The rongorongo tablet from Berlin and the time-depth of Easter Island’s writing system

Abstract

Rongorongo is a non-deciphered writing system from Rapa Nui (Easter Island). Because the island was isolated from the outside world until relatively recently, rongorongo has the potential of being one of only a few instances in human history of an independent invention of writing. However, no scientific consensus exists regarding the time span for when rongorongo was used. Its cessation in the 1860s is well-known but its origins are not. Here, we report on detailed analysis of one of the 23 existing rongorongo artifacts—the Berlin Tablet—including botanical wood identification, radiocarbon dating, and photogrammetric study. The wood used to create the tablet was identified as Pacific rosewood, Thespesia populnea, a species that once grew on Rapa Nui, which counters previous theories that the tablet was made from salvaged driftwood. The radiocarbon date, adjusted in accordance to the ethnographic data, suggests that the tablet was made some time between ca. AD 1830 and 1870. Prior to its collection, the tablet had spent a significant amount of time within a cave context that destroyed around 90% of its content. The text is estimated to have been over 5000 signs long, more than double the length of the next longest rongorongo text.

Keywords:

• Sub-fossilized wood
• origin of writing systems
• Rapa Nui
• Polynesia
• Pacific Islands

Introduction

Rapa Nui¹ (Easter Island) is located in the southeastern Pacific Ocean and was one of the most remote islands ever settled by humans when Polynesians arrived during their widespread expansion across this vast ocean. Dates for colonization have ranged from AD 690 (Bahn and Flenley 2011, 125) to AD 1200 (Hunt and Lipo 2006, 1606; Mulrooney 2013), though recent analysis using a stringent selection of radiocarbon dates puts the initial settlement some time between AD 1150 and 1280 (DiNapoli et al. 2020). After arrival, Rapanui civilization developed in complete isolation until European contact in AD 1722.

While Easter Island is primarily known for its megalithic stone statues (moai) and ceremonial platforms (ahu) constructed along the island's shores, lesser known but equally impressive achievements concern the development of a unique writing system called rongorongo. The presence of writing on such a remote island is remarkable as it not only might represent an independent script invention—one of the few acknowledged cases of writing that include ancient Mesopotamia, Egypt, China, and Mesoamerica (Davletshin 2017; Guy 1985; Horley et al. 2018)—but also is the only known in all of Remote Oceania.

While a writing system is not mentioned in any expedition report before the arrival of the first resident missionary Brother Eugène Eyraud in 1864, it was suggested that rongorongo could have been inspired by European contact, in particular by the Spanish annexation ceremony of 1770 when islanders witnessed a reading of the annexation document by officials and were invited to sign the document themselves (Emory 1968; Fischer 1997; Haberlandt 1886). In spite of intensive scholarly debate, the Rapanui signatures on the Spanish annexation treaty cannot be considered as irrefutable evidence either pro or contra script existence in 1770. As such, the time depth of rongorongo remains unknown (Melka 2009b, 118; Robinson 2002; Wieczorek 2011).

Rongorongo itself is a very pictorial script with many signs representing recognizable objects like humans, birds, different kinds of fish, turtles, and many glyphs presumably reminiscent of various cultural objects. Most, however, are difficult to interpret. The number of distinct signs could reach several hundred though the precise number of these, and the deciphering of different textual fragments (or reading values of particular signs) are not agreed upon by scholars. For an extensive review of rongorongo, including ethnohistoric reports, ethnographic information, and epigraphic analysis, see Horley et al. (2018).

The demise of the script began with Peruvian slave raids in 1862 after a large group of Rapanui, including elites versed in writing, were carried to the continent. The survivors later returned to the island with new diseases that triggered devastating epidemics (Fischer, 1997, 8). As a result of these tragic events, traditional knowledge and lore were almost completely lost. The missionary observations of 1864 mentioned inscribed tablets as a common household item, saying briefly that the islanders carved them with "sharp stones" (i.e., obsidian tools) (Altman 2004, 21). A fragment of a tablet was given as a gift to the Bishop Tepano Jaussen of Tahiti in 1869; upon his encouragement, the best-quality inscriptions were collected in 1870. In that same year, the situation on the island had deteriorated to a degree that missionaries were forced to return to Mangareva and Tahiti together with a half of the island’s native population. Although a few tablets were eventually collected by the end of the 19^th century, no new inscriptions were apparently produced (except for a few late imitative attempts), which marks the terminal period of script use ca. 1862–1870.

