Hi Huck, et al.,
In the light of this discussion about the number of trumps, here's an unfinished paper discussing the three readings of the "70 cards" 1457 reference, starting with autobis' theory and moving on to Hurst and my own.
(the part that is most unfinished is that dealing with Michael Hurst, but I think that the captions are clear enough to give the idea of what I meant).
"Big Triumph Cards, 70 per pack"
The earliest mention of a pack of Triumph cards (February 10, 1442) says it is composed of 4 suits and figures, without describing which figures and how many. But the earliest mention of the _number_ of cards in a Triumph pack (July 21, 1457) immediately poses a problem of interpretation.
The document is an account of payment to the d'Este court artist specializing in Triumph cards, Gerardo de Andrea da Vincenza, for two packs of "big triumph cards, which have 70 cards per pack." (para due de carte grande da trionfi, che sono carte 70 per zogo)
First noted by Campori in 1874, then confirmed and republished by Franceschini in 1996, this reference in the d'Este accounts was first considered troubling by Gherardo Ortalli in 1996 (neither Campori nor Dummett commented on the number when mentioning it previously). Although he noted the problem, Ortalli proposed no solution.
Four solutions have since been offered:
1. The trumps are fewer in number;
2. It is a reduced pack;
3. It is a scribal error.
4. It is another kind of "Triumph cards".
1. In 2003, autorbis suggested that the reference refers to a Triumph card pack with 14 trump cards, which combined with a pack of 4 suits of 14 cards would make 70.
2. A critic of autobis, Michael J. Hurst, proposed another solution in 2003, that it refers to a reduced pack, in which two pip cards of each suit are not present.
3 & 4. I myself suggested the third and fourth interpretations.
Follows the evidence adduced by the authors of the solutions to the "70 cards" problem, in the order given above (Autorbis, Hurst, Caldwell A, Caldwell B).
Autorbis' primary evidence consists of two things:
A. The 14 Trumps painted by Bonifacio Bembo remaining in the Visconti-Sforza or Pierpont-Morgan-Bergamo (PMB - named for the museums among which the full set of cards is currently divided) pack;
B. The Ferrara 1457 mention of 70 cards.
Additionally he refers to:
C. 14 pictures painted on card-stock in 1441;
D. A German playing card pack from the late 15th century containing 5 suits of 14 cards.
Discussion of the evidence.
A. The PMB is considered to have been painted by Bonifacio Bembo for a member of the Sforza household between 1450 and 1453. The surviving pack contains 68 cards painted by Bembo, and 6 Trumps painted approximately two decades later by another artist.
The question is which suggests itself is: were the latter six cards newly introduced to the pack, or did they replace lost or damaged cards originally painted by Bembo?
To answer this question, appeal must be made to the earlier packs painted for the Visconti, at least one of which is by Bembo, and to theories of the original design and development of the standard model of 22 trumps.
The Visconti di Modrone or Cary-Yale (CY) pack, also painted by Bonifacio Bembo, contains 11 trumps. Of these trumps, 6 are common to the PMB and CY, 2 are present in the 22 standard trump sequence, and 3 are not present in the standard sequence but do appear in the Minchiate pack of Florence, attested a century later. Additionally, this pack contains 6 court cards per suit, a unique feature. These qualities have given rise to various interpretations.
Several early commentators believed that the CY was an early Minchiate pack, and that the standard model was formed from it. However, because Minchiate packs are not known in the 15th century, and because Minchiate also does not possess 6 court cards, as far as I know this theory has no adherents today.
