The Game of Tarot: from Ferrara to Salt Lake City 
by Michael Dummett.
Duckworth, 600 pp., £45, August 1980, 0 7156 1014 7
Show More
Twelve Tarot Games 
by Michael Dummett.
Duckworth, 242 pp., £5.95, August 1980, 0 7156 1488 6
Show More
Show More

During recent decades a variety of very distinguished academics have taken time off from their learned pursuits to write imitation Agatha Christie detective stories, so when I first learned that Michael Dummett, widely regarded as the most formidable philosopher of his generation, was about to publish a book about Tarot cards, I rather naturally assumed that it must be an exercise of this same recreational sort. In a certain very off-centre sense, my assumption was correct. The Preface to The Game of Tarot explains the origins of Dummett’s project roughly as follows.

In 1968, he spent three months in the United States. It was a year of political disaster. Nixon became President; Martin Luther King and Robert Kennedy were both assassinated. But for Dummett things were even worse at home. He was ‘deeply involved in work to combat that racism which has, over the past fifteen years, disfigured our national life and dishonoured our country’, and in 1967 the Campaign against Racial Discrimination (CARD) had disintegrated. Dummett was in deep personal distress. ‘I found it almost impossible to do any more work on philosophy or logic than my teaching duties made essential ... But when one is engaged in what produces constant emotional anxiety, there is need for some kind of refuge, and my new hobby became for me a refuge.’

The new hobby concerned cards of a different denomination. It seems to have been sparked off partly by the fact that when Dummett’s son Andrew came across a book of rules relating to a French version of Tarot, father and son had together learned how to play the game and found that it was a very good one, and partly by the fact that his present collaborator, Sylvia Mann, who is the foremost collector of playing-cards in Britain, had recently published a book on her subject.

The resulting product is altogether astonishing: encyclopedic in its scope; monumental in its learning. Indeed, were it not for the lucidity and elegance of Dummett’s style, one would inevitably describe the general manner as teutonic. The 600 pages of The Game of Tarot are printed two columns to the page with learned footnotes by the hundred, some of them over a column in length.

Let me emphasise from the start that the book is a history of the game of Tarot; it is not a history or analysis of the designs of the Tarot cards, a quite different topic which might turn out to be more interesting than Dummett himself considers at all likely. It should also be stressed that Dummett is the sole author: Mann’s main role was to suggest literary sources which might prove worth investigating and to put Dummett in touch with other playing-card experts throughout the world. The fact that, at the end of the day, a whole galaxy of scholars of the most diverse sorts should have collaborated in the enterprise speaks volumes for Dummett’s powers of persuasion and for his capacity to arouse interest in unlikely lines of argument.

The Shorter Oxford English Dictionary entry for Tarot reads:‘Also Taroc. 1598. [F., ad. It. tarocco (pl. tarocchi), of unkn. origin.] a. One of a set of playing-cards, first used in Italy in the 14th c. Also used in fortune-telling. b. pl. The game played with these’. The general effect of Dummett’s study is to show, first, that while the game of Tarot (initially called trionfi (Triumphs – hence ‘trumps’) did in fact originate in Italy in the 15th century, it later developed many different forms in many different countries, some of which are still played, and, secondly, that, insofar as the Tarot cards have any connection with fortune-telling, this is only a recent (and largely British) development.

The bulk of the book consists of a meticulous description (types of pack, values and denominations of the cards, rules of play) of all the variations of the game of Tarot so far as they are now identifiable. One way or another, that includes pretty well every card game involving several players that anyone has ever heard of, as well as a great many that they have not. The reference to Salt Lake City in Dummett’s subtitle concerns contemporary American card games known as Frog and Solo. These require a pack of 36 cards and three players, which puts them in the same family as Tapp, a South German game, which is in turn closely related to Württembergiseher Tarock, which, apart from the name, seems to bear hardly any resemblance to the early Italian games, most of which used packs of either 78 or 62 cards. In all, 211 types of game from 11 different countries are fully described, along with a further scatter which are ‘not fully described or only conjectured’.

Insofar as these matters might aid our understanding of the rules of the different games, Dummett also discusses the structure of the different types of pack and, in a general way, the typology of the suit designs and of the pictures on the various triumph cards. Both books are generously illustrated. But this aspect of Dummett’s researsh was secondary. The books are not a guide to the specialist problems faced by collectors of playing cards.

