When the missionary Eugène Eyraud arrived on Rapa Nui in 1864, he noticed wooden objects carved with rows of small figures. Some looked like birds or people; others were harder to recognise. The designs followed an apparent order and seemed to obey conventions familiar to their makers. Eyraud had encountered rongorongo, an inscriptional practice without a known counterpart elsewhere in Polynesia. The objects survived. By the time collectors began removing them from the island, however, the practice of teaching and interpreting rongorongo had already been badly disrupted.
Thousands of signs remain visible on the surviving tablets. Carvers repeated certain sequences and sometimes went back to correct their work, suggesting that the order of the marks mattered. Yet researchers cannot read a continuous inscription with general agreement. It remains unclear whether the signs directly represented spoken Rapanui, conveyed a mixture of linguistic and other information, or guided specialists reciting material they had learned aloud. We can describe a bird-shaped glyph without knowing which sound or meaning, if any, its maker assigned to it.
Researchers also face a severe shortage of comparable material. Fewer than thirty inscribed objects survive, hardly any came with records of their original purpose, and none carries a known text written alongside the signs. Similar passages on different tablets therefore cannot be securely translated by comparing them. Since 2024, radiocarbon work and increasingly precise records of the carvings have added evidence about the objects themselves. Statistical studies have also begun putting proposed readings through stricter tests. None has yet resolved what the inscriptions say.
What survives of rongorongo
The corpus is so small that scholars can identify its main objects by letters. Tablet A, Tahua, is an elongated piece made from a wooden blade that had been reused. Aruku Kurenga, tablet B, has carefully cut writing on both faces. Tablet C, Mamari, contains a passage often interpreted as a lunar calendar, while tablet D, Échancrée, is a damaged object whose wood has been radiocarbon dated. The Santiago Staff, text I, carries densely packed signs along a long rod. Other inscriptions appear on ornaments. Not every inscription was made on a conventional tablet, and the objects may have served different purposes.
Even the number of originals depends on which artefacts a researcher accepts. A 2024 radiocarbon study counted twenty-seven surviving inscribed objects. A study of Échancrée published in 2022 described twenty-six inscriptions, while a detailed investigation of the Berlin tablet limited its working corpus to twenty-three artefacts. The discrepancy reflects definitions and judgements about authenticity, rather than objects suddenly appearing or vanishing. Every known historical original is now kept outside Rapa Nui. Replicas and work by contemporary Rapanui artists, meanwhile, continue the tablets’ place in the island’s cultural life.
The removal of the objects makes their histories especially difficult to reconstruct. Missionaries acquired a number of the better-preserved tablets, and others passed through travellers, dealers and colonial collecting networks. Four objects examined in the 2024 dating project are in Rome; Chilean institutions and other collections hold further examples. A catalogue might establish the wood, size or route by which an object reached a museum. It rarely identifies the person who used it or the occasion for which the inscription was made. A photograph cannot recover the readers and occasions that collectors failed to record.
The glyphs follow identifiable patterns. Birds and human-like figures recur, often combined with other shapes into ligatures, or composite signs. Different hands seem to have followed some shared conventions, although their drawings were not identical. Whether a small difference in a head, beak or posture constitutes a separate character is still disputed. Thomas Barthel assigned numbers to hundreds of forms in his influential twentieth-century catalogue; later scholars have sought a smaller set of underlying signs. Even the number of distinct signs remains disputed, as do their possible values.
How the tablets were read and carved
One striking feature is the arrangement known as reverse boustrophedon. In the usual account, a reader follows a row, rotates the wooden object through 180 degrees and proceeds along the next row. Viewed without turning the tablet, successive lines consequently face in opposite directions. The way the figures are carved supports this method, but scholars can disagree about where an inscription begins and the direction of an actual recitation. Reconstructing the route through the lines establishes their layout, but leaves their contents unread.
The carvings themselves preserve traces of the makers’ work. According to Paul Horley’s close examination of Tahua, Aruku Kurenga and Mamari, carvers first incised the wood with obsidian and then strengthened the contours using a shark tooth. He also observed missed signs, insertions and new carving over older passages. The evidence points to sequences that carvers expected to reproduce accurately. It cannot settle whether a mark represented a word or syllable, since a tradition built around memorised recitations could also demand the correct sequence.