In 1974 Ronald Decker proposed that the CY lacked a Fool, and had only 14 trumps. In exchanges with Michael Dummett in the _Journal of the Playing Card Society_ in 1974-1975, this theory was debated, and it seems that Decker no longer believes it. Dummett's account of the issues and summary of the exchange, in 1980, deserves to repeated in full, because it also contains a theory of the evolution of the number and subjects of the trumps:
"The Visconti di Modrone pack is the only Tarot pack, of any kind, in which the suits include court cards other than the usual King, Queen, Cavalier and either Jack or Maid. There must have been sixty-four suit cards originally, and whether a Fool was included, it is impossible to say. Ronald Decker has suggested that there may originally have been only fourteen triumphs, and no Fool, so as to make up the usual total of 78 cards (footnote, cited after); but the total number of cards in the pack is unlikely to have been seen as a significant feature. Since four of the stock set of seven Virtues were included among the triumphs, it seems probable that the other three were also: Temperance and Justice, which belong to the standard list of triumph subjects, and Prudence, which does not. It is just possible, on the other hand, that what was held constant was the ratio between the number of triumphs and the number of cards in each suit, which, in the 78-card Tarot pack, is 3:2; if this was also so in the Visconti di Modrone pack, it would have had twenty-four triumph cards, in which case it could have contained all save one of the usual subjects, making, if the Fool was included, a pack of 89 cards, altogether; indeed, if we do not suppose that it included Prudence, it could have had all of the usual subjects.
(footnote discussion of Decker-Dummett debate: ) Mr Decker presumes that the Visconti di Modrone pack had only 78 cards, like other Tarot packs; since it must have had 64 suit cards, that leaves only 14 triumph cards and no Fool. There can, on this reasoning, have been no Fool, since Mr Decker accepts my view that the three missing Virtues must originally have been present, and, if we add these to the eleven surviving triumphs, we already obtain 14, and there is no room for the Fool. Mr Decker then takes the illogical step of arguing that, since there are only 13 (surviving) triumph cards in the Visconti-Sforza pack that were painted by Bembo, perhaps these, together with the Fool, were all that the pack originally contained. This is illogical because in this pack there are only the usual 56 suit cards, so that he is suggesting an original pack of only 70 cards, whereas the original premiss (Decker's) was that all Tarot packs had 78 cards. He attempts to rescue his hypothesis by conjecturing that the Visconti-Sforza pack had originally six court cards in each suit; but this is obviously very special pleading. On his hypothesis, there would, besides the suit cards, have been been seven cards in common between the two packs: the Empress, the Emperor, Love, Justice, the Chariot, Death and the Judgment. Seven of the triumphs present in the Visconti di Modrone pack would then have been removed, namely the World and the six Virtues other than Justices, when the Visconti-Sforza pack was painted, to make room for the Fool, the Bagatto, the Popess, the Pope, the Wheel of Fortune, the Hermit (which originally represented Time), and the Hanged Man. Later, when the number of triumphs was increased by eight, this was done by restoring from the original set of subjects, the World and two of the Virtues, Temperance and Fortitude, but not the other four, and adding the Devil, the Tower, the Star, the Moon and the Sun. All this makes so little sense, and is so grossly implausible, that the hypothesis that demands it is not to be entertained. What is impressive about the fifteenth-century Tarot packs that have come down to us is not the variation in subjects, but, on the contrary, their invariance, given the fact that no pack has survived complete. Certainly we must allow that, after the Visconti di Modrone pack was made, four of the seven Virtues were removed: the advantage of the hypothesis that that pack contained twenty-four triumph cards (not including the Fool as a triumph) is that it gives a reason for the removal of at least three of them when the number was reduced to twenty-one."
(Game of Tarot, pp. 77-78, and note 34).
Autorbis' theory bears little relation to Decker's, so that the latter cannot really be considered a precursor. Autorbis' theory posits that Triumph packs were made to order with whatever subjects were demanded by those who commissioned them, so that notions such as "standard subjects" and "standard number" or ratio of trumps to suit cards do not apply until some time after 1457.
(An interesting point from Dummett's account is the final sentence, and the implications of the whole passage: that the CY represents a previous standard of sorts, and bears some kind of direct relationship with the PMB, so that by removing the three theological Virtues and rejecting Prudence from ever having been part of pack, the "reduced" standard pack results. Instead of taking the CY as an unusual experiment, inspired *by* the standard model, it is regarded as a former standard model, before the emergence of the standard trump series. This gives some context to statements throughout his 1980 work, and the assertion in 1996, that the "number and subjects of the trump series were fixed by about 1450", which is the earliest date for the PMB. That is, why would would anything have to be removed from the CY, if the standard trump sequence already existed, and the PMB faithfully represents it? I don't know if Dummett still holds that the Visconti court was the place where the tarot was invented and was its "testing ground", so to speak, until the PMB was painted. I tend to think not.)
Thus most authorities assume some kind of evolution of the pack before a certain date. The question is which date, and what forms did the pack have before then? In their 1996 work "A Wicked Pack of Cards", Decker, Depaulis and Dummett argued for 1450. This date appears to have been chosen because of the discussion of the relationship of the CY to the PMB already quoted, as well as then-newly discovered references to Triumph cards which showed that they were known generically in Milan (Marcello's letter (1449)) and Sforza's letters (1450) and that the game was known in Florence (1450). The implication is that a standard kind of Triumph game, including a standard kind of Triumph cards, was known in diverse places such as Ferrara, Milan, and Florence, by 1450. If this standard pack changed its form in Milan, it is difficult to believe it would also change form in Ferrara and Florence to match. In other words, the game was *popular* by 1450, which precludes luxury models influencing the standard popular type of pack. Therefore, the standard pack of 21 sequential trumps and a Fool existed by 1450 at the latest. Moreover, it appears to have been known in Siena in 1452, and is referred to in Bologna in 1459. Thus at least five cities knew the card game of Triumphs during the 1450s. It seems to me highly implausible that a standard model of 14 trumps in circa 1452 Milan would also be the standard model in all of those places, assuming that a simultaneous change happened to all of them and more at some later date, when the game was already established.
In addition, the consensus is changing on the dating and provenance of the so-called Charles VI pack, and its sister pack in Catania. Comparisons of the designs with the Rosenwald sheets, from Florence (or nearby in Tuscany), as well as the numbering later placed on them, indicate a Florentine provenance (Done by Caldwell and Depaulis, 2005). Independently, Cristina Fiorini concluded from an art historical perspective that the cards are Florentine, and date to around 1450. Taken together, this information changes radically our notions of the origin of these cards, and also indirectly affects Teikemeier's 5x14 hypothesis. This is because the Charles VI cards contain the standard three Virtues, the Fuoco (Tower or House of God), Moon and Sun cards, as well as *both* of the Angel (or Judgment) and World cards. Even if their numbering is late enough to accomodate autorbis' theory comfortably, their composition at this early date is not.
So, what does the foregoing have to say about autorbis' use of the 1457 Ferrara reference in support of the 5x14 theory? In my judgment, it makes the reference less explicable as a reference to a 14 trump pack, and makes its interpretation more difficult.
B. In the light of the weight of evidence for a widespread standard game by 1450, it seems that the best interpretation of the PMB pack's 14 trumps from 1450-53 is that they are fragmentary of a once-complete standard pack. Otherwise, that the pack is an unusual experiment. But it does not seem that it can be used as evidence for a widespread standard 14-trump tarot pack.
C. The 14 images on card stock painted by Jacomo Sagramoro for Bianca Maria Visconti, for which he was paid 2 lire on January 1st 1441, remain obscure. They were made to have fun or play with, it is true. They pre-exist the earliest triumph card reference, in the same records, also true. They also concern key players in early tarot history, both in places where tarot might have been invented, or even by whom it was not inconceivably invented.
But autorbis' theory suggests that they were the prototypes of the PMB images. This goes very far beyond the evidence, and demands improbably narrow conditions for tarot's early development and distribution.
In the first place, if CY were a gift to Francesco Sforza and Bianca Maria on their wedding of October, 1441 (a widespread opinion), the CY model must have been conceived sometime before that, and it bears little resemblance to the PMB.
Secondly, if Sagramoro and Bianca Maria worked together to produce the first tarot pack, with 70 cards, but was already sufficiently altered to produce the CY within a few months, how likely is it that the original structure or model would survive until the 1450s as it is represented in the PMB? And how could it become mass-produced enough to make it into common parlance by 1449, and be as widespread as to be in Milan and Florence by 1450?
That is, why and how was it relatively mass-produced? Sforza and Bianca Maria lived a fairly unsettled life between the time of their marriage and late 1447. There would be little time to develop and commission dozens of decks and send them places, let alone to have them take root and develop a following large enough to gain the notice of lawmakers.
Note also that Marcello in 1449 calls Filippo Maria's invention a "new kind of Triumphs", implying that it was apparent to him that the others were an old kind (that Filippo Maria's pack was probably not new is beside the point). That is, that before 1447, he thought Filippo Maria had already designed a "new kind" of triumphs. Since he is a good witness and no doubt inquired a little into the game (he sought out the artisans of triumph cards), this narrows our window of time for Bianca Maria to propagate her new invention, to between 1441 and 1447. Did she really have the resources or opportunity for this?
To me, the 1441 notice remains as obscure as ever, and despite the circumstances and numbering, is not a proof of the 5x14 theory.
D. Master PW's Cologne pack has five suits with 14 cards. It illustrates that someone could conceive of a pack of cards with 5 suits and 14 cards in each suit. But the suits have an identical structure, with 10 pips and 4 court cards, and it is not apparent that any suit is a permanent trump suit. It doesn't resemble a pack of Triumph cards, in which the permanent trump suit contained an allegorical series of cards.
Therefore, this pack doesn't seem to support the idea that the original structure of the Triumph pack was 5x14. It appears to be a coincidental number on an unrelated pack, and raises more questions than it answers.
These considerations are hardly definitive, and are merely my opinion.
Autorbis' theory is widely discussed on internet groups, and appears to be gaining adherents. It appeared in print, unattributed, in Giordano Berti's "Storia di tarocchi" (Mondovi, 2007). However, it has not been critically assessed by recognized authorities on early tarots such as Michael Dummett or Thierry Depaulis.
For support of a shortened pack Hurst adduces the later attested practices of either removing pip cards from a full pack or manufacturing shortened (or augmented) packs for various kinds of Tarot games, i.e.
97-card Minchiate (Florence, early 16th century)
66-card Rules of Tarot (France, 1637/1585?)
64-card Sicilian Tarot (Sicily, 17th century)
62-card Tarocchino (Bologna, 17th century)
54-card Italian-suited German decks (mid 18th century)
54-card French-suited modern decks (mid 18th century)
42-card Hungarian decks (late 18th century)
(from Hurst's webpage "Collected Fragments of Tarot History," page 2, "1440-1479: Decembrio to Boiardo" at
At the head of this list, Hurst puts "70-card _carte grande da trionfi_ (Ferrara, 1457)", to indicate conformity with a pattern.
In my opinion, this presentation is potentially misleading.
1. Distance in time.
2. Lack of attestation of any shortened packs in the 15th century.
3. My principal and final objection is the improbability of making two very luxurious, custom designed packs, with two pips from each suit missing. If there were a popular Triumph game which demanded a reduced pack, wouldn't it be easier to simply lay aside these two cards from each suit, as demanded in some later games?
Also, if the original game were desired, the whole pack would still be there; and if a regular game without trumps were desired, that would be possible as well. One of the advantages of a Triumph pack was that a regular card pack was included.
Since both Campori and Franceschini independently transcribed the number in the manuscript as the Arabic number "70", it is unlikely that they both misread it; any error would therefore have to be in the autograph.
This is admittedly the weakest hypothesis a priori; since the autograph is not published, it is impossible to discuss it further on the basis of direct evidence.
However, scribes often make mistakes, and with a hapax legomenon like this, it is not to be ruled out. (refer to habit of rounding off numbers - Berni says that the tarocchi player has "hundreds" of cards; Orioli says that Bolognese packs have sixty cards, etc.)
This possibility is, that the "big Triumph cards" refers to another kind of Triumph cards than the presumed standard model, perhaps with different subjects and a different number of subjects. This might be the case if it were an isolated experiment, like the Visconti di Modrone (Cary-Yale Visconti) seems to be.
I consider this unlikely because Triumph cards appears to be a generic type of product, even if modifications were sometimes made in the subjects and number of the trumps, or the size of the cards. The game would have to be intelligible, and to be so it has to match or only vary slightly from some kind of standard.
Two kinds of variations in the number of the trumps are known, an augmentation and a diminution. The Minchiate (or Germini) of Florence, augmented the complete trump series with 20 cards. The Lucca variation (Dummett and McLeod 2004, pp. 353-354), in the 17th century used only 12 trumps: those numbered VIIII (Wheel) to XV (Tower) and the five arie, unnumbered, and the Matto. Thus 69 cards.
Both of these packs are clearly modifications of the standard series, not prototypes of it, so they cannot be used as evidence that a 14-trump pack once existed.