Luminescence Dating in Archaeology: from Origins to Optical

Home , Guitarrero Cave, Koonalda Cave, Luminescence dating

LUMINESCENCE DATING IN ARCHAEOLOGY: FROM ORIGINS TO OPTICAL

RICHARD G. ROBERTS School of Earth Sciences, La Trobe University, Melbourne, Victoria 3083, Australia

AbstractÐLuminescence dating has a proud history of association with archaeology, beginning almost half a century ago. The subsequent decades of research have seen a range of archaeometric applications of luminescence dating: from ®red pottery and burnt ¯ints to sediments incorporated into occupation deposits and earthen constructions. Important contributions have been made to topics as diverse as modern human origins, continental colonisations and the dating of prehistoric rock art. This paper provides an overview of these applications, with a particular focus on recent ®ndings such as those from Tabun Cave in Israel, Diring Yuriakh in Siberia, and Jinmium in Australia. # 1998 Elsevier Science Ltd. All rights reserved

1. INTRODUCTION the key papers published over the past few decades. While some of the more recent and controversial contributions are discussed at length, mention is also made of the less publicised luminescence stu- That they were repositories of the dead, has been obvious to all: but on what particular occasion constructed, was matter dies that have laid many of the technical foun- of doubt. (Thomas Jeerson, 1787, Query XI ``Aborigines'') dations for the dating method. Attention is drawn also to studies that have tackled a long-standing archaeological problem using a novel luminescence More than two centuries have elapsed since technique, or that have reported a technical devel- Thomas Jeerson contemplated the antiquity of opment with implications beyond the speci®c prehistoric burial mounds in North America and archaeological application. Some contributions of became ``the ®rst scienti®c digger'' in the history of luminescence dating to archaeologically-relevant archaeology (Wheeler, 1956, p. 58) when he dug a issues (e.g. environmental change, faunal extinc- trench through a mound in Virginia and examined tions) are discussed brie¯y, as are a variety of po- its internal strata (Jeerson, 1787). Such anthropo- tential avenues for pro®table future archaeometric genic structures have only recently been investigated research. using luminescence methods (Feathers, 1997a,b; The paper is organised into sections based loosely Saunders et al., 1997), whose pedigree in archaeolo- on archaeological ``themes'' or periods, rather than gical research extends back to the pioneering work on luminescence dating methods (TL or OSL) or of Daniels et al. (1953). This team recognised the procedures (palaeodose or dose rate determi- potential for dating heated artefacts using thermolu- nations). I have attempted to make the technical minescence (TL), an archaeometric tradition sub- discussion in each section as self-contained as poss- sequently continued by Ugumori and Ikeya (1980) ible, but the details of speci®c luminescence dating in the ®rst application of optically stimulated lumi- techniques are given elsewhere in this special nescence (OSL) to archaeologically-relevant ma- volume (Prescott and Robertson, 1997; Wintle, terial (CaCO3). Following these novel insights, TL 1997) and in more than 20 reviews, several written and OSL techniques have been developed and speci®cally for archaeologists, published over the tested extensively on a vast array of archaeological past two decades (Aitken, 1977, 1985, 1989, 1990, materials and across a wide range of archaeological 1994, 1998; Berger, 1988, 1995; Duller, 1996; issues; from human origins to human constructions, Feathers, 1997a; Fleming, 1979; Forman, 1989; from rock shelters to rock art. Huntley and Lian, 1997; Lamothe et al., 1984; The aim of this paper is to present a broad over- Mejdahl, 1986; Mejdahl and Wintle, 1984; Roberts view of these archaeometric applications of lumines- and Jones, 1994; Seeley, 1975; Singhvi and Mejdahl, cence dating, from its origins in pottery dating to 1985; Singhvi and Wagner, 1986; Wagner et al., the application of optical dating to archaeological 1983; Wintle, 1980, 1987, 1993; Wintle and sediments. The huge volume of literature on the Huntley, 1982). subject precludes a truly comprehensive review, but A variety of archaeological themes are pursued in I have attempted to illustrate the diversity of the this paper. The choice is idiosyncratic and biased archaeological applications by reference to many of towards those in which luminescence dating has

819 820 R. G. ROBERTS thrown the greatest light on a major problem or body of the pot, thereby zeroing the luminescence period in archaeology. The theme of Australian ``clock'' which, for dating purposes, is based on colonisation is given particular prominence, not electron traps in the temperature range 300±4508C only because of my familiarity with the subject but (e.g. the 3758C peak in quartz). At the outset, pot- also because TL and optical dating of archaeologi- tery of ``known'' antiquity (established on the basis cal sediments in Australia provides a striking coun- of pottery style, site stratigraphy, and uncalibrated terpoint to TL dating of pottery and burnt ¯int, radiocarbon (14C) ages) was tested to verify the age which has dominated luminescence studies in determinations made using the infant TL methods Europe and the Middle East. I have attempted to (Fig. 1). Pot sherds up to 8000 years old from the make this overview appealing to graduate students Middle East (Iran, Turkey and the Levant), Britain and professional practitioners in both archaeology (particularly from the Roman period), Italy, and Quaternary geochronology, but I apologise to Denmark, Peru and Thailand were tried initially those whose work has been overlooked, misrepre- (Aitken et al., 1963, 1964, 1968; Mazess and sented, or cited in an improper context. Zimmerman, 1966; Mejdahl, 1969; Ralph and Han, 1966; Ralph and Han, 1969; Zimmerman and Huxtable, 1969). The bleaching of TL signals by

2. MODERN TO MESOLITHIC sunlight was explicitly reported for both CaCO3 and sediment in the ®rst of these trials (Aitken et 2.1. Pots and potentates al., 1963), but Soviet scientists are generally credited Artefacts provide crucial evidence of human ac- with being the ®rst to recognise the potential to tivity, and the abundant remains of ancient pots date solar-reset sediments using TL (Morozov, and ceramic vessels are among the most valuable to 1968; Shelkoplyas, 1971; Shelkoplyas and Morozov, archaeologists. Luminescence dating made its ear- 1965). Not until the 1980s was optical dating of cal- liest forays into archaeology through the appli- cite and sediment attempted (Huntley et al., 1985; cation of TL methods to heated pottery and Ugumori and Ikeya, 1980). ceramics from ancient Kingdoms, Empires, TL dating continued to develop during the 1960s Dynasties and civilisations. Firing of pottery (to and 1970s, largely through the eorts of the above 5008C) evicts the electrons trapped over geo- research team led by M. Aitken at Oxford logical time in the minerals which make up the University, and the technique became increasingly

150

100

50 SPECIFIC GLOW (Arbitrary Units)

1000 1000 2000 3000 4000 5000 6000 A.D. B.C. ARCHAEOLOGICAL AGE Fig. 1. An early attempt to compare the TL from 18 heated pot sherds with their known archaeological ages. The ``speci®c glow'' is de®ned as the natural TL (in the 350±4508C temperature region of the glow curve) divided by the product of the dose-regenerated TL and the sherd radioactivity. The speci®c glow is approximately proportional to age (after Aitken et al., 1963). LUMINESCENCE DATING IN ARCHAEOLOGY 821 recognised as an independent method for authenti- tary deposits (Wintle, 1978; Wintle and Huntley, cating the age of ®red pottery, ceramics and the 1979). The latter techniques were ®rst applied in clay cores of bronze casts (e.g. Zimmerman et al., archaeological contexts by Debenham (1983) and 1974). Progress was not made without the oc- Readhead (1982), both workers independently casional stumble, such as the notorious ``Glozel'' devising a combined additive/regenerative TL controversy which threw the validity of TL dating method that is now generally called the `Àustralian into some doubt (see Aitken, 1985 for an historical slide'' method (Prescott et al., 1993). An earlier ver- account). But the re®nement of quartz inclusion sion of this approach had been proposed for lava- (Fleming, 1970) and ®ne-grain (Zimmerman, 1971) baked quartz pebbles by Valladas and Gillot (1978), TL techniques, and the development of feldspar in- and extended to burnt ¯ints by Valladas (1978). clusion (Mejdahl, 1972), pre-dose (Fleming, 1973a) The second signi®cant feature of the Zimmerman and zircon (Sutton and Zimmerman, 1976; and Huxtable (1971) paper is the comparison made Zimmerman et al., 1974) dating methods, culmi- between TL and 14C ages for a soil layer deposited nated in a successful series of forgery identi®cations, more than 30,000 years ago, from which animal and including the ``Hacilar'' ceramic wares from Turkey human ®gurines had been recovered. They showed (Aitken et al., 1971), and the supposedly T'ang that uncalibrated 14C ages underestimated TL ages Dynasty ceramics (Fleming, 1973b) and Hui Hsien (corrected for supralinearity) by 04000 years at style pottery (Fleming and Sampson, 1972) from 033,000 calendar years before present. Samples of China; additional examples are given by Fleming this antiquity had never before been dated using (1979). TL, so the discrepancy between the 14C and TL As a result, TL is now used routinely in authen- ages was regarded as only an `àpparent dierence''. ticity testing (Baili, 1994; Stoneham, 1991) and Since then, a similar degree of 14C age underestima- gained early service as an independent test of con- tion (of up to 7000 years) for the period 30,000± troversial 14C age estimates. For example, TL was 40,000 years before present has been reported on used to date heated materials associated with the the basis of TL dating of ®replaces (Bell, 1991; Jomon culture in Japan, the ®rst people in the Huxtable and Aitken, 1977), burnt ¯ints (BoeÈ da et world to make pots from burnt clay. Uncalibrated al., 1996; Mercier et al., 1991) and unheated sedi- 14C ages of 12,000±13,000 years for two cave sites ments (Nanson and Young, 1987; Smith et al., had been hotly disputed, but were con®rmed by TL 1997); OSL dating of unheated quartz grains dates of 12,000±14,000 years on a pot sherd (using (David et al., 1997; Roberts et al., 1994a); the ``subtraction'' technique; Fleming and 234U/230Th dating of lake deposits (Peng et al., Stoneham, 1973) and four blocks of baked sand- 1978), coral reefs (Bard et al., 1990, 1993) and car- stone from a hearth (using quartz inclusions; bonate tufas (Bischo et al., 1994); electron-spin Ichikawa and Nagatomo, 1978). At a third site, 14C resonance (ESR) dating of tooth enamel (Rink et ages tallied with quartz inclusion TL dates on al., 1996); and geomagnetic intensity variations Jomon pottery from later periods (2300±7600 years recorded in marine sediments (Guyodo and Valet, BP; Ichikawa et al., 1978). Other early examples of 1996; Laj et al., 1996). 14C/TL age comparisons can be found in Sampson et al. (1972), Whittle and Arnaud (1975) and Zimmerman and Huxtable (1969, 1971). 2.3. Recent TL applications Luminescence dating of pottery and ceramics has played a lesser role in archaeological research since 2.2. Path®nders of the Palaeolithic the 1980s than has TL dating of burnt ¯ints and The TL study made by Zimmerman and TL and OSL dating of unheated sediments. Huxtable (1971) of burnt clay lumps at DolnõÂ Nonetheless, there have been several important ap- Vestonice is notable to archaeologists in at least plications of TL dating to pottery and ceramics in two important respects. First, it represents the in- archaeological contexts where no other numerical itial attempt to extend luminescence dating into the age estimates (sensu Colman et al., 1987) are avail- Palaeolithic period before the Last Glacial able or where the available 14C chronology is Maximum: a crucial time period in terms of human ambiguous or controversial. For example, TL dat- evolution, artistic development and stone tool man- ing was deployed on the earliest pottery yet found ufacture, in Europe as elsewhere. This push into the in the Western Hemisphere, and compared with Palaeolithic, and beyond the limit of 14C dating, accelerator mass spectrometry (AMS) 14C dates for became the major thrust of TL dating over the next associated shell and charcoal samples (Roosevelt et decade, with the development and application of al., 1991). The ®ne-grain TL technique was used to methods to date burnt ¯ints (GoÈ ksu et al., 1974; derive a date of 07000 years for a single pot sherd Wintle and Aitken, 1977) and baked sand from ®re- from a shell midden in the Brazilian Amazon, a places (Adams and Mortlock, 1974; Huxtable and date consistent with the multiple AMS 14C age de- Aitken, 1977), as well as natural cave and sedimen- terminations. The existence of pottery of such anti- 822 R. G. ROBERTS quity in this tropical forest habitat indicated that, there is a lack of associated material suitable for contrary to the prevailing notion, such environ- 14C or other forms of dating. Dunnell and Feathers ments were not unfavourable for permanent settle- (1994) demonstrated how TL dating can be applied ment, population growth and cultural development to pot sherds exposed on the ground surface in the (Roosevelt et al., 1991). central Mississippi Valley, and surface ceramics There has also been renewed interest in TL dat- from the atomic test site in arid southern Nevada ing of ceramics from Central and North America have since been dated by luminescence (Feathers (e.g. Dunnell and Feathers, 1994; Feathers, 1997a,d; and Rhode, 1997). In the latter study, protohistoric Feathers and Rhode, 1997; Godfrey-Smith et al., ``brownware'' pottery was dated by the ®ne-grain 1997). At the site of St. Croix in Nova Scotia, TL TL method, with one sherd giving a comparable dating of ®ve sherds has produced the ®rst numeri- OSL age from quartz inclusions dated using cal age estimates for the decorative styles used on Duller's (1991) single-aliquot approach. The the local and regional ceramics. Fine-grain TL ages `Àustralian slide'' method was adopted, although of between 1150150 and 2620290 years were luminescence multiplication (``scaling'') factors of obtained, the oldest being in close agreement with up to 1.93 were required to match the dose-shifted an associated 14C age of 2500120 years BP regenerative-dose data with their additive-dose (Godfrey-Smith et al., 1997). TL dating of pottery counterparts. Diculties were encountered in the has also provided archaeologists with a chronologi- determination of the dose rate, not because the cer- cal framework for the pre-hispanic Classic and amics were exposed to fallout from nuclear weap- Postclassic periods in western Mexico (Duverger et ons testing but because of disequilibrium in the al., 1993), the Ochre Colour Ware and megalithic uranium decay series. Nonetheless, the TL and OSL cultures in India (Singhvi et al., 1983), and dates of circa AD 1450±1840 accorded well with Neolithic to Bronze Age sites in Sicily (Troja et al., age controls based on associated organics (dated by 1996), Hungary and Croatia (BenkoÈ et al., 1989), 14C) and other artefacts (projectile points and his- where 14C dates are sparse and sometimes at odds toric beads). with expectations of age based on stratigraphy and TL and 14C ages have been compared in many typology. studies, and a few examples only are listed below to il- Over the timespan of the past few centuries, lustrate the types of heated materials and anthropo- where the use of charcoal from long-lived tree genic structures that have been investigated. Guibert species creates problems for 14C dating of recent et al. (1994a) provided con®rmation of 14C ages on archaeological events, TL dating of heated materials shell, thought to be erroneously old, from ®ne-grain (and now also optical dating of unheated sediments) TL dates on pottery from two Neolithic cemetries in is especially valuable. Human contact with New central Sudan, whereas Buschbeck and Flettner Zealand has been brief, extending over perhaps the (1994) found that TL dates for the sintered remains of last millennium (Anderson, 1991) or two (R. a Neolithic loam hut in Germany con®rmed archaeo- Holdaway, 1996). In a novel attempt to date earth logical expectations that the 14C ages on charcoal ovens made by the early Maori, Fankhauser (1990) were too young. The latter study employed coarse- extracted quartz inclusions from oven stones and grain feldspars for analysis, as did Sanderson et al. obtained TL ages (corrected for supralinearity) of (1988) and Strickertsson et al. (1988) in their dating of up to 0770 years. In contrast, charcoal collected the sintered remains of 10 vitri®ed forts in Scotland. from these ovens gave calibrated 14C ages that were Both groups made corrections for anomalous fading greater by 100±300 years for long-lived tree species. (Wintle, 1973) but Sanderson et al. (1988) did not use This study demonstrates the necessity of matching quartz inclusions for dating because of acute suprali- the dating method to the archaeological event of nearity problems. The four oldest forts were TL dated interest: TL dating here provides a more accurate to between 3100 and 4100 years, thereby extending estimate than 14C dating for the elapsed time since the known timespan for these monuments. Still these earth ovens were last used. Understanding the further north, Huxtable et al. (1976) obtained TL relation between the dated event (e.g. exposure of dates of 1000±400 years BC for the ``mounds of burnt mineral grains to heat or sunlight, or the death of a stones'' that constitute the most numerous class of tree) and the target (i.e. archaeological) event is antiquity in the Orkney Islands. Fine grains were clearly central to archaeometry, and Dunnell and extracted from cooking stones, with those that exhib- Readhead (1988) and Feathers (1997a) provide ited short-term anomalous fading being rejected. The further discussion on this issue. TL dates compared favourably with two calibrated By virtue of being able to date the target event 14C dates of 01180 and 01065 years BC, at which directly, TL dating has been used to maximum ad- time the mounds were built as peat-®red cooking vantage in tests of pottery and ceramic typological places for boiling meat in response to the decline of sequences, and at sites where the existing 14C chron- forest cover in Orkney (Huxtable et al., 1976). ologies are suspect or where artefacts are exposed Other monuments and settlements dated recently on the ground surface. Sur®cial artefacts are often by TL include: three Minoan palaces in Crete dicult to date because, unlike buried artefacts, (dated to 3300±3800 years using quartz inclusions LUMINESCENCE DATING IN ARCHAEOLOGY 823 extracted from the kiln walls; Liritzis and Thomas, These problems are most often associated with the 1980); the Terracotta Army in Xi'an, China (dated estimate of the dose rate, owing to radioactive dise- to 02200 years using the ®ne-grain TL from cer- quilibrium in the uranium-series (238U) decay chain. amics and baked soil; Lu et al., 1988); stoneware Even after ®ring (and particularly if to temperatures from the period of the early Thai kingdom of below 10008C), the clay matrix of pottery and cer- Sukhothai (dated using the pre-dose technique to amics retains an appreciable cation exchange ca- 600±800 years; Robertson and Prescott, 1987, 1988); pacity, and may interact with radionuclides moving temples and citadels in Sri Lanka (dated to the last through the soil pro®le (Hedges and McLellan, 2500 years using quartz inclusions in pottery and 1976; Murray, 1981). Radon (222Rn) loss is a com- ®red bricks; Abeyratne, 1994); ®fteen Mesolithic to mon form of disequilibrium, particularly at sites medieval habitation sites in Denmark, Sweden and with porous pottery (e.g. Meakins et al., 1979) and Finland (dated using quartz and feldspar from cer- igneous geology (Aitken, 1985), and was deemed re- amics and burnt stones; Mejdahl, 1989); the sponsible for the 20% age underestimate obtained remains of the strategically-located Roman city of from pot sherds in volcanic terrain in the American Carnuntum in Austria (dated using the ®ne-grain south-west (Kojo, 1991). Problems have been technique on ceramic tiles from public baths, a lime reported in dating pottery from volcanic islands in kiln and the ®ll of a post-hole; Erlach and Vana, the south-western Paci®c Ocean, but for reasons of 1988); and the foundations of the archaic forti®ca- extremely weak TL emissions from the extracted tions on Palatino Hill in Rome itself (dated to ®ne grains, rather than dosimetric diculties 02600 years using quartz inclusions from pottery; (Prescott, 1982; Prescott et al., 1982). To correct for Bacci et al., 1990). Even the Spanish exploration of the extreme (050%) loss of 222Rn from pottery of western Canada in the 18th century AD has been the Hong Kong region, the thick-source alpha the subject of TL investigation, by virtue of an counting ``gas cell'' system was devised (Aitken, olive jar hauled up from the sea-bed o the Queen 1978; Huxtable and Aitken, 1978) and high-resol- Charlotte Islands and dated to 23517 years ution gamma spectrometry was later employed (Williams et al., 1991). (Huxtable et al., 1982±3; Murray and Aitken, An unusual ®nal example is that of Liritzis (1994) 1982). These techniques were used to infer radon who used TL to date the unburnt limestone blocks of a enrichment in surface ceramics from the Great Mycenean wall in Greece, in this instance the zeroing Basin (the additional radon possibly being absorbed mechanism being insolation at the time of construc- from the atmosphere; Feathers and Rhode, 1997), tion of the wall. He took three 1 mm thick slices of cal- and they have been widely adopted for lumines- cite from the contact surface of a limestone block and cence dating of sediments. determined that only the outermost layer of calcite Another form of disequilibrium is uranium had been bleached by sunlight during the period of enrichment, which Guibert et al. (1994b) detected in wall construction; this exterior layer gave an age of ceramics from Niger, West Africa using high-resol- 3700450 years, in accord with archaeological ex- ution gamma spectrometry. Leaching of uranium pectations, while the inner layers yielded geological from the granitic rock shelter and deposition in the TL signals. This method has since been extended to sediments and ceramics was proposed as a probable the dating of limestone pyramids (Theocaris et al., explanation for the excess of 238U (up to 90%) 1997) and the temple of Apollo in Delphi (Liritzis et compared to 226Ra. TL ages were then calculated al., 1997). (Luminescence methods could also be used for the ceramics assuming uranium enrichment had to date ``standing stones'' and other stone arrange- occurred immediately after burial or more recently, ments, where the exterior grains on the base of the in a similar manner to early and later uranium- stones were last exposed to sunlight when the stones uptake models used in the ESR dating of tooth were emplaced. Potential applications include mega- enamel (GruÈ n et al., 1987). Based on the data pre- lithic graves and monolithic structures in western sented by Carriveau and Harbottle (1983), uranium Europe (W. J. Rink, Personal Communication, 1994), mobilisation also appears to seriously aict pot stone lines and cairns on the Arnhem Land plateau sherds from Ban Chiang in Thailand. The calcu- in northern Australia (Chaloupka, 1993, p. 237) and lated TL dates (Bronson and Han, 1972; Mortlock the variety of stone monuments (Djarlarlar), burial and Price, 1980±1) make no allowance for this dise- rings (Wundudmool) and ``sneezing heaps'' quilibrium, and could be seriously in error (Aitken (Djindjudngarri) found in the Kimberley region et al., 1983; Carriveau and Harbottle, 1983). of Western Australia (G. Walsh, Personal Uranium-series disequilibria have been reported Communication, 1996). also for archaeological sediments on the karst Nullarbor Plain in South Australia. In the Allen's Cave deposit, there was a preferential loss of uranium (compared to 230Th) in association with 2.4. Problems and prescriptions carbonate complexes derived from the surrounding There have, however, been TL studies in which limestone (Roberts et al., 1996; Olley et al., 1997a). diculties were encountered in dating pottery. Several other types of disequilibrium were also 824 R. G. ROBERTS observed, and the evolution of the dose rate zuela and the American south-west. Optical dating through time was modelled to improve the accuracy of heated artefacts has also capitalised on the new of the OSL dates (Olley et al., 1997a). For samples generation of single-aliquot techniques, which oer of Holocene age, disequilibrium due to 226Ra is of two substantial advantages: ®rst, the mass of particular concern because the majority of the dose sample required is reduced to a minimum, which rate in the uranium chain originates from the decay may be critical in certain situations; and second, the of 226Ra and its ``daughter'' products, and because palaeodose can be calculated with improved pre- the half-life of this radionuclide (01600 years) cision as each aliquot is self-normalising and as sev- means that even a single episode of leaching has eral independent estimates of the palaeodose can be long-lived (5±6 half-lives) consequences (Meakins et obtained at least as quickly as one multiple-aliquot al., 1978; Aitken, 1985; Olley et al., 1996). palaeodose estimate. Liritzis et al. (1994) dated two Disequilibrium in the thorium (232Th) decay series ceramic sherds using the single-aliquot additive- is rare, or at least rarely reported. Dunnell and dose technique (Duller, 1991, 1995) and obtained Feathers (1994) observed an excess of 232Th over its OSL dates in close accord with TL dates, and of daughters in pot sherds from south-eastern equal or better precision. Two aliquots of quartz Missouri, which they attributed to leaching of 228Ra; (each weighing only 5±10 mg) were used for each owing to the short half-life (06 years) of this nuclide, sherd, one to characterise the dose-response curve the extant disequilibrium implies that the sherds and the other to describe the decay rate due to the have acted as chemically `òpen'' systems during at repeated OSL measurements and preheats. least the past 30 years. In contrast, these two radio- There are potential pitfalls with single-aliquots, nuclides were shown to be in secular equilibrium in however, particularly in the use of regenerative-dose the deposits at Allen's Cave (Olley et al., 1997a). procedures (Duller, 1991; McKeever et al., 1996, 1997; Richardson, 1994; Tso and Li, 1994) which, unlike single-aliquot additive-dose methods, 2.5. Let there be light! measure the total luminescence signal. Mejdahl and The use of TL methods in Mesolithic to medieval Bùtter-Jensen (1994) observed signi®cant sensitivity archaeology has recently been supplemented by changes in quartz and feldspars resulting from the optical dating methods. Optically stimulated lumi- erasure of the ``natural'' signal by TL, OSL or nescence (OSL, using green or green/blue wave- IRSL, and proposed a novel method of correcting lengths for stimulation) and infrared-stimulated for these changes using a hybrid of additive-dose luminescence (IRSL) have been applied to both and regenerative-dose protocols. Their single-ali- heated materials and unheated sediments in archae- quot/regeneration and added dose (SARA) tech- ological contexts (the latter commonly termed nique is not strictly a single-aliquot method, but it `àrchaeological sediments'' or `àrchaeosediments''). does require many fewer aliquots than conventional Given the potential for vertical displacement of multiple-aliquot methods and can accommodate artefacts in unconsolidated sedimentary deposits sensitivity changes provided they are independent of (Cahen and Moeyersons, 1977; Hughes and the added dose. Initial tests of the SARA technique Lampert, 1977; Richardson, 1992), it is clearly pre- on quartz extracted from 600±1300 year-old cer- ferable to date artefacts directly rather than date amics and kiln bricks showed good agreement the archaeosediments in which they lie, all other between OSL ages and independent age estimates factors (e.g. matching the dated event to the target (Mejdahl and Bùtter-Jensen, 1994). Another event; accuracy and precision of the dating method; approach, requiring only three aliquots, was pro- and so on) being equal. Dating of archaeosediments posed by Tso and Li (1994) who adapted Duller's is especially valuable, however, when heated arte- (1991) single-aliquot regenerative-dose IRSL proto- facts are unavailable or when the target events are col to bracket the true palaeodose in ®ne grains periods of major landscape disturbance (e.g. river extracted from 3500 year-old Chinese pottery. and hillsltope sedimentation) triggered by human Volcanic ash (tephra) is a type of heated sediment activities, such as forest clearance and agriculture. found at some archaeological sites. Minerals in ash may include glass, feldspar and quartz, whose lumi- 2.5.1. Heated materials. Bùtter-Jensen and Duller nescence ``clock'' is reset thermally during an erup- (1992) made the ®rst attempt to date heated mation. Volcanic feldspars suer from anomalous terial by OSL, using quartz extracted from burnt fading (Wintle, 1973) but silt-sized glass shards stone at a Viking Age site in Sweden. The additive- appear to yield reliable TL and IRSL dates for dose method was used in conjunction with 24 ali- tephras in the time range 500±400,000 years quots to derive an OSL age in accord with the TL (Berger, 1991, 1992; Berger and Huntley, 1994). age for the site, and of similar precision. Similarly Quartz is useful in proximal volcanic contexts, and close agreement between TL and OSL dates on Liritzis et al. (1996) have recently applied ®ne-grain heated quartz has been reported by Feathers and quartz inclusion TL methods to a tephra (1997a), who noted that OSL gave the more precise deposited on the island of Yiali in the Aegean Sea. estimates for four out of six pot sherds from Vene- The four dates obtained (using the 3258C TL signal LUMINESCENCE DATING IN ARCHAEOLOGY 825 and the additive-dose method, with corrections for (including projectile points) and hearths; for a supralinearity) ranged from 01000 to 02100 years 5000 year-old sample, good agreement was obtained BC, but with errors that overlapped; unusually for between OSL and TL ages on quartz, but both quartz, the grains extracted from the two oldest were double the 14C age and the TL age obtained samples exhibited signi®cant internal radioactivity. from ®ne-grain polyminerals (Huntley et al., 1983). The mean age of 01460 years BC indicates that the Greater success was achieved at three archaeologi- hitherto overlooked Yiali eruption, as well as the cal sites in Morocco (Chaperon Rouge, Skhirat and destructive Thera (Santorini) eruption also in the Tahadart), where dune sands had been deposited 2nd millennium BC (Renfrew and Bahn, 1991), between 6000 and 24,000 years ago: multiple-aliquot should be considered as a possible cause of the col- additive-dose OSL dates on coarse-grain quartz lapse of the Minoan civilisation in Crete. There from these archaeosediments compared favourably have been few other archaeological applications of with 14C ages and with TL ages on pottery and luminescence dating to tephras, although great po- burnt ¯ints (Rhodes, 1988; Smith et al., 1990a,b). tential exists in regions with thick accumulations of The latter optical dating team also collected samples volcanic ash; examples from Papua New Guinea from the Upper Palaeolithic to Roman age site at (Groube et al., 1986; Pavlides, 1993; Pavlides and Hengistbury Head in Dorest, England (Smith et al., Gosden, 1994) are discussed later in connection 1990a,b), an important site in terms of linking the with the human colonisation of Melanesia. British stone tool sequence with that found in north-west Europe (Barton and Huxtable, 1983). 2.5.2. Unheated sediments. TL dating of archaeo- The quartz OSL palaeodose determinations (mul- sediments was attempted in the early 1980s using tiple-aliquot, additive-dose) matched those obtained techniques that were equally applicable to geologi- by TL on sediment and burnt ¯int, and the ages cal sediments (Readhead, 1982; Singhvi et al., were consistent with the 14C chronology. 1982), and optical dating of archaeosediments has OSL and TL dating methods have also been used likewise employed procedures developed originally to check the 14C ages obtained previously for for OSL and IRSL investigations of natural depos- anthropogenic earthen mounds and concentric its. At both archaeological and geological sites, ex- ridges in north-east Louisiana (Feathers, 1997b; posure to sunlight is needed to reset the Saunders et al., 1997). For such structures, lumines- luminescence ``clock''. cence methods hold the advantage over 14C determi- It is advantageous, however, to use optical, rather nations by dating the target archaeological event: than TL, methods to date unheated archaeosedi- the time of mound or ridge construction. It is im- ments, as the electron traps sampled in optical dat- portant to demonstrate, however, that the lumines- ing are reset by a few minutes, rather than a few cence ``clock'' was reset when the earthworks were hours, of solar exposure (Godfrey-Smith et al., built, otherwise too great an age will be estimated. 1988). Single-aliquot methods are especially useful Sand-sized quartz grains were extracted from buried at habitation sites where archaeosediments in the surfaces within mound and ridge structures, and upper few decimetres of the deposit have been their OSL and TL palaeodoses were determined, the mixed as a result of human activity (Chawla and latter from the hard-to-bleach (3758C) peak which Singhvi, 1989). This ``zone of disturbance'' is analo- had been isolated by ®rst erasing the easy-to-bleach gous to the bioturbation of soil A-horizons (Huntley TL signal using green (55020 nm) light. The dose et al., 1983), both processes resulting in the re-ex- rates were derived mainly from high-resolution posure and mixing of sur®cial sediments. gamma spectrometry, which indicated no disequili- Conventional multiple-grain, multiple-aliquot brium in either the 232Th chain or, for most samples, methods will yield only an average age for those the 238U chain. The combined additive/regenerative grains present at any particular depth, whereas `Àustralian slide'' procedure (Prescott et al., 1993) single-grain optical dating methods (Galloway, yielded TL ages that re¯ected the Pleistocene anti- 1996; Lamothe et al., 1994; Lamothe and Auclair, quity of the geologic features on which the earth- 1997; Murray and Roberts, 1997; Murray et al., works stood, indicating that exposure to ultraviolet 1997) present the opportunity to distinguish between wavelengths of sunlight had been too brief during ``modern'' grains that have in®ltrated downwards ridge and mound construction to even partially and `òld'' grains that have moved upwards but erase the hard-to-bleach signal (Feathers, 1997b). have not been re-exposed on the ground surface. OSL palaeodoses were determined using the OSL and IRSL methods have seen surprisingly ``dose corrected'' additive and regenerative single- little application to Holocene and terminal aliquot procedures of Duller (1994, 1995), and a Pleistocene archaeosediments since Huntley et al. preheat of 2208C for 300 seconds. For each sample, (1985) ®rst tested the accuracy of optical dating on the natural OSL intensity of each aliquot was a 6000 year-old aeolian dune sequence in the Peace plotted against its palaeodose, following Li (1994). River valley of British Columbia, Canada. The bur- Two samples showed an increase in palaeodose ied soil horizons in these dunes had been dated by with OSL intensity, again suggestive of insucient 14C and contained artefacts such as stone tools bleaching in antiquity (of even the most light-sensi- 826 R. G. ROBERTS tive signal); a third sample showed no such relation and an age of circa 4000±5000 years was calculated for this earthen ridge. Two ages were calculated for both partially bleached samples: a maximum age was obtained from the smallest single-aliquot palaeodose (Duller, 1995) and the approximate true age was inferred from the palaeodose intercept on the OSL intensity/palaeodose plot (Li, 1994). One of the samples (a mound) gave ages of 05500 and 04600 years, in close accord with the 14C date of 5300 years BP for charcoal from the construction ®ll. The other sample (a ridge at Poverty Point) yielded OSL ages of 02100 and 01800 years, which compare favourably with a 14C date of 2100 years BP for the same level (Feathers, 1997b). A luminescence chronology for the ``moundbuilders'', who so intrigued Jeerson and 19th century American archaeologists (Renfrew and Bahn, 1991, pp. 26± 27), now appears to be within reach using single-ali- Fig. 2. Topographic map and ground plan of the quot optical dating methods to deal with partially ``Michelsberg'' defence trenches, showing sampling site lo- bleached sediments. cations. Contours are in metres above mean sea-level IRSL made its debut in archaeosediment dating (after Lang and Wagner, 1996). at a Romano-British site near Hereford in England ecacy of bleaching was tested experimentally by (Spooner et al., 1990). A mixture of ®ne-grain polyminerals was extracted from an alluvial sedi- Lang and Wagner (1996), who showed that the ment layer sandwiched between two occupation IRSL signal was eectively zeroed after 30 min of levels containing pottery, stylistically dated to 1700 exposure to sunlight ®ltered through cloud and fog. 50 years. Ages consistent with this were obtained by A more recent sample of heated colluvium from a IRSL (1900200 years), using 880 nm diodes for ®replace was also found to yield an IRSL age in stimulation, and by OSL (1500200 years), using accord with its stratigraphic position and a TL age the 514 nm (green) line from an argon-ion laser; in determination; the IRSL age was signi®cantly more each case, the additive-dose method was applied to precise than its TL counterpart, indicating the ben- 50 aliquots. e®ts of using optical dating for burnt archaeosedi- Infrared stimulation has since continued to be ments. The youngest sample dated (780280 years) used to obtain optical dates from feldspars further demonstrates the potential for optical dating extracted from archaeosediments that may have of archaeological events as recent as medieval. been poorly bleached at deposition, and hence liable Despite these promising ®ndings, Wiggenhorn et to yield gross TL age overestimates. For example, al. (1994) expressed concern that the IRSL ages Fuller et al. (1994) employed a Mesolithic burial may have been too young owing to interference site on the banks of the River Danube in Romania from unstable IRSL emissions in the broad optical to constrain the age of the overlying alluvium; this detection window (330±600 nm) used for dating. As they dated using a ``partial bleach'' approach, to a test of this proposition, Lang and Wagner (1996) simulate the bleaching conditions experienced by extended their IRSL investigations to a nearby ¯uvially transported silt grains, and obtained an Early Neolithic site, where a pit had been in®lled IRSL age consistent with the archaeological age. with sediment and pot sherds from the ``Mittlere Even slopewash deposits transported distances of Bandkeramik'' period (Fig. 3). Ages obtained using less than 100 m have been successfully dated using the broad detection band were compared with those IRSL (Lang, 1994; Lang and Wagner, 1996, 1997; obtained using a narrow detection band, centred on Wiggenhorn et al., 1994; ZoÈ ller et al., 1996). 402 nm. Only minor dierences were observed, Among the remains of a ``Michelsberger Kultur'' although the narrow band was recommended for hillslope settlement at Bruchsal Aue in southern future studies. The IRSL ages obtained for the pit- Germany are two sets of defence trenches which ®ll (6400800 years) and oldest colluvium were in®lled by colluvium after abandonment of the (75001300 years) were consistent with each other settlement in the 4th millennium BC (Fig. 2). The and with archaeological expectations, and suggest IRSL ages of silt-sized feldspars obtained from that the advent of agriculture with the these colluvial in®lls (05000 years) post-date the ``Bandkeramik'' people was accompanied by accel- abandonment of the Late Neolithic settlement, and erated soil erosion and sedimentation. indicate that the IRSL signal was satisfactorily zer- Further indications of the magnitude and timing oed in antiquity, despite the sediments being trans- of soil erosion resulting from ancient human activi- ported by water over a very short distance. The ties have been gleaned from archaeosediments by LUMINESCENCE DATING IN ARCHAEOLOGY 827

mates for a variety of human activities since the Mesolithic. Heated artefacts have been dated directly using TL and, increasingly, by OSL and IRSL methods, while unheated sediments in archaeological contexts (archaeosediments) have enabled human disruptions to the landscape to be discerned. These kinds of applications have also been made for time periods preceding the Mesolithic. To these we shall now turn, beginning with the debate on the origins and evolution of modern humansÐan arena in which luminescence dating has made, in my view, Fig. 3. Schematic cross-section of the ``Bandkeramik'' its most spectacular and in¯uential contributions to sampling site, showing sediment type, sample locations, and IRSL ages in thousands of years (after Lang and archaeology. Wagner, 1996).

3. MODERN HUMAN ANTIQUITY AND Lang and Wagner (1997) and Rees-Jones and Tite ANCESTRY (1997a). Again using a narrow detection band (390± 450 nm) for IRSL dating of ®ne-grain feldspars, The origins and evolution of anatomically mod- Lang and Wagner (1997) found that the greatest ern humans, Homo sapiens sapiens, is a theme in rate of colluvial accumulation at a site near archaeology and palaeoanthropology that fascinates Wiesenbach, south-west Germany, occurred in re- laymen and scholars alike. Some of the greatest sponse to intensive agriculture in medieval times, scientists of the 1860s were drawn into this debate, ceasing around 54045 years ago; surprisingly, such as the comparative anatomist Richard Owen, rather little sedimentation was associated with farm- the geologist Charles Lyell, and the biologists ing during the Roman period. In contrast, periods Charles Darwin, Thomas Henry Huxley and Alfred of sedimentation in the river valleys of the Nene Russel Wallace (Desmond and Moore, 1991). Lyell and Thames in England were found to coincide only became convinced of a pre-glacial antiquity for with forest clearance in the Bronze Age and inten- modern humans in 1859 after a tour of archaeologi- sive farming in Roman and Saxon times (Rees- cal sites in England and France, during which he Jones and Tite, 1997a). The latter results were acknowledged the co-existence of ¯int tools with obtained using the IRSL from ®ne-grain polymin- the remains of extinct animals (Lyell, 1863). At the erals and the, rarely exploited, OSL from ®ne-grain same time, the ®rst discovery of a Neanderthal quartz. A regenerative-dose correction procedure skull was made in a cave near DuÈ sseldorf in was devised to account for the hard-to-bleach and Germany. This species of human, Homo sapiens preheat-induced IRSL signal, which constituted 7± neanderthalensis, added an extra element to the 30% of the total IRSL in these samples; by com- question of human evolution which entangled both parison, palaeodose corrections of less than 4% Huxley (1863) and Darwin (1871) and which con- were applied to the ®ne-grain quartz OSL, for tinues to intrigue to this day (Krings et al., 1997). which alpha-eciency ``a'' values of 0.032±0.043 Over the past quarter century, TL has played a (0.002) were measured. central chronological role in the debate on modern Rees-Jones and Tite (1997a) reported less success, human antiquity and ancestry. Stone tools are one however, in dating colluvial deposits from an of the most enduring artefacts found at archaeologi- in®lled Iron Age ditch in Hampshire and an 8th cal sites (e.g. the 2.5 million year-old tools from century AD earthwork (Wat's Dyke) in North northern Ethiopia; Semaw et al., 1997) and their Wales. In both cases, the IRSL and OSL ages typological dierences have been used to recognise obtained from ®ne-grain feldspar and quartz were distinctive stone tool ``industries'' and to construct much older than the archaeological evidence, insuf- relative chronologies for the emergence and spread ®cient bleaching of the sediments at deposition of humans around the world. Since the work of being the proposed explanation. A similar lack of GoÈ ksu et al. (1974), burnt ¯int (a form of chert) success with IRSL dating of ®ne-grain feldspars in has taken on additional signi®cance as a reliable colluvium has been reported by Wintle et al. (1993). medium for obtaining numerical age estimates using It now appears advantageous to use coarse-grain TL. Recent applications of TL dating to heated feldspars allied with single-aliquot techniques to ¯ints has forced a re-evaluation of the evolutionary identify those colluvial deposits that contain a pro- relationship between anatomically modern humans portion of poorly bleached grains (Duller et al., and Neanderthals, and their association with par- 1995; Li, 1994; Wintle et al., 1995). ticular stone tool industries, in Western Europe and In conclusion, luminescence dating methods have the Middle East. Much of this section of the paper played a major role over the past three decades in will review these outstanding and often controver- providing archaeologists with numerical age esti- sial contributions. 828 R. G. ROBERTS

Luminescence dating of unheated archaeosedi- been measured in situ and with high-resolution ments has made a similar dramatic impact on the gamma spectrometry to check for radioactive dise- chronology of modern human origins in regions as quilibrium in this groundwater-aected site. far ¯ung as Africa, Siberia, Australia and the To my knowledge, these are the oldest OSL ages Americas. The Australian and American appli- yet reported for quartz and, despite dose-saturation cations will be discussed in the next section on con- problems giving rise to the wide error margins, they tinental colonisations, but we shall begin here with demonstrate that the OSL and associated easy-to- luminescence studies of sediments from the pre- bleach 3258C TL signals in quartz (Kaylor et al., sumed birthplace of modern hominidsÐAfrica. 1995; Smith et al., 1986; Spooner, 1994a; Wintle and Murray, 1997) are stable over at least the last 300,000 years. This ®nding con¯icts with the upper limit of 0150,000 years suggested by Huntley et al. 3.1. Africa (1993a, p. 18), who observed the lack of a palaeo- The point in time at which modern humans can dose plateau over the 3258C TL region in quartz be distinguished anatomically from their archaic sand deposited as recently as 120,000 years ago (see ancestors is a matter of considerable interest, as it Huntley et al., 1994a). They attributed this fore- relates directly to arguments about the morphologi- shortened plateau to instability of the 3258C TL cal and genetic lineage of ancient human popu- peak but, in such young samples, an alternative ex- lations. Some palaeoanthropologists (e.g. Frayer et planation may be the decay of the overlapping al., 1993; Templeton, 1993; Wolpo, 1989) have 2808C TL peak, whose lifetime (at 15±208C) is argued the case for ``regional continuity'' over the probably on the order of 300,000 years (Aitken, past million years, where hominid populations in 1985; Roberts et al., 1993b) or less (Murray et al., more or less geographically isolated regions of the 1997; Prokein and Wagner, 1994). world have evolved their own distinctive, regional The date of 0260,000 years for the Florisbad traits. Other palaeoanthropologists and most geneti- hominid provides a slightly longer timeline for ana- cists (e.g. Cann et al., 1987; Groves, 1994; Hammer, tomically modern humans than that estimated by 1995; Stringer and McKie, 1996; Tishko et al., geneticists. But it supports the recent suggestion of 1996) have championed the ``replacement'' view a >250,000 years age for the transition from that modern humans everywhere are derived from a archaic Homo sapiens to the ®rst fully modern common genetic stock that evolved in Africa humans, the Proto-Cro-Magnons, in the Middle between 100,000 and 200,000 years ago, and then East (Mercier et al., 1995a,b). The skull of the swept across the globe; over time, these relatively ``Man of Galilee'' was collected from Zuttiyeh Cave recent migrants completely replaced the existing in Israel and was described originally by Arthur archaic hominid populations. It is therefore import- Keith (1927), the prominent British prehistorian ant to pinpoint when modern humans ®rst emerged who was then embroiled in the ``Piltdown Man'' in Africa, as this will place a lower limit on the date fraud. Some palaeoanthropologists (e.g. of their departure from the continent and their dis- Vandermeersch, 1989) consider the skull as being persal around the world. the close ancestor of Proto-Cro-Magnons and, on the basis of the associated stone tool industry, 3.1.1. Southern Africa. Hominid remains at Flor- would appear to correlate with an age of more than isbad spring in South Africa show a combination of 250,000±270,000 years (Mercier et al., 1995a,b). The archaic and modern characteristics, and may be latter age estimate is derived from TL dates ``among the direct forerunners of anatomically obtained on burnt ¯ints from stone tool industries modern humans'' (BraÈ uer, 1989, p. 129). The throughout the Levant (see below). Both TL and remains were recovered from a spring vent, which OSL have therefore played their part in de®ning the presented severe problems for reliable dating. Early length of time over which modern humanity has dating attempts were made on the overlying and existed. adjacent peat layers, which produced 14C and TL and optical dating have also contributed to uranium-series ages of >42,600 years BP and questions concerning the Middle Stone Age in >100,000 years, respectively. Recently, however, Africa. This period is crucial in terms of the evol- ESR dating of a hominid tooth from the spring ution of anatomically modern humans, yet its vent assemblage and OSL dating of the adjacent chronology is poorly constrained. At Florisbad, peat and brown sand layers have provided ®nite quartz sediments from a Middle Stone Age horizon age estimates for this specimen (GruÈ n et al., 1996). gave an OSL age of 133,00031,000 years, in close The ESR age of 259,00035,000 years (using a agreement with a mean ESR age of novel non-destructive method) compared favour- 121,0006000 years from associated tooth frag- ably with the OSL ages of the deposits through ments (GruÈ n et al., 1996). This last interglacial age which the spring had vented (281,00073,000 and is similar to the ESR ages obtained for Middle 279,00047,000 years). The latter were obtained Stone Age artefacts and remains of anatomically from quartz extracts, the gamma dose rate having modern humans found in Border Cave and the cave LUMINESCENCE DATING IN ARCHAEOLOGY 829 complex at the mouth of Klasies River in South (and thus age) overestimation if contaminant grains Africa (GruÈ n et al., 1990; GruÈ n and Stringer, 1991). form a signi®cant (>10%) proportion of each However, the Border Cave dates (including those of sample and if each aliquot holds more than a few 45,000±75,000 years for the distinctive Howiesons tens of grains. Contaminant grains may then be Poort lithic industry) are too young by 030% com- present in roughly equal numbers on all aliquots pared with amino-acid racemization dates on and this will go undetected, even by the method ostrich eggshell (Miller et al., 1992); the latter proposed by Li (1994). Only analyses using fewer suggest an age of more than 106,000 years, and than 10±20 grains per aliquot, extending ultimately possibly >145,000 years, for the base of the to the dating of single grains (e.g. Lamothe et al., sequence. Attempts to resolve this discrepancy 1994; Murray and Roberts, 1997), can circumvent through TL and optical dating were unsuccessful, this thorny problem. owing to the powdery nature of the cave sediments (Wintle, 1996), but luminescence dating of the 3.1.2. Central and northern Africa. Katanda in Klasies River cave sediments would appear war- eastern Zaire is another Middle Stone Age site ranted, as the ESR ages of the Howiesons Poort where luminescence has been used to date the ®rst layers are similar to those obtained at Border Cave behavioural traits of fully modern humans, as evi- and may likewise be too young. denced by bone points carved more than Luminescence methods have also been used to 90,000 years ago (Brooks et al., 1995; Yellen et al., date the Middle and Late Stone Age levels at White 1995). TL and OSL dates were obtained for quartz Paintings Rockshelter, located near the Okavango grains extracted from well-sorted alluvial sands. Delta in north-west Botswana (Feathers, 1997c). The easy-to-bleach OSL and TL signals in quartz This site is important for the early appearance of were exploited initially, but unusual dose-response barbed bone points and ®sh exploitation in Africa behaviour prevented de®nitive palaeodose determi- (Robbins et al., 1994). As deposits in the shelter nations (Kaylor et al., 1993). Consequently, the were a mix of wind-blown and colluvial quartz hard-to-bleach TL peak was used for analysis, a sand, some being derived from roof collapse, steps less-than-ideal choice given its slow bleaching re- were taken to determine the adequacy of sediment sponse to un®ltered sunlight and particularly those bleaching at deposition. The rapidly bleached OSL wavelengths that penetrate through a column of and 3258C TL signals and the hard-to-bleach water (Berger, 1990; Spooner et al., 1988). Fortu- (3758C) TL peak in quartz were examined, the lat- nately, the TL palaeodose estimate of 17010 Gy ter by the ``Australian slide'' combination of addi- was similar to that suggested by OSL (0190 Gy), tive-dose and regenerative-dose procedures to providing some con®dence that the sample minimise extrapolation errors (Prescott et al., 1993). had been well bleached before deposition. The He also constructed a ``master'' additive-dose TL palaeodose corresponds to an age of growth curve from all samples at the site, as each 82,0008000 years. The likely prospect of radio- exhibited a similar dose-response; this provided a second means of restricting extrapolation errors for active disequilibrium in these ¯uvial sediments was the older samples. For the OSL analyses, Feathers examined directly by using high-resolution gamma (1997c) employed the single-aliquot additive-dose spectrometry, and none was observed in the and regenerative-dose protocols in tandem with the present-day deposits. ``dose correction'' method that Duller (1994) Independent con®rmation of the great antiquity devised for feldspars. of the bone harpoons at Katanda is given by the A similar palaeodose was obtained from the three ESR age of 089,000 years obtained from the enamel signals using the dierent methods, which was of hippopotamus teeth recovered from the same taken as indicating sucient bleaching of the sedi- level (Brooks et al., 1995). This ESR age is calcu- ments at deposition (Feathers, 1997c). For example, lated assuming that uranium was absorbed by the the sample marking the transition from the Middle teeth soon after burial, whereas the assumption of to the Late Stone Age yielded mean palaeodose linear-uptake produces a much greater age estimates of 55 Gy (Australian slide) and 63 Gy (0155,000 years). The latter falls between mass- (``master'' curve) for the hard-to-bleach TL peak, spectrometric U-series ages of 0140,000 and 71 Gy (additive-dose) for the rapidly bleached TL 0174,000 years on dentine from two teeth, but peak, 56 Gy from single-aliquot regenerative-dose these samples are considered to have suered from OSL, and three determinations of 54, 57 and 72 Gy the loss of uranium late in their burial history, from single-aliquot additive-dose OSL. The resulting in age overestimates (Brooks et al., 1995). weighted mean of these determinations corresponds By way of contrast, uranium loss (such as that to an age of 55,4004700 years, in reasonable detected by Feathers (1997c) from high-resolution accord with 14C and amino-acid racemization dates gamma spectrometry analyses of deposits at White for the Middle to Late Stone Age transition at Paintings Rockshelter) has a minor eect on lumi- Border Cave (Miller et al., 1992). There remains the nescence age determinations because the bulk of the possibility, however, of a systematic palaeodose dose rate in the 238U series is derived from 226Ra 830 R. G. ROBERTS and its decay products (Aitken, 1985; Olley et al., OSL ages, as did TL analysis of the 3758C TL peak 1996). in two burnt sediment layers from Bir Sahara East The Katanda study demonstrates the value of and one burnt core from Bir Tarfawi (Huxtable, using multiple dating methods at an archaeological 1993). site to avoid de®ciencies in any particular technique. The burnt samples and the surrounding unheated Despite the apparent concordance between the sediments were analysed by high-resolution gamma luminescence and ESR ages, however, there seem to spectrometry to check for disequilibrium in the be some problems with the original study, and 238U decay series. None was observed, nor did further OSL analysis of the sediment samples using thick-source alpha counting show any evidence of a single-aliquot additive-dose protocol (Duller, radon escape (Huxtable, 1993). Past episodes of dis- 1994) suggests that the material may not have been equilibrium, however, may have occurred during fully bleached in atiquity (Feathers, 1997c; wetter, lake-full periods. Diculties encountered in Gibbons, 1997c). Another example of multi-method U-series dating of lacustrine marls clearly suggest dating comparisons is the study of two adjacent fos- that these sediments have not always behaved as sil lakes (Bir Tarfawi and Bir Sahara East) in the chemically ``closed'' systems (Schwarcz and Egyptian portion of the Sahara Desert (Wendorf et Morawska, 1993). TL, OSL and in®nite 14C dates al., 1993, 1994). Stone artefacts have been recovered for lake sediments, and OSL ages for overlying aeo- from the lacustrine sediments, and a variety of nu- lian sands, indicate that the most recent Pleistocene merical dating methods (14C, uranium-series, ESR, wet episode occurred more than 60,000 years ago TL, OSL and amino-acid racemization) have been (Wendorf et al., 1993, 1994). If this and earlier epi- applied to a range of materials to obtain a chronol- sodes were characterised by leaching of 226Ra, for ogy for these archaeological and palaeoclimatic example, then such disequilibrium would since have events. A neat chronology was not obtained, how- disappeared, leaving no sign that dose rates deter- ever, leading Wendorf et al. (1993, p. 558) to con- mined from present-day radionuclide concentrations clude ``We expected grossly similar results and, to are in error. This problem plagues all luminescence some extent, that expectation has been ful®lled. dating studies, but especially those involving river However, there are discrepancies between the dates and lake sediments which are proof of former obtained by dierent methods, and none of them hydrologic activity and which are sampled, often appears to yield completely reliable or consistent many millennia later, when dry. results, although some are better than others.'' The Despite the chronological uncertainties, lumines- latter category included optical dating methods, cence and other dating methods have placed con- which provided ``the most internally consistent suite straints on the timing of archaeological and of dates for the Tarfawi and Sahara East palaeohydrologic events for the duration of the sequences.'' (Wendorf et al., 1993, p. 560). Middle Stone Age at Bir Tarfawi and Bir Sahara The luminescence chronologies for Bir Tarfawi East. The earliest deposits contain Late and Final and Bir Sahara East were obtained using sand-sized Acheulian hand axes, followed by Middle Stone quartz grains from unheated sediments as well as Age artefacts in deposits no older than burnt sediments. The unheated sediments were 0175,000 years. At least ®ve lacustrine phases examined by TL (Bluszcz, 1993) and OSL (Stokes, occurred between then and 60,000 years ago, with 1993), both using the regenerative-dose method. As ®rm evidence for major wet episodes in the interval the initial bleaching step in this method may induce 100,000±140,000 years ago. For much of the Middle changes in TL and OSL sensitivity (e.g. Berger, Stone Age, therefore, the environment diered con- 1988; Bowall et al., 1987; Smith et al., 1990b), the siderably from the desert of today, permanent or additive-dose technique was also used for a series of seasonal lakes being present for most of this im- OSL samples. The calculated palaeodoses were portant period of human behavioural and biological found to correspond closely to their regenerative- development. Which species of human inhabited dose counterparts (using the 514.5 nm line from an these sites is not known, but the limited variation in argon-ion laser for optical stimulation), demonstrat- stone tool typology and the consistent pattern of ing the absence of signi®cant sensitivity change in site exploitation throughout the Middle Stone Age these samples. As the TL palaeodoses were deter- suggest that ``the acceleration of cultural develop- mined principally from the hard-to-bleach peak ment that supposedly accompanied the appearance (judging by the palaeodose plateaux obtained), sen- of the modern form of human had not begun by sitivity change was less of a concern than adequate the end of the Last Interglacial.'' (Wendorf et al., bleaching in antiquity. Ultraviolet laboratory 1993, p. 573). bleaches of between 2 and 6 h duration, and the The suggestion that a marked change in human plateau method of Mejdahl (1988a), were applied to behaviour can be perceived in artefacts produced determine the ``residual'' TL level reached by the only since the last interglacial is consistent with the sediments at deposition. This variant of the ``partial luminescence dates of 70,000±90,000 years (or less; bleach'' method (Berger, 1988; Wintle and Huntley, J. Feathers, Personal Communication, 1997) for the 1982) produced ages in broad agreement with the bone implements found at Katanda, to the south- LUMINESCENCE DATING IN ARCHAEOLOGY 831 west. It also sits comfortably with the longevity of of tooth enamel, and TL dating of ®ne-grain quartz certain lithic industries at modern hominid sites in sediments (Bowman, 1985) and, most recently, the Middle East, to the north-east. At Tabun Cave, burnt ¯ints (Mercier et al., 1995b). Particular atten- a mandible assigned to Homo sapiens sapiens was tion has been paid to Layer C (in Garrod's scheme, recovered from Layer C and seven associated burnt equivalent to Jelinek's Unit I) which contained ¯ints were TL dated to 171,00017,000 years Mousterian artefacts together with the remains of (Mercier et al., 1995b). This lithic industry persisted two hominids, one a Neanderthal and the other in the Levant for a further 80,000 years, being pre- possibly a Proto-Cro-Magnon (Mercier et al., sent at Qafzeh Cave in deposits dated to 1995b). 92,0005000 years (Valladas et al., 1988). Fully The ®rst numerical age estimates for Tabun were modern human behaviour thus appears to have obtained by 14C which yielded ®nite dates of developed only since the onset of the most recent between 040,900 and 051,000 years BP for Layer C glaciation, a period during which at least two homi- (Weinstein, 1984). An apparent stratigraphic inver- nid species inhabited Europe and the Middle East. sion was indicated by a younger age of Our understanding of the temporal overlap between 35,400900 years BP for the underlying Layer D. Neanderthals and anatomically modern humans has These determinations lie at the limits of the 14C been revolutionised in the past decade, due largely method, however, and should be viewed as mini- to the application of TL dating methods. These mum ages only (GruÈ n and Stringer, 1991). notable contributions are the main subject of the Subsequent mass-spectrometric U-series dating of next section. dental fragments produced ages of circa 98,000± 105,000 years for Layer C and 110,700900 years for Layer D (McDermott et al., 1993). These are 3.2. Eurasia comparable to the ESR ages of 102,00017,000 For hominids leaving Africa, southwest Asia has and 122,00020,000 years for Layers C and D re- been the gateway to Europe. While Homo erectus spectively, assuming early uranium-uptake by teeth may have been living 1.0±1.4 million years ago in after burial (GruÈ n et al., 1991); the linear-uptake the Jordan Valley (Bar-Yosef, 1994) and 1.6± ESR ages were substantially older, being 1.8 million years ago in the Caucasus (Gabunia and 0119,000 years for Layer C and 0166,000 years for Vekua, 1995), the earliest remains of Homo in Layer D. The agreement between the U-series and Europe are those of H. antecessor from Gran early-uptake ESR ages was taken as providing con- Dolina in northern Spain (BermuÂ dez de Castro et ®dence in both sets of age determinations. al., 1997; Gibbons, 1997b). This newly-named Recent TL dates from burnt ¯ints, however, species may be the last common ancestor of modern seriously con¯ict with the ESR and U-series humans and Neanderthals, and the fossil deposits age estimates, Layer C being dated to have been dated by palaeomagnetism to more than 171,00017,000 years and Layer D (Jelinek's Unit 780,000 years (PareÂ s and PeÂ rez-GonzaÂ lez, 1995). IX) to 263,00027,000 years (Mercier et al., The prehistory of these archaic hominids lies at or 1995b). The base of the sequence (Layer E) was TL beyond the dating limit of 0800,000 years for cur- dated to 330,000±350,000 years, which is double the rent luminescence methods (Berger et al., 1992; U-series and early-uptake ESR ages for this level. Huntley et al., 1993a, 1994a,b). TL dates of How can these dierences be explained? The TL 850,000 years for the sites of Korolevo in Ukraine dates were obtained using well-tested methods for and Kuldara in Tajikistan (cited in discussion by burnt ¯int (Aitken and Valladas, 1992; Bowman, Carbonell et al., 1995, pp. 828±829) should be 1982; GoÈ ksu OÈ gelman, 1986; Huxtable, 1982; regarded with caution until details of the methods Mercier et al., 1995a; Valladas, 1978, 1992) which, used are made available for scrutiny. Luminescence at other sites, have produced ages in good accord dating eorts have instead been concentrated on the with 14C (e.g. BoeÈ da et al., 1996; GoÈ ksu et al., chronology of modern human evolution in this key 1974; Roosevelt et al., 1996; Valladas, 1992; region, with a particular focus on the eastern Valladas and Valladas, 1987), ESR (Mellars and Mediterranean region (known variously as the GruÈ n, 1991; Mercier et al., 1993; Schwarcz et al., Middle East, the Near East, and the Levant). 1988; Schwarcz et al., 1989), U-series on tooth (McDermott et al., 1993) and calcite (Green, 1984; 3.2.1. The Levant. Tabun Cave is the key site for Green et al., 1981), OSL of unheated quartz the Late Pleistocene cultures of the Levant (Bar- (Michab et al., 1997; Roosevelt et al., 1996; Smith Yosef, 1989). It lies on the western edge of Mount et al., 1990b), TL of calcite (Green, 1984; Huxtable Carmel in Israel and a variety of lithic industries and Aitken, 1986) and quartz in heated pebbles and are represented in its deposits (Jelinek, 1982). A ceramics (Valladas, 1992; Valladas and Valladas, chronology for this important site has been erected 1987). These comparisons cover the period 4000± from 14C dating of charcoal and black soil (Wein- 200,000 years, the oldest site being Pontnewydd stein, 1984), ESR (GruÈ n et al., 1991) and mass-spec- Cave in north Wales (Green, 1984; Green et al., trometric U-series (McDermott et al., 1993) dating 1981). 832 R. G. ROBERTS

Although the dates for Layers D and E extend are collected) showed a similarly non-uniform distri- beyond this range, ages of 0230,000 and bution of dose rate (Mercier et al., 1995c). 0285,000 years have been reported also for burnt A condition of secular equilibrium prevailed in ¯ints from Terra Amata in southern France (Wintle the 238U decay series at Tabun, there being no evi- and Aitken, 1977) and BelveÂ deÁ re in the southern dence of signi®cant uranium migration nor radon Netherlands (Huxtable and Aitken, 1986), respect- escape (226Ra/210Pb activity ratios consistent with ively. Furthermore, the trapped electrons that give unity) at the present day (Mercier et al., 1995b). rise to the ¯int dating peak at 370±3908C have a The latter result is at odds with the 10±30% emana- lifetime in excess of 50 million years (Wintle and tion of 222Rn inferred from ``sealed'' versus Aitken, 1977; Bowman, 1982) and appear not to ``unsealed'' thick-source alpha counting (Bowman, suer from anomalous fading (Bowman et al., 1982; 1985). If the cave deposit only began to emanate Huxtable and Jacobi, 1982; Valladas, 1985), so the radon after being disturbed during ¯int collection, practical upper limit to ¯int dating is governed not then the activity of 210Pb (half-life of 022 years) by signal instability but by dose-saturation. As determined by gamma spectrometry gives a better methods to obtain reliable age estimates for ¯ints indication of the in situ 222Rn activity than does approaching dose-saturation have been developed thick-source alpha counting, which instead indicates (Mercier, 1991; Mercier et al., 1992), an unforeseen the post-disturbance level of 222Rn escape. bias in palaeodose determinations would appear The validity of the external dose rates determined unlikely. Incomplete heating of ¯int in antiquity by Mercier et al. (1995b) is most convincingly will yield an overestimate of the age, but analysis of shown by their application of the ``isochron test'' at the TL glow curves (Rowlett et al., 1974) and appli- Tabun (Fig. 4). This method allows the external cation of the ``plateau test'' (Melcher and dose rate to be deduced independently, from a plot Zimmerman, 1977) provide a check on this proof palaeodose versus internal dose rate for a suite blem. The plateau test is performed routinely in of contemporaneous ¯ints. The only requirements ¯int dating (Aitken, 1985; Aitken and Valladas, 1992), leading to the rejection of a signi®cant frac- are that the ¯ints have received a similar environ- tion of collected ¯ints (Mercier et al., 1995a). Partial mental radiation dose and possess a wide range of bleaching of ¯int by sunlight (Wintle and Aitken, internal dose rates. The isochron age calculated for 1977; Huxtable, 1981; Bowman and Sieveking, the ¯ints is then based solely on their internal radio- 1983) would reduce, rather than enhance, the TL activity, which is unaected by uranium-series dise- signal and thus the age. To avoid the latter problem, quilibrium and variations in soil water content ¯ints should be at least 10 mm thick and be sealed (Mercier et al., 1995a) and thus avoids many of the immediately in a black bag after collection; as a complications that plague the determination of the further precaution, and to remove the ``rind'' irra- external dose rate. The approach was ®rst suggested diated by alpha and beta particles, the outer 2 mm for TL by Mejdahl (1982), and was demonstrated is removed in the laboratory (Aitken, 1985; using a variety of minerals (alkali and plagioclase Huxtable and Jacobi, 1982; Mercier et al., 1995a). feldspars, and quartz) of dierent grain size (0.1± As regards the dose rate at Tabun, Mercier et al. 2 mm diameter) extracted from a burnt stone (1995b) conducted a set of careful ®eld and laboratory analyses of the ¯ints and the surrounding sedi- 500 ments. The environmental dose rate due to 450 sediments and cosmic rays is especially important in 400 ¯int dating studies because of the low internal 350 radioactivity of ¯int; at Tabun, the environmental 300 dose rate constituted 42±83% of the total (Mercier et al., 1995b). The external dose rate was deduced 250 from two independent methods: thermolumines- 200 150 cence dosimetry (TLD) capsules (using CaSO4:Dy Paleodose (Gy) as the dosimeter) buried in the cave deposits for 100 one year, and high-resolution gamma spectrometry 50 measurements of discrete sediment samples. The combination of these methods allowed a check on 0 100 200 300 400 500 600 700 800 the homogeneity of radionuclide distribution in the Internal dose (µGy yÐ1) cave deposit and any extant disequilibrium in the Fig. 4. Isochron plot for ®ve burnt ¯ints from Jelinek's 238 232 U and Th decay chains. Dose rates varied by Unit XI at Tabun Cave. The ¯int TL palaeodoses are a factor of two between adjacent units, with local plotted against their internal radioactivities, and a best-®t ``hot spots'' of 40K concentration being partly re- linear regression is made. The average age of the ¯ints (287,00020,000 years) is calculated from the slope of the sponsible. At the western Galilee site of Hayonim line, and the external dose rate (80090 mGy per year) is Cave, neutron activation analysis of sediments in obtained from the intercept on the horizontal axis (from hearth-dominated layers (from which burnt ¯ints Mercier et al., 1995b). LUMINESCENCE DATING IN ARCHAEOLOGY 833

(Mejdahl, 1983). The average TL age of 916 years Precipitation of uranium salts might be expected for six separate determinations closely matched the also in sediments adjacent to bone, owing to the isochron age of 918 years, and the external dose local chemical environment created by decompo- rate measured by in situ spectrometry likewise sition of animal tissue (Mercier and Valladas, agreed with the value inferred from the isochron 1994). Consequently, the same caution applies to analysis. A related approach, called the ``subtrac- the determination of the U-series dose rate for sedi- tion technique'', had been proposed previously for ments extracted from the bones of museum speci- pottery (Fleming and Stoneham, 1973); with this mens for direct dating by single-aliquot technique, the dierence in palaeodose measured luminescence methods. An additional consideration from ®ne and coarse grains of quartz is due only to is that often only ®ne-grain sediments remain their dierent eective alpha dose rates, and the attached to teeth and bones in museum collections. need for any knowledge about the environmental As silts and clays tend to have a greater dose rate dose rate is eliminated. per unit mass than coarser fractions, then the beta The isochron test has since been applied to TL dose rate in the original deposit may also be signi®- dates obtained from burnt ¯ints at Qafzeh (Aitken cantly overestimated (Mellars et al., 1997). and Valladas, 1992), Kebara (Mercier et al., 1995a) Problems of sediment inhomogeneity should not, and Tabun. For Jelinek's Unit XI (equivalent to however, aect the mass-spectrometric U-series ages Garrod's Layer E and dated to 0300,000 years), the (McDermott et al., 1993). To facilitate a corre- external annual dose computed from TLD capsules sponding increase in the latter, Mercier et al. and high-resolution gamma spectrometry (1995b) challenged the assumption of early-uptake (74342 mGy) accords closely with the estimate de- of uranium by the teeth and postulated instead a rived by isochron analysis (80090 mGy), and the linear-uptake model. This doubled the reported U- age comparison was similarly encouraging (Mercier series ages and brought them into closer alignment et al., 1995a,b; Fig. 4). This result indicates that the with the ¯int ages. The dating discrepancies appear, present-day environmental dose rate is typical of therefore, to be resolved if early-uptake ESR and the average dose rate for the last 300,000 years, so U-series ages at Tabun are viewed as minima the possibility of the ¯int ages being distorted sig- (Mercier et al., 1995b). This view is consistent with ni®cantly by radionuclide leaching or ingress over other comparative TL/ESR results from caves in this period can be discounted. the Levant (Es Skhul, Kebara, Qafzeh) and western To explain the discrepancy between the seemingly Europe (Le Moustier), where TL ages agree at least reliable TL ages and the previous ESR determi- as well with linear-uptake ESR ages as with their nations, Mercier et al. (1995b) compared the early-uptake counterparts (Mellars and GruÈ n, 1991; 232Th/238U ratios of the sediments that comprised Mercier et al., 1993; Schwarcz et al., 1988, 1989; the bulk of the deposit (used to calculate the TL Stringer et al., 1989; Valladas et al., 1986, 1987, dates) and the sediments that were adhered to the 1988). teeth used for ESR and U-series dating. This ratio is a relative measure of the heterogeneity of radio- TL dating of sediments does, on the other hand, nuclide distribution in sediments and, being dimen- raise concerns about the validity of the ¯int dates sionless, can accommodate dierences in absolute for Tabun Cave. Two sediment samples were col- radionuclide concentrations; a related ratio lected from Layer D and a mix of ®ne-grain min- (232Th/230Th) has been used to identify archaeologi- erals was extracted for TL dating (Bowman, 1985). cal sediments derived from dierent sources (Olley The Tabun sediments are dominated by quartz and et al., 1997a). The sediments stuck to the teeth had clay minerals (GruÈ n et al., 1991) and quartz (rather generally lower 232Th/238U ratios than the ``bulk'' than feldspar) appeared to be the dominant TL sediments, which they attributed to uranium enrich- mineral, yielding palaeodoses of 0240 Gy (sample ment in the vicinity of the teeth. The precipitated TBN 4) and 0390 Gy (sample TBN 5) from a com- uranium will contribute an insigni®cant additional bined additive-dose and regenerative-dose pro- dose (for the observed relative enrichment), but will cedure. Bowman (1985) was reluctant, however, to bias U-series dose rate determinations made from calculate their corresponding ages, owing to the parental concentrations (e.g. by ICP-MS or neutron extent of radon escape inferred from thick-source activation). alpha counting. But as discussed above, the high- The ESR gamma dose rates were derived from resolution gamma spectrometry, TLD capsule, and neutron activation analysis of sediment attached to isochron-derived estimates of the environmental teeth, which would give an overestimate of the true dose rate calculated by Mercier et al. (1995b) do ``bulk'' soil gamma dose rate and, thus, an underes- not indicate signi®cant 222Rn emanation, either at timate of the ESR ages. This problem could have the present day or during the past 300,000 years. been avoided by using methods such as alpha and Moreover, the `ùnsealed'' count rates of the Layer gamma spectrometry to directly measure the activi- D samples examined by Bowman (1985) correspond ties of 226Ra and its decay products, which account to uranium and thorium dose rates (using the con- for the majority of the U-series dose rate. version factors of Nambi and Aitken, 1986) similar 834 R. G. ROBERTS those calculated by Mercier et al. (1995b); her pot- anomalous fading in these samples was not dis- assium concentrations are likewise comparable. counted by Bowman (1985), who measured fading Bowman's `ùnsealed'' thick-source alpha counts of 55% after a 12-day delay between irradiation and K2O data (combined with a mean water con- and heating, but quartz does not generally suer tent of 27.5% and an annual cosmic-ray dose of from severe fading (Aitken, 1985; Fragoulis and 70 mGy, as reported by Mercier et al. (1995b), and Readhead, 1991; Wintle, 1973). an assumed alpha-eciency `à'' value of 0.1) yield A complex site stratigraphy may provide an eective annual doses of 2.23 mGy (TBN 4) and alternative explanation. Mercier et al. (1995b) col- 3.46 mGy (TBN 5). The corresponding TL dates for lected ¯ints from Jelinek's excavation in the inter- Layer D are 0108,000 and 0113,000 years, respect- mediate chamber of Tabun Cave, whereas the ESR ively. These dates are remarkably close to the mass- and U-series age determinations were made on spectrometric U-series and early-uptake ESR dates teeth collected from Garrod's excavation of the of 110,700900 and 122,00020,000 years, re- outer chamber; the provenance of Bowman's spectively (GruÈ n et al., 1991; McDermott et al., samples is not speci®ed. Certainly, the cave deposits 1993), but are only half as old as the TL ages on at Tabun have been periodically eroded and burnt ¯int (Mercier et al., 1995b). I cannot resolve deformed (Jelinek, 1982), with the sedimentation this dierence to my satisfaction. The occurrence of being described as ``chaotic'' in places (GruÈ n et al.,

Fig. 5. Summary diagram of the ¯int TL chronology constructed for some key Palaeolithic sites in the Levant and western Europe. The annotations to the left of each bar indicate the archaeological levels, and the oxygen isotope scale is shown at far left. The time scale is in thousands of years (from Mercier et al., 1995a). LUMINESCENCE DATING IN ARCHAEOLOGY 835

1991, p. 233). Perhaps none of the numerical age 3.2.2. Western Europe. A radically dierent situ- estimates are faulty, their apparent discrepancy ation was revealed for western Europe. Evidence of faithfully recording local stratigraphic complications modern humans here extends back only 30,000± hitherto unrecognised in the cave deposit. 40,000 years (Stringer, 1988; Stringer and GruÈ n, Dierences in the 232Th/238U ratios of the sediments 1991) and the remains of Neanderthals and modern noted by Mercier et al. (1995b) may likewise re¯ect humans do not occur in association with the same variations in radionuclide abundances between the lithic industry (Mercier et al., 1995a). The youngest outer and intermediate chambers. Maybe a ®nal dates for Neanderthal remains are those from the attempt should be made to date the sediments in French sites of Saint-CeÂ saire, Arcy-sur-Cure and Le both cave chambers, applying OSL methods to both Moustier (Fig. 5). At the latter site, TL ages of the ®ne-sand and silt-sized quartz fractions. circa 40,300±42,500 years were obtained for layers I have discussed Tabun Cave at length for several that had yielded Mousterian tools and a Nean- derthal skeleton (Valladas et al., 1986), and these reasons: it is a pivotal archaeological site for the dates were later con®rmed by ESR (Mellars and Levant; numerical ages have been estimated using a GruÈ n, 1991). The most recent Neanderthal remains variety of dating methods and are much older than are found, however, in association with Upper expected by the archaeologists; and it falls into the Palaeolithic ChaÃ telperronian tools at Saint-CeÂ saire rare category of cases where TL dating of ¯int and Arcy-sur-Cure. TL dating of burnt ¯ints appears to have produced a con¯icting chronology from Saint-CeÂ saire produced an age of to that obtained from ESR and U-series dating of 36,3002700 years for the hominid-bearing layer teeth, and from luminescence dating of sediment. (Mercier et al., 1991), and an uncalibrated 14C age Despite the latter conundrum, the ¯int dating study of 033,800 years BP has been obtained from Arcy- at Tabun represents a ®ne example of the appli- sur-Cure (Hedges et al., 1994; Hublin et al., 1996). cation of state-of-the-art TL methods, including iso- These two dates are comparable, if allowance is chron analysis, detailed geochemical investigation, made for calibration of the 14C timescale (e.g. Bard and multiple methods of dose rate determination. et al., 1990, 1993). These attributes have characterised the series of In comparison, Upper Palaeolithic Aurignacian benchmark studies conducted by the ¯int dating tools (thought to be produced by fully modern team at Gif-sur-Yvette since their three iconoclastic humans) appeared as early as 40,000 years ago in papers were published a decade ago (Valladas et al., western Europe. At El Castillo Cave, L'Abreda 1986, 1987, 1988). An excellent summary account Cave and Abric RomanõÂ in northern Spain, uncali- of these and their subsequent TL investigations of brated AMS 14C dates of circa 38,500±39,700 years burnt ¯int is given by Mercier et al. (1995a), and BP (Bischo et al., 1989; Cabrera Valdes and their results for the Levant and western Europe are Bischo, 1989; Hedges et al., 1994), ESR dates of illustrated in Fig. 5. 039,900 years (calculated using ``one-group'' theory These Middle Palaeolithic dating studies have for beta dose attenuation; Rink et al., 1996) and demonstrated that both Neanderthals and modern mass-spectrometric U-series dates of 042,600 years humans inhabited the Levant between 0120,000 and (Bischo et al., 1994) have been obtained for the 060,000 years ago, and are associated with the basal Aurignacian levels (which contain artefacts but not human remains). Taken together with the same (Mousterian) lithic industry. Burnt ¯ints as- 14 sociated with Proto-Cro-Magnon remains in Israel TL, C and ESR results for Saint-CeÂ saire, Arcy- sur-Cure and Le Moustier, these dates suggest that were dated to 119,00018,000 years at Es Skhul Neanderthals co-existed with modern humans for (Mercier et al., 1993) and 92,0005000 years (or several millennia in south-west Europe (Mercier et 88,0009000 years by the isochron method) at al., 1991; Straus, 1989), having been the sole occu- Qafzeh (Aitken and Valladas, 1992; Valladas et al., pants of the continent since at least the last inter- 1988). At the nearby cave site of Kebara, the layer glacial (Mercier et al., 1995a). containing the skeleton of an adult Neanderthal TL dating of burnt ¯ints has also demonstrated was TL dated to 59,9003500 years (or that the Mousterian sequence at Combe Grenal, a 57,0008000 years by isochron analysis) and the famous rock shelter in the Dordogne district of most recent unequivocally Mousterian layer was France, spans the period circa 44,000±63,000 years dated to 051,900 years (Mercier et al., 1995a; (Bowman and Sieveking, 1983) and, therefore, mar- Valladas et al., 1987). These ®rst reported ®ndings ginally precedes the nearby Le Moustier sequence of a substantial (but not necessarily continuous) (Mellars, 1986). For the ``Charentian'' Mousterian temporal overlap between Neanderthals and mod- lithic industry in the Massif Central, ¯int dates ern humans were supported by ESR dating suggested a minimum age of 078,500 years, this (Schwarcz et al., 1988, 1989; Stringer et al., 1989), being the earliest known occurrence of this type of and provided the basis for discarding ideas of a des- industry in France (Raynal and Huxtable, 1989). cendant evolutionary relationship between the two However, the Mousterian site of Pech de l'AzeÂ IV, hominid types. in the Dordogne, yielded TL dates that the authors 836 R. G. ROBERTS deemed to be ``too low to be acceptable'' (Bowman lithic industry (the ``Colombanian'') and the ®rst et al., 1982, p. 368), although no physical causes use of ®re by Europeans (possibly Homo erectus). could be identi®ed. Their cautious view is contin- TL is now being used to check the controversial gent on the climatically assigned ages for the ESR dates of circa 380,000±465,000 years obtained Mousterian layers being correct. Six of the seven from burnt quartz pebbles and sediments. The ESR dated ¯ints gave ages of 043,000 years, a result not palaeodoses were determined using the Al centre in con¯ict with the abrupt Mousterian/Aurignacian (FalgueÁ res et al., 1991), the absence of the E' centre transition recorded at circa 40,000±43,000 years in providing an assurance that the samples had been northern Spain (Bischo et al., 1988, 1989, 1994; heated in antiquity to above 5008C (FalgueÁ res et Straus, 1989). al., 1994a). A possible hearth of similar antiquity The Lower Palaeolithic site of Hoxne in eastern has been excavated at SchoÈ ningen in Germany, England provides another example of ¯int dates together with numerous ¯int tools, bones of butch- that appear to be younger than independent age ered mammals, and the world's oldest hunting estimates: the average TL date of 210,00020,000 spears (Thieme, 1997; Dennell, 1997). TL dating of years for two burnt ¯ints (Bowman, 1993) diers the ®re-baked sediments and burnt ¯ints would be markedly from the mean ``modelled'' ESR date of instructive as the site has not been dated by numeri- 319,00038,000 years (Schwarcz and GruÈ n, 1993). cal age methods. The latter was calculated from combined ESR and The ®rst appearance of another developmental U-series dating of two horse teeth that exhibited trait in human behaviour, the attachment of late uptake of uranium. The cause of the chrono- handles to stone tools using a mastic, has also been logical discrepancy seems to lie in the estimates of the subject of a recent TL dating study. At the the external dose rate: the ESR study used parental open-air site of Umm el Tlel in Syria, the presence radionuclide concentrations in the sediments of bitumen on stone tools has been interpreted as adhered to the teeth, whereas Bowman (1993) made evidence for the manufacture of hafting material by in situ gamma spectrometry measurements on the Middle Palaeolithic people (BoeÈ da et al., 1996; S. deposits adjacent to the original excavation (made Holdaway, 1996). A minimum age for this techno- several years earlier). The calculated ESR external logical innovation was established by TL dates of dose rate was only 65% of the on-site determi- 034,000 and 036,000 years for burnt ¯ints from nation, the latter being broadly supported by lab- 14 oratory analyses of the in situ sediments. two overlying layers, and by uncalibrated C ages for these same layers of 032,000 and 034,530 years Radionuclides are, therefore, distributed non-uni- 14 formly at the site or the sediments stuck to the BP, respectively. The dates given by C are teeth are not representative of the majority of the younger by 1500±2000 years, a shortfall similar in deposit. The second possibility has been discussed magnitude to previous suggestions for this time above for Tabun Cave, but in contrast to Tabun interval which were based on comparisons with (and other sites; Mellars et al., 1997), the dose rate calendrical age methods (e.g. luminescence, ESR of the sediment attached to the Hoxne teeth is less and U-series) and variations in geomagnetic inten- than that of the bulk deposit. The TL and ESR sity (see Section 2.2). chronologies can be brought into coincidence by using the same external dose rate, but an indepen- 3.2.3. Archaeosediment applications. Compared dent evaluation of the latter cannot be reliably with the rich human history that TL dating of obtained from isochron analysis of the two ¯ints. anthropogenically burnt objects has revealed for Bowman (1993) was unsuccessful in her attempt to Eurasia, TL and optical dating of archaeosediments obtain a third chronology by TL dating of the ®ne- have so far played a minor role. An early attempt grain sediments: the sediment date (circa 145,000± to date wind-blown archaeosediments was made at 170,000 years by my calculation, depending on dose Plaidter Hummerich in north-west Germany, using rate assumptions) was lower than the burnt ¯int ®ne-grains and a regenerative-dose procedure ages, a result in keeping with many reports of TL (Singhvi and Wagner, 1986; Singhvi et al., 1986b). age shortfalls in >100,000 year-old samples com- The in situ remains of a Neanderthal, together with posed of ®ne-grain feldspars (Berger, 1994; stone artefacts, had been discovered in loess depos- Debenham, 1985; Wintle, 1990). its, beneath a humus layer estimated to be TL techniques have also been applied recently to 080,000 years old. The archaeological horizon was a fourth site in the Dordogne, the strati®ed cave underlain by loess dated by TL to 0135,000 years, site of Grotte XVI, where baked sediments from a so an age bracket of 80,000±135,000 years was pro- Mousterian ®replace were dated to 060,000 years posed for the Neanderthal. Given the propensity (Rigaud et al., 1995). Another TL investigation of for anomalous fading in feldspars (Spooner, 1994b), ®replaces is currently underway at Menez-Dregan 1, and the fading of 015% exhibited over a nine-week a collapsed sea cave on the southern coast of period by a second sample from the 0135,000 year- Brittany (Balter, 1995; Monnier et al., 1994). This old horizon (Singhvi et al., 1986b), these ages site has yielded evidence of a distinctive, archaic should perhaps be regarded as minima. LUMINESCENCE DATING IN ARCHAEOLOGY 837

