Beyond the radiocarbon barrier in Australian prehistory.

Roberts, Richard G. ; Jones, Rhys ; Smith, M.A. 等

The team that has been dating early Australian sites by luminescence methods replies to Allen's (1994) view of the continent's human chronology, published in the June ANTIQUITY (68: 339-43). They argue the strength of the long chronology with their new optical dates.

The issue

Systematic application of radiocarbon dating to archaeological sites in Australia and Papua New Guinea during the 1960s revolutionized knowledge concerning the antiquity of human presence in the region, with established values being extended from mid-Holocene times (Clark 1961: 243) to c. 33,000 radiocarbon years ago (Jones 1973) in little over a decade. It became apparent during the next 15 years that an apparent 'ceiling' had been reached, whereby radiocarbon dates of between 35,000 years and just short of 40,000 years were obtained from a number of disparate locations across the continent.

Two 1989 papers interpreted these data differently (Allen 1989; Jones 1989). Allen (1989), taking this limit literally, argued that some of the oldest dates came from stratigraphically less secure contexts, such as river terraces and other open deposits. However, one of us (RJ) had been concerned for several years that dates of this order of magnitude were close to the theoretical limits of the method and that contamination by even a tiny amount of modern carbon could change a sample of 'infinite' age into one with an apparent age of 40,000 years or less (Jones 1982: 30).

Geomorphological examples

The same issue has been discussed in geomorphological research, for example by Thom (1973) concerning evidence for old sea levels.

More generally, as Chappell (1991) points out, if a deposit which extends in age across the range 20,000 to 100,000 years were to be uniformly sampled, then some 70% of the results ought to be infinite in radiocarbon terms. Were only 1% modern carbon to be added, no sample would give an infinite result and 80% would appear to have an age of about 35,000-40,000 years; this Chappell (1991: 378) calls an 'event horizon'.

Deep lacustrine sequences in Australia, at Pulbeena Swamp in northwestern Tasmania (Colhoun et al. 1982; Colhoun 1985: 48-9), Lake Terang in western Victoria (D'Costa & Kershaw in press) and Lake Eyre in South Australia (Magee et al. in press), all show this phenomenon: 14C age increases steadily with depth back to about 35,000 years BP, and this apparent finite 14C age then continues into deposits at least 80,000-90,000 years old. At a level in the Lake Eyre sequence at Williams Point, for example, Genyornis eggshell was dated by accelerator mass spectrometry (AMS) radiocarbon, by thermal ionization mass spectrometry (TIMS) uranium series, and by amino-acid racemization; the surrounding sediment was dated by thermoluminescence (TL). While the three latter methods yielded ages of c. 50,000 years, the 14C age was only c. 40,000 years BP. AMS radiocarbon values of c. 45,000 years BP were also obtained from fine-grained charcoal and pollen collected from deposits that are at least last interglacial in age (Magee et al. in press).

The famous pollen sequence from Lake George, on the Southern Tablelands of New South Wales, further illustrates this point. 14C dating of organics yielded ages in correct stratigraphic order back to a value of c. 30,000 years BP at a depth of 2 m (Singh & Geissler 1985: 396). 14C ages continued to fluctuate around this value to a depth of 7 m in deposits, that on other grounds, are believed to date to the last interglacial.

In the geological sequences discussed above, few of the 14C ages are stated to be equivalent to 'background'.

Archaeological examples

Among Australian archaeological sequences, Allen (1994: 341) states he has not seen evidence that, when plotted against depth, 14C ages start to 'flatten out' at slightly less than 30,000 radiocarbon years. However, the Devil's Lair site in southwestern Australia (Dortch 1979; 1984: 40-41) shows just such a pattern, with frequent stratigraphic inversion of radio-carbon ages of 30,000-38,000 years BP in deposits of possibly significantly greater age.

At the Mungo lunette site in southwestern New South Wales, stone artefacts were excavated to a depth of 1.5 m below the Mungo palaeosol which has been 14C dated to c. 30,000 years BP. This palaeosol horizon is cemented with calcium carbonate, which reduces the possibility of any contamination from higher levels. A small sample of charcoal (ANU-1263) obtained from a stratigraphic position above the lowest artefact gave a value indistinguishable from background; the result being somewhat coyly reported as being in excess of 40,000 years (Shawcross 1975; Shawcross & Kaye 1980; Mulvaney 1975: 153). The implications of this date, which is still not fully published, were too great for the intellectual milieu of the early 1970s.

The nature of the archaeological material dated

At most Australian sites, both geomorphological and archaeological, the 14C chronology is derived largely from samples composed of charcoal and finely comminuted soil organics, which are more susceptible to contamination than shell or eggshell samples.

This problem has been recognized by some for at least a decade. In the Willandra Lakes area of New South Wales, Bowler & Wasson (1984: 191) noted:

recent evaluation of the reliability of some radio-carbon dates obtained, casts doubt on the validity of organic dates [from charcoal specks; J. Bowler pers. comm. 1994] in the time range greater than 25,000 years

and

consistent evidence suggests that all organic samples older than 20,000 to 25,000 BP are subject to substantial and irregular patterns of contamination by younger organic complexes. This makes them appear too young. In some cases, the errors involve differences of up to 10,000 apparent years between dates on shell and organic samples from the same midden. In all tests of consistency, the unionid [that is, freshwater) shells provide more reliable results in this part of the time scale near the limits of radio-carbon detection.

In discussing contamination, Allen wrongly presumes that we believe that all Australian 14C samples older than c. 35,000 years BP are contaminated to some degree. We do not. We believe only that ages of 35,000 years BP are either that age or more, and that problems of contamination are apt to compress the early part of the 14C chronology for human occupation of the continent. While some such ages may be correct (after the half-life and any calibration corrections), others may be considerably in error due to contamination. Not all samples need be contaminated -- in this respect, it is Allen who presupposes ubiquitous contamination. However, because the Libby half-life of 5568 years (rather than the correct half-life of 5730 years) is used to calculate conventional 14C ages, all uncalibrated 14C ages require a correction of c. 3%, irrespective of any additional adjustments for changes in the atmospheric production rate of 14C. Conventional ages may differ from calendrical years by as much as a few millennia in the time period up to 40,000 years but could differ negligibly at 45,000-50,000 years (Mazaud et al. 1991).

Luminescence ages in relation to radiocarbon ages

During the 1981 excavation of the Lindner Site, Nauwalabila I, in Deaf Adder Gorge in the Kakadu region of the Northern Territory, one of us (RJ) faced what, at that time, seemed an insoluble dilemma. Here was a deep stratified sequence of artefacts in sands, with the quantity of charcoal decreasing exponentially with depth and no charcoal in the basal third of the deposit (Jones & Johnson 1985). On geomorphological grounds, the basal sands were believed to be of considerable antiquity. TL dating of naturally deposited sands presented a potential solution (Jones & Johnson 1985: 183) and became feasible through a geomorphological research programme to TL-date sand aprons in the Kakadu region (Roberts et al. 1991). Charcoal persisted only a few centimetres below the surface of these natural mantles of unconsolidated sand, so the much older charcoal surviving at archaeological sites offered the best prospect of calibrating TL ages against 14C ages. It is ironic, given the present calls for calibration by Allen (1994), that this has been a keystone of our research methodology from the beginning.

In our work, the best match between 14C (scintillation and AMS) and luminescence (TL and optical) ages is from Allen's Cave. This rock shelter is situated in a shallow, collapsed doline on the Nullarbor Plain, a karst limestone area in South Australia, and receives wind-blown sands from the Great Victoria Desert. In the cave deposit, a hearth with an antiquity of c. 10,000 years yielded calibrated radiocarbon ages in excellent agreement with the luminescence ages for the unheated sediments immediately beneath (Roberts et al., in press b). The integrity of this feature precludes any possibility of post-depositional charcoal or sediment translocation. We intend to submit full details of this inter-comparison for publication in ANTIQUITY.

