Introduction

In the process of dissemination, barkcloth beaters evolved typologically. Based on morphology, function and ethnographic accounts from across the Asia Pacific, Tang (Reference Tang2000) delineates this evolution into two stages: an earlier composite type with a wooden handle and a later club type with an integrated handle (Tang Reference Tang2000, Reference Tang and Zhang2003; Tang et al. Reference Tang, Nguyen, Tang, Tang, Wu and Rolett2019; Wu Reference Wu2021: 157–58). Conversely, drawing on extensive archaeological evidence, Cameron (Reference Cameron and Howard2006, Reference Cameron, O’Connor, Clark and Leach2008) identifies eight distinct types, seven of which are found in Southeast Asia. This Cameron typology is important for understanding the evolution of this iconic tool and its spread across the Asia-Pacific region (see Figure 1).
Typological spectrum of barkcloth beaters encountered throughout Southeast Asia. Depicted are: I) the unworked type; II) the horned type; III) grooved pebbles; IV) the flattened grooved type; V) the handled type; VI) the hafted type; and VII) the Malay type (reproduced with permission from Cameron Reference Cameron, O’Connor, Clark and Leach2008: fig. 3).

This article focuses on type IV barkcloth beaters, defined as flattened stones with one face entirely covered in parallel rows of worked grooves. Used during Phase I of the Pearl River culture, between 6500 and 5700 BP, beaters of this type have been identified in various locations, including Taiwan, Hong Kong, northern Vietnam, the Philippines and Sulawesi (Soejono Reference Soejono, Poesponegoro and Notosusanto1984: 193–94; Cameron Reference Cameron and Howard2006: 67–68; Anggraeni 2012, 2022; Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018; Adhityatama et al. Reference Adhityatama2022), indicating a broad geographical spread. Cameron (Reference Cameron and Howard2006: 68) emphasises the unique technological advancement and specificity of type IV beaters compared to other types developed by Late Neolithic (3500–2500 BP) Southeast Asian populations. This advancement is evident in both form and function, indicating a substantial leap in barkcloth production techniques.
Type IV is the sole category of barkcloth beater identified in the archaeological and ethnographic record of Sulawesi. Though generally scarce in the archaeological record, excavations at Buttu Batu in South Sulawesi revealed 16 type IV beaters. Consideration of these artefacts alongside an extensive ethnographic collection from Seko and Rampi Valley in the Toraja highlands offers a robust basis for studying barkcloth beaters in the region. This article aims to deepen the regional understanding of Austronesian cultural dynamics by tracing the development and use of type IV barkcloth beaters from Southeast Asia to the Pacific, thus highlighting their role in technological innovation and cultural adaptation.
Results of excavations at Buttu Batu
Located within the Enrekang limestone formation in the mountainous area of Bambapuang, the Buttu Batu rockshelter (3°27′46.5″S, 119°44′38.6″E) is 383m above sea level, with the Saddang River flowing to the east (Figure 2). According to local, Torajan mythology, Mount Bambapuang plays an important role as the ladder of spirits leading to the realm of Puya (nirvana) (Nur Reference Nur2011; Duli Reference Duli2012). The rockshelter cavity is 14m long, 4m wide and 0.8m high, though the internal height is lower. The current internal surface is littered with limestone chunks that have fallen from the inlet wall and the land in front is used by the local community for teak (Tectona grandis) and candlenut (Aleurites moluccanus) gardens (Figure 3). Surface-collected artefacts include pottery sherds, stone artefacts, shells, fragments of animal bones, and pig teeth. Nine trenches were excavated inside the rockshelter during five fieldwork seasons between 2013 and 2022.
Aerial view of the Saddang River, Enrekang, taken using an unmanned aerial vehicle from approximately 250m above sea level (figure by authors).

Front view (looking west) of the Buttu Batu rockshelter, taken in 2018 (reproduced from Hasanuddin Reference O’Connor, Bulbeck and Meyer2018: fig. 11.5).

