Gameplay as Foreplay at a Medieval Indian Court: Translation and Discussion of Mānasollāsa 5:16, Phañjikākrīḍā

This study focuses on the singular courtly game of phañjikā described in the 12th-century Mānasollāsa attributed to King Someśvara III of the Western Cālukya Empire. It shows that phañjikā belongs to the family of cruciform race games, which also counts the famous games of caupaṛ and paccīsī among its members. Phañjikā, however, predates the earliest evidence for both of those games by several centuries, and should therefore be considered an early indication of the popularity that cruciform race games would come to enjoy in elite and royal households from at least the 15th century onward. The study also shows that phañjikā did not enjoy the same status at court as other board games, such as chess and backgammon, also described in the Mānasollāsa. It was primarily associated with the women at court, and only engaged in by the king for the pleasure of witnessing the passionate emotion that it stirred in them. Based on the low status of the game, and the prevalence of race games in all levels of society, the study argues that phañjikā was likely an elaborate courtly adaptation of a simpler folk game. This would explain its absence from the literature outside the Mānasollāsa, as well as its many correspondences with a wide range of cruciform, square, and single-track race games only documented in more recent sources. The study suggests that more scholarly attention should be paid to the regional literatures of India, as they developed in the first half of the 2nd millennium CE, for a more detailed understanding of the early history of medieval Indian race games to be arrived at.

parts known as viṃśatis, or collections of twenty, referring to the number of chapters they contain. The first part concerns the acquisition of a kingdom, the second part the maintenance of a kingdom, and the third, fourth, and fifth parts the various enjoyments to be had in a kingdom once acquired and properly maintained. 4 The latter three parts are divided into pleasures (upabhogaviṃśati), entertainments (vinodaviṃśati), and games (krīḍāviṃśati). They are mostly concerned with various physical sports and pastimes, such as animal fights, wrestling, hunting, and drinking. The Krīḍāviṃśati also contains chapters on what is sometimes referred to as sedentary games, or games that are played with a combination of dice, pawns, and game boards. The chapters on chess (caturaṅgakrīḍā, Mān. 5.12) and backgammon (pāśakakrīḍā, Mān. 5.13) have been discussed in detail by 5 while the chapter on a third board game known as phañjikā has only received scant attention (phañjikākrīḍā, Mān. 5.16). 6 This is likely due to the terminological difficulties and corrupt passages of the text, further complicated by the lack of additional information about the game outside of the Mānasollāsa. 7 As a result, authoritative discussions of phañjikā are still lacking. 8 The inclusion of phañjikā in the Mānasollāsa is surprising, as it belongs to a family of primarily folk games otherwise not described in the literature until several centuries later. 9 Games belonging to the same extended family are among the earliest board games recorded anywhere in the world, tracing their origins 4 As suggested by McHugh (2013: 155-56), the division of the work can be seen as a reflection of the three traditional aims of life (trivarga), with the first section falling within the category of dharma, or right conduct, the second within that of artha, or worldly affairs, and the final three within that of kāma, or sensual pleasure. 5 Bock-Raming 1995;1996. 6 I have previously written about phañjikā in relation to the games of gyān caupaṛ (Schmidt-Madsen 2019) and also caupaṛ (Schmidt-Madsen 2021). The present study expands upon my earlier arguments and in some cases reaches different conclusions. 7 The vast majority of surviving manuscripts of the Mānasollāsa break off before the Krīḍāviṃśati section, and the few that include it suffer from corruptions and scribal errors. G. K. Shrigondekar, who critically edited the Mānasollāsa, used only three manuscripts (two of which were copied from the same source) for his edition of the Krīḍāviṃśati section. He noted in his pre-face that "[a]s all the three mss. are corrupt the constitution of the text became very difficult" (Shrigondekar 1925-61: v. 1, ix). The earliest of the manuscripts (Shrigondekar MS D) dates from Śaka 1592 (1671 CE), some five and a half centuries after the original text was written (Shrigondekar 1925-61: v. 1, v). All references to the Mānasollāsa in the present study follow the edition of Shrigondekar (1925-61). 8 The paraphrases by Mishra (1966), Samaddar (2000), and Arundhati (2004) leave out or misinterpret several passages. The identification of phañjikā by V. Raghavan (1979: 81) as being a variation of saptalekha cannot be sustained, as the latter is a pure gambling game without pawns or board (5.14.782cd-795ab), and thus has little in common with phañjikā. 9 Backgammon, which the Mānasollāsa described in full for the first time in South Asian literature, is categorized as a race game like phañjikā, but appears to have been associated more exclusively with the upper back to ancient Egypt and Mesopotamia, but detailed descriptions of them in a South Asian context only begin in the regional literatures around the mid-second millennium CE. As such, the chapter on phañjikā provides a unique glimpse into the history of these games as they appeared at a powerful Hindu court in the early twelfth century. A key observation about phañjikā is that it seems to have been primarily played by the women at court. The king's main interest in the game was not the game itself, but rather its propensity for bringing forth emotion in the women who played it and its function as a pretext for sexual encounters. This might well explain the lower status awarded the game as compared to chess and backgammon, as well as the incomplete description of its layout and rules. It also hints at the possibility that phañjikā had its origins in contemporary folk games, and that the version encountered in the Mānasollāsa was merely an elaborate adaptation of a much more widely played game.
The present study is divided into three main parts followed by an appendix. The first part provides an overview of the chapter on phañjikā in the Mānasollāsa, and extracts all relevant information about the material components of the game and the rules that govern its operation. The second part engages with the social context of the game, focusing on the interactions between the king and the women who played it, and the amorous purposes to which it was put. The argument for an erotic component in the relationship between the king and the women is further supported by a discussion of the location of the phañjikā chapter among the overtly sexual final chapters of the Krīḍāviṃśati. The third part focuses on the ludic context of the game and its place in the wider history of traditional South Asian board games. Drawing upon the theory of ludemes as conceptual units of game-related information capable of traveling between games, it proposes a strong link between phañjikā and later games within the categories of cruciform, square, and single-track race games. The study concludes with an appendix which revisits the original text of the phañjikā chapter as critically edited by Shrigondekar. It carefully considers the emendations suggested by him, and translates the entire chapter into English for the first time.
2 P H A Ñ J I KĀ T HE CHAPTER ON PHAÑJIKĀ consists of 47½ verses corresponding to verses 5.16.816-63ab in the edition of Shrigondekar. 10 The verses are written in the śloka metre, and will be referenced in parentheses throughout the study.
classes of Indian society (Soar 2007: 228). See, however, my discussion of the relationship between backgammon and the phañjikā-related games of caupaṛ and paccīsī (Schmidt-Madsen 2021: 41-44). 10 Shrigondekar 1925-61: v. 3, 253-257. Paraphrases and discussions are based on my translation of the text, which can be consulted for further commentary in the Appendix. The chapter on phañjikā can be divided into three main sections, consisting of a contextual introduction to the game (816-26ab), an overview of its components (826cd-836), and a description of its rules of play (837-63ab). Though the rules section is longer than the other two sections combined, it does not go into the same level of detail as the rules sections in the chapters on chess and backgammon, and does not allow for a complete reconstruction of the game in all its details. Similarly, the components section leaves several questions unanswered, such as the placement of safe squares on the game board and the use of notational signs for recording dice throws. The introductory section, however, provides information about the female players and the king's underlying reasons for playing the game which is not found in the chapters on chess and backgammon, nor in the related chapter on gambling (varāṭikākrīḍā, Mān. 5.14). The overall impression left by the text is one of aesthetic and social above ludic concerns. This can be partly explained by the reliance on implicit knowledge about the game no longer accessible to the modern reader. While this explanation strengthens our assumption that the game, or at least some less elaborate version of it, was known beyond the courtly context in which it appears, it does not explain the king's focus on the physical appearance and flirtatious nature of the women who played it. A further part of the explanation, as discussed in the second part of the study, therefore seems to be that the king regarded the game as a ludic means to an amorous end.
Leaving aside the aesthetic and social concerns of the text, and extracting only the information relevant to the game itself, we are able to reconstruct it in some detail. It is played on four 6 × 6 grids forming the arms of a cruciform game board. Up to sixteen players can take part, with each player, or team of players, controlling a set of five identically colored cowrie shells which function as pawns. Seven larger cowrie shells are used as binary dice. The pawns move along an unspecified route on the arms of the board according to the throws of the dice, beginning and ending the game in the center of the board. A throw of five cowrie shells face-down and two cowrie shells face-up, known as a phañjikā, is required to enter the pawns into and exit them from the game. If a pawn lands on the square of a pawn belonging to another player, the latter pawn is captured and returned to start. An unspecified number of squares are identified as safe squares, indicating that pawns resting on them cannot be captured. The player who first manages to exit all their pawns wins the game. The game, however, continues after a winner has been declared, and only stops when a loser has been singled out from among the remaining players.
The above description makes it clear that phañjikā belongs in the category of race games defined by board game historian H. J. R. Murray as games "in which teams of equal size race one another along a given track, and the first player to complete the course with his team wins". 11 The shape of the phañjikā board further identifies the game as belonging in the subcategory of cruciform race games, 12 which also includes the popular Indian games of caupaṛ and paccīsī dating back to at least the fifteenth century. 13 Other subcategories with which phañjikā is affiliated include square and single-track race games, 14 which are also widely attested in India, albeit only with any certainty from the nineteenth century onward. The affinities between phañjikā and the various subcategories of race games will be explored in detail in the third part of the study.
Before entering into the specifics of components and rules, a brief note on the word phañjikā should be given. The text refers to the game variously as phañjikā, phañjī, and phañji, which the dictionaries gloss as botanical expressions for one of several plants. Given the ludic context of the word in the Mānasollāsa, a more plausible explanation would be to regard phañjikā as a vernacular variant of pañcikā, or a collection of five. 15 This interpretation is supported by the fact that a throw of five cowrie shells face-down is referred to as a phañjikā. Naming a game after a special throw of the dice is common practice in traditional South Asian games, with a close parallel being found in paccīsī, or a collection of twenty-five. The likewise related single-track game of pañci, or a collection of five, may either derive its name from a throw of the dice, or from the spacing of safe squares along the game track. Though neither paccīsī nor pañci is attested contemporaneously with phañjikā, they both belong to the same ludic family, which likely reaches back far beyond the Mānasollāsa. An even closer linguistic parallel to phañjikā is the game of pañcikā, which also appears to derive its name from a throw of five binary dice landing either face-up or face-down. The game is mentioned in the seventh-century Kāśikāvṛtti, which, however, makes it clear that it is a simple dice game without pawns or game board. 16 11 Murray 1952: 4-5. Murray's categories of race, war, hunt, mancala, and alignmentand-configuration games have often been criticized as being arbitrary and lacking a sound theoretical basis. Still, they continue to be used by board game historians, and remain the preferred mode of reference. David Parlett's attempt at replacing Murray's categories with race (race), space (alignment and configuration), chase (war, hunt), and displace (mancala) games in his Oxford History of Board Games never appears to have caught on (Parlett 1999: 8-14). An attempt at a new system of categorization was recently suggested by Thierry Depaulis, but its wider application remains to be seen (Browne et al. 2019: 5-7). 12 Murray 1952: 132-140. 13 Schmidt-Madsen 2021. 14 Murray 1952 If phañjikā was indeed the vernacular word by which the game was known in common speech, it would make good sense to keep the original spelling, and hence pronounciation. The phonological reaons for aspirating pa to pha and voicing ñci to ñji are unclear to me, but the manuscript reading pañcikā (838c) does indeed appear to be synonymous with phañjikā. 16 KV 2.1.10: Pañcikā nāma dyūtaṃ pañcabhir

