Two-Slit Experiments: Partitioning, indeterminacy and initial research into the

use of point gestures to determine macro- and micro-level parametric

assignments

Introduction

Composed in 2010, Two-Slit Experiments (TSE) takes as its principal source of material(s) a
pictorial representation of the two-slit experiment, shown below (Ex.1).

Ex.1

As is shown in the above picture, there are four stages to the experiment that have been
demarcated A, B, C and D. From this are derived four material-types that were in turn
assigned to bars of a specific denominator class (dc):

A. /8 dc. 1-12 pitch classes (pcs), always realised as a sustained single pitch or
chromatic cluster.
B. /16 dc. 2-24 pcs, always realised as sustained points.
C. /32 dc. 4-48 pcs, always realised as demisemiquaver dyads and are
staccato in nature.
D. /32 dc also. Realised as 6- or 5-note semitonal clusters spanning the entire
range of a standard piano and always played staccato.

Each dc consists of 18 values, or time signatures (t-sigs): 2/8… 19/8, 11/16… 28/16 and
20/32… 37/32. As there are 3 groups, each consisting of 18 values and with each being
realised only once in the work, the work consists of 54 bars. These are subsequently
partitioned into 18 3-bar subsections, discussed below (see Partitioning).

As was the case in Probability Interpretation, there are instances in which t-sigs from
different classes share the same temporal space value as one or two others, and some t-
sigs that do not. These are shown in Table 1, below. Therefore there can be 1, 2, 3 or 4
material-types (A, B, C and D, above) within a given time-space.

None With 1 other With 2 others

2/8 25/16 5/8 – 20/32 20/32 – 5/8 6/8 – 12/16 – 24/32
3/8 27/16 10/8 – 20/16 20/16 – 10/8 7/8 – 14/16 – 28/32
4/8 21/32 11/8 – 22/16 22/16 – 11/16 8/8 – 16/16 – 32/32
15/8 23/32 12/8 – 24/16 24/16 – 12/8 9/8 – 18/16 – 36/32
16/8 25/32 13/8 – 26/16 26/16 – 13/16 12/16 – 6/8 – 24/32
17/8 27/32 14/8 – 28/16 28/16 – 14/8 14/16 – 7/8 – 28/32
17/8 29/32 11/16 – 22/32 22/32 – 11/16 16/16 – 8/8 – 32/32
19/8 31/32 13/16 – 26/32 26/32 – 13/16 18/16 – 9/8 – 36/32
19/16 33/32 15/16 – 34/32 30/32 – 15/16 24/32 – 6/8 – 12/16
21/16 35/32 17/16 – 34/32 34/32 – 17/16 28/32 – 7/8 – 14/16
23/16 37/32 32/32 – 8/8 – 16/16
36/32 – 9/8 – 18/16

Table 1

1
Partitioning

In the initial stages of the compositional process the focus was on how to partition the piano
in an effective way in order to present the different material-types objectively and cohesively.
Taken as a whole, the standard piano has 88 keys. These are partitioned into successively
smaller segments, shown below (Ex.2).

1-88

1-22 23-44 45-66 67-88

1-6 7-11 12-17 18-22

Ex.2

As one can see from the flow diagram above, the 88 keys of the piano are partitioned into 4
regions consisting of 22 keys, with 1 being the lowest A of the piano and 88 being the
highest C. These regions are then partitioned further into 6- and 5-note semitonal clusters, or
interference clusters (ics), of which there are a total of 16. The micro-partitioning in 1-22,
above, is repeated in the other 3 partitions (23-44, 45-66 and 67-88). As an example, Table 2
shows the lowest 4 interference clusters which appear in the 1-22 pitch range. Highlighted in
red are the 6-note clusters, in black the 5-note clusters.

1-22
11 – G 22 - Fs
10 – Fs 21 – F
9–F 20 – E
8–E 19 – Ds
7 – Ds 18 – D
6–D 17 – Cs
5 – Cs 16 – C
4–C 15 – B
3–B 14 – As
2 – As 13 – A
1–A 12 – Gs
Table 2

The specific assignments of interference clusters within a time-space are determined by the
use of stochastic processes, or random number generators, which are discussed on pages 6
and 7.

