How these five games are related
Four of the five games on this site are cousins. They descend, with a great deal of argument about the details, from a single ancestor. The fifth shares nothing with them but the idea of a board.
Updated
One game, somewhere in northern India
The oldest game anyone can point to as the ancestor of chess is chaturanga, played in northern India by about the sixth century of the common era. The name is a Sanskrit word for an army of four parts: foot soldiers, horse, elephants and chariots. Those four parts, plus a king and a counsellor, are the shape every later game in this family still keeps.
That is roughly where the certainty stops. No rule book survives from the period. What survives is a handful of literary references, a scatter of archaeological pieces that may or may not belong to the game, and a set of later texts written by people who were themselves guessing about the past. Everything reconstructed about chaturanga is inference from what its descendants have in common, and the descendants do not agree.
West: shatranj, then chess
The game reached Persia, where it became shatranj and acquired the vocabulary that still echoes in English. Shah is the king. Checkmate is a worn-down version of a Persian phrase for the king being defeated. The counsellor beside him was the ferz, a piece so weak it could only shuffle one square diagonally.
After the Arab conquest of Persia, shatranj became a written game. Arabic-speaking players produced problem collections, endgame studies and treatises on opening arrays, and it is from those manuscripts that we know the mediaeval rules in detail. Shatranj is the earliest form on this site whose rules are genuinely attested rather than reconstructed.
It entered Europe by more than one road: across the Mediterranean through Islamic Spain and Sicily, and overland through Byzantium and the Rus. For four hundred years European chess was recognisably shatranj with different pieces on it. Then, in the last quarter of the fifteenth century, players in Spain and Italy gave the ferz the run of the board and the alfil a full diagonal. The counsellor became the queen, the elephant became the bishop, and the game roughly doubled in speed.
The rest is tidying up. Castling, the pawn double step and en passant settled over the following two centuries. Stalemate stopped being a win and became a draw. Written rules were standardised in the nineteenth century by clubs and then by the international federation, which is the version most people mean when they say chess.
East: xiangqi, janggi, shogi
The eastern line is harder. Xiangqi, the Chinese game, is played on the intersections of a grid rather than on squares, cuts the board with a river, and locks each general inside a nine-point palace he may never leave. It has a piece nothing in the western game has: a cannon that moves like a chariot but can only capture by leaping over exactly one other piece.
Two accounts compete. One says xiangqi is chaturanga arriving overland and being rebuilt to Chinese taste. The other says the Chinese game has native roots in older board games and merely borrowed pieces. The cannon is a useful clue in one direction at least: whatever it depicts, it cannot be older than the siege weapons it is named for, which dates the modern form rather than the game.
Janggi, the Korean game, sits on the same grid as xiangqi and is plainly derived from it, but the changes are deliberate rather than accidental. The river is gone, so the elephants and the soldiers are free of it. The palace diagonals are real lines that pieces travel along. The cannon must leap to move at all, not merely to capture. And before the first move each player chooses how to arrange his own horses and elephants, which makes the very first decision of the game a matter of taste.
Shogi, the Japanese game, took a different turn entirely. Its board is nine squares by nine, its pieces are flat wedges marked with characters rather than carved figures, and — the change that matters — a captured piece changes sides. It goes into the capturer’s hand and can be dropped back onto the board as a friendly man. Nothing is ever permanently removed, so a shogi game does not thin out into an endgame the way a chess game does. It gets sharper.
And then there is tafl, which is not related at all
Hnefatafl and its relatives were played across Scandinavia, Ireland, Britain and the Sami north before chess arrived, and they have no connection to the Indian game whatever. Grouping them here is a convenience of this site, not a claim about ancestry.
Tafl is asymmetric, which no game in the chess family is. One side has a king and about a dozen men and starts in a block at the centre of the board. The other side has roughly twice as many men and starts around the edges. The defenders win by getting the king out. The attackers win by pinning him down. The two players are not playing the same game, and a fair rated result needs two games with the sides swapped.
If tafl has an ancestor it is probably the Roman game of soldiers, ludus latrunculorum, carried north along trade and military routes — but that is a hypothesis built on a family resemblance in the capture rule, not a documented descent. When the Norse world took up chess in the later Middle Ages, tafl faded, and no complete set of its rules was ever written down.
How to read the diagram on this site
The lineage diagram on the front page draws five arrows. Each one means "most scholars think this game is downstream of that one", and each one has a literature arguing about it. It is a map of a consensus, and a consensus is not a proof.
| Line | Confidence | Where the argument is |
|---|---|---|
| Chaturanga to shatranj | High | The Persian and Arabic sources describe the transmission themselves. |
| Shatranj to chess | High | Continuous European evidence; only the timing of the fifteenth-century reforms is contested. |
| Chaturanga to xiangqi | Contested | Rival case for native Chinese roots; the cannon and the river have no Indian parallel. |
| Xiangqi to janggi | High | The board is the same board; the differences read as reform, not as independent invention. |
| Chaturanga to shogi | Moderate | The route through China or Korea is not settled, and the drop rule is a Japanese invention with no ancestor at all. |
| Tafl | Unrelated | A separate northern European tradition. Nothing in common but the board. |
Each game on this site has its own history page, which goes further into that game’s evidence and says where it runs out.
Common questions
- Which came first, chess or xiangqi?
- Neither, on the evidence we have. The Indian game is the older attested one, but the case that xiangqi descends from it rather than from native Chinese board games is a reconstruction from resemblances, not a documented transmission. Anyone who tells you the question is settled is telling you which side of the argument they are on.
- Is hnefatafl related to chess?
- No. The tafl games were played in northern Europe before chess arrived there, they are asymmetric where every chess-family game is symmetric, and they have no piece, rule or vocabulary in common with the Indian line. They are on this site because they are a board game tradition worth playing, not because they are cousins.
- Why does shatranj feel so slow?
- Because its two diagonal pieces are nearly useless. The counsellor moves one square diagonally and the elephant leaps to one of eight squares in the whole game. Without a long-range queen or bishop, attacks take many more moves to build, which is exactly why fifteenth-century European players changed the two pieces and got modern chess.
- Do any original rules for these games survive?
- For mediaeval shatranj, yes — the Arabic manuscripts describe it in detail. For the modern Asian games, yes, in the shape of twentieth-century federation rule books. For chaturanga and for every tafl game, no. Those are reconstructions, and the pages for them on this site say so and say where the reconstructions disagree.
Sources
Everything on this page is written from scratch. These are the works it leans on, and what each one settles.
H. J. R. Murray, A History of Chess (1913)
Still the standard survey of the whole family, and the source of most of what is confidently said about mediaeval shatranj. Murray worked from the Arabic manuscripts directly. His readings of the Indian material are the ones most often challenged since.
H. J. R. Murray, A History of Board-Games Other Than Chess (1952)
Where the tafl games are surveyed, and where the ludus latrunculorum hypothesis is set out as a hypothesis.
Andreas Bock-Raming, on the Indian sources
A later re-reading of the Sanskrit material that disagrees with Murray on several details of early Indian play, including how the elephant moved. This site exposes that disagreement as a setting on its chaturanga preset rather than picking a winner.
Carl Linnaeus, journal of the Lapland journey (1732)
The only account of a tafl game written by someone who watched one being played. It is our single best evidence for tablut, and it still leaves several rules unstated.