ISSN 1897 -1695 (online), 1733 8387 (print) 2015 C. Burow et al. This work is licensed under the Creative Commons Attribution-NonCommercial NoDerivatives 3.0 License. GEOCHRONOMETRIA 42 (2015): 107â€“ 125 DOI 10.1515/geochr 20150010 Available online at http://www.degruyter.com/view/j/geochr LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN CHRISTOPH BUROW1, MARTIN KEHL1, ALEXANDRA HILGERS1, GERD CHRISTIAN WENIGER2, 3, DIEGO E. ANGELUCCI4, VALENTN VILLAVERDE5, JOSEFINA ZAPATA6, JOO ZILHO7 1University of Cologne, Institute of Geography, Albertus -Magnus -Platz, 50923 Cologne, Germany 2Neanderthal Museum, Talstrae 300, 40822 Mettmann, Germany 3University of Cologne, Institute of Prehistoric Archaeology, Albertus -Magnus -Platz, 50923 Cologne, Germany 4Dipartimento di Lettere e Filosofia, Universit degli Studi di Trento, via Tommaso Gar 14, 38122 Trento, Italy 5Departament de Prehistria i d'Arqueologia, Universitat de Valncia, Av. Blasco Ibaez 28, 46010 Valncia, Spain 6rea de Antropologa Fsica, Facultad de Biologa, Universidad de Murcia, Campus Universitario de Espinardo, 30100 Murcia, Spain 7Seminari dâ€™Estudis i Recerques Prehistriques, Departament de Prehistria, Histria Antiga i Arqueologia, Facultat de Geografia i Hist ria, Universitat de Barcelona/ICREA, Spain Received 5 M arch 2014 Accepted 6 January 2015 Abstract: The fluvial sediments at Cueva Antn, a Middle Palaeolithic rock shelter located in the valley of the River Mula (Southeast Spain), produced abundant lithic assemblages of Mousterian affinities. Radiocarbon dates are available for the upper part of the archaeological succession, while for the middle to lower parts chronometric data have been missing. Here we present luminescence dating results for these parts of the succession. Quartz OSL on small aliquots and single grain measurements yield ages ranging from 69 7 ka to 82 8 ka with a weighted mean of 72 4 ka for sub-complexes AS2 to AS5. Equivalent dose estimates from large aliquots were highest a nd inconsistent with those from single grains and small multiple grain aliquots. This is probably caused by the presence of oversaturating grains, which have been quantified by single grain measurements. Additional post -IR IRSL measurements on coarse grai ned feldspar give strong support to a well -bleached quartz OSL signal. While independent chronometric control is missing, the results are within the expected age range and support the notion of a rapid accumulation of the fluvial deposits. Keywords: Cueva Antn, Middle Palaeolithic , Luminescence dating , fluvial sediment , single grain dating , post -IR IRSL . 1. INTRODUCTION Cueva Antn is a Palaeolithic rock shelter located in the region of Murcia in Southeast Spain. The several meter thick sedimentary fill is mainly of alluvial origin deposited by the adjoining River Mula (Angelucc i et al. , 2013 ) (Fig. 1 ). Based on early findings during salvage excavations in 1991, where Middle Palaeolithic occup ations were identified at the basal levels of the alluvial succession (Martnez, 1997), more recent fieldwork b etween 2006 and 2012 gave evidence that human occupation also occurred during the accumulation of the upper parts of the succession (Zilho et al. , 2010, 2012 ). A rchaeological findings comprise abundant lithic artefacts, hearth features and a perforated Pecten maximus shell bearing residues of pigment (Zilho et al. , 2010 ). Rad i ocarbon dating on wood charcoal from Cueva Antn using the ABOx SC pre treatment protocol (Acid Base Oxidation Stepped Combustion; Brock et al. , 2010 ) dates Corresponding author: C. Burow e -mail: email@example.com
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 108 unit II b in the upper part of the archaeological succe ssion ( sub complex AS1, see Fig. 2 ) to 32890 2 00 BP (OxA21244, 38440â€“ 36810 cal BP ; Zilho et al. , 2010, Angelucci et al. , 2013 and Wood et al. , 2013 ). Further radiocarbon dates are available for units I k of AS1 (31070 1 70 BP, OxA 20882) and II h/i of AS2 (39650 5 50 BP , OxA18672), but are considered unr eliable due to the applied ABA (Acid Base Acid) pre treat ment method that is shown to be less reliable in r emoving contaminants from charcoal (Jris and Street 2008 ; Higham et al. , 2009 ; Higham 2011; Mar oto et al. , 2012 and Wood et al. , 2013). This study focuses on the application of luminescence dating to the fluvial sediments of the middle to lower parts of the archaeological succession (AS2 to AS5) at Cueva Antn. Optically stimulated luminescence (OSL) provides the tools to not only cross check the existing chronological framework, but also to provide first chro nometric age estimates for the archaeological succession where radiocarbon dating failed due to insufficient yields of the charcoal samples (Zilho et al. , 2010 ). Moreover, the fluvial nature of the sediments (Angelucci et al. , 2013 ) as well as the protected depositional environment in the rock shelter facilitates a challenging test bed to investigate the potential of various luminescence dating techniques. Since the advent of OSL dating (Huntley et al. , 1985) there has been substantial progress both in instrumentation (cf. Btter Jensen, 1997 ; Duller et al. , 1999 ; Btter Jensen et al. , 1999 , 2000, 2003 , 2010 ; Thomsen et al. , 2008a and Lapp et al. , 2012 ) and in methodological aspects; for the latter most notably by the introduction of the single aliquot regenerative dose pr otocol (Murray and Wintle, 2000 , 2003 and Wintle and Murray , 2006 ). One of the major challenges in lumine scence dating with respect to fluvial sediments is the que stion of whether the luminescence signal was completely or only partia lly reset (also termed â€œbleachedâ€) during transport prior to deposition (Wallinga, 2002 ; Jain et al. , 2004 and Rittenour, 2008). The use of increasingly smaller sub samples (â€œaliquotsâ€) (Olley et al. , 1998 and Duller, 2008) and single grain measurements (Duller et al. , 1999 and Btter Jensen et al. , 2000 ) greatly improved the reliability of differentiating fully from partially bleached grains. Single grain dating has also been proved to be valuable for sediments in archaeological contexts, e.g. to identify the mixing of sediments due to human activity (Jacobs and Roberts, 2007 ). Likewise, substantial progress has been achieved in infrared stimulated lum inescence (IRSL) dating of feldspar, which was long su bordinate because of anomalous fading ( Spooner, 1992, 1994 ), i.e. the loss of signal over time due to quantum mechanical tunnelling (Poolton et al. , 2002). Besides methods for correcting anomalous fading (Huntley and Lamothe, 2001; Auclair et al. , 2003 and Kars et al. , 2008) the development of elevated temperature post IR IRSL (pIRIR) dating protocols (Buylaert et al. , 2009 , 2012 and Thiel et al. , 2011 ), which measure non fading feldspar signals, initiated a series of studies to test their potential in various contexts (e.g. Kars et al. , 2012 ; Lowick et al. , 2012 ; Roberts, 2012 and Vasiliniuc et al. , 2012). Considering all this, there is now a wide variety of luminescence dating techniques available, each with their own assets and drawbacks. The present study applies and compares OSL dating on multiple and single grains of quartz. Additional post IR IRSL dating on K rich fel dspars are carried out to shed further light on the bleaching history of the sediments. The analysis of va rious lum inescence characteristics through a series of laboratory experiments and cross validation by comparison of the performance of the dating techniques used was carried out in order to ascertain the most appropriate method for establishing a reliable chronostratigraphy for the archae ological record at Cueva Antn. 2. STUDY SITE AND SAMPLES The Cueva Antn rock shelter (383'51.84"N, 129'47.20"W), located in the province of Murcia (Sout heast Spain), approximately 3 km north of the town of Mula and dir ectly adjoining the River Mula, is one of several Middle Palaeolithic sites in the Murcia region ( Zilho and Villaverde, 2008) . Situated in the vicinity of the Betic Cordillera to the north and the Mula Pliego Basin to the south the rock shelter itself is located in the Mula valley at an altitude of 356 m above sea level. The local geology of Cueva Antn is mainly conditioned by the Paleocene to Miocene rocks that form the local Mula unit, with lower Miocene mar ls and marly limestones, which outcrop along several thrust faults, dominating ( Fig. 1 ). The rock shelter itself is located at the base of a 25 â€“ 3 0 m high escarpment of mainly middle to upper Eocene limestones comprising calcareous breccias and conglomerates, calcarenites as well as micritic and nu mmulitic limestones (Angelucci et al. , 2013). The stratigraphic layout of the succession in Cueva Antn can be divided in four main complexes (from top to bottom): i) DD (Dam Deposits), made of fine, mostly silty beds that accumulated over the last decades (caused by occasional floodings of the rock shelter due to the construction of the La Cierva Dam started in 1929), ii) TL (Transitional Layers), fo rmed of disturbed layers of uncertain age in intermediate position between DD deposits and underlying Pleistocene sediments (in part, backdirt from the 1991 trench), iii) AS (Archaeological Succession), formed of a number of Upper Pleistocene superposed al luvial sequences featuring distinct sedime ntary facies, lateral variations, and including intercalations of slope material (particularly near the back wall), and iv) FP (Fine Palustrine), a weakly bedded, fine, organic rich sediment forming the base of the exposed succession ( Zilho et al. , 2010 and Angelucci et al. , 2013 ). It is the AS complex which is of particular geoarcha eological interest and is best described as a well prese rved alluvial sequence. A continuous accumulation of alluvial
C. Burow et al. 109 beds with high sedimentation rates, the protective effect of the rock shelter and the incision of the River Mula preserved the sedimentary facies and archaeological el ements from active surface dynamics, limited postdepositional dynamics and soil formation processes and spared the deposit from the subsequent action of fluvial dynamics (Angelucci et al. , 2013). Based on sedimentary and stratigraphic criteria the AS complex can further be subdivided into five subcomplexes (further comprising 48 different units). For a detailed synopsis of the stratigraphic complexes the reader is referred to Angelucci et al. ( 2013). While AS1 could successfully be dated by the radi ocarbon method, no radiometric ages are available for the subcomplexes AS2 to AS5. Attempts at radiocarbon dating the base of the succession have failed so far due to insufficient yields of the charcoal sa mples ( Zilho et al. , 2010 ). In order to provide first radiometric age estimates a total of nine samples were taken from the archaeolog ical succession for luminescence dating by pushing opaque stainless steel tubes into various layers of two profiles ( Fig. 2 ). The tubes were closed with opaque plastic caps, thus preventing an early reset of the luminescence signal. The first set of samples was taken during excavations in 2010 and a second during excavations in 2012. To determine the apparent residual dose in feldspar an additional modernanalogue sample (CA7) was taken from a subrecent fluvial terrace directly adjacent to the cave entrance of Cueva Antn. 3. EXPERIMENTAL DETAILS Sample preparation All sample preparation procedures were conducted under subdued red illumination, following routine proc edures (Wintle, 1997 ). Chemical treatment of the bulk samples involved hydrochloric acid (HCl 10%), hydrogen peroxide (H2O2 10%) and sodium oxalate (Na2C2O4). During this procedure any carbonates were dissolved, organic matter oxidi sed and the clay fraction dispersed. Coarse grained quartz and K rich feldspar were separated using sodium polytungstate with densities of 2.58 g cm, 2.62 g cm and 2.68 g cm. The enriched quartz fraction was further etched in hydrofluoric acid (HF 40%, 40 min) to remove any remaining feldspars and the outer layer of the quartz grains that received an alpha dose from the environment . Precipitated fluorides were removed by HCl (10%) wash for one hour. Depending on the amount of remai ning sample material purified quartz and feldspars were sieved to 100â€“1 50 m or 100 â€“2 00 m. For single Fig. 1. Geological setting and photographs of the Cueva Antn rock shelter. A) Geological context of the Cueva Antn rock shelter in the geological map, scale 1:50 000 (IGME, 1972 ). Legend: A: Concordant conta ct; B: Discordant contact; C: Fault; D: Thrust; 1, undifferentiated, 2, limestones (Toarcian -Oxfordian, Tithonian), 3, polygenetic conglomerates, arenites and marls (Oligocene), 4, arenites, limestones and conglomerates (Torton ian), 5, marls and marly lime stones (Lower Miocene), 6, marls with intercalations of nummulitic limestones (Lutetian), 7, limestones with Discocyclina (foraminifera) (Priabonian), 8, limestone conglomerates and marls (Paleocene -Lower Eocene), 9, nummulitic limestones (Lutetian), 10, l imestones and marls (Ypresian Lutetian), 11, marls with intercalations of nummulitic limestones (Priabonian). B) and C) photos of the Cueva Antn rock shel ter.
