Confined to the dry interior of peninsular India, Neolithic ashmounds have been controversial features since their first description (Foote Reference Foote1887). Ashmounds, which can reach up to 10 m in height, are formed of stratified layers of vitrified cow dung as well as other material. Ongoing discussion of these dramatic and enduring landscape features, however, focuses primarily on a limited sample of the larger and best-known features, with little analysis of formation processes. Newly documented ashmound deposits at the site of Brahmagiri have the potential to significantly expand our understanding of the role of ashmounds in the Neolithic of southern India. In this article, we report on new work, using Electrical Resistance Tomography (ERT) to document the extent, shape, and depth of buried ashmounds—evidence crucial for understanding their depositional histories. This less-commonly used technique is cost effective and portable, providing a viable alternative to other forms of remote sensing.
Ashmounds and Ashmound Research
Ashmounds were described as early as 1887 (Foote Reference Foote1887, Reference Foote1916) and are found in a limited region of southern India. The term “ash” derives from early ideas about their role as funeral pyres (Morrison Reference Morrison2016); the more apropos “cinder mound” initially proposed by Foote never caught on. Intensive discussions of ashmound function and meaning (e.g., Allchin Reference Allchin1963; Johansen Reference Johansen2004; Paddayya Reference Paddayya1991, Reference Paddayya1998) have dwarfed consideration of their formation histories, with most scholars focusing on ashmounds in their completed form and seeing them as prominent landscape features, if not monuments (Johansen Reference Johansen2004). In contrast, Paddayya (Reference Paddayya2019a:50) has recently highlighted the work of Allchin, noting that “his detailed excavations at the site of Utnur in the Mahbubnagar district of Telangana confirmed its character as a cattle-pen by producing evidence of multiple stages of wooden stockade preparations . . . cattle occupation [hoof impressions], dung accumulation and burning, and ash formation.” Rajala et alia (Reference Rajala, Madella, Korisettar, Niccolucci and Hermon2010) raise important unresolved issues of mobility and seasonality associated with herding and ashmound formation.
The ongoing importance of cattle and their dung in contemporary India has inspired many scholars to posit ritual motivations for ashmound construction (e.g., Boivin Reference Boivin2004; Boivin et al. Reference Boivin, Korisettar and Fuller2005, Reference Boivin, Fuller, Korisettar and Petraglia2008), though Paddayya (Reference Paddayya2019a:57–59) also adds to this additional utilitarian factors, such as the value of burning to reduce dung volume, the “clean and hygienic appearance of burnt surfaces” and, with reference to a contemporary festival in which cattle are led though dung fires, potential veterinary hygienic effects.
Excavated ashmounds at Utnur, Piklihal, Palavoy, and Budihal (Allchin Reference Allchin1961; Mujumdar and Rajaguru Reference Mujumdar and Rajaguru1966; Paddayya Reference Paddayya1998, Reference Paddayya2019b; Reddy Reference Reddy1976), among others, show complex stratigraphic histories with both thick and thin layers of fired dung and other material, indicating multiple episodes of deposition and burning. Ashmounds are found both in isolation—without evidence for permanent settlement—and in association with Neolithic villages, and they are restricted to the semiarid interior of peninsular India (Paddayya Reference Paddayya2019b). Allchin (Reference Allchin1961, Reference Allchin1963) reported the presence of “vallums” (walls or enclosures) of ashmound material at some sites, but most or all of these seem to have been destroyed by the late twentieth century (e.g., Morrison Reference Morrison2009). Foote (Reference Foote1916) also noted vallums of ashmound material, referring to them as “cinder camps.” These “walls” (Allchin Reference Allchin1961:69) are interesting as examples of ashmound material formed into nonmounded shapes, but for the most part, ashmounds have been conceived as isolated hills or mounds.
