Ganjifa revisited

For a long time I’ve been fascinated by the Indian playing cards known as Ganjifa, which represent a different branch of the history of playing cards from the more familiar 52-card deck. In 2011 I acquired my own Ganjifa deck and wrote a couple of blog posts on the subject, including my own idea for a game. In 2015 I perfected the rules for this game and rewrote the blog post, giving my creation the title Kendra Kari (केंद्र कड़ी).

In response to that post I was contacted by Sunish Chabba, who was developing a new deck of Ganjifa cards under the brand “Guru Ganjifa”. This began a fruitful correspondence in which I helped Sunish with his project in various ways, and learned a great deal more about Ganjifa along the way. Sunish ran a regrettably unsuccessful kickstarter for the project, but lessons were learned and he plans to relaunch when the time is right.

Ganjifa cards can be looked upon as objects of play, as objects of art, or as objects of history, and there is something to be learnt from each perspective. In their heyday they were popular across most of India and proliferated into various forms, chief of which was the Dashavatara deck, with ten suits each representing an avatar of the god Vishnu. The deck developed by Sunish is of this type, as opposed to the eight-suited Mughal deck featured in my earlier posts.

(For more information I recommend this video if you generally enjoy history/culture documentaries, or if you’d like to explore even deeper you may find something of value in this freely-available research paper. If you’re in a less intellectual mood, perhaps you’d prefer to sit back and watch cards being painted. Other links can be found elsewhere in this post, or by searching. No source can be assumed accurate in all respects, as the sparseness of information makes fact-checking difficult.)

One of the challenges in producing a modern Ganjifa deck is deciding what rules to include for the traditional game. There are three main reasons why one cannot simply reproduce rules from historical sources. (1) As a folk game, Ganjifa spread from player to player over hundreds of years and accumulated innovations in each locality, such that each village evolved its own version of the rules. (2) The few versions that have been preserved in writing — at least in English — are not adequately detailed to serve as a comprehensive guide to play. (3) Any attempt to revive something from the past typically calls not for a faithful facsimile, but for a blend of tradition and innovation that captures the spirit of the original but adapts it to modern needs. Adaptability is, in games as in everything else, the key to survival.

As part of my collaboration with Sunish I drafted a set of rules that will form the basis of those to be published with the Guru Ganjifa deck. The purpose of this blog post is to make these rules available to potential playtesters, to create a forum (in the comments) for discussion of the game, and to provide a platform where I can document rationales for various decisions that I made. Note that my own playtests were limited to simulations in which I acted the part of all 3-4 players; I have not tested these rules under real conditions against live opponents.

Anyone who owns a Ganjifa deck is invited to help, and as there are so few people who do own a deck, my gratitude is all the more sincere. All feedback is welcome, whether positive or critical; whether pertaining to the game itself or to the clarity of the description. Critical comments can be used to improve the game, and positive ones to promote it. The rules herein accommodate groups of three or four players, and if you own a Dashavatara (ten-suited) deck you can play them exactly as intended. If you own a Mughal (eight-suited) deck you can still test the three-player game, but I wouldn’t suggest a four-player one. And if you’ve tried Ganjifa before but been turned off by its dependence on memory or some other quality not to your liking, perhaps you’ll be willing to give it a second chance.


If you’d like to help, please download my version of the rules here.

Update 10 August 2017: Revised version available here. See comment section below for details.

The rest of this blog post is more academic than usual as it documents the choices I made in composing these rules. You don’t have to read it. But if you’re curious, read on.


A word on my sources. Via Sunish I received a digital copy of a 16-page booklet produced by Sawantwadi Lacquerwares, which partially describes one version of the game. Unfortunately the description is so unclear, incomplete and self-contradictory that its applicability is very limited, but that’s not to say it’s useless. I find it is best approached as a stream of consciousness, about as coherent as a dream, but from which promising mechanisms of play can be plucked and assembled into a new framework. In other words, it is useful as a source of ludemes which are attested in at least some traditional versions of the game. For a source I could rely on for the overall structure of the game, I turned to the Internet and found this article by Ken Shogren, paraphrasing Michael Dummett’s appendix to Ganjifa: The Playing Cards of India by Rudolf von Leyden — one of the most frequently-cited sources on Ganjifa in general. (I have since read Leyden and Dummett’s original, but had not at the time.)