The substantial impact of taphonomic processes on the deposition of cultural materials within the soil matrix is apparent under the rockshelter. Rockfalls from both the ceiling and walls precipitated subsidence of the soil layers, to the extent that the excavation yielded a predominance of rock debris over soil (Figure 4). The discovery of small pottery sherds (ranging from 10–25mm in length and 5–15mm in width) suggests that the area underwent intensive, stratified habitation, adversely affecting the preservation of in situ finds. This stratification and subsequent disturbance led to instances of soil layer intermixing. Despite these challenges, it is possible to delineate four cultural strata, each characterised by distinct features.
Layers of rock that had fallen from the ceiling and walls of the rockshelter were uncovered during excavation, as shown in this photogrammetric model (figure by authors).

The uppermost layer (phase IV), radiocarbon dated to 460–301 BP (see Table 1), includes a solitary interment of cremated human remains and artefacts including an iron adze and pottery sherds. An earlier Palaeometallic layer (phase III), dated to c. 2060–1820 BP, contains the skeletal remains of three individuals; the distinctive shovel shape of the lingual surface of the incisor teeth in each individual suggests East Asian ancestry. Additionally, the layer includes 14 barkcloth beaters and faunal remains from pigs, deer and wild buffalo, indicating established pig domestication practices. A Neolithic layer (phase II), providing radiocarbon dates in the range of 3561–3405 BP and 3067–2883 BP, contains two subadult burials in a foetal position. These individuals also have shovel-shaped upper incisors. Pottery, mostly plain and cord-marked fragments, comprise the bulk of the material culture from both the Neolithic and the Palaeometallic layers. A pre-Neolithic layer (phase I), dated to 5900–5663 BP, includes lithic artefacts, Tylomelania sp. shells and pig bones. Starting at a depth of 1.1m in Trench C, this earliest phase of activity remains partially unexplored because of rock deposits but no pottery sherds have yet been identified. During five seasons of excavation at Buttu Batu, a total of 16 barkcloth beaters were discovered (see Table 2)—two in the historical layers (phase IV) and 14 in the Palaeometallic layers (phase III).
Representation of pottery types in the four cultural phases at Buttu Batu and associated radiocarbon dates.

Note: All radiocarbon analyses described here were obtained from research conducted at the Buttu Batu site, Enrekang in 2017 and 2022 by Hasanuddin and his team.
* Calibrated with CALIB 8.2 (Reimer et al. Reference Reimer2020).
Detailed information on the 16 barkcloth beaters from archaeological layers at Buttu Batu.

