2. Sub-Saharan Africa: Bell Patterns and Polymetric Layering

If there is a single tradition that Poly was designed to model, it is the polymetric drumming of West and Central Africa (Kubik 1999 traces rhythmic practice across this belt and into the diaspora that Chapter 4 picks up). The principle of layering independent rhythmic cycles — each with its own length, density, and social function — did not arrive in Western music theory until the twentieth century. In the Ewe communities of southern Ghana it is the organising logic of ensemble music. The practice has been documented in the ethnomusicological literature since the mid-twentieth century — A. M. Jones’s two-volume Studies in African Music is built largely on Ewe material⁶ — and Locke’s analysis of Southern Ewe dance drumming sets out the offbeat-timing and cross-rhythm principles this chapter builds on (Locke 1982).

Theory deep dive

This chapter stays broad and practical. For the explicit counterpoint rules behind it — what must stay fixed, what may vary and by how much, and what breaks the idiom — see Timeline Counterpoint: Sub-Saharan Africa.

The Gankogui Bell: E(7,12)

Every Ewe drum ensemble begins with the gankogui, the double iron bell. The bell player establishes a fixed timeline pattern that does not vary. Other drummers orient their parts relative to it, entering at different points, filling different rhythmic spaces. The bell does not lead in a melodic sense — it defines the temporal grid against which everything else is measured.

The standard gankogui pattern distributes seven strokes across twelve pulses. It is Euclidean rhythm E(7,12), which Toussaint describes as the most commonly used timeline pattern in sub-Saharan Africa — a descriptive claim about how often it appears, not a ranking of importance.¹

E(7,12)

Notice the asymmetry. The seven hits do not divide twelve evenly — they create an uneven grouping of long and short intervals (2-1-2-2-1-2-2). This asymmetry is not an accident. It is what gives the pattern its characteristic tension, the sense that it is always pulling forward toward the next downbeat without ever quite resolving.

Role Separation: Timeline, Support, Lead

West African ensembles organise their parts into functional layers. Understanding these roles is the key to building convincing polymetric patches in Poly.

The timeline (bell) is fixed and metronomic. It never changes. In Poly, this is your anchor lane — high velocity, no mutation, no phrase gating. To lock a specific bell pattern into place so the engine cannot modulate it, use timeline mode; see Chapter 18 (Editors and Advanced Views) for how to enter timeline mode and draw the pattern step by step.

The support drums play shorter, sparser patterns that interlock with the bell. In Ewe drumming, the kidi and sogo fill the spaces the bell leaves open. These parts respond to the timeline but remain relatively stable. In Poly, these lanes use fewer hits and lower velocity, with modest mutation values if you want subtle variation.

The lead drum (atsimevu in Ewe tradition) improvises. It plays across and against the other parts, creating rhythmic tension and release. In Poly, you model this with higher mutation percentages, phrase gating, and probability — the lead lane should breathe and vary while the timeline holds steady beneath it.

Polymetric Stacking

The power of the Ewe approach is that support drums often run at different cycle lengths from the bell. A 12-step bell layered with a 7-step support pattern creates a composite rhythm that does not repeat for 84 pulses. Add a 5-step third voice and the full cycle extends to 420 pulses — over thirty bars at 1/8 subdivision. The groove is always familiar but never quite the same.

E(5,12)

E(5,12) distributes five hits across twelve steps — a sparser, more open pattern that leaves room for the bell’s density.

E(3,12)

E(3,12) is the sparsest of the three, striking only on beats 1, 5, and 9. It provides a deep structural pulse beneath the busier patterns above.

Ewe-Inspired Polymetric EnsembleSub-Saharan: Agbekor

Lane	Role	Steps	Hits	Rotation	Subdivision	Timeline	Note	Velocity	Ghost	Kotekan
1	Bell timeline	12	7	0	1/8	on	56	110	0	off
2	Dance beat	12	4	0	1/8	off	41	95	0	off
3	Support drum	12	3	0	1/8	off	43	90	40	off
4	Responding drum	5	3	0	1/8	off	45	85	50	off
5	Lead complement	7	5	2	1/8	off	47	75	60	L1

Ewe Polymetric EnsembleEwe-inspired polymetric ensemble — gankogui bell timeline with kidi, sogo, and lead drum

PlayTry it

Lane 4 uses the Kotekan setting L1, which generates a complementary pattern interlocking with Lane 1’s bell. This mirrors the way a support drummer listens to the bell and fills its gaps. Set mutation to 15-25% on Lane 4 to give it the slight variability of a responding part.

What to Listen For

The 5-step and 7-step lanes create shifting accents against the steady 12-step bell. Pay attention to the moments of convergence — when all three cycle lengths align on the same pulse. These rare unison points feel like structural downbeats, and the tension between them is what drives the groove forward.

Manding Traditions: Djembe Ensembles

West of the Ewe heartland, the Manding traditions of Mali, Guinea, and Senegal build their ensembles around the djembe. Where Ewe drumming stacks many independent cycles, Manding patterns typically lock to a shared cycle length but vary in density and timbre. The dunun bass drums provide a sparse timeline, the sangban plays a mid-density response, and the djembe soloist improvises over both.

A simplification

Presenting Manding ensembles as sharing one cycle length is a pedagogical simplification, not a description of the whole tradition. Charry’s survey of Maninka and Mandinka practice (Charry 2000) treats jembe drumming as one of four distinct spheres, and is where to go for what the tradition actually contains. Manding dunun ensembles do in many contexts play parts of distinct cycle lengths, and the relationship between dunun, sangban and kenkeni varies by repertoire and region. The shared cycle is used here because it is the shortest route to a playable patch that sounds idiomatic; Charry (2000), Mande Music, is where to go for what it leaves out.

You can model this in Poly by keeping all lanes at the same step count but varying hits and velocity. The effect is closer to hocketing — rhythmic interlocking within a single metric frame.

Manding Djembe Ensemble

Lane	Role	Steps	Hits	Rotation	Subdivision	Note	Velocity	Ghost
1	Dunun bass	8	3	0	1/8	36	100	0
2	Sangban mid	8	5	1	1/8	43	85	45
3	Djembe lead	8	7	0	1/8	50	95	55

Manding DjembeManding djembe ensemble — three lanes sharing an 8-step cycle with varying density

PlayTry it

Building Improvisatory Feel

Traditional lead drumming is never fully random — it follows shared vocabulary and responds to dancers and other musicians. In Poly, you approximate this not by cranking mutation to 100%, but by combining moderate mutation (15-30%) with phrase gating. Set a phrase length of 4-8 beats with a gap of 2-4 beats on the lead lane, and the pattern will breathe — playing for a phrase, resting, then re-entering with variation. The timeline and support lanes, ungated, continue without interruption.

This call-and-response structure is fundamental to West African music. The bell calls. The support responds. The lead converses with both. In Poly, that conversation emerges from the interaction of independent Euclidean cycles, and no two bars are quite the same.