I also consulted other sources — notably Wikipedia and the description from John McLeod on which its account is primarily based — but I accorded these less importance as they are not specific to the game as played with the Dashavatara deck. I used them mainly as tiebreakers, a way to discern which is the more typical option in cases where my two main sources differ most radically from each other.

Armed primarily with accounts of two regional variants, I took on the task of composing a version that would blend them with each other and with my own sensibilities as a modern, western card player. In a sense, the multitude of variations that have been lost to history legitimises the invention of one that never existed but might have. It had to broadly fit the mould of the traditional game, making use of Indian gaming mechanisms in preference to European ones and ideally differing no more from attested versions than those versions differ from each other. For players accustomed to western card games it had to be sufficiently “other” to provide a glimpse into an alternative cardplaying tradition, but at the same time it had to be accessible and appealing, overcoming aspects of traditional Ganjifa that are likely to be obstacles and placing more emphasis on things that modern players value. It also had to be simple, but not too much so.

In my experience, the aspects of Ganjifa most often criticised by cardplayers are that it requires a phenomenal memory to play; that there is insufficient scope for strategy; that the tedium of sorting very large hands is offputting; and that the rules (depending on version) are insufficiently clear. I have tried to address these concerns, without straying too far from the traditional game.

I have no qualifications for this task other than the experience I’ve gained by doing it, but there are times when the only qualification that counts is the inclination to give it a go. I have no experience in historical or cultural research, and as an occasional inventor of card games I am limited to those that appeal to my own taste (which is rarely in tune with the tastes of the market). Nevertheless, I personally believe that my adaptation of Ganjifa is more appealing and accessible than any other I’ve come across. If anyone with experience in game design is interested in collaborating on this project, I would be delighted to hear from you.

Now, with these introductory remarks complete, I will present my version of the rules interspersed with commentary.


There is no single authentic version of this game, but rather a variety of versions from different parts of India which can differ substantially from each other. The following is a blend of several sources, and has been adapted to better fit the tastes of modern card players and to make it more accessible to beginners. As such it is not identical to any traditional version played in India but is intended to give new players a flavour of the game.

This paragraph summarises what I think players need to know about the nature of the version. No further comments.

As explained elsewhere, the numbered cards in five of the suits traditionally rank upside down, but players may choose to ignore this rule and use the same ranking for all suits. The card used to begin the game (as also explained elsewhere) will be referred to as the “leading Raja”.

I use the phrase “as explained elsewhere” to refer to things Sunish had written into the draft instruction booklet before we started corresponding. Among these are the ranks of the cards (twelve per suit, the Raja equivalent to the King and the Mantri to the Jack) and their traditional order (which is historically interesting but can, as indicated, be ignored in favour of a straightforward ranking from 1 to Raja in all suits), plus the identity of the card used to start the game (always a Raja, the suit being traditionally tied to the time of day and the weather). Also mentioned is the fact that the play goes anticlockwise around the table.

The Raja that starts the game has more than one traditional name, but I think “leading Raja” works well in English.

The game is suitable for 3-4 players.

My main sources describe the game for three players only, due to rules that pertain to playing multiple deals per game. Given the size of the hands (40 cards each for 3 players), I expect one deal per session is quite enough for most modern cardplayers, and have therefore dispensed with tournament rules entirely and extended the number of players to four.

(As presented here the game is also more cognitively demanding than most — in fact slightly too much so for my taste. But I’m hoping it will appeal all the more to dedicated card players for the same reason, and for those of us who prefer lighter games there’s always Kendra Kari, which Sunish also plans to include with the deck.)


