Navigation – Plan du site
Dossier Rongorongo Tablet Keiti: a discussion

Tablet Keiti and calendar-like structures in Rapanui script

Konstantin Pozdniakov
p. 39-74

Résumés

L’écriture Rapa Nui n’est toujours pas déchiffrée à ce jour, malgré quelques déclarations triomphantes affirmant le contraire. Pour ce qui est du contenu sémantique des textes rapanui, le seul point qui fait consensus est l’existence d’un calendrier dans l’un des fragments du texte appelé Mamari. Il a pu être identifié grâce à la structure particulière de ce texte, mise en évidence par Thomas Barthel et Jacques Guy. Je montre dans cet article que la structure en question se retrouve également dans des fragments de la plupart des textes rongorongo. J’analyse également ici certaines autres structures largement représentées dans l’écriture rapanui, et je pose des principes pour l’analyse structurale d’un texte rongorongo, appliqués à l’analyse du texte dit Keiti. Le choix de ce texte vient de ce qu’il a été au centre d’une polémique entre chercheurs confrontant leurs approches théoriques pour le dechiffrement de l’écriture Rapa Nui. La très grande majorité des chercheurs s’appuie sur le catalogue de Thomas Barthel (500 graphèmes), qui, on le sait, comprend non seulement des signes, mais aussi des ligatures, c’est-à-dire des combinaisons de signes. Les résultats présentés ici s’appuient au contraire, sur un catalogue de 50 signes (annexe 1). C’est la découverte de plusieurs séquences parallèles de signes dans différents textes et leur analyse qui ont permis de remettre en cause le catalogue de Barthel. Vingt d’entre elles sont présentées dans l’annexe 3 de l’article, où elles sont, pour la plupart, publiés pour la première fois.

Haut de page

Dédicace

To Igor Pozdniakov, father and coauthor

Texte intégral

1What does it mean to describe the structure of the text for Rapa Nui script?

2To my opinion, this question deserves a detailed answer in light of recent publication of two papers (Melka, 2008; Wieczorek, 2011) that discuss actively the “structure” of tablet Keiti (text E in Barthel’s nomenclature). In addition to these, a detailed palaeographic study of the tablet was published by Horley (2010). Thus, text E was quite lucky (in contrast to the original tablet, which perished in the fires of the First World War) to be the subject of three research papers, and even a chapter of a monograph (de Laat, 2009). In all known rongorongo corpus, perhaps only tablet Mamari and Santiago staff were honored with such concentrated attention of the scholars. Such an arduous discussion about text E, to my opinion, was triggered by Melka’s paper. From semiotics we know that one of the main functions of the text is to generate other texts. The polemics concerning the tablet Keiti makes a good illustration of this idea: today, if a scholar working in rongorongo field does not express an opinion about text E, it may be considered as bad manners. Fulfilling the “requirement” to write a paper about this particular inscription, I would like to use this opportunity to define the principal bases for structural analysis of Rapa Nui script in general, to outline the minimum criteria that should be met for any scholarly description of text structure, and finally, to emphasize the importance of proper structural analysis, which can be extremely useful for a potential deciphering.

3To catch up with modern trends in rongorongo studies, it was decided to dedicate a special attention to a popular question about the potential calendars in the Rapa Nui script, which became the focus of two other papers published in this issue (Wieczorek, 2011; Horley, 2011).

The structure of rongorongo inscription

4Can we judge about the structure of the text, if we can’t read it? The text is composed of signs that can be grouped into a sign catalogue. If the things were that simple, the question would be much easier to solve. The problem is that today we don’t have any widely-accepted catalogue, which means that the unified version of text E is nonexistent. According to Barthel (1958), text E has 880 signs, according to Horley (2005)–982, according to Pozdniakov-Pozdniakov (2007), there are 1115 signs. Barthel’s catalogue includes over 500 different signs. At the modern state of knowledge, it seems incorrect to call it a catalogue – perhaps, rather a first approximation of a catalogue – because it contains so many duplicated glyphs, allographs and ligatures that definitely should not be there. The interested reader is advised to read Guy (2006) for a critical review.

5Both Melka and Wieczorek are working with Barthel’s catalogue, knowing that it is heavily flawed. Why one cannot simply remove ligatures and allographs from Barthel’s catalogue? Well, because no people are working in this field of rongorongo studies – I don’t know any publication dedicated to the systematical improvement of Barthel’s catalogue, while there are dozens of publications with “final and complete decipherment ” of rongorongo. The papers by Melka and Wieczorek are no exception – they are based on Barthel’s catalogue for the want of any better sign list. The problem here is that to distill a real catalogue, one should make – manually, on pre-computer stage – a very complicated and painstaking work on comparison of known parallel texts (H/P/Q and Gr/K), as well as almost a hundred of parallel fragments, which again never appeared in a systematic publication (this important lacuna is partially filled by the Appendix 3 of this paper). And this may be the only way to separate the meaningful and non-meaningful variations of the graphemes, finally arriving to a formal structural analysis of rongorongo texts. To the contrary to Barthel’s sign list, our catalogue Pozdniakov-Pozdniakov (2007) has been thoroughly filtered, and now it contains only several dozens of confirmed independent signs (see Appendix 1). Therefore, looking on the same inscription, we actually study differenttransliterated texts that naturally have differenttextstructures.

6Another important question: let us imagine that we finally came to an agreement about the sign catalogue and paved the way to the studies of text structure, and, in particular, for the structure of text E. In accordance with the papers published on the topic, I think it is imperative to consider two principal points.

7The first: Yuri Knorozov, who proposed a possible genealogy in Gv, already knew in the 1950s that text E features ten occurrences of the same fragment with small variations (Butinov and Knorozov 1957: 10): “Text VII (Keiti) has a row of the initial double combination of the signs of the earth and the rat” – which are nothing else than glyphic groups 4.430-22.380, mentioned by Melka as “sequence beta”. How it comes that in 21st century Wieczorek writes the following:

«In a recent article, Melka (2008) put tablet Keiti…under careful structural analysis. He identified three types of glyphic sequences that repeat several times in the recto side of the tablet, which were earlier independently retrieved by Sproat (2003) and Horley (2007: 26). However, Melka (2008) submitted them to more dedicated study, and named each one of them; sequence alpha 1-10, sequence beta 1-7 and sequence gamma 1-10 according to the number of times they are repeated on the tablet.»?

8This is a serious problem, which cannot remain unnoticed, at least because of required honors to the memory of professionals working in rongorongo field. I think that everyoneof them knew about these ten fragments of text E; they also knew, for example, that in the inscription of the famous Santiago staff (text I according to Barthel) the signs can be grouped in graphical triads (we were discussing these triads with Fischer and Barthel in Leningrad many years before the publication of Fischer’s “discovery”). There are hundreds of such particular observations in collections of every specialist, which does not make it ready for publication for the simple reason that a publisher would rather accept the next sensational decipherment rather than a paper with a detailed structural observations without any “cosmic” conclusions.

9The second: Wieczorek’s paper uses many times the words “structural analysis”. He believes that Melka presented the detailed structural analysis of the text E and made several important discoveries; now, it’s Wieczorek’s turn to improve the work of his forerunner by discovering new structural properties characteristic to this inscription. Actually, both Melka and Wieczorek reduce the structural analysis to the study of ten widely known repetitive fragments “alpha-beta-gamma” (here I use their terminology to facilitate discussion) and ten fragments “beta” repeated separately. Apart from this, the authors point to the obvious and well-known (since 19th century) frequent occurrence of sign combination 380.1 (see Harrison, 1873: 379):

«In the four middle lines (of tablet G, recto side)…the signs are arranged in compartments or paragraphs, each of which commences (or ends) with the sitting figure of a negrito holding a staff…there are in all thirty-one of these figures, and consequently the same number of compartments.»

10Grounding their own analysis on these “discoveries”, Melka and Wieczorek propose the hypothesis that the texts on recto and verso sides of tablet E are different, allowing Wieczorek to exclude verso side from his analysis. These are practically all the observations about the structure of the text E, to say nothing about incomprehensible comments about the sequence “beta” on the recto side of the tablet.