Rongorongo glyphs were carved in wood, and around two dozen of these inscribed wooden objects are preserved in museums worldwide, although the authenticity of some is disputed. The choice of wood as a medium makes it amenable for determining its age using radiocarbon dating. However, until now, only one rongorongo tablet has been dated—the Small St. Petersburg tablet,² also known as rongorongo text Q, which was collected in AD 1871 (Miklucho von Maclay 1872; Piotrowski 1925) from Rapanui expatriates on Mangareva or Tahiti. The dating of this specimen resulted in an age of 80 ± 40 years before the present (BP). The 2-sigma calibrated age produced three date ranges with 95% probability: AD 1680–1740, AD 1800–1930, and AD 1950–1960 (Orliac 2005, 118), with the second range corresponding with its original acquisition. However, the generally poor physical condition of the tablet (wood degradation, insect damage) could indicate an earlier date of manufacture (Orliac 2005, 118) prior to the Spanish annexation ceremony.

Other scholars have argued for a greater antiquity of the tablets. Analysis of the wood from the Mamari tablet (rongorongo text C) by Orliac (2005, 117) reached the conclusion that this object was sculpted from Pacific rosewood Thespesia populnea (called mako’i in the Rapanui language), which once grew on the island. Judging from the size of the tablet and the marked presence of sapwood, the trunk of the tree from which the tablet was made would have been about 20 cm in diameter, which corresponds to a tree of around 15 m in height. The earliest European visitors in the eighteenth century noted the lack of tall trees on the island, and a comprehensive study of charcoal has shown that Rapa Nui was largely devoid of forests by the mid-seventeenth century (Orliac and Orliac 2000, 216). As a result, a tree large enough to provide suitable wood for the Mamari tablet would have had to be growing before this date. However, more recent studies have shown that deforestation was not necessarily uniform through time or across different parts of the island (Rull 2020), paving the way for the possibility that small tracts of forest could have persisted somewhat longer.

To improve our understanding of the rongorongo script, we report on analysis of the Berlin tablet, also known as rongorongo text O. The acquisition of the Berlin tablet is related to the visit of the German HMS Hyäne, which reached Rapa Nui in AD 1882. Although several tablets were known to exist at this time and were in possession of the chief Hangeto, the presence of a foreign ship incited the islanders to elevate the price. To address this issue, the Germans made an advance payment to Alexander Salmon, the manager of the sheep station on Rapa Nui, who agreed to ship the tablets to Germany with the German Imperial Consul Gustav Godeffroy of Tahiti. Salmon managed to acquire three tablets, which he passed to Heinrich August Schlubach, the German Consul-General at Valparaiso and the husband of Salmon's niece. Schlubach moved from Chile to Hamburg by March 1883, informing Adolf Bastian of Königlisches Museum für Völkerkunde (Berlin) about a number of Rapanui objects that were sent to the museum (Fischer 1997, 79–80). The shipment, including one inscribed tablet, arrived in April 1883 (Fischer 1997, 498). The tablet was given an inventory number (VI 4878) and is now preserved in the Ethnological Museum Dahlem in Berlin. This tablet has been discussed in previous publications (Barthel 1958; Bastian 1883; Fischer 1997; Horley 2012; Imbelloni 1951), with the other two tablets sold as part of the Schlubach collection that eventually reached the Museum of Ethnology in Vienna (Fischer 1997, 501).

The tablet measures 103 × 12.5 × 6 cm and weighs 2.6 kg. It is the heaviest surviving rongorongo artifact followed by the Santiago staff, which is 126 cm long and 6.5 cm in diameter and weighs 2 kg (Fischer 1995a, 305). Unfortunately, most of the inscription that once covered the Berlin tablet was damaged by erosion. Only one side of the tablet contains clearly legible glyphs, and even these cover less than a half of the side’s surface (Figure 1). It was initially postulated that the Berlin tablet was made from a piece of driftwood (Fischer 1997, 497). This is certainly plausible given that wood was extremely scarce and highly valued, particularly during the later stages of occupation when forest cover was declining (Bahn and Flenley 2011, 259–268; Wieczorek 2017, 68). Moreover, wood washed to shore was considered a gift from the gods, imbued with the sacred power mana (Orliac and Orliac 2008, 268). Reusing such wood for inscribing would have amplified the power of the glyphs by virtue of its origin. At the same time, driftwood is more problematic for radiocarbon dating. Even for tablets carved from a local wood, dating results may be difficult to interpret if they were carved from thick branches or trunks of ancient trees leading to inbuilt age (the “old wood” problem) (Schiffer 1986). To reduce this ambiguity, ¹⁴C dating of short-lived plant species or small fragments such as twigs or seeds is preferred. In addition, older wood could also have been used in a more recent carving, with the radiocarbon age not adequately capturing the cultural event (i.e., carving). As a result, it was imperative to identify the wood taxon prior to dating.

Figure 1. The Berlin rongorongo tablet. Orthographic views are rendered from a textured 3D model obtained with photogrammetric reconstruction from a set of digital photographs. The better-preserved side features patches of brown wood (illustrated in close-up photograph that covers the area around the largest knot) with lines of rongorongo glyphs arranged in inverse boustrophedon fashion (the neighboring lines go in the opposite direction and are turned upside-down in respect to each other). Glyphs were carved with a sharp implement into a set of shallow grooves (flutes).