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 110 grain measurements of quartz the 200 â€“2 50 m grain size fraction was extracted. Additionally, for sample CA 9 the 40 â€“6 3 m quartz grain size fraction was prepared, appl ying the same chemical treatment as above. Mineral sep aration was achieved by etching the bulk sample in he xafluorosilicic acid (34%, 14 d). Precipitated fluorides were removed by HCl 10% wash. Instrumentation Luminescence measurements were conducted on Ris TL/OSLDA15/20 readers equipped with 90Sr/90Y beta sources for irradiat ion and delivering dose rates between 0.08 Gy s and 0.15 Gy s. Optical stimulation of quartz multiple grain samples was performed at 80 â€“9 0% power using blue diodes (47 0 3 0 nm) for 40 s (Btter Jensen et al. , 1999). Single grains of quartz were stimulated for 2 s with a green Nd:YVO4 diode pumped laser (532 nm) delivering a power density of ~50 W cm (Btter Jensen et al. , 2000 ). Feldspar samples were measured at 90% power using infrared diodes (88 0 8 0 nm) for 200 s. Luminescence signals were detected with an EMI 9235 photomultiplier through a 7.5 mm thick Hoya U340 (quartz) and 410 nm interference filter (feldspar). Fig. 2. Luminescence sampling at Cueva Antn. A) Site plan and excavated areas of the Cueva Antn rock shelter. Elevations are given in metres above sea level (modified from Angelucci et al. , 2013 ). B) The W wall of trench L (square L20 to L21). Samples and excavation units: CA -9, II -; CA 10, III -e/h; CA -11, III-k/l; CA -12, IIIm; CA -6, IIIm. C) The E wall of trench J (square J19 to J20). Samples and excavation units: CA -1, II-e; CA -2, II e; CA -4, II-y; CA-5, III-f. Dashed lines indicate stratigraphic complexes and subcomplexes. AS = Archaeological Succession, DD = Dam Deposits, FP = Fine Palustrine.
C. Burow et al. 111 The beta sources of three individual Ris TL/OSL reader with single grain attachments were calibrated using the BAG478 calibration quartz , which was thermally sensitised and homogeneously dosed to 8.03 Gy using a 60 source . Following the approach of Ballarini et al. ( 2006) about 500 individual De values (in seconds) in average were obtained for each reader. The relative standard deviation of De values obtained for each single grain position remained <20% in all cases. The spatial v ariability in the delivered dose rate was assessed by 3D plots ( Fig . S 1 in the Appendix). It is evident that the spatial distribution of delivered dose rate is very specific to individual beta sources (cf. Spooner and Allsop, 2000 ; Ballarini et al. , 2006 and Lapp et al. , 2012). While two beta sources have a near uniform dose rate distribution, a clear trend in delivered dose was identified for one of the beta sources and reconfirmed a previous calibration of the same source by Lomax ( 2009 ). When applying individual dose rates to calculate single grain De estimates, the ove rdispersion (OD), calculated using the central age model (CAM, Galbraith et al. , 1999), could be reduced by up to 6.5 percentage points ( Table 1 ). However, even after correction of the non uniform beta dose rate an OD of 8 â€“ 1 2% remained, which is in the range or slightly larger than the values of 6.9% and 7.3% reported in Thomsen et al. ( 2005). Similar values of 11.4% and 12% for gamma dosed single grain distrib utions are also reported in Thomsen et al. ( 2007 ). Equivalent dose estimation Quartz OSL was measured using a standard quartz SAR protocol after Murray and Wintle ( 2000, 2003 ) employing preheat temperatures of either 200C or 220 C for 10 s ( see Section 4 â€” Luminescence characteristics ). The cutheat temperature was always chosen to be 20 C lower than the preheat temperature. Values for Lx and Tx were derived from the initial 0.8 s of the OSL signal, minus a background estimated from the last 5.0 s of the stimulation curve. For single grain measurements the OSL signal was derived from the first 0.054 s of stimul ation, minus a background of the las t 0.4 s. IRSL on fel dspars was measured using the pIRIR225 and pIRIR290 protocols ( Buylaert et al. , 2009 and Thiel et al. , 2011) employing preheat temperatures of 250C and 320C f or 60 s, respectively. For both protocols the pIRIR stimul ation time was 200 s, and after every SAR cycle the sa mples were IR stimulated at 290 C (pIRIR225) and 325 C (pIRIR290) for 100 s to minimise any buildup of charge giving rise to a recuperated sign al . The pIRIR signals were derived from the first 2.4 s of the IR stimulation with a subtracted background of the last 40 s. For single and multiple grains of quartz test doses of 20 Gy were applied, whereas a higher test dose of 40 Gy was used for all fel dspar measurements. All dose response curves were fitted with a single saturating exponential function. Dosimetry Samples for dose rate determination were dried at 105 C for 24 h and finally homogeni sed with a mortar. The homogenis ed samples were filled in Marinelli bea kers of different standardised geometries filled with either 830 g or 1459 g sample material . After storing the sa mples for at least four weeks to allow equilibrium re establishment of 226Ra and its daughter nuclides, the sa mples we re measure d for 20 h in a high resolution gamma ray spectrometer with a coaxial P type high purity ge rmanium (HPGe) detector . The uranium (238U) and thor ium (232Th) contents were calculated by measuring the gamma rays of the corresponding daughter nuclides 226Ra, 214Pb, 214Bi and 228Ac, 212Pb , 208Tl, respectively. For all samples the specific activity (Bq kgâ€“1) of these n uclides were compared to check for indications of a dis equilibrium in the 238U and 232Th decay series. None of the samples showed aberrant behaviour , hence a secular equilibrium was assumed. The potassium content (40K) was determined by measuring the 1460 keV gamma ray that is emitted during the electron capture decay to 40Ar. The nuclide activities of 238U, 232Th and 40K were co nverted to dose rates using the conversion factors given in Gurin et al. ( 2011). To account for the attenuation of particles in grains correction factors of Mejdahl ( 1979 ) were used . The cosmicray dose rates were calculated according to Prescott and Hutton ( 1994 ), taking into consideration the altitude and geomagnetic latitude of the sampling site and the thickness, density and water content of the ove rlying sediments. For near surface samples (CA 1, CA 2) the cosmic ray dose rate was modelled using a polynom ial component fitted to data presented by Prescott and Hutton ( 1988 ). The attenuation of cosmic rays in the Table 1. Summary of beta dose rate assessment for three Ris TL/OSL readers with single grain attachments. Calibration of the beta sourc es was done using a calibration quartz that was thermally sensitised and homogeneously dosed to 8.03 Gy using a 60 Co gamma source. The overdispersion was calculated using the central age model (Galbraith et al. , 1999 ). The variation in delivered beta dose rate was highest for Reader C (see also Fig. S1 in the Appendix ). Reader Number of grains accept ed/ measured Mean dose rate (Gy s â€“ 1 ) L owest individual dose rate (Gy s â€“ 1 ) H ighest individual dose rate (Gy s â€“ 1 ) RSD (%) O verdispersion in D e values (mean dose rate) Overdispersion in De values (indi vidual dose rate) B 635/1700 0.088 0.066 0.107 9.2 12.4 0.2 11.1 0.2 C 553/1700 0.148 0.104 0.201 15.4 18.2 0.3 11.7 0.2 F 415/2400 0.081 0.072 0.097 6.3 9.1 0.2 8.1 0.2
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 112 overlying rock was accounted for and a correction for the geometric shielding by the rockshelter was incorporated using the equations given by Dunne et al. ( 1999). To account for the external alpha dose rate of the ou ter rim of the mineral grains a lpha efficiencies (avalues) of 0.0 7 0 .02 and 0.03 5 0 .003 were used for feldspar and quartz, respectively (Preusser et al. , 2005 and Lai et al. , 2008 ) . In addition, f or K rich feldspar an internal beta dose rate due to an internal 40K content has to be consi dered for age calculation. As the internal K content was not specifically measured, a po tassium concentration of 12. 5 0 .5% (Huntley and Baril, 1997 ) was assumed. While indirect (Reimann et al., 2012a ) and direct (Zhao and Li, 2005 and Smedley et al., 2012 ) determinations confirm a large variability of internal K contents in fel dspar, a mean potassium concentration of 12. 5 0 .5% for multiple grain measurements is still likely to be a valid estimate (Smedley et al. , 2012 ). Because water has a higher radiation absorption coe fficient than air it is necessary to estimate the water co ntent of the sample. Measured p resent day water contents varied from 1% to 19%. However, these values were assumed to be not representative for the water content over the burial time. O n the basis of soil texture and pore space, assuming predominantly cold and dry climate conditions during the Late Pleistocene (Fletcher and Snchez Goi, 2008 and Vegas et al ., 2010 ) and taking into consideration the sheltering from direct precipitation, a longterm water content of 5 3 % was estimated i nstead and used to correct the alpha, beta and gamma dose rates with correction factors given in Aitken ( 1985) . D osimetry data are summarised in Table 2 . 