11Apart from the aforementioned problem with a use of different glyph catalogues, it is difficult to accept Wieczorek’s analysis of the fragments “alpha”, “beta” and especially “gamma” because the proposed description of the structureis unacceptably simplified (this will be discussed in more detail below). To my opinion, the structure analysis of rongorongo text, including text E, should contain at least the following aspects:

12There are many sign sequences frequently repeated in different texts. Text E is no exception, the sequences “alpha”, “beta” and “gamma” should be analyzed (what is partially done by Wieczorek), but, to reiterate, this could be done only by basing oneself on the firm catalogue, where meaningful differences would be separated from insignificant ones and freed from ligatures and numerous allographs (the key issues causing serious problems with Barthel’s catalogue).

13The texts may “favor” certain signs. The same sign is more frequent in some inscriptions and virtually disappears from the others. There also exist atypically rare signs. Text E is no exception to this, featuring very frequent sign 63 in four initial lines of Er. To discuss the frequency of occurrence for each sign, it is insufficient to publish a catalog different from Barthel’s – it is necessary to perform a clear transliteration showing how exactly each Barthel’s sign re-codes in each particular place. The detailed description of this point goes far beyond the format of a paper. Therefore, I will only make a complete re-coding of text E to our catalogue without detailed argumentation in Appendix 2. In each particular case I am willing to present the supporting evidence confirming re-coding of each particular glyph. These arguments are based on the detailed analysis (on a pre-computer stage) of sign use in the parallel texts (H=P=Q, Gr=K) and dozens of parallel fragments.

14Regularuse of two different (according to Barthel) signs in the same context (the same position in parallel texts or parallel fragments) was used to declare two Barthel’s signs as allographs of the same sign. Similarly, one can be sure that a common ligature should be split if in the same context it appears in joint writing in one text, while it is represented by separate signs on another tablet.

15The second strong criterion in favour of meaningful separation of the ligatures concerns sign collocation. I will repeat the example from our previous publication. How will respond the frequency of sign collocation to separation of letters Q and R into “ligatures” “O\” and “P\”? The “excessive” backslash sign “\” will be quite frequent, but it will combine only with two signs – O and P, which will “overload” all the possible limits for collocations of the “normal” signs. Due to this criterion, we restrained from further splitting the signs 240, 380, 91 and 99. Theoretically, one can see sign 99 as a ligature 14+200 (also 91 as 62+280), but these glyphic components will occur so frequently in these exclusive combinations that it will be inconceivable to suggest that they represent separate signs.

16The structure of the text is not merely defined by the repetitive fragments within it and the characteristics of particular signs, but also by parallel fragments shared with other inscriptions. Tablet Keiti is common in this sense – considerably long sign sequences from text E can be found in the other texts, sometimes on a single tablet and sometimes on most of them. Such parallel fragments can be composed from 5 or 6 signs, but some of them may have scores of glyphs. Alas, it is impossible to present all discovered parallel fragments in a single paper – it can be possible only in monographic publication. However, as this paper focuses on text E, I decided to present here 20 parallel fragments from this text, which are listed in Appendix 3 as they appear. Publication of these fragments, to my opinion, is a priority for improving future studies of Rapa Nui script.

17In each rongorongo text one can found passages (sometimes several lines long) that do not occur anywhere else except in a single text. These sign sequences should also be highlighted in the structural analysis.

18Some text passages of Rapa Nui script has a particular structure, in which the text is divided in very short segments by regularly combining two or three signs, for example, the combination 380.1. Text E is also common in this sense, with its lines Ev1-Ev5 organized just in this way. Jacques Guy called them “harmonic sequences” (2006: 59). I will rather use here the terminology proposed by Yuri Knorozov, who called the sign groups bracketed with delimiters as “blocks”, and the whole sequence forming “block sequences”.

19The composition of the aforementioned five factors determines, to my opinion, the structureof rongorongo text. Comparing the distribution of repetitive sign fragments, clusterings of particular signs, sequence and distribution of the fragments appearing in several texts, the “voids” filled by the unique sign combinations, block sequences and graphical delimiters standing between them, we will obtain full description of general structure of the text, which then can be compared with structure of the other texts.

20Analysis of text E structure performed by Wieczorek is mainly limited to the first point only. Additionally, to my opinion, the definition of repetitive fragments of Er requires significant improvement. In this paper I will describe the structure of text E as a whole. Taking into account the fact that Wieczorek discovered a calendar in text E (the most fashionable topic in Rapa Nui studies!) I will dedicate a special attention to the important problems caused by such interpretations.

Sub-structures “Alpha” and “Beta”

21Wieczorek uses slightly modified subdivision scheme suggested by Melka, defining three repetitive segments that occur ten times in Er: the “sequence alpha-beta: composed of glyphs 300.028x-004.430-022.430y”, as well as “the final sub-sequence alpha-gamma: composed solely of an athropomorphic glyph with various suffixes” (Wieczorek 2011). However, in his table 3 entitled “Sequence gamma 1-10 ”, Wieczorek presents sequences 004.430-022.430y, which are interpreted in the text as “alpha-beta”. Naturally, this issue does not facilitate the discussion. Barthel’s transliteration of the fragments 1-10 can be found in Wieczorek’s table 1; the transliteration of the same segments to our catalogue (Pozdniakov-Pozdniakov) is given below.

22Apart from sign codes used, the most pronounced difference between the data in table 1 and the table from Wieczorek’s paper concerns inclusion of the signs 63 into segments “alpha-gamma”. Wieczorek says that sequences “alpha-gamma” are “composed solely of an athropomorphic glyph with various suffixes”, excluding sign 63 from the segment. This result is the consequence of using Barthel’s transliteration, which for the segments 1, 2 and 9 shows sign 63, while segments 3-8 feature glyphs 203 or 203s that actually have sign 63 included. As it may be confusing for the reader, here are the glyphs from the first three sequences “alpha-gamma” accompanied with Barthel’s and Pozdniakovs’ encoding:

(Pozdniakov-Pozdniakov):

(Pozdniakov-Pozdniakov):

23In this way it becomes clear that sequence “alpha-gamma” does not go with whatever athropomorphic glyph, but with the same sign 200 in all ten cases.

24I would like to emphasize that the signs were interpreted as ligatures in our transliteration because of criterion independent on text E: there are many examples in rongorongo corpus showing regular correspondence of sign 203 to the combination 200.63, proving the necessity to separate a ligature and exclude sign 203 from the catalogue.

25Importantly, the last sign closing the segments 1-10 is unusual. The final ligature (204.077 in Barthel’s transliteration) transcribes as 200.6.44 in our catalogue (with somewhat questionable identification of glyph 44).

26Perhaps, we are dealing here with a “marker glyph” signaling the closure of a homogeneous set of segments, so that the last sign can actually be downward-pointing “adze” glyph 63. As we know from the statistical studies, anthropomorphic signs “prefer” facing to the right with their hands raised. But in some cases these glyphs “turn” to the left and their hands are depicted dangling. The exact function of these deviations is not known. I would like to suggest a hypothesis based on dozens of examples (the discussion of which goes beyond the scope of this paper): the deviation from the standard orientation of the signs may have a special “marker” function signaling the beginning/ending of a complete textual passage.

principallyrdththrdththththth

Statistics

27One can choose among a significant number of statistical characteristics to study rongorongo (Pozdniakov-Pozdniakov 2007). Here we will consider the most important of these – the overall occurrence frequency of the signs. Let us compare sign frequencies in text E with average ones for the reference corpus (Cor) formed with inscriptions of the tablets A, B, C, E, G/K, H/P/Q, N, R, and S. Text E in our transliteration contains 1,155 signs, while the reference corpus is composed of 12,414 sings, Table 2 presents glyph list sorted over their occurrence frequency in the reference corpus.

