Mahjong(g) Before Mahjong(g) Part 1

Michael Stanwick

Both Stewart Culin in America and William H. Wilkinson in China and England, were two of the earliest Western students of Chinese games. From their literature we have a view that both a tile set and most probably a game played with it, now associated with the game known as Mahjong(g), originated in China in the late nineteenth century. In part 1 and 2 of this article I will present new documentation, either material and/or archival, that sheds new light on the early development of different features of this game.

In Part 1 I will present an extract from an unnoticed and/or unpublished early record about a late nineteenth century game-play. I will also present and discuss three very early tile sets, together with an unpublished and hitherto unseen document describing one of these sets and the name associated with it. These new documents substantially modify our views about the early development of various features of the game of Mahjong(g).

1. The Name of the Game.

The term ‘Mahjong(g)’ was coined by Joseph P Babcock circa 1920 and was, at that point, directly associated or fixed to what I shall stipulate as the three basic features that formed the circa 1920 game; (1) a double sinogram 麻雀 ma que in modern Pinyin1 – literally ‘hemp bird’, but in the bonded form meaning ‘sparrow’; (2) a game-play and (3) a specific tile set. However, using the term ‘Mahjong(g)’ when investigating any of the proposed archetypes of (1), (2) or (3) that existed before circa 1920, may give the misleading impression that they are directly related to, or are identical to, their namesakes in the circa 1920 game. These relationships can only be established from analysis of, and between, the documentary evidence of the proposed archetype and the circa 1920 Mahjong(g) feature. Once a relationship is established and it is found that the archetype of the feature lacks or has additional elements compared to those that make up the circa 1920 feature, that archetype is then stipulated as a ‘precursor’ to Mahjong(g).

  • [1]

    Throughout this article I have chosen to use Pinyin romanisation of Chinese. Where other romanisations are used the pinyin form is in parentheses.

2. The Game-Play.

In this article I have stipulated the unique element of the game-play to be the formation of melds of two, three or four identical units from a group or a suit, in addition to a sequence of consecutive units from a suit, using the draw and discard method. However, where and when this game-play originated in China, is still unclear in my view. With reference to the development of the game-play though, one piece of relevant documentary evidence is an unpublished paper by Wilkinson (1925)(recently unearthed by Thierry Depaulis), in which he quotes directly from his Manuscript notes of 1890 and describes a game-play called chungfa (zhongfa), listing the combinations or melds and comparing the game to the Chinese card game of Khanhoo (K’anhu)(See Lo, 2003 and Berry, 2003).

(1) “Chungfa. The game is one for four players, neither more nor less. These bear as names the points of the compass, the leader being east, the player to his right south, the third player west, and the fourth north. The same number of cards (viz. 15) is dealt to each player as in Khanhoo, and the object of each (to “fill” his hand) is the same… …Sequences help towards filling a hand, but are not regarded as tricks, and score nothing. The tariff, so to speak, is not fixed, but is determined by the players before commencing.”

The number of player’s and nearly all of the ‘tricks’ 2 are in direct agreement with the corresponding features of the game-play of Mahjong(g). Most of these combinations involve melds of identical pieces (by ‘cards’ Wilkinson means tiles), which are made possible by the quadruplication of all the tiles. However, meld # 8 is most interesting; “any 3 aces, or any 3 nines (not all of the same suit)”. This involves melds of similar pieces – that is, of the same rank but not necessarily the same suit. This is a feature not present in Mahjong(g) and, in my view, makes this a precursor game-play to Mahjong(g). Further, it should also be noted that all but three of the melds listed involve the non-suit groups. The three just mentioned only involve the ‘aces’ and ‘nines’. This also distinguishes this game-play as a precursor.

  • [2]

    What Wilkinson calls ‘tricks’, in this instance, are what we now call melds, but these come with a points value.

3. The Tile Set.

The basic components of a Mahjong(g) tile set are: three quadruplicated Suits(based on money); four quadruplicated Directions; three quadruplicated ‘Honours’; two quartets usually, but not always, called ‘Flowers’ and ‘Seasons’ and 4 spare Blanks (Babcock, 1920). Thus 148 pieces.

3.1 The Two Glover Sets Figures 1 and 2.


Figure 1a. The American Museum of Natural History (AMNH) set.


