V002 5 March 2000, 5rth Revision 27 May 2001

The Phaistos Disk: A Statistical Decipherment

J. Faucounau
Member of the Linguistic Society of Paris, France

The Phaistos Disk
A puzzling discovery

On July 3d, 1908, the Italian archaeologist Luigi Pernier found at Phaistos, an ancient city in the south of Crete, Greece, that he was excavating, a strange disk of baked clay. Its diameter was about 16cm and its thickness a bit more than 2cm. Both its faces were covered with hieroglyphs of an unknown kind, running in a spiral.

Since its discovery, no archaeological item has been a bigger source of controversy than the Phaistos Disk. Its origin has been supposed to be Crete, Anatolia, Cyprus, Rhodos, Egypt, Africa or even China. It has been read either from the center outwards or from the edge to the center. The "commas" attached to some hieroglyphs have been considered as a punctuation, as "viramas", or as verse-marks. Various decipherment results have been proposed, based on languages as different as Greek, Hittite, Basque, Ancient Egyptian, Sumerian, Latin, Germanic, various Semitic languages, etc. It has even been supposed by some authors, in desperation, that it could be a musical partition, an astronomical calendar, or perhaps a fake !

These multifarious and contradictory opinions have led to a regrettable consequence: the neglecting of the Disk by "serious scholars". The Disk was declared to be indecipherable and therefore useless.

A decipherment through Statistics

In 1975, a decipherment was succinctly published in a French scientific journal by the present writer, who, not being on the faculty of any university, had been free to solve the enigma without being considered as a fool by his colleagues. This publication went almost unnoticed because linguists did not believe, at that time, that the language found, proto-Ionic, was possible (cf the so-called "Risch-Chadwick" theory). The decipherment was, nevertheless, quite different from others, because it was not starting from some postulate concerning the origin of the Disk or the language used or the type of script, but was just based upon a 7-year statistical work.

The statistical, probabilistic method used could not warrant the correctness of the decipherment which, moreover, was quite surprising: a proto-Ionic GREEK LANGUAGE and a SYLLABARY of an ARCHAIC TYPE ! A verification was therefore needed. This last step lasted some 20 years, about three times the duration of the decipherment itself.

A book was published in 1999 (Le déchifferement du disque de Phaistos. Preuves et conséquences by Jean Faucounau, edit. L'HARMATTAN, 5 rue de l'École Polytechnique, F-75005 Paris harmat@worldnet.fr), mainly aiming at linguists (not at mathematicians!), summing up the basis of the deciphering method, giving the translation of the text with the corresponding philological commentary, and enunciating the diverse proofs of the correctness of the decipherment.

A decipherment which may deeply modify the Mediterranean History

Apart from the fact that the new decipherment of the Phaistos Disk may have some notable repercussions on Greek Linguistics, its most important consequences may be archaeological. Unexpectedly, it may lead to a REVISED HISTORY of many archaeological problems. We will cite, in particular: a) the HISTORY of the GREEKS, which can now be traced from c. 3000 BCE on b) the history of the oldest relations between Greeks and the other Mediterranean peoples: Egyptians, Anatolians, Semitic Peoples, etc. Some difficult problems, like the Troyan War or the Origin of the Philistines, may now have to be deeply revised, in the light of the Proto-Ionian theory, a consequence of the proposed new decipherment.

Complementary Comments


When I wrote my book, I have certainly under-estimated how severe was the controversy concerning the origin of the Disk (Cretan or not ?) or the type of script. I have also underrated that some well-established facts, like the direction of reading, were - surprisingly - not universally recognized as such. A few remarks on these points seem, therefore, useful :


It has been written by some authors that the datation of the Disk (terminus ante quem) was "uncertain" because the layer in which it was found was disturbed. This is not correct. In his well-known book on the Phaistos Disk, Yves Duhoux(1) has made a careful examination, "shred of pottery by shred of pottery" if I may say so, of what Luigi Pernier has written (He is the Italian archaeologist who made the discovery). The conclusion is unequivocal : the layer was undisturbed in the vicinity of the Disk and entirely Middle Minoan.


Because of the corrections he made, it is obvious that the scribe who printed the Disk was not copying a model : these corrections are, by their number (too great for being attributed to a lack of attention in copying!), by their nature (too often the same "mistake"!) and by the way they have been done, most "on the spot" - what means : by the scribe himself - showing that the scribe was not copying a model, but composing a psalm.

As a consequence, there is no way to dissociate the direction of reading from the direction of printing. And, as I have shown in my book, there is ONLY ONE WAY to reconstitute the movements of the printer, explaining at the same time all the epigraphical facts : the scribe has alterned tracing of the spiral and printing of the signs. He began from the outside, tracing first the most exterior spire ( = segment of spiral running from an angulous point to the next one), then printing the signs in the track so-created. Once this track was filled up with signs, he traced the second spire, etc.


In the book already quoted, Yves Duhoux has difinitively shown that, for epigraphical reasons, the Face A has been printed before the Face B.


It seems, for obvious reasons, that an Aegean origin of the Disk cannot be denied. But, for several motives, a Minoan origin is impossible, as I have demonstrated in a recent paper (See Faucounau (3) in Bibliography).

In fact, the Disk appears as a late remnant (c.1800 BC) of the "Maritime Troia-Kultur, as defined by Manfred Korfmann, which was replaced at Troy by the "Anatolische Troia-Kultur" c.2200 BC, but which survived in some of the southern Cycladic Islands until being destroyed by the Minoan expansion (to Thera and Rhodes in particular) at the beginning of the Middle Minoan Period.


The problem is a very difficult one and it took me several years to reach the correct solution, which is : a syllabary of archaic type. To answer a remark I've received, I would like to mention that this type of script is not as strange as it seems at first sight. It was not obvious for the first inventor of a syllabic script to render Greek KROKOS by KO-RO-KO-SE, as Linear B and the Cypriot Script will do at the end of a lenghty evolution. It is more natural to write : KRO-KO-? , the problem being the translation of the final consonant. Two solutions were both used by the scribe who printed the Disk : a)- not to write it (what he systematically did for final -N) b)- to render it by a sign -(A)S or -(A)R.


Concerning the decipherment itself, there is little to add to what I wrote in my book. The language (Proto-Ionic) is probably a surprise for many. But should it be really a surprise ? I don't thing so. A few scholars, like Vl.Georgiev and C.Ruijgh, have already demonstrated the fallacious character of the so-called "Risch-Chadwick Theory". The decipherment of the Disk is just a confirmation that they were right : Ionic is not a "late dialect" !

It seems more important to me to answer to the questions concerning the DECIPHERING METHOD. As I have written, my book is mainly dealing with the linguistical - and not the mathematical! - aspect of the decipherment, although statistical methods have been involved. The reason of this is obvious : explaining and discussing the mathematical part of the deciphering process would need me to write a totally different book. Maybe I will publish it some day. But, for the present, my main concern has been not about explaining the deciphering process - which cannot probably be duplicated (as I wrote, "each decipherment is a unique adventure") - but about stating the PROOFS that the method has lead to the right solution ! So, I would like to say to people who have written to me : "I don't understand why you gave THIS phonetic value to THIS sign" that I am sorry to answer that their lack of understanding is a normal thing. They'll have to wait for my calculations to be published ! Then, they will know why I could not for statistical reasons give a NA-value to Sign 25 or a KE-value to Sign 26 !..

An "a priori objection" I have frequently heard of, is that the text of the Disk would be too short for statistical method to lead to a decipherment. If one uses Ventris' or Ipsen's method, surely ! But there are, happily, more efficient methods when one dealswith a TEXT (not a LIST!), because a text is a collection of quite lenghty Markov's chains. (I cannot elaborate more, here, on the subject).

One reader pointed to me that the reasons of the final choice concerning the language (i.e. Proto-Ionic) were not very apparent. This is a very good remark and I warmly thank my correspondent for making it. It is true that, as I did not publish the details of the DECIPHERING METHOD in my book, the evidence concerning the Proto-Ionic language is not clear enough. To fulfil this absence, here are a few words of explanation about the way it came : During about 4 years, I made statistical comparisons, using "mild sieving criteria", in order to approximately determine the type of script and the language. Those criteria were able to help in discriminating between very different types of script (e.g. syllabic v./ ideographic) and languages (e.g. Greek v./ Akkadian), but not between too close scripts or languages. The conclusion of this first step being in favor of some type of syllabic script with open syllables, and of a Greek language, I adopted the "Greek Hypothesis". Then, during a Second Step (which lasted about 3 years), I used "sharp criteria" in order to get a more precise idea of the SCRIPT. This is the way I came to a SCRIPT with obvious Ionic characteristics (e.g. 5 vowels : A,E,AE,I,O/U), leading to conclude in a PROTO-IONIC language. This was the lucky part of the deciphering process : finding a language which could easily be "reconstruct" from what we know from Ionic !.. Once the language found, the rest was quite easy...


Diverse proofs that the decipherment is correct are mentioned in my book. The most badly understood seems to be the "mathematical" one. It can be enunciated as follows : After 7 years of calculations, I could determine - in total disregard of acrophony - the "probable phonetic values" of some eleven signs. I noticed, then, that those values "seemed" to be acrophonic. The question was : Is it mere illusion ? or the confirmation to be on the right track ?.. Let's suppose that it was just an illusion. What was the probability of finding BY MERE CHANCE eleven "plausible acrophonic values" ? (A "plausible acrophonic value" is a value which could be true if the script was really acrophonic. For instance, for Sign 29 (the "Cat"), the "plausible values" correspond to the plausible identifications, like CAT, LIONESS, LYNX, etc.). As for each sign, the number of "plausible values" does not generally exceed 4 or 5, an easy calculation shows that an illusion is "statistically impossible".

The other proofs can be divided into INTERNAL and EXTERNAL evidence.

The INTERNAL proofs are linked to the decipherment itself. They are :

a)- The philological coherence of the resulting text. I would like to point out that this internal philological coherence cannot be affected by some etymological errors I may have made, concerning the Proto-Ionic names of the signs, for instance (To give an example, I recognize to-day that supposing a form *YUDRIA - as I did on p. 86 - is etymologically impossible).

b)- The plausible explanation of all the obliterated signs. Here again, some possible errors (the obliterated signs are very difficult to read !) would have no real consequence.

c)-The explanation of most, even minute, epigraphical details. E.g. the cramming of the signs in A29, but not in the adjacent boxes A28, A30, A31.

The EXTERNAL proofs are by far the most decisive. They are totally independent from the Disk. They are showing : a)- The existence of a Proto-Ionic language, before (and during) Mycenaean times , in the Aegean area (="Linguistical proofs") b)- The existence c. 2500 BC of Greeks in the Aegean ( = "Astronomical proof") c)- the historical and archaeological coherence of the "Proto-Ionian Theory" ( = "Archaeological proofs" : Sword from Dorak, Seal from Tarsus, Lion from Kea, "Maritime Troia Kultur" from M. Kaufmann, etc., i.e. c.20 pieces of evidence all together)."


These consequences - known as the "Proto-Ionian Theory"- are more important than the decipherment itself. Because the existence of the Proto-Ionians, at such an early date, forces to revise the HISTORY of the GREEKS and of their relations with the other Mediterranean peoples : Egyptians, Anatolians, Semitic peoples, etc.

As an illustration, I would like to give here the reconstructed frame of the very early History of the Greeks that I propose, following the Kretschmerian ideas that I have advocated for more than 30 years. In his famous paper published in GLOTTA 1925, Paul Kretschmer distinguished three linguistic layers in the European languages : a)- Pre-Indoeuropean b)-" Proto-Indoeuropean" (Proto-IE) c)-"Indoeuropean (IE) stricto sensu". This linguistical division was roughly corresponding to three periods : a)- Paleolithic (with late remnants like Basque, Hurrian and some Caucasic languages, impossible to classify) b)- Neolithic (with "proto-Ie" languages, like Lycian, Etruscan and the hypothetical "Pelasgic") c)- Early Bronze Age, with an expansion from 3000 BCE on, of the Indoeuropean languages, like Greek, Sanskrit or Celtic. It was also corresponding to different geographical origins, the birth-place of "Proto-IE" being Anatolia and the Danubian Valley, the one of the "IE stricto sensu" the steppes of Southern Russia, north of the Black Sea (Kurgane Civilization). Archaeologically and linguistically, this scheme is a lot more satisfactory than the "unique homeland" hypothesis for the Indoeuropeans, and it is unfortunate that it has not been universally accepted! (although some slight modifications to the initial Kretschmer's scheme are needed : for instance, Kretschmer's exagerated reliance upon the -NTH- and (principally) -SS- suffixes as "linguistical markers").

The decipherment of the Phaistos Disk enters very well in the Kretschmerian frame : c.3000 BCE, the "first Greeks", the Proto-Ionians, came by sea from the Danubian Delta and settled at Troy, then a few centuries after in the Cycladic Islands and on some coastal places in Greece (Attica and Euboea, in particular). The "cultural unity" noticed during the Early Bronze Age is explained by the Proto-Ionian influence in the Aegean at that time. A cultural unity, which cannot mask, nevertheless, some strong differences between Minoan Crete, Continental Greece, Cyclades and coastal Anatolia at the beginning of the Bronze Age. Between Crete and Greece, for instance, this difference will last until the Mycenaean conquest of the island c.1450 BCE.


1- DUHOUX Yves, "Le Disque de Phaestos", Louvain 1977

2- FAUCOUNAU Jean, "Le déchifferement du Disque de Phaistos", Paris 1999.

3- ______________,"Du Disque de Phaistos aux écritures proto-Philistines" (in print)

4- OLIVIER Jean-Pierre, "Le disque de Phaistos. Edition photographique, B.C.H. 99, Paris 1975

Back to Cover