Materials and methods

Materials

The Berlin tablet was examined macroscopically with a magnifying glass and a movable light source. The photographic documentation of the Berlin tablet was carried out in two sessions (February 2017 and August 2019) using a 24 megapixel Nikon^TM D7200 digital camera. General views and close-ups of the tablet were taken under direct and slanting illumination to emphasize the faint glyph contours and surface texture, including line fluting. In total, more than 2100 photographs were taken of the object.

3D modeling

A 3D model of the tablet was created to help identify the wood taxon vis-a-vis volume and density estimation. A high precision rendering of the tablet’s surface is also useful for determining the total length of the inscription. The 3D model was produced using photogrammetric reconstruction from a set of digital photographs with Agisoft PhotoScan^TM 1.4.0 (Agisoft LLC 2018). The resulting model contains about 5 × 10⁵ data points (approximately 10⁶ faces) with precision sufficient enough for studying principal surface formations such as knotholes, fluting, and traces of erosion. The mesh produced by PhotoScan was analyzed with VCL–ISTI–CNR MeshLab^TM 2016.12 (Cignoni et al. 2017) where it was aligned to the principal axes and rescaled so that the longest bounding box edge of the model was equal to the tablet length. MeshLab was used to calculate tablet cross-sections and the main geometrical parameters of the 3D model, including mesh bounding box dimensions, mesh volume, and mesh surface area. To reduce calculation time required for application of a texture (obtained by image averaging from a selected set of photographs), the mesh size was reduced to 2 × 10⁴ faces. Figure 1 shows the principal views of the textured mesh rendered in orthographic projection. The 3D model of the tablet is included as supplementary material.

Sampling

For botanical identification and radiocarbon dating, several wavers (about 100 mg of wood) were cut off with a scalpel from the side of the tablet that was severely damaged. Wood fragments partially split by a crack were chosen for this purpose (Figure 2A–B). The general size and number of collected wood wavers are shown in Figure 2C. The single biggest piece of wood was used for microtome sectioning, producing thin samples for botanical identification. All unused fragments were subsequently collected and sent to the Radiocarbon Dating Laboratory at the University of Waikato.

Figure 2. Details of the sample collection: (A–B) before and after views illustrating removal of a wood splinter marked with an arrow; (C) the collected wood sample placed in a plastic phial; (D) results of ¹⁴C dating calibrated with SHCal20.

Botanical identification

Botanical identification is critical for helping to determine whether the origin of the wood for making the tablet was indigenous to Rapa Nui or relic driftwood. For taxonomic analysis, the wood sample was boiled in distilled water until it sank. We then obtained transverse and longitudinal sections approximately 25–30 μm thick with a sledge microtome (Leica Biosystems^TM SM2010R, Wetzlar, Hesse, Germany). The sections were stained with a 0.5% w/v aqueous solution of safranin and 0.5% w/v aqueous solution of alcian blue. These were subsequently dehydrated in ethanol solutions in increasing concentrations and mounted in Euparal (Carl Roth, Karlsruhe, Baden-Württemberg, Germany). Wood anatomy was examined using light microscopy and descriptive terminology that followed the nomenclature of the International Association of Wood Anatomists (IAWA Committee 1989). Wood maceration was not performed due to scarcity of the material.

For wood identification, we obtained a list of species that shared the suite of diagnostic traits of wood from the Berlin tablet using the InsideWood database (InsideWood 2004; Wheeler 2011). We then compared our wood sections with anatomical descriptions and microphotos of the secondary xylem in species from our list and their relatives which are available in the wood anatomical databases (FFPRI Wood Identification Database Team 2002; InsideWood 2004; Wheeler 2011) and the literature (Chattaway 1933; Dhikale, Jadhav, and Nirmal 2019; Friday and Okano 2006; Kukachka and Rees 1943). Finally, we compared the structure of our sample with that of all woods previously identified in Rapanui artifacts (Orliac 1993, 1998, 1999, 2005, 2007, 2010) as well as all woody plants reported on Easter Island that could potentially be used for carving (Zizka 1991).