Discussion
The type IV barkcloth beater
Barkcloth beaters from archaeological contexts on Sulawesi are rare and most lack associated dates (van Stein Callenfels Reference van Stein Callenfels1951; van Heekeren Reference van Heekeren1972; Soejono Reference Soejono, Poesponegoro and Notosusanto1984; Kadir Reference Kadir1986; Simanjuntak Reference Simanjuntak and Simanjuntak2007). Conversely, ethnographic data on barkcloth beaters are abundant (Kruyt Reference Kruyt1920; Raven Reference Raven1932; Kooijman Reference Kooijman1963; Aragon Reference Aragon1990; Kusakabe Reference Kusakabe, Prasetyo, Nastiti and Simanjuntak2016). All documented barkcloth beaters found on Sulawesi, both archaeological and ethnographic, are identified as type IV (van Heekeren Reference van Heekeren1972; Soejono Reference Soejono, Poesponegoro and Notosusanto1984; Kadir Reference Kadir1986; Anggraeni 2012, 2022; Duli & Nur Reference Duli and Nur2016; Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018), with the single exception of a type I beater at Kamasi site (Kalumpang area) (van Stein Callenfels Reference van Stein Callenfels1951). The only dated archaeological examples come from the Karama River basin (West Sulawesi Province), specifically from the sites of Kamasi (2500–2000 BP; Anggraeni 2012, 2022), Sakkara (2000 BP; Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018) and Palemba (1710–1550 BP; Anggraeni 2022). Although not explicitly discussed by Suryatman and colleagues (Reference Hakim, O’Connor, Bulbeck and Meyer2018) and Anggraeni (2012, 2022), these sites have been instrumental in establishing a Palaeometallic date (2000 BP) for Sulawesi type IV barkcloth beaters. Given the vast geographic spread of this artefact type and the complexity of its ethnogenesis (Bellwood Reference Bellwood1979; Kwang Reference Kwang and Tang1994; Zang Reference Zhang1999; Jiao Reference Jiao2002; Wu Reference Wu2010; Deng Reference Deng2011; Tang Reference Tang and Tang2011; Chang et al. Reference Chang, Liu, Moncada, Seelenfreund, Seelenfreund and Chung2015), the available archaeological data are hardly representative of Sulawesi’s strategic position as a cultural conduit of Austronesian migration towards the Pacific.
The oldest known type IV beaters, dating from 6500–5700 BP, originate from the Xiantouling and early Austronesian cultural sites along the Pearl River in southern China. In Taiwan, examples from the Shuntanpu culture date from 4300–4200 BP (Liu Reference Liu2008), including the Chungleng (Tang Reference Tang2012) and the Dingchiaotzu sites (Chu Reference Chu2011), suggesting that type IV was in use for more than 2000 years at Xiantouling before spreading to Taiwan by 4300 BP. In Hong Kong, the two sites of Fu Tei Wan (Meacham Reference Meacham1994) and Man Kok Tsui (Chiu & Ward Reference Chiu and Ward1979) have also produced type IV beaters, suggesting their development from the Late Neolithic period onwards (4000–3000 BP). Type IV is also reported in northern Vietnam (Tan Reference Tan and Học1980) and Cebu (Philippines), though without associated dates (Cameron Reference Cameron and Howard2006: 69). In Sulawesi, type IV beaters are found both inside and outside of stone burials (kalamba) at three sites within the Kalumpang area and Central Sulawesi, likely dating to the early Christian era 2000 years ago (Anggraeni 2012, 2022; Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018).
The distribution of type IV barkcloth beaters—originating in southern China and later found in Taiwan, the Philippines and Sulawesi—underscores the artefact’s role in Austronesian cultural migration towards the Pacific (see Figure 5). The discovery at Phung Nguyen in northern Vietnam suggests that this type also spread to Mainland Southeast Asia, but the timing of this spread remains to be identified (Khoach Reference Khoach1980). Supporting this dispersal pattern, a study of paper-mulberry (Broussonetia papyrifera) chloroplast DNA sequences indicates that the plant was carried from southern China to northern and southern Taiwan and further to distant Pacific islands via Sulawesi and New Guinea (Chang et al. Reference Chang, Liu, Moncada, Seelenfreund, Seelenfreund and Chung2015). On Sulawesi, paper-mulberry bark has long been a crucial material for barkcloth (Anggraeni 2022).
Distribution and dating of type IV barkcloth beaters from South China and their spread to Taiwan, the Philippines and eventually Sulawesi (figure by authors).

