Australia: The Land Where Time Began

A biography of the Australian continent 

Malakunanja II Arnhem land (more recently called Madjedbebe)

This is a shallow sandstone rock shelter near Ngarradji Warde Djobkeng and Ja Ja Billabong, south of Malangangerr. It has faded paintings on its overhanging walls. The first excavation found charcoal dating to 18,000 years ago. Associated with the charcoal were a grinding hollow and 2 flattish mortars, one of which had clear traces of ochre. There are many Aboriginal burials in this rock shelter.

Later excavations in the 1980s established Malakunanja as the oldest dated site in Australia. The first signs of  Human occupation appear 2.6 m below the surface. The layers showing signs of human occupation were TL dated to between 61,000 and 52,000 years ago. Humans apparently appeared abruptly dated to 61,000 +9,000/-13,000. The sand below this layer was devoid of any signs of human activity. From a depth of 2.5-2.3 m there was dense occupation, from between 52,000 +7,000/-11,000 BP and 45,000 +6,000/-9,000 BP. More than 1500 artefacts were found in the lowest occupation layer.

Artefacts included those made from silcrete, quartzite and white quartz, a grindstone, pieces of dolerite and ground haematite, chlorite and mica and red and yellow ochre. The researchers allowed for the earliest occupants to have trodden artefacts into the soft sand of the floor, putting the first occupation of the site at a conservative time of 52,000 years BP. Below the earliest occupation there was 2 m of sand that were deposited gradually over a period of 110,000 years.

The presence of high grade haematite in this deposit indicates that long distance exchange or transport took place during the Pleistocene, as the nearest known possible sources for the haematite are long distances from the site (Jones & Johnson, 1985b; Jones & Negerevich, 1985; Chaloupka, 1993).

See Stone Tools


According to Roberts et al. (1994) the date of the first arrival of humans on the Australian continent has important implications for the debate on human origins. Later evidence for the timing of the entry of humans into Australia was provided by the optical dating of unburnt quartzose sediments from Nauwalabila I, Lindner Site in Deaf Adder Gorge, 70 km to the south of Malakunanja II. Optical dates were obtained from several stratigraphic levels in an excavation that was 3 m deep where flaked stone artefacts and ground pigments were found in a primary depositional setting. Dates of 53.4 ± 5.4 ka and 60.3 ± 6.7 ka bracket the lowest human occupation levels, and that there is good agreement between optical and 14C age estimates has been demonstrated by the upper levels. Associated directly with the 53 ka level is a high quality haematite with ground facets and striations that indicates the earliest Aboriginals were already using pigments. The authors3 suggest evidence for the colonisation of northern Australia a short time after 60 ka should be seen in the context of this region being the likely entry route of the first human movement into Sahul.

Malakunanja II (Madjedbebe) site in northern Australia associated with early colonisation – Archaeology, Chronology and Stratigraphy4  

As well as having been dated to more than 50 ka BP Malakunanja II in northern Australia also contains the largest assemblage of stone artefacts and other important archaeological components of any site known in Sahul, and these have never been described in detail, which has led to doubts about the stratigraphic integrity of the site. In this paper Clarkson et al. report their recent analysis of the stone artefacts and faunal and other materials that have been recovered from the site during the excavations in 1989, as well as the stratigraphy and history of deposition that was recorded by the original excavators. This study has demonstrated that the technology and raw materials of the early assemblage are distinctive from those recovered from the upper layers. In the early assemblage silcrete and quartzite are common raw materials, and there are also fragments of an edge-ground axe and ground haematite. The lower stone assemblage, which comprised a mix of long convergent flakes, some radial flakes that had platforms that were faceted, and many small thin flakes of silcrete that Clarkson et al. interpreted as thinning flakes. Occasional grinding of haematite and woodworking are indicated by residue and use-wear analysis, as well as frequent abrading of the edges of platforms on thinning flakes. This study led to the conclusion that there may have been a degree of overstatement in previous claims of displacement of artefacts being extensive and post-deposition disturbance. The earlier claims of human occupation 50-60 ka BP are supported by the stone artefacts and the stratigraphic details, and also showed that human occupation at this time differed from occupation in later periods. In this study Clarkson et al. discuss the implications of these new data for understanding the earliest colonisation of Sahul by humans.