Table 2

Cor

E

Cor

E

Cor

E

Cor

E

Cor

E

sign

%

%

sign

%

%

sign

%

%

sign

%

%

sign

%

%

6

9.9

9.2

700

2.6

2.4

730

1.3

1.6

99

0.7

0.7

28

0.3

1.0

200

8.4

9.3

4

2.6

3.0

901

1.3

1.5

76

0.7

0.4

71

0.3

0.8

10

6.6

8.7

41

2.5

3.1

95

1.2

0.8

45

0.7

0.7

999

0.3

0.0

400

5.9

4.8

660

2.3

2.4

44

1.2

1.5

60

0.7

0.2

91

0.3

0.3

1

5.6

7.1

22

2.0

4.2

7

1.0

0.9

67

0.6

0.5

25

0.2

0.1

3

5.4

2.9

9

1.8

1.7

34

1.0

1.3

53

0.5

0.9

15

0.2

0.0

2

4.1

3.2

63

1.7

4.1

69

0.9

0.4

52

0.5

0.0

720

0.2

0.2

62

4.0

2.3

240

1.6

1.1

48

0.8

0.6

74

0.5

0.2

530

0.1

0.0

380

3.2

4.8

5

1.6

1.6

70

0.8

0.6

16

0.4

0.3

14

0.1

0.0

61

3.2

2.0

8

1.6

0.7

59

0.8

0.3

27

0.4

0.2

280

3.0

3.8

66

1.4

0.5

50

0.7

0.7

38

0.3

0.3

28Text E in comparison features a much lower number of glyphs 3, 62, 61, 400, 8, 66 and 2 in comparison to that of the reference corpus. At the same time, the inscription of Keiti definitely “favors” the signs 63, 22, 10, 380, 1, 200, 280 and 28. The positive deviation of occurrence frequencies for these glyphs is understandable – especially for signs 380 and 28 entering the fragments “alpha/beta”. But even the presence of ten sequences “alpha” spotting adze glyph 63 does not explain its exceptional frequent occurrence in text E, which is more than twice higher than in the reference corpus. The actual problem is even deeper – the distribution of the sign 63 within Keiti inscription is very heterogeneous. Out of 47 occurrences, 42 appear on recto side and only 5 on verso side. Therefore, if we aim to describe the structure of the text E, we should also consider this anomalously high concentration of sign 63 literally “peppering” the short passage of Er.

Parallels between Keiti inscription and other inscribed artifacts

29As mentioned before, the text E contains at least 20 fragments shared with other rongorongo inscriptions (these are listed in Appendix 3 and referenced as F.1–F.20 in their occurrence order). It is important that in many cases, apart from sharing individual fragments, rongorongo inscriptions feature complete sequences thereof. Detailed comparison of these sequences (which by itself would deserve a separate publication) will pave the way for reconstructing the most stable “texts within the texts”, allowing to classify the texts that have survived in Rapa Nui script.

“Block sequences” and text structure

Calendar-like structures

30Wieczorek’s calendar interpretation of a fragment from text E does not convince me. I agree completely with the arguments of Horley (2011), who says that L-turned glyphs (which otherwise are known in their predominant R-turned forms) are quite frequent, so that the phenomena is not restricted to crescent signs. The uncommon glyph orientation can be found in many signs, and these can easily be found in the passages that preclude any reference to calendar topic. Moreover, the different orientation of the signs is noted in many parallel fragments. The examples shown by Horley can be easily expanded, but even those illustrated are enough to postulate that unusual sign orientation is not phonetically meaningful. Instead, it may have some complementary function, which is unknown to us – perhaps, that of “mini-texts” delimiter. Therefore, Wieczorek’s appeal that the word kokore (which is used in the names of several moon nights) means ‘without’ is far from convincing. In the fragment of text C which, according to the majority of the specialists, contains a lunar calendar of some kind, there are four fish signs in the first part of the calendar inscription (corresponding to the rising moon phase), which are oriented head-up – the usual way the fish glyphs are seen in rongorongo script (a special case of a fish sign incised on an edge of the tablet is discussed in Horley 2009: 255). Four fish signs for the second part of the calendar (denoting the setting moon) use the uncommon head-down orientation. This particular usage of fish glyphs constitutes the strongest difference between the delimiter groups of the calendar. In this example it is obvious that the uncommon orientation of the sign is connected to iconic (but not phonetic) function. Moreover, this distinct head orientation of fish sign, acting as a marker for two halves of lunar month offers the main supporting argument for identifying lines Ca6-Ca9 with a calendar inscription.

The name kokore

31Let us consider the problem of the word kokore, used for the names of several lunar nights, in conjunction with a potentially iconic interpretation of fish signs appearing in calendar inscription. Could it be that the orientation of fish signs characterizes only the notions for the rising / setting moon? It is accepted that the occurrence of the word kokore (translated as ‘without’) in the names of the nights of Eastern Polynesian languages can be explained by “namelessness” of these nights (as, for example, the name of the ring finger in Russian is “bezym’annyj”, which literally means “nameless”). Is it really so?

32In the classic paper by Stimson about the names of lunar nights recorded in Tahiti, there are unique data about the connection of each night with concepts of fishing and fertility. Stimson says that there are nights when fishing is easy (because the fish rises up to the surface), and there are nights when fishing is useless (the fish goes to the depth). In some nights it is acceptable to fish, but in some nights it is prohibited (and this attribute not necessarily coincides with fish accessibility: sometimes fishing is prohibited when the fish is readily available; in other days it is possible to go fishing, but the fish roams deeply and is difficult to catch). Let us generalize Stimson’s data (Table 3).

33According to this author, after the full moon and before the end of the month the fish goes to the deep. In the waxing moon, the fish goes up except for the 7th and 8th nights, when it disappears. Importantly, only these two nights in the waxing moon part of the calendar are called ore’ore (which corresponds to Rapanui kokore). In the waning moon part of the month, the name ore’ore appears for the nights 21-23, for which fish also disappears. Could it be that the word kokore ‘without’, usually interpreted as “nameless”, actually means “fishless”?

34As one can see, the data of the column “fishing outcome” does not coincide with the data in the column “fish” – the former has more frequent oscillations. There are many fishes in nights 13-16/17, and then in the end of the month (nights 25-28/29). The fish is scarce on the nights 8-12 (opening with two ore’ore), nights 18-23 (ending with three ore’ore nights), and in the last 30th night maari-mate,‘the death of the Moon’ (Stimson). The full moon, to the contrary, is very favorable for many things: the fish abounds, the planting is successful and large-eyed children are born.

35To my opinion, it is extremely important that fishing is tightly bound to the calendar, and the presence of fish signs “swimming” up and down in eight delimiter fragments of the Mamari calendar can be directly related with the concepts of fish and fishing.

36Speaking about the perspectives to find other calendars in rongorongo, Horley (2011) says: “Therefore, while it is completely reasonable to expect that Easter Island tablets may contain references to individual lunar nights / month names, it seems that the crescent series appearing in lines Ca6-9 of tablet Mamari form the only complete lunar calendar in the survived rongorongo corpus”. I think that this statement is open for discussion. Below I will show that there are several unusual segments that feature many similar characteristics with the famous calendar inscription of tablet Mamari.

Calendar in Ca5-9 and ... in Er1-Er3?

37First of all, let us study a schematic chart of lines Ca5-Ca9 (Fig. 1), which allows easy comparison with another potential calendar fragment. The question about the exact number of nights appearing in this list is still open for discussion. I think that one should count at least 31 nights (28+3) numbered in the figure. The ligature 1.6 serves as a “graphical frame” – the “forward” form 1.6 opens it and the “mirrored” form 6.1 closes the calendar.

38Letters x inFig. 1 denote sequences of glyphs not shown here for the sake of presentation clarity. The single letters x stands for of about 10 signs long.

39Now let us return to the tablet Keiti. I think that it is a much better candidate for the calendar than the sequences “alpha-alpha studied by Wieczorek. Curiously, this potential “calendar” fragment (Fig. 3) appears within the lines Er1-Er3 analyzed by Wieczorek.

The figure shows eight fragments “alpha”. Why only eight out of ten? Because only these fragments written in lines Er1-Er3 enclose the tight grouping of sign 63, which becomes considerably rare in the rest of the tablet. The figure numbers 31 occurrences of the “adze” signs, omitting those belonging to delimiter sequences “alpha” (in a similar way, we do not count the crescent signs appearing in delimiter sequences of Mamari calendar).

40Let us consider first the structural differences between Er1-Er3 and Ca5-Ca9:

  1. Text C features clearly ideographic calendar signs – the crescents are recognizably moon-like; text E uses completely different sign 63 and there is no straightforward evidence to consider it similar (or related) to the crescent glyph 41.

  2. Text C uses ideographic fish sign that changes its orientation for the moon-waxing and moon-waning part of the month; nothing similar appears in text E.

41Now, let us list structural similarities between these two fragments, which are too numerous to be ignored:

  1. In both cases we are dealing with very common rongorongo signs appearing in every survived text; these glyphs also appear in contexts that clearly do not have a slightest relation to calendar texts.

  2. At the same time, the occurrence of the signs 41 and 63 in aforementioned passages of Mamari and Keiti tablets is so high that it practically precludes the possibility of their phonetic reading.

  3. In both fragments the corresponding sign occurs 31 times.

  4. In both fragments the passage including 31 occurrences of the signs 41 and 63 is delimited with eight glyphic sequences – perhaps forming 8 parts of lunar month?

