Rongorongo: The Unread Script of Rapa Nui

Many rongorongo signs are recognisable depictions. Human and human-like figures, birds, fish, lizards, plants and geometric shapes all appear. The glyphs are incised into the wood rather than painted or burned onto it, and according to Eyraud's own account the islanders cut them with a sharp stone. Popular accounts often specify obsidian flakes or shark teeth; that particular attribution was not verified for this article and is not asserted here. What can be said is that the marks are cut, and that a contemporary observer said a stone did the cutting.

Thomas Barthel's 1958 study Grundlagen zur Entzifferung der Osterinselschrift established the standard catalogue that specialists still use: each surviving inscribed object carries a single-letter designation from A to Z, and Barthel's reference catalogue lists 599 sign shapes. The surviving corpus is roughly two dozen objects. Not one of them is on Rapa Nui. They are held in Santiago, Rome, London, Vienna, Berlin, St Petersburg, Washington, Honolulu, Paris and elsewhere, dispersed by the collecting of the nineteenth century.

The exact number is not fixed, and any page that gives you a confident single figure is overstating the case. Silvia Ferrara and colleagues, writing in 2024, say the script survives on twenty-seven wooden objects and elsewhere describe the total as fewer than thirty. Paul Horley's 2021 corpus edition is described in the review literature as covering the objects labelled A to Z plus one more. Tomi Melka and Robert Schoch argued in 2024 for a further piece. The safe and accurate statement is that roughly two dozen inscribed objects survive, with the exact count still moving as pieces are identified and argued over.

Rongorongo is written in reverse boustrophedon, and this is one of the very few things about the script that is not in dispute. The reader begins at the bottom left of the object and reads to the right along a line, then rotates the object 180 degrees to read the next line, and alternates that way to the end. Alternate lines are therefore upside down relative to their neighbours. Note how the direction was established: not by anybody's testimony, but by the physical evidence of the same text recurring on three different objects (H, P and Q) with lines of different lengths, which constrains how the lines must run. It is an internal, structural result.

How many signs are there? The honest answer is a range, and the range exists because specialists genuinely disagree about how to count allographs (variant shapes of one sign) and ligatures (fused compounds). Barthel's 1958 catalogue lists 599 sign shapes. Our own research file gives roughly 120 basic elements combining into roughly 600 compounds. Konstantin and Igor Pozdniakov's statistical reduction arrives at about 52 basic signs accounting for 99.7 percent of the texts. Current summaries of Horley's 2021 corpus edition put the underlying inventory at roughly 130 basic shapes. Total glyph tokens across the whole corpus run to on the order of 14,000 to 15,000.

One passage, on one object, is read with broad specialist agreement. The Mamari tablet (text C) carries a sequence across lines Ca6 to Ca9, and some scholars include Ca5, in which a crescent glyph changes shape progressively across roughly 29 to 30 positions, punctuated by other glyphs. It is read as a lunar calendar, matching the nights of a synodic month. THOMAS BARTHEL identified it in his 1958 Grundlagen. JACQUES GUY (1990) reconstructed its structure and proposed phonetic readings for several of the accompanying signs. PAUL HORLEY (2011) argued that the calendar is a complete thirty-night sequence in full agreement with astronomy and with other Polynesian calendars. The attribution matters: the identification is Barthel's, and crediting it to Guy, as is sometimes done and as our own research file did until this page corrected it, is a factual error, and it sits on the one reading this whole subject is able to offer.

Between 1869 and 1874 Bishop Florentin-Etienne Jaussen, known as Tepano Jaussen, worked in Tahiti with Metoro Tau'a Ure, a Rapanui man said to know the inscriptions, and recorded what Metoro chanted while looking at four tablets. It is tempting to treat this as the last living contact with the script, and popular accounts often present Metoro as the last man who could read rongorongo. Specialists have concluded otherwise: that Metoro was describing the pictures rather than reading a text, because his chanted lines do not hold up under later analysis. Jacques Guy devoted a 1999 paper in the Journal de la Societe des Oceanistes to the question of whether Metoro's testimony can be used to decipher the script at all, and the answer developed in the specialist literature is essentially no.

