Navigation – Plan du site

AccueilNuméros162Dossier thématiqueEmbodied and materialized cosmolo...

Dossier thématique

Embodied and materialized cosmological principles: a graph-theoretic approach to cycles and layers in North Pentecost Sia Raga sand drawing

Des principes incarnés et matérialisés : approche des cycles et des couches dans les dessins sur le sable par la théorie des graphes (Sia Raga, nord de la Pentecôte)
Alban Da Silva
p. 118-144

Résumés

Cet article propose une analyse ethnomathématique des dessins sur le sable des Sia Raga du Nord-Pentecôte (Vanuatu). Fondée sur une ethnographie réalisée en 2019, une modélisation mathématique basée sur la théorie des graphes permet de formaliser les règles orales et les gestes observés chez la plupart des praticiens lors de la réalisation des bwatuli. L’application d’un théorème majeur de la théorie des graphes révèle que ces dessins se décomposent en cycles et en couches – des concepts qui structurent par ailleurs de multiples domaines de la vie Sia Raga, notamment les cycles agricoles, l’organisation des espaces et les relations de parenté. Cette convergence suggère que certains principes cosmologiques, en particulier les processus de croissance et de transformation progressives, sont à la fois incarnés dans les gestes des praticiens et matérialisés dans certains dessins, comme le bebe ure, le papillon porteur de la création du monde. D’un point de vue méthodologique, cet article s’efforce de montrer que les approches ethnomathématiques offrent de nouvelles prises sur des problématiques déjà ouvertes en anthropologie, notamment la place du cycle dans les cosmologies, ou encore les catégories autochtones de créativité et de complexité.

Haut de page

Notes de l’auteur

Figures: All figures generated in Python in this article are licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0; figures from the author, 2025).

Texte intégral

Graph theory: from kinship systems to sand drawing in Vanuatu

Graph-theoretic approach in anthropology

1Mathematical approaches in anthropology, despite generating considerable expectations, have gradually lost momentum to now reside at the discipline’s margins. The purpose of this article is not to revisit the reasons for this trajectory, which have been extensively discussed (Selz-Lauriere, 1988; Casajus, 1994; Fabry, 2023). Instead, it asks whether ethnomathematics—a field whose methodological foundations are becoming increasingly established—might offer fresh analytical pathways for anthropological research. In particular, through an ethnomathematical case study, this article reconsiders the heuristic value of mathematical modeling for anthropology.

  • 1 Throughout his work, Lévi-Strauss continued to invoke mathematics, either by employing its formalis (...)
  • 2 This article directly employs graph theory in its mathematical generality. Thus, readers familiar w (...)
  • 3 This is not always the case: in some graphs, vertices can be positioned according to GPS coordinate (...)

2The convergence between mathematics and anthropology has historically followed two distinct epistemological trajectories. The first concerns the study of mathematical ideas developed outside formal academia, exemplified by ethnomathematics—a field approached below through the pioneering work on Vanuatu sand drawings. The second involves deploying formal mathematics as a methodological tool for ethnographic analysis, with roots in Lévi-Strauss’s structural anthropology. For Lévi-Strauss, the development of new “human mathematics” specific to the social sciences represented an opportunity to grant the social sciences their true scientificity, moving beyond the necessarily reductive use of statistics and the law of large numbers (Lévi-Strauss, 1954: 648).1 This vision found perhaps its most influential early expression in the famous appendix to Elementary Structures of Kinship written by the mathematician André Weil (Weil, 1979: 563), which became a classic example that inspired extensive literature on the mathematics of kinship (Courrège, 1965; Haddad & Sureau, 1993; Rauff, 2016). In particular, mathematical graphs,2 which both the present article and Eric Vandendriessche’s contribution extensively recruit (Vandendriessche, 2026), have proved particularly fruitful for modeling certain kinship relationship structures in Melanesian social systems (Guilbaud & Lévi-Strauss, 1970; Glowczewski, 1989). A graph is a mathematical structure composed of objects and a relation between these objects. The objects are usually represented by points called vertices, and the relation between two objects (or vertices) by curved or straight lines called edges. The shape of each edge and the position of the vertices are not important,3 and such a drawing is called a representation. Figure 1 is a representation of a graph with six vertices and six edges.

Figure 1. A graph with six vertices and six edges.

Figure 1. A graph with six vertices and six edges.
  • 4 Faced with the complexity of the Murngin marriage system (North Australia), Lévi-Strauss engaged in (...)

3An edge may be directed, indicated by an arrow, if the relation exists only from one vertex to another. This is the case for the graph of Figure 2, which could be perceived as a refinement of Figure 1. It also has six vertices, but twelve directed edges, which means that the relation from A to B may differ from the one from B to A. Introduced by the mathematician Guilbaud to describe the marriage systems of North Ambrym (Vanuatu),4 this graph is of particular interest for its striking methodological choice. Rather than representing individuals as vertices (as in traditional genealogical trees), Guilbaud chose to represent marriages as vertices and kinship relations as directed edges. This seemingly counterintuitive encoding revealed a cyclic structure with exactly six positions—a pattern that would have remained invisible using conventional tree diagrams. As we will highlight in this article, such a shift in what graph vertices represent can prove fruitful to reveal underlying structures.

Figure 2. A directed graph representing a six-class kinship system in North Ambrym.

Figure 2. A directed graph representing a six-class kinship system in North Ambrym.

Source: Guilbaud & Lévi-Strauss, 1970: 21.

4The relevance of such an approach stems from a fundamental property of graphs highlighted by some anthropologists: their capacity to represent discrete relational structures. Depending on the research context, the vertices may represent individuals, human groups, words, myths, or other entities (Hage, 1979). It helps to “get away from excessive entanglement in empirical facts and value-loaded concepts” (Leach, 1961: 12), revealing structural patterns that would otherwise remain hidden from ethnographers. Remarkably, the use of graphs in anthropology realizes the heuristic power that mathematicians themselves hoped graphs would offer: the capacity to “connect quite abstract things with clear ideas, whereby new connections among concepts and problems seemingly distant from one another often come to light” (König, 1936: 47). In cognitive anthropology, graphs have shown promise in analyzing social networks, allowing to focus on “structural properties (local and global) that have important feedback on behavior and cognition” (White & Johansen, 2005: 8).

Before modelling: fieldwork in ethnomathematics

5Outside such an approach, however, graph theory has found limited traction in anthropological practice. Lévi-Strauss himself cautioned against the tension between mathematical models and the reality of social life. He pointed to the lack of realism of “non-temporal structures in [Guilbaud’s] diagram,” noting the risk of misalignment with “the study of real history” of Ambrymese societies (Guilbaud & Lévi-Strauss, 1970: 30). These objections raise a broader epistemological question about the validity of formal models in the social sciences: they are never neutral but reflect the modeler’s own frameworks. This risks straying from indigenous conceptualizations, running counter to anthropology’s core commitment to capturing the emic perspective.

  • 5 Marcia Ascher (1935-2013) was a mathematician whose husband, Robert Ascher, was an anthropologist. (...)

6It is precisely this tension that motivates the trajectory of ethnomathematics, a research approach that seeks to illuminate mathematical ideas as they develop within cultural practices. Marcia Ascher,5 a pioneer in this field, approached this challenge by recruiting mathematical tools to capture the underlying structures of diverse cultural practices—divination systems, calendars, kinship organizations, and ground-tracing arts. In analyzing Vanuatu sand drawing, she articulated a theoretical position that guides this article:

Keep in mind that tracing procedures are hand motions that leave transient marks in the sand. One way to talk about them or write about them is to refer to them by symbols. The symbols are ours, not theirs. But, as often happens in mathematics, if we assign our symbols with care, we can highlight logical and structural relationships that may otherwise escape us. It is the Malakula who created the tracing system, but it is we who are introducing symbols in an attempt to capture and convey the structure of their systems. (Ascher, 1991: 48)

7Following this principle, Ascher deployed graph theory as her symbolic system to classify sand drawings and illuminate what she identified as the practitioners’ underlying procedures and algorithms (Ascher, 1988). Her analysis revealed Eulerian graphs within these drawings—a mathematical structure we discuss later—which she then classified by their structural properties to distinguish types across the available ethnographic corpora. While Ascher’s work opens rich methodological avenues, her graph-theoretical approach did not account for what I call the “forgotten rule”—a crucial sand drawing principle detailed further in this article. This relative oversight could be partly explained by the fact that Ascher worked solely from second-hand ethnographic sources without conducting fieldwork. The rule in question, essentially gestural rather than verbal, emphasizes the significance of the renewed ethnographic fieldwork that I carried out in North Pentecost. While direct fieldwork was not always standard practice in the early years of ethnomathematics (Vandendriessche & Petit, 2017), it has since become a well-established methodological stance in the current to which this article belongs, as Vandendriessche has shown through fieldwork conducted in Vanuatu and Papua New Guinea on string figures (Vandendriessche, 2015). The approach adopted here is that of classical anthropological fieldwork: participant observation guided by a holistic perspective, attentive to all aspects of social life, including those that may seem remote from sand drawing practice. Throughout, the aim has been to remain as close as possible to the practitioners’ own categories and procedures––an emic commitment that, as we shall see, bore directly on the mathematical modelling itself.

8However, I did encounter legitimate reluctance to share sensitive knowledge of sand drawing. In Vanuatu societies, traditional knowledge is commonly governed by codes of prohibition: the right to perform certain songs (Ammann, 2012) or to produce specific artefacts such as pottery or mats (Bertin, 2019) is protected by intellectual property norms and may be subject to exchange or monetization. As Monika Stern, whose work focuses on musical practices in Vanuatu, observes: “a principle very close to intellectual rights has existed for centuries” across songs, dances, sand drawings, and other intangible knowledge (Stern, 2013). Sand drawings are largely performed in public, yet the knowledge underpinning certain figures remains sensitive in this sense.

9The field sites—the islands of Pentecost, Maewo and Efate—were selected in agreement with the Vanuatu Kaljoral Senta (VKS) and the relevant customary chiefs. The VKS provided local fieldworkers who greatly facilitated my integration into the communities and helped convey the scope of the project and the intended use of the data to the communities. Beyond my own research, I have committed to the ethical principle that all research findings will be shared with the VKS, particularly with a view to their educational implications. Of the three field sites, Pentecost proved the most productive. Following the intercession of Kolombas Todali, my fieldworker in Pentecost, with the chiefs’ council, I was granted authorization to work on sand drawings. I collaborated with at least five expert practitioners—to whom I return below—bringing together a primary corpus of over fifty drawings. Most were recorded using an action camera mounted on a tripod, capturing the full sequence of gestures rather than the sole traces left in the sand. This attention to gesture is not incidental: it lies at the heart of the approach. Drawing on Bloch’s cognitive anthropology, the methodology deployed here rests on the premise that language alone cannot provide full access to practitioners’ knowledge and attends closely to gesture as a complementary and analytically productive avenue of inquiry (Bloch, 2013; 2018). Sand drawing illustrates this point: drawers do not always feel the need to verbalize what they are doing, and an essential part of the knowledge resides in the gesture itself. This ethnographic approach has made it possible to assemble a corpus of data that we now describe in detail.

The Raga uliuli in North Pentecost: preliminary ethnographic considerations

Sand Drawing in Vanuatu: An Overview

  • 6 I will not address this issue but I have demonstrated that the number of possible “paths” for compl (...)
  • 7 The conservation of traditional knowledge, and particularly that related to sand drawing, is the su (...)

10Vanuatu has been an attractive place for anthropologists for more than a century, and prominent figures in the discipline have made it a privileged field site. Indeed, from north to south, this archipelago of 83 islands spanning more than 900 kilometers offers a wide range of cultural variations and anthropological issues. Surprisingly, until recently, the practice of sand drawing was lightly documented, apart from ethnographic studies (Deacon & Wedgwood, 1934; Layard, 1942; Baron, 2020) and books by enlightened enthusiasts (Cabane, 1997; 2016). However, this topic has gained scholarly interest in recent years within different domains of anthropology: anthropology of memory (Baron, 2020), cognitive semiotics (Devylder, 2022), and anthropology of art (Lind, 2018). This ephemeral art—the drawing is swept away once it is finished—stimulates storytelling and is usually concluded with the telling of a myth or a tale encompassing diverse themes such as special places, characters, animals, chieftainship, kinship system, or social organizations. Drawings can be the object of local reappropriation, in which case they can bear completely different names, and very different drawing procedures can be observed6. This traditional graphic art has a “strong mnemonic dimension involved in the recollection of ritual, mythological, and environmental knowledge” (Vandendriessche & Da Silva, 2022; Baron, 2023). Although the simplest drawings are freely circulating (most of them known to children), it is hard to say exactly how many of these drawings are still used today, with new drawings appearing in some societies and disappearing in others.7 The community acknowledges a drawer’s expertise according to their ability to execute or create drawings that cannot be learned merely by reproducing gestures. Only a few practitioners achieve this status, regardless of age: young innovators creating new designs or elders with encyclopedic knowledge.

Sand drawing from North Pentecost

  • 8 Its current name derives from the date of its discovery by French explorer Louis-Antoine de Bougain (...)
  • 9 The other part was carried out in North and South Maewo.

11Pentecost8 Island lies in the central Vanuatu archipelago, situated between Maewo Island and Ambrym. The island extends approximately 60 kilometers from north to south and just over 10 kilometers at its widest point. Five main vernacular languages are spoken across Pentecost, including Raga [Ra-ra] in the north. In this article, all translations of terms in Raga are based on the extensive documentation of this language (Hardacre, 1924; Yoshioka, 1987; Yoshioka & Leona, 1992; Taylor 2008; Vari-Bogiri, 2011; Duhamel, 2020). The term Raga designates both the place and the language, while the society as a whole is referred to as Sia Raga. Part of my fieldwork was carried out in the northern villages of Avatvotu, Loltong, and Labwatgogoru (shown in the green area in Figure 3)9. This region has been the focus of New Zealand anthropologist John Taylor, who conducted a remarkable long-term ethnographic fieldwork (Taylor, 2005; 2008). His detailed documentation of Sia Raga provides essential foundations for the present study. Sections 3 and 4 of this article engage with and place in new perspective certain results from Taylor’s work, which are then discussed in Section 5.

Figure 3. Map of Pentecost Island showing fieldwork sites (in green).

Figure 3. Map of Pentecost Island showing fieldwork sites (in green).

Map reproduced with the permission of CartoGIS Services, Scholarly Information Services, The Australian National University. CC BY SA 4.0 – CartoGIS CAP 18-253_KP.

  • 10 Linguist M-F Duhamel provides me the following interpretation: bwat(u) means “head” and, by semanti (...)

12In Raga, the activity of sand drawing bears the name uliuli, a reduplication of the action verb uli, which means to write, paint (a house), or draw. The word designating a drawing is bwatuli10 and for Sia Raga, they are traces of words of “the land”. Indeed, Chief Ruben Todali (†), a ratahigi (the highest grade in the chieftaincy system of Sia Raga) explained to anthropologist J-P Taylor and later to me in similar words:

He told me that in the past, many centuries before the arrival of tuturani (whites, foreigners like myself), the people of north Pentecost could not speak. They communicated by way of designs that they inscribed on the ground with their fingers. Instead of people, the sentient and mobile rocks and stones were talkative. The dark soil of the hills and valleys also spoke. So too did the winds, the rains, and the salt water that lapped against the sand and coral of the island’s coast. But some time ago this situation reversed, and now it is people who talk, while the land, winds, rain, and sea remain wordless. Nowadays, said Ruben, Sia Raga people sometimes say of their island, we must speak for the land, because the land cannot speak for itself. (Taylor, 2008: 1)

Figure 4. Ratahigi Ruben Todali.

Figure 4. Ratahigi Ruben Todali.

Photo by the author, Pentecost, 2019.

Figure 5. Practitioners from Avatvotu (North-Pentecost), 2019: Left: Edgar Butangi. Right: unknown practitioner.

Figure 5. Practitioners from Avatvotu (North-Pentecost), 2019: Left: Edgar Butangi. Right: unknown practitioner.

Photos by the author, Pentecost, 2019.

  • 11 Unfortunately, I cannot elaborate further on this aspect. It should be noted that this assertion is (...)

13Generally speaking, for Sia Raga, the bwatuli are not merely representations but can be manifestations—ancestral beings or mythic heroes made visible—that carry multifaceted significance.11

14To appreciate how mathematical modeling illuminates this practice, one must examine the rules and constraints that govern it.

Way of tracing a bwatuli and the forgotten rule

15In most cases, a bwatuli begins with a grid drawn on the ground covered with sand or ash (Figure 6). This grid will act as a framework serving as a guide for tracing. Grids manifest in diverse configurations and dimensions (rectangles, rectangular combinations, concentric circles, etc.).

Figure 6. Kolombas Todali performs a drawing on ash-covered ground.

Figure 6. Kolombas Todali performs a drawing on ash-covered ground.

Photo by the author, Pentecost, 2019.

  • 12 This is a mathematical idealization. Since sand drawing is performed without geometric instruments, (...)
  • 13 The term “forgotten rule” is not intended to diminish her foundational contribution, but rather to (...)

16Following grid completion, the expert executes a continuous line using the fingertip. The finger remains in contact with the surface throughout the drawing process, though it may occasionally be raised for decorative elements before resuming from the point of departure. The line travels exclusively from one grid intersection to another, aligning at each node with one of two diagonal trajectories (45° or -45°)12. These directions are represented in Figure 7, the color scheme (red and blue) being maintained throughout. This directional constraint constitutes what I call the forgotten rule, a crucial principle overlooked in Ascher’s analysis introduced earlier13. The finger traverses intersections smoothly without altering its trajectory, and avoids retracing previously drawn segments, although intersection points may be crossed multiple times. Upon completion, the line returns to its initial position following the same trajectory. In most cases (exceeding 90% of the Sia Raga collection), the patterns exhibit symmetrical characteristics.

Figure 7. The line crosses each node along one of two diagonal trajectories without altering direction. Practitioner: Edgar Hinge, a VKS facilitator, drawing: bebe ure, “the world of the butterfly”.

Figure 7. The line crosses each node along one of two diagonal trajectories without altering direction. Practitioner: Edgar Hinge, a VKS facilitator, drawing: bebe ure, “the world of the butterfly”.

Photo by the author, Port Vila, 2019.

A dedicated terminology

17In North Pentecost, to my knowledge, there is no specific term designating the grid itself. However, the vertical lines of the grid are called gaina, a term referring to the vertical posts used in traditional hut construction (Figure 8). The word gai designates both a stick / wood used to cut vegetation and a large stick used in traditional dances. The horizontal lines for their part bear the name Gai Vol(o)volo, which is also used to designate the wooden crossbeams placed perpendicular to the beams. It should be noted that the term volovolo refers to the notion of “cross/crossing” (Hardacre, 1924: 94). Finally, the corners of the grid are named sigu, a term that seems exclusively used to designate the “corner” of a room (in a house), delimited by a right angle.

Figure 8. Vocabulary for the grid.

Figure 8. Vocabulary for the grid.
  • 14 It should be noted that the present analysis does not differentiate between these categories, as th (...)

18A specialized lexicon refers to various patterns and forms, ranging from publicly shared terms to more personal designations used by individual practitioners14 (Table 1).

Table 1. Vernacular terms for patterns.

Name (in raga) Pattern Comments
Gabau – a wing Employed in diverse contexts beyond iconic reference to actual wings
Bul taiwage – to join From bulu (to join) + taiwage (parcel of land). Most direct method for connecting two grid nodes

Gagaravwe –

knot

Borrowed from construction: vine knots used to assemble beams in traditional dwellings (imwa)
Huri – to follow

Polysemous verb meaning “to follow/participate”. Key property: path continues

without deviation when closing the loop

Gai balasi From gai (stick, mark) + balas (to separate). Used to mark passages or delimit restricted areas

Livon gahi –

ant’s teeth

May also be designated by terms referring to butterfly

wings

19This vernacular terminology underscores the significance of movements between grid nodes. Coupled with the directional constraint that governs each transition—that is, the specific orientations in which nodes are approached and departed—these two observations proved crucial for developing an adequate mathematical formalization, as the following section demonstrates.

Graph-theoretic modeling and Sia Raga ontology of cycles

From bwatuli to mathematics: the graph Gmod

  • 15 Contrary to popular belief, the concept of graph was not originally introduced by Euler. The term E (...)
  • 16 This operation, consisting in duplicating vertices, is generally called “vertex-splitting”; it is u (...)
  • 17 The reason why it is not necessary to orient the graph is too long to explain in a footnote; for th (...)

20Given the tracing method of bwatuli, almost all of them can be compared directly with graphs, and more precisely with Eulerian graphs. A graph is said to be Eulerian15 if one can traverse all its edges without repeating any edge while returning to the starting point. This approach was previously developed (Ascher 1988; 1991) but failed to resonate with vernacular conceptions. As mentioned in the Introduction, graphs constitute powerful tools for modeling relationships between entities of any kind. However, by directly identifying grid nodes as graph vertices and curve segments as edges, the nature of the relationship between vertices becomes overlooked or even tautological: “two vertices are related if they are connected to each other”. As we demonstrated in a previous article (Vandendriessche & Da Silva, 2022), a crucial observation emerges: nodes are connected one to another according to both departure and arrival directions. One must therefore shift perspective from considering the graph represented by the grid itself to conceptualizing an abstract graph whose vertices consist of all grid nodes, each paired with one of the two diagonal directions—red for the first and blue for the second (Figure 7). The vertices of this modeling graph denoted Gmod are thus pairs (node, direction)16. To facilitate visualization, I assign distinct colors to each diagonal orientation yielding a bicolored non-oriented graph17 structure. Figures 9 and 10 exemplify this representation. Although visualizing such an abstract graph may present challenges, this formalism offers a crucial analytical tool for our analysis by faithfully encoding the finger’s trajectory through its successive (node, direction) transitions.

Figure 9. Construction of vertices and edges in the graph Gmod. The tracing joins two nodes of the grid, transitioning from a red to a blue orientation. This corresponds to an edge in Gmod between a blue and a red vertex.

Figure 9. Construction of vertices and edges in the graph G        mod. The tracing joins two nodes of the grid, transitioning from a red to a blue orientation. This corresponds to an edge in G        mod between a blue and a red vertex.

Figure 10. On the left: a simplified version of the sand drawing umwa ata Lando.

Figure 10. On the left: a simplified version of the sand drawing umwa ata Lando.

On the right: the corresponding graph Gmod. Each node of the left-side grid gives rise to two vertices in the graph (one red, one blue). The edges are then built following the rules of Figure 9, as illustrated by the edges labeled 4.

21The concept of cycle, which is central to this article, is used from now on in its broadest sense, referring both to what is round in shape and to a series of recurring events. This polysemy is not coincidental, and this concept enabled us to establish connections between our mathematical model and the spatial structures and temporal processes that are fundamental to Sia Raga thought and practice. Cycles thus manifest in three interconnected dimensions: 1. the mathematical notion of cycle in graph theory; 2. the material realization of cycles in bwatuli; and 3. the indigenous conception of cycles. A cycle in a graph is a path that starts and ends at the same vertex without repeating any edges.

Figure 11. The pattern huri allows the tracing to continue without deviation.

Figure 11. The pattern huri allows the tracing to continue without deviation.

Figure 12. A cycle, the starting and arrival point/direction (S, d) and (S', d') are equal.

Figure 12. A cycle, the starting and arrival point/direction (S, d) and (S', d') are equal.

