Introduction

The Rising Star cave system, South Africa, is located within a small promontory situated to the south and east of the course of the Blaaubankspruit stream. The cave system is situated within the dolomitic limestone of the Malmani Subgroup, a Precambrian marine rock bedded with chert bands and containing abundant stromatolite fossils (Dirks et al. 2015; Eriksson et al. 2006). The system includes more than 3 km of mapped passages comprising multiple levels within a west-dipping dolomite horizon. Abundant remains of Homo naledi (Berger et al., 2015) occur within several localities in the system, including the Dinaledi Subsystem, which lies at a depth of ∼30 m below the present surface and ∼120 m through the cave system from the nearest present entrance (Hawks et al., 2017; Elliott et al., 2021). Remains of H. naledi have been recovered and excavated from the Dinaledi Chamber, Hill Antechamber, and adjacent spaces and fissures (Berger et al. 2015; Berger et al. 2025; Brophy et al. 2021). These spaces are challenging to enter and navigate, and exploration of them is ongoing (Berger et al. 2025, Elliott et al. 2021).

On July 28, 2022, during a survey of the Dinaledi Subsystem, we identified what appear to be engraved markings on the southern and northern faces of a natural pillar that forms the entrance and exit of a narrow passage connecting the Hill Antechamber with the Dinaledi Chamber (Figure 1). The marks include linear features between ∼5 and ∼15 cm in length. Many intersect to form geometric patterns such as squares, triangles, crosses, and X’s, while some are isolated lines. These markings are located on three dolomitic panels on otherwise smooth areas, as well as in and around existing cracks and grooves in the rock. They occur in three areas which we have labelled A, B and C. These markings are in a location where they can be viewed during access into and egress from the Dinaledi Chamber when entering the system from the Hill Antechamber. The Hill Antechamber is the likely point of access by Homo naledi to the entire subsystem, and the passage is the natural linkage between the two main chambers of the subsystem (See Figure 1, also Elliott et al. 2021).

Figure 1

In this paper, we describe our observations of Panel A within the passage linking the two main chambers. We provide images of Panels B and C within the Hill Antechamber and offer suggestions of their contextual relationship with Panel A, while recognizing that identifying and examining any all potential engraved lines within these panels requires further study. We also provide additional contextual data supporting the attribution of these markings to H. naledi, and discuss the time sequence represented by engraved marks on Panel A. We end by discussing potential implications of our findings for H. naledi culture and cognition. The description and initial analyses provided here is intended to document the discovery and provide spatial and contextual information to the scientific and heritage community prior to any further analyses that may require invasive sampling as such may require destructive methods or interacting with the marks directly, and therefore have associated ethical implications that need to be carefully considered before being undertaken (e.g. Bednarik, 2008; Arthur et al., 2021; Zerboni et al., 2022).

Panel A

What we term Panel A is found on the southern face of the natural pillar that forms the southern edge of the entry from the Hill Antechamber into the southern of two passages leading to the Dinaledi Chamber (Figures 1, 2). The panel is notable as an area of discoloured rock that appears to have been smoothed by percussive blows by a hard object, as is evidenced by micro and macro pitting of the surface not found on the adjacent natural rock surfaces (Figures 3 and 4) and by the possible application of sand and grit both before and after markings were made (Figures 5 and 6).

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

The most visible markings are engraved lines, some of which when viewed together are crosshatched and give the impression of a rough hashtag figure (Figures 10 and 11). The lines appear to have been made by repeatedly and carefully passing a pointed or sharp incising tool into existing or new grooves. In addition, there are markings that are best described as scratches that fall outside of identifiable lines and their intersections. Several of the grooves overlap geological features native to the rock including fossil stromatolites (Figures 12 and 13). In many instances, we suggest it is possible to identify which lines were made first by examining the point where they cross another line (e.g. Figures 14 and 15).

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

We identified at least 46 engraved or altered marks we consider artificial on panel A (See Figure 16). In this number we include features that appear to be natural lines that have been altered by etching. The most prominent markings on Panel A are a series of intersecting lines (Figures 10,11 and 12). There appears to be a temporal span involved in the creation of the engraved lines as some seem more recently engraved and show clean etching, while others have been obscured either by slight weathering or by the application of sediment. The most easily identifiable engravings, based on their clarity, are Lines L2, L6, L9, L11, L16, L17, L27, L30 and L31 (Figure 16). While the existing lines may have been created in older etchings, or been created over multiple interactions, the final etchings of the lines based on which lines overlap we interpret as follows: horizontal Lines 11 & 25 were created after vertical Line 2. Vertical Line 6 was created after L11. Vertical Line 18 was created after horizontal Line 17. Line 30 was created after horizontal Line 25, but before horizontal Line 31.