Most scholars think that if rongorongo is writing at all, it records Rapanui, an Eastern Polynesian language still spoken today. Konstantin and Igor Pozdniakov's statistical analysis (2007) reduced the sign inventory to about 52 basic signs covering 99.7 percent of the texts, observed that this is close to the 54 or 55 syllables of Rapanui, and read that as evidence for a mainly syllabic system with some word-signs. This is a real and respectable inference, and it is not a decipherment. A statistical resemblance between two inventory counts is suggestive; it does not tell you what any particular sign says.

Barthel, and later Paul Horley in papers of 2005 and 2007, argued from recurring glyph patterns that some texts may contain genealogical chants or creation narratives, which would be consistent with the widespread Polynesian practice of preserving genealogy by recitation. Repeated anthropomorphic glyphs occur in formulaic sequences that resemble the structure of Rapanui oral chants recorded by ethnographers. Nobody has read a genealogy off a tablet. The claim is that the structure looks like chant structure, which is a real published inference and is not a reading.

The claim that rongorongo is related to the Indus Valley script is refused. Guillaume de Hevesy proposed in 1932 that some forty rongorongo signs have counterparts in the Indus script and inferred a common origin. It was rejected by Alfred Metraux, whose Ethnology of Easter Island (1940) dealt with the comparison, and by Barthel. The two traditions are separated by roughly 4,000 years and half the planet, and no structural relationship survives systematic comparison. The visual overlap is exactly what one expects from any two independently developed sign systems drawing on a limited repertoire of distinguishable shapes. The same refusal applies to the wider family of claims that rongorongo resembles Chinese or Sumerian writing, or evidences trans-oceanic contact.

The claim that rongorongo encodes advanced astronomical calculations or mathematical systems is refused. No peer-reviewed support exists. This sits close to a real finding and the distinction is worth making carefully, because this page has already told you there is a lunar calendar on Mamari. Counting the nights of the moon is observation, and every seafaring people in the Pacific did it. Predictive or computational astronomy is a different claim altogether, and it has no support. No mathematical content has been identified by any rigorous analysis.

Ferrara and colleagues state their limits explicitly, and those limits travel with the result. First, radiocarbon dates wood and never dates carving: the authors write that the wood's age serves primarily as a terminus post quem for the tablet's manufacture, and that the inscription may be more recent than the wood it is cut into. Second, the wood of tablet D, Podocarpus latifolia, is native to southeastern Africa and never grew on Rapa Nui, and the wood of tablet A is likewise foreign; the paper says both reached the island from elsewhere and that the driftwood hypothesis gains significant traction. Third, the authors phrase their own conclusion conditionally rather than as a finding.

Before 2024 only one rongorongo object had ever been radiocarbon dated: text Q, the Small St Petersburg tablet. Catherine Orliac (Archaeology in Oceania, 2005) reported an age of 80 plus or minus 40 years BP, whose calibration gives three windows at 95 percent probability: AD 1680 to 1740, AD 1800 to 1930, and AD 1950 to 1960. The second window fits the tablet's documented acquisition in 1871. Orliac's botanical work identified the wood of the objects she examined as Thespesia populnea, Pacific rosewood, a tree of high ceremonial standing across eastern Polynesia, which counted against the idea that the tablets were made from salvaged driftwood. This page corrects our own research file on three points here: the dated tablet is Q, the Small St Petersburg tablet, not the Small Santiago tablet, which is G; Orliac identified Thespesia populnea and not toromiro; and her radiocarbon result pointed at the nineteenth century, not at a pre-contact origin.

Whether the script existed before Europeans arrived in 1722 is a live and genuinely important question, and as of 2026 the strongest evidence for a pre-contact origin is the tablet D wood date, carried with every caveat above. Direct dating of the glyphs is not possible: carving removes wood rather than adding datable material, and destructive sampling of museum objects is heavily constrained. Readers may encounter a claim that rongorongo originated between the thirteenth and sixteenth centuries. That range appears in our own research file, no source anywhere supports it, and this page does not assert it.