In a recent reevaluation of ashmound chronology, Fuller et alia (Reference Fuller, Boivin and Korisettar2007) argue that ashmound construction typically took place over a relatively short period of time—100 to 200 years, only a few human generations—with specific construction dates varying by site but generally occurring between 2500 and 1250 BC. Their arguments are based primarily on their own excavations of two ashmounds, Sannarachamma and Hiregudda, and on Bayesian modeling of existing dates from several other sites, most notably Budihal (Paddayya Reference Paddayya1998). Although Paddayya (Reference Paddayya1998) saw Neolithic settlement at Budihal as contemporaneous with ashmounds there, reanalysis of the Budihal dates suggests that the ashmound may predate those settlement deposits, as they also did at Sannarachamma. Fuller et alia (Reference Fuller, Boivin and Korisettar2007) argued that (later) Neolithic settlements were attracted by (earlier) Neolithic ashmounds, suggesting that ashmound construction predated the establishment of permanent villages across the Southern Neolithic. This argument is based on a small sample, given that most other excavated ashmounds are not well dated.
The connection between ashmounds and settlements has been widely disputed (Allchin Reference Allchin1963; Fuller et al. Reference Fuller, Boivin and Korisettar2007; Paddayya Reference Paddayya1991, Reference Paddayya2019b), as has the very nature of Southern Neolithic land use, with interpretations ranging from highly mobile pastoralists practicing some farming, to agropastoralism, to settled village agriculturalists keeping large flocks (which possibly grazed at some distance from the village). Although we should certainly expect strategies to vary across time and over the large region covered by Southern Neolithic sites, few studies have focused specifically on seasonality and mobility, making it difficult to define the range of past land-use practices beyond the general understanding that Southern Neolithic people grew dry-farmed millets and pulses and kept herds of sheep, goats, and cattle. However, see Fuller (Reference Fuller2005) for a discussion of the relationship between seasonality and agriculture in the Southern Neolithic and Korisettar et alia (Reference Korisettar, Venkatasubbaiah, Fuller, S. and Ravi2002) for arguments about the potential relationship between ashmound deposits and seasonal transhumance.
Site Background
The archaeological site of Brahmagiri, in the Chitradurga District, Karnataka, is one of the most famous sites in South Indian prehistoric archaeology. It occupies around 3 km2 on a granitic outcrop more than 100 m higher than the surrounding plain and is adjacent to the Chinna Hagari stream. Brahmagiri gained early prominence as the location of several Ashokan Minor Rock Edicts of the third century BC—proclamations about Buddhism made by Ashoka, ruler of the distant north Indian Mauryan empire (Rice Reference Rice1903). Initial archaeological fieldwork by the Mysore State Archaeological Department (Krishna Reference Krishna1934, Reference Krishna1941) established the presence of substantial deposits from what is now known as the Southern Indian Neolithic (ca. 3000–1200 BC), as well as later deposits extending to the sixteenth century AD. In many places on the site, deposits from these later towns cover those of the Neolithic, making spatial analyses of earlier periods challenging.
In work that has continued to define the basic sequence of South Indian prehistory, Brahmagiri was excavated for a single season by the Archaeological Survey of India (ASI) under the direction of R. E. Mortimer Wheeler in 1947 (Wheeler Reference Wheeler1948). On the east side of the Brahmagiri outcrop, Wheeler’s (Reference Wheeler1948) excavations in what he called the “town site” of Isila (mentioned in one of the edicts) exposed strata that he interpreted in terms of three successive cultures: the Stone Axe Culture, the Megalithic Culture, and the Andhra Culture—periods now known as the Southern Neolithic, the Iron Age, and the Early Historic. Much of Wheeler’s (Reference Wheeler1948) focus at Brahmagiri was on dating the extensive megaliths at the site, using the units in the Early Historic “town site” primarily to situate the kinds of ceramics associated with the megaliths into a stratigraphic sequence. Although ashmounds had already been objects of study for more than 50 years at the time of excavation, Wheeler (Reference Wheeler1948) did not recognize the layers of ashmound material exposed in the “town site” excavation units as such—an oversight made clear during our section cleaning of units Br-17 and Br-21 in 2023. Wheeler’s enduring impact in the field (Morrison Reference Morrison2025; Ray Reference Ray2007) may account for the long oversight of these deposits.
Although no previous excavation at Brahmagiri identified ashmounds, the presence or possible presence of ashmounds at the site has been mentioned in the literature (Fuller et al. Reference Fuller, Boivin and Korisettar2007), with ashmound deposits directly noted by Arjun (Reference Arjun2022). One of the goals of new survey work at Brahmagiri by the South Indian Landscape Trajectories (SILT) project has been to identify the distribution of ashmound features across the site. This was accomplished through multiple approaches: mapping visible features via satellite and drone imagery, surface artifact survey via transect walking, intensive feature recording and documentation, and remote sensing.
Based on our preliminary work, we estimate that Brahmagiri has at least nine distinct ashmounds and perhaps as many as 10. All of these are located close to the high rocky outcrop that dominates the site. Because of this, several of the ashmound deposits lie underneath a layer of colluvial overburden washed from upslope, material that has partially protected them from erosion. We traced the Brahmagiri ashmounds in part by their exposure in recent drainage channels, though some also outcrop slightly above the present ground surface and can be documented by surface observation alone. Ashmounds farther away from the outcrop have been exposed by recent field leveling, terracing, and plowing. Our analysis of exposed sections in both farmers’ channels and old excavation trenches suggests a complex stratigraphy with multiple ashmound layers alternating with layers of brown fill. Only some of these features are accurately described as “mounds,” but all consist of fired dung—what we refer to here as “ashmound material.” Hardness varies significantly, with some deposits that are soft enough to be cut with a trowel and others that are rock hard. In softer areas, ashmound material is visibly fibrous, and in all cases, it is white in color. The presence of multiple ashmounds at Brahmagiri is significant, in part because Brahmagiri is also the location of Neolithic habitation deposits, including discrete Early Neolithic levels and an as-yet unexcavated later Neolithic settlement.
The Brahmagiri ashmounds are highly unusual in that most of them cluster along the base of the high granitic outcrop rather than being scattered about. Many lie close together, either overlapping or nearly overlapping; in total, they cover an extremely large area (Figure 1), though the presence of later, overlying deposits makes it difficult to estimate this precisely. Unlike many studied ashmounds, some of the Brahmagiri ashmounds appear to be relatively flat rather than mounded, although others exposed in recent cuts may be more mounded, with deeper (higher) deposits more than 3 m thick. Despite the challenges posed by this large and complex set of deposits, the Brahmagiri ashmounds present us with a critical opportunity to resolve many outstanding issues about the South Indian Neolithic, such as the relationship between (1) the timing of megaliths and ashmounds and (2) ashmounds and settlements (Morrison Reference Morrison, Jarrige and Lefèvre2005).
Drone-derived base map of Brahmagiri with locations of identified ashmounds (red), megalith areas fenced by ASI (purple), and the ERT survey location (yellow). Inset: site location. (Color online)
Methodology
New work at Brahmagiri as part of the SILT project began in 2023, with a focus on surface description, mapping, photography, remote sensing, and cleaning of already exposed sections, laying the groundwork for future test excavations. As part of the planning for future work at the site, we focused our limited remote-sensing work on resistivity mapping.
Archaeologists have been using soil resistivity to remotely identify anthropogenic features for decades (Atkinson Reference Atkinson1953; Carr Reference Carr1977, Reference Carr1982; Clark Reference Clark1990; Gaffney Reference Gaffney2008; Kvamme Reference Kvamme2003). It is a powerful and cost-effective tool that can be deployed relatively easily to map landscapes. However, traditional soil resistivity surveys are generally limited to two-dimensional surveys at a given depth based on the probe spacing and array selected. Less common in archaeological research has been the use of resistivity to explore deposits at depth. Although the idea of recording vertical resistivity profiles is not new (Clark Reference Clark1990) and has been used extensively in geology and related environmental research (Edwards Reference Edwards1977), the use of pseudosections to investigate archaeological data has been relatively less common. Electrical Resistance Tomography (ERT) is commonly used in geographical and environmental surveys because it offers excellent ability to resolve large depositional features over a wide area (Arjwech et al. Reference Arjwech, Sriwangpon, Somchat, Pondthai and Everett2020; Gourdol et al. Reference Gourdol, Clément, Juilleret, Pfister and Hissler2018). Because it is generally less able to resolve small anthropogenic features, ERT has had slower uptake among archaeologists—although it is being deployed with increasing frequency, often in conjunction with other remote sensing techniques with finer resolution, such as ground-penetrating radar (GPR; Balkaya et al. Reference Balkaya, Yalçın Kalyoncuoğlu, Özhanlı, Merter, Çakmak and Talih Güven2018, Reference Balkaya, Ekinci, Çakmak, Blömer, Arnkens and Kaya2021; Casana et al. Reference Casana, Herrmann and Fogel2008; Rabbel et al. Reference Rabbel, Erkul, Stümpel, Wunderlich, Pašteka, Papco, Niewöhner, Bariş, Çakin and Pekşen2015; Ullrich et al. Reference Ullrich, Günther and Rücker2008).
In our first season at Brahmagiri in 2023, we undertook a small ERT test in the hopes that large ashmound deposits would be resolvable. For this survey, we used a Frobisher Tar-3 resistance meter. The Tar-3 is designed to be used primarily in twin-probe or Wenner-array traditional surface surveys, but it can be configured for pseudosection (ERT) recording. This two-dimensional visualization of apparent resistivity is called a “pseudosection” because it assumes a constant relationship between probe spacing and depth (Schmidt Reference Schmidt2013). In this mode, an ERT transect is laid out with a tape measure and then metal stakes are placed at the designated spacing—in this case, 0.5 m along the transect. The four cables of the Tar-3 are then connected, in order, to every combination of pins to create the overlapping readings necessary. We opted for 7 m max spacing, which should have the potential to visualize a depth of up to 3.5 m.
We were not sure if ERT would be useful in this context, so we were glad to be able to experiment with this using the Tar-3 that the project had already acquired for traditional resistivity mapping, rather than needing to first acquire an expensive dedicated system that might not, ultimately, prove useful. As a test to see if ERT would allow this sort of ashmound mapping at Brahmagiri, we surveyed a single transect that spanned 35 m. This resulted in 609 individual resistivity readings, each of which required the four probe wires to be manually switched to connect to the appropriate stakes. The Tar-3 works well for collecting this sort of data, but it is slow. We worked as a team of four, with three people moving the clips between each reading, and one person operating the instrument (see Figure 2). At the end of the survey, an elevation was recorded for each point along the transect so that topography could be incorporated into the analysis. Elevation data was recorded with an Emlid Reach RS3 and with an Emlid Reach RS2+ acting as a base station on a known point.
View looking west of the set-up for pseudosection recording in the field. The author (Hill) stands with the Tar-3 instrument while two crew members prepare to start moving the probes between sets of pins placed at regular intervals along the pseudosection to be recorded (marked with pin flags and a yellow measuring tape). The drainage ditch with an exposed profile of Ashmound 1 can be seen in the background. Photo courtesy of Moriah McKenna. (Color online)
The transect selected for this survey was chosen in a field where our surface survey suggested buried ashmound deposits might be close enough to the surface to be visible in a pseudosection, and where we hoped to be able to detect the edge of the deposits. During the terrestrial survey of the northeast side of the outcrop, we identified nine locations where ashmound deposits were visible, either on the surface or exposed by cuts (see Figure 1), either old excavation units or recent agricultural drainage cuts. Ashmound 1 is a large feature, exposed on its west and part of its north side and visible in a recent drainage ditch. Its eastern edge is not visible on the surface. We placed the ERT transect adjacent to the exposed west side of Ashmound 1, where it was visible approximately 2 m below the surface. The transect was oriented east–west away from the outcrop in the hopes of finding the eastern edge of Ashmound 1.
ERT data must be postprocessed for interpretation, and the inversion algorithm for this is nontrivial (Morelli and LaBrecque Reference Morelli and LaBrecque1996; Tsourlos et al. Reference Tsourlos, Papadopoulos, Papazachos, Myeong-Jong and Kim2014; Zhang et al. Reference Zhang, Zhang, Chen and Jia2015). For these data, we used the free and open source ResIPy inversion software package (Blanchy et al. Reference Blanchy, Saneiyan, Boyd, McLachlan and Binley2020) based on the R family of code written by Andrew Binley (http://www.es.lancs.ac.uk/people/amb/Freeware/Freeware.htm). When the data were modeled and inverted in ResIPy, and elevation data were added to account for elevation change, it produced the visualization shown in Figure 3.