22A cycle in a bwatuli is a pattern that departs from a vertex in a given direction and returns to that same vertex in the same direction, corresponding to a cycle in the graph Gmod (Figure 12). The motivation for this definition, which may not be immediately apparent, will become clear further on. For example, the patterns huri (Figure 11) and livon gahi (Table 1) are cycles. A bwatuli itself can be a cycle, as exemplified by umwa ata Lando (Figure 13). This concept finds profound resonance across multiple dimensions of Sia Raga life and cosmology, where cyclical processes structure temporal, spatial, social, and economic relationships. The imperative of completing the drawing at the starting point and in the same direction exemplifies this cyclical worldview. Agricultural cycles, particularly yam cultivation, structure annual life and stand as one of the pillars of Sia Raga temporal organization, conceived not merely “as a series of substitutions, but rather as layers of renewal” (Taylor, 2008: 101). This temporal cyclicity extends to spatial organization: the cyclic management of garden plots through fallow systems is exemplified by the drawing umwa ata Lando. Remarkably, this drawing is itself a cycle in the sense defined in this article (Figures 13–15).

Figure 13. Umwa ata Lando – “Lando’s fallow fields”.

Figure 13. Umwa ata Lando – “Lando’s fallow fields”.

This is a cycle in the mathematical sense, as is Sia Raga, which refers to the cultivated soil cycle. To help the reader navigate Gmod, the first ten edges have been numbered; the continuation follows by symmetry.

Figure 14. The modeling graph Gmod corresponding to umwa ata Lando.

Figure 14. The modeling graph G        mod corresponding to umwa ata Lando.

Figure 15. The same graph Gmod, redrawn with vertices rearranged but edges unchanged, so as to reduce crossings and make the cycle structure more visible.

Figure 15. The same graph G        mod, redrawn with vertices rearranged but edges unchanged, so as to reduce crossings and make the cycle structure more visible.

23Beyond agricultural and spatial organization, the cyclical principle extends to systems of value and exchange. Pig tusks, manipulated through ancestral processes into spiral shapes (Jolly, 1984; Bedford, 2018), occupy a central place in Sia Raga imagery. Notably, Taylor has suggested that the spiral shape of these artifacts (Figure 16) echoes the way in which “they conceive exchange relations, not simply a repetitive process, but also productive and regenerative across space and time” (Taylor, 2008: 105–06). A pig with full-circle tusks is called livoala, derived from alo (sun), which also means “to circle/encircle,” or “to circulate” (Taylor, 2008: 121). This semantic field aligns directly with our definition of cycles as closed curves that enable further elaboration. The kinship system within Sia Raga provides a final illustration of this cyclical logic. The term huri—which, as already mentioned, designates both the circle in sand drawing and the action “to follow”—also refers to “the husband of my father’s sister” (Figure 17), a pivotal kinship position that ensures lineage continuity through cycles of alliances, marriages, gifts, and ritual or economic support between extended families (Löffler, 1960).

Figure 16. A pig’s tusk from Vanuatu.

Figure 16. A pig’s tusk from Vanuatu.

Author’s illustration, digitally processed.

Figure 17. Circular representations of Sia Raga relationship terminology and the pivotal role of Huri for ego.

Figure 17. Circular representations of Sia Raga relationship terminology and the pivotal role of Huri for ego.

Source: Taylor, 2005: 118, figure 20. Courtesy of the author.

Revealing hidden structures: the analytical power of cycle decomposition

  • 18 I have encoded the model and built an IT tool to perform this analysis. See (Da Silva, 2022: Annex (...)
  • 19 Avato—Endoxyla leucomochla—is the white grub consumed in many parts of Melanesia, notably (...)
  • 20 The realization of this cycle is constrained by an arithmetic condition on grid dimensions (detaile (...)

24The definition of cycles in bwatuli finds its justification in Veblen’s classical theorem on Eulerian graphs. This theorem states that any such graph decomposes into disjoint elementary cycles, where disjoint means that these cycles have no common edges (but can have common vertices) (Veblen, 1912). As Gmod is an Eulerian graph, it can be broken down into such a disjoint union of cycles. Returning from Gmod to bwatuli, this means that each one can be decomposed as such. This result has profound implications: it reveals that cycles are the basic building blocks of bwatuli, since any sand drawing (following the classic tracing rules) can be broken down into such cycles. Such a decomposition can be carried out quite easily by following the drawing procedure. Each time the finger returns to an already-traversed node in the same direction, a cycle can be marked. This approach provides a systematic tool18 for analyzing drawing construction according to the practitioner’s method. The drawing avato19 demonstrates the analytical power of cycle decomposition (Figure 18) by revealing a prominent cycle (numbered 1 in blue) that corresponds to an algorithm executable across various grid sizes20. Variants of this algorithmic procedure appear in drawings from other cultural areas in Vanuatu, particularly in depictions of animals such as turtles and grasshoppers (Da Silva, 2023), suggesting inter-island circulation of drawing procedures. Although this recurrent pattern lacks a vernacular name (to our knowledge), its systematic reappearance across multiple drawings suggests it may function as a foundation for creating new drawings—a creative dimension addressed in the next section.

Figure 18. The drawing avato, “a white grub”. 18a: Avato by Edgar Hinge. 18b: The decomposition into cycles of Edgar Hinge’s method.

Figure 18. The drawing avato, “a white grub”. 18a: Avato by Edgar Hinge. 18b: The decomposition into cycles of Edgar Hinge’s method.

Photo by the author, Port Vila, 2019.

Layers and grafting as Sia Raga cosmological principles

Tagure, vanua, and bwatuli as layers

  • 21 Tagure is the raga word for the sago palm, the leaves of which are used to make roof that (...)

25As Taylor has highlighted, the ethnographic analysis of the cycle as a vernacular concept is deeply interlaced with that of layers. This interweaving is embedded in the raga language through the term hou, as Hardacre’s translation demonstrates (Taylor, 2005: 81). Thus, depending on the context, hou can refer to “in sequence, generation,” to “layer in tagure (salgo palm)”,21 or to “respect/honor” (Hardacre, 1924: 56). The genealogical dimension has been emphasized by Duhamel, whose translations are “ancestry line” or “descent line” (Duhamel, 2020: 76). According to Taylor, this polysemy reveals a distinctively non-linear temporality in which “the vertical [genealogical succession] and lateral [spatial co-presence]—the past and the present—merge into one (diagonal) ‘line’ or totality” (Taylor, 2005: 81). Just as successive layers of tagure form the stratified roof (hou) of a hut, each new generation does not replace but rather overlays the previous one, such that “the relationship between layers is characterized by identification rather than differentiation” (ibid.). Ancestors are thus not relegated to a distant past but remain co-present within the vanua (place), their “place-substance” flowing continuously through the living (ibid.). As illustrated in Figure 19, spaces therefore constitute superimposed “layers of memory that are etched into the landscape” (Taylor, 2008: 135).

Figure 19. Abandoned Imwa (“traditional huts”) that contribute to the layered collective memory of places.

Figure 19. Abandoned Imwa (“traditional huts”) that contribute to the layered collective memory of places.

Photo by the author, Pentecost, 2019.

26This ontology of layers within Sia Raga finds an echo in the mathematical concept of layers that I will now define. If cycles constitute the elementary building blocks of bwatuli, the concept of layers provides a complementary analytical refinement grounded in ethnography. Just as hou (layers of salgo palm) form the stratified roof of a house, I propose first to define layers in a mathematical sense as unions of cycles corresponding to a symmetric portion continuously traced by the practitioner. However, building on the vocabulary of sand drawing grids borrowed from that of house construction (Figure 8), I further stipulate that a layer must traverse all nodes of the grid. This requirement mirrors the architectural principle whereby each layer of thatch must rest upon all the gai (supporting posts) to ensure structural integrity.

Figure 20. A decomposition into cycles of bebe ure, “The world of the butterfly”.

Figure 20. A decomposition into cycles of bebe ure, “The world of the butterfly”.

The upper double red loop drawn by Edgar Hinge (figure 7) has been replaced by a single loop. This change does not alter the overall procedure, and the drawing exists in this version as well.

Grafting cycles to materialize the world

27While creation remains a challenging and multidimensional topic in anthropology that calls for caution (Lavie et al., 1993; McLean, 2009), I conclude by sketching how the model could help address some specific aspects of sand drawing creation. This approach complements existing perspectives that have emphasized the role of dreams and imagination (Baron, 2020: 207, 212–17) by offering new analytical tools to examine the procedural foundations of these drawings. This section focuses here on the drawing bebe ure (Figure 20), whose terminology refers to the world (ure) of the butterfly (bebe or beba). This drawing occupies a central place in Sia Raga cosmogony and imaginary. According to Taylor, Chief Ruben Todali explained that bebe ure is the first animal to emerge on Raga through processes of “incremental growth and transformation” from lumute (moss), which he described as “the basis of all life”. He emphasized that humans themselves are ultimately derived from lumute through this same process of incremental growth and transformation, with the proof of this genealogy retained in the physical anatomy of bebe ure (Taylor, 2008: 79).

