At the base of a rock-hewn church in Lalibela, Ethiopia, shallow oval depressions are cut into the volcanic tufa floor. They are not ritual basins or drainage channels. They are mancala holes, twelve in two parallel rows, each no wider than a thumbprint, worn smooth by centuries of finger-rubbed seed movement. Nearby, in a monastery library in Mysuru, India, a palm-leaf manuscript from the 14th century sketches the same configuration, but with annotations in Kannada about turn order and capture conditions. Between these two points lies over 5,000 kilometres, several mountain ranges, and more than six hundred years. Yet the board layout is nearly identical. What moved so quickly across such distances was not grain, nor cloth, nor even coin, but the idea of a board itself.
This pattern repeats: a standardised grid or sequence of pits appears far from its earliest known context, often centuries before any surviving rule text names it. The board travels first. The rules follow, slowly, unevenly, sometimes never arriving at all. To trace that movement is to map not just leisure, but contact, between merchants and monks, pilgrims and porters, scribes and soldiers, along the ancient networks that stitched together Africa, Asia, and the Mediterranean.
Stone and seed: the material evidence of mancala history
Mancala is not one game but a family of count-and-capture games sharing a common architecture: a board with two or more parallel rows of pits, filled with seeds, pebbles, or cowrie shells. Over two hundred regional variants exist today, from Oware in Ghana to Bao in Zanzibar to Sungka in the Philippines. Their shared structure suggests deep historical connection, but the physical evidence for that connection is not parchment or papyrus. It is stone. In Ethiopia’s highland churches, mancala boards appear as simple pairs of rows carved directly into flagstones, some dated by stratigraphy to the 12th century. In Yemen’s ancient port city of Aden, similar pits are incised into the thresholds of merchant houses built before the 10th century. In Karnataka, India, temple walls bear relief carvings of mancala boards beside depictions of chess and dice, their pits rendered as precise hemispherical hollows in soapstone.
These stone versions endure because they require no maintenance, no replacement parts, no literacy. A trader could replicate the layout in sand or ash using only memory and fingertip measurement. That durability explains why mancala boards appear earlier and farther afield than any associated rule set. The earliest known written description of mancala play comes from a 17th-century Arabic treatise in Cairo, centuries after the Ethiopian and Indian stone examples, and even then, it describes only one variant, with no claim to universality. The board, meanwhile, needed no translation. Its logic was spatial, tactile, immediate: count forward, drop one per pit, capture when the last seed lands in an occupied pit opposite. No script required.
The chessboard’s slow migration along the Silk Road
Chess offers a starker contrast. Its ancestor, chaturanga, emerged in northern India around the 6th century CE, not as a carved floor pattern, but as a wooden board with painted squares and figurative pieces representing infantry, cavalry, elephants, and chariots. Unlike mancala’s minimal pits, chaturanga demanded precision: an 8×8 grid, distinct piece types, and layered movement rules. When it reached Persia by the early 7th century, it became shatranj, retaining the board but simplifying the elephant into a piece that moved diagonally two squares, and the chariot into a rook-like mover. Persian traders carried shatranj westward; by the 9th century, it appeared in Baghdad, where scholars began writing systematic analyses of openings and endgames.
Yet the board itself changed little during this journey. Excavated shatranj sets from Samarkand, ivory pieces found alongside a lapis-inlaid wooden board, show the same 8×8 geometry as Indian chaturanga reliefs at the Ajanta caves. The consistency suggests that the board’s proportions were tied to its function: the grid enabled calculation of relative force and control of space. Altering square size or row count would disrupt the balance between mobility and confinement that made the game intellectually coherent. So while rules evolved, Persian players dropped the dice-based element of early Indian play, and later Arab theorists formalised checkmate conditions, the board remained stable. It was the fixed reference point across languages, scripts, and legal systems.
