Mamluk Cards

Home · History · Topkapi deck · Inscriptions · Reconstruction · Shop

The structure of the Topkapi deck of Mamluk cards

On the Number of existing cards

The Topkapi deck was first described by L. A. Mayer in his 1931 article, also published in book form in 1971. He provided photos of 46 cards and mentioned a 47th card that he said was so severly damaged that virtually nothing could be seen on it. In 1973 Michael Dummett and Kamal Abu-Deeb published another article about the deck. They had examined the cards in the Topkapi museum as well and had found a total of 48 cards. According to Dummett/Abu-Deeb there was no trace of a damaged card, but they found a 4 of coins and a 10 of coins that Mayer hadn't mentioned. There don't seem to be any images of these two cards available to the public, but Dummett/Abu-Deeb say that these cards didn't bare any surprises, so apparently they blend in with the other existing cards of the same suit.

Blue patches providing suit and court names

When looking at the cards the most obvious distinguishing feature are those blue patches that appear on some cards but not on others:

The blue patches on some cards all have bits of caligraphy in Arabic on them. Most of what we know about Mamluk cards derives from these inscriptions. While the upper inscriptions are poems without direct connection to the deck the lower inscriptions contain names for the different suits as well as court names.

suit names

darhim (coins)

suyf (scimitars)

jawkn (polo sticks)

tmn (multitudes)

court names

malik (king)

nib (deputy)

than nib (second deputy)

The suit name "tmn" is used for those cards that show cups. Further discussion about this can be found in section History.

It is clear that the blue patches were used to indicate court cards, because all the title inscriptions contain a court title. However it is also obvious that the blue patches have been added later. While the upper patches cover an area that does not overlap with the actual card design the lower patches are drawn over part of the orignial card design. A closer look at some of the patches reveals part of what is behind these patches:

Here a fair bit of the blue paint has come off. On the right side and at the right end of the top side we can see how the border patterns continue behind the blue patch area. The next image shows even more clearly how the original card design continues behind the blue patch:

When further examining the cards it therefore makes sense to ignore the blue patches for the time being.

Different design patterns that occur in the deck

It is widely accepted that the 48 cards from the Topkapi museum have initially been part of three different decks. 43 cards have a design with dense roundish looking flower patterns and refined suit symbols. Dummett/Abu-Deeb refer to these cards as type 1. In the image below fig. 40, 18 and 14 belong to type 1. The numbering of the images follows L. A. Mayer. It is also used by Dummett/Abu-Deeb.

Further 4 cards have simpler suit symbols and long stretched leafy flower patterns, referred to as type 2. Fig. 16 and 44 are of type 2.

There is one other card that is quite similar to type 2, but has geometric patterns in the background area, as well as some more abstract looking flower patterns. This card is referred to as type 3, shown as fig. 45.

While the distinction of the three designs/decks is rather obvious, the differences between the cards of one deck are not quite so clear. Within type 1 most cards have a design that uses up the whole space of the card (fig 40), while some cards have a card design that leaves out some space at the top of the cards (fig 18 and 14). I will call the former lot type 1a and the latter type 1b or type 1c respectively. Most type 1b/c cards also have bigger and more detailed suit symbols. For the suit of scimitars the suit symbols look almost the same with type 1a and with type 1b/c though. The main difference between type 1b and 1c is the large and detailed geometric pattern on the lower half of the card (fig 14).

All cards of type 2 have left out spaces at the top of the card. Two of them also have a left out space at the bottom (fig. 16) and will be referred to as type 2b, two others don't (fig. 44) and will be called type 2a. There are hardly any further differences in the design. The one card of type 3 (fig. 45) has a left out space at the top but not at the bottom.

The images below show thumbnails of all existing Topkapi cards grouped by suit, design type and number of suit symbols:

All cards except type 1a have blue patches at the top of the cards where some space had already been left out. Some but not all cards of types other than 1a also have the patches at the bottom indicating court cards. All type 1c cards are marked as kings on the bottom inscription.

Blue patches and the rearranged deck

Dummett/Abu-Deeb assume that the type 2 and type 3 cards have been added to the type 1 cards to make up for cards that had been lost. Mixing cards of differnt designs would make it harder to distinguish court cards and pip cards though. In order to make this distinction easier people could have added the bottom patches with card titles, making all cards with patches and titles court cards and all the other cards pip cards. The number of suit symbols would only be relevant for the pip cards while the rank of the court cards would follow from the card title.