This precaution was recommended by ZoÈ ller et (Rees-Jones et al., 1997). The coarse-grain quartz al. (1991), who observed anomalous fading over a component of these deposits has been investigated nine-month period in ®ne grains extracted from using OSL and varying the width of the stimulation loess deposits at two Middle Palaeolithic sites in the waveband. Palaeodoses were determined using the Middle Rhine valley of Germany. TL ages were multiple-aliquot additive-dose method, but no dates determined, using a regenerative-dose procedure, were calculated. For one of their samples (GOR1), for the open-air sites of ToÈ nchesberg and Remagen- the two narrowest wavebands yielded palaeodoses Schwalbenberg. The oldest archaeological horizon of 477 Gy (514.5 nm from an Ar-ion laser) and at ToÈ nchesberg was bracketed by a minimum TL 4015 Gy (51417 nm from a ®ltered halogen age of 0120,000 years and a maximum age of lamp), both of which dier markedly from the 0200,000 years (from 40Ar/39Ar laser dating of the palaeodose of 135 Gy obtained using a broader underlying volcanic tephra, e.g. Bogaard et al., halogen-lamp passband (440±560 nm). 1987), indicating that archaic Homo sapiens inhab- Their observation contrasts with my own for an ited this region during the harsh climatic conditions Australian quartz sample (K166) examined in an of the penultimate glaciation. At Remagen- identical manner (i.e. multiple aliquots, the addi- Schwalbenberg, the loess layer containing transi- tive-dose method, integration of the total light sum tional Middle/Upper Palaeolithic artefacts was TL (minus background), saturating exponential curve dated to 31,3002600 years and this date was sup- ®ts (with each aliquot weighted by the inverse ported by two uncalibrated 14C ages (on snail shell) square of its OSL intensity), and a preheat of of 028,000 years BP, making it the youngest known 2208C for 300 seconds). This sample gave a TL lithic industry transition in Germany. Taken palaeodose of 23.11.0 Gy (no preheat, 270±4308C together with the contemporaneous occurrence of plateau region), and OSL palaeodoses of 24.60.7, Aurignacian tools elsewhere in the region and in 24.61.5 and 22.91.9 Gy using the 514.5 nm south-west Europe, these ®ndings suggest that a laser line and wavebands of 50040 nm and 420± ``mosaic'' pattern of cultural evolution prevailed at 550 nm, respectively (Roberts et al., 1994a and this time (ZoÈ ller et al., 1991). unpublished data). The degree of scatter among Optical dating has only recently contributed to normalised aliquots is much smaller, and the sensi- the debate on the timing of stone tool transitions tivity to radiation is much greater, in the Australian and modern human occupation of the Levant and quartz, resulting in palaeodose uncertainties of 3± western Europe. At Holon in Israel, red loamy soil 8% for K166 compared with 15±38% for GOR1. containing Lower Palaeolithic ¯int tools has been Nevertheless, the quartz OSL exhibited by the dated by both OSL (on quartz) and IRSL (on feld- Gibraltar coastal cave deposits appears suciently spar) to 0220,000 years, a result in agreement with reproducible and dose-responsive to encourage dat- the age of 0215,000 years obtained from ESR dating of hominid activities and lithic industries at one ing of bovid teeth (Porat et al., 1996). This date for of the most likely points of entry for early hominids a Late Acheulian lithic industry is consistent with into Europe (Dennell and Roebroeks, 1996). the date of 0200,000 years for Acheulian handaxes The potential for optical dating of unheated, as at Pontnewydd Cave in north-west Europe (Green, well as burnt, ¯int tools has also been investigated 1984; Green et al., 1981) and with the beginning of recently (Poolton et al., 1995). Flakes and grains of the Middle Palaeolithic in Africa by 200,000± unheated ¯int from France, Israel and Florida were 230,000 years ago (Wendorf et al., 1993, 1994). It is, tested for their dose-response and OSL behaviour however, much younger than the ¯int dates of under 420±550 nm stimulation. The ¯ints exhibited 300,000±350,000 years obtained for the Acheulian- an extremely rapid decay of OSL when illuminated, bearing units at Tabun Cave (Mercier et al., 1995b). and a growth curve was generated with added dose. If the ¯int tools recovered at Holon had been burnt The source traps for the OSL may be those associ- suciently, then comparative TL dates would be ated with the TL peak observed at 1808C, no pre- instructive; further collection of material would be heat having been administered. This peak showed impractical as the site is now built over, but the en- no fading over a period of 24 h, but its long-term vironmental dose rate could be determined from stability remains in doubt (given the 700-year life- previous in situ gamma spectrometry measurements time of the 1908C trap in quartz; Aitken, 1985). and radionuclide analyses of the sediment samples The dissimilar behaviour of ¯int and sedimentary (Porat et al., 1996). quartz was attributed to the existence of dierent Optical dating oers the prospect of dating the crystal lattice defects in the two materials, despite termination, rather than the initiation, of the both being composed largely of SiO2 (Poolton et Middle Palaeolithic on the island of Gibraltar, al., 1995). located at the western extremity of the On the other hand, a rather more optimistic out- Mediterranean Sea. Middle and Upper Palaeolithic come has been reported for optical dating of quart- stone artefacts are present at Gorham's Cave and a zite pebbles from the archaeological site of Diring variety of lithics occur in Vanguard Cave, deposited Yuriakh in Siberia (Richards, 1994). Quartz grains in a matrix of silt, sand and raised beach sediments were extracted from the sun-bleached exteriors of 838 R. G. ROBERTS the pebbles in an attempt to date them directly directly if the ¯uvial deposit and late Pleistocene using OSL and IRSL methods, and then extend the Homo erectus fossils are contemporaneous, or if the method to quartzite artefacts. This study is dis- fossils are reworked from an older deposit (as some cussed further in the next section, together with have argued, see Gibbons, 1996c, and GruÈ n and other contributions that luminescence dating has Thorne 1997). It could also provide an independent made to the chronology of human prehistory in the check on the validity of (a) the uranium-uptake Orient. models used to derive the ESR ages (which increase by 50±70% from 027,000 years to 40,000± 46,000 years under assumptions of early-uptake and 3.3. East Asia linear-uptake, respectively) and (b) the application South-east Asia has vied with Africa as the of `òne-group'' theory to model beta dose attenu- favoured location for the birthplace of Homo since ation in enamel, which increases the quoted ESR Alfred Russel Wallace returned to England in 1862, ages by 04000 years (Swisher et al., 1996). after 8 years in the Malay Archipelago, and per- Homo erectus is generally thought to have lacked suaded Charles Lyell that `òur progenitors'' might the necessary intellectual, linguistic and technologi- be discovered in the caves of Borneo (Desmond and cal capacity to make sea voyages across Wallace's Moore, 1991, p. 523). The island of Borneo lies on Line, but contrary evidence has recently emerged the Sunda Shelf, the name given to the emergent with the discovery of ¯aked stone tools on the landmass which existed to the east and south of the island of Flores (Morwood et al., 1997; Sondaar et Malay Peninsula when sea-level has been lower; al., 1994). The artefacts were recovered from a ¯u- Borneo, Sumatra and Java were connected to the vial sand and silt layer, whose age is estimated to mainland, while the Philippines, Sulawesi and the be just less than 780,000 years, based on the pre- Indonesian island chain as far east as Timor were sence of a palaeomagnetic reversal (presumed to be separated by short water crossings. These water the Brunhes/Matuyama boundary) in the under- barriers, although narrow, have been sucient to lying deposits. If Homo erectus was able to make restrict the movement of land animals and birds, such sea crossings, then perhaps this species arrived resulting in distinct faunal distributions on either in Australia before Homo sapiens, and at a much side of ``Wallace's Line'' (Wallace, 1860; van earlier date than either the currently accepted Oosterzee, 1997). Borneo, Sumatra and Java lie to ®gures of 40,000±60,000 years (Allen, 1989, 1994; its west, while to the east are Sulawesi and the Roberts et al., 1990a, 1994a) or the more controver- chain of islands stretching from Lombok to Timor. sial claims of circa 120,000±180,000 years (Fullagar Still further east lies the continent of Australia and et al., 1996; Kershaw, 1994; Singh and Geissler, the islands of Irian Jaya/Papua New Guinea, which 1985). have been connected by dry land (the Sahul Shelf) Before such far-reaching questions are con- at times of lower sea-level. Human colonisation of sidered, however, the artefact-bearing layer on Sahul is discussed later; this section is concerned Flores must be dated directly or be bracketed by re- with the chronology and taxonomy of hominids in liable age determinations. A variety of numerical the Sunda region. dating methods are now being applied to this site: luminescence dating of the ¯uvial artefact-bearing 3.3.1. The Malay Archipelago. In the early 1890s, sediments (whose expected age lies close to the limit the Dutch palaeontologist Eugene Dubois discov- of reliability claimed for current luminescence ered the remains of Homo erectus (then called Pithe- methods, e.g. Berger et al., 1992; Huntley et al., canthropus erectus) in Java (Dubois, 1894). These 1993a, 1994a,b), ESR dating of associated animal fossils were promoted as the ``missing link'' between teeth, and ®ssion-track dating of overlying and humans and apes, and placed this region at the underlying volcanic tus. This latter study has been forefront of the debate on hominid evolution completed recently (Morwood et al., 1998), and in- (Lewin, 1989). The anthropological signi®cance of dicates zircon ®ssion-track ages of circa 800,000± south-east Asia has since been maintained, boosted 880,000 years for stone tools. This work also throws by recent revelations of Homo erectus habitation on light on the role played by Homo erectus in the Java from as long ago as 1.8 million years (Swisher extinction of endemic island fauna, as the disap- et al., 1994) or one million years (de Vos and Son- pearance from the fossil record of giant tortoise daar, 1994), to as recently as 27,000±53,000 years and pygmy Stegodon, a type of elephant, appears to ago (Bartstra, 1988; Bartstra et al., 1988; Swisher et coincide with the human colonisation of Flores al., 1996). These dates have been derived from the (Morwood et al., 1998; Sondaar et al., 1994). application of 40Ar/39Ar, ESR, U-series, ®ssion- track and palaeomagnetic methods, but lumines- 3.3.2. The Indian sub-continent. Luminescence cence dating could be used to provide additional dating has already provided some important clues chronological control on those sediments deposited to the timing of human occupation of the Asian in the last 800,000 years. For example, optical dat- mainland. Homo may have colonised the Indian ing of the Solo River terraces in Java could show sub-continent as early as 2 million years ago, based LUMINESCENCE DATING IN ARCHAEOLOGY 839 on palaeomagnetic and stratigraphic analyses of a and detrital contamination of the tephra (Shane et sedimentary sequence in northern Pakistan that has al., 1995, 1996). This study shows how luminescence yielded in situ stone artefacts (Dennell et al., 1988; dating of sun-bleached sediments can provide a Rendell et al., 1987). These artefacts are similar in valuable check on the stratigraphic integrity of a age and technological simplicity to other recent deposit whose age has been inferred using methods ®nds in Java (Swisher et al., 1994) and China (W. that give the time of mineral formation and not Huang et al., 1995; Larick and Ciochon, 1996), but necessarily the most recent episode of deposition. con®rmation of their great antiquity by numerical The same concerns have been expressed in connec- age estimates would be desirable. Luminescence tion with the early hominid sites in Java (Swisher et methods have not been tested on these Plio-Pleisto- al., 1994), where 40Ar/39Ar dating of hornblende cene sediments but they have been applied to crystals will overestimate the true age of the fossils nearby late Pleistocene archaeosediments (Rendell if the crystals have been reworked from older and Dennell, 1987). TL dates of 24,000± deposits (Gibbons, 1994b); the cystals would then 64,000 years were obtained from ®ne-grain loess date the volcanic event that reset the argon ``clock'' using the additive-dose method and, for one sample but not the target event of archaeological interest. also, the ``partial bleach'' method (Berger, 1988; Acheulian tools have not been found in Sri Wintle and Huntley, 1982). Anomalous fading was Lanka (Kennedy and Deraniyagala, 1989), but TL not detected over a 4-week period, but concerns dating of a later lithic tradition has proved feasible were raised about contamination of the older loess at Beli Cave (Abeyratne, 1994). This rainforest cave units by poorly bleached sediments derived from has yielded human remains from a level previously local gullying. A conservative age of 42,000± dated by 14C to 012,000 years BP (Kennedy and 45,000 years was thus proposed for the associated Deraniyagala, 1989), while geometric microliths ¯aked stone tools (including blades) and the have been recovered from layers with 14C ages of remains of a small structure, which may have been 3000±27,000 years BP (Abeyratne, 1994). These 14C used for the processing of animal hides (Rendell ages, obtained on charcoal, were broadly con®rmed and Dennell, 1987). by TL dating of quartz grains extracted from Remarkably few applications of luminescence heated sediments, although the four Pleistocene TL dating to archaeosediments have been made in dates were consistently younger than their uncali- India, despite its strategic location between Africa brated 14C counterparts (by 1000±3000 years, and and East Asia and the diverse types lithic industry calibration would increase this disparity). The lack that are poorly dated. The earliest appearance of of in situ gamma dose rate estimates may be partly Acheulian tools is presently placed at circa responsible, if the non-uniform distribution of dose 540,000±670,000 years in the Pune District of west- rate in the hearth-dominated layers at Hayonim central India (Mishra et al., 1995, Mishra and Cave (Mercier et al., 1995c) is any guide; so too Rajaguru, 1996). This ®gure is based on ®ssion- may the repeated use of ®res which would reset the track, K±Ar and 40Ar/39Ar ages (Horn et al., 1993; TL (but not the 14C) ``clock'' on each occasion. Mishra et al., 1995), and minimum U-series ages of The latter was considered as a possible expla- 350,000 years (Mishra and Rajaguru, 1996), for a nation for the OSL age underestimation observed at tephra that is overlain by an artefact-bearing ¯uvial Batadomba Cave in Sri Lanka, where 14C ages on gravel unit. Acheulian handaxes of similar antiquity charcoal, shell and bone, and ESR ages on shell had been reported previously from northern and bone, had also been obtained (Abeyratne et al., Pakistan (Rendell and Dennell, 1985), their age 1997). The most self-consistent suite of results was being constrained by the underlying Brunhes/ generated from 14C dating of charcoal and optical Matuyama palaeomagnetic boundary at dating of heated quartz sediments. The OSL chron- 0780,000 years (Shackleton et al., 1990; Spell and ologies are, however, systematically younger than McDougall, 1992; Tauxe et al., 1992) and the onset the calibrated 14C ages; the greatest discrepancies of folding of the artefact-bearing strata at (of 7000±12,000 years) occur in the lower, inten- 0400,000 years. sively ®red levels, which were 14C dated to circa The Indian dates and previous stratigraphic in- 23,000±31,000 years. The basal level at Batadomba terpretations have been challenged, however, by the Cave yielded geometric microliths and human correlation of the tephra mineralogy, rare earth and remains coated with yellow pigment (Kennedy and trace element chemistry, and shard morphology Deraniyagala, 1989). Given the calibrated 14C age with the 074,000 year-old Youngest Toba Tu in of 031,000 years and two OSL ages of 019,000 and Indo-Malaysia (Shane et al., 1995). Luminescence 024,000 years for this level, geometric microliths methods have also entered this debate by providing appear to have predominated at an earlier date in a TL date of 023,000 years for the tephra (Horn et southern Asia than in Europe, where they did not al., 1993), suggesting that this unit was last rise to prominence until after the Last Glacial reworked in the late Pleistocene. Bias in the earlier Maximum (Gamble, 1986, p. 120). age determinations, as well as the stratigraphic Further support for the antiquity of microlithic inconsistencies, are consistent with ¯uvial reworking technology in Sri Lanka is given by the TL dates of 840 R. G. ROBERTS circa 22,700±28,300 years obtained for in situ micro- to 475,000125,000 years (Ciochon et al., 1996; W. liths in the coastal dunes at Bundala (Singhvi et al., Huang et al., 1995). 1986a). This study is also notable in two other In northern China, a variety of numerical dating respects. First, it demonstrated that red iron-oxide methods have been employed to provide a reliable coatings on quartz grains can dramatically reduce chronology for the famous ``Peking Man'' site at the bleaching of the TL signal at both the easy-to- Zhoukoudian. This cave site has yielded skulls, bleach (3258C) and hard-to-bleach (3758C) peaks; bones and teeth of Homo erectus, more than exposing red-stained grains to a sunlamp for 100,000 stone tools, the remains of fossil animals, 1000 min resulted in only 30% and 20% reductions and claimed evidence (in the form of ash and hearth in the TL signals at 3258C and 3758C, respectively. lenses) for the use of ®re by prehistoric humans (It should be remarked, however, that the residual (Black et al., 1933; Jia and Huang, 1990). A group TL at 3258C is probably composed mostly of the of Chinese geochronologists have recently published low-temperature ``tail'' of the 3758C signal, and a multi-method dating comparison for this import- that the ``pure'' 3258C (and hence OSL) signal may ant site, including a single TL date of circa have been erased completely.) Second, it considered 290,000±310,000 years for a baked hearth in Layer the possibility that uranium may be precipitated 4 (Huang et al., 1993). This date agrees well with with iron in the red coatings, resulting in a time- the ESR, U-series and ®ssion-track dates for this dependent change in the dose rate as the coatings layer, although the reliability of the TL age deter- thicken with burial time. Singhvi et al. (1986a) con- mination cannot be gauged as no procedural details cluded that the major eect of uranium precipi- are given. The lowest ashy layer, interpreted as a tation would be on the external alpha dose rate, hearth, occurs at the boundary of Layers 9 and 10, which is mostly removed during etching of the the latter having a TL date of 420,000± grains with HF acid. They also showed that if the 590,000 years (cited in Chen and Zhang, 1991, red coatings were acquired after deposition of the again with no supporting data) and ®ssion-track and amino-acid racemization dates of grains, then reliable TL ages can be obtained from 0460,000 years. The most complete skull found at the application of ``total bleach'' methods to sand- this site was retrieved from Layer 11, whose age is sized quartz grains prepared in the standard manner estimated to be 460,000±585,000 years (using ESR (the stains being removed during acid-etching). and ®ssion-track methods). These deposits lie within Grains that were transported and deposited with an the age range of luminescence methods and a new existing red coat, however, are apt to have been study has begun to date them using OSL for the incompletely zeroed in antiquity, resulting in gross unburnt sediments and TL for the baked hearth age overestimates using ``total bleach'' TL methods. sediments; ESR dates on tooth enamel will be avail- Optical dating should be more accurate and, fur- able for comparison (Huang and GruÈ n, 1996). thermore, as every grain in a sample is unlikely to Luminescence methods also have unful®lled po- be coated to the same thickness, single-aliquot and tential to date the transition between Homo erectus single-grain techniques should produce a broad dis- and Homo sapiens in China. Several uranium-series tribution of palaeodoses if the grains were not and ESR dating studies suggest that this transition bleached uniformly, as has been reported for other took place between 200,000 and 350,000 years ago, partially bleached sediments (e.g. Duller et al., with the possibility of coexistence of the two species 1995; Li, 1994; Murray et al., 1995; Olley et al., (Brooks and Wood, 1990; Chen and Yuan, 1988; 1997b; Wintle et al., 1995). Chen and Zhang, 1991; Chen et al., 1994; Huang and GruÈ n, 1996). Luminescence dating methods 3.3.3. China. The fossil record of human evol- have so far contributed little to the chronology of ution in the more northern parts of Asia is poorly this crucial period of human prehistory, but they constrained by numerical age estimates. The pre- should be applied more often to check the validity sence of Homo (a more archaic form than Homo of the U-series and ESR age determinations, es- erectus) in southern China has recently been dated pecially given the detection of widespread `òpen to 1±2 million years (W. Huang et al., 1995), an age system'' uranium migration in fossil bones and consistent with early hominid sites in Java and teeth (Chen and Yuan, 1988). Pakistan. The ages are derived from the palaeomag- The Hexian Man site in Anhui Province, east- netic pro®le and ESR dating of tooth enamel at central China, has yielded fossils of late Homo erec- Longgupo Cave, where primitive stone tools have tus and has been TL dated to less than been found in association with fragmentary homi- 200,000 years (cited in Brooks and Wood, 1990 and nid remains and the teeth of Gigantopithecus, a Chen and Zhang, 1991). No details are given, how- large primate. Hominids appear to have coexisted ever, to allow a check of the internal consistency of with Gigantopithecus for more than one million the TL dating procedure, although the age is not years in south-east Asia, as Homo erectus and dissimilar to those obtained from associated fossil Gigantopithecus fossils are found together at Tham teeth by U-series (150,000±200,000 years: Brooks Khuyen Cave in Vietnam, in levels dated by ESR and Wood, 1990; Chen and Yuan, 1988) and ESR LUMINESCENCE DATING IN ARCHAEOLOGY 841

(170,000±400,000 years: P.-H. Huang et al., 1995; reverted to a permanent seaway at 011,000 years Huang and GruÈ n, 1996). The cave sediments at this BP (Elias et al., 1996). The oldest ``Palaeoindian'' site have, unfortunately, been excavated completely sites in Alaska have been 14C dated to almost and none remain for further TL or OSL dating (P.- 12,000 years BP (Hoecker et al., 1993; Kunz and H. Huang, Personal Communication, 1994). The Reanier, 1994), and a ¯uted bifacial point (a tech- same is not true at other cave sites, such as that nology thought to be restricted to the Americas) near Nanjing where two skulls attributed to Homo has been discovered in north-eastern Siberia, erectus were discovered in 1993 (Etler, 1996); sedi- beneath a tephra dated by 14C to 08300 years BP ments and teeth have recently been collected from (King and Slobodin, 1996); a maximum age for the this site for optical and ESR dating. Another im- ¯uted point, and thus its temporal relation to lithic portant site is that of Yunxian Man, where two cultures in North America, could be established by skulls with a mix of Homo erectus and archaic luminescence dating of the sand deposit on which Homo sapiens features were discovered recently on the artefact rested. a terrace of the Han River (Li and Etler, 1992). The As with the date for human colonisation of the age of the terrace is estimated to be circa 580,000± Americas (the subject of the next section), the time 850,000 years, based on palaeomagnetic analysis of of initial human arrival in Siberia remains a matter the sediments and ESR dating of associated teeth of ®erce debate. Some archaeologists have claimed (Etler, 1996; Chen et al., 1997). To re®ne this age 14C dates of up to 45,000 years BP for palaeolithic estimate, luminescence dating of the terrace sedi- sites in Siberia (see review by Dolitsky, 1985), ments has begun. whereas a recent evaluation of 14C dates for the Much also remains to be learnt about the timing region place the earliest human occupation at of the later palaeolithic sequences in China. Some 12,000±14,000 years BP (Kuzmin and Tankersley, U-series ages have been obtained from sites with 1996). Mitochondrial DNA data support the view fossil animal remains, but the geochronologists re- that north-east Asia provided the founding popu- sponsible have warned that several of these may not lation for North America, but indicate that prob- be reliable owing to secondary mobilisation of ably only one ``wave'' of people (rather than the uranium, particularly in samples from open-air sites four waves proposed by linguists) emigrated from on river terraces (Chen and Yuan, 1988). Siberia, at circa 20,000±25,000 years ago (Gibbons, Luminescence dating of archaeosediments can play 1996b). an important role at such open-air sites. For Into this debate has stepped luminescence dating. example, in situ stone tools were discovered in 1987 In the mid 1980s, TL dates of 1.5 million years at Xiangyang, on an abandoned terrace of the (with an uncertainty of 25%) and 02 million Shuiyang River in Anhui Province (Fang et al., years were claimed by Y. Mochanov for the sites of 1992). ESR dating of the artefact-bearing quartz Ulalinka and Diring Yuriakh, respectively sediments has produced ages of 300,000± (Ackerman and Carlson, 1991; Dolitsky, 1985, pp. 800,000 years for a 10 m-thick section composed of 364 and 375; Mochanov, 1992). The great antiquity loess and alluvium (Fang et al., 1992; P.-H. Huang, that these TL dates bestowed on the associated peb- Personal Communication, 1996). The well-known ble tool industry led to much controversy, and the ESR signals in quartz cannot be reset completely by anthropogenic origin of the Ulalinka ``artefacts'' sunlight (Brumby and Yoshida, 1994; Schwarcz, has been challenged. The details of the dating 1994) and only a few studies have reported ESR methods and results are not available in English, dating of solar-reset sediments based on these but as the dates lie well outside the demonstrated incompletely-bleached signals (e.g. Laurent et al., limit of reliability (0800,000 years) for current TL 1994; Tanaka et al., 1995; Yokoyama et al., 1985). dating practices, they should still be regarded as It is likely, therefore, that these age determinations questionable. Recently, however, the Diring are maxima, so luminescence dating methods have Yuriakh site on the Lena River has been re-exam- now been deployed at this site. TL dates for other ined using optical methods on quartzite pebbles palaeolithic sites along the Shuiyang River are cited (Richards, 1994) and TL methods on the artefact- by Fang et al. (1992), but procedural details are not bearing sediments (Morell, 1994; Waters et al., provided. Sites throughout this region of central 1997a,b). These new results, supported by full docu- China also warrant further investigation, given its mentation of the experimental procedures, suggest strategic location on the boundary between the dis- an age in excess of 110,000 yearsÐan order-of-mag- tinctive north and south Asian lithic industries nitude greater than the currently accepted chronol- (Fang, 1994). ogy for human occupation based on 14C determinations. 3.3.4. Siberia. Palaeolithic cultures in the far Over 4000 artefacts have been collected at Diring north of Asia have also come under the lumines- Yuriakh, from an occupation surface which is bur- cence ``spotlight'' recently. Siberia is a key region in ied beneath a variable thickness of sediment (Fig. 6). terms of human migration from Asia into North Waters et al. (1997a,b) provide evidence that the America, across the Bering land bridge that only covering deposit is composed of aeolian sand and 842 R. G. ROBERTS

Fig. 6. Generalised cross-section of the stratigraphy at Diring Yuriakh and the associated TL ages in years (and sample codes). Solid triangles indicate the lag surface on which stone artefacts have been found (from Waters et al., 1997a. Copyright 1997 American Association for the Advancement of Science). loess, in contrast to the suggestion made by the allowances made for the cosmic-ray dose rates and original excavator that the artefacts are buried by sample water contents. Measurements of the in situ alluvium (Mochanov, 1992). The artefacts are gamma dose rate were not made but, fortunately, mostly made from quartzite and are also the subject the annual doses show little variation (<10%) of dispute among archaeologistsÐa natural origin within any particular aeolian unit (e.g. 2.96± being argued by some (e.g. R. Klein in Holden, 3.14 mGy in Unit IV, and 3.68±3.99 mGy in Unit 1997), while others view them as the product of IIIe). The absence of signi®cant 222Rn escape was human hands (e.g. Ackerman and Carlson, 1991; inferred from the ``unsealed'' versus ``sealed'' alpha Mochanov, 1992; Waters et al., 1997a,b). The latter count ratio of 0.96±1.01, but no checks were per- group has argued that discrete clusters of unifacial formed to detect other forms of disequilibrium in choppers, cores and scrapers, ¯ake debris, and large the 238U chain. To overestimate the TL ages by an anvils occur on a ``lag'' surface of wind-abraded order-of-magnitude, however, requires that the pebbles, produced by de¯ation of the underlying measured (modern) dose rates be one-tenth the water-lain sands. long-term average; neither the total loss of uranium Owing to the lack of materials suitable for other in the distant past, nor the recent leaching of 226Ra numerical dating methods, luminescence dating was or emanation of 222Rn, would produce a sucient the sole technique available to determine the ages eect (Olley et al., 1996). By my calculation, 50± of the sedimentary units above and below the 70% of the total dose rate is derived from potass- archaeological horizon. The ®ne-grain sediments ium, and doubling the entire U-series contribution were extracted from these units and their TL would raise the total dose rates (and thus reduce palaeodoses were determined using the additive- the TL ages) by only 10±25%. dose method (Waters et al., 1997a,b). The TL in In terms of the TL measurements, the sand units polymineral extracts is dominated by the signal immediately above and below the artefact horizon from feldspars, but ®ne grains of quartz were ad- yielded palaeodoses of 0770 Gy and 01150 Gy, re- ditionally analysed for one sample. Given a sub- spectively, corresponding to bracketing ages of aerial mode of transport and deposition, Waters et 267,00024,000 years and 366,00032,000 years al. (1997a,b) assumed that the TL signal had been (Waters et al., 1997a,b). The additive-dose growth reset in antiquity to a low ``residual'' level, which curves and palaeodose plateaux are shown for both was estimated by exposing grains to both natural these key samples by Waters et al. (1997b). In both sunlight and an ultraviolet sunlamp. Tests for cases, the ``natural'' TL intensity is 060% of the anomalous fading indicated little or no fading over saturation intensity, so the dose-response curves periods of 1±6 months, but the TL ages would be require substantial extrapolations to intersect the too young, not too old, if long-term fading is a pro- ``residual'' level. Furthermore, the palaeodoses tend blem. to rise with temperature across the glow curve The environmental dose rates were deduced from region 250±4008C, rather than form a perfectly ¯at thick-source alpha counting (uranium and thorium) plateau. The palaeodoses used in the age calcu- and ICP emission spectrometry (potassium), with lations are conservative estimates, however, as they LUMINESCENCE DATING IN ARCHAEOLOGY 843 are biased towards the smaller palaeodoses associ- attempted at Diring Yuriakh by S. Forman (who ated with the TL peak at 03008C; the residual TL found the IRSL signal was in dose-saturation; levels are low (<10% of the natural TL) at this Gibbons, 1997c), and in a pilot IRSL dating study temperature also, so the signal-to-noise ratios are of the water-lain sands beneath the artefact layer optimal. Waters et al. (1997b) recognise the TL (Hu, 1994). The lack of dependence of IRSL inten- ages of these two samples may not be accurate, and sity on radiation dose precluded an estimate of the instead suggest that the ®ve TL dates obtained palaeodose from feldspar inclusions, but a mini- from the overlying loess (Unit IIIe) provide a re- mum age of 0290,000 years was calculated for the liable minimum age for the Diring Yuriakh arte- same layer using coarse grains of potassium feld- facts. spar and assuming that these grains had been well These ®ve dates range from 240,00019,000 bleached at deposition (Hu, 1994). It has been years to 284,00024,000 years (the latter date is found, however, that even wind-blown sediments given in Waters et al., 1997b, and is not shown in have not always been exposed to sucient sunlight Fig. 6), and the dose-response curve and palaeodose to reset the IRSL signal (Huntley and Lian, 1997; plateau are illustrated for one of these samples. Lamothe and Auclair, 1997). This sample shows a similar dose-response beha- On the other hand, D. Huntley and M. Richards viour to the older pair discussed above, but the (Personal Communication, 1997) have cautioned palaeodoses form an eectively ¯at plateau (at that the Diring Yuriakh artefacts may be much 01000 Gy) from 2508C to at least 3508C (Waters et older than the TL ages suggested by Waters et al. al., 1997b), inspiring greater con®dence in the age (1997a,b). They have argued that, provided the sedi- determination. Intriguingly, the ®ne-grain quartz ments received adequate solar exposure, then the extracted from this same sample gave an identical TL ages measured by Waters et al. (1997a,b) should palaeodose of 0990 Gy. In my experience, this rep- be treated as lower limits. This is based on the ob- resents an extraordinarily large palaeodose for servation that feldspars suer from long-term quartz grains, whether they be silt or sand sized (cf. instability of the TL signal, resulting in substantial Huntley et al., 1993a, 1994a; Rees-Jones, 1995). The age underestimates (Mejdahl, 1988b). The presence palaeodose plateau region for this sample (250± of a palaeomagnetic reversal, interpreted as the 4008C) is also unlike that commonly reported for Brunhes/Matuyama boundary (e.g. Alekseev et al., quartz; for a sample estimated to be 0250,000 years 1990) or an earlier event (e.g. Ackerman and old, the partial or complete decay of the 2808C Carlson, 1991; Mochanov, 1992) in the sediments peak would yield a plateau that starts at 03408C above the artefact-bearing layer suggests that the and, provided the hard-to-bleach 3758C peak had artefacts may be older than 0780,000 years. been fully reset in antiquity, extends to at least Although burnt ¯ints are not available at Diring 4508C (e.g. Huntley et al., 1993a). Yuriakh for comparative dating, steps toward lumi- It is possible that Waters et al. (1997a,b) have nescence dating of unheated quartzite artefacts have instead exploited the presence of micro-inclusions been made. (presumably feldspars) within the ®ne-grain quartz, Richards (1994) made several innovative investi- as has been reported for sand-sized quartz from gations on the feasibility of using TL and optical South Australia (Huntley et al., 1993b) and inferred dating methods to measure the time since buried for other silt-sized quartz data (see comments by D. quartzite pebbles were last exposed to sunlight. For Huntley following Rees-Jones, 1995). Support for the Diring Yuriakh pebbles, an extended period of this proposition is given by the ®ndings of Richards sub-aerial (and thus solar) exposure is indicated by (1994), who used the IRSL signal to date sand-sized their wind-abraded surfaces. Natural river pebbles grains of quartz extracted from the Diring Yuriakh were collected at night (by D. Huntley) from Diring pebbles. As quartz does not respond to infrared Yuriakh, to characterise the luminescence behaviour stimulation at room temperature (Short and of the local quartzite and to develop suitable dating Huntley, 1992; Spooner, 1994a), feldspar inclusions procedures. In addition, Richards examined a peb- were deemed responsible for the IRSL signal, their ble which had been excavated one year earlier and presence being detected also by X-ray diraction had since been exposed to sunlight on the ground analysis (Richards, 1994). surface. Initial petrographic and XRD analyses of The TL dates for Diring Yuriakh are spectacu- three quartzite pieces revealed that two were com- larly old, and require con®rmation. If the overlying posed wholly of quartz, whereas one also contained sediments were emplaced by water, rather than by albite, a feldspar species which appears not to suer wind, then it becomes critical to demonstrate that severely from anomalous fading (Spooner, 1994b). the sediments were well bleached at deposition; the As the Diring Yuriakh pebbles are not translu- TL ages might otherwise be considerable age over- cent, only the external portion of each pebble estimates. Optical dating of feldspars and feldspar would be suitable for optical dating of the solar- inclusions in quartz grains, to exploit the rapidly reset minerals. It was therefore necessary to adopt a bleached IRSL signal, would provide a useful check ``micro-stratigraphic'' approach and remove grains (W.J. Rink in Holden, 1997). The latter has been from the surface of the pebbles and from increasing 844 R. G. ROBERTS depths, to determine the maximum depth to which each ``HF layer'' corresponded to a thickness of sunlight had penetrated (akin to the methodology 250 mm. of Liritzis, 1994). By etching the pebbles in 48% The sand-sized grains from each HF layer were HF acid for successive periods of 30 min, Richards then examined for their TL, OSL and IRSL bleach- (1994) obtained a layer-by-layer suite of granular ing and dose-response characteristics. Although extracts, the silica matrix having been dissolved by each quartzite behaved dierently, OSL and IRSL the acid. He calculated that 10% of the mass decay curves were generated using 514 nm and removed was recovered as quartz grains, and that 880 nm stimulation, respectively, and sunlight was

Fig. 7. (a) Shine-down curves for quartzite pebble DY-1d under 2.4 eV (514.5 nm) stimulation (22 mW cm2 on 10 mg aliquots). Grains of 90±150 mm diameter were extracted from each ``HF layer'' after the pebble had been exposed to direct sunlight for 12.5 h and to cloudy skies for 5.5 h. The data for ``no sun exposure'' are from grains removed before the pebble was exposed to sunlight, and the ``24 hr bleach (grains only)'' data are from grains exposed to a sunlamp for 24 h. (b) The same data as in (a) but expressed as the sum of photon counts for the ®rst 10 s of stimulation (from Richards, 1994). LUMINESCENCE DATING IN ARCHAEOLOGY 845 found to be eective in bleaching the TL and OSL great an impact on the palaeolithic archaeology of signals. For the purpose of dating the pebbles, how- northern Asia as TL dating has done in western ever, it was necessary to establish the penetration Europe and the Levant. The early human occu- depth of sunlight, so experiments were conducted pation of Siberia also provides an increased time- with quartzite slices of varying thickness. For one frame for these ®rst Siberians to travel across the sample, light transmission declined to 8% of the in- Bering land bridge into North America. The earliest itial intensity using a 0.6 mm slice while, for traces of human arrival in the Americas, and the another sample, the amount of light transmitted chronological contributions made by luminescence through a 0.25 mm slice was reduced to less than methods, are considered in the following section. 1%. A similar result was obtained from a comparison of OSL signal intensities for individual HF layers removed from quartzite pebbles: the eect of 4. CONTINENTAL AND ISLAND 8±12 h of solar exposure was shown to extend COLONISATIONS through the outermost 6±8 layers (equivalent to 4.1. The Americas 1.5±2 mm), with the remanent signal increasing with depth (Fig. 7). As the same treatment removed This paper began with a quote from Thomas 020% of the TL signal (at 350±4008C) in the outer- Jeerson concerning the age of some archaeological most HF layer but had nil eect on the deeper structures in Virginia. His interests extended also to layers, sun-bleached pebbles and artefacts should be the antiquity of the indigenous American people dated using optical, rather than thermal, stimu- (Jeerson, 1787), the so-called ``Palaeoindians''. lation. The results of these laboratory experiments The time of their arrival in the Americas, and the were supported also by OSL tests on an excavated number of founding populations, is still being hotly pebble left exposed on the ground surface for one contested two centuries later by archaeologists, lin- year. The top (exposed) surface of the pebble was guists and geneticists (Bednarik, 1989; Bray, 1988; bleached through to the eighth HF layer, whereas Gibbons, 1993, 1996b; Gruhn, 1988; Lynch, 1990; the underside of the pebble showed no bleaching of Marshall, 1990; Meltzer, 1989, 1995). Perhaps the any of the outer ®ve layers. only point of agreement is that the ``Clovis'' culture Richards (1994) then proceeded to examine a was present in North America by 11,000± quartzite pebble from the archaeological horizon, 11,500 years BP (e.g. Haynes, 1993; Holliday, 1997; using the single-aliquot additive-dose IRSL pro- Taylor et al., 1996). A recent multi-method dating cedure of Duller (1991); regenerative-dose exper- study of a Palaeoindian campsite in the Brazilian iments on other quartzite pebbles had shown the Amazon has shown that people were present also in OSL signal was too weak and unresponsive to South America by this time (Roosevelt et al., 1996). dierent treatments, or the OSL data were too scat- tered, to make palaeodose determinations. Owing 4.1.1. Pre-Clovis colonisation. Earlier human oc- to the small size of the pebble, only 20 mg of 106± cupation has been claimed for sites such as Mea- 250 mm diameter grains could be extracted. Five ali- dowcroft Rockshelter in Pennsylvania (Adovasio et quots were used, one as a ``monitor'' for signal ero- al., 1978; Adovasio et al., 1990), Monte Verde in sion due to the repeated preheat (1208C for 16 h) Chile (Dillehay and Collins, 1988), Guitarrero Cave and short-shine (5 s) stimulation cycles. The IRSL in Peru (Lynch et al., 1985), Taima-taima in Vene- signal, however, remained constant with added zuela (Bryan et al., 1978), and a numer of sites in dose, indicating the sample was in dose-saturation. Brazil, including Alice BoeÈ r (BeltraÄ o et al., 1982; A minimum palaeodose of 0400 Gy was determined Danon et al., 1982), Pedra Furada (Bahn, 1993; by comparing the ``natural'' IRSL signal of each Delibrias et al., 1988; Guidon and Delibrias, 1986; aliquot with the corresponding signal regenerated Guidon and Arnaud, 1991; Guidon et al., 1994) after an initial 3 h bleach (using a quartz-halogen and Toca da Esperanc a (de Lumley et al., 1988). At lamp ®tted with a Schott RG-715 ®lter). This yields each of these sites, except the last, 14C determi- a minimum elapsed time of 110,000 years since the nations have been made on organic remains. The pebble was last exposed to sunlight, given a total reliability of these 14C ages and their association annual dose of 03.6 mGy (deduced from thick- with the cultural remains has been disputed at some source alpha counting, atomic absorption and neu- sites (e.g. Meadowcroft Rockshelter; Tankersley tron activation, with adjustments for cosmic rays, and Munson, 1992, and rejoinder by Adovasio et beta-dose attenuation and soil water content). al., 1992), while the human manufacture of ¯aked The minimum IRSL age for this pebble is consist- stones has been queried at others (e.g. Pedra Fur- ent with the TL age determinations for the enclos- ada; Meltzer et al., 1994, and reply by Parenti et ing sediments (Waters et al., 1997a,b). The al., 1996). The latter is an archaeological not a archaeological horizon at Diring Yuriakh appears, chronological issue, but luminescence dating has therefore, to be reliably dated to at least been deployed at remarkably few sites to check 110,000 years, and possibly 0260,000 years or more. ambiguous 14C dates and provide a more detailed These luminescence dates will doubtless make as chrono-stratigraphy. 846 R. G. ROBERTS

Luminescence methods could be used to good facts, which Reagan et al. (1978) interpreted as the eect at the river terrace site of Monte Verde, result of insucient heating in antiquity. Although which Meltzer (1995) considered to be the most the lowest lithic culture was not dated directly, the viable pre-Clovis candidate. The oldest cultural overlying TL ages indicate that initial site occu- layer is sandwiched between ¯uvial sand and rede- pation must be older than 15,000 years, possibly posited volcanic ash units, and has yielded one dating back into the last glaciation (Reagan et al., ®nite 14C age of 033,400 years BP and an in®nite 1978). As TL dating of cherts was then a new devel- determination of >33,000 years BP (Dillehay and opment and no material was available for 14C dat- Collins, 1988). Given the latter result, and the poten- ing, this site has received scant attention from tial for trace amounts of modern carbon to greatly archaeologists (MacDonald, 1983; Stanford, 1983), reduce the apparent 14C ages of samples older than even though there now seems little reason to doubt 30,000 years (Aitken, 1990), the lowest cultural layer the accuracy of the published TL ages. at Monte Verde may be of similar antiquity to the TL dating of heated lithics has also been exploited basal levels at Pedra Furada, dated to >48,000 years at the Brazilian sites of Alice BoeÈ r, Pedra Pintada BP (Guidon and Arnaud, 1991; Guidon et al., 1994). and Pedra Furada. The southern site of Alice BoeÈ r To resolve this uncertainty, the 14C ages of the low- was examined initially as 14C dates of up to est cultural layer and the upper 012,500 years BP 14,200 years BP had been obtained previously from level (the latter now generally accepted as pre-Clovis: the upper cultural layer (BeltraÄ o et al., 1982; Danon Gibbons, 1997a; Adovasio and Pedler, 1997; et al., 1982). Heated cherts were again used, Meltzer, 1997; Meltzer et al., 1997) could be checked although only 20% of the total number tested by optical dating of the adjacent sediments. proved suitable for dating. The internal radioactivity Luminescence dating could also be deployed to of the cherts was measured and the environmental advantage at the Brazilian cave site of Toca da dose rate was determined from in situ gamma spec- Esperanc a, where ¯aked stone tools have been trometry and from lithium ¯uoride and calcium ¯u- found in association with faunal remains (de oride dosimeters. Checks for anomalous fading were Lumley et al., 1988). Three bones from the artefact- made, and none was detected after a period of bearing layer at the base of the deposit were dated 3 months. TL ages of 2000±11,000 years were to circa 200,000±300,000 years using alpha and obtained for Layer III, but the artefacts from the gamma spectrometry U-series methods. Provided deepest levels of this layer had not been heated su- these dates are accurate, this is by far the oldest ciently to reset the TL clock. These ages are consist- archaeological site yet discovered in the Americas. ent with the 14C determinations, and suggest that The site has been given little credence by most the lower cultural stratum (Layer V) might have archaeologists, but if humans were present in been deposited more than 20,000 years ago (BeltraÄ o Siberia more than 260,000 years ago (Waters et al., et al., 1982). Numerical age estimates for the latter 1997a,b) then perhaps the U-series dates for Toca await the deployment of AMS 14C methods and a da Esperanc a are not so outlandish. Optical dating renewed search for burnt cherts (Danon et al., of the basal clayey sand layer could be used to 1982), but it would now seem prudent to add TL check the U-series dates, including those determined and optical dating of sediments to this list. by non-destructive high-resolution gamma spec- TL dating of sediments received early application trometry; recent applications of the latter technique to Palaeoindian sites in San Diego, California to hominid bone have yielded dates of variable (Berger and Huntley, 1986). Burnt clay from a quality (e.g. Barton and Stringer, 1997; Berzero et hearth preserved in terrace sediments of the Otay al., 1997; Latham, 1997). River was dated to 014,000 years using convention- TL dating of heated chert (i.e. ¯int) artefacts al TL techniques for pottery, while a date of from the Shriver site in Missouri (Reagan et al., 020,000 years was obtained from unheated sedi- 1978) represents, to my knowledge, the earliest ments associated with human occupation of the involvement of luminescence methods with the Scripps Tower site. The latter date was considered American colonisation question. Burnt cherts from unreliable, as it had been obtained from a buried the top and bottom of a Palaeoindian ¯uted point Ah soil developed on mixed colluvial/aeolian sedi- horizon yielded TL ages of 10,6401000 and ments and as it con¯icted with a 14C date of 14,8001500 years. Few procedural details are 09900 years BP for the same occupation. At a third given, other than the palaeodoses were determined site (Fletcher Wash), artefact-bearing gravels were from thin polished slices (following GoÈ ksu et al., bracketed by TL dates of 09000 and 28,000 years. 1974) and the annual environmental dose These dates were also treated with caution as they (1.78 mGy) was deduced from calcium ¯uoride dosi- had been determined from slopewash and alluvial meters buried on site for six months. The sediments whose TL signals may have been incom- Palaeoindian horizon overlay the lowest cultural pletely reset by sunlight at the time of deposition. layer, which contained a dierent lithic assemblage To diminish the latter problem, optical dating of composed mainly of unifacial tools and ¯akes. A sediments has been exploited recently in the search wide range of TL ages were obtained for these arte- for the ®rst Americans. At the famous Clovis site LUMINESCENCE DATING IN ARCHAEOLOGY 847

(Blackwater Draw No. 1 locality) in New Mexico, 33330 mGy, respectively, while the corresponding aeolian deposits containing Palaeoindian artefacts ages are 10,570800 and 10,580660 years have been dated by 14C and OSL methods, the lat- (Feathers, 1997a; Michab et al., 1997; Roosevelt et ter employing coarse-grain quartz stimulated by an al., 1996). The three older lithics (14,800± Ar-ion laser (Stokes, 1992). The 14C chronology 17,500 years) in the same layer were postulated to obtained for the youngest artefact-bearing unit indi- be remnants of an earlier period of site habitation cated a terminal date for deposition of either circa which had been incorporated into the artefact 6200±6500 or 7900±8800 years BP (Haynes and assemblage during the subsequent occupation Agogino, 1966; Stokes, 1992). Support for the (Michab et al., 1997). The quartz sediment dates younger determination was given by an optical date (017,000 years) determined using the hard-to-bleach of 5950600 years for the base of the overlying TL signal could be interpreted as supporting this unit, the dose rate being deduced from thick-source theory. If the sediments deposited at the site had alpha counting (uranium and thorium) and ¯ame been fully bleached 017,000 years ago (when the photometry (potassium). The lower units with site was ®rst occupied), but only a brief exposure to ``Folsom'' and ``Clovis'' artefacts were not optically sunlight had occurred during later sedimentation dated, but any such attempt may be foiled by events, then only the easy-to-bleach TL and OSL reworking of the sediments and associated artefacts signals would record the time since the most recent via the action of ancient groundwater springs and episode of insolation. trampling by mammoths. The ®rm chronology established at Pedra Pintada shows the unequivocal presence of humans on the 4.1.2. The early Amazonians. The ideal combi- Amazon River by the terminal Pleistocene. The re- nation of optical dating of sediments, TL dating of lation of these forest dwellers to the Clovis people heated lithics, and AMS 14C dating of short-lived is, however, more contentious (Gibbons, 1996a; plant remains (such as fruit seeds) has been success- Fiedel, 1996; Roosevelt, 1996). There would appear fully employed at Caverna da Pedra Pintada in the to be no climatic or biogeographic reason to Amazonian rainforest (Michab et al., 1997; Roose- exclude the possibility of earlier occupation of the velt et al., 1996). The 56 AMS 14C dates ranged rainforest which has persisted throughout the last from 10,00060 years BP to 11,145135 years 40,000 years, the Last Glacial Maximum being BP, which corresponds to a calibrated age range of characterised by a cooling of 5±68C and an increase circa 10,500±14,200 years. The TL dates for 10 in rainforest species diversity (Colinvaux et al., burnt artefacts (composed of quartz sandstone) 1996; Stute et al., 1995). were divided into two groups: circa 9500± The site of Pedra Furada oers the strongest case 11,900 years (seven specimens) and circa 14,800± for 50,000 years or more of human colonisation of 17,500 years (Michab et al., 1997). All 10 artefacts the Americas (Bahn, 1993; Delibrias et al., 1988; were dated using methods described previously Guidon and Arnaud, 1991; Guidon et al., 1994; (Mercier et al., 1995a; Valladas, 1992), and no Parenti et al., 1996). Forty-six of the 55 14C deter- cause could be found to reject the three oldest TL minations made on abundant pieces of charcoal are ages. The three optical (OSL) dates centered on the currently accepted by the excavators, the oldest period between 11,400 and 12,000 years ago. The sample returning an uncalibrated AMS age of OSL palaeodoses were determined using a com- >48,000 years BP. Like Caverna da Pedra Pintada, bined additive/regenerative single-aliquot procedure this large rock shelter has been painted, and frag- with sand-sized quartz grains that had possibly ments of painted rock were found in deposits dated been heated in antiquity (Michab et al., 1997). to 032,000 years BP, with a pictograph (two red Attempts to use the hard-to-bleach 3758C TL signal painted lines) at the 017,000 years BP level (Guidon in the sediments were again unsuccessful: the regen- and Delibrias, 1986). The 5 m-deep sandy deposit erative-dose palaeodoses were 45% greater than also contains 0160 hearths and more than 7000 their OSL counterparts, as a result of the TL signal stone artefacts, the basal assemblage being made of having been incompletely bleached or heated in ¯aked quartz and quartzite. antiquity (Feathers, 1997a; Michab et al., 1997). Critics have argued that iron pigments from the The environmental dose rates were calculated nests of mud-daubing insects could have created the from high-resolution gamma spectrometric analyses pictograph (Bednarik, 1989), that the operation of of collected sediment samples, and no disequili- geological processes could have produced chipped brium in either the 238U or 232Th decay chains was stones, and that natural bush-®res could have detected. On-site measurements of the environmen- formed the hearths and introduced the charcoal tal dose rate were not made at Pedra Pintada, but (e.g. Lynch, 1990; Meltzer et al., 1994). These possi- isochron analysis of the group of seven burnt lithics bilities have been addressed by the excavators lends support to the laboratory determinations of (Guidon and Arnaud, 1991; Guidon et al., 1994), the external dose rates and to the TL ages of the in- with TL analysis of hearth stones providing one dividual specimens: the measured and isochron esti- line of evidence against the bush-®re argument mates of the annual external doses are 33622 and (Parenti et al., 1990, 1996). The method proposed 848 R. G. ROBERTS by Valladas (1981) to determine the temperature the arrival of people, in this region (FalgueÁ res et reached in prehistoric hearths indicated that 78% of al., 1994b). quartz cobbles and 14% of sandstone fragments Humans were hunting big game in Venezuela by from a hearth feature had been heated to above 13,000±14,000 years BP, however, as shown by the 4508C. This is considerably greater than the maxi- presence of a quartzite projectile point in the pelvis mum mean surface temperature reached during of a juvenile mastodon at Taima-taima (Bryan et natural ®res in Wyoming (225±2508C; Bellomo, al., 1978). Both human predation and climatic 1993) and Western Australia (2908C; Koch and change may have played a role in the demise of the Bell, 1980). Ground surface temperatures of 5508C South American megafauna since the Last Glacial have been reported, however, for bush-®res in the Maximum (Dillehay et al., 1992), but if people ®rst open forests of northern Australia (Braithwaite and arrived tens of millennia earlier then perhaps there Estbergs, 1985), although such high temperatures was a period of faunal attenuation, as yet unrecog- are reached only brie¯y during the passage of the nised and undated, that set the stage for their post- ®re. Moreover, as the immediate sub-surface sedi- glacial collapse. Faunal remains are not preserved ments are heated to less than 1008C (e.g. Koch and in the acid sediments at Pedra Furada (Guidon et Bell, 1980), natural bush-®res are unlikely to be of al., 1994), but rich and diverse megafaunal assem- sucient severity or duration to erase the high-tem- blages have been found in association with stone perature TL of the interior portion of a cobble- tools in several neighbouring limestone caves sized stone. Further support for an anthropogenic (Parenti et al., 1996). As organic material is too origin for the hearths is given by the spatial distri- scarce for 14C dating (Parenti et al., 1996), lumines- bution of heated stones: these were clustered on one cence methods could be applied to these mega- side of a hearth (14C dated to 042,400 years BP), fauna-bearing sediments, as well as those at Toca suggestive of site cleaning and re-use, rather than da Esperanc a (de Lumley et al., 1988), to determine being distributed uniformly as might be expected if the late Pleistocene chronology of mammalian they had been heated by a bush-®re (Parenti et al., extinctions in Brazil. 1990, 1996). TL dating has been deployed recently in this ca- TL dating of the Pedra Furada hearth stones is pacity in North America, where the poorly pre- in progress, and optical dating of the sediments is served remains of an adult mastodon were found in also being considered (F. Parenti, Personal 1973 in a loess section on the banks of the Missouri Communication, 1996). Any application of optical River (Berger, 1996; Chapman, 1975; Dunnell and dating will need to take account of in situ rubble Hamilton, 1995). Accelerator 14C dating of the disintegration, which would liberate unbleached bone collagen fraction had provided two dates of grains into the deposit. Multiple-aliquot dating pro- 035,800 years BP, and TL dating of the immedi- cedures would be susceptible to sample contami- ately underlying loess gave a ``partial bleach'' age nation by these unbleached grains, and could yield of 41,7006100 years. The dierence of considerable age overestimates (Roberts and Jones, 05900 years is consistent with other uncalibrated 1994). Single-aliquot or single-grain optical dating 14C age comparisons for this time period (see methods should be employed under such circum- Section 2.2). The TL sample was obtained from the stances (e.g. David et al., 1997) to ensure accurate sediment ``pedestal'' attached to the plaster-cast age determinations. ®eld specimens held at a museum, the ®nd site having been completely destroyed. The uranium- 4.1.3. Man meets beast. Luminescence methods enriched bone was shown to have a negligible eect also have unrealised potential to elucidate the tim- on the gamma dose rate (Berger, 1996), and the ing of human arrival in the Americas and the substantial dierence in 238U concentrations in the extinction of many species of large mammal (the bone and TL samples (010.6 and 03.1 ppm, re- so-called ``megafauna''). If Pedra Furada was ®rst spectively) argues against signi®cant uranium occupied by humans more than 50,000 years ago, enrichment of the sediment close to the bones (cf. then it becomes less likely that megafaunal extinc- Mercier and Valladas, 1994; Mercier et al., 1995b). tions in the terminal Pleistocene were due mainly to The apparent association of the mastodon with a hunting by the ®rst Americans (Bray, 1986). The ¯ake knife-scraper made of chert (Chapman, 1975, impact of massive climatic change has been impli- pp. 53±54) opens up the intriguing possibility that cated in North America, the homeland of the people were present in North America, and hunting ``overkill'' hypothesis (Martin, 1973, 1990), by the big game, by 42,000 years agoÐa situation reminis- lack of synchroneity between human arrival and cent of Australia, the subject of the next section. mammalian extinctions (Grayson, 1989, 1991). In northern Peru also, U-series dating of fossil bones 4.2. Australia has indicated that extinction of the megafauna occurred 15,000±16,000 years ago, coincident with The question has frequently presented itselfÐAre there any geological traces of man on this Continent, such as deglaciation and the onset of aridity, but predating exist in other countries, and whereby the presence of a former race, or the antiquity of the present fast disappearing LUMINESCENCE DATING IN ARCHAEOLOGY 849 one, can be traced? The answer given by those most com- Americas. Applications have been concentrated at petent to judge isÐNo! (Etheridge, 1890, p. 259) sites where other numerical dating methods cannot be applied, owing to the paucity of suitable materials, and where the practical limits to 14C dating To what cause, it may be asked, is due the extinction in have been reached. The scarcity of heated ¯ints has Australia of the genera Diprotodon, Nototherium, prevented direct dating of stone artefacts, but the Thylacoleo, Phascolonus, Thylacinus, Sarcophilus, dating of archaeosediments has made as great an Palorchestes, Procoptodon, Pachysiagon, Protemnodon, impression on Australian archaeology as ¯int dating Sthenurus, with the larger species of existing genera of Kangaroos and Wombats? No adequate cause suggests has in the Middle East and Europe. A discussion of itself to my mind save the hostile agency of man. (Owen, these and other luminescence contributions to 1877, v. 1: viii) Australian archaeology is presented in this section; site locations are shown in Fig. 8. The antiquity of the human colonisation of A century of research has produced no concen- Australia, and its link with the extinction of the sus, however, on the contention by Owen (1877), a giant marsupials, reptiles and birds, has been specu- prominent anatomist and adversary of Charles lated upon for more than a century. Since this time, Darwin, that human predation was the primary much has been learnt on both matters. The view cause of megafaunal extinctions. Some researchers expressed by Etheridge (1890), the palaeontologist hold that extinction followed immediately after, or to the Australian Museum, was ®nally abandoned within a few millennia of, human landfall in the 1960s when 14C age determinations pushed a (Flannery, 1990, 1994; Jones, 1968), whereas others human presence back towards 40,000 years (see consider that megafauna and humans co-existed for reviews by Jones, 1989, 1993). Luminescence dating tens of millennia (Diamond, 1989; Merrilees, 1968; has, over the past decade, extended this timeframe Wright, 1986a) with climate change, or a combi- still further, often fomenting controversy equal to nation of factors, dealing the mortal blow (e.g. that surrounding claims for pre-Clovis sites in the Dodson, 1989; Gorecki et al., 1984; Horton, 1984).