Discussion

New Ireland and New Guinean sites

Allen's 14C ages from New Ireland were obtained from marine shell. While they are the oldest yet reported from that island, they are not the oldest in that tropical region. TL and uranium series ages for archaeological deposits on the Huon Peninsula in northeastern Papua New Guinea indicate an age of 40,000-60,000 years for artefact-bearing deposits (Groube et al, 1986). More recent TIMS uranium series determinations for the coral-reef terrace supporting these deposits indicate an age of 52,000-61,000 years (Chappell et el. 1994). One of us (RGR) has recently collected further sediment samples from this site for optical dating, to reduce the uncertainties associated with the earlier TL determinations. At the present state of knowledge, the age of the Huon site is consistent with the ages for first occupation of the two northern Australian sites. Allen and his team apparently have not applied luminescence dating techniques to the sands underlying the midden at the New Ireland site, nor did they excavate to the base of the sands. Had they done so, we would be in a better position to know the time-period during which the site was not occupied by people, as we have done at Malakunanja II. We do not presuppose that the oldest sites in New Ireland have necessarily been located by archaeologists.

Tasmanian and South Australian sites

Luminescence ages generally have error margins that are too large to 'fine-tune' the 14C time-scale. However, constraints on the magnitude of radiocarbon calibration can be made by selecting samples from contexts in which luminescence dating is considered to yield reliable ages and radiocarbon contamination can be discounted. Because luminescence dating of unheated sediments relies on the dating signal being zeroed by sunlight prior to sediment deposition, limestone cave deposits such as those in southwest Tasmania proposed by Allen are not ideal for age inter-comparisons. The true age of the deposit (that is, the elapsed time since the deposit was last reworked) will be overestimated if the luminescence 'clock' is not reset completely prior to sample burial. Optical dating has been shown to overestimate the age of cave deposits in deep karst systems where, en route to the cave floor, sediment is stored for long periods and remobilized intermittently in the darkness of the cave (e.g. Koonalda Cave in South Australia; Roberts et al. in press b). TL dating is apt to fare worse. In addition to concerns about radiocarbon calibration and contamination, incomplete zeroing of the luminescence dating signal is a potential contributor to discrepancies between 14C and luminescence ages.

Northern Territory sites

For Malakunanja II and Nauwalabila I, Allen is concerned about possible systematic luminescence age overestimation and post-depositional movement of artefacts. Our confidence in the luminescence chronology at both sites is based not only on the sequence of multiple, closely-spaced samples in correct stratigraphic order but also on the luminescence chronology being pinned at three points in both profiles. Near-modern TL and optical ages were obtained close to the ground surface; at two deeper sampling locations, luminescence ages accorded closely with the calibrated 14C ages for associated charcoal pieces (Roberts et al. 1990a; 1990b; in press a).

Allen appears not to have understood the purpose or significance of the regression between TL age and depth at Malakunanja II (Roberts et el. 1990b). He comments that, 'by itself, depth-age correlation is no demonstration of real age, but merely of consistency between samples'. Our TL ages are reported in calendrical years, and the total uncertainties incorporate all known sources of random and systematic error, including those associated with the laboratory and environmental dose rates. The regression, based on these TL ages, indicates not only stratigraphic consistency between samples but also the general relation between the depth of the deposit and its age (in calendar years) at any chosen level.

Allen errs in claiming that we have not considered the role of taphonomic processes in artefact displacement, other than the mixing of sediments and artefacts in the 'kick zone'. In rejecting the possibility of major downward displacement of artefacts at Malakunanja II (Roberts et al. 1990a; 1990b), we noted no indications of such movement -- neither a decline in artefact concentration nor sorting by artefact size or density. The greatest concentration of artefacts was between 2.3 m and 2.5 m depth, and there was no apparent sorting by size or density over this range. Furthermore, sample KTL164 directly overlies a small pit filled with rubble, stone artefacts and haematite. It is highly improbable that such a feature could have been created by post-depositional displacement. This feature gives us confidence that the TL age for KTL164 of 45,000[+ or -]9000 years is a reliable minimum date for human occupation of Malakunanja II.

To cross-check our results at Malakunanja II, we investigated the Nauwalabila I site (Roberts et al. 1993; in press a). There, the earliest artefacts (described by Jones & Johnson 1985), bracketed by optical dates of 53,400[+ or -]5400 and 60,300[+ or -]6700 years, were recovered from a layer of sand and interlocked rubble into which the post-depositional movement of artefacts from younger levels can be discounted.