If rongorongo was invented on Rapa Nui with no model from any outside writing system, it would be one of a very small number of independent inventions of writing in human history, alongside Mesopotamian cuneiform, Egyptian hieroglyphs, Chinese script and Mesoamerican writing. Ferrara and colleagues draw exactly this parallel and treat the tablet D date as support for it, while framing the conclusion conditionally. The standard argument in its favour is the lack of structural resemblance to any known script.

Steven Roger Fischer proposed the alternative: that the idea of writing reached Rapa Nui in 1770, when a Spanish expedition asked island chiefs to sign a written declaration of annexation. Under this reading the islanders took the idea of marking meaning on a surface from that encounter and then built something entirely original in form. The 1770 document survives and the marks the chiefs made on it are preserved, and they are sometimes described as the earliest record of rongorongo-like signs made off the island, though that identification is itself argued. Do not read this as settled either way. A tablet whose wood was felled around 1500 sits awkwardly with an idea that arrived in 1770, but the wood is not the carving, so neither hypothesis closes the other.

The corpus is not closed. Tomi Melka and Robert Schoch published a description in 2024 of a previously undocumented piece, the Topaze stick fragment, a small inscribed wooden relic collected during HMS Topaze's 1868 visit to Rapa Nui. A separate object, the Rangitoki fragment, carries a rongorongo sequence in ink on bark cloth dated to 1869, a useful reminder that the script was not only incised into wood. Both are peripheral to the canonical A to Z corpus and neither is universally accepted as part of it. The right way to hold this is that new pieces still turn up, not that the corpus is now twenty-eight objects.

There is a structural reason nothing works, and it is not a lack of cleverness. With on the order of 14,000 to 15,000 glyph tokens spread across roughly two dozen objects, and no bilingual text of any kind, most computational and statistical decipherment methods cannot produce reliable results. Barthel himself acknowledged that the corpus sits near the minimum threshold for meaningful frequency analysis. Our own research file on writing systems puts the same point more bluntly: even if the correct reading were proposed tomorrow, it might be impossible to verify scientifically.

Martha Macri and others have argued that rongorongo may not be writing in the full sense at all, but a mnemonic system, comparable to the Naxi Dongba script, whose signs prompt a reciter who already knows the chant rather than encoding language sound by sound. If that is right, decipherment in the ordinary sense may be impossible, because what would be needed is not a key but the specific oral traditions the signs accompanied. Those traditions were broken in the 1860s, which is the subject of the final section.

No Rapanui islander could read the tablets when European missionaries asked in the late nineteenth century, and the reason is documented colonial violence rather than mystery. Peruvian slave raiders operating between October 1862 and 1863 removed a large part of the island's population, and a smallpox epidemic followed when the few survivors were repatriated. The precise scale is not settled between the sources: our own research file says approximately 1,500 islanders, roughly half the population, while Moreno-Mayar and colleagues, writing in Nature in 2024, say a third of the population was kidnapped. Those figures are not reconcilable from what was checked for this article, so this page gives the range and attributes it rather than picking one. The figure that is externally sourced and safe is the endpoint: the Rapanui population fell to an estimated 110 individuals.

This has to be said plainly, because the popular literature on Easter Island is saturated with a register that treats the island as a place of vanished mysterious islanders. It is not, and they did not vanish. Moreno-Mayar and colleagues sequenced the genomes of 15 ancient Rapanui individuals radiocarbon dated between 1670 and 1950 CE and concluded, in the paper's own words, that these results do not support a major population collapse on Rapa Nui after its initial peopling and before the 1800s, and rather suggest that the island was home to a small population whose effective size steadily increased until the 1860s. There was no pre-European ecological suicide. There was a slave trade and an epidemic, in one decade, and the population went from many to about 110. The Rapanui are alive today, numerous and self-governing, the island is a Chilean special territory with an elected Rapa Nui presence, and Rapanui is a language people speak.