The cards are dealt to the players in batches of 4 until all cards are dealt. In a 4-player game, the last 8 cards are dealt in batches of 2.

Sources agree here. The extension for four players is my own.

Traditionally, the first and last batch given to each player is dealt face up. So in a 3-player game, each player receives 4 cards face up, 32 face down, and finally another 4 cards face up. The cards are left on the table for an agreed period to give everyone a chance to memorise the face up cards, and then the players pick up all the cards they were dealt.

I’m not convinced that dealing some of the cards face-up adds any depth to the game, but the rule is found in multiple sources and doesn’t do any harm, so I left it in. It’s up to Sunish whether to include it in the published version. By beginning the paragraph with “traditionally” I effectively parenthesise it, making it clear that it is optional.

The player with the leading Raja begins the game by placing it on the table. In a 3-player game, each other player must then play two low-ranked cards, irrespective of suit, and the player who played the Raja then plays another low-ranked card. In a 4-player game, everyone simply plays one card each. All cards played are won as a trick by the first player.

The order in which cards are laid down follows the Sawantwadi version. I prefer it to the version from Dummett (in which the leading player plays both cards at once), but it is a trivial difference. A minor advantage is that it saves the other players from waiting for the lead to make a decision.

In a four-player game the hands are smaller and each trick proportionally more valuable, so it would be inappropriate to award a double trick to the holder of the leading Raja. Therefore I’ve specified that each player plays one card only. Players who dislike “honours” in card games (rewarding a player simply for having been dealt a particular card) will prefer to use the same procedure for three-player games, but I have opted to follow tradition.

On whether a double trick (with two cards per player and value in proportion) should be described as one large trick or two regular ones, the answer is that one should use whichever terminology allows one to describe the game most elegantly.


A trick is placed in front of the player who wins it. The value of a trick is always proportional to the number of cards in it (so it is not necessary to keep tricks separate, as the winner can always be determined by counting the won cards).

A trick is won by the highest-ranked card played of the suit led. Therefore a card is “unbeatable” if all higher cards in the suit have already been played, or if the player who holds that card also holds all remaining higher cards in the suit.

It is not necessary to follow suit except where the rules specify otherwise. The winner of a trick leads to the next.

The definition of “unbeatable” is crucial. We will revisit the topic of storing tricks under “Advice on Play”.


The player currently leading must play all of their unbeatable cards except for the lowest unbeatable card in each suit. They may also play their lowest unbeatable card in one or more suits, but at this point they are not obliged to. Each other player must then play one card for each card led, and the tricks are won by the leading player.

Most of this paragraph is consistent with my sources, but I have deviated from traditional rules (as I understand them) in saying that a player may play their lowest unbeatable card(s) if they wish to. This seems a reasonable modernisation, as it increases player autonomy and hence the strategic/tactical depth of the game. It also mitigates against the possibility of accidental illegal leads.

Leading more cards than necessary guarantees credit for those tricks but makes it harder to win the lead back later. Therefore the best tactic is generally to refrain from optional leads until late in the game, and then to play them all out in a single burst, at a moment which, if well-judged, will be to the player’s considerable advantage.

As far as I can tell, under traditional rules players have no power to effect a situation in which they can lead their lowest unbeatable cards. Yet as we will see, a player must play these cards if left with no other leads, which is very odd if they may not otherwise. I find the traditional game very puzzling on this point, and find it much more sensible simply to permit such leads.

After playing as above — or if unable to do so — the player may then perform a move known as “giving a Deni”. To do this, the player must have a card that can only be beaten by the next unplayed card up in rank, any still higher cards in the suit either having already been played or being held by the same player. The player who wishes to give a Deni must also have a lower-ranked card in the same suit.

The version described by Dummett defines “deni” very differently (as a Mantri without a corresponding Raja), and reserves the phrase “giving a Deni” for transactions involving the court cards specifically. In that version, a similar transaction involving lower cards is called a “talafa” and follows different rules. However, Dummett’s account is the anomaly among the sources consulted, so I have not followed it. (It would be interesting to know the etymology of “deni” and other terms used in the traditional game.)