We are therefore confident of the stratigraphic integrity and chronological coherence at the Malakunanja II and Nauwalabila I sites. The individual TL age determinations and the corresponding regression analysis for the Malakunanja II samples, together with the optical ages obtained for the Nauwalabila I samples, strongly supports our view that initial human colonization of the northern part of Australia took place between 53,000 and 60,000 years ago. We favour a date of c. 60,000 years for first landfall (Roberts et al. 1993; in press a).

In conclusion

As Jones (1989; 1993) has stated before, and as Allen (1994) reiterates, we are in the midst of a significant dating revolution. With few extinct faunal successions, precise lithic technologies and geomorphic benchmarks to guide us at Australian sites, this process includes a diversification of dating methods used in Australian archaeology. We do not advocate reliance on any single dating method, be it radiocarbon or luminescence. Pleistocene chronologies should be constructed using the widest range of appropriate dating techniques -- such as radiocarbon, luminescence, electronspin resonance, uranium series and amino-acid racemization. We have adopted this approach in our previous investigations and have a programme in place to continue such chronometric comparisons to elucidate the date of initial occupation, and the relationship with mega-faunal extinctions, at sites elsewhere around the continent, such as at Devil's Lair in Western Australia, Wood Point in South Australia, Cuddie Springs and Tambar Springs in New South Wales (Dodson et al. 1993; Furby et al. 1993; R. Wright pers. comm. 1993), and sites in far north Queensland and northwest Western Australia where conventional radiocarbon ages exceed 37,000 years BP (David 1993; S. O'Connor pers. comm. 1994). The phenomenon of the radiocarbon barrier, discerned in the Australian record, is likely to be a general problem and warrants close attention by scholars working in other parts of the world.

References

ALLEN. J. 1989. When did humans first colonize Australia? Search 20: 149-54.

1994. Radiocarbon determinations, luminescence dating and Australian archaeology, Antiquity 68: 339-43.

BOWLER, J.M. & R.J. WASSON. 1984. Glacial age environments of inland Australia, in J.C. Vogel (ed.), Late Cainozoic palaeoclimates of the Southern Hemisphere: 183-208. Rotterdam: A.A. Balkema.

CHAPPELL, J. 1991. Late Quaternary environmental changes in eastern and central Australia, and their climatic interpretation, Quaternary Science Reviews 10: 377-90.

CHAPPELL, J., A. OMURA, M. MCCULLOCH, T. ESAT, Y. OTA & J. PANDOLFI. 1994. Revised Late Quaternary sea levels between 70 and 30 ka from coral terraces at Huon Peninsula, in Y. Ota (ed.), Study on coral reef terraces of the Huon Peninsula, Papua New Guinea: establishment of Quaternary sea level and tectonic history: 155-65. Yokohama: Department of Geography, Yokohama National University. Contribution to IGCP 274.

CLARK, G. 1961. World prehistory: an outline. Cambridge: Cambridge University Press.

CALHOUN, E.A. 1985. Radiocarbon dates for Tasmania, 1956-1984, Papers an d Proceedings of the Royal Society of Tasmania 119: 39-54.

CALHOUN, E.A., G. VAN DE GEER & W.G. MOOK. 1982. Stratigraphy, pollen analysis and palaeoclimatic interpretation of Pulbeena Swamp, northwestern Tasmania, Quaternary Research 18: 108-26.

DAVID, B. 1993. Nurrabullgin Cave: preliminary results from a pre-37,000 year old rockshelter, North Queensland, Archaeology in Oceania 28: 50-54.

D'COSTA, D.M. & A.P. KERSHAW. In press. A Late Quaternary pollen diagram from Lake Terang, western plains of Victoria, Australia, Palaeogeography, Palaeoclimatology, Palaeoecology.

DODSON, J., R. FULLAGAR, J. FURBY, R. JONES & I. PROSSER. 1993. Humans and megafauna in a late Pleistocene environment from Cuddle Springs, north western New South Wales, Archaeology in Oceania 28: 94-9.

DORTCH, C.E. 1979. 33,000 year old stone and bone artifacts from Devil's Lair, Western Australia, Records of the Western Australian Museum 7: 329-67.

1984. Devil's Lair, a study in prehistory. Perth: Western Australian Museum.

FURRY, J.H., R. FULLAGAR, J.R. DODSON & I. PROSSER. 1993. The Cuddle Springs bone bed revisited 1991, in M.A. Smith et al. (ed.), Sahul in review: Pleistocene archaeology in Australia, New Guinea an d island Melanesia: 204-10. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. Occasional Papers in Prehistory 24.

GROUBE, L., J. CHAPPELL, J. MUKE & D. PRICE. 1986. A 40,000 year-old human occupation site at Huon Peninsula, Papua New Guinea, Nature 324: 453-5.

JONES, R. 1973. Emerging picture of Pleistocene Australians, Nature 246: 278-81.

1982. Ions and eons: some thoughts on archaeological science and scientific archaeology, in W. Ambrose & P. Duerden (ed.), Archaeometry: an Australasian perspective: 22-35. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. Occasional Papers in Prehistory 12.

1989. East of Wallace's line: issues and problems in the colonisation of the Australian continent, in P. Mellars & C. Stringer (ed.), The human revolution: behavioural and biological perspectives on the origins of modern humans: 743-82. Edinburgh: Edinburgh University Press.

1993. A continental reconnaissance: some observations concerning the discovery of the Pleistocene archaeology of Australia, in M. Spriggs et al. (ed.), A community of culture: the people and prehistory of the Pacific: 97-122. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University. Occasional Papers in Prehistory 21.

JONES, R. & I. JOHNSON. 1985. Deaf Adder Gorge: Lindner Site, Nauwalabila I, in R. Jones (ed.), Archaeological research in Kakadu National Park: 165-227. Canberra: Australian National Parks and Wildlife Service. Special Publication 13.

MAGEE, J.W., J.M. BOWLER & D.L.G. WILLIAMS. In press. Stratigraphy, sedimentology, chronology and palaeohydrology of Quaternary lacustrine deposits at Madigan Gulf, Lake Eyre, South Australia, Palaeogeography, Palaeoclimatology, Palaeoecology.

MAZAUD, A., C. LAJ, E. BARD, M. ARNOLD & E. TRIC. 1991. Geomagnetic field control of 14C production over the last 80 ky: implications for the radiocarbon time-scale, Geophysical Research Letters 18: 1885-8.

MULVANEY, D.J. 1975. The prehistory of Australia. 2nd edition. Melbourne: Penguin.

ROBERTS, R.G., R. JONES & M.A. SMITH. 1990a. Thermoluminescence dating of a 50,000 year-old human occupation site in northern Australia, Nature 345: 153-6.

1990b. Stratigraphy and statistics at Malakunanja II: reply to Hiscock, Archaeology in Oceania 25: 125-9.

1993. Optical dating at Deaf Adder Gorge, Northern Territory, indicates human occupation between 53,000 and 60,000 years ago, Australian Archaeology 37: 58-9.

ROBERTS, R.G., G.C. NANSON & T.J. EAST. 1991. Sediment budgets and Quaternary history of the Magela Creek catchment, tropical northern Australia. Sydney: Supervising Scientist for the Alligator Rivers Region. Open Fmle Record 80.

ROBERTS, R.G., R. JONES, N.A. SPOONER, M.J. HEAD, A.S. MURRAY & M.A. SMITH. In press a. 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.

ROBERTS, R.G., N.A. SPOONER & D.G. QUESTIAUX. In press b. Palaeodose underestimates caused by extended-duration preheats in the optical dating of quartz, Radiation Measurements.

SHAWCROSS, F.W. 1975. Thirty thousand years and more, Hemisphere 19: 26-31.

SHAWCROSS, F.W. & M. KAYE. 1980. Australian archaeology: implications of current interdisciplinary research, Interdisciplinary Science Reviews 5: 112-28.

SINGH, G. & E.A. GEISSLER. 1985. Late Cainozoic history of vegetation, fire, lake levels and climate, at Lake George, New South Wales, Australia, Philosophical Transactions of the Royal Society of London Series B 311: 379-447.

THOM, B.G. 1973. The dilemma of high interstadial sea levels during the last glaciation, Progress in Geography 5: 167-246.