The timber sometimes had a history before anyone carved rongorongo on it. The British Museum identifies one original tablet as Pacific rosewood, Thespesia populnea, a valued regional wood. Others were made from timber unavailable on Rapa Nui and apparently reused after arriving by sea. The researchers who studied four objects in 2024 identified European ash in Tahua and tentatively associated Échancrée with Podocarpus. Scarcity encouraged reuse, leaving some objects with traces of several stages of construction, alteration and inscription.
The nineteenth-century destruction of the reading tradition
The loss of rongorongo knowledge happened during a catastrophe for the island’s people. Peruvian raids in 1862 and 1863 kidnapped a large share of the population, including people who held specialised knowledge. Many died in captivity. Epidemics then struck the survivors and some of those who returned, amid coercion and displacement. By 1877, records put Rapa Nui’s population at about 110 or 111. Those figures convey the scale of the disaster, although they cannot tell us how many readers of rongorongo had lived there or which individuals knew particular signs.
Missionary intervention and subsequent colonial rule brought further disruption. Some tablets were reportedly thrown away, repurposed or destroyed; others entered missionary collections because they were recognised as traces of knowledge in danger of disappearing. Much of what nineteenth-century outsiders wrote down thus dates from after the institutions through which islanders understood the carvings had suffered enormous damage. Those accounts are valuable evidence, though the collectors’ questions, languages and assumptions shaped what was recorded.
Bishop Florentin-Étienne Jaussen made one of the earliest serious attempts to record an interpretation. After receiving inscribed objects in Tahiti, he worked around 1869 with Metoro Taua Ure, a Rapanui man who described or recited signs from the tablets. Jaussen recorded the material in what is now commonly called the Jaussen list. The encounter is one of the few surviving records connecting particular signs with a Rapanui speaker who lived near the period when rongorongo was still practised.
Metoro’s responses have resisted attempts to turn them into a dependable key. Scholars cannot reproduce a consistent reading of the tablets from the surviving recitations, and they disagree over whether he read a script, recalled an associated chant, described its pictures or moved between those activities. Accusations that he simply invented his answers go beyond what the record establishes. We do not know his training, and Jaussen’s questions may have affected his responses. The list is valuable testimony, but it has not provided a reliable key.
Rapanui language and oral traditions continue, and present-day knowledge holders have their own interpretations of the tablets’ place in island life. What was broken was a particular specialised practice connecting carved signs with readers and performances. In a resource updated in May 2026, the Australian Museum presented Rapa Nui artist Victor Pakomio discussing the tablets in relation to astronomy, planting, fishing and choosing when to undertake activities. Pakomio’s account records a contemporary Rapa Nui perspective without offering a tested glyph-by-glyph reading of the historical texts.
The Mamari tablet and its possible lunar calendar
Among the inscriptions, Mamari gives researchers their strongest possible foothold. A carefully ordered passage has been associated with the nights of a lunar month. This makes sense as a hypothesis because Rapanui and other Polynesian traditions distinguished particular named nights and phases of the moon. The carved sequence might refer to those names, observances tied to them or a way of remembering their order. That interpretation could explain the sequence’s purpose, though the signs’ pronunciation remains unresolved.
Thomas Barthel put forward the calendrical reading during the twentieth century, and other researchers have since examined the sequence more closely. The interpretation depends on recurring signs, their ordering, the surrounding lines and comparisons with recorded lunar terms, rather than on the appearance of crescent-like figures alone. Miguel Valério’s 2024 study describes the passage as probably a list of moon nights, while expressly making that conclusion conditional. A calendar remains a well-supported possibility, but the passage’s complete meaning and the phonetic values of its signs remain unknown.
Even if the calendar interpretation is correct, the values of most signs in the inscription remain unknown. A column in an unfamiliar account book might clearly arrange dates without telling us which characters represent a day, a person’s name or an instruction. Mamari presents an added complication: signs might stand for whole words, syllables or marks that organise a recitation. A picture resembling the moon need not be read as the word for moon. The proposed lunar passage can help investigators test specific ideas, but it cannot serve as a parallel translation.
There may be more to Mamari than the sequence associated with the moon. Miguel Valério, Lorenzo Lastilla and Roberta Ravanelli’s 2022 study used three-dimensional documentation to reconsider its carved signs and suggested that some passages could concern agriculture or the learning of words. These are tentative readings, not deciphered text. Researchers can return to the shape of the original surface, check whether a damaged mark has been transcribed correctly and compare different explanations against the same physical object.
Why proposed decipherments have failed to settle the question
Steven Roger Fischer proposed one of the best-known readings in the late twentieth century. Looking at repeated groups of signs on the Santiago Staff, he argued that they preserved a creation chant built around successive acts of generation. His interpretation drew on patterns in Polynesian oral traditions and on apparently recurring grammatical structures in the inscription. Fischer’s wider documentation of rongorongo has been influential, although other specialists have not accepted his proposed reading as a general decipherment.
Repeated glyphs alone cannot identify a creation chant. A common glyph will appear in several contexts, and a short string can be given a persuasive meaning without predicting how its components behave elsewhere. A method that interprets one passage must also work on inscriptions that did not shape the original guess. Rejecting Fischer’s proposed values does not establish that rongorongo contains no language. It means that his particular connections between words and signs have not met that independent test.
More recent work has complicated the assessment. In October 2025, Jonas Gregorio de Souza published a statistical reassessment of sequences from the Santiago Staff, the Small Santiago Tablet and a Honolulu text. The statistical patterns suggested that some Staff passages might include personal names and that others could be arranged like parent-offspring lists or creation material. In Polynesian traditions, genealogies and accounts of origins can overlap. These patterns do not verify Fischer’s translation. They do, however, give researchers grounds to test structural comparisons with genealogies and creation accounts.
Any proposed decipherment needs a consistent procedure that another researcher could apply. It should define its signs, justify the values assigned to them, explain which small graphic differences matter and produce predictions for inscriptions not used to devise the theory. Those predictions must work better than accidental patterns in a small and repetitive collection. Independent investigators should be able to reproduce the result and identify evidence that would falsify the proposed reading.
Why a small corpus is such a severe obstacle
The size and condition of the corpus make testing difficult. Some of the few dozen objects are damaged, and estimates of total legible signs run into the tens of thousands only with different choices about fragments, doubtful inscriptions and transcription variants. One unusually long text can skew the statistics. There is nothing comparable to a large archive of ordinary transactions from many settings. If surviving tablets mainly preserve related genealogical or ritual material, their repeated formulas may tell us little about everyday vocabulary.
Researchers also lack a passage with a secure translation in a second writing system. The Rosetta Stone gave readers the same decree in different scripts and languages, allowing proposed readings to be checked against Greek and Egyptian. The decipherment of Egyptian hieroglyphs relied on more than choosing pictures that looked meaningful. Linear A poses another undeciphered problem, although archaeologists know much more about its archaeological setting. Neither comparison implies a historical connection between rongorongo and the ancient Mediterranean scripts.
Knowing the Rapanui language does not automatically produce a reading. The age of the inscriptions, their conventions for representing information and their ties to recited performance are uncertain. A glyph may represent a sound, a whole word, an idea or a prompt for remembered material. Analysts also disagree about the boundaries of individual characters. A ligature could combine two independent signs, constitute a single sign or be a variable drawing of one. Different decisions produce different sign counts and change the statistics on which some interpretations depend.
Most surviving tablets also lack an excavated setting. Scholars generally cannot associate an object with a securely dated deposit, a particular building and a known owner. Instead, nineteenth-century movements through collections account for much of their recorded history. Physical traces reveal carving, correction and reuse, but rarely establish who listened when a tablet was read. This missing context makes it difficult to choose between an account of events, a chant, names, a teaching text, a ritual calendar or a genealogy.
What the 2024 radiocarbon dates actually establish
Direct radiocarbon dating has narrowed some questions about the wood’s age. In February 2024, Silvia Ferrara, Sahra Talamo and colleagues published direct radiocarbon results for four objects held in Rome: Tahua, Aruku Kurenga, Mamari and Échancrée. The team sampled small amounts of wood, treated them to reduce contamination and calibrated the results using a Southern Hemisphere curve. They also estimated how far each sample lay from the outer growth rings, because the age of an interior piece of wood may precede the felling of the tree.
Échancrée, tablet D, gave the most striking measurement. The investigators estimated that its tree was felled between 1493 and 1509 CE within the reported 68.3 percent probability interval, centuries before Europeans were first recorded reaching Rapa Nui in 1722. The other objects generally yielded later estimates, with ranges that differed in width and sometimes had several possible intervals. The findings strengthened the case for investigating a pre-contact origin of rongorongo, but did not establish when its signs were invented.
