Where xiangqi came from

A shared ancestor, or a separate invention?

Every game in the chess family shares a skeleton. Two armies face each other. Each has one piece that must be protected and whose loss ends the game. Each has a piece of unlimited straight-line reach, a piece that hops in a bent path, and a rank of weak forward-only footmen. Xiangqi has all of that, and so do western chess, shogi and the games of Southeast Asia. A shape that specific, repeated across thousands of miles, is not a coincidence.

The usual explanation is that all of them descend from chaturanga, played in northern India by around the sixth century, which spread west through Persia and east along the trade routes. H. J. R. Murray argued this case at length in A History of Chess in 1913, and his framework still organises most of the discussion a century later. On that reading, xiangqi is a Chinese redesign of an imported game.

A number of Chinese scholars have argued the opposite: that xiangqi grew from native Chinese board games and divination boards, and that the resemblance to chaturanga is either shallow or runs the other way. The honest position is that the transmission is not documented. Nobody has a text describing the game arriving, or a board that can be dated to the moment of contact. What we have are two games that look like relatives and a gap in the record between them. Anyone who tells you the question is settled is overstating the evidence in one direction or the other.

What the board itself suggests

The physical design of xiangqi is not a variation on a chequered eight by eight board; it is a different object. The pieces stand on intersections, not in cells. The board is nine files by ten ranks, an odd number of files so that there is a true centre line, and an even number of ranks so that the two armies are symmetrical about the middle. It is divided by a river and marked with two fortresses.

Playing on intersections is the Chinese board convention. It is how weiqi is played, and how several older Chinese games were played, long before any chess-shaped game appears in the record. That is a genuine argument for local design: whatever the ancestry of the pieces, the board is Chinese in its bones.

The river and the palace are game-design choices with no counterpart in the Indian game. They give xiangqi its character. Confining the general and its advisors to nine points makes the endgame technical rather than a chase; forbidding the elephants to cross the river permanently splits each army into an attacking half and a defending half. These are not accidents of transmission. Somebody designed them.

The texts, and how much weight they bear

The word xiangqi appears in Chinese sources well before the modern game existed. That is the first trap. An early mention of a game by that name is not evidence that the game was this game, any more than a medieval reference to football tells you about the modern sport.

The most quoted piece of evidence is a story in the Xuanguai lu, a Tang dynasty collection of strange tales usually attributed to Niu Sengru in the ninth century. In it a traveller dreams of a battle between miniature armies, and the account names ranks that sound like chess pieces manoeuvring in a describable way. It is regularly presented as proof that a chess-like game with a cavalry piece was played in Tang China.

It is worth being careful here. This is a supernatural anecdote in a collection of supernatural anecdotes, written to entertain. Its author was not describing a game he was playing; he was describing a dream in a story. A literary allusion can tell you that a game was familiar enough to be recognisable to readers, which is a real fact and worth having. It cannot tell you the rules, the board size, or which pieces were on it. Dating a game from a ghost story is the weakest kind of argument, and the confident reconstructions built on this passage go far beyond what the passage supports.

Archaeology gives firmer ground. Round pieces with characters on them, of the kind still used today, turn up in datable contexts, and the more of them that are found the tighter the window becomes. But loose pieces without a board tell you a game existed, not how it was played.

The cannon, the Song, and the modern rules

One piece dates the game, and it is the cannon. No other chess in the family has a piece that captures by leaping a screen, and the cannon is unmistakably named after a siege weapon. Whatever kind of engine the name first referred to, a piece called a cannon cannot have been designed before China had cannon to name it after. That places the invention of this piece in the era of gunpowder siege weapons and not before, which is centuries after the earliest chess-like games are supposed to have reached China.

This is the strongest single argument in the whole subject, and it cuts cleanly. Whatever came before, the game we call xiangqi today, with cannons on the board, is a product of roughly the Song period. By the end of the Song the shape is recognisable: the nine by ten board, the river, the palace, the seven piece types, the opening array we still use. From that point onward the record is continuous. Manuals of openings and endgame studies survive, some of them still studied, and the game becomes a subject of serious literature rather than incidental mention.

What was not fixed for a long time was the awkward part: what to do about a player who checks forever, or chases a piece round the board forever, to avoid losing. Local traditions differed, and disputes were settled by the judgement of whoever was refereeing. The twentieth century brought national and then international federations, and with them a written code. The Asian Xiangqi Federation rules, used for international play, set out which repeated sequences are forbidden and who is penalised, so that a game between players from different traditions can be decided by rule instead of by argument. That code is what the federation preset on this site follows.

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 (Oxford, 1913)

  The foundational comparative history of the chess family. It settles the framing used on this page: that xiangqi belongs to a group of related games rather than standing alone, while leaving the direction of transmission genuinely open.

• Asian Xiangqi Federation, tournament rules

  The modern written code for international play. It settles the last section of this page: the repetition and chase judgements were codified by federations in the twentieth century, not inherited from the Song.

• Xuanguai lu, attributed to Niu Sengru (ninth century)

  The Tang collection of strange tales containing the dream-battle story. Cited here as the example of literary evidence and its limits, not as a description of the rules.