Gameboard
The text explains that the game board (maṇḍala) used in phañjikā should be drawn and decorated by a person skilled in the game (phañjiviśārada) (828cd). As no physical materials are mentioned, we should probably assume that it was drawn on the ground with chalk, powder, or similar. At its most basic, it consists of five quadrants arranged in a cruciform pattern. The central quadrant, measuring 18 × 18 aṅgulas (c. 29 × 29 cm), is drawn first, 17 and then the outer quadrants (bhadraka), also measuring 18 × 18 aṅgulas, are drawn in extension of its four sides (826cd-827ab). 18 The whole board is ruled crosswise and lengthwise with lines set at an interval of 3 aṅgulas (c. 5 cm), resulting in each of the five quadrants being overlaid with a grid of 6x6 squares for a total of 180 squares (827cd-28ab). The central quadrant, referred to as the geha, or home square, is further divided by four half-moons (ardhacandra) (828cd), which can either be understood in the figurative sense of semi-circles, or in the abstract sense of triangles. In the former case, the solution would probably be to draw a semi-circle extending from the top of each of the outer quadrants into the central quadrant ( Figure 1). In the latter case, two diagonal lines crossing between opposite corners would suffice ( Figure 2). Regardless of the method applied, the four divisions (bhadra) created by the half-moons are decorated with illustrations of a palace (prasāda), a lotus (paṅkaja), a crescent moon (khaṇḍa), or a swan (haṃsapakṣiṇa) (829). 19 The text states that the initial throw (pūrvadāya) entering the pawns onto the squares of the outer quadrants is counted from the central quadrant (gehagaṇana) (848), indicating that the half-moons function as home spaces for the pawns. The grid squares in the central quadrant would therefore seem to have been of no consequence to the game, and might even have been left out entirely. On the other akṣaiḥ śalākābhir vā bhavati / tatra yadā sarve uttānāḥ patanti avāñco vā tadā pātayitā jayati tasyaivāsya vighāto 'nyathā pāte sati jāyate / (KV 2.1.10) [There is a game called pañcikā (played) with five akṣas or śalākās. When they all fall face-up or face-down, then the thrower wins. When the fall is otherwise, he loses.] Akṣa is often used as a general term for "dice," while śalākā is often used as a specific term for "stick dice." Since the description of the game makes it clear that the dice used were binary, perhaps we should understand akṣa in the sense of "binary dice" and śalākā in the sense of "binary throwing sticks." 17 An aṅgula, or a finger-breadth, is not an exact unit of measurement in origin, and the extent to which it was standardized during the rule of King Someśvara III is unclear. Common estimates include c. 1.6 cms (e.g., Michaels 1978: 156-57) and c. 1.9 cms (e.g. R. Bhattacharya 2019: 33). I lean toward the lower number to ensure that the game would be playable on the smaller surface. 18 In comparison, the backgammon board described in the chapter on pāśakakrīḍā is 20 aṅgulas (c. 32 cm) wide and 48 aṅgulas (c. 77 cm) long (Mān. 5.13.634), meaning that the phañjikā board is approximately the size of two backgammon boards placed crosswise. 19 See the Appendix for further details on this corrupt and difficult verse. hand, ruling the whole game board at once would probably have resulted in a more uniform ruling than leaving the central quadrant blank and ruling each of the outer quadrants separately. In later cruciform race games, the central square is also sometimes found to be on a grid, even when its sole purpose is to hold the pawns before and after they circumambulate the arms of the board (Figure 3). More commonly, however, the grid is replaced by four triangular home spaces, corresponding to the stylized version of the half-moons in phañjikā (Figure 4). Whether the grid was maintained in the central quadrant of the phañjikā board or not, the total number of operational squares should probably be adjusted from Another feature referred to in the text is that of safe squares (śaraṇāgāra) (852), which are usually marked with an "X" on traditional game boards. The text, however, does not mention how the squares should be marked, nor where on the game board they should be located.

Dice
Phañjikā is played with seven large cowrie shells (varāṭakā) for dice (838ab). Cowrie shells are often used as binary dice in traditional South Asian games, including race games related to phañjikā. They have a flat side with a teethed opening and a rounded side with a smooth surface, and when thrown they can either land face-down on the flat side, or face-up on the rounded side. The count of a throw is usually based on the number of cowrie shells landing face-up, but in phañjikā the cowrie shells landing face-down are counted instead. 20 Throws are generally referred to as dāyas, but dāya also has more specific meanings in the game. A distinction is made between phañjikā throws used to enter and exit 20 Whether the reversed method of counting in phañjikā reflects an earlier practice, or whether it is merely an idiosyncracy of the game, cannot be determined from the available evidence. It should, however, be noted that the same method of counting is implied in the description of the related game of pañci in the game manuscripts of Kṛṣṇarāja Oḍeyar III (1794-1868) (Vasantha 2006: 32). pawns, and dāya throws used to move pawns along the squares in the outer quadrants. A further distinction is made between dāya throws used to move pawns in general, and pūrvadāya throws specifically used to move pawns from the central to the outer quadrants . 21 The text refers to throws of 1, 2, 3, 4, 6, and 7 cowrie shells face-down as dāyas, throws of 5 cowrie shells face-down as phañjikā, and throws of 0 cowrie shells face-down as kalasaptaka, or "sweet seven" in the sense of "lucky seven" (838cd-46).  As shown in Table 1, all throws except phañjikā are associated with a specific notational sign. The text does not explain the rationale behind the signs, but perhaps the most obvious use would be to keep track of the results thrown. In later South Asian race games, certain throws usually allow players additional throws, sometimes leading to long sequences of throws which are only applied to the pawns at the end of the sequences. In order to remember the sequences and avoid error, or even cheating, some sort of mnemonic device, such as notational signs, is required to keep track of them. The practice of awarding additional throws is not mentioned directly in the text, but one verse implies that multiple phañjikā throws can be made in sequence (847), and another that they might be cancelled if too many are made (853). 23 Another possible use of notational signs 21 Confusing as this may sound, it follows the use of dāya in later race games, where the term is alternately applied to a game, a turn, a die, a throw of the dice, a throw of 1 on the dice, and the entry or exit of a pawn by means of a throw of 1 on the dice. The term is especially common in Tamil where it is known as tāyam (Tam. தாயம்). See Parker 1909: 617-20, Bell 1969: I, 17-20, and Balambal 2005 The calculation is based on a simple empirical experiment with seven cowrie shells brought home from India. It shows that the cowrie shells have approximately a 45% chance of landing face-up and a 55% chance of landing face-down. I am grateful to Toke Lindegaard Knudsen for carrying out the experiment and making the calculations (cf. Schmidt-Madsen 2019: 206, footnote 345). 23 In later South Asian race games, an ex-is to record bets on the result of individual throws during the game. 24 However, since phañjikā appears to have been played in an intimate setting for purposes quite different from material gain, it is perhaps unlikely that betting would have been involved. Two throws merit special attention as they are both assigned properties beyond that of mere calculation. The throws in question are the phañjikā and kalasaptaka throws, neither of which are identified as dāya throws. The main function of phañjikā throws is to enter pawns into and exit them from the game (857). In order to enter a pawn into the game, a player has to throw a phañjikā followed by a dāya, referred to as a pūrvadāya, or initial dāya. The phañjikā throw activates the pawn, so to speak, and the pūrvadāya throw determines how many squares it moves from the central onto the outer quadrants (847-48). Later, when the pawn has completed the track on the outer quadrants, another phañjikā throw is required for it to reenter the central quadrant (856cd). Once the pawn has reentered the central quadrant, a final phañjikā throw is required to bear it off the board (858ab). 25 If, as seems to be the case, phañjikā throws are never used to move pawns along the track, but only to enter and exit them, this may explain why they are the only throws not associated with a notational sign. They simply do not figure in the calculation of moves, and may even have been applied immediately upon being thrown rather than form part of any longer sequences of throws.
Contrary to the phañjikā throw, which occurs in about one in five throws, the kalasaptaka throw only occurs in about one in two hundred throws (see Table 1). When it does finally occur, the effect seems to be that a player lucky enough to have thrown it immediately completes the track with one of their pawns currently in play and returns it to the central quadrant, from where it can then be borne off with a phañjikā throw. The pawn is said to "possess the half-square," with reference to the half-moon-shaped home space where it began its journey (855). 26 Since kalasaptaka throws are marked with a doubling of the notational sign for a throw of seven (saptakadviguṇa), it seems possible that it could also be used to move a pawn two times seven, or fourteen, squares. This would correscess of certain throws awarding additional throws will often cancel the entire sequence of throws. An early formulation of this rule with regard to paccīsī is found in the Cetovinodanakāvya, or poem on the enjoyments of the mind, written by Dājī Jyotirvid in 1822. There it is stated that three sequential throws of 10 or 25 makes the entire sequence void (nirarthaka) (CVK: 380 and cf. footnote 148 below). Cf. Williamson 1801: 16-17 andTemple 1884: 245. 24 Betting on individual throws was reported by R. C. Temple in late nineteenthcentury Punjab (Temple 1884: 244-245). Unfortunately, he did not explain how the bets were kept track of. 25 This rule is also found in many later South Asian race games (e.g., Williamson 1801: 22;Bell 1969: I, 18 pond with what we know from later sources, 27 but it is unclear when the move would be applied, as kalasaptaka throws are not designated as dāya throws, and hence cannot be used to enter pawns from the central quadrant. A simpler solution would therefore be to regard the notational sign as an indication of a "lucky seven."

Pawns
The cowrie shells used as pawns in phañjikā are distinguished by size and color from the cowrie shells used as dice. The pawns are smaller than the dice, and they come in sets of five with the same shape and color. Each set is referred to as a dhāman (family), and individual pawns as dhāmakas (belonging to a dhāman), or simply cowrie shells (kapardikā, varāṭikā). Players choose a set according to their liking at the beginning of the game, and place the pawns outside the outer quadrants (835cd-37). The text lists sixteen different sets of pawns (831-35ab), corresponding to the maximum number of players that can participate in the game (see Players below). The detailing of the various shapes and colors of the pawns, together with the instructions for decorating the game board, shows the importance given to the aesthetic quality of the components. This conforms to what we know from other sources about game equipment in elite and royal households, which often included pawns and dice made of precious materials and studded with pearls. 28 As noted above, the text states that the pawns begin the game outside the outer quadrants (bahirbhadram) [837cd, 851ab]. It is unclear whether this means that the pawns are positioned in the half-moons in the central quadrant, or outside the game board as a whole. Taking into account the rule that players enter pawns into the game by making a phañjikā throw followed by a pūrvadāya throw, it seems most likely that the pawns are initially kept outside the game board. They will then enter into the central quadrant by means of phañjikā throws, and continue onto the squares in the outer quadrants by means of pūrvadāya throws. This will also help avoid confusing them with pawns that have already completed the game track and returned to the central quadrant but have yet to be borne off by a final phañjikā throw. 29 27 Throws of zero, which is essentially what a kalasaptaka throw is, often count as double the amount of binary dice thrown (e.g., Balambal 2005: 37-38). Thus, in the earliest known complete set of rules for paccīsī, a throw of seven cowries face-down counts as 14 (Williamson 1801: 12), while in later sets of rules, seven cowries faceup sometimes count as 14 (e.g. Temple 1884: 244; R. K. Bhattacharya et al. 2011: 35, 44, 56, 99). 28 See G. N. Sharma 1968: 132 andBalambal 2005: 57-58.
The method of preparing cowrie shells for the gambling game of puñjikā, described in the chapter on varāṭikākrīḍā, prescribes that the shells should be polished (ślakṣṇa) and filled with cow dung (gomaya) . 29 In other cruciform race games where RULES Players Phañjikā should primarily be played by women (strī), though young boys (mānava) can also take part. 30 The only persons who should not take part are men (nara) (819). The role of the king is not clearly stated in the text. On the one hand, he is described as taking stock of whether the women in the game wants to defeat him or be defeated by him (822cd-23ab), and on the other hand, the women are portrayed as asking him for the phañjikā throws they require to proceed in the game (824cd-825ab). 31 The possibility that the king throws some or all of the dice for the women finds support in the chapter on backgammon (pāśakakrīḍā, Mān. 5.13), where a person other than the two players throws the dice. In the case of backgammon the purpose clearly is to avoid cheating, as stakes are wagered on the game, and as the person throwing the dice is required to be impartial to the players. 32 In phañjikā the situation is different, as no stakes are wagered, and as showing partiality is consciously employed by the king as a stratagem for evoking emotive responses in the women. 33 Whether the king actively participates in the game, or merely presides over it, it seems likely that he manipulates the throws of the dice for emotional effect. His ability to do so is demonstrated in the preceding chapter on phaṇidākrīḍā, or the game of placing bets, where he demonstrates sleight-of-hand tricks to impress the gamblers in his audience (Mān. 5.15.800).
The number of players in the game is given as "five, seven, six, eight, nine, (or) sixteen" (818cd). Confusing as this may sound, we should probably just understand it to mean that phañjikā is a multiplayer game with a variable player count. 34 Judging from the cruciform design of the game board with its four outer quadrants, we might have expected a four-player game, or at least a game played by multiples of four. While the enumeration of sixteen different types of cowrie pawns enter and exit the game via the same route, they are often distinguished by placing the returning pawns on their side (Williamson 1801: 13-14). This might also have been the case in phañjikā, though the text does not mention it. 30 In modern-day Karnataka, the square race game cauka bāra is reported as being mostly played by women, but sometimes also by young boys (Kulirani and Vijayendra 2011: 112). In Tamil Nadu, the related game of tāyam is reported as being played by women and young girls (Bell 1969: I, 17). 31 A similarly ambiguous role is assigned to the king in a physical race game described in the chapter on jyotsnākrīḍā (see footnote 49). There the king can apparently choose whether he wants to run in the game, spur on the women who run, or take on the role of umpire . 32 Bock-Raming 1995: 7. 33 See the section on Amorous Play below for further discussion of this and other stratagems employed by the king. 34 In the physical race game mentioned above (footnote 31), the number of players on each of the two teams is similarly described as "five, six, ten, eight, ten [sic], seven, or nine" (pañca ṣaḍ vā daśāṣṭau vā daśa sapta navāpi vā, Mān. 5.8.359cd). shells used as pawns suggests that the game can indeed involve sixteen individual players, making for a dizzying total of eighty pawns, it is also possible that multiple players would join up to form teams, each controlling a shared set of five pawns. 35 This is a well known practice in many traditional South Asian race games even to this day. 36 It was recorded by Henry Parker in Sri Lanka in the early twentieth century, where up to eight people would play the four-player cruciform race game of pahaḍa keḷiya (Sin. පහඩ ෙකලිය), 37 and even the two-player single-track race game of pañca keḷiya (Sin. පංච ෙකලිය). 38