Indeed, a great deal of Dummett’s presentation is pretty tedious stuff except for fanatics. This presumably accounts for the issue of the shorter and cheaper volume. If you simply want to learn how to play Tarot, then a selection of the more interesting variations is provided in Twelve Tarot Games in paperback. This consists of extracts from the larger work pasted up with connecting argument. But you must also, of course, find (or construct for yourself) appropriate packs of cards. And this may prove quite a problem. The 12 games described require packs of various sizes: 78 cards, 62 cards, 63 cards, 54 cards, 42 cards, 36 cards. But that is only the beginning.

Here, in summary, is a description of the Tarot pack known as Tarocco Bolognese, which you can, it seems, readily acquire if you first make your way to Bologna. The pack has 62 cards. There are four suits: Swords, Batons, Cups, Coins, each consisting of four court cards, Ace and 6 to 10. In Swords and Batons these rank: King (high), Queen, Cavalier, Jack, 10, 9, 8, 7, 6, Ace (low). In Cups and Coins the order is: King (high), Queen. Cavalier, Jack, Ace, 6, 7, 8, 9, 10 (low). There are 21 trumps which rank: Angel (high), World, Sun, Moon, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, four Moors (of equal rank), Begato. Finally, there is the Matto (Fool), which has a quite special role. By comparison chess begins to seem simplicity itself, while Contract Bridge is mere child’s play.

But all that is for the addicts. What about the fortune-telling, and the fables that Tarot originated in Ancient Egypt or Persia or India or China or wherever? Here Dummett is ruthless and devastating. His guiding principle has been total scepticism. Believe nothing unless there is clear positive evidence. If you find such evidence, look at it through a microscope. All, or very nearly all, conjectural history must be totally rejected. Working through the thickets of Tarot legend with this bulldozing machinery, Dummett reached the conclusion that virtually everything that was ‘generally known to be the case’ with regard to Tarot was either demonstrably false or so improbable, in the light of the evidence, as to be not worth serious consideration.

The question of just where card-playing, as such, may have originated depends to some extent upon our definition or terms. What constitutes a game of cards? What constitutes a pack? Dummett cites evidence which shows that a ‘regular’ pack, very similar to our modern four-suited pack of 52 cards, is likely to have been in use in the early part of the 14th century in Persia, and perhaps in other parts of the Islamic world. Card-playing in this form probably first reached Italy around 1375, by way of Mamluk Egypt and Venice. Dummett makes a further case, very elaborately argued, for supposing that the card game that was then introduced was a trick-taking game which shared a common ancestry with games played in India with a 96-card pack (ganjifa).

But this was not Tarot. On the contrary, it is quite clear that the Tarot pack, much as we now know it in England, was invented in Italy around 1430. It was an enlargement of the ‘regular’ pack which had already been in use for about seventy years. The innovation was evidently associated with the simultaneous introduction of a new trick-taking game for several players which required the use of 21 trump cards (triumphs). The triumphs were a sequence of picture cards not identified by a suit sign, like the ordinary suits, and not divided into numeral cards and court cards.

There was nothing mystical about the designs on these cards: but they needed to be easily identifiable, easily distinguished from each other, and easily named. The triumph cards, like the names of the suits and their designs, have varied considerably over the years, and the history of these matters, carefully spelled out in Dummett’s book, has varied from country to country. But fortune-telling is an extremely peripheral element. The earliest reference to the use of cards for fortune-telling in a European source is in an Italian work published in 1540, but the cards used came from a regular pack and not from a Tarot pack. The earliest recorded English publication of a pack of cards specifically intended for fortune-telling dates from 1690. Again it was not a Tarot pack. Cartomancy with regular playing-cards became regularly established around 1760, but the Taro pack was not used for this purpose until about twenty years later.

The contemporary belief, which is current throughout the Anglo-Saxon world, that the Tarot pack was specifically designed for use in occult practices or in divination is not only quite without foundation but also quite recent. Although it has some weak roots in French Masonic writings of the end of the 18th century, it is primarily the product of lunatic-fringe Cabalistic and Rosicrucian materials which were derived, during the latter part of the 19th century, from the bizarre publications of a Frenchman, Alphonse-Louis Constant, born in 1810 but known after 1853 as Eliphas Lévi (Zahed). Lévi wrote about alchemical magic in books which, in Dummett’s words, ‘range from the unintelligible, through the obscure, to the frankly puerile’.