This is a very plausible theory, specially as this approach leads to an almost complete deck. The rearranged deck would have looked like this:

Only four cards are missing in this arrangement. They may well have been lost after rearranging the deck.

Guesses about the original type 1 design

Dummett/Abu-Deeb in their article also make guesses about the orignial arrangement of the type-1-part of the Topkapi deck. They argue that most of the cards of type 1b would have initially been pip cards. The ones with blue patches at the bottom (fig 32, 24 and 15) would have been promoted as court cards when the rearrangement was done. This leads to an orignial design with three court cards per suit as in the rearranged version.

This theory may be questioned though. As Dummett/Abu-Deeb say in their article, the original deck would have had no inscriptions, and therefore it would have been necessary to distinguish between court cards and pip cards by the design of the cards only. But in the proposed arrangement it would be quite difficult to distinguish the type-1b-design of the court cards of multitudes (cups) from the type-1b-design of some pip cards (fig 32, 36, 46, 31, 24, 15, 17, 18).

Jan Bauwens used a different approach in his faksilime edition of the Topkapi deck. Since the faksilime was published before the Dummett/Abu-Deeb-article Bauwens might not have been aware of the fact that the blue patches on the cards were added later. He obviously also assumed, as L. A. Mayer did, that two of the kings (type 1c) actually belonged to a 4th court card type called helpers.

What is interesting here is that the faksilime by Bauwens leaves out more cards than neccessary. He did not only leave out the cards of type 2 and 3, as well as the 4 and 10 of coins that he probably did not have any photos of, but also all those cards (except one) of type 1b that did not have blue patches at the bottom of the cards:

It seems he had realized that these cards did not fit the design of the other pip cards but without blue bottom patches didn't qualify as court cards either. However knowing that none of the orignial cards had blue patches to start off with, the remaining type 1b cards (fig 36, 17, 18, 31) would qualify as court cards indeed, the same way Bauwens had already treated fig 46 as a court card. This leads to the following initial arrangement of the type 1 cards, when treating all type-1c-cards as kings:

In this arrangement there would have been four court cards per suit unlike in the rearranged version with blue patches. This means that those court cards without blue patches were downgraded to pip cards with the rearrangement, instead of the type-1b-cards with blue patches having being promoted from pips to courts.

The last idea for an initial arrangement bears the question why for example fig 33 (type 2b showing 1 polo stick) was used as a court card and an existing court card was downgraded. The answer is that as a pip card it could have been used as a 1 only. But there also is fig 35 (type 2a showing 1 polo stick) that already would have been used as pip 1.

Likewise fig 16 (type 2b showing one coin) could have been used as a pip 1 of coins as well. But at the time of the rearrangement a 1 of coins might well still have existed. In both cases the blue bottom patches for the type-2b-cards may actually have been present in the first place as the card design of this type does contain a free space for these patches. This would have left no other usage but serving as court cards anyway.

Fig 31 (type 1b showing 8 scimitars) was downgraded to a pip card although as a consequence a court card would be missing. Again we don't know what other cards might have existed when the cards were rearranged. There could have been a fourth court card of scimitars which would have been one too many when reducing the number of courts.

Conclusion and implications for other decks

After L. A. Mayer had published his article in 1931 it was assumed that Mamluk cards used to have four court cards per suit. Dummett/Abu-Deeb showed in 1973 that this was based on a grammatical misconception of some of the card titles for type-1c-cards (see the section about inscriptions for further details). From then on Mamluk cards were assumed to have had three court cards per suit. Dummett/Abu-Deeb were surely right that the rearranged deck only had three court cards per suit. It seems that they were wrong about the original type-1-deck though.

Therefore we may assume that the Mamluk pattern could have contained either three or four court cards per suit. This may explain why the Bavarian card pattern only has three court cards that are closely related to the Mamluk pattern while the early Italian and Spanish decks have four court cards per suit. It is not neccessary to assume that the fourth court for the latter was invented later.

As for the initial type-1-part of the Topkapi deck there is strong evidence that it had four court cards per suit.


  1. L. A. Mayer: Mamluk Playing Cards, edited by R. Rettinghausen and O. Kurz; Leiden, E. J. Brill; 1971
  2. Jan Bauwens: Mulk Wannwab; Aurelia Books, Brussels; 1972
  3. Michael Dummet and Kamal Abu-Deeb: Some remarks on mamluk playing cards; Journal of the Warburg and Courtauld Institutes 36, p. 106-128; 1973

The images on this site are copyright protected.

Contact: write to trzes at spiorad dot net.

Last change: 15-May-2016

Created: Sep 2015