Movement
The movement of the pawns is determined by the fall of the dice, with phañjikā throws being used to enter and exit pawns, and dāya throws being used to move pawns along the squares of the game track. Unfortunately, the exact route followed by the pawns is not described in the text. All we can say with certainty is that the pawns enter the outer quadrants from the central quadrant, and that they later return to it after completing an unspecified track. The verse on kalasaptaka throws discussed above further suggests that the pawns move around (pari+√hṛ) the game board by way of circumambulating the central quadrant as in other cruciform race games (855). We can also infer from the rule for capturing pawns that the track is unidirectional, as pawns are described as being captured from behind (849-50). A unidirectional track, however, does not necessarily mean that all pawns follow the exact same route. On analogy with later cruciform, square, and single-track race games, pawns may have entered and exited the outer quadrants in different places, and only have shared part of their individual tracks with each other. This will be further discussed in part three of the study.
Regardless of the route followed by the pawns, we can safely assume that players alternate taking turns throwing the dice and moving their pawns. We know that the initial throw (pūrvadāya) is counted from the central quadrant (848), and it seems obvious that the same method of counting is applied when moving pawns already in play. The number of squares moved on a simple dāya throw probably corresponds to the number of cowrie shells falling face-down, as this is the number after which the throws are named (see Table 1 above). However, since the range of available numbers is relatively low, starting from 1 and 35 That sixteen-player race games did in fact exist is evidenced by the game of "chandal mandal," said to have been invented by Akbar sometime in the second half of the sixteenth century. The game is a 16-armed variant of caupaṛ allowing for up to sixteen individual players. The game board, how-ever, bears little resemblance to the game board in phañjikā (Blochmann and Jarrett 1873-94: 304-5). 36 E.g., Singh et al. 2016: 60. 37 Parker 1909: 612. 38 Parker 1909 ending at 7, and the total number of squares is relatively high, amounting to 144 across the four outer quadrants, it is possible, as discussed above, that certain throws award additional throws. Another possibility is that some throws count for more than the number of cowrie shells falling face-down. We have already suggested that kalasaptaka throws may have been counted as 14, and it is possible that other throws, too, translated into higher numbers. This is a common practice in later South Asian race games, such as paccīsī, where throws of 10, 14, 25, and 30 can result from a single throw of seven cowrie shells. 39 Without the inclusion of additional throws and higher counts, possibly in combination, phañjikā would likely have been quite slow and dull. On the other hand, we are told that it was played in the evening or at night when time might not have been of any great concern to the players involved (826ab).

Capture and Safety
A passage in the rules section of the text introduces the concepts of "squares of death" (mṛtyugeha) and "squares of refuge" (śaraṇāgāra) (849-52). Squares of death do not appear to be specific squares marked on the game board, but rather any square in which a pawn can be captured. Once a pawn has entered the outer quadrants, it is liable to be captured on any square other than a square of refuge. This happens if an opposing pawn, moving up from behind, ends its move on the same square as the pawn in question. The pawn is then returned to start, and has to begin the game anew. Since squares of refuge are specific squares protecting pawns against capture, they must have been marked on the board in advance, though no information is offered about their number or placement. Rules for capture and safe squares are common to most South Asian race games, and the usual way to mark the latter is by means of an "X," sometimes referred to as a goose-foot (haṃsapāda) (e.g., KK: 158). The placement of safe squares can vary greatly between cruciform race games, 40 and if this was already the case at the time of the Mānasollāsa, it may have contributed to the lack of instructions on where to place them. It might also be noted, though no such rule is found in the text, that many South Asian race games require players to capture at least one opposing pawn before they can begin to bear off their own pawns. 41

Winning and Losing
The text states that a player who exits all their pawns from the game wins, and a player who does not loses (858cd). While it seems obvious that the first player to exit their pawns is declared the winner, the final passage of the text, discussed in more detail below, makes it clear that there is only one loser in the game. This 39 E.g. Temple 1884: 244. 40 E.g. Balambal 2005: 47-48, 53-59, 61-64. 41 E.g. Balambal 2005 means that in a sixteen-player game, fifteen players must complete the game with their pawns before it is over. We might be excused for thinking that once the excitement of finding a winner is relieved, there is little point in continuing the game, but this is not necessarily how traditional games were or are played. Early descriptions of four-player paccīsī record that it was played until only one player was left. 42 Similarly, the four-player game of nāṉ ku kaṭṭam tāyam (Tam. நான்கு கட்ட தாயம்) played in modern day Tamil Nadu may continue until both a winner and a loser have been declared. 43 In modern day Karnataka, the related fourplayer game of cauka bāra (Kan. ಚೌಕ ಬಾರ) sometimes continues until all players have completed the game, whereupon they are assigned titles of king (rāja), minister (mantrin), general (senāpati), and soldier (sainika), depending on whether they completed the game first, second, third, or fourth (Kulirani and Vijayendra 2011: 113).

S O C I A L CON T E X T T HE SECTIONS ON PLEASURES (upabhogaviṃśati), entertainments (vinodaviṃśati),
and games (krīḍāviṃśati) in the Mānasollāsa reveal much more than the activities themselves. They allow us to enter into the daily life of a twelfth-century king, and introduce us to some of the many ways in which he would spend his leisure time at court. We often find him surrounded by a host of women, and even when they are not at the center of his attention, they tend to figure somewhere in the background. Daud Ali has shown how the ubiquitous presence of "palace women" around the king is a fixture in courtly sources from at least the Gupta period onward. 44 According to the Mānasollāsa, the women would take their places behind the king in the assembly hall (sabhā), and "frequently cast glances in his direction to cause him joy". 45 Interaction between the sexes at court would often take place in the form of play, and courtship was indeed conceived of as a game or a contest. 46 That board games were a part of this repertoire of playful interaction from early on is confirmed by the Kāmasūtra which makes several mentions of them. They are counted among the sixty-four arts (kalā) to be mastered by the urban elite (KS 1.3.15); they are found among the accoutrements of a gentleman's household (KS 1.4.4); and they constitute one of the tools used by him to win over women from a young age (KS 3.3.6-7). 47 Depictions of 42 Williamson 1801: 22;Shurreef 1832: liii. 43 Balambal 2005: 50. 44 Ali 2004: 114. 45 Ali 2004: 113. 46 Ali 2004 The Jayamaṅgalā commentary on the Kāmasūtra glosses the unidentified board game ākarṣakrīḍā as pāśakakrīḍā (KS 1.3.15), identified as backgammon in the Mānasollāsa. Since Yaśodhara wrote his commentary about a century after the Mānasollāsa, it is possible that he, too, intended pāśakakrīḍā to mean backgammon. This, however, does not mean that the original ākarṣakrīḍā, as it appears in the Kāmasūtra, can be identified as backgammon.
board games in sculpture, painting, and story literature are also partial to the motif of a loving couple throwing dice and moving pawns. 48 It should therefore come as no surprise that board games form the subject of several chapters in the Mānasollāsa.
The location of the chapter on phañjikā alerts us to the fact that it was not solely, and perhaps not even primarily, occupied with how the game was played. An initial clue is provided by the preceding chapter on phaṇidākrīḍā, or the game of placing bets. 50 Here the king surrounds himself with a select audience of wo-48 See, for example, the sculptures of Śiva and Pārvatī playing a backgammon-like game (Soar 2007), the miniature paintings of royal couples engaged in games of especially caupaṛ (e.g., Biswas and Chopra 1982: 7-8, pl. 5; Figure 3 in the present study; cf . Finkel 2004: 51), and the episodes of both gods and men playing board games in the regional literatures (e.g., Handelman and Shulman 1997). 49 Shrigondekar identified the second of the three games described in the chapter on jyotsnākrīḍā as the physical race game known as āṭyāpāṭyā in Marathi and āgarapāṭa in Gujarati (Shrigondekar 1925-61: v. 3, ix). Āṭyāpāṭyā (spelled aṭyāpātyā) was glossed as "[a] play amongst children" in Molesworth's Marathi-English dictionary (Molesworth 1857: 14), and was described in detail in Deodhar's Sacitra marāṭhī kheḷāñceṁpustak, or "Illustrated Book of Marathi Games" (Deodhar 1905: 15-13;cf. Kamath 2020: 22-23). A Tamil-Sinhala version was also described by Parker (1909: 627-28). 50 Following the example of phañjikā in deaspirating the initial pha, I derive phaṇidā from the nominal stem paṇa (bet, wager) and the verbal suffix -dā (giving, granting). Another, albeit less plausible, derivation would be from phāṇita (sugarcane juice), with reference to the sugarcane stalks used for target practice in the game. Mishra (1966: 501) derived phaṇidā from phaṇīndra, another name for the cosmic serpent Śeṣa which Viṣṇu rests upon during the dissolution of the universe. This, however, does not fit the context of the game in any meaningful way. men (preyasī), poets, musicians, intellectuals, gamblers, wrestlers, and others. He then proceeds to display his skills at sleight-of-hand tricks and expertly chopping up sugarcane stalks, coconuts, oranges, and other fruits. The audience is encouraged to place bets on the success of his endeavors (Mān. 5.15.806), but the text ascribes other motives to the king than winning bets. It states that he should engage in the exploits for the sake of enjoying himself and others, receiving praise from his bards, inspiring his fighters and wrestlers, and increasing the women's love (premavṛddhi) for him (Mān. 5.15.800-1). As we shall see below, the loving attention sought by the king in phaṇidākrīḍā also plays an important role in the game of phañjikā. A further contextual clue to the underlying reasons for playing phañjikā is found in the succeeding chapter on timirakrīḍā, or the game of darkness. Here the king first fills an underground chamber or curtained-off room with attractive young women. He then ushers in a group of children to perform various antics, such as pulling the hair of the women, squeezing their breasts, and partly disrobing them, all with the express purpose of breaking down their barriers of courtesy (dākṣiṇya) . When a sufficiently boisterous mood has been stirred up, the king himself enters the scene and starts groping the women until, finally, they return his favors and bring him to a state of supreme pleasure (paramāṃ prītim) (Mān. 5.17.912). A similar concept of gameplay as foreplay is found in the chapter on phañjikā.
In summary, we can see that phañjikā is grouped not only with ludic but also with erotic pastimes. In the following brief sections, the social context of phañjikā will be further explored with regard to the setting of the game, the king's motives for engaging in it, and the rituals of shaming associated with it.

THE WOMEN'S QUARTERS
The only thing we know for certain about the setting of phañjikā is that it was played in the evening or during the night (826ab), signaling that it was played before bedtime and possibly used by the king as a pretext for choosing which woman to sleep with. The text does not state where the game was played, but given that the women summoned by the king should be skilled in the game (phañjīkrīḍāviśārada) (817cd), it seems obvious that they were familiar with the game outside of the specific context in which the Mānasollāsa describes it. The long hours spent by the women in the antaḥpura, or women's quarters, would have made it ideally suited to the playing of games, and we know from later sources that the family of race games to which phañjikā belongs was often associated with women and children. Miniature paintings from the Mughal period provide several examples of women playing games in the antaḥpura, 51 and the 51 See, for example, Patnaik 1985: 72, fig. 15, Finkel 2004 fig. 3.14, and Figure 4 in the present study. Cf. Topsfield 2002: 58. study of the period in Rajasthan by G. N. Sharma (1968: 132) specifically identified caupaṛ as "a common pastime in the harem". 52 We might therefore easily imagine a scenario where the women would habitually play phañjikā and other games in the antaḥpura, and where the king would sometimes join in as a playful yet amorous way of interacting with them. Though only dating from 1743 CE, a miniature painting of Jagat Siṃh II of Mewar (r. 1734-1751 CE) playing eighthanded caupaṛ with 18 women seems to be portraying a closely related situation ( Figure 5).

AMOROUS PLAY
The chapter on phañjikā begins with a description of the women the king should call upon to play it. They should be young, beautiful, charming, flirtatious, full of laughter, and infused with the emotion of love (premabhāva) (816cd-817). In fact, the text implies that they should be comparable to the milkmaids (gopī) with whom we are told the god Kṛṣṇa used to play the game (820ab). 53 Modeling the playing of the game on the encounters between Kṛṣṇa and the milkmaids not only legitimizes the king's engagement with it, but also creates an amorous context for what might ensue during and possibly after the game. Contrary to the other sedentary games described in the Krīḍāviṃśati, the king's purpose is not merely to enjoy the game and succeed in it. Just as the game is said to have delighted the hearts of the milkmaids, the king should play it with the women at his court to see their various emotional states (bhāva) (825cd) and "their many flatteries pregnant with feelings of sexual desire" (cāṭūn bahūn ... śṛṅgārarasagarbhitān) (820cd-821ab). One of the stratagems employed by the king is to show partiality toward certain women and take note of their reactions. While some become affectionate when he favors them, others become jealous when he does not favor them (821cd-822ab). Another stratagem is to touch and let himself be touched by the women, and look out for similar expressions of love and jealousy among them (823cd-824ab). He should also be mindful of how the women behave toward him with respect to the game, and observe whether they want him to win or be defeated (822cd-823ab). Besides the sheer enjoyment of riling the emotions of the women, the king would likely use their various intimations to 52 Kalpna Chaudhry's doctoral thesis on the depiction of women in the Mughal period also noted that the women of the antaḥpura would be engaged in playing games during their leisure hours (Chaudhry 2014: 3). 53 Several games described in the Krīḍāviṃśati section are associated with Kṛṣṇa. He is said to have invented one of the games described in the chapter on jyotsnākrīḍā (Mān. 5.8.366cd), and to have played the other two games described in the same chapter . Additionally, in the chapters on toyakrīḍā (Mān. 5.5.259cd) and madirāpānakrīḍā (Mān. 5.10.513), the king is likened to Kṛṣṇa playing with the milkmaids. form his opinion about them and decide which of them to favor sexually. According to Ali, referring to the Kāmasūtra, the king would select which woman to sleep with at night based on little gifts received from them, 54 and it certainly does not defy the realm of possibility that phañjikā could have been used as a similar pretext for choosing among the women.

SHAMING THE LOSER
The chapter concludes with a passage on how to ridicule and shame the loser of the game. The other players should draw mocking images of her, climb onto her back and drive her around "like a beast of burden" (vāhavat), blindfold her and direct her to an agreed upon spot, or lead her away to the sound of clapping and singing like "someone possessed of inauspicious marks" (vilakṣaṇa) (859-62). Similar practices are recorded for other games described in the Mānasollāsa. In the chapter on kukkuṭavinoda, or cock-fighting, the winning team should climb onto the backs and shoulders of the losing team and shame them with ridicule (Mān. 4.7.1131), or mark their chests and foreheads with saffron images of cocks (Mān. 4.7.1166). The practice of climbing onto the backs of the losers is repeated in the description of the running-and-feinting game described in the chapter on jyotsnākrīḍā (Mān. 5.8.375ab), and seems to have survived into much later times. 55 Curiously, an ethnographical study of traditions associated with an isolated survival of paccīsī in the Khorezm region of Uzbekistan in the 1950s records similar and even harsher forms of punishment and humiliation for the losers. 56 A further study of shaming the losers in the aftermath of ritual games associated with the goddess Pattini (Sin. පත්තිනි) in Sri lanka was undertaken by Obeyesekere. 57 It thus becomes apparent that the specific practices described in the Mānasollāsa form part of a larger cultural practice related to play, games, and ritual.
An unexplained peculiarity of the final passage is the use of the masculine singular in reference to the loser. Not only are women described as the primary players of the game, but the verse immediately preceding the passage specifically refers to women as winning or losing the game (858cd). It is, of course, possible 54 Ali 2004: 225. 55 Kittel's Kannada-English dictionary describes a similar practice for the outdoor jumping game paṟ inuṟ i (Kan. ಪಱಿನುಱಿ) recorded in the thirteenth-century Kannada grammar Śabdamaṇidarpaṇa and later known as hāṟ uguppē (Kan. ಹಾಱುಗುಪೆಪ್ ) (Kittel 1894(Kittel : 956, 1651. 56 Practices reminiscent of those described in the Mānasollāsa include sitting on the back of the loser and driving him around like a donkey, blackening his face with soot, and clapping at him mockingly. Among the more extreme examples given is that of tying the loser to a ladder used to carry corpses, and leaving him outside overnight in the cold of winter. A practice that would allegedly sometimes result in his death (Snesarev 1963: 10-11). 57 Obeyesekere 1984: 499-508. that the masculine is simply used as a neutral way of referring to the loser, but one almost gets the sense that the passage could have been directly copied from somewhere else. It does not refer to phañjikā specifically or to any practices related to it, and comes across as sufficiently generic to be appended to the description of any other game. This is further strengthened by the fact, as shown above, that the practices described in it apply to several other games, and perhaps even to a wider tradition of shaming. It should, however, be noted that an early verse in the text speaks of women wanting to ridicule the king if he should lose the game (823ab). Whether ridiculing the king according to the practices described in the text would be considered socially acceptable is unclear to me, though the physicality of some of the practices might be seen as appropriate in the context of amorous play. As Ali noted, courtship was often conceptualized in antagonistic terms with lovers physically assaulting each other (Ali 2004: 236 f), and exactly such a situation is in fact described at length in the chapter on timirakrīḍā .

LU D I C CON T E X T
T HE CATEGORY OF RACE GAMES to which phañjikā belongs is by far the earliest category of board games attested anywhere in the world. Race games like the Egyptian senet and the Mesopotamian game of twenty squares date back to between the fourth and third millennium BCE (Crist et al. 2016: 41-44, 82-84). Fragments of what may have been boards for the game of twenty squares were found at various Harappan sites (ibid. 82), which would make it the earliest known race game in South Asia. The famous aṣṭāpada board, consisting of an 8 × 8 grid, was first mentioned in Buddhist and Jain texts dating to the early centuries BCE, 58 and though the board later came to be associated with chess, it originally appears to have been used for race games. 59 From the mid-1st millennium CE until the early centuries of the second millennium CE, backgammon is the race game most frequently alluded to in Indian art and literature, including the Mānasollāsa where it is referred to as pāśakakrīḍā (Mān. 5.13). 60 Backgammon shares many ludic concepts with phañjikā, but appears to have enjoyed a higher status at court, which places it at a further remove than phañjikā from the race games played outside the court. Backgammon, however, would eventually fall out of favor with the Hindu courts, and the inclusion of phañjikā in the Mānasollāsa may be seen as a sign of things to come. Certainly, from the fifteenth century onward, the predominant game associated with first Hindu and later Mughal 58 Bock-Raming 1999: 43-44. 59 Murray 1952. The early history of the aṣṭāpada board was outlined by Bock-Raming (1999: 43-48). 60 A comprehensive survey and discussion of early visual and textual references to backgammon-like games in South Asia is offered by Soar (2007). Cf . Syed 1998. courts was the cruciform game of caupaṛ, which is much closer related to phañjikā than to backgammon. 61 The inherent difficulty in situating phañjikā in the wider history of traditional South Asian board games is the complete lack of references to it outside the Mānasollāsa. While this might lead us to conclude that it was a unique game confined to the court of King Someśvara III, later references to a wide range of related games suggest that it is the reference to phañjikā rather than the game itself which should be viewed as an anomaly. Considering that caupaṛ would later appear as a game played both inside and outside the royal household, and that several other related games are primarily documented as folk games in vernacular sources, a plausible explanation is that phañjikā was a courtly adaptation of a simple race game deemed unfit for inclusion in the Sanskrit sources which form the basis of most studies of traditional South Asian board games. In this connection, it is worth remembering that board games, and especially board games played by common people, constitute what is essentially an oral tradition. Even to this day, they are often sketched on the ground with a piece of chalk or similar, and played with pebbles, shells, seeds, and other ephemeral materials substituting for pawns and binary dice. 62 Rules are rarely written down, but rather communicated and agreed upon between players, which accounts for the bewildering array of variant rules sometimes encountered within the confines of even a single game. 63 The malleability and ever-changing nature of traditional games is important to keep in mind as we delve deeper into the relations between phañjikā and other games for which we only have later evidence.
Games and the relations between them can be productively described in terms of ludemes. 64 Ludemes are "conceptual units of game-related information" that can be used to define the "form" and "function" of a game. 65 They can 61 Schmidt-Madsen 2021: 35-36. 62 Balambal 2005cf. Sinha 2013: 23-24. 63 This has often been commented upon in relation to caupaṛ and paccīsī, none of which conform to any standard set of rules (e.g., Parlett 1999: 43). 64 The word "ludeme" was coined in the early 1970s to signify "a constituent of a set of rules in a game" (Depaulis 2019: 23), but has only recently caught on among board game scholars. A working definition was offered by David Parlett in 2016: "[A] ludeme or 'ludic meme' is a fundamental unit of play, often equivalent to a 'rule' of play; the conceptual equivalent of a material component of a game. A notable characteristic is its mimetic property -that is, its ability and propensity to pass from one game or class of game to another." (Parlett 2016: 82). 65 Piette et al. 2021: 7. The definition given here is adopted from the current work of Cameron Browne and his colleagues in the Digital Ludeme Project (http://ludeme. eu/). The project uses ludemes to define and model traditional strategy games digitally and allow them to be played and analyzed by an artificial intelligence (https: //ludii.games/). The ludemes are then mapped geographically and chronologically to suggest ways in which they may have traveled between cultures and influenced the spread of games and game concepts, allowing for a better understanding of the universal history of board games. The project either be self-contained or composed of multiple lower-level ludemes, each of them expressing a rule or a concept. Examples of ludemes include the number of players, the distribution of squares, the application of dice, the movement of pawns, etc. Every game can be broken down into its constituent ludemes, which can then be compared to the ludemes of other games. A game need only differ from another game in a single ludeme or two for it to be considered a variant. Variants are often fairly obvious, such as whether or not players are required to capture an enemy pawn before they can enter the final square of the game track, or whether or not they have to make a special throw in order to bear off their pawns from the final square. Other variants, such as changes to the layout of the game board, can make games look and feel completely different, even when they are in fact closely related. The advantage of a ludemic approach is the ability to discern similarities and differences at a level of detail which not only allows for analyzing individual games, but also for determining their relationship to other games with which they share key ludemes.
Consider the following ludemic description of phañjikā: Phañjikā is played by multiple players on a cruciform game board consisting of a central square and and four arms with 6 × 6 squares. Each player controls a group of five identical pawns moving along a unidirectional track according to the throws of seven binary dice. Certain squares on the track are marked as safe squares, indicating that pawns resting in those squares cannot be captured by other pawns. A pawn resting in any other square is captured and sent back to the beginning of the track if a pawn controlled by an opposing player ends its move on the same square. A special throw termed phañjikā is required for entering a pawn into the game, as well as exiting a pawn from the game. The first player to exit all their pawns wins the game, while the last player to exit all their pawns loses the game.
Disregarding the specifics of how the game board is laid out, and the exact number of pawns and dice used to interact with it, the above description is true not only of phañjikā, but also of numerous other South Asian race games documented in textual, visual, and ludic sources from at least the fifteenth century onward. While changing the specifics and adding or removing minor rules may impact the length of the game and the level of interaction between the players, it does not change the overall concept of a race toward a goal, or the key mechanics of promises to provide new insights into the playing of traditional games for which we only possess an incomplete set of ludemes, such as phañjikā, as well as furthering our understanding of how ludemes travel horizontally and vertically between games and families of games.
entering, exiting, protecting, and capturing pawns. The anthropologist Charles J. Erasmus warned us that comparisons based on surface resemblances can be misleading, causing cases of independent origination to be confused with cases of cultural diffusion. 66 Erasmus, however, was specifically commenting on the suggestion that the Mesoamerican game of patolli might be related to the Indian game of paccīsī, for which there is little, if any, supporting evidence. In the case of phañjikā and the games proposed here as being related to it, the ludemic evidence is much stronger, and the region in which the games appear, albeit several centuries apart, is one and the same. The argument against cultural diffusion, or rather cultural continuity, would therefore seem more difficult to make than the argument for it. Though we cannot trace the exact genealogy of the family of games to which phañjikā belongs, nor claim with certainty that games only attested several centuries later were contemporaneous with it, we can demonstrate a strong degree of similarity between the ludemes that inform them. Table 2 presents a detailed ludemic comparison between the description of phañjikā as it appears in the Mānasollāsa and the descriptions of three related games as they appear in later sources. Each description is based on a single historical source to avoid conflating multiple variants of the same game, and should by no means be considered "standard" descriptions. The games chosen represent the three South Asian families of cruciform, square, and single-track race games, i.e., the ones that have the most in common with phañjikā. 67 The families derive their names from the fact that the shape of their game boards is the key ludeme that separates them from each other. Other ludemes may vary between them, but never beyond the observation that they all derive from the same limited pool of ludemic information.
The following three sections discuss the families of cruciform, square, and single-track race games in more detail, aiming to provide an overview of the various families and how they relate to phañjikā. While the discussions provide useful concepts for thinking about phañjikā, we should be careful not to retroject those concepts onto it, as this may obscure and distort what is truly unique about it. Thus, an important difference is that many later race games are primarily folk games, while phañjikā is a court game. And judging from other known courtly adaptations of folk games, such as the sixteen-armed variant of caupaṛ designed by Akbar in the late sixteenth century, 68 and the complicated karmic and astrological games designed by Mahārāja Kṛṣṇarāja Oḍeyar III in the nineteenth 66 Erasmus 1950: 382. 67 The names of the families are adopted from Murray who distinguished between "Games on Cruciform Boards" (Murray 1952: 132-140), "Games on Square Boards" (ibid. 129-32), and "Single-Track Games" (ibid. 140-44). 68 Blochmann and Jarrett 1873-94: 304-5. century, 69 we should expect phañjikā to be more elaborate and refined than the games from which it was adapted.
In Table 2 below (p. 196), the analysis of phañjikā is based on the chapter on phañjikākrīḍā in the Krīḍāviṃśati (Mān. 5.16). The analysis of the pan-Indian cruciform race game paccīsī is based on the earliest known set of rules for the game paccīsī, 70 probably recorded in Bengal where Williamson served as an officer in the army of the East India Company. 71 The analysis of the Tamil square race game tāyam is based on the testimony of a Tamil informant recorded by Bell. 72 The reason for using a Tamil game is that the only early descriptions of the related Kannada game of cauka bāra are too brief to form the basis of a detailed ludemic analysis. See, however, the section on Square Race Games below, p. 202. The analysis of the primarily South Indian single-track race game pañci is based on the early to mid-nineteenth-century descriptions found in several manuscripts of Kṛṣṇarāja Oḍeyar III. 73