  5. (line Ca6) and (line Er1). They include two anthropomorphic signs showing head in profile accompanied with two crescent signs that form ABAB pattern. In the first delimiter of both texts the heads of the signs are looking towards each other, because the second glyph 300 is depicted in uncommon left-turned form. It is also important that both initial sequences are composed from three graphemes: the first glyph 41 is included into ligature and the second one is written separately.
  6. In both fragments the key sign that repeats itself 31 times enters into corresponding delimiters: there are two crescents 41 in each calendar delimiter of text C and one “adze” 63 closing every sequence alpha.

  7. ) after 15th occurrence of sign 63, which also may have an iconic interpretation as a full moon. In the 14th position there is glyph 16 depicting two halves of a whole (). In the text E the first night of the waxing moon is marked by an uncommon left turn of the sign 63. In the same way, the delimiter group following the full moon in Mamari calendar includes the bird sign 631 turned to the left (Guy 1990: 141).
  8. Text C features sign # 3 before the 23rd crescent. In our calendar-like structure of text E this glyph appears only once, following 22nd occurrence of the adze glyph!

  9. . In text E the signs for the nights 30-31 are written smaller in comparison with the others (an iconic way to approximate 29.5 nights of the lunar month?). After the 29th adze sign there is a rare grapheme 71.66.71 that will be discussed below.
  10. The eight fragments from Mamari calendar is divisible in two groups (halves of the month) by orientation of the fish glyph. In text E four first delimiter groups (“alpha-alpha”) distinguish themselves by orientation of the crescent sign. Namely in the 4th sequence “alpha-alpha” the first crescent is turned to the left, while in the three previous sequences it was turned to the right. Namely in the 8th sequence (and only there!) both signs 41 are turned to the left. Let us note that in the text C the last 31st crescent is the only one rotated to the left. The 9th sequence in Keiti inscription is special due to the unique position of sign 380 – it is the only example where this sign opens block “alpha-alpha – alpha-beta” instead of closing it.

42I am completely aware that some of the discussed properties may be occasional. However, in analyzing them in totality, one is convinced to be dealing with ordered structures that have to be considered in detail if one pretends to describe the structure of the text E, as it is claimed by Melka (2008) and Wieczorek (2011).

Calendar structure in Ev7?

43Let us consider first a very important question (also discussed in other rongorongo papers appearing in this issue): how many nights should be in the lunar calendar? From an astronomical point of view the answer is obvious and known for centuries: the synodic month (connected to the changes of visible phases of the moon) is 29.5 nights long. For the cultural calendar there is no (and can not be) any definitive answer. Moreover, cultural calendars may become completely unrelated to the astronomy. Sometimes, this “independence” is one of the main purposes of calendar creation – as it was in French revolution calendar with its year composed by ten months, each month composed by ten days, each day by ten hours; another example is the 1919 Soviet calendar with its five-day weeks. Leaving aside these exaggerated cases of radical reformations one will nevertheless find out that the most commonly-implemented chronology includes a dialogue between the cultural and natural calendars. The possible month lengths are:

4428 nights is rather a cultural interpretation than an astronomic one. The underlying idea concerns the possibilities to obtain two equal halves of the month (14 nights), which can be further subdivided in half to get the “classical” week of 7 nights.

4529 nights in the month is closer to astronomical reality. It may have place in some cultures, but each culture uses its own way to round the lunar month to an integer number of nights.

4630 nights, perhaps, is the ideal mixture of astronomical and cultural interpretations. The number is good for a realistic observable moon cycle and also fits well into cultural requirements – it can be divided into two halves and three equal parts of ten nights each, which looks so natural to human beings using decimal counting system based on ten fingers on both hands. The tripartite month with ten-day weeks is known in China and Polynesia.

ststst()stMamari

4732-night month, alongside with 28-night month, is very attractive due to its high divisibility into halves and quarters. Such month will have four weeks of 8 nights each, which is quite wide-spread in the calendar cycles of the World. Only this variant is subdivisible into eighths composed with an integer number of nights.

48Polynesian cultures use all the aforementioned possibilities. In the most complete collection of the Tahitian night lists (Roberts, Weko and Clarke 2006) one can find months ranging from 28 to 32 nights. We know that in Rapanui culture there were also different night nomenclatures that are already profusely discussed in the literature (see, for example Horley (2011)).

49Therefore, looking for calendar structures in rongorongo script, one should pay attention to the anomalously tight grouping of signs in quantities ranging from 28 to 32. It should be desirable to have this sequence of signs divided in half by a specific marker. Most remarkably, one should also expect to see the sequence “framed” with iconic markers denoting the beginning and ending of the calendar. All these properties are characteristic not only of the renowned Mamari calendar Ca5-Ca9, but of Keiti “calendar” Er1-Er3 as well.

Mamari Keiti KeitiMamari

Calendar structure in Ca9-Ca12?

50The repetitive sequence Ca9-Ca11 (Fig. 5) also features structural peculiarities that can be interpreted as a calendar structure:

  1. interpreted as a closing “frame” of Ca5-Ca9 calendar (Fig. 1).
  2. There are 31 occurrences of sign 2 (in Barthel’s nomenclature) here. This number can be related to the number of nights in lunar month. Such high occurrence of the same sign in a short fragment precludes the possibility of their phonetic reading. This situation is highly reminiscent of the discussed structures in texts C and E.

  3. The text shown in Fig. 5 can be divided into six delimiters (marked with frames in Fig. 5), which are distributed evenly in two halves of would-be “lunar month”, similarly to that in the calendar Ca5-Ca9: there are three delimiters for the nights 1-15 and the other three for the nights 16-31.

  4. The last delimiter appears after the sign corresponding to the 28th night, splitting the month into 28+3 nights, which is also the case with the calendar structures in the texts C and E.

  5. In the last sixth delimiter – and only there – the hands of the lizard glyph point downwards and the head of the bird looks to the left, marking the end of a four-week cycle (28 nights).

  6. The first delimiter features the separate writing of signs 10 and 70; in five further delimiters these glyphs are written together.

  7. appears 10 times in the figure, if we count glyphs and to be its allographic forms (in the similar manner, text Er features 10 occurrences of the sequence “beta” (a part of sequence “alpha”) and 10 occurrences of sequence alpha.
  8. The final sign 2 (corresponding to 31st night) has an uncommon “profile” depiction (sign 20 in Barthel’s nomenclature). The dozens of regular correspondences in the parallel texts allow to firmly associate signs 2 and 20 as allographs. Therefore, this “turn” of the sign can be considered as a specific beginning/ending marker for a mini-text – in our case, it is a sign standing for the “last night”.

  9. Immediately after the 15th occurrence of sign 2, closing the first half of the month, there are several “turtle” signs 280. In a petroglyph panel located near Ahu Ra’ai (Horley 2011, Fig. 5) the turtle divides the lunar month in two halves, similarly to the discussed fragment.

  10. There are three turtles in the aforementioned petroglyph, superimposed over a sequence of crescents (the central turtle divides the month in two halves, plus two turtles marking the beginning and end of the month corresponding to the 1st and 28th-30th nights). In our Fig. 5, there are also three turtle signs.

  11. The position of only one (central) turtle directly corresponds to its position in the petroglyph. Two other turtles take other positions, but they also fit well into the logics of a calendar cycle. Combination of the signs 2 and 280 forms a special graphical mark of the third week enclosing the full moon (nights 15-21), where the sign 2 acts as a “frame” embracing the signs of the “turtle week”:

  12. Similarly to the third week marked with three signs 280, there are three signs 4 marking the fourth week, nights 22-28. Apart from this usage, the sign 4 appears nowhere else in the sequence illustrated in Fig. 5.

  13. . These three “diamonds” may have smaller “beads” attached, most frequently to its right side or to the both sides (and, respectively); in fewer cases, the beads are attached to the left side only (). The signs 2 belonging to the first half of lunar month (nights 1-15) are sporting 35 “beads”, 20 of which are attached to their left side and 15 to their right side. The signs belonging to the second half of the month feature 14 “beads” (8 to the left and 6 to the right).
  14. The signs depicting the closing nights (29th to 31st) of the lunar month are drawn shorter than the rest of the “lozenge” glyphs 2; they are composed only of two “diamonds” in place of the usual three. Similarly, in the line Er3 two last “adze” signs – 30th and 31st in a sequence – are drawn smaller than the preceding signs 63. This can be interpreted as intended to bring closer the cultural month composed of 31 to the astronomical month (29.5 nights); alternatively, it can be interpreted as a graphical representation of the “moonless” nights.