Fig. 8. Location map of archaeological and megafaunal sites discussed in the text. 850 R. G. ROBERTS

As faunal attenuation over several glacial-intergla- life, is 28,650 years BP. TL ages obtained by cial cycles may have preceded the ®nal extinction Huxtable and Aitken (1977) using their ®ne-grain events, it is crucial to investigate the time period samples were considered to be overestimates, owing beyond the range of 14C dating. Faunal remains to spurious TL from carbonates. Again in contrast, suitable for direct dating by ESR, U-series or Adams and Mortlock (1974) reported no such pro- amino-acid racemization are not found at every site blems and produced a ®ne-grain TL date of in Australia, so luminescence methods are particu- 30,4003300 years, in close accord with the quartz larly well placed to provide the ``baseline'' chronol- TL ages of Huxtable and Aitken (1977) and the 14C ogy for Pleistocene faunal extinctions; this date of 031,000 years BP for this hearth. This good chronology can then be compared with that of agreement would appear to be fortuitous, as Bell human colonisation, based on 14C and luminescence (1991) subsequently reported that these ®ne-grain ages. Within this archaeological context, some sediments suer from anomalous fading of 10±30% recent and forthcoming applications of lumines- over a period of 3 months. He managed, however, cence dating to megafauna-bearing sediments are to obtain TL dates from baked quartz sand from mentioned below. four ®replaces (mean 14C age of 28,220 years BP), including the one that had frustrated Adams and 4.2.1. Lake Mungo. The ®rst archaeological site Mortlock (1974). Neither an explanation for this to capture the attention of luminescence daters was discrepancy, nor the temperature regions of his Lake Mungo, which at times during the Pleistocene palaeodose plateaux, are supplied by Bell (1991), formed part of a chain of freshwater lakes along whose arithmetic mean age of 33,500 years is identi- Willandra Creek in south-western New South cal to that derived independently by Huxtable and Wales (Bowler, 1971). Long crescent-shaped aeolian Aitken (1977). Both of these studies show that dunes (or ``lunettes'') developed along the eastern uncalibrated 14C determinations underestimate TL shorelines of these lakes, to which people and a var- ages by 05000 years at circa 33,500 years ago, a iety of fauna (including megafauna) were attracted ®nding that was ®rst noted by Zimmerman and during more humid periods. Severe erosion of the Huxtable (1971) and which has since been broadly Mungo lunette (at a section known as the ``Walls of con®rmed by other calendar-year dating methods China'') in the late 1960s revealed animal fossils, as (see Section 2.2). well as stone artefacts, ®replaces, shell middens and The lunette sediments at Lake Mungo also the cremated remains of two people (Bowler et al., formed the basis of Readhead's (1982) pioneering 1970). Soon after these sensational discoveries, the eorts to apply TL dating techniques to unburnt ``Mungo I'' female was dated by 14C to archaeosediments in Australia. Readhead (1982, 25,5001000 years BP, and the skeleton of a third 1988, 1990) used sand-sized quartz grains together anatomically modern human (named ``Mungo III'') with an early version of the ``Australian slide'' com- was located in ochre-stained grave ®ll (Bowler and bination of ``total bleach'' additive-dose and regen- Thorne, 1976). A 14C date of 033,000 years BP was erative-dose protocols. A series of aeolian sand obtained from freshwater shells, which lay alongside samples were collected from the three major sedi- stone artefacts on a de¯ated surface, and charcoal mentary units exposed in the ``Walls of China'' sec- from several Aboriginal ®replaces yielded ages of tion: the stratigraphically youngest Zanci unit, the circa 26,000±31,000 years BP, during which period underlying archaeological Mungo unit, and the a local geomagnetic polarity excursion also basal, culturally-sterile Golgol unit. Dose rates were occurred (Barbetti and Allen, 1972; Barbetti and calculated from laboratory measurements only, McElhinny, 1972; Bowler et al., 1972). using ``sealed'' versus ``unsealed'' thick-source alpha The global signi®cance of the Mungo ®nds pre- counting to gauge the degree of radon escape; 0± sented an ideal opportunity to test TL dating 26% emanation was observed, whereas none was methods on the baked hearth sediments and oven- detected by thick-source alpha counting of the stones. The ®rst TL application was made to baked baked hearth sediments from the Mungo unit (Bell, sand from a hearth dated by 14C to 031,000 years 1991). Readhead (1982, 1988) obtained TL dates of BP. Both ®ne-grain and quartz inclusion methods 022,000 years for the Zanci unit (compatible with were attempted, although the quartz samples failed the 14C ages for this layer) and circa 24,500± the ``plateau test'' for constancy of palaeodose 29,300 years for the upper part of the Mungo unit, above 3508C (Adams and Mortlock, 1974). This but reliable TL ages could not be determined for implies that the 3758C TL peak in baked quartz the Golgol unit samples, owing to in¯exions in their sand had not been thermally erased in antiquity, dose-response curves. He later ventured estimates of whereas the same signal had been reset in coarse- circa 220,000±290,000 years for the Golgol sedi- grain quartz extracted from the ovenstones. The lat- ments (Readhead, 1990). These ages are consider- ter result was reported by Huxtable and Aitken ably greater than those of circa 140,000± (1977) who computed an average TL age of 180,000 years obtained recently (Oyston, 1996) for 33,5004300 years from 7 determinations; the cor- the base of the lunette using the same TL dating responding average 14C age, using the Libby half- protocols. LUMINESCENCE DATING IN ARCHAEOLOGY 851

Oyston (1996) also applied the ``selective bleach'' unconformity, the TL and 14C dates may relate to TL method (Franklin and Hornyak, 1990; Prescott dierent events; subsequent TL dating of the and Fox, 1990; Prescott and Mojarrabi, 1993; unheated dune sands beneath the ®replace produced Prescott and Purvinskis, 1991; Prescott and ages of circa 8000±14,500 years (see below), compar- Purvinskis, 1993) to quartz sand grains from the able with the 14C age for the charcoal. An oven Mungo and Zanci units; the easy-to-bleach 3258C mound on the Riverine Plain in the central Murray signal being subtracted (selectively bleached) from valley has since been TL dated to 02650 years, using the total TL signal by exposing aliquots to the ®ne grains extracted from baked clay balls used as >475 nm component of sunlight. This approach is heat-retainers during cooking (Downey and Frankel, akin to optical dating and exploits the most light- 1992); this date con®rmed three calibrated 14C deter- sensitive signal, which has a much greater prob- minations on charcoal of circa 2600±2750 years. ability of having been zeroed in antiquity than the TL dating of unheated sediments also received harder-to-bleach signals additionally sampled by early application at Roonka (Prescott, 1983), where total bleach TL methods. The improved accuracy is cultural material and the remains of 216 people shown clearly in the dating of the Zanci and upper (most of which are late Holocene in age) have been Mungo units, for which total bleach ages were excavated from aeolian sand dunes bordering the 14 greater than the selective bleach (and C ages) by river (Pretty, 1988). On the west bank, the dune is 12,100 and 9400 years, respectively (Oyston, 1996). underlain by a culturally-sterile ``terra rossa'' soil. The selective bleach procedure was also used to This terminus post quem for human occupation was bracket the age of the Mungo III burial: an age of dated to 65,00012,000 years using sand-sized 43,3003800 years was determined for the lower quartz grains and a variant of the `Àustralian part of the Mungo unit (into which the burial had slide'' method (whereby the second glow data, been dug) and an age of 24,6002400 years was rather than bleached grains, were used to construct obtained from the upper Mungo unit sediments, regenerative-dose growth curves). On the east bank, which capped the burial. These age constraints for the poorly preserved skeletons were overlain by the Mungo III burial indicate that it is contempora- dune sand dated to 14,5002000 years, although neous with, or older than, the Mungo I skeleton, as Prescott (1983) raised doubts about the degree to anticipated by Bowler and Thorne (1976). No other which the TL signal had been zeroed at deposition. Australian site boasts such ancient human remains Exposing the acid-etched quartz grains to direct (Brown, 1992; Pardoe, 1995). sunlight for a minimum of 60 h reduced the TL sig- The Lake Mungo lunette is probably the most nal to a much lower ``residual'' level than that securely dated archaeological site in Australia. attained by grains collected from the surface of a Radiocarbon methods have been applied to a var- spoil heap formed 5 years earlier. Using the latter iety of materials (freshwater shells providing the residual level instead of the former reduced the age more reliable results for samples older than 20,000± of the sample to 08000 years, more in keeping with 25,000 years BP; Bowler and Wasson, 1984) and 14 comparable TL dates have been obtained from the the C date of 09200 years BP for the oldest burial 3758C and 3258C glow peaks in baked and on the west bank. Palaeodose plateaux extended unheated quartz, respectively; in situ measurements from 3008C to 4208C in both cases, however, which of the dose rate and a direct demonstration of secu- super®cially suggests that both the 3258C and lar equilibrium in the 238U decay series would 3758C peaks had been reset in antiquity. further strengthen the TL claims. A clue to the solution of this riddle was given by Prescott and Purvinskis (1991, 1993), who used the 4.2.2. Roonka and Urana. Prehistoric human bur- ``selective bleach'' method to abstract the ``pure'' ials, ®replaces and stone artefacts have also been 3258C signal from two near-modern Roonka yielded at Roonka, a site on the Murray River in samples. They found that the 3258C signal had been south-eastern South Australia (Pretty, 1988). Radio- fully zeroed but that a signi®cant residual TL signal carbon dates of up to 18,000 years BP have been remained at temperatures as low as 2508C due to obtained from ®replaces, with a late Holocene the hard-to-bleach component. A palaeodose pla- example producing matching 14C and TL ages of teau extending to temperatures as low as 260±2808C 01000 years BP, the TL date being obtained from can be expected in quartz if both the easy-to-bleach heated calcrete ovenstones (Prescott et al., 1983). and hard-to-bleach signals have been fully bleached Spurious TL halted any attempt to date the calcite in antiquity (e.g. Chawla et al., 1992; Roberts et al., directly, but acid dissolution of the ovenstones pro- 1990a, 1994a, 1996). A plateau starting at tempera- duced a sucent quantity of ®ne-grain minerals and tures above 3008C implies dierential bleaching (or sand-sized quartz for dating. A second ®replace was dierential thermal stability, in much older samples) TL dated to circa 2000±2500 years, in contrast to the of the two components, with the hard-to-bleach sig- charcoal 14C date of 011,300 years BP (Prescott et nal being incompletely zeroed. The fact that the pla- al., 1983). The source of this discrepancy is unclear teau extends as low as 3008C does not mean it but, as the ®replace was situated on a stratigraphic includes the 3258C peak: it shows only that the 852 R. G. ROBERTS same palaeodose is obtained from the low-tempera- (Thorne and Macumber, 1972). This site is pre- ture ``tail'' and the peak of the 3758C signal. sently being considered for optical dating, to check The dierence in the residual signals measured the validity of the early Holocene 14C determi- for the acid-etched and spoil heap grains can be sat- nations made previously (T. Stone, Personal isfactorily explained by the yellow-brown coatings Communcation, 1996). on the hominid-bearing dune sands at Roonka. If the grains had been transported with their extant 4.2.3. Graveyards of the giants. South-eastern coatings, then the hard-to-bleach signal is unlikely Australia also boasts a wealth of fossil megafaunal to have been reset completely at deposition (see material that could shed light on the timing of Singhvi et al., 1986a), and the spoil heap grains pro- extinction events. Many of the lunettes bordering vide the best estimate of the residual level. The re- the lakes along Willandra Creek (e.g. Mungo) and sidual level attained by the etched grains would be other nearby lakes (e.g. Menindee, Tandou, Vic- appropriate only if the grains had been stained after toria) contain abundant faunal remains, as well as deposition: if this were true, and solar exposure had recording the palaeohydrology of the region (Bow- been prolonged, then the hard-to-bleach signal ler, 1971, 1976; Bowler and Wasson, 1984; Hope, should also have been fully reset, resulting in a 1982; Hope et al., 1983). These sites will shortly be palaeodose plateau extending to 2708C. The lack of sampled for luminescence dating, in a joint project such a plateau indicates that the grains were stained between the author and T. Flannery, a specialist in prior to transport or that the unstained grains were extant and extinct Australasian mammals. Samples exposed only brie¯y to sunlight. In either case, the have already been collected from two sites (Cuddie 8000 year-old TL date is preferred for the Roonka Springs, Tambar Springs) in northern New South burial, although this may still be an overestimate Wales (Fig. 8), where artefacts have been found in (the ``plateau test'' having failed). Optical or selec- association with megafaunal remains (Dodson et tive bleach TL dating methods should be able to al., 1993; Furby et al., 1993; Wright, 1986a). For resolve this issue and provide reliable age estimates the samples examined thus far, additive-dose OSL for the fossil hominid levels at this important site. methods have produced comparable palaeodoses Human remains have also been discovered in the using multiple and single aliquots of quartz (Mur- lunette at the southern end of Lake Urana, which is ray et al., 1997), indicating that problems of insu- located on the Riverine Plain 0300 km south-east cient bleaching do not appear to plague these of Lake Mungo (Page et al., 1994). The skull and water-lain sediments. post-cranial remains are those of an adult female, A single-aliquot check of multiple-aliquot OSL similar in morphology to the Mungo I female, but palaeodose determinations shall be made also on no evidence was found of grave excavation. Two the megafauna-bearing sediments collected from large grindstones and numerous quartzite artefacts Cox's Creek, where the almost complete skeleton of were later recovered from the site, but the absence a Diprotodon was excavated in 1979 from an erod- of charcoal deposits precluded 14C dating. TL ing stream bank (Wright, 1986a), and Naracoorte samples were collected from the dull orange aeolian Caves in south-east South Australia, which contains sand units above and below the skeleton, to bracket Australia's largest known collection of Pleistocene its age. Sand-sized quartz grains were dated using fossil vertebrates (Wells et al., 1984). The 100 and the combined additive/regenerative ``total bleach'' 200 mm-diameter quartz grains in the entombing method (Readhead, 1982, 1988), but no correction sand deposit at Cox's Creek have yielded prelimi- was made for the residual TL of a surface sample nary OSL dates of 075,000 and 095,000 years, re- (Page et al., 1994). As discussed above, large age spectively, while provisional OSL ages in excess of overestimates have been made for hominid-bearing 170,000 years have been estimated for the Victoria aeolian sands at Lake Mungo (Oyston, 1996) and Cave deposit at Naracoorte. At the latter site, how- Roonka (Prescott, 1983) using the total bleach ever, it remains to be shown that the sediment method in tandem with the residual TL level entered the deep cave system in a fully bleached reached by acid-etched grains. The two dates of state and has not been periodically transported and 25,6007300 and 32,4008000 years (Page et al., redeposited by underground streams. Single-aliquot 1994) should be regarded, therefore, as maximum methods should be able to detect infringement of age estimates; no reassurance is given by the uniform bleaching before the sediment entered the palaeodose plateaux, which begin only at 3008C for cave, but they cannot provide any information on both samples, and no independent age control is subterranean remobilisation. In some chambers, the available at this site. contemporaneity of the megafauna and the cave OSL or selective bleach TL methods could again deposits can be independently tested by U-series be used to eliminate this chronological uncertainty, dating of speleothems and ESR dating of fossil and determine if the Urana hominid is contempora- teeth (K. Moriarty, Personal Communication, neous with the Mungo I specimen. Kow Swamp, a 1996). site 0200 km south-west of Lake Urana, has yielded For a more complete picture of megafaunal at- skeletons of a more archaic type of hominid tenuation and extinction during the latter half of LUMINESCENCE DATING IN ARCHAEOLOGY 853 the Quaternary, sites have been selected from known as the Kimberley. Or they may have jour- around the continent to accommodate possible neyed through to Irian Jaya/Papua New Guinea dierences in timing between dierent climatic and and then made their way to far north Queensland, ecological zones. Several cave and open-air sites through the Cape York Peninsula, or to Arnhem across the south of the continent will be investi- Land at the ``top end'' of the Northern Territory. gated, as shall a number of previously identi®ed This route necessitated only a short sea passage fossil deposits in the deserts of outback Australia during interglacial periods, while the exposed Sahul (Tedford and Wells, 1990; J. Magee, Personal Shelf functioned as a land bridge around the time Communication, 1996). Only those sites with mega- of the glacial maxima. fauna in articulated anatomical position will be The identi®cation of the region of ®rst landfall sampled, to exclude the possibility that the skeletal lies beyond the reach of 14C methods, however, due remains have been reworked from older deposits. to the paucity of organic material in the basal cul- Whenever feasible, luminescence dating of mega- tural levels of the deepest archaeological sequences fauna-bearing sediments will be complemented by in northern Australia (e.g. Malakunanja II, direct dating of fossil remains, such as the eggshell Nauwalabila I, Mushroom Rock). By the late of Genyornis (a genus of extinct giant ¯ightless 1980s, the chronology for continental colonisation bird), using AMS 14C, U-series and amino-acid rested on 14C dates (on charcoal) of up to racemization methods (e.g. Magee et al., 1995; 38,000 years BP from southern sites, such as Lake Miller et al., 1997), and ESR and U-series dating of Mungo, Upper Swan (Pearce and Barbetti, 1981) tooth enamel. and Devil's Lair (Dortch, 1979), and on 14C dates Fossils from several sites have already been (on marine shell) of up to 33,000 years BP from described and some museum specimens have a su- sites in island Melanesia, to the north-east of cient thickness of sediment attached to the outside Australia (e.g. Allen et al., 1988; Wickler and or inside of bones for the application of single-ali- Spriggs, 1988). Some archaeologists (e.g. Allen, quot optical dating methods; enough sediment is 1989) considered these dates to re¯ect initial occu- occasionally fastened also to artefacts stored in pation of the continent, whereas others (e.g. Jones, museums (R. Jones, Personal Communication, 1982, 1989) viewed some of these dates as minimum 1997). The principal bene®ciaries of such methods age estimates, owing to sample contamination by are sites that have been destroyed or are inaccess- modern carbon. The existence, or otherwise, of a ible. A major disadvantage, which applies also to 14C ``glass ceiling'' at circa 35,000±40,000 years BP the dating of museum collections of teeth by ESR continues to be a source of lively debate (e.g. Allen, and pottery by luminescence, is the lack of environ- 1994; Allen and Holdaway, 1995; Chappell et al., mental dosimetry data for the original surrounding 1996a; Roberts et al., 1994b). The application of deposit and the possible bias introduced by estimat- TL dating to archaeosediments was proposed by ing this dose rate from the sediment adhered to the Jones and Johnson (1985) as a means to resolve this specimen (e.g. Mellars et al., 1997; Mercier et al., chronological conundrum and to obtain basal ages 1995b). But if more than one mineral, or size frac- for the north Australian sequences. As a conse- tion of the same mineral, can be extracted from the quence, a TL dating project was begun by the residual sediment, then application of `ìsochron'' or author and R. Jones in 1988 (Jones, 1989, p. 762), ``subtraction'' techniques should eliminate the need and the third member of the team, M. Smith, joined for any knowledge of the external dose rate (see the following year. Since then, TL and optical dat- Section 3.2.1). ing methods have been deployed at archaeological sites in Arnhem Land, Cape York and the 4.2.4. The ®rst Australians. In contrast to Kimberley, with the recent TL claims for Jinmium southern and central Australia, fossil remains of rock shelter (Fullagar et al., 1996) being the latest megafauna and humans are much less abundant to spark controversy over the early human occu- north of the Tropic of Capricorn. The weathering pation of northern Australia. regime of high temperature and humidity, and the naturally acidic monsoonal rainfall (Likens et al., 4.2.5. Western Arnhem Land. The ®rst sites to 1987; Noller et al., 1990), act against the long-term receive attention lay at the foot of the 200 m-high preservation of organic materials such as bone and escarpment that marks the western edge of Arnhem charcoal. Stone artefacts are preserved in tropical Land, 0250 km east of Darwin. Denudation of the environments, however, and these form the basis escarpment and adjoining plateau has resulted in for understanding the archaeology of northern Aus- the formation of sand `àprons'' along the base of tralia, the region through which the ®rst human the cli (Nott and Roberts, 1996; Roberts, 1991, arrivals most likely entered the continent (Jones, published in toto as Roberts et al., 1991). At the 1989). The initial colonists may have travelled from head of two sand aprons lay the rock shelter sites south-east Asia along the chain of Indonesian of Malangangerr and Malakunanja II, both of islands to Timor, and then negotiated the wide sea which had been excavated previously and had crossing to the north of Western Australia, a region yielded 14C dates of circa 18,000±23,000 years BP 854 R. G. ROBERTS for the non-basal levels (Kamminga and Allen, tive/regenerative procedure to determine the palaeo- 1973; White, 1971). In 1988, sand-sized quartz doses and high-resolution gamma spectrometry to grains were extracted from these archaeosediments estimate the environmental dose rates (Roberts et and TL dates were derived using a combined addi- al., 1990a). Thick-source alpha counting of samples

Fig. 9. (a) TL glow curves for natural aliquots of Malakunanja II sample KTL167, prepared under red (solid lines) and yellow (dashed lines) laboratory illumination. Heating rate is 2.5 K s1. (b) Palaeodose ``plateau test'' for aliquots of sample KTL167 processed in red (solid triangles) and yellow (open circles) light. Note partial erasure of the 3258C peak, and failure of the plateau test, under yellow illumination (from Roberts, 1991, and Roberts et al., 1991). LUMINESCENCE DATING IN ARCHAEOLOGY 855 before and after acid etching was used to deduce processed in red light (KTL167r) and the other in the magnitude of the internal dose rate, but no in yellow light (KTL167y). The results obtained were situ measurements of the external dose rate were striking. The red-lit samples invariably passed the made. The laboratory measurements showed, how- ``plateau test'', giving palaeodose plateaux that ever, that the dose rate did not change abruptly began at 260±2708C and extended across both the between levels at any particular site, and that a con- easy-to-bleach and hard-to-bleach TL peaks. For dition of secular equilibrium prevailed in the KTL167, a ``double'' plateau was obtained from the uranium and thorium decay chains. The palaeo- yellow-lit portion and only the palaeodose for the doses were determined from interpolation of the upper plateau agreed with that derived from the regenerative-dose growth curves, as the latter co- long plateau (270±5008C) exhibited by the red-lit incided with the shifted additive-dose data, indicat- portion (Fig. 9b; Roberts et al., 1991); a compari- ing that no signi®cant sensitivity change had son of the natural glow curves shows the magnitude occurred upon sunlight or sunlamp bleaching. of reduction (but not elimination) of the 3258C sig- An interesting anomaly was reported, however, nal due to a few hours of yellow illumination for the palaeodose plateaux of samples processed [Fig. 9(a)]. under yellow (>470 nm) laboratory illumination. This experiment demonstrated that yellow ®lters These showed a distinct ``double'' plateau, the ®rst can result in the inadvertent bleaching of the 3258C at 260±3208C and the second at 340±5008C. Only signal and a foreshortening of the plateau tempera- the latter plateau produced TL dates in correct stra- ture region, but that if red ®lters are installed then tigraphic order, but the failure to obtain the same the palaeodose plateau should start at 02608C for palaeodose from both the 3258C and 3758C regions quartz samples that have not been partially reset raised grave concerns that the easy-to-bleach signal since deposition. The plateau region will be aected had been partially reset, either in the ®eld or in the slightly by heating rate, the 3258C peak shifting to laboratory. Other samples processed in this labora- lower temperatures at slower heating rates. A dier- tory show similar plateaux characteristics (e.g. ence of 0108C might be expected by heating at Murray et al., 1992; Nanson et al., 1993), and a 2.5 K s1 (as used for the Malakunanja II samples) double plateau had also been reported previously instead of 5 K s1 (Spooner, 1994a: Fig. 2), the lat- for quartz sediments collected from the Spencer ter rate being used for the Nauwalabila I and Gulf in South Australia (Smith et al., 1982); in the Allen's Cave samples described below; these were latter instance, the lower plateau (associated with processed in red light and showed a prominent the 3258C peak) was thought to represent the last natural TL peak at 3258C, with plateaux beginning time that the marine sediments had been trans- at 265±2708C (Murray et al., 1997; Roberts et al., ported underwater, while the last exposure to un®l- 1994a, 1996). In well bleached samples older than tered sunlight was recorded by the upper plateau the last interglacial, however, the plateau may not (associated with the 3758C peak). For the archaeo- start until higher temperatures (e.g. Huntley et al., logical samples, various possible mechanisms to pre- 1993a, 1994a; see Section 3.1.1). The Malakunanja ferentially reduce the 3258C palaeodose were II sequence is mostly too young to check the gener- considered (Roberts et al., 1991, 1993a): partial ality of this observation, but the basal TL sample, bleaching of the TL in water-transported grains or prepared under red light and dated to oxide-coated grains exhumed by termite or human 0110,000 years, has a plateau which begins only at activities; thermal erasure of the 3258C peak, and 3258C (Roberts et al., 1990a); palaeodoses in the subsequent increase in TL sensitivity, due to bush- 285±3108C region are 015% smaller (Roberts et al., ®res; thermal instability of the 3258C peak at ambi- 1991). Some of these points will arise later, in dis- ent temperatures of 25±308C (the modern range of cussion of the TL dates for Jinmium (Fullagar et mean annual temperatures across tropical northern al., 1996). Australia) giving rise to a shortened trapped-elec- TL dating of the archaeosediments at tron lifetime; and inadvertent bleaching during Malakunanja II indicated that humans ®rst occu- sample collection or laboratory preparation. pied the site between 61,00010,000 and Cautions had earlier been expressed about the 52,0008000 years ago, the older date being as- use of yellow ®lters for laboratory illumination sociated with the lowest stone artefact (Roberts et (Smith, 1988; Spooner and Prescott, 1986), but pre- al., 1990a). The random and systematic uncertain- liminary tests had given no indications of labora- ties are here added in quadrature, following the tory bleaching (A. Murray, Personal standard practice in luminescence dating (Aitken, Communication, 1987). Nonetheless, acting on the 1985); in Roberts et al. (1990a), the uncertainties suggestion of B. Smith and A. Wintle, red ®lters were summed arithmetically, this being the more were installed and the key samples from conservative approach to total error estimation by Malakunanja II were processed under the >590 nm assigning a maximum limit to the systematic error, transmitted light (Roberts et al., 1990a). rather than assuming a Gaussian distribution Furthermore, to facilitate a direct comparison, one (Roberts et al., 1990b; Roberts and Jones, 1994). sample was split into two portions: one portion was As there is only 10 cm depth of sand deposit 856 R. G. ROBERTS between these two sample mid-points, the possibility to rock shelters, however, it seems improbable that of downward displacement of artefacts cannot be all traces of human activity would be restricted to discounted (e.g. Cahen and Moeyersons, 1977; the deepest recesses (M. Smith, Personal Hughes and Lampert, 1977; Richardson, 1992), Communication, 1997). even though such a mechanism seems inadequate to Certainly, the Malangangerr rock shelter, situated account for the presence of artefacts which dier 10 km north-east of Malakunanja II, warrants markedly in size and density (e.g. silcrete ¯akes, further study. The basal sediments in the shelter pieces of dolerite and ground ochre, including a were TL dated to 32,0007000 years and an over- lump of high-grade haematite, a grindstone, and lying unit was dated to 16,0004000 years, the lat- numerous quartzite and quartz artefacts) in the ter being slightly younger than, but within two zone of ®rst occupation. standard errors of, the calibrated age range of A secure terminus ante quem for occupation is 20,300±27,000 years for four 14C samples (Roberts given, however, by the concentration of artefacts in and Jones, 1994). The basal age may, however, rep- a small pit feature overlain by sediments dated to resent only a minimum age for occupation as the 45,0007000 years. A weighted least-squares linear lowest artefacts rest directly on bedrock and, thus, regression was also performed on the nine could be a ``lag'' assemblage buried by much Pleistocene TL dates in the sequence to obtain a younger sediments (as may also have occurred at second estimate of the age of the lowest artefact: Pedra Pintada; see Section 4.1.2). The base of the this was determined to be 49,9004700 years adjoining sand apron was TL dated to (Roberts et al., 1990b), the data being consistent 0120,000 years, similar to the date of initiation of with a linear rate of sand accumulation (as had the Malakunanja II apron, so future archaeological been noted previously for unoccupied sand aprons excavations should perhaps focus on the sedimen- in the region; Roberts et al., 1991). Independent tary sequence immediately outside the support for the TL dates in the upper 2 m of Malangangerr shelter (Roberts and Jones, 1994). deposit is given by their comparability with 14C To test our theory that people ®rst arrived in this ages: one TL date of 15,0002000 years was region between 50,000 and 60,000 years ago, we bracketed by calibrated 14C ages of 14,700±17,000 engaged luminescence dating methods at a second and 17,500±18,200 years (2s range, following deep archaeological sequence at the foot of the Stuiver and Reimer, 1993), while a second TL date Arnhem Land escarpment: the Nauwalabila I of 24,0004000 years matched, within the estimate (Lindner Site) rock shelter in Deaf Adder Gorge, of uncertainty, a calibrated 14C age of 20,600± 70 km south of Malakunanja II (Roberts et al., 22,300 years (Roberts et al., 1990b; Roberts and 1994a; Roberts and Jones, 1994). A series of 14C de- Jones, 1994). On the basis of these results, we pro- terminations had already been made at this site posed that this part of northern Australia had been (Jones and Johnson, 1985) and additional 14C dates colonised by 50,000 years ago. were obtained to compare against luminescence The culturally-sterile layers at Malakunanja II dates for the upper 2 m of the 3 m-deep sand were dated to between 0110,000 years and deposit. Sand-sized quartz grains were extracted to 065,000 years, which suggested to us (Roberts et determine the TL and OSL palaeodoses using the al., 1990a) that people did not arrive in this region multiple-aliquot additive-dose protocol (the OSL during the penultimate glacial maximum, when aliquots being held at 2208C for 300 s before lower sea-level might have facilitated open-water 514.5 nm illumination). As at Malakunanja II, the crossings from the north-west and Australia could environmental dose rate was estimated from high- have been reached by land bridge from Irian Jaya/ resolution gamma spectrometry, which again Papua New Guinea. But could we have missed evi- showed no evidence of radionuclide disequilibrium dence of older occupations in western Arnhem in the 238U or 232Th decay chains. On-site dosimetry Land? It is unlikely that the ®rst colonists from measurements were not taken, but laboratory ana- Indo-Malaysia used tools made exclusively from lyses of the bedrock and compacted rubble in the perishable materials, such as wood and bamboo, as basal quarter of the deposit were made to recon- stone artefacts have been found at early Pleistocene struct the ``in®nite matrix'' gamma dose rate experi- sites in south-east Asia and at middle Pleistocene enced by the two lowermost samples. These sites on Flores (see Section 3.3). Early habitation analyses indicated that, despite the heterogenous sites may exist on the now-submerged Sahul Shelf, texture of the deposit, the rubble and bedrock had and new excavations at Malakunanja II may reveal similar radionuclide activities to the neighbouring a large rock shelter hidden behind the 4.6 m-deep sand samples and, therefore, the calculated dose sand apron. If such a shelter exists, then an older rates should be accurate (Roberts et al., 1994a). occupation may be recorded only in its deepest The presence of decomposed rubble posed a recesses, with subsequent in®lling of the shelter hav- much greater threat to the accurate determination ing forced the later human occupants onto the sand of the palaeodose in the two lowest samples. Sand apron, where our excavation was placed. As occu- grains liberated by the disintegration of buried rub- pation debris is commonly found near the entrances ble will not have been exposed to sunlight since the LUMINESCENCE DATING IN ARCHAEOLOGY 857 time of rock formation, resulting in a luminescence employed for TL and OSL palaeodose determi- signal that is either dose-saturated or, if the ambi- nations, each aliquot holding 04000 quartz grains. ent temperature of burial is suciently high (e.g. Examination of the dose-response curves showed no Prokein and Wagner, 1994), in a less-than-saturated evidence of increasing relative scatter with added condition of dynamic equilibrium. Even at a burial dose, implying that low-level contamination was temperature of 308C, the 3258C and 3758C TL not a problem. The possibility of high-level con- peaks in quartz have sucient lifetimes (>4 million tamination was checked indirectly, by dating the years, calculated from the values listed by Aitken, near-basal sediments of the sand apron immediately 1985, p. 272) for late Quaternary sediments to outside the rock shelter. These sediments have been remain in dose-saturation and, hence, to appear deposited directly onto bedrock and are not subject essentially unbleached. Mixing such grains (e.g. to rubble contamination. The OSL date of through bioturbation) with a population of recently 58,3005800 years for the inception of the sand and completely bleached grains will produce grossly apron coincides with the OSL date of in¯ated TL and optical dates if multiple-grain, mul- 60,3006700 years for the start of sediment ac- tiple-aliquot methods are employed. cumulation in the rock shelter, indicating that the A low level of rubble contamination (e.g. less latter date is not grossly distorted by severe rubble than 10%) will likely result in aliquots containing contamination (Roberts et al., 1994a, p. 582); a unequal numbers of contaminant grains: multiple- direct test of this proposition is planned using grain OSL aliquots that have been short-shine nor- single-grain dating methods (Murray and Roberts, malised should produce multiple-aliquot growth 1997). The indurated nature of the Mesoproterozoic curves that exhibit increasing relative scatter in sandstone bedrock might explain the absence of OSL intensity with added dose (Rhodes, 1990), rubble and saprolite contamination at Nauwalabila while a suite of single aliquots (each composed of I. Similar good fortune is expected at Malangangerr many grains) should show a correlation between the and at Malakunanja II, where the lowest artefacts measured palaeodoses and the corresponding natu- are, moreover, safely positioned 2 m above the ral OSL intensities (Li, 1994). Single-grain analyses nearest rubble or bedrock. would permit the contaminant grains to identi®ed The 060,300 year-old sample underlies the basal directly, and are the only solution to the problem occupation level at Nauwalabila I, while the over- of high-level (e.g. more than 10%) contamination. lying sample gave an OSL date of 53,4005400 The latter would result in contaminant grains being years. Furthermore, the lowest artefacts at present on each multiple-grain aliquot that con- Nauwalabila I are tightly bound within the sand sisted of more than a few tens of grains; and a and interlocked rubble unit, so we can discount the Gaussian distribution is rapidly generated as the possibility of their post-depositional movement, level of contamination increases. Severe contami- unlike at Malakunanja II. This age bracket for the nation might be expected close to decayed rubble zone of initial occupation almost exactly matches but would not be detected using the approach of Rhodes (1990) or Li (1994) if each aliquot was uni- that obtained at Malakunaja II, reinforcing our ear- formly contaminated. Dating of single grains or lier conclusion that ®rst settlement of western single-aliquots composed of few grains (e.g. 10 or Arnhem Land occurred between 50,000 and less) should be used in such instances; otherwise the 60,000 years ago. only means of recognising contamination by rubble At the three archaeological sites investigated in or saprolite (in situ weathered bedrock) is from a the region, rubble contamination appears to be comparison of luminescence and independent age insigni®cant, and three main lines of evidence argue estimates. against insucient bleaching in antiquity. First, the Although these concerns were raised by Roberts near-surface samples yielded small TL and OSL and Jones (1994), single-aliquot and single-grain palaeodoses: 1±2 Gy and 00.16 Gy, respectively, the methods of palaeodose determination were not latter corresponding to an age of only 29060 available in 1991, when the Nauwalabila I optical years. Second, the palaeodose plateaux of the red-lit dating study was begun. Sediment samples were col- TL samples (which includes the pair at Nauwalabila lected from several levels, including one above and I) incorporate both the 3258C and 3758C peaks, one below the rubble and sand unit that contained which implies a suciently prolonged solar exposure the lowest assemblage of artefacts. The latter is to have reset the hard-to-bleach signal as well as the similar in composition to that found at easy-to-bleach signal. Third, there is good agree- Malakunanja II, and included ¯akes of quartz, ment between luminescence dates and calibrated 14C quartzite and chert, a grindstone and scraper, and dates for charcoal from the upper occupation levels pieces of dolerite and abraded haematite. The rub- at Malakunanja II and Malangangerr (see above); ble unit was not directly sampled for luminescence the same is true also at Nauwalabila I where TL dating, but the possibility remained that the two and OSL dates of circa 13,500±14,600 and 28,000± adjacent samples may be contaminated. 30,000 years compare favourably with calibrated Conventional multiple-aliquot methods were 14C ages of 10,500±13,500 (2s range) and circa 858 R. G. ROBERTS

24,000±27,000 years, respectively (Roberts et al., penultimate glacial maximum requires that the 1994a; Roberts and Jones, 1994). palaeodoses be greatly underestimated and/or that A third possible mechanism to obtain age overes- the dose rates be signi®cantly overestimated. The timates is thermal transfer of charge from light- TL and OSL palaeodoses are unlikely to suer insensitive traps to light-sensitive traps during the from anomalous fading, as none was detected in OSL preheat treatment. Thermal transfer eects, samples that were dosed and then stored at room and the means to overcome them, have been dis- temperature for periods of up to 400 days (Roberts cussed recently for potassium feldspars (Huntley et al., 1990a, 1994a). The high-resolution gamma and Clague, 1996) and quartz (Rhodes and Bailey, spectrometry facilities are calibrated against inter- 1997). Berger (1995, p. 95) suggests that thermal national standards, rendering a factor-of-two error transfer may account for the non-zero OSL age of improbable, and the thickness of the rock shelter the near-surface sample at Nauwalabila I, and he roof was factored into the estimate of the cosmic- also comments that the rise in OSL palaeodose with ray dose rate (which would otherwise have been increasing illumination time (for the 030,000 year- overestimated). The luminescence dates will increase old sample) may indicate the presence of poorly or by 00.8% for each 1% rise in sample water con- non-bleached grains. We did not anticipate a zero tent, so the circa 60,000 year-old dates at age for the near-surface sample which was buried to Nauwalabila I and Malakunanja II would increase a depth of 1±6 cm and, therefore, had been exposed by only 7000 and 10,000 years, respectively, if the for some time to the surrounding environmental samples had been permanently saturated since depo- radiation ®eld and cosmic rays (Roberts et al., sition; this proposition is hardly credible for a habi- 1993b, 1994a). The OSL age for this sample instead tation site on highly porous and permeable sands accords well with the age expected from the rate of which, if saturated, would turn into a slurry. sediment accumulation deduced from the Holocene The western Arnhem Land sites have been dis- 14C chronology and the depth at which glass shards cussed in detail because they represent still, in my appear in the deposit (Jones and Johnson, 1985). view, the oldest reliable evidence for human occu- Also, as just discussed, contamination of the older pation of the continent. Malakunanja II and samples would appear to be minimal, and the oper- Nauwalabila I also contain the earliest evidence for ation of non-®rst-order kinetics during illumination the use of pigments (presumably for rock art or (i.e. complex interactions arising from optical trans- body decoration), in the form of red and yellow fer of charge) is a more likely cause of rising ochres, and haematite ``crayons'' with ground facets palaeodose (see Roberts et al., 1994a). and striations, excavated from the 50,000± The magnitude of thermal transfer of charge due 60,000 year-old levels (Morell, 1995; Roberts et al., to a preheat of 2208C for 300 s was examined expli- 1990a, 1994a). Recent claims have been made for citly by Roberts et al. (1993b, pp. 51±52). earlier landfall and rock art in the Kimberley region Negligible transfer in the 030,000 year-old sample of northern Australia (Fullagar et al., 1996), but was inferred from two experiments. First, preheat- the TL dates for Jinmium rock shelter which sup- ing of the ``natural'' aliquots caused a decrease in port these claims are of questionable validity (see OSL intensity (consistent with thermal erosion of below). Before scrutinising the Jinmium data, how- part of the OSL signal) and not a transfer-induced ever, it is pertinent to mention the optical dates increase, such as that reported by Godfrey-Smith et obtained from archaeosediments at two painted al. (1988). Second, natural aliquots that had been rock shelters in Cape York: Ngarrabullgan Cave bleached for 20 h and then preheated gave rise to (David, 1993; David et al., 1997) and Mushroom an OSL signal three orders-of-magnitude smaller Rock (Wright, 1971b; Morwood and Hobbs, 1995; than that produced by natural aliquots which had Morwood et al., 1995b). been bleached, given a dose of 11.2 Gy and then preheated. The resulting thermal transfer in the 4.2.6. Cape York. The basal strata of the 40 cm- bleached-then-preheated aliquots is thus equivalent thick deposit in Ngarrabullgan Cave have been to 00.01 Gy, which is a trivial correction to the intensively dated by AMS and beta-counting 14C measured palaeodose for this sample (024 Gy) and methods, with 20 dates (18 from individual pieces is an order-of-magnitude less than the palaeodose of charcoal) currently available for the reported from the near-surface sample. 032,500 years BP unit; an underlying stratum has Furthermore, this bleached-then-preheated value is yielded a 14C age of >37,200 years BP, although directly comparable to the ®rst point on the dosed- this may represent an outlier of the 032,500 years bleached-preheated growth curve which may be BP suite of ®nite age determinations, which spans used as a non-zero OSL baseline (Huntley et al., the period circa 28,800±36,100 years BP (David et 1993b; Huntley and Clague, 1996), an approach al., 1997). Sand-sized quartz grains were extracted strongly advocated by Berger (1995, p. 95). from the basal strata but, as the sample was posi- If the western Arnhem Land dates are not too tioned only 5 cm above the bedrock ¯oor, concerns old, might they be too young? To make the dates were raised about the possibility of contamination consonant with initial colonisation during the by saprolite. Consequently, additive-dose OSL LUMINESCENCE DATING IN ARCHAEOLOGY 859 palaeodoses were determined using not only con- experimental conditions to those used at ventional multiple-aliquot procedures (short-shine Ngarrabullgan Cave), and the environmental dose normalisation, preheat of 2208C for 300 s, signal in- rates were also determined using the same methods, tegration over the full period of stimulation (minus and with the same outcome, as at Ngarrabullgan. background) by a halogen lamp ®ltered to An optical date of 090,000 years was indicated for 50040 nm) but also the newly-devised single-ali- the level of the lowest artefact. This remarkable quot protocol (Murray et al., 1997); the latter ®nding was soon thrown into doubt, however, by involved a ``preheat plateau'' test (Aitken, 1994) an apparent age of 060,000 years for a sample only using 24 multiple-grain aliquots, and a series of 15 cm higher up the pro®le. As no stratigraphic brief stimulations by 420±550 nm ``green'' light. The unconformity was visible, the dierence in age environmental dose rate was estimated from in situ between these two closely-spaced samples seemed at and high-resolution gamma spectrometry, the two odds with the rapid rate of sedimentation (1 m per measures being accordant and secular equilibrium 4000 years) implied by the 14C chronology for the being observed. The average palaeodose of upper levels. The lowermost sample had been col- 482 Gy for the six aliquots in the preheat plateau lected from only 50 cm above the top of a strongly region (280±3008C for 10 s) showed agreement with mottled zone, which is either saprolite or, from the multiple-aliquot palaeodose of 474 Gy, indi- its micromorphology (J. Magee, Personal cating that low-level contamination of the sample Communication, 1994), an ancient colluvial or allu- by decomposed rubble or saprolite is not prevalent vial deposit. Mixture of the latter with recently (David et al., 1997). Single-aliquot and single-grain bleached sediments was considered as the probable analyses of this sample using a novel regenerative- cause of the problem, the samples nearest the dose protocol (Murray and Roberts, 1997) have mottled zone being contaminated to a greater con®rmed these ®ndings (J. Olley, Personal Com- degree than those further away. munication, 1997). High-level contamination can At that stage, true single-aliquot methods had also be discounted, as the corresponding weighted not been devised for quartz, but an attempt was mean OSL age (34,7002000 years) is a few mil- made to examine the palaeodose distribution for in- lennia older than the average 14C age of dividual grains using a simple regenerative-dose 032,500 years BP, as expected for uncalibrated 14C technique (see Murray et al., 1995; Olley et al., determinations of this antiquity (see Section 2.2). 1997b). The ``natural'' OSL intensities of (pre- Although Ngarrabullgan Cave is the oldest heated) single grains were measured, beta doses human occupation site so far recorded in Cape were then administered, and the total OSL signals York, Mushroom Rock boasts one of the deepest were again measured (after a second preheat); the archaeological deposits in Australia, containing ratio of the ®rst and second light sums multiplied stone artefacts to depths of 4.4 m (Morwood et al., by the given dose provides a ®rst-order approxi- 1995b). Optical dating of the Mushroom Rock sedi- mation of the palaeodose (the non-linear dose-re- ments has, unlike Ngarrabullgan Cave, revealed sponse of quartz, and possible sensitivity changes, severe problems of contamination of the lowest preventing an accurate determination). Single grains artefact-bearing levels. Radiocarbon determinations from the ``60,000 year-old'' sample proved, unfortu- had been made previously on charcoal preserved in nately, to be too dim, so the process was repeated the upper 2 m of deposit, and TL dates had been obtained from the quartz sediments (using a combined additive/regenerative approach and thick- source alpha counting; Morwood et al., 1995b). The ®rst warning signs were given by the discord between the 14C and TL dates for the habitation deposits: calibrated 14C dates of 07700 and 02350 years were poorly matched by TL dates of 011,400 and 06800 years, respectively. In the underlying carbon-sterile deposits, a TL date of 37,4003900 years was obtained from a depth of 4 m (i.e. above the level of the lowest artefacts). Owing to the uncertainty surrounding the validity of these TL dates (Morwood et al., 1995b), an optical dating study was initiated by the author, R. Fig. 10. Frequency distribution of apparent OSL ages (in Jones and M. Morwood in 1994 to con®rm or thousands of years) obtained using single aliquots of refute the apparent great antiquity of the archaeolo- quartz from Mushroom Rock, Cape York. Each aliquot gical sequence. holds 10 grains. The inset ®gure shows the sum of OSL counts (minus the background count rate) for each aliquot The OSL palaeodoses were obtained, in the ®rst plotted against its apparent age. Note the increase of OSL instance, from multiple-aliquot additive-dose pro- intensity with age, indicating the presence of poorly cedures applied to coarse-grain quartz (identical bleached grains on at least some of the aliquots. 860 R. G. ROBERTS on single aliquots, each composed of 10 grains. estimates, as initially suspected from the mis- This number was chosen to increase the OSL matched 14C/TL dates for the overlying deposits. A intensity, yet maintain a low probability of con- date of 30,000±40,000 years for ®rst occupation of tamination. The apparent palaeodoses and corre- this rock shelter would be consistent with other evi- sponding ages (calculated using the mean dose rate) dence in Cape York, such as the 14C date of for nine aliquots showed a wide distribution: six ali- 032,000 years BP for the lowest artefacts at the quots clustered between circa 21,000±38,000 years, nearby site of Sandy Creek 1 (Morwood and while the remaining three aliquots produced appar- Hobbs, 1995; Morwood et al., 1995a), and the 14C ent ages of 66,000, 72,000 and 108,000 years and OSL ages for Ngarrabullgan Cave (David et (Fig. 10). Furthermore, the oldest ages are associ- al., 1997). ated with the brightest aliquots (Fig. 10, inset), as 4.2.7. Jinmium. With these ®ndings in mind, it is predicted for aliquots contaminated by poorly appropriate to review the TL evidence for human bleached grains (Li, 1994). For this sample, the occupation of Jinmium rock shelter (Fullagar et al., youngest group of apparent ages (mean 1996). Claims for occupation before 030,000 years) may represent an uncontaminated 116,00012,000 years, and possibly as early as population (although this remains to be shown), 176,00016,000 years, made media headlines whereas the older aliquots contain one or more con- around the world, as did the dates of circa 58,000± taminant grains. 75,000 years for buried circular engravings (pecked Optical dating of the lowest artefact-bearing cupules) on the rock shelter wall and on a fallen levels at Mushroom Rock is in progress, using sandstone fragment (Bahn, 1996; Holden, 1996). single-aliquot additive-dose (Murray et al., 1997) When R. Fullagar ®rst told me of these extraordi- and regenerative-dose (Murray and Roberts, 1997) nary results in 1994, I recounted the problems protocols for quartz. Multiple-aliquot TL and OSL encountered at Mushroom Rock and suggested that methods will clearly generate considerable age over- optical dating of the Jinmium archaeosediments be

Fig. 11. Stratigraphic section of the north face of Trench C1, squares II and III, at Jinmium rock shelter. Locations of TL samples (including W1647) are shown with their published ages, and locations of OSL and elemental carbon 14C samples are indicated by their COOR ®eld codes. LUMINESCENCE DATING IN ARCHAEOLOGY 861 carried out to check the TL dates; the latter had Prescott and Purvinskis, 1991, 1993; Spooner et al., been obtained from quartz sand grains using a com- 1988). When the ``plateau test'' is performed, the bined ``total bleach'' additive/regenerative multiple- Trench C1 samples have foreshortened plateaux aliquot procedure (each aliquot holding 03000 that begin at 3758C, whereas the samples collected grains, heated at 5 K s1) and thick-source alpha from the sand apron outside the rock shelter (e.g. counting. Fresh sediment samples were later col- W1752) exhibit plateaux extending over the tem- lected by the author, R. Jones and R. Fullagar, and perature region 300±5008C. Sample W1752 has a in situ measurements of the gamma dose rate were prominent natural 3258C signal [Fig. 12(b)], but as made. Dating of the Jinmium deposit using state- the plateaux for this and other sand apron samples of-the-art OSL and AMS 14C methods is underway do not extend to 2708C, they cannot include all of (Gibbons, 1997c), the latter exploiting the pro- the 3258C TL peak and, thus, should not be used cedures developed by Bird and GroÈ cke (1997) to to infer adequate bleaching in antiquity. If the extract elemental carbon from the traces of soot Jinmium sand apron is constructed in the same preserved in marine cores. In the absence of these manner as those in western Arnhem Land, however, alternative chronologies, what do the data pub- then the constituent sand grains are apt to be well lished by Fullagar et al. (1996), my ®eld obser- bleached at deposition; perhaps the use of yellow vations at Jinmium, and experience gleaned from laboratory illumination, which caused a ``double luminescence studies of other Australian quartz plateau'' in the Arnhem Land samples, is partly to sediments reveal about the probable accuracy of the blame for the plateaux starting at only 3008C. The TL ages? conspicuous failure of the Trench C1 samples to Several lines of evidence suggest that the TL pass the ``plateau test'' is more alarming. By start- dates published for the key samples at Jinmium (i.e. ing at 3758C, these plateaux do not even include all those from Trench C1, squares I to III; see Fig. 11) of the 3758C TL peak and, thus, strongly indicate are substantial overestimates of the true ages of this hard-to-bleach signal was not completely reset sediment deposition. From an examination of their in antiquity. TL data, the Trench C1 samples do not present an Fullagar et al. (1996, p. 768) explain the fore- internally consistent picture of having been well shortened plateaux in terms of ``partial re-exposure bleached in antiquity. The natural TL glow curve following deposition rather than insucient ex- shown for sample W1647 [Fig. 12(a)] shows no evi- posure prior to deposition'' (their emphases). This dence of a TL signal at 3258C, despite having sup- explanation is inadequate, however, as a brief re-ex- posedly been buried for 50,000 years. The glow posure would zero only the easy-to-bleach (3258C) curve is dominated instead by the hard-to-bleach signal and, upon reburial, a substantial hard-to- peak at 3758C and is comparable to the TL signal bleach (3758C) signal would be retained. In such that typically remains after selective bleaching (e.g. instances, the elapsed time since last exposure to

Fig. 12. (a) TL age ``plateau test'' for Jinmium sample W1647, with the natural TL glow curve (solid line) shown for a heating rate of 5 K s1. This sample was collected from the archaeological excavation (see Figure 11) and exhibits a foreshortened plateau. (b) Plateau test and natural TL glow curve (solid line) for sample W1752. This sample was collected from the sand apron (at 4 m depth) outside the rock shelter, and has a longer plateau . TL ages are in thousands of years (after Fullagar et al., 1996). 862 R. G. ROBERTS sunlight can be accurately determined only from the why the 3758C peak is so poorly bleached in the easy-to-bleach signal. In addition, the re-exposure rock shelter deposits, yet is adequately bleached on argument is dicult to reconcile with the observed the adjacent sand aprons. The aprons, as in western increase of TL age with depth. For the dates to Arnhem Land (Roberts et al., 1991), are composed remain in correct stratigraphic order requires that of sand grains eroded from the local sandstone bed- quartz grains be raised to the ground surface, rock outcrops and transported by slopewash exposed to sunlight for a brief period (so as not to (Fullagar et al., 1996, p. 766). This process usually reset the 3758C signal), and then returned to the results in the 3258C and 3758C peaks both being same level from which they originated. Such a pro- reset before the sediments are ®nally buried cess of reworking seems unlikely. If grains from (Roberts et al., 1990a, 1991, 1994a). In both deep in the section had been redeposited at a higher regions, modern grains are deposited without sur- level, and near-surface grains had been redeposited face stainings or oxide coatings, the latter being further down the pro®le, then the entire deposit acquired post-depositionally on the sand aprons; would be characterised by a single, average age. the Jinmium rock shelter samples are, moreover, A ®rst-order approximation of the time since only lightly stained. An alternative explanation for sample W1647 was last exposed to sunlight can be the TL age overestimates involves a combination of estimated from the size of the easy-to-bleach TL two factors: contamination of the Trench C1 signal [Fig. 12(a)]. The natural glow curve for this samples by decomposed rubble and inadvertent sample is composed mainly of the residual 3758C bleaching of all samples during laboratory pretreat- peak but, as this sample has been buried for at least ment. This is my preferred interpretation of the 1800 years (as indicated by the 14C date on charcoal Jinmium data, as it better ®ts the geomorphological from an overlying level), some fraction of the TL evidence and additionally accounts for the slightly signal at 3258C must represent the ``pure'' 3258C foreshortened plateaux of the Jinmium sand apron peak that has been regenerated since the last samples and other quartz samples processed by the bleaching event. Based on the data presented in same laboratory. Spooner et al. (1988) for selectively bleached quartz, The Jinmium rock shelter is composed of Upper and the probable contribution of the 3758C peak to Carboniferous sandstone and conglomerate, which the natural TL at 3258C for sand apron sample weathers readily and produces rubble that is easily W1752 [Fig. 12(b)], we might expect that 10±50% cut with a hand trowel. A similar situation pre- of the TL at 3258C is due to regrowth of the vails at Mushroom Rock, whereas the Arnhem ``pure'' 3258C peak. As Fig. 12(a) shows the TL age Land sites are characterised by indurated at 3258C to be 021,000 years, then this fractional Mesoproterozoic sandstone bedrock. Disintegration adjustment implies that only 2100±10,500 years of buried rubble at Jinmium will liberate dose-satu- have elapsed since the easy-to-bleach signal was last rated (i.e. eectively ``unbleached'') grains, which reset by sunlight. A similar argument was presented can then become mixed into the sediment column in February 1997 by N. Spooner at the Sixth through bioturbation. The most contaminated sedi- Australasian Archaeometry Conference in Sydney, ment samples are apt to be those collected closest the details of which are being prepared for publi- to the rubble. As the amount of rubble increases cation with J. Prescott (Personal Communication, with depth at Jinmium, the ages of the basal 1997; Gibbons, 1997c). samples are likely to be the most distorted and These ages are much younger than the those higher up the pro®le should be less severely 50,000 years claimed for this sample by Fullagar et contaminated, giving rise to a series of ages that are al. (1996), but they are consistent with the rate of in correct stratigraphic order but are all biased to sand accumulation outside the rock shelter and some extent. The adjacent sand aprons are not con- with the lack of ®eld evidence (such as a strati- taminated by rubble, so their TL ages should be graphic unconformity) for a hiatus of 040,000 years more accurate. at 050 cm depth; the latter was inferred by Fullagar The degree of contamination of the Trench C1 et al. (1996) on the basis of a 2300 year-old date for samples cannot be evaluated from the published a TL sample positioned 020 cm above the supposed data, but one feature is clearÐthe saturated 3258C 50,000 year-old level. The foreshortened TL age pla- signal associated with any contaminant grains is lar- teaux are not illustrated for the deeper samples, so gely absent in the Trench C1 samples, yet a 3258C the above fractional adjustment cannot be applied peak is present in the sand apron samples. In both directly to recalculate their TL ages from the cases, multiple-grain aliquots and yellow laboratory ``pure'' 3258C signal. Nonetheless, assuming the light ®lters were used. The latter may provide a sol- same proportional error applies throughout the pro- ution to this puzzle. If the well bleached quartz ®le, the 0116,000 and 0176,000 year-old samples grains extracted from the sand apron behave in a may be as young as 5000±24,000 and 7000± similar fashion to their Arnhem Land counterparts, 37,000 years, respectively. then their preparation under yellow illumination This explanation for the TL age overestimates at could reduce the natural 3258C signal by 050% [see Jinmium fails, however, to account satisfactorily for Fig. 9(a)]; some 3258C signal will remain, but a LUMINESCENCE DATING IN ARCHAEOLOGY 863 large proportion will be inadvertently bleached in necessitates two compounding in¯uences, neither of the laboratory. Variations in the magnitude of this which are yet proven for Jinmium. Price (1994) eect are expected between samples, owing to dier- reported that several quartz samples yielded the ences in bleaching behaviour and duration of illumi- same palaeodose from the TL signals at 3258C and nation (e.g. Spooner and Prescott, 1986; Roberts et 3758C, indicating that his yellow laboratory lights al., 1991). The foreshortened plateau (starting at are safe. This evidence is insucient, however, as 03258C) of sand apron sample W1752 is consistent all sample preparations were conducted in yellow with this proposition, the lack of a ``double pla- light, so the ``pure'' 3258C peak may have been teau'' perhaps indicating a greater degree of inad- bleached to a greater or lesser extent in every vertent bleaching. sample before the ®rst TL measurements were The glow curve of the rock shelter sample W1647 made; the obtainment of similar palaeodoses at [Fig. 12(a)] can be likewise explained. If this sample 3258C and 3758C could be argued to show only consisted entirely of contaminant grains, then its that the low-temperature ``tail'' and the peak of the TL signal should be fully saturated with a dose of hard-to-bleach signal contain the same palaeodose. 0250 Gy in both the 3258C and 3758C peaks. The Fullagar et al. (1996, p. 758) claim support for the latter will not be aected by yellow illumination, so TL dating procedures used at Jinmium from the its palaeodose of only 50 Gy indicates that 20%, or studies of Nanson et al. (1991) and Shepherd and less, of the grains on each aliquot of this sample are Price (1990). The plateaux published for these contaminants. Consequently, if the 3258C signal samples begin at 3508C and 2758C, respectively: the was inadvertently bleached in sample W1647 by the latter is an acceptable plateau, the former is not. same absolute amount as in sample W1752 (whose Such sample-to-sample dierences prevent me from palaeodose is almost double that of W1647), then making a de®nitive judgement on the issue of lab- the ``pure'' 3258C peak would be almost completely oratory illumination, but a convincing test would erased, as is observed [Fig. 12(a)]. Older samples in involve a comparison of samples prepared under the Jinmium sequence may have some natural red and yellow light (e.g. Fig. 9). 3258C peak remaining, but these samples, together Optical dating of the Jinmium sediments, using with all contaminated Trench C1 samples, will fail fresh samples prepared under red light, should show the total bleach ``plateau test'' because the ``pure'' which (if either or a combination) of the above two 3258C signal has been completely or partially reset interpretations of the TL data is correct. If the ®rst and the 3758C signal is derived from a mix of interpretation is correct (i.e. solar resetting of the bleached (i.e. sand apron) and unbleached (i.e. rub- 3258C peak but not the 3758C peak), then multiple- ble) grains. The true age of sample W1647 depends aliquot OSL methods should yield the true palaeo- critically on the degree of rubble contamination. A dose. On the other hand, these procedures will yield late Holocene age is calculated assuming each mul- palaeodose overestimates if the samples are contami- tiple-grain aliquot is composed of 20% rubble nated with rubble (the pretreated grains not being grains, whereas an age of 40,000 years is derived for inadvertently bleached by red light); the dose 5% contamination. acquired by the non-contaminant grains since last The foregoing two interpretations of the Jinmium exposure to sunlight can, in such circumstances, data have been proposed to explain the predomi- only be revealed by single-aliquot or single-grain nance of the hard-to-bleach TL signal and the fore- protocols. These predictive tests are currently being shortened palaeodose plateaux in the Trench C1 performed on the Jinmium samples. samples. Both interpretations imply that the published TL ages are too old, a conclusion supported 4.2.8. The hand of man? TL dating has also fea- by three independent forms of geomorphological tured in previous proposals for human occupation evidence: the lack of a stratigraphic unconformity of Australia prior to the last interglacial. At Point between the closely-spaced 2300 and 50,000 year-old Ritchie, on the coast of Victoria, an unusual shell levels; the absence of post-depositional red staining deposit (showing traits of an Aboriginal midden) of the Trench C1 sand grains compared to that and discoloured cobbles (suspected of being oven developed in the supposedly coeval sand aprons; stones) were investigated using a variety of numeri- and the apparent sedimentation rate inside the rock cal dating methods (Prescott and Sherwood, 1988; shelter being an order-of-magnitude less than the Sherwood et al., 1994). The shell bed (composed rates for the adjacent sand apron and square C1/ chie¯y of Turbo undulatus) had yielded ®nite 14C IV, excavated only 10 m outside the shelter. dates of circa 25,000±42,000 years BP from char- The degree of age overestimation is contingent on coal, calcrete, rhizomorphs, and calcite and arago- the mechanism invoked to explain the inadequate nite from T. undulatus shells. A total bleach TL ``bleaching'', be it through insucient sunlight ex- date of 080,000 years (later revised to posure of all the grains at deposition or through 67,00010,000 years) was obtained from unheated rubble contamination in tandem with yellow labora- quartz sediments extracted from the poorly cemen- tory illumination. The former runs counter to geo- ted shell-bearing calcarenite, while quartz grains morphological expectations, but the latter separated from the presumed hearth stones (found 864 R. G. ROBERTS in the same stratigraphic layer) gave ages of circa 4.2.9. Kimberley to Koonalda. At the present time, 155,000±170,000 years, assuming the TL signals therefore, claims for occupation of Australia prior were zeroed by heating. As these ages are much to 60,000 years are unsupported by compelling greater than that obtained for the sediment, it was archaeological or chronological proof. The oldest concluded that the discoloured cobbles were not reliable dates are those of 50,000±60,000 years from burnt suciently to reset the TL ``clock'' and, western Arnhem Land, although two rock shelter therefore, cannot be classed as unequivocal artefacts sites in the Kimberley region of Western Australia 14 (Sherwood et al., 1994). If sunlight is the resetting hold promise of great antiquity. Uncalibrated C mechanism, then the non-zero ``residual'' TL level dates of up to 39,7001000 and 28,060600 years reduces the dates to circa 115,000±125,000 years, in- BP have been obtained from the non-basal cultural dicative of a last interglacial age for the sediment levels at Carpenter's Gap 1 and Widgingarri 1, re- from which the cobbles formed. As a similar or spectively (O'Connor, 1995; Veth, 1995), and OSL dating of quartz sediments from these and the younger age for the T. undulatus shells was inferred underlying deposits is in progress. South of the Tro- from amino-acid racemization and ESR analyses pic of Capricorn, no archaeological site has yet (Goede, 1989; Sherwood et al., 1994), this supposed been dated to more than 45,000 years by 14C or ``midden'' may have been formed naturally during luminescence methods (bearing in mind that uncali- high sea-levels of the last interglacial. brated 14C ages of 038,000 years BP correspond to Other claims for human landfall in Australia at calendar-year ages of 40,000±45,000 years; see Sec- around the time of the penultimate glacial maxi- tion 2.2). As with the sites in northern Australia, mum are based on marked changes in the types of the author is involved in a program of luminescence vegetation and the increased abundance of charcoal dating of southern sites that have yielded 14C dates in pollen cores (Singh and Geissler, 1985; Kershaw, in excess of 30,000 years BP. Con®rmation or refu- 1994). These changes have been attributed to the tation of the existing 14C chronology by lumines- deliberate ignition of bush®res by humans, although cence methods will provide a test of the various no associated archaeological evidence has yet been theories proposed for the rate of continental coloni- found. In addition, numerical dating methods have sation and the routes taken by these ®rst arrivalsÐ not been used to date the crucial horizons in either whether they clung to the coastline and inland the Lake George core (from south-eastern New waterways (e.g. Bowdler, 1977), spread rapidly South Wales; Singh and Geissler, 1985) or the through all major ecological zones (e.g. Jones, Ocean Drilling Program Site 820 marine core (from 1989), or took further tens of millennia to settle the the edge of the continental shelf o north-east southern fringe and arid interior of Australia. Queensland; Kershaw et al., 1993). These cores Devil's Lair is one of the oldest archaeological have been correlated with the oxygen isotope deep- sites in south-west Western Australia, having yielded sea record, although sedimentation rates are vari- 14C ages in excess of 30,000 years BP for ®rst occu- able at both locations (Singh et al., 1981; pation (Dortch, 1979; Dortch and Dortch, 1996). Peerdeman et al., 1993). A linear extrapolation of Optical dating of these cave deposits is being the 14C age-versus-depth curve derived from the attempted using single-aliquot, as well as multiple- upper part of the Lake George core indicates an aliquot, methods to circumvent problems of incom- age of 060,000 years at the depth of the supposed plete bleaching of the sediments washed into the penultimate glacial maximum (Wright, 1986b). A cave. The short distance (015 m) from cave entrance TL study of this 18 m core was undertaken by to deposition site should not thwart dating at Devil's Lair, but the much longer (0150 m) entrance Mortlock and Price (1984). The natural TL of slope into Koonalda Cave, on the karst Nullarbor quartz sand grains, collected at depth intervals Plain, was held responsible for OSL age overesti- down the core, showed a steady increase in the mates at the ``Gallus site'' (Roberts et al., 1996). 3758C signal intensity to a depth of 04 m, followed Sediment samples collected from the latter exca- by a 4 m section of fairly constant TL, and then a vation (Wright, 1971a) produced multiple-aliquot renewed increase in signal intensity down the lowest quartz OSL dates of 60,000±70,000 years, compared 10 m of core. TL ages were not calculated, but the with calibrated 14C ages of circa 21,000± uniform TL signal for the middle section of core 28,000 years for multiple charcoal samples. The was tentatively attributed to a large and sudden OSL date of 09200 years for a ¯uvial sample depos- depositional event. As the major change in pollen ited recently on the cave ¯oor further demonstrated type and charcoal abundance occurs at a depth of the inadequacy of sediment bleaching in this deep 04 m, con®rmation of their ®ndings could resolve cave system. Aeolian sediments may enter the doline the chronological uncertainties of the Lake George in a fully bleached state but they acquire a signi®- core. Optical dating of single grains of quartz sand cant palaeodose en route to the cave ¯oor owing to from these lake and marine cores is planned to their intermittent storage along the entrance slope obtain numerical age estimates for the key hor- and their remobilisation by rivulets in the dimness izons. of the cave. Furthermore, the degree of age overesti- LUMINESCENCE DATING IN ARCHAEOLOGY 865 mation will vary from sample to sample, depending rived from the local limestone. High-resolution on the speci®c history of the transport route and gamma and alpha spectrometric analyses of the cumulative residence time. Palaeodose determi- ``pure'' carbonate and aeolian fractions have shown nations on individual grains from a single habitation that their annual beta doses dier suciently level may show signi®cant dierences between (00.14 and 02.6 mGy, respectively) to account for grains, but identi®cation of the ``true'' palaeodose the observed spread in palaeodose (Olley et al., requires that at least one of these grains was trans- 1997a). The magnitude of heterogeneity in the beta ported rapidly to the site of ®nal depositionÐa dose rate will vary from location to location, but is requirement that may not be satis®edÐand that all apt to plague all sedimentary deposits and, hence, other factors contributing to the observed spread in complicate the interpretation of single-grain palaeo- single-grain quartz palaeodoses (Murray and dose distributions. Nonetheless, a population of Roberts, 1997; Roberts et al., 1997; J. Olley, well bleached grains with variable microdosimetry Personal Communication, 1997) are held constant. might be expected to produce an approximately normal, uni-modal distribution, whereas the in- 4.2.10. Allen's Cave. Optical dating of archaeose- clusion of poorly bleached grains should give rise to diments proved more successful at Allen's Cave, a skewed or multi-modal distribution. located 070 km west of Koonalda Cave. This shal- In addition to the inhomogenous distribution of low rock shelter has been studied in detail for the radionuclides through the Allen's Cave deposit, sev- variety of radionuclide disequilibria observed in the eral types of disequilibria were measured in the 238U chain (Olley et al., 1997a; Roberts et al., 1996) 238U decay series (Olley et al., 1997a). Some of and for the application of multiple-aliquot (Roberts these disequilibria, such as the excess of 230Th over et al., 1996), single-aliquot (Murray et al., 1997) both its parent 238U (by up to 450%) and its and single-grain (Murray and Roberts, 1997) daughter 226Ra (by up to 190%), cause time-depen- methods of palaeodose determination. Charcoal dent variations in the dose rate during the period of samples from a well-preserved hearth had been sample burial. The evolution of the dose rate dated previously by 14C to circa 9300±9500 years through time was therefore modelled, and thorium BP (2s calibrated age range of circa 10,000± isotope ratios (230Th/232Th) were used to demon- 11,000 years), and the quartz sediments collected strate that the deposit is composed of sediments de- from 5 cm beneath this hearth yielded closely rived from two distinct sources. The near-surface matching OSL dates of 10,100600 (multiple-ali- sediments are derived from an unidenti®ed source quot, additive-dose), 9700600 (single-aliquot, and exhibit the greatest excesses of 230Th, as well as additive-dose), 10,300500 (single-aliquot, regen- 210Pb excesses from atmospheric fallout. The bulk erative-dose), 9400700 (single-grain, additive- of the deposit (from 10 cm depth to the base of the dose) and 10,300700 (single-grain, regenerative- excavation at 510 cm) has 230Th/232Th ratios com- dose) years. The same coarse-grain quartz gave an parable with those of the sediments on the sur- additive-dose TL date of 11,100900 years (pla- rounding Nullarbor Plain, and 210Pb/226Ra ratios teau region 265±3808C, which includes all of the consistent with an average of 0.740.03, implying dominant 3258C peak) and ESR dates of circa 025% radon escape irrespective of depth. 9800±13,300 years using three newly discovered The lowest artefact and hearth at Allen's Cave light-sensitive ESR signals, observed at g = 1.9870, have been dated to 39,8003,100 years, using mul- 1.9842 and 1.9162 at a temperature of 77 K tiple-aliquot additive-dose OSL procedures (Roberts (Yoshida, 1997). The excellent agreement found et al., 1996). An independent check on this date is between these 14C, OSL, TL and ESR age determi- forthcoming via AMS 14C dating of elemental car- nations is further enhanced by the young OSL age bon extracted from the basal hearth (using the proof 21030 years obtained from the near-surface tocols devised by Bird and GroÈ cke, 1997). This sediments. These data, in stark contrast to those for OSL date is the oldest yet reported for human oc- Koonalda Cave, provide compelling evidence for cupation of the Australian arid zone, slightly pre- adequate bleaching of the aeolian sediments depos- ceding the TL and calibrated 14C dates of 30,000± ited in Allen's Cave. 35,000 years for ®rst human presence at Lake The additive-dose and regenerative-dose single- Mungo (see Section 4.2.1) and Puritjarra rock shel- grain OSL dates are calculated using the average ter in the ``red centre'' (Smith et al., 1997). palaeodose from 28 grains (022 Gy) and 25 grains (024 Gy), respectively, although the individual 4.2.11. Puritjarra. The application of TL methods palaeodoses show a broad, quasi-Gaussian distri- to unburnt archaeosediments at Puritjarra began in bution with standard deviations of 5±6 Gy for both 1987, after publication of the ®rst set of 14C ages data sets (Murray and Roberts, 1997). A major for this site (Smith, 1987). A decade of research at cause of this spread is probably the variable beta Puritjarra has inspired the development of the microdosimetry experienced by the grains, some of ``selective bleach'' TL technique and has revealed which may be embedded in aeolian sediments some interesting discrepancies between the 14C whereas others may be surrounded by carbonate de- chronology derived from charcoal and the lumines- 866 R. G. ROBERTS cence dates obtained from heated and unheated soil fauna or humans. Although contamination of quartz sediments. Both methods agree on a date of the older 14C samples remains a possibility, such an 035,000 years for initial habitation of the rock shel- explanation does not account for the large age ter, with the underlying culturally-sterile sediments dierences in the Holocene and terminal Pleistocene being TL dated to 045,000 years using both total levels. The stone artefact typology and the phytolith bleach and selective bleach procedures in combi- (polymerised biogenic silica) record of the Last nation with the additive/regenerative ``Australian Glacial Maximum support the 14C chronology, so a slide'' technique (Smith et al., 1997). Within the cul- number of options were explored to explain the tural levels, however, the 14C and TL chronologies aberrant TL ages. Clay coatings on the quartz sand diverge markedly. Both sets of dates are in correct grains were considered as a possible cause of incom- stratigraphic order, but the age-versus-depth curve plete bleaching in antiquity. For the Holocene for the 14C dates is slightly concave in shape, levels, the selective bleach ages were younger than whereas TL dates indicate a linear rate of sediment their total bleach counterparts but still too old com- accumulation (Fig. 13). The two data sets agree pared to the 14C ages. The two TL methods, sup- best where their age-versus-depth curves intersect: plemented by OSL dating at three dierent at the 35,000 year-old level and near the ground laboratories (Adelaide, Oxford and Simon Fraser), surface (where selective bleach TL yielded a zero gave similar results for the Pleistocene sediments. age for grains removed by sticky tape from the Beta microdosimetry eects (due to the clay coat- modern surface). At intermediate depths, the selec- ings), water content variations, and radioactive dise- tive bleach TL ages are 5000±10,000 years older quilibrium were also considered, but were found to than their calibrated 14C counterparts. be inadequate to reconcile the 14C and luminescence In their search for an answer to this riddle, Smith chronologies. The only remaining possibility would et al. (1997) canvassed a series of possibilities, appear to be contamination of the deposit by in situ including contamination of 14C samples by modern disintegration of the local Devonian sandstone: as carbon, vertical displacement of artefacts and char- few as 5% contaminant grains would bring the dis- coal, and large-scale disturbance of the deposit by cordant TL ages into alignment with the calibrated

Fig. 13. Age-versus-depth plot of calibrated 14C determinations and ``selective bleach'' TL ages from Puritjarra rock shelter. TL ages are shown by solid circles, AMS 14C ages by open diamonds, and radiometric 14C ages by crossed diamonds. Ages are in thousands of years. The depth range is indicated for each sample, together with its 2s age uncertainty (from Smith et al., 1997). LUMINESCENCE DATING IN ARCHAEOLOGY 867

14C ages (Smith et al., 1997), and the use of single- Australia may also have been made by accom- aliquot OSL methods has been implemented to plished seafarers, whose deliberate ocean crossings identify such material (M. Smith, Personal might have been prompted by the sighting of natu- Communication, 1996). ral bush®res, ignited by lightning strikes, which can create smoke plumes that are visible at sea-level 4.2.12. Terrace teasers. Along the humid-temper- from distances of up to 260 km (Dortch and Muir, ate fringe of south-eastern Australia, human arrival 1980). Crossings to the eastern end of the Solomon as early as 40,000±47,000 years has been claimed by Islands can be made with at least one island always Nanson et al. (1987), on the basis of TL and 14C in sight. Once east of this island chain, however, dates for the Cranebrook Terrace river gravels from the sea gaps extend to more than 300 km, making which stone tools had been collected a decade ear- this truly the realm of the ancient mariners. lier (Stockton and Holland, 1974). The in situ status 4.3.1. The ®rst Melanesians. The oldest dates for of the artefacts has been questioned, however, and initial occupation of Melanesia are those from the the acceptance of these dates has been postponed Huon Peninsula, on the north-east coast of Papua until controlled excavations can locate artefacts in New Guinea (Fig. 8). A number of stone tools, primary stratigraphic position (Allen, 1989; Jones, including ``waisted axes'' that may have been 1989). The age of the river terrace was constrained hafted, were found in volcanic tephras deposited on by four ®ne-grain TL dates of circa 41,000± a raised coral reef terrace (Groube et al., 1986). 47,000 years, one sample giving the same age by This terrace (labelled IIIa) was dated by alpha-spec- ``partial bleach'' and ``total bleach'' methods (Nan- trometric U-series methods to circa 45,000± son and Young, 1987). These ages are consistent 53,000 years (Chappell, 1974), but this timespan has with the seven oldest ®nite 14C determinations on since been revised to circa 52,000±61,000 years, on wood and charcoal (circa 37,000±42,000 years BP), the basis of mass-spectrometric and new alpha-spec- although contamination of the submerged logs by trometric U-series determinations (Chappell et al., younger carbon in the groundwater can be inferred 1996b). TL dating was also undertaken, using silt- from the ®ve ages of circa 27,000±34,000 years BP sized quartz grains extracted from the set of three and the obtainment of older ages with more rigor- heavily weathered tephras. Ages of circa 42,000± ous pretreatment of selected samples (Nanson and 60,000 years were derived from the palaeodoses Young, 1987). (350±4758C plateau region) and the measured Human arrival by 45,000 years ago has also been radionuclide concentrations in completely dry sedi- suggested from the degree of pedogenesis of the ment (Groube et al., 1986). The most important artefact-bearing ``D Clay'' at Keilor, near artefacts, including waisted axes, were recovered Melbourne (Bowler, 1976). Optical dating of these from the two lowest tephras, which both gave TL river terrace sediments is planned (J. Allen, ages of 060,000 years. As typically observed for Personal Communication, 1995) to provide a quartz, no anomalous fading of the TL signal was further test of the present picture of continental observed over a period of one month, but unusually colonisationÐthat northern Australia was settled for quartz, and especially for ®ne-silt grains, the 10,000±20,000 years earlier than in the south. dose-response curves were reported as being linear, Across the Sahul Shelf in Melanesia, a region poss- even though palaeodoses of 0100 Gy were ibly traversed by the ®rst Australians, early occu- obtained. Perhaps feldspar inclusions were respon- pation sites have also been investigated by sible for the TL signal; this possibility will be exam- luminescence methods. These studies are discussed ined directly by optical dating of the sediments in the next section, together with those pertaining collected from this site by the author (Roberts et to the timing of settlement of the remote Polynesian al., 1994b). islands in the Paci®c Ocean, some of the last places Time-dependent variations in the dose rate gave on the planet to be colonised by people. much greater cause for uncertainty in these TL age estimates. The ingrowth of 230Th in the neighbouring coral reef would give rise to a measured U- series dose rate that is higher than the long-term 4.3. Melanesia and Polynesia average (and, thus, yield TL ages that are too Melanesia can be broadly de®ned to encompass young), but the measured potassium concentrations those islands running eastward from Irian Jaya/ posed a more serious problem. These were an Papua New Guinea to Fiji, while Polynesia includes order-of-magnitude lower than found in other those islands falling within the triangle created by tephras in the region, which Groube et al. (1986) Hawai'i, Easter Island and New Zealand. Irian attributed to leaching during strong weathering of Jaya/Papua New Guinea and Australia formed a the tephra. Using the higher potassium concen- single landmass during glacial periods of lowered tration reduced the TL ages of the lowest two sea-level, but the human colonisation of islands tephras to 037,000 years, but these represent mini- further east required open-sea voyages of several mum ages because they are calculated for zero tens to hundreds of kilometres. First landfall in water content in the tephra. Given the wet con- 868 R. G. ROBERTS ditions typical of tropical Melanesia, Groube et al. Pleistocene, a sea crossing of less than 50 km being (1986) concluded that the tephras are at least needed to make landfall on New Britain. Here, in 40,000 years old (but did not quote the water- the highland rainforests at Yombon, chert artefacts adjusted ages), consistent with the U-series dates have been found sealed beneath tephra layers for the underlying coral reef. For a water content (Pavlides and Gosden, 1994). Lumps of charcoal as- of 25%, the minimum ages of the basal tephras are sociated with the lowest artefacts have given three 047,000 years (using the higher K values, and 14C ages of circa 32,600±35,600 years BP, and this assuming an alpha-eciency ``a'' value of 0.1 and clay layer is unconformably overlain by a series of equal activities of uranium and thorium), and these Holocene tephras. Calibration of these dates gives ages increase to 076,000 years using the measured an age for artefact deposition of perhaps 38,000± K concentrations. A maximum age of 61,400600 42,000 years, although contamination of the years for the artefact-bearing tephras is indicated, samples cannot be dismissed given the proximity of however, by the latest mass-spectrometric U-series the dates to the 14C ``glass ceiling'' and their pro- date for the supporting reef (Chappell et al., curement from charcoal in a region with an annual 1996b). The probable age of the waisted axes from rainfall of more than 6 m. the lowest pair of tephra beds is, therefore, circa A short distance to the east of New Britain lies 47,000±61,000 years, a time period remarkably simi- New Ireland, where the basal cultural deposits at lar to that for ®rst human occupation of western the cave sites of Buang Merabak and Arnhem Land (Roberts et al., 1990a, 1994a). Matenkupkum (Fig. 8) have produced 14C dates of Other sites of comparable antiquity have not 032,000 and circa 31,000±33,000 years BP, respect- been discovered in Melanesia, the next most ancient ively (Allen, 1991; Allen et al., 1988; Gosden, 1995). archaeological sites being dated by 14C to no more The latter dates were obtained on marine shells than 36,000 years BP. One such site (Lachitu) lies from the dense shell midden which forms the ear- on the coast 800 km north-west of the Huon liest occupation layer at Matenkupkum; the dated Peninsula, but the others are found on the islands event, in this instance, coincides with the target of New Britain and New Ireland, which constitute event. These 14C ages may also be subject to con- the Bismarck Archipelago (Gosden, 1995). These tamination, as suggested by the increase in age of islands face the Huon Peninsula but they were 03000 years when the same shell species was pre- never joined by a land bridge during the pared in a nitrogen environment to minimise uptake

Fig. 14. Location map of the south-western Paci®c Ocean, with arrows showing sites from which pottery samples have been obtained for TL dating. The dash±dot line marks the approximate eastern boundary of continental rocks, and the dashed line indicates the ``andesite line''. Islands to the south and west of this line are characterised by andesitic rocks of relatively low uranium and thorium content, while islands to its north and east are based on basaltic oceanic volcanoes or coral built on volcanic basement (from Prescott et al., 1982). LUMINESCENCE DATING IN ARCHAEOLOGY 869 of modern atmospheric carbon (Allen, 1994). To arations contaminated by CaCO3 can be normalised check for 14C age underestimation at the New successfully (Li and Wintle, 1994). Ireland and New Britain sites, optical dating of Obsidian is a glass of volcanic origin that was sediments is foreshadowed (J. Allen, Personal widely traded in Melanesia and Polynesia, and by Communication, 1995). prehistoric people in other regions of the world (see Renfrew and Bahn, 1991). Although the timing of 4.3.2. The peopling of Polynesia. Other types of the volcanic event which formed the obsidian may artefacts have been examined by luminescence bear no relation to that of artefact manufacture, methods, an early attempt being made to obtain TL TL has been used to supplement the many chemical dates from the distinctive decorated pottery associ- methods of source identi®cation. Pieces of obsidian ated with the ``Lapita'' culture (Prescott, 1982; from a single source typically exhibit small vari- Prescott et al., 1982; see Fig. 14). There is ®erce dis- ations in their natural and laboratory-dosed TL sig- agreement among archaeologists, linguists and gen- nals, but between-source variations are sucient to eticists about the place of origin of the Lapita permit characterisation of obsidian quarries in New culture (be it south-east Asia or the Bismarck Zealand, North and Central America, the Archipelago), the rate of spread of Lapita people Mediterranean and Kenya (Carriveau and Nievens, across Melanesia and into Polynesia, and their 1979; GoÈ ksu and TuÈ retken, 1979; Huntley and ancestral relationship to the Polynesians (e.g. Allen, Bailey, 1978; Leach and Fankhauser, 1978; Rendell 1991; Bellwood, 1991; Diamond, 1988; Gibbons, et al., 1985). TL has yet to be used to characterise 1994a; Gosden et al., 1989; Terrell, 1989). Lapita obsidian in Melanesia, but given the stratigraphic pottery has been found as far east as Fiji and the sequences of ashfall layers and soils that occur western Polynesian islands of Samoa and Tonga, throughout this region, a detailed chronology could and some of these sites have been dated by 14C to be obtained by TL dating of the heated glass shards as early as 3500 years BP. (e.g. Berger, 1991; Berger, 1992) or quartz grains TL analysis of Lapita pottery was recognised as a (e.g. Liritzis et al., 1996), and by optical dating of means of dating the target archaeological event, the unheated sediments. Some West New Britain rather than relying on association-based 14C chron- sites have 14C age control for several of the ologies (Prescott, 1982; Prescott et al., 1982). Of the Holocene tephras (Machida et al., 1996; Pavlides, pot sherds examined, however, most were composed 1993). Direct dating of the volcanic eruptions of volcanic minerals ill-suited to TL dating, and remains a priority, however, to establish their only a few sherds yielded ®ne-grain potassium feld- impact on human activities in this important source spars and coarse-grain quartz. The latter category area for chert and obsidian, and possible Lapita includes pottery from Viti Levu, the main island of ``homeland'' (Gosden et al., 1989). To this end, the Fiji group, which has a core of relict continental luminescence samples were collected recently from rock. Some of these sherds proved dicult to date one such site (Namundo), where Lapita pottery and because of the coralline setting, which results in a obsidian tools occur between tephras which are low environmental dose rate (and the relatively lar- believed to be older than 2000 years BP (R. ger contribution from cosmic rays), weak TL emis- Torrence, Personal Communication, 1997). sions (owing to the small accumulated doses), and the prevalence of spurious TL from CaCO3. Dates 4.3.3. New Zealand. Luminescence methods have were, nonetheless, obtained for several sherds from also been largely overlooked in attempts to date two Fiji sites (Naqarai and Yanuca). TL dating of palaeoenvironmental evidence for human colonisa- one sherd con®rmed its age based on ceramic type, tion of the remote Melanesian and Polynesian both estimates being signi®cantly younger than the islands. Anthropogenic disturbance of island ecosys- calibrated 14C determinations on charcoal. The lat- tems may be inferred from vegetation changes ter were made at the Gakushuin (GaK) Laboratory, documented by pollen and charcoal records (e.g. whose dates for Paci®c and Asian samples are fre- Kirch and Ellison, 1994; but see Anderson, 1994), quently anomalous when compared with 14C dates animal extinctions after human landfall (e.g. Dia- from other laboratories (Spriggs and Anderson, mond, 1991; Steadman, 1995), and the introduction 1993). Two other Lapita sites on Viti Levu of non-domesticated animals transported by (Sigatoka and Natunuku) were sampled recently by humans (e.g. Flannery et al., 1988). Fossil remains the author and A. Anderson, to attempt optical of the commensal Paci®c rat (Rattus exulans) have dating of the beach-dune deposits that have pre- been found throughout the region, with AMS 14C viously yielded 14C dates suggestive of early coloni- dates of up to 02000 years BP being reported sation (Davidson et al., 1990). Low environmental recently for rat remains in New Zealand (R. Hold- dose rates (as shown by in situ gamma spec- away, 1996). These bone-gelatin dates are much trometry) may again be problematic, but quartz older than the ®rst unequivocal archaeological evi- OSL emissions should be much brighter than the dence for human colonisation at 0850 years BP ``thermally quenched'' 3258C TL signals (Aitken, (Anderson, 1991), and are considered to re¯ect the 1994) and the IRSL emissions from ®ne-grain prep- introduction of rats by transient human visitors 870 R. G. ROBERTS more than 1000 years before the permanent Polyne- the petroglyph and on the adjacent rock surface (J. sian settlement of New Zealand (R. Holdaway, Stone, Personal Communication, 1995). Geological 1996). The validity of these early 14C dates has been micro-erosion methods have been used for some challenged by Anderson (1996), who argues instead engravings (Bednarik, 1992, 1995b) and, for petro- for rat-bone contamination. To address the chrono- glyphs covered by mineral accretions, minimum logical uncertainty, further AMS 14C analyses are ages have been reported using cation-ratio and being done and, in collaboration with R. Hold- AMS 14C methods on the inorganic and organic away, sediment samples have been collected for components, respectively (e.g. Dorn et al., 1988, optical dating from the key rat-bone sites. 1989; Nobbs and Dorn, 1993; Watchman, 1993a,b). A resolution to this matter is vital not only for Individual quartz or feldspar grains trapped in Paci®c prehistory, but also for models of island these precipitate ``skins'' may be amenable to opti- zoogeography as, on the current evidence, the rat- cal dating (Bednarik, 1996), provided that modern induced extinctions of small native animals should sunlight does not pentrate far through the opaque have preceded the elimination of larger fauna by rock ``varnish''. human predation and habitat destruction (R. Luminescence methods have recently contributed, Holdaway, 1996). Luminescence methods should be albeit indirectly, to the debate over the antiquity of used more often to construct Quaternary chronolo- the CoÃ a petroglyphs in north-eastern Portugal. On gies for palaeontological sequences, given their fre- stylistic grounds, these pecked ®gures are assigned quent employment as proxy archaeological data an Upper Palaeolithic age by some archaeologists (e.g. Diamond, 1991). Optical dating methods also (e.g. ZilhaÄ o, 1995), whereas micro-erosion studies of deserve more widespread application to hitherto the petroglyphs and AMS 14C dating of organic in- disregarded areas of archaeometric research, such clusions in the overlying silica accretions suggest an as the dating of prehistoric art. To my knowledge, age of less than 03000 years (Bednarik, 1995a,b; only a few luminescence studies have contributed to Watchman, 1995). A late Holocene age is supported the latter themeÐstudies in which, as discussed in by preliminary IRSL dates of 01000 years and circa the following and penultimate section of this review, 4000±6000 years for the ¯uvial terrace sediments the dating methods used are ®t to be acclaimed that underlie the site (M. Lamothe in Watchman, ``state of the art''. 1995). These IRSL ages were obtained using single- grain methods for feldspars (Lamothe et al., 1994), which revealed a population of uniformly well 5. PAINTINGS AND PETROGLYPHS bleached, young grains (M. Lamothe, Personal Communication, 1995). Moreover, if some of the One of the most striking features that dis- grains had been poorly bleached at deposition then tinguishes modern humans from other hominids their apparent ages would have been too old, rather and the rest of the animal kingdom is the ability to than too young. transform thoughts into pictures (Valladas et al., These late Holocene age determinations have, 1992). These pictures may be produced by an addi- however, proven highly controversial (Fischman, tive process, such as painting, drawing, or the appli- 1995; ZilhaÄ o, 1995). A second set of similarly recent cation of a residue such as beeswax, or they may be AMS 14C ages has been explained in terms of created by a reductive process, such as engraving, sample contamination by younger carbon (Dorn, pounding or pecking (Bednarik, 1993). The latter 1997a; see also Dorn, 1997b), and it now appears group of pictures are termed ``petroglyphs''. Until that the rock panels bearing the petroglyphs were the advent of numerical dating methods, rock art available for inscription during the Upper and other rock markings could be placed only in a Palaeolithic, as shown by the late Pleistocene ``ex- relative age sequence, based on criteria such as posure ages'' determined using 36Cl (Phillips et al., style, content (e.g. extant or extinct animals), order 1997). A moderately stable landscape is also of superposition, and degree of weathering. Much suggested by the latter study, although the frequent eort has, therefore, been made to directly date ¯oods that sweep through the CoÃ a gorge might rock pictures, and luminescence dating is being used account for the recent deposition of the inset river increasingly to assist in this challenging endeavour. terraces. Direct dating of the CoÃ a petroglyphs seems, therefore, the only way to resolve this chronological disputeÐperhaps optical dating of 5.1. Petroglyphs the overlying silica skins should be attempted, using The direct dating of petroglyphs has been es- the ``micro-stratigraphic'' approach of Richards pecially intractable, because the reductive process (1994). involved in their manufacture leaves behind no sub- Optical dating methods have been exploited stance to signify ``time zero'' (Bednarik, 1993). recently to directly date an unusual type of petro- Markings are generally too shallow to distinguish glyph: the White Horse chalk ®gure at Ungton, dierential production rates of cosmogenic radio- near Oxford (Rees-Jones and Tite, 1997a,b). This nuclides (e.g. 36Cl, 10Be, 26Al) in the deepest part of elegant ®gure was made on a hillside by cutting a LUMINESCENCE DATING IN ARCHAEOLOGY 871