  • 22 The grafting operation, which I have analyzed in detail elsewhere (Da Silva, 2026), allows for the (...)

28This cosmogony invites us to question the drawing itself as a material manifestation of such incremental growth. To do so, let us aggregate the cycles (except the red and green ones) into a single pattern (in blue, Figure 21), which corresponds to the first portion traced by Edgar Hinge. Being symmetric and passing through all nodes, it is a layer in the sense defined in this article. The decomposition into cycles reveals that Edgar Hinge’s tracing can be interpreted as a succession of cycles, all grafted on to a main cycle (in violet in Figure 18). More precisely, following Hinge’s procedure, this main cycle can be considered as a framework upon which four huri cycles and two livon gahi cycles are progressively grafted22 during the tracing (Figure 22).

Figure 21. The first step of the drawing corresponds to all cycles from Figure 20 (except the green and red ones) colored here in blue. We obtain three layers executed in blue-green-red order.

Figure 21. The first step of the drawing corresponds to all cycles from Figure 20 (except the green and red ones) colored here in blue. We obtain three layers executed in blue-green-red order.

Figure 22. Reconstruction of the first layer of bebe ure by successively grafting cycles onto a main cycle: 22a. The principle of grafting a huri on to the violet cycle. 22b. Grafting four huri. 22c. Grafting two livon gahi.

Figure 22. Reconstruction of the first layer of bebe ure by successively grafting cycles onto a main cycle: 22a. The principle of grafting a huri on to the violet cycle. 22b. Grafting four huri. 22c. Grafting two livon gahi.

29This interpretation makes the drawing bebe ure a material extension of the cosmogonic principles and the processes of incremental growth and transformation. In this view, the drawing does not represent the cosmogony; it materially realizes it through these very operations. This analysis invites us to step back and consider its broader implications, which are addressed in the following discussion.

Discussion

Cycles and layers as key-structures for Sia Raga

30The decomposition of bwatuli into elementary cycles transforms what appears as a continuous gesture into a structured composition of elementary units. The model presented in this article suggests that formal mathematical frameworks can reveal structural principles that profoundly resonate with fundamental vernacular principles. Strikingly, an abstract mathematical theorem has brought to light categories—cycles and layers—that echo vernacular terminology and practice. In particular, the term huri—designating both the simplest cyclic pattern and a central kinship position—exemplifies this convergence. Cycles and layers are not concepts introduced artificially; they emerge from modeling that is carefully grounded in ethnography. In this respect, they can be considered as a materialization of organizing principles across multiple domains. In particular, the discussion has emphasized agricultural cycles that organize annual life as layers of renewal, spatial management through fallow systems, landscapes as layers of memories or the exchange systems materialized in circular pig tusks (livoala).

Creation as incremental and growing process

31The decomposition of bebe ure (“the world of the butterfly”) into cycles suggests a way to access the creative logics underlying this drawing. This cyclic structure reveals a strong connection with the mythic time of creation in which the butterfly emerged through an incremental growth process. This invites a parallel with graphic art practices across Melanesia and Aboriginal societies that materialize cosmogonic narratives. The wondjina rock paintings of the North Kimberley (Elkin, 1930) provide a notable parallel: for Aboriginal practitioners, these are not representations of the dreamtime (the time of creation) but direct material connections to it (McLean, 2009: 220–21). The same ontological principle applies to the walkala—the sand drawings of Kukatja societies of Gibson Desert—which are “not merely a representation, but a manifestation of the presence and the power of the creative beings” (Poirier, 2013: 156). In both cases, these practices operate not through individual human agency but as channels connecting the present to primordial creative forces. The decomposition of bebe ure into cycles echoes this anthropological analysis, and Figure 22 might materialize the creative power of mythic heroes. Moreover, the three successive layers in Figure 21 invite us to open a research avenue: could Chief Ruben Todali’s reference to “the physical anatomy” of bebe ure be interpreted as metamorphic stages—caterpillar, chrysalis, and butterfly? The existence of another drawing referring to the butterfly, bebe gaga, offers an alternative way to address this question: what distinguishes these two bwatuli referring to the same animal?

Figure 23. Bebe gaga drawn by a child on Laone beach.

Figure 23. Bebe gaga drawn by a child on Laone beach.

Photo by the author, Pentecost, 2019.

From visual to procedural complexity: expanding sand drawing ethnography

32Decomposing bwatuli into cycles yields an important finding: any given drawing admits multiple execution paths. As the number of vertices and edges in Gmod increases, the number of possible decompositions grows exponentially. Moreover, these multiple paths represent fundamentally distinct procedures rather than trivial variations—they do not derive from one another through circular permutation or reversal of the tracing sequence.

33This finding has direct ethnographic significance. It formalizes what practitioners already know: a single drawing can be executed through radically different procedures. As documented in a previous article (Vandendriessche & Da Silva, 2022), practitioners are or have been actively interested in finding out new procedures. Discovering new execution paths enables two interconnected processes: the reappropriation of drawings as they circulate across cultural areas—often accompanied by renaming—and the creative elaboration of entirely new drawings. This resonates with Dan Sperber’s epidemiological model of cultural transmission (Sperber, 1996): what circulates is not a fixed procedure but a family of procedures, transformed and reconstructed as they pass through successive practitioners. The rules of sand drawing function as what Sperber calls “cultural attractors” (ibid.: 111)—stable principles that structure the practice while permitting procedural diversity. This adherence to core principles over time is itself ethnographically significant. In contrast, deviations from these rules constitute equally important ethnographic data, as Eric Vandendriessche demonstrates in his contribution to this special issue. Understanding when and why practitioners depart from these principles illuminates the boundaries of the practice and the social dynamics of transmission.

  • 23 For example, I showed the drawing umwa ata Lando (Figure 13) to Edgar Hinge, who, although original (...)

34These considerations give us an opportunity to discuss the originality of the ethnographic data presented in this article in relation to my predecessor JP Taylor, whose fieldwork on the Raga area provides valuable data for the present article. Taylor mentions the existence of sand drawing (Taylor, 2008: 88, 109, 194) and draws parallels between the labyrinthine lines of sand drawing and the systems of exchange and relationships in Sia Raga (ibid.: 105). Drawing on Gell’s anthropology of art, he establishes a parallel between the visually entangled closed lines of sand drawings and the raga term wasi meaning “closely bound”, “entangled”, or “stuck” (ibid.: 55). His arguments, although highly relevant, are based primarily on visual considerations and rely on iconic arguments, the visual complexity of sand drawing echoing the constraining nature of social obligations. This perspective strives to convey the indigenous viewer’s point of view, whereas the present article focuses on the cognitive level of practitioners. In fact, procedural and visual complexity in sand drawing are two dimensions that can sometimes conflict with each other. Certain drawings that appear highly entangled can be executed through relatively simple procedures for experienced practitioners.23 This tension between visual and procedural complexity would deserve careful attention and could offer a research avenue to better understand how experts are spontaneously designated by the rest of the community. Furthermore, the relationship between mathematical structure and practitioner expertise requires further investigation.

Conclusion

35This article exemplifies the use of graphs beyond their classical anthropological applications in kinship and network analysis. When grounded in the ethnographic observation of gestures and material traces, graph-theoretic modeling can bring into new perspective fundamental organizing and cosmological principles already documented through classical ethnographic methods. A crucial methodological shift made this possible: just as Guilbaud astutely constructed an unconventional graph where vertices represented marriages rather than individuals, we chose to represent (node, direction) pairs as vertices. This proved decisive for applying Veblen’s theorem—a result elaborated in entirely different mathematical contexts. The decomposition of bwatuli into elementary cycles is a finding that converges remarkably with John Taylor’s highlighting of cycles and spirals across multiple Sia Raga domains, including agricultural cycles, spatial organization, and kinship relations. This convergence suggests that certain drawings such as bebe ure may materialize cosmogonic principles, particularly processes of incremental growth and transformation from lumute (moss) to bebe, the butterfly that emerged as the first animal on Raga and from which humans themselves ultimately derive. This approach to reconstructing creative processes is comparable to how archaeologists infer knapping sequences and technical choices from lithic debris, reconstructing gestures and cognitive processes from material evidence. Finally, the method deployed in this article could inspire the study of other ground-tracing practices such as Tamil Nadu kolams or Chokwe sonas in Angola. While these traditions have already attracted the attention of ethnomathematicians (Gerdes 1989; 1990), graph-theoretic approaches adapted to each ethnographic context could renew this research.

36However, it is equally important to recognize that, like all formalization, this modeling approach proceeds from analytical choices. While it has captured structural principles that would likely have remained inaccessible through classic ethnographic methods, it abstracts from certain dimensions. For instance, narrative performance—the stories practitioners weave around drawings—remains beyond the current framework. Indeed, experts typically narrate while drawing, anchoring crucial moments of the story to specific details of the traced form. Moreover, sand drawing unfolds as a dynamic practice, yet temporality remains absent from the formalization. The sequential order in which nodes are crossed might influence performance, particularly given the material constraints of sand or ash. Adding these narrative and temporal dimensions could refine the model and deepen our understanding of the cognitive processes through which Sia Raga practitioners materialize cosmological principles in ephemeral traces—a practice that has yet to reveal all its secrets.

Acknowledgments

37I wish to express my sincere thanks to all the members without whom this article would not have been possible, and especially to the VKS director, Richard Japuneyo, as well as to practitioners Edgar Hinge and Julia Toto for their expertise and commitment. I would also like to express my appreciation to the talented Raga sand drawers: Kolombas Todali, Edgar Butangi, and Lilibeth from the village of Latano. My thanks also go to the Todali family from the village of Avatvotu, and to the Wesom family from Naone village in the Maewo islands.

38I must also add a special mention to the council of chiefs held in Avatvotu in October 2019, which authorized the use of all the collected data for research purposes, including the images featured in this article. I extend a particular and heartfelt thought to the late Chief Ruben Todali, whose exchanges with me—though at times difficult owing to his ageing speech—proved to be of invaluable worth to my research.

Haut de page

Bibliographie

Ammann Raymond, 2012. Sounds of secrets: field notes on ritual music and musical instruments on the Islands of Vanuatu, Zürich/Münster, LIT Verlag Münster.

Ascher Marcia, 1988. “Graphs in Cultures: A Study in Ethnomathematics,” HistoriaMathematica 15 (3), pp. 201–27.

Ascher Marcia, 1991. EthnoMathematics: A Multicultural View of Mathematical Ideas, Brooks/Cole Publishing Company.

Ascher Marcia & Robert Ascher, 1981. Code of the Quipu: A Study in Media, Mathematics, and Culture, Ann Arbor, University of Michigan Press.

Barbosade Almeida Mauro W., 1990. “Symmetry and Entropy: Mathematical Metaphors in the Work of Levi-Strauss”, Current Anthropology 31 (4), pp. 367–85.

Baron Jacopo, 2020. Gardens of the Mind: A Study on Vanuatu Sand-Drawing, PhD, Paris, EHESS.

Baron Jacopo, 2023. “Ephemeral Images, Lasting Memories: The Sand-Drawings of Ambrym Island, Vanuatu”, Res: Anthropology and Aesthetics 79/80, pp. 114–32.

Beauguitte Laurent, 2016. “L’analyse de réseaux en sciences sociales et en histoire”, in R. Letricot, M. Cuxac, M. Uzcategui, & A. Cavaletto (eds.), Le réseau. Usages d’une notion polysémique en sciences humaines et sociales, Louvain-la-Neuve, Presses universitaires de Louvain, pp. 9–24.

Bedford Stuart, 2018. “Modified Canines: Circular Pig’s Tusks in Vanuatu and the Wider Pacific”, in M.C. Langley, M. Litster, D. Wright, & S.K. May (eds.), The Archaeology of Portable Art: Southeast Asian, Pacific, and Australian perspectives, London/New York, Routledge.

Bertin Marion, 2019. “Allier kastom et tabu au musée: gestion et exposition des objets du Vanuatu”, ICOFOM Study Series 47 (1-2), pp. 41-55.

Bloch Maurice, 2013. L’anthropologie et le défi cognitif, Paris, Odile Jacob.

Bloch Maurice, 2018. How we think they think: Anthropological approaches to cognition, memory, and literacy, 2nd ed., New York, Routledge.

Cabane Jean-Pierre, 1997. Ululan : les Sables de La Mémoire, Nouméa, Grain de sable.

Cabane Jean-Pierre, 2016. La Parole des sables. Dessins sur sable du Vanuatu, Port Vila, Alliance française éditions.

Casajus Dominique, 1994. “L’Hypercube est-il bon à penser?”, L’Homme 130, pp. 111–18.

Courrège Philippe, 1965. “Un modèle mathématique des structures élémentaires de parenté”, L’Homme 5 (3-4), pp. 248–90.

Da Silva Alban, 2022. Une étude ethnomathématique du dessin sur le sable du Vanuatu. De l’ethnographie à la modélisation mathématique, regards croisés sur la pratique des Uli-Uli chez les Raga de Nord-Pentecôte, doctoral thesis in history and philosophy of science, Université Paris Cité. Online: https://theses.fr/2022UNIP7005.

Da Silva Alban, 2023. “Les dessins sur le sable du Vanuatu et ‘l’algorithme de la tortue’”, Images des mathématiques 1, pp. 207–36. Online: https://doi.org/10.60868/ra32-jx25.

Da Silva Alban, 2026. “Group actions at the fingertip: A 4E approach to mathematical cognition in Vanuatu sand drawing”, The Journal of Mathematical Behavior 82 (101313). https://doi.org/10.1016/j.jmathb.2025.101313.

Deacon A. Bernard & Camilla Hildegarde Wedgwood, 1934. “Geometrical Drawings from Malekula and Other Islands of the New Hebrides”, with a note by A.C. Haddon, The Journal of the Royal Anthropological Institute of Great Britain and Ireland 64, pp. 129–75.

Devylder Simon, 2022. “The Archipelago of Meaning: Methodological Contributions to the Study of Vanuatu Sand Drawing”, The Australian Journal of Anthropology 33 (2), pp. 279–327.

Duhamel Marie-France, 2020. Variation in Raga - a quantitative and qualitative study of the language of North Pentecost, Vanuatu, PhD, Canberra, Australian National University.

Eades Peter & Candido Ferreira Xavier deMendonça, 1995. “Vertex Splitting and Tension-Free Layout”, in F.-J. Brandenburg, GD ’95. Proceedings of the Symposium on Graph Drawing, Berlin/Heidelberg, Springer, pp. 202–11.

Elkin Adolphus Peter, 1930. “Rock-Paintings of North-West Australia”, Oceania 1 (3), pp. 257–79.

Fabry Lucie, 2023. “Quelles mathématiques pour l’anthropologie de la parenté et de l’alliance?”, Terrain. Anthropologie & Sciences Humaines. Online: https://doi.org/10.4000/terrain.25479.

Gerdes Paulus, 1989. “Reconstruction and Extension of Lost Symmetries: Examples from the Tamil of South India”, Computers & Mathematics with Applications 17 (4), pp. 791–813.

Gerdes Paulus, 1990. “On Mathematical Elements in the Tchokwe ‘Sona’ Tradition”, For the Learning of Mathematics 10 (1), pp. 31–34.

Glowczewski B., 1989. “A topological approach to Australian cosmology and social organization” L'Homme 19 (3), pp. 227–40.

Guilbaud Georges & Claude Lévi-Strauss, 1970. “Système parental et matrimonial au Nord Ambrym”, Journal de la Société des Océanistes 26, pp. 9–32.

Haddad Labib & Yves Sureau, 1993. “Les groupes, les hypergroupes et l’énigme des Murngin: The Murngin Case”, Journal of Pure and Applied Algebra 87 (3), pp. 221–35.