  15. The pre-final 30th sign 2 is ligatured with a crescent glyph 41, which, except for the “official” calendar Ca5-Ca9, does not appear at all on this side of the tablet. Could it be a hint that the sequence of signs 2 should be treated in the frame of the calendar cycle, similarly to a sequence of crescents 41 in the calendar Ca5-Ca9?

51Even if many of the highlighted structural characteristics are purely coincidental, such a peculiar structured fragment directly following the “official” calendar requires much attention.

52The similar high concentration of signs 2 can be found on tablet Tahua, line Ab6 (Fig. 6). Counting the “diamonds” forming the “lozenges”, one will obtain the number 2+28=30. The fragment opens with five signs 70 (Guy 2006: 60), which ironically can be interpreted as depictions of the moon.

Calendar structure in Gr / K ?

verso

53Most frequently these delimiters are seen in the texts Gr/K, illustrated side-by-side in Appendix 4. It is easy to see that the total number of delimiters 380.1.3 is 31, which is the number of occurrences of the crescent sign 41 in Mamari calendar, the number of “adze” glyphs 63 in Er1-Er3 and the number of “lozenge” sign 2 in Ca9-Ca11. Therefore, one has considerable bases to assume that the passages marked with 31 delimiters 380.1(.3) may represent, for example, the names of the lunar nights or the omens connected with them, similar to those presented by Stimson. Moreover, we know that sign 380 is directly related to the ideogram of the full moon in the “official” calendar Ca5-Ca9 (see Barthel 1958: 245, Guy 1990: 136, and Horley 2011: Fig. 9). These observations motivate ourselves to study the sequence Gr/K in detail, because it may offer some hints for the phonetic reading (which is definitely not the case with the multiple inline repetitions of the same sign).

54Let us consider the passages presented in Appendix 4. Before the first “block” following the delimiter 380.1 in texts Gr/K one can see the fragment F.12, one of the most common parallel sequences in the Easter Island script. As it appears in the text E, the reader may find it in Appendix 3. Table 4 gives only the beginning of this fragment, which is sufficient to see the correspondence of the signs Gr/K standing before the 1st block.

reimiro afterfirstfirstopens a first

55In conclusion, one can say that the long block sequences delimited with the glyph combination 380.1 or its analogs opens in the same way, independently on the contents of these chains; they are introduced by the same sign sequence listed as F.12 in Appendix 3. Moreover, the inverse is also true: with a rare exception (the case of Ev3) the appearance of the fragment F.12 suggests that immediately after it we will find a long chain of signs delimited with combination 380.1 or variations thereof.

Mamari thth

56Finally, the new 28th block includes turtle sign 280 – similarly to the moon-related petroglyph (see Horley 2011, Figs. 5 and 6). Three turtles (which in accordance to petroglyph are quite a pronounced hint to a calendar!) appear in the 20th block; there are turtles in 18th block as well – which are highly reminiscent of the turtle signs in lines Ca9-Ca11 standing by the 20th, 17th and 15th glyphs of a potential “calendar” based on sign 2.

Figure 7: Fragment F.9

Figure 7: Fragment F.9
open with the same fragmentstndrd

57The difficulty of such comparison (and especially of the presentation of results) is caused by the fact that these sequences should be compared by taking into account the block sequences of the other inscriptions, in the first place including those from the texts N, Ca, Cb, Ab. To prepare the reader for a better understanding of this multi-layer comparative analysis, let us build the discussion in the following order: 1) comparison of the inscriptions E and N; 2) comparison of E, N with G/K ; 3) comparison of the block sequences C with those from E, N, G/K as well as with block sequences from the text A.

The block sequences in texts E and N

58The sequence of 23 blocks written on the verso side of tablet Keiti contains 7 blocks from the text tablet N in the same order (Pozdniakov 1996, Horley 2010). In several cases two blocks of Ev correspond to a single block in text N, which means that the delimiting group 380.1.52 is sometimes omitted in the inscription of the Small Vienna tablet. Comparing block sequences with text E, one can conclude that text N actually has 10 blocks in place of seven explicitly marked with delimiters. To address the “compound” block in text N I will use numbers and letters. For example, the initial part of the 3rd block of text N corresponds to the 11th block in Ev, while the final part of the same block corresponds to the 12th block in Keiti text. Due to this, the 3rd block of text N should be split in two – blocks 3A and 3B, respectively.

59Let us compare the parallel sequence of blocks in the texts E and N. These open with the same fragment F.9 (see Appendix 3), which in addition to Ev and N also appears in line Cb3 (Figure 7) inside the delimited sequence of blocks, occupying the 4th position there.

60The figure emphasizes the importance of parallel fragments in the comparative structural study of the texts. The frames shown for lines Ev2 and Na2 denote the first occurrence of the delimiter group, so that one may think that the glyphs following it represent the first block in the sequence. However, the real situation is different since the sequence starts with fragment F.9 preceding the first delimiter, and this detail was impossible to determine without a comparison with parallels in text C. Thus, all three sequences include the same fragment: it appears in full form in Ev2 and Na2 and is given in truncated form (beginning only) in line Cb3.

61Actually, fragment F.9 could have been expanded with the 2nd block that follows the inscription illustrated in Figure 7:

62Just after the fragment F.9 in text Ev one can find the fragment F.10. In the Small Vienna tablet the same block appears on the 6th position (Na5).

63Immediately after the 6th block in text N (and 3rd block in text N) one finds the composite block named 7A, which corresponds to the 5th block in the inscription of Keiti. However, between them appears the 4th block of the text E, the “formal” place of which, as we will see from the texts Gr/K, should be “deeper” in the block sequence. Thus, removing block 4 from the chain, one obtains the identical sequence of blocks in the texts E and N: blocks 3-5 (Keiti) corresponding to blocks 6-7A (Small Vienna tablet). The inscription of Keiti continues with blocks 6-8, two of which belongs to the fragment F.12. Text N lacks the corresponding parallels. Therefore, the initial group of eight blocks from the text E are parallel to two pairs of blocks in the inscription of N (1st and 2nd blocks in N correspond to the 1st and 2nd blocks in E; 6th and 7thA in text N correspond to the 3rd and 5th blocks in E). The inscription of Gr/K does not show parallels to this part of the text, except for the important fragment F.12 (1st block in Gr/K and 6thblock in the text E).

Comparison of block chains in text E and Gr/K

64Implementing new numeration Gr/K to include fragment F.12 as an initial block, one obtains 32 blocks in this sequence. Comparison of blocks Gr/K with those of Ev and N leads to very interesting results:

65None of the initial blocks (occupying 1st to 16th positions) in sequence Gr/K appears in the inscriptions E and N, except for fragment F.12.

66The majority of the blocks from the second half of the sequence Gr/K (blocks 17-32) have the corresponding passages in Ev, which, moreover, appear in the same order. The inscriptions are similar to such a degree that one can suggest that we are dealing with the same text (Figure 8).

67The full validation of each correspondence is too cumbersome to be presented in this paper because it includes analysis of the parallel fragments from other texts and many other factors. However, I hope that the reader will easily spot various key signs in the compared blocks. The abridged scheme of Gr/K block sequences compared to the text E is given in Table 5.

68This particular distribution of the blocks stimulates further detailed analysis of block sequences delimited with glyph combination 380.1 (and variations thereof) due to their potential relation to the calendar structures. It can be, for example, that the text Gr/K presents two halves of the lunar month, while text Ev features only one half of it.

69Considering 32 blocks in the lists Gr/K (Figure 8, Appendix 4), I would like to make another comment about the length of cultural months. In his analysis of moon petroglyph from Ahu Ra’ai, Horley (2011, fig. 5) counts 30 nights, omitting four vertical lines in the upper central part of the figure, but including an extra night 16 that is absent from petroglyph tracing (though it may be present in the original carving). He also assumes that the crescent for the 28th night coincides with the outlines of the turtle, which is a point open to discussion. Under these assumptions, Horley obtains 30 nights characteristic of an astronomical month. However, counting all the lines that do appear in the tracing of the petroglyph, one obtains 32 night marks – a number that permits easy subdivision in halves, quarters and eighths. In this case, the right turtle splits the calendar into 30+2 nights, representing a good “junction” between astronomical and cultural calendars.