Fig. 15. Stratigraphic section at the belly of the Ungton White Horse, showing the sample locations and the dates obtained from ®ne-grain quartz (OSL) and polymineral (IRSL) extracts (after Rees-Jones and Tite, 1997b). series of ditches and packing them with chalk. It is 5.2. Paintings the only prehistoric equine hill-®gure in Britain These innovative applications of optical dating to and, on stylistic evidence, was thought to be either petroglyphs have only recently been matched by Celtic (late Iron Age) or Saxon in age. OSL and optical dating of prehistoric paintings. Accelerator IRSL dates were obtained from ®ne-grain colluvial 14C dating has been used for pictures that contain sediments that had washed onto the ®gure and been subsequently buried by chalk added to the ®gure organic materials such as charcoal (Clottes et al., during repairs (Fig. 15). Several aspects of this 1995; David, 1992; McDonald et al., 1990; Valladas novel dating problem were of concern to the geo- et al., 1992), blood proteins (Loy et al., 1990), plant chronologists: adequate bleaching of the sediments ®bres and other organic binders (Russ et al., 1990; at deposition; the magnitude of the palaeodose cor- Watchman and Cole, 1993), and beeswax (Nelson 14 rection needed for such young samples; the di- et al., 1995). AMS C methods have also been culty in measuring accurately the dose rate in a applied to the organic fraction of mineral accretions low-activity material such as chalk; and the uncer- formed on top of rock paintings in northern tainty surrounding the water content. Australia (Morwood et al., 1994; Watchman, Reassurance that the sediments were well 1993a,b; Watchman et al., 1997). To date the target 14 bleached at deposition is given by the close corre- (archaeological) event by C requires that the car- spondence between the IRSL date (3230430 bon ``clock'' starts to tick at the same time as the years) and the two OSL dates (2890340 and picture is created, a condition which may not be 3020360 years), as quartz and feldspar bleach at met if ancient charcoal is used for drawing dierent rates (Godfrey-Smith et al., 1988). All (Bednarik, 1996) or if the source of the organic three ages are, however, very sensitive to small vari- matter is ambiguous (e.g. Loy, 1994; Nelson, 1993) ations in the annual dose: this is less than 0.5 mGy, or subject to contamination by younger carbon (e.g. 14 of which the cosmic-ray dose rate must contribute a discrepant C dates for the same picture; sizeable (but unreported) proportion. The water McDonald et al., 1990). content is too low to introduce any signi®cant bias, Optical dating could, in principle, be used to while any inhomogenity of the gamma dose rate in determine the age of rock paintings at sites exposed this shallow (<70 cm thick) deposit and any dise- to indirect or direct sunlight. The surface grains on quilibrium due to leaching of uranium (as observed the rock shelter would be eectively ``zero-age'' im- in other carbonate-rich deposits, e.g. Olley et al., mediately prior to painting, but a thick application 1997a) should be adequately accommodated by the of paint or charcoal might be sucient to seal a use of in situ gamma spectrometry. The dates of few grains from further sunlight exposure. These 03000 years place the construction of the White grains would then accumulate a radiation dose, Horse in the Bronze Age or early Iron Age, older which could be determined using single-grain opti- than originally anticipated by the archaeologists but cal dating methods. Even pictures in caves may be coincident with other evidence for intense Bronze amenable to optical dating, because torches were Age activity in the region. used to illuminate caves for painting and drawing 872 R. G. ROBERTS

(e.g. Clottes et al., 1995). Exposure to the light bleached subsequently by light transmitted through from a naked ¯ame held close to the rock surface the paint. immediately before the application of paint or char- Optical dating may also be suitable for quartz coal might be sucient to reset the luminescence and feldspar grains embedded in opaque mineral ``clock'' of the surface grains; this proposition is skins, and other accretions, which overlie and easily tested experimentally. Optical dating holds underlie paintings in rock shelters: minimum and the advantage over 14C methods in these situations, maximum ages, respectively, are thereby obtained as the target event (the date of application of paint) for the target event. The fossilised remains of insect is contemporaneous with the optically dated event nests represent one such type of ``accretion''. (the time since the painted grains were last exposed Northern Australia has some of the most spectacu- to light). In practice, a ``micro-stratigraphic'' strat- lar, and possibly among the earliest, rock art in the egy (e.g. Liritzis, 1994; Richards, 1994) would need world (e.g. Chaloupka, 1993; Flood, 1997; Walsh, to be adopted, to demonstrate that the sun-lit (or 1988; Watchman, 1993b). The nests of mud-wasps torch-lit) grains had been reset initially and not are often built on top of these rock paintings

Fig. 16. Simpli®ed version of the Kimberley rock art sequence proposed by Walsh (1994). Note the hiatus anticipated between the ``Wandjina'' and ``Bradshaw'' period motifs, and that predicted between the ``Bradshaw'' and ``Irregular In®ll Animal Period'' paintings (after Walsh, 1994, and Morwood et al., 1994). LUMINESCENCE DATING IN ARCHAEOLOGY 873

(Naumann, 1983; Waterhouse, 1991), while the face (Naumann, 1983; Waterhouse, 1991). Although ``stumps'' of some nests have been painted over. By some grains will have been exposed to sunlight for dating the remnants of nests above and below a 0340 years on the stump of the earlier nest, most painting, its age can be bracketed. This approach will have been derived from the unbleached interior has recently been taken by Roberts et al. (1997), of these remnant portions. A more enigmatic result who extracted sand-sized quartz grains from mud- is the non-zero palaeodose obtained for the outer wasp nests in the Kimberley region of northern portion, and for the core of the modern nest (which Western Australia. These muddy sediments would gave a palaeodose of 0.150.02 Gy). Further in- have been collected by wasps from the banks of vestigations should reveal whether this represents streams or pools, and then carried to the rain-pro- an artefact of thermal transfer, the inclusion of a tected ceilings and recesses of rock shelters where small proportion of incompletely bleached grains, the nests are constructed. Final solar exposure of or another cause. the quartz grains thus occurs when the mud nest is The two nests associated with the possible built, an archaeologically `ìnstantaneous'' event. ``Bradshaw'' ®gure were dated using only the single- To permit optical dating of the grains concealed aliquot protocol, owing to the paucity of grains for within the body of a nest, centimetre to decimetre comparative multiple-aliquot analysis. The core of sized nests were removed at night using only red the larger nest yielded a palaeodose of 033 Gy, for (>590 nm) light (Roberts et al., 1997). Five of the which the above non-zero ``residual'' OSL palaeo- nests examined were built over paintings, whose dose is a trivial fraction. The optical date of relative ages could be estimated from a superimpo- 17,5001800 years provides a minimum Last sition sequence of motif styles (Morwood et al., Glacial Maximum age for the underlying painting, 1994; Walsh, 1994; see Fig. 16). Three lightly- making it the world's oldest dated painting of a cemented nests were constructed on parts of human ®gure. The second nest associated with this ``Wandjina'' ®gures (which are believed to be painting demanded a novel optical dating approach, Holocene in age) and two petri®ed nests were built because of its small size. Quartz grains were over the head-dress of a human ®gure, possibly re- extracted by disaggregation of the whole nest and, lated to the more archaic ``Bradshaw'' style. Two as there were too few grains for multiple-grain other nests, not clearly associated with any paint- analysis, individual grains were dated using addi- ings, were also collected: a heavily cemented nest in tive-dose and regenerative-dose single-aliquot proto- the ``Wandjina'' rock shelter and a modern nest cols (Murray and Roberts, 1997; Roberts et al., (<2 years old), the latter providing a check that the 1997). Most grains gave palaeodoses of 14±59 Gy, quartz grains had been bleached completely when forming broad, approximately Gaussian, frequency the nest was made. distributions. The spread in the distributions prob- Initial tests were conducted on the depth of pen- ably arises from grain-to-grain dierences in dose- etration of modern sunlight into one of the response behaviour and beta microdosimetry, as ``Wandjina'' nests. The largest nest collected was reported for other analyses of individual quartz divided into seven portions, the ®rst being com- grains (Murray and Roberts, 1997; J. Olley, posed of the outer few millimetres of nest mud and Personal Communication, 1997). A few grains had successive portions consisting of mud hidden deeper weak natural OSL signals, presumably because they within the nest; the ``core'' portion had been are insensitive to radiation or are derived from the attached to the shelter wall. The palaeodose of each sun-bleached surface of the nest. These ``dim'' portion was determined using conventional mul- grains were further analysed using only the regen- tiple-aliquot additive-dose OSL procedures, and erative-dose protocol, in which the total light sum concordant palaeodoses were obtained from two is measured. Palaeodoses in the range 0.5±8 Gy portions examined using a single-aliquot additive- were obtained. Such values are intermediate dose protocol (Murray et al., 1997). The outer layer between those expected from grains exposed to yielded a palaeodose of only 0.140.02 Gy (corre- modern sunlight (00.15 Gy) and the >14 Gy sponding to an age of 0110 years), the next four palaeodoses obtained from the majority (75%) of portions had palaeodoses of 0.35±0.43 Gy (giving a the single grains examined. These grains are weighted mean age of 27020 years), and the thought to have originated from the ``twilight zone'' palaeodoses for the innermost two portions jumped of the nest, where the growth in palaeodose is oset to 00.85 Gy (equivalent to 61040 years). These by partial bleaching from silica-®ltered sunlight. ``micro-stratigraphic'' data are interpreted as show- The equilibrium conditions existing in this twilight ing that modern sunlight penetrated only as far as zone remain an area for future investigation. Grains the outer layer, and that two ``generations'' of nest with palaeodoses of >14 Gy yielded regenerative- are preserved: a 0610 year-old basal portion and a dose and additive-dose optical dates of 0270 year-old central portion. This ®nding is con- 016,400 years (average of 15 grains) and sistent with the observation that mud-wasps prefer 023,800 years (average of 21 grains), respectively to build nests near existing nests and on the stumps (Roberts et al., 1997). The latter determination is of abandoned nests, rather than on a bare rock sur- in¯uenced by the inclusion of three grains that con- 874 R. G. ROBERTS tain about twice the average palaeodose; their omis- obtained from stick-nest rat middens (Pearson and sion would reduce the mean age to 021,200 years. Dodson, 1993), but they also are amenable to AMS Despite these uncertainties, the single-grain OSL 14C dating (Brown et al., 1989; Kelly et al., 1991), dates con®rm the Pleistocene antiquity of the thereby providing a possible cross-check on the underlying painting and demonstrate the utility of optical dates. Roberts et al. (1997) report 14C/OSL single-grain analyses in situations (e.g. the dating of age comparisons for recent mud-wasp nests, but the painted nest stumps) where sample material is too Last Glacial Maximum nests, which are of most sparse for multiple-grain methods. interest to archaeologists, yielded an insucient A Pleistocene age for this painting, and a young mass of pollen grains and phytoliths for AMS 14C age for two other nests, is independently indicated dating (although enough could be extracted for by the radionuclide data (Roberts et al., 1997). palaeoenvironmental work). The potential for lumi- Nests built during the last 500±600 years exhibit a nescence dating of phytoliths certainly warrants clo- 40±110% excess of 226Ra over its parent 230Th, ser attention. whereas secular equilibrium prevails in the Pleistocene nest mud. If the latter had been deposited with a similar initial excess of `ùnsupported'' 5.3. Phytoliths 226Ra, then equilibrium would have been attained after a period of 8000 years (5 half-lives of 226Ra). Rowlett and Pearsall (1993) made a pioneering, This represents a minimum age for the nest mud, and seemingly successful, attempt to date burnt and and hence for the underlying painting, as secular unburnt phytoliths using TL. Phytoliths are com- equilibrium could have been maintained thereafter posed of microcrystals of biogenic silica (but are for an inde®nite period. At the other end of the age often described as being amorphous in structure) spectrum, the optical dates of <200 years for two and thus, Rowlett and Pearsall (1993) surmised, nests are supported by the presence of a 160±270% they could be as amenable to luminescence dating excess of 210Pb on the sediments: this excess is prob- as quartz and chert. Given their small size ably due to atmospheric fallout and indicates that (<200 mm diameter) and translucence, I can envi- the nests are built from mud collected within the sage three mechanisms by which the luminescence last 100 years (5 half-lives of 210Pb). AMS 14C de- signal could be reset: by secretion of the phytolith terminations on the pollen grains extracted from in the plant, by exposure to sunlight after plant de- these nests also give dates that are indistinguishable composition (and during any later episodes of inso- from modern (Roberts et al., 1997). lation), and by heating in anthropogenic or natural As with the White Horse study, the mud-wasp ®res. Burnt phytoliths from two archaeological sites nests presented a challenge to the accurate determi- in coastal Ecuador were investigated using a regen- nation of the dose rate. The gamma-ray contri- erative-dose procedure and 00.8 g aliquots. The bution from the wall of the rock shelter, and the procedural details are sketchy, however, and no TL dose due to cosmic rays, was estimated by in situ growth curves or palaeodose plateaux are given. and high-resolution gamma spectrometry, and from Nonetheless, the phytoliths appear to luminesce and calcium ¯uoride TL dosimetry capsules placed on- respond to applied radiation. TL dates of circa site for one year. The beta dose rate, almost all of 1000±3000 years BC were derived from the 350± which is derived from the nest mud, was deduced 3808C glow curve region, which agreed with the from XRF and alpha spectrometry, while U and Th ages assigned to the associated ceramics and 14C concentrations in the acid-etched quartz grains were samples. TL signal integration over the 200±3808C measured by laser-ablation ICP-MS. Disequilibria region gave rise to similar results, although the TL in the 238U decay series included radon gas escape in the 200±2758C region appears to be thermally of 020% in a recent nest, and the aforementioned unstable (Rowlett and Pearsall, 1993). `ùnsupported'' excesses of 226Ra and 210Pb. Intriguingly, slightly younger ages were obtained Potassium concentrations exhibited the greatest from the unburnt phytoliths (Rowlett and Pearsall, variation from nest to nest, however, ranging from 1993), which raises the prospect that a light-sensi- 20 Bq kg1 (in the modern nest) to 500 Bq kg1 (in tive signal may be present and this was partially the Pleistocene nest mud). Given these ®ndings and bleached during laboratory pretreatment (no special the need for single-aliquot or single-grain methods illumination precautions having been taken). of palaeodose determination, optical dating of mud- Phytoliths may thus exhibit luminescence signals wasp nests and sediment accretions created by other upon thermal and optical stimulation. The electron animals (e.g. ants, termites, mud-nesting birds) is traps involved in luminescence production need to never likely to be straightforward. be investigated, but perhaps, in the new millennium, Many mud-wasp nests also contain fossil pollen, optical dating of individual phytoliths will be a spores and phytoliths (Roberts et al., 1997). These complementary technique to optical dating of single not only provide palaeovegetation information for grains of quartz and feldspar. Even in the absence the ``snapshot'' of time represented by a nest, in a of a light-sensitive signal, TL dating could be use- similar manner to the palaeoecological data fully applied to unburnt phytoliths whose TL LUMINESCENCE DATING IN ARCHAEOLOGY 875

``clock'' starts when the silica microcrystals are weathering of the paint and rock substrate, and the formed; TL analysis may be feasible when AMS formation of mineral accretions. Taphonomic pro- 14C dating is not, because the latter exploits only blems are equally likely to hinder the long-term sur- the tiny amounts of occluded carbon. Using mud- vival of cave paintings (Clottes et al., 1995). wasp nests as an example, the phytoliths concealed Petroglyphs are a more durable type of rock art. in the mud are likely to be contemporaneous with Circular rock engravings (pecked cupules) are wide- the time of nest construction (AMS 14C dating spread across northern Australia and they are con- shows this is true for pollen grains deposited in sidered to predate paintings in the Kimberley modern nests; Roberts et al., 1997). The palaeodose region (Morwood et al., 1994; Walsh, 1994; see acquired by the phytoliths since crystallisation Fig. 16). In the sandy deposits at Jinmium rock could thus provide a check on the solar-reset OSL shelter, a fallen fragment of sandstone marked with palaeodoses in the quartz grains, but their common pecked cupules was buried by sediments TL dated dose rate precludes an independent age comparison. to circa 58,000±75,000 years (Fullagar et al., 1996); 58,000 years represents a minimum age for the cupules, which could have been made many millennia before the fragment fell from the shelter wall. 5.4. The dating palette The lowest pecked cupules engraved on the wall The above discussion has focussed on experimen- also occurred between these chrono-stratigraphic tal applications of luminescence dating to rock art. levels, while pieces of ochre were recovered from The expression ``necessity is the mother of inven- the underlying deposits, dated to more than tion'' can be justly applied to these studies, each of 075,000 years. But the TL ages would appear to which required a novel approach. Optical dating of extend the timeframe for rock art in Australia, ®ne-grain quartz from the White Horse, and single- almost doubling its duration if the maximum age grain dating in the CoÃ a valley and northern for the pigments is accepted. But the TL ages for Australia, are not without their problems, and the the Jinmium deposits, and the antiquity of the rock dose rate determinations for the mud-wasp nests art contained therein, are likely to be gross overesti- proved unusually troublesome. But further testing mates for the reasons I have described at length in of these methods, and continued creative appli- Section 4.2.7. cations, should elevate luminescence dating to a Appropriate luminescence methods are, nonethe- more prominent position in the ®eld of rock art less, a valuable tool in the arsenal of the rock art dating. New possibilities, such as luminescence dat- chronologist. They can be used to good eect to ing of phytoliths and optical dating of individual constrain the ages of pieces of pigment, painting or grains hidden beneath mineral accretions or paint, petroglyph that may have fallen into an archaeolo- could be investigated. The time of emplacement of gical deposit. Not only can optical dating be prehistoric stone arrangements can also be estab- applied to the art-bearing sediments, but TL dating lished using the ``micro-stratigraphic'' luminescence of associated heated artefacts can provide ad- approach of Liritzis (1994), who determined the ditional age control. Both strategies were employed date of construction of a Mycenean wall (see at Pedra Pintada in Brazil (Michab et al., 1997; Section 4.2.5); in these situations, the surface grains Roosevelt et al., 1996). The continuation of such on the undersides of the stones are last exposed to applications, and the development of innovative sunlight when the stones are emplaced. methods for the direct dating of paintings and pet- The contributions made to rock art dating by tra- roglyphs, promises to make rock art research an ditional multiple-aliquot luminescence methods exciting new theatre of luminescence activity. should not, however, be overlooked. Considerable controversy has been generated from the indirect dating of petroglyphs and ochres in northern 6. RETROSPECT AND PROSPECT Australia (Bahn, 1996; Holden, 1996; Morell, 1995). Lumps of red and yellow ochre and abraded hae- Luminescence dating can claim a place in matite were found at Malakunanja II and archaeometry that stretches back nearly half a cen- Nauwalabila I in quartz sediments dated to circa tury. In this overview, I have attempted to cover 50,000±60,000 years using TL and OSL, respectively some of the earliest applications and several of the (see Section 4.2.5). Whether these pigments were latest contributions to world archaeology. A few used for rock art or body decoration is unknown, but studies have been discussed in detail, particularly their presence in the initial occupation levels implies the recent controversial ®ndings from Tabun Cave that rock paintings may have been made in western in Israel (Mercier et al., 1995b), Diring Yuriakh in Arnhem Land more than 20,000 years before the Siberia (Waters et al., 1997a,b), and Jinmium rock earliest dated paintings in Europe (Clottes et al., shelter in Australia (Fullagar et al., 1996). These 1995). It is doubtful, however, that paintings could TL contributions to the story of human antiquity remain visible after many tens of millennia in an and ancestry are complemented by the novel open rock shelter, given the natural processes of deployment of optical dating methods to anthropo- 876 R. G. ROBERTS genic earthen mounds (Feathers, 1997b), quartzite ported by a Queen Elizabeth II Fellowship awarded by pebbles (Richards, 1994) and rock art (Rees-Jones the Australian Research Council. and Tite, 1997a,b; Roberts et al., 1997). There are also archaeometric applications of luminescence dating that have not been canvassed REFERENCES in this review. TL investigations of fossil bone, Abeyratne, M. (1994) TL dating of Sri Lankan archaeolo- tooth and shell (Christodoulides and Fremlin, 1971; gical sites. Quaternary Science Reviews (Quaternary Driver, 1979; Jasinska and Niewiadomski, 1970), Geochronology) 13, 585±588. Abeyratne, M., Spooner, N. A., GruÈ n, R. and Head, archaeometallurgical slags (Elitzsch et al., 1983), J. (1997) Multidating studies of Batadomba Cave, Sri ancient man-made glass (Sanderson et al., 1983) Lanka. Quaternary Science Reviews (Quaternary and calcite deposits in caves (e.g. Aitken and Geochronology) 16, 243±255. Bussell, 1982; Debenham, 1983; Debenham and Ackerman, R. E. and Carlson, R. L. (1991) Diring Aitken, 1984; Wintle, 1978) have been largely dis- Yuriak: an early Paleolithic site in Yakutia. Current Research in the Pleistocene 8, 1±2. continued during the last decade, in favour of ¯int Adams, G. and Mortlock, A. J. (1974) Thermoluminescent and sediment dating. Our current knowledge of the dating of baked sand from ®re hearths at Lake potential and pitfalls of these applications is well Mungo, New South Wales. Archaeology and Physical summarised by Aitken (1985) and little more can be Anthropology in Oceania 9, 236±237. Adovasio, J. M. and Pedler, D. R. (1997) Monte Verde added here. The call by Wintle (1980) for further and the antiquity of humankind in the Americas. studies of calcite and biological materials remains Antiquity 71, 573±580. as appropriate now as it was then, given their undi- Adovasio, J. M., Gunn, J. D., Donahue, J. and minished archaeological importance. Stuckenrath, R. (1978) Meadowcroft Rockshelter, What are the new archaeological frontiers for 1977: an overview. American Antiquity 43, 632±651. Adovasio, J. M., Donahue, J. and Stuckenrath, R. (1990) luminescence dating, as we approach the 21st cen- The Meadowcroft Rockshelter radiocarbon chronol- tury? A reliable method to date individual grains of ogy 1975-1990. American Antiquity 55, 348±354. quartz and feldspar would allow a variety of con- Adovasio, J. M., Donahue, J. and Stuckenrath, R. (1992) tentious issues to be addressed, such as the biotur- Never say never again: some thoughts on could haves and might have beens. American Antiquity 57, bation of archaeological deposits, the dating of 327±331. rock art and standing stones, and the identi®cation Aitken, M. J. (1977) Thermoluminescence and the archae- and rejection of ``contaminant'' grains. Such a ologist. Antiquity L1, 11±19. move would bring luminescence dating into line Aitken, M. J. (1978) Radon loss evaluation by alpha with other numerical dating methods, which have counting. Journal of the Council of Europe PACT 2, 104±114. already reaped the bene®ts from using small Aitken, M. J. (1985) Thermoluminescence Dating. samples (Wintle, 1996). Novel materials, such as Academic Press, London. phytoliths, and the extraction of tried-and-tested Aitken, M. J. (1989) Luminescence dating: a guide for minerals from unusual contexts (e.g. quartz grains non-specialists. Archaeometry 31, 147±159. Aitken, M. J. (1990) Science-based Dating in Archaeology. from mud-wasp nests and quartzite artefacts) Longman, London. should be further investigated, as should the pursuit Aitken, M. J. (1994) Optical dating: a non-specialist of new signals and procedures to allow lumines- review. Quaternary Science Reviews (Quaternary cence dating to contribute to questions of early Geochronology) 13, 503±508. Quaternary prehistory. The direct dating of heated Aitken, M. J. (1998) An Introduction to Optical Dating. Oxford University Press, Oxford (in press). artefacts, the historical foundation of luminescence Aitken, M. J. and Bussell, G. D. (1982) Dating of fallen dating, will continue to play a major role in archae- stalactites. Journal of the Council of Europe PACT 6, ology. But while developments of long-range and 550±554. high-precision luminescence chronologies pose the Aitken, M. J. and Valladas, H. (1992) Luminescence dating relevant to human origins. Philosophical hardest challenges, they also oer the greatest pro- Transactions of the Royal Society of London spects to open new windows onto our human past. 337(Series B), 139±144. Aitken, M. J., Tite, M. S. and Reid, J. (1963) Thermoluminescent dating: progress report. AcknowledgementsÐThis review would not have been Archaeometry 6, 65±75. possible without the generous supply of reprints, preprints Aitken, M. J., Tite, M. S. and Reid, J. (1964) and illustrations provided by many of my colleagues. I Thermoluminescent dating of ancient ceramics. thank them for their support and apologise if their work Nature 202, 1032±1033. has been misrepresented or misinterpreted. Special appreci- Aitken, M. J., Zimmerman, D. W. and Fleming, S. ation is extended to the following people, whose comments J. (1968) Thermoluminescent dating of ancient pot- on a complete draft of this review have greatly improved tery. Nature 219, 442±445. its clarity and accuracy: Jim Feathers, Simon Holdaway, Aitken, M. J., Moorey, P. R. S. and Ucko, P. J. (1971) Jon Olley, Mike Smith and Ann Wintle. Several sections The authenticity of vessels and ®gurines in the were read also by other colleagues, whose input is grate- Hacilar style. Archaeometry 13, 89±141. fully acknowledged. Special thanks are given to Rhys Aitken, M., Huxtable, J. and Murray, A. (1983) Jones, who introduced me to Australian archaeology, and Thermoluminescence dating of Ban Chiang pottery. to Andrew Murray, who encouraged me to pursue lumi- Antiquity 57, 217±218. nescence dating. It has been a pleasure and an inspiration Alekseev, M. N., Kamaletdinov, V. A., Siegert, C., with them both during the past decade. The author is sup- Grinenko, O. V., Gnibidenko, Z. N., Gravis, LUMINESCENCE DATING IN ARCHAEOLOGY 877

G. F. and Shamshina, E. A. (1990) Problems of the of Modern Humans, eds. P. Mellars and C. Stringer, geology of the Diring Yuryakh Paleolithic site. pp. 589±610. Edinburgh University Press, Edinburgh. Report to the Ministry of Geology of the USSR and Bar-Yosef, O. (1994) The Lower Paleolithic of the Near Academy of Sciences of the USSR, Siberian East. Journal of World Prehistory 8, 211±265. Department, Yakutsk. Bednarik, R. G. (1989) On the Pleistocene settlement of Allen, J. (1989) When did humans ®rst colonize Australia? South America. Antiquity 63, 101±111. Search 20, 149±154. Bednarik, R. G. (1992) A new method to date petro- Allen, J. (1991) Introduction. In Report of the Lapita glyphs. Archaeometry 34, 279±291. Homeland Project, eds. J. Allen and C. Gosden, pp. Bednarik, R. G. (1993) The calibrated dating of petro- 1±8. Occasional Papers in Prehistory 20, Department glyphs. The Artefact 16, 32±38. of Prehistory, Research School of Paci®c Studies, Bednarik, R. G. (1995a) The CoÃ a petroglyphs: an obituary Australian National University, Canberra. to the stylistic dating of Palaeolithic rock-art. Allen, J. (1994) Radiocarbon determinations, luminescence Antiquity 69, 877±883. dating and Australian archaeology. Antiquity 68, Bednarik, R. G. (1995b) The age of the CoÃ a valley petro- 339±343. glyphs in Portugal. Rock Art Research 12, 86±103. Allen, J. and Holdaway, S. (1995) The contamination of Bednarik, R. G. (1996) Only time will tell: a review of the Pleistocene radiocarbon determinations in Australia. methodology of direct rock art dating. Archaeometry Antiquity 69, 101±112. 38, 1±13. Allen, J., Gosden, C., Jones, R. and White, J. P. (1988) Bell, W. T. (1991) Thermoluminescence dates for the Lake Pleistocene dates for the human occupation of New Mungo Aboriginal ®replaces and the implications for Ireland, northern Melanesia. Nature 331, 707±709. radiocarbon dating. Archaeometry 33, 43±50. Anderson, A. (1991) The chronology of colonization in Bellomo, R. V. (1993) A methodological approach for New Zealand. Antiquity 65, 767±795. identifying archaeological evidence of ®re resulting Anderson, A. (1994) Palaeoenvironmental evidence of from human activities. Journal of Archaeological island colonization: a response. Antiquity 68, 845± Science 20, 525±553. 847. Bellwood, P. (1991) The Austronesian dispersal and the Anderson, A. (1996) Was Rattus exulans in New Zealand origin of languages. Scienti®c American 265, 70±75. 2000 years ago? AMS radiocarbon ages from Shag BeltraÄ o, M. C., Danon, J., Enriquez, C. R., Poupeau, River Mouth. Archaeology in Oceania 31, 178±184. G. and Zuleta, E. (1982) On the arrival of men in Bacci, C., Bernardini, P., Furetta, C., Rispoli, B., Sanipoli, America: dating by thermoluminescence of heated C., Scacco, A., Carafa, P., Carandini, A. and cherts from archaeological site Alice Boer (Brazil). Esposito, A. (1990) The Via Nova±Via Sacra archaic Comptes Rendus de L'AcadeÂmie des Sciences, Paris site on Palatino Hill: ®rst thermoluminescence 295(SeÂ rie II), 629±632. archaeometric results. Radiation Protection Dosimetry BenkoÈ , L., HorvaÂ th, F., Horvatincic, N. and Obelic, 34, 61±64. B. (1989) Radiocarbon and thermoluminescence dat- Bahn, P. G. (1993) 50,000 year-old Americans of Pedra ing of prehistoric sites in Hungary and Yugoslavia. Furada. Nature 362, 114±115. Radiocarbon 31, 992±1002. Bahn, P. G. (1996) Further back down under. Nature 383, Berger, G. W. (1988) Dating Quaternary events by lumi- 577±578. nescence. In Dating Quaternary Sediments, ed. D. Baili, I. K. (1994) The pre-dose technique. Radiation J. Easterbrook, Special Paper 227, pp. 13±50. Measurements 23, 471±479. Geological Society of America, Boulder. Balter, M. (1995) Did Homo erectus tame ®re ®rst? Science Berger, G. W. (1990) Eectiveness of natural zeroing of 268, 1570. the thermoluminescence in sediments. Journal of Barbetti, M. and Allen, H. (1972) Prehistoric man at Lake Geophysical Research 95(B8), 12375±12397. Mungo, Australia, by 32,000 years BP. Nature 240, Berger, G. W. (1991) The use of glass for dating volcanic 46±48. ash by thermoluminescence. Journal of Geophysical Barbetti, M. and McElhinny, M. (1972) Evidence of a geo- Research 96(B12), 19705±19720. magnetic excursion 30,000 yr BP. Nature 239, 327± Berger, G. W. (1992) Dating volcanic ash by use of ther- 330. moluminescence. Geology 20, 11±14. Bard, E., Hamelin, B., Fairbanks, R. G. and Zindler, Berger, G. W. (1994) Thermoluminescence dating of sedi- A. (1990) Calibration of the 14C timescale over the ments older than 1100 ka. Quaternary Science past 30,000 years using mass spectrometric U-Th Reviews (Quaternary Geochronology) 13, 445±455. ages from Barbados corals. Nature 345, 405±410. Berger, G. W. (1995) Progress in luminescence dating Bard, E., Arnold, M., Fairbanks, R. G. and Hamelin, methods for Quaternary sediments. In Dating B. (1993) 230Th-234U and 14C ages obtained by mass Methods for Quaternary Deposits, eds. N. W. Rutter spectrometry on corals. Radiocarbon 35, 191±199. and N. R. Catto, GEOtext 2, pp. 81±104. Geological Barton, J. C. and Stringer, C. B. (1997) An attempt at dat- Association of Canada, St. John's. ing the Swanscombe skull bones using non-destruc- Berger, G. W. (1996) Limiting age of the Miami mastodon tive gamma-ray counting. Archaeometry 39, 205±216. by thermoluminescence. Current Research in the Barton, R. N. E. and Huxtable, J. (1983) New dates for Pleistocene 13, 77±78. Hengistbury Head, Dorset. Antiquity 57, 133±135. Berger, G. W. and Huntley, D. J. (1986) San Diego Bartstra, G.-J. (1988) The age of Solo Man and Java research and chronology revisited. Current Research Man. In Early Man in the Southern Hemisphere, ed. in the Pleistocene 3, 39±40. J. R. Prescott, pp. S12±S22. Supplement to Berger, G. W. and Huntley, D. J. (1994) Tests for opti- Archaeometry: Australasian Studies 1988, Department cally stimulated luminescence from tephra glass. of Physics and Mathematical Physics, University of Quaternary Science Reviews (Quaternary Adelaide, Adelaide. Geochronology) 13, 509±511. Bartstra, G.-J., Soegondho, S. and van der Wijk, A. (1988) Berger, G. W., Pillans, B. J. and Palmer, A. S. (1992) Ngandong man: age and artifacts. Journal of Human Dating loess up to 800 ka by thermoluminescence. Evolution 17, 325±337. Geology 20, 403±406. Bar-Yosef, O. (1989) Geochronology of the Levantine BermuÂ dez de Castro, J. M., Arsuaga, J. L., Carbonell, E., Middle Palaeolithic. In The Human Revolution: Rosas, A., MartõÂ nez, I. and Mosquera, M. (1997) A Behavioural and Biological Perspectives on the Origins hominid from the lower Pleistocene of Atapuerca, 878 R. G. ROBERTS

Spain: possible ancestor to Neandertals and modern pp. 127±138. Australian Institute of Aboriginal humans. Science 276, 1392±1395. Studies, Canberra. Berzero, A., Caramella-Crespi, V. and Cavagna, P. (1997) Bowler, J. M. and Wasson, R. J. (1984) Glacial age en- Direct gamma-ray spectrometric dating of fossil vironments of inland Australia. In Late Cainozoic bones: preliminary results. Archaeometry 39, 189± Palaeoclimates of the Southern Hemisphere, ed. J. C. 203. Vogel, pp. 183±208. A. A. Balkema, Rotterdam. Bird, M. I. and GroÈ cke, D. R. (1997) Determination of Bowler, J. M., Jones, R., Allen, H. and Thorne, A. the abundance and carbon-isotope composition of el- G. (1970) Pleistocene human remains from Australia: emental carbon in sediments. Geochimica et a living site and human cremation from Lake Cosmochimica Acta 61, 3413±3423. Mungo, western New South Wales. World Bischo, J. L., Julia, R. and Mora, R. (1988) Uranium- Archaeology 2, 39±60. series dating of the Mousterian occupation at Abric Bowler, J. M., Thorne, A. G. and Polach, H. A. (1972) RomanõÂ , Spain. Nature 332, 68±70. Pleistocene man in Australia: age and signi®cance of Bischo, J. L., Soler, N., Maroto, J. and JuliaÁ , R. (1989) the Mungo skeleton. Nature 240, 48±50. Abrupt Mousterian/Aurignacian boundary at c. Bowman, S. G. E. (1982) Thermoluminescence studies on 14 40 ka bp: accelerator C dates from L'Arbreda Cave burnt ¯int. Journal of the Council of Europe PACT 6, (Catalunya, Spain). Journal of Archaeological Science 353±361. 16, 563±576. Bowman, S. G. E. (1985) Thermoluminescence character- Bischo, J. L., Ludwig, K., Garcia, J. F., Carbonell, E., istics of sediments from the Tabun cave, Israel. Vaquero, M., Staord, T. W. Jr. and Jull, A. Nuclear Tracks and Radiation Measurements 10, 731± J. T. (1994) Dating of the basal Aurignacian sand- 736. wich at Abric RomanõÂ (Catalunya, Spain) by radio- Bowman, S. G. E. (1993) Thermoluminescence dating of carbon and uranium-series. Journal of Archaeological the Lower Industry. In The Lower Paleolithic Site at Science 21, 541±551. Hoxne, England, eds. R. Singer, B. G. Gladfelter and Black, D., de Chardin, T., Young, C. C. and Pei, W. C. J. J. Wymer, pp. 208±210. University of Chicago (1933) Fossil Man in China: the Choukoutien Cave Press, Chicago. Deposits with a Synopsis of our Present Knowledge Bowman, S. G. E. and Sieveking, G. de G. (1983) of the Late Cenozoic in China. Geological Memoirs Thermoluminescence dating of burnt ¯int from Series A, No. 11, Geological Survey of China, Combe Grenal. Journal of the Council of Europe Peiping. PACT, 9, 253±268. Bluszcz, A. (1993) Thermoluminescence dating of deposits Bowman, S. G. E., Loosemore, R. P. W., Sieveking, G. de from the area of Bir Sahara East and Bir Tarfawi. In G. and Bordes, F. (1982) Preliminary dates for Pech Egypt during the Last Interglacial: the Middle de l'Aze IV. Journal of the Council of Europe PACT, Paleolithic of Bir Tarfawi and Bir Sahara East, eds. 6, 362±369. F. Wendorf, R. Schild, A. E. Close and Associates, Braithwaite, R. W. and Estbergs, J., A. (1985) Fire pat- pp. 224±226. Plenum Press, New York. terns and woody vegetation trends in the Alligator BoeÈ da, E., Connan, J., Dessort, D., Muhesen, S., Mercier, Rivers region of northern Australia. In Ecology and N., Valladas, H. and TisneÂ rat, N. (1996) Bitumen as Management of the World's Savannas, eds. J. C. a hafting material on Middle Palaeolithic artefacts. Tothill and J. J. Mott, pp. 359±364. Australian Nature 380, 336±338. Academy of Science, Canberra. Bogaard, P. v. d., Hall, C. M., Schmincke, H.-U. and Brauer, G. (1989) The evolution of modern humans: a York, D. (1987) 40Ar/39Ar laser dating of single È grains: ages of Quaternary tephra from the East Eifel comparison of the African and non-African evidence. volcanic ®eld, FRG. Geophysical Research Letters 14, In The Human Revolution: Behavioural and Biological 1211±1214. Perspectives on the Origins of Modern Humans, eds. Bùtter-Jensen, L. and Duller, G. A. T. (1992) A new sys- P. Mellars and C. Stringer, pp. 123±154. Edinburgh tem for measuring optically stimulated luminescence University Press, Edinburgh. from quartz samples. Nuclear Tracks and Radiation Bray, W. (1986) Finding the earliest Americans. Nature Measurements 20, 549±553. 321, 726. Bowall, L., Frean, R., McKeogh, K. J., Rendell, H. Bray, W. (1988) The Palaeoindian debate. Nature 332, M. and Townsend, P. D. (1987) Sensitivity changes 107. in the TL of quartz and feldspar after bleaching. Bronson, B. and Han, M. (1972) A thermoluminescence Crystal Lattice Defects and Amorphous Materials 16, series from Thailand. Antiquity 46, 322±326. 37±43. Brooks, A. S. and Wood, B. (1990) The Chinese side of Bowdler, S. (1977) The coastal colonisation of Australia. the story. Nature 344, 288±289. In Sunda and Sahul: Prehistoric Studies in Southeast Brooks, A. S., Helgren, D. M., Cramer, J. S., Franklin, Asia, Melanesia and Australia, eds. J. Allen, J. Golson A., Hornyak, W., Keating, J. M., Klein, R. G., and R. Jones, pp. 205±246. Academic Press, London. Rink, W. J., Schwarcz, H., Leith Smith, J. N., Bowler, J. M. (1971) Pleistocene salinities and climatic Stewart, K., Todd, N. E., Verniers, J. and Yellen, change: evidence from lakes and lunettes in south- J. E. (1995) Dating and context of three Middle eastern Australia. In Aboriginal Man and Stone Age sites with bone points in the Upper Environment in Australia, eds. D. J. Mulvaney and Semliki Valley, Zaire. Science 268, 548±553. J. Golson, pp. 47±65. Australian National University Brown, P. (1992) Recent human evolution in East Asia Press, Canberra. and Australasia. Philosophical Transactions of the Bowler, J. M. (1976) Recent developments in reconstruct- Royal Society of London 337(Series B), 235±242. ing late Quaternary environments in Australia. In Brown, T. A., Nelson, D. E., Mathewes, R. W., Vogel, The Origin of the Australians, eds. R. L. Kirk and A. J. S. and Southon, J. R. (1989) Radiocarbon dating G. Thorne, Human Biology Series 6, pp. 55±77. of pollen by accelerator mass spectrometry. Australian Institute of Aboriginal Studies, Canberra. Quaternary Research 32, 205±212. Bowler, J. M. and Thorne, A. G. (1976) Human remains Brumby, S. and Yoshida, H. (1994) An investigation of from Lake Mungo: discovery and excavation of Lake the eect of sunlight on the ESR spectra of quartz Mungo III. In The Origin of the Australians, eds. R. centres: implications for dating. Quaternary Science L. Kirk and A. G. Thorne, Human Biology Series 6, Reviews (Quaternary Geochronology) 13, 615±618. LUMINESCENCE DATING IN ARCHAEOLOGY 879

Bryan, A. L., Casamiquela, R. M., Cruxent, J. M., H. and Reagan, M. (1996) Dated co-occurrence of Gruhn, R. and Ochsenius, C. (1978) An El Jobo Homo erectus and Gigantopithecus from Tham mastodon kill at Taima-taima, Venezuela. Science Khuyen Cave, Vietnam. Proceedings of the National 200, 1275±1277. Academy of Sciences (USA) 93, 3016±3020. Buschbeck, H. M. and Flettner, S. (1994) TL dating of a Clottes, J., Chauvet, J.-M., Brunel-Deschamps, E., destructive ®re in a Neolithic settlement. Quaternary Hillaire, C., Daugas, J.-P., Arnold, M., Cachier, H., Science Reviews (Quaternary Geochronology) 13, Evin, J., Fortin, P., Oberlin, C., Tisnerat, N. and 569±573. Valladas, H. (1995) The palaeolithic paintings of the Cabrera Valdes, V. and Bischo, J. L. (1989) Accelerator Chauvet-Pont-d'Arc Cave, at Vallon-Pont-d'Arc 14C dates for early Upper Paleolithic (basal (ArdeÁ che, France): direct and indirect radiocarbon Aurignacian) at El Castillo Cave (Spain). Journal of datings. Comptes Rendus de L'AcadeÂmie des Sciences, Archaeological Science 16, 577±584. Paris 320(SeÂ rie IIa), 1133±1140. Cahen, D. and Moeyersons, J. (1977) Subsurface move- Colinvaux, P. A., de Oliveira, P. E., Moreno, J. E., Miller, ments of stone artefacts and their implications for M. C. and Bush, M. B. (1996) A long pollen record the prehistory of Central Africa. Nature 266, 812± from lowland Amazonia: forest and cooling in glacial 815. times. Science 274, 85±88. Cann, R. L., Stoneking, M. and Wilson, A. C. (1987) Colman, S. M., Pierce, K. L. and Birkeland, P. W. (1987) Mitochondrial DNA and human evolution. Nature Suggested terminology for Quaternary dating 325, 31±36. methods. Quaternary Research 28, 314±319. Carbonell, E., BermuÂ dez de Castro, J. M., Arsuaga, J. L., Daniels, F., Boyd, C. A. and Saunders, D. F. (1953) DõÂ ez, J. C., Rosas, A., Cuenca-BescoÂ s, G., Sala, R., Thermoluminescence as a research tool. Science 117, Mosquera, M. and RodrõÂ guez, X. P. (1995) Lower 343±349. Pleistocene hominids and artifacts from Atapuerca- Danon, J., Enriquez, C. R., Zuleta, E., BeltraÄ o, M. M. TD6 (Spain). Science 269, 826±830. C. and Poupeau, G. (1982) Thermoluminescence dat- Carriveau, G. W. and Harbottle, G. (1983) Ban Chiang ing of archaeologically heated cherts. Journal of the pottery: thermoluminescence dating problems. Council of Europe PACT 6, 370±383. Antiquity 57, 56±58. Darwin, C. (1871) The Descent of Man, and Selection in Carriveau, G. W. and Nievens, M. (1979) Guatemalan Relation to Sex. John Murray, London. obsidian source characterization by thermolumines- David, B. (1992) An AMS date for north Queensland rock cence. Journal of the Council of Europe PACT 3, art. Rock Art Research 9, 139±141. 506±509. David, B. (1993) Nurrabullgin Cave: preliminary results Chaloupka, G. (1993) Journey in Time: the World's from a pre-37,000 year old rockshelter, North Longest Continuing Art Tradition. Reed, Sydney. Queensland. Archaeology in Oceania 28, 50±54. Chapman, C. H. (1975) The Archaeology of Missouri, I. David, B., Roberts, R., Tuniz, C., Jones, R. and Head, University of Missouri Press, Columbia. J. (1997) New optical and radiocarbon dates from Chappell, J. (1974) Geology of coral terraces, Huon Ngarrabullgan Cave, a Pleistocene archaeological site Peninsula, New Guinea: a study of Quaternary tec- in Australia: implications for the comparability of tonic movements and sea-level changes. Geological time clocks and for the human colonisation of Society of America Bulletin 85, 553±570. Australia. Antiquity 71, 183±188. Chappell, J., Head, J. and Magee, J. (1996a) Beyond the Davidson, J., Hinds, E., Holdaway, S. and Leach, radiocarbon limit in Australian archaeology and F. (1990) The Lapita site of Natunuku, Fiji. New Quaternary research. Antiquity 70, 543±552. Chappell, J., Omura, A., Esat, T., McCulloch, M., Zealand Journal of Archaeology 23, 121±155. Pandol®, J., Ota, Y. and Pillans, B. (1996b) de Lumley, H., de Lumley, M.-A., Beltrao, M., Reconciliation of late Quaternary sea levels derived Yokoyama, Y., Labeyrie, J., Danon, J., Delibrias, from coral terraces at Huon Peninsula with deep sea G., Falgueres, C. and Bischo, J. L. (1988) oxygen isotope records. Earth and Planetary Science Discovery of ¯aked tools associated with Middle Letters 141, 227±236. Pleistocene fauna in the Toca da Esperanc a, Bahia Chawla, S. and Singhvi, A. K. (1989) Thermoluminescence State, Brazil. Comptes Rendus de L'AcadeÂmie des dating of archaeological sediments. Naturwissen- Sciences, Paris 306(SeÂ rie II), 241±247. schaften 76, 416±418. de Vos, J. and Sondaar, P. (1994) Dating hominid sites in Chawla, S., Dhir, R. P. and Singhvi, A. K. (1992) Indonesia. Science 266, 1726±1727. Thermoluminescence chronology of sand pro®les in Debenham, N. C. (1983) Reliability of thermolumines- the Thar desert and their implications. Quaternary cence dating of stalagmitic calcite. Nature 304, 154± Science Reviews 11, 25±32. 156. Chen, T. and Yuan, S. (1988) Uranium-series dating of Debenham, N. C. (1985) Use of U.V. emissions in TL dat- bones and teeth from Chinese palaeolithic sites. ing of sediments. Nuclear Tracks and Radiation Archaeometry 30, 59±76. Measurements 10, 717±724. Chen, T. and Zhang, Y. (1991) Palaeolithic chronology Debenham, N. C. and Aitken, M. J. (1984) and possible coexistence of Homo erectus and Homo Thermoluminescence dating of stalagmitic calcite. sapiens in China. World Archaeology 23, 147±154. Archaeometry 26, 155±170. Chen, T., Yang, Q. and Wu, E. (1994) Antiquity of Homo Delibrias, G., Guidon, N. and Parenti, F. (1988) The Toca sapiens in China. Nature 368, 55±56. do BoqueiraÄ o do Sitio da Pedra Furada: stratigraphy Chen, T.-M., Yang, Q., Hu, Y.-Q., Bao, W.-B. and Li, and chronology. In Early Man in the Southern T.-Y. (1997) ESR dating of tooth enamel from Hemisphere, ed. J. R. Prescott, pp. S3±S11. Yunxian Homo erectus site, China. Quaternary Supplement to Archaeometry: Australasian Studies Science Reviews (Quaternary Geochronology) 16, 1988, Department of Physics and Mathematical 455±458. Physics, University of Adelaide, Adelaide. Christodoulides, C. and Fremlin, J. H. (1971) Dennell, R. (1997) The world's oldest spears. Nature 385, Thermoluminescence of biological materials. Nature 767±768. 232, 257±258. Dennell, R. and Roebroeks, W. (1996) The earliest coloni- Ciochon, R., Long, V. T., Larick, R., GonzaÂ lez, L., GruÈ n, zation of Europe: the short chronology revisited. R., de Vos, J., Yonge, C., Taylor, L., Yoshida, Antiquity 70, 535±542. 880 R. G. ROBERTS