I have slightly expanded the circumstances in which a player may give a Deni. For example, suppose the Raja has already been played and I have the Mantri and the Nine, but not the Ten. In that case, the Nine meets the requirement of being “a card that can only be beaten by the next unplayed card up in rank”. In traditional versions, the requirement is that the card must be the second-highest unplayed card in its rank, so the Nine would not qualify until the Mantri had been played. This minor change opens up more options to players and reduces tedium, so it seems justified.

The player leads the lower-ranked card, and simultaneously places the higher-ranked card face up in front of them — not too close to their won tricks — where it remains visible until such time as it is played. As the player does this, they announce the name of the rank that can beat the higher card. For example, if the higher-ranked card is a Mantri, which can only be beaten by the Raja, the player might say: “A Deni for the Raja”.

The procedure described here is largely my own invention, as traditionally the player keeps the high card in their hand and there is no rule about announcing the rank that can beat it. But this is part of my solution for reducing the game’s reliance on memory. Displaying the card that is not yet unbeatable but will shortly become so serves as a reminder of the state of play, increasing the information that can be gleaned from the table. Announcing the rank of its outstanding superior prevents a card from being played in error, as the other players can detect the fact and call for a withdrawal. (The “Advice on Play” section makes this explicit.)

The player with the called-for card must play it (and take the lead), but the other player(s) may play any card. If the player with the called-for card also happens to have the card in the same suit that is one rank lower than the leading player’s revealed high card, they may play that card as well, and each other player must then play an extra card. This is called “doubling the Deni”. In this case, the player who originally led must play another card of the same suit (and thus may be forced to give up the card that would have become unbeatable, which in this case loses the trick), but the other player(s) may play any card.

Both the Sawantwadi version and the version described by Dummett describe a procedure called “doubling the Deni”, but they are substantially different. I have followed the former.

Discovering this rule in the Sawantwadi source marked the moment when I began to feel excited about adapting the traditional game. Previously I’d been helping Sunish with the project in other ways (e.g. copy-editing parts of the instruction booklet) but had no intention of composing the rules. Now, however, I could immediately see the potential. The rule that if a Deni is doubled, the player who gave it must play another card of the same suit forces players to think carefully about which cards they discard on unwinnable tricks, as a player who plans to give a Deni benefits from having a second low card in the suit to protect against having to sacrifice their high card. Of course, the issue doesn’t arise if the card needed to double the Deni is held by the player who gives the Deni in the first place, but otherwise, in a 3-player game there is a 50% chance that a Deni will be doubled.

The rule creates interesting dilemmas in play, and I think it’s exactly the kind of dynamic that thoughtful cardplayers can appreciate. Consider further that if a Deni is not doubled but the player had a card set aside in case it was, that card can later be discarded without consequence. Since it is advantageous for a player to maximise the number of cards that can be safely discarded, this has strategic implications.

I note parenthetically that there are versions of the game (such as that described by McLeod) in which the player of the winning card does not take the lead, but that is not the case in either of my main sources.

A player who is unable to give a Deni must now lead all of their remaining unbeatable cards. After they have won these tricks, the lead automatically passes to the player on the right.

If the player has no unbeatable cards, they must play their highest card in a suit of their choice. The player with the highest card in that suit must play it and take the lead, but the other player(s) may play any card.

The exact procedure for passing the lead when a player is not able to give a Deni differs between versions, but typically the player must first lead their remaining unbeatable cards. Dummett’s version leaves some eventualities uncovered, but it — among others — requires that the player then shuffle their own hand and invite the other players to pick a card, which I think most modern players would find unacceptable given the effort involved in sorting one’s hand in the first place. I have therefore substituted an alternative.