Madjedbebe (MJB), or Malakunanja II, its previous name, has attracted much attention as a result of its claims of early human occupation at the site between 50 and 60 ka BP (Roberts et al., 1990a). The scientific significance of the site was established by previous work at the site, especially in connection with an understanding of the timing of human colonisation of Sahul. A dense lower cultural assemblage, which includes evidence of early complex technological, subsistence and artistic behaviours, which has implications for understanding the economic and symbolic dimensions of the earliest societies to occupy Sahul. Thermoluminescence (TL) and Optically Stimulated Luminescence (OSL) ages of 52 ± 11 and 61 ± 13 ka BP brackets the lowest artefacts in the MJB site (Roberts et al., 1990a). At Nauwalabila, a nearby site, similar OSL ages were found to bracket the ages of the lowest artefacts at between 53 ± 5 and 60.3 ± 6 ka BP (Roberts et al., 1994; Bird et al., 2002). Potentially, both of these sites predate Lake Mungo, Devil’s Lair, Nawarla Gabarnmang, Riwi, Lake Menindee Lunette, and Carpenter’s Gap 2 by 5-15 krs (Bowler & Price, 1998; Roberts et al, 1998; Balme, 2000; Turney et al., 2001; Bowler et al., 2003; O’Conner & Veth, 2005; Cupper and Duncan, 2006), which therefore increases substantially the period of human occupation.

Clarkson et al. refer to Sahul as a geographic terminus for migrating modern humans as they moved out of Africa along the southern arc through south and Southeast Asia. Modern genetic analyses also supports such a dispersal (Huoponen et al., 2001; Macaulay et al., 2005; Liu et al., 2006; Sun et al.,2006; Friedlaender et al., 2007; Hudjashov et al., 2007; Oppenheimer, 2009, 2012; Rasmussen et al., 2011), as well as remains of anatomically modern humans that have been recovered from sites such as Liu Jiung [?Liujiang] in China, with an estimated age of 65 ka BP, in Laos at Tam Pa Ling, 46-63 ka BP, Borneo at Niah Cave, 40 ka BP, and in Australia, Lake Mungo, 40 ka BP; Shen et al., 2002; Barker et al., 2007; Demeter et al., 2012; Veth & O’Connor, 2013). The possibility of contact and gene flow has been raised by an archaic species on Flores (Brown et al., 2004), and an unidentified species of Homo in the Philippines (Mijares et al., 2010) species of humans), as well as a modern human presence that was potentially sparse and patchy in the region before  the colonisation of Sahul.

It is suggested by early dates for colonisation that before modern humans colonised Europe they had reached the end of the southern dispersal route, which means that the European Upper Palaeolithic would have had little to do with understanding on the development of modern technology and symbolic expression in South and Southeast Asia and Oceania (e.g. Brumm & Moore, 2005; Habgood & Franklin, 2008; Davidson, 2010; Langley et al., 2011). The period of contact between humans and megafauna in Sahul would also have been lengthened substantially by a ‘long’ chronology for Sahul (cf. O’Connell & Allen, 2004) of 50-60 kyr, which would require further consideration of the nature of this interaction and role of predation versus climate change in leading to their demise.

An opportunity to examine closely the nature of the lithic technology that was employed by early colonists would also be made available by a colonisation date of 50-60 ka BP. It has been argued by Mellars and colleagues (Mellars, 2006; Mellars et al., 2013) that modern humans left Africa with microlithic technology, artistic conventions, and bead-making technologies that were similar to those present in eastern and southern Africa after 60 Ka BP. In sites that have been dated to more than 40 ka BP there is, however, little evidence for this along likely dispersal routes between Africa and Sahul. There is another possibility, that when modern humans left Africa they had technology that was from the African Middle Stone Age (MSA), which included prepared core technology and projectile points, and that this technology is antecedent to those technologies that have been found in Southern Asia and Sahul (Clarkson et al., 2012; Clarkson, 2014). The assemblage at MJB is suggested by Clarkson et al. to be ideal to investigate the nature of the earliest stone technologies in Australia, given the high density of artefacts and the presence of a stone technology that was hitherto not documented in the earliest occupation period in Sahul.

According to Clarkson et al. no detailed report or stratigraphy or assemblage at MJB has ever been published, in spite of the significance of MJB for addressing questions of chronology, origins of modern humans, and early complex behaviour. As a result of this there have been persistent concerns regarding the chronology of human occupation at the site and the extent of any post-deposition disturbance and movement of artefacts that may have obscured patterns of cultural change. In this paper Clarkson et al. address some of these concerns with a detailed examination of evidence, published as well as unpublished. The specific questions addressed by this paper are:

1 The chronology of the archaeological materials that were excavated in 1989,  

2 evidence of change over time of human activities, stone artefact technology in particular, at the site,

3 and the implications of the stone artefact assemblage data for post-depositional disturbance and the movement of artefacts.

Clarkson et al. present new data on the chronology and stratigraphy of the site, the size and diversity of the lithic assemblage, and the change throughout the sequence at MJB in the pattern of technological change. A better understanding of the age and formation of this site, its stratigraphic integrity, the nature of the early lithic industry, and the subsequent technological changes over time was afforded by these data. Re-examination of the assemblage that was recovered during the excavations in 1989, as well as new information about the chronology, stratigraphy, biological components, and the changing nature of the deposition of artefacts, that were obtained from field records that had not been published, were the basis for this study.

Conclusion of Clarkson et al.

According to Clarkson et al. their re-analysis of data from the excavations from 1973 and 1989 at MJB shows that a detailed sequence of industrial succession and palaeoecological change, that overlaps with the regional pattern that is evident at other archaeological sequences in western Arnhem Land, has been preserved at this site. This study found that there is not much evidence that indicates the process of displacement of artefacts or post-depositional disturbance to the deposit that is exceptional at this site. It appears that any post-depositional movement had little impact on the integrity of discrete cultural units, even at the lowest occupation level.

The level of resolution in age estimation that can be achieved for the initial levels of occupation, particularly whether any material can be assigned with confidence to the period prior to 45-46 Ka BP, is the major outstanding issue. With regards to the first specific question asked in the introduction, a date of 50-55 ka BP, based on current evidence, is likely to be a conservative estimate of age for the lower occupation level at MJB. There is a dense horizon of debris of occupation at the 45-55 ka BP level, and there are in situ artefacts below this level. The status of MJB, as being among the earliest archaeological sites in Australia, is supported by the data that has been presented in this paper. The key issue, therefore is whether a series of luminescence ages that are provenanced more tightly, by the use of an OSL technique that reduces uncertainties on individual estimates of age and assesses if there is any mixing of sand from different levels, will provide a chronology that is sufficiently fine-grained to provide a high-resolution age-depth curve for level from 2.4-2.9 m below the surface. There is also a need to assess further intrinsic differences between the assemblages at the 45 ka BP levels and material recovered from earlier levels over a wider area of the site.

The cultural change over time, the second question, has been addressed by examining the assemblages below and above the lens at 2.39 m below the surface, which suggests the levels below 45 ka BP (i.e. below spit 40) display differences in raw materials used and technological composition, which includes a discrete and differentiable lens feature. The raw material preferences, for stone artefact manufacture, display stark changes over time, with quartzite and silcrete being dominant in the early phase and quartz and chert being dominant in the later phase. Technological changes are also associated with these raw material changes, such as ground-edge artefacts, convergent and radial flakes, and in the lower levels, thinning flakes, and points in the upper levels. One of the priorities for analysis of the 2012 excavation is further analysis of the spatial structure of the lowest level of occupation. Clarkson et al. described a major change in the use of the site with the appearance of a shell midden in the Holocene, which reflects adaptation to local variability in mangrove environments and/or a different utilisation of the landscape and the resources that were available after the rise of the sea level and landscape evolution that occurred in the mid-Holocene.

After the report of the brief excavation in 1989 at the MJB site, critiques of the MJB site to a large extent focussed on clarification of dating, artefacts and stratigraphy. Some of the original doubts surrounding the possible inversions and displacement downwards of artefacts have been allayed by this new analysis that is presented in this paper. Several issues that are critical to understanding the chronology, composition and formation of the MJB site were solved by re-examination of the unpublished ages, field notes and lithic assemblages from the 1989 excavations at the MJB site. With reference to the 3rd question about the implications for understanding post depositional disturbance and movement of artefacts of the archaeological data, there is evidence of major changes over time in the preferences for raw materials, which would not be visible if the deposit was greatly disturbed. There is also evidence that within 3 spits in the lower deposit which Clarkson et al. were able to refit flakes, which also suggests that these spits have not been highly disturbed. According to Clarkson et al. a robust response to concerns about the age and integrity can be provided only with field collection of additional material, though these results are insightful. The use of OSL dating at this site was one of its first applications of OSL dating for an Australian archaeological site, and since that time the OSL dating technique has undergone significant technological and methodological developments (see Jacobs et al., 2008, Wintle, 2008), therefore it is strongly suggested that MJB should be re-dated with modern luminescence dating techniques, as is presently being carried out on samples from the 2012 excavation.

Though much more will be revealed about the structure and formation of the deposit by the new campaign, as well as gathering a much larger sample of artefacts, it is possible to deduce from the 1989 excavation material that the site contains succession of industrial changes, a radiocarbon and OSL chronology that is consistent, a record of ecological changes in the Holocene that is consistent with the Arnhem Land sequence that has been documented at other sites, and the presence of artefact refits and cultural features in the lower layer. Though it should be expected that all sites show some degree of mixing which is consistent with deposition being in a predominantly sandy matrix, there is also no a priori reason to suspect that the artefacts at MJB are disturbed heavily or seriously mixed. Understanding of the formation and chronology will be enhanced by the results of the new dating campaign and the new geoarchaeological investigations, such as micromorphology.

Burial Practices in Western Arnhem Land, Australia – a Ground Penetrating Radar Study5

Prior to new archaeological excavations being carried out at Madjedbebe (Malakunanja II) Rockshelter, western Arnhem Land, Australia, a ground penetrating radar (GPR) survey was carried out. It was revealed by the GPR that there was subsurface patterning of rocks in the deposits in the shelter, and it was demonstrated by archaeological excavations that these were related to burials. Following excavation GIS (geographic information system) and statistical analysis elucidated further the relationship between the rocks and the human burials. An opportunity to test a method of identification of unmarked burials by the use of GPR in sandstone rockshelters, and document a marker for burial identification in the region was provided by this integration of mapping, GPR and excavation.

The density of burials tends to strongly correlate with population densities in Australia, and where burials may be found within residential spaces, the development of methods to detect burials is an area of keen interest for researchers as well as managers. A non-invasive way of investigating subsurface features is provided by geophysical techniques (Gaffney & Gater, 2003; Johnson, 2006; Witten, 2006), and therefore these techniques, especially GPR, are now very popular in projects where it is anticipated there may be burials.

It has been suggested (Conyers, 2006: 66) that the physical features that are frequently associated with burials that it is possible to identify by GPR include:

1.      Sediment that has not been disturbed below and surrounding the shaft of a grave;

2.      A coffin or human body as well as associated artefacts;

3.      Sediment that has been used to fill the shaft of the grave that has not been disturbed;

4.      Any surface sediments that have accumulated above the shaft and surroundings following interment (Conyers, 2006: 66).

Also of particular relevance are areas of compacted soil and void spaces, especially in Indigenous burials.

GPR sometimes produces false negatives or no results, and sometimes it produces false positives as a result of other sources of disturbance, in cases in which the graves are not distinguishable from the surrounding strata, therefore GPR is not a foolproof method of detection (Bevan, 1991; Dalan et al., 2010; Davenport, 2001; Nobes, 1999). In Australian historical archaeology, burials that are unmarked are common, and almost exclusively unmarked in Australian Indigenous archaeology, and this presents specific challenges. The identification of these burials with GPR is often impeded by the particular form of these burials, such as bundle, cremation, limited grave goods, shallow depth, no coffin etc. (see Meehan, 1971) and the nature of sediments which are geologically ancient that the interment occurs into. Also, there can be significant “distortions” in the data for both the area that is disturbed of the grave shaft and areas that are not disturbed adjacent to the grave, where the sedimentary matrix is comprised of gravelly, shelly or sediments that are rich in cobbles, which adds to the complexity of the interpretation. The interpretation is often speculative, and excavations are only rarely carried out in order to confirm the specific nature of the anomalies that are identified by GPR, as a result of there being limited case studies with which the Australian results can be compared.

In this paper Lowe et al. detail how a combination of GPR and archaeological excavation was combined with a GIS approach to test and identify many unmarked burials in the context of a rock shelter. Statistical analysis was also used to test the results in order to confirm that the association that has been documented was deliberate rather than random. Across Arnhem Land burial methods are known ethnographically to include secondary burials in rock shelters, excarnation (removing the flesh from corpses), tree burial and hollow log coffins (Meehan, 1971), though as to why certain individuals might receive particular treatment there is little known evidence, and it is also not known if this changed over time. None have been reported from Lowe et al.’s study site, though several accounts have been documented in the region of the study.

Additionally, legal codes have changed over the last 30 years, which define Indigenous peoples as the primary rights holders with regard to decision making in respect to their heritage have been instrumental in improving relationships between archaeologists and the Traditional Owners, though they have also resulted in a lower number of burial sites being investigated in Australia. The Gundjeihmi Aboriginal Corporation (GAC) – which represents the Traditional Owners of the study area, the Mirarr, have granted permission to study the Madjedbebe Rockshelter in Arnhem Land as part of broader heritage initiatives, with the result that was a rare opportunity to carry out a detailed geophysical survey, and then ground disturbance by archaeological excavations.

Madjedbebe site

This is a narrow rock shelter, that is a sandstone overhang that faces to the northwest dating to the Pleistocene, located at the base of the Arnhem Land Plateau escarpment, about 40 km to the west of the East Alligator River.  The floor of the shelter is generally flat, sandy and mostly free of vegetation, and the wall contains a gallery of pigment art. At Madjedbebe the archaeological deposits comprise a thick shell midden unit that is about 70 cm thick dating to the Holocene, below which is a further about 3 m of cultural deposits from the Pleistocene (Kamminga & Allen, 1973). The subsoil parent material is mix of sand and silt that has weathered from the adjoining quartzose sandstone escarpment of the Kombolgie Formation dating from the Middle Proterozoic (East, 1996: 40). In this study it is only the shell midden unit that is of interest.

Madjedbebe has been investigated several times since its discovery in 1972 (Kamminga & Allen, 19723), and in 1989 (Roberts et al., 1990); the 1989 excavation yielded luminescence dates of 50,000-60,000 BP. Investigations involving small test pits revealed that within the midden unit there were burials present, though there were assumed to be only a few mainly secondary bundle burials (Smith, 1989). Concerns were raised by the prior identification of burials when the site was to be re-investigated; therefore a geophysical survey was conducted before work began on the excavation so the researchers would be prepared for what they might find.

The excavations carried out in 2012 recovered 17 individuals that coded as skeletal remains (SR) in various states of completeness. These burials were mainly primary interments (n=13) that had been dug into, or just through, the shell midden into the uppermost level of the sand unit on which the midden had been deposited. All the burials contained minimal amounts of grave goods, and the SR were found in both flexed and extended positions.

Rocks were found to be associated with at least 9 of the burials, which is a tradition that had been documented at the Nawamoyn site which was near the Madjedbebe site (Schrire, 1982). It was found that at Madjedbebe most of the rocks were placed on the individual’s heads and, in 2 instances, rocks had been placed on the head and the feet (SR1 and SR5), and 1 burial (SR4) the rock placed only on the feet. Apart from 2 burials in a single grave, SR3 and SR14, the rocks were of a similar size in each burial, with an average diameter of 20 cm, a size which would allow the rock to be moved by a single person, though not likely to be moved by animal activity or bioturbation, as is indicated by the relatively intact and articulated bones in the burial. When these rocks were plotted during excavation it was revealed that they coincided with the burials, therefore when these rocks showed up in the GPR survey it corresponded with primary interments.

Concluding remarks

The importance of detailed recording of data and integration when attempting to investigate and map complex archaeological sites is highlighted by this excavation. This study demonstrates the potential value of GPR surveys, though GPR surveys have been extremely rare in rock shelter studies in Australia. The use of GIS in the integration of the results of GPR and excavations has proved to be very beneficial in the understanding of burial practices at Madjedbebe because of the specific way the individuals were interred at this particular site. The presence of many subsurface rocks of unknown origin was found by the initial GPR survey; they were subsequently identified by excavation to be associated with 17 burials, and it was indicated by statistical analysis that the association was deliberate, and therefore not random.

Ngarrabullgan Cave, a Pleistocene Archaeological Site, Australia - New Optical and Radiocarbon Dates, Implications for Comparability of date and Human Colonisation of Australia

See Aboriginal Occupation - Populating the Continent - The Evidence

Sources & Further reading

  1. Flood, Josephine, 2004, Archaeology of the Dreamtime, J. B. Publishing
  2. Phillip J. Habgood & Natilie R. Franklin, The revolution that didn't arrive: A review of Pleistocene Sahul, Journal of Human Evolution, 55, 2008
  3. Roberts, Richard G., Rhys Jones, Nigel A. Spooner, M. J. Head, Andrew S. Murray, and M. A. Smith. "The Human Colonisation of Australia: Optical Dates of 53,000 and 60,000 Years Bracket Human Arrival at Deaf Adder Gorge, Northern Territory." Quaternary Science Reviews 13, no. 5–7 (// 1994): 575-83.
  4. Clarkson, C., M. Smith, B. Marwick, R. Fullagar, L. A. Wallis, P. Faulkner, T. Manne, E. Hayes, R. G. Roberts, Z. Jacobs, X. Carah, K. M. Lowe, J. Matthews and S. A. Florin (2015). "The archaeology, chronology and stratigraphy of Madjedbebe (Malakunanja II): A site in northern Australia with early occupation." Journal of Human Evolution 83(0): 46-64.

  5. Lowe, K. M., L. A. Wallis, C. Pardoe, B. Marwick, C. Clarkson, T. Manne, M. A. Smith and R. Fullagar (2014). "Ground-penetrating radar and burial practices in western Arnhem Land, Australia." Archaeology in Oceania 49(3): 148-157.


  1. Australian Aboriginal prehistoric sites
  2. Radiocarbon determinations, luminescence dating and Australian Archaeology


Author: M. H. Monroe
Last updated: 28/04/2016


Aboriginal Australia
Anthropological History
Aboriginal History
Aboriginal Occupation Sites-Tasmania
Aboriginal physical type
Archaeological Sites
Birrigai Shelter
Fire-Stick Farmers
Genetic Evidence
H. erectus near Australia
Cloggs Cave
The First Boat People
Evidence from Lake George
Regional Continuity Theory
Social Organisation
Trade - Macassan Traders
Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading