Sit down at a chessboard today — in Reykjavík, Manila, or a Renaissance faire in Pennsylvania — and you are playing the great-grandchild of a game that took shape in northern India roughly fifteen centuries ago. Its name was chaturanga, and from it descends not only the Western chess we know but a whole family of games across Asia. Here is where it came from, what the pieces meant, and how it was most likely played — along with its wilder four-player cousin, chaturaji.
A game named for an army
The word chaturanga is Sanskrit for “four limbs” or “four divisions” — chatur (four) and anga (limb). It did not originally describe a game at all. It was a term for the four branches of an ancient Indian army: infantry, cavalry, chariots, and war elephants. That military image is the whole key to the game. Each side fields the four arms of a real battlefield force, drawn up behind its king and his counsellor. Chess, in other words, began as a model of war built on the armies of Gupta-era India.
How old is it, really?
Here honesty matters, because the history of ancient games is littered with confident guesses. The earliest solid evidence for chaturanga comes from northern India during the Gupta Empire, in about the sixth century CE. The name itself is first clearly recorded in Bana’s Harshacharita, written around 625 CE, while Subandhu’s romance Vasavadatta, dated somewhere between the fifth and seventh centuries, carries an early allusion to the game. You will sometimes read that chess is thousands of years older — tied to the Indus Valley city of Lothal, for instance — but those claims are unproven and rest on pieces whose purpose is unknown. The defensible statement is the modest one: a recognizable ancestor of chess existed in India by the 500s–600s CE.
The board and the pieces
Chaturanga was played on the ashtapada, an 8×8 board of sixty-four squares — but with no light-and-dark checkering. The alternating colors we take for granted were a later European addition. The ashtapada was an older Indian board, apparently first used for race games, which chaturanga simply borrowed.
Each player commanded six types of piece, and every one of them survives, transformed, on a modern board:
- Raja (king) — moved one square in any direction, exactly like today’s king.
- Mantri or senapati (counsellor / general) — a weak piece that moved only one square diagonally. This is the ancestor of the queen, which gained its sweeping modern move only in late-medieval Europe.
- Ratha (chariot) — moved in straight lines like the rook. (“Rook” comes from the Persian rukh, chariot.)
- Gaja or hasti (elephant) — became the bishop. In between it was the Arabic alfil, “the elephant,” which is why the bishop is still called an elephant in many languages.
- Ashva (horse) — the knight, and it moved then as now, in its unmistakable L.
- Padati or bhata (foot soldier) — the pawn.
Reconstructed rules of play
We do not have a Gupta-era rulebook. What follows is the reconstruction accepted by most historians — above all H. J. R. Murray, whose monumental A History of Chess (1913) remains the standard reference — pieced together from later Sanskrit, Persian, and Arabic sources. Treat the broad strokes as reliable and the fine detail as scholarly best-guess.
- Setup: two players, the pieces arranged much as in modern chess — king and counsellor in the center of the back rank, flanked outward by elephant, horse, and chariot, with the row of foot soldiers in front.
- Movement: king, horse, and chariot moved essentially as their modern descendants do. The counsellor crept one diagonal square at a time. The elephant’s move is the least certain point — the most common reconstruction has it jumping exactly two squares diagonally, leaping over the square between.
- Pawns: one square forward, capturing one square diagonally, with no initial two-square option. A pawn that reached the far side promoted — but only to a counsellor, the weak piece, not to a powerful queen.
- Winning: as in chess, by trapping the enemy king. Because the counsellor and elephant were so feeble, games unfolded more slowly and with more maneuvering than the modern sport.
Chaturaji: the four-player game of dice and kings
There was also a striking four-handed version, chaturaji (“four kings”). The Persian scholar al-Biruni described it in detail around 1030 CE, and in his account it was partly a game of chance: players threw dice to decide which piece they were allowed to move — one throw freed the pawn or king, another the horse, another the elephant, and so on. Four small armies in red, yellow, green, and blue fought around the board, turns passing clockwise, and players scored points for the pieces they captured.
For a long time scholars had this backwards. The Cox–Forbes theory — begun by Hiram Cox in the late 1700s, elaborated by Duncan Forbes, and pushed further by Stewart Culin — argued that the four-player dice game was the original chaturanga and that two-player chess descended from it. Murray dismantled that idea in 1913, and historians today side with him: the two-player game came first, and chaturaji is a later, dice-driven offshoot. It is a fine game in its own right — just not the parent.
From one Indian game to the world
Chaturanga traveled. Carried into Sassanid Persia, it became chatrang; after the Arab conquests it was reshaped into shatranj, the medieval game that spread across the Islamic world and into Europe, where it slowly evolved into the chess played today. (If you want the next link in that chain, see our companion post on how to play Shatranj.) Other branches of the same family tree are debated by scholars but unmistakably related — Chinese xiangqi, Japanese shogi, Korean janggi, Thai makruk. From one war game on an uncheckered Indian board grew a worldwide tradition.
At Ancient.Games we handcraft the living branches of this family tree — from Western chess sets to Japanese Shogi and Chinese Xiangqi. Explore the collection and hold a piece of a 1,500-year-old game in your own hands.
Hero image: early ivory chess piece, 7th–8th century, The Metropolitan Museum of Art (accession 64.262.1), Open Access (CC0).