Figure 1b. Side view of various tiles from the AMNH set.


Figure 2. The Brooklyn Museum of Art (BMA) set.


3.1.1 Acquisition and Accession Histories.

In Stewart Culin’s Smithsonian Institution Report (Culin 1893) there is a section called ‘Dominoes from Fuhchau’ (pages 518 – 521). This contains a detailed description of one tile set that bears some resemblance to the circa 1920 Mahjong(g) tile set. The second footnote on page 519 reveals that this set was acquired by the “Hon. George B. Glover, formerly US Consul at Fuhchau”, or Foochow, (Fuzhou) and was gifted by him to the museum of the Long Island Historical Society (LIHS), Brooklyn, New York, now called the Brooklyn Historical Society (BHS). Upon my request, a search of their archives by museum staff showed no record of the tile set in question.

Further investigation of this second footnote reference finally led Thierry Depaulis and myself to the letters of Robert Hart in ‘The I. G in Peking; the letters of Robert Hart’ (1975). These allowed us to build a sketch of Glover’s occupation and where and when he was stationed in China. George B. Glover is one George Bunker Glover, an American citizen, who joined the Chinese Customs Service at Canton in 1859, contrary to Culin’s reference. The data from Hart’s letters also shows that Glover was stationed at Fuhchau (Fuzhou) between 22nd February 1872 and 12th September 1873, (supporting Culin’s assertion that the ‘Dominoes’ were from there). Further, it allows us to infer that they were acquired during that period and were most probably donated to the LIHS at the end of that period.

Furthermore, this second footnote also alludes to a second set collected by Glover, which went to the American Museum of Natural History (AMNH), Central Park, New York. My investigation of this lead using the AMNH’s on-line collections database revealed that this set [Figure 1] was still with the museum and was acquired as part of a much wider collection of games donated to the AMNH under accession # 1869.90.19 (during the time period between the dates 1869 and 1890, the objects were the 19th lot accessioned). The tile set is now located in the Division of Anthropology and has the AMNH catalogue # 1/2377. I discovered, subsequently, that this collection is accompanied by a covering letter (also with the same accession number) written by Glover, dated Shanghai 18th August 1875. This describes in detail the tile set that he labels ‘a species of dominoes’.

Following this discovery, I began a further search for the missing LIHS tile set and found that Culin had started work in 1903 with the Brooklyn Museum in New York City, now the Brooklyn Museum of Art (BMA). Since he had described the LIHS Glover set I was searching for, I decided to investigate the BMA collections as well. With the help of the museum staff, an uncatalogued tile set nearly identical to the AMNH set was discovered in a box together with other uncatalogued Chinese games. Further investigation revealed a list of uncatalogued objects the LIHS had donated to the Museum in 1928, and on that list appeared: “Domino counters, Cantonese dominoes, Chinese game, 2 boxes Chinese dominos, Chinese dice, 4 boxes Chinese playing cards, and a Chinese game.” Comparing (1) this list with the contents of the box of uncatalogued Chinese games, (2) the description of the LIHS Glover set given by Culin (1893) with the BMA tile set and (3) the AMNH Glover tile set with the BMA tile set, reveals such a very high number of similarities that the BMA Registrar for the permanent collection, together with the Curator of Asian Art and myself, are led to conclude that there are good reasons to consider that the tile set in question [Figure 2], together with other objects, has a very high probability of being the one gifted by Glover to the LIHS, who subsequently donated it to the BMA in 1928. It now has the accession # 28.779.

3.1.2 The AMNH Set [Figure 1] and the BMA Set [Figure 2].

The AMNH tile set (see Figure 1a) now consists of 143 pieces composed of bone (indicated by the presence of Haversian Canals3) and bamboo, with dimensions: Length 19mm. Width 15mm. Height 12mm. Each of the pieces has a concave bone top, joined with an unusual longitudinal dovetail joint to a similar concave bamboo base (see Figure 1b). The missing pieces of the set are one 9 of Cash, one 6 Strings of Cash, one Season (‘Spring’) and two Rulers – ‘Heavenly Ruler’ and ‘Northern Ruler’. The ‘Earthly Ruler’ is in what would be the ‘Northern Ruler’ position. The ‘Winter’ tile is upside down.

The BMA tile set (see Figure 2) also now consists of 143 pieces composed of ivory (described by the museum) and bamboo with dimensions of: Length 18mm. Width 14mm. Height 10mm. As above, each piece has a concave joined with a similar longitudinal dovetail joint to a similar shaped bamboo base. The set has one 1 of Cash and four Blanks missing (the latter probably existed since one of them was used as a replacement for a 9 of Cash and possibly a 2 of Cash and a 9 of Strings of Cash).

Taking these missing pieces into account, both tile sets each have a total of 148 pieces. However, it should be noted that Culin (1893) lists five ‘Direction Rulers’ in the BMA set – the extra being the ‘Middle Ruler’. This tile may also be missing, making the BMA tile set 149 pieces.

Since the AMNH and BMA sets contain identical numbers of pieces and identical groups, we can apply Glover’s description in his 1875 letter to both tile sets. Glover begins his handwritten description of the AMNH tile set thus (for Chinese characters Glover used Mandarin which he romanised using Wade-Giles).

“No. 17 Set of a species of Dominoes, called 拷家雀 K’ao Chiá Ch’iao, snatching the house sparrow, composed of 148 pieces, the bottom of each being made of bamboo, and the top of bone.”

It should be noted that this name is only applied to the AMNH set, to which it directly refers.

拷 K’ao becomes kao using modern Pinyin romanisation, meaning ‘to beat’, ‘flog’ or ‘torture’. This seems a long way from Glover’s interpretation ‘snatching’. 家 chia becomes jia and means ‘home’, ‘family’ or ‘household’ (so house would seem to be an appropriate meaning in this context) and 雀 ch’iao is qiao in pinyin – another romanisation of the sinogram que ‘sparrow’ that is found in ma que. This appears to be the earliest association of the sinogram for ch’iao, “Sparrow” with a tile set.

Thierry Depaulis has reported to me that according to the recent Big Chinese-French dictionary the Grand Dictionnaire Ricci de la langue chinoise, Taipei, 1999, 7 volumes, which still uses Wade-Giles, “chia ch’üeh” (jia que) is the name given to the species of Sparrow, Passer domesticus, while ma ch’üeh (ma que) is given to the species of Sparrow Passer montanus. However, the dictionary says both names are synonyms. This suggests the name of Glover’s game might also have been known as “Snatching the Sparrow”. Glover continues his letter with a detailed description of the tile set:

“Of 40 pieces four are blank, and of the remaining 36, four have on their faces round ones, four round twos, and so on, running up to four round nines. Of 40 other pieces, four are blank, and four have on their faces oblong ones, running up to four oblong nines. Of 36 pieces four have 一万 ten thousand. Four have 二万 twenty thousand, and so on, running up to four of 九万 ninety thousand.”

The positions of the 8 Blanks in this description do not correspond to any feature in 19th century three-suited money-packs, from which the three suits most probably were derived, as Dummett (1980, 1996) has pointed out. This is possibly a mere accident of Glover’s way of describing the set. However, seeing how carefully and methodically the rest of the set is categorised, the possibility he has accurately described the composition of the suits cannot be ruled out at this stage. Contrary to Glover, when describing the BMA set, Culin (1893) places all the blanks together as a separate category.

With reference to Figure 1 and 2, the first suit in Glover’s description is Cash, (with one 1 of cash missing in figure 2.). The ‘cash’ or ‘coins’ are arranged as circles in patterns, with some resemblance to domino dots, but unlike those of most families of three-suited money cards.

Next is the Strings of Cash suit. The 1 String of Cash represents loops of strings, one of which is full of cash, all threaded through a red ‘toggle’ through which the base of the string holding the cash is also threaded. The loose ends are left to hang down from the ‘toggle’. The four single vertical lines therefore represent these loose strings. The rest of the suit consists of ‘rods’ or ‘oblongs’, uniquely arranged, to represent strings of cash.

For the Myriads of Cash suit the representations are directly copied from the sinograms found at the top of the Myriads of Cash suit in three-suited money cards.

