Coan Ki

The name of this Chinese game means "the bottle game", and reflects the traditional shape of the pieces. It is a race game with a slight resemblance to backgammon, but it is played by completely different rules. The rules are only partially recorded, but some credible attempts have been made at reconstruction.

Players aim to race their pieces anti-clockwise around the board in response to the rolls of two dice. The game is played for a stake, and a certain throw incurs a financial penalty on the player. It was first recorded in the west in the seventeenth century.

History of Coan Ki

By the late nineteenth century, the American anthropologist Stewart Culin studied the game, but could not find any Chinese native who had ever heard of it. This may suggest that, whatever its age, the game had died out two hundred years after Hyde was told of it.

Hyde's account is incomplete and is written in vulgar Latin. This has meant that some effort has to be made into reconstructing the game. The English game historian R. C. Bell published a simple but workable reconstruction in 1969. Jack Botermans, a Dutch game historian, published a more elaborate reconstruction twenty years later, and a revised version in 2008.

Rules for Coan Ki

1. The game is played on a board of eight rows. Each row is notionally split into two, one end belonging to each player. There are sixteen reversible pieces per player, and two dice to control movement.

2. Players sit opposite one another, at the ends of the marked rows. They each pay an agreed stake into a pot.

3. The game begins with all pieces on the board, each player having eight pieces at his end of each of the two rows furthest to his left.

4. Each player rolls one die, re-rolling if the two dice show the same number. The player who rolls highest has the first turn, and takes the value shown on the two dice as his first throw.

5. The direction of movement is anti-clockwise around the board, first from row to row until the player's last row is reached, then along that row (counting as one space) to the opponent's end, then back along the opponent's rows until their last row is reached, then back along that row back to the player's own side.

6. A throw of double one does not allow a piece to move. Instead, the player must pay a penalty to the opponent of 10% of the stake. As a compensation he removes one of his pieces from the board, meaning that fewer pieces have to complete the course.

7. A throw of two consecutive numbers, say 5 and 6, allows the player to (a) move one piece by the lower number and the other by the total (5 and 11 in this example), or (b) move both pieces by one less than the total, and then one of them a further space (10 and 11 in the example).

8. A throw of doublets, e.g. 4 and 4, allows the player to (a) move one piece by half of the total (in this example 4), or (b) move two pieces each by the full total (in this example 8).

9. A throw that does not consist of a doublet, or of two consecutive numbers, allows the player to choose: (a) to move one piece around the board by the total number of spaces shown on the dice, or (b) to move two pieces, one by the number of spaces indicated on each die.

10. There are no captures in this game. A piece is only removed on a throw of double one.

11. When a piece completes its course, it is reversed (i.e. placed upside down or on its side) in its final position, to distinguish it from any pieces that have not yet set out on their journey.

12. A throw must be used in full. If it cannot be used, then the entire throw is lost; it is not possible to utilise one die and not the other. Neither is it allowable to refuse a throw if it can be used.

13. A player wins the game when he returns all of his (remaining) pieces to their starting positions.

The "penalty" of throwing a double one seems unsatisfactory in these rules. Alan Wykes (The Complete Illustrated Guide to Gambling, 1964) suggests that a player with an incomplete set of pieces can only win the game if his opponent is similarly reduced.

Jack Botermans (The World of Games, 1989) suggests instead that a piece once removed must be entered on a roll of one, and no other piece moved in the mean time. He also suggests that the 10% stake is paid into the pot, and that a 20% stake may instead be paid in order to avoid the loss of a piece.

Strategy in Coan Ki

There is no capture in coan ki, in contrast to backgammon. So there may be advantages in spreading out one's pieces; this allows more choice of move than keeping the pieces bunched up together. This is especially valuable near the game's end, when some pieces are very close to their destination and may therefore be unable to use large rolls.

The player has a real choice in interpreting the dice. It is not always advantageous to move the furthest distance that the dice allow. Care must be made not to move into a position from which it is difficult or impossible to advance (for instance, having the last two pieces in play on the opponent's last row).