4. RESULTS Luminescence characteristics Quartz and feldspar from Cueva Antn showed bright OSL and IRSL signals , respectively ( Fig. 3 ) . Multiple grain aliquots were rejected when the recycling ratio exceeded 1.1 or was less than 0.9 (Murray and Wintle, 2000 ). Single grains of quartz were discarded when the recycling ratio exceeded 1.2 or was les s than 0.8 (cf. Jacobs et al. , 2003) . The error on the recycling ratio was not considered. Multiple grain aliquots of quartz and feldspar and single quartz grains were rejected when the r ecuperation ratio , i.e. the quotient of the Lx/ Tx ratio of the zero dose and the Ln/ Tn ratio of the natural dose, e xceeded 5% (Murray and Wintle, 2000 ). Furthermore, only aliquots and single grains of quartz having an OSL IR depletion ratio (Duller, 2003) between 1.1 and 0.9 were accepted. Further aliquots and single quartz grains were rejected if the net OSL signal was not at least three stan dard deviations above background, the test doses error was >5%, or when no meaningful curve fit of the dose response curve could be calculated. To assess the variability in the saturation dose D0 of quartz from Cueva Antn a single saturating exponential was fitted to all aliquots and single grains that passed the reject ion criteria. Mean D0 values varied between 63 and 67 Gy when measured on multiple grain aliquots. Ho wever, for a specific sample D0 values varied significantly from aliquot to aliquot with standard deviations ranging from 7 Gy (8 mm aliquots) to 22 Gy (1 mm aliquots). The variability in D0 values determined on single grains of quartz on samples CA 1 and CA5 is even larger with mean saturation doses of 6 3 2 8 Gy and 6 0 2 7 Gy, respectively. Table 2 . Dosimetry data for quartz and feldspar samples from Cueva Antn. Nuclide contents of U, Th and 40 K were derived from measurements using a high-resolution gamma-ray spectrometer with a coaxial P -type high purity germanium (HPGe ) detector . Measured water contents were assumed to be not representative for the wat er content over the burial time and a long-term water content of 5 3% was estimated instead . The water content is expressed as mass of dry sediment. Q = quartz, FS = fel dspar , WCM = w ater content measured, AWC = a ssumed water content . Sample Depth (m) Mineral Grain size (m) Uranium (ppm) Thorium (ppm) Potassium (%) WCM (%) AWC (%) Cosmic dose rate (Gy ka â€“ 1 ) Total dose rate D0 (Gy ka â€“ 1 ) CA -1 0.27 Q 100 â€“ 150 200 â€“ 250 1.15 0.06 2.02 0.13 0.40 0.02 18.9 5 3 0.103 0.87 0.10 CA 1 0.27 FS 100 â€“ 150 1.15 0.06 2.02 0.13 0.40 0.02 18.9 5 3 0.103 1.35 0.14 CA 2 0.38 Q 100 â€“ 150 1.18 0.08 2.49 0.17 0.46 0.03 4.1 5 3 0.100 0.97 0.12 CA 4 1.54 Q 100 â€“ 200 1.38 0.07 2.88 0.17 0.54 0.02 16.7 5 3 0.073 1.08 0.13 CA -5 1.88 Q 100 â€“ 200 200 â€“ 250 1.01 0.06 1.50 0.10 0.33 0.01 8.4 5 3 0.066 0.70 0.09 CA 5 1.88 FS 100 â€“ 150 1.01 0.06 1.50 0.10 0.33 0.01 8.4 5 3 0.066 1.18 0.16 CA 6 2.78 Q 100 â€“ 150 1.15 0.08 2.45 0.17 0.42 0.02 12.3 5 3 0.048 0.86 0.12 CA 9 2.26 Q 100 â€“ 150 1.02 0.06 1.84 0.11 0.27 0.01 1.5 5 3 0.058 0.67 0.08 CA -10 2.51 Q 40 â€“ 63 100 â€“ 150 0.93 0.07 1.41 0.11 0.27 0.02 1.9 5 3 0.053 0.65 0.08 CA 11 2.66 Q 100 â€“ 150 1.01 0.07 1.56 0.12 0.30 0.02 2.3 5 3 0.050 0.66 0.08 CA 12 2.78 Q 100 â€“ 150 1.22 0.09 2.32 0.18 0.39 0.02 3.3 5 3 0.048 0.85 0.10
C. Burow et al. 113 The signal composition of quartz was checked by li nearly modulated (LM )OSL measurements conducted on five 8 mm aliquots for samples CA 1 and CA 6, respe ctively ( Fig . S 2 in the Appendix). Deconvolution of the LM OSL curves was done using the fit_LMCurve() fun ction of the R package â€˜Luminescenceâ€™ (Kreutzer et al. , 2012 ). Analysis of the deconvolved LM OSL curves revealed the presence of four signal components. Mean values for photoionisation cross section of com ponents were (from component 1 to component 4): 2.7 3 0 .12 10â€“17 cm, 2.6 3 0 .21 10â€“18 cm, 2.3 7 0 .13 10â€“19 cm, and 2.32 0 .4 10â€“20 cm. Co mponent 1 can be associated with the fast co mponent . For these samples the fast component is dominant and makes up 80â€“ 8 5% of the net OSL signal. A p reheat plateau test was done on 8 mm aliquots from sample CA 1 ( Fig. 4 ). The De was measured in groups of five for different preheat temperatures in steps of 20 C ranging from 180C to 300 C (held for 10 s). The cutheat temperature was always chosen to be 20 C lower than the preheat temperature. Fig. 4 indica tes a preheat plateau between 200 â€“ 2 40 C . Additionally, a set of combined dose recovery preheat plateau tests were carried out on small aliquots (1 and 2 mm) of samples CA 1, CA 2 and CA9 ( Fig. 5 ). The naturally trapped charge was removed by blue stimulation for 150 s at room temperature. T he samples were then given a labor atory dose of 60 Gy. For all samples the dose recovery ratios were in agreement within 10% of unity for preheat temperatures from 180 C to 280C . Fig. 3 . Representative luminescence decay curves and corresponding dose response curves for quartz and feldspar samples from Cueva Antn. A) Quartz OSL intensity of a multiple grain aliquot (1 mm) from sample CA 6 during stimulation with blue diodes. B) OSL intensity of a single quartz gra in from sample CA -1 during stimulation with a green laser. C) and D) show the pIRIR decay curves during elevated stimulation temperature readouts at 225C and 290C, respectively. For quartz multiple grain measurements the signal of the initial 0.8 s were used minus a background of the last 5 s of stimulation. For single grain measurements the OSL signal was derived from the first 0.054 s of stimulation m inus a background of the last 0.4 s. Both pIRIR signals were derived from the first 2.4 s of the IR stim ulation with a subtracted background of the last 40 s. The insets show the sensitivity corrected dose response curves. Vertical dashed lines indicate the initial integral used to obtain the luminescence signal and the time interval used for background subt raction. Sensitivity-corrected OSL (Lx/Tx) Sensitivity-corrected OSL (Lx/Tx)Stimulation time (s)Stimulation time (s)OSL intensity (counts / 0.018 s) 020000400006000080000100000120000 01234 A 010203040Equivalent dose (Gy)Equivalent dose (Gy)020406080100120010015050 Sensitivity-corrected pIRIR225 signal (Lx/Tx)Stimulation time (s) OSL intensity (counts / 0.16 s) pIRIR225 intensity (counts / 0.8 s) 040008000120001600020000Equivalent dose (Gy) 00.51.01.52.02.53.0 C050100150200050100150 Sensitivity-corrected pIRIR290 signal (Lx/Tx)Stimulation time (s) pIRIR290 intensity (counts / 0.8 s) 0500010000150002000025000Equivalent dose (Gy) 012543 D050100150200050100150200
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 114 Based on these tests a preheat temperature in the range of 200 â€“2 40 C was considered to be appropriate for the samples under study. For all subsequent quartz OSL measurements a preheat/cutheat combination of either 200/180 C or 220/200C was used. The same thermal treatment was applied in the SAR protocol for single grain measurements. To check the applicability of all applied protocols for quartz and feldspar a series of dose recovery tests were conducted. Details on the dose recovery tests and t he ir results are given in Table 3 . In all cases the administered dose could be recovered within 10% of unity. For quartz the mean dose recovery ratio was 0.9 9 0 .04 and 1.0 4 0 .06 for feldspar. The dose recovery tests on fel dspar indicate a better reproducibility of the pIRIR225 si gnal over the pIRIR290 signal. Fig. 5. Dependency of De on preheat temperature deduced from â€œdose recovery preheat plateauâ€ tests for samples CA -1, CA -2 and CA 9. Filled circles indicate the mean D e of three or four aliquots measured for each preheat temperature. The error bars are the standard error of dose e stimates obtained for each temperature. Open squares are the mean recuperation rates. Solid line represents the given laboratory dose; dashed lines repr esent 10% uncertainty. For all three samples the laboratory dose of ~60 Gy could be recovered within 10% of unity for preheat temperatures from 180C to 280C. In all cases the mean recuperation rate remained below 5% of the natural signal. Table 3. Summary of dose recovery tests. Small aliquots of quartz were optically bleached for 150 s at room temperature with blue d iodes. The natural OSL signal of single quartz grains was removed by a repeated green laser stimulation for 4 s (at 25% las er power) at room temperature with an intermediate pause of 10000 s (cf. Duller, 2012 ) . Cross checking the OSL decay curves of both stimulations revealed that the natural signal of all grains was fully reset prior to irradiation. Feldspar aliquots were bleached for 1 h in a Hnle SOL2 solar simul ator ( a) and in the Ris TL/OSL reader by a repeated 200 s IR stimul ation with an intermediate pause of 1000 s ( b). For the aliquots bleached in the solar simulator residual doses of 4.2 Gy (pIRIR 225 ) and 12.6 Gy (pIRIR 290 ) were subtracted, which were measured on separate aliquots after the same bleach ing conditions (see Fig. S5 in the Appendix ). In all cases the applied laboratory doses could be recovered within 10% of unity . SG = single grain. Sample Mineral Protocol Grain size (m) Aliquot di ameter (mm) Given dose (Gy) n Dose recovery ratio CA 1 Q SAR 100 â€“ 150 2 60 24 0.94 0.03 CA 1 Q SAR 200 â€“ 250 SG 50 27 0.97 0.02 CA 1 FS pIRIR 225 100 â€“ 150 1 87 3 0.96 0.07 a CA 1 FS pIRIR 225 100 â€“ 150 1 87 3 0.99 0.01 b CA 1 FS pIRIR 290 100 â€“ 150 1 117 3 1.09 0.03 a CA 1 FS pIRIR 290 100 â€“ 150 1 117 3 1.10 0.03 b CA 9 Q SAR 100 â€“ 150 1 60 17 1.04 0.02 CA 10 Q SAR 40 â€“ 63 1 60 20 0.99 0.01 Fig. 4. Dependency of De on preheat temperature deduced from a preheat plateau test for sample CA 1. Filled circles indicate the mean D e of five aliquots measured for each preheat temperature. The error bars are the standard error of dose estimates obtained for each te mperature. Open squares are the recuperation rate , i.e. the quotient of the L x/Tx ratio of the zero dose and the Ln/Tn rati o of the natural dose, expressed in percent. The preheat plateau test was conducted on large 8 mm aliquots and shows a well defined preheat plateau ranging from 200C to 240C. The mean recuperation rate always remained below 5% of the natural signal.