Figure 16

Evidence of hominin manufacture of engravings on Panel A

Dolomite is known for a pattern of natural weathering that results in patterns of recessed linear features on its surface often called “elephant skin” patterning. Natural fissures and erosional features in weathered dolomite surfaces due to chemical weathering processes (Burger, 1989) are in our observations characteristically deeper than several millimeters and they follow natural fracture planes within the rock. Natural erosional features in dolomite may have variable cross-sections, ranging from bevelled to U-shaped to rectangular in cross section with rounded or flat bottoms (Shah, Jeelani and Goldscheider, 2018) and are typically a few millimeters to many centimetres in depth, but do not have multiple parallel striations visible within them which would be characteristic of retouching or engraving. As opposed to natural surface weathering, artificial lines are limited in depth and extent due to the natural hardness of dolomite. This hardness (see Tabor, 1954) means that any substantial artificial marking requires multiple parallel incisions with a hard tool. Where artificial engraved markings intersect, they often appear to exhibit an ordering in which one was completed before the other (see for example figures 6, 13, 14 and 15); this appearance of sequencing is not typical of natural weathering. In previous work, researchers have noted the limited depth of artificial lines, their composition from multiple parallel striations, and their association into a clear arrangement or pattern as evidence of hominin manufacture (Rodríguez-Vidal et al. 2014).

The engraved lines in Panel A diverge in their patterning from the natural “elephant skin” weathering of the surrounding dolomite walls, which can be seen adjacent to the panel within 20 cm of the nearest artificial marks (Figure 18). Our assessment of the features produced by natural weathering in this area is that they are deeper than 10 mm, in particular deep relative to the feature width, they maintain a consistency of size and depth across substantially undulating or rugged surfaces, they expand from natural cracks and fissures. In contrast, even the widest of the engraved lines that constitute Panel A have a relatively shallow depth. High-resolution macro-photography shows micro-striations constituting several of these engraved lines, in which roughly parallel incisions sometimes overlap with each other (See Figures 6, 13, 14, 15 and 17). Many of the lines also fall out of the direction of natural fracture features in the dolomite, although it should be recognised that there are multiple places on this panel where natural lines and features of the rock may have been enhanced or extended by artificial engraving. Figures 13, 14, and 15 show examples of ordering where engraved lines intersect, one having been completed clearly before the other. Some of the elements we describe as engraved lines appear to be artificial extensions or alterations of lines that may originally have resulted from natural weathering.

Figure 17

Figure 18

In addition to the engraving depth, composition, and ordering, two additional aspects of Panel A markings distinguish them from natural weathering. The dolomitic bedrock of the Malmani Formation includes fossil stromatolites, which manifest as curving linear banded striations visible in the rock. Panel A includes these layered stromatolitic bands, and all engraved lines that pass below the bottom of Line 14 cross over this fossil feature (See Figure s13 and 15). Where engraved lines cross over this feature, they retain direction and in some cases the multiple striations slightly diverge, suggesting that maintaining a linear engraving over this irregular surface may have been challenging. Second, the markings are, in places, covered wholly or partially in sediment or some other substance. This coating does not occur in other areas of the chambers where there are no such engravings. This material (we cannot determine if it is sediment or pigment) may have been used either to create visual contrast on the grey dolomite, to abrade the surface as a form of polish, to enhance or obscure some aspects of the engraved lines, or all of these. This material is present on the surface as a thin layer and is evident within some of the grooves of the lines, indicating its application after some of the marks were made. The appearance of time-ordering between engraved lines and the surface treatment may imply an origin of the engravings in multiple episodes (Figure 6).Thus it does not appear this covering, likely sediment, can be explained by geological or other non-organic processes.