However, it is the ‘extra’ tile groups that are remarkably unique since they have sinograms that appear to be unrelated to any of the cards in surviving three-suited money packs. As Glover describes:

“Then there are four pieces each with 中 Chung central, four with 東 Tung east, four with 南 Nan south, four with 西 Hsi west, and four with 北 Pei north. The four seasons come next, each season being marked on a piece as follows 春 Ch’un spring, 夏 Hsia summer, 秋 Ch’u autumn, and 冬 Tung winter, Then there are the rulers of the four points of the compass, as follows, 東王 Tung Wang, eastern ruler, 南王 Nan Wang, southern ruler, 西王 His Wang, western ruler, and 北王 Pei Wang, northern ruler, and lastly there are the following four 天王 Tien Wang, heavenly ruler, 地王 Ti Wang, earthly ruler, 人王 Jen Wang, ruler of men, and 和王 Ho Wang, peaceful ruler.”

Some of the sinograms – Central, North, South, East and West, and those for the four Seasons – do survive in tile sets to this day, although the so-called ‘’Dragons’ are conspicuously absent in this set. However, in the course of later development the sinogram for ‘central’ or ‘middle’ may have been annexed to the ‘Honours’ group (the so-called ‘Dragons’), to serve as the ‘Red Dragon’.

The five ‘Directions’ and the four ‘Seasons’ form an ancient grouping related to the wu xing, the Five Processes, as discussed by Graham (1989) and also described by Te-K’un (1957). The four ‘Rulers of the Four Points of the Compass’ and the four additional Rulers may also be symbolic manifestations of popular Chinese Cosmology. The presence of these Rulers and the absence of the ‘Honours’ or ‘Dragons’ triad also designate these sets as precursors.

3.2 The Himly Set.

A very similar bamboo tile set was also collected and described by Karl Himly (1889, 1901, translated by Thierry Depaulis). According to Himly (1901) the tile set was called Ning bo zhu pai, “bamboo cards of Ningbo”. Since he left China in 1876 he must have acquired it in China during his stay in Shanghai (1868 – 1876), although where he acquired it remains undetermined. This places it contemporaneously with the AMNH Glover tile set.

Each of the 148 tiles in this set has the dimensions: Length 14.4mm. Width 12mm. Height 9.6mm. It has the three quadruplicated suits – Cash (同 tong), Strings of Cash (索 suo) and Myriads of Cash (万 wan). Each suit has one ‘Flower’ (化 hua 4) tile associated with it – thus 同化 tong hua, 索化 suo hua and 万化 wan hua. There are four Directions (points of the compass), quadruplicated, and four ‘Kings of the Four Heavenly Directions,’ 北王 bei wang, 南王 nan wang, 西王 xi wang and 東王 dong wang, to which is added a 摠王 zong wang or “Almighty King”. Further, there are an additional four ‘Kings’, with three named after the 三才 san cai or the “three powers” – thus Heavenly King (天王 tien wang), Earthly King (地王 di wang), Man’s King (人王 ren wang) – and the fourth called 和王 he wang, King of Harmony. Finally there is a ‘Seasons’ quartet and eight blank tiles. All of the non-suit groups have coloured frames. It appears the colours blue, red and green are used in the same manner as in the Glover sets.

Interestingly, this tile set does not have the four ‘Centre’ (中 zhong) tiles that are present in the two Glover sets, but it does have a 摠王 zong wang ‘ Almighty King’. Further, the three hua tiles, each associated with a suit, are also absent from the Glover sets. Again, the presence of these Rulers and the absence of the ‘Honours’ or ‘Dragons’ triad also designates this set as a precursor to Mahjong(g).

The absence of the ‘Honours’ or ‘Dragons’ in this set, gives us good reason to suppose, at least as far as the Glover sets and the Himly set are concerned, the quadruplicated ‘Points of the Compass’ or ‘Directions’ were present before the three quadruplicated ‘Honours’ or ‘Dragons’.

The game-play of chungfa (zhongfa) and the tile sets of Karl Himly, the BMA and the AMNH, I have labelled as a precursor game-play and precursor tile sets to Mahjong(g). From extant documentation we can infer that the BMA and AMNH Glover sets and the Himly set, are the earliest precursor tile sets, with the AMNH set having the earliest use of the sinogram for ‘sparrow’ associated with it. That both the BMA and AMNH sets are almost identical in construction and patterns and come from the same donor, does not suggest that they were collected at the same date or location. However, this possibility cannot be ruled out either. The date of the game-play of chungfa, plus the name of ‘sparrow’ associated with the AMNH tile set and the acquisition date of the BMA tile set, push the development history of these features back to 1890 and 1875 and 1872/73, respectively.

  • [3]

    The Haversian Canals carry blood vessels throughout compact bone. The black ‘pitting’ on the top of the tile’s bone surface is where dirt has filled in the pits created by the obliquely cut Canals.

  • [4]

    Himly notes that although hua means “to change”, in this context he considers the sinogram to be an abbreviation of hua, meaning “flower”.

I would like to thank the American Museum of Natural History and the Brooklyn Museum of Art for permission to illustrate their sets;

Figure 1a & 1b: Catalogue No. 1/2377 Courtesy of the Division of Anthropology, American Museum of Natural History, New York, N.Y. See;

Figure 2: Accession number 28.779, Brooklyn Museum of Art. Gift of the Long Island Historical Society. See;


Cheng Te-K’un. ‘Yin-Yang Wu-Hsing and Han Art’. In: Harvard Journal of Asiatic Studies, Harvard-Yenching Institute, 1957, p. 162 –186.

Babcock, Joseph Park. Rules for Mah-Jongg. n.p. San Francisco : W.A. Hammond Co., 1920. 12 p. (Sloper Coll.); reprints 1921, 1922, etc.

Berry, John. ‘Chinese Money-Suited Cards’. In: The Playing-Card, Vol. 31, No.5. 2003.

Brunner, Katherine F., Fairbank, John K., Matheson, Elizabeth (eds). The I. G. in Peking; the Letters of Robert Hart, Chinese Maritime Customs, 1868 – 1907. Cambridge Mass., and London, 1975. 2 Vols. 1625 pages.

Culin, Stewart. ‘Dominoes from Fuhchau’ under ‘Chinese Games with Dice and Dominoes’. In: Annual Report of the Board of Regents of the Smithsonian Institution for the year ending June 30, 1893 (Report of the US National Museum), Washington 1895, p. 491-537 (reprint : Seattle, Wash.: Shorey Books, 1972).

Culin, Stewart. ‘The Game of Ma-Jong, its Origin and Significance’. In: Brooklyn Museum Quarterly, Brooklyn, NY, Vol. XI, 1924, p. 153-168.

Dummett, Michael. The Game of Tarot, London, 1980.

Dummett, Michael. ‘A Vietnamese Card Game’. In: Ludica, 2, 1996, p. 255-261.

Graham, Angus C. Disputers of the Tao. Philosophical Arguments in Ancient China, Open Court, 2001. (1st ed. 1989).

Himly, Carl. ‘Morgenlandisch oder abendlandisch? Forschungen nach gewissen Spielausdrucken’. In: Zeitschrift der Deutschen Morgenlandischen Gesellschaft, XLIII, 1889, p. 415-463 + 555-578.

Himly, Carl. ‘Die Abteilung der Spiele im ‘Spiegel der Mandschu-Sprache’, VII’. In: T’oung Pao, 2nd series, Vol.II, 1901, p. 20-3.

Lo, Andrew. Pan Zhiheng’s ‘Xu Yezi Pu’ (Sequel to a Manual of Leaves) – Part 1. In: The Playing-Card, Vol.31, No.5. 2003.

Wilkinson, William H. (1890): Published in 1901 as p. 184-194 of Catalogue of the Collection of Playing Cards Bequeathed to the Trustees of the British Museum, F. M. O’Donoghue.

Wilkinson, William H. (1893): Published in Culin, Games of the Orient, Tuttle, 1958. First published under the title Korean Games, with Notes on the Corresponding Games of China and Japan, University of Pennsylvania, 1895.

Wilkinson, William H. ‘Chinese Origin of Playing Cards’. In: The American Anthropologist, vol. VIII, 1895, p. 61-78.

Wilkinson, William H. ‘Mah-Jongg: A Memorandum’ Copyright by The Continental Mah-Jongg Sales Co. Amsterdam, 1925. Not for Publication.

Le Grand Dictionnaire Ricci de la Langue Chinoise. Paris : Desclée de Brouwer : Institut Ricci de Paris ; Taipei : Institut Ricci de Taipei, 2001 (8 vol.).