Australia: The Land Where Time Began

A biography of the Australian continent 

Moyjil Site, Southwest Victoria, Australia –a prologue of people, birds, shell and Fire

An unusual shell deposit is located at Moyjil (aka Point Richie), southwest Victoria that shows many characteristics of a midden. An age of 60,000 years (60 Ka) or older was established by earlier research for the shell deposits but was not able to establish if it was humans or animals, such as seabirds, that were responsible for its formation. This paper, the first of 6 in this special issue of Proceedings of the Royal Society of Victoria, summarises the most recent phase (~10 years) of investigations. The site is confirmed to be a midden by the fragmentation and limited size of the dominant shellfish Lunella undulata syn. Turbo undulates). There is also evidence of fire (charcoal and discoloured, fractured stones), as well as 2 hearth-like features, 1 of which has been excavated archaeologically. None of the evidence could demonstrate conclusively a human versus animal origin for the site. It is considered significant that a human origin remains to be disproved. These papers provide the basis for a new phase of research into the possible cultural status of the Moyjil site.

This series of papers investigates an unusual shell deposit on the bank of the Hopkins River mouth, at Warrnambool in southwest Victoria. The site has been the subject of a long-running research project, about 10 years, by the current researchers, their work building on the earlier work of Edmund Gill and others.

A Warrnambool naturalist and historian, Jim Henry, first led Gill to the shell scatter on West Stack in 1981. Subsequently, Gill in collaboration with Sherwood, the author of this paper, began a wider investigation into the Moyjil (Point Ritchie) headland area, which included correlating the deposit at West Stack with shells within a sand layer between 2 calcretes on the headland cliff. He also identified several beach deposits from the Last Interglacial (LIG) around the headland. On the West Stack shells the lack of rounding resulting from abrasion was in contrast with the extensive rounding of shells of the same species (dominantly Lunella undulata syn. Turbo undulates) in the LIG beach deposit. In many Aboriginal middens from the Holocene near rocky coasts in southeastern Australia, this species also dominates. An age beyond the Holocene is suggested by the cementation of the shells at West Stack in their sandy matrix, which was subsequently confirmed by radiocarbon dating which gave ages interpreted as being beyond the limits of the radiocarbon dating method (>40 ka). It was indicated by charcoal and terrestrial snails that were found with the marine shells in the sand layer of the headland that the marine shells had been transported inland. Gill was led by these characteristics to suggest the site was an ancient midden. A workshop was organised by Gill in 1986 in order to examine the Moyjil as well as a second upstream shell deposit (the Hopkins Estuary site) that was later determined to be a natural shell deposit (Gill et al., 1981). The 22 attendees at the workshop included leading geologists, archaeologists, traditional owners and representatives of the Victorian Archaeological Society. It was agreed in the final summary statement that the Moyjil site was more than 40,000 years old based on radiocarbon dating; that it was of human origin and therefore one of the oldest archaeological sites known at present in Australia; and that it warranted further investigation.

Further collaborative work was carried out to refine the age of the Moyjil site by the use of a range of techniques that were relatively new (thermoluminescence (TL), amino acid racemisation (AAR), uranium/thorium radiometric dating and electron spin resonance) suggested an age of 67 ± 10 ka – beyond the limit of the archaeologically accepted time of arrival of modern humans into Australia from Southeast Asia (Prescott & Sherwood, 1988; Sherwood et al., 1994). It was suggested by an ESR study of a shell (Goede, 1989) and another TL analysis (Oyston, 1996) that the age was even older corresponding to the early LIG. In 2006 Hannah Nair investigated further shell speciation and taphonomy. An interpretation that the site was a midden was supported by a number of factors, such as the lack of water wear on the shells, as well as their fragmented nature, and the presence of charcoal and blackened stones, a fish otolith, and fragments of crustacean exoskeletons, though it was also recognised that the deposit could have been formed by either humans or seabirds (Nair & Sherwood, 2007).

The substantial evidence that it was a midden, together with its potential great age, was strong incentive to investigate the site more thoroughly in order to determine how it was formed, and to repeat the age determinations using techniques that had been developed or refined in the 30 years since the initial work. According to Sherwood, this series of 6 papers reports a suite of studies, which were designed to further investigate the age and origin of the shell deposit. In these papers the stratigraphy, chronology, shell taphonomy, the evidence for fire and the origin of discoloured (blackened) stones, and excavation of a hearth-like feature, are detailed. In order to place these studies in context the major conclusion of each study are summarised in the first paper, as well as recommendations for research in the future.

The Site

The Hopkins River enters the Bass Strait at the eastern edge of Warrnambool. A cliffed headland on the western side of the river mouth, that is more than 12 m high and there are 2 seawards stacks. On the flat surface of West Stack there is a scatter of marine shells, stones (some that are blackened) and sand. A 2-4 m wide bench at ~8 m above sea level, at the same level as the surface of the West Stack, which hosts stones that are similarly discoloured, and 2 hearth-like (charcoal and burnt stone) features. A layer of sand up to 2 m thick overlies the headland bench and partially buries it. Dispersed within this sand layer there are marine shells and (rarely) discoloured stones. The Aboriginal name for the headland area is Moyjil, and it is called Point Ritchie by Europeans. The shell scatter on West Stack and the headland is referred to as the Moyjil site

Stratigraphy and the age of the deposit

It has been revealed by stratigraphic and geomorphic analysis of the cliff below the shell bed that there are at least 4 calcarenite/palaeosol couplets (units V,T,S, and R: Carey et al., 2018, this volume) which have been interpreted by Sherwood et al. as evidence of at least 3 interglacial sea level highstands that date back to possibly as much as 500 ka. At Moyjil a valley or swale in unit C is occupied by units T, S, and R. To form the flat surface of West Stack and the headland cliff bench, all have been planated. The flat surface that has been designated Ground surface alpha (Gsα), supports the shell deposit and on the headland discoloured stones and a calcarenite (unit Q2) that contains isolated shells and fragments of charcoal. Analysis by optically stimulated luminescence (OSL) yielded an age of 239 ± 17 ka for the unit R sand that lies immediately below Gsα (Sherwood et al., 2018a, this volume) which corresponds to a penultimate (Stage 7) highstand. The calcrete cap of unit R presumably prevented its removal by marine erosion during sea level maximum of the LIG (MIS stage 5e), as unit R is cemented rather weakly. A sea level history for stage 5e has been constructed by Hearty et al. (2007). The sea level stood at a level about 2-5 m above the level of the present for most of the period between 135 and 118 ka. Visored notches were cut on West Stack, East Stack, a nearby island and the cliff during this time. A brief interval of a few thousand years around 120 ka when the known seal level peaked at about 6-9 m was identified (Hearty et al., 2007). A beach deposit at 7.5 m at Port Fairy, 20 km west of Warrnambool was recognised (Gill & Amin, 1975). A small shell-gravel deposit at Moyjil is situated at 5.8-6.1 m on the ridge between West Stack and the headland. Both deposits may have formed during this sea level peak. Whatever the case, at this time the sea was close to the cliff of the present, and therefore would have had a strong influence on the former cliff top. The West Stack and cliff top would have been overtopped if the sea level reached 8 m and they would have been stripped down to calcrete Rep to form Gsα. Potholes on West Stack, and marine abrasion and potholes on 2 fallen slabs (Blocks B and F) (Carey et al., 2018) is evidence of this overtopping. Even if the sea was 1-2 m below the top of the cliff a hostile environment would have been formed for the accumulation of sediment, and the growth of plants on its surface.

Sherwood et al. believe the accumulation of shells began on this bare surface of Gsα, which was accompanied by the accumulation of blackened stones. A complex history of deposition was revealed by OSL analysis of the sand (unit Q2) which buried Gsα. A broad overdispersion is shown by quartz grains in the sand that was largely calcareous (Sherwood et al., 2018a). It was indicated by the use of a 3-component finite mixing model to model this overdispersion that most of the sand was of the LIG age, which is consistent with the stratigraphic interpretation. Incorporation of this sand as a consequence of mixing from below (bioturbation?) or by mass movement from an upslope source indicates that there is a significant component of older sand (from Stage 7). It is believed a younger component of unit Q2 sand (50-60 ka) represents the time at which a soil calcrete that was developing on unit Q2 (unit Q2cs) achieved closure, which sealed the sand from further turbation.

It was confirmed by re-determination of ratios of amino acids of Lunella undulata syn. Turbo undulates shells from unit Q2, as well as first time determinations of 3 storm-beach deposits from notches on West Stack and East Stack, that they belonged to the same LIG aminozone, which is consistent with the OSL analysis (Sherwood et al., 2018, this volume).

Additional evidence of an age in the LIG is provided by the molluscan fauna of the West stack. A single embedded specimen of Lunella torquata syn Turbo torquatus, a warm water species that is not found in shell beds from the Holocene or along the present coast, was present within the shell assemblage. This species has been regarded as an index fossil for the LIG in southwest Victoria (Valentine, 1965).

The results that were derived from the dating techniques are consistent with the stratigraphy. The most likely time of deposition of unit Q2 is following the retreat of the sea from its maximum in Stage 5e (i.e., post ~120 ka; Hearty et al., 2007) from stratigraphic considerations (Carey et al., 2018; this volume). This estimate cannot be refined by OSL or AAR – in the case of overdispersion precludes sufficient temporal resolution and a large range is shown by AAR ratios, even for shells within the same deposit. Dune formation and burial of Gsα, with its blackened stones, and shells on West Stack, resulted from a retreating sea. The shell and stone arrangements on West Stack capture1 other event. According the Sherwood et al. this is identified as a major disturbance event, which is called by Sherwood et al. the Z event, which shattered calcrete and transported shells and stones (some of which were discoloured) in a matrix of pinkish sand and mud across Gsα. This mixture moved on West Stack as a debris flow to the south across to and into pothole-like depressions. As a result of sorting coarser stones were left on the northern part of the stack. Similar sand and mud on the headland forms the basal component of Q2. The assemblage of stones and shells were disrupted by the Z event, though it did not mark the end of the accumulation of shells.

Origin of marine shells

On Gsα marine shells occur at an elevation of 8.0-8.4 m Australian Height Datum (AHD) on west Stack, and up to 9.5-10.0 m AHD in the headland’s unit Q2 sands. There are not many whole shells. The fragments have sharp edges and no indication of rounding. early investigators (Prescott & Sherwood, 1988) noted their differences from shells of similar age in the storm beach deposits dating to the LIG. It was suggested (Nair & Sherwood, 2007) the Moyjil deposits could be attributed to either humans or animals such as seabirds that are known to form such middens. Middens made by pacific gulls (Larus pacficus) resemble the shell deposits on West Stack. In South Australia examples are dominated by Lunella undulata syn. Turbo undulates, though fragments of the exoskeletons of crustaceans and molluscs such as abalone also occur (Sherwood et al., 2016). It was shown by taphonomic analysis of Lunella opercula that gulls are selective, as they prefer larger shellfish. At low tide on rocky shores gulls forage for shellfish and crustaceans and drop them on rocky surfaces of anvil rocks), to break their shells to access the flesh. The process of shell breakage forms sharp edged fragments and opercula with rim damage mainly concentrated on the edge originally closest to the mollusc’s outside body whorl. This is the location where anvil impact damage mostly occurs (Sherwood et al., 2016).

Near rocky shores in western Victoria, Aboriginal middens dating to the Holocene Lunella undulata syn. Turbo undulatus were the dominant species of shellfish. When Aboriginal middens are compared to those of pacific gulls it is observed that those of humans do not exhibit the same degree of size selectivity, the Aboriginal middens frequently containing smaller individuals and smaller species such as Cominella lineolata and Benbicium nanum. Also, Aboriginal middens tend to have a much higher proportion of opercula with rims showing no sign of damage resulting from the particular way the shells were broken open with hammer stones. The location on impacted rims, when it happens is, however, similar to that seen on opercula from pacific gull middens (Sherwood et al., 2016; Sherwood et al., 2018b, this volume).

No significant difference has been observed (>0.050) between the West Stack in situ (embedded) opercula population and a Moyjil Aboriginal midden from the Late Holocene and 2 pacific gull middens at Point Avoid and Golden Island Lookout in South Australia. However, the in situ population differs very significantly (P<0.01) from the Cape Duquesne, Victoria, Early Holocene Aboriginal midden population. The West Stack population of loose (not in situ) opercula differs at least highly significantly (P<0.01) from all others with the exception of that from golden Island Lookout seabird midden, from which it does not differ significantly (P>0.05). These, together with other observations (presence of smaller shellfish species, presence of smaller L. undulata individuals, and the limited areal distribution of shells) do not allow for a confident assignment of the midden deposit to a singular agency of either humans or pacific gulls (Sherwood et al., 2918b, this volume).

Fire Evidence

The presence of charcoal indicates evidence that there was fire at Moyjil, and fire is also implied by darkened (possibly) burnt stones. Charcoal has been preserved as macroscopic pieces (often 1 cm³ or larger) in lensoidal or horizontal clusters within unit Q2 (Nair & Sherwood, 2007). Darkened stones on Gsα show blackening of varying degrees. Some have darkened surfaces and when broken show a gradation of darkening with depth into the stone. There is a general correspondence between the size of the stones and the colour, with larger stones being pale or partially darkened while smaller stones are dark throughout.

White calcrete that is experimentally heated in a wood fuelled camp fire of modest size resulted in the fragmentation and darkening of the calcrete. Smaller fragments were darkened throughout after 30-60 minutes in the fire. Other researchers have reported similar results for limestone (e.g., Gonzales-Gomez et al., 2014).

A correlation with the size and colour of stones is also shown by magnetic susceptibility (MS) measurements of samples of calcrete. Lower MS is exhibited by larger pale stones compared to small dark calcrete stones. Thermal effects on MS that result from fire are known (Oldfield & Crowther, 2007; Herries & Fisher, 2010) and have been ascribed to mineralogical changes in iron bearing rocks at the temperatures that are reached during combustion (>700^oC – Gonzales-Gomez, et al., 2015). Other circumstances have been attributed to the blackening of limestone (Miller et al., 2013). During the formation of calcrete by fungal activity dark rocks can be produced that have higher magnetic susceptibility as a result of the anaerobic deposition of iron minerals. There is a possible example of such deposition within the palaeosols of unit T in the form of many small (1-3 cm diameter) sub-angular pebbles that have an appearance that is quite different from the stones on Gsα. No black calcrete has been identified at Gsα that could serve as a source for stones on Gsα.

There are 3 dark stones from the surface of Gsα that were subjected to TL analysis in order to test further whether the darkening of the calcrete could be attributed to heating. For these stones the LIG ages that were obtained (93 – 143 ka), are the same as the surface on which they lay, which suggests strongly that TL had been reset by heating (i.e., Sherwood et al. interpreted the TL findings as being to heating of older calcrete during LIG).

Controlled use of fire

Bowler observed 2 structures that had the appearance of fireplaces and he designated them Fp1 and Fp2. The Fp1 was on Gsα embedded within the basal sands of Q2. An archaeological excavation was undertaken under cultural heritage 12/006690 McNiven et al., 2018, this volume). Fp2, which is about 50 m to the east of Fp1 within unit Q2, has not yet been excavated. Fp1 was designated Charcoal and Burnt Stone Feature ♯1 (CBS1) by McNiven et al. (2018, this volume), is in a roughly circular depression in Gsα. There is a dark sand within the depression that contains blackened stones, some of which were fractured with pieces still in jigsaw fit. At the top of CBS1 there is a large flat calcrete stone that has a terra rossa pebble cemented to its surface. Immediately adjacent to the pebble there is a small piece of terra rossa, which suggests thermal spalling (Bowler et al., 2018, this volume).

A wide range of criteria to test more human versus natural processes (such as bushfire) behind the origin of CBS1 was used by McNiven et al. (2018), this volume). Critically, a sample of 30 charcoal fragments that were excavated from the feature were identified, half (N=16) by microscopic examination as possibly to definitely derived from roots. Minute root hairs were found on 2 fragments of charcoal which suggests they had not been disturbed during combustion (i.e., in situ burning within CBS1). Almost all specimens of roots appeared to belong to the same taxon. Confident conclusions about whether the charcoal originated from in situ burning of roots or from root wood that had been brought to CBS1 was not allowed by the small size of the sample (McNiven et al., 2018, this volume).

Exposure to very high temperatures for relatively long periods (>30 minutes at >700^oC) is required for blackening of calcrete as was observed. In a wildfire such conditions are not likely to be present, particularly as the fuel was likely to be heath vegetation. Terrestrial snails within unit Q2 are found in drier coastal woodland or heath environments (Nair & Sherwood, 2007; McNiven et al., 2018, this volume) and within unit Q2 rhizomorphs that were found are of a diameter (<1-2 cm) which suggests a shrub rather than a tree community. Also, the root penetration by large trees would be hindered severely by the well-developed calcrete Rcp which covers the top of the cliff. According to Sherwood et al. they believe the coastal heathland of the present resembles the heathland that existed in the area at the time CBS1 was formed. It is also possible that Gsα was a bare stone pavement following the retreat of the LIG sea. Fracturing and/or blackening of stones by wildfire is not likely beneath a heath vegetation or bare stone pavement (Bowler et al., 2018, this volume).

An evaluation of the arguments

At Moyjil the shell deposit has been subjected to a degree of scrutiny that is rare if not unprecedented in the archaeology of Australia. Its age has now been assigned with confidence to the LIG, which is much older than the currently accepted ages of the oldest known human sites in Australia and New Guinea (i.e., 45-65 ka; i.e., O’Connell & Allen, 2015; Clarkson et al., 2017). For this reason the present researchers have strived to test rigorously the hypothesis of a human origin for the site. The shell opercula of the L. undulata show clear evidence of the selection of size and Sherwood et al. believe that the deposit can be labelled confidently as a midden. It cannot be established with confidence from the evidence that is available whether humans or sea birds are responsible. Some features, such as the presence of small species of shellfish and small L. undulata individuals and the topographically limited area of the scatter are not expected in a midden formed by seabirds. Against this, the distribution of size of L. undulata opercula is more similar to that of seabird middens of the present than Aboriginal middens of the Holocene (Sherwood et al., 2018b, this volume).

The evidence from fire, which includes the TL of dark stones, and particularly the evidence of in situ fire in CBS1 (and possibly Fp2) is similarly equivocal in terms of human versus natural  agency. It is certain that these combustion features are not part of a seabird midden. Evidence for the charcoal of burnt roots, especially in situ burnt root charcoal, is consistent with a feature of natural origin. Though the vegetation at the time the shells were accumulating was most likely coastal heath with a low wood fuel load not capable of reaching the intense heat that would be necessary to facture and blacken stones.

Sherwood et al. suggest that the weight that the reader gives to the various pieces of evidence that is presented in the following papers will depend on individual knowledge and experience. Among the group involved in this project the extent to which available evidence is currently considered supportive of a human agency ranges from ‘weak’ (McNiven) to ‘strong’ (Bowler). It is important, and in spite of these differences, that they all agree that available evidence has failed to demonstrate conclusively that the site is of natural origin. In southern Australia a human site of Last Interglacial age (~120 ka) would be of international significance for its implications for the movement of modern humans out of Africa. According to Sherwood et al. they accept that, as a result, the requirement of a high level of confidence in the evidential basis of human agency is necessary. The presence of stone artefacts is considered by many as a minimum for demonstrating human agency, but it has been noted that many coastal shell middens in southeastern Australia do not contain such materials (McNiven et al., 2018, this volume.

The last phase of long-running research into the potential human origins of the Moyjil site are presented in the 6 papers of this volume. The degree of uncertainty in the conclusions of these studies, together with the potential national and international cultural significance of this site, call for another phase of research at the Moyjil site employing a range of new techniques that will allow better differentiation between human and natural processes of site formation. Sherwood et al. suggest that this work could include further research on differentiating between human and bird middens (Sherwood et al., 2018b, this volume), differentiating between human and natural combustion features (McNiven et al., 2018, this volume), and differentiating between calcrete stones that were blackened by burning versus other known processes such mineralisation and organic impregnation (Bowler et al., 2018, this volume). Fp2 needs to be excavated archaeologically, at the very least, and analysed using a battery of chemical and micromorphological techniques (Bowler et al., 2018, this volume; McNiven et al., 2018, this volume).

Other papers in this volume

Sherwood, J. E., et al. (2019). "The Moyjil site, south-west Victoria, Australia: shells as evidence of the deposit&#x2019;s origin." Proceedings of the Royal Society of Victoria 130(2): 50-70.

McNiven, I. J., et al. (2019). "The Moyjil site, south-west Victoria, Australia: excavation of a Last Interglacial charcoal and burnt stone feature &#x2014; is it a hearth?" Proceedings of the Royal Society of Victoria 130(2): 94-116.

Sherwood, J. E., et al. (2019). "The Moyjil site, south-west Victoria, Australia: chronology." Proceedings of the Royal Society of Victoria 130(2): 32-49.

the LIG sea-level maximum (120&#x2013;125 ka).

Bowler, J. M., et al. (2019). "The Moyjil site, south-west Victoria, Australia: fire and environment in a 120,000-year coastal midden &#x2014; nature or people?" Proceedings of the Royal Society of Victoria 130(2): 71-93.

Carey, S. P., et al. (2019). "The Moyjil site, south-west Victoria, Australia: stratigraphic and geomorphic context." Proceedings of the Royal Society of Victoria 130(2): 14-31.

Sherwood, J. E. (2019). "The Moyjil site, south-west Victoria, Australia: prologue &#x2014; of people, birds, shell and fire." Proceedings of the Royal Society of Victoria 130(2): 7-13.

Sources & Further reading