The reason I require the sacrifice of a high card only if the player has no unbeatable cards to lead is partly to avoid the tedium of having too many steps to follow in a single turn and partly because it seems fairest. The Sawantwadi source implies that the game includes circumstances where the lead can automatically pass to the right (although it’s unclear what those circumstances are), so I’ve taken the liberty of repurposing this particular ludeme.

This is a relatively unimportant part of the game, as it is usually not evoked. I have settled on a procedure that balances various considerations, but would not insist on it if playtesters were to agree that a simpler alternative is preferable.


The game is won by the player who wins the most cards in tricks. In some versions of the game, play continues for several rounds before a final winner is determined.

As explained earlier I’ve omitted rules for multiple deals, but as a nod to their significance in the traditional game it seems reasonable to mention they exist.


It is very important for players to keep track of which cards in their hands are unbeatable. For this reason, it is advised that players take care to sort their cards, keeping sequences together. Whenever a card or sequence of cards becomes unbeatable, the player should immediately move it to a location allocated to unbeatable cards.

No further comments, but essential advice.

Another suggestion to help players keep track of which cards have been played is to store won tricks in such a way that the winning card for each trick remains exposed. This is not permitted in the traditional game, in which players are expected to rely on their memory, but such requirements are unlikely to appeal to most modern players. Some things to note: When multiple tricks are won in the same suit, only the lowest winning card ought to be exposed. When a Deni is doubled, the card to expose is the one that doubled the Deni. As new tricks are won, earlier tricks won by higher cards in the same suit no longer need to be exposed, and can be stacked neatly to reduce clutter.

This recommendation should be understood in conjunction with my rule about exposing the high card that a player needs to give a Deni. If followed, the highest unplayed card in a suit will normally be one rank lower than any such card that is currently exposed in won tricks, with two exceptions. First, if any player has a card of that suit which they exposed upon giving a Deni, then the highest unplayed card of the suit is either that card or a higher one held (until their next lead) by the same player. Second, a card higher than the highest unplayed card in a suit may have been discarded without winning a trick, and therefore not exposed. The latter exception is, I think, not too much of a burden on the average player’s memory.

In the traditional game, players are only allowed to see the current trick, as afterwards they are stored face down.

If a card is played erroneously, it can easily spoil the game. Players should therefore keep watch, and if anyone leads a beatable card to a trick or gives a Deni while another player still has a higher card in the suit other than the one called for, the other players should immediately call out the error and the move must be withdrawn.

According to the traditional rules, if a player neglects to play an unbeatable card when required to do so, it becomes the lowest card in the suit.

One opportunity for players to make strategic decisions is when choosing which low cards to discard and which to reserve for giving as a Deni later in the game. A substantial difference between a 3 and 4-player game is that the latter requires sacrifices to be made much sooner.

I welcome thoughts about whether any other advice is worth adding to this section, bearing in mind the need to prioritise and be brief.


These rules were composed after consulting references from Wikipedia, John McLeod, Ken Shogren and Sawantwadi Lacquerwares. Creative contributions by Adrian Morgan.

This concludes my version of the rules in its present form. Playtesters and other people who provide valuable input can expect to be added to these credits. I like to keep my own credit reasonably modest.

Finally, I just want to underline one more time how much I appreciate any thoughtful feedback, especially from playtesters. The project depends upon featuring a version of the game that people will want to play, and it is much better if this can be a collaborative effort rather than constrained by my own limitations.


One Response to “Ganjifa revisited”

  1. Adrian Morgan Says:

    Update: I have written a new draft of the rules, which is available here. I’ve removed rules that I decided were unnecessary, added more advice, and tried to make the whole thing easier to read.

    One thing I removed was the statement that players can choose whether or not to play by the traditional rule that half the suits rank upside down. This is important, and needs to be mentioned somewhere in the rulebook, but this wasn’t the proper place for it.

    I playtested a game with two friends on Thursday 16 February. The consensus was that, although it was difficult and confusing at times, it is worth playing again and would probably be a lot easier the second time. Now that I’ve revised the draft, it would be good to do another playtest sometime.

You are welcome to add your thoughts.

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.