Deciphering the Phaistos

Disk and other Cretan

Hieroglyphic Inscriptions
Epigraphic and Linguistic Analysis
of a Minoan Enigma

Andreas Fuls

Hamburg: tredition 2019

 MATHEMATICA EPIGRAPHICA
1
Andreas Fuls (Editor)

Fuls, Andreas: Deciphering the Phaistos Disk and other Cretan Hieroglyphic Inscriptions -
Epigraphic and Linguistic Analysis of a Minoan Enigma.
Hamburg: tredition (Mathematica Epigraphica, No. 1) 2019.

Verlag: tredition GmbH, Halenreie 40-44, D-22359 Hamburg

ISBN 978-3-7482-5919-0 (Paperback)
ISBN 978-3-7482-5972-5 (Hardcover)
 Andreas Fuls, Germany.
Contents

PREFACE ............................................................................................................................ IX
ABBREVIATIONS .............................................................................................................. XII

INTRODUCTION ................................................................................................................... 1

1 METHODS OF DECIPHERMENT ...................................................................................... 5

§ 1 Writing systems ..................................................................................................... 5
§ 2 Classifying undeciphered writing systems ............................................................ 8
§ 3 Introduction to successful decipherments ............................................................ 22
§ 4 Unsuccessful decipherments of the Phaistos disk................................................ 36
§ 5 Methods of text segmentation .............................................................................. 38
§ 6 Summary .............................................................................................................. 50

2 EPIGRAPHIC ANALYSIS OF THE PHAISTOS DISK .......................................................... 53

§ 1 The Phaistos disk ................................................................................................. 53
§ 2 The sign list of the Phaistos disk ......................................................................... 54
§ 3 Sign positions ....................................................................................................... 59
§ 4 Paradigmatic analysis .......................................................................................... 65
§ 5 Structural analysis ................................................................................................ 75
§ 6 Word length ......................................................................................................... 82
§ 7 Type of language ................................................................................................. 82
§ 8 Epigraphic results ................................................................................................ 85

3 DECIPHERMENT OF THE PHAISTOS DISK ..................................................................... 87

§ 1 Introduction .......................................................................................................... 87
§ 2 The Luwian Hypothesis ....................................................................................... 87
§ 3 Comparison of signs on the Phaistos disk with Luwian hieroglyphs ................. 89
§ 4 Decipherment by means of phonetic complements and sign comparison .......... 94
§ 5 Decipherment of signs by means of contextual analysis ................................... 109
§ 6 Identified syllables, logograms and determinatives ........................................... 115
§ 7 Summary ............................................................................................................ 116

v
4 TRANSLATING THE PHAISTOS DISK .......................................................................... 119
§ 1 Side A ................................................................................................................ 120
§ 2 Side B ................................................................................................................ 124
§ 3 Undeciphered signs............................................................................................ 128
§ 4 Specific features of the Phaistos disk writing system........................................ 130

5 VERIFICATION OF THE DECIPHERMENT .................................................................... 133

§ 1 Verification by context ...................................................................................... 133
§ 2 The Arkalochori axe .......................................................................................... 135
§ 3 The Malia stone block ....................................................................................... 138
§ 4 Cretan Hieroglyphic writing .............................................................................. 143
§ 5 The language on the Phaistos disk and the Luwian language ........................... 149
§ 6 New readings of Luwian hieroglyphs ................................................................ 151
§ 7 Assessing the reliability of the decipherment .................................................... 153
§ 8 Criteria for a successful decipherment .............................................................. 158
§ 9 Summary............................................................................................................ 163

6 THE HISTORICAL CONTEXT OF THE PHAISTOS DISK ................................................ 165

§ 1 The Aegean Bronze Age.................................................................................... 165
§ 2 Cretan and Anatolian writing systems during the Bronze Age ......................... 172
§ 3 Dating the Phaistos disk ................................................................................... 181
§ 4 The Origin of the Phaistos disk ......................................................................... 181
§ 5 Summary............................................................................................................ 188

APPENDICES ................................................................................................................... 193

I SIGN LIST ............................................................................................................ 194
II NORMALISED WEIGHED SIGN POSITION HISTOGRAMS OF THE PHAISTOS DISK ..... 198
III MULTIVARIATE SEGMENTATION TREES OF THE PHAISTOS DISK ......................... 200
IV WORD LIST......................................................................................................... 221

REFERENCES .................................................................................................................. 224

INDEX ............................................................................................................................. 239