Hage Per, 1979. “Graph Theory as a Structural Model in Cultural Anthropology”, Annual Review of Anthropology 8, pp. 115–36.

Hardacre Miss, 1924. “Dictionary of Raga (North Pentecost) Language”, Unpublished, Port Vila, Vanuatu National Library.

Jolly Margaret, 1984. “The Anatomy of Pig Love: Substance, Spirit and Gender in South Pentecost, Vanuatu”, Canberra Anthropology 7 (1-2), pp. 78–108.

König Dénes, 1990 [1936]. “Theory of Finite and Infinite Graphs”, in D. König (ed.), Theory of Finite and Infinite Graphs, Boston/Basel/Berlin, Birkhäuser, pp. 45–121.

Lavie Smadar, Kirin Narayan & Renato Rosaldo, 1993. Creativity/anthropology, Ithaca, Cornell University Press.

Layard John, 1942. Stone Men of Malekula, London, Chatto & Windus.

Leach Edmund Ronald, 1961. Rethinking Anthropology, London/New York, University of London/The Humanities Press.

Lévi-Strauss Claude, 1954. “Les mathématiques de l’homme”, Bulletin international des sciences sociales 6 (4), pp. 643–53.

Lind Craig, 2018. “The Beauty of Sand-Drawing in Vanuatu: Kinship and Continuity on Paama Island”, in S. Bunn (ed.), Anthropology and Beauty: From Aesthetics to Creativity, London/New York, Routledge, pp. 418–33.

Löffler Lorenz G, 1960. “The Development of the Ambrym and Pentecost Kinship Systems”, Southwestern Journal of Anthropology 16 (4), pp. 442–62.

McLean Stuart, 2009. “Stories and Cosmogonies: Imagining Creativity Beyond ‘Nature’ and ‘Culture’”, Cultural Anthropology 24 (2), pp. 213–45. Online: https://doi.org/10.1111/j.1548-1360.2009.01130.x.

Petitot Jean, 1988. “Approche morphodynamique de la formule canonique du mythe”, L’Homme 106–107, pp. 24–50.

Poirier Sylvie, 2013. “Traces, bâtons à fouir et dessins sur le sable. Chez les Aborigènes du désert occidental australien”, Techniques & Culture 2 (61), pp. 144–65.

Rauff James V., 2016. “The Algebra of Marriage: An Episode in Applied Group Theory”, BSHM Bulletin: Journal of the British Society for the History of Mathematics 31 (3), pp. 230–44.

Selz-Lauriere Marion, 1988. “Les Mathématiques En Ethnologie”, L’Homme 108, pp. 147–55.

Sperber Dan, 1996. La contagion des idées: théorie naturaliste de la culture, Paris, Odile Jacob.

Stern Monika, 2013. “Music in Traditional Exchanges in North Vanuatu”, Pacific Studies 36 (1/2), pp. 59–76.

Taylor John, 2005. “Paths of Relationship, Spirals of Exchange: Imag(in)ing North Pentecost Kinship”, The Australian Journal of Anthropology 16 (1), pp. 76–94.

Taylor John, 2008. The Other Side: Ways of Being and Place in Vanuatu, Honolulu, University of Hawai’i Press.

Vandendriessche Eric, 2015. String figures as mathematics? An anthropological approach to string figure-making in oral tradition societies, Cham, Springer.

Vandendriessche Eric, 2026. “Understanding the underlying system of sand drawing practices: collaborative experiments with Ambrymese experts”, Journal de la Société des Océanistes 162, this issue.

Vandendriessche Eric & Alban Da Silva, 2022. “Les dessins sur le sable du Nord de l’île Ambrym. Une étude ethnomathématique”, Ethnographiques.org 43. Online: www.ethnographiques.org/2022/Vandendriessche_DaSilva.

Vandendriessche Eric & Céline Petit, 2017. “Des prémices d’une anthropologie des pratiques mathématiques à la constitution d’un nouveau champ disciplinaire : l’ethnomathématique”, Revue d’histoire des sciences humaines (31), pp. 189–219.

Vari-Bogiri Hannah, 2011. “Phonology and Morpho-Syntax of Raga, Vanuatu”, PhD, Port Vila, University of the South Pacific.

Veblen Oswald, 1912. “An Application of Modular Equations in Analysis Situs”, Annals of Mathematics 14 (1/4), pp. 86–94.

Weil André, 1979. Œuvres scientifiques/Collected Papers. Volume 1 (1926-1951), Berlin/Heidelberg, Springer.

White Douglas & Ulla Johansen, 2005. Network Analysis and Ethnographic Problems: Process Models of a Turkish Nomad Clan, Lanham, Lexington Books.

Yoshioka Masanori, 1987. “The Story of Raga: A Man’s Ethnography on His Own Society (I) the Origin Myth”, Journal of the Faculty of Liberal Arts 21, pp. 1–66.

Yoshioka Masanori & Richard Leona, 1992. “A vocabulary of the north Raga language: olgeta tok long lanwis blong not Pentekost”, Kindai 72, pp. 1–739.

Haut de page

Notes

1 Throughout his work, Lévi-Strauss continued to invoke mathematics, either by employing its formalisms through the famous “canonical” formula for the structural analysis of myths (Petitot, 1988), or by summoning it for metaphorical purposes (Barbosa de Almeida, 1990).

2 This article directly employs graph theory in its mathematical generality. Thus, readers familiar with the field of network analysis—which encompasses “a set of methods, notions, and concepts based on graph theory for studying a given relational phenomenon” (Beauguitte, 2016)—may be surprised by the unconventional choices made for the vertices and edges of the graphs presented in this article. I thank the anonymous reviewer for this remark.

3 This is not always the case: in some graphs, vertices can be positioned according to GPS coordinates. Moreover, for readability purposes, it is often useful to position vertices in such a way that edge crossings are minimized.

4 Faced with the complexity of the Murngin marriage system (North Australia), Lévi-Strauss engaged in direct collaboration with the mathematician Guilbaud, as attested in the online video at www.canal-u.tv/video/ehess/mathematique_sociale_entretien_avec_georges_theodule_guilbaud.14090. Guilbaud’s model surprisingly revealed an algebraic structure well known to mathematicians as the dihedral group of order 6.

5 Marcia Ascher (1935-2013) was a mathematician whose husband, Robert Ascher, was an anthropologist. They co-wrote many articles and a book, Code of the Quipu: A Study in Media, Mathematics, and Culture (Ascher & Ascher, 1981).

6 I will not address this issue but I have demonstrated that the number of possible “paths” for completing a drawing increases exponentially with the size of the grid (Da Silva, 2022).

7 The conservation of traditional knowledge, and particularly that related to sand drawing, is the subject of a sustainable development policy by the Vanuatu authorities. The Vanuatu Cultural Center of the capital Port Vila plays a key role, offering demonstrations, courses, and festivals that bring together the best experts from each cultural area. See also the introduction to this special issue.

8 Its current name derives from the date of its discovery by French explorer Louis-Antoine de Bougainville on Pentecost Sunday, May 22, 1768.

9 The other part was carried out in North and South Maewo.

10 Linguist M-F Duhamel provides me the following interpretation: bwat(u) means “head” and, by semantic extension, also “foundation/base”. For example, the expression bwatun vanua is used for the original place of a family or clan. The meaning of bwatuli could be “foundational drawings” because they were/are the support of oral tradition.

11 Unfortunately, I cannot elaborate further on this aspect. It should be noted that this assertion is based on comments surrounding several drawings featuring two mythological figures, Bwatmahana and Tagaro, from the islands of Pentecost and Maewo (Taylor, 2008: 94–95). These heroes embody contrasting cosmogonic principles: Bwatmahana, the primordial donor whose sacrificial death creates volcanic islands, and Tagaro, the trickster demiurge who brings cultural knowledge through deception. In particular, during my fieldwork, Jean-Louis Toa from the village of Nasawa in southern Maewo commented on a drawing similar to Figure 5b in these terms: “The line in the drawing is Tagaro jumping from branch to branch in pursuit of Bwatmahana”.

12 This is a mathematical idealization. Since sand drawing is performed without geometric instruments, the actual angles naturally deviate from precisely ±45°. What proves crucial for the analysis is not the exact angular measurement, but rather the consistent distinction between two discrete diagonal orientations.

13 The term “forgotten rule” is not intended to diminish her foundational contribution, but rather to identify a principle that, once recognized, substantially refines her original model.

14 It should be noted that the present analysis does not differentiate between these categories, as the ethnographic investigation did not permit access to such granular distinctions.

15 Contrary to popular belief, the concept of graph was not originally introduced by Euler. The term Eulerian became standard in homage to Euler’s elegant solution to the famous “Königsberg Bridges problem” in 1741, which contained seminal ideas of graph theory.

16 This operation, consisting in duplicating vertices, is generally called “vertex-splitting”; it is used notably to make graph representation more flexible (Eades & De Mendonça, 1995: 202).

17 The reason why it is not necessary to orient the graph is too long to explain in a footnote; for this technical point, see (Da Silva, 2022: 157).

18 I have encoded the model and built an IT tool to perform this analysis. See (Da Silva, 2022: Annex section).

19 Avato—Endoxyla leucomochla—is the white grub consumed in many parts of Melanesia, notably by Sia Raga. A recording of Edgar Hinge performing this drawing is available here: https://vimeo.com/564900487?share=copy&fl=sv&fe=ci.

20 The realization of this cycle is constrained by an arithmetic condition on grid dimensions (detailed discussion appears in Da Silva, 2023).

21 Tagure is the raga word for the sago palm, the leaves of which are used to make roof thatch, as well as for the woven sections of thatch themselves (Taylor, 2005: 92, note 13).

22 The grafting operation, which I have analyzed in detail elsewhere (Da Silva, 2026), allows for the progressive elaboration of complex patterns from simpler cyclic structures without altering the fundamental topology of the base cycle.

23 For example, I showed the drawing umwa ata Lando (Figure 13) to Edgar Hinge, who, although originally from Pentecost, did not know it. He drew it spontaneously without any mistakes and even suggested versions that he considered more complete (interview at VKS, 2019).

Haut de page

Table des illustrations

Titre Figure 1. A graph with six vertices and six edges.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-1.jpg
Fichier image/jpeg, 86k
Titre Figure 2. A directed graph representing a six-class kinship system in North Ambrym.
Crédits Source: Guilbaud & Lévi-Strauss, 1970: 21.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-2.jpg
Fichier image/jpeg, 129k
Titre Figure 3. Map of Pentecost Island showing fieldwork sites (in green).
Crédits Map reproduced with the permission of CartoGIS Services, Scholarly Information Services, The Australian National University. CC BY SA 4.0 – CartoGIS CAP 18-253_KP.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-3.jpg
Fichier image/jpeg, 473k
Titre Figure 4. Ratahigi Ruben Todali.
Crédits Photo by the author, Pentecost, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-4.jpg
Fichier image/jpeg, 505k
Titre Figure 5. Practitioners from Avatvotu (North-Pentecost), 2019: Left: Edgar Butangi. Right: unknown practitioner.
Crédits Photos by the author, Pentecost, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-5.jpg
Fichier image/jpeg, 361k
Titre Figure 6. Kolombas Todali performs a drawing on ash-covered ground.
Crédits Photo by the author, Pentecost, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-6.jpg
Fichier image/jpeg, 814k
Titre Figure 7. The line crosses each node along one of two diagonal trajectories without altering direction. Practitioner: Edgar Hinge, a VKS facilitator, drawing: bebe ure, “the world of the butterfly”.
Crédits Photo by the author, Port Vila, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-7.jpg
Fichier image/jpeg, 312k
Titre Figure 8. Vocabulary for the grid.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-8.jpg
Fichier image/jpeg, 60k
Titre Figure 9. Construction of vertices and edges in the graph Gmod. The tracing joins two nodes of the grid, transitioning from a red to a blue orientation. This corresponds to an edge in Gmod between a blue and a red vertex.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-9.jpg
Fichier image/jpeg, 100k
Titre Figure 10. On the left: a simplified version of the sand drawing umwa ata Lando.
Légende On the right: the corresponding graph Gmod. Each node of the left-side grid gives rise to two vertices in the graph (one red, one blue). The edges are then built following the rules of Figure 9, as illustrated by the edges labeled 4.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-10.jpg
Fichier image/jpeg, 95k
Titre Figure 11. The pattern huri allows the tracing to continue without deviation.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-11.jpg
Fichier image/jpeg, 159k
Titre Figure 12. A cycle, the starting and arrival point/direction (S, d) and (S', d') are equal.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-12.jpg
Fichier image/jpeg, 180k
Titre Figure 13. Umwa ata Lando – “Lando’s fallow fields”.
Légende This is a cycle in the mathematical sense, as is Sia Raga, which refers to the cultivated soil cycle. To help the reader navigate Gmod, the first ten edges have been numbered; the continuation follows by symmetry.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-13.jpg
Fichier image/jpeg, 390k
Titre Figure 14. The modeling graph Gmod corresponding to umwa ata Lando.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-14.jpg
Fichier image/jpeg, 143k
Titre Figure 15. The same graph Gmod, redrawn with vertices rearranged but edges unchanged, so as to reduce crossings and make the cycle structure more visible.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-15.jpg
Fichier image/jpeg, 113k
Titre Figure 16. A pig’s tusk from Vanuatu.
Crédits Author’s illustration, digitally processed.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-16.jpg
Fichier image/jpeg, 256k
Titre Figure 17. Circular representations of Sia Raga relationship terminology and the pivotal role of Huri for ego.
Crédits Source: Taylor, 2005: 118, figure 20. Courtesy of the author.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-17.jpg
Fichier image/jpeg, 184k
Titre Figure 18. The drawing avato, “a white grub”. 18a: Avato by Edgar Hinge. 18b: The decomposition into cycles of Edgar Hinge’s method.
Crédits Photo by the author, Port Vila, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-18.jpg
Fichier image/jpeg, 178k
Titre Figure 19. Abandoned Imwa (“traditional huts”) that contribute to the layered collective memory of places.
Crédits Photo by the author, Pentecost, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-19.jpg
Fichier image/jpeg, 427k
Titre Figure 20. A decomposition into cycles of bebe ure, “The world of the butterfly”.
Légende The upper double red loop drawn by Edgar Hinge (figure 7) has been replaced by a single loop. This change does not alter the overall procedure, and the drawing exists in this version as well.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-20.jpg
Fichier image/jpeg, 294k
Titre Figure 21. The first step of the drawing corresponds to all cycles from Figure 20 (except the green and red ones) colored here in blue. We obtain three layers executed in blue-green-red order.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-21.jpg
Fichier image/jpeg, 316k
Titre Figure 22. Reconstruction of the first layer of bebe ure by successively grafting cycles onto a main cycle: 22a. The principle of grafting a huri on to the violet cycle. 22b. Grafting four huri. 22c. Grafting two livon gahi.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-22.jpg
Fichier image/jpeg, 110k
Titre Figure 23. Bebe gaga drawn by a child on Laone beach.
Crédits Photo by the author, Pentecost, 2019.
URL https://journals.openedition.org/jso/docannexe/image/18141/img-23.jpg
Fichier image/jpeg, 412k
Haut de page

Pour citer cet article

Référence papier

Alban Da Silva, « Embodied and materialized cosmological principles: a graph-theoretic approach to cycles and layers in North Pentecost Sia Raga sand drawing », Journal de la Société des Océanistes, 162 | 2026, 118-144.

Référence électronique

Alban Da Silva, « Embodied and materialized cosmological principles: a graph-theoretic approach to cycles and layers in North Pentecost Sia Raga sand drawing », Journal de la Société des Océanistes [En ligne], 162 | 2026, mis en ligne le 30 septembre 2026, consulté le 12 novembre 2026. URL : https://journals.openedition.org/jso/18141 ; DOI : https://doi.org/10.4000/16uzw

Haut de page

Auteur

Alban Da Silva

Department of Innovation in Medical Education (DIEM), Faculty of Medicine, Laval University, Canada

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search