C. Burow et al. 115 Estimating the amount of grains on a sample disc Multiple grain De estimation techniques are known to be prone to averaging effects (e.g. Duller, 2008 and A rnold and Roberts, 2009 ) and there is an increasing nu mber of single grain studies wher e aberrant grain behaviour was found to be the cause for De over and underestim ations on the multiple grain scale (e.g. Arnold et al. , 2012 ; Stone and Bailey, 2012 and Demur o et al. , 2008 , 2013 ). However, when dealing with averaging effects in mult iple grain measurements there is the obvious question on how many grains are actually present on an aliquot that potentially contribute to the measured OSL signal. Here, we adopted the approach of Heer et al. ( 2012), who presented empirical data on this issue, and estimated the number of grains n on an aliquot by a ssuming a t wo dimensional â€˜Packing Equal Circles in a Circleâ€™ (PECC) problem (cf. Huang and Ye, 2011 ) and using the follo wing equation d r r ng c 2 2 (4.1) where rc is the radius of the aliquot (mm), rg is the mean radius of the grain size fraction (mm) and d is the packing density (value between 0 and 1). Instead of assuming 907.012d (e.g. Rhodes, 2007 and Heer et al. , 2012 ), a value that can only be achieved in a densest circle packing on an infinite plane (Chang and Wang 20 10 ), packing density values were taken from Specht ( 2012) . Here, packing density values vary between 0.656 and 0.875 for n up to 1500 (omitting values for n = 1 and 2). In addition, multiple grain aliquots in 1, 2 and 8 mm diameter were prepared as described in Heer et al. ( 2012) and photographed with a digital Keyence VHX2000 microscope in order to count the number of grains on the di sc. The results are shown in Table 4 . It is evident that there is a substantial discrepancy b etween the theoretical and actual number of grains on an aliquot, even when assuming maximum packing densities in a PECC problem. In addition to the possible reasons given by Heer et al. ( 2012 ) it has to be considered that preparing the aliquot by sprinkling the grains on the disc is a highly random process. Thus, the probability of a rranging the grains in the densest possible configuration on the disc can be considered zero. One exception to this observation is the calculated packing density of 40 â€“ 6 4 m grains on 1 mm aliquots. Possible reasons for a packing density d eter is in reality larger than 1 mm (e.g. due to imprecise disc preparation), b) the grains are not arranged in a commonly assumed mono layer, or c) a combination of both. Aggregating the results of Heer et al. ( 2012) and of Table 4 a mean empirical packing density value of d = 0.65 is proposed. Using this value and assuming a mean grain size of 125 m it was estimated that 1, 2 and 8 mm aliquots contain 42, 166 and 2662 grains, respe ctivel y (see Fig . S 3 in the Appendix ). Proportion of grains emitting luminescence The brightness distribution of quartz from Cueva Antn was investigated by constructing a cumulative light sum curve (Duller et al., 2000) for the natural signal ( Ln) of samples CA 1 and CA 5 ( Fig. 6). However, vari ations in grain to grain brightness of the natural signal can be the result of external sources of variation such as pa rtial blea ching or beta dose heterogeneity . Therefore, a second plot was done for the natural test dose signal ( Tn), where the ratio of bright to dim grains ought to be the result of inherent quartz grain characteristics. In addition, the signal intensity distributi on of two quartz samples used for beta source calibration (RisoeQ and BAG478) was analysed in the same way. As both samples were thermally sensitised and homogenously dosed to 4.81 Gy and 8.03 Gy, respectively, in a uniform 60Co radiation field prior to this study, a smaller variation in signal i ntensity was expected. The cumulative light sum curves revealed that 95% of the total natural signal intensity is emitted by 7.4% and 9.2% of the total grain population for samples CA 1 and CA 5, respectively. In combination with the estimated amount of grains on a sample disc it is estimated that the natural signal from 1, 2 and 8 mm aliquots is mainly derived from 3, 13 and 213 grains in average. Furthe rmore, for a typical 1 mm aliquot used in this study containing 40 â€“ 4 5 grains about 80% of the signal is likely to come from a single grain. After applying the test dose the proportion s of light emitting grains increase to 13.6% and 14.3%. As expected, the signal intensity distributions of the calibration quartz samples were more uniform. About 26.4% (RisoeQ) and 50.9% (BAG478) of the grains emit 95% of the bulk OSL signal. Table 4. Results of counting grains on single aliquot discs. Values for maximum packing density d were taken from Specht ( 2012 ). A mean packing density of 0.68 was calculated, which is significantly lower than the commonly assumed packing density of ~0.907, a value that can only be achieved in the densest hexagonal circle packing on an infinite plane (Chang and Wang , 2010). Grain size (m) Aliquot diameter (mm) Maximum packing density d Calculated max imum number of grains Number of coun ted al iquots Counted average number of grains Standard devi ation Packing density d 40 â€“ 63 1 0.85 320 3 387 10 1.03 40 â€“ 63 2 0.86 1304 3 807 62 0.54 100 â€“ 150 1 0.78 55 6 34 14 0.53 100 â€“ 150 2 0.83 221 6 180 36 0.70 100 â€“ 200 1 0.78 35 1 31 0.70 100 â€“ 200 2 0.82 146 1 101 0.57
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 116 Presence of oversaturating grains The same single grain data sets used to investigate the brightness distribution were systematically survey ed for â€œover saturatingâ€ (â€˜o sâ€™) grains (Stone and Bailey, 2012) , i.e. grains with a natural signal Ln/ Tn which does not intercept the dose response curve (Yoshida et al. , 20 00 and Jacobs et al. , 2003 ) . This kind of behaviour has been predicted in the dose absorption model of Bailey ( 2004 ) as a result of the difference in dose rates between natural and labor atory irradiation with regard to the R1 hole trapping centre. During natural irradiation the relatively thermally unstable R1 centre ( E = 1.43 eV) is in a state of low equilibrium concentration, but in a considerably higher equilibrium concentration during laboratory irrad iation. During the lat t er, the resulting increased compet ition for free electrons during irradiation effectively lo wers the dose response curve so that the natural dose point, unaffected by the laboratory dose rate, can be above the fitted asymptotic signal level (Bailey, 2004 ). The asymptotic regeneration value was established for those grains for which a meaningful exponential growth curve could be fitted. To identify and distinguish â€˜o sâ€™ grains from those grains whose natural signal is above the dose response curve due to random counting errors, the natural signal had to be at least two standard deviations above the asymptotic regeneration value. Proportion and brightness of over saturating grains Out of the 2728 grains measured in total of sample CA 1 only 20 grains (0.73%) have shown a similar b ehavi our attributed to â€˜o sâ€™ grains. A further 10 grains (0.37%) were close to saturation, but had an Ln/ Tn value that still intercepted the dose response curve in the satur ation area above 2 D0 and hence were not classified as â€˜o sâ€™ grains. For sample CA 5 only 15 grains ( n = 2598) were identified as â€˜o sâ€™ grains (0.5 8 %). A comparable amount of grains ( n = 16) had an Ln/ Tn value that intercepted the dose response curve above 2D0 (0.62%). Dividing the sum of net OSL intensities of all â€˜o sâ€™ grains in a sample by the total light sum revealed that the â€˜o sâ€™ grains are responsible for 32% (CA 1) and 11% (CA 5 ) of the total emitted OSL signal. While on the single grain scale â€˜o sâ€™ grains can easily be identified and discarded from further analysis, for multiple grain aliquots the presence of â€˜o sâ€™ grains is expected to lead to an overestimation of the De due to their disproportional contribution to the natural signal compared to the regenerated signals (Yoshida et al. , 2000 ). Howev er, the impact of â€˜o sâ€™ grains on mult iple grain De estimates depends on the probability of i nclusion, the degree of over saturation and the relative brightness (Stone and Bailey, 2012 ). While the contrib ution of all â€˜o sâ€™ grains to the total light sum appears large, the â€˜o sâ€™ grain OSL intensities were found to vary in orders of magnitude. While the brightest â€˜osâ€™ grain of sample of CA 1 had a net OSL signal of ~400000 counts in the first 0.05 s of stimulation, the dimmest â€˜o sâ€™ grain only had ~400 counts. Hence, it appears that the contrib ution of â€˜o sâ€™ grains to the total light sum is not evenly shared by all â€˜o sâ€™ grains, but rather dominated by a smaller fraction of disproportional bright grains. While less t han 0.8% of all measured grains were identified as â€˜o sâ€™ grains and despite the low probability of an â€˜osâ€™ grain on a small aliquot (<20%) when assuming a binomial distribution, the influence of â€˜o sâ€™ grains on De estimates may very well increase for larg er al iFig. 6 . Proportion of grains emitting luminescence, visualised as a) a cumulative light sum curve after Duller et al. ( 2000 ), and as b) the absolute brightness of each grain as described by Duller ( 2006 ). Both plots show the brightness distribution of the natural signal Ln and after the natural test dose T n of 15 Gy. The dashed line represents a population of grains with all grains having the same brightness. The cumul ative light sum curves revealed that 95% of the total natural signal intensity is emitted by 7.4% and 9.2% of the total grain population for samples CA -1 and CA -5, respe ctively. A typical 1 mm aliquot used in this study that contains 40 45 grains about 80% of the signal is likely to come from a single grain. After applying the test dose proportion of light emitting grains increase to 13.6% and 14.3%. NetOSLsignal (counts / 0.05 s)01020304050607080901000102030405060708090100 Proportionof totallightsum (%) CA-1 (Ln)CA-1 (Tn= 15 Gy) CA-5 (Ln)CA-5 (Tn= 15 Gy) RisoeQ (Ln) BAG478 (Ln)a)b) CA-1 (Ln)CA-1 (Tn= 15 Gy) CA-5 (Ln)CA-5 (Tn= 15 Gy) RisoeQ (Ln) BAG478 (Ln) Proportion of grains (%)Proportion of grains (%)
C. Burow et al. 117 quots. For a 1 mm aliquot covered by 40 â€“6 3 m grains (CA 10) there is only a 18% probability that none of the 240 grains is an â€˜o sâ€™ grain. In contrast, there is a prob ability of 31% of one â€˜osâ€™ grain, 26% of two â€˜osâ€™ grains and 14% of three â€˜o sâ€™ grains to be present o n the aliquot. For CA 1 virtually all large 8 mm aliquots (2662 grains per disc) contain at least five â€˜o sâ€™ grains. The effect of over saturating grains on synthetic aliquot De To further investigate the effect of â€˜o sâ€™ grains on multiple grain De estimates synthetic aliquots were co nstructed by summing up the signals of all grains of a single grain disc where â€˜o sâ€™ grains were identified. For sample CA 1 13 synthetic aliquots were constructed where each disc contained one â€˜o sâ€™ grain. The relative contribution of the â€˜o sâ€™ grain to the total light sum of each synthetic aliquot varied between 1 to 55%, while the proportion of grains emitting 95% of the total light sum ranged from 6 to 28% with an average of 12. 2 7 .3%. For three of the synthetic aliquots the â€˜o sâ€™ grains were bright enough to dominate the dose response, so that no De could be calculated. Similar values were determined for the synthetic aliquots of sample CA 5 ( n = 7), where the relative contribution of â€˜o sâ€™ grains ranged from 3 to 64% and the proportion of grains emitting 95% of the total light sum varied from 3 â€“2 0% with an average of 10. 0 5 .7%. The obtained De values from the multiple grain synthetic aliquots were then compared to the De that was obtained when excluding the â€˜o sâ€™ grains ( Fig. 7 ). Exclusion of these grains revealed that the presence of â€˜o sâ€™ on a multiple grain aliquot directly effects the calcula ted De value. After rem oval of the â€˜osâ€™ grains from the synthetic aliquot the maximum observed decrease in De was ~66%. It further appears that for the samples under study the relative decrease in De is linearly proportional to the relative contribution of the â€˜o sâ€™ grain to the total light sum ( Fig. 7 ). Another observation is that almost half of the identified â€˜o sâ€™ grains only amount for less than 12% of the total light sum and hence lead to a decrease in De of similar proportion after removal from the synthetic aliquot. Most of the remaining â€˜o sâ€™ grains lead to a d ecrease in De of 10 â€“2 5% after removal, while two distinct outliers more than halved the De of the synthetic aliquot due to their high contribution to the total light sum of about 65%. As a c onsequence, the presence or absence of â€˜o sâ€™ grains has to be considered as an additional source of intrinsic scatter between multiple grain aliquots when explaining the observed scatter in De distributions of the samples under study. 5. DISCUSSION Dose distributions The observed dispersion in natural dose distributions is a composite of intrinsic and extrinsic sources of error. The former arises e.g. from variations in photon counting statistics and curve fitting uncertainties (Galbraith and Roberts, 2012). In the absence of any extrinsic factors such as partial bleaching, post depositional sediment mixing or variations in beta microdosimetry, and when all intrinsic sources of uncertainty have been determine d and properly accounted for, no spread in dose distributions should be observed. However, it is common that even in dose recovery experiments, where any extrinsic sources of scatter that affect natural samples are effectively excluded, and after taking all intrinsic errors into account, a substantial spread in De values remains (Galbraith et al., 2005 ). To assess whether the assigned uncertainties on the individual De estimates are sufficient to explain the o bserve d variability we calculated the overdispersion using the central age model. For the quartz samples OD values ranged from 11% to 33%. The general assumption of less variation in De with an increasing number of simultaneously measured grains (Duller, 2008 and Arnold and Roberts, 2009 ) also applied to this study ( Fig . S 4 in the Appendix). Single grain De distributions exhibit the hig hest mean OD (29%), which continuously decreases for multi grain aliquots with increasing number of simult aneously measured grains. The decrease in OD approx imately follows the prediction of Cunningham et al. ( 2011). However, in te rms of absolute values, the o bserved OD in single grain and multiple grain De distrib uFig. 7. Effect of over -saturating (â€˜o -sâ€™) grains on the De of synthetic multiple grain aliquots. For each of the single grain discs where an â€˜o sâ€™ grain was identified the single grain signals of all grain hole positions were summed up and a D e value was calculated before and after excluding the â€˜o -sâ€™ grain from the synthetic aliquot. The decrease in De , expressed as the ratio of the D e without the â€˜o-sâ€™ grain and the D e including all grains, is plotted against the contribution of the â€˜o sâ€™ grain to the total light sum of the synthetic aliquot. An apparently linea r proportionality between light sum contribution and decrease in D e was observed for the samples under study.
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 118 tio ns is higher than those summaris ed in Arnold and Roberts ( 2009). In any case, the calculated OD values hence indicate that the observed spread in dose distrib utions cannot be explained by intrinsic uncertainties alone. This may either be due to the fact that the intrinsic vari ability has been underestimated or that additional intrinsic and extrinsic sources of scatter have to be considered. Extrinsic sources of uncertainty Samp les analysed in this study are known to be of fluvial origin (Angelucci et al. , 2013 ), hence partial bleaching might affect the De distributions to some e xtent. However, LM OSL measurements have shown a clear dominance of the fast component and routine a ssessment of De( t ) plots (Bailey, 2003 ) constructed for samples CA 1, CA 2 and CA 6 on different aliquot sizes and single quartz grains provide no evidence for a signif icant depe ndence of De on the integration interval over the initial part of the signal . Contamination of the integrated OSL signal used for De determination by a partially bleached medium component is hence regarded unlikely. Another method for identifying partial bleaching of quartz from samples CA 1 and CA 5 was applied by measuring the much harder to bleach pIRIR signals of feldspars (Thomsen et al. , 2008b and Buylaert et al. , 2012 ; see Fig . S 5 in the Appendix ). As dose rates to quartz and feldspar and hence the De differ due to the internal K content of feldspar we calculated the age of these samples in order to compare the OSL and pIRIR results. Prior to that r esidual doses of 3.7 Gy and 11.4 Gy, determined on the modern sample CA 7 , were su btracted from the pIRI R225 and pIRIR290 equivalent dose, respectively. Applying the homogeneity test after Ga lbraith ( 2003 ) indicate s that t he pIRIR290 age (74. 1 7 .9 ka) for sample CA 1 almost perfectly agrees ( P = 0.96) with the quartz OSL ages (70. 9 7 .6 ka and 74. 1 7 .9 ka). However, the pIRIR225 age (54. 5 4 .9 ka) seems to underestimate the burial age ( P = 0.06). For sample CA 5 the pIRIR225 age (83. 7 8 .9 ka) is slightly larger than the quartz OSL age estimates (69. 9 7 .6 ka and 68. 7 6 .8 ka) , but all age estimates still ag ree with each other ( P = 0.36). The pIRIR290 signal is regarded to yield the more reliable age estimate due to its higher stability over the pIRIR225 signal (Thiel et al. , 2011 ; Thomsen et al. , 2011 and Buylaert et al. , 2012 ). Note that no additional tests to correct for anomalous fading ( e.g. Huntley and Lamothe, 2001 ; Auclair et al. , 2003 and Kars et al. , 2008 ) were conducted as fading correction methods are problematic (e.g. Wallinga et al. , 2007 and Reimann et al. , 2011). However, s ince the quartz OSL age estimates from single grains and small aliquots perfectly agree with the age derived from the much harder to bleach pIRIR290 signal, this gives strong support to a well bleached quartz OSL signal. This confirm s the assumption that partial bl eaching is not necessarily an impediment to obtaining acc urate chronologies for Late Pleistocene alluvial samples (Jain et al. , 2004 ; Martins et al. , 2010 ; Sohbati et al. , 2012 and Medialdea et al. , 2014). Another source of extrinsic variability might arise from post depositional sediment mixing that is often associated with archaeological deposits (cf. Jaco bs and Roberts, 2007). However, the single grain De distrib utions from sample CA 1 and CA 5 provide no evidence for distinct dose components that might be related to the intrusion of younger or older grains ( Fig. 8) . The lack of Fig. 8 . Single grain De distributions of samples CA -1 and CA -5 visualised in a combined radial plot (Galbraith, 1988 ) and kernel density estimate (KDE) plot, generated with the plot_AbanicoPlot() function of the R package â€˜Luminescenceâ€™ (Kreutzer et al. , 2012 ) . The grey bar in the radial plot is centred on the weighted D e calculate d by the central age model (CAM, Galbraith et al. , 1999 ). The calculated overdispersion values indicate that the observed spread in the dose distributions cannot be explained by intrinsic uncertainties alone. Standardised estimate 051015 PrecisionRelative error (%) 20106.7 02 20406080100120140 Equivalent dose (Gy) 00.132Density CA-1Single grain200-250 mn = 50CAM De = 61.6 4.2 GyOD = 30.2 1.9% Standardised estimate 051015 PrecisionRelative error (%) 20106.7 02 20406080100 Equivalent dose (Gy) 00.112Density CA-5Single grain200-250 mn = 65CAM De = 46.9 2.9 GyOD = 28.1 1.5%
C. Burow et al. 119 post depositional modifications at Cueva Antn is also confirmed by Angelucci et al. ( 2013 ). The use of a finite mixture model (Galbraith and Green, 1990 and Roberts et al. , 2000) was hence considered to be inap propriate. The observed spread in the dose distributions may a lso partly be explained by small scale variations in the beta radiation dose to which the samples have been e xposed during burial (Olley et al., 1997 and Duller, 2008). Modelling studies (e.g. Nathan et al. , 2003) have shown that beta heterogeneity may significantly influence De distributions. However, there is no straightforward met hod to identify beta dose heterogeneities and their impact on dose estimates (cf. Gurin et al. , 2013 ). At this stage we cannot assess whether the observed overdispersion can at least partly been explained by microdosimetry or not. Considering that our samples were probably not affected by partial bleaching, the CAM may yield the most appropriate burial dose estimate if beta dose heter ogeneity is probable (Gurin et al. , 2013). Intrinsic sources of uncertainty While extrinsic sources of error potentially account for some of the observed scatter, fur ther intrinsic factors may have not been accounted for yet. While low OSL intensity of quartz (cf. Duller, 2006 ; Preusser et al. , 2006 and Klasen et al. , 2007 ) as a source of additional scatter to De distributions can be excluded due to the bright si gnals, it was shown that even for heated and homogenously irradiated quartz a n OD of 8 â€“ 1 2% remained in single grain De datasets after correcting for the spatial non uniformity of the beta sources. This additional intrinsic uncertainty is comparable to values reported earlier by Thomsen et al. ( 2005, 2007 ) and Reimann et al. ( 2012b ). However, unlike Thomsen et al. ( 2007) and Reimann et al. ( 2012b) we did not add this uncertainty onto the u ncertainties of the single grain De estimates because this value was derived from a gamma dose recovery exper iment for calibrating the beta sources on calibration samples of different provenance. Beyond that, recent single grain studies have demonstrated t hat different â€˜grain behavioural typesâ€™ with regard to luminescence properties such as brightness and dose response characteristics can have a significant impact on both single and multiple grain dose distributions and have to be considered as additional s ources of intrinsic scatter (e.g. Yoshida et al. , 2000 ; Duller, 2012 ; Stone and Ba iley, 2012 and Demuro et al. , 2008 , 2013). By single grain measurements the variability of the saturation dose D0 and the presence of â€˜over saturatingâ€™ was investigated. The former has been reported to affect the ability to su ccessfully recover a laboratory dose in a dose recovery experiment as the given dose approaches or exceeds the saturation dose (Duller, 2012). A mean D0 of ~63 Gy was determined for quartz from Cueva Antn, whi ch falls within the range of saturation doses of quartz reported in Wintle and Murray ( 2006). Despite the large variability of D0 values similar to the values reported by Duller ( 2012) all conducted dose recovery experiments on single grains and multiple grain aliquots performed well ( Table 3 ). Hence, the observed variability in D0 values is not expected problematic for the samples from Cueva Antn. Fol lowing the suggestion of Stone and Bailey ( 2012) the single grain data sets were monitored for the presence of â€˜o sâ€™ grains. By making synthetic aliquots consisting of 100 grains each it was assessed that most â€˜o sâ€™ grains only have a minor influence on the calculated De. Ho wever, some â€˜o sâ€™ grains more than doubled the De or were even so bright to saturate the synthetic aliquot so that no De could be calculated. While â€˜o sâ€™ grains are easily r ejected from single grain data sets, there is no formal ind ication whether a multiple grain aliquot and its De that otherwise passes all rejection criteria is affected by the presence of one or more â€˜o sâ€™ grains. Hence, for multiple grain aliquots systematically higher De and age estimates should be expected (Stone and Bailey, 2012 ). However, statistically indistinguishable single grain and small al iquot CAM De estimates of 61. 6 4 .2 Gy and 61. 5 3 .4 Gy for CA 1 and 46. 9 2 .9 Gy and 48. 9 2 .8 Gy for CA 5, respectively, were obtained. It is only for the 8 mm aliquots of CA 1 that a significantly higher mean De 74. 7 4 .3 Gy was calculated. A similar behaviour was observed for sample CA 10, where 1 mm aliquots covered with 100 â€“ 1 50 m quartz grains yielded a De of 44. 4 2 .7 Gy, but where the same aliquot size covered with 40â€“ 6 3 m quartz grains yielded a conside rably higher De of 62. 4 3 .6 Gy. Assuming that the De increase can indeed be attributed to the presence of â€˜o sâ€™ grains, t hey would be responsible for an increase of about 1 2 â€“ 1 8 Gy. Considering that single grain and small aliquot De distributions of CA 1 and CA 5 yield indistinguish able mean De estimates it appears that the presence of â€˜o sâ€™ grains only has a relevant impact for larger aliquot s (cf. Stone and Bailey, 2012). While it is not possible to definitely state the causes for the observed spread in data and the De and age ove re stimation of larger multiple grain aliquots, the single grain and synthetic aliquot analyses indicate that multiple grain averaging effects are a complex system of various (aberrant) OSL signal characteristics. Here, these pro blematic averaging effects seem to become noticeable when the number of simulta neously measured grains exceeds a certain threshold value. Small aliquot De est imates are seemingly not affected by the presence of â€˜o sâ€™ grains, which might be the result of a low probability of having an â€˜o sâ€™ grain and an overall low percentage of bright grains that contribute to the total light sum. When a small aliquot contains one or more very bright â€˜osâ€™ grains, which have been shown to have the highest i mpact on the De about equal to their share of the total light sum, there is a higher probability that the â€˜o sâ€™ grain even saturates the small aliquot with only a few other grains. As a result, the small aliquot would not pass the rejection criteria. But as the amount of simultaneously measured grains increases, it is less likely that the â€˜o sâ€™ grain s is
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 120 able to saturate the multiple grain aliquot. Depending on its relative contribution to the multiple grain OSL signal the â€˜o sâ€™ may then lead to a systematic overestimation of the equivalent dose. Palaeodose and age estimates Concluding the previous sections we consider the CAM as the most appropriate age model to calculate an equivalent dose representative for the â€˜trueâ€™ burial age of the samples from Cueva Antn. Furthermore, only the quartz single grain and small aliquot (1 mm and 2 mm with 100 â€“ 1 50/200 m grains) data sets are considered to yield reliable age estimates, as larger aliquots are suppo sedly affected by a systematic overestimation of the De due to the presence and influence of over saturating grains. CAM De estima tes obtained from these data sets show only little variation on a sample to sample basis and no obvious correlation with sample depth or stratigraphic order. The CAM De values range from 44 Gy to 75 Gy and yield luminescence ages between 69 ka and 82 ka. R esults of equivalent dose determination and age calcul ation are summarised in Table 5 . Archaeological and sedimentological implications A coherent quartz OSL based chronostratigraphy for the archaeological succession (AS2 AS5) at Cueva Antn is provided ( Fig. 9 ). Due to the highly significant agre ement of age estimates ( P = 0.95), an error weighted mean and associated standard deviation of 71. 0 1 .8 ka is suggested for unit II e. Likewise, weighted mean ages of 69. 2 0 .9 ka ( P = 0.91) and 72. 1 5 .8 ka ( P = 0.42) for unit III f and III m are suggested, respectively. Taki ng all age estimates into account, the high agreement of age estimates ( P = 0.99) reduces to an error weighted mean of 72. 0 4 .2 ka for sub complexes AS2 to AS5. The age estimates for unit II e (~71 ka) at the top of AS2 clearly contradicts the ABA radiocarbon age of 39. 7 0 .6 ka BP (~43.5 ka cal BP, OxA18672; Zilho et al. , 2010) obtained from the underlying unit II h/i. Ho wever, the radiocarbon date is regarded as a preliminary minimum age estimate for unit II h/i (Zilho et al. , 2010). The very low variation in quartz OSL age estimates throughout the entire archaeological succession suggests a fairly rapid fluvial accumulation of the sediments, which is in very good accord with the sedimentological and micromorphological findings of Angelucci et al. ( 2013). However, given that all age estimates are statist ically indistinguishable it is neither possible to discern individual flood events nor to constrain the involved time span. Furthermore, the suggested time frame for the depos ition of the middle to lower part of the alluvial succession also implies a major hiatus between sub complexes AS2 and AS1, which are separated by an erosional cut. Ho wever, the interpretation of such a large hiatus in the sed imentary record remains open and requires additional investigation. 6. CONCLUSIONS The aim of this study was to establish a reliable chronostratigraphy for the archaeological succession from sub complexes AS2 to AS5 of Cueva Antn. For that purpose, the most reliable luminescence dating technique for estimating the age of fluvial deposits in a rock shelter was ascertained through a series of laboratory exper iments and by cross validating the performance of the dating methods used. From a methodological perspective the following conclusions can be drawn: i) The multiple grain approach utilising quartz grains on small (preferably 1 mm) al iquots is a viable method to determine the burial age of f luvial deposits in rock shelters that are unaffected by post depositional mixing. ii) While the same results on the interpretation of the succession at Cueva Antn would have been achieved if exclusively small aliquots were used, s ingle grain measurements of quartz provided val uable information for the analysis of De distributions . St atistically indistinguishable De and age estimates from Table 5 . Results of equivalent dose determination and age calculation . Q = quartz . Unit Lab . Code Sample Mineral Aliquot diameter (mm) Grain size fraction (m) N r of aliquots/grains accepted/measured Age model Over dispersion (%) Equivalent dose (Gy) Total dose rate (Gy ka 1 ) Luminescence age (ka) II e C L2941 CA 1 Q 2 100 â€“ 150 42/50 CAM 15.3 0.7 61.53.4 0.870.10 70.97.6 Q SG 200 â€“ 250 50/3100 CAM 30.21.9 61.64.2 0.840.11 72.97.7 II e C L3137 CA 2 Q 1 100 â€“ 150 51/60 CAM 29.31.6 66.94.4 0.970.12 69.37.7 II C L3375 CA 9 Q 1 100 â€“ 150 46/58 CAM 22.61.2 55.03.3 0.670.08 82.28.0 II y C L2942 CA 4 Q 2 100 â€“ 200 41/48 CAM 17.00.8 75.14.3 1.080.13 69.47.2 III f C L2943 CA 5 Q 2 100 â€“ 200 45/48 CAM 17.90.8 48.92.8 0.700.09 69.97.6 Q SG 200 â€“ 250 65/2800 CAM 28.11.5 46.92.9 0.680.08 68.76.8 Q 1 100 â€“ 150 46/62 CAM 23.51.3 44.42.7 0.620.08 72.27.8 III k/l C L3377 CA 11 Q 1 100 â€“ 150 60/84 CAM 27.11.3 49.03.0 0.660.08 74.17.9 III m C L3138 CA 6 Q 1 100 â€“ 150 66/84 CAM 32.71.7 66.94.3 0.860.12 77.59.8 III m C L3378 CA 12 Q 1 100 â€“ 150 45/70 CAM 25.01.4 58.43.7 0.850.10 69.17.3
C. Burow et al. 121 single grains and small aliquots give strong support for the potential of small aliquot measurements for studies in comp arable contexts. Furthermore, single grain mea surements allowed the detection of aberrant grain beha viour and by constructing synthetic aliquots their impact on multiple grain De estimates c ould be quantified. iii) The presence of â€˜over saturatingâ€™ grains, i.e. grains whose natural signal is significantly above the dose r esponse curve, affected multiple grain De estimates and can lead to a systematic overestimation of the De as the number of simultaneously measured grains increases. Besides the probability of having an â€˜osâ€™ grain on the aliquot, the degree of overestimation appears to be pr imarily controlled by the relative contribution of the â€˜o sâ€™ grain to the total light sum of a multiple grain aliquot . iv) The comparison of different luminescence dating tec hniques utilising different minerals and luminescence signals with varying bleaching rates is a powerful tool to identify partial bleaching and to strengthen the reliability of the obta ined chronologies (cf. Murray et al. , 2012). Finally, the results of the present study are a valuable contribution to the chronological framework and interpr etation of the Middle to Upper Palaeolithic transition in the Iberian Peninsula. The quartz OSL chronology for the middle to lower parts of the fluvial succession at Cueva Antn suggests a fairly rapid deposition of the sediments during late MIS 5 and possibly throughout MIS 4. Consequently, Mousterian occupations o f the rock shelter primarily documented in the lower parts of the succession fall wit hin that particular time frame. Fig. 9 . OSL chronostratigraphy for the archaeological succession (AS2AS5) at Cueva Antn using multiple and single grains of quartz. GISP2 oxygen isotope curve from Grootes et al. ( 1993) with timescale of Meese et al. (1997 ). Approximate timing of Heinrich Event 6 (HE 6) taken from Hemming ( 2004) (see also Cayre et al. , 1999 and references therein). Dashed lines on y -axis denote unlisted archaeological units (see Angelucci et al. , 2013 for a comprehensive list). Grey bars on x axis are the standard deviation (dark) and twice the standard deviation (light) on the error weighted mean of all age estimates. Inset: Radial plot (Galbraith, 1988 ) of quartz OSL age estimates. The reference value is the error weighted mean of age estimates. The grey bar denotes the 2 (95% confidence interval) band on the reference value. II-eII-II-yIII-fIII-e/hIII-k/lIII-mUnit(s) Stratigraphic sub-complex SampleAS5 AS4 AS2 GISP218O ()MIS3MIS4MIS5 HE6? Standardised EstimatePrecisionRelative Error (%)361812903691212088725640
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 122 APPENDIX Five figures are available as Supplementary Material in electronic version of this article at http://dx.doi.org/10.1515/geochr 20150010. Fig. S 1: 3D Plots of beta source dose rate distributions from three Ris TL/OSL readers; Fig. S 2 : Deconvolved LMOSL curves from samples CA 1 and CA 6; Fig. S 3: Number of mineral grains o n a sample disc as a function of the al iquot diameter, mean grain size and packing density; Fig . S 4: Dependency of overdispersion on the amount of grains on a sample disc; Fig . S 5: Bleaching rates of the pIRIR225 and pIRIR290 signals from sample CA 1. ACKNOWLEDGEMENTS Arch a eological fieldwork and research at Cueva Antn were funded, over the years, by the Fundacin Sneca (Murcia), the University of Murcia, the Spanish Ministerio de Ciencia e Innovacin (grant HAR2011 24878), the Leakey Foundation, the Direccin General del Medio Natural de la Regin de Murcia and the M unicipality of Mula, and actively supported by the Museo de Arte Ibrico El Cigarralejo de Mula. We are grateful for the financial support by the C1 and F2 projects of the CRC 806 â€œOur Way To Europe. Culture Environment Interaction and Human Mobility in the Late Quaternaryâ€, funded by the German Research Foundation (DFG). Melanie Bartz is thanked for providing additional data on the amount of grains on a sample disc. REFERENCES Aitken MJ, 1 985. Thermoluminescence Dating. Academic Press, Lo ndon. Angelucci DE, Anesin D, Susini D, Villaverde V, Zapata J and Zilho J, 2013. Formation processes at a high resolution Middle Paleolithic site: Cueva Antn (Murcia, Spain). Quaternary International 315 : 24â€“ 4 1, DOI 10.1016/j.quaint.2013.03.014. Arnold LJ and Roberts RG, 2009. Stochastic modelling of multigrain equivalent dose ( De) distributions: Implications for OSL dating of sediment mixtur es. Quaternary Geochronology 4: 204 â€“ 2 30, DOI 10.1016/j.quageo.2008.12.001. Arnold LJ, Demuro M and Navazo Ruiz M, 2012. Empirical insights into multi grain averaging effects from â€˜pseudoâ€™ singl egran OSL measurements. Radiation Measurements 47 : 652â€“ 6 58, DOI 10.1016/j.radmeas.2012.02.005. Auclair M, Lamothe M and Hout S, 2003. Measurement of anomalous fading for feldspar IRSL using S AR. Radiation Measurements 37: 487â€“ 4 92, DOI 10.1016/S13504487(03)000180 . Bailey RM, 2003. Paper I: The use of measurement time dependent singlealiquot equivalentdose estimates from quartz in the identification of incomplete signal resetting. Radiation Measurements 37: 673 â€“ 6 83, DOI 10.1016/S13504487(03)000787 . Bailey RM, 2004. Paper I â€“ simulation of dose absorption in quartz over geological timescales and its implications for the precision and accuracy of optical dating. Radiation Measurements 38: 299â€“ 3 10, DOI 10.1016/j.radmeas.2003.09.005. Ballarini M, Wintle AG and Wallinga J, 2006. Spatial variation of dose rate from beta sources as measured using single grains. Ancient TL 24(1): 1â€“ 8 . Btter Jensen L, 1997. Lumin escence techniques: instrumentation and methods. Radiation Measurements 27(5/6) : 749 â€“ 7 68, DOI 10.1016/S13504487(97)002060 . Btter Jensen L, Duller GAT, Murray AS and Banerjee D, 1999. Blue light emitting diodes for optical stimulation of quartz in retrospective dosimetry and dating. Radiation Protection Dosimetry 84: 335â€“ 3 40. Btter Jensen L, Bulur E, Duller GAT and Murray AS, 2000. Advances in luminescence instrument systems. Radiation Meas urements 32: 523â€“ 5 28, DOI 10.1016/S13504487(00)000391 . Btter Jensen L, Andersen CE, Duller GAT and Murray AS, 2003. Developments in radiation, stimulation and observation facilities in luminescence measurements. Radiation Measurements 37: 535â€“ 5 41, DOI 10.1016/S13504487(03)000209 . Btter Jensen L, Thomsen KJ and Jain M, 2010. Review of optically stimulated luminescence (OSL) instrumental developments for retrospective dosimetry . Radiation Measurements 45: 253 â€“ 2 57, DOI 10.1016/j.radmeas.2009.11. 030. Brock F, Higham T, Ditchfield P and Ramsey CB, 2010. Current pr etreatment methods for AMS radiocarbon dating at the Oxford R adiocarbon Accelerator Unit (ORAU). Radiocarbon 52(1) : 103â€“ 1 12. Buylaert J P, Murray AS, Thomsen KJ and Jain M, 2009. Testing the potential of an elevated temperature IRSL signal from Kfeldspar. Radiation Measurements 44(5) : 560â€“ 5 65, DOI 10.1016/j.radmeas.2009.02.007. Buylaert J P, Jain M, Murray AS, Thomsen KJ, Thie l C and Sohbati R, 2012. A robust feldspar luminescence dating method for Middle and Late Pleistocene sediments. Boreas 41: 435 â€“ 4 51, DOI 10.1111/j.15023885.2012.00248.x. Chang H C and Wang L C, 2010. A simple proof of Thueâ€™s Theorem on Circle Packing. WEB site: . Accessed 2014 July 24. Cunningham AC, Wallinga J and Minderhoud PSJ, 2011. Expectations of scatter in equivalentdose distributions when using multi grain aliquots for OSL dating. Geochronometria 38(4): 424 â€“ 4 31, DOI 10.2478/s13386011 0048z. Cayre O, Lancelot Y and Vincent E, 1999. Paleoceanographic reco nstruction from planktonic foraminifera off the Iberian Margin: Temperature, salinity, and Heinrich events. Paleoceanography 14(3): 384â€“ 3 96, DOI 10.1029/1998PA900027. Demuro M, Roberts RG, Froese DG, Arnold LJ, Brock F and Bronk Ramsey C, 2008. Optically stimulated luminescence dating of single and multiple grains of quartz from perennially frozen loess in western Yukon Territory, Canada: Comparison with r adiocarbon chronologies for the late Pleistocene Dawson tephra. Quaternary Geochronology 3: 346â€“ 3 64, DOI 10.1016/j.quageo.2007.12.003. Demuro M, Arnold LJ, Froese DG and Roberts RG, 2013. OSL dating of loess deposits bracketing Sheep Creek tephra beds, northwest Canad a: Dim and problematic singlegrain OSL characteristics and their effect on multi grain age estimates . Quaternary Geochronology 15: 67â€“ 8 7, DOI 10.1016/j.quageo.2012.11.003. Duller GAT, 2003. Distinguishing quartz and feldspar in single grain luminescence measurements. Radiation Measurements 37: 161 â€“ 1 65, DOI 10.1016/S13504487(02)001701 . Duller GAT, 2006. Single grain optical dating of glacigenic deposits. Quaternary Geochronology 1: 296â€“ 3 04, DOI 10.1016/j.quageo.2006.05.018. Duller GAT, 2008. Singlegrain optical dating of Quaternary sediments: why aliquot size matters in luminescence dating. Boreas 37: 589 â€“ 6 12, DOI 10.1111/j.15023885.2008.00051.x. Duller GAT, 2012. Improving t he accuracy and precision of equivalent doses determined using the optically stimulated luminescence signal from single grains of quartz. Radiation Measurements 47: 770â€“ 7 77, DOI 10.1016/j.radm eas.2012.01.006. Duller GAT, Btter Jensen L, Murray AS and Truscott AJ, 1999. Single grain laser luminescence (SGLL) measurements using a novel automated reader. Nuclear Instruments and Methods in Physics R esearch B 155: 506 â€“ 514, DOI 10.1016/S0168583X(99)004887 . Duller GAT, Btter Jensen L and Murray AS, 2000. Optical dating of single sand sized grains of quartz: sources of variability. Radiation
C. Burow et al. 123 Measurements 32: 453â€“ 4 57, DOI 10.1016/S13504487(00)00055X. Dunne J, Elmore D and Muzikar P, 1999. Scaling factors for the rates of production of cosmogenic nuclides for geometric shielding and a ttenuation at depth on slope d surfaces. Geomorphology 27: 3â€“ 1 1, DOI 10.1016/S0169555X(98)000865 . Fletcher WJ and Snchez Goi MF, 2008. Orbitaland suborbitalscale climate impacts on vegetation of the western Medite rranean basin over the last 48,000 yr. Quaternary Research 70: 451 â€“ 4 64, DOI 10.1016/j.yqres.2008.07.002. Galbraith RF, 1988. Graphical Display of Estimates Having Differing Standard Errors. Tech nometrics 30(3): 271â€“ 2 81, DOI 10.1080/00401706.1988.10488400. Galbraith RF, 2003. A simple homogeneity test for estimates of dose obtained using OSL. Ancient TL 21(2): 75â€“ 7 7. Galbraith RF and Green PF, 1990. Estimating the component ages in a finite mixture. Nuclear Tracks and Radiation Measurements 17(3): 197â€“ 2 06, DOI 10.1016/13590189(90)90035V. Galbraith RF and Roberts RG, 2012. Statistical aspects of equivalent dose and error calculation and display in OSL dating: An overview and some recommendations. Quaternary Geochronology 11: 1â€“ 2 7, DOI 10.1016/j.quageo.2012. 04.020. Galbraith RF, Roberts RG, Laslett GM, Yoshida H and Olley JM, 1999. Optical dating of single and multiple grains of quartz from Jinm ium rock shelter, Northern Australia: Part I, Experimental design and statistical models. Archaeometry 41(2): 339â€“ 3 6 4, DOI 10.1111/j.14754754.1999.tb00987.x . Galbraith RF, Roberts RG and Yoshida H, 2005. Error variation in OSL palaeodose estimates from single aliquots of quartz: a factorial experiment. Radiation Measurements 39: 289 â€“ 3 07, DOI 10.1016/j.radmeas.2004.03.023. Grootes PM, Stuiver M, White JWC, Johnsen S and Jouzel J, 1993. Comparison of oxygen isotope records from the GISP2 and GRIP Greenland ice cores. Nature 366: 552â€“ 5 54, DOI 10.1038/366552a0 . Gurin G, Mercier N and Adamiec G, 2011. Doserate conversion factors: update. Ancient TL 29(1): 5â€“ 8 . Gurin G, Murray AS, Jain M, Thomsen KJ and Mercier N, 2013. How confident are we in the chronology of the transition between Ho wiesonâ€™s Poort and Still Bay? Journal of Human Evolution 64: 314 â€“ 3 17, DOI 10.1016/j.jhevol.2013.01.006. Heer AJ, Adamiec G and Moska P, 2012. How many grains are there on a single aliquot? Ancient TL 30(1): 9 â€“ 1 6. Hemming SR, 2004. Heinrich Eve nts: Massive Late Pleistocene detritus layer of the north Atlantic and their global climate imprint. R eviews of Geophysics 42: RG1005, DOI 10.1029/2003RG000128. Higham T, 2011. European Middle to Upper Palaeolithic radiocarbon dates are often older than they look: Problems with previous dates and some remedies. Antiquity 85(327): 235â€“ 2 49. Higham T, Brock F, Peresani M, Broglio A, Wood R and Douka K, 2009. Problems with radiocarb on dating the Middle to Upper Palaeolithic transition in Italy. Quaternary Science Reviews 28: 1257â€“ 1 267, DOI 10.1016/j.quascirev.2008.12.018. Huang W and Ye T, 2011. Global optimization method for finding dense packings of equal circles in a circle. European Journal of Operational Research 210(3): 474â€“ 4 81, DOI 10.1016/j.ejor.2010.11.020. Huntley DJ and Baril MR, 1997. The K content of the K feldspars being measured in optical dating or in thermoluminescence dating. A ncient TL 15(1): 11 â€“ 1 3. Huntley DJ and Lamothe M, 2001. Ubiquity of anomalous fading in K feldspars and the measurement and correction for it in optical d ating. Canadian Journal of Earth Sciences 38: 1093 â€“ 1 106, DOI 10.1139/cjes 387 1093. Huntley DJ, Godfrey Smith DI and Thewalt MLW, 1985. Optical dating of sediments. Nature 313: 105 â€“ 1 07, DOI 10.1038/313105a0. IGME, 1972. Ma pa Geolgico de Espaa. Escala 1:50.000. 912(2636). Mula. Instituto Geolgico y Minero de Espaa, Madrid. Jacobs Z and Roberts RG, 2007. Advances in Optically Stimulated Luminescence Dating of Individual Grains of Quartz from Archeological Deposits. Evolutionary Anthropology 16: 210 â€“ 2 23, DOI 10.1002/evan.20150. Jacobs Z, Duller GAT and Wintle AG, 2003. Optical dating of dune sand from Blombos Cave, South Africa: II â€“ single grain data. Journal of Human Evolution 44: 613 â€“ 6 25, DOI 10.1016/S00472484(03)000496 . Jain M, Murray AS and Btter Je nsen L, 2004. Optically stimulated luminescence dating: how significant is incomplete light exposure in fluvial environments?. Quaternaire 15(1/2): 143â€“ 1 57. Jris O and Street M, 2008. At the end of the 14C time scale â€“ the Middle to Upper Paleolithic reco rd of western Eurasia. Journal of H uman Evolution 55: 782â€“ 8 02, DOI 10.1016/j.jhevol.2008.04.002. Kars RH, Wallinga J and Cohen KM, 2008. A new approach towards anomalous fading correction for f eldspar IRSL dating â€“ tests on samples in field saturation. Radiation Measurements 43: 786â€“ 7 90, 10.1016/j.radmeas.2008.01.021. Kars RH, Busschers FS and Wallinga J, 2012. Validating post IR IRSL dating on K feldspars through comparison with quartz OSL ages. Quaternary Geochronology 12: 74 â€“ 8 6, DOI 10.1016/j.quageo.2012.05.001. Klasen N, Fiebig M, Preusser F, Rei tner JM and Radtke U, 2007. Lum inescence dating of proglacial sediments from the Eastern Alps. Quaternary International 164 â€“ 1 65: 21â€“ 3 2, DOI 10.1016/j.quaint.2006.12.003. Kreutzer S, Schmidt C, Fuchs MC, Dietze M, Fischer M and Fuchs M, 2012. Introducing an R package for luminescence dating analysis. Ancient TL 30(1): 1â€“ 8 . Lai ZP, Zller L, Fuchs M and Brckner H, 2008. Alpha efficiency determination for OSL of quartz extracted from Chinese loess . R adiation Measurements 43: 767 â€“ 7 70, DOI 10.1016/j.radmeas.2008.01.022. Lapp T, Jain M, Thomsen KJ, Murray AS and Buylaert J P, 2012. New luminescence measurement facilities in retrospective dosimetry. Radiation Measurements 47: 803â€“ 8 08, DOI 10.1016/j.radmeas.2012.02.006. Lomax J, 2009. Palaeodunes as archives of environmental change â€“ A case study from the western Murray Basin (South Australia) based on optically stimulated luminescence (OSL) dating of single and multiple grains of quartz . Ph.D. Thesis, University of Cologne, Germany: 265pp. Lowick SE, Trauerstein M and Preu sser F, 2012. Testing the application of post IR IRSL dating to fine grain waterlain sediments. Quaternary Geochronology 8 : 33â€“ 4 0, DOI 10.1016/j.quageo.2011.12.003. Maroto J, Vaquero M, Arrizab alaga , Baena J, Baquedano E, Jord J, Juli R, Montes R, van der Plicht J, Rasines P and Wood R, 2012. Current issues in late Middle Palaeolithic chronology: New a ssessments from Northern Iberia: The Neanderthal Home: spatial and social behaviours. Quaternary International 247: 15â€“ 2 5, DOI 10.1016/j.quaint.2011.07.007. Martins AA, Cunha PP, Buylaert J P, Huot S, Murray AS, Dinis P and Stokes M, 2010. Kfeldspar IRSL dating of a Pleistocene river terrace staircase sequence of the Lower Tejo River (Portugal, western Iberia). Quaternary Geochronology 5: 176â€“ 1 80, DOI 10.1016/j.quageo.2009.06.004. Martnez C, 1997. El yacimiento musteriense de Cueva Antn (Mula, Murcia) (The Mousterian site of Cueva Antn (Mula, Murcia)). Memorias de Arqueologa de la Regin de Murcia 6: 18â€“ 4 7 (in Spanish). Medialdea A, Thomsen KJ, Murray AS and Benito G, 2014. Reliability of equivalentdose determination and agemodels in the OSL d ating of hi storical and modern palaeoflood s ediments . Quaternary Geochronology 22: 11â€“ 2 4, DOI 10.1016/j.quageo.2014.01.004 . Meese DA, Gow AJ, Alley RB, Zielinski GA, Grootes PM, Ram M, Taylor KC, Mayewski PA and Bolzan JF, 1997. The G reenland Ice Sheet Project 2 depth age scale: Methods and results. Journal of Geophysical Research 102(C12): 26411â€“ 2 6423, DOI 10.1029/97JC00269. Mejdahl V, 1979. Thermoluminescence dating: beta attenuation in quartz grains. A r chaeometry 21: 61â€“ 7 3 , DOI 10.1111/j.14754754.1979.tb00241.x.