The Tarot cards were incorporated into this rigmarole, and it was in this form that Tarot occultism reached England under the patronage of Kenneth Mackenzie, who, in 1866, had been one of the founders of a Masonic-cum-Rosicrucian Society known as the Societas Rosicruciana in Anglia. It was, however, another of Mackenzie’s occultist inventions, the Hermetic Order of the Golden Dawn, which was mainly responsible for the original British vogue for the Tarot mystique.

Dummett’s account of these goings-on is to be found around page 150 and gives cause to wonder just how mad Englishmen of the affluent classes of that period needed to be before they were put under restraint. This is, incidentally, the only point in the book where Dummett, the philosopher, and Dummett, the historian of Tarot, come face to face. He remarks of a certain Dr Wynn Westcott, who supplied ‘the entitlement to set up the Order of the Golden Dawn’, that ‘like the great philosopher Gottlieb Frege, whom he in no other way resembled’. Westcott ‘lived from 1848 to 1925’. Dummett’s growing renown as a professional philosopher is largely derived from his several studies of Frege.

But even this is relatively ancient history. The now widespread British belief that Tarot is an ancient form of cartomancy dates, in the main, from after the Second World War. This very recent efflorescence seems to be the result of some skilful operations in the publishing trade which Dummett understandably dismisses as unworthy of consideration.

I have only one comment to make on the history side of the argument. This is a point of etymology and I would be the first to agree that etymological evidence is, on its own, quite notoriously weak. For all I know, Dummett may have good reason for rejecting it. As emphasised by Dummett in many places, the class of card games to which Tarot belongs is that of ‘trick-taking games’. Dummett’s definition of such a trick-taking game fills half a page (page 167), and I will not attempt to condense it. Whist and Bridge are both trick-taking games. My dictionary offers no explanation as to why the word ‘trick’ should be used in this sense. The Continental equivalents are quite different: German stich, French levée. So is ‘trick’ simply a version of ‘taroc’? Unimportant, but, if valid, worthy of inclusion in the record.

On the anthropological side my views as an anthropologist are, perhaps naturally, rather more mixed. I miss any account of participant observation. Dummett started out with a contemporary form of Tarot as played in France and worked backwards, so it is not surprising that he ended up by making a survey of European card games resembling modern French Tarot in one or more fairly obvious ways. But it is not self-evident that either card games, as such, or European card games, as such, make a particularly interesting category. The point I am making here is implicit in Dummett’s remarks at page 63, where he notes with some surprise (and perhaps irritation) that although trick-taking is not a characteristic of card games played in China, it does feature as a characteristic of T’ien chiu, played with dominoes. But how do we define dominoes?

I have no idea as to what kinds of game (card or other) are permitted in contemporary Communist China, but in the backwoods rural areas of that country which I visited at various dates between 1933 and 1942 it was normal to find gambling marquees set up in market centres or travelling fairgrounds. As in a European casino, the visitor was offered the choice of a great variety of ways in which to lose his money. Some of these games were played with playing-cards, some with bamboo chips, some with ‘dominoes’, some with cowrie shells, and so on. Some games were much more popular than others, and in some games the stakes were much higher than in others, but the rules of the different games were often very similar.

What I am getting at is that I do not feel convinced that the two hundred-odd games which Dummett describes in detail really belong to a closed set. If he had considered European games which are not played with cards at all, and also early Italian games which were not played with cards at all, he might have found the linkages with 14th-century Ferrara working out to a different pattern. Or, to put it differently, despite the vast learning that Dummett displays, he never gives the reader a feeling of what the game might mean to the players. What is the social context in which the action takes place?

The Tarot packs which have survived from early times are mostly fragmentary and most of the fragments come from upper-class households. For this and other reasons, we can learn very little about how the game might then have been played simply from a study of the cards. But much that is relevant might have been learned from an observation of modern Tarot players. Dummett gives us the formal rules of how a great variety of modern derivatives of early Tarot ought to be played, but he does not tell us anything about how these games are actually played. He does not describe the atmosphere of play.

Perhaps I am asking too much, but since Dummett’s project was encyclopedic, he might well have paid closer attention to the relationship between the games he was considering and gambling. Gambling of one sort or another is found in all forms of human society. Moreover, wherever there is gambling you will also find magical procedures for controlling the laws of chance and for making predictions about particular gambling strategies. And that brings us back to fortune-telling.

Dummett could not go back to 14th-century Ferrara to watch the action, but Tarot, he tells us, is still played in modern Bologna. And if he had given us more detail of the gambling environment of that context and of how the Tarot triumphs enter into it, I would have felt that I had a better understanding of why such an exceedingly complicated game should have become so popular, and why, even now, the triumph cards should carry the peculiàr designs that they do.

But that is a carping comment. Other hobbyists may envy Professor Dummett for having a publisher willing to invest capital in such an enterprise, and maybe this is not the book for which he would wish to be remembered in future years. But in its own very special way it is a major achievement of which its author can be proud.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.


Vol. 2 No. 18 · 18 September 1980

SIR: It is improbable that, as Michael Dummett appears to say in his books on Tarot (LRB, 4 September), the Tarot pack was used merely for games. Our familiar pack was symbolical to start with; nobody in the Renaissance would invent such a random thing without making it symbolical, or claiming to. And the picture cards of the Tarot are rather aggressively mysterious.

William Empson
London NW3

Vol. 2 No. 20 · 16 October 1980

SIR: The trouble with Michael Dummett’s book on the Tarot, to which William Empson referred in a recent letter (Letters, 18 September), is that while it leaves us dispossessed of some of our favourite myths it does not go far enough in explaining (or attempting to explain) the origin of these cards.

Professor Dummett flails the 18th century and later occultists mercilessly for their duplicity in propounding an antique. Egyptian origin for the Tarot and for their belief that the Tarot trumps embodied the secrets of an ancient wisdom. My own view is that the delights which they have produced (and who dare say these are not creations?) are a sufficient reward and recompense for their naughtinesses. The occultist’s viewpoint is well-comprehended and expressed by Kathleen Raine in Yeats, the Tarot and The Golden Dawn. She sees the occultist movements in terms of an imaginative creativity, and this is arguably at least as important as the social intercourse, intellectual exercise or sheer escapism of card-playing.

That card-playing was of pre-eminent importance in relation to the history of the Tarot Michael Dummett leaves us in no doubt. Indeed, it is his contention that the Tarot was invented as a new card game embodying a hitherto unknown feature, the principle of trumps, and he stresses that it was not until the late 18th century (when occult revelation pronounced the cards to be of near-diluvian antiquity) that the Tarot was used for any purpose other than card-playing. Now, the invention of trumps must rank as the most significant development in the history of card-playing since the introduction of playing-cards to the West in the mid-14th century. It is not strange that someone should have conceived of the use of master cards in a trick-taking game, but what I do find strange is the form in which the invention is embodied and the fact that the inventor is not known. On Dummett’s theory the inventor of this new game produced a whole new pack of cards incorporating four suits of 14 (corresponding to the existing playing-cards but with an additional court card in each suit) and a set of 22 picture cards, the trumps, trionfi or atouts, known to occultists as the Major Arcana. The pictures, says Dummett, are standard Renaissance subjects: The Emperor, The Pope, The Last Judgment, Temperance, Fortune etc. (It would burden the argument here to mention the inclusion of such non-standard subjects as The Hanged Man and The Tower.)

His theory is based, not only on surviving cards from the 15th century, and on written testimony in the sermons (dated 1450-1480) of a Dominican friar, but on the lack of any comparable cards outside Italy from the same or an earlier period. But lack of evidence cannot be used to support a theory, and all Michael Dummett has really proved is that the earliest known location for the game of Tarot was Italy (1442, possibly even a little earlier). Thus, while the theory of an Italian invention is in accord with the known facts, it cannot be regarded as proven and may well be wrong. I think it is, primarily because the subjects and hierarchy of the trionfi appear to have been standardised from the beginning. The list of trumps which appears in the Dominican’s sermon would serve as an adequate inventory for a standard pack of the present day; apart from occasional (often politic) variations, the subjects and order of the Tarot trumps have remained unchanged since their introduction. This argues a high stage of development of the Tarot pack when it occurred in Italy, and it is just this condition in respect of playing-cards which Dummett uses to adduce that these were not a Western European invention but a foreign introduction.

But if the Tarot is not an Italian invention, where did it come from? I would suggest a sorely beleaguered Byzantium, which was to afford many a refugee and many a manuscript in the decade or so following Dummett’s 1442. Survivals of an Eastern Imperial tradition are said to persist on some packs in the details of the papal vestments and regalia and in the head-gear of the Emperor.