Cruciform Race Games
The family of cruciform race games is named after their characteristic crossshaped game boards, consisting of a central square with a single rectangle, or arm, extending from each of its four sides ( Figure 6). They are usually played by two or four players, or teams of players, with each player controlling the pawns of a single arm, or sometimes two arms in a two-player game. In four-player games, partnerships are often formed by players seated across from each other, requiring them to carefully coordinate their moves with their partner. The pawns either start on predetermined squares on their own arm, or in the central square above their own arm. They are usually moved by throws of six or seven binary dice, or two or three four-sided stick dice. The goal is to move the pawns down the central column of their own arm, counter-clockwise around the outer columns of all four arms, and then back up the central column of their own arm and into the square at the center. The player -or players, in case of a partnership game -first to complete the circumambulation of the game board with all their pawns wins the game. As with most other traditional race games there is no standard set of rules, but rather a shared pool of ludemes from which specific instances of the game can be created. 69    Paccīsī is the cruciform race game that resembles phañjikā the most. 74 The most striking similarity between them, apart from their cruciform boards and basic method of play, is their names and the special throws to which they refer. Just as a throw of five cowrie shells face-down in phañjikā is called phañjikā, or a group of five, so a throw of five cowrie shells face-up in paccīsī is called paccīsī, or a group of twenty-five. Both throws are considered special throws, which in the case of phañjikā means that a player can use them to enter or exit a pawn, and in the case of paccīsī to move one pawn twenty-five squares and another pawn a single square. The single-square move, which is also awarded by certain other throws, is sometimes required to enter a pawn onto the board, and is the only way a pawn situated on the final square before the central square can move into it. Though the rules of paccīsī come with numerous other subtleties and variations, which are obviously lacking in the case of phañjikā, the naming of two cruciform race games after identically configured throws (i.e., five cowrie shells face-up or face-down) with identical key features (i.e., entering and exiting pawns) seems an unlikely coincidence. If paccīsī represents a later stage in the development of phañjikā, or perhaps rather games related to phañjikā, we might speculate that the calculation of the earlier throw of five cowrie shells was multiplied by itself for a total of twenty-five to speed up the game and increase the drama. Since twenty-five also plays a key role in the layout of the paccīsī board, which counts twenty-five squares from the safe square at the bottom of one arm to the central square at the top of the next, it is possible that the change in layout from phañjikā to paccīsī was influenced by a change in the value of the eponymous die rolls.
The main difference in layout between the game boards is the arms, which consist of four 6 × 6 grids in phañjikā and four 3 × 8 grids in paccīsī. 75 This means that the route followed by the pawns on a paccīsī board cannot easily be transferred to a phañjikā board. On analogy with cruciform race games in general, it seems plausible that the pawns in phañjikā would have circumambulated the central square, but exactly how remains unclear. Since the game accommodates up to sixteen players, we might speculate that two or four columns on each of the outer quadrants functioned as home columns, whereby pawns would enter and exit the game. Though the former solution would only result in eight home columns, the earliest known set of rules for paccīsī does in fact state that if five or more players take part individually, some players will simply share their home columns. 76 If we consider the possible routes on a phañjikā board with 8 versus 16 home columns, it quickly becomes apparent that the former would make for a much better game than the latter. A board with 16 home columns would lead to a very crowded route around the edges of the board, with pawns constantly being captured and returned back home (Figure 7). A board with 8 home columns, on the other hand, would result in a long but potentially interesting game with many 75 A cruciform game from Syria and Lebanon identified as "Edris a jin," or the game of the genie, by Stewart Culin in the late nineteenth century is played on four arms divided into grids of 4 × 8 squares and interconnected at the bottom by further rows of 8 squares (Culin 1898: 857;cf. Bell 1969: II, 2-3). Details about how the game was played is unfortunately lacking. 76 Williamson 1801: 9. difficult decisions of when to hold back and when to press on ( Figure 8). However, before we get too carried away, we should remember that neither routes nor home columns are mentioned in the description of phañjikā. 77 Unless further evidence turns up, it is therefore unlikely that we will ever know the exact route followed by the pawns.
The earliest datable reference to a cruciform race game other than phañjikā is found in the Sūfī romance Mirigāvatī written in Avadhī by Qut .. ban Suhravardī 77 For the possible translation of talaṭallaka or talakūṭaka (856) as "home column," see footnote 153. in 1503 CE, more than three-and-a-half centuries after the Mānasollāsa. 78 The romance refers to the game of caupaṛ, 79 which also finds mention in several other contemporary sources. 80 Caupaṛ is a more complex game than paccīsī, and differs from it in several respects, including the use of stick dice instead of cowrie shells and the initial placement of pawns on the arms of the board instead of in the central square. The first detailed description of caupaṛ appears in the Ain-i-Akbari written by Akbar's court historian Abul Fazl around 1590 CE. 81 Interestingly, Fazl noted that "[f]rom times of old, the people of Hindústán have been fond of this game [caupaṛ]," but without additional evidence we can only speculate as to how old those times actually were. As for paccīsī, the earliest references I am aware of date from the late eighteenth century, 82 while the earliest detailed description dates from an apparently unsuccessful attempt at introducing the game to the British public by Thomas Williamson in 1801. 83 However, given its status as "the poor man's chaupar", 84 it is quite possible that it existed silently alongside the more respectable caupaṛ from much earlier times. 85 When caupaṛ first enters the historical record, it does so as a game played both inside and outside the courts. In Mirigāvatī (c. 1503) and Padmāvatī (c. 1540), it is played by kings, princes, and princesses, and in the latter work it is even said that there is a caupaṛ board in every palace. 86 Metaphorical mentions of the game in contemporary bhakti, or devotional, poetry aimed at a broader swathe of the population indicate that it was also played by common people. 87 This is exemplified in the seventeenth-century Caurāsī vaiṣṇavan kī vārtā, where we find a group of men playing caupaṛ by the side of the road. 88 The Cetovinodanakāvya from 1822 distinguishes between three forms of the cruciform game (dyūta), stating that the learned (śiṣṭas) play it with three dice, the twice-born with two dice, 78 An unpublished doctoral thesis on the history of games in Andhra Pradesh by Mopidi Kallappa claims that earlier references to cruciform race games can be found in works of Telugu literature dating from the late twelfth/thirteenth to the late fifteenth centuries (Kallappa 2006: 203, 205). According to Nagaraj Paturi, who kindly looked into the relevant passages for me, there is no clear indication that the games referred to are cruciform race games (personal communication). I will return to the question of vernacular sources and their untapped potential at the end of this study. 82 Two early references to "paucheess" can be dated to 1792 (Hunter 1801: 314) and 1798 (Tennant 1804: 366). 83 Simplified and vastly more successful forms of the game would appear as Parcheesi in the US in 1867 (Whitehill 1999: 118) and as Ludo in England in 1886 (Copisarow 2010: 208). 84 Brown 1968: 49. 85 Williamson confirmed Brown's appellation in describing the game as "one of the principal recreations of the poorer classes" (Williamson 1801: 4). He also noted that the game is "in high vogue among the ladies" (Williamson 1810: I, 362  Graphic design by the author. and the śūdras with cowrie shells (CVK 257cd-58). The appearance of phañjikā as a cruciform race game played with cowrie shells at a twelfth-century Hindu court suggests that cowrie-based versions of cruciform games only came to be frowned upon by the upper classes later. At the same time, the status of phañjikā as a game primarily played by women, and only engaged in by the king to stir up their sexual desire, suggests that it was considered a lesser game than chess and backgammon, and quite probably one also enjoyed outside the palace walls in one form or another. While cruciform games would never lose their attraction among the general populace, as evidenced by their prevalence during festival celebrations (Rettberg 2008: 37), it may only have been in the fifteenth century that they came to enjoy their status as the royal game par excellence.
Another theme that connects phañjikā with caupaṛ is the idea that it might be played for purposes other than the game itself. While King Someśvara uses phañjikā to look into and open up the hearts of the women who play it, the king in Mirigāvatī and the princess in Padmāvatī both use the game to divulge whether their opponent is truly the prince that he claims to be. In Mirigāvatī, the prince proves himself by playing the game four-sided (caubaghī), meaning that two players each control two sets of pawns, and have to exit one set from the board before the other (Doniger and Behl 2012: 95). 89 In Padmāvatī, the princess describes the game as a metaphor for love (Shirreff 1944: 189-90, v. 27.23), which the prince then promptly expands upon (ibid. 190-91, v. 27.24), thereby proving his identity and proceeding to make love to the princess in a passage which itself uses the language of caupaṛ as a metaphor for love-making (ibid. 194, v. 27.31). In the Ain-i-Akbari, too, Fazl insists that Akbar has "higher aims" with the game, and that he uses it to "[weigh] the talents of a man" and "[teach] kindness". 90 Clearly, caupaṛ was considered to have special properties beyond that of a mere game, especially when it came to revealing the true nature of a person's character, whether in terms of love, talents, or identity. 91