Radiocarbon dating

One of the major hurdles in establishing the time-span of the rongorongo phenomenon and whether it predates or postdates European contact is the lack of hard scientific data allowing for secure anchoring in time of different rongorongo objects. To help address this issue, we submitted a sample of the Berlin tablet for radiocarbon dating. The sample was cleaned and ground and contaminants removed by soxhlet extraction with xylene, toluene, ether, acetone, and distilled water (in an elutrope sequence). The sample was then washed in hot HCl, rinsed, and treated with multiple hot NaOH washes. The NaOH insoluble fraction was treated with hot HCl, filtered, rinsed, and dried. The sample was converted to CO₂ by oxidation at 800 °C overnight and subsequently graphitized. The concentration of ¹⁴C in the graphite was analyzed using accelerator mass spectrometry (AMS) and radiocarbon results fractionation-corrected using the measured on-line AMS δ¹³C values. Results were calibrated using OxCal^TM v.4.3.2 (Ramsey 2017) and the southern hemisphere correction curve (SHCal20) (Hogg et al. 2020) (Figure 2D). Previous radiocarbon measurement of the Small St. Petersburg tablet (Orliac 2005) was also recalibrated with SHCal20 to provide a comparator.

Results and discussion

Description of wood and its identification

Micrographs of wood sections from the Berlin tablet are shown in Figure 3. Growth rings were not observed, perhaps due to narrowness of the sample available for study. The wood is most likely diffuse-porous. Vessels are circular to oval and rounded in outline, (50–)91(–140) μm wide, and scarce (ca. 16 vessels mm⁻²), often filled with dark brown contents. Vessels are partly solitary (38%) and mostly arranged in small groups of (2–)3(–6). Vessel walls were (4.9–)7.4(–9.3) μm thick. Perforation plates were not observed directly because material was not preserved sufficiently well, but judged from the transverse cell wall remnants’ shape, the perforation plates were exclusively simple. Intervessel pitting is alternating (i.e., pits that form on lateral walls of vessels are arranged in more or less diagonal rows). Pitting between vessels and ray parenchyma (i.e., vessel-ray pitting) is similar to the intervessel one. Helical thickenings were not found (i.e., the internal wall seems smooth in light microscopy). Fibers are non-septate and mostly moderately thick-walled (thin- to thick-walled according to the IAWA Committee [1989]) with minute pits. The fiber wall thickness range was (1.9–)2.8(–3.8) μm. Axial parenchyma is diffuse-in-aggregates and diffuse as well as scanty paratracheal to vasicentric (i.e., seen in the transverse section parenchyma cells are randomly scattered solitarily or in small groups, or occasional parenchyma cells are associated with the vessels, sometimes forming a complete sheath around them). Observed in the longitudinal section, parenchyma cells are fusiform and are arranged in short strands of up to three cells. All rays are short and narrow (6 mm⁻¹). Most rays are bi- and tr-iseriate, (78–)119(–188) µm high, and with some sheath cells. Very few of the observed rays could possibly be uniseriate. Most ray cells are procumbent. The square and upright ray cells either flank the ray or form rows between rows of procumbent cells. The tile cells of Pterospermum type occur in the ray composition. Many ray cells contain prismatic crystals or druses.

Figure 3. Microscopic sections of Berlin tablet wood sample: (A–B) transverse sections showing the lack of growth rings, diffuse porosity, thin- to thick-walled fibers, deposits in vessels, diffuse-in-aggregates, diffuse and vasicentric axial parenchyma; (C–D) tangential sections with storied fibers, axial parenchyma strands and narrow rays (panel C) and lateral wall of a vessel element with alternate pitting (panel D); (E–G) radial sections showing short vessel elements apparently with simple perforation plates on transverse end walls as well as the rays with procumbent, square (occasionally upright) and tile cells in their composition. A row of possible Pterospermum-type tile cells is indicated with an arrow in panel F. Prismatic crystals in square and upright ray cells are indicated with arrows and a close-up image in panel (G).

Fibers, strands, and fusiform cells of axial parenchyma as well as the rays are in storied arrangement. The combination of simple perforation plates, alternate intervessel pitting, non-septate libriform fibers with simple to minutely bordered pits, diffuse-in-aggregates and vasicentric axial parenchyma, 1–3-seriate rays, tile ray cells, and prismatic crystals in upright and/or square ray cells (InsideWood search: 13p 22p 61p 66p 77p 79p 97p 111p 137p with 0 allowable mismatches) has been reported in Desplatsia chrysochlamys Mildbr. and Burret, Desplatsia dewevrei (De Wild. and T. Durand) Burret, Desplatsia subericarpa Bocq., Duboscia macrocarpa Bocq., Duboscia viridiflora (K. Schum.) Mildbr., Grewia brideliifolia Baill., Grewia triflora Walp., Grewia excelsa Vahl, Grewia mollis Juss., Pterospermum Schreb. sp., and Thespesia populnea (L.) Soland. ex Corrêa belonging to the plant family Malvaceae (InsideWood 2004–onwards). The wood of the rongorongo tablet is distinctive, however, from Desplatsia and two Grewia species (G. excelsa and G. mollis) by having narrower (mostly 2–3-seriate) rays. It also differs from G. brideliifolia and G. triflora by its scanty apotracheal parenchyma, and from Pterospermum sp. in the lack of rays having uniseriate portions of the same width as multiseriate ones. The wood sample under study shows the greater similarity to the wood of T. populnea and two Duboscia species. Unlike Duboscia (Kukachka and Rees 1943), however, the brown contents are found in vessels of the wood used for the Berlin tablet. As such, we can convincingly attribute this wood to T. populnea—the Pacific rosewood—as much of this is the only one of the aforementioned species naturally occurring on Easter Island (Zizka 1991).