Artefact associations at Buttu Batu
The material culture associated with type IV beaters at Buttu Batu indicates their technical function, rather than their status as a grave good. The beaters from the Palaeometallic phase were associated with plain pottery fragments, and the peak use of cord-marked pottery coincides with the onset of iron smelting technology at the site. This differs from contemporaneous sites elsewhere in Sulawesi, Indonesia and the Philippines; thus, the association of 14 type IV barkcloth beaters with the period of peak use of cord-marked pottery at Buttu Batu is novel and requires further consideration.
Cord-marked pottery was widely used by Austroasiatic populations inhabiting Mainland Southeast Asia between 9000 and 3000 BP. While the authors lack evidence to refute the entry of type IV barkcloth beaters and red-slipped pottery into Sulawesi via Taiwan and the Philippines, we also lack evidence to assert that type IV barkcloth beaters, red-slipped pottery and cord-marked pottery were introduced as a cultural package at Buttu Batu. Although it seems premature to initiate a discussion on the arrival of cord-marked pottery through a western route—Sumatra, Java, Borneo (Simanjuntak Reference Simanjuntak, Piper, Matsumura and Bulbeck2017) and then Sulawesi—the available evidence suggests that such a discussion has prospects.
The Southeast Asian Island Palaeometallic period began approximately 2000 years ago—the early Christian era—aligning with the dating of the third cultural phase at Buttu Batu, within which barkcloth beaters and evidence for their on-site production are associated with iron objects. Bulbeck (Reference Bulbeck and Barnard1996–1997: 1044) suggests that the inception of metallurgical technology in Sulawesi took place among the highland Toraja communities. Buttu Batu village (and the Buttu Batu rockshelter site) was one such community inhabiting the Bambapuang mountain area, where local mythology places the original settlement of the Toraja people before they spread to the broader Toraja mountainous regions (Duli Reference Duli2012). To date, the finds from the Buttu Batu rockshelter represent the earliest metal artefacts in the Sulawesi highlands, the area inhabited by the Toraja culture. Archaeological evidence therefore compels us to consider the validity of local mythologies.
Over five seasons of research across nine excavation trenches (25% of the site), no spindle whorls were found at Buttu Batu, indicating the absence of weaving activities or skills. This mirrors the findings from the Neolithic–Palaeometallic period sites along the Karama River, where spindle whorls are also absent (Anggraeni 2012), except for one fragment at Palemba (Anggraeni 2022). Similarly, Neolithic–Palaeometallic sites in Mallawa (Maros Regency, South Sulawesi) report no spindle whorls despite the presence of numerous adzes and pottery fragments (Simanjuntak Reference Simanjuntak and Simanjuntak2007; Hakim & Nur Reference Hakim and Nur2009; Hasanuddin et al. 2020). Sulawesi is notably excluded from a distribution map of spindle whorls across Southeast Asia to the Pacific (Buckley Reference Buckley2023), and the archaeological evidence supports the assertion that spindle whorls did not spread beyond Luzon during the Neolithic (Bellwood Reference Bellwood2007). As such, Cameron’s (Reference Cameron2002: 196; Reference Cameron, Bellwood and Duzon2013: 115) hypothesis that not all barkcloth-producing groups had textile technology, while all textile-producing groups knew barkcloth production, appears to be validated for Sulawesi.
Development of type IV barkcloth beaters in Sulawesi
The type IV barkcloth beaters from Buttu Batu show meticulous craftsmanship with precision in size and clear lines. Crafted from sandstone and limestone, each beater features two striking surfaces with grooves of varying shapes, sizes and depths, and all show signs of intensive use. The beaters were deliberately concaved for cord attachment and connected to a rattan handle. Iron ore was also associated with the beaters at the site, suggesting that the manufacturing tools are iron.
We compared type IV barkcloth beaters from across Sulawesi, including archaeological discoveries and dozens of ethnographic objects from the Rampi and Seko villages. Analysis of morphology indicates that the evolution of type IV barkcloth beaters in Sulawesi can be divided into two phases.
The first phase (from around 2000 years ago to the late nineteenth century AD, corresponding to the Palaeometallic period) is characterised by longitudinal groove patterns. Among the 16 examples available from Buttu Batu, variation is minimal and limited to the size, number of grooves and groove depth. The grooves are exclusively longitudinal and aligned with the longest axis on each beater (Figure 6a–d). Beaters with this longitudinal groove design are found at Buttu Batu, Sakkara (Figure 6e; Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018), Kamasi (Figure 6f; Anggraeni 2012) and Palemba (Anggraeni 2022) and continue to be found in the historic period (446–356 BP); the two type IV barkcloth beaters from phase IV at Buttu Batu are morphologically similar to the examples from phase III.
Examples of barkcloth beaters featuring longitudinal groove patterns from the Buttu Batu site (a–d), from the Sakkara site (e) (as described by Suryatman et al. Reference Hakim, O’Connor, Bulbeck and Meyer2018) and from the Kamasi site (f) (Anggraeni 2012). The lower row illustrates the evolution of groove motifs on type IV barkcloth beaters around the early twentieth century: g) a beater with an oblique horizontal pattern and a combination of deep and shallow grooves; h) one with a simple horizontal oblique pattern; i) another with a trapezoidal cross-groove motif; and j) a further example featuring a trapezoidal cross motif with additional deeper vertical grooves (figure by authors).

The second phase (early twentieth century to the present) is marked by a diversification of groove patterns, including longitudinal, horizontal, diagonal, cross and chequerboard motifs with combinations of deep and shallow grooves. This phase is supported by ethnographic records and materials, specifically seven sets of barkcloth beaters from the Rampi and Seko villages. Although still classified as type IV based on gross morphology and hafting style, these beaters differ from earlier examples in displaying a more comprehensive range of groove motifs.
During the first phase, which spanned approximately 1900 years, type IV barkcloth beaters were produced and used in Sulawesi, indicating the key role of barkcloth production in the late prehistoric era. The primary reliance on barkcloth can be attributed to the absence of alternative textiles, making barkcloth essential for clothing and utilitarian items such as blankets. Technological advancements during the Palaeometallic phase, including the production of metal tools, likely improved the efficiency and quality of barkcloth production, allowing for more precise crafting of beaters and easier processing of bark fibres.
Currently, the seven sets of barkcloth beaters from Rampi and Seko are used only for exhibitions or village parties. Consultation with village members indicates that the production of one barkcloth requires up to 15 beaters, two of which are made of hardwoods for the first stage, while up to 13 stone beaters may be used for the later stages. Each of the 13 stone beaters have two faces bearing different groove motifs and depths; including the hardwood beaters, this totals 28 different processing stages (see Figure 7). It can take two to three days to beat one piece of bark into cloth using the 15 beaters, and this cloth is then further processed, for example, to produce traditional garments such as those shown in Figure 8.
The arrangement of 15 barkcloth beaters from Rampi based on the stages of their use (figure by authors).

A girl from Rampi wearing an ornate bast shirt, skirt and hat made using barkcloth beaters (figure by authors).

Kooijman’s (Reference Kooijman1963: 18–19, 56) observations highlight the remarkable sophistication of barkcloth production techniques in the Sulawesi highlands, underscoring the integral role of the cloth in social status and the ritualistic practice of headhunting. At Rampi, barkcloth has been transformed into various garments, including jackets, skirts, shirts, shorts, shawls, sarongs and hats, with shorts being particularly prevalent. While everyday attire is generally unadorned, colourful and patterned clothing is reserved for ceremonial occasions and is typically associated with the higher social echelons.
Consideration of the Buttu Batu beaters and the demonstration of evolving groove motifs on Sulawesi leads us to question the catalyst behind the diversification of groove patterns in the early twentieth century. Insights from informants in Rampi and Seko suggest that the advent of varied groove motifs and depths was aimed at enhancing the decomposition of bark fibres, thereby producing a denser, finer cloth (see Figure 9). Although these responses do not entirely resolve our query, they indicate that the introduction of varied groove designs in a single beater set leads to more efficient processing and improved fabric quality.
Variation in grooved patterns of Sulawesi type IV beaters. The longitudinal line motif (a) is the only pattern observed in the first phase, while more complex motifs are observed in the second phase (figure by authors).

The critical role of barkcloth at the end of the prehistoric era can also be understood in the context of external influences and technological transitions. As Sulawesi began to interact more with neighbouring regions through trade and cultural exchange, new technologies such as weaving were introduced. However, these technologies initially penetrated only the more accessible lowland and coastal areas, leaving highland communities reliant on traditional barkcloth production. Therefore, at the end of the prehistoric era, barkcloth remained essential in these highland regions because of geographic isolation and the delayed adoption of weaving technologies.
To understand the development of barkcloth beaters, ethnographic records by Kruyt (Reference Kruyt1920) are invaluable. During his visit to Seko, Kruyt noted that over half of the women wore sarongs woven in Rongkong, accompanied by jackets made from barkcloth (Kruyt Reference Kruyt1920: 399). Meanwhile, in Rongkong, which is in a lowland area that is easier to access from the coast, the manufacture of clothing from bark was unknown (Kruyt Reference Kruyt1920: 374). This indicates that by the early twentieth century, weaving technology had substantially displaced the production of barkcloth in Seko, while in Rongkong, barkcloth clothing had been abandoned altogether.
Kusakabe (Reference Kusakabe, Prasetyo, Nastiti and Simanjuntak2016: 379) later distilled Kruyt’s (Reference Kruyt1920) descriptions of the complex barter of barkcloth beaters, woven fabric and buffaloes between the people of Rampi, Rongkong and Seko (Figure 10). Kusakabe (Reference Kusakabe, Prasetyo, Nastiti and Simanjuntak2016: 379) interpreted the buffalo exchange for barkcloth beaters as evidence of the coexistence of barkcloth making and cloth weaving in the Sulawesi highlands. In relation to the evolution of barkcloth beaters on Sulawesi, we view this complex exchange as indicating that the diversification of groove motifs in the early twentieth century was a response to the increasing trade in woven fabrics reaching Seko and Rampi, the most remote highlands of Sulawesi.
The Rongkong people (To Rongkong) exchanged ikat (woven cloth) for barkcloth beaters with the Rampi people. The barkcloth beater (pasasa) was then brought by the Rongkong people to Seko to be exchanged for buffalo (modified from Kusakabe Reference Kusakabe, Prasetyo, Nastiti and Simanjuntak2016: 379, diag. 1).

The introduction of weaving technology and the increasing availability of woven fabrics posed a substantial challenge to traditional barkcloth production. To remain competitive and meet the evolving demands of their communities, barkcloth producers innovated by diversifying groove patterns on beaters to enhance barkcloth quality. The adoption of various groove motifs allowed the production of finer and denser fabrics, making barkcloth an attractive alternative to woven textiles.
The coexistence of barkcloth making and cloth weaving in the Sulawesi highlands (Kusakabe Reference Kusakabe, Prasetyo, Nastiti and Simanjuntak2016: 379) presents one model of interaction, which contrasts with the situation in Sumatra, where bark fibres woven into fabric were used as the inner lining of clothing, with cotton forming the outer layer (Marsden Reference Marsden2013). There are no ethnographic reports on the combination of barkcloth and woven fabric as clothing materials on Sulawesi.
Conclusion
The discovery of 16 type IV barkcloth beaters at the Buttu Batu rockshelter in South Sulawesi provides support for the early adoption of this technotype on the island. Associated radiocarbon dates and artefacts indicate that the beaters were used during the Palaeometallic era and were probably produced using metal tools, indicating a level of technological advancement comparable to earlier examples from South China, Taiwan and the Philippines. Beater morphology remained remarkably consistent over 1900 years of use on Sulawesi before groove motifs began to diversify in the early twentieth century, potentially reflecting a need to produce higher-quality cloth and remain competitive as weaving technologies gained traction. Further comprehensive studies integrating a larger collection of barkcloth beaters and extensive ethnographic work are crucial to fully grasp the evolution and importance of this fabric-production method, not just on Sulawesi but across Austronesia more generally.
Acknowledgements
We express our sincere appreciation to the communities of Buttu Batu, Rampi and Seko for their assistance during our research. Furthermore, we extend our gratitude to Truman Simanjuntak for his guidance and contributions to our research, and to Akin Duli for his willingness to discuss the ethnographic aspects of barkcloth beaters.
Funding statement
The results presented here are the outcome of the investigative efforts led by Hasanuddin and the team in 2022 in Enrekang Regency, South Sulawesi, Indonesia. This study was supported financially by The National Research and Innovation Agency (BRIN) of Indonesia.