Block sequences in text Ca

70Side Ca of Mamari tablet also contains glyphic sequences delimited with nine combinations 380.1. Except for initial and final combinations (located in Ca1 and Ca14), seven combinations are clustered in lines Ca2-Ca3. In relation to the calendar studies, block sequences in text Ca are important because they appear on the same side as the “official” calendar (lines Ca5-Ca9) and a “possible calendar” based on tightly clustered signs 2 (lines Ca9-Ca12). In a certain sense, the delimited sequences 380.1 frame these calendar structures.

71Curiously, four initial blocks of the sequence have undeniable parallels with the similar block sequence in text A (line Ab4), which uses slightly different delimiters (Figure 9). Namely the parallels between Ca2-3 and Ab4 prove that the combination 1.3 is a “modified” delimiter 1.52, so that the 3rd and the 5th blocks are definitely separate entries. Starting from the 6th block, the sequence in line Ab4 does not have parallels in Ca2-Ca3.

72The second part of block sequence in Ca has parallels with text Gr, also in its second half, where one observes the systematic parallels of Gr/K and Ev/N (Figure 10).

ththth

73The sequences delimited with sign combination 1.5.9 require much attention; however, it is better to leave a detailed discussion of them for another occasion.

Results and discussion

Structure of the text E

74Basing ourselves on the aforementioned discussion, the structure of the text E can be presented in the following compact way (Figure 11).

x xx xins

75As one can see from Figure 11, the inscription of the tablet Keiti starts with a segment (lines Er1-3) that has a similar structure to that of the acknowledged calendar Ca5-9. Line Er4 is practically devoid from original passages – it has several sequences “alpha” and “beta” and two parallel fragments (F.3 and F.4). On the contrary, the following lines Er5-9 are practically unparalleled in the other texts, which is denoted by numerous letters x in the figure. Namely we find here almost all segments “beta” and the final segment “alpha-gamma”. The famous fragment F.7, seen at the beginning of the parallel texts P/H/Q and calendar inscription C, starts in Er9 and continues to Ev1. This observation proves that recto and verso sides were assigned correctly by Barthel, at the same time presenting a counter-argument to the suggestion by Melka (2008) and Wieczirek (2011) that each side of the tablet Keiti was inscribed with an independent text. The same fragment F.7 is also partially reproduced in the end of line Ev1. The sequence with delimiters 380.1 (also possibly related to the calendar-like structure) follows in lines Ev2-9. It is important that contents and also order of the blocks have multiple parallels in other texts. Line Ev6 contains five parallel fragments. Line Ev7 features a structured inscription that may be related to calendar. The final line Ev8 is closed with tree clustered glyphs 19 (in Barthel’s nomenclature), which may possibly function as iconic signs representing the halves of a lunar month (Table 5).

Potential calendars and the problem of phonetic reading

76I would like to stress that this paper is not intended to sell a “boatload” of new calendars from Easter Island to the reader. The author is completely aware that the most difficult thing in hypotheses like these is to stop at the appropriate moment. Each proposed interpretation of rongorongo glyphs influences our understanding of the script and possibility of its decipherment. The accurate understanding and rebuttal of hastily-made and poorly-based conclusions may cost much time and effort for the future researchers. Allowing ourselves to step away from strict phonetic reading (e.g., claiming that the reading of a glyph varies with context) and to depart from sign catalogue founded on comparative study of multiple parallel fragments (e.g., counting the individual “diamonds” and “beads” composing Barthel’s glyph 2) means that we are actually trying to deny that rongorongo represents a written system – the solid fact that was already firmly proven. Thus, if we hold that rongorongo signs correspond to the syllables of a spoken language and hence are phonetic (namely this conclusion follows from the statistical analysis), we cannot interpret the same signs (and moreover, their parts) as ideograms. Namely because of this we avoided the passages with questionable chances of phonetic reading during our analysis and identification of the potential syllabic signs (Pozdniakov 1996, Pozdniakov-Pozdniakov 2007). However, the presence of such segments is obvious – for example, the “official calendar” in lines Ca5-9 – and one only wonders how this (or similar) passage could be ever read and translated (e.g., lines Er1-3 “read” by de Laat and Fedorova).

77When we say that certain glyphs are not phoneticor not only phonetic in specific contexts, we refer to the following cases:

  1. Extremely high concentration of the same sign in the short fragment of the text (as in the examples of Ca5-9, Ca9-12, Er1-3).

  2. Delimited sequence of blocks (e.g., with a separator 380.1). These delimiters may be devoid of phonetic reading, rather acting as determinatives, markers of the proper names, lunar months, toponyms and other specific words.

  3. Conspicuous ordering and “sign mirroring” in the sequence. Horley (2011) suggests that these “symmetric glyph arrangement, considerably appreciated and employed by the rongorongo men (were used) to improve the visual appearance of their texts”, giving an example from the line Bv3 (ibid., Fig. 11). I would like to emphasize that symmetric / mirrored placement of glyphs / alloglyphs is far from being marginal exotic phenomena – the surviving rongorongo corpus counts dozens of such examples. Let us consider a single case related to the block sequences G/K = E = N (Figure 7), focusing attention to the block from text N (Figure 12). It is quite probable that the scribe created the elaborated block in line Na3 being motivated by a necessity (functional or aesthetic) to extend some basic structure to a mirror-like one: the bird sign (600) is set in the center of the composition, surrounded by three pairs of mirrored graphemes according to the pattern A B C D C B A. Such “graphical spoonerisms”, most possibly, do not have any phonetic basis. Importantly, the “hand and stick” ligature 1.6 and 6.1 “embracing” the sequence are the same signs used to create a graphical frame for the Mamari calendar in lines Ca5-9.

  4. The previous issue is closely related to the problem of right- and left-facing orientation of the signs and their elements, as well as their up-down “flipping”. Let us sum up the main aspects of the problem. The regular use of “uncommon” sign orientation (especially in adjacent or almost-neighboring graphic forms) precludes the conclusion about its phonetic meaning – which becomes the main counterargument for the hypothesis proposed by Wieczirek. However, the question remains: what could be the possible function of sign orientation, if this function is not phonetic? There are dozens of examples bringing us to the conclusion that uncommon orientation of the signs (which is most frequently manifested by glyphs facing to the left) most probably forms a graphic frame to mark the end of a mini-text and to separate it from the next meaningful passage. If this hypothesis is true, sign orientation might function as a clever analog of punctuation signs (a coma or a period) in rongorongo script. Perhaps, there could be other explanations, but I would rather refrain from voicing them at the moment. In any case, the systematic analysis of this particular glyph use (that should be addressed in depth in a separate publication) may significantly improve our understanding of rongorongo script, the function of allographs and the bases of the sign catalogue.

78Finally, I would like to stress that there are no obvious reasons to claim that the tablet Keiti (text E) is the most interesting or, let us say, most promising for the decipherment in comparison with other rongorongo inscriptions. Therefore, the appearance of so many papers dedicated to this particular text can be interpreted as a positive signal marking the beginning of a new epoch in the studies of Easter Island script – when the times of individual “enlightening” and “revelations” are over, when the specialists finally started to talk with each other, sometimes even coming to an agreement.

I would like to thank Paul Horley for his useful comments and critics, as well as for the tracings of rongorongo glyphs shown in the figures of this paper.

Haut de page

Bibliographie

Barthel Thomas, 1958. Grundlagen zur Entzifferung der Osterinselschrift, Hamburg, Cram, de Gruyter.

Butinov Nikolay A. and Yuri V. Knorozov, 1957. Preliminary report on the study of the written language of Easter Island, Journal of the Polynesian Society 66, pp. 5–17.

Fedorova Irina K., 2001. “Talking boards” from Easter Island: deciphering, reading, translation. St. Petersburg: Kunstkamera)

Guy Jacques, 1990. The lunar calendar of tablet Mamari, Journal de la Société des Océanistes 91, pp. 135-149.

—, 2006. General properties of the rongorongo writing, Rapa Nui Journal 20, pp. 53-66.

Harrison J. Park, 1874, The hieroglyphs of Easter Island, Journal of the Royal Anthropological Institute of Great Britain and Ireland 3,pp. 370-383.

Horley Paul, 2005. Allographic variations and statistical analysis of the Rongorongo script, Rapa Nui Journal 19, pp. 107-116.

—, 2007. Structural analysis of rongorongo script, Rapa Nui Journal 21 (1), pp. 25-32.

—, 2010. Rongorongo tablet Keiti, Rapa Nui Journal 24 (1), pp. 45-56.

—, 2011. Lunar calendar in rongorongo texts and rock art of Easter Island (current issue of the Journal).

Laat M. (de) , 2009. Words out of wood: proposals dor the decipherment of the Easter Island script, Eburon Delft.

Melka Tomi S., 2008. Structural observations regarding rongorongo tablet ‘Keiti’, Cryptologia 32, 2, pp. 155-179.

Pozdniakov Konstantin, 1996. Les bases du déchiffrement de l'écriture de l'île de Pâques, Journal de la Société des Océanistes 103 (2), pp. 289-303.

Pozdniakov Igor and Konstantin Pozdniakov, 2007. Rapanui writing and the Rapanui language: preliminary results of a statistical analysis, Forum for Anthropology and Culture 3, pp. 3-36.

Mere Robert, Frank Weko, Liliana Clarke. Maramataka: the Maori Moon Calendar. Research Report 283. August 2006. Matauranga Maori and Bio Protection Research Team. National Centre for Advanced Bio-Protection Technologies. Lincoln University, Canterbury, New Zealand. http://www.lincoln.ac.nz/Documents/2333_RR283_s6506.pdf

Stimson J. Frank, 1928. Tahitian names for the nights of the moon. Journal of the Polynesian Society 37(147), pp. 326-337.

Wieczorek Rafal M., 2011. Astronomical content in rongorongo tablet Keiti (current issue of the Journal).

Haut de page

Annexe

Appendix 1. – Catalogue of rongorongo signs (Pozdniakov-Pozdniakov 2007)

Appendix 2 and 3 (see pdf)

Appendix 4. – Sequences of blocks delimited with ligature 380.1 in parallel texts Gr and K

A word of dedication

Igor Konstantinovich POZDNIAKOV (1/06/1927-16/01/2010)

Recently I lost my father and coauthor, Igor Konstantinovich Pozdniakov. He was working on decipherment of Easter Island script for more than a quarter of a century.

As a boy, he survived by a miracle the Siege of Leningrad (1941–1943). After the Second World War Igor became a naval officer. In 1963 he obtained Ph.D. diploma on phase measurements at the Research Institute for Metrology and some years after he became a director of Research Institute for Scientific Equipment.

In 1984 my father started active research on decipherment of the Easter Island script, and this was the start of our joint work.

At that time we had a Soviet computer «Iskra» with 16 kilobytes of RAM. On this computer I.K. Pozdniakov started to study the distribution of signs in the rongorongo texts. It is worth saying that in those times the computers were completely different from the multi-processor workstations of today. For example, to save the files one should use a common tape recorder, which was painfully slow and could only manage small files. As computers were gradually improving, my father developed hundreds of files addressing a wide spectrum of statistical issues that could be useful for the decipherment. I think that the true importance of his studies will be understood only when the Easter Island script will be deciphered. I am sure that the decipherment will show that my father was many times close to the correct solution, or even discovered correct readings for many signs, but did not have a chance to complete his monumental task.

After publication of our joint paper, we heard many critical comments saying something like: «if the statistical characteristics of the rongorongo signs and the syllables of the Rapanui language coincide so well, why did not the authors come to the decipherment for such a long time»? My father used to answer this question: «Finally, it is not so bad to remain the only ones who still have NOT deciphered rongorongo». Fortunately, this is not quite correct: there are other scholars who delve into meticulous and thorough work on rongorongo and do not wait for any cosmic revelations – these are, for example, Jacques Guy, Paul Horley and some other specialists.

My father was much annoyed by the comment repeated many times by Richard Sproat: “Pozdniakov would appear to have merely re-discovered the Zipf’s law (well, not quite since the populations of syllables are too small for the curves to be truly Zipfian)”. This misleading critique can even be found in  the Wikipedia: «the results from the frequency distributions are nothing more than an effect of the Zipf’s law, and furthermore that neither rongorongo nor the old texts were representative of the Rapanui language, so that a comparison between them is unlikely to be enlightening». I.K.Pozdniakov had several good replies for a rebuttal of Sproat’s critique:

1) The pronounced similarity of usage frequency distribution of the rongorongo signs (according to our catalogue) and the Rapanui syllables. As Sproat briefly mentions, this distribution has nothing to do with the Zipf’s law at all. Look how enlightening the comparison is: the most frequent hand sign 6 covers about 10% of the Easter Island texts, like the most frequent syllable A of the Rapanui language. The similar distribution curves for signs and syllables are actually quite sufficient. These two curves would not coincide if we tried to compare rongorongo with the syllables of any unrelated languages – let it be Russian, Wolof or Abkhazian. For example, Russian allows combinations of several consonants; as a result, the number of possible syllables is so great that none of them could have the occurrence of 10%. Thus, if we plot an (analogous) distribution curve for Russian syllables, it will be far lower and flatter than the distribution of the Rapanui syllables. Of course we checked this for different languages before the publication. Sproat’s comparison of the occurrence distribution of the syllables from the short Rapanui text Apai with that of English letters forming first 12,000 words of Genesis is pointless, because comparing a syllabary to an alphabet is senseless. The curve characterizing English letters obviously goes below the curve for the Rapanui syllables in the plot supplied by Sproat, showing that his “reference” English system with capital and small letters and punctuation marks contains a larger number of elements, which is reflected by their lower usage frequencies.  I think that the similarity of the distribution curves for the rongorongo signs and the Rapanui syllables proves that the structure of the Easter Island script – with its phonologic glyph set and phonotactic rules – is remarkably similar to that of the East Polynesian languages. It is also important that the hypothesis about the predominantly syllabic nature of rongorongo clearly explains that glyph ligatures represent multisyllabic words, and that spaces between the glyphs actually separate these words. If we treat rongorongo as a logographic system, we will not be able to explain the function of spaces.

2) The Zipf’s law is completely unrelated to other statistical properties that had been studied in detail by Igor Pozdniakov. All these statistical properties also show good correlation between rongorongo and Rapanui – similar distribution of signs (initial, final, median), similar occurrence frequencies of independent signs, syllables and reduplicated ABAB structures (ro-ngo-ro-ngo, a-ku-a-ku), etc.

We lost a prominent and strongly self-disciplined scholar in the field of rongorongo. Igor Pozdniakov had a great gift to generate new brilliant hypotheses and an even greater gift – to discard these hypotheses if they could not be confirmed (structurally or statistically):  one of his values was the strict discipline of thought. Hundreds of his files still await a detailed analysis and will definitely contribute to the decipherment of rongorongo.

Haut de page

Documents annexes

Haut de page

Table des illustrations

Titre Table 1
URL http://jso.revues.org/docannexe/image/6371/img-1.png
Fichier image/png, 87k
Titre (Barthel):
URL http://jso.revues.org/docannexe/image/6371/img-2.png
Fichier image/png, 22k
Titre (Pozdniakov-Pozdniakov):
URL http://jso.revues.org/docannexe/image/6371/img-3.png
Fichier image/png, 8,0k
URL http://jso.revues.org/docannexe/image/6371/img-4.png
Fichier image/png, 17k
URL http://jso.revues.org/docannexe/image/6371/img-5.png
Fichier image/png, 1,6k
URL http://jso.revues.org/docannexe/image/6371/img-6.png
Fichier image/png, 1,4k
URL http://jso.revues.org/docannexe/image/6371/img-7.png
Fichier image/png, 3,6k
URL http://jso.revues.org/docannexe/image/6371/img-8.png
Fichier image/png, 3,1k
URL http://jso.revues.org/docannexe/image/6371/img-9.png
Fichier image/png, 564 octets
URL http://jso.revues.org/docannexe/image/6371/img-10.png
Fichier image/png, 825 octets
URL http://jso.revues.org/docannexe/image/6371/img-11.png
Fichier image/png, 576 octets
URL http://jso.revues.org/docannexe/image/6371/img-12.png
Fichier image/png, 5,4k
URL http://jso.revues.org/docannexe/image/6371/img-13.png
Fichier image/png, 5,9k
URL http://jso.revues.org/docannexe/image/6371/img-14.png
Fichier image/png, 2,9k
URL http://jso.revues.org/docannexe/image/6371/img-15.png
Fichier image/png, 6,4k
URL http://jso.revues.org/docannexe/image/6371/img-16.png
Fichier image/png, 4,2k
URL http://jso.revues.org/docannexe/image/6371/img-17.png
Fichier image/png, 4,4k
URL http://jso.revues.org/docannexe/image/6371/img-18.png
Fichier image/png, 4,8k
URL http://jso.revues.org/docannexe/image/6371/img-19.png
Fichier image/png, 5,0k
URL http://jso.revues.org/docannexe/image/6371/img-20.png
Fichier image/png, 3,3k
Titre Table 3
URL http://jso.revues.org/docannexe/image/6371/img-21.png
Fichier image/png, 27k
Titre Figure 1
URL http://jso.revues.org/docannexe/image/6371/img-22.png
Fichier image/png, 454k
URL http://jso.revues.org/docannexe/image/6371/img-23.png
Fichier image/png, 1,3k
Titre Figure 2
URL http://jso.revues.org/docannexe/image/6371/img-24.jpg
Fichier image/jpeg, 560k
Titre Figure 3
URL http://jso.revues.org/docannexe/image/6371/img-25.png
Fichier image/png, 911k
URL http://jso.revues.org/docannexe/image/6371/img-26.png
Fichier image/png, 6,8k
URL http://jso.revues.org/docannexe/image/6371/img-27.png
Fichier image/png, 6,4k
URL http://jso.revues.org/docannexe/image/6371/img-28.png
Fichier image/png, 3,5k
URL http://jso.revues.org/docannexe/image/6371/img-29.png
Fichier image/png, 4,0k
URL http://jso.revues.org/docannexe/image/6371/img-30.png
Fichier image/png, 3,6k
URL http://jso.revues.org/docannexe/image/6371/img-31.png
Fichier image/png, 6,3k
URL http://jso.revues.org/docannexe/image/6371/img-32.png
Fichier image/png, 3,6k
URL http://jso.revues.org/docannexe/image/6371/img-33.png
Fichier image/png, 1,4k
URL http://jso.revues.org/docannexe/image/6371/img-34.png
Fichier image/png, 1,6k
URL http://jso.revues.org/docannexe/image/6371/img-35.png
Fichier image/png, 3,6k
Titre Figure 4
URL http://jso.revues.org/docannexe/image/6371/img-36.png
Fichier image/png, 228k
URL http://jso.revues.org/docannexe/image/6371/img-37.png
Fichier image/png, 3,6k
URL http://jso.revues.org/docannexe/image/6371/img-38.png
Fichier image/png, 3,3k
URL http://jso.revues.org/docannexe/image/6371/img-39.png
Fichier image/png, 43k
URL http://jso.revues.org/docannexe/image/6371/img-40.png
Fichier image/png, 3,3k
URL http://jso.revues.org/docannexe/image/6371/img-41.png
Fichier image/png, 3,4k
URL http://jso.revues.org/docannexe/image/6371/img-42.png
Fichier image/png, 2,2k
URL http://jso.revues.org/docannexe/image/6371/img-43.png
Fichier image/png, 2,6k
URL http://jso.revues.org/docannexe/image/6371/img-44.jpg
Fichier image/jpeg, 20k
URL http://jso.revues.org/docannexe/image/6371/img-45.png
Fichier image/png, 3,5k
URL http://jso.revues.org/docannexe/image/6371/img-46.png
Fichier image/png, 3,5k
URL http://jso.revues.org/docannexe/image/6371/img-47.png
Fichier image/png, 114k
URL http://jso.revues.org/docannexe/image/6371/img-48.jpg
Fichier image/jpeg, 16k
URL http://jso.revues.org/docannexe/image/6371/img-49.png
Fichier image/png, 2,1k
URL http://jso.revues.org/docannexe/image/6371/img-50.png
Fichier image/png, 2,4k
URL http://jso.revues.org/docannexe/image/6371/img-51.png
Fichier image/png, 2,0k
Titre Figure 5
URL http://jso.revues.org/docannexe/image/6371/img-52.png
Fichier image/png, 878k
Titre Figure 6
URL http://jso.revues.org/docannexe/image/6371/img-53.png
Fichier image/png, 214k
URL http://jso.revues.org/docannexe/image/6371/img-54.png
Fichier image/png, 2,9k
URL http://jso.revues.org/docannexe/image/6371/img-55.png
Fichier image/png, 3,5k
URL http://jso.revues.org/docannexe/image/6371/img-56.png
Fichier image/png, 4,0k
URL http://jso.revues.org/docannexe/image/6371/img-57.png
Fichier image/png, 2,4k
URL http://jso.revues.org/docannexe/image/6371/img-58.png
Fichier image/png, 2,0k
Titre Table 4
URL http://jso.revues.org/docannexe/image/6371/img-59.png
Fichier image/png, 43k
URL http://jso.revues.org/docannexe/image/6371/img-60.png
Fichier image/png, 1,9k
URL http://jso.revues.org/docannexe/image/6371/img-61.png
Fichier image/png, 2,5k
URL http://jso.revues.org/docannexe/image/6371/img-62.png
Fichier image/png, 1,6k
URL http://jso.revues.org/docannexe/image/6371/img-63.png
Fichier image/png, 2,3k
URL http://jso.revues.org/docannexe/image/6371/img-64.png
Fichier image/png, 4,0k
URL http://jso.revues.org/docannexe/image/6371/img-65.png
Fichier image/png, 3,7k
URL http://jso.revues.org/docannexe/image/6371/img-66.png
Fichier image/png, 1,8k
URL http://jso.revues.org/docannexe/image/6371/img-67.png
Fichier image/png, 7,8k
URL http://jso.revues.org/docannexe/image/6371/img-68.png
Fichier image/png, 3,2k
Titre Figure 7: Fragment F.9
URL http://jso.revues.org/docannexe/image/6371/img-69.png
Fichier image/png, 280k
URL http://jso.revues.org/docannexe/image/6371/img-70.png
Fichier image/png, 4,0k
URL http://jso.revues.org/docannexe/image/6371/img-71.png
Fichier image/png, 2,1k
URL http://jso.revues.org/docannexe/image/6371/img-72.png
Fichier image/png, 3,5k
URL http://jso.revues.org/docannexe/image/6371/img-73.png
Fichier image/png, 73k
Titre Fragment F.10
URL http://jso.revues.org/docannexe/image/6371/img-74.png
Fichier image/png, 79k
Titre Figure 8
URL http://jso.revues.org/docannexe/image/6371/img-75.jpg
Fichier image/jpeg, 156k
Titre Table 5
URL http://jso.revues.org/docannexe/image/6371/img-76.png
Fichier image/png, 33k
Titre Figure 9
URL http://jso.revues.org/docannexe/image/6371/img-77.jpg
Fichier image/jpeg, 744k
Titre Figure 10
URL http://jso.revues.org/docannexe/image/6371/img-78.jpg
Fichier image/jpeg, 320k
URL http://jso.revues.org/docannexe/image/6371/img-79.png
Fichier image/png, 3,1k
URL http://jso.revues.org/docannexe/image/6371/img-80.jpg
Fichier image/jpeg, 132k
Titre Figure 11
URL http://jso.revues.org/docannexe/image/6371/img-81.jpg
Fichier image/jpeg, 988k
URL http://jso.revues.org/docannexe/image/6371/img-82.png
Fichier image/png, 1,4k
Titre Figure 12
URL http://jso.revues.org/docannexe/image/6371/img-83.jpg
Fichier image/jpeg, 168k
URL http://jso.revues.org/docannexe/image/6371/img-84.png
Fichier image/png, 49k
URL http://jso.revues.org/docannexe/image/6371/img-85.jpg
Fichier image/jpeg, 1,0M
URL http://jso.revues.org/docannexe/image/6371/img-86.jpg
Fichier image/jpeg, 1,2M
URL http://jso.revues.org/docannexe/image/6371/img-87.png
Fichier image/png, 325k
Haut de page

Pour citer cet article

Référence papier

Konstantin Pozdniakov, « Tablet Keiti and calendar-like structures in Rapanui script », Le Journal de la Société des Océanistes, 132 | 2011, 39-74.

Référence électronique

Konstantin Pozdniakov, « Tablet Keiti and calendar-like structures in Rapanui script », Le Journal de la Société des Océanistes [En ligne], 132 | 1er semestre 2011, mis en ligne le 20 juin 2014, consulté le 25 septembre 2017. URL : http://jso.revues.org/6371 ; DOI : 10.4000/jso.6371

Haut de page

Droits d’auteur

© Tous droits réservés

Haut de page