Radiocarbon dating

The Berlin table sample was radiocarbon dated to 117 ± 14 years BP (i.e., 1950). The result after calibration at 2σ produced four date ranges (Figure 2D): AD 1706–1721 (5.5% probability), AD 1811–1838 (23.1%), AD 1848–1868 (5.5%), and AD 1878–1928 (61.4%). As the tablet was collected in 1882, we can confidently reduce the upper limit to 1882. Moreover, taking into account that the knowledge of rongorongo went into a sharp decline during the years AD 1862–1870 (Fischer 1997, 8; Horley et al. 2018, 325), we tentatively suggest that the upper limit can be further reduced to AD 1870 or perhaps even to AD 1862. The "ethnographically-adjusted" terminus ante quem of AD 1870 and "collection-adjusted" terminus ante quem of AD 1882 differ by 12 years, which accounts for the considerable surface erosion acquired by the tablet upon its prolonged unattended storage and/or abandonment.

Discussion and conclusions

The Berlin tablet is made of a large curved tree branch that was partially “flattened” by removing wood from its sides. After this, it was polished and a set of shallow sunken grooves (flutes) was created, perhaps using shark skin in lieu of sandpaper. The inscription was carved inside these flutes, which might have offered some protection for the signs. It is unclear whether the ceremonial life of the Berlin tablet was prolonged or not. Any possible traces of frequent handling (which are visible on other tablets) were erased by subsequent erosion.

The tablet was likely stored in a cave for a prolonged period of time. The side facing the soil was almost completely obliterated by rotting, retaining only faint traces of fluting. The better-preserved side also features extensive damaged areas. The most notable erosion occurred in the central part of the tablet, as if an area about 20 cm wide was prone to retaining moisture. We speculate that soil or dirt washed in by rain covered this area, or perhaps some plants growing nearby spread their leaves over the tablet's surface. The tablet also features a few carbonized spots at the edges, most notably wide notches on the tablet's convex edge are carbonized along their bottom edges.

Arthropod damage to the tablet is also evident, which was primarily inflicted by woodlice. The space under the tablet seemed to provide shelter for these organisms as is evident by at least 17 woodlice melts preserved in the crevices of wood (Figure 4B,C). It is likely that at the time of collection, the number of woodlice melts was even higher; they are only loosely connected to the wood and can fall off even after gentle handling. Comparison of the photographs taken in two documentation sessions (2017, 2019) revealed one such dislodged melt (Figure 5).

Figure 4. Details of arthropod damage: (A) dorsal view of Porcellionides pruinosus (Walker 2008)—the isopod species that seems to be principally responsible for active modification of the tablet; (B) close-up to woodlice melt (cuticle) still attached to the wood; (C) melt surrounded with a sunken area—a woodlice burrow; (D) two neighboring woodlice burrows, one with melt skin still attached, and the other featuring an area of different coloring corresponding to a dislodged melt; (E) series of woodlice burrows aligned along tablet's edge; (F) a cluster of wormholes with diameter 0.3–0.5 mm; (G) several wormholes with diameter 1.0–2.0 mm. The bottom part of the figure provides a visual guide for the areas covered in panels (B–G) and marks locations of woodlice burrows, empty or with cuticles still attached.

Figure 5. Dislodging of woodlice melts: (A) photograph taken in 2017, showing two melts in a large round concavity (marked with arrows); (B) photograph taken in 2019 shows only one melt in the same place (marked with the upmost arrow).

The exposed side of the tablet was also attractive to woodlice, most prominently along its concave edge. Some 33 depressions—the traces of woodlice burrows—are scattered across the tablet, each measuring about 0.7 × 0.4 cm. We are confident that these burrows were made by woodlice because their melts (cuticles) are preserved in such depressions at the tablet's edge (Figure 4D). An ordered row of concavities running along these edges (Figure 4E), pictured in a photograph under low-angle illumination, might have been perceived as a row of holes. Indeed, some scholars expressed the opinion that the Berlin tablet was re-used in marine carpentry (Barthel 1958, 27; lmbelloni 1951, 101). Fischer (1997, 497–498) was the first to note that the tablet's shape and thickness do not meet the requirements of such use, and furthermore, the tentative "lashing holes" observed in the photographs do not completely perforate the object.

