2

r ttCbb ttEbb tGb tBb D F# A# tc#s r

s ttFbb ttAbb tCb tEb G B
E
tm s
ttBbbb tDbb tFb tAb C twn
t
u ttEbbb tGbb tBbb Db F A m t
u

w ttAbbb tCbb tEbb Gb Bb tD tA# w

X tDbbb tFbb tAbb Cb Eb tG tB tfD# tm$ X

y tGbbb tBbbb Dbb Fb Ab tc tE tm tm y

z tCbbb tEbbb Gbb Bbb tDb tF tA tm tm

The tones in the small central oblong form the duodene of which C is the root.

The tones in the large central oblong form the duodenes which have at least one
all

tone in common with the central duodene of C, forming the limits of radical duodena-
tion from C.

The complete Table contains all the duodenes which have at least one tone in com-
mon with duodenes whose roots are tones in the duodene of C, forming the limits of
general radical duodenation.

The 48 tones in columns I. to VI., between the dotted lines, are those considered

sufficient for instruments with fixed tones in skhistic or unequally just intonation.
 30 Mr. A. J. Ellis on Musical Duodenes. [Nov. 19,

Table II.

List of Trines and Duodenes in order of Fifths.

The Capitals point out the 48 tones in Table I. ; the small letters are synonyms

Name. No. Trine Name. No. Trine. Name. No. Trine.

tBb 1 ttAb to tE ttAS 1 ttgS ttbS tdSS

tEb 2 tDb tP tA tDS 2 ttcS tteS tgSS
tAb 3 tab tBb D tGS 3 ttfS ttaS tcSS
Db 4 tCb tEb ^ tcs 4 ttb tdS tfSS
Gb 5 tFb tAb c tFS 5 ttc tgS tbS
.
Cb 6 tBbb Db F 1B 6 tta tcS te#
Fb 7 tEbb Gb Bb tE 7 td tfS taS
Bbb 8 fAbb Cb Eb tA 8 +g tb dS

Ebb 9 dbb fb ab D 9 to tE GS tcss 9 ttbS tdSS tfSSS

Abb 10 gbb bbb {db G 10 tF tA cs tFSS 10 tteS tgSS tbSS
{Dbb 11 ebb ebb tg* C 11 tBb D FS tBS 11 tta* tcSS teSS
{Gbb 12 fbb abb {cb F 12 tEb G B tB# 12 tdS tfSS taSS
{Cbb 13 bbbb {dbb {fb Bb 13 tAb E tAS 13 tgS tbS dSS
14 ebbb {gbb {bbb 1 Eb 14 Db F A DS 14 tcS teS gSS
15 abbb {ebb {ebb Ab 15 Gb Bb {D G| 15 tfS taS cSS
16 {dbbb {fbb {abb j
{Db 16 Cb Eb {G cs 16 tb ds m
tESS 17 tdss tfsss tasss {Gb 17 fb ab {c FS 17 tE GS BS
tASS 18 tgss tbss dSSS {Cb 18 bbb {db {f B 18 tA CS ES
dss 19 tcss tess g-sss {Fb 19 ebb {gb {bb E 19 D FS AS
on 20 tfss taSS csss {Bbb 20 abb {cb {eb A 20 G B }DS
ess 21 tbS an fsss {Ebb 21 {dbb {fb {ab {D 21 C E {GS
fss 22 tefll g## hU 22 {gbb {bbb {{db {G 22 F A {CS
bs 23 ta# on m 23 {ebb {ebb {{gb {C 23 Bb {D {FS
m 24 <3g m ass 24 {fbb {abb {{cb {F 24 Eb {G {B

AS 25 GS bs {dss {Bb 25 ab {c {e

{DS 26 cs m {ass {Eb 26 {db {f {a

m
{cs
27
28
.
m
B
AS
'\m
{CSS
{ESS
{Ab
{{Db
27
28
{gb

{cb
{bb

{eb
{{d •

{{g

tn 29
30
E
A
m
{cs
{BS
{ES
{{Gb 29 {fb {ab {{c

30 {bbb{{db {{f
31
32
£D
}G
mB {AS 31 {ebb {{gb {{bb
{{DS 32 {abb {{cb {{eb
»K*=EB33SV£STC=r=:icctaTssssrwnA'zuTJzjzsjKXJi TS5ESg5gscg?5aacaaBcaiK:aBiiMma«aEa
 1874.] Mr, A. J. Ellis on Musical Duodenes, 31

Table III.

Manuals for Duodenary Instruments.

The Capital letters indicate broad and low, small letters narrow and high, digitals.

Digitals,
Nanles Duodenes
of containing the tones of the Duodenes displayed horizontally.
No.
wi th Synonyms.
ft
1 1 2 3 4 5 6 8 9 10 11 | 12

tBb tta# 1 to tdb D teb tE tP tgb a ttab tA tbb ttB

tBb td# 2 C tdb D teb ttB tF tgb a tab tA tbb ttB

tAb +g* 3 db D teb ttE P tgb a tab ttA tbb ttB

Db teg 4 c db tD teb ttE P gb a tab ttA bb ttB

Gb m o clb tD eb ttE P gb ta tab ttA bb 1B

Cb tb 6 to db tD eb tE F gb ta ab ttA bb IB

Fb te 7 to tdb tD eb tE tP gb ta ab tA bb tB

Bbb ta 8 to tdb D eb tE tP tgb ta ab tA tbb tB

ebb D tcSS 9 to tdb D teb tE tF tgb a ab tA tbb B

abb G tf*# 10 tdb D teb E tP tgb a tab tA tbb B
Jdbb C tbS 11 db D teb E F tgb a tab A tbb B
tgbb p teg 12 db JD teb E F gb a tab A bb B
|cbb Bl? ta# 13 • db ^.-L' eb E F gb ta tab A. bb tB

Eb d# 14 {0 db tD eb tE P gb ta ab A bb tB

Ab g# 15 |0 tdb tD eb tE JF gb ta ab tA bb tB

±XJ u c# 16 \G tdb D eb tE JF tgb ta ab tA tbb tB

tgb Ftf teffff 17 to tdb D teb tE tP tgb a ab tA tbb B

Jcb B ta## 18 tdb D Jeb E tP tgb a Jab tA tbb B
ffb E d'P 19 tf.db D :feb E P te b a Jab A tbb B
tbbb A. g## 20 ttdb tD teb E P ftgb a {ab A bb B
Jebb P c## 21 ttdb tD eb E P tfeb ta tab A bb tB

ta fS# 22 \G ttdb P eb tE P ttgb ta ab A bb tB

to b# 23 JO tdb tD eb tE tP ttgb ta ab tA bb tB

j:F e$ 24 to tdb fP eb tE JP tgb ta ab J.-GL tbb tB

|bb A# 25 to Jdb tP Jeb J-E tP tgb tta ab tA tbb B

|eb £D* 26 %\c tdb \%D teb E JP tgb tta tab tA tbb B
JJab JG# 27 JJO ttdb Jeb E tfP tgb tta tab .ol tbb B
||db JCfl 28 +fO ttdb tD teb E JJF ttgb ^.!j.v4 tab A JJbb B

i
ftgb
9Q ttdb !

i
P tteb :
E ttP ttgb ta
i
tab A iibb
T
IT
.j.ij