vi
Preface
This book is devoted to the decipherment of the Phaistos disk. The topic of the book
can of course give rise to mistrust and arouse doubts in the mind of the well-informed
reader, since there have been no less than a hundred most equivocal attempts of
deciphering the disk. A claim to suggest a new, for once correct interpretation must
certainly seem suspicious. I can fully understand this reaction, since I am also inclined to
take a cautious attitude towards such claims and always tend to treat them with suspicion.
At the same time, I am far from thinking that the decipherment of the Phaistos disk is
quite impossible and therefore should not be carried on. It should for the simple reason that
not all of the existing possibilities of its exploration have been exhausted. What is
suggested here is the method of combining epigraphic and linguistic analysis. In fact, the
decipherment of the Phaistos disk presented in this book is intended to demonstrate the
effectiveness and potentialities of this method.
The epigraphic research was part of my PhD project on Maya astronomy and
chronology. At that time (back in 1999) I was getting more and more interested in Maya
hieroglyphic writing, since it led to a better understanding of sophisticated astronomical
inscriptions with their complex writing system. My attention was immediately drawn to the
fact that the mechanism of logographic and syllabic sign usage in Maya hieroglyphic
writing was similar to other ancient writing systems. Working as an engineer at the
Department of Geodesy and Geoinformation Science of the Berlin University of
Technology (Technische Universität Berlin), I managed to create an epigraphic database
for Indus writing in cooperation with Dr. Bryan Wells, whose dissertation was dedicated to
the epigraphic analysis of Indus writing (Wells 2011). Our collaboration resulted in the
Online Indus Writing Database that facilitated the search for Indus texts with any sign
combinations and the analysis of spatial and temporal sign distribution.
While working on the project, Dr. Bryan Wells and I discussed possibilities of using
computer programmes that would allow us to perform and visualize paradigmatic and
syntagmatic sign analysis. By means of mathematical methods I managed to develop
statistical algorithms and tools in order to analyse the undeciphered Indus writing system.
This made it possible to draw Normalized Weighed Sign Position histograms as well as to
create an improved Multivariate Segmentation method and a new statistical method of
classifying both known and undeciphered writing systems. These were the new epigraphic
methods that actually laid the basis of the epigraphic analysis of the Phaistos disk.
Teaching a seminar on the decipherment of writing systems, I got familiar with several
deciphered and undeciphered ancient writing systems such as Egypt hieroglyphs, Linear A
and B, Cuneiform, Luwian hieroglyphs, ancient Chinese, Proto-Elamite, Aztec, Maya and
Isthmian writing, Rongorongo and Proto-Byblos writing. The list, of course, included also
the Phaistos disk writing. The first thing that struck me most was a large number of works
claiming to crack the mystery of the disk and suggesting various far-fetched hypotheses of
how to read the pictographic signs. Dismissing scores of ungrounded theories not
withstanding any criticism, I could not but clearly see that in fact there were only a few

ix
contributions that treated the Phaistos disk inscription earnestly enough. Those were the
cases when scholars really tried to focus on the analysis of sign frequencies, their positions
and the internal structure of the writing system. What concerns my own research of the
Phaistos disk, at first I was far from thinking of tackling the problem of deciphering it.
Working on an epigraphic database since 2012 and applying the statistical methods
previously developed for the Indus script to the Phaistos Disk, I was just trying to see if
arbitrary assumptions and misleading conclusions could be avoided in the epigraphic
analysis.
Soon, however, I managed to find a way of identifying some of the suffixes and
prefixes and define the type of writing system on the Phaistos disk as a mixed logographic-
syllabic one. Then Dr. Bryan Wells suggested that I should compare the Phaistos disk signs
with Luwian hieroglyphs, and it was very helpful that he pointed out the similarity of the
Phaistos disk and Hittite seals. Since there is no bilingual inscription that has been found
yet, the comparison with a known writing system is the only hope to decode the Phaistos
disk. The graphic comparison of signs proved to be very fruitful, since previously
identified affixes of the Phaistos disk corresponded to Luwian grammar, which gave rise to
the hypothesis that the root language of the Phaistos disk was no other than Luwian.
The subsequent linguistic analysis together with the sign comparison and the
epigraphic analysis finally made it possible to start a gradual and consistent decipherment.
The identification of a frequent sign and a frequent sign pair both functioning as a
demonstrative pronoun which always occur at the beginning of sentences (‘this’ and ‘that’,
respectively) proved quite promising. Another breakthrough at the very start of my
investigation was the identification of phonetic complements. Thus, the logogram of the
Sun-god ‘Tiwazas’ was prefixed by the syllable ti- and suffixed by the syllable -sa. The
subsequent search for meaningful sign sequences led to the identification and verification
of other sign values. As a result, about 85% of the signs on the Phaistos disk for the most
part can now be read with a high degree of confidence. Therefore, it is possible to
understand most parts of the inscription, and there is good reason to consider the text to be
a letter written in Luwian more than 3500 years ago.
The decipherment offered in this book was stimulated by various discussions and
exchange of ideas with different scholars. Above all, I’m deeply indebted to Bryan Wells,
an archaeologist and epigrapher. We spent long hours on phone calls and exchanged
extensive emails discussing many epigraphic issues on ancient writing systems in general
and on the Maya, the Indus, the Proto-Byblos and the Phaistos disk writing systems, in
particular. He always kept me from going off the track of precise scientific thinking, from
hasty conclusions and wishful thinking. Any failure or mistake in the work is, of course,
entirely my responsibility.
I would also like to thank Karl Maroldt, a linguist from the Berlin University of
Technology. We were discussing many issues concerning writing systems and their
relationship to languages. His suggestions on linguistic terminology proved extremely
useful.

x
 Special thanks to Annick Payne, an expert on Luwian hieroglyphic writing at the Free
University of Berlin, who now works at the University of Bern. She was always very
helpful in clarifying many issues on Luwian hieroglyphic writing.
Last but not least I would like to thank my wife Ekaterina Poljakova and her mother
Irina Poljakova. I cannot thank them enough for their incredibly meticulous reading of the
final proofs.

xi
Abbreviations
Ø zero morpheme
/ alternative reading
[...] reconstructed reading
?? unknown reading
1. 1st person
2. 2nd person
3. 3rd person
A. accusative
AD Anno Domini
act. active
BC before Christ
c. common gender
caus. causative
CHIC Corpus Hieroglyphicarum Inscriptionum Cretae
CV consonant – vowel
CVC consonant – vowel - consonant
D. dative
EC Early Cycladic
EH Early Helladic
EM Early Minoan
eth. suff. ethnical suffix
FT free translation
G. genitive
imp. imperative
LH Late Helladic
LM Late Minoan
loc. Locative
MH Middle Helladic
MM Middle Minoan
N. nominative
pl. plural
PN personal name

xii
 A BBREVIATIONS

prs. present
prt. preterite
pt. particle
qpt. quotative particle
sg. singular
SOV Subject – Object – Verb word order
SVO Subject – Verb - Object word order
V vowel
WT word-for-word translation

xiii
Figure 1: Minoan sites of Bronze Age Crete.

xiv
 Introduction
Phaistos is one of five large palaces of the Minoan culture in Crete. Around 1925 BC
administrative centres had developed in Knossos, Malia, Zakros, Archanes, and Phaistos
dominating their surrounding region. The hill on which the palace of Phaistos was built
overlooks the fertile valley of the Mesara plain. Phaistos thus controlled the southern part
of Crete, while Knossos and Archanes dominated its northern part, Malia its northeastern,
and Zakros its eastern part (Figure 1). Besides the palaces there were many smaller
settlements, caves used for rituals, villas, peak sanctuaries and cemeteries. The distribution
of the Minoan sites shows that mainly the eastern half of Crete was populated.
The palaces had public areas such as spacious courts on the one hand and many private
rooms connected by corridors and staircases on the other. There were also storage
magazines, royal apartments, rooms with benches around the walls, altars, workshops,
water pipes and cisterns for water supply. Each palace thus fulfilled stately, administrative,
economic and religious functions.

Figure 2: View over the palace of Phaistos and the Mesara plain.

The palaces were destroyed more than once and rebuilt on the same site immediately
afterwards. Only the last destruction around 1650 BC, probably triggered by an earthquake,
put an end to the Minoan culture. Thus at least two palace periods can be distinguished, the
Proto-Palatial period of the Old Palaces until ca. 1700 BC and the New Palace period after
1700 BC.
The Phaistos disk was found in a small room at the north side of the palace complex of
Phaistos which belongs to the Old Palace period. Together with Middle Minoan pottery
and a tablet inscribed in Linear A, it had been lying on the ground beneath the ruins of a
collapsed presumably two-storey building.

1
INTRODUCTION

Figure 3: Part of the Northeast complex of the Old Palace in Phaistos with storage
repositories. The Phaistos disk was found in one of them (marked by x) together with other
valuable objects.

Since the time when the Phaistos disk was found in 1908, it has intrigued and baffled
scholars investigating ancient writing. What makes the object so attractive is its uniqueness.
It is unique in several ways. Firstly, it is disk-shaped (16 cm in diameter) and is inscribed
on both sides. Then, the inscriptions themselves are arranged in spirals, starting from the
rim of the disk and running towards its centre. Another unusual characteristic of the disk is
that the writing on it is not inscribed but rather impressed with intaglio blocks into the
softened clay, one block for each sign. Fourth, the text is divided into groups of signs,
presumably words or phrases, by linear lines forming boxes around each group. Finally,
that there is no other example of such a script.
It is obvious that one should be prepared to overcome a number of serious obstacles
deciphering the Phaistos disk beyond the fact of its uniqueness. Probably the most daunting
of them is the issue of the root language of the Phaistos script. As we shall see in the
course of the analysis that follows, the language of the Phaistos disk is not easy to discern,
still it can be identified by means of epigraphic methods developed by the author. The
resulting identification allows us to go beyond a simple statistic analysis of signs and deal
with their meaning. What is suggested here is a systematic decipherment that starts with
the descriptive details of the inscription and progresses step-by-step by means of structural
analysis, cross-cultural comparison and linguistic analysis to arrive at a sensible
explanation of the text on the Phaistos disk.
The first chapter of this book is called ‘Methods of Decipherment’. It examines new
methods of analysing undeciphered writing systems including the method of classifying