Eight woodlice species have been reported on Rapa Nui (Taiti and Wynne 2015). Styloniscus manuvaka and Hawaiioscia rapui are endemic, and the rest are cosmopolitan. The cuticles still attached to the wood of the Berlin tablet measure 6–11 mm (Figure 4B–D), which is too big for the native species and also Trichorhina tomentosa and Venezillo parvus; none of them exceed 7 mm in body length. Three species (Porcellio scaber, Porcellio laevis and Armadillidium vulgare) may reach 15 mm or more. The only reported woodlice species that fits the observed marks is Porcellionides pruinosus (Figure 4A), which has a body length generally ranging from 6 to 11 mm (Walker 2008). Unless other larger native isopod species existed in the nineteenth century, the evidence suggests that P. pruinosus—a species of Mediterranean origin—was already present on Rapa Nui at the time when the Berlin tablet was stored in a cave prior to its collection in 1882.

Although woodlice damage is more prominent, other small animals also left traces on the tablet. A few wormholes were observed which cluster into two principal groups by size: 0.3–0.5 mm (Figure 4F) and 1.0–2.0 mm in diameter (Figure 4G). Since the two sets of wormholes appear in distinct regions of the tablet, we assume that they were left by two different animal species, though it was not possible to identify these species with any certainty. Rapa Nui caves shelter many arthropods, which are currently restricted to a fraction of their former range due to extensive anthropogenic disturbance (Taiti and Wynne 2015, Wynne et al. 2014). The tablet contains about 387 glyphs in total, including 15 glyphs preserved on a tiny patch of wood on the eroded side (Figure 6A) in close vicinity to the flat extremity, and about 30 glyphs scattered on islets or semi-eroded wood along the tablet's edges (Figure 6B–D), none of which have been discussed or published before. Although fragmentary, the remains of these inscriptions clearly show the inverse boustrophedon arrangement of lines which is characteristic of rongorongo. The presence of scattered inscription fragments on the edges confirms Fischer's assumption that the entire surface of the Berlin tablet was originally inscribed, possibly totaling some 26 lines of text (Fischer 1997, 496–497).

Figure 6. Half-eroded inscription fragments that were not noticed in the previous studies: (A) at the badly-damaged side of the tablet; (B) on the concave edge; (C–D) on the convex edge. Tentative tracings of these fragments are provided side-by-side for comparison. Arrows are used for marking correspondence to extremely faint glyphs.

To address this question in more detail, we studied cross-sections of the 3D model. The perimeter of the tablet generally ranges around 29–30 cm (Figure 7). The surviving lines feature glyphs about 1 cm in height; some extra space is also occupied by the fluting ridges that separate the lines. Although the flutes are quite shallow (only about 0.4–0.5 mm), they nevertheless can be distinguished in the tablet's cross-sections. Fluting ridges are discernible not only on the better-preserved parts of the tablet, but also the severely eroded side.

Figure 7. Cross-sections of a 3D model with perimeter values marked inside. The schematic diagram of the tablet at the bottom of the figure marks location of the cutting planes (A–C). For cross-section (A), the reconstructed set of lines is visualized with colored segments going around its perimeter. The numbers set inside the cross-section denote quantities of tentatively identified glyphs per line. The colors of the segments mark line directions; these are also shown with arrows in the schematic diagram of the tablet at the plane corresponding to cross-section A.

To help identify how many lines were originally present, the average line height recovered from a well-preserved inscribed patch (Figure 7, shaded area on tablet sketch at the cross-section A) was used as a unit of measurement along the entire perimeter of the cross-section. As the edges are badly damaged, there was no way of estimating the measuring unit in this area. However, upon examination of the eroded side, the position of the unit was still in good agreement with surviving fluting ridges, suggesting that the same line height was likely maintained on the edges as well. Proceeding in this way, we concluded that the original tablet likely had 28 lines, which formed an unbroken, inverse boustrophedon arrangement over its entire surface.

Fischer (1997, 497) was the first to observe some geometric similarity between the Berlin tablet and Santiago staff, the latter of which is too thick to be properly classified as a tablet. The cross-sections of the Berlin tablet are elliptical; they are even more rounded upon approaching to the central part of the object (Figure 7B). The artifact is so thick that no single edge line can be picked in favor of its neighbors; even when viewed strictly in profile, four to five lines can be seen. Given these characteristics, the Berlin tablet lies somewhere between what would normally be referred to as a “tablet” or “staff.”