Square Race Games
Square race games are often described as simplified versions of cruciform race games, and in the late nineteenth-century Krīḍākauśalya, which refers to cruciform race games as dyūta, square race games are known as dyūtārdha, or halfdyūta (KK 303-5). 92 They are played by two, three, or four players on a square grid with an odd number of rows and columns (usually 5 × 5, 7 × 7, or 9 × 9), though even-numbered grids are also sometimes found (Figure 9). Each player controls a set of pawns which move according to the throws of binary dice, with the exact number of pawns and dice depending on the size of the game board. 93 The pawns enter play on the outermost central squares along the four sides of the game board, and move counter-clockwise around the edge of the board until 89 Note that the translation of sukaṭhā sorahī as "sixteen cowries" in verse 144h is probably wrong. Sukaṭhā, tentatively identified as the name of a particular throw by Doniger and Behl (2012: fn. 121), is more likely to be cognate with Marathi sōkaṭī (Skt. sārī + kāṣṭha) in the sense of a wooden pawn. The translation "sixteen pawns," referring to the total number of pawns in a four-sided game, would be a better match for the context. 90 Blochmann and Jarrett 1873-94: 304. 91 Simlar properties are also attributed to other traditional South Asian games, such as, for example, the mancala game of cenne maṇe (Kan. ಚೆನೆನ್ ಮಣೆ) played in the Tuluspeaking region of Karnataka (Claus 1986). 92 For a general description of square race games, see Parlett 1999: 54-56. 93 Among the Tamils and Sinhalese of Sri Lanka, two stick dice configured as 1 : 3 : 6 : 4 are sometimes used on the large 9 × 9 boards (e.g. Parker 1909: 605). they arrive back at the square immediately to the left of their entry square. They then move up into the next concentric circle of the board, and follow it around in a clockwise direction. Depending on the size of the game board, they keep on spiraling inward in a clockwise direction until they arrive in the square at the center of the board. The goal of the players is to get all their pawns into the central square, from where they sometimes also have to be borne off by special throws of the dice. The first player to complete the route with all their pawns wins the game. Since each player enters the concentric circles from different sides of the grid, the ways in which the routes of the pawns overlap can be difficult to discern at a glance, adding an extra element of unpredictability and drama to the game. 94 It has been suggested that cruciform race games developed from square race games as a way of making the route followed by the pawns easier to track (Murray 1913: 40, Parlett 1999, which begs the question as to whether the influence of square race games would be more pronounced in an early cruciform race game such as phañjikā. As we have seen, the phañjikā board consists of five 6x6 grids arranged in a cross-like formation. It might therefore be suggested that phañjikā developed as an attempt at making a cruciform race game by combining the grids of five square race games. This would explain our difficulty in tracing the route of the pawns across and between the outer quadrants of the phañjikā board. The late Rangachar Vasantha speculated that the outer quadrants might have been used to play a series of square race games, with the completion of one game sending pawns directly into the next. 95 While the quadrangular shape of the grids and the low range of possible throws would fit the context of a square race game better than that of a cruciform race game, the suggestion raises numerous other questions, such as the role of the central quadrant and the semi-circles or triangles drawn within it. If phañjikā was indeed influenced by square race games, we will need more than a simple retro-projection of later games onto it to establish what exactly that influence was. 96 94 The route followed on the popular 5 × 5 grids, with its single change of direction from counter-clockwise to clockwise, is reminiscent of the labyrinthine military formation known as the cakravyūha, or wheel formation, associated with the killing of Arjuna's son Abhimanyu in the Mahābhārata (Smith 2009: 418-32). Cakravyūha patterns are found among the royal games of Kṛṣṇarāja Oḍeyar III (Gowda 2020: 81), as well as on a crudely embroidered playmat in the Shreyas Folk Museum in . A further compar-ison between square race games and the architectural grids of the vāstupuruṣamaṇḍala is found in R. G. Raghavan 2020: 213-16. 95 I am grateful to Vasantha's son Raghuveer for providing me with this information during a private meeting in Bangalore in early 2020. 96 In my doctoral dissertation on the original Indian game of snakes and ladders (gyān caupaṛ), I argued for a similar interpretation of the phañjikā board, with the central quadrant serving as the playing surface and the outer quadrants as stor- Figure 9: Sample 7 × 7 square race game indicating the route followed by the pawns belonging to the lower side. Graphic design by the author.
Murray believed that the aṣṭāpada, or 8 × 8 grid usually associated with chess, was originally used for square race games (1913: 32ff.). The aṣṭāpada traces its history back to about a millennium before the emergence of chess around the sixth century CE, and while literary and material evidence indicate that games were played on square grids of different dimensions in the early period, we do age spaces for the pawns (Schmidt-Madsen 2019: 63-68). I have since revised that hypothesis, and in a more recent article on the early history of caupaṛ and paccīsī, I argue for an interpretation closer to the one suggested in the present study (Schmidt-Madsen 2021: 58-59). not know if games similar to later square race games were among them. 97 The earliest references to square race games are in the form of visual representations of games which, however, cannot be identified with certainty. Two reliefs on the railing of the Vaneśvara Mahādeva temple in Dungarpur, Rajasthan, dating to at least the sixteenth century, depict a king and a queen playing a game with conical pawns on what in one case appears to be a 5 × 5 board (Figure 10). 98 Similarly, a miniature painting from eighteenth-century Jaipur, Rajasthan, shows a king and a queen engaged in a game on a 4 × 4 board with two diagonal lines drawn between opposite corners to create four triangular shapes reminiscent of those found in the central quadrant of the phañjikā board. 99 Several graffiti boards incised in stone, many of them from the Deccan Plateau, may be even older than the above examples, but as is often the case with graffitis, it is impossible to know exactly how old. 100 97 Game boards with 8 × 8 and 10 × 10 squares are mentioned in early Buddhist and Jain literature (Bock-Raming 1999: 43-44). A bas-relief from the Bhārhut Stūpa Figure 11: Relief of a royal couple playing a game on a 5 × 5 board (cf. Figure 10), a small cruciform board, or similar. Vaneśvara Mahādeva temple, Dungarpur, Rajasthan. Photo by Leander Feiler.
The earliest textual reference to a square race game that I am aware of is the mention of "āthā chumuk" in a list of games played by South Indian Muslims in the early nineteenth century. 101 The name as written would seem to be an alternative rendering of Telugu aṭṭa chemma (Tel. అటట్ చెమమ్), commonly used as a name for square race games in Andhra Pradesh. Clearly identifiable references to square race games only appear in the game manuscripts of Kṛṣṇarāja Oḍeyar III, written in a mixture of Sanskrit and Kannada in the mid-nineteenth century. Several of the manuscripts include illustrations and brief descriptions of cauka bāra (Kan. ಚೌಕ ಬಾರ), or four-twelve, as played on a 5 × 5 grid. 102 An ethnographcal account of a similar Bengali game known as aṣṭākaṣṭe, also played on a 5 × 5 grid, is found in an uncredited article by Lāl Bihārī De, often transliterated as Lal Behari Day, in a contemporary issue of the Calcutta Review. 103 dating from the second half of the second century BCE shows four men, or two men and two attendants, engaged in an unidentified game on a 6 × 6 grid. Another bas-relief from the Bodhgayā Stūpa dating from the sixth or seventh century CE shows a man and a horse-headed woman playing a like-The evidence for square race games, though later and less substantial than the evidence for cruciform race games, agrees in associating them with people both inside and outside the royal courts. Shurreef & Herklots identified "āthā chumuk" as being played by the children of the "vulgar" classes and Day listed aṣṭākaṣṭe as a game played by the women of the "middling" classes. 104 Though a version of the game also made its way into Kṛṣṇarāja's courtly manuscripts, it is obvious from their encyclopedic nature that the game might very well have been a folk game played in contemporary Mysuru. The two reliefs from the Vaneśvara Mahādeva temple in Dungarpur and the miniature painting from Jaipur, if indeed they depict square race games, provide more certain evidence that the game was also played in royal households in Rajasthan. The reliefs from Dungarpur are especially interesting in the present context, as they appear among a sequence of reliefs showing the king enjoying the company of several voluptuous and scantily clad women. One of the reliefs have three women watching the game between the king and the queen, and while the queen seems focused on what appears to be a four-or six-sided die in her hand, the king is clearly fondling the breasts of one of the women in the audience (Figure 11). Given the amorous context in which phañjikā was played, we cannot help but wonder whether the Mahārāja of Dungarpur, too, played race games with the women of his court for purposes other than the games themselves.