Dennell, R. W., Rendell, H. M. and Hailwood, E. (1988) Duller, G. A. T. (1995) Luminescence dating using single Late Pliocene artefacts from northern Pakistan. aliquots: methods and applications. Radiation Current Anthropology 29, 495±498. Measurements 24, 217±226. Desmond, A. and Moore, J. (1991) Darwin. Penguin Duller, G. A. T. (1996) Recent developments in lumines- Books, Harmondsworth. cence dating of Quaternary sediments. Progress in Diamond, J. M. (1988) Express train to Polynesia. Nature Physical Geography 20, 127±145. 336, 307±308. Duller, G. A. T., Wintle, A. G. and Hall, A. M. (1995) Diamond, J. M. (1989) Quaternary megafaunal extinc- Luminescence dating and its application to key pre- tions: variations on a theme by Paganini. Journal of Late Devensian sites in Scotland. Quaternary Science Archaeological Science 16, 167±175. Reviews 14, 495±519. Diamond, J. M. (1991) Twilight of Hawaiian birds. Dunnell, R. C. and Feathers, J. K. (1994) Nature 353, 505±506. Thermoluminescence dating of sur®cial archaeologi- Dillehay, T. D. and Collins, M. B. (1988) Early cultural cal material. In Dating in Exposed and Surface evidence from Monte Verde in Chile. Nature 332, Contexts, ed. C. Beck, pp. 115±137. University of 150±152. New Mexico Press, Albuquerque. Dillehay, T. D., CalderoÂ n, G. A., Politis, G. and BeltraÄ o, Dunnell, R. C. and Hamilton, T. M. (1995) Age of the M. C. M. C. (1992) Earliest hunters and gatherers of Miami mastodon. Current Research in the Pleistocene South America. Journal of World Prehistory 6, 145± 12, 91±92. 204. Dunnell, R. C. and Readhead, M. L. (1988) The relation Dodson, J. R. (1989) Late Pleistocene vegetation and en- of dating and chronology: comments on Chatters vironmental shifts in Australia and their bearing on and Hoover (1986) and Butler and Stein (1988). faunal extinctions. Journal of Archaeological Science Quaternary Research 30, 232±233. 16, 207±217. Duverger, C., LeÂ vine, D., Ney, C., Guibert, P., Schvoerer, Dodson, J., Fullagar, R., Furby, J., Jones, R. and Prosser, M. and Mook, W. G. (1993) A chronological I. (1993) Humans and megafauna in a late approach to the site of Coamiles (Nayarit, Mexico): Pleistocene environment from Cuddie Springs, north dating by thermoluminescence and radiocarbon. western New South Wales. Archaeology in Oceania Journal de la SocieÂteÂ des AmeÂricanistes 79, 105±139. 28, 94±99. Elias, S. A., Short, S. K., Nelson, C. H. and Birks, H. Dolitsky, A. B. (1985) Siberian Paleolithic archaeology: H. (1996) Life and times of the Bering land bridge. approaches and analytic methods. Current Nature 382, 60±63. Anthropology 26, 361±378. Elitzsch, C., Pernicka, E. and Wagner, G. A. (1983) Dorn, R. I. (1997a) Constraining the age of the CoÃ a valley Thermoluminescence dating of archaeometallurgical (Portugal) engravings with radiocarbon dating. slags. Journal of the Council of Europe PACT 9, 271± Antiquity 71, 105±115. 286. Dorn, R. I. (1997b) Uncertainties in 14C ages for petro- Erlach, R. and Vana, N. (1988) TL-dating of archaeologi- glyphs from the Olary province, South Australia. cal samples from Carnuntum. Nuclear Tracks and Archaeology in Oceania 32, 214±215. Radiation Measurements 14, 295±298. Dorn, R. I., Nobbs, M. and Cahill, T. A. (1988) Cation- Etheridge, R. Jr. (1890) Has man a geological history in ratio dating of rock-engravings from the Olary Australia. Proceedings of the Linnean Society of New Province of arid South Australia. Antiquity 62, 681± South Wales 15, 259±266. 689. Etler, D. A. (1996) The fossil evidence for human evol- Dorn, R. I., Jull, A. J. T., Donahue, D. J., Linick, T. ution in Asia. Annual Review of Anthropology 25, W. and Toolin, L. J. (1989) Accelerator mass spec- 275±301. trometry radiocarbon dating of rock varnish. FalgueÁ res, C., Yokoyama, Y. and Miallier, D. (1991) Geological Society of America Bulletin 101, 1363± Stability of some centres in quartz. Nuclear Tracks 1372. and Radiation Measurements 18, 155±161. Dortch, C. (1979) Devil's Lair, an example of prolonged FalgueÁ res, C., Miallier, D., Sanzelle, S., FaõÈ n, J., Laurent, cave use in south-western Australia. World M., Montret, M., Pilleyre, T. and Bahain, J. J. (1994a) Archaeology 10, 258±279. Potential use of the E' center as an indicator of initial Dortch, C. E. and Dortch, J. (1996) Review of Devil's resetting in TL/ESR dating of volcanic materials. Lair artefact classi®cation and radiocarbon chronol- Quaternary Science Reviews (Quaternary Geo- ogy. Australian Archaeology 43, 28±32. chronology) 13, 619±623. Dortch, C. E. and Muir, B. G. (1980) Long range sight- FalgueÁ res, C., Fontugne, M., Chauchat, C. and Guadelli, ings of bush ®res as a possible incentive for J.-L. (1994b) Radiometric dating of the extinction of Pleistocene voyages to Greater Australia. The the large Pleistocene fauna in Peru. Comptes Rendus Western Australian Naturalist 14, 194±198. de L'AcadeÂmie des Sciences, Paris 319(SeÂ rie II), 261± Downey, B. and Frankel, D. (1992) Radiocarbon and 266. thermoluminescence dating of a central Murray Fang, Y. (1994) The pebble tool or chopper/chopping tool mound. The Artefact 15, 31±35. industry in China. Human Evolution 9, 263±272. Driver, H. S. T. (1979) The preparation of thin slices of Fang, Y., Yang, D., Han, H. and Zhou, L. (1992) bone and shell for thermoluminescence. Journal of Taphonomic study of paleolithic site group of the Council of Europe PACT 3, 290±297. Shuiyang River. Acta Anthropologica Sinica 11, 134± Dubois, E. (1894) Pithecanthropus erectus: eine 142. Menschenaehnliche Uebergangsform aus Java. Fankhauser, B. (1990) Dating earth ovens with thermolu- Landesdruckerei, Batavia. minescence and radiocarbon. In Quaternary Dating Duller, G. A. T. (1991) Equivalent dose determination Workshop 1990, ed. R. Gillespie, pp. 41±44. using single aliquots. Nuclear Tracks and Radiation Department of Biogeography and Geomorphology, Measurements 18, 371±378. Research School of Paci®c Studies, Australian Duller, G. A. T. (1994) Luminescence dating of sediments National University, Canberra. using single aliquots: new procedures. Quaternary Feathers, J. K. (1997a) The application of luminescence Science Reviews (Quaternary Geochronology) 13, dating in American archaeology. Journal of 149±156. Archaeological Method and Theory 4, 1±66. LUMINESCENCE DATING IN ARCHAEOLOGY 881

Feathers, J. K. (1997b) Luminescence dating of early Furby, J. H., Fullagar, R., Dodson, J. R. and Prosser, I. mounds in northeast Louisiana. Quaternary Science (1993) The Cuddie Springs bone bed revisited, 1991. Reviews (Quaternary Geochronology) 16, 333±340. In Sahul in Review: Pleistocene Archaeology in Feathers, J. K. (1997c) Luminescence dating of sediment Australia, New Guinea and Island Melanesia, eds. M. samples from White Paintings Rockshelter, A. Smith, M. Spriggs and B. Fankhauser, Occasional Botswana. Quaternary Science Reviews (Quaternary Papers in Prehistory 24, pp. 204±210. Department of Geochronology) 16, 321±331. Prehistory, Research School of Paci®c Studies, Feathers, J. K. (1997d) Luminescence applications in Australian National University, Canberra. American archaeology. Presentation at the 62nd Gabunia, L. and Vekua, A. (1995) A Plio-Pleistocene Annual Convention of the Society for American hominid from Dmanisi, East Georgia, Caucasus. Archaeology (Nashville, April 1997). Nature 373, 509±512. Feathers, J. K. and Rhode, D. (1997) Luminescence dating Galloway, R. B. (1996) Equivalent dose determination of protohistoric pottery from the Great Basin. using only one sample: alternative analysis of data Geoarchaeology (in press). obtained from infrared stimulation of feldspars. Fiedel, S. (1996) Paleoindians in the Brazilian Amazon. Radiation Measurements 26, 103±106. Science 274, 1821±1822. Gamble, C. (1986) The Palaeolithic Settlement of Europe. Fischman, J. (1995) Portuguese rock art gets younger. Cambridge University Press, Cambridge. Science 269, 304. Gibbons, A. (1993) Geneticists trace the DNA trail of the Flannery, T. F. (1990) Pleistocene faunal loss: implications ®rst Americans. Science 259, 312±313. of the aftershock for Australia's past and future. Gibbons, A. (1994a) Genes point to a new identity for Archaeology in Oceania 25, 45±67. Paci®c pioneers. Science 263, 32±33. Flannery, T. F. (1994) The Future Eaters: an Ecological Gibbons, A. (1994b) RewritingÐand redatingÐprehis- History of the Australasian Lands and People. Reed, tory. Science 263, 1087±1088. Sydney. Gibbons, A. (1996a) First Americans: not mammoth hun- Flannery, T. F., Kirch, P. V., Specht, J. and Spriggs, ters, but forest dwellers? Science 272, 346±347. M. (1988) Holocene mammal faunas from archaeolo- Gibbons, A. (1996b) The peopling of the Americas. gical sites in island Melanesia. Archaeology in Science 274, 31±33. Oceania 23, 89±94. Gibbons, A. (1996c) Homo erectus in Java: a 250,000-year Fleming, S. J. (1970) Thermoluminescent dating: re®ne- anachronism. Science 274, 1841±1842. ment of the quartz inclusion method. Archaeometry Gibbons, A. (1997a) MonteVerde: blessed but not con- 12, 133±145. ®rmed. Science 275, 1256±1257. Gibbons, A. (1997b) A new face for human ancestors. Fleming, S. J. (1973a) The pre-dose technique: a new ther- Science 276, 1331±1333. moluminescent dating method. Archaeometry 15, 13± Gibbons, A. (1997c) Doubts over spectacular dates. 30. Science 278, 220±222. Fleming, S. J. (1973b) Thermoluminescence and glaze stu- Godfrey-Smith, D. I., Huntley, D. J. and Chen, W.- dies of a group of T'ang Dynasty ceramics. H. (1988) Optical dating studies of quartz and feld- Archaeometry 15, 31±52. spar sediment extracts. Quaternary Science Reviews 7, Fleming, S. J. (1979) Thermoluminescence Techniques in 373±380. Archaeology. Clarendon Press, Oxford. Godfrey-Smith, D. I., Deal, M. and Kunelius, I. (1997) Fleming, S. J. and Sampson, E. H. (1972) The authenticity Thermoluminescence dating of St. Croix ceramics: of ®gurines, animals and pottery facsimiles of chronology building in southwestern Nova Scotia. bronzes in the Hui Hsien style. Archaeometry 14, Geoarchaeology 12, 251±273. 237±244. Goede, A. (1989) Electron spin resonanceÐa relative dat- Fleming, S. J. and Stoneham, D. (1973) The subtraction ing technique for Quaternary sediments near technique of thermoluminescent dating. Warrnambool, Victoria. Australian Geographical Archaeometry 15, 229±238. Studies 27, 14±30. Flood, J. (1997) Rock Art of the Dreamtime: Images of GoÈ ksu, H. Y. and TuÈ retken, N. (1979) Source identi®- Ancient Australia. Angus and Robertson, Sydney. cation of obsidian tools by thermoluminescence. Forman, S. L. (1989) Applications and limitations of ther- Journal of the Council of Europe PACT 3, 356±359. moluminescence to date Quaternary sediments. GoÈ ksu, H. Y., Fremlin, J. H., Irwin, H. T. and Fryxell, Quaternary International 1, 47±59. R. (1974) Age determination of burned ¯int by a Fragoulis, D. V. and Readhead, M. L. (1991) Feldspar in- thermoluminescent method. Science 183, 651±654. clusions and the anomalous fading and enhancement GoÈ ksu OÈ gelman, H. Y. (1986) Thermoluminescence dat- of thermoluminescence in quartz grains. Nuclear ing: a review of application to burnt ¯ints. In The Tracks and Radiation Measurements 18, 291±296. Scienti®c Study of Flint and Chert, eds. G. de G. Franklin, A. D. and Hornyak, W. F. (1990) Isolation of Sieveking and M. B. Hart, pp. 263±267. Cambridge the rapidly bleaching peak in quartz TL glow curves. University Press, Cambridge. Ancient TL 8, 29±31. Gorecki, P. P., Horton, D. R., Stern, N. and Wright, R. Frayer, D. W., Wolpo, M. H., Thorne, A. G., Smith, F. V. S. (1984) Coexistence of humans and megafauna H. and Pope, G. G. (1993) Theories of modern in Australia: improved strati®ed evidence. human origins: the paleontological test. American Archaeology in Oceania 19, 117±119. Anthropologist 95, 14±50. Gosden, C. (1995) Arboriculture and agriculture in coastal Fullagar, R. L. K., Price, D. M. and Head, L. M. (1996) Papua New Guinea. Antiquity 69, 807±817. Early human occupation of northern Australia: Gosden, C., Allen, J., Ambrose, W., Anson, D., Golson, archaeology and thermoluminescence dating of J., Green, R., Kirch, P., Lilley, I., Specht, J. and Jinmium rock-shelter, Northern Territory. Antiquity Spriggs, M. (1989) Lapita sites of the Bismarck 70, 751±773. Archipelago. Antiquity 63, 561±586. Fuller, I. C., Wintle, A. G. and Duller, G. A. T. (1994) Grayson, D. K. (1989) The chronology of North Test of the partial bleach methodology as applied to American Late Pleistocene extinctions. Journal of the infra-red stimulated luminescence of an alluvial Archaeological Science 16, 153±165. sediment from the Danube. Quaternary Science Grayson, D. K. (1991) Late Pleistocene mammalian Reviews (Quaternary Geochronology) 13, 539±543. extinctions in North America: taxonomy, chronology, 882 R. G. ROBERTS

and explanations. Journal of World Prehistory 5, Hedges, R. E. M. and McLellan, M. (1976) On the cation 193±231. exchange capacity of ®red clays and its eect on the Green, H. S. (ed.) (1984) Pontnewydd Cave, a Lower chemical and radiometric analysis of pottery. Palaeolithic Hominid Site in Wales: the First Report. Archaeometry 18, 203±207. Quaternary Studies Monographs 1, National Museum Hedges, R. E. M., Housley, R. A., Bronk Ramsey, C. and of Wales, Cardi. van Klinken, G. J. (1994) Radiocarbon dates from Green, H. S., Stringer, C. B., Collcutt, S. N., Currant, A. the Oxford AMS system: Archaeometry datelist 18. P., Huxtable, J., Schwarcz, H. P., Debenham, N., Archaeometry 36, 337±374. Embleton, C., Bull, P., Molleson, T. I. and Bevins, Hoecker, J. F., Powers, W. R. and Goebel, T. (1993) R. E. (1981) Pontnewydd Cave in WalesÐa new The colonization of Beringia and the peopling of the Middle Pleistocene hominid site. Nature 294, 707± New World. Science 259, 46±53. 713. Holdaway, R. N. (1996) Arrival of rats in New Zealand. Groube, L., Chappell, J., Muke, J. and Price, D. (1986) A Nature 384, 225±226. 40,000 year-old human occupation site at Huon Holdaway, S. (1996) Tool hafting with a mastic. Nature Peninsula, Papua New Guinea. Nature 324, 453±455. 380, 288±289. Groves, C. P. (1994) The origin of modern humans. Holden, C. (1996) Art stirs uproar down under. Science Interdisciplinary Science Reviews 19, 23±34. 274, 33±34. Gruhn, R. (1988) Linguistic evidence in support of the Holden, C. (1997) Tooling around: dates show early coastal route of earliest entry into the New World. Siberian settlement. Science 275, 1268. Man 23, 77±100. Holliday, V. T. (1997) Paleoindian Geoarchaeology of the GruÈ n, R. and Stringer, C. B. (1991) Electron spin reson- Southern High Plains. University of Texas Press, ance dating and the evolution of modern humans. Austin. Archaeometry 33, 153±199. Hope, J. (1982) Late Cainozoic vertebrate faunas and the GruÈ n, R. and Thorne, A. (1997) Dating the Ngandong development of aridity in Australia. In Evolution of humans. Science 276, 1575. the Flora and Fauna of Arid Australia, eds. W. R. GruÈ n, R., Schwarcz, H. P. and Zymela, S. (1987) Electron Barker and P. J. M. Greenslade, pp. 85±100. Peacock spin resonance dating of tooth enamel. Canadian Publications, Adelaide. Journal of Earth Sciences 24, 1022±1037. Hope, J. H., Dare-Edwards, A. and McIntyre, M. GruÈ n, R., Beaumont, P. B. and Stringer, C. B. (1990) L. (1983) Middens and megafauna: stratigraphy and ESR dating evidence for early modern humans at dating of Lake Tandou lunette, western New South Border Cave in South Africa. Nature 344, 537±539. Wales. Archaeology in Oceania 18, 45±53. Horn, P., Muller-Sohnius, D., Storzer, D. and ZoÈ ller, GruÈ n, R., Stringer, C. B. and Schwarcz, H. P. (1991) ESR L. (1993) K±Ar, ®ssion-track, and thermolumines- dating of teeth from Garrod's Tabun cave collection. cence ages of Quaternary volcanic tus and their Journal of Human Evolution 20, 231±248. bearing on Acheulian artifacts from Bori, Kukdi val- GruÈ n, R., Brink, J. S., Spooner, N. A., Taylor, L., ley, Pune District, India. Zeitschrift der Deutschen Stringer, C. B., Franciscus, R. G. and Murray, A. Geologischen Gesellschaft 144, 326±329. S. (1996) Direct dating of Florisbad hominid. Nature Horton, D. R. (1984) Red kangaroos: last of the 382, 500±501. Australian megafauna. In Quaternary Extinctions: a Guibert, P., Ney, C., Bechtel, F., Schvoerer, M. and Geus, Prehistoric Revolution, eds. P. S. Martin and R. G. F. (1994a) TL and radiocarbon dating of Neolithic Klein, pp. 639±680. University of Arizona Press, sepultures from Sudan: intercomparison of results. Tucson. Radiation Measurements 23, 393±398. Hu, J. (1994) Infrared optical dating of organic-rich sedi- Guibert, P., Schvoerer, M., Etcheverry, M. P., Szepertyski, ments. M.Sc. thesis, Simon Fraser University, B. and Ney, C. (1994b) IXth millennium B.C. cer- Vancouver. amics from Niger: detection of a U-series disequili- Huang, P.-H. and GruÈ n, R. (1996) ESR dating and brium and TL dating. Quaternary Science Reviews chronological framework of Homo erectus sites in (Quaternary Geochronology) 13, 555±561. China. Abstracts of the 8th International Conference Guidon, N. and Arnaud, B. (1991) The chronology of the on Luminescence and Electron Spin Resonance Dating New World: two faces of one reality. World (Canberra, April 1996), pp. 167±168. Archaeology 23, 167±178. Huang, P.-H., Jin, S.-Z., Peng, Z.-C., Liang, R.-Y., Lu, Z.- Guidon, N. and Delibrias, G. (1986) Carbon-14 dates J., Wang, Z.-R., Chen, J.-B. and Yuan, Z.-X. (1993) point to man in the Americas 32,000 years ago. ESR dating of tooth enamel: comparison with U- Nature 321, 769±771. series, FT and TL dating at the Peking Man site. Guidon, N., Parenti, F., da Luz, M. de F., GueÂ rin, C. and Applied Radiation and Isotopes 44, 239±242. Faure, M. (1994) Le plus ancien peuplement de Huang, P.-H., Zheng, L.-Z., Quan, Y.-C., Liang, R.-Y., l'AmeÂ rique: le PaleÂ olithique du nordeste BreÂ silien. Xu, Y.-H., Fan, Y.-S. and Fan, D.-S. (1995) Bulletin de la SocieÂteÂ PreÂhistorique Franc aise, 91, Preliminary study on ESR dating of Hexian Man 246±250. and its fauna. Nuclear Techniques 18, 491±494. Guyodo, Y. and Valet, J.-P. (1996) Relative variations in Huang, W., Ciochon, R., Gu, Y., Larick, R., Fang, Q., geomagnetic intensity from sedimentary records: the Schwarcz, H., Yonge, C., de Vos, J. and Rink, past 200,000 years. Earth and Planetary Science W. (1995) Early Homo and associated artefacts from Letters 143, 23±36. Asia. Nature 378, 275±278. Hammer, M. F. (1995) A recent common ancestry for Hublin, J.-J., Spoor, F., Braun, M., Zonneveld, F. and human Y chromosomes. Nature 378, 376±378. Condemi, S. (1996) A late Neanderthal associated Haynes, C. V., Jr. (1993) Clovis-Folsom geochronology with Upper Palaeolithic artefacts. Nature 381, 224± and climatic change. In From Kostenki to Clovis: 226. Upper Paleolithic-Paleo-Indian Adaptations, eds. O. Hughes, P. J. and Lampert, R. J. (1977) Occupational dis- Soer and N. D. Praslov, pp. 219±236. Plenum turbance and types of archaeological deposit. Journal Press, New York. of Archaeological Science 4, 135±140. Haynes, C. V. Jr. and Agogino, G. A. (1966) Prehistoric Huntley, D. J. and Bailey, D. C. (1978) Obsidian source springs and geochronology of the Clovis site, New identi®cation by thermoluminescence. Archaeometry Mexico. American Antiquity 31, 812±821. 20, 159±170. LUMINESCENCE DATING IN ARCHAEOLOGY 883

Huntley, D. J. and Clague, J. J. (1996) Optical dating of from Senpukuji Cave. Journal of the Council of tsunami-laid sands. Quaternary Research 46, 127± Europe PACT 2, 174±179. 140. Ichikawa, Y., Nagatomo, T. and Hagihara, N. (1978) Huntley, D. J. and Lian, O. B. (1997) Determining when a Thermoluminescent dating of Jomon pattern pottery sediment was last exposed to sunlight by optical dat- from Taishaku valley. Archaeometry 20, 171±176. ing. In Holocene Climate and Environmental Changes Jasinska, M. and Niewiadomski, T. (1970) in the Palliser Triangle, Southern Canadian Prairies, Thermoluminescence of biological materials. Nature eds D. S. Lemmen and R. E. Vance. Geological 227, 1159±1160. Survey of Canada Bulletin (in press). Jeerson, T. (1787) Notes on the State of Virginia. John Huntley, D. J., Berger, G. W., Divigalpitiya, W. M. R. and Stockdale, London (reprinted by The University of Brown, T. A. (1983) Thermoluminescence dating of North Carolina Press, Chapel Hill, 1955). sediments. Journal of the Council of Europe PACT 9, Jelinek, A. J. (1982) The Tabun Cave and Paleolithic Man 607±618. in the Levant. Science 216, 1369±1375. Huntley, D. J., Godfrey-Smith, D. I. and Thewalt, M. L. Jia, L. and Huang, W. (1990) The Story of Peking Man: W. (1985) Optical dating of sediments. Nature 313, From Archaeology to Mystery. Oxford University 105±107. Press, Oxford. Huntley, D. J., Hutton, J. T. and Prescott, J. R. (1993a) Jones, R. (1968) The geographical background to the arri- The stranded beach-dune sequence of south-east val of man in Australia and Tasmania. Archaeology South Australia: a test of thermoluminescence dating, and Physical Anthropology in Oceania 3, 186±215. 0±800 ka. Quaternary Science Reviews 12, 1±20. Jones, R. (1982) Ions and eons: some thoughts on archae- Huntley, D. J., Hutton, J. T. and Prescott, J. R. (1993b) ological science and scienti®c archaeology. In Optical dating using inclusions within quartz grains. Archaeometry: an Australasian Perspective, eds. W. Geology 21, 1087±1090. Ambrose and P. Duerden, Occasional Papers in Huntley, D. J., Hutton, J. T. and Prescott, J. R. (1994a) Prehistory 12, pp. 22±35. Department of Prehistory, Further thermoluminescence dates from the dune Research School of Paci®c Studies, Australian sequence in the southeast of South Australia. National University, Canberra. Quaternary Science Reviews 13, 201±207. Jones, R. (1989) East of Wallace's line: issues and pro- Huntley, D. J., Lian, O. B., Hu, J. and Prescott, blems in the colonisation of the Australian continent. J. R. (1994b) Tests of luminescence dating making In The Human Revolution: Behavioural and Biological use of paleomagnetic reversals. Ancient TL 12, 28± Perspectives on the Origins of Modern Humans, eds. 30. P. Mellars and C. Stringer, pp. 743±782. Edinburgh Huxley, T. H. (1863) Evidence as to Man's Place in University Press, Edinburgh. Nature. Williams and Norgate, London. Jones, R. (1993) A continental reconnaissance: some ob- Huxtable, J. (1981) Light bleaching of archaeological ¯int servations concerning the discovery of the Pleistocene samples: a warning. Ancient TL 16, 2±4. archaeology of Australia. In A Community of Huxtable, J. (1982) Fine grain thermoluminescence (TL) Culture: the People and Prehistory of the Paci®c, eds. techniques applied to ¯int dating. Journal of the M. Spriggs, D. E. Yen, W. Ambrose, R. Jones, A. Council of Europe PACT 6, 346±352. Thorne and A. Andrews, Occasional Papers in Huxtable, J. (1993) Thermoluminescence dates for burnt Prehistory 21, pp. 97±122. Department of Prehistory, earth samples from Bir Sahara East and a burnt core Research School of Paci®c Studies, Australian from Bir Tarfawi. In Egypt during the Last National University, Canberra. Interglacial: the Middle Paleolithic of Bir Tarfawi and Bir Sahara East, eds. F. Wendorf, R. Schild, A. E. Jones, R. and Johnson, I. (1985) Deaf Adder Gorge: Close and Associates, pp. 227±228. Plenum Press, Lindner Site, Nauwalabila I. In Archaeological New York. Research in Kakadu National Park, ed. R. Jones, Huxtable, J. and Aitken, M. J. (1977) Thermoluminescent Special Publication 13, pp. 165±227. Australian dating of Lake Mungo geomagnetic polarity excur- National Parks and Wildlife Service, Canberra. sion. Nature 265, 40±41. Kamminga, J. and Allen, H. (1973) Alligator Rivers Huxtable, J. and Aitken, M. J. (1978) Environmental Fact-Finding Study: Report of the Thermoluminescence dating of sherds from Sham Archaeological Survey. Australian Government Wan. In Sham Wan, Lamma Island. An Publishing Sevice, Canberra. Archaeological Site Study, ed. W. Meacham, Journal Kaylor, R., Feathers, J., Gottfried, M., Hornyak, W. Monograph 3, pp. 116±124 and 291±292. Hong Kong F. and Franklin, A. D. (1993) Important date/strange Archaeological Society, Hong Kong. material.. Ancient TL 11, 40±44. Huxtable, J. and Aitken, M. J. (1986) Dating of European Kaylor, R. M., Feathers, J., Hornyak, W. F. and ¯int by thermoluminescence. In Proceedings of the Franklin, A. D. (1995) Optically stimulated lumines- 24th Archaeometry Symposium, eds. J. S. Olin and cence in Kalahari quartz: bleaching of the 3258C M. J. Blackman, pp. 465±472. Smithsonian peak as the source of the luminescence. Journal of Institution, Washington DC. Luminescence 65, 1±6. Huxtable, J. and Jacobi, R. M. (1982) Keith, A. (1927) A report on the Galilee skull. In Thermoluminescence dating of burned ¯ints from a Researches in Prehistoric Galilee: 1925±1926, ed. F. British Mesolithic site: Longmoor Inclosure, East Turville-Petre, pp. 53±106. British School of Hampshire. Archaeometry 24, 164±169. Archaeology in Jerusalem, London. Huxtable, J., Aitken, M. J., Hedges, J. W. and Renfrew, Kelly, E. F., Amundson, R. G., Marino, B. D. and A. C. (1976) Dating a settlement pattern by thermo- Deniro, M. J. (1991) Stable isotope ratios of carbon luminescence: the burnt mounds of Orkney. in phytoliths as a quantitative method of monitoring Archaeometry 18, 5±17. vegetation and climate change. Quaternary Research Huxtable, J., Murray, A. S. and Aitken, M. J. (1982-3) 35, 222±233. Thermoluminescence dating of sherds from Cheung Kennedy, K. A. R. and Deraniyagala, S. U. (1989) Fossil Chau. Journal of the Hong Kong Archaeological remains of 28,000 year-old hominids from Sri Lanka. Society, 10, 84±89. Current Anthropology 30, 394±399. Ichikawa, Y. and Nagatomo, T. (1978) Kershaw, A. P. (1994) Pleistocene vegetation of the humid Thermoluminescence dating of burnt sandstones tropics of northeastern Queensland, Australia. 884 R. G. ROBERTS

Palaeogeography, Palaeoclimatology, Palaeoecology thermoluminescence. Journal of the Royal Society of 109, 399±412. New Zealand 8, 331±342. Kershaw, A. P., McKenzie, G. M. and McMinn, A. (1993) Lewin, R. (1989) Bones of Contention: Controversies in the A Quaternary vegetation history of northeastern Search for Human Origins. Penguin Books, Queensland from pollen analysis of ODP Site 820. Harmondsworth. Proceedings of the Ocean Drilling Program. Scienti®c Li, S.-H. (1994) Optical dating: insuciently bleached Results 133, 107±114. sediments. Radiation Measurements 23, 563±567. King, M. L. and Slobodin, S. B. (1996) A ¯uted point Li, S.-H. and Wintle, A. G. (1994) Use of infra-red stimu- from the Uptar site, northeastern Siberia. Science lated luminescence signals from pellets of compressed 273, 634±636. loess from Rocourt, Belgium. Quaternary Science Kirch, P. V. and Ellison, J. (1994) Palaeoenvironmental Reviews (Quaternary Geochronology) 13, 529±532. evidence for human colonization of remote Oceanic Li, T. and Etler, D. A. (1992) New Middle Pleistocene islands. Antiquity 68, 310±321. hominid crania from Yunxian in China. Nature 357, Koch, J. M. and Bell, D. T. (1980) Leaf scorch in 404±407. Xanthorrhoea gracilis as an index of ®re intensity. Likens, G. E., Keene, W. C., Miller, J. M. and Galloway, Australian Forest Research 10, 113±119. J. N. (1987) Chemistry of precipitation from a Kojo, Y. (1991) The reliability of thermoluminescence dat- remote, terrestrial site in Australia. Journal of ing: a pilot experiment. Geoarchaeology 6, 367±374. Geophysical Research 92(D11), 13299±13314. Krings, M., Stone, A., Schmitz, R. W., Krainitzki, H., Liritzis, I. (1994) A new dating method by thermolumines- Stoneking, M. and PaÈ aÈ bo, S. (1997) Neandertal cence of carved megalithic stone building. Comptes DNA sequences and the origin of modern humans. Rendus de L'AcadeÂmie des Sciences, Paris 319(SeÂ rie Cell 90, 19±30. II), 603±610. Kunz, M. L. and Reanier, R. E. (1994) Paleoindians in Liritzis, Y. and Thomas, R. (1980) Palaeointensity and Beringia: evidence from arctic Alaska. Science 263, thermoluminescence measurements on Cretan kilns 660±662. from 1300 to 2000 BC. Nature 283, 54±55. Kuzmin, Y. V. and Tankersley, K. B. (1996) The coloniza- Liritzis, I., Galloway, R. B. and Theocaris, P. S. (1994) tion of eastern Siberia: an evaluation of the Thermoluminescence dating of ceramics revisited: Paleolithic Age radiocarbon dates. Journal of optical stimulated luminescence of quartz single ali- Archaeological Science 23, 577±585. quot with green light-emitting diodes. Journal of Laj, C., Mazaud, A. and Duplessy, J.-C. (1996) Radioanalytical and Nuclear Chemistry, Articles 188, Geomagnetic intensity and 14C abundance in the at- 189±198. mosphere and ocean during the past 50 kyr. Liritzis, I., Guibert, P., Foti, F. and Schvoerer, M. (1997) Geophysical Research Letters 23, 2045±2048. The temple of Apollo (Delphi) strengthens novel Lamothe, M. and Auclair, M. (1997) Assessing the dat- thermoluminescence dating method. Geoarchaeology ability of young sediments by IRSL using an intrinsic 12, 479±496. laboratory protocol. Radiation Measurements 27, Liritzis, I., Michael, C. and Galloway, R. B. (1996) A sig- 107±117. ni®cant Aegean volcanic eruption during the second Lamothe, M., Dreimanis, A., Morency, M. and Raukas, millennium B.C. revealed by thermoluminescence A. (1984) Thermoluminescence dating of Quaternary dating. Geoarchaeology 11, 361±371. sediments. In Quaternary Dating Methods, ed. W. C. Loy, T. H. (1994) Direct dating of rock art at Laurie Mahaney, Developments in Palaeontology and Creek (NT), Australia: a reply to Nelson. Antiquity Stratigraphy 7, pp. 153±170. Elsevier, Amsterdam. 68, 147±148. Lamothe, M., Balescu, S. and Auclair, M. (1994) Natural Loy, T. H., Jones, R., Nelson, D. E., Meehan, B., Vogel, IRSL intensities and apparent luminescence ages of J., Southon, J. and Cosgrove, R. (1990) Accelerator single feldspar grains extracted from partially radiocarbon dating of human blood proteins in pig- bleached sediments. Radiation Measurements 23, ments from Late Pleistocene art sites in Australia. 555±561. Antiquity 64, 110±116. Lang, A. (1994) Infra-red stimulated luminescence dating Lu, Y., Zhang, J., Xie, J. and Wang, X. (1988) TL dating of Holocene reworked silty sediments. Quaternary of pottery sherds and baked soil from the Xian Science Reviews (Quaternary Geochronology) 13, Terracotta Army site, Shaanxi Province, China. 525±528. Nuclear Tracks and Radiation Measurements 14, 283± Lang, A. and Wagner, G. A. (1996) Infrared stimulated 286. luminescence dating of archaeosediments. Lyell, C. (1863) The Geological Evidences of the Antiquity Archaeometry 38, 129±141. of Man, with Remarks on Theories of the Origin of Lang, A. and Wagner, G. A. (1997) Infrared stimulated Species by Variation. John Murray, London. luminescence dating of Holocene colluvial sediments Lynch, T. F. (1990) Glacial-age man in South America? A using the 410 nm emission. Quaternary Science critical review. American Antiquity 55, 12±36. Reviews (Quaternary Geochronology) 16, 393±396. Lynch, T. F., Gillespie, R., Gowlett, J. A. J. and Hedges, Larick, R. and Ciochon, R. L. (1996) The African emer- R. E. M. (1985) Chronology of Guitarrero Cave, gence and early Asian dispersals of the genus Homo. Peru. Science 229, 864±867. American Scientist 84, 538±551. MacDonald, G. F. (1983) Eastern North America. In Latham, A. G. (1997) Uranium-series dating of bone by Early Man in the New World, ed. R. Shutler, Jr., pp. gamma-ray spectrometry: comment. Archaeometry 97±108. Sage Publications, Beverly Hills. 39, 217±219. Machida, H., Blong, R. J., Specht, J., Moriwaki, H., Laurent, M., FalgueÁ res, C., Bahain, J.-J. and Yokoyama, Torrence, R., Hayakawa, Y., Talai, B., Lolok, D. and Y. (1994) GeÂ ochronologie du systeÁ me de terrasses ¯u- Pain, C. F. (1996) Holocene explosive eruptions of viatiles Quaternaires du bassin de la Somme par Witori and Dakataua caldera volcanoes in West New datation RPE sur quartz, deÂ seÂ quilibres des familles Britain, Papua New Guinea. Quaternary International de l'uranium et magneÂ tostratigraphie. Comptes 34-36, 65±78. Rendus de L'AcadeÂmie des Sciences, Paris 318(SeÂ rie Magee, J. W., Bowler, J. M., Miller, G. H. and Williams, II), 521±526. D. L. G. (1995) Stratigraphy, sedimentology, chron- Leach, B. F. and Fankhauser, B. (1978) The characteriz- ology and palaeohydrology of Quaternary lacustrine ation of New Zealand obsidian sources by use of deposits at Madigan Gulf, Lake Eyre, South LUMINESCENCE DATING IN ARCHAEOLOGY 885

Australia. Palaeogeography, Palaeoclimatology, Mellars, P. (1986) A new chronology for the French Palaeoecology 113, 3±42. Mousterian period. Nature 322, 410±411. Marshall, E. (1990) Clovis counterrevolution. Science 249, Mellars, P. and GruÈ n, R. (1991) A comparison of the elec- 738±741. tron spin resonance and thermoluminescence dating Martin, P. S. (1973) The discovery of America. Science methods: the results of ESR dating at Le Moustier 179, 969±974. (France). Cambridge Archaeological Journal 1, 269± Martin, P. S. (1990) 40,000 years of extinctions on 276. the ``planet of doom''. Palaeogeography, Mellars, P. A., Zhou, L. P. and Marseglia, E. A. (1997) Palaeoclimatology, Palaeoecology (Global and Compositional inhomogeneity of sediments and its Planetary Change Section) 82, 187±201. potential eects on dose rate estimation for electron Mazess, R. B. and Zimmerman, D. W. (1966) Pottery dat- spin resonance dating of tooth enamel. Archaeometry ing from thermoluminescence. Science 152, 347±348. 39, 169±176. McDermott, F., GruÈ n, R., Stringer, C. B. and Meltzer, D. J. (1989) Why don't we know when the ®rst Hawkesworth, C. J. (1993) Mass-spectrometric U- people came to North America? American Antiquity series dates for Israeli Neanderthal/early modern 54, 471±490. hominid sites. Nature 363, 252±255. Meltzer, D. J. (1995) Clocking the ®rst Americans. Annual McDonald, J., Ocer, K., Jull, T., Donahue, D., Head, Review of Anthropology 24, 21±45. J. and Ford, B. (1990) Investigating 14C AMS: dating Meltzer, D. J. (1997) Monte Verde and the Pleistocene prehistoric rock art in the Sydney sandstone basin, peopling of the Americas. Science 276, 754±755. Australia. Rock Art Research 7, 83±92. Meltzer, D. J., Adovasio, J. M. and Dillehay, T. D. (1994) McKeever, S. W. S., Bùtter-Jensen, L., Agersnap Larsen, On a Pleistocene human occupation at Pedra Furada, N., Mejdahl, V. and Poolton, N. R. J. (1996) Brazil. Antiquity 68, 695±714. Optically stimulated luminescence sensitivity changes Meltzer, D. J., Grayson, D. K., Ardila, G., Barker, A. W., in quartz due to repeated use in single aliquot read- Dincause, D. F., Haynes, C. V., Mena, F., Nunez, out: experiments and computer simulations. L. and Stanford, D. J. (1997) On a Pleistocene anti- Radiation Protection Dosimetry 65, 49±54. quity of Monte Verde, southern Chile. American McKeever, S. W. S., Agersnap Larsen, N., Bùtter-Jensen, Antiquity 62, 659±663. L. and Mejdahl, V. (1997) OSL sensitivity changes Mercier, N. (1991) Flint palaeodose determination at the during single aliquot procedures: computer simu- onset of saturation. Nuclear Tracks and Radiation lations. Radiation Measurements 27, 75±82. Measurements 18, 77±79. Meakins, R. L., Dickson, B. L. and Kelly, J. C. (1978) Mercier, N. and Valladas, H. (1994) Thermoluminescence The eect on thermoluminescent dating of disequili- dates for the Paleolithic Levant. In Late Quaternary brium in the uranium decay chain. Journal of the Chronology and Paleoclimates of the Eastern Council of Europe PACT 2, 97±103. Mediterranean, eds. O. Bar-Yosef and R. S. Kra, pp. Meakins, R. L., Dickson, B. L. and Kelly, J. C. (1979) 13±20. Radiocarbon, Department of Geosciences, Gamma ray analysis of K, U and Th for dose-rate University of Arizona, Tucson. estimation in thermoluminescent dating. Mercier, N., Valladas, H., Joron, J.-L., Reyss, J.-L., Archaeometry 21, 79±86. LeÂ veÃ que, F. and Vandermeersch, B. (1991) Mejdahl, V. (1969) Thermoluminescence dating of ancient Thermoluminescence dating of the late Neanderthal Danish ceramics. Archaeometry 11, 99±104. remains from Saint-CeÂ saire. Nature 351, 737±739. Mejdahl, V. (1972) Progress in TL dating at RisoÈ . Mercier, N., Valladas, H. and Valladas, G. (1992) Archaeometry 14, 245±256. Observations on palaeodose determination with Mejdahl, V. (1982) An automated procedure for thermolu- burnt ¯ints. Ancient TL 10, 28±32. minescence dating of pottery and burnt stones. Mercier, N., Valladas, H., Bar-Yosef, O., Vandermeersch, Journal of the Council of Europe PACT 7, 83±96. B., Stringer, C. and Joron, J.-L. (1993) Mejdahl, V. (1983) Feldspar inclusion dating of ceramics Thermoluminescence date for the Mousterian burial and burnt stones. Journal of the Council of Europe site of Es-Skhul, Mt. Carmel. Journal of PACT 9, 351±364. Archaeological Science 20, 169±174. Mejdahl, V. (1986) Thermoluminescence dating of sedi- Mercier, N., Valladas, H. and Valladas, G. (1995a) Flint ments. Radiation Protection Dosimetry 17, 219±227. thermoluminescence dates from the CFR laboratory Mejdahl, V. (1988a) The plateau method for dating par- at Gif: contributions to the study of the chronology tially bleached sediments by thermoluminescence. of the Middle Palaeolithic. Quaternary Science Quaternary Science Reviews 7, 347±348. Reviews (Quaternary Geochronology) 14, 351±364. Mejdahl, V. (1988b) Long-term stability of the TL signal Mercier, N., Valladas, H., Valladas, G., Reyss, J.-L., in alkali feldspars. Quaternary Science Reviews 7, Jelinek, A., Meignen, L. and Joron, J.-L. (1995b) TL 357±360. dates of burnt ¯ints from Jelinek's excavations at Mejdahl, V. (1989) A survey of archaeological samples Tabun and their implications. Journal of dated in 1988. Risù-M-2807, Risù National Archaeological Science 22, 495±509. Laboratory, Roskilde. Mercier, N., Valladas, H., Joron, J. L., Schiegl, S., Mejdahl, V. and Bùtter-Jensen, L. (1994) Luminescence Bar-Yosef, O. and Weiner, S. (1995c) dating of archaeological materials using a new tech- Thermoluminescence dating and the problem of geo- nique based on single aliquot measurements. chemical evolution of sedimentsÐa case study: the Quaternary Science Reviews (Quaternary Mousterian levels at Hayonim. Israel Journal of Geochronology) 13, 551±554. Chemistry 35, 137±141. Mejdahl, V. and Wintle, A. G. (1984) Merrilees, D. (1968) Man the destroyer: late Quaternary Thermoluminescence applied to age determination in changes in the Australian marsupial fauna. Journal of archaeology and geology. In Thermoluminescence the Royal Society of Western Australia 51, 1±24. and Thermoluminescent Dosimetry, ed. Y. S. Michab, M., Feathers, J. K., Joron, J.-L., Mercier, N., Horowitz, Volume III, pp. 133±190. CRC Press, Selo, M., Valladas, H., Valladas, G., Reyss, J.-L. and Boca Raton. Roosevelt, A. C. (1997) Luminescence dates for the Melcher, C. L. and Zimmerman, D. W. (1977) Paleoindian site of Pedra Pintada, Brazil. Quaternary Thermoluminescent determination of prehistoric heat Science Reviews (Quaternary Geochronology) (in treatment of chert artifacts. Science 197, 1359±1362. press). 886 R. G. ROBERTS