2
 INTRODUCTION

writing systems, methods used for analyzing sign positions (paradigmatic analysis), and
methods of text segmentation (syntagmatic analysis). The methods of text analysis are by
no means subjective as they are mathematically based. The chapter also examines some
problems encountered by would-be decipherers and discusses reasons for unsuccessful
decipherments of the Phaistos disk, such as the incorrect identification of the type of
writing system, the reading direction, and the root language.
The epigraphic methods are applied to the inscription of the Phaistos disk in Chapter 2.
The epigraphic methods include statistical and structural analysis of the writing system
without using any specific language or sign function. Its comparison with well-known
writing systems shows that the Phaistos disk uses a mixed logographic-syllabic writing
system where most signs are syllables and only a few are logograms. To classify signs as
initial, medial or terminal, it was necessary to use the calculation of Normalised Weighed
Sign Position histograms. The analysis of internal sign patterns is based on Multivariate
Segmentation trees. This results in a list of paradigmatic sign groups and the identification
of probable prefixes and suffixes. In the final epigraphic analysis, text fields are arranged
in syntactic fields to form potential sentences. In general, epigraphic methods are aimed at
understanding an undeciphered writing system as best as possible avoiding any biased
assumptions or presuppositions. Therefore, as it is shown in Chapter 3, the results of the
epigraphic methods could be used to verify any decipherment of the Phaistos disk.
In Chapter 3 the Phaistos disk and its sign list are compared with Hittite seals and
Luwian hieroglyphs. About 50% of all Phaistos disk signs are graphically similar or
identical to Luwian hieroglyphs and may have the same meaning. Prefixes and suffixes
identified in the previous epigraphic analysis of the Phaistos disk correspond to Luwian
grammar. This leads to the hypothesis that the language of the Phaistos disk is Luwian. A
few logograms are identified by phonetic complements derived after comparing Phaistos
disk signs with identical or similar Luwian hieroglyphs. Hence, one can rely on reading
about 85% of all Phaistos disk signs. The partial decipherment confirms that the writing
system of the Phaistos disk is a mixed logographic-syllabic one and that it makes use of
some determinatives.
The translation of both sides of the Phaistos disk is presented in Chapter 4. However,
there are some signs that remain undeciphered and their meaning can not be determined
with certainty. The grammatical features of the Phaistos disk language are discussed in
detail in the final section of the chapter.
This is followed by the verification of the meaning of the deciphered signs within the
context of the Phaistos disk text. An important text related to the validity of the Phaistos
disk decipherment is the Arkalochori axe. Found in Crete, it is kept in the same museum as
the Phaistos disk. This votive item is inscribed with 18 signs that can also be read in
Luwian.
Another item is the stone altar found near Malia, Crete. It bears a vertical inscription
with 16 signs. Other inscriptions with the same type of writing can be found on seals and

3
INTRODUCTION

sealings with Cretan hieroglyphs. They were excavated in Knossos, Malia, Petras, and
other sites on Crete. If the decipherment of the Phaistos disk can be applied to the
Arkalochori text, the Malia stone altar, and the Cretan hieroglyphic inscriptions, then the
decipherment should be considered far more reliable. This is demonstrated in Chapter 5
where the partial decipherment of Cretan hieroglyphic writing allows one for the first time
to look at the Minoan culture from a historical perspective. The new readings can also be
applied to Luwian hieroglyphic inscriptions, for example, the Karkamiš stone bowl. The
chapter proceeds with an assessment of the quality of the decipherment. The criteria for a
successful decipherment are discussed here in relation to the decipherment of the Phaistos
disk.
The last chapter (Chapter 6) puts the Phaistos disk in the cultural context of the Aegean
Bronze Age. It gives a short overview of the chronology and cultural development during
the Bronze Age in Greece and Anatolia and discusses the writing systems used in this area.
The origin and dating of the Phaistos disk are highlighted in the conclusion of the chapter.
Whether the Minoan palaces were ruled by secular or more religious rulers is still
unclear, although the throne excavated in Knossos, as shown at the end of the book,
resembles the sign for 'high-priest' and thus points to a religious ruler. On the other hand,
the text on the Phaistos disk deals with secular matters such as the construction and
occupation of a throne, succession to the throne, construction measures and a death
sentence, as well as religious topics such as listing various gods and who follows which
god. We can therefore assume that the Minoan palaces were ruled by rulers of both secular
and religious power.

4
Chapter 1
Methods of Decipherment
At present several ancient writing systems remain either undeciphered or their
proposed decipherments are controversial. For example, the decipherment of the Isthmian
script from Mesoamerica by Kaufman and Justeson (2001) was severely criticized by
Houston and Coe (2003) and by Wells (unpublished manuscript 2008). Several attempts
have been made to crack the Indus script (Possehl 1996), but no decipherment has received
a widespread academic support. In some cases, such as the Phaistos disk, only a limited
corpus of inscriptions is known. It is clear that this reduces the probability of successful
decipherment, since there is not enough comparative material available. Therefore, it is
important to start with a critical discussion of writing systems and the methodology used
for their successful decipherment.

1. WRITING SYSTEMS
There are several types of writing systems that were used by ancient cultures. The most
common among them are: alphabetic, syllabic, and logographic writing systems, on the
one hand, and a mixture of logograms and syllables called logographic-syllabic writing, on
the other. An abjad is a special type of writing system that uses only consonants and no
vowels. Phoenician writing, a predecessor of our western alphabet, belongs to this
consonantal type of writing. Vowels were introduced later by the Greeks. Sampson (1985)
classifies abjads and alphabets as segmental writing, since their signs represent phonemes
as their smallest units (Figure 1.1).

Figure 1.1: Classification of writing systems according to Sampson (1985).