Why boards outpace rules in ancient games trade routes
The asymmetry between board and rule transmission rests partly in materiality, partly in use. A board is a tool: portable, reusable, legible without instruction. A 3rd-century BCE terracotta mancala board excavated near Merv (in modern Turkmenistan) fits neatly in one hand, its twenty-four pits arranged in four rows of six, distinct from the Ethiopian two-row form, yet clearly related by proportion and spacing. It could be carried in a saddlebag, set on a caravan rest-stop, and played within minutes. No explanation needed. Rules, by contrast, depend on shared language, pedagogical context, and social permission. Teaching oware in coastal Ghana involved elders demonstrating capture sequences during dry-season evenings; teaching shatranj in Abbasid Baghdad required literate tutors, paper, ink, and time, resources concentrated in urban centres and religious institutions.
This disparity created pockets of divergence. In Madagascar, the game of katro uses a board with three rows of nine pits, structurally closer to East African bao than to West African oware, but its capture rules resemble those of Southeast Asian congkak. That hybridity likely reflects maritime contact between Swahili coast traders and Austronesian sailors, rather than direct Indian Ocean transmission. The board arrived first, via ship timbers or woven mat patterns; local players then adapted existing counting logic to fit it. Such adaptations rarely leave written traces. They survive only in oral practice, or not at all. Which means the archaeological record privileges the board: durable, silent, and endlessly reproducible.
Temple floors and trade posts: reading geography in game layouts
A board’s physical layout encodes clues about its origin and adaptation. Consider the Ethiopian mancala variant called gabata. Its board has two rows of six pits plus two larger end pits, called ‘stores’, cut into the same stone as the playing surface. This extended layout appears nowhere else in Africa but shows strong parallels with the giuthi boards of central Kenya, which also feature enlarged end pits. Both regions sit along historic salt-trade corridors linking the Afar Depression to the Great Rift Valley. Salt was carried in woven baskets lined with animal hide; the gabata board, small enough to fit inside such a basket, may have served as both pastime and tally device, counting loads, measuring portions, or settling disputes mid-journey. Its elongated shape suits horizontal placement on a donkey pack-saddle, unlike the compact square boards of Persian shatranj, designed for tabletop use in settled dwellings.
Similarly, the 12×12 grid of the Cambodian game ouk chaktrung mirrors the dimensions of traditional rice-storage granaries in the Tonlé Sap region, structures built with precisely spaced bamboo supports. That correlation suggests the board may have originated as a schematic overlay for managing grain distribution, later abstracted into competition. These material resonances are not coincidental. They reflect how games emerge from infrastructure: road widths dictate board portability; storage units suggest grid proportions; climate determines material, sand for Sahelian play, polished wood for monsoon-cooled temples, slate for highland churches. The board is not neutral ground. It is shaped terrain.
When the rules finally arrive
Rules do eventually catch up, but rarely as wholesale imports. They arrive piecemeal, filtered through local priorities. The Arabic word shatranj entered Persian as shātrang, then Turkish as şatranç, each shift accompanied by subtle rule shifts: Persian players introduced the concept of ‘check’, but not ‘checkmate’ as a forced win; Turkish variants later added castling-like moves for the king and rook. In West Africa, Portuguese traders recorded oware being played with strict prohibitions against ‘sowing’ into empty pits on the opponent’s side, a restriction absent in early Arabic descriptions. That innovation likely arose from competitive play in village squares, where spectators enforced fairness through consensus, not codified law.
One telling example comes from the Maldives. A 19th-century British colonial survey notes islanders playing a mancala variant on coral-stone boards with thirteen pits per row, unusual, since most Indian Ocean forms use twelve or fourteen. Local elders explained the number referred to the thirteen atolls of the archipelago, each pit symbolising a jurisdiction. Here, the board had become a political map; the rules, though unrecorded, were embedded in communal memory. No manuscript preserved them. No foreign trader transcribed them. They lived only in repeated action, in the rhythm of seed-drop, the pause before capture, the nod of agreement between players. That kind of knowledge does not travel well. It stays put. Which is why, when we find matching boards across continents, we are seeing not uniformity, but conversation: fragmented, persistent, and written in dust, stone, and seed.