Single-Track Race Games
The game boards of single-track race games are much more varied than the game boards of cruciform and square race games. This owes to the free-form method of combining the five-square seqments that make up the track, which can be extended and looped at will to lengthen the game and change the dynamics of player wise unidentified game on an 8 × 8 grid. Both scenes are discussed at length in Bock-Raming 2000. 98 The dating is based on a eulogistic inscription (praśasti) in the temple from 1561 CE (Ojhā 1936: 90  interaction. 105 The shape most frequently encountered consists of five such segments laid out one after the other at perpendicular angles ( Figure 12). Every turn of the track, including the square at the very end, is marked with an "X," indicating that pawns resting on those squares cannot be captured by other pawns. Two further segments, starting from opposite sides and joining in the middle, are added to the beginning of the track as lead-ins for the two players, or teams of players, competing in the game. Each player controls a set of pawns which move along the track according to the throws of binary dice. Special throws are usually required to enter the track, and pawns can only be borne off by throws of exactly one more than required to reach the final square of the track. The first player to complete the track with all their pawns and bear them off wins the game.
Parker noted that the name may either derive from a throw of five cowrie shells face-up (pañca) or from the positioning of safe squares at intervals of five squares each along the track. 107 Circumstantial evidence would seem to favor the former explanation, which also applies to the related games of phañjikā and paccīsī. In fact, Kittel's Kannada-English dictionary glosses pañci (Kan. ಪಂಚಿ) as: "Five or a cinque, at a game". 108 Interestingly, one of the references cited by Kittel is the Basavapurāṇa, which was translated from Telugu to Kannada by Bhīmakavi in 1369 CE. 109 Though the exact game played in the Basavapurāṇa remains to be identified, it narrows down the gap between the throws of phañjikā and pañci to some two-and-a-half centuries. 110 As previously mentioned, a dice game called pañcikā, named after a throw of five binary dice face-up or face-down, was already known in the seventh century CE. It would therefore seem that phañjikā, despite the lack of references to it outside the Mānasollāsa, belonged to a larger family of dice and board games centered around the number five. Norman Brown, in his seminal article on caupaṛ and paccīsī, suggested that the earliest versions of paccīsī might have been played with five instead of the usual six or seven cowrie shells. 111 And if we go back even further, we find that five dice in the form of vibhītaka nuts were handed over to the king before the commencement of the ritual dice game during the Vedic Rājasūya ritual, 112 and that another Vedic dice game played with two cubical dice also revolved around the number five. 113 Contrary to paccīsī and cauka bāra, the track in pañci traces its course along a sequentially arranged line of squares rather than around a grid. This makes comparison with the 6 × 6 grids in phañjikā difficult, but a few points are still worth noticing. The five-square track segments in pañci, and the right angles at which they are joined, can be said to form a series of invisible 6 × 6 grids where only the squares along the edges are used. This is particularly clear in variant layouts seen in undated graffiti boards which loop the track back onto itself, thus effectively creating one or more 6 × 6 grids with the central 4 × 4 squares left blank ( Figure 13). 114 Some of the resulting shapes resemble crude versions of cruciform and square race games (Figure 14), and it has indeed been suggested that single-track race games led to the development of more complex race games. 115 Since the routes followed by the pawns in games like paccīsī and cauka 107 Parker 1909: 610. 108 Kittel 1894: 924. 109 Rice 1921 In this connection, it should be remembered that the Mānasollāsa also refers to phañjikā as phañji (e.g., v. 818) and phañjī (e.g., v. 817). 111 Brown 1968: 50. 112 Heesterman 1957: 141, 143-45. 113 Falk 1986 In this connection, it should be noted that the evolution of race games has often been traced back to dice games. The general idea is that the need for keeping track of scores prompted the need for score boards, which then, in turn, developed into simple race games (Parlett 1999: 35-36). 114 For on-site documentation of singletrack graffiti board variations, see Rogersdotter 2015. 115 Parlett 1999 bāra do not allow us to reconstruct the route followed in phañjikā with any certainty, the evidence gathered from pañci should also be included in our considerations. Could it, for example, be that the pawns in phañjikā did not move along all the squares in the 6×6 grids, but only used the squares along the outer edges? This would significantly shorten the number of moves required to cover the entire game board, which would fit well with the low range of throws in the game. Following the example of pañci, this might also mean that the safe squares were placed in the corners of the grids at a distance of five squares each. Furthermore, if phañjikā throws were not only used to enter and exit pawns, but also to move already entered pawns a distance of five squares, as is usually the case in related race games, it would mean that a pawn positioned on a safe square could move directly to the next safe square on a throw of phañjikā. This, of course, is only speculation, but it helps to build the context required to arrive at a more detailed understanding of phañjikā and its place in the wider history of traditional South Asian games.
The earliest reference to a single-track race game I am aware of is the inclusion of pañci among the mid-nineteenth-century games and game manuscripts of Kṛṣṇarāja Oḍeyar III. 116 Kṛṣṇarāja's game is a more complex version of the basic form of the game described above, and includes 12 five-square segments, eight of which are arranged in a loop requiring enemy pawns to move past each other in opposite directions. As noted above, single-track race games are also known from graffitis etched into the floors of caves, temples, and ruined structures across India, with particularly large concentrations found in South India. 117 Extensive documentation of graffiti boards among the ruins of Vijayanāgara, which served as the capital of the eponymous empire from 1336-1565 CE, led Vasantha to state that single-track race games were found "[p]ractically everywhere". 118 The same conclusion was reached by Elke Rogersdotter who showed them to be the most prevalent type of game at the site, with 203 occurrences among 576 samples. 119 Unfortunately, as in the case of other graffiti games, they cannot be dated with any certainty, and may only have been made by local villagers, shepherds, soldiers, and others long after the city was abandoned in 1565. Attempts have also been made at tracing single-track race games back to the Mesopotamian game of twenty squares, tentatively 116 See, for example, p. 313b of the Caturaṅgasārasarvasva, dated 1843 CE, in the British Library (microfilm IOL Neg 4589), and p. 2a of the Caduraṅga cakra (Kan. ಚದುರಂಗ ಚಚಕ ), dated 1859 CE, in the Kuvempu Institute of Kannada Studies, Mysuru (acc. no. KB515). Earlier still would be the announcement of "The New Game of Paunchee, or Twenty-Five" in the London newspaper The Morning Post in 1824 (Copisarow 2010: 195). While the accompanying description and a surviving illustration clearly identifies the game as paccīsī (ibid. 196), it opens up the interesting possibility that pañci was used interchangeably with paccīsī. 117 Marin 1942. 118 Vasantha 2003: 33. 119 Rogersdotter 2015  associated with the Indus Valley Civilization, but that would lead us too far into the realm of speculation for our present purposes. 120 Little can be said about the social status of single-track race games, except that the frequent occurrence of graffiti boards indicates that they must at one time have been a popular pastime among common people. I am not aware of any representations of them in sculpture or painting, nor of any references to them in primary sources other than the game manuscripts of Kṛṣṇarāja Oḍeyar III mentioned above. They do. however, form part of modern celebrations of the Tamil

CO NC LU S IO N T HIS STUDY HAS ATTEMPTED to shed light on a little known medieval Indian
board game and the social and ludic contexts in which it appears. The study has shown that the game belongs to the category of race games in which two or more players, or teams of players, move their pawns along a track and compete to be the first to complete it. While the shape of the game board would place it in the subcategory of cruciform race games, the size of the grids that make up the arms of the cross, and the lack of information about the route followed by the pawns, prevent us from relating it directly to later cruciform games such as caupaṛ and paccīsī. Comparison with other subcategories of race games, such as square and single-track race games, further suggests that no single game known from later sources can be successfully retro-projected onto phañjikā. As demonstrated by the numerous similarities between phañjikā and later race games, this does not mean that phañjikā should be treated as an anomaly in the history of South Asian board games. Rather, it should be seen as having grown from a large and varied body of undocumented race games which only entered the historical record several centuries after the Mānasollāsa was written in 1129 CE.
Perhaps the strongest evidence for explaining the apparent vacuum in which phañjikā exists comes from the scarcity of evidence pertaining to other race games even after they enter the historical record. The one exception is caupaṛ which became increasingly associated with royal households in the fifteenth century, and became the courtly game par excellence during the rule of the Mughal emperor Akbar in the second half of the sixteenth century. This would seem to indicate that race games primarily existed as folk games, and only found representation in art and literature when adopted by courtly culture. Race games played on aṣṭāpada and backgammon boards preceded phañjikā in this regard, and while phañjikā never seems to have enjoyed any great popularity outside the court of King Someśvara III, it may have prepared the ground for the later success of caupaṛ at Hindu and Mughal courts alike.
It is important to understand that the mere inclusion of phañjikā in the Mānasollāsa does not necessarily mean that it enjoyed the same status as other games, such as chess and backgammon, also included in the work. As the study has demonstrated, phañjikā was primarily associated with women, and only engaged in by the king for amorous purposes. As such, the description of phañjikā is incidental to descriptions of other erotic pastimes enjoyed by the king. This hints at the possibility that phañjikā was considered a women's game, and likely one used to while away the leisure time in their own quarters. The fact that young boys were also counted among those eligible for playing the game further associates it with children, which not only strengthens the supposition that it was not considered a serious game, but also increases the likelihood that it was played by common people outside the court. The form that it would have taken outside the court can only be speculated upon, but since the game board was drawn on the ground and only required cowrie shells or other binary dice to play, we can easily imagine that it also existed in simpler and less elaborate forms than the one described in the Mānasollāsa. Whether these forms were closer to later cruciform, square, and single-track race games remains to be seen, but evidence clearly shows that race games adopted for courtly play underwent complex transformations allowing for more players and longer play times.
By way of conclusion, I would like to suggest that more satisfying answers to the questions raised in this study may be found in the regional literatures of India as they developed in the first half of the second millennium CE. As Sheldon Pollock and others have shown (Pollock 2003), the emergence of regional literary cultures during this period facilitated a localization of Sanskrit culture, evidenced, for example, by the numerous adaptations of the Mahābhārata into regional languages. This, in turn, gave rise to a whole new body of literature expressive of local customs and traditions, opening up a window into the everyday life of cultures largely ignored by earlier literature. One aspect of this literature is the description of games and pastimes prevalent among people from widely different levels of society both urban and rural. Research into traditional South Asian games has largely based itself on Sanskrit and Prakrit sources, with little attention paid to the vernacular sources which became increasingly dominant in the centuries after the Mānasollāsa was written. A closer examination of the regional literatures, including those that existed contemporaneously with the Mānasollāsa, promises to fill out many of the blanks that exist around medieval Indian board games in general, and around phañjikā in particular. It could also help bridge the gap between textual and archaeological sources, finally allowing the innumerable graffiti boards their rightful place in the conversation.
Significant work in this direction has been undertaken by the late Vasantha (2006) and her doctoral student Kallappa (2006) with regard to Telugu literature, and by Amit S. Deshmukh (2021) with regard to Marathi literature. While the value of this work cannot be underestimated, it also shows the inherent difficulties that must be contended with. Literary references to games usually lack the precision that would allow us to identify the exact games alluded to. Just because a text mentions game objects like boards, dice, and pawns, it does not mean that we can name the games played with them, and even in rare cases when names of games are actually given, we cannot be certain that those names referred to the same games then as they do now. A case in point is the description of a game played between Kṛṣṇa and his consorts Satyabhāmā and Rukmiṇī in the Uttaraharivaṃśamu written in Telugu by Nācana Sōmana around the midfourteenth-century. Despite the claims of Vasantha and Kallappa that the description represents the earliest known literary reference to caupaṛ or paccīsī, 122 it does not provide much information beyond the mention of a board (palaka), pawns (sāre), and dice (pāsika) (UHV 3.120). A verse detailing the range of results obtained by throwing the dice allows us to infer that the game was played with two four-sided stick dice configured with the numbers 1, 3, 4, and 6 (UHV 3.121), but further details about the game are lacking. While the description certainly does not preclude the possibility that the game could be a version of caupaṛ, sometimes played with two stick dice similar to those inferred, 123 it could also be practically any other race game known or unknown to us. Instead of jumping to conclusions and begin retro-projecting knowledge of later games onto vague descriptions of earlier games, we have to patiently collect the available evidence and say as much as we can without saying too much. Only then can we hope to unravel the mysteries of phañjikā and all the other medieval Indian games awaiting discovery. [हिरत] Emendation by Shrigondekar (accepted) [रत] Emendation by Shrigondekar (rejected) [नील] Emendation by the author added in square brackets (see Table 3). It is unclear which of his readings derive from MS D (1671 CE), and which from MS A (1873-74 CE) or the closely related MS G (undated). 125 Alternative readings supplied by Shrigondekar in footnotes sometimes make this apparent, but I have not included them here to avoid unnecessary clutter. Readers interested in consulting the footnotes can easily do so in the original edition. It should also be noted that some of the emendations suggested by Shrigondekar are in fact silently adopted from MS A (MS Pune BORI 115 of 1873-74 (Shrigondekar A) Having summoned (a group of) young and beautiful women, infused with the emotion of love, coquettish, flirtatious, fond of laughter, charming, and skilled in the game of phañjī, the king should play the game of phañji with them (816cd-818ab). (There should be) five, seven, six, eight, nine, (or) sixteen players (818cd). 126 Women and young boys should both take part in phañjikā; women, in particular, should take part; only men should not take part (819).