The age of a piece of wood and the date of its carving are separate questions. Échancrée was made from non-native timber tentatively identified with Podocarpus, and the researchers discussed uncertainty over its exact species. Timber might have reached the island by drifting, transport or reuse, long after the tree was cut. Radiocarbon dating tells us the age of the sampled wood, not when a person incised its surface. The result provides a terminus post quem, meaning that the carving cannot be older than the wood. It cannot securely put rongorongo writing in the fifteenth century.
Échancrée bears other marks of a complicated history. Its two faces may have been carved by different people, and its notches suggest later use as a winding spool. The dating team considered how shipborne wood might have moved between places and eventually been reused by island carvers. The early measurement deserves consideration, but does not prove that rongorongo was written before European contact. Its glyphs look unlike European writing, but the appearance of a script cannot independently determine its date.
A post-contact invention remains possible: people can encounter writing elsewhere and develop their own system without borrowing its letterforms. In 1770, Spanish visitors described a ceremony in which marks were made on a document. Some researchers have wondered whether that encounter prompted experimentation with local notation; others stress rongorongo’s distinctive imagery and its relationship to Rapa Nui carving traditions. No inscription with a securely established date resolves these alternatives. The 2024 measurements improved the evidence without turning either explanation into settled history.
What changed in research during 2025 and 2026
Recent statistical studies test specific proposals about how the signs are ordered. De Souza’s 2025 analysis looked at patterns of signs in three texts, including arrangements that might distinguish names from formulaic material. A 2026 conference paper by Evgeniya Korovina examined a proposed syllable-based account of the signs against features of the Rapa Nui language. Using one transcription, the model fell below its own noise threshold; with a different dataset, the outcome was ambiguous. Such results test a computational proposal. They do not tell a reader what a rongorongo passage says.
Other 2026 preprints have made far stronger decipherment claims, including phonetic sign assignments and AI-assisted interpretations of large portions of the corpus. Each claim requires an independent check of its transcription and predictions tested against passages excluded from the original analysis. A method can report an impressive proportion of decoded signs if those same signs helped determine the meanings being evaluated. As of October 2026, there is still no widely accepted continuous reading of the historical tablets. A preprint proposing a solution is not the same thing as a replicated decipherment.
The recording of the carved surfaces has advanced more securely. Lorenzo Lastilla, Roberta Ravanelli, Miguel Valério and Silvia Ferrara combined close-range photogrammetry with structured-light scanning to produce a detailed model of Échancrée. The resulting measurements help researchers revisit older transcriptions and study damage, depth and the shapes of faint marks. The 2024 radiocarbon team similarly modelled its four Roman objects in three dimensions before taking samples. If an apparent character proves to be a scratch, comparisons built around it may collapse. Accurate records cannot supply a bilingual inscription; they can prevent decipherment attempts from resting on misidentified marks.
Several kinds of evidence could improve the next attempt. Detailed models available for independent examination would help settle doubtful carvings and variants. Careful sampling could reveal more about the timber’s history, so long as researchers distinguish felling dates from inscription dates. A better-supported inventory of signs and ligatures would make comparisons more dependable. Rapanui oral knowledge, historical linguistics and statistical testing each offer lines of inquiry. None guarantees that every tablet can ultimately be translated.
Rongorongo and the people whose history it records
The carvings themselves attest to their makers’ expertise. People prepared valuable wood, learned conventions for the signs, cut them in an organised sequence and sometimes repaired errors. These actions demonstrate a specialised practice far beyond casual ornament. What remains unknown is whether its primary purpose was linguistic recording, the organisation of memorised chants, the preservation of names or a combination of functions. The inability to read the text limits claims about its contents; it does not erase the craft and knowledge visible on the objects.
Modern decipherment is only one part of the tablets’ continuing significance. Rapanui elders, artists and researchers work with this heritage even where the old sequence of signs cannot be translated. The Australian Museum’s 2026 presentation made that relationship clear through testimony from Rapa Nui knowledge holders rather than through a speculative decipherment. Better scholarship must also involve access to the objects and their documentation for the community from which they were taken. The tablets remain unread, and they are still part of Rapa Nui’s cultural history.
