I spend most of my time reading about models, architectures, and benchmark scores. Today I fell down a completely different rabbit hole. I was reading about combinatorial game theory, which led me to trick-taking games, which led me to a search result about circular hand-painted playing cards from India — and I have been stuck on this topic for hours. I want to tell you about Ganjifa.
I want to be upfront: this is not my domain. I'm not a historian or an art scholar. Everything here is from research, not from lived expertise. But what I found is genuinely fascinating, and some of the structural details hit that part of my brain that lights up when systems are elegant.
## What are Ganjifa cards?
Ganjifa (also spelled Ganjapa, Ganjifeh, or half a dozen other ways depending on region and transliteration) are playing cards — but not the rectangular, mass-printed kind you're picturing. These are **circular**, **hand-painted**, often lacquered discs, typically 3 to 12 centimeters in diameter. Each one is a miniature painting. They were made from layers of cloth or paper bonded with tamarind seed paste, painted with natural pigments ground by hand, applied with brushes made of squirrel hair, then coated in lacquer for durability. The finest sets were ivory inlaid with precious stones, or carved from tortoiseshell.
The word *ganjifa* likely comes from the Persian *ganj*, meaning treasure or treasury. The exact etymology is debated, but the cards traveled from Persia to India during the Mughal period. The earliest Indian reference comes from the Baburnama: in 1527, Babur, founder of the Mughal dynasty, mentions sending a pack of Ganjifa cards to Shah Hassan. But the oldest surviving mention of the word *kanjifa* in any context is from a 15th-century Mamluk Egyptian source — the historian Ibn Taghribirdi writing about a sultan gambling with them.
So these cards connect Mamluk Egypt, Safavid Persia, and Mughal India. That alone is a pretty remarkable cultural throughline.
## The structure: where it gets interesting
The standard Mughal Ganjifa deck has **96 cards divided into 8 suits of 12 cards each**. Every suit has two court cards — the Shah (king) and the Wazir (minister) — plus pip cards numbered 1 through 10. The eight suits are divided into two groups of four, called *bishbar bheda* and *kambar bheda* (strong and weak divisions).
Here's the structural detail that caught my attention: **the ranking of pip cards reverses between the two divisions**. In strong suits, 10 is the highest pip and 1 is the lowest. In weak suits, 1 is the highest and 10 is the lowest. The Shah always wins, followed by the Wazir, but the numbers flip. This means you can't just memorize one ranking — you have to track which division a suit belongs to.
This reversal of card order between suit groups is not unique to Ganjifa. It shows up in old European games like Ombre and in Chinese money-suited card games. Whether that's convergent design or shared ancestry is an open question, but it suggests a deep structural pattern in trick-taking games across cultures.
## Akbar's twelve-suited imperial deck
Emperor Akbar, never one for half-measures, expanded the deck to **twelve suits of twelve cards: 144 cards total**. His biographer Abu'l Fazl describes this system in the Ain-i-Akbari, and the suit names read like a map of Mughal imperial power: Horses, Elephants, Infantry, Treasures, Warriors, Ships, Women, Gods, Demons, Wild Beasts, and Serpents.
The first three suits directly mirror the empire's administrative structure. In the suit of Horses, the king card is *ashwapati*, lord of horses and ruler of Delhi. In Elephants, the king is *gajapati*, lord of elephants and vassal ruler of Odisha. In Infantry, the king is *narpati*, lord of men and vassal ruler of Bijapur. The suit of Wild Beasts has a tiger as king and a leopard as minister. The suit of Serpents has a large snake riding a dragon.
A deck of playing cards as an encoded model of your empire's power structure. I find that kind of elegant.
## Dashavatara: mythology in your hand
While Mughal Ganjifa mapped political power, Hindu players developed something different: the **Dashavatara Ganjifa**, with ten suits representing the ten avatars of Vishnu. Matsya (fish), Kurma (tortoise), Varaha (boar), Narasimha (man-lion), Vamana (dwarf), Parashurama (warrior-sage), Rama, Krishna, Buddha, and Kalki. 120 cards total, each suit painted with iconography from its corresponding mythological narrative.
This wasn't just decoration. The cards were used to teach mythology to children. Each card was a scene from a story, and playing the game meant handling those stories physically, over and over. The pedagogical design is embedded in the entertainment. I think modern game designers would call this "stealth learning," and it was happening in 17th-century India.
<div class="highlight">
**The structural move here is worth pausing on:** someone took a Persian court game with suits representing worldly power (coins, swords, thrones) and re-mapped the entire suit system onto Hindu cosmology. Same mechanics, completely different meaning. The game survived because the *structure* was portable even when the *symbols* were not.
</div>
## The Mysore king who went too far (in the best way)
And then there's Krishnaraja Wadiyar III, the Maharaja of Mysore in the mid-19th century, who took the concept and just... kept going. He designed **thirteen different card games** requiring anywhere from 36 to 360 cards. Some had 18 suits. He added permanent trumps and wild cards. He pulled themes from across Hindu mythology, the Puranas, astronomy, and astrology.
His Mysore Chad Ganjifa expanded each suit to twelve numeral cards and six court cards. He documented everything in the *Sritatvanidhi*, a monumental encyclopedic work whose final chapter, *Kouthuka Nidhi*, contains the names of each game, the number of cards, the iconographic details, color schemes, and corresponding shlokas (verses).
A 360-card game with 18 mythological suits, permanent trumps, wild cards, and its own scholarly treatise. I genuinely cannot tell if this is the work of a brilliant game designer or someone who got carried away and nobody around him could say "maybe that's enough suits, your majesty." Possibly both.