The metric data obtained from the 3D model can be also used for estimation of the original inscription length. Before proceeding with the results, a short explanation about the counting of rongorongo glyphs is necessary. Some of these are apparently composite, but no consensus has been reached so far about the catalog of independent signs. Barthel's sign catalog can be used for this purpose, but it mixes independent signs and compound signs (Pozdniakov and Pozdniakov 2007, 5). On the other hand, to avoid problems with sign definitions, glyphs can be counted (i.e., inscription units that are separated from each other by spaces) (Wieczorek and Horley 2015, 132). Because some glyphs are composed of several signs, glyph counts will produce lower numbers than sign counts. By comparing published glyph (Davletshin 2013–14, 46) and sign counts (Barthel 1958, 14–33) per tablet, the largest and most well-preserved inscriptions have sign/glyph ratios around 1.3. For example, the largest surviving rongorongo inscription on the Santiago staff is composed of 2320 signs (Barthel 1958, 24; Melka 2009a, 32) and 1752 glyphs (Davletshin 2013–14, 46), with a sign/glyph proportionality coefficient of 1.32. Barthel (1958, 28) noted that 90 signs documented by him on the Berlin tablet correspond to about 7% of the original inscription covering the better-preserved side of the object, which he estimated to contain some 1200–1300 signs. Doubling that for the entire surface of the tablet, there would be about 2400–2600 signs in Barthel's estimation, which is comparable to the sign count on the Santiago staff.

In our study, we performed a more rigorous estimation based on the geometry of the tablet. As an average glyph height on the Berlin tablet was estimated to be about 1 cm, we searched for tablets with a similar average glyph height, which include the Large Santiago tablet (1.02 cm), Small Vienna tablet (1.1 cm), Small Washington tablet (1.08 cm), and Honolulu tablet (1.1 cm). All of these objects are damaged to different degrees. One side of the Large Santiago tablet bears a carbonized scar, the Small Vienna and Small Washington tablets are broken on one of their ends, and the Honolulu tablet is considerably eroded. Among these tablets, only the Large Santiago tablet has one side almost intact, featuring 644 glyphs (Davletshin 2013–14, 46). The area of this side was estimated from the photograph (Horley and Pozdniakov 2018, 87) to be 475.82 cm², permitting calculation of the glyph density for the Large Santiago tablet as 644/475.82, or 1.353 glyphs per cm². The surface area of the Berlin tablet, obtained from its 3D model, is 2999 cm². Multiplying this value by the obtained glyph density, we estimate that the glyph count for the undamaged Berlin tablet might have been as high as 4058 and the sign count as up to 5275. The 387 glyphs documented on the Berlin tablet thus correspond to 9.54% of its original inscription, which contained more than twice the glyphs appearing on the Santiago staff, which total 1752 (Davletshin 2013–14, 46).

Identification of the Berlin tablet as Pacific rosewood (T. populnea) is significant given that it once grew on the island. Brought by the first Polynesian settlers, it was the wood of choice for Easter Island carvers (Orliac 1993; Orliac and Orliac 2008, 265–266). According to botanical identifications (Orliac 2010) and analysis of wood density (Horley 2011, 35), Pacific rosewood seems to be the only native wood used for carving rongorongo—no less than 11 inscribed objects were made from this wood, including the Berlin tablet. Six other tablets with their wood identified were made from foreign species: Podocarpus sp. (four objects), Fraxinus sp. (one object), and Proteaceae spp. (one object). The earliest documentation of the wood employed in rongorongo (Lavachery 1934), published almost a century ago, does not include microscopic images and so the identification remains equivocal. The volume of the 3D model of the Berlin tablet was calculated to be 5019 cm³, which, considering a tablet weight of 2.6 kg, provides a wood density of 518 kg/m³. The specific density for T. populnea is reported to vary significantly between 440 and 890 kg/m³ and averaging at 600 kg/m³ for dry wood (Friday and Okano 2006, 10). The densities of five wooden pendants collected on Easter Island, all made from Pacific rosewood, was calculated as 517, 550, 555, 556, and 603 kg/m³ (Horley 2011, 34–35), with an average density of 556.2 kg/m³. As a result, the wood density of 518 kg/m³ obtained for the Berlin tablet agrees reasonably well with the data reported for the dry Pacific rosewood used by the Rapanui carvers.

It should be noted that botanists and woodworkers use different approaches to quantify wood densities. Wood moisture content during measurement will affect the results, but a universally accepted procedure does not exist. Therefore, different studies report wood densities obtained with moisture content varying anywhere between 0% (oven-dried, basic density), 8%, and 12% (most commonly used to calculate weight density), up to the content moisture of fresh material; and in many cases moisture content is unknown (air-dry density) or unreported (Barnett and Jeronimidis 2003, 90). Since the moisture content of the Berlin Tablet—a sub-fossilized wood specimen that spent almost 140 years in the museum—is unknown, the obtained density may differ from other published results. Additionally, intraspecific density variation is affected by a multitude of biological factors, among them early- to late-wood ratio (Barnett and Jeronimidis 2003), which in turn is influenced by plant growing conditions. The identification of the wood as T. populnea is important given that it is a small tree some 10 m high with a trunk diameter of 20–30 m. Its growth rates of 1–3 cm of stem diameter/year mean that it reaches full size in around 30–40 years (Friday and Okano 2006). Thus, we can assume that the old wood problem in the case of Pacific rosewood would not typically exceed 40 years. Extended storage time in the moist climates like that of Rapa Nui does not favor Pacific rosewood; when left in a soil context, its lighter sapwood begins to discolor in a matter of weeks (Friday and Okano 2006, 11). The tree tends to develop crooked stems, which may explain the longitudinal boomerang-like shape of the tablet (Figure 1). The object is 12.5 cm wide, constraining the minimal size of the trunk or a branch. The sufficient stem size is reached by T. populnea in about 10 years of growth.