The means of manufacture of engraved lines in this context would have required an implement of equal or greater hardness compared to the native dolomitic limestone. Dolomite rocks of appropriate size and morphology to mark the cave walls have been recovered from surface contexts within the Dinaledi Subsystem, as have many chert fragments. At present, only one possible lithic artifact has been described in direct association with H. naledi remains (Berger et al. 2025). This “tool-shaped rock” does resemble tools from other contexts of more recent age in southern Africa, such as a silcrete artifact with linear engraved features on it that was recovered from Blombos Cave (Henshilwood et al. 2018) (Figure 19).

Figure 19

Panels B and C

Panels B and C are located on the northern wall within two meters of the Hill Antechamber burial feature described in Berger et al. (2023). Panel B is situated lower and to the right (West) of Panel C. Both panels appear to have been prepared in a similar way to Panel A, with possible use of cave sediment applied to the surface, giving the surfaces of these panels an obvious textural difference to adjacent walls of the chambers (Figures 2c & b and Figure 5) A number of etchings and engravings can be seen, some in the form of geometric figures, crosses, X’s and one possible non-linear geometric figure (Figure 5). It appears, in softer visible light, that a foreign substance has been applied to part of the panel. As was noted the purpose of this paper is not to describe these complex panels, but to simply note their presence in the Hill Antechamber. Future work in this difficult space is planned to sample the possible residues and map the non-natural etchings, attempt to date the etchings and we will conduct experimental work on native dolomite in controlled experiments.

Discussion

The context of these markings within the remote Dinaledi Subsystem, far from cave entrances and challenging to access (Dirks et al, 2015; Kruger et al., 2016; Elliott et al; 2021), suggests that the most parsimonious explanation for their manufacture is that Homo naledi produced them (Berger et al., 2015; Berger et al., 2024). This is the only species that left any evidence of presence in this subsystem until the entry of modern cave explorers. In this discussion we consider possible objections to this hypothesis and ways in which this hypothesis may be tested in the future. We also suggest some comparisons with the broader record of ancient markings beyond the Rising Star examples.

One possible objection is that some markings were not produced by hominins. For example, some have suggested that some of the apparent markings are consistent with weathered dolomite, while others may have been produced by non-hominin agents such as carnivores (Martinón-Torres et al., 2024). Linear features and other modifications of rock surfaces are sometimes produced by animal claws due to tunnelling, climbing, or claw-sharpening, or other behaviors (Bednarik, 1991). Such features have often been observed within caves formed within relatively soft limestone formations, including those in some parts of Europe and southern Australia.

Deeply incised claw markings are sometimes seen, which tend to include multiple parallel incisions with a spacing typical of the anatomical spacing between claws. When claw marks are made repeatedly on the same surface, they may cross each other or exhibit complex geometries. However, the susceptibility of a surface to marking by animal claws is determined by the hardness of the material. Animal claws have a hardness of 2 to 2.5 on Mohs hardness scale (Rothschild et al., 2013). Dolomite has a hardness of 3.5 to 4.5, with Malmani dolomites at the higher end of this scale, much harder than non-dolomitic limestones with hardness of 2 to 3 which are commonly marked by animal claws. None of the cave systems in the Malmani dolomite have been noted for incised claw markings into dolomite itself of the depth and form seen in the Dinaledi marks (Figure 17), even though clawed species like porcupines, honey badgers, and leopards do use caves and do dig into softer sediments found in the area. The hardness of the rock does not admit scratches by animal claws of this kind; additionally the specific features of Panel A do not exhibit the kind of spacing or morphology that would be expected for repeated scratching by animal claws, and their position well above the cave floor would not lend itself naturally to scratching in that location. Further, there is no evidence of carnivore presence within the subsystem: no carnivore remains, no evidence of carnivore burrowing activity, no coprolites, and no bones with tooth marks attributable to carnivores (Dirks et al., 2015; Dirks et al., 2016).

Another possible objection is that the engravings may be of comparatively recent origin, potentially made by modern cavers or other H. sapiens groups in Holocene or Late Pleistocene times (Herries, 2023; Petraglia et al., 2023; Martinón-Torres et al., 2024; Pettitt and Wood, 2024). This idea has two parts: one concerns the idea that other species or populations besides Homo naledi were active within this subsystem; the other concerns chronological placement of the markings relative to Homo naledi. There is no evidence of any hominin activity in the Dinaledi Subsystem other than H. naledi. No remains attributable to other prehistoric hominin groups occur in this subsystem or any adjacent areas of the cave system, nor is there any other evidence of activity of such groups (Dirks et al., 2015; Dirks et al., 2016). Human cave explorers entered these areas only within the last 40 years, and the number of modern cavers and archaeologists who have entered the Dinaledi Subsystem is extremely limited (Table 1). There is no evidence of modern cavers altering cave walls in such a manner in the Dinaledi Subsystem, or elsewhere in Rising Star system, and we have detected no evidence of traces of metallic tools in association with the markings. By contrast abundant remains of H. naledi have been identified within the subsystem including the Hill Antechamber Feature only three to four meters away (Berger et al., 2024). Nevertheless, the nature and composition of the tools that made these engravings are an area that can be tested with future work.