Miller, G. H., Beaumont, P. B., Jull, A. J. T. and Morwood, M. J., O'Sullivan, P. B., Aziz, F, and Raza, Johnson, B. (1992) Pleistocene geochronology and A. (1998) Fission track ages of stone tools and fossils palaeothermometry from protein diagenesis in ostrich in central Flores, Indonesia. Nature (in press). eggshells: implications for the evolution of modern Murray, A. S. (1981) Environmental radioactivity studies humans. Philosophical Transactions of the Royal relevant to thermoluminescence dating. D.Phil. thesis, Society of London 337(Series B), 149±157. Oxford University, Oxford. Miller, G. H., Magee, J. W. and Jull, A. J. T. (1997) Low- Murray, A. S. and Aitken, M. J. (1982) The measurement latitude glacial cooling in the Southern Hemisphere and importance of radioactive disequilibria in TL from amino-acid racemization in emu eggshells. samples. Journal of the Council of Europe PACT 6, Nature 385, 241±244. 155±169. Mishra, S. and Rajaguru, S. N. (1996) Comment on ``New Murray, A. S. and Roberts, R. G. (1997) Determining the geochemical evidence for the Youngest Toba Tu in burial time of single grains of quartz using optically India''. Quaternary Research 46, 340±341. stimulated luminescence. Earth and Planetary Science Mishra, S., Venkatesan, T. R., Rajaguru, S. N. and Letters (in press). Somayajulu, B. L. K. (1995) Earliest Acheulian Murray, A., Wohl, E. and East, J. (1992) industry from peninsular India. Current Anthropology Thermoluminescence and excess 226Ra decay dating 36, 847±851. of late Quaternary ¯uvial sands, East Alligator Mochanov, Y. A. (1992) The Earliest Palaeolithic of River, Australia. Quaternary Research 37, 29±41. northeastern Asia and the problem of an extratropi- Murray, A. S., Olley, J. M. and Caitcheon, G. G. (1995) cal cradle of Man. Presentation at the 45th Annual Measurement of equivalent doses in quartz from con- Northwest Anthropological Conference (Simon Fraser temporary water-lain sediments using optically stimu- University, Vancouver, April 1992). lated luminescence. Quaternary Science Reviews Monnier, J.-L., HalleÂ goueÈ t, B., Hinguant, S., Laurent, M., (Quaternary Geochronology) 14, 365±371. Auguste, P., Bahain, J.-J., FalgueÁ res, C., Gebhardt, Murray, A. S., Roberts, R. G. and Wintle, A. G. (1997) A., Marguerie, D., Molines, N., Morzadec, H. and Equivalent dose measurement using a single aliquot Yokoyama, Y. (1994) A new regional group of the of quartz. Radiation Measurements 27, 171±184. Lower Palaeolithic in Brittany (France), recently Nambi, K. S. V. and Aitken, M. J. (1986) Annual dose dated by electron spin resonance. Comptes Rendus de conversion factors for TL and ESR dating. L'AcadeÂmie des Sciences, Paris 319(SeÂ rie II), 155± Archaeometry 28, 202±205. 160. Nanson, G. C. and Young, R. W. (1987) Comparison of Morell, V. (1994) Did early humans reach Siberia 500,000 thermoluminescence and radiocarbon age-determi- years ago? Science 263, 611±612. nations from Late-Pleistocene alluvial deposits near Morell, V. (1995) The earliest art becomes olderÐand Sydney, Australia. Quaternary Research 27, 263±269. more common. Science 267, 1908±1909. Nanson, G. C., Young, R. W. and Stockton, E. D. (1987) Morozov, G. V. (1968) The relative dating of Quaternary Chronology and palaeoenvironment of the Ukrainian sediments by the thermoluminescence Cranebrook Terrace (near Sydney) containing arte- method. In Proceedings of the 8th International facts more than 40,000 years old. Archaeology in Quaternary Association Congress (Paris, France). Oceania 22, 72±78. United States Geological Survey Library Catalogue Nanson, G. C., Price, D. M., Short, S. A., Young, R. No. 208 M8280, Washington DC. W. and Jones, B. G. (1991) Comparative uranium± Mortlock, A. J. and Price, D. M. (1980±1) A new group thorium and thermoluminescence dating of weathered of Ban Chiang potsherds dated by thermolumines- Quaternary alluvium in the tropics of northern cence. Journal of the Hong Kong Archaeological Australia. Quaternary Research 35, 347±366. Society, 9, 86±90. Nanson, G. C., East, T. J. and Roberts, R. G. (1993) Mortlock, A. J. and Price, D. M. (1984) Unusual features Quaternary stratigraphy, geochronology and evol- of the thermoluminescence signal pro®le for sedi- ution of the Magela Creek catchment in the monsoon ments from beneath Lake George NSW. Ancient TL tropics of northern Australia. Sedimentary Geology 2, 10±13. 83, 277±302. Morwood, M. J. and Hobbs, D. R. (1995) Themes in the Naumann, I. D. (1983) The biology of mud nesting prehistory of tropical Australia. Antiquity 69, 747± Hymenoptera (and their associates) and Isoptera in 768. rock shelters of the Kakadu region, Northern Morwood, M. J., Walsh, G. L. and Watchman, A. (1994) Territory. In The Rock Art Sites of Kakadu National The dating potential of rock art in the Kimberley, ParkÐSome Preliminary Research Findings for their N.W. Australia. Rock Art Research 11, 79±87. Conservation and Management, ed. D. Gillespie, Morwood, M. J., Hobbs, D. R. and Price, D. M. (1995a) Special Publication 10, pp. 127±189. Australian Excavations at Sandy Creek 1 and 2. In Quinkan National Parks and Wildlife Service, Canberra. Prehistory: the Archaeology of Aboriginal Art in S.E. Nelson, D. E. (1993) Second thoughts on a rock-art date. Cape York Peninsula, Australia, eds. M. J. Morwood Antiquity 67, 893±895. and D. R. Hobbs, Tempus 3, pp. 71±91. Nelson, D. E., Chaloupka, G., Chippindale, C., Alderson, Anthropology Museum, University of Queensland, M. S. and Southon, J. R. (1995) Radiocarbon dates Brisbane. for beeswax ®gures in the prehistoric rock art of Morwood, M. J., L'Oste-Brown, S. and Price, D. M. northern Australia. Archaeometry 37, 151±156. (1995b) Excavations at Mushroom Rock. In Quinkan Nobbs, M. and Dorn, R. (1993) New surface exposure Prehistory: the Archaeology of Aboriginal Art in S.E. ages for petroglyphs from the Olary Province, South Cape York Peninsula, Australia, eds. M. J. Morwood Australia. Archaeology in Oceania 28, 18±39. and D. R. Hobbs, Tempus 3, pp. 133±146. Noller, B. N., Currey, N. A., Ayers, G. P. and Gillett, R. Anthropology Museum, University of Queensland, W. (1990) Chemical composition and acidity of rain- Brisbane. fall in the Alligator Rivers region, Northern Morwood, M. J., Aziz, F., van den Bergh, G. D., Territory, Australia. The Science of the Total Sondaar, P. Y. and de Vos, J. (1997) Stone artefacts Environment 91, 23±48. from the 1994 excavation at Mata Menge, west cen- Nott, J. and Roberts, R. G. (1996) Time and process rates tral Flores, Indonesia. Australian Archaeology 44, over the past 100 m.y.: a case for dramatically 26±34. increased landscape denudation rates during the late LUMINESCENCE DATING IN ARCHAEOLOGY 887

Quaternary in northern Australia. Geology 24, 883± Poolton, N. R. J., Bùtter-Jensen, L. and Rink, W. J. (1995) 887. An optically stimulated luminescence study of ¯int O'Connor, S. (1995) Carpenter's Gap Rockshelter 1: related to radiation dosimetry. Radiation 40,000 years of Aboriginal occupation in the Napier Measurements 24, 551±555. Ranges, Kimberley, WA. Australian Archaeology 40, Porat, N., Horwitz, L., Zhou, L. P., Chazan, M., 58±59. Saragosti, I. and Noy, T. (1996) ESR and lumines- Olley, J. M., Murray, A. and Roberts, R. G. (1996) The cence dating of the Paleolithic site at Holon, the eects of disequilibria in the uranium and thorium Coastal Plain, Israel. Abstracts of the Annual decay chains on burial dose rates in ¯uvial sediments. Meeting of the Israel Geological Society (Elat, March Quaternary Science Reviews (Quaternary 1996), p. 76. Geochronology) 15, 751±760. Prescott, J. R. (1982) Thermoluminescence dating in the Olley, J. M., Roberts, R. G. and Murray, A. S. (1997a) South-West Paci®c region. Journal of the Council of Disequilibria in the uranium decay series in sedimen- Europe PACT 7, 67±73. tary deposits at Allen's Cave, Nullarbor Plain, Prescott, J. R. (1983) Thermoluminescence dating of sand Australia: implications for dose rate determinations. dunes at Roonka, South Australia. Journal of the Radiation Measurements 27, 433±443. Council of Europe PACT 9, 505±512. Olley, J., Caitcheon, G. and Murray, A. (1997b) The dis- Prescott, J. R. and Fox, P. J. (1990) Dating quartz sedi- tribution of apparent dose as determined by optically ments using the 3258C TL peak: new spectral data. stimulated luminescence in small aliquots of ¯uvial Ancient TL 8, 32±34. quartz: implications for dating young sediments. Prescott, J. R. and Mojarrabi, B. (1993) Selective bleach: Quaternary Science Reviews (Quaternary an improved partial bleach technique for ®nding Geochronology) (in press). equivalent doses for TL dating of quartz sediments. Owen, R. (1877) Researches on the Fossil Remains of the Ancient TL 11, 27±30. Extinct Mammals of Australia; With a Notice of the Prescott, J. R. and Purvinskis, R. A. (1991) Zero thermo- Extinct Marsupials of England (2 volumes). luminescence for zero age. Ancient TL 9, 19±20. J. Erxleben, London. Prescott, J. R. and Purvinskis, R. A. (1993) A problem in Oyston, B. (1996) Thermoluminescence age determinations the thermoluminescence dating of recent sediments. for the Mungo III human burial, Lake Mungo, In Archaeometry: Current Australasian Research, eds. southeastern Australia. Quaternary Science Reviews B. L. Fankhauser and J. R. Bird, Occasional Papers (Quaternary Geochronology) 15, 739±749. in Prehistory 22, pp. 110±113. Department of Page, K., Dare-Edwards, T., Thorne, A., Webb, S. and Prehistory, Research School of Paci®c Studies, Price, D. (1994) Pleistocene human occupation site at Australian National University, Canberra. Lake Urana, New South Wales. Australian Prescott, J. R. and Robertson, G. B. (1997) Sediment dat- Archaeology 38, 38±44. ing in review. Radiation Measurements 27, 893±922. Prescott, J. R. and Sherwood, J. E. (1988) Pardoe, C. (1995) Riverine, biological and cultural evol- Thermoluminescence ages for an unusual shell ution in southeastern Australia. Antiquity 69, 696± deposit at Point Ritchie, Warrnambool, Australia. In 713. Archaeometry: Australasian Studies 1988, ed. J. R. Parenti, F., Mercier, N. and Valladas, H. (1990) The old- Prescott, pp. 61±69. Department of Physics and est hearths of Pedra Furada, Brasil: thermolumines- Mathematical Physics, University of Adelaide, cence analysis of heated stones. Current Research in Adelaide. the Pleistocene 7, 36±38. Prescott, J. R., Robertson, G. B. and Green, R. C. (1982) Â Parenti, F., Fontugue, M. and Guerin, C. (1996) Pedra Thermoluminescence dating of Paci®c Island pottery: Furada in Brazil and its ``presumed'' evidence: limi- successes and failures. Archaeology in Oceania 17, tations and potential of the available data. Antiquity 142±147. 70, 416±421. Prescott, J. R., Polach, H. A., Pretty, G. L. and Smith, B. PareÂ s, J. M. and PeÂ rez-GonzaÂ lez, A. (1995) Paleomagnetic W. (1983) Comparison of 14C and thermoluminescent age for hominid fossils at Atapuerca archaeological dates from Roonka, South Australia. Journal of the site, Spain. Science 269, 830±832. Council of Europe PACT 8, 205±211. Pavlides, C. (1993) New archaeological research at Prescott, J. R., Huntley, D. J. and Hutton, J. T. (1993) Yombon, West New Britain, Papua New Guinea. Estimation of equivalent dose in thermoluminescence Archaeology in Oceania 28, 55±59. dating: the Australian slide method. Ancient TL 11, Pavlides, C. and Gosden, C. (1994) 35,000 year-old sites in 1±5. the rainforests of West New Britain, Papua New Pretty, G. L. (1988) Radiometric chronology and signi®- Guinea. Antiquity 68, 604±610. cance of the fossil hominid sequence from Roonka, Pearce, R. H. and Barbetti, M. (1981) A 38,000 year-old South Australia. In Early Man in the Southern archaeological site at Upper Swan, Western Hemisphere, ed. J. R. Prescott, pp. S32±S52. Australia. Archaeology in Oceania 16, 173±178. Supplement to Archaeometry: Australasian Studies Pearson, S. and Dodson, J. R. (1993) Stick-nest rat mid- 1988, Department of Physics and Mathematical dens as sources of palaeoecological data in Physics, University of Adelaide, Adelaide. Australian deserts. Quaternary Research 39, 347±354. Price, D. M. (1994) TL dating of ¯uvial quartz sands: a Peerdeman, F. M., Davies, P. J. and Chivas, A. R. (1993) comparison of ages obtained at 3258C and 3758C. The stable oxygen isotope signal in shallow-water, Ancient TL 12, 20±23. upper-slope sediments o the Great Barrier Reef Prokein, J. and Wagner, G. A. (1994) Analysis of thermo- (Hole 820A). Proceedings of the Ocean Drilling luminescent glow peaks in quartz derived from the Program. Scienti®c Results 133, 163±173. KTB-drill hole. Radiation Measurements 23, 85±94. Peng, T.-H., Goddard, J. G. and Broecker, W. S. (1978) A Ralph, E. K. and Han, M. C. (1966) Dating of pottery by direct comparison of 14C and 230Th ages at Searles thermoluminescence. Nature 210, 245±247. Lake, California. Quaternary Research 9, 319±329. Ralph, E. K. and Han, M. C. (1969) Potential of thermo- Phillips, F. M., Flinsch, M., Elmore, D. and Sharma, luminescence in supplementing radiocarbon dating. P. (1997) Maximum ages of the CoÃ a valley (Portugal) World Archaeology 1, 157±169. engravings measured with Chlorine-36. Antiquity 71, Raynal, J.-P. and Huxtable, J. (1989) First thermolumines- 100±104. cence dating of Charentian Mousterian in Velay 888 R. G. ROBERTS

(Massif Central, France). Comptes Rendus de Richardson, N. (1992) Conjoin sets and stratigraphic L'AcadeÂmie des Sciences, Paris 309(SeÂ rie II), 157± integrity in a sandstone shelter: Kenni Cave 162. (Queensland, Australia). Antiquity 66, 408±418. Readhead, M. L. (1982) Extending thermoluminescence Rigaud, J.-P., Simek, J. F. and Ge, T. (1995) Mousterian dating to geological sediments. In Archaeometry: an ®res from Grotte XVI (Dordogne, France). Antiquity Australasian Perspective, eds. W. Ambrose and P. 69, 902±912. Duerden, Occasional Papers in Prehistory 12, pp. Rink, W. J., Schwarcz, H. P., Lee, H. K., Cabrera ValdeÂ s, 276±281. Department of Prehistory, Research School V., Bernaldo de QuiroÂ s, F. and Hoyos, M. (1996) of Paci®c Studies, Australian National University, ESR dating of tooth enamel: comparison with AMS Canberra. 14C at El Castillo Cave, Spain. Journal of Readhead, M. L. (1988) Thermoluminescence dating study Archaeological Science 23, 945±951. of quartz in aeolian sediments from southeastern Robbins, L. H., Murphy, M. L., Stewart, K. M., Australia. Quaternary Science Reviews 7, 257±264. Campbell, A. C. and Brook, G. A. (1994) Barbed Readhead, M. (1990) Thermoluminescence dating of sedi- bone points, paleoenvironment, and the antiquity of ments from Lake Mungo and Nyah West. In ®sh exploitation in the Kalahari Desert, Botswana. Quaternary Dating Workshop 1990, ed. R. Gillespie, Journal of Field Archaeology 21, 257±264. pp. 35±37. Department of Biogeography and Roberts, R. G. (1991) Sediment budgets and Quaternary Geomorphology, Research School of Paci®c Studies, history of the Magela Creek catchment, tropical Australian National University, Canberra. northern Australia. Ph.D thesis, University of Reagan, M. J., Rowlett, R. M., Garrison, E. G., Dort, W. Wollongong, Wollongong. Jr., Bryant, V. M. Jr. and Johannsen, C. J. (1978) Roberts, R. G. and Jones, R. (1994) Luminescence dating Flake tools strati®ed below Paleo-Indian artifacts. of sediments: new light on the human colonisation of Science 200, 1272±1275. Australia. Australian Aboriginal Studies 1994/2, 2±17. Rees-Jones, J. (1995) Optical dating of young sediments Roberts, R. G., Jones, R. and Smith, M. A. (1990a) using ®ne-grain quartz. Ancient TL 13, 9±14. Thermoluminescence dating of a 50,000 year-old Rees-Jones, J. and Tite, M. S. (1997a) Optical dating human occupation site in northern Australia. Nature results for British archaeological sediments. 345, 153±156. Archaeometry 39, 177±187. Roberts, R. G., Jones, R. and Smith, M. A. (1990b) Rees-Jones, J. and Tite, M. S. (1997b) Optical dating of Stratigraphy and statistics at Malakunanja II: reply the Ungton White Horse. In Archaeological to Hiscock. Archaeology in Oceania 25, 125±129. Sciences 1995, ed. A. Sinclair, pp. 171±174. Roberts, R. G., Nanson, G. C. and East, T. J. (1991) Proceedings of the British Archaeological Sciences Sediment budgets and Quaternary history of the Conference 1995, Oxbow Archaeological Magela Creek catchment, tropical northern Monographs, Oxbow Books, Oxford. Australia. Open File Record 80, Supervising Scientist for the Alligator Rivers Region, Sydney. Rees-Jones, J., Hall, S. J. B. and Rink, W. J. (1997) A lab- Roberts, R. G., Uren, C. J. and Murray, A. S. (1993a) oratory inter-comparison of quartz optically stimu- Thermoluminescence dating techniques at the lated luminescence (OSL) results. Quaternary Science Alligator Rivers Region Research Institute. Technical Reviews (Quaternary Geochronology) 16, 275±280. Memorandum 41, Supervising Scientist for the Rendell, H. and Dennell, R. W. (1985) Dated Lower Alligator Rivers Region, Australian Government Palaeolithic artefacts from northern Pakistan. Publishing Service, Canberra. Current Anthropology 26, 393. Roberts, R. G., Spooner, N. A. and Questiaux, D. Rendell, H. M. and Dennell, R. W. (1987) G. (1993b) Cautions on the use of extended duration Thermoluminescence dating of an Upper Pleistocene preheats in the optical dating of quartz. Ancient TL site, northern Pakistan. Geoarchaeology 2, 63±67. 11, 47±54. Rendell, H. M., Steer, D. C., Townsend, P. D. and Roberts, R. G., Jones, R., Spooner, N. A., Head, M. J., Wintle, A. G. (1985) The emission spectra of TL Murray, A. S. and Smith, M. A. (1994a) The human from obsidian. Nuclear Tracks and Radiation colonisation of Australia: optical dates of 53,000 and Measurements 10, 591±600. 60,000 years bracket human arrival at Deaf Adder Rendell, H. M., Hailwood, E. A. and Dennell, R. Gorge, Northern Territory. Quaternary Science W. (1987) Magnetic polarity stratigraphy of Upper Reviews (Quaternary Geochronology) 13, 575±583. Siwalik Sub-Group, Soan Valley, Pakistan: impli- Roberts, R. G., Jones, R. and Smith, M. A. (1994b) cations for early human occupance of Asia. Earth Beyond the radiocarbon barrier in Australian prehis- and Planetary Science Letters 85, 488±496. tory. Antiquity 68, 611±616. Renfrew, C. and Bahn, P. (1991) Archaeology: Theories, Roberts, R. G., Spooner, N. A., Jones, R., Cane, S., Methods and Practice. Thames and Hudson, London. Olley, J. M., Murray, A. S. and Head, M. J. (1996) Rhodes, E. J. (1988) Methodological considerations in the Preliminary luminescence dates for archaeological optical dating of quartz. Quaternary Science Reviews sediments on the Nullarbor Plain, South Australia. 7, 395±400. Australian Archaeology 42, 7±16. Rhodes, E. J. (1990) Optical dating of quartz from sedi- Roberts, R., Walsh, G., Murray, A., Olley, J., Jones, R., ments. D.Phil. thesis, Oxford University, Oxford. Morwood, M., Tuniz, C., Lawson, E., Macphail, M., Rhodes, E. J. and Bailey, R. M. (1997) The eect of ther- Bowdery, D. and Naumann, I. (1997) Luminescence mal transfer on the zeroing of the luminescence of dating of rock art and past environments using mud- quartz from recent glacio¯uvial sediments. wasp nests in northern Australia. Nature 387, 696± Quaternary Science Reviews (Quaternary 699. Geochronology) 16, 291±298. Robertson, G. B. and Prescott, J. R. (1987) Richards, M. P. (1994) Luminescence dating of quartzite Thermoluminescence: a key element in the Thai cer- from the Diring Yuriakh site. M.A. thesis, Simon amics archaeological project. In Archaeometry: Fraser University, Vancouver. Further Australasian Studies, eds. W. R. Ambrose Richardson, C. A. (1994) Eects of bleaching on the sensi- and J. M. J. Mummery, Occasional Papers in tivity to dose of the infrared-stimulated luminescence Prehistory 14, pp. 195±202. Department of of potassium-rich feldspars from Ynyslas, Wales. Prehistory, Research School of Paci®c Studies, Radiation Measurements 23, 587±591. Australian National University, Canberra. LUMINESCENCE DATING IN ARCHAEOLOGY 889

Robertson, G. B. and Prescott, J. R. (1988) The Thai cer- Kebara Cave, Israel. Journal of Archaeological amics archaeological project: TL characteristics of Science 16, 653±659. the artifacts. Nuclear Tracks and Radiation Seeley, M.-A. (1975) Thermoluminescent dating in its ap- Measurements 14, 299±305. plication to archaeology: a review. Journal of Roosevelt, A. C. (1996) Paleoindians in the Brazilian Archaeological Science 2, 17±43. Amazon: response. Science 274, 1823±1825 and 1934. Semaw, S., Renne, P., Harris, J. W. K., Feibel, C. S., Roosevelt, A. C., Housley, R. A., Imazio da Silveira, M., Bernor, R. L., Fesseha, N. and Mowbray, K. (1997) Maranca, S. and Johnson, R. (1991) Eighth millen- 2.5 million-year-old stone tools from Gona, Ethiopia. nium pottery from a prehistoric shell midden in the Nature 385, 333±336. Brazilian Amazon. Science 254, 1621±1624. Shackleton, N. J., Berger, A. and Peltier, W. R. (1990) An Roosevelt, A. C., Lima da Costa, M., Lopes Machado, alternative astronomical calibration of the lower C., Michab, M., Mercier, N., Valladas, H., Feathers, Pleistocene timescale based on ODP Site 677. J., Barnett, W., Imazio da Silveira, M., Henderson, Transactions of the Royal Society of Edinburgh, Earth A., Sliva, J., Cherno, B., Reese, D. S., Holman, Sciences 81, 251±261. J. A., Toth, N. and Schick, K. (1996) Paleoindian Shane, P., Westgate, J., Williams, M. and Korisettar, cave dwellers in the Amazon: the peopling of the R. (1995) New geochemical evidence for the Americas. Science 272, 373±384. Youngest Toba Tu in India. Quaternary Research Rowlett, R. M. and Pearsall, D. M. (1993) Archaeological 44, 200±204. age determinations derived from opal phytoliths by Shane, P., Westgate, J., Williams, M. and Korisettar, thermoluminescence. In Current Research in Phytolith R. (1996) Reply to comments by S. Mishra and S. N. Analysis: Applications in Archaeology and Rajaguru on ``New geochemical evidence for the Paleoecology, eds. D. M. Pearsall and D. R. Piperno, Youngest Toba Tu in India''. Quaternary Research MASCA Research Papers in Science and Archaeology 46, 342±343. 10, pp. 25±29. University Museum, University of Shelkoplyas, V. N. (1971) Dating of the Quaternary Pennsylvania, Philadelphia. deposits by means of thermoluminescence. In Rowlett, R. M., Mandeville, M. D. and Zeller, E. J. (1974) Chronology of the Glacial Age, eds. V. A. Zubakov The interpretation and dating of humanly worked and V. V. Kochegura, pp. 155±160. Geographical siliceous materials by thermoluminescent analysis. Society of the USSR, Leningrad. Proceedings of the Prehistoric Society 40, 37±44. Shelkoplyas, V. N. and Morozov, G. V. (1965) Some Russ, J., Hyman, M., Shafer, H. J. and Rowe, M. results of an investigation of Quaternary deposits by W. (1990) Radiocarbon dating of prehistoric rock the thermoluminescence method. In Materials on the paintings by selective oxidation of organic carbon. Quaternary Period of the Ukraine, pp. 83±90. 7th International Quaternary Association Congress, Nature 348, 710±711. Kiev. Sampson, E. H., Fleming, S. J. and Bray, W. (1972) Shepherd, M. J. and Price, D. M. (1990) Thermoluminescent dating of Colombian pottery in Thermoluminescence dating of late Quaternary dune the Yotoco style. Archaeometry 14, 121±128. sand, Manawatu/Horowhenua area, New Zealand: a Sanderson, D. C. W., Warren, S. E. and Hunter, comparison with 14C age determinations. New J. R. (1983) The TL properties of archaeological Zealand Journal of Geology and Geophysics 33, 535± glass. Journal of the Council of Europe PACT 9, 287± 539. 298. Sherwood, J., Barbetti, M., Ditchburn, R., Kimber, R. W. Sanderson, D. C. W., Placido, F. and Tate, J. O. (1988) L., McCabe, W., Murray-Wallace, C. V., Prescott, Scottish vitri®ed forts: TL results from six study J. R. and Whitehead, N. (1994) A comparative study sites. Nuclear Tracks and Radiation Measurements 14, of Quaternary dating techniques applied to sedimen- 307±316. tary deposits in southwest Victoria, Australia. Saunders, J. W., Mandel, R. D., Saucier, R. T., Allen, Quaternary Science Reviews (Quaternary E. T., Hallmark, C. T., Johnson, J. K., Jackson, E. Geochronology) 13, 95±110. H., Allen, C. M., Stringer, G. L., Frink, D. S., Short, M. A. and Huntley, D. J. (1992) Infrared stimu- Feathers, J. K., Williams, S., Gremillion, K. J., lation of quartz. Ancient TL 10, 19±21. Vidrine, M. F. and Jones, R. (1997) A mound com- Singh, G. and Geissler, E. A. (1985) Late Cainozoic his- plex in Louisiana at 5400±5000 years before present. tory of vegetation, ®re, lake levels and climate, at Science 277, 1796±1799. Lake George, New South Wales, Australia. Schwarcz, H. P. (1994) Current challenges to ESR dating. Philosophical Transactions of the Royal Society of Quaternary Science Reviews (Quaternary London 311(Series B), 379±447. Geochronology) 13, 601±605. Singh, G., Opdyke, N. D. and Bowler, J. M. (1981) Late Schwarcz, H. P. and GruÈ n, R. (1993) Electron spin reson- Cainozoic stratigraphy, palaeomagnetic chronology ance (ESR) dating of the Lower Industry. In The and vegetational history from Lake George, N.S.W. Lower Paleolithic Site at Hoxne, England, eds. R. Journal of the Geological Society of Australia 28, Singer, B. G. Gladfelter and J. J. Wymer, pp. 210± 435±452. 212. University of Chicago Press, Chicago. Singhvi, A. K. and Mejdahl, V. (1985) Schwarcz, H. P. and Morawska, L. (1993) Uranium-series Thermoluminescence dating of sediments. Nuclear dating of carbonates from Bir Tarfawi and Bir Tracks and Radiation Measurements 10, 137±161. Sahara East. In Egypt during the Last Interglacial: Singhvi, A. K. and Wagner, G. A. (1986) the Middle Paleolithic of Bir Tarfawi and Bir Sahara Thermoluminescence dating and its application to East, eds. F. Wendorf, R. Schild, A. E. Close and young sedimentary deposits. In Dating Young Associates, pp. 205±217. Plenum Press, New York. Sediments, eds. A. J. Hurford, E. JaÈ ger and J. A. M. Schwarcz, H. P., GruÈ n, R., Vandermeersch, B., Bar-Yosef, Ten Cate, CCOP Technical Publication 16, pp. 159± O., Valladas, H. and Tchernov, E. (1988) ESR dates 197. CCOP Technical Secretariat, Bangkok. for the hominid burial site of Qafzeh in Israel. Singhvi, A. K., Sharma, Y. P. and Agrawal, D. P. (1982) Journal of Human Evolution 17, 733±737. Thermoluminescence dating of sand dunes in Schwarcz, H. P., Buhay, W. M., GruÈ n, R., Valladas, H., Rajasthan, India. Nature 295, 313±315. Tchernov, E., Bar-Yosef, O. and Vandermeersch, Singhvi, A. K., Sharma, Y. P. and Agrawal, D. P. (1983) B. (1989) ESR dating of the Neanderthal site, Thermoluminescence dating of Indian archaeological 890 R. G. ROBERTS

sites. Journal of the Council of Europe PACT 9, 365± Stockton, E. D. and Holland, W. (1974) Cultural sites and 370. their environment in the Blue Mountains. Singhvi, A. K., Deraniyagala, S. U. and Sengupta, Archaeology and Physical Anthropology in Oceania 9, D. (1986a) Thermoluminescence dating of 36±65. Quaternary red-sand beds: a case study of coastal Stokes, S. (1992) Optical dating of independently dated dunes in Sri Lanka. Earth and Planetary Science Late Quaternary eolian deposits from the Southern Letters 80, 139±144. High Plains. Current Research in the Pleistocene 9, Singhvi, A. K., Sauer, W., Wagner, G. A. and KroÈ ger, 125±129. K. (1986b) Thermoluminescence dating of loess Stokes, S. (1993) Optical dating of sediment samples from deposits at Plaidter Hummerich and its implications Bir Tarfawi and Bir Sahara East: an initial report. In for the chronology of Neanderthal man. Egypt during the Last Interglacial: the Middle Naturwissenschaften 73, 205±207. Paleolithic of Bir Tarfawi and Bir Sahara East, eds. Smith, B. W. (1988) More cautions on laboratory illumi- F. Wendorf, R. Schild, A. E. Close and Associates, nation. Ancient TL 6, 9. pp. 229±233. Plenum Press, New York. Smith, B. W., Prescott, J. R. and Polach, H. (1982) Stoneham, D. (1991) Authenticity testing. In Scienti®c Thermoluminescence dating of marine sediments Dating Methods, eds. H. Y. GoÈ ksu, M. Oberhofer from Spencer Gulf. In Archaeometry: an Australasian and D. Regulla, Eurocourses Advanced Scienti®c Perspective, eds. W. Ambrose and P. Duerden, Techniques 1, pp. 175±192. Kluwer Academic Occasional Papers in Prehistory 12, pp. 282±289. Publishers, Dordrecht. Department of Prehistory, Research School of Paci®c Straus, L. G. (1989) Age of the modern Europeans. Studies, Australian National University, Canberra. Nature 342, 476±477. Smith, B. W., Aitken, M. J., Rhodes, E. J., Robinson, P. Strickertsson, K., Placido, F. and Tate, J. O. (1988) D. and Geldard, D. M. (1986) Optical dating: meth- Thermoluminescence dating of Scottish vitri®ed forts. odological aspects. Radiation Protection Dosimetry Nuclear Tracks and Radiation Measurements 14, 317± 17, 229±233. 320. Smith, B. W., Rhodes, E. J., Stokes, S. and Spooner, N. Stringer, C. (1988) The dates of Eden. Nature 331, 565± A. (1990a) The optical dating of sediments using 566. quartz. Radiation Protection Dosimetry 34, 75±78. Stringer, C. B. and GruÈ n, R. (1991) Time for the last Smith, B. W., Rhodes, E. J., Stokes, S., Spooner, N. Neanderthals. Nature 351, 701±702. A. and Aitken, M. J. (1990b) Optical dating of sedi- Stringer, C. and McKie, R. (1996) African Exodus: the ments: initial quartz results from Oxford. Origins of Modern Humanity. Jonathan Cape, Archaeometry 32, 19±31. London. Smith, M. A. (1987) Pleistocene occupation in arid Stringer, C. B., GruÈ n, R., Schwarcz, H. P. and Goldberg, Central Australia. Nature 328, 710±711. P. (1989) ESR dates for the hominid burial site of Es Smith, M. A., Prescott, J. R. and Head, M. J. (1997) Skhul in Israel. Nature 338, 756±758. Comparison of 14C and luminescence chronologies at Stuiver, M. and Reimer, P. J. (1993) Extended 14C data Puritjarra rock shelter, central Australia. Quaternary base and revised CALIB 3.0 14C age calibration pro- Science Reviews (Quaternary Geochronology) 16, gram. Radiocarbon 35, 215±230. 299±320. Sondaar, P. Y., van den Bergh, G. D., Mubroto, B., Aziz, Stute, M., Forster, M., Frischkorn, H., Serejo, A., Clark, F., de Vos, J. and Batu, U. L. (1994) Middle J. F., Schlosser, P., Broecker, W. S. and Bonani, Pleistocene faunal turnover and colonization of G. (1995) Cooling of tropical Brazil (58C) during the Flores (Indonesia) by Homo erectus. Comptes Rendus last glacial maximum. Science 269, 379±383. de L'AcadeÂmie des Sciences, Paris 319(SeÂ rie II), Sutton, S. R. and Zimmerman, D. W. (1976) 1255±1262. Thermoluminescent dating using zircon grains from Spell, T. L. and McDougall, I. (1992) Revisions to the age archaeological ceramics. Archaeometry 18, 125±134. of the Brunhes±Matuyama boundary and the Swisher, C. C., III, Curtis, G. H., Jacob, T., Getty, A. G., Pleistocene geomagnetic polarity timescale. Suprijo, A. and Widiasmoro (1994) Age of the ear- Geophysical Research Letters 19, 1181±1184. liest known hominids in Java, Indonesia. Science 263, Spooner, N. A. (1994a) On the optical dating signal from 1118±1121. quartz. Radiation Measurements 23, 593±600. Swisher, C. C., III, Rink, W. J., AntoÂ n, S. C., Schwarcz, Spooner, N. A. (1994b) The anomalous fading of infrared- H. P., Curtis, G. H., Suprijo, A. and Widiasmoro, stimulated luminescence from feldspars. Radiation (1996) Latest Homo erectus of Java: potential con- Measurements 23, 625±632. temporaneity with Homo sapiens in southeast Asia. Spooner, N. A. and Prescott, J. R. (1986) A caution on Science 274, 1870±1874. laboratory illumination. Ancient TL 4, 46±48. Tanaka, K., Machette, M. N., Crone, A. J. and Bowman, Spooner, N. A., Prescott, J. R. and Hutton, J. T. (1988) J. R. (1995) ESR dating of aeolian sand near The eect of illumination wavelength on the bleach- Tennant Creek, Northern Territory, Australia. ing of the thermoluminescence (TL) of quartz. Quaternary Science Reviews (Quaternary Quaternary Science Reviews 7, 325±329. Geochronology) 14, 385±393. Spooner, N. A., Aitken, M. J., Smith, B. W., Franks, Tankersley, K. B. and Munson, C. A. (1992) Comments M. and McElroy, C. (1990) Archaeological dating by on the Meadowcroft Rockshelter radiocarbon chron- infrared-stimulated luminescence using a diode array. ology and the recognition of coal contaminants. Radiation Protection Dosimetry 34, 83±86. American Antiquity 57, 321±326. Spriggs, M. and Anderson, A. (1993) Late colonization of Tauxe, L., Deino, A. D., Behrensmeyer, A. K. and Potts, East Polynesia. Antiquity 67, 200±217. R. (1992) Pinning down the Brunhes/Matuyama and Stanford, D. (1983) Pre-Clovis occupation south of the ice upper Jaramillo boundaries: a reconciliation of orbi- sheets. In Early Man in the New World, ed. R. tal and isotopic time scales. Earth and Planetary Shutler, Jr., pp. 65±72. Sage Publications, Beverly Science Letters 109, 561±572. Hills. Taylor, R. E., Haynes, C. V. Jr. and Stuiver, M. (1996) Steadman, D. W. (1995) Prehistoric extinctions of Paci®c Clovis and Folsom age estimates: stratigraphic con- island birds: biodiversity meets zooarchaeology. text and radiocarbon calibration. Antiquity 70, 515± Science 267, 1123±1131. 525. LUMINESCENCE DATING IN ARCHAEOLOGY 891

Tedford, R. H. and Wells, R. T. (1990) Pleistocene depos- bon dates for prehistoric paintings at the Altamira, its and fossil vertebrates from the ``dead heart of El Castillo and Niaux caves. Nature 357, 68±70. Australia''. Memoirs of the Queensland Museum 28, van Oosterzee, P. (1997) Where Worlds Collide: the 263±284. Wallace Line. Reed, Melbourne. Templeton, A. R. (1993) The ``Eve'' hypotheses: a genetic Vandermeersch, B. (1989) The evolution of modern critique and reanalysis. American Anthropologist 95, humans: recent evidence from southwest Asia. In The 51±72. Human Revolution: Behavioural and Biological Terrell, J. (1989) Commentary: what Lapita is and what Perspectives on the Origins of Modern Humans, eds. Lapita isn't. Antiquity 63, 623±626. P. Mellars and C. Stringer, pp. 155±164. Edinburgh Theocaris, P. S., Liritzis, I. and Galloway, R. B. (1997) University Press, Edinburgh. Dating of two Hellenic pyramids by a novel Veth, P. (1995) Aridity and settlement in northwest application of thermoluminescence. Journal of Australia. Antiquity 69, 733±746. Archaeological Science 24, 399±405. Wagner, G. A., Aitken, M. J. and Mejdahl, V. (1983) Thieme, H. (1997) Lower Palaeolithic hunting spears from Thermoluminescence Dating. Handbooks for Germany. Nature 385, 807±810. Archaeologists 1, European Science Foundation, Tishko, S. A., Dietzsch, E., Speed, W., Pakstis, A. J., Strasbourg. Kidd, J. R., Cheung, K., BonneÂ -Tamir, B., Wallace, A. R. (1860) On the zoological geography of the Santachiara-Benerecetti, A. S., Moral, P., Krings, Malay Archipelago. Journal of the Proceedings of the M., PaÈ aÈ bo, S., Watson, E., Risch, N., Jenkins, T. and Linnean Society of London 4, 172±184. Kidd, K. K. (1996) Global patterns of linkage dise- Walsh, G. L. (1988) Australia's Greatest Rock Art. E. quilibrium at the CD4 locus and modern human ori- J. Brill-Robert Brown and Associates, Bathurst. gins. Science 271, 1380±1387. Walsh, G. L. (1994) Bradshaws: Ancient Rock Paintings of Thorne, A. G. and Macumber, P. G. (1972) Discoveries of North-West Australia. The Bradshaw Foundation, late Pleistocene man at Kow Swamp, Australia. Geneva. Nature 238, 316±319. Watchman, A. (1993a) Perspectives and potentials for ab- Troja, S. O., Cro, A., Gueli, A. M., La Rosa, V., solute dating prehistoric rock paintings. Antiquity 67, Mazzoleni, P., Pezzino, A. and Romeo, M. (1996) 58±65. Characterization and thermoluminescence dating of Watchman, A. (1993b) Evidence of a 25,000 year-old pic- prehistoric pottery sherds from Milena. Archaeometry tograph in northern Australia. Geoarchaeology 8, 38, 113±128. 465±473. Tso, M.-Y. W. and Li, S.-H. (1994) Equivalent dose esti- Watchman, A. (1995) Recent petroglyphs, Foz CoÃ a, mation for pottery by single disc regeneration Portugal. Rock Art Research 12, 104±108. method. Radiation Measurements 23, 451±454. Watchman, A. and Cole, N. (1993) Accelerator radiocar- Ugumori, T. and Ikeya, M. (1980) Luminescence of bon dating of plant-®bre binders in rock paintings CaCO3 under N2 laser excitation and application to from northeastern Australia. Antiquity 67, 355±358. archaeological dating. Japanese Journal of Applied Watchman, A. L., Walsh, G. L., Morwood, M. J. and Physics 19, 459±465. Tuniz, C. (1997) AMS radiocarbon age estimates for Valladas, G. and Gillot, P. Y. (1978) Dating of the Olby early rock paintings in the Kimberley, N.W. lava ¯ow using heated quartz pebbles: some pro- Australia: preliminary results. Rock Art Research 14, blems. Journal of the Council of Europe PACT 2, 18±26. 141±150. Waterhouse, D. F. (1991) Insects and humans in Valladas, H. (1978) Thermoluminescence dating of burnt Australia. In The Insects of Australia: a Textbook for stones from prehistoric site. Journal of the Council of Students and Research Workers, ed. CSIRO, (2nd edi- Europe PACT 2, 180±184. tion, Volume I) pp. 221±235. Melbourne University Valladas, H. (1981) Thermoluminescence de gres de foyers Press, Melbourne. prehistoriques: estimation de leur temperature de Waters, M. R., Forman, S. L. and Pierson, J. M. (1997a) chaue. Archaeometry 23, 221±229. Diring Yuriakh: a Lower Paleolithic site in central Valladas, H. (1985) Some TL properties of burnt prehisto- Siberia. Science 275, 1281±1284. ric ¯ints. Nuclear Tracks and Radiation Measurements Waters, M. R., Forman, S. L. and Pierson, J. M. (1997b) 10, 785±788. Geoarchaeological studies at Diring Yuriakh: a Valladas, H. (1992) Thermoluminescence dating of ¯int. Lower Paleolithic site in central Siberia. Quaternary Quaternary Science Reviews 11, 1±5. Research (submitted). Valladas, H. and Valladas, G. (1987) Thermoluminescence Weinstein, J. M. (1984) Radiocarbon dating in the dating of burnt ¯int and quartz: comparative results. southern Levant. Radiocarbon 26, 297±366. Archaeometry 29, 214±220. Wells, R. T., Moriarty, K. and Williams, D. L. G. (1984) Valladas, H., Geneste, J. M., Joron, J. L. and Chadelle, The fossil vertebrate deposits of Victoria Fossil Cave J. P. (1986) Thermoluminescence dating of Le Naracoorte: an introduction to the geology and Moustier (Dordogne, France). Nature 322, 452±454. fauna. The Australian Zoologist 21, 305±333. Valladas, H., Joron, J. L., Valladas, G., Arensburg, B., Wendorf, F., Schild, R. and Close, A. E. (1993) Summary Bar-Yosef, O., Belfer-Cohen, A., Goldberg, P., and conclusions. In Egypt during the Last Laville, H., Meignen, L., Rak, Y., Tchernov, E., Interglacial: the Middle Paleolithic of Bir Tarfawi and Tillier, A. M. and Vandermeersch, B. (1987) Bir Sahara East, eds. F. Wendorf, R. Schild, A. E. Thermoluminescence dates for the Neanderthal burial Close and Associates, pp. 552±573. Plenum Press, site at Kebara in Israel. Nature 330, 159±160. New York. Valladas, H., Reyss, J. L., Joron, J. L., Valladas, G., Bar- Wendorf, F., Schild, R., Close, A. E., Schwarcz, H. P., Yosef, O. and Vandermeersch, B. (1988) Miller, G. H., GruÈ n, R., Bluszcz, A., Stokes, S., Thermoluminescence dating of Mousterian ``Proto- Morawska, L., Huxtable, J., Lundberg, J., Hill, C. L. Cro-Magnon'' remains from Israel and the origin of and McKinney, C. (1994) A chronology for the modern man. Nature 331, 614±616. Middle and Late Pleistocene wet episodes in the east- Valladas, H., Cachier, H., Maurice, P., Bernaldo de ern Sahara. In Late Quaternary Chronology and Quiros, F., Clottes, J., Cabrera ValdeÂ s, V., Paleoclimates of the Eastern Mediterranean, eds. O. Uzquiano, P. and Arnold, M. (1992) Direct radiocar- Bar-Yosef and R. S. Kra, pp. 147±168. Radiocarbon, 892 R. G. ROBERTS

Department of Geosciences, University of Arizona, cedures applied to late-Holocene colluvium near St Tucson. Paul's Mission, Natal, South Africa. The Holocene 5, Wheeler, M. (1956) Archaeology from the Earth. Penguin 97±102. Books, Harmondsworth. Wolpo, M. H. (1989) Multiregional evolution: the fossil White, C. (1971) Man and environment in northwest alternative to Eden. In The Human Revolution: Arnhem Land. In Aboriginal Man and Environment in Behavioural and Biological Perspectives on the Origins Australia, eds. D. J. Mulvaney and J. Golson, pp. of Modern Humans, eds. P. Mellars and C. Stringer, 141±157. Australian National University Press, pp. 62±108. Edinburgh University Press, Edinburgh. Canberra. Wright, R. V. S. (1971a) The archaeology of Koonalda Whittle, E. H. and Arnaud, J. M. (1975) Cave. In Aboriginal Man and Environment in Thermoluminescent dating of Neolithic and Australia, eds. D. J. Mulvaney and J. Golson, pp. Chalcolithic pottery from sites in central Portugal. 105±113. Australian National University Press, Archaeometry 17, 5±24. Canberra. Wickler, S. and Spriggs, M. (1988) Pleistocene human oc- Wright R. V. S. (1971b) Prehistory in the Cape York cupation of the Solomon Islands, Melanesia. Peninsula. In Aboriginal Man and Environment in Antiquity 62, 703±706. Australia, eds. D. J. Mulvaney D. J. and J. Golson, Wiggenhorn, H., Lang, A. and Wagner, G. A. (1994) pp. 133±140. Australian National University Press, Infrared stimulated luminescence: dating tool for Canberra. archaeosediments. Naturwissenschaften 81, 556±558. Wright, R. (1986a) New light on the extinction of the Williams, H., Beasley, T., Huntley, D. and Newton, Australian megafauna. Proceedings of the Linnean W. (1991) A Spanish jar from the Queen Charlotte Society of New South Wales 109, 1±9. Islands. The International Journal of Nautical Wright, R. (1986b) How old is zone F at Lake George? Archaeology 20, 259±261. Archaeology in Oceania 21, 138±139. Wintle, A. G. (1973) Anomalous fading of thermolumines- Yellen, J. E., Brooks, A. S., Cornelissen, E., Mehlman, M. cence in mineral samples. Nature 245, 143±144. J. and Stewart, K. (1995) A Middle Stone Age Wintle, A. G. (1978) A thermoluminescence dating study worked bone industry from Katanda, Upper Semliki of some Quaternary calcite: potential and problems. Valley, Zaire. Science 268, 553±556. Canadian Journal of Earth Sciences 15, 1977±1986. Yokoyama, Y., FalgueÁ res, C. and Quaegebeur, J. P. (1985) Wintle, A. G. (1980) Thermoluminescence dating: a review ESR dating of quartz from Quaternary sediments: of recent applications to non-pottery materials. ®rst attempt. Nuclear Tracks and Radiation Archaeometry 22, 113±122. Measurements 10, 921±928. Wintle, A. G. (1987) Thermoluminescence dating of loess. Yoshida, H. (1997) Quaternary dating studies using ESR Catena Supplement 9, 103±114. signals, with emphasis on shell, coral, tooth enamel Wintle, A. G. (1990) A review of current research on TL and quartz. Ph.D thesis, Australian National dating of loess. Quaternary Science Reviews 9, 385± University, Canberra. 397. ZilhaÄ o, J. (1995) The age of the CoÃ a valley (Portugal) Wintle, A. G. (1993) Luminescence dating of aeolian rock-art: validation of archaeological dating to the sands: an overview. In The Dynamics and Environmental Context of Aeolian Sedimentary Palaeolithic and refutation of ``scienti®c'' dating to Systems, ed. K. Pye, Special Publication 72, pp. 49± historic or proto-historic times. Antiquity 69, 883± 58. Geological Society of London, London. 901. Wintle, A. G. (1996) Archaeologically-relevant dating Zimmerman, D. W. (1971) Thermoluminescent dating techniques for the next century. Journal of using ®ne grains from pottery. Archaeometry 13, 29± Archaeological Science 23, 123±138. 52. Wintle, A. G. (1997) Luminescence dating procedures and Zimmerman, D. W. and Huxtable, J. (1969) Recent appli- laboratory protocols. Radiation Measurements 27, cations and developments in thermoluminescent dat- 769±817. ing. Archaeometry 11, 105±108. Wintle, A. G. and Aitken, M. J. (1977) Zimmerman, D. W. and Huxtable, J. (1971) Thermoluminescence dating of burnt ¯int: appli- Thermoluminescent dating of Upper Palaeolithic cation to a Lower Palaeolithic site, Terra Amata. ®red clay from DolnõÂ Vestonice. Archaeometry 13, Archaeometry 19, 111±130. 53±57. Wintle, A. G. and Huntley, D. J. (1979) Zimmerman, D. W., Yuhas, M. P. and Meyers, P. (1974) Thermoluminescence dating of a deep-sea sediment Thermoluminescence authenticity measurements on core. Nature 279, 710±712. core material from the Bronze Horse of the New Wintle, A. G. and Huntley, D. J. (1982) York Metropolitan Museum of Art. Archaeometry Thermoluminescence dating of sediments. Quaternary 16, 19±30. Science Reviews 1, 31±53. ZoÈ ller, L., Conard, N. J. and Hahn, J. (1991) Wintle, A. G. and Murray, A. S. (1997) The relationship Thermoluminescence dating of Middle Palaeolithic between quartz thermoluminescence, photo-trans- open air sites in the Middle Rhine Valley, Germany. ferred thermoluminescence, and optically stimulated Naturwissenschaften 78, 408±410. luminescence. Radiation Measurements 27, 611±624. ZoÈ ller, L., Lang, A. and Wagner, G. A. (1996) Wintle, A. G., Li, S. H. and Botha, G. A. (1993) Luminescence dating in archaeology and Quaternary Luminescence dating of colluvial deposits from geology. In Archaeometrical and Archaeological Natal, South Africa. South African Journal of Science Research in Macedonia and Thrace, pp. 427±444. 89, 77±82. Proceedings of the 2nd Symposium (26±28 March Wintle, A. G., Li, S. H., Botha, G. A. and Vogel, 1993), Hellenic Archaeometrical Society, J. C. (1995) Evaluation of luminescence-dating pro- Thessaloniki.