Pseudosection resistance profile (in ohm/m) of the survey transect (facing north). (Color online)
The processed pseudosection appears to show a low-resistance band (shown in red in the visualization) that starts on the west side of the transect about 2 m below the surface. This places the red band at the same elevation as the ashmound deposit visible in the drainage channel profile approximately 3 m to the west. This band of lower resistance is likely the ashmound deposit. It runs horizontally for 30 m before it intersects the surface, which slopes from west to east. It is reasonable to expect that ashmound deposits would be visible in the pseudosection. Resistivity variability is most sensitive to moisture content in the soil (Schmidt Reference Schmidt2013), and the ash deposits are very likely to retain moisture at different rates from the surrounding soil. ERT studies of naturally occurring volcanic ash deposits were similarly able to discriminate ashy deposits from the surrounding matrix (Xia et al. Reference Xia, Ludvigson, Miller, Mayer and Haj2010). Although the Brahmagiri ashmounds are not naturally occurring, like volcanic ash deposits, their high silica content is similar, so it should not be surprising to see the ashmound deposits show up in the pseudosection.
The ability to discern the ashmound layer in the pseudosection data is encouraging, establishing the viability of this method for subsurface depth estimation of ashmounds. Given surface evidence for multiple ashmounds along the base of the Brahmagiri hill and the proximity of these features to one another, establishing the locations, depths, and sequencing of these features is critical. The large size of the ashmound zone (around 11 ha) means that excavation can only be a partial solution, with ERT providing critical subsurface evidence. The ability to remotely sense these features gives us the ability to begin to explore a larger variety of ashmound forms and to dissect their specific histories of construction. Consequently, we may be able to better understand the “life histories” of a set of Neolithic ashmounds, linking them to habitation deposits and to the wider depositional history of the site and region.
Being able to discern the frequency and extent of ashmound deposits is critical for understanding the depositional history at the site. Until now we could identify ashmound deposits only where they happened to be exposed naturally by erosion or accidentally by modern agricultural practices. Therefore, we do not know how many individual ashmounds there are, if there are continuous deposits over a wide area, or what the extent of those deposits are. But the ability to potentially visualize these features via ERT can help us better understand this. It has already helped us identify other relevant data as well. For instance, the resistivity data appears to show the eastern side of Ashmound 1 coinciding with the surface at around the 30 m mark. On the ground, in person, when the earth had been recently turned for planting, there was no obvious hint that this area might be the eroding edge of the deposit, even though ashmound deposits are significantly lighter in color than other cultural deposits or the natural soil. However, once we identified this as the spot where the ashmound might be on the surface, we looked at historic aerial imagery of the same field at different times of year, and it is sometimes possible to see a distinct boundary between darker soil to the west and lighter soil to the east that lines up directly with this same location. Using the Google Earth time slider, this boundary is visible in February 2017 and March 2012 (Figure 4). This seems to help confirm that the eastern edge of Ashmound 1 is indeed eroding in the field here. The curve of the feature is visible as it bends around to the north, suggesting the limit of this ashmound deposit. The area defined by this visible color change is approximately 1,600 m2, suggesting a minimum size for this ashmound deposit. Further soil-color variation in the adjacent fields also suggests the extent of additional deposits at the site.
Apparent soil color change line visible in the same field as the ERT survey. This shows a comparison of 2023 drone imagery (top) and Google Earth imagery from 2012 (middle) and 2017 (bottom). The apparent round shape of the light soil edge visible in the earlier imagery (middle and lower) is marked in red (top) on the more recent imagery where it is not visible. Satellite imagery has had contrast enhanced slightly to exaggerate the color change. Note the white color of the fields to the southeast of Ashmound 1 where another ashmound is located. (Color online)
Chipped stone tools were used in both the Neolithic and earlier Iron Age (Morrison et al. Reference Morrison, Sinopoli, Black, Trivedi, Bauer, Lycett, Haricharan, Bates, Reddy and Korisettar2022) in this region. Our surface artifact mapping consisted of counts along transects spaced 250 m apart across both the east and west sides of the outcrop. We found a significant clustering of lithic artifacts—many likely to be Neolithic in date and contemporary with the ashmound deposits—at the base of the outcrop in the same location as the ashmound deposits discussed here, further supporting our feature mapping (Figure 5).
Density distribution map for lithics counted during the surface artifact survey. Note the heavy concentration northeast of the outcrop, in the vicinity of the ashmounds. (Color online)
Discussion
Ashmounds may have first drawn attention because of their monumental aspect. The most well-known ashmounds are prominent landscape features, rising up to 10 m above the surface, representing large-scale communal enterprise. Much discussion of the ritual and social meaning of ashmounds has centered on the scale of the largest deposits, the long-term utilization of the features, and the relationship to sedentary villages (Allchin Reference Allchin1963; Fuller et al. Reference Fuller, Boivin and Korisettar2007; Paddayya Reference Paddayya2019a, Reference Paddayya2019b). Although the scale of deposits at sites such as Wandalli, Kupgal, Palavoy, Kudatini Utnur, and Budihal suggests that the construction of ashmounds functioned as a symbolic communal activity (Allchin Reference Allchin1963; Boivin et al. Reference Boivin, Fuller, Korisettar and Petraglia2008; Johansen Reference Johansen2004; Morrison Reference Morrison2009), as yet we have little understanding of how this worked or how ashmound building and burning were linked to livestock herding, farming, or patterns of mobility (Rajala et al. Reference Rajala, Madella, Korisettar, Niccolucci and Hermon2010). It is critical to document the variability of ashmounds, including smaller ones (Arjun Reference Arjun2017, Reference Arjun2022) and ones with more truncated sequences.
Although the number of sites with identified ashmound deposits is high (at least 150; Paddayya Reference Paddayya2019b; Venkatasubbaiah Reference Venkatasubbaiah2012), relatively few excavations of buried ashmound deposits have occurred to date. Where ashmound deposits are more dispersed as more ephemeral buried features—as they are at Brahmagiri—understanding the formation processes, extent of deposits, and evidence for reuse, damage, and modern disturbance is crucial for understanding their wider variability. Documentation of more ephemeral or buried ashmound deposits will also provide novel insight into the variability of ashmound deposits and may even find undisturbed “vallum” deposits such as those that have been described by Allchin (Reference Allchin1963) and Foote (Reference Foote1916).
Conclusion
By demonstrating that we can remotely sense the stratigraphic location of ashmound deposits, this research enables a whole new potential avenue for investigating ashmounds at sites such as Brahmagiri and Watgal (Arjun Reference Arjun2017), where, until now, buried deposits have only been visible in excavation trenches or by the fortuitous agency of modern agricultural processes. This relatively small demonstration that we can track ashmounds with ERT will act as a pilot study for a more complete survey at Brahmagiri, which will allow 3D reconstructions of the shape and distribution of buried deposits, volumes estimates, and potentially, reconstructions of the impact of deposits on soil flow and local hydrology. Although there are many remote-sensing technologies available to archaeologists that have the potential to reveal buried features, ERT provides a great and relatively low-cost tool for mapping larger homogeneous subsurface anthropogenic geomorphic features that might be difficult to study otherwise.
Acknowledgments
Thanks go to the entire crew of the 2023 season at Brahmagiri, including Mutharasu Anbalagan, Gopalakrishna Umapathy, Sampurna Bordoloi, Siddharth Kutty, and Ritvik Chaturvedi. We acknowledge and appreciate permission and support by the Government of India and Archaeological Survey of India.
Funding Statement
Thanks to the University of Pennsylvania; the Penn Museum; and the Indian Institute of Technology, Bombay, for research funding.
Data Availability Statement
The ERT data presented here is available at https://doi.org/10.6084/m9.figshare.31902280.
Competing Interests
The authors declare none.
Open access