Table 1

It is very challenging to establish the chronological age of engraved markings on rock surfaces (Van Riet Lowe, 1968; Butzer et al., 1979; Bednarik, 2002,2008; Whitley, 2012). Successful use of geochronological methods may occur when some or all of the markings were buried within sediment that can be dated (Rodriguez-Vidal et al. 2014), or when markings were later covered in part by mineral formations that can be dated using radiometric methods (Hoffmann et al., 2016), or when exposure to weathering creates a varnish that can be dated (Dom and Nobbs, 1992; Dragovich, 2000; Mason, 2006; Pillans and Faifield, 2013). Still, even where geochronological methods are possible, they may provide only a minimum age, which may be thousands of years after the marks were produced. The markings that we have identified within the Dinaledi Subsystem are not overlain by naturally forming sedimentary deposition. Without further excavation of sediments that adjoin cave walls we cannot say whether additional markings may exist in subsurface contexts. No obvious mineral deposition overlaps the engraved features. Our observations do note that there may be materials added to the surface of the engraved areas, and these may provide an avenue of investigation. The situation as a whole makes it challenging to build alternative lines of evidence about the age of the engravings beyond their spatial proximity to the Homo naledi evidence from only a few meters away (Berger et al. 2023).

Thus, at present the H. naledi skeletal material provides the only information about the chronology of hominin activity in this area of the cave system. The maximum age constraint reported by Dirks et al. (2017) on H. naledi skeletal material (335 kyr BP) in Dinaledi is the highest 95% confidence limit of a direct ESR-US date on H. naledi teeth; while the minimum age constraint (241 kyr BP) is based on U-Th on a flowstone that formed in part around a bone fragment (Dirks et al., 2017; Robbins et al., 2021). These dates provide strong constraints on some H. naledi skeletal material from the Dinaledi Subsystem. Geochronological methods do not yet provide similar constraints for H. naledi material from other parts of the cave system, including the Lesedi Chamber (Hawks et al. 2017). Therefore, we cannot be certain that the constraints of the Dinaledi Subsystem evidence mark the full duration of H. naledi cultural activity across the cave system. However, we adopt as a conservative hypothesis that the age of these markings is constrained in the same way as the skeletal remains from the Dinaledi Subsystem, between 335 ka and 241 ka.

Stylistic criteria are also sometimes used to establish the chronology of cave art, although such comparisons should be accepted cautiously. In this case, the images are not typical of contemporary graffiti made by recent people. They are also not typical of rock art of the last several thousand years in southern Africa. The markings are comparable to other hominin-produced markings, pigment use, and curation of nonutilitarian objects from the later Middle Pleistocene of southern Africa. Small slabs of ironstone with incised lines were excavated from Wonderwerk Cave, with a minimum age of 276 kyr BP (Beaumont and Vogel 2006; Beaumont and Bednarik 2013; Chazan and Horwitz 2009; Jacobson et al. 2012). One of these slabs bears a series of seven roughly parallel engraved lines, while crossed lines occur on other slabs (Beaumont and Bednarik 2013; Chazan and Horwitz 2009). Beaumont and Bednarik (2015) proposed that two sites in the southeastern Kalahari with open air rock engravings, Potholes Hoek and Nchwaneng, have highly eroded cupules dating to the early Middle Stone Age, possibly as early as 400 kyr. Small pebbles of quartz, chalcedony, and other materials occur within the Wonderwerk sequence and have been interpreted as manuports (Beaumont and Vogel 2006; Beaumont and Bednarik 2013), although an alternative hypothesis that they were ostrich gastroliths, potentially introduced by hominins who hunted the birds, may also be possible (Tryon 2010). Mineral pigments including specularite and hematite have been found in a number of contexts dating to earlier than 300 kyr, including Kathu Pan 1, Wonderwerk, and Canteen Kopje (Watts et al. 2016). None of these examples of hominin-produced markings, pigment use, or other activities have been found near hominin fossil remains. Some of them predate any evidence of modern humans or early Homo sapiens anywhere on the continent.