The question of when rongorongo script began in use and its duration is important. If developed prior to European contact with Rapa Nui, it would add significantly to our understanding of Polynesian cultural behaviors and complexity, but also the science behind the development of script and writing systems (Robinson 2002; Horley et al. 2018, 324; Davletshin 2017, 61). The dating of rongorongo tablets Q (Orliac 2005) and O does not solve this question; the obtained non-calibrated dates of 80 ± 40 years BP and 117 ± 14 years BP fall mostly into the nineteenth century (Figure 8). Knowing that the production of the tablets ceased after AD 1870 (or perhaps earlier after AD 1862), the most conservative conclusion is that the Berlin tablet was created some time between AD 1706 and AD 1870, but most likely in the nineteenth century. The full radiocarbon results, however, do still leave the possibility open of a pre-contact origin of rongorongo as the range cross the date of AD 1770 when islanders participated in the Spanish ceremony and were observed writing and practicing with a quill. The probability of obtaining such a large branch of T. populnea seems also have been more likely in the first half of the eighteenth century versus the first half of the nineteenth century.

Figure 8. Comparison of SHCal20-calibrated radiocarbon dates corresponding to the Berlin tablet (O) and the Small St. Petersburg tablet (Q). Darker shading indicates the epoch for which the use of the rongorongo script is supposed to be very active. Lighter shading starts with the terminal period of 1862–1870, when the script gradually fell into disuse and no new inscriptions were likely produced.

Five more rongorongo tablets (Tahua, Echancrée, the Small Vienna tablet, Large St. Petersburg tablet, and the Large Washington tablet) are carved from foreign woods of possibly European origin (Orliac 2010, 133), so that they may post-date the regular visits of foreign vessels to the shores of Rapa Nui. At the same time, this prediction may not hold for all aforementioned objects because some foreign wood could have feasibly arrived as driftwood. Only further radiocarbon analysis will help to resolve these chronological questions.

Efforts to anchor the ages of rongorongo based on paleographic differences between the tablets has been attempted (Guy 1985; Fischer 1995b; Wieczorek 2011). However, given that there is no solid point of reference for connecting different calligraphic forms to specific dates, such analysis is only relative and of limited benefit. Moreover, the observed calligraphic differences might have been caused not only by temporal changes, but also spatial distribution if there were different scribal schools operating on the island. Further studies of rongorongo as a social and artistic phenomenon require better understanding of its timeline, for which it will be important to continue with radiocarbon dating of the surviving rongorongo objects. With more tablets dated, we will have a firmer basis for determining whether this unique Polynesian script represents a fully-independent ancient pre-contact creation or constitutes a later contact-inspired development.

Acknowledgements

The authors are very grateful to Dorothea Deters (Ethnological Museum Dahlem, Berlin) for her kind permission to study, document, and obtain wood samples from the Berlin rongorongo tablet, which would not have been possible without her help. Many thanks to Ronny Heuschneider (Ethnological Museum Dahlem, Berlin) for his kind assistance with sampling the wood from the tablet. We would also like to thank Krzysztof Spalik (University of Warsaw, Warsaw) for making sledge microtome available to us and Alan Hogg (University of Waikato, Waikato) for his detailed advice regarding the radiocarbon dating and preparation of the samples. We would also like to thank two anonymous reviewers for their comments and suggestions that helped improve the paper.

Disclosure statement

The authors declare no financial or other competing interests.

Additional information

Funding

A.O. acknowledges financial support by the National Research Foundation of South Africa (incentive grant No. 109531), the Russian Foundation for Basic research (grant No. 19-04-00714), the University of Johannesburg and the Komarov Botanical Institute (institutional research Project No. AAAA-A19-119030190018-1).

Notes

1 According to the established naming convention, the island name is written as two words Rapa Nui; the name of the people living on the island, their culture and language are written with a single word, Rapanui.

2 Many inscribed objects are named after the city in which they are located; if several tablets belong to the same museum, they are usually distinguished as "Large" and "Small." The nomenclature of rongorongo corpus was established by Barthel (1958), who suggested the use of capital letters for denoting the inscribed objects.