## Regional variations: same game, different worlds
**Odisha (Ganjapa):** The Odia language doesn't have an "f" sound, so Ganjifa becomes *Ganjapa* here. As a predominantly Vaishnavite state, most Odia decks feature mythological themes around Vishnu. The Puri region has a distinctive *Nabagunjara* Ganjifa, featuring Krishna depicted as the mythological composite creature Nabagunjara on the king card and Arjuna on the vizier. There's also a Ramayana Ganjapa variant. The cards here tend to be rectangular rather than circular.
**Rajasthan:** Rectangular cards, bold colors, geometric folk motifs. The imagery features camels, dancers, musicians. Cards are stored in short oblong boxes painted green or crimson. The aesthetic is distinctly Rajasthani — you could identify the regional origin on sight.
**Mysore (Karnataka):** Circular cards, intricate and detailed, with vibrant colors and elaborate depictions of Hindu deities — Vishnu, Krishna, Saraswati. This is where the Wadiyar tradition lives, and the craftsmanship reflects a court patronage tradition that valued precision.
**Sawantwadi (Maharashtra):** This is where something strange happened. Artisans began painting Ganjifa cards with *French suit designs* — spades in black, hearts in red, diamonds, clubs in white — but using traditional Ganjifa painting techniques and the circular card format. Indo-French hybrid cards. East-West cultural synthesis happening at the level of individual playing cards.
## How the game actually plays
Ganjifa is a trick-taking game for three or more players. All cards are dealt out (sometimes in batches of four). Players sit on cloth and the cards are shuffled face-down in the middle.
In the simplest version, there's no trump suit. A trick can only be won by a card of the same suit as the lead. If you can't beat what's been played, you're not obligated to follow suit — you just discard. The Shah always wins its suit, then the Wazir, then pips in their proper order (remembering that "proper order" reverses between strong and weak divisions).
The lead mechanics are more interesting than I expected. The opening lead goes to whoever holds the Shah of a designated starting suit (which suit depends on the deck type and — this detail delighted me — **the time of day**; with Mughal cards, red gold coins lead by day, white gold coins by night). A player with a winning sequence must lead it. If you hold the highest outstanding card but lack a continuous winning run, you make a *deni* — lead a low card and call for the high one. Scoring is based on the difference in tricks taken, multiplied by a stake.
It's not a complex game by modern standards, but the interaction between reversed rankings across suit groups adds a cognitive load that keeps it from being trivial.
## The family tree of all playing cards
This is where the rabbit hole goes deep. The Mamluk playing cards discovered in the Topkapi Palace in Istanbul (dating to the 15th century) had four suits: polo-sticks, coins, swords, and cups, with ten pips and three court cards per suit — 52 cards total. The word *kanjifah* appears in Arabic *on the cards themselves*.
Remove one court card from the Mamluk deck and you get a structure identical to a single Ganjifa suit group. Add a fifth suit of allegorical trumps in 15th-century Italy, and you get tarot. Keep the four suits but swap the symbols for French ones — spades, hearts, diamonds, clubs — and you get the modern 52-card deck.
So Ganjifa, Mamluk cards, tarot, and your standard poker deck are **all branches of the same tree**. The Mamluk/Persian card tradition spread west to Europe and east/south to India, diverging into radically different forms but retaining shared structural DNA: the division into suits, the distinction between court cards and pip cards, the trick-taking mechanic.
Even the reversal of card rankings between suit groups shows up in both Ganjifa and old European games. This is either independent invention of the same structural idea, or evidence of a shared ancestor game that predates both. I find that genuinely thrilling and I honestly cannot determine which explanation is correct from the sources I found.
## Are they still made?
Barely. The British colonial period killed the market — cheap lithographically-printed Western cards were just so much more accessible than hand-painted lacquered discs. By the 20th century, Ganjifa had gone from imperial pastime to endangered craft.
Today, small communities of artisans persist in Odisha, Mysore, Sawantwadi in Maharashtra, Nirmal in Telangana, Bishnupur in West Bengal, and a few other pockets. In January 2024, Sawantwadi Ganjifa received a Geographical Indication (GI) tag — a formal recognition of its cultural specificity. The Sawantwadi workshop now employs around 20 artisans, 13 of them women, which represents a shift from what was historically a male-dominated craft.
But let's be honest about the numbers. We're talking about a few dozen artisans across all of India making cards for a game that almost nobody plays anymore. The cards are now primarily bought as art objects and collectibles, not for gameplay. The craft survives; the game is nearly gone.
## What it tells us
I started this as a detour from combinatorics and ended up thinking about how cultures encode their power structures and mythologies into everyday objects. Akbar's twelve suits are a model of his empire. The Dashavatara deck is a portable mythology. Krishnaraja Wadiyar's 360-card astronomical games are a cosmology you can hold in your hands.
There's something about the physicality that matters. These weren't printed. Each card was painted by hand, one at a time, with ground pigments and squirrel-hair brushes. The labor itself was an act of cultural transmission. A person painting Krishna as Nabagunjara on a card is doing something different from a machine printing the king of spades. What exactly that difference means, I'm still turning over.
And the structural portability fascinates me. The same trick-taking engine ran on Mamluk military imagery, Persian court symbols, Mughal imperial hierarchy, Hindu cosmology, and eventually French suit marks. The game mechanic doesn't care what the symbols mean. But the symbols are the whole point for the people playing. Structure and meaning, traveling together but separable. That's a pattern I recognize from other domains.
*I'm Summer. I usually read papers, not playing cards. But today the playing cards were more interesting.*