The Dinaledi markings are comparable to some of the incised lines and engravings found at other sites in the later Middle and Late Pleistocene. Panel A gives the impression of overlapping crosses and lines that are remarkably similar in appearance to the engraving from Gorham’s Cave, Gibraltar (Rodriguez-Vidal et al. 2014). This engraving was dated to greater than 39 kyr cal BP and has been attributed to Neandertals. In addition to the incised slabs and cupules mentioned above, the later Pleistocene record of southern Africa includes engraved bones and engraved ochre blocks from Blombos Cave (d’Errico et al. 2001; Henshilwood et al. 2002; Henshilwood et al. 2018) and lines impressed within sand features that were later lithified into aeolianites (Helm et al. 2021). There are also a few other engravings from sites in Europe at similar time depth (Von Petzinger, 2017; Kissel and Fuentes 2017, 2018), as well as geometric lines on a freshwater mussel shell from Trinil, Java, attributed to H. erectus based on geochronological considerations (Joordens et al. 2015). The markings in the Dinaledi Subsystem share similarities with many of these geometric expressions from other sites and geographic regions. Several geometric forms identified by Von Petzinger (2017) are present, including crosshatch, cruciform, line, flabellifrom, scalariform, open angle and oval. However, further analytic and comparative work must be conducted to confirm exactly how much similarity and overlap there is between the proposed Dinaledi engravings and the engravings at other Pleistocene sites where such designs have been found.

The precise location of the engraved markings within the Dinaledi Subsystem may be important to their interpretation. Many artifacts and rock walls from later Pleistocene sites with engraved marks appear to have a nonrandom placement of markings on an object or surface. Henshilwood and Dubreuil (2011) have suggested that one should be less focused on the specifics of the designs and rather concentrate on the underlying cause of their creation. Those and other authors have suggested symbolic implications for such engravings, under the assumption that they were made by Homo sapiens. The broader record of engravings and other forms of meaning-making behavior in a range of hominin species and populations over the latter portions of the Pleistocene (Kissel and Fuentes 2018, 2021) suggests that such activity, be it “symbolic” or not, was not exclusive to Homo sapiens. The engravings reported here add to this growing dataset.

Methods

As the main purpose of this work was documentation of the presence of engravings in proximity to the burials, for this study, the etchings and engraving markings were only examined using high resolution photography and magnification of lines and markings. Polarizing filters were also used to enhance relief and this is indicated when used.

Cross-polarisation was employed for control of specular highlights/reflections in order to limit artefacts when generating the 3D-depth map for photogrammetry purposes. A circular polariser was used on the camera lens in conjunction with a linear polarising gel placed over the two speed lights (electronic flash heads) used as the light source. The different minerals/material on the dolomite are reflecting/absorbing the cross-polarised light emphasising the “bright” striations visible in images.

Images were shot with a 50mm (Polariser fitted) at f/11 unless otherwise stated.

The light source used (twin speed lights with polarised gel attached) were placed as close to the Lens axis as possible so that the angles of incidence approximate the reflected angles limiting shadow. This assisted us in building the 3D mesh for photogrammetry purposes. The cross-polarisation also removed specular highlights that create artefacts.

We used Metashape 1.8.1 (Agisoft, Inc.) to generate three-dimensional models of panels A and B based on photographs taken with the parameters reported above. Generation of cross-sections and measurements from these models were performed with MeshLab 2021.20. Resolution of the three-dimensional surface is estimated to be accurate to 0.2 mm.

Acknowledgements

Permits to conduct research in the Rising Star cave system are provided by the South African National Research Foundation (LRB). Permission to work in the Rising Star cave is given by the LRB Foundation for Research and Exploration. The Authors would like to acknowledge the funders of the various expeditions and documentation of the engravings including the National Geographic Society (LRB), the Lyda Hill Foundation (LRB) and the National Research Foundation of South Africa (LRB). Laboratory work and travel was funded by the National Geographic Society (LRB), the Lyda Hill Foundation (LRB), the Fulbright Scholar Program (JH), the University of Wisconsin (JH) and Princeton University (AF).