A more telling anomaly (for Michael Dummett’s thesis) is the number of the trumps: 22 (or better, 21 + 1, since the 22nd trump, The Fool, is usually unnumbered and has a specific role in play). Dummett hedges the question of why such a large number of trumps should be necessary. I have looked carefully into the play of the games represented in his book and can find no cogent reason why there should be 22 trumps. It has been suggested that this number was chosen so that the ratio of trumps to any suit should be 3:2; other investigators have suggested that the number 78 was important; but considered dispassionately, neither of these reasons makes sense. Playing with Tarot cards is very tiresome because of the cumbrously large hands involved (25 cards in the three-handed games). If we consider the difficulty of playing with a pack which had no indices and in which the trumps were not numbered, we must revere all the more the mnemonic powers of the peoples of the Renaissance.

I think there are 22 Tarot trumps because the subjects already existed as a set of 22, that this set was adopted by a card-player used to a 56-card pack and that the idea of trumps was born from the cards rather than that the cards were an implementation subsequent to an invention of the principle of trumps. Michael Dummett is adamant that the trumps are not to be separated from the rest of the pack – the Tarot pack is a whole, he says, and was invented as a whole. Once again he is rescued by lack of evidence. I like to believe in the pre-Tarot existence of a set of 22 designs because, unlike Michael Dummett, I have a weakness for occultists. It is axiomatic for them that each card of the Tarot is associated with one of the 22 letters of the Hebrew alphabet and hence with one of the 22 paths joining the ten Sephiroth of the Tree of Life. The cards may thus be used in meditative cabalism as mandalas for traversing the paths of the Tree.

The basis for the present cabalist tradition is undoubtedly Eliphas Lévi (late 19th-century) and since his time the designs of the cards have incorporated the shapes of some Hebrew letters; others have been interpreted as being present where they obviously are not. But quite apart from this late adoption of the Tarot into occultism, I consider that two of the trumps (The Tower, trump XVI, and The Moon, trump XVIII) have, from at least the 16th century, and throughout changes in design, persistently shown an iconographic association with the Hebrew letters ayin and tzaddi respectively (the 16th and 18th letters of the Hebrew alphabet).

This may, of course, he coincidence allied to wishful thinking. Even if there were a proven correspondence between the Tarot trumps and the letters of the Hebrew alphabet I should hesitate to suggest that this indicated that the cards were ever used for cabalistic purposes, but a set of playing-cards that could afford secret communication between cognoscenti in public must have had their use. If this letter/card association is well-founded, it is strongly supportive of the notion of a fixed set.

John Henry Jones
London NW3

Vol. 2 No. 21 · 6 November 1980

SIR: Sir William Empson considers it improbable that ‘the tarot pack was used merely for games’ (Letters, 18 September). I think that the clue to his opinion lies in his use of the word ‘merely’. Intellectuals, scholars and other serious-minded people are prone to consider playing games a trivial occupation on which no one would expend any genuine effort; for some reason that entirely escapes me, they do not consider pretentious rigmaroles purporting to reveal the future to be equally trifling. Hence, when they contemplate an artefact as beautiful and intricate as the Tarot pack, they cannot bring themselves to believe it to have been invented for play: it must have been intended for some serious purpose such as predicting future events. I think their estimate wrong, both in itself and historically. Attempting by non-rational means to divine what is to happen is one of the most absurd of human activities; devising and playing games, on the other hand, is a manifestation of ingenuity and of delight in order, an art form as worthy of respect as that of the dance. The game of Tarot was born in the courts of early Renaissance Italy; and in those courts men and women did not despise games as trifling, but cultivated them and took them seriously, by which I do not mean portentously.

These are a-priori arguments, intended only to rebut Sir William’s a-priori argument. For my assertion that no one associated Tarot cards with the occult or with fortune-telling until the 1780s, I relied on empirical facts, which I marshalled. I cited a string of references, from the 15th and 16th centuries, to the use of Tarot cards for play and pointed to the utter absence, from before 1781, of any mention of their use for any other purpose or of their bearing any esoteric meaning. I mentioned two texts, attacking or ridiculing the game, whose authors could hardly have failed to mention their occult associations, if they had been thought to have any. I observed that Court de Gebelin, who initiated the whole occultist fantasy concerning the Tarot pack, believed himself to be the first since ancient times to recognise in it more than an instrument of play, and that Eliphas Lévi and his followers, who were anxious to attribute to their occultist predecessors their own beliefs about it, failed miserably, as A.E. Waite also remarked, to cite one plausible reference to it from their works. If my conclusions are to be rejected, some explanation is needed why it took 350 years for the occult significance of the Tarot pack to come to light; it is surely not enough just to insist that something that makes so mysterious a first impression could not have been originally used ‘merely for games’.