AP P E N D I X A : T E X T A N D TR
[ Kṛṣṇa played this game, 127 delighting the hearts of the milkmaids [gopī]; therefore, women should be foremost (among those who play it) (820abc). The king should play (the game) with them in the same way (as Kṛṣṇa) in order to witness their many flatteries pregnant with feelings of sexual desire (820d-21ab): "Which (women) regard (my) partiality for another (woman) with jealousy? Which woman appears full of love because of (my) partiality (for her)? Which (woman), partial to me, wants me to win? 128 Which woman wants 126 An apparent confusion between the words khelaka/lekhaka (player/writer) and khelana/lekhana (playing/writing) runs throughout this and other chapters on games in the Mānasollāsa.
This might simply be due to metathesis (i.e., the accidental switching around of syllables), perhaps aggravated by the fact that the chapter on varāṭikākrīḍā, or gambling with cowrie shells, does indeed refer to writers or notaries (lekhaka) responsible for keeping track of throws, stakes, fines, and outcomes . It is also possible that the physical act of noting down signs for various throws resulted in players (khelaka) and playing (khelana) sometimes being referred to as writers (lekhaka) and writing (lekhana). 127 Kṛṣṇeneyaṃ kṛtā krīḍā could also be taken to mean that Kṛṣṇa invented the game. 128 The verse may be taken in the double sense of the woman wanting the king to win both the game and herself. to ridicule (me) after defeating me? Which (woman) watches me jealously because another woman's hand touches (me)? Which (woman) has the hairs of her body stand on end because my hand touches her (821cd-24ab)? One longing and affectionate woman asks (me) for a phañjikā (throw), and another becomes hostile toward me when a phañji is thrown (824cd-25ab)." 129 The king desires to see these and various other emotions (825cd).
[  129 If we reject the emendation in v. 824d, and instead emend te to me, the verse could also be translated as: "One woman (develops) a longing (for me) and (becomes) affectionate toward me during (the game of) phañjikā, and another (woman) becomes hostile toward me upon losing in phañji." However, translating phañjipāte as "upon losing in phañji" would run counter to the otherwise consistent use of the causative of √pat as referring to a throw of the dice. 130 The use of bhadraka, or bhadra as it is also termed in later verses (cf. vv. 837cd, 851ab), derives from architecture, where a bhadra signifies a projection from a central square (Kramrisch 1946: v. 1, 210). In an earlier chapter of the Mānasollāsa on architecture, the terms caturbhadra and bhadra are indeed used interchangeably to denote the fourfold projections from the central square of a building  I will now explain the different types of cowrie shells (used as pawns and dice) by the names of their colors (830cd): the yellow ones, the white ones, the black ones, the red ones, the pink ones, the variegated ones, the shiny ones, the tawny ones, the yellow-bellied ones, the ones that have the color of the spotted tiger, the saffron-red ones, the ones that have the color of the black antelope and the donkey, the ones 132 I take khaṇḍa (part) in the sense of khaṇḍendu (crescent moon that have (the color of) the belly of a snake, the ones that shine like ghee and milk, the ones (that look) like bhallātaka nuts, 135 and the ones marked with black and white stripes are known as soṃllala (831-33); 136 some (of them) have the splendor of the mongoose, (others) the same luster as the jackal, (some are) like kalāya seeds, (others) like rice grain, (and some are like) various (kinds of) pearls, captivating and differently shaped (834-35ab). One should arrange (the cowrie shells) into groups of five with the same form and color; a combined group of five is called a dhāman [i.e., a "family" of pawns] (835cd-36ab).
[RULES] The seven large, uniform cowrie shells mentioned (above) are (used as dice) for the purpose of playing. By means of phañjikā (throws), dāya (throws) should be made according to the combinations (resulting from the fall of the cowrie shells) (838): 138 If all (the cowrie shells) are face-down, the dāya (throw) is known as saptaka [i.e., seven]; it is marked with a line moving like cow's urine [i.e., a wavy line] (839). If one cowrie shell is face-up (and) six are face-down, the dāya (throw) is called ṣaṭka [i.e., six]; in this case, a straight line is drawn (840). If two (cowrie shells) appear face-up (and) five (appear) with their face to the ground, a phañjikā [i.e., five] is declared; it is counted without (drawing) a line (841). If three (cowrie shells) are face-up (and) four face-down, the dāya 138 The idea seems to be that a phañjikā throw allows a player to make a dāya throw, which can then be used to move a pawn from the central to the outer quadrants of the game board (cf. v. 847-48). An alternative translation of phañjikāsaṅkhyayā might be "according to the (system of) enumeration in (the game of) phañjikā." (throw) amounts to catuṣka [i.e., four]; it is marked with a circle (842). If four (cowrie shells) are face-up and three are positioned face-down, the dāya (throw) is called trika [i.e., three], and should be marked with a goose-foot (843). 139 If five (cowrie shells) are face-up (and) two have their face to the ground, the dāya (throw) is called dvika [i.e., two]; it should be marked with a goad (844). 140 If six (cowrie shells) are face-up and one has its face to the ground, the dāya ( As many phañjikās as (a player) throws on the ground, that many dāyas they should throw at the same time before ceasing (to throw) (847). 141 The initial dāya (throw) [pūrvadāya] (for moving a pawn from the central to an outer quadrant) is seen (on any throw) other than a phañjikā. 142 The position of a pawn (entering an outer quadrant) is in accordance with the counting of squares (from the central quadrant); 143 this is the rule (848). 139 In ludic contexts, a goose-foot (haṃsapada) usually refers to an "X" drawn in a square on a game board to indicate that pawns resting in that square cannot be captured (cf. v. 852). A similar mark may be intended here. 140 Goads (aṅkuśa) are hybrid tools with a pointed end and a hook, which may explain their use as representative marks for the number two. 141 The meaning seems to be that phañjikā throws award additional throws, and that every phañjikā throw should have a matching dāya throw.
Instead of translating ni+√vṛt as "ceasing (to throw)," we might translate it as "moving down (from the central to the outer quadrants)." The meaning would essentially be the same, i.e., that phañjikā throws (used for entering pawns) must be followed by dāya throws (used for moving pawns). 142 Perhaps a slightly better reading would be to emend pradṛśyate to na dṛśyate: "Without a phañjikā (throw), the initial dāya (throw) is not seen." The meaning, however, would remain the same, i.e., that the initial dāya throw (used for moving pawns from the central to the outer quadrants) must be a dāya throw following upon a phañjikā throw. 143 Taking  As the pawns of the players enter (the outer quadrants) one by one, they should return each other from the "square of death" (849). 144 If a pawn arrives at that square (of death) from behind by means of dāya (throws), then it claims [i.e., returns] the previous (occupant of the square) and is (itself) not returned from that (square) (850). A pawn returned (from a square of death) should go back outside the outer quadrants, 145 and move sequentially through the squares once more by means of dāya (throws) (851).

[Safe squares] फिजकायां िथतायां [°ता या] च शरणागारगा च या। न कदािचभवे तया दायापाते [दाय°] िनवतर् नम ् ॥ ८५२॥
When a (pawn) has entered (the game of) phañjikā and is abiding in a "square of refuge," 146 it should never be returned (from that square) on the fall of a dāya ( If, by means of the throws, no phañjis or three phañjikās come up, then 144 Presumably a square occupied by one player's pawn and landed upon by another player's pawn, resulting in the newly arrived pawn returning the previous occupant to its starting position outside the game board (cf. v. 850). 145 Cf. v. 837cd. 146 Presumably a safe square on the game board marked with an "X" or similar.
147 If we reject Shrigondekar's emendation of phañjikāyāṃ sthitāyāṃ, we might translate: "If a pawn goes to a 'square of refuge' on the throw of a phañjikā, it should never be returned (from that square) on the fall of a dāya (throw)." This, however, would imply that phañjikā throws allow pawns to jump between safe squares, which is not seen in other traditional South Asian race games. all (the throws) become void (853). 148 When by fate (only) a single pawn remains (available) for moving, (then) on a throw of more phañjis (than required) for entering (the central quadrant), the exit (of the pawn from the outer quadrant) is in vain ( If a pawn has gone out from the central quadrant [talageha] 150 when a kalasapta(ka) (throw) appears, it should move around (the game board) and return (to the half-moon-shaped home space in the central quadrant); 151 it [i.e., the pawn] is known as "possessing the half-square" [ardhagṛhiṇī, i.e., the half-moon-shaped home space] (855). 152 148 The suggested emendation of daśa (ten) to trayaḥ (three) is informed by the often repeated rule that three special throws in sequence become void. This is recorded in the earliest known set of rules for paccīsī (Williamson 1801: 16-17). Cf. this verse on paccīsī in the Cetovinodanakāvya: daśa pañcaviṃśatir vā pratyekaṃ samuditāś ca patitāḥ / avyavahitaṃ trivāraṃ cet tarhy ete nirarthakāḥ syur iha // (CVK 380) [If (throws of) ten or twenty-five turn up one after the other and fall three times without interruption, then in that case they all become void]. It is unclear to me why a throw of no phañjikās should also become void, as this would make the game extremely frustrating and slow, unless, of course, the rule only applied to a player who had not yet entered any pawns into the game. 149 Since every phañjikā throw awards an additional throw, it would indeed be possible for a player to throw more phañjikās than required for exiting their last pawn. The rule is also found in the earliest known set of rules for paccīsī (Williamson 1801: 17-18). 150 The exact meaning of tala in the present context is uncertain, but talageha appears to be synonymous with geha in vv. 828cd, 848cd and tala in v. 856 as a reference to the central quadrant of the game board.
Tala may derive its meaning from Kannada tala (ತಲ) or tale (ತಲೆ), glossed as "head" or "being uppermost and principal" by Kittel 1894: 699. The word talak was recorded as having a similar meaning in the isolated survival of paccīsī in the Khorezm region of Uzbekistan in the 1950s (Snesarev 1963: 8;cf. Finkel 2002: 71-74). Alternatively, the unemended daladehagatā (i.e., gone to the body of a petal) or partly emended dalagehagatā (i.e., gone to the squares on a petal) might be taken in the similar sense of a pawn that has entered the outer quadrants. The use of dala (i.e., petal) with reference to the "arm" of a cruciform board is attested in the Kautukanidhi section of Kṛṣṇarāja Oḍeyar III's Śrītattvanidhi written in Mysore c. 1825-50 (ŚTN 9.17.1). 151 Note that returning (punar + √gam) a pawn to the central quadrant by means of a kalasaptaka throw is different from returning it by capturing it (cf. vv. 849-51). A pawn returned by a kalasaptaka throw is considered to have completed the game track and to be ready for bearing off (cf. v. 858ab). 152 It is possible, although in my opinion less likely, that the effect of the kalasaptaka throw should be understood in the negative sense of removing a pawn (pari+√hṛ) from the game board and returning it (punar + In that case, we would have to understand kalasaptaka as a "low seven" in the sense of an "unlucky seven," and ardhagṛhiṇī as "half a housewife" in the sense of a pawn that has not fully completed the track around the game board. 153 The exact meaning of talaṭallaka is unclear to me, but from the context I understand it as signifying a smaller square (ṭalla < tala? + diminutive suffix -ka) on the outer quadrants adjacent to the larger square of the central quadrant (tala). Shrigondekar's MS. A reads talakūṭaka, where kūṭaka could be taken in the same sense of a "small square" (1925: v. 3, 257, fn. 4). An altogether different possibility would be to understand talageha in v. 855 as "the lower squares," i.e., the squares below the central quadrant, and talaṭallaka/talakūṭaka as a "heap" (cf. ṭalla in Turner 1966: 302) of lower squares (tala) in the sense of a home column for the entry and exit of pawns as seen in later cruciform games. This, however, is too speculative to be accepted without further evidence. 154 This clearly indicates that a phañjikā throw is required for a pawn to enter the central quadrant from the last square of the game track. Whether there are other ways to enter the central quadrant (apart from the kalasaptaka throw described in v. 855) is unclear (cf. v. 857). 155 The meaning seems to be that an additional phañjikā throw is required for a pawn to be borne off after returning to the central quadrant. If we accept Shrigondekar's emendation, the throw would be required at the beginning of the game, which would fit well with the usual practice of invoking Gaṇeśa at the outset of an enterprise. This is supported by Molesworth, whose Marathi-English dictionary glosses devagaṇapata as an initial throw (ḍāva) forfeited and given up to the gods (1857: 423). Against this is the occurrence of the verse at the very end of the rules explanation, which would seem an unusual place to introduce a rule applied at the very beginning of the game. If (the pawns of) the women leave (the game board) they win, and if they remain they lose (858cd). (The winner) should sketch the loser on the ground in a state of dejection, (and) draw (him) as untouchable, tarnished, abandoned, and deformed (859); or climb onto the back (of the loser) and cause him to move like a beast of burden, (and) cover his eyes with a piece of cloth and make (him) reach a designated spot ( Shrigondekar described the manuscripts on which he based his editorial work. However, his descriptions were not particularly clear. Additionally, in volume three he regrettably re-used the siglum "A" for a different manuscript than that referred to in volume one. The table on the next page clarifies these issues as far as has been possible without personal inspection.