As well as the partitioning of the piano keys to create interference clusters, the keys are also
partitioned further for each dc, determining the vertical pitch displacement of the pcs
assigned. The temporal space assignments for each dc are a reflection of the general
trajectory of material relations in the work, whereby the /8 dc has the least amount of
material (1-12 pcs), the /16 dc having 2-24 pcs with finally the /32 dc having the most with 4-
48 pcs. This relates back to Ex.1, whereby as one moves from A through to C/D there is a
move away and spreading out from an initial centre (A) towards an area encompassing all
possibilities (C/D). Table 3 below shows the temporal-space regions for each dc (highlighted
in red).

2
 /8 /16 /32
1-22 23-44 45-66 67-88 1-22 23-44 45-66 67-88 1-22 23-44 45-66 67-88
22 44 66 C8 88 22 44 66 C8 88 22 44 66 C8 88
21 43 65 87 21 43 65 87 21 43 65 87
20 42 C6 64 86 20 42 C6 64 86 20 42 C6 64 86
19 41 63 85 19 41 63 85 19 41 63 85
18 C4 40 62 84 18 C4 40 62 84 18 C4 40 62 84
17 39 61 83 17 39 61 83 17 39 61 83
C2 16 38 60 82 C2 16 38 60 82 C2 16 38 60 82
15 37 59 81 15 37 59 81 15 37 59 81
14 36 58 80 14 36 58 80 14 36 58 80
13 35 57 79 13 35 57 79 13 35 57 79
12 34 56 78 12 34 56 78 12 34 56 78
11 33 55 77 11 33 55 77 11 33 55 77
10 32 54 C7 76 10 32 54 C7 76 10 32 54 C7 76
9 31 53 75 9 31 53 75 9 31 53 75
8 30 C5 52 74 8 30 C5 52 74 8 30 C5 52 74
7 29 51 73 7 29 51 73 7 29 51 73
6 C3 28 50 72 6 C3 28 50 72 6 C3 28 50 72
5 27 49 71 5 27 49 71 5 27 49 71
C1 4 26 48 70 C1 4 26 48 70 C1 4 26 48 70
3 25 47 69 3 25 47 69 3 25 47 69
2 24 46 68 2 24 46 68 2 24 46 68
1 23 45 67 1 23 45 67 1 23 45 67
Table 3

Reading from left to right, one can see that the temporal-space allowance for each dc is
double that of the previous class and spreads out from the initial central 22 keys in the /8 dc,
to the 22 keys either side of the /8 dc giving the /16 dc a total of 44 keys, with finally the /32
dc having all 88 keys, partitioned into 4 regions of 22 keys (1-22, 23-44, 45-66, 67-88).

As was mentioned in the introduction above, the work itself consists of 54 bars. The 54 bars
consist of 3 dcs, each with 18 values: 2/8… 19/8, 11/16… 28/16 and 20/32… 37/32. These
54 t-sigs are grouped into 18 3-bar micro-groups, each of which contains one of each dc.
These were assigned by combining both indeterminate (random number generators) and
determinate (point gestures) systems. The outcomes of this process are shown in Table 4,
below.

3 2 6 4 5 1 5 3 2 4 6 1 4 1 5 3 2 6
16 8 32 16 32 8 32 16 8 16 32 8 16 8 32 16 8 32
8 32 16 32 8 16 8 8 32 32 16 16 32 16 8 8 32 16
32 16 8 8 16 32 16 32 16 8 8 32 8 32 16 32 16 8
Table 4

By coupling these systems one is able to create multiple assignments from only a limited set
of values, thus reducing the time it takes to assign macro- and micro-level parametric values.

The use and combination of indeterminate and determinate systems for the creation of
music has been a focus of mine and is employed on several levels in the work. As
determinate and indeterminate systems are so closely related within the work it would be
futile to discuss one without referencing the other, which is why they are clarified together.

3
Indeterminacy and Point Gestures

There are two ways in which indeterminacy is employed in this work: either random
processes alone (such as interference cluster assignments, impulse points and tuplet
placements) or random processes coupled with 3- or 4-point gestures (such as in Table 4
above or fundamental pitch-class set assignments, below). Random processes alone are
realised in two different ways and can be Markovian or non-Markovian in nature.

Markovian and non-Markovian processes often appear within the same dc, with the dc being
split into 2 groups of 9 t-sigs (1-9, 10-18). A Markovian process is one in which the outcomes
of one group are dependent on the outcomes of a previous group.

In Table 4 there are 18 columns containing one of each dc, representing the 18 macro-
partitions of the work (see previous page). The 3 dcs are assigned by coupling point
gestures with a random number generator.

The point gestures used to assign dcs are 3-point gestures (3-pgs) of which there are 6:
1 2 3 4 5 6
1
. . . . . .
2
. . . . . .
3
. . . . . .
[1,2,3] [1,3,2] [2,1,3] [2,3,1] [3,1,2] [3,2,1]
Ex.3

Each denominator class is assigned a point gesture value: /8 – 1, /16 – 2, /32 – 3. As there
are 6 3-pgs, when all of these are assigned it will set the denominator classes for 18 bars. A
different number sequence is used for bars 19-36 and 37-54. In Table 4, the shaded area in
the top row is the outcome of the random number generator, being:

1. 3,2,6,4,5,1 Bars 1-18

2. 5,3,2,4,6,1 Bars 19-36
3. 4,1,5,3,2,6 Bars 37-54

The 6 numbers correspond to the 6 3-pgs shown in Ex.3. Bars 1-18 are therefore:

3: [2,1,3] - /16, /8, /32

2: [1,3,2] - /8, /32, /16
6: [3,2,1] - /32, /16, /8
4: [2,3,1] - /16, /32, /8
5: [3,1,2] - /32, /8, /16
1: [1,2,3] - /8, /16, /32

This can be verified in the first 18 bars of the score. As the t-sigs have been removed from
the score for editorial reasons, the t-sig assignments for the first 18 bars are shown in Table
5, below. These assignments are treated as macro-level assignments, whereby what they
have assigned are fundamental to how the time-space of the work is constructed, and form
the basis from which micro-level assignments are derived.

4
 Bar T-Sig Bar T-Sig Bar T-Sig Bar T-Sig Bar T-Sig Bar T-Sig
1 17/16 4 3/8 7 33/32 10 19/16 13 32/32 16 4/8
2 6/8 5 28/32 8 26/16 11 20/32 14 9/8 17 15/16
3 21/32 6 22/16 9 2/8 12 14/8 15 14/16 18 23/32
Table 5

Whilst the use of 3-pgs is limited to the assignment of macro-level variables, the use of four-
point gestures (4-pgs) in the work takes this further by assigning both macro- and micro-level
variables.

There are 24 permutations of the numbers 1, 2, 3 and 4, of which the first 6 are shown
below.

A1 A2 A3 A4 A5 A6
1 . . . . . .
2 . . . . . .
3 . . . . . .
4 . . . . . .

[1,2,3,4] [1,2,4,3] [1,3,2,4] [1,3,4,2] [1,4,2,3] [1,4,3,2]

Ex.4

As there are 4 types of set (P, I, R and RI), 4-pgs were used to determine the fundamental
set-types for /16 and /32 dcs (the pitch-class content of the /8 dc is derived using different
principles than that of the /16 and /32 dcs so is discussed separately (see /8 Material, page
7)).

Combining the /16 and /32 classes gives 36 t-sigs. To assign fundamental sets to these
therefore requires 9 4-pgs. Each set-type is assigned a point gesture value: P – 1, I – 2,
R – 3 and RI – 4, with the 9 4-pgs assigned in two groups of 4 and a 1. Within each group of
4, each set-type appears only once. The outcome of this is:

C-16: [3,2,4,1] = R,I,RI,P

B-7: [2,1,3,4] = I,P,R,RI
A-5: [1,4,2,3] = P,RI,I,R
D-21: [4,2,1,3] = RI,I,P,R
A-3: [1,3,2,4] = P,R,I,RI
C-13: [3,1,2,4] = R,P,I,RI
B-11: [2,4,1,3] = I,RI,P,R
D-20: [4,1,3,2] = RI,P,R,I
A-4: [1,3,4,2] = P.R.RI,I

To determine the transposition level (T-level) for each bar, a chromatic scale starting on C
was projected onto the linear realisations of the /16 and /32 classes, giving:

5
 /16’s /32’s
11 R-0 20 I-6
12 I-1 21 RI-7
13 RI-2 22 R-8
14 P-3 23 P-9
15 I-4 24 I-t
16 P-5 25 RI-e
17 R-6 26 I-0
18 RI-7 27 RI-1
19 P-8 28 P-2
20 RI-9 29 R-3
21 I-t 30 RI-4
22 R-e 31 P-5
23 RI-0 32 R-6
24 I-1 33 I-7
25 P-2 34 P-8
26 R-3 35 R-9
27 P-4 36 RI-t
28 R-5 37 I-e
Table 6

The inherent efficiency of this process results in the fundamental architecture of the work
being constructed in a relatively short time period. This reduction in the workload is reflected
at the micro-level, where 4-pgs are used to assign the gestural content within /32 bars.

In Table 3 (page 3, above), the /32 dc is shown to employ the full range of the piano,
partitioned into four 22-key regions. To each of these regions is assigned a 4-pg value:

1. 67-88
2. 45-66
3. 23-44
4. 1-22

This allows the key regions in which a dyad is to appear to be determined by assigning 4-
pgs to each bar in the class. The pc content for /32 material appears in multiples of 4, giving
4-48 pcs in a given bar. As these are dyads the impulse content is 2-24, giving between 0.5
and 6 4-pgs needing to be assigned to each bar. The way in which these were derived is
similar to that used for deriving fundamental set-types. In this case, once all 24 4-pgs have
been assigned, the process starts again until all the material has been assigned. Table 7
shows the outcomes for the t-sigs 20/32 to 24/32.

T-Sig Impulse content No. of 4-pgs 4-pg assignments

20/32 24 6 C15, A5, B9, D21 C16,D20
21/32 24 6 A6, B12 D19, C17, A3, B11
22/32 22 5.5 C18, B10, A1, D24 B7, D23(0.5)
23/32 20 5 C14, A4 C13, B8, D22
24/32 20 5 A2 D19, B12, C15, A6
Table 7

The D19 in 24/32 marks the first of the next group of 24 4-pg assignments as it is the first
repeated point gesture.

Interference clusters (ics) are assigned using both Markovian and non-Markovian systems.
As mentioned above, the dc is initially split into two groups of 9: 20/32-28/32 (A) and 29/32-
37/32 (B). A is determined by a Markovian system like that used to assign 4-pgs (above) and
B has in some instances the same clusters assigned in consecutive bars, making it non-
Markovian (see Table 8, below).

6
 A B
T-Sig ic assigned T-Sig ic assigned
20/32 None 29/32 9,8,3,213,10,1,14
21/32 11 30/32 9,12,3,6,5,16,15,2,7
22/32 8, 13 31/32 6,7,12,3,10,11,4,9,8,1
23/32 16,3,6 32/32 14,15,8,13,12,3,4,9,16,7,2
24/32 15,14,1,10 33/32 15,8,1,6,5,12,11,2,7,4,13,14
25/32 5,12,7,2,9 34/32 1,10,15,8,13,2,11,6,5,4,9,12,7
26/32 4| 11,10,13,16,9 35/32 4,9,12,1,14,7,8,15,6,5,16,13,2,3
27/32 2,1,6,5,14,3,8 36/32 8,1,14,3,4,15,16,13,10,7,2,11,6,5,12
28/32 15,4,7,12| 5,16,11,6 37/32 1,14,7,6,13,4,3,12,11,2,5,10,9,8,15,16
Table 8

As the total ic-content increases, it becomes impractical for there to be a Markovian system
in place to assign ics as there is an inherent increase in invariance between successive t-
sigs in a linear realisation, given the finite set from which they can be chosen. Due to other
systems governing the global t-sig assignments of a dc (such as that shown in Table 5),
these minor similarities in content between bars is not easily noticeable to the listener.

/8 material

The pc content of 8 bars was determined by employing a system that successively adds a
semitone up from the pitch-class assigned, followed by a semitone down, until the pc
aggregate is complete. Table 9 shows how this would be realised if based on 0 (C natural):
No. of pitch-classes Pitch-classes assigned
1 0
2 0,1 (one up)
3 e,0,1 (one up, one down)
4 e,0,1,2 (two up, one down)
5 t,e,0,1,2 (two up, two down)
6 t,e,0,1,2,3 (three up, two down)
7 9,t,e,0,1,2,3 (three up, three down)
8 9,t,e,0,1,2,3,4 (four up, three down)
9 8,9,t,e,0,1,2,3,4 (four up, four down)
10 8,9,t,e,0,1,2,3,4,5 (five up, four down)
11 7,8,9,t,e,0,1,2,3,4,5 (five up, five down)
12 7,8,9,t,e,0,1,2,3,4,5,6 (six up, five down)
Table 9

This is then elaborated further by projecting P-0 onto the linear realisation of the /8 dc, using
the T-levels assigned as the fundamental pitch class for the pitch class(es) of the cluster.
With P-0 being [0,1,4,2,3,5,e,t,7,9,8,6], the result of this mapping is:

2/8 3/8 4/8 5/8 6/8 7/8 8/8 9/8 10/8 11/8 12/8 13/8 14/8 15/8 16/8 17/8 18/8 19/8

0 0 1 4 4 2 3 3 5 e e t 7 7 9 8 8 6

Table 10

7
And with T-level assignments taken into account, the resulting pitch-classes are:
No. of pitch-classes Time signature Pitch-class assignments
1 2/8, 3/8 0
2 4/8 1,2 (one up)
3 5/8,6/8 3,4,5 (one up, one down)
4 7/8 1,2,3,4 (two up, one down)
5 8/8, 9/8 1,2,3,4,5 (two up, two down)
6 10/8 3,4,5,6,7,8 (three up, two down)
7 11/8, 12/8 8,9,t,e,0,1,2 (three up, three down)
8 13/8 7,8,9,t,e,0,1,2 (four up, three down)
9 14/8, 15/8 3,4,5,6,7,8,9,t,e (four up, four down)
10 16/8 5,6,7,8,9,t,e,0,1,2 (five up, four down)
11 17/8, 18/8 3,4,5,6,7,8,9,t,e,0,1 (five up, five down)
12 19/8 1,2,3,4,5,6,7,8,9,t,e,0 (six up, five down)
Table 11

A note on tuplets

As is the case in many of my works, each dc is assigned specific tuplets (or lack thereof) as
part of its global (linear) trajectory, with each dc having specific constructive principles,
outlined below.

In the most broad and basic sense, each dc is split into 2 groups of 9 t-sigs, with the 9
smallest t-sigs assigned a tuplet each and the 9 largest assigned no tuplets, shown in Table
12, below.