Sir William says that ‘nobody in the Renaissance would invent such a random thing without making it symbolical’. Quite so: Renaissance people delighted in symbolism, and used it in all possible contexts, so that its presence needs no special purpose to explain it. There is obviously symbolism in the subjects of the Tarot trumps, but its presence does not imply that the cards were used for anything but play. Most of the symbols were familiar ones: the Wheel of Fortune, for instance, is to be seen on churches in France, England, Sweden and, no doubt, many other countries.

There may even be symbolism there that is not apparent to us. But if there was any hidden symbolism in the pack, the secret was quite soon lost, and this is in itself enough to show that the use to which the cards were put was one to which the symbolism was inessential. If Sir William will look at Alberto Lollio’s mock-serious verse diatribe against the game, published in 1550 and discussed and quoted in Chapter 21 of The Game of Tarot, he will see that the poet, who confesses to having earlier been a devotee of the game, had no conception of any symbolic significance for the trump cards, the meaninglessness of which he derides. In Florence, they removed the Popess, turned the Pope into some kind of temporary ruler and stuck in an extra 20 subjects, to bring the number of trumps to 40; and later changes of subject elsewhere likewise suggest a complete indifference to any supposed symbolism, and, instead, a desire to have easily recognisable and namable figures. A duchess in Sicily objected to the presence of the Devil, and had him replaced by a Ship. In Normandy and in what is now Belgium they replaced the Pope by Bacchus and the Popess by Captain Fracasse from the Commedia dell’Arte.

Whoever first devised the pack may or may not have had symbolic intentions that go beyond what is obvious to us: but, if so, they had, at least by the 16th century, and probably earlier, slipped from the consciousness of those who used the cards, as much as had the military significance of the chess pieces for the Indian inventors of chess from that of chess-players in Europe and the Islamic world. We are disposed to believe otherwise because we have been exposed to the propaganda of the occultists and because we do not live, as did people of the Renaissance, in a world in which the figures on the Tarot trumps are of familiar occurrence in other contexts. But the history of the Tarot pack to be found in the works of occultist writers is pseudo history, bearing no relation to the actual facts; and when the historical facts are looked at, they tell an altogether different story.

Mr J.H. Jones has done me the courtesy to respond (Letters, 16 October), from a standpoint more favourable to the occultists than mine, to the arguments concerning the origin of the Tarot pack which I set out in my book. As I understand him, he does not maintain that the Tarot cards were known, in Western Europe, anywhere outside Italy before the 16th century (or 1480 at the earliest), but is conjecturing a Byzantine origin for the trump cards. Likewise, I do not take him to be suggesting that the composite pack (trumps plus suit cards) was ever used, before the 18th century, for any purpose other than play, or was generally interpreted as having some occult meaning. Rather, I understand him as endorsing the view, rejected by me but propounded by Robert Steele and Gertrude Moakley, that the set of 22 trumps originally existed as an independent set before being joined to the suit cards to form the composite pack, and as coupling this with the occultist idea that the cards correspond to the letters of the Hebrew alphabet.

If he were right on both points, there would remain much in common between us, supposing that I have interpreted him correctly. It would still hold good that there has been no continuous tradition of occultist interpretation of the cards, and that the composite pack, from its first formation, was used only for play, until the late 18th century. But I venture to say that Mr Jones’s theories are more speculative than any suggestion of mine. There is a question here of burden of proof. Mr Jones says that I am ‘rescued by lack of evidence’ in holding that the trumps were, from the outset, part of a composite pack. My point was that there is virtually no evidence for the opposing theory: it so happens that every one of the many incomplete hand-painted 15th-century sets that survive contains at least one suit card. Since we have imperfect knowledge, the idea that the trumps originally formed an independent set cannot be ruled out as impossible: but, in the absence of any evidence for it, it seems to me unreasonable to believe it.

The same goes for the Italian origin of the cards. We have many cards, and many references to them, from 15th-century Italy; we have many references, and one set of cards, from 16th-century France, as well as ones from Italy in that century: the natural conclusion is that they originated in Italy and spread to France about the turn of the century. I gave specific reasons for supposing that this happened during the French occupation of Milan. I never considered Byzantium, where, so far as I am aware, playing-cards of any kind were unknown. Mr Jones’s conjecture is worth investigating, but, without evidence, it is mere conjecture.