5
CHAPTER 1: METHODS OF DECIPHERMENT

Sampson (1985) defines another phonetic writing system as ‘featural’ because it marks
phonetic features, such as initial and final sounds. A well-known example is Hangeul from
Korea.
The composition of syllabic scripts depends on the language they express (Coulmas
2003: 65). Different languages have different phonetic inventories (phonemes). Tonal
languages (Chinese) differentiate between identical sounds with different tones. Many
syllabic writing systems use only syllables of the ‘consonant + vowel’ (CV) shape or pure
vowel (V) shape. Other writing systems can express more complex syllables such as CVC
(e.g. Cuneiform). The sound inventory of the language has a great influence on the number
and inventory of syllables. Persian Cuneiform has only 40 syllables, while Thai writing
uses 90. This is the range of most syllabic writing systems (between 40 and 90 signs).
Other pure syllabic writing systems that fall within that range are Cherokee with 85
syllables, Linear B (written in Greek) with 87 syllables, and the Cypriot script with 56
signs.
Logograms are signs representing either an entire word or a word stem. For example,
sign K’IN in Maya hieroglyphic writing means ‘sun’ or ‘day’ and has calendric usages.
Conventionally, most transcriptions of ancient writing use capital letters to indicate a
logogram in contrast to syllables written in lower case letters. The logogram K’IN can take
a numeric prefix to count days and can be used to express derived meanings. For example,
K’inich means ‘Sun-Face’ and Lahun K’in means ’ten days’.
Phonetic complements are often attached to logograms to clarify the phonetic reading
of logograms. Sometimes phonetic complements are used for logograms with multiple
meanings, but quite often they are redundant and helpful in the process of decipherment.
The meaning of a logogram is independent of the root language. The pronunciation of
a logogram depends on the dialect of the speaker, but the logogram itself remains
semantically unaffected. For example, logographic writing can help two Chinese speaking
two different dialects, when both of them understand the same Chinese signs.
Most ancient writing systems are logographic-syllabic, i.e. they use a mixture of
logograms and syllables. Every logographic-syllabic writing system has its own ratio of
logograms to syllables, and spelling rules can approximate the linguistic value of a word.
There is often some degree of flexibility of how a specific word can be spelt (the degree of
its phonetization). For example, the word balam (‘jaguar’) in Maya hieroglyphic writing
can be written in many different ways (Figure 1.2).

6
 WRITING SYSTEMS

a) b) c)
Figure 1.2: Spelling variations of balam (‘jaguar’) in Maya hieroglyphic writing: a) logogram
BALAM, b) logographic-syllabic spelling ba-BALAM-ma, c) syllabic spelling ba-la-ma.
Another problem is the use of graphic variants with the same phonetic value
(allographs). Maya hieroglyphic writing uses at least 13 different allographs of the syllable
/u/ (Coe and van Stone 2001: 157), and Linear B uses three different signs representing
vowel /a/ (Figure 1.3: sign a, a2, a3).

Figure 1.3: Allographs of vowel /a/ in Linear B.

Another hindrance to decipherment is that a sign can have more than one value
(polyvalence of a sign). For example, in Sumerian cuneiform the sign DINGIR for the
logogram ‘sky’ functions as a semantic determinative for the names of deities and at the
same time as the syllable for /an/ (Coulmas 2003: 47). A few centuries later, after
Sumerian cuneiform had come into use, Sumerian cuneiform was adapted to the Akkadian
language. The Akkadians assigned new phonetic values to the signs in their own language
but retained their Sumerian meaning. This resulted in the Akkadian cuneiform sign
DINGIR being used as a logogram meaning ‘god, goddess’ and ‘sky, heaven’, as a
determinative for the names of deities, and as the syllable /an/ and /il/. Therefore, it had 5
different values or functions (Figure 1.4).

Figure 1.4: Polyvalence of the DINGIR sign in Sumerian (left) and Akkadian cuneiform
(right).
We must keep in mind that writing differs from what we usually mean by ‘language’.
Writing is discrete, timeless, autonomous, permanent, visible and is produced by hand.

7
CHAPTER 1: METHODS OF DECIPHERMENT

One of the aspects of language is speech which is continuous, bound to time of utterance,
contextual, temporary, audible and is produced by the organs of speech (Coulmas 2001:
11). Nevertheless, there is a complex interaction between writing and language.

2. CLASSIFYING UNDECIPHERED WRITING SYSTEMS

One of the first steps in analysing an undeciphered writing system is the identification,
tabulation, and count of signs (graphemes). The distinction between signs and graphemes
requires some explanation. Inscriptions consist of graphemes which are graphic
representations of signs. They differ by the method of writing (painting, carving,
scratching), writing material, font, size, and type of hand writing (if not impressed as is the
case with the Phaistos disk signs). A sign, by contrast, is the interpretation of one or a list
of similar graphemes as an abstract and unique symbol representing a specific meaning
and/or a phonetic value. The process of sign identification is complex because of
allographs and polyvalent signs, and requires the paradigmatic analysis of all inscriptions
available. An example of sign analysis is described by Wells (2015: 13-18).
The first step results in a list of signs with unknown meanings and sound values. While
alphabetic or consonantal scripts have no more than about 36 signs, “pure” syllabic writing
systems have between 40 and 90 signs (Coe 1999: 43; Robinson 2002: 41-42). If there are
several hundred signs or more, the script is sure to contain many logograms. The
classification of logographic-syllabic writing systems with an unknown ratio of logograms
to syllables is problematic. The number of syllables depends on the language and varies
between 40 and 90 syllabic signs. This approach also ignores polyvalence and allographic
variation. Therefore, it is inappropriate to classify and describe the degree of phonetization
of an undeciphered writing system by using only the given number of distinct signs alone.
Counting the frequency of each sign provides a more precise assessment of a writing
system. Writing systems consisting mostly of logograms (Chinese or Proto-Sumerian
cuneiform) have several thousand signs, although most of them occur very seldom (Figure
1.5a). In syllabic or mixed writing systems some signs are very frequent, but the frequency
for the majority of signs ranges between medium and low (Figure 1.5a and b). In
alphabetic writing, signs (actually characters) range between medium and high frequency
(Figure 1.5b). As the comparison of histograms shows, the distribution of sign frequencies
depends on the type of writing system and is therefore a measure to determine the type of
unknown writing systems. We will make use of this observation in the next section.

8
 CLASSIFYING UNDECIPHERED WRITING SYSTEMS

a)

b)
Figure 1.5: Histogram of (a) logographic and logographic-syllabic writing systems, (b) syllabic
and alphabetic writing systems.

9
CHAPTER 1: METHODS OF DECIPHERMENT

The frequency is affected by allographs and polyvalence. Polyvalence increases the

frequency of signs. By contrast, the frequency of signs with many allographs is split into
low frequencies for each allograph. Therefore, histograms give a statistical overview of
sign frequencies related to the graphic representation of signs and not of their phonetic or
semantic values.

2.1. Sign frequency

The histogram of sign frequencies (Figure 1.5) is useful for distinguishing between
different types of writing systems. In order to be more precise in classifying writing
systems, it is necessary to apply a mathematical law which would describe the sign
frequency distribution of each writing system and reduce the data to one parameter.
George Kingsley Zipf (1935) proposed a power law to be used in counting the
frequency of words in different languages (Zipf’s law). Testing Zipf’s law for the
frequency of signs in writing systems shows that the rank distribution of sign frequency
does not fit a simple power law of:
f(r) = A / r = A * r-1 (1.1)
where r is the rank, A is a constant, and the sign frequency is f(r). Instead, one may
count the number of signs with the same frequency that follows a Modified Power Law
(MPL):
w(r) = B * rG (1.2)
where r is a rank, B is a constant, G is an exponent, and the number of signs with the
same frequency is w(r) (Naranan and Balasubrahmanyan 2005). Ranking is performed by
sorting w(r) in descending order. In a double-logarithmic diagram each MPL is represented
by a straight line (a linear regression line), where the slope equals exponent G (Figure 1.6).
Since each line declines from left to right (with increasing rank), the slope is, by definition,
a negative value.
As can be seen in Figure 1.6, exponent G is a parameter which describes the sign
frequency distribution of each writing system (Fuls 2015c). The exponent varies between
about -2 for pure logographic writing systems and for word counts (in an alphabetic script
but treating each word as a logogram), about -1.0 to -0.6 for pure syllabic writing systems,
and zero for alphabets (counting the frequency of characters). Logographic-syllabic writing
systems, such as Classic Maya glyphs (G = -1.4) or Old-Akkadian cuneiform (G = -1.1),
fall in between the extreme values. Egypt hieroglyphic writing, a logographic-consonantal
writing system, also falls within the same range, since its exponent G equals -1.4.

10
CHAPTER 1: METHODS OF DECIPHERMENT

published in the 16th century edition of Landa’s Relation (Tozzer 1941), as a partial
syllabary of Maya hieroglyphics (Robinson 2002: 122-123).

4. UNSUCCESSFUL DECIPHERMENTS OF THE PHAISTOS DISK

Since its discovery, the Phaistos disk has attracted many scientists and amateurs who
have been working on its decipherment. The proposed decipherments are based on various
assumptions and methods, however, none of them has been universally accepted.
There are three important questions to be answered before starting any decipherment.
First: is the type of writing system of the Phaistos disk an alphabetic, a syllabic, a
logographic, or a mixed logographic-syllabic one? Second: what is the root language of the
text on the Phaistos disk? Third: what is the reading order, from left to right or from right
to left? Each question must be analysed before decipherment makes any progress. Any
wrong answer to at least one of the three questions would inevitably lead to incorrect
decipherment.
Until now, many of the works on the Phaistos disk have often been based on untested
assumptions. The type of writing system is often assumed to be syllabic because of the
number of 45 distinct signs (and an attached stroke). The problem is that the text of the
Phaistos disk (totalling 241 signs) is a small sample. Comparing all signs on the Phaistos
disk results in a sign list which is shorter than is usually expected from a large text corpus.
The repetition of some high frequency signs and sign pairs probably contradicts pure
logographic writing, but it does not contradict a mixed logographic-syllabic writing system.
For example, one Maya inscription totalling 241 signs has only 135 distinct signs, while
the Maya sign list includes about 831 signs besides many rare signs (mostly logograms).
There arises a need for a new method which would account for different number of signs in
the corpus of inscriptions. In Chapter 1 §2, we introduce a new method for classifying
unknown writing systems based on comparative statistical analysis. In Chapter 2, this
method is then applied to the Phaistos disk (Figure 2.6).
There are many root languages proposed for the Phaistos disk: Greek, Sumerian,
Semitic, Hittite, Luwian, Slavonic, Old Colchian, and Polynesian, among others (Eisenberg
2008, Timm 2008). They belong to different language types. To identify the root language
correctly is in fact the most important task. Still, the proposed cases of decipherment often
assume a root language regardless of the structural analysis of the Phaistos disk text. In this
book, we have developed a new method which makes it possible to distinguish between
agglutinative, fusional, and isolating languages (Figure 1.11). This method is applied to the
Phaistos disk in Chapter 2 §7 (Figure 2.33).
There are also non-linguistic interpretations of the Phaistos disk. Pomerance (1976)
believes that it is an astronomical document, Hagen (2001) makes it a calendar, Mosenkis
and Pereverziev (2010) consider the Phaistos disk to be a star compass, Wenzel (2012)
regards it as a planetarium, and Aleff (1982) interprets the Phaistos disk as a game board.

36
 UNSUCCESSFUL DECIPHERMENTS OF THE PHAISTOS DISK

Yet, the analysis of the entropy of the Phaistos disk shows that the sign sequences belong
to a writing system and not to a system of symbols.
Alphabetic Syllabic Logographic- Ideographic
Language syllabic
Greek Ohlenroth Hempl 1911, Stawell 1911, Matev 1984, Delekat
1996, Ephron 1962, Dogas 1986, Bowden 1992 1979
Martin Fischer 1988, Otto 1989,
2000 Szalek 1999, Faucounau
1999
Hebrew Egert 1989
Hittite Davis 1967, Georgiev 1976 Prendergast
Luwian Best & Woudhuizen 1988 Fuls (this book)
Old Kvashilava 2006
Colchian
Polynesian Fell 1976
Semitic Blaufuss 1928, Gordon
1966, Aartun 1992, Dürr
2000
Slavonic Rjabchikov 1998
Sumerian O’Brian
1985
Table 1.6: Proposed cases of decipherment of the Phaistos disk comparing the language and
type of writing. The proposed reading directions are right/left (inwards to the centre) or
left/right (centre-outwards to the rim), or not known to the author.
Timm (2009: 17-20) distinguishes two groups of decipherments. In the first group of
26 proposed decipherments, the reading of the Phaistos disk starts in the centre and runs to
the rim (from left to right). The second group (14) reads the Phaistos disk from the rim
towards the centre (from right to left). If the Phaistos disk represents a writing system with
a specific reading direction, it follows that one of the groups must be wrong. There are
several arguments suggesting that the reading direction is from right to left, which is
discussed in detail in Chapter 2 §3.1.
Another problem arising in the process of decipherment is connected with the applied
methods and with the way the results and the proposed readings of signs are verified. The
methods frequently used are the acrophonic principle and the try-and-error use of phonetic
values. Both methods involve a high risk of deriving false conclusions. Verification within
small data sets runs the risk of being circular, but an external verification increases the
reliability of the proposed sign values. External verification can be performed with the help
of iconography coming from other sources or new inscriptions, such as the Arkalochori axe
and the Malia stone block discussed in Chapter 5.

37
Chapter 2
Epigraphic Analysis of the Phaistos Disk
Epigraphic methods presented in Chapter 1 are applied here to the inscription of the
Phaistos disk. The first step is to compile an accurate sign list and to analyse the sign
frequency by comparing it to already known writing systems. This allows us to estimate
the type of writing system and its degree of phonetization (§2). The positional analysis of
each sign gives an idea of possible suffixes, prefixes, and syllables (§3). The next step is to
analyse sign combinations in order to show their paradigmatic relationships (§4). Sign
pairs can be distinguished by their frequency and connectivity value as well as by their
positional behaviour. This makes it possible to identify strong sign pairs as morphological
units. The final step is to analyse the syntagmatic relationships of text fields in order to
understand the structure of similar sentences and the syntax of the Phaistos Disk root
language (§5). The epigraphic results allow us to determine the average word length (§6)
and the type of the root language (§7) of the Phaistos disk text. All this analysis is an
important and necessary preparation to start the actual decipherment of the Phaistos disk in
Chapter 3.

1. THE PHAISTOS DISK

The Phaistos disk was found in 1908 together with a Linear A tablet at the Minoan
palace in Phaistos, Crete. The dating is rather vague but is based on the archaeological
context, therefore the disk presumably dates back to a period between 1850 and 1600 BC
(Chapter 6 §3; Robinson 2002: 303; Timm 2008: 14). It is a circular clay disk (about 16 cm
in diameter and 2cm thick) impressed with unknown signs on both sides (Figure 2.1).
Because of its uniqueness and corpus of a single text, the Phaistos disk has remained
undeciphered until now. The enigmatic signs indicate some typical pattern of sign usage,
but one of the major questions still remains: What is the root language of the Phaistos text?
To approach this problem we need a detailed morphological analysis of the text. This will
help us to understand the nature and the root language of the Phaistos disk writing system.

Figure 2.1: Scheme of the Phaistos disk sides A and B with the numbered text fields.

53

Index
A
Abjad ............................................... 5, 35
Abugida ............................................... 32
acrophonic principle .................... 35, 160
Aegean sea ......................................... 165
affix...................................................... 33
affixation.............................................. 15
allographes ......................... 7, 10, 29, 162
Alphabet ..................................... 5, 18-19
agglutination .......................... 13, 15, 161
Anatolia ..................................... 165, 181
Archanes ........................................ 1, 164
Arkalochori axe .......... 132, 135-138, 179
Attica ................................. 113, 122, 162
B H
boustrophedon ............. 30, 143, 177, 180
C Hittite ............. 87, 96, 103, 106, 119, 120
case ending ........................................ 150
Cherokee ............................................ 6, 9
Chinese writing .............................. 13, 18
chronology
-, Egyptian .................................. 166-167
-, high .................................. 166-168, 181
corpus of inscriptions .................. 29, 177
Cretan hieroglyphic writing ....... 143-149,
.................... 172-173, 179, 184, 196-197
Cuneiform ........................................ 7, 24
-, Akkadian ............................................ 7
-, Hittite.......................................... 9, 171
-, Luwian ............................................ 109
-, Old-Akkadian ............................. 10, 28
-, Persian .......................................... 6, 59
-, Proto-Sumerian .................................. 9
-, Sumerian ............................................ 7
-, Ur III................................................. 18
Cyclades .................................... 165, 186
Cycladic civilization .................. 166, 169
Cypriot script ................................... 6, 34
D -, root ............................... 24, 36, 93, 162
decipherment ............ 22-36, 94-116; 153
-, reliability of .................................... 156
degree of fusion ................................... 15
determinative ... 7, 93, 116, 130, 180, 221
INDEX
Devenagari ........................................... 31
disharmony .......................................... 93
DNA .................................................. 185
E
Egyptian hieroglyphic writing . 10, 14, 24
Epi-Olmec writing → Isthmian writing
exponent G ................... 10-19, 57-58, 82
F
fusional → language, fusional
G
Ge’ez ............................................. 14, 19
graphem ................................................. 8
graphic universals ................................ 34
Hangeul ................................................. 6
Hiragana .............................................. 56
homophony .......................................... 33
I
i-mutation ............................ 97, 100, 137
Indus writing... 12, 20, 28, 35, 38-50, 158
inflection ........................................ 13, 15
information theory ............................... 43
isolation ............................................... 13
Isthmian writing ..................... 13, 21, 27
K
Kanji .................................................... 56
Karkamiš stone bowl ......................... 151
Knossos ................................. 1, 169, 192
L
language
-, agglutinative ................. 16, 82, 84, 163
-, fusional ....................................... 16, 82
-, isolating ...................................... 16, 82
-, type of ........................................ 16, 82
Linear A ....................................... 20, 173
Linear B .......... 6, 24, 28, 33-34, 170, 174
loan word ........................................... 181
logogram ........... 6, 93, 115-117, 180, 221
-, marker of ........................................ 179
239
Luwian hieroglyphic writing .... 31, 87-91 suffix ............................... 67, 84, 92, 150
..................................... 151-153, 175-180 syllable ............................... 115-117, 180
Luwian language ..... 37, 89, 93, 149-151, synharmony ......................................... 93
.............161, 171-172, 175, 181-182, 187 synonym .............................................. 33
syntagmatic analysis............................ 32
M
Malia................................ 1, 88, 138, 169 T
Malia stone block ............... 138-142, 179 Thera eruption ................................... 166
Maya hieroglyphic writing .... 6-7, 10, 32, Troy ........................................... 165, 170
................................. 35, 36, 93, 158, 159
Miletos ......................................... 89, 171 W
Minoan ........................................ 88, 170 word length.......................... 13-19, 82-83
Minoan Bronze Age .................. 166, 181 word order ......................................... 151
Minoan sites .............................. 183, 185 writing ................................................... 7
Minotauros ................................ 113, 162 -, logographic ............................ 9, 16, 56
Modified Power Law ........................... 10 -, logographic-syllabic..... 6, 9-10, 18, 37,
morpheme .................... 13, 40, 66, 73, 74 ....................................... 58, 87, 221-222
morphological boundary ..................... 78 -, syllabic ............................... 6, 8, 14, 37
- system ....................... 5, 36, 55, 86, 160
P
paradigmatic analysis ....... 32, 65-70; 153 Z
particle ............................................... 149 Zakros .................................................... 1
period Zipf’s law ............................................ 10
-, New Palace................................. 1, 166
-, Post-Palatial ................................... 170
-, Proto-Palatial.............................. 1, 166
Phaistos.......................................... 1, 169
Phaistos disk ..................... 52-53, 87, 181
phonetic complement ... 6, 111, 131, 154,
................................................... 161, 180
phonetization .......................... 11-13, 160
plene writing ...... 102, 154, 156, 175, 180
polyvalence...................................... 7, 10
prefix ....................................... 67, 85, 91
Pre-Greek language ........ 89, 94, 182-183
pronoun.............................................. 150
Proto-Byblos writing ............... 12, 19, 30

R
reading direction ....... 30-32, 59, 160, 180

S
segmentation........................ 38-50, 68-70
Semitic................................................. 19
sign frequency ..................................... 10
sign connectivity ................ 46-49, 70, 78
sign list ................. 8, 24, 29, 54, 194-197
speech ................................................. 7-8
structural analysis ................32-33, 75-81

240