/8 class /16 class /32 class

T-Sig Tup T-Sig Tup T-Sig Tup T-Sig Tup T-Sig Tup T-Sig Tup
2/8 7:4 11/8 None 11/16 2:11 20/16 None 20/32 19:10 29/32 None
3/8 7:6 12/8 None 12/16 3:10 21/16 None 21/32 18:10 30/32 None
4/8 6:4 13/8 None 13/16 4:9 22/16 None 22/32 17:11 31/32 None
5/8 5:5 14/8 None 14/16 5:8 23/16 None 23/32 16:12 32/32 None
6/8 5:6 15/8 None 15/16 6:7 24/16 None 24/32 15:12 33/32 None
7/8 4:7 16/8 None 16/16 7:6 25/16 None 25/32 14:13 34/32 None
8/8 3:8 17/8 None 17/16 8:5 26/16 None 26/32 13:14 35/32 None
9/8 3:9 18/8 None 18/16 9:4 27/16 None 27/32 12:14 36/32 None
10/8 2:10 19/8 None 19/16 10:3 28/16 None 28/32 11:15 37/32 None
Table 12

Each dc is assigned distinct constructive principles in regards to tuplet placement and

realisation:

 /8 class tuplets last for the full duration of the bar, partitioning it into successively
larger and fewer partitions.
 The /16 class begins by encompassing the whole bar, but where it differs from the /8
class is that non-tuplet points are successively added either side of the tuplet as-
signed, whilst the area in which the tuplet appears is reduced.
 The /32 class has its tuplets successively reduce in total impulse content whilst suc-
cessively increasing the time-space (in demisemiquaver units) in which it is realised.

The basic realisations of the above bullets are shown in Ex.5, below.

8
Ex.5

A Markovian system was used to assign the tuplet positions in the /32 class and impulse
points within each class.

/8 class: As the total number of possible impulse points is successively reduced as the t-sig
becomes larger, there becomes a point in 10/8 where there is only one possible impulse
point. The impulse point assignments are non-repetitive so Markovian in nature, with the
outcomes being 3,5,4,2,5,3,2,3,2. For 11/8 to 19/8 a non-repetitive random number
generator was used to determine the impulse point (in quaver units) of the cluster assigned.

/16 class: The impulse/pc content for 11/16 to 19/16 is 2,2,4,6,6,8,10,10,12, with as many of
these impulses as possible being assigned within the tuplet. When there are fewer
impulses/pc than there are tuplet points, the first is omitted (as is the case in the 12/16 bar in
the score). When there are more than there are within the tuplet, a non-repetitive random
number generator was used to assign which of the non-tuplet impulse point(s) are assigned.
For 20/16 to 28/16, a non-repetitive random number generator was used that assigned the
omitted impulse points.

/32 class: The impulse points at which the /32 class tuplets are assigned is determined by a
random number generator, giving: 3,7,9,6,1,2,5,8,4 (20/32-28/32). 29/32-37/32 has no
tuplets assigned. All impulse points are assigned using a random number generator, with the
numbers assigning omitted impulse points for the first 9 t-sigs (20/32-28/32) and for 29/32-
37/32, the outcomes assign impulse points used

In order for there to be no simultaneities, ic impulse points are assigned to the impulses
omitted from the /32 assignments. These are also assigned using a non-repetitive random
number generator, whereby consecutive t-sigs in a linear realisation aren’t assigned the
same impulse point values.

9
Summary

Two-Slit Experiments laid the foundations from which the use of point-gestures to derive
macro- and micro-level parametric assignments became a key constructive device in my
works. Stemming from initial inquiries into Gordon Downie’s Piano Piece 3, where Downie
uses 3-pgs to determine the sonic contours of trichordal arrays within the work (see pages 2
and 3 of linked PDF), it became evident to me that this mode of production and realisation
can be taken further to incorporate all macro- and micro-level variables pertaining to the
construction of a musical object, providing a means by which to create works that had an
inherent structural and conceptual integrity, with the various sections of the work relating
back to a relatively small set of 3-pgs and/or 4-pgs.

The works that follow this piece (Entanglement, Objects: Object Distribution, Objects 2:
Four-Point Gestures, Objects 3: Vocalised Objects (plus its satellite works Objects 3a-d) and
Objects 4a: Solo piece 1 all build upon the systems first extensively used in Two-Slit
Experiments, incorporating them on every level of the works’ construction, leading to a
reduction in the use of stochastic processes to the point where in Objects 4a there are none.

10