Mr Jones is right to say that the set of subjects for the trumps was standardised at a very early date, but wrong to say the same about their order. The order used everywhere outside Italy (with small variations) was probably derived from Milanese practice, but in Ferrara, a different order was employed, and, in Bologna and Florence, yet a third. This puts in doubt any firm association (for which, again, there is no evidence from pre-18th-century sources) between them and the Hebrew letters. Of course, it does not show that there was not originally such an association, which was never generally known or was quickly forgotten: but, to establish what it was, or to make the theory plausible, one has first to decide which of the different orders was the original one. The fact is, as it seems to me, that, if there were ever such an association, it is very hard to discern. The occultists, who have all believed in such an association, have been quite unable to agree what it is – that is, which card should be taken to correspond with which letter. Mr Jones favours associations between trumps XVI and XVIII which agree with Eliphas Lévi’s system; but, as he well knows, there are rival systems. The Tower and the Moon have not always been numbered XVI and XVIII respectively; if Mr Jones will look at the tables on pp. 399 and 400 of my book, he will see that they have often been numbered XV and XVII. I am not sure why he is so convinced that they correspond to the two Hebrew letters he associates with them. The iconography of these cards, especially of the Tower (whose original meaning is probably the House of the Devil), has varied enormously, as can be seen from plates 6, 13, 18, 26, 27, 30 and 31 in my book: thus, in the Sicilian pack, there is no lightning on the Tower, while, in the Belgian one and some French predecessors, there is lightning but no tower on the corresponding card.

In any case, it is not certain that 22 is the original number of trump cards (counting the Fool as a trump): the earliest surviving hand-painted pack may have had more (it had six instead of four court cards per suit). Mr Jones wants to press the question: why 22? He asks more generally: ‘Why so large a number of trumps?’ In many games there are more trumps than cards per plain suit: in Skat, for example, there are 11 trumps and seven cards in each of the three plain suits. In Tarot, where there are four plain suits and it is compulsory to play a trump when unable to follow suit, at least a similar proportion is essential. The tendency has, in fact, always been to increase the proportion of trumps to suit cards, by reducing the number of suit cards to 41 in Sicily, to 40 in Bologna, and to 32 or even 20 in Central Europe, or by increasing the number of trumps from 2l + 1 to 40 + 1 in Florence. So the only remaining question is: why specifically 22? I do not know how much of an answer the question merits: in devising the pack, its inventor had to pick on some number, and the precise ratio of 3:2 between trumps proper and cards in each plain suit, to which I called attention, seems to me sufficient explanation for his choice. I do not see that the thought, ‘It can hardly be a coincidence that 22 is the number of letters in the Hebrew alphabet,’ is much more impressive in this case than in that of the chapters of the Book of Revelation.

I do not wish to be dogmatic. It is conceivable that whoever first fixed the number of Tarot trumps at 22 had the Cabala in mind: but if so, his idea was very rapidly forgotten, and he failed to leave unmistakable clues to it. I argued, in my book, that, in the absence of the generic idea of playing-cards, the 52-card pack has too complex a structure to have been invented de novo without any evolutionary forebears: but my thesis that playing-cards were introduced into Europe from the Islamic world had also positive evidence as well as this a-priori argument to support it. I do not think there is a similar a-priori argument concerning Tarot cards. Certainly playing-card packs consisting solely of picture cards were known in 15th-century Italy. It does not strike me as improbable that, given both these and the regular four-suited pack, someone should have had the idea of adding picture cards, to serve as trumps, to the regular pack; there is no need to suppose either that that specific set of picture cards had already existed as an independent unit, or that the idea was imported from Byzantium or elsewhere. Given such an invention, in which the trumps did not form a suit in the ordinary sense, there would be nothing especially natural in having the same number of them as of cards in an ordinary suit: what special explanation do we then need for there being 22 of them?

It is not surprising that we do not know the name of the inventor of the game of Tarot and of the pack with which it is played: we know the names of the inventors of exceedingly few games (not even of that great benefactor of mankind, the inventor of chess).

Michael Dummett
New College, Oxford

send letters to

The Editor
London Review of Books
28 Little Russell Street
London, WC1A 2HN

Please include name, address and a telephone number

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences