The recto side of tablet Keiti contains a number of repetitive phrases which suggest a tightly structured text. It can, for example, be divided into 22 segments delimited by two similar looking composites consisting of two bird-like and two abstract signs (3). Although the beginning glyphs are damaged, it is clear that a variant of this phrase also starts the inscription. One remarkable feature is that the first 11 segments all contain one or more independent or fused occurrences of the so-called adze-glyph, some 38 in total, whereas the sign occurs only once in the rest of the inscription (in line Ev5). In all probability, this sign which has phonetic value pu, is used so profusely here to refer directly and indirectly to the "hole" (pu) of 'Ana O Keke or a similar place to seclude the neru (4). The syllabic sign appears in a number of terms which describe the circumstances of the enclosed children: pupa-pupa, "cold"; punga-punga, "to grow fat"; puhi-puhi, "to bloat"; kopu, "belly"; pura-pura, "to blind"; puha-puha, "fat"; pu roa, "isolated hole"; puku, "cliff".

Another sign that is used very frequently on this side is the "crescent moon" which appears in the vocative construction e ... e and in the verb hehe, "to dazzle" (5). Part of line Er4 has a parallel in line Ab6 of tablet Tahua. This is immediately followed by a segment that is also present on tablet Aruku Kurenga (line Bv11). The final part of the last line on the recto side is similar to the start of texts C and H/P/Q. Fragments of this sequence also appear in lines Ra5-6 and Sa7 of the Washington tablets.

The verso side of Keiti continues with an independent text until halfway through line 2 another structured sequence begins that ends in the last part of line 5. This has 23 segments that are delimited by the "sitting man with staff'"-sign. Similar series – in different order and with many variations – are found on several other tablets. The sign of the "sitting man" which reads tetahi 'a, "those others", is the subject of a series of questions concerning the dire circumstances and the negative consequences of a prolonged seclusion of the neru. After this long sequence most of the rest of the verso side of Keiti contains an independent text. The exception is the first part of line Ev6, fragments of which have parallels in texts B, C, G, H, M, and R. The existence of all these parallels clearly shows that the text of Keiti is closely related to the other inscriptions.


Fig. 1  Drawing of the verso side of tablet Keiti (Thomson,1891:Plate 37)

The transcription of the tablet has been compiled primarily from drawings by Paul Horley published in Horley (2010a; 2010b), Pozdniakov (2011), and Davletshin (2014). This diversity of sources accounts for some variation in size and resolution. For the parts where these very accurate transcriptions were not available to the present writer, Barthel's have been used (indicated by a green line). When these proved to be faulty or incomplete, details of Pinart's rubbing as published by Horley (2010a:49) (fig. 3) have been inserted. These have been underlined in orange.


Tablet Keiti belonged to the set of rongorongo artifacts that was collected for bishop Jaussen by Roussel in 1868. It measured about 39 x 13 cm and contained 9 and 8 lines, respectively. These consist of some 441 signs or 972 elements (Horley 2005:112). The tablet perished during the German siege of Louvain in August 1914 when the Catholic University and its famous library were destroyed. Fortunately, most of the inscription has been preserved in photographs taken by Hoare around 1873 and Weisser around 1882, paper rubbings made by Pinart in 1877, and drawings (after photographs) published by Thomson in 1891 (fig. 1) (1).

Fig. 3  Detail of Pinart's rubbing (verso side) (from Horley, 2010a:49, fig. 4)

Tablet Keiti (Text E)


Added: 2013-09-08  Modified: 2017-08-07


(1) Hoare's photographs – for which the glyphs were enhanced with white filling resulting in some blurring – were first published by Lavachery (1933: fig. 2-3) and Stéphen-Chauvet (1935: figs. 157b; 158; 159), In the latter version, the blurred parts have been retouched, resulting in some incorrect outlines (Horly 2010a 45). Weisser's photographs were first published by Thomson (1891:Plate 46) and the invaluable rubbings of Pinart by Horley (2010a:49). Apart from photographs, rubbings, and drawings, there apparently existed a thin tissue impression of the tablet. This belonged to the collection of the SSCC in Rome before it was eaten by moths (Fischer 1997a:652, note 15).

(2) While writing Words out of wood (De Laat 2009a), I was still under the impression that the transcription of Keiti in Barthel's Grundlagen was accurate in general, although comparison to Barthel's own numerical transcription and the photographs published by Stéphen-Chauvet indicated that there were a number of issues. The publications of Horley have made it abundantly clear that neither Barthel's nor Fischer's transcription of Keiti stands much chance of going down in history as a prime example of scholarly Gründlichkeit.

(3) For an exhaustive structural analysis of Keiti and a comparison with other texts, see Pozdniakov 2011. Sequences which (appear to) consist of segments beginning with the same sign such as the "sitting man with stick" are frequently taken to be lists of some kind, even though there are many – Polynesian as well as Rapanui – examples of repetitious narrative, chants with refrains, etc., which display the same patterns but have nothing to do with lists. Text segments with a specific structure or with a repetition of certain glyphs also run the risk of being labeled as a separate text or as belonging to a different genres. For Keiti, it has, for example, been suggested that the recto and verso sides contain different texts or textual sub-categories (e.g., Melka 2008:170; Wieczorek 2011:7; Davletshin 2014:55). However, other than "structural" proof for these claims has not been offered.

(4) This density of the pu-sign is only matched by some 40 appearances in lines Br2-6 of tablet Aruku Kurenga.

(5) The presence of structured sequences of "crescent"-signs is likely to trigger a comparison with the so-called "lunar calendar" on tablet Mamari. Proposals of this type for text E and critical reviews can be found in Melka (2008), Wieczorek (2011), Pozdniakov (2011) and Horley (2011b).

Fig. 2  Examples of errors in the transcriptions of Barthel (1958) and Fischer (1997)

Although Barthel (1958:20) and Fischer (1997:435) have reported the presence of casts of the tablet in the Smithsonian Museum and the Museé de l'Homme, Horley (2010a:47) has argued convincingly that both must have been mistaken and that casts of this tablet probably never existed.

Barthel does not mention which sources were at his disposal, but his transcription is marred by numerous omissions and mistakes (e.g., fig. 2b, f, j) (2). A few of these have been corrected by Fischer (1997:433-434), but the bulk was simply repeated by him (e.g., fig. 2c, g, k) and some new ones were added (e.g., fig. 2o). Recently, this unsatisfactory situation has been remedied to a large extent by Horley (2010a). A meticulous study of all remaining documents has enabled him to rectify the errors in Barthel's and Fischer's transcriptions (e.g., fig. 2d, h, l, p). His paper also provides a comparison of the many parallels between the verso side of the Keiti tablet and the poorly preserved Small Vienna tablet (text N).