I have a conjecture about the origin of Chinese Chess. It is known that there was an earlier version, played on a board ten squares by ten, uncheckered; Murray reproduces a reconstructed arrangement by Karl Himly, with the 'king' and 'queen' arranged fore-and-aft in the nine-castle. This is a crucial (in many ways!) error. The current Chinese Chess board, eight squares by nine with a nine-castle at each end and a river across the middle, is known to be older than Chinese Chess, and to have been used for two previous games. There is therefore no basis for drawing the nine-castle on the ten-by-ten-square board for the earlier version. There is also no basis for believing that the earlier board contained a central river. Take them away, and we have the plain ten-by-ten-square board of so many variants, including Shatranj al-Kamil I. It also has the same pieces as Shatranj al-Kamil I: A king, his attendant minister, two elephants moving as alfils, two knights or horses, two rooks, and two orthogonal leapers, with a front rank of pawns. I think therefore that when Chaturanga was introduced into China in the time of the Wei-ti Emperor, and he had the two players beheaded and forbade the use of any game with a piece representing an emperor or called such, Chaturanga was indeed driven out of China. A couple of centuries later Shatranj al-Kamil Type One was introduced along another trade route from Persia. Perhaps the players were informed of the previous edict, or perhaps it was just their native prudence that persuaded them to demote one king to governor and the other to general, each with his appropriate officer. They then moved the game to a native board, abandoning the race game for which that board must have been quite inconvenient. Since the commander-in-chief and his adjutant were now inside a fortress, they were forced to stay within its walls. The elephant, huge, heavy, and one imagines heavily laden, was ruled unable to cross the river. The orthogonal leaper was changed from a camel to a catapult, or cannon, capable of destroying its victim even past a screen, but moving along the ground. The rook, or chariot, was left unchanged, and the pawn and horse were slightly modified for reasons that I don't see. <p>The odd thing is that Murray almost worked this out himself; he remarked on the great similarity between the earlier Chinese game of chess and the Persian variants. I think it was only the spurious nine-castle on Himly's diagram that prevented him from seeing the obvious.