LUMINESCENCE DATING OF FLUVIAL DEPOSITS IN THE ROCK SHELTER OF CUEVA ANTâ€N, SPAIN 124 Murray AS and Wintle AG, 2000. Luminescence dating of quartz using an improved single aliquot regenerativedose protocol. Radiation Measurements 32: 57â€“ 7 3, DOI 10.1016/S13504487(99)00253X. Murray AS and Wintle AG, 2003. The single aliquot regenerative dose protocol: potential for improvements in reliability. Radiation Measurements 37: 377â€“ 3 81, DOI 10.1016/S13504487(03)000532 . Murray AS, Thomsen KJ, Masuda N, Buylaert JP and Jain M, 2012. Identifying wellbleached quartz using different bleaching rates of quartz and feldspar luminescence signals. Radiation Measurements 47: 688 â€“ 6 95, DOI 10.1016/j.radmeas.2012.05.006. Nathan RP, Thomas PJ, Jain M, Murray AS and Rhodes EJ, 2003. Environmental dose rate heterogeneity of beta radiation and its implications for luminescence dating: Monte Carlo modelling and experimental validation . Radiation Measurements 37: 305 â€“ 3 13, DOI 10.1016/S13504487(03)000088 . Olley JM, Roberts RG and Murray AS, 1997. Disequilibria in the uranium decay series in sedimentary deposits at Allenâ€™s cave, Nullarbor Plain, Australia: Implications for dose rate determin ations. Radiation Measurements 27: 433â€“ 4 43, DOI 10.1016/j.jhevol.2013.01.006. Olley JM, Caitcheon G and Murray AS, 1998. The distribution of apparent dose as determined by optically stimulated luminescence in small aliquots of fluvial quartz: implications for dating young sediments. Quaternary Geochronology 17: 1033â€“ 1 040, DOI 10.1016/S02773791(97)000905 . Poolton NRJ, Wallinga J, Murray AS, Bulur E and Btter Jensen L, 2002. Electrons in feldspar I: on the wavefunction of electrons trapped at simple lattice defects. Physics and Chemistry of Minerals 29: 210 â€“ 2 16, DOI 10.1007/s0026900102173 . Prescott JR and Hutton JT, 1988. Cosmic ray and gamma ray dosimetry for TL and ESR. Nuclear Tracks and Radiation Meas urements 14(1/2): 223â€“ 2 27, DOI 10.1016/1359 0189(88)900696 . Prescott JR and Hutton JT, 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: large depths and long term time variations. Radiation Measurements 23(2/3): 497â€“ 5 00, DOI 10.1016/13504487(94)900868 . Preusser F, Andersen BG, Denton GH and Schlchter C, 2005. Lum inescence chronology of Late Pleistocen e glacial deposits in North Westland, New Zealand. Quaternary Science Reviews 24: 2207 â€“ 2 227, DOI 10.1016/j.quascirev.2004.12.005. Preusser F, Ramseyer K and Schlchter C, 2006. Characterisation of low OSL intensity quartz from the New Zealand Alps. Radiation Measurements 41: 871â€“ 8 77, DOI 10.1016/j.radmeas.2006.04.019. Reimann T, Tsukamoto S, Naumann M and Frechen M, 2011. The potential of using Krich feldspars for optical dating of young coastal sediments â€“ A test case from Darss Zingst peninsula (southern Baltic Sea coast) . Quaternary Geochronology 6: 207 â€“ 2 22, DOI 10.1016/j.quageo.2010.10.001. Reimann T, Thomsen KJ, Jain M, Murray AS and Frechen M, 2012a. Singlegrain dating of young sediments using the pIRIR signal from feldspar. Quaternary Geochronology 11: 28â€“ 4 1, DOI 10.1016/j.quageo.2012.04.016. Reimann T,Lindhorst S, Thomsen KJ, Murray AS and Frechen M, 2012b. OSL dating of mixed coastal sediment (Sylt, German Bight, North Sea). Quaternary Geochronology 11: 52 â€“ 6 7, DOI 10.1016/j.quageo.2012.04.006. Rhodes EJ, 2007. Quartz single grain OSL sensitivity distributions: implications for multiple grain single aliquot dating. Geochron ometria 26: 19â€“ 2 9, DOI 10.2478/v1000300700025 . Rittenour TM, 2008. Luminescence dating of fluvial deposits: applications to geomorphic, palaeoseismic and archaeological research. Boreas 37: 613 â€“ 6 35, DOI 10.1111/j.15023885.2008.00056.x . Roberts HM, 2012. Testing PostIR IRSL protocols for minimising fading in feldspars, using Alaskan loess with independent chron ological control. Radiation Measurements 47: 716 â€“ 7 24, DOI 10.1016/j.radmeas.2012.03.022. Roberts RG, Galbraith RF, Yoshida H, Laslett GM and Olley JM, 2000. Distinguishing dose populations in sediment mixtures: a test of singlegrain o ptical dating procedures using mixtures of laboratory dosed quartz. Radiation Measurements 32: 459 â€“ 4 65, DOI 10.1016/S13504487(00)001049 . Smedley RK, Duller GAT, Pearce NJG and Roberts HM, 20 12. Determining the Kcontent of single grains of feldspar for luminescence dating. Radiation Measurements 47: 790â€“ 7 96, DOI 10.1016/j.radmeas.2012.01.014. Sohbati R, Murray AS, Buylaert J P, Ortuo M, Cunha PP and Masana E, 2012. Luminescence dating of Pleistocene alluvial sediments affected by the Alhama de Murcia fault (eastern Betics, Spain) â€“ a comparison between OSL, IRSL and postIR IRSL ages. Boreas 41: 250 â€“ 2 62, DOI 10.1111/j.15023885.2011.00230.x. Specht E, 2012. The best known packings of equal circles in a circle. WEB site: . Accessed 2014 July 24. Spooner NA, 1992. Optical dating: preliminary results on the anom alous fading of luminescence from feldspars. Quaternary Science Reviews 11: 139â€“ 1 45, DOI 10.1016/02773791(92)90055D. Spooner NA, 1994. The anom alous fading of infrared stimulated lum inescence from feldspars . Radiation Measurements 23: 625â€“ 6 32, DOI 10.1016/1350 4487(94)901112 . Spooner NA and Allsop A, 2000. The spatial variation of do se rate from 90Sr/90Y beta sources for use in luminescence dating. Radiation Measurements 32: 49â€“ 5 5, DOI 10.1016/S13504487(99)002528 . Stone AEC and Bailey RM, 2012. The effect of single grain luminescence characteristics on single aliquot equivalent dose estimates. Quaternary Geochronology 11: 68 â€“ 7 8, DOI 10.1016/j.quageo.2012.03.014. Thiel C, Buylaert J P, Murray A, Terhorst B, Hofer I, Tsukamoto S and Frechen M, 2011. Luminescence dating of the Stratzing loess pr ofile (Austria) â€“ Testing the potential of an elevated temperature post IR IRSL protocol. Quaternary International 234 : 23â€“ 3 1, DOI 10.1016/j.quaint.2010.05.018. Thomsen KJ, Murray AS and Btter Jensen L, 2005. Sources of variability in OSL dose measurements using single grains of quartz. Radiation Measurements 39: 47â€“ 6 1, DOI 10.1016/j.radmeas.2004.01.039. Thomsen KJ, Murray AS, Btter Jensen L and Kinahan J, 2007. Determination of burial dose in incompletely bleached fluvial samples using single grains of quartz. Radiation Measurements 42: 370â€“ 3 79, DOI 10.1016/j.radmeas.2007.01.041. Thomsen KJ, Btter Jensen L, Jain M, Denby PM and Murray AS, 2008a. Recent instrumental developments for trapped electron d osimetry. Radiation Measuremen ts 43: 414â€“ 4 21, DOI 10.1016/j.radmeas.2008.01.003. Thomsen KJ, Murray AS, Jain M and Btter Jensen L, 2008b. Labor atory fading rates of various luminescence signals from feldspar rich sediment extracts. Radiation Measurements 43: 1474â€“ 1 486, DOI 10.1016/j.radmeas.2008.06.002. Thomsen KJ, Murray AS and Jain M, 2011. Stability of IRSL signals from sedimentary Kfeldspar samples. Geoch ronometria 38(1): 1â€“ 1 3, DOI 10.2478/s133860110003z . Gabor A, Panaiotu C, Cosma C and van den Haute P, 2012. Testing the potential of elevated temperature postIR IRSL signals for dating Romanian loess. Qu aternary Geochronology 10: 75 â€“ 8 0, DOI 10.1016/j.quageo.2012.02.014. Vegas J, RuizZapata B, Ortiz JE, Galn L, Torres T, Garca Corts A, GilGarca MJ, PrezGonzlez A and Gallardo Milln JL, 2010. Identification of arid phases during the last 50 cal. ka BP from the Fuentillejo maar lacustrine record (Campo de Calatrava Volcanic Field, Spain). Jour nal of Quaternary Science 25(7): 1051 â€“ 1 062, DOI 10.1002/jqs.1262. Wallinga J, 2002. Optically stimulated luminescence dating of fluvial deposits: a review. Boreas 31: 303 â€“ 3 22, DOI 10.1111/j.15023885.2002.tb01076.x. Wallinga J, Bos AJJ, Dorenbos P, Murray AS and Schokker J, 2007. A test case for anomalous fading correction in IRSL dating. Quaternary Geochronology 2: 216â€“ 2 21, DOI 10.1016/j.quageo.2006.05.014.
C. Burow et al. 125 Wintle AG, 1997. Luminescence dating: laboratory procedures and protocols. Radiation Measurements 27(5/6): 769â€“ 8 17, DOI 10.1016/S13504487(97)002205 . Wintle AG and Murray AS, 2006. A review of quartz optically stim ulated luminescence characteristics and their relevance in singlealiquot regeneration dating protocols. Radiation Measurements 41: 369â€“ 3 91, D OI 10.1016/j.radmeas.2005.11.001. Wood RE, Barroso Ruz C, Caparrs M, Jord Pardo JF, Santos BG and Higham TFG, 2013. Radiocarbon dating casts doubt on the late chronology of the Middle to Upper Palaeolithic transition in sout hern Iberia. Proceedings of the National Academy of Sciences of the United States of America 110(8): 2781 â€“ 2 786, DOI 10.1073/pnas.1207656110. Yoshida H, Roberts RG, Olley JM, Laslett GM and Galbraith RF, 2000. Extending the age range of optical dating using single â€˜supergrainsâ€™ of quartz. Radiation Measurements 32: 439 â€“ 4 46, DOI 10.1016/S13504487(99)002875 . Zhao H and Li SH, 2005. Internal dose rate in Kfeldspar grains from radioactive elements other than potassium. Radiation Measur ements 40: 84â€“ 9 3, DOI 10.1016/j.radmeas.2004.11.004. Zilho J and Villaverde V, 2008. The Middle Paleolithic of Murcia. Treballs dâ€™Arqueologia 14: 229 â€“ 2 48. Zilho J, Angelucci DE, BadalGarca E, dâ€™Errico F, Daniel F, Dayet L, Douka K, Higham TFG, Martnez Snchez MJ, Montes Bernrdez R, Murcia Mascars S, PrezSirvent C, Roldn Garcaj C, Vanhaerenk M, Villaverdec V, Wood R and Zapata J, 2010. Sy mbolic use of marine shells and mineral pigments by Iberian Neandertals. Proceedings of the National Academy of Sciences of the United States of America 107(3): 1023 â€“ 1 02 8, DOI 10.1073/pnas.0914088107. Zilho J, Angelucci DE, Burow C, Hilgers A, Kehl M, Villaverde V, Wood R and Zapata J, 2012. From Late Mousterian to Evolved Aurignacian: New Evidence for the Middleto Upper Paleolithic Transition in Mediterranean Spain. Abstracts European Society